Back to the programme printer.gif Print

Thursday, 13 March 2014
11:15 - 12:45 Advanced operation & maintenance
Science & Research  

Room: Llevant
Session description

The session covers the entire area of wind farm and wind turbine operation and maintenance, e.g. how to access, repair and organise operation and maintenance logistics onshore and offshore. In order to keep in hand the current health of turbines and farms, failure detection, identification and prognosis methods are also presented. Maintenance operations are also addressed from the viewpoints of required activities and efficiency. In order to cover management aspects, operation and lifetime cost calculation methodologies are also introduced. Experts from various European countries share their results during the session.

Learning objectives

  • Advanced operation and maintenance
  • Fault detection methods
  • Reliability calculation techniques
  • Monitoring on the field
  • Statistical and artificial intelligence-based solutions for diagnostics data- and model-based solutions for fault detection
Lead Session Chair:
Zsolt Viharos, Hungarian Academy of Sciences, Hungary

Christopher J. Crabtree , University of Durham, United Kingdom
Christopher Crabtree Durham University, United Kingdom
Christopher Crabtree (1) F P Donatella Zappal (1) Peter Tavner (1) Simon Hogg (1)
(1) Durham University, Durham, United Kingdom

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Christopher Crabtree is Lecturer in Wind Energy Systems in the School of Engineering and Computing Sciences. Having received the MEng in Engineering from the University of Durham in 2007, he began a PhD in the School of Engineering and Computing Sciences at Durham and graduated with a thesis on condition monitoring techniques for wind turbines. His research focuses on operation and maintenance aspects of wind energy, onshore and offshore, with regard to improving reliability and availability, and reducing the cost of energy from wind through condition monitoring. He is currently involved in the UK EPSRC Supergen Wind Energy Technologies Consortium.




1. As large-scale wind farms move further offshore into more inhospitable environments, achieving a high availability and capacity factor is an essential in ensuring a competitive cost of energy. The cost of operations and maintenance (O&M) has been shown to be anything between 15% and 35% of the cost of energy from wind making this a clear target for cost reduction. One approach to reducing the cost of O&M is to move away from reactive maintenance strategies to planned, proactive and preventative strategies.


This requires the use of remote condition monitoring (CM) of the individual wind turbines (WT) in informing operators of the health of each WT at any point in time. In order to allow for planned maintenance, a CM system (CMS) must be able to indicate the severity of a fault so that a judgement can be made as to when maintenance should take place. A particular challenge facing wind farm operators is that of automation of fault detection as manual interpretation of large amounts of data from multiple WTs is costly. Ideally, an automated system should present a clear detection or health signal to the operator who could then choose to examine the WT signals in more detail, if required.

This paper applies a frequency tracking algorithm to mechanical and electrical monitoring signals from a small-scale test to examine how the various fault signatures compare between signals when generator electrical faults are present. In particular, the relative sensitivities of the various signals are examined.

2. Monitoring of Electrical Faults

Being such a key drive train component, faults in a WT’s generator can have catastrophic effects that result in costly and lengthy repairs, particularly offshore. Nevertheless, monitoring of electrical faults in generators has not yet become standard practice in the wind industry. Reliability studies [2] [3] have shown that generator defects make a significant contribution to WT downtime with [2] showing that 30% of total annual downtime resulting from power conversion failures. Of this, 30% of the downtime resulted directly from generator failures. A recent survey of failed commercial generators [4] showed that brush-gear and slip-ring failures in wound rotor induction generators (WRIG) accounted for 16 % of 2MW range generator failures. In smaller machines, 50% of failures originated from rotor unbalance.

Main body of abstract

Generator rotor electrical unbalances such as brush-gear and slip-ring damage have been shown to exhibit certain characteristics in the stator terminal electrical current and power signals [5] [6] for machines operating at constant speed and load. These signals can be analysed by Fourier transform approaches and give clear indications of the presence of electrical faults. However, the majority of modern WTs operate at variable speed and so current and power spectra change rapidly over time making conventional analysis challenging. Research presented in [7] describes a frequency tracking approach which can be applied to non-stationary signals on the assumption that they are effectively stationary if examined over very short periods. The algorithm is discussed in detail in [7] but can be summarised as extracting solely the signal’s amplitude corresponding to a particular frequency of interest. In WRIGs, electrical rotor unbalances manifest themselves as a function of the machine slip as shown in Table 1 where s represents the per unit slip, fs is the stator supply frequency (50 Hz) and fc is the fault frequency of interest.

Table 1: Fault frequencies in stator electrical signals

In the majority of modern WTs, however, rotor current signals are only monitored for control purposes and operators often have difficulty in obtaining permission to use these signals for condition monitoring purposes. However, mechanical signals are often recorded by the CMS. Electrical faults effectively manifest themselves as torque pulsations as they pass through the stator magnetic field and so can be expected to produce mechanical vibration signals. In the case of rotor electrical unbalance, the fault manifests itself in speed and torque signals at the same frequency as in the stator total power signal ().

3. Physical Test Rig

Since CM data from large scale operational WTs is no readily available, due to operators’ concerns about confidentiality, the data used in the paper is recorded from a physical test ring at Durham University. Details of the test rig are given in [7] and [8]. The test rig features a grid-connected 30kW WRIG that is driven by a DC motor according to speed profiles derived from a WT model. The driving conditions used in this paper are shown in figure 1. Rotor electrical faults are introduced by external resistances. A base resistance of 1.3Ω is included to allow for a wide generator speed variation and faults of 23% and 46% are introduced to represent the development of rotor electrical unbalance faults.

Figure 1: Generator variable speed profile

4. Results and Discussion

4.1. Analysis of Electrical Signals

To verify the approach, measurements for the generator stator total power and line current were taken. The frequency tracking algorithm was applied to search for frequency content as defined in Table 1. Three fault levels were applied: healthy, 23% unbalance and 46% unbalance. The result for total power is shown in figure 2.

Figure 2: Frequency tracking of the stator total power signal

The point at which the fault level is changed is very clear for both the 26% and 46% unbalances. This follows from earlier work and suggests that the data is valid. It should be noted that the data was recorded in a noisy test environment as would be experienced on an operational WT so, keeping this in mind, the clarity of the results is very positive. The line current signal results are not given in this abstract due to space limitations but were found to be similar to the total power signal but with a slightly higher noise level.
As would be expected, the electrical unbalances are clearly visible in the total power signal because of the direct link between the fault itself and the measured signals.

4.2. Analysis of Mechanical Signals

The same approach is now applied to the mechanical torque and speed measurements. The power frequency from Table 1 was taken as the fault frequency of interest in power speed and torque.
The frequency tracking result for the machine speed is shown in figure 3.

Figure 3: Frequency tracking of the generator speed signal


As expected, the speed signal is extremely noisy in comparison to the power. Nevertheless, a change is visible in the fault frequency magnitude as the fault increases in severity. Given the low power levels applied to the test rig, in the 5kW region, it is not unsurprising that this mechanical signals has not responded as strongly as the total electrical power signal. However, the electrical power fault frequency stems from an induced pulsation in torque as the unbalance develops so it would be expected that the torque signal should offer a stronger detection signal that the speed. The frequency tracking result for generator shaft torque is shown in figure 4.

Figure 4: Frequency tracking of the generator shaft torque signal

The detection signals shows a response which, whilst not as clear as the power signal in figure 2, is distinct and clearly defined. It is expected that when higher generator powers are experienced, the torsional response will increase significantly.

5. Conclusions

This paper presents work completed at Durham University on the use of electrical and mechanical signals for the detection of generator electrical unbalances. It can be concluded that:
• Generator electrical unbalances cause electrical and mechanical torque pulsations that should be detectable in both electrical and mechanical systems.
• Generator electrical signals give the most distinct response to changes in fault magnitude but that these signals are not widely available in existing monitoring systems.
• Generator speed measurements show a change as a fault is introduced but the low power levels on the test rig seem to be limiting the magnitude of this response.
• Generator shaft torque measurements show a distinct change when an electrical fault is present and this would be expected to increase in magnitude for higher power machines.


This work was funded under the UK EPSRC SUPERGEN Wind Energy Technologies programme, EP/H018662/1.

Learning objectives
This paper examines the important area of wind turbine condition monitoring. The paper presents the idea of frequency tracking as applied to monitoring signals for the detection of generator electrical faults.

- Condition monitoring
- Generator electrical faults


[1] Feng, Y., Tavner, P. J., Long, H., “Early experiences with UK Round 1 Offshore Wind Farms”, Proc. Institution of Civil Engineers, Energy, Vol. 163, Iss. EN4 pp. 167-181, 2010.
[2] M. Wilkinson, B. Hendriks, F. Spinato, E. Gomez, H. Bulacio, J. Roca, P. Tavner, Y. Feng, H. Long, “Methodology and Results of the Reliawind Reliability Field Study”, Scientific Proc. European Wind Energy Conf. (EWEC), Warsaw, Poland, March 2010.
[3] F. Spinato, P. J. Tavner, G. J. W. Van Bussel, E. Koutoulakos, “Reliability of Wind Turbine Subassemblies”, IET Renewable Power Generation, Vol. 3, Iss. 4, pp. 1-15, 2009.
[4] K. Alewine, W. Chen, "Wind Turbine Generator Failure Modes Analysis and Occurrence," Windpower 2010, Dallas, Texas, May, 2010.
[5] S. Djurović, S. Williamson, A. Renfrew, “Dynamic Model for Doubly-fed Induction Generators with Unbalanced Excitation, both With and Without Faults”, IET Electric Power Applications, Vol. 3, Iss. 3, pp. 171-177, 2009.
[6] Djurovic, S., Crabtree, C. J., Tavner, P. J., Smith, A. C., Condition Monitoring of Wind Turbine Induction Generators with Rotor Electrical Asymmetry, IET Renewable Power Generation, Vol. 6, Iss. 4, pp. 207-216, 2012.
[7] Crabtree, C. J., Tavner, P. J., Condition Monitoring Algorithm Suited to Wind Turbine Use, IET Renewable Power Generation Conference, Edinburgh, September, 2011.
[8] Zappala, D., Tavner, P. J., Crabtree, C. J., Sheng, S., Sideband Algorithm for Automatic Wind Turbine Gearbox Fault Detection and Diagnosis, Proc. Scientific Track, EWEA 2013, Vienna, February 2013.