Condition monitoring of wind turbines, Part II: SCADA-based condition monitoring
Since 2017, it is mandatory that newly planned wind energy projects have to be announced in an auction system leading to an increasing cost pressure. Therefore every cost factor needs to be considered and reduced if possible; especially yearly expenses such as maintenance. To reduce the (un-)scheduled maintenance time and resulting down time, it is relevant to ensure a proper monitoring of the wind turbine.
Condition monitoring of technical assets aims at detecting changes and trends that represent deviations from normal operational behaviour and thus indicating a developing fault. In case of wind turbines, the monitoring of structural components as the support structure, the tower or rotor blades is often referred to as Structural Health Monitoring (SHM), while systems for monitoring other components, like e.g. the rotating drive train, are usually called Condition Monitoring Systems (CMS).
The utilization of CMS for condition-based maintenance (CBM) can be considered as a three-step process, illustrated below, consisting of data acquisition (1), data processing and diagnosis (2) and residual-life prognosis (3). Due to the fact that the interpretation of CMS data requires specific expert knowledge, condition monitoring is mainly carried out from dedicated service centres.
Three-step process of CMS-data utilization for condition-based maintenance
In the VGB research project 383 Condition Monitoring Systems of Wind Power Plants: Status Quo, User Experience, Recommendations, Part I, a study was carried out to identify the state of the art and main challenges related to the use of CMS for wind turbines. The main challenges can be summarized as follows:
- Sensor reliability and accuracy
Little to no trustable performance data of CMS / SHM sensors is available.
- Limited fault-detection performance (false alarms, missed faults)
The detection performance of current CMS is often not optimal. Straightforward condition-monitoring approaches require less detailed turbine-specific information for diagnosis. As these models are mainly data-driven their performance is highly depending on the size of the database.
- Purely subjective prognosis of remaining useful life
The severity assessment in case of a detected fault, i.e. the prognosis concerning the allowable time to perform maintenance activities is limited to the knowledge and experience of the monitoring company as quantitative methods for this purpose have not been developed yet.
Conventional CMS and SHM require a certain amount of sensors. An additional approach would be a CBM based on Supervisory Control and Data Acquisition (SCADA) signals. It has the advantage of utilizing existing data acquisitions. Published literature states SCADA signals in a temporal resolution of ~1 Hz have considerably higher information content than 10min-averaged SCADA data. However, through statistical distribution (depending on the current operating state), the 10 min data can be extrapolated to higher resolutions as 1 Hz data leading to a representative estimation. This extrapolation as well as the use of SCADA signals as a CMS are considered an important subject for future work and are the main focus of this project.
The general objective of the project is to develop methods to reduce the levelised cost of energy of wind turbines through reduced maintenance cost by a data-driven approach. This will be achieved by "normal behaviour models" and "alarm and fault analysis" based on the historic SCADA data available of field turbines.
The project is carried out by the Center for Wind Power Drives of the RWTH Aachen University (Prof Ralf Schelenz) supported by the following VGB member companies:
- E.ON Climate & Renewables GmbH
- ENTEGA AG
- Fortum Corporation
- Innogy SE
- MVV Energie AG
- STEAG GmbH
- swb Crea GmbH
- SWM Services GmbH
- Vattenfall Nederland BV/Nuon
- VERBUND Hydro Power GmbH