WindCast - Advanced Forecasting Methods for Efficient Wind Power Market Integration

• Applicants: Professor Dr. Carlo L. Bottasso; Professor Dr. Christoph Goebel
• Subject Area: Electrical Energy Systems, Power Management, Power Electronics, Electrical Machines and Drives
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 569719872

To be profitable in modern electricity markets, energy producers need to rely on accurate forecasts, because unprecise bids may come with heavy economic penalties. While forecasting the power generation of fossil-based plants on day-long horizons is trivial, it poses a formidable and still open scientific challenge for renewable and highly volatile energy sources like wind. In fact, wind power forecasting needs to account for a multitude of complex and interconnected factors. While wind speed is the primary driver of power generation, several other atmospheric parameters play determining roles and need to be predicted with accuracy: wind direction, because it affects wake trajectories; atmospheric stability, because it affects the speed of recovery of wakes; temperature, because it affects air density; precipitation, because it may trigger shutdowns to mitigate blade erosion. Furthermore, bats and birds active in the proximity of a wind power plant can require shutdowns. The accurate prediction of all these parameters not only enables the forecasting of energy production, but also of the services that a wind plant can deliver to the grid. In fact, due to its growing importance in the energy mix, wind power is increasingly being called upon to play a more active role in ensuring the secure, efficient, and stable operation of the grid. This includes providing reserve margins, inertia, frequency support, and black-start capabilities. The provision of these services generates additional revenue streams for operators but can also be a prerequisite for grid connection. The combination of all these factors clearly highlights the highly interdisciplinary scientific challenge of developing forecasting methods enabling participation of wind power plants in modern electricity markets. The objective of the WindCast project is to advance the state of the art in wind power forecasting by developing holistic forecasters capable of predicting the power generation ability of wind power plants on day(s)-ahead and intra-day horizons. Using advanced physics-informed machine learning, these forecasters will generate probabilistic forecasts of all plant-internal and external factors that influence power generation, including atmospheric conditions and environmental wildlife-related effects, as well as generated wind power itself. The open-source WindCast holistic forecasting tools will enable the wind industry to take full part in modern electricity markets, ensuring profitability and facilitating the further penetration of wind in the energy mix.

• DFG Programme: Research Grants