Generating synthetic power curves from real turbine data: a scalable approach based on specific power density

Accurate modeling of wind turbines requires precise power curves, which often remain unavailable for innovative turbine technologies, prototypes, or undocumented plant configurations. This study develops a data-driven methodology for generating synthetic power curves that serves as a foundation for advanced energy system analyses and wind farm planning. Based on 509 quality-filtered real power curves, we develop a two-stage approach that creates weighted reference curves from similar turbines and subsequently applies physically-based scaling via specific power density. Our methodology achieves an RMSE of 2.14% with global training data and up to 1.27% with manufacturer-specific data, representing significant improvements over established physical methods (Sørensen 4.56% RMSE, Saint-Drenan 2.37% RMSE). The consistent performance across all technology segments from 150–650 W/m2 specific power density enables precise applications in energy system modeling, grid planning optimization, and economic evaluation of wind energy projects. This foundational methodology can continuously expand with new market data, thus providing a robust basis for complex energy system analyses and future scenarios.