Effects of numerical weather predictions on wind power forecasts

Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Graduate School of Natural and Applied Sciences, Graduate School of Natural and Applied Sciences, Turkey

Approval Date: 2018

Thesis Language: English


Principal Supervisor (For Co-Supervisor Theses): İsmail Yücel

Co-Supervisor: Ramazan Sarı


Wind energy investments are rapidly increasing in Turkey. The prediction of the electrical power generated from the wind is also gaining importance in this field because of the complexity of meteorological parameter wind. In this context, Wind Energy Monitoring and Forecasting Center (RITM) project has been initiated within the scope of the General Directorate of Renewable Energy (YEGM), in 2010. The final hourly wind energy predictions are generated by using the combination of the production data from Wind Power Plants and different numerical weather prediction models with this project. In this Thesis Study, 6 Wind Power Plants are selected according to their high wind potential and their terrain structure (complex or flat) from 3 geographical regions (Marmara, Mediterranean, Aegean) in Turkey. The terrain structures of Wind Power Plants are determined by using Geographical Information System Models which give two maps: digital elevation and roughness. The long term (3-4 year) observed wind speed data of the wind power plants from each region are compared with 3 different Meso Scale Numerical Weather Forecast Model (ECMWF, GFS, ALADIN) outputs and final wind power predictions which mean a combination of RITM power forecast system, compared to actual energy productions. The analyses are made for diurnal, seasonal, monthly basis and different grid points that belongs to each NWP model. Obtained results which is determined by using RMSE, bias and Correlation Coefficients for each time scales are used for determining best grid points for each model. This study aims to compare the performance of each Numerical Weather Prediction Models in the RITM system which has different terrain and climate structures, at different time scales and at different energy thresholds. In addition to numerical weather prediction analysis, Turkish Electricity Market prices according to Renewable Energy Supporting Mechanism and Day Ahead Market Prices have also been calculated for 6 wind power plants in this study in order to research effects of wind power forecasts to the income and market prices. It is foreseen that the study will research and analyze the performance of different numerical weather forecasts in the wind forecasting system of different climate and terrain conditions and importance of wind power forecasts in electricity market.