Comparision of Weibull distribution parameters estimating methods for wind energy generating in East Azerbaijan Province

Introduction In order to determining wind energy potential, it is necessary to use statistical analysis. Wind energy is one of the main necessities for economical and industrial development for every society. In order to accurate assessment of wind energy features and wind potential long term records of wind speed is needed. Increase of environmental pollutions resulting from fossil fuels sources, global warming issue and green house phenomenon effect and acid rains fossil fuels and more extra attention to use renewable energies. In this regard wind as a renewable source of energy interested by many researchers in many countries. 2- Methodology There are many methods in order to estimating Weibull distribution parameters. Because these parameters were used in estimating parameters related to wind energy and determining sites which have wind energy potential, so it is important their estimating using proper method. In this paper, 5 distinct methods for estimating of parameters of Weibull distribution considered. For this purpose, 6 synoptical station which have adequate 3 hours wind speed data from 1987 to 2009 (23 years) were selected. These stations are: Ahar, Jolfa, Maragheh, Mianeh, Sarab and Tabriz. For estimating scale and shape parameters of Weibull distribution the method of moments, empirical, graphical, energy pattern factor and maximum likelihood methods were used. For determining the best parameters estimating method using cumulative distribution function of the Weibull distribution (F(v)), expected values were generated. The Chi square test was used to select the appropriate method. For determing the method which has high level of significance among other methods, CHIDIST command in Excel software was used. Graphical and maximum likelihood methods gave significance level about zero representing low level of correspondence so these methods removed from further analysis. Among the methods which gave significance level about 1, the method which had low value of Chi square were selected as the best one. Other computations related to wind energy characteristics in 10, 20 and 40 m height and wind speed values corresponding to return period of 10, 25, 50 and 100 years were analysed. 3- Discussion Among the considered methods, the method of moment dut to having high level of significance and low Chi square were selected as a suitable method for estimation of the Weibull distribution parameters. Using this method, scale and shape parameters of Weibull distribution at 20 and 40 m height was estimated, too. Then wind energy characteristics, namely, wind power density (Wm-2), wind energy density (Kwh-1m-2), the most probable wind speed (ms-1) and the maximum energy carrying wind speed (ms-1) were computed. In order to determine the station having wind energy potential the distribution parameters were computed. At monthly time scale and in 10m height, maximum value of k observed in Tabriz station on July and the lowest value of k observed in Mianeh station in January. The maximum value of c was observed at Jolfa in July and the lowest value of c observed at Mianeh in December. Such situation was observed at 20 and 40 m heights. Generally, potential of Jolfa station is in the first rank from the wind energy characteristics viewpoint in East Azerbaijan. Using the wind power density all stations were classified into distinct groups. Jolfa station and Ahar station had good power for using wind energy but the other stations namely, Mianeh, Sarab, Maragheh and Tabriz stations had poor wind power potential. In wind speed with different return periods in 10m height and monthly time scale point of view, maximum and minimum values were observed at Jolfa and Mianeh stations, respectively. In 10m height and annual time scale we can classify stations as following rank: Ahar, Tabriz, Maragheh, Jolfa, Sarab and Tabriz. 4- Conclusion We can summarize main conclusions drawn from this investigation as following: • Investigation of monthly average of wind speed in studied stations shows that it increases in all stations on June, July and August months except Ahar station. • Five methods, namely Maximum Likelihood, Method of Moment, Graphical, Empirical and Energy Pattern Factor methods were investigated and using Chi sqare test the most suitable method was selected. This method is Method of Moment. Using parameters obtained from this method, parameters related to wind energy in studied related were estimated. • When the variation of shape and scale parameters during the yearis high, then the parameters related to wind energy vary more. For example, in Ahar station parameters related to wind energy don’t vary during the year. • The highest value of wind power density equal to 300 Wm-2 observed in Jolfa station and this station has a good situation in wind energy harnessing. • Wind speed analysis in different return periods showed that in monthly time scale, the highest value of wind speed will occur in Jolfa station.