Power curves are crucial for describing the relationship between wind speed and the electrical output of wind turbines, forming the foundation for energy yield assessments, grid integration models, and system planning studies. In these analyses, planners or researchers often rely on turbine types that do not yet exist, whether due to projected technology development, regional adjustments, or design variations. To enable realistic modeling under these conditions, we present a scalable methodology for generating synthetic power curves based on real power curve data.

Our approach utilizes a dataset of over 800 real-world power curves from various manufacturers and turbine types, which were harmonized, normalized, and categorized according to rotor area, rated power, and wind speed resolution. We developed a transformation method that allows the derivation of a synthetic curve from any given reference curve using a scaling function. One core assumption is that the wind speed axis shifts proportionally to the cube root of the ratio between the specific power densities of the reference and target turbines.

This method directly addresses a critical gap in system modelling as it enables to create realistic and physics-informed power curves for hypothetical or future turbine designs, grounded in empirical turbine data. In contrast to many current approaches, which attempt to derive generalized turbine behaviour from first-principles models or simplified aerodynamic assumptions, our method leverages observed power curves and systematically scaled specific power density. Existing methodologies often rely on coarse or idealized input parameters, which may not capture the nuanced design variations or site-specific performance characteristics of real turbines.

By using a large dataset of actual turbine performance curves and applying strict filtering criteria, our approach ensures both technical realism and economic relevance. It enables the generation of synthetic curves for high- or low-wind-class turbines, manufacturer specific, emerging rotor configurations, or upscaled designs based on proven concepts, thereby improving the fidelity and applicability of future-oriented energy system models.

The resulting synthetic curves maintain physical consistency while offering the flexibility to adapt to technological developments. They can be seamlessly integrated into grid integration studies, GIS-based siting tools, and market modeling frameworks. Moreover, the method enables automation and opens pathways for machine-learning-assisted turbine modeling.

As a further outcome, we plan to publish the methodology as an open-source tool, potentially as a Git-hosted repository, providing researchers and practitioners with a transparent and extensible framework for the creation and comparison of wind turbine power curves.