Current signature analysis to diagnose doubly-fed induction generators of in-service wind turbines

• Autores: Estefania Artiago
• Directores de la Tesis: Emilio Gómez Lázaro (dir. tes.)
• Lectura: En la Universidad de Castilla-La Mancha ( España ) en 2018
• Idioma: español
• Tribunal Calificador de la Tesis: Rubén Puche Panadero (presid.), Fausto Pedro García Márquez (secret.), Olimpo Anaya Lara (voc.)
• Programa de doctorado: Programa de Doctorado en Ciencias y Tecnologías aplicadas a la Ingeniería Industrial por la Universidad de Castilla-La Mancha
• Enlaces
  • Tesis en acceso abierto en: RUIdeRA
• Resumen

  • Of all types of renewable energy sources, wind power has experienced the greatest growth, contributing with more new power generation than any other renewable source in the last year. A global cumulative wind capacity of nearly 540 GW was achieved by the end of 2017, of which, over 18 GW was located oshore. Under this scenario, Wind Turbine (WT) power continues to evolve, with larger machines and new oshore developments increasingly being installed. Such growth comes along with important challenges to be addressed. First, accessing offshore machines is more expensive than onshore ones and may require long periods of work. Second, larger WTs exhibit more faults when compared to smaller ones. These arguments, together with the ageing of the European WT fleet, mean the development of new maintenance strategies have become the focus of attention in recent years.

    Operation and Maintenance (O&M) activities are key to improving reliability and availability of WTs, but the costs these entail account for up to 25% of the total costs of a wind farm project. This percentage can rise up to 35% in the case of offshore developments. Access and repair activities are the leading categories of overall O&M costs. Onshore availability has been found to be over 97%, achieved through appropriate O&M activities.

    Dierent numbers are obtained for offshore sites, of about 85% availability, due to inaccessibility issues and the current uncertainty surrounding offshore failure characteristics. Three main types of maintenance strategies are commonly implemented, based on preventive, corrective or predictive actions. Preventive maintenance consists of a series of periodic actions independent of the WT health; while downtime is avoided, resources are used unnecessarily and heavy expenses are incurred. Corrective maintenance is applied once a failure has occurred, and hence only the required resources are used. However, this can result in excessive downtime periods. Conventional onshore O&M activities consist of a combination of preventive and corrective maintenance actions, but this approach has room for improvement and is unwise offshore. Under this scenario, the trend is moving from corrective and preventive actions towards predictive ones, where the condition of a component is continuously monitored and maintenance activities are planned accordingly, referred to in the literature as Condition Monitoring (CM).

    The principle behind CM draws on the fact that faults develop from an early (or incipient) stage to a critical condition, and, consequently, critical faults and downtime can be reduced by premature detection of anomalies.

    Nevertheless, the secrecy present in the wind energy sector makes it dicult to understand which components are critical for CM. Based on this premise, the first objective of the present Doctoral Thesis was to carry out a review of wind turbine reliability studies. Thirteen reliability studies were identified in the scientic literature, highlighting the lack of public reliability data. The differences across the studies were investigated, with great effort being made to unify the various studies to obtain comparable results. To this end, a wind turbine taxonomy common across the different studies was developed, formed by thirteen WT assemblies, and the failure rates and downtimes from each study were normalised. The results established differences between the least reliable assemblies, categorised as critical, and the most reliable ones.

    The drive train is among the top three contributors to failure rates and downtime of WTs, together with induction generators. Early detection of gear, bearing and generator faults is therefore crucial if predictive maintenance is to be implemented in order to reduce corrective and preventive actions. Early development of Condition Monitoring System (CMS) focused on vibration techniques applied on the drive train, including gearbox and bearings. More recently, Supervisory Control And Data Acquisition (SCADA)-based methods have been implemented on real-time CMSs of WTs, with further fault detection capabilities (such as pitch, yaw, and hydraulic faults). However, none of these methods provides accurate information on generator health. Current Signature Analysis (CSA) has been identied as a cost-effective and non-intrusive technique that can monitor both mechanical and electrical faults within the induction generator, as well as bearing and gearbox related faults. It can be applied to Doubly Fed Induction Generator (DFIG), which are the dominant technology employed in variable-speed WTs. Failure data analyses applied specically to in-service WT DFIGs are scarce in the scientic literature. Moreover, unlike studies based on computer simulations or laboratory benches with known and induced faults where the authors know which fault frequencies to target, a whole set of potential frequencies had to be calculated in order to diagnose operating WT DFIGs under unknown health status. Under this framework, the second objective of this Doctoral Thesis was to validate CSA on actual in-service WT DFIGs. Up to three dierent WT DFIGs were analysed, and their diagnoses successfully achieved. This outcome will therefore contribute to the development of Advanced CMSs, and therefore to improving reliability and availability of WT generators.