Influence of the fitted probability distribution type on the annual mean power generated by wind turbines: A case study at the Canary Islands

This paper aims to quantify the influence that probability distribution selected to fit wind speed data has on the estimation of the annual mean energy production of wind turbines. To perform this task, a comparative analysis between the well-known two parameter wind speed Weibull distribution and alternative mixture of finite distribution models (less simple but providing better fits in many locations) is applied, in order to contrast simplicity versus accuracy. Data fitted from a set of weather stations located at the Canary Islands and a representative sample of commercial wind turbines are taken into account to carry out this analysis. The calculations provide a wide variety of numerical results but, as a general conclusion, the analysis evidences that any improvement in wind data fits given by the use of a mixture of finite distributions, instead of the standard Weibull distribution, is partially or even totally lost as the annual mean energy production is worked out, practically regardless the weather station, the wind speed distribution model, the turbine size or the turbine concept.