This paper presents the advantages of using wind speed time series models from ARMA-GARCH class. The models are found using good statistical practice and are able to capture the most important characteristics of the data like distribution, time dependence structure and periodicity in a satisfying manner. It is shown that the models offer several crucial advantages. The artificial wind speeds simulated from the obtained models are statistically indistinguishable from the wind speed time series measurements recorded in other years than the original data. Moreover, models can contribute to the considerations of extreme scenarios of wind power generation by simulating wind speeds characterized by a very high energy content. Thanks to the use of continuous cdf transformations, the synthetic time series do not possess the measurement error.