Quantitative performance evaluation of a wind turbine generator cluster using statistical techniques

Abstract

This report dissects the analysis of annual performance of a cluster of wind turbine generators operated by Suzlon Energy Limited to troubleshoot shortfalls in the predicted generation of individual machines for the fiscal year of 2011-12. The first phase involved an estimation of the annual energy production using wind resource assessment on the commercial software Wind Atlas, Analysis and Application Program (WAsP). Comparison of the Annual Energy Production (AEP) with direct extrapolation of monthly generation to centum machine and grid availability highlighted negative inconsistency between the two in the case of five machines. Detailed study of their data histories indicated the reason to be coarse extrapolation and generation curtailment, although the latter is a minor contributor. The second phase consists of quantification of these losses for the concerned machines using statistical extrapolation and a variety of data approaches. A qualitative comparison of the methods is presented based on accuracy and utility. It is concluded that correlation of average hub-height wind velocities with the concerned machine's generation yields the most reliable and professionally useful results. Also, the advantages and shortcomings of the other methods have been discussed. © 2014 The Authors.