UC San Diego

Solar and wind energy have the potential to power the world's energy needs. However, the variable and uncertain power generation from these sources are posing a major challenge for the reliable and economic integration in the existing electric power system. For solar energy, the problem consists of two related parts, (1) variability in the resource (determined by the location of a solar plant) and (2) uncertainty in power output, (determined by the local meteorological conditions). First, this work presents a verification of the accuracy of satellite image based irradiance models, used to globally assess the solar resource. The focus is placed on the direct normal irradiance (DNI) component of solar radiation and its variability. Second, we develop two solar forecasting methods, necessary for grid integration and market participation of solar energy generators. For intra-day forecasting, a satellite imagery based global horizontal irradiance (GHI) forecast methodology is proposed. For day -ahead forecasting, we present a numerical weather prediction (NWP) based model to predict hourly values of DNI, necessary for power output scheduling of concentrated solar power (CSP) plants. The proposed day-ahead forecast is extensively validated for regions in North America with high and medium potential for the deployment of CSP. The benefits of this forecast for large scale grid integration of CSP plants, combined with optimized siting to reduce variability and uncertainty, are shown. Results include the quantification of errors in satellite based DNI assessment, the successful application of cloud tracking in satellite images for forecasts up to 3h ahead and the significant reduction of power output uncertainty for day- ahead market participation of CSP plants