SOLAR IRRADIANCE FORECASTING AND MODELING FOR SMART CITIES WITH HIGH PHOTOVOLTAIC PENETRATION

Abstract

This thesis focuses on the development of solar irradiance forecasting methods. Spatio-temporal modeling and forecasting of irradiance will be essential requirements for the management of electricity grids in all cities with high photovoltaic (PV) penetration. Temporal variability of the harvestable solar energy on earth remains a great challenge facing widespread deployment of PV. To increase solar energy competitiveness, this work aims to provide tools to mitigate the problems caused by variability. Accurate near future forecasts and predictions of geographical smoothing effects of distributed systems over a certain area enhance the energy manageability, but the precise manner in which this occurs must be quantified. In this thesis, very short-term (≤ 5 minutes ahead) solar irradiance forecasting and modeling of irradiance geographical smoothing are performed using the anisotropic kriging method. Probabilistic forecasts on accumulated irradiance (6 and 12 hours ahead) are also generated using various ensemble methods. Finally, preliminary results of day-ahead irradiance forecast in Singapore using the Weather Research and Forecasting (WRF) model are presented.