Statistical Characteristics of Wind Power in Kenya

Abstract

The statistical characteristics of wind power in Kenya were investigated in this study with the data consisting of daily maximum, minimum and mean wind speed values from 24 sites in Kenya. The method of Principal Component Analysis (PCA) was initially applied to the wind speed data in order to determine the spatial similarity in the wind characteristics over Kenya. The results from PCA indicated that the method was able to describe the spatial patterns of the wind by some few uncorrelated factors (eigenvectors). Eleven homogeneous wind categories could be delineated from the spatial patterns of the dominant eigenvectors. Detailed characteristics of the~wind power within the eleven regions were then investigated. The second part of the study fitted several statistical models to the wind speed data. The fitted models included the Lognormal distribution with parameters 2 ~nd 3, the Pearson III and Log Pearson III distribution~ as well as the Weibull 2 and 3 parameter distributions. It was noted that t~e 3 parameter Weibull ~ distribution ·was~the best distribution since it fitted the data well at all of the locations considered. The model was subsequently used to estimate the wind power potential at the varibus homogeneous sites. It was observed from the weibull estimations that maximum wind power were located around Marsabit/ Maralal regions as well as along the coastal strip of Kenya. Substantial powers were also obtained around Nairobi and Eldoret. Tne seasonal variability of wind powers at tne various locations indicated tnat the patterns of tne .totaL wind power_closely resembled the seasonal characteristics of the winds. Finally, the variations of wind powers with neignt were examined at tne various locations. Tne best vertical wind power profile were obtained using tne tnree parameter Weibull distribution. Tnese results generally indicated tnat tne optimum level for wind power generation over Kenya was approximately Between 2S-30m above the ground level. ... Tne cost and benefit factors were however, not considered here.