The share of the global energy production coming from solar power is increasing, and forecasting solar power production is the key for a successful integration of solar power into the existing electricity grid. SOLARFOR (Solar Power Forecasting) is a flexible tool for online prediction of the solar power production coming from e.g. PV (photo-voltaic) based power systems. The tool can be used both to deliver forecasts for single solar power plant, and to deliver forecast for a large geographical region with a huge number of both small and larger solar power installations.
SOLARFOR use meteorological forecasts as input and hence the maximal prediction horizon is determined by the meteorological forecasts.
Solar power forecasting is a relatively new discipline and SOLARFOR is based on research mainly carried out on DTU from 2005 and onwards. The models have been further developed in order to precisely describe the path of the sun over the sky, support sun tracker technologies, support global meteorological models with relatively low temporal resolution, and include optimal combination of several meteorological models, statistical downscaling, shadow detection, and more.
Input to SOLARFOR
SOLARFOR requires the following input:
- Measurements of actual production, preferably received online
- Meteorological forecasts from one provider
Preferably measurements of availability and curtailment should also be supplied as input to SOLARFOR. Furthermore, SOLARFOR is able to handle scheduled availability and curtailment. In order to increase prediction performance one or more additional providers of meteorological forecasts can be used.
The data-driven approach used by SOLARFOR limits the static data requirements. The precise requirements depends on plant type, e.g. for a PV plant with fixed panels the static data requirements includes the plant capacity and the approximate orientation and tilt of the panels. For regional forecasting similar information is inferred from historic production measurements.