Wind farm production estimation under multivariate wind speed distribution

An efficient wind power generation assessment needs a proper probabilistic characterization of the wind sources, also taking into account correlation between wind speeds at different sites. Several models for wind speed (WS) were proposed in literature, such as Weibull, Gamma, Burr, etc. Different models, although capable of a global good fitting of the WS data at a certain site, may lead to rather different estimates of the producibility. To demonstrate this, the impact of various multivariate WS distributions on line transfer capability is evaluated in the paper. This is done by comparing the distribution quantiles of the whole wind farm producibility, estimated through adequate numerical methods and extensive Monte Carlo simulations.