Parsimonious modeling of wind turbine power curves with explainable parameters

The problem of approximating the shape of wind turbine power curves through analytical formulas has been broadly investigated, resulting in a plethora of alternative expressions of varying complexity, ensuring satisfactory fitting across different curve geometries. However, it seems that minimal attention was paid to the issue of parameter explainability, while most evaluation studies employed so far rely on small samples or specific systems in situ. Taking advantage of a flexible parsimonious relationship from the family of logistic functions, this research aims at investigating whether its two shape parameters depend, and eventually can be inferred, on the grounds of easily retrievable information, in terms of essential properties of commercial wind turbines (rated power, diameter, characteristic wind speed values). This is formalized through a twofold analysis, namely the local calibration of the proposed formula against a large sample of 150 commercial models extending over all available scales, and the global calibration stage, resulting in empirical relationships across three clusters of turbine scales, which allow for direct estimation of reliable parameter values. The derived tools and lessons learned offer several practical implications and future research perspectives, towards the overall goal of improving strategic planning, feasibility assessment and technical design of wind energy systems.