Stochastic study of extreme wind velocity values on global scale using K-moments

Currently, more countries make a swift in the renewable energy sources to reduce the environmental impact of the use of fossil fuels. The wind energy has a significant position in this hierarchy, as is one of the most efficient to be converted to electric energy, covering the society’s needs in the fields of transportation, trade, and consumption. From the other hand, extreme-oriented wind speed could be the cause of life-threatening phenomena, such as tornados and typhoons. Assessment of extremes in hydrological processes is crucial (extreme rainfalls, floods) as well as in engineering design when choosing the appropriate return periods. As a result, the critical importance when choosing the most suitable distribution to imitate the behaviour of wind speed is crucial, when focusing on extremes which also correspond to high return periods. A variety of distributions, from the literature is used, and by the estimations of the goodness of fit, it seems that the Pareto-Burr-Feller (PBF) distribution fits better to data. Additionally, the usage of Κ-moments “which are particularly strong for an extreme-oriented modelling” (Koutsoyiannis 2019) is used to evaluate the parameters that focus on the extreme wind speed values of the dataset. The analysis for all available stations on global scale and especially those of which the number of data is more than 30 years of observations, is coupled with the parameters from K-moments, to help us to transform it into PBF with 2 constraint parameters. The proposed distribution with only 1 free parameter is compared with the Weibull and the Rayleigh distributions, that are proposed in the IEC 61400-1. Finally, all the above distributions are compared, in terms of the energy production estimate, among three wind turbines of different wind classes (more efficient in high, medium, and low wind speed, respectively).