Faulstich2011.pdf

WIND ENERGY Wind Energ. 2011; 14:327–337 Published online 26 July 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/we.421

RESEARCH ARTICLE

Wind turbine downtime and its importance for offshore deployment S. Faulstich1, B. Hahn1 and P. J. Tavner2 1 Fraunhofer Institute for Wind Energy and Energy System Technology, Kassel, Germany 2 Energy Group, School of Engineering, Durham University, Durham DH1 4RL, United Kingdom

ABSTRACT While the performance and the efficiency of wind turbines and their energy yields have been improved with time, their reliability still needs improvement, particularly when considering their deployment offshore. IWES has been gathering operational experience from wind turbines since 1989, being involved in different projects dealing with the topic of availability and reliability. This paper draws statistical data from Germany’s ‘250 MW Wind’ programme, evaluated by IWES. The prime objective of the survey was to extract information about the reliability characteristics of wind turbines. The main purpose of this paper is to discuss the frequency of failures and duration of downtimes for different wind turbine subassemblies based on existing onshore experience and point out the likely outcomes when turbines are deployed offshore. Copyright © 2010 John Wiley & Sons, Ltd. KEYWORDS wind turbine; reliability; mean time between failures, failure rate; mean time to repair, downtime Correspondence S. Faulstich, Fraunhofer Institute for Wind Energy and Energy System Technology, Kassel, Germany. E-mail: [email protected] Received 30 June 2009; Revised 16 June 2010; Accepted 22 June 2010

1. INTRODUCTION In the period from 1989 to 2006, a large monitoring survey for onshore wind turbines (WTs) in Europe, the Scientific Measurement and Evaluation Programme (WMEP), had been managed by IWES (The Fraunhofer IWES consists of the former Institute for Solar Energy Supply Technology and the Fraunhofer Center for Wind Energy and Maritime Technologies) under the German publicly funded programme ‘250 MW Wind’. The WMEP survey collected 64,000 maintenance and repair reports from 1500 WTs that have been captured and analysed,1 covering approximately 15,357 operational turbine-years. Hence, the WMEP database contains detailed information about both the reliability and availability of WTs. It provides the most comprehensive worldwide study on the long-term reliability behaviour of WTs and the most trustworthy characteristic reliability parameters—mean time between failure (MTBF) and mean time to repair (MTTR)— published to date. These two parameters, described in more detail in this contribution, are useful for answering the following questions: ‘how often does a WT fail?’ and ‘which WT downtimes are associated with which failure?’ The definitions used in the WMEP survey are set out in detail in the WMEP annual reports and previous publications set out in Ref. 2, but important definitions for this paper are set out in Appendix A. An incident report from WMEP containing definitions of different WT subassemblies can be found in Appendix B. Evaluations of this survey show that modern onshore WTs in Europe achieve a high availability of 95–99%, as exemplified in Ref. 3. However, despite WT technology progress, in terms of economy and performance, WT reliability has declined with growing turbine size.4 Electrical and electronic subassemblies, in particular, fail more frequently, leading to higher failure rates for WTs of higher complexity.5,6 An increasing number of failures cause unplanned downtimes up to 10 times per turbine per year,7,8 resulting in high maintenance effort and production loss. Copyright © 2010 John Wiley & Sons, Ltd.

327

Wind turbine downtime

S. Faulstich, B. Hahn and P. J. Tavner

Figure 1. Representativeness of WTs in the WMEP programme to WTs installed throughout Germany.

Unfortunately, the WMEP database does not contain sufficient information to assess fault severity by repair cost, but analysis of downtime durations may indicate fault severity, giving an indication of downtimes to be expected from offshore WTs. The representativeness of the WMEP survey population to the German WT population is exemplified by Figure 1, which shows the distribution of the WMEP WTs (15,357 turbine-years) compared with the distribution of the whole German WT population (181,560 turbine-years). This is amplified in Figure 1(a) with respect to technical concept: • stall or pitch control, • constant or variable speed, • gearbox or direct drive, and in Figure 1(b) with respect to WT location: • North German plain, • German coast, • German highlands. These demonstrate that the WMEP survey contains a range of WTs, representative of the whole German population. Based on data from Ref. 2, Figure 2 shows the failure rates and downtimes for different subassemblies from WTs in the WMEP survey. Here, the annual failure rate λ (the reciprocal of MTBF) is plotted alongside the downtime per failure, MTTR. This figure highlights in abscissa lengths the significance of failures in different WT subassemblies. Previous publications on this subject7,8 concentrate on failure rate rather than downtime, whereas it is clear from Figure 2 that both are important. From Figure 2, electrical and electronic subassemblies fail more frequently than mechanical ones, but the mechanical subassemblies experience longer downtimes. From WMEP, WT electrical and electronic subassemblies fail on average every 2–2.5 years, whereas a WT drive train, excluding the gearbox and brake, only fails every 19 years. Nevertheless, it is clear that the less frequent drive train failures result in much longer WT downtimes, and the more frequent failures of the electrical and electronic subassemblies are qualified by shorter downtimes. The total annual WT downtime, due to individual subassemblies, varies between 0.3 day (for blades) and nearly 0.9 day (for the electrical system). Currently, the wind industry focuses strongly on improving the rotor blade, gearbox and other mechanical subassembly reliability using appropriate condition monitoring systems (CMS). However, these results show that electrical and electronic subassemblies also cause significant downtimes, which will be extended in offshore application.

328

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

S. Faulstich, B. Hahn and P. J. Tavner

Wind turbine downtime

Figure 2. Reliability characteristics for different subassemblies in the WMEP programme.

2. RELIABILITY CHARACTERISTICS 2.1. Failure rate The term MTBF is frequently used to describe reliability (see the definitions in Appendix A) and is the average period between unplanned stoppages. MTBF is an average, statistical value for the time-dependent probability of a system failure. For a typical failure distribution with some variance, the MTBF represents a top-level aggregate statistic and is unsuitable for predicting specific time to failure, so it cannot give information about the time to failure for a particular turbine. In the case of an exponentially distributed failure statistic, the probability of an individual subassembly being operational for a time equal to its MTBF is only 36.8%. Thus, when using MTBF, it is important to consider certain preconditions. The MTBF is the inverse of failure rate λ, which is only applicable if the failure intensity is constant with time. The development of failure intensity with time for non-repairable systems is well known and is often described by the bathtub curve, which divides the lifetime of a technical system into three phases. A theoretical bathtub curve is shown in Figure 3. The first phase is marked by falling failure intensity due to ‘early failures or teething problems’. This is followed by a longer second phase, when failure intensity is constant due to ‘intrinsic or random failures’, which can be called failure rate. This is followed by a period of rising failure intensity as damage accumulates with operational age due to ‘wear out’. The development of failure rate with operational time for different WT subassemblies has been analysed from the WMEP programme, and the variation of failure rate for different subassemblies is clear from Figure 4. For some subassemblies, failure intensity is decreasing with time, for example, in the control system. In some cases, the failure intensity is increasing with time, for example, for the electrical system. However, the general form of the bathtub curve can be recognized from Figure 4. Early failures dominate the first year of operation before the failure rate becomes constant in the intrinsic failures phase of the curve. The wear out phase can be distinguished, for example, starting after 11 years for the gearbox, some years before the supposed of 20 year lifetime of a WT. This could be an artefact due to the smaller number of WTs in their late operating years in the WMEP programme, although it is confirmed for some subassemblies after a similar period of 10 years by results from a different onshore WT survey, Landwirtschaftskammer Schleswig-Holstein.8 Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

329

Wind turbine downtime

S. Faulstich, B. Hahn and P. J. Tavner

Figure 3. The intensity function of machinery.

Figure 4. Development of the failure rate with time of operation.

Nevertheless, failures appear to be occurring at an approximately constant failure rate over a large period of turbine life, and calculating MTBF as the inverse of failure rate λ appears reasonable. There are several factors that influence the WT failure rate,9 such as wind speed, turbine concept and climatic conditions, which should be part of any appropriate reliability analysis, and these are considered here. The dependence of reliability on wind speed is analysed in general in Ref. 10. The relation between failure rate and wind energy index is shown for a population of Danish WTs in Ref. 11, with failure rate increasing at higher wind speeds and electrical subassembly failure rates showing the strongest dependency. An overview from WMEP data of mean annual failure rate for various WT concepts and different locations, with varying climatic conditions, is shown in Figure 5. With increasing the WT concept complexity, a general trend towards a higher failure rate can be observed, which is particularly noticeable for the electric system, electronic control, sensors, yaw system, rotor blades, generator and drive

330

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

S. Faulstich, B. Hahn and P. J. Tavner

Wind turbine downtime

Figure 5. Failure rates of WTs in the WMEP with respect to technical concepts and turbine location.

train. The only downward trends observed are for the mechanical brake and hydraulic system. Other subassemblies (such as the support and housing, rotor hub and gearbox) show no significant trend. With regard to WT location, Figure 5 shows that turbines located near the coast and in the highlands suffer higher failure rates. 2.2. Downtimes, MTTR It is not possible to extract fault severity information, such as repair costs per incident, from the WMEP data. Therefore, the authors used the downtime duration to assess failure severity. A statistical value for downtime is the MTTR (see Appendix A), which is the average time that it takes for a subassembly to recover from any failure. The reciprocal of MTTR is the repair rate μ. In the following, the downtime durations are analysed in more depth. Figure 6 illustrates the relative distribution for all the downtimes of the WTs in the database, and this is asymmetric with a strong weight towards downtimes of short duration. The distribution could be approximated to a Weibull or exponential function, although tests of different equations did not lead to a satisfactory fit. However, for qualitative evaluation, the Weibull function has been used in Figure 6 because the WMEP data reflect a large number of short repair and a small number of long repair periods. In the case of many failures, only a small repair, or manual WT reset or change in the control parameters was needed to reactivate the WT. From Figure 6, time-consuming repairs with downtimes of several days appear much less frequently than would be supposed. Since the severity of a failure and its downtime depends on the affected subassembly, investigation of different downtimes was performed. It can be seen that the downtime distributions have similar shapes for all subassemblies. The classification of failures, as will be described later, has been made according to the accumulation of the downtime frequency distribution for different subassemblies, as shown in Figure 7. The heavy black line represents the distribution according to Figure 6. The dark-shaded area shows the range of distributions for complete subassemblies, such as the generator or gearbox. The wider, light-shaded area shows the range of distributions of components, such as bearings (being part of the complete gearbox subassembly), or generator windings (being part of the complete generator subassembly). From Figure 7, it can be seen that 65–85% of subassembly failures are repaired in less than 1 day, dependent on the subassembly, but in Figure 2, the average downtime per failure is 1–7 days, although many failures can be repaired in a shorter time than this. Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

331

Wind turbine downtime

S. Faulstich, B. Hahn and P. J. Tavner

Figure 6. Distribution of downtimes in the WMEP programme.

Share of faults not exceeding the downtime [%]

100

80

60

40

20

0 0

1

2

3

4

5

Downtime [days]

Figure 7. Cumulative frequency of downtimes for several subassemblies. Heavy black line: distribution as in Figure 6. (Dark-shaded area: spread of complete subassembly downtimes. Light-shaded area: spread of component downtimes.)

332

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

S. Faulstich, B. Hahn and P. J. Tavner

Wind turbine downtime

3. DIVISION INTO MAJOR AND MINOR FAILURES It is clear from the previous section that there are substantial differences in downtime between failures. In order to distinguish more severe failures from those which are less severe, the WMEP failure data have been divided into major failures (occurring infrequently but with long downtimes) and minor failures (occurring frequently but with short downtimes). A downtime duration of 1 day has been used to divide between these short and long downtimes because with a downtime longer than 1 day, it is certain that the service team will travel at least twice to the WT site, increasing production losses and costs compared with shorter downtimes failures. Therefore, • Failures with downtimes ≤1 day are assumed to be minor. • Failures with downtimes >1 day are assumed to be major. The characteristic failure rates and downtimes for these two classes of failures are shown for all subassemblies in Figure 8. The ratio between the total downtimes of minor failures to the total downtimes of major failures varies from 0.10 for the electrical system to 0.02 for the gearbox. It should be noted that the proportion of minor failures is significantly larger for the electrical and electronic subassemblies. It can be seen that electrical system minor failures cause eight times the gearbox downtime and about twice the rotor downtime. Thus, to improve the WT availability, operators should attend to minor failures, particularly concentrating on electrical and electronic subassemblies. For onshore application, minor failures (representing about 75% of the total number) are responsible for only 5% of the downtime, whereas major failures (representing 25% of failures) are responsible for 95% of the downtime. The implication for onshore WTs is that maintenance and condition monitoring effort should concentrate on the 25% of failures causing the majority of downtime. However, Figure 8 shows that these failures are not concentrated on a few subassemblies but are spread amongst a number, and work is needed to investigate these failures in more detail. Assuming the gearbox, generator, yaw system and rotor failures cause long downtimes but also high repair costs, reliability improvement and condition monitoring development onshore should concentrate especially on these subassemblies.

0,45

Electrical System

0,34

Electronic Control

0,17

0,12

0,15

0,09

Sensors

0,20

Hydraulic System

0,18

0,18

0,05

6,87

0,11 0,09

0,06 0,03

Gearbox Minor failures

Generator Support & Housing

0,15

0,08

0,14

0,5

0,25 Annual failure rate

0,01 0,58

14,34

0,01 0,46 0,01 0,30

28,01

0,17

0

0,01 0,59

18,38

0,02 0,03 0,02

Drive Train

0,02 0,28

11,86

0,17

0,07 0,04

Major failures

13,08

0,18

0,02

0,02 0,62 0,02 0,37

10,93

0,16

0,03

Rotor Blades

0,02 0,46

10,09

0,06

Mechanical Brake

0,03 0,28

5,93

0,18

0,12

0,03 0,32

6,41

0,16

0,05

Rotor Hub

0,08 0,80 0,05 0,63

Minor failures

Major failures 0,16

0,05

0,13

Yaw System

6,55

15,47

7,5 Downtime per failure [days]

15 Mean annual Downtime [days]

Figure 8. Reliability characteristics for different subassemblies in the WMEP programme dividing faults into minor and major failures. [Minor failures: 1.8 per year (75%) causing 0.3 day downtime (5%). Major failures: 0.6 per year (25%) causing 5.7 days downtime (95%).]

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

333

Wind turbine downtime

S. Faulstich, B. Hahn and P. J. Tavner

Furthermore, electronic control and electrical system subassemblies are responsible for a large portion of the total downtime shown in Figure 8. Assuming that these subassembly faults are seldom detected in advance, more detailed statistical analyses should concentrate on them.

3.1. Offshore WTs Onshore WTs suffer from a large number of faults, which are easy to resolve with a small effect on downtime. As offshore WT technology has been directly derived from onshore technology, similar faults can be expected, but under offshore conditions, the downtime due minor faults will be increased due to limited accessibility, and it is expected that increased downtime will result. The dilemma of maintaining offshore WT has already been described in Ref. 12. It has been shown that the availability of an offshore wind farm, which is depicted as a function of site accessibility, will decrease. Table I confirms that the availability of existing European offshore wind projects, based on current published data, is low and needs improvement.

3.2. Outlook The analyses presented in this paper are based on data from the onshore WMEP programme. A more detailed analysis of downtime would need to take more factors into account.9 However, because of the diversity of these factors, a subdivision of WTs into different groups could lead to single groups with an inadequate statistical basis for analysis, even from a broad database like WMEP.15 To overcome this limitation, a collaborative reliability database is proposed, using as much experience as possible, where standardized data structures are required and consideration is extended to offshore turbines. Joint activities to standardize operation and maintenance measures, documentation and data structures for onshore WTs have commenced on a national German basis. First steps have also been made for offshore WTs. A group of planners and operators has confirmed support for a new German survey to monitor the development of offshore WTs as an aid to improving the availability of offshore wind power plants. This new project is named ‘Offshore~WMEP’ following the former German monitoring survey for onshore turbines and is currently developing the concept for the project as shown in Figure 9. Table I. Availability of existing wind farms onshore and offshore. Wind farm

Distance offshore (km)

Average technical availability (%) Years of operation 1

European wind farms onshore North Hoyle, UK, offshore Scroby Sands, UK, offshore Kentish Flats, UK, offshore Egmond aan zee, Netherlands, offshore Barrow, UK, offshore

— 8 2 8.5 11

84.0 84.2 87.0 81.4

10

67.4 (Ref. 13)

(Ref. (Ref. (Ref. (Ref.

13) 13) 13) 14)

2 98.2 (average from Ref. 2) 89.1 (Ref. 13) 75.1 (Ref. 13) 73.5 (Ref. 13) — —

3

87.4 (Ref. 13) 90.4 (Ref. 13) — — —

Figure 9. Concept of the Offshore–WMEP database.

334

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

S. Faulstich, B. Hahn and P. J. Tavner

Wind turbine downtime

4. CONCLUSIONS 4.1. Onshore This paper has extracted average failure rates and downtimes from maintained, onshore German WTs in the WMEP programme, distinguishing between major and minor failures. The paper has demonstrated the following:

• Durations of downtimes due to failures for different WT subassemblies vary from a few hours to months, and the distribution is strongly asymmetric. Failures can be grouped into ‘minor’ failures (which can be resolved within one day) and ‘major’ failures (which require longer). • Minor failures (representing 75% of all failures) are responsible for only 5% of the downtime, whereas major failures (representing 25% of all failures) are responsible for 95% of the downtime. • Minor failures cause relatively little downtime but require considerable maintenance attention and significant repair effort. • The development of CMS needs to concentrate on these major failures. • Particular attention must be paid in the future to improving the reliability of electrical and electronic subassemblies. An important element of this must be the application of reliability-based maintenance.

4.2. Offshore Regarding offshore wind energy use, the paper has shown the following:

• The conclusions above are likely to be more significant offshore, where longer waiting, travel and work times will amplify the influence of minor failures on offshore WT availability. • Preliminary results from existing offshore wind farms confirm the likely increase in annual downtimes caused by these minor failures. However, current offshore wind farm experience has been with WTs no more than 12 km offshore. For wind farms planned at ≥50 km from shore, availability may decline even further.

4.3. Outlook • In Germany, it is proposed that in order to address the different issues of offshore wind energy application, an Offshore–WMEP should supersede the former WMEP programme. Collecting reliability data in a standardized way and to improve the maintenance and availability of offshore WTs will be some of the core issues of the Offshore–WMEP.

APPENDIX A Definitions used in the WMEP survey • Reliability—the probability of a device performing its purpose adequately for the period of time intended under the operating conditions encountered. • Availability—the probability of finding a system in the operating state at some time into the future. • Mean time to failure (MTTF). • Mean time to repair, or downtime—the average time it will take for a subassembly to recover from any failure (MTTR). • Mean time between failures or reliability—the average period between unplanned stoppages of a subassembly (MTBF): MTBF = MTTF + MTTR = 1/λ + 1/μ • Failure rate: λ = 1/MTBF • Repair rate: μ = 1/MTTR • Availability: A = (MTBF − MTTR)/MTBF = 1 − (λ/μ) Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

335

Wind turbine downtime

S. Faulstich, B. Hahn and P. J. Tavner

APPENDIX B WMEP incident report

Maintenance and Repair Report

report-nr.

work carried out

day

month

year

WMEP 250 MW-Wind cause of malfunction post code

plant identification number

operator

high wind

malfunction of control system

grid failure

component wear or failure

lightning

loosening of parts

icing

other causes cause unknown

manufacturer and model

effect of malfunction

reason for repair scheduled maintenance

overspeed

reduced power

scheduled maintenance with replacement of worn parts or repair of defects

overload

causing follow-up damage

unscheduled reapir after malfunction

down time

noise

plant stoppage

vibration

other consequences

removal of malfunction perfect functioning of plant after

not stopped

stopped control reset

to day

month

time

reading of hour counter

costs stated on bill material

Euro

labour

Euro

journey

Euro

total cost incl. VAT

Euro

hub

gear box

hub body

bearings

pitch mechanism

wheels

pitch bearings

gear shaft

rotor blades

sealings

blade bolts

mechanical brake

blade shell

brake disc

aerodynamic brakes

brake pads

generator

brake shoe

generator windings

drive train

generator brushes

rotor bearings

bearings

drive shafts

electric

couplings

converter

hydraulic system

fuses

hydraulic pump

switches

pump motor

cables/connections

valves

sensors

hydraulic pipes/hoses

anemometer/wind vane

yaw system

comments

operator place/date

signature

changing of control parameters

repaired or replaced components

from

vibration switch

yaw bearings

temperature

yaw motor

oil pressure switch

wheels and pinions

power sensor

structural parts/housing

revolution counter

foundation

control system

tower/tower bolts

electronic control unit

nacelle frame

relay

nacelle cover

mesurement cables and connections

ladder

Replaced main components nacelle

yaw system

rotorblade/-blades

tower

hub

control system cabinet

gear box

transformer

generator W&I /ISET 10/02

336

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

S. Faulstich, B. Hahn and P. J. Tavner

Wind turbine downtime

REFERENCES 1. Faulstich S, Durstewitz M, Hahn B, Knorr K, Rohrig K. Windenergie Report Deutschland 2008, Institut für solare Energieversorgungstechnik (Hrsg.), Kassel, 2008. 2. [Online]. Available: http://www.iset.uni-kassel.de/pls/w3isetdad/www_iset_new.main_page?p_name=7261110&p_ lang=eng (Accessed 9 August 2009). 3. Hahn B. Zuverlässigkeit, Wartung und Betriebskosten von Windkraftanlagen. Proceedings of the First Rheiner Windenergie-Symposium, Kötter Consulting Engineers, Rheine, 2003. 4. Hahn B, Durstewitz M, Rohrig K. ‘Reliability of wind turbines’ in wind energy. Proceedings of the Euromech Colloquium 464b 2005 Oldenburg. Springer-Verlag: Berlin, 2007; 329–332. 5. Faulstich S, Hahn B. Comparison of different wind turbine concepts due to their effects on reliability. UpWind, EU supported project No. 019945(SES6), deliverable WP7.3.2, public report, Kassel, 2009. 6. Echavarria E, Hahn B, van Bussel GJW, Tomiyama T. Reliability of wind turbine technology through time. Journal of Solar Energy Engineering 2008; 130: 031005-1–031005-8. 7. Tavner PJ, Xiang J, Spinato F. Reliability analysis for wind turbines. Wind Energy 2007; 10: 1–18. 8. Spinato F, Tavner PJ, van Bussel GJW, Koutoulakos E. Reliability of wind turbine subassemblies. IET Renewable Power Generation 2009; 3: 1–15. 9. Faulstich S, Hahn B. Appropriate failure statistics and reliability characteristics. Proceedings of the German Wind Energy Conference (DEWEK), Bremen, 2008. 10. Hahn B. Zeitlicher Zusammenhang von Schadenshäufigkeit und Windgeschwindigkeit. FGW-Workshop ‘Einfluß der Witterung auf Windenergieanlagen’, Institut für Meteorologie, Leipzig, 6 May 1997. 11. Tavner P, Edwards C, Brinkman A, Spinato F. Influence of wind speed on wind turbine reliability. Wind Engineering 2006; 1: 55–72. 12. Bussel Gv. Offshore wind energy, the reliability dilemma. Proceedings of the First World Wind Energy Conference, Berlin, 2002. 13. UK DTI Capital Grant Scheme Annual Reports: North Hoyle Offshore Wind Farm: July 2004–June 2005, North Hoyle Offshore Wind Farm: July 2005–June 2006, North Hoyle Offshore Wind Farm: July 2006–June 2007, Scroby Sands Offshore Wind Farm: January 2005–December 2005, Scroby Sands Offshore Wind Farm: January 2006– December 2006, Scroby Sands Offshore Wind Farm: January 2007–December 2007, Kentish Flats Offshore Wind Farm: January 2006–December 2006, Kentish Flats Offshore Wind Farm: January 2007–December 2007, Barrow Offshore Wind Farm: July 2006–June 2007. 14. NoordzeeWind Report OWEZ_R_000_20081023. Netherlands, 2008. 15. Faulstich S, Hahn B, Jung H, Rafik K. Suitable failure statistics as a key for improving availability. Proceedings of the European Wind Energy Conference (EWEC), Marseille, 2009.

Wind Energ. 2011; 14:327–337 © 2010 John Wiley & Sons, Ltd. DOI: 10.1002/we

337