Conference programme

Back to the programme printer.gif Print

Thursday, 19 November 2015
17:00 - 18:30 Condition-based decision support
O&M & logistics  
Onshore      Offshore    

Room: Montparnasse

This session will describe current and explore application of new methods and sensors for the condition monitoring of wind turbines. Laboratory and field-based technologies will be explored and their operational effectiveness and expected added value quantified.

Learning objectives

  • Analyse new methods for monitoring turbine health and performance
  • Quantify the operational effectiveness of condition-based monitoring
  • Explore new sensors for turbine health monitoring
  • Quantify the value of prognostics
Lead Session Chair:
Simon Watson, Loughborough University, United Kingdom
Jan Helsen OWI-lab/VUB, Belgium
Jan Helsen (1) F Wout Weijtjens (1) Gert De Sitter (1) Christof Devriendt (1)
(1) OWI-lab/VUB, Brussel, Belgium

Share this presentation on:

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2015 and are published here unedited

2007: graduated as Master in Engineering Sciences: Mechatronics
2007-2012: Phd on the dynamics of high power density wind turbine gearboxes funded by IWT and ZF Wind Power. Focus on multibody simulations of wind turbine drivetrains
2012-2014: Post-doc on model based monitoring of wind turbine drivetrains. Funded by ZF Wind Power and Siemens
2014-... Coordinator drivetrain monitoring at OWI-lab


SCADA analysis for Condition Monitoring


There is increasing attention for condition monitoring techniques to minimize the influence of downtime on turbine revenue. The goal is to get early warning about failing subcomponents such as gearboxes, main bearings and generators in order to cluster maintenance actions during low wind days. Thus minimizing the influence of the downtime due to the corresponding repairs on total turbine production. A possibility to get the necessary insights in the initiation of these failures is equipping turbines with oil monitoring and vibration monitoring systems. These insights can be used in the definition of alarms. However, these devices have a cost and need to be retrofitted. To minimize the cost of Condition Monitoring Systems (CMS) this paper proposes the use of the already available Supervisory Control and Data acquisition (SCADA) system. Both sensor data and status codes are used.


This paper discusses an integrated approach to predict failure in wind turbine drivetrain components by means of available SCADA data. All SCADA sensor data and SCADA status codes are integrated in one coherent dataset. Data-models are used to represent healthy turbine behavior. Statistical tools detect failure initiation. Turbine status codes are used to link operating conditions to failure initiation.

Main body of abstract

To describe healthy turbine behaviour a data-model is constructed on the historical turbine data. Automated algorithms are used to find the optimal model definition period for each turbine in the farm. Due to e.g. wake effects each turbine in the farm has unique behaviour. Therefore, it is necessary to construct dedicated models for each turbine to represent healthy behaviour. This paper illustrates these differences by means of real life wind farm data. Moreover, it discusses three metrics to assess the quality of the constructed models.
Once failure is detected it is necessary to gain insights in failure propagation. Alarm propagation is not straightforward to assess. This paper discusses different alarm metrics based on the physical propagation of the fault and illustrates these for a specific real life example: detection of a bearing fault 6 months prior to failure.
Once the failure initiation is detected it is necessary to link this initiation to operating conditions. The paper suggests the use of the turbine status codes. Turbine events directly prior to failure initiation are investigated to gain insights in the failure initiation process.


This paper showed an approach to use the available SCADA data for the prediction of mechanical drivetrain failures. It illustrated the methodology to automatically create and optimize the data-model creation process and corresponding validation. Moreover, it discussed the use of status codes to link operating conditions to failure initiation.

Learning objectives
Understand the potential of SCADA based analysis for CMS analysis
Show the need for individual turbine models within the farm
Develop techniques to assess failure propagation
Use status codes to understand failure initiation