Performance evaluation of a wind farm using different power density distributions

Abstract

ABSTRACT Several wind resource assessment studies showed that the various power density functions (PDFs) were not always able to reproduce the wind regime available in different regions. Moreover, the impact of the PDFs on the assessment of wind farm performance was not widely investigated. Thereby, in the present study, the generalized extreme value (GEV), Weibull and Normal distributions are considered to investigate whether the expected performances of an actual wind farm are well predicted. This study is applied to Kaberten wind park located at Adrar, in south of Algeria. First, the selected PDFs are applied to analyze wind data collected at the meteorological station of Adrar. Then, the expected annual energy production (AEP), capacity factor and number of operating and non-operating hours of the wind park are computed using the PDFs. Thereafter, these predicted performances are compared to actual data of Kaberten wind park, and errors are assessed for every PDF. The results show that the errors on the mean wind speed are less than 0.5% for the three PDFs, whereas the available wind power density relative error varies from about 2.5% for Weibull to 20.9% for GEV, while it is about 5.9% for Normal distribution. Nevertheless, the AEP errors vary from about 8% for the GEV distribution to more than 24% for the two other distributions. Thus, these results indicate that for the studied case, the three PDFs are comparable for estimating the average wind speed, while the GEV distribution is more suitable for estimating the AEP.