Improved models for increasing wind penetration, economics and operating reliability. Final report

The need for wind power prediction in order to enable larger wind power penetrations and improve the economics and reliability of power system operation is discussed. Methods for estimating turbulence and prediction of diurnal wind power prediction are reviewed from the literature. A method is then presented to predict meteorological event induced wind power variation from measurements of wind speed at reference meteorological towers that encircle all wind turbine clusters and from sites within the wind turbine clusters. The methodology uses a recursive least squares model and requires: (1) detection of event propagation direction; and (2) determination of delays between groups of measurements at reference meteorological towers and those measurements at towers in the array. Proper filtering of the data and methods for switching reference sites and delays for the transition from one frontal system to another is also discussed. The methodology is thoroughly tested on data from the SESAME array of meteorological towers in Oklahoma and at the Goodnoe Hills MOD-2 wind turbine cluster in Washington. The performance of the prediction methodology on data sets from both sites was quite good and indicates one or more hour ahead prediction of wind power for meteorological events is feasible.