Evaluation of Synthetic Wind Speed Time Series for Reliability Analysis of Offshore Wind Farms

Abstract

A method for synthesising wind speed time series (WSTS) from limited data is required that can be used for reliability examination of wind farms and maintenance strategies for a range of wind speed scenarios. Key characteristics of the wind resource need to be captured, including energy availability and maintenance weather windows. 4 WSTS simulators were used to produce synthetic WSTS based on benchmark data from a meteorological mast data at the offshore Egmond aan Zee wind farm in the Netherlands. These synthetic WSTS were compared with test criteria to determine their suitability for reliability analysis. This included comparing the synthetic WSTS to the benchmark data in terms of the energy availability in the wind and from a typical turbine, residence time at wind speeds, number of transitions between 1m/s wind speed bins, replication of seasonal characteristics including weather windows, and underlying statistical properties. Based on the chosen criteria, the most appropriate WSTS simulator was the modified Markov process. However, no modelling technique performed best against all criteria and none capture the autocorrelation function (ACF) as closely as desired. Therefore, there is scope for a more advanced technique for wind speed modelling for reliability analysis which combines the best aspects of the models used in this work.