Posters

Abstract

Rotor imbalance is a very well-known and commonly observed phenomenon that can affect a wind turbine at any time of its life cycle. Its origins can be multiple: mass imbalance on the rotor blade structure (e.g., an uneven mass distribution during manufacturing or due to gradual water penetrating) or aerodynamic imbalance caused by a pitch offset leading to an uneven torque intake on the blades. Regardless its origin, rotor imbalance systematically introduces additional vibrations on the turbine structure increasing the loads and the occurrence of rotor-related failures and ultimately reducing the turbine lifetime. Moreover, aerodynamic imbalance directly translates into under performance, a pitch offset of 1° being quantified by an annual energy production (AEP) loss ranging from 2% to 7% in the literature. There are two types of methods commonly used to detect rotor imbalance. On the one hand, onsite measurement campaign-based methods, such as photometric or laser measurements, provide accurate but one-time diagnoses. Such methods may result in production losses due to downtime during measurement. On the other hand, data-driven approaches use different signals in order to identify the signature of characteristic vibrations caused by the rotor imbalance. Although no downtime is caused, most of these approaches require the installation of specific sensors, making their large-scale deployment on thousands of turbines very challenging. The present work aims to demonstrate that standard high-frequency SCADA data can be used for rotor imbalance monitoring. We discuss how a method backed-up by scientific literature has been successfully implemented by i4SEE and how its simplicity and low requirements make it particularly suitable for large-scale deployment.