Posters

Abstract

The performance of wind turbines is expected to decline with age, similarly to what happens with most technical systems, but there are no consolidated theoretical estimates of how much this should occur. The only possibility is learning from experience, which means from data: in the latest years, this has become possible because there are plenty of wind turbines operating since a large number of years. Studies based on cumulative data provide controversial indications: on one side, there are evidences that the evolving technology mitigates the performance decline with age, on the other side it is suspected that the determining variable is the age of the asset rather than its size and the technology specifications. Basing on these considerations, the objective of the present study is developing appropriate SCADA data analysis methods for the long-term investigation of wind turbine performance, with a particular focus on the interpretation by the point of view of characteristic times and sub-components behavior. The test cases selected for this study cover different wind turbine sizes (850 kW and 2 MW) and types of control (hydraulic and electric blade pitch). The analysis of operation curves is the keystone for a meaningful interpretation of the lifecycle of wind turbines: in particular, curves related to the blade pitch, the rotor speed and the power are fundamental. In this study, the operation curves are analyzed qualitatively through the method of bins and quantitatively through data-driven non-linear regressions: the rationale for the latter analysis is that, selecting opportunely the training data set and simulating the output on target data sets, it is possible to estimate the performance change from one period to another. Applying the above method to several operation curves on the same data sets, it is possible to separate the contributions of the main sub-components to the performance decline with age. In particular, for all the considered test case it is observed that the gearbox aging does not cause performance worsening. Furthermore, it is questionable to expect that the aging of wind turbines manifests as a certain regular decline rate per year: actually, most of the considered test cases display negligible performance decline, while few are affected by remarkable worsening. The critical point is therefore the interpretation of the latter and the individuation of counteracting actions. The SCADA data analysis developed in this work support the hypothesis that the hydraulic blade pitch is a critical component which is likely associated to the most noticeable performance worsening in time. The performance degradation actually manifests mainly as a shift in the wind speed – blade pitch curve and, consequently, as diminished extracted power for given rotational speed, which is equivalent to a power coefficient – tip speed ratio curve operating at a non-optimal working point. Therefore, from the analysis collected in the present study, it arises that actions counteracting wind turbine performance decline with age should be primarily focused on solving issues dealing with the rotor: for example, mass or pitch unbalances, pitch system deterioration, systematic yaw error.