Comparison of Wind Scenarios Generation Methods-A Case Study From Ecuador

Abstract

Wind scenarios generation is a critical instrument for power systems scheduling and planning. Therefore, reliable approach methods are required to address this issue. This paper analyzes three scenarios generation methods accuracy to represent wind variability. For this purpose, a wind speed database of the Villonaco wind farm located in Loja-Ecuador is analyzed. The wind speed is characterized, and the generation of scenarios is performed through methods based on ARMA models, Monte Carlo, and Moment-Matching techniques. Quality analysis of the scenarios generated is carried out using Weibull probability plots, Kolmogorov-Smirnov Goodness-of-Fit test, and error analysis. It is found that the set of scenarios generated by the Monte Carlo method evidence better performance when both Weibull probability plots and Kolmogorov-Smirnov Goodness-of-Fit test are analyzed. However, set based on the ARMA model exhibit lower Root Mean Squared and Mean Absolute errors. Finally, the wind speed sets are transformed into wind power scenarios considering the power curve and characteristics of the wind turbines.