Comparison of seven numerical methods for determining Weibull parameters of wind for sustainable energy in Douala, Cameroon

Purpose The purpose of this paper is contribution to estimate the potential of wind energy in Douala in Cameroon, by modeling and predicting the regime of wind. The paper deals with the analysis and comparison of seven numerical methods for the assessment of effectiveness in determining the parameters for the Weibull distribution, using wind speed data collected at Douala International Airport in Cameroon, in the period from September 2011 to May 2013, obtained by meteorological equipment belonging to the Laboratory of Energy Research of the Institute of Geological and Mining Research. Design/methodology/approach By using ANOVA, root mean square error and chi-square tests to compare the proposed methods, this study aims to determine which methods are effective in determining the parameters of the Weibull distribution for the available data, in an attempt to establish acceptable criteria for better usage of wind power in Douala, which is the economic capital and ought to have prominence in the use of renewable sources for electricity generation in Cameroon. Findings The study helps to determine that moment, empirical and energy pattern factor methods used to determine the shape parameter k and the scale parameter c of the Weibull distribution present a better curve fit with the histogram of the wind speed. This fact is clearly validated by means of the statistical tests. But, all the seven methods gave excellent performance. Then, k reaching levels ranging from 3.5 to 5.5 and c range from 1.7 to 2.4. Originality/value Then as far as we are concerned, for a significant contribution, it could be more effective to have a model for prediction of wind characteristics using wind data collected per hour, one at least three years. A comparison of results obtained from lots of other methods (seven in this case) is necessary before an efficient discussion. Standard deviations and errors between measured and predicted data must also be presented.