Development of a power system operator model for short-term variable renewable energy forecasting

Project Status: 

South Africa exhibits one of the most carbon-intensive electricity supply systems in the world and is currently in a transition towards one which is less carbon-intensive. During the last several years South Africa has developed 3.8 GW of combined wind power, solar photovoltaic (PV) and concentrated solar power (CSP). The South African Integrated Resource Plan foresees a further growth of these so-called VRE generation sources by 2.6 GW each year for the next 30 years. This has a massive knock-on benefit for the environment and those communities that historically suffered the adverse effects of pollution, are likely to benefit.

Eskom’s challenge, as the country’s system operator, is maintaining grid stability and the security of supply as the energy mix diversifies. In the past, renewable energy was considered an unpredictable energy source, fully dependent on the varying weather conditions and has thus not contributed to the system reliability (adequacy or security). The increasing penetration of renewable energy into the grid, makes VRE forecasting critically important in Eskom’s scheduling of load requirements and ancillary reserve power in the South African power system.

The CSIR is spearheading the effort to develop a system operator model for short-term variable VRE forecasting to assist Eskom with its short-term generation dispatch decisions.




Eskom CSIR

Contact Person

Greg Landwehr

Key Concept

VRE Forecasting
VRE Forecasting is a field of expertise that uses meteorology and machine learning to predict on an intraday and day ahead basis the amount of energy that can be supplied by such VRE generation plants. Two basic forecast models exist for consideration i.e. statistical and physical each being relevant for a particular time horizon.