The impact of wind conditions on wind turbines Lovisa Eriksson Petersen

Master of Science Thesis KTH School of Industrial Engineering and Management Energy Technology EGI-2016-037 MSC Division of Heat and Power Technology SE-100 44 STOCKHOLM

-2-

Master of Science Thesis EGI 2016: 037 MSC

The impact of wind conditions on wind turbines

Lovisa Eriksson Petersen Approved

Examiner

Supervisor

7/6-16

Björn Laumert

Björn Laumert

Commissioner

Contact person

Vattenfall AB

Niclas Broman

Abstract The world is facing global warming and the challenge to reduce greenhouse gas emissions. Wind power is a renewable source of energy with no greenhouse gas emissions when operating. Therefore, it could contribute in this challenge. Vattenfall is a Swedish energy company that invests in the wind power business and have many wind turbines in operation. On behalf of Vattenfall this study has been performed with the aim to find how wind conditions affect wind turbines and how it is correlated to amount of alarms, time loss, energy availability and time availability in a wind turbine. Improving this knowledge will be an advantage when making investments in wind turbines and their maintenance. A statistical analysis was conducted in order to examine correlations of wind speed and turbulence intensity with the number of alarms, time loss, energy availability and time availability of a turbine. A case study of Lillgrund, an offshore park in Sweden, was performed since the park layout is tightly spaced and hence interesting in sense of turbulence intensity. Lillgrund suffered to a lot of blade vibration alarms and therefore these alarms were investigated deeper in terms of the wind conditions one hour, one day and one week before the alarm occurred. Four additional parks with other turbine types and manufacturers than Lillgrund’s were also included in the correlation analysis. The amount of alarms per year of each turbine was examined in order to compare this with the failure curve of a wind turbine. The purpose was to see if the wear-out period had started earlier for turbines with certain wind conditions but unfortunately the turbines were not old enough to draw a conclusion. The analysis resulted in positive correlations for high turbulence and high amount of alarms and blade vibration alarms for the Siemens SWT-2.3-93 turbines investigated while there were no clear correlations for time loss, energy availability and time availability. Also, the other turbine types had no strong correlations for the investigated parameters. From the results it can be concluded that there are no strong correlations for the wind conditions investigated and energy availability and time availability. It can also be said that the amount of alarms in Siemens SWT-2.3-93 turbines increases with higher turbulence. Hence, alarms do not influence the energy availability noticeably for this type of turbine. -3-

Summary in Swedish Världen står inför en global uppvärmning och utmaningen att minska utsläppen av växthusgaser. Vindkraft är en förnybar energikälla utan utsläpp av växthusgaser vid drift som skulle kunna bidra positivt i denna utmaning. Vattenfall är ett svenskt energiföretag som investerar i vindkraftsverksamhet och äger många vindkraftverk. Denna studie genomförts på uppdrag av Vattenfall med syfte att finna hur vindförhållanden påverkar vindkraftverk och hur de är korrelerade till mängden av alarm, tidsförlust, energitillgänglighet och tidstillgänglighet i ett vindkraftverk. Att förbättra denna kunskap kommer att vara en fördel när investeringar i vindkraftverk och deras underhåll görs. En statistisk analys genomfördes för att undersöka korrelationer av vindhastighet och turbulensintensitet med antalet alarm, tidsförlust, energitillgänglighet och tidstillgänglighet hos en turbin. En studie av Lillgrund, en havsbaserad park i Sverige, genomfördes då parkens turbiner är placerade tätt intill varandra och därmed ger ett intressant perspektiv på turbulensintensitet. Lillgrund har en stor mängd bladvibrationsalarm och därför har dessa alarm undersökts djupare när det gäller vindförhållandena en timme, en dag och en vecka innan alarmet inträffade. Ytterligare fyra parker med andra turbintyper och tillverkare än Lillgrunds ingick i korrelationsanalysen. Mängden alarm per år för varje turbin undersöktes för att jämföra detta med felintensitetskurvan för ett vindkraftverk. Syftet var att se om utslitningsperioden hade börjat tidigare för turbiner med vissa vindförhållanden men tyvärr var de studerade turbinerna inte tillräckligt gamla för att dra slutsatser kring detta. Analysen resulterade i positiva korrelationer för hög turbulens och hög mängd av alarm och bladvibrationsalarm för de undersökta Siemens SWT-2.3-93-turbinerna medan det inte fanns några tydliga korrelationer för tidsförlust, energitillgänglighet och tidstillgänglighet. Även de andra turbintyperna saknade starka samband för de undersökta parametrarna. Av resultaten kan slutsatsen dras att det inte finns några starka samband för de undersökta vindförhållandena och energitillgängligheten och tidstillgängligheten. Det kan också sägas att mängden alarm i Siemens SWT2.3-93-turbiner ökar med högre turbulensintensitet. Därför påverkar inte alarm energitillgängligheten märkbart för denna typ av turbin.

-4-

Acknowledgement First of all I would like to thank Niclas Broman, Head of Development & Tendering at Vattenfall, for guiding me during the process of my master thesis and for discussing and forming the aim of this master thesis. Secondly, I would like to thank all the people at Vattenfall who has taken the time to help me and shared their knowledge about wind turbines and wind conditions, especially Jan-Åke Dahlberg, Francois Besnard, Thomas Stalin, Vincent Baron, Tor Söderlund, Villads Haar Jakobsen and Jesper Kristoffersen Runge. I would also like to thank Carles Campos at Vattenfall for providing me with data from the parks which made it possible to do my study. Thirdly, I would like to thank my supervisor from Royal Institute of Technology, Björn Laumert, Doctorate of Philosophy & Associate Professor, who have supported me and given me new perspectives and ideas of the objectives and the method. Lastly, I would like to thank my family and friends who has been a great support during the challenge of writing a master thesis.

-5-

Nomenclature The Nomenclature chapter contains abbreviations, symbols, subscripts and definitions used in the thesis. The first time an abbreviation is mentioned in the thesis it will be spelled out in order to ease the understanding while reading.

Abbreviations IEC – International Electrotechnical Commission rpm – revolutions per minute SCADA – Supervisory Control And Data Acquisition SLG – Lillgrund wind park, Sweden SRL – Stor-Rotliden wind park, Sweden WH1 – Horns Rev 1 wind park, Denmark WLY – Lyngsmose wind park, Denmark WNE – Nørrekær Enge wind park, Denmark WPDC – Wind Power Data Centre

Symbols Symbol

Unit

Description

Symbol

Unit

Description

𝒎̇

kg/s

Mass/second

𝜌

kg/m3

Density

𝒏

-

𝜎𝑖

-

Standard deviation of a variable i

𝑷

W

Power

𝜎𝑖𝑗

-

Covariance of two variables i and j

𝒓

-

Correlation coefficient

𝒖

m/s

Wind speed

̅ 𝒙

-

Mean value

Subscripts 𝑎𝑣𝑒 – Average of variables 𝑖 – Name of a variable 𝑗 – Name of a variable 𝑢 – Wind speed (meters/second) 𝑟𝑒𝑓 – Reference value of a variable

-6-

Definitions Alarms – logged alarms in SCADA system Energy Availability - based on actual and lost production. It is affected by: curtailment during failures or service, scheduled maintenance, damage, faults, alarms, manual stops, service, suspension due to personal safety or equipment integrity (e.g. ice) Population – a finite or infinite number of individuals subject to a statistical study Sample – a selection of a population Time Availability – based on installed power. It is affected by: curtailment during failures or service, scheduled maintenance, damage, faults, alarms, manual stops, service, suspension due to personal safety or equipment integrity (e.g. ice) Time loss – number of seconds in a 10 minutes interval that the wind turbine is not fully operational, visualized in percentage in graphs in this report Wind conditions – in this report it mostly refers to wind speed and turbulence intensity

-7-

Table of Content 1

2

Introduction ........................................................................................................................................................13 1.1

Background ................................................................................................................................................13

1.2

Aim and objectives....................................................................................................................................13

1.3

System boundary and limitations ............................................................................................................14

1.4

Thesis overview .........................................................................................................................................14

Theory ..................................................................................................................................................................15 2.1 2.1.1

Rotor ..................................................................................................................................................15

2.1.2

Drivetrain ..........................................................................................................................................17

2.1.3

Transformer ......................................................................................................................................18

2.1.4

MV switchgear ..................................................................................................................................18

2.1.5

Control systems ................................................................................................................................18

2.1.6

Tower and foundation ....................................................................................................................19

2.2

Failure curve .....................................................................................................................................19

2.2.2

Maintenance ......................................................................................................................................19 Wind conditions ........................................................................................................................................20

2.3.1

Power in the wind & Betz limit .....................................................................................................20

2.3.2

Turbulence ........................................................................................................................................21

2.3.3

IEC wind classes ..............................................................................................................................21

2.4

Wind turbine loads ....................................................................................................................................22

2.5

Lillgrund .....................................................................................................................................................23

2.6

Other investigated parks ..........................................................................................................................24

2.7

Statistical methods ....................................................................................................................................27

2.7.1

4

Operation and maintenance ....................................................................................................................19

2.2.1 2.3

3

Wind turbines ............................................................................................................................................15

Correlation coefficient ....................................................................................................................27

Literature review .................................................................................................................................................29 3.1

Wind power in forests ..............................................................................................................................29

3.2

Fatigue loads on wind turbine blades in a wind farm .........................................................................29

3.3

Study of weather and location effects on wind turbine failure rates ................................................29

Methodology .......................................................................................................................................................30 4.1

Method of case study................................................................................................................................30

4.1.1

Data collection..................................................................................................................................30

4.1.2

Data analysis .....................................................................................................................................30

4.2

Method of complementary study ...........................................................................................................32

4.2.1

Data collection..................................................................................................................................32

4.2.2

Data analysis .....................................................................................................................................33 -8-

5

Results and discussion .......................................................................................................................................34 5.1 5.1.1

Summary of collected data .............................................................................................................34

5.1.2

All alarms ...........................................................................................................................................40

5.1.3

Selected alarms .................................................................................................................................41

5.1.4

Blade vibrations ................................................................................................................................42

5.1.5

Time loss ...........................................................................................................................................48

5.1.6

Energy availability and time availability ........................................................................................49

5.2

6

Results of case study .................................................................................................................................34

Results of Siemens SWT-2.3-93 turbine................................................................................................51

5.2.1

Lyngsmose ........................................................................................................................................51

5.2.2

Nørrekær Enge .................................................................................................................................51

5.2.3

Combination of Siemens SWT-2.3-93 turbines ..........................................................................52

5.3

Results of Stor-Rotliden ...........................................................................................................................56

5.4

Results of Horns Rev 1 ............................................................................................................................57

5.5

Results of all turbines combined ............................................................................................................58

5.6

Summative discussion ..............................................................................................................................60

5.7

Uncertainties ..............................................................................................................................................61

Conclusions .........................................................................................................................................................62 6.1

Future work................................................................................................................................................62

6.2

Suggestions for Vattenfall ........................................................................................................................62

7

References ...........................................................................................................................................................64

8

Appendices ..........................................................................................................................................................66 8.1

Figures.........................................................................................................................................................66

8.1.1

Data summary of Lillgrund ............................................................................................................66

8.1.2

Correlations .......................................................................................................................................69

8.2

Average wind speed and turbulence intensity in Lillgrund 2008-2015.............................................79

8.3

Average wind speed and turbulence intensity in Lillgrund 2009-2015.............................................81

8.4

Average wind speed and turbulence intensity in Lillgrund 2010-2015.............................................82

8.5

Excluded alarms Lillgrund .......................................................................................................................84

-9-

Index of Tables Table 1. IEC wind classes from 2005 standard, table modified from IEC (International Electrotechnical Commission, 2005)......................................................................................................................................................22 Table 2. Table of Siemens SWT-2.3-93 turbine specifics. ....................................................................................24 Table 3. Table of Vestas V80-2000KW-2G turbine specifics. .............................................................................26 Table 4. Table of Vestas V90-2000KW-2G & Vestas V90-1800KW-2G turbine specifics. ..........................27 Table 5. Collected data specifics. ..............................................................................................................................30 Table 6. Cardinal and degree directions. ..................................................................................................................31 Table 7. Data time interval for Lyngsmose, Nørrekær Enge, Stor-Rotliden and Horns Rev 1......................32 Table 8. Most frequent errors of all turbines in Lillgrund 2008-2015. ...............................................................37

Index of Figures Figure 1. The aerodynamic forces acting on a wind turbine blade......................................................................16 Figure 2. The bathtub curve. .....................................................................................................................................19 Figure 3. Lillgrund layout based on coordinates of the turbines. ........................................................................23 Figure 4. Lyngsmose layout based on coordinates of the turbines. ....................................................................24 Figure 5. Nørrekær Enge layout based on coordinates of the turbines. .............................................................25 Figure 6. Horns Rev 1 layout based on coordinates of the turbines...................................................................25 Figure 7. Stor-Rotliden layout based on coordinates of the turbines. ................................................................26 Figure 8. Nacelle direction distribution average of all Lillgrund turbines, 2012-2015. ....................................34 Figure 9. Average wind speed per unit in Lillgrund (meters/second), 2008-2015. ..........................................35 Figure 10. Average turbulence intensity per unit in Lillgrund (in percentage), 2008-2015. ............................35 Figure 11. Average annual alarms per unit in Lillgrund, 2008-2015. ..................................................................36 Figure 12. Average annual alarms per unit in Lillgrund, 2008-2015. ..................................................................36 Figure 13. Total amount of alarms per unit in Lillgrund every year since commission year 2007. ................37 Figure 14. Average annual time loss in Lillgrund, 2008-2015. .............................................................................38 Figure 15. Average annual energy availability per unit in Lillgrund, 2009-2015................................................38 Figure 16. Time availability per unit in Lillgrund, 2009-2015. .............................................................................39 Figure 17. Annual amount of repairs in Lillgrund. ................................................................................................39 Figure 18. Annual amount of replacements in Lillgrund. .....................................................................................40 Figure 19. Average annual alarms compared with average wind speed for each unit in Lillgrund, 2008-2015. ........................................................................................................................................................................................40 Figure 20. Average annual alarms compared with turbulence intensity for each unit in Lillgrund, 2008-2015 ........................................................................................................................................................................................41 Figure 21. Average annual amount of selected alarms per unit in Lillgrund, 2009-2015. ...............................41 Figure 22. Average annual selected alarms and average wind speed for each unit in Lillgrund, 2009-2015. ........................................................................................................................................................................................42 Figure 23. Average annual selected alarms and average turbulence intensity for each unit in Lillgrund, 20092015. ..............................................................................................................................................................................42 Figure 24. Blade vibrations per month, accumulated for the years 2010-2015 in Lillgrund. ..........................43 Figure 25. Average annual amount of blade vibration alarms per unit in Lillgrund, 2010-2015. ...................43 Figure 26. Average annual amount of blade vibrations and average wind speed for each unit in Lillgrund, 2010-2015. ....................................................................................................................................................................44 Figure 27. Average annual amount of blade vibrations and average turbulence intensity for each unit in Lillgrund, 2010-2015. ..................................................................................................................................................44 Figure 28. Percentage of total average wind speed the hour before a blade vibration alarm in Lillgrund, 20102015. ..............................................................................................................................................................................45 -10-

Figure 29. Percentage of total average wind speed the day before a blade vibration alarm in Lillgrund, 20102015. ..............................................................................................................................................................................45 Figure 30. Percentage of total average wind speed the week before a blade vibration alarm in Lillgrund, 2010-2015. ....................................................................................................................................................................46 Figure 31. Percentage of total average turbulence intensity the hour before a blade vibration alarm in Lillgrund, 2010-2015. ..................................................................................................................................................46 Figure 32. Percentage of total average turbulence intensity the day before a blade vibration alarm in Lillgrund, 2010-2015. ..................................................................................................................................................47 Figure 33. Percentage of total average turbulence intensity the week before a blade vibration alarm in Lillgrund, 2010-2015. ..................................................................................................................................................47 Figure 34. Annual average time loss in percent and average wind speed for each unit in Lillgrund, 20082015. ..............................................................................................................................................................................48 Figure 35. Average annual time loss in percent and average turbulence intensity for each unit in Lillgrund, 2008-2015. ....................................................................................................................................................................48 Figure 36. Energy availability in percentage and average wind speed for each unit in Lillgrund, 2009-2015. ........................................................................................................................................................................................49 Figure 37. Average annual energy availability in percentage and average turbulence intensity for each unit in Lillgrund, 2009-2015. ..................................................................................................................................................49 Figure 38. Time availability in percentage and average wind speed for each unit in Lillgrund, 2009-2015..50 Figure 39. Time availability in percentage and average turbulence intensity for each unit in Lillgrund, 20092015. ..............................................................................................................................................................................50 Figure 40. Amount of alarms per unit every year in Lyngsmose since commission year 2008. .....................51 Figure 41. Amount of alarms per unit every year in Nørrekær Enge since commission year 2009...............51 Figure 42. Amount of annual alarms and average wind speed for each unit in SLG, WLY and WNE. ......52 Figure 43. Amount of annual alarms and average turbulence intensity for each unit in SLG, WLY and WNE. ........................................................................................................................................................................................52 Figure 44. Amount of annual blade vibration alarms and average wind speed for each unit in SLG, WLY and WNE. .....................................................................................................................................................................53 Figure 45. Amount of annual blade vibration alarms and average turbulence intensity for each unit in SLG, WLY and WNE. ..........................................................................................................................................................53 Figure 46. Time loss and average wind speed for each unit in SLG, WLY and WNE....................................54 Figure 47. Time loss and average turbulence intensity for each unit in SLG, WLY and WNE. ....................54 Figure 48. Energy availability and average wind speed for each unit in SLG, WLY and WNE. ...................55 Figure 49. Energy availability and average turbulence intensity for each unit in SLG, WLY and WNE. ....55 Figure 50. Time availability and average wind speed for each unit in SLG, WLY and WNE. .......................56 Figure 51. Time availability and average turbulence intensity for each unit in SLG, WLY and WNE. ........56 Figure 52. Amount of alarms per unit every year in Stor-Rotliden since commission year 2010. .................57 Figure 53. Amount of alarms per unit in Horns Rev 1, data missing 2002-2006. ............................................57 Figure 54. Time loss and wind speed for all studied turbines. .............................................................................58 Figure 55. Time loss and turbulence intensity for all studied turbines. ..............................................................58 Figure 56. Energy availability and wind speed for all studied turbines. ..............................................................59 Figure 57. Energy availability and turbulence intensity for all studied turbines. ...............................................59 Figure 58. Time availability and wind speed for all studied turbines. .................................................................60 Figure 59. Energy availability and turbulence intensity for all studied turbines. ...............................................60 Figure 60. Wind speed distribution in Lillgrund, 2010-2015................................................................................66 Figure 61. Turbulence intensity distribution in Lillgrund, 2010-2015. ...............................................................66 Figure 62. Average annual time loss in Lillgrund, 2008-2015. .............................................................................67 Figure 63. Average annual energy availability in Lillgrund, 2008-2015...............................................................67 Figure 64. Average annual time availability in Lillgrund, 2008-2015. .................................................................68 Figure 65. Average wind speed the hour before a blade vibration alarm in each unit, 2010-2015. ...............68

-11-

Figure 66. Average turbulence intensity the hour before a blade vibration alarm in each unit in Lillgrund, 2010-2015. ....................................................................................................................................................................69 Figure 67. The amount of software update alarms and wind speed in Lillgrund. ............................................69 Figure 68. The amount of software update alarms and turbulence intensity in Lillgrund...............................70 Figure 69. Amount of blade vibration alarms per year and the wind speed an hour before a blade vibration alarm for each unit in Lillgrund, 2010-2015. ...........................................................................................................70 Figure 70. Amount of blade vibration alarms per year and the turbulence intensity an hour before a blade vibration alarm for each unit in Lillgrund, 2010-2015...........................................................................................71 Figure 71. Amount of alarms per year and wind speed for each unit in Stor-Rotliden, 2012-2015. .............71 Figure 72. Amount of alarms per year and turbulence intensity for each unit in Stor-Rotliden, 2012-2015. ........................................................................................................................................................................................72 Figure 73. Time loss and wind speed for each unit in Stor-Rotliden, 2012-2015. ............................................72 Figure 74. Time loss and turbulence intensity for each unit in Stor-Rotliden, 2012-2015. .............................73 Figure 75. Energy availability and wind speed for each unit in Stor-Rotliden, 2012-2015..............................73 Figure 76. Energy availability and turbulence intensity for each unit in Stor-Rotliden, 2012-2015. ..............74 Figure 77. Time availability and wind speed for each unit in Stor-Rotliden, 2012-2015. ................................74 Figure 78. Time availability and turbulence intensity for each unit in Stor-Rotliden, 2012-2015. .................75 Figure 79. Amount of alarms and wind speed for each unit in Horns Rev 1....................................................75 Figure 80. Amount of alarms and turbulence intensity for each unit in Horns Rev 1. ....................................76 Figure 81. Time loss and wind speed for each unit in Horns Rev 1. ..................................................................76 Figure 82. Time loss and turbulence intensity for each unit in Horns Rev 1. ...................................................77 Figure 83. Energy availability and wind speed for each unit in Horns Rev 1....................................................77 Figure 84. Energy availability and turbulence intensity for each unit in Horns Rev 1. ....................................78 Figure 85. Time availability and wind speed for each unit in Horns Rev 1. ......................................................78 Figure 86. Time availability and turbulence intensity for each unit in Horns Rev 1. .......................................79

-12-

1 Introduction 1.1 Background The world is facing a great challenge: global warming. Many countries have started to take action against this problem by trying to reduce greenhouse gas emissions. A way of reducing greenhouse gas emissions is to decrease use of fossil energy and this can be achieved by introducing renewable energy. Wind power is a renewable energy source that has been used by humans ever since the invention of sails (Wizelius, 2007). Windmills were introduced around 3,000 years ago in China and Japan, but the first mill to be documented was located in Persia and it had a vertical axis (Wizelius, 2007). The wind mills were spread to Europe as late as in the end of the 12th century and these had improved technology with horizontal axis (Wizelius, 2007). The technique to use turbines to produce electricity is younger and the first wind turbine to produce electricity was built in 1892 (Wizelius, 2007). In the end of 2015 the total installed wind turbine capacity of the world was 432,419 MW, most of which was located in Asia and Europe (Global Wind Energy Council, 2016). Wind power is a fast growing energy source in Europe and one of the largest operators in the wind energy market is Vattenfall (Vattenfall AB, 2015). Vattenfall owns about 1,000 turbines in Denmark, Germany, Great Britain, Netherlands and Sweden, 130 of which are installed in Sweden (Vattenfall AB, 2015). The capacity of Sweden in the end of 2015 was 6,025 MW (Global Wind Energy Council, 2016), which came from 3,233 turbines (Svensk Vindenergi, 2016). Prior to the decision to invest in a wind farm it is very important to understand the assessment of the profitability of the investment and the risks and opportunities. A parameter that affects the profitability is the maintenance cost of the wind turbines. A turbine consists of many components; typically there are the rotor, gearbox, generator, yaw system, transformer and anemometer (Wizelius, 2007). Replacing components will lead to higher maintenance cost and Vattenfall would like to limit this cost as much as possible. One factor that could be examined in order to reduce maintenance costs is the connection between the wind conditions at the site (wind speed and turbulence) and how these affect the wear on the wind turbine and thus the need for maintenance. This will be investigated in this report. However, it needs to be kept in mind that maintenance may be required on a regular basis, so called scheduled maintenance, for example replacement of wear parts, refill of oil and other care of the materials. The focus of this study will be on unscheduled maintenance.

1.2 Aim and objectives The aim of this master thesis is to investigate the influence of wind conditions on wind turbines. To achieve this, the correlation between the wind conditions and alarms, time loss, energy availability and time availability need to be examined. Key question: o

How do wind conditions (wind speed and turbulence intensity) affect wind turbines?

Auxiliary questions: o o o o

Which errors may occur in the wind turbines that Vattenfall owns? What wind conditions occur at the turbines? What is the frequency of the alarms in the turbines? How do wind conditions (wind speed and turbulence intensity) correlate to alarms, time loss, energy availability and time availability?

The expected result is an indication of how wind speed and turbulence intensity affects wind turbines.

-13-

1.3 System boundary and limitations Vattenfall owns turbines in Denmark, Germany, Great Britain, Netherlands and Sweden. There are turbines of several different models and manufacturers. This study focuses on five parks owned, entirely or partly, by Vattenfall. The case study includes data from Lillgrund in Sweden. This park is an offshore park with turbines of type Siemens SWT-2.3-93. The case study was complimented by a study of data from Lyngsmose and Nørrekær Enge in Denmark. They have turbines of the same type as Lillgrund, Siemens SWT-2.3-93. In addition, data from Horns Rev 1 in Denmark and Stor-Rotliden in Sweden was collected to see the trends of other turbine types. Horns Rev 1 have Vestas V80-2000KW-2G turbines and Stor-Rotliden Vestas V902000KW-2G and Vestas V90-1800KW-2G. The wind conditions this report will focus on are wind speed and turbulence. Unfortunately, it was not possible to investigate shear since lack of wind data from different heights at the turbines.

1.4 Thesis overview Chapter 1 Introduction contains the background to the project, the aim and objectives and the system boundaries. It introduces the reader to the main focus of the thesis. Chapter 2 Theory contains the theory of wind turbines, wind conditions, a description of the studied wind parks with emphasis on Lillgrund and statistical methods. Chapter 3 Literature review contains the literature review of previous studies in the areas. Chapter 4 Methodology contains the method used in the case study and the complementary studies. Chapter 5 Results and discussion contains the results and a discussion of the results of the case study and the complementary studies. Chapter 6 Conclusions contains the conclusions of the results of the case study and the complementary studies. In the end there is a list of references in chapter 7 and appendices are attached in chapter 8.

-14-

2 Theory 2.1 Wind turbines Wind makes the blades and the hub, which together is called rotor, of the turbine to rotate and when it is connected to a generator on a turbine axis, it is possible to get electricity as an output. Usually a gearbox is installed between the rotor and the generator (Vattenfall AB, 2013), if not the turbine is called a direct drive turbine. The gearbox increases the number of revolutions (Vattenfall AB, 2013), in order to get same frequency as the grid. A yaw system helps to turn the turbine into the direction of the wind in order to maximize the output from the wind. Wind turbines exist in many forms, with both horizontal and vertical axis. Among the horizontal turbines there are micro turbines with powers lower than 1 kW, farm-based turbines, medium turbines with 50-1,000 kW power, megawatt-turbines with 1-2 MW power and multi-megawatt turbines with powers larger than 2 MW (Wizelius, 2007). Wind turbines can have different shapes with different amount of blades and turbines considered is this report are mainly horizontal turbines with three blades. Wind parks can consist of many turbines in connection in a transmission system. The distance between the turbines should be four to ten times the rotor diameter depending on wind conditions at the site in order to minimize power losses (Vattenfall AB, 2013). A turbine starts to produce at cut-in speed and this is usually around 4 m/s. If less wind speed, the turbine will be in standby. The turbine delivers full power at its rated speed, which is around 12-14 m/s. At cut-out speed it shuts down to protect the turbine from damage and heavy wear and this is commonly speeds higher than 25 m/s (Vattenfall AB, 2013). The technical lifetime of a wind turbine is calculated to 20 or 25 years (Wizelius, 2007). 2.1.1 Rotor A rotor consists of a hub and blades. The rotor diameter is the common measure of rotor size. The commercially available turbines often have three blades. In order to maximize the wind energy the rotor should have the correct tip speed ratio, which is the relation between the tip speed and the wind speed. A higher tip speed ratio is needed when the blades are fewer (Wizelius, 2007). The blade bearing connects the blade to the hub and makes the twist of the blades possible. The aerodynamic forces make the rotor rotate. The two most central forces in aerodynamics are lift and drag. Lift force is force that pulls upwards and causes a lift, while drag force is a force that pulls backwards causing a resistance (see Figure 1). The shape of the blade determines its characteristics. The shape of the blade and the angle of the incoming wind (angle of attack) will determine the lift force. If the angle of attack is too large the air stream may not follow the profile of the blade and instead cause a swirl starting at the separation point, which is called stall and it causes less lifting force (Wizelius, 2007).

-15-

Figure 1. The aerodynamic forces acting on a wind turbine blade.

Stall may be used in power regulation, which is done in order to decrease the power from a turbine when the wind speed is too high. Power regulation by blade profile is called stall regulation. Power regulation is needed when the wind speed increases so much that the power output needs to be regulated in order to avoid to heavy loading on the turbine (Wizelius, 2007). Another way of regulating the power is pitch-regulation. With pitch-regulation the rotor blades are versatile at the hub which makes it possible to change the angle of attack into a desired angle which will give no more power than the rated power. At cut-out speed the blades turns into a position where the wind can pass through without causing lift forces (Wizelius, 2007). The rotor suffers from dynamic load changes which can lead to high strain (Wizelius, 2007) and in worst case blade failures. Therefore, the rotor need to be constructed of a material that can manage alternating loads (Wizelius, 2007). The turbines owned by Vattenfall usually have blades made of strong glass fibre material with built-in lightening protection (Vattenfall AB, 2013), but other materials are also commercially available, for example carbon fibre (Wizelius, 2007). Wind turbines are classed to withstand loads (see chapter 2.3.3 IEC wind classes) and the blades as well, particularly to extreme loads, fatigue loads and normal conditions (see 2.4 Wind turbine loads). Normal conditions demands that the blades need to have enough distance between blade tip and tower in order to ensure free passage of the blade. The Eigen frequency of the blade should not match the frequency of the operating turbine. If turbulence intensity becomes too high or low it may conflict with Eigen frequencies and this contributes to failures (Freudenreich, 2016). Higher loading than design conditions or wind shear may cause vibrations in the blades. High turbulence intensity shakes the turbine as well as frequent shutdowns and restarts. Vibrations can also occur because of wrong pitch angle of the blades – if it is too high there will be a stalled flow which causes vibrations and loads (Freudenreich, 2016). When the turbine detects vibrations it will trigger a blade vibration alarm. An alarm level set too sensitive may set off more blade vibration alarms than necessary. Another cause of vibrations may be dirt and ice on blades (Freudenreich, 2016). Rotors to be placed in arctic climate, for example in northern Sweden, will need a de-icing system. The ice will change the blade profile and the aerodynamic characteristics and this may make the turbine stop itself. In order to prevent this blades are heated by heating coils controlled by an ice detector (Wizelius, 2007). A blade failure can occur for many reasons, the most common are manufacturing problems and lack of proper maintenance. There can also be overloading, higher wind speeds (annual and extreme) and turbulence intensities than designed for. The exceedance of extreme loads may lead to failures. Extended fatigue loads have not been seen much yet since many of the existing turbines have not yet reached their -16-

designed life time. Gusty winds and shear is not covered by the International Electrotechnical Commission (IEC) wind classes (see chapter 2.1.7) and are usually caused by complex terrains. This is a factor that contributes to blade failures. Another cause of failure may be yawing problems since a mismatch of main wind direction and nacelle direction increases loadings (Freudenreich, 2016). An example of a failure in the rotor is cracks on the blades. They should be noticed during scheduled maintenance or service in the turbine and be repaired before the blade is severely damaged. 2.1.2 Drivetrain The drivetrain includes the generator, often a gearbox and other smaller components (Gipe, 2004), such as brakes (Manwell, et al., 2009). This is placed in the nacelle connected by the main shaft with its bearing mounted on a bed plate (Wizelius, 2007). The main shaft is also connected to a yaw motor which turns the nacelle in the wind direction (Wizelius, 2007). Common failures at the main shaft are the main bearing, especially at high wind speeds when it suffers from high stresses. Maintenance of main bearings is complicated and when failures occur the bearing is often not possible to repair and a full replacement is needed (Lindqvist & Lundin, 2010). 2.1.2.1

Generator

Generators convert the mechanical energy from the rotor into electricity. A simple form of generator contains a rotating coil of wire within a magnetic field (Gipe, 2004). The magnets in the rotor create a magnetic field which carries current and they induce voltages through the coils of the stator. A generator usually creates alternating current (Wizelius, 2007). The most common generator is the asynchronous alternator (induction), which is used in turbines with high rotational speed. Another kind of generator is the synchronous generator which has low rotational speed; hence it does not need a gearbox. Ring generators have many poles and a large diameter which enables production without a gearbox. A direct drive turbine has lower required maintenance since it does not have a gearbox but a disadvantage is that the generator is heavier (Wizelius, 2007). Some turbines contain two generators with different rotational speeds. Usually dual generators consist of one with six poles and 1,000 revolutions per minute (rpm) and one with four poles and 1,500 rpm. Lower rpm gives lower noise, which is good when constructing sites close to residential buildings (Wizelius, 2007). The coils may melt because of high temperature if the wind speed becomes too high and the turbine does not have any pitch or stall regulation, but most wind turbines are constructed such as it will withstand higher wind speed than rated for a while when the regulation system is settled (Wizelius, 2007). A common failure of the generator is corrosion due to lack of lubrication. Corrosion may lead to failures on the magnets since they have ferrous content. Mechanical stresses are common in newer turbine designs. Permanent magnet generators have problems with bearing failures. Vibration and periodical electrical tendering tests will lower the time of outages (Alewine & Chen, 2012). 2.1.2.2

Gearbox

The gearbox is needed in order to increase the rotational speed from 20-30 rpm into the required rotational speed of the generator. The required rotational speed is 1000 rpm for a generator with six poles and 1500 rpm for a generator with four poles when the frequency of the generator is 50 hertz which is the frequency of the grid in Sweden (Wizelius, 2007). Most gearboxes have three stages. Each stage in the gearbox causes losses, around 1% per stage. Thus, a three stage generator has to an efficiency of around 97%. The gearbox need to be maintained with oil as lubrication and cooled by a cooling system (Wizelius, 2007). Failure of the gearbox is more critical than the other components since it has long maintenance time per failure (Ribrant & Bertling, 2007). Elforsk investigated failure of gearboxes more closely and found that the -17-

stresses may cause that the cogs gets in contact with each other and causes lubrication problems. The cogs may fail totally leading a need to replace the entire gearbox. The most critical parts are the high speed shaft and its bearing. The lubrications are important in order to get longer lifetime. The lifetime is around 7-8 years and a change or renovation during the turbines lifetime should be planned for (Horste & El-Thalji, 2011). 2.1.3 Transformer The transformer is used to distribute the electricity from the turbine to the grid since its voltage need to be converted into the voltage of the grid. Transformers are rated according to its apparent power in kVA. The range is often 5-50 kVA (Manwell, et al., 2009). Many turbines today have 690 volt and if they were to be connected to the grid a transformer would be needed to convert the voltage into high voltage at 10 or 20 kV. The transformer is usually placed on ground level inside or outside the tower. However, some transformers are placed in the nacelle as a counterweight to the rotor (Wizelius, 2007). 2.1.4 MV switchgear A switchgear is needed in order to connect and disconnect the wind park from the grid. It is usually made of large contactors and is controlled by electromagnets. It is recommended that the switchgear operate automatically in order to fast disconnect if there is a failure in the grid or in a turbine (Manwell, et al., 2009). 2.1.5 Control systems There is a computer system in order to control the functionality of the turbine. Temperature, pressure, voltage, frequency and vibrations in different components of the turbine are measured among many other parameters. It also measures availability, production and amount of stops (Wizelius, 2007). An anemometer placed behind the rotor on top of the nacelle measures wind speed and nacelle direction. When the measures exceed the safety range the turbine is stopped and sends an alarm to the surveillance centre. The problem can often be solved manually remotely (Wizelius, 2007). 2.1.5.1

Brake system

There need to be two brake systems; aerodynamic and mechanical brakes. The mechanical brakes are placed in the nacelle and are used when the turbine needs service or when the aerodynamic brakes do not work when the wind has reached cut-out wind speed. The mechanical brake is a disc brake at main shaft or between gearbox and generator (Wizelius, 2007). It is most common to place the brake on the high speed shaft (after the gearbox) since they can be smaller and thus less expensive than the large disk brakes on the main shaft (Gipe, 2004). 2.1.5.2

Yaw system

In order to gain maximum power the rotor need to be perpendicular to the wind. This can be controlled by a yaw system. A computer system gathers information from the anemometer and sends the information of how much the yaw system shall rotate the rotor. Since the wind is changing almost every second the system cannot send a signal to twist for every change. The computer system ignores the changes until the wind direction has stabilized and been in the same average direction for some minutes and then it sends a signal to the yaw system to turn the nacelle (Wizelius, 2007). The major component is a bearing that connects the bed plate to the tower. It is driven by a yaw motor that is controlled by the yaw system. There is often a yaw brake that is used to keep the nacelle still when not yawing. This helps mitigating the common failures of rapid wear and breaking of the yaw drive (Manwell, et al., 2009).

-18-

When the nacelle has turned three complete rotations it will need to untwist the cables in order to save them from breaking. The turbine stops itself automatically, untwists the cables and then it restarts itself (Wizelius, 2007). 2.1.6 Tower and foundation Most towers are of conical structure, with a wider base than top, and are made in steel. Tower height differs, where old ones may not be higher than 30 meters high and new ones can be high as 160 meters. When the towers are higher than 40 meters they need to be created in different sections and mounted with bolts at site because of transportation problems (Wizelius, 2007). The foundation need to bear the weight of the tower and ensure that the turbine will not collapse. Hence the foundation needs to be adjusted to the specifics of the ground. There are two common offshore foundation; gravitation and monopoles (Wizelius, 2007).

2.2 Operation and maintenance 2.2.1 Failure curve Failures in a wind turbine are often predicted over the time of its operating life. The first years of a turbine (the infancy period) often have early failures with a higher rate than during the useful life period where the failures are occur at almost constant rate (intrinsic failures). When the turbine is reaching the end of its designed life time the wear-out period starts with a higher amount of failures (deterioration). This curve is often called the bathtub curve.

Figure 2. The bathtub curve.

2.2.2 Maintenance Scheduled maintenance is usually performed every 6 and 12 months with additional extended controls later in the life time at certain chosen years. The inspection occurring every 12 months includes inspections and tests of the components. For instance:   

the bolts are inspected and tightened tests are performed for brakes, sensors, controller system, pitch system, temperatures and signals etcetera oils and other lubricants are refilled, for example in the bearings -19-

 

blades are checked for cracks filters are checked and changed

Also, there is a test of emergency and safety functions, check of safety equipment and the turbine is cleaned among many other things. When the service is ended the turbine is restarted. It takes around 1.5 day in total per turbine of which the turbine is shut down 9-10 hours. The 6 month maintenance is similar to the 12 month but is less extensive. Instead it takes around 1 day per turbine of which the turbine is shut down around 5 hours. In addition, the transformers are inspected and it takes 0.5 day and leads to 2-3 hours downtime (Vilminko, 2016). Besides maintenance of turbines the grid stations (one for each turbine) and substation (which delivers the production from all turbines to the grid) needs inspections as well. Maintenance of a grid station takes about 0.5 hours and can be done during turbine operation every 6 and 12 months. The substation is inspected every 12 months which do not need a shutdown of the turbines but every second year it need a service which contribute to a shutdown of all turbines in the park for around 8 hours. Every fourth year it will need a longer service during 3 days with around 24 hours of park downtime (Vilminko, 2016). There is also unscheduled maintenance. This is needed when a turbine suffers to a failure which needs to be solved outside the scheduled service time slots in order to be able to get production from the turbine again. When a failure is registered by the control system in the turbine or by a technician unscheduled service is ordered for the turbine.

2.3 Wind conditions Wind is air that moves and a wind turbine is using this kinetic movement and turning it into rotational work (mechanical work) and thereafter into electrical work. Wind is formed by pressure differences created by temperature differences. Difference in pressure between one place and another will cause a pressure gradient. The Coriolis force will bend the winds towards the equator from northeast and southeast instead of orthogonally (Wizelius, 2007). Friction will affect the wind close to the ground. The friction will work in the opposite direction of the wind direction and hence slow it down and cause changes in the balance between the pressure gradient and the Coriolis force. The friction force will be larger closer to the ground and hence slow the wind down and also change its direction. This change in direction and speed is called shear (Wizelius, 2007). The turbines have increased in height but no matter how high they are they will always be affected by friction. The wind at over 100 meters height over ground will be affected by the shape of the terrain within twenty kilometres. The friction will be larger in a forest than compared to an open area (Wizelius, 2007). The direction of the wind is also changing, not only the speed. In Sweden most of the wind comes from west but it is not as dominating as one may think. It depends on the location in Sweden, at the west coast west winds are dominating but on Gotland most come from south and in the north the direction is varying a lot. The prevailing wind is an important factor when preparing the placement of wind turbines in a wind park (Wizelius, 2007). The easiest way to find the energy content of the wind is to measure the wind with an anemometer. The anemometer can measure direction and speed at one or more heights. These measurements can be used to calculate frequency distribution, energy content and distribution of direction. These calculations can be used to forecast the wind during the lifetime of the turbine (Wizelius, 2007). 2.3.1 Power in the wind & Betz limit Air has the mass of about a kilo per cubic meter. When it is moving it has kinetic energy. This energy can be converted to electricity by a wind turbine. The power in the wind can by calculated by (Wizelius, 2007): 1

𝑃𝑘𝑖𝑛𝑒𝑡𝑖𝑐 = 2 𝑚̇ ∙ 𝑢2 .

(1) -20-

Where, (2)

𝑚̇ = 𝜌 ∙ 𝐴 ∙ 𝑢.

The power is proportional to the power of three of the velocity, hence it is of importance to place a turbine where there is good wind conditions (Wizelius, 2007). The density is dependent of temperature and pressure; it decreases with increasing temperature (Gipe, 2004). The turbine can theoretically exploit 59.3% of the wind power (from the Betz coefficient 16/27≈0.593) and this is called Betz limit. Albert Betz showed that a turbine is delivering largest power when the wind is decelerated with 1/3 by the rotor and 1/3 after the rotor before returning to its original wind speed. The power will naturally be lower than 59.3% of the full potential since there exist aerodynamic and mechanical losses (Wizelius, 2007). 2.3.2 Turbulence Turbulence occurs when the wind meets obstacles in its direction. These obstacles causes swirl and changes the direction from the horizontal plane (laminar wind) into many other directions than the main direction, this is called turbulent wind. Temperature differences can also cause turbulence (Wizelius, 2007). Turbulence is also created when the wind passes the rotor of a turbine and this is called wake and it may cause turbulence downwind of the rotor up to ten rotor diameters away from the turbine (Wizelius, 2007). Hence, it is of importance to have a certain distance in between the turbines in a wind park. A rule of thumb is that an object of height H implies strong turbulence up to a height of twice the height H and to a distance of twice the height H before object and twenty times the height H after the object (Wizelius, 2007). A site with strongly hilly terrain and high mountains, deep valleys and steep inclinations is called complex terrain and these terrains cause many special wind phenomena (Wizelius, 2007). Turbulence will increase wear on the turbine and decrease its effectiveness (Ackerman, 2014). The turbulence intensity (TI) is defined as the ratio between the standard deviation of wind speed and the wind speed (Manwell, et al., 2009): 𝑇𝐼 =

𝜎𝑢 . 𝑢

(3)

Where 𝑢 is the average wind speed and 𝜎𝑢 is the standard deviation of the wind speed of a chosen time interval (see equation 7). An interval of ten minutes is often chosen in wind calculations. The turbulence intensity is often in the range of 10-40%. The highest turbulence intensity is found at low wind speeds (Manwell, et al., 2009). Lower turbulence intensity will give the turbine longer lifetime (Ackerman, 2014). 2.3.3 IEC wind classes Turbines are categorized after which wind conditions it shall endure. A site is also categorized according to the same standard. The categorization is done in order to enable a selection of a suitable turbine type to the site that is robust enough and can endure the loading from the wind at the site. One standard is IEC wind turbine class and it includes reference wind speed and turbulence intensity at hub height. The reference wind speed is rated in classes I, II or III and turbulence in A, B or C (see Table 1). Reference wind speed is the average wind speed during an interval of 10 minutes. The turbine shall be able to endure a climate where the highest speed is the reference wind speed (the extreme average wind speed) with an occurrence of one time in 50 years. Class A means higher turbulence conditions, B medium and C lower (International Electrotechnical Commission, 2005). There is also a class called S. This class is for special wind and external conditions. The design of a S-turbine is chosen by the designer and is specified in its documentation. This class is required for offshore sites and sites with wind conditions such as tropical storms (International Electrotechnical Commission, 2005). -21-

The annual average wind speed was considered as a basic parameter in the earlier standard. This was calculated as (International Electrotechnical Commission, 2005): 𝑢𝑎𝑣𝑒 = 0.2 ∙ 𝑢𝑟𝑒𝑓 ,

(4)

Where 𝑢𝑟𝑒𝑓 is the reference wind speed. A wind turbine of class I is more robust than one of class II or III, hence it will endure higher loads. The size of the rotor can also affect the class of the turbine; a bigger rotor increases the loads on the turbine and the wind class may need to be lowered. If a turbine is exposed to higher turbulence and wind speed than its classification the turbine will most likely see a shorter lifetime of components, higher failures rates and suffer to more breaks and repairs since a turbine with wrong wind class has been chosen. Components that may be more influenced by wrong wind class than others are yaw gear, blades, main bearing and gear boxes (Jakobsen, 2016). Table 1. IEC wind classes from 2005 standard, table modified from IEC (International Electrotechnical Commission, 2005).

Wind turbine class

I

II

III

S

𝒖𝒓𝒆𝒇

50

42.5

37.5

𝒖𝒂𝒗𝒆 *

10

8.5

7.5

Chosen by the designer

A

0.16

B

0.14

C

0.12

*𝑢𝑎𝑣𝑒 has been calculated from equation 4 and is not included in the table from the 2005 standard.

2.4 Wind turbine loads A wind turbine is exposed to many kinds of loads: steady, cyclic, transient, stochastic and resonance-induced loads. When steady wind blows at a stationary or rotating turbine it will induce static loads on various parts of the turbine, especially on the blades when the turbine is rotating (Manwell, et al., 2009). Cyclic loads occur in a cyclic behaviour, these loads are mainly the loads caused by the rotation of the rotor. Transient loads vary with time and occur due to temporary happenings outside the turbine. Stochastic loads are also time varying but occur randomly. Turbulence is a contributing factor to stochastic loads (Manwell, et al., 2009). Resonance induced loads are cyclic loads that occur at the natural frequency of the wind turbine structure and may cause a dynamic response of the structure. These loads may occur due to poor design and should be avoided as much as possible since the vibrations may reach high magnitudes (Manwell, et al., 2009). There are different sources of loads; aerodynamics, gravity, dynamic interactions and mechanical control. Among aerodynamics there are loads that arise from high wind speeds. When the turbine is stopped the drag force contributes most while when rotating the lift force contributes most to the loads of concern (Manwell, et al., 2009). The turbine needs to withstand high loadings that occur over time. The loads are important to know in order to avoid fatigue and to define the ultimate strength of the turbine. During normal operation the turbine suffers to loadings that occur from its starting and stopping, yawing and the constantly changing wind passing through the blades. Fatigue may cause an earlier end of lifetime of the components (Manwell, et al., 2009).

-22-

2.5 Lillgrund Lillgrund is a project started by Eurowind AB in 1997. Vattenfall bought it in 2004 and has since then had the responsibility for procurement and construction of the wind park. At Lillgrund the average wind speed was measured to be 8.5 m/s at 65 meters height during planning of the project. The sea depth is 4-10 meters (Vattenfall, 2009). A request for permission of maximum 48 turbines with 1.5 MW power and a rotor diameter of 65 meters was sent to the government in 1998. It was approved in 2001 and by that time turbines of this size were not available on the market any more. The government removed the power limitation and remaining limits was number of turbines and the size of the area. Vattenfall chose to keep the layout and the number of turbines and built the farm with larger turbines than planned for in the beginning (Vattenfall, 2009). With these new turbines the distance between the turbines would be much tighter than planned and production calculations concluded that the turbines would stand too close to each other to get optimal production. Vattenfall proceeded with unchanged number of turbines with supposed higher production instead of getting maximum revenue with a lower number of turbines (Vattenfall, 2009). The building of the park was started in March 2006. The first turbine was connected to the grid in October 2007 and all 48 turbines were connected in November 2007 with final testing in December 2007. Layout of the park is shown in Figure 3. The predicted production from Lillgrund is 0.33 TWh, which is electricity to 60 000 homes (Vattenfall, 2009). Turbine specifics are shown in Table 2. Lillgrund is referred to as SLG in figures in this report.

Lillgrund layout

Figure 3. Lillgrund layout based on coordinates of the turbines.

-23-

Table 2. Table of Siemens SWT-2.3-93 turbine specifics.

Siemens SWT-2.3-93 Rated power [kW]

2,300 (Siemens, n.d.)

Rotor diameter [m]

93 (Siemens, n.d.)

Hub height [m]

80 or site specific heights (Siemens, n.d.), 69 in Lillgrund, 80 in Lyngsmose and Nørrekær Enge

Generator

Asynchronous (Siemens, n.d.)

Drive

Gearbox (Siemens, n.d.)

Power regulation

Pitch regulation (Siemens, n.d.)

Remote control

WebWPS SCADA system (Siemens, n.d.)

Controller type

WTC-3

IEC-class

IIA (The Wind Power, 2016))

2.6 Other investigated parks Lyngsmose (WLY in figures) is an onshore park in Denmark with two Siemens SWT-2.3-93 turbines (see Figure 4). It was commissioned in 2008. Turbine specifics are explained in Table 2. The site is far from the sea with forests in the surrounding area.

Lyngsmose layout

Figure 4. Lyngsmose layout based on coordinates of the turbines.

Nørrekær Enge (WNE in figures) is an onshore park in Denmark with thirteen Siemens SWT-2.3-93 turbines (see Figure 5). It was commissioned in 2009. Turbine specifics are explained in Table 2. The site is close to the sea with fields in the surrounding area.

-24-

Nørrekær Enge layout

Figure 5. Nørrekær Enge layout based on coordinates of the turbines.

Horns Rev 1 (WH1 in figures) is an offshore park in Denmark with seventy-nine (previously eighty) Vestas V80-2000KW-2G turbines (see Figure 6). It was commissioned in 2002. Turbine specifics are explained in Table 3.

Horns Rev 1 layout

Figure 6. Horns Rev 1 layout based on coordinates of the turbines.

-25-

Table 3. Table of Vestas V80-2000KW-2G turbine specifics.

V80-2000KW-2G Rated power [kW]

2,000 (The Wind Power, 2016)

Rotor diameter [m]

80 (The Wind Power, 2016)

Hub height [m]

60-100 (The Wind Power, 2016), 70 in Horns Rev 1

Generator

Asynchronous (The Wind Power, 2016)

Drive

Gearbox (The Wind Power, 2016)

Power regulation

Pitch regulation (The Wind Power, 2016)

Controller type

VMP5000.2

IEC-class

IA (The Wind Power, 2016)

Stor-Rotliden (SRL in figures) is an onshore park in Sweden with forty turbines (see Figure 7), twenty-nine Vestas V90-2000KW-2G and eleven Vestas V90-1800KW-2G. It was commissioned in 2010. Turbine specifics are explained in Table 4. The site is situated on hills in the forest.

Stor-Rotliden layout

Figure 7. Stor-Rotliden layout based on coordinates of the turbines.

-26-

Table 4. Table of Vestas V90-2000KW-2G & Vestas V90-1800KW-2G turbine specifics.

V90-2000KW-2G

V90-1800KW-2G

Rated power [kW]

2000 (Vestas, n.d.)

1800 (Vestas, n.d.)

Rotor diameter [m]

90 (Vestas, n.d.)

90 (Vestas, n.d.)

Hub height [m]

80-105 (Vestas, n.d.)

80-105 (Vestas, n.d.)

Generator

Asynchronous (The Wind Power, 2016)

Asynchronous (The Wind Power, 2016)

Drive

Gearbox (Vestas, n.d.)

Gearbox (Vestas, n.d.)

Power regulation

Pitch regulation (The Wind Power, 2016)

Pitch regulation (The Wind Power, 2016)

Controller type

VMPGlobal

VMPGlobal

IEC-class

IIA/IIIA (The Wind Power, 2016)

IIA/IIIA (The Wind Power, 2016)

2.7 Statistical methods A statistical analysis often consist of four steps; planning, data collection, processing and presentation. The presentation includes a summary of results with diagrams and further recommendations (Blom, et al., 2005). In order to see if measurements of two variables are correlated there are many methods to use. A scatter diagram gives a rough idea of the relationship (if any) between the measurements. Curve fitting can be used to give the scatter diagram a function. If a straight line can be fitted into the scatter it is said to be a linear relationship, otherwise it is non-linear (Chatfield, 1983). 2.7.1 Correlation coefficient When two variables are random and interrelated they are correlated. The correlation may be strong or weak. If the correlation is linear it is possible to find the correlation coefficient between the two variables. If the scatter diagram indicates a non-linear relationship the correlation coefficient may be misleading and should not be calculated. If the correlation coefficient is positive: large values of both variables correlates. Whereas it is negative if small values of one variable correlates with large values of the other. The correlation is high if the observations lie close to a straight line and it is low if scattered. If there is no relation in between the variables they are uncorrelated. One has to consider that the coefficient may be high even though the correlation does not necessarily indicate an underlying relationship. The data may be influenced by a third variable that causes a simultaneous change in the studied variables (Chatfield, 1983). The correlation coefficient is calculated by the following equation (Chatfield & Collins, 1986): (5)

𝑟𝑖𝑗 = 𝜎𝑖𝑗 /(𝜎𝑖 𝜎𝑗 )

Where 𝜎𝑖𝑗 is the covariance of 𝑖 and 𝑗, 𝜎𝑖 is the standard deviation of 𝑖 and 𝜎𝑗 the standard deviation of 𝑗. Covariance is the measure of association between two variables and is given by (Chatfield, 1983):

-27-

𝜎𝑖𝑗 =

1 ∑𝑛 (𝑥 𝑛−1 𝑖=1 𝑖

− 𝑥̅ )(𝑦𝑖 − 𝑦̅).

(6)

Standard deviation is calculated by (Chatfield, 1983): 2

∑(𝑥𝑖 −𝑥̅ ) 𝜎𝑖 = √ 𝑛−1 .

(7)

The covariance is divided by the standard deviations in order to avoid influence of the scales the measurements are made in (Chatfield, 1983).

-28-

3 Literature review 3.1 Wind power in forests The report Wind power in forests by Elforsk (Bergström, et al., 2013) investigates wind speed and turbulence, especially in forests, by looking at data from forty-two sites and by mesoscale modelling and wind tunnel measurements and simulation of loads from turbulent winds. Results from site data show that the structure in forests has an impact on average wind speed and turbulence. The wind tunnel studies agreed with this at some points and showed that lower wind speed and higher turbulence can be found after the beginning of a forest compared to area upstream of the forest. They also found that turbulence intensity decreases after the forest ends. Clearings affect the wind speed mostly close to the area around the tree tops and wind speed increased over the clearing while turbulence was reduced (Bergström, et al., 2013). Most turbines are certified by the IEC 61400-1 standard from 2005. A model were created to calculate loads and the results shows that turbines in forests may suffer greater fatigue than the IEC61400-1 class turbines can hold (Bergström, et al., 2013). The damage equivalent loads in forest conditions are 35% larger than for turbines of all wind classes (I, II and III) for IEC-A for blade and tower loads. This results in 5% shorter lifetime for blades and 30% for the tower. Consequently, life time critical dimensions have to be increased by 35% to bear the increased loads. Overall, the results indicates that forest wind climate with high turbulence may cause higher fatigue than the current IEC turbine classes can withstand and they conclude that site assessment is an important factor in development of wind turbines in forests (Bergström, et al., 2013).

3.2 Fatigue loads on wind turbine blades in a wind farm The report Fatigue loads on wind turbine blades in a wind farm by Jan-Åke Dahlberg and Maria Poppen investigates four turbines in order to see the influence of wakes on the flap moment in the blades of a turbine. They conclude that wakes are important to consider, foremost in offshore sites where the turbulence and load cycles are low. It also says that complex terrain may diminish the influence of wakes (Dahlberg & Poppen, 1992).

3.3 Study of weather and location effects on wind turbine failure rates The report Study of weather and location effects on wind turbine failure rates by Tavner et al. investigates the influence of weather and location on failure rate and downtime in the turbines in order to understand root causes and consequences. Failure and weather data from three wind turbine sites has been analysed and crosscorrelated; specifically wind speed, maximum temperature and humidity. This study is a deepened study based on the previous study Influence of wind speed on wind turbine reliability by Tavner et al. in 2006. Results from the previous study showed that reliability is reduced when there is high wind speed and between wind speed and failures they found a cross-correlation of 43% with a confidence level of 95%. This study focused on the following farms: Fehmarn, Krummhörn and Ormont in Germany. The results showed that the highest cross-correlation for Krummhörn was maximum wind speed and humidity, for Ormont it was standard deviation of wind speed and for Fehmarn it was maximum wind speed. In conclusion their results support their hypothesis that location and weather affect failures (Tavner, et al., 2012).

-29-

4 Methodology In the initial phase of the project a literature review was conducted to gain knowledge about wind turbines and wind conditions (chapter 2 Theory) and about previous studies (chapter 3 Literature review). Research about theory has been done at the library of Royal Institute of Technology and the library of Stockholm city. Also, knowledge has been gathered by interviewing experts at Vattenfall. The literature review has been conducted by searching in the databases of Royal Institute of Technology and at the following sites: Vindforsk and Elforsk.

4.1 Method of case study The statistical analysis has been conducted in four steps; planning, data collection, processing in Microsoft Excel 2010 and presentation (through summaries and diagrams). The case study includes samples of data from Lillgrund which has Siemens SWT-2.3-93 turbines. 4.1.1 Data collection Turbine anemometer measurements data for Lillgrund was available from 29/09/2007 21:30 to 08/03/2016 10:10 (with an exception for nacelle direction, which was available from 18/05/2011 07:50). The treated data consists of data from 01/01/2008 00:00 to 31/12/2015 23:50, since low availability in the beginning and the benefit of using full years in comparisons. The data was retrieved from the Supervisory Control And Data Acquisition (SCADA) department at Vattenfall and the Vattenfall database Wind Power Data Centre (WPDC). The sample should be representative for all turbines of type Siemens SWT-2.3-93 but not to be compared with the whole fleet of Vattenfall. Table 5 shows the specifics of the collected data. Table 5. Collected data specifics.

Collected data

Interval

Sample rate

Alarms

10 minutes

1 Hz

Wind speed average

10 minutes

1 Hz

Wind speed standard deviation

10 minutes

1 Hz

Time loss

10 minutes

1 Hz

Nacelle direction

10 minutes

1 Hz

Energy availability

Yearly

1 Hz

Time availability

Yearly

1 Hz

4.1.2 Data analysis 4.1.2.1

Summary of data

The variates of the sample data was summarized as an average per turbine unit per year. These were shown graphically in a layout of the park. The model of the layout was created from given coordinates of the turbines. A distribution for the nacelle direction was calculated by finding the amount of angle measurements in each direction in Table 6. The percentage of each direction was shown in a wind rose of the park.

-30-

Table 6. Cardinal and degree directions.

Cardinal Direction

Degree Direction

N – North

348.75 - 11.25

NNE – North-northeast

11.25 - 33.75

NE – Northeast

33.75 - 56.25

ENE – East-northeast

56.25 - 78.75

E – East

78.75 - 101.25

ESE – East-southeast

101.25 - 123.75

SE – Southeast

123.75 - 146.25

SSE – South-southeast

146.25 - 168.75

S – South

168.75 - 191.25

SSW – South-southwest

191.25 - 213.75

SW – Southwest

213.75 - 236.25

WSW – West-southwest

236.25 - 258.75

W – West

258.75 - 281.25

WNW – West-northwest

281.25 - 303.75

NW – Northwest

303.75 - 326.25

NNW – North-northwest

326.25 - 348.75

Distributions of wind speed and turbulence intensity was calculated for the years 2010-2015 in order to compare with distributions of wind speed and turbulence intensity one hour, one day and one week before certain alarms. The distribution was calculated as a summation of all measures of a certain category of wind speed or turbulence intensity, e.g. wind speed 5-6 m/s and turbulence intensity 15-20%. Thereafter, the distribution was plotted in bar diagrams for these categories in percentage. The turbulence intensity (TI) was calculated by equation 3 for each turbine unit in Lillgrund. Then, it was presented as annual averages of the year intervals 2008-2015, 2009-2015 and 2010-2015. The average wind speed (WS) and its standard deviation used in the calculation was of an interval of 10 minutes. Some intervals were missing (lack of data available) and these slots were not possible to use for turbulence intensity calculations. Hence, they are not included in the average turbulence intensity. The wind speed was summarized as an average for the years 2008-2015, 2009-2015 and 2010-2015. The most frequent alarms were found. The average amount of alarms per turbine per year was calculated in order to see the change of amount of alarm during the turbines lifetime. Also, the duration of an alarm were calculated. The time loss was summarized as an average in percent of the years 2008-2015. The data was given in amount of lost seconds in 10 minutes. The seconds of one year were summarized and divided by the total amount of seconds for all 10 minutes intervals that year resulting in a percentage of lost seconds in one year. These yearly percentages were used to calculate the average time loss of 2008-2015. The energy and time availability was summarized as an average of the years 2009-2015.

-31-

4.1.2.2

Correlations

All alarms, time loss, energy availability and time availability of each unit were compared to the wind speed and turbulence intensity in scatter diagrams. Thereafter, the correlation coefficient was calculated by equation 5 by the “correlation” function in Excel. Alarms were compared to wind speed and turbulence intensity in terms of all alarms, selected alarms and blade vibrations only. Selected alarms excluded the alarms that are not influenced by wind conditions and the alarms with undefined nature and posts without information. An example of an alarm not correlated to the studied wind conditions is software updates (see Appendix 8.1.2.1); hence this alarm was excluded among many other similar alarms. The remaining alarms were analysed called “selected alarms” presented in chapter 5.1.3. This was done in order to exclude as many uncertain alarms as possible. Excluded alarms can be found in Appendix 8.5. A large amount of the alarms were blade vibration alarms (8%) and therefore they were compared with the wind conditions as well in order to find the correlation in between. The wind conditions one hour, one day and one week before the blade vibration alarm occurred were also calculated. 4.1.2.3

Average and peak wind speed and turbulence

Average and peak wind speed and turbulence were calculated for the hour, day and week before blade vibrations. The average and peak value for wind speed and turbulence intensity the hour, day and week before was calculated for all 10 minute intervals. With this calculated the blade vibration alarms were connected by its start time to the corresponding average and peak value for the hour, day and week before its start by “IF” and “LOOKUP” functions in Excel. 4.1.2.4

SAP data

There was some data available over the repairs and replacements done in the park. The data range was 20122015. The repairs and replacements were shown in graphs as amount per turbine and year.

4.2 Method of complementary study The complementary statistical analysis includes samples of data from Lyngsmose, Nørrekær Enge, StorRotliden and Horns Rev 1. 4.2.1 Data collection Data was collected the same way as Lillgrund for Lyngsmose, Nørrekær Enge, Stor-Rotliden and Horns Rev 1 (see chapter 4.1.1 and Table 5. Collected data specifics.), but with other time intervals shown in Table 7. Table 7. Data time interval for Lyngsmose, Nørrekær Enge, Stor-Rotliden and Horns Rev 1.

Start date

Stop date

Lyngsmose

01/02/2008 09:30

31/12/2015 23:50

Nørrekær Enge

20/05/2009 10:50

31/12/2015 23:50

Stor-Rotliden

01/01/2012 00:00

31/12/2015 23:50

Horns Rev 1

01/01/2010 00:00

31/12/2015 23:50

-32-

4.2.2 Data analysis The same analysis was performed in the complementary study as for the case study (see chapter 4.1.2):   

Calculation of turbulence intensity Calculation of averages of wind speed, turbulence intensity, time loss, energy availability and time availability Correlations for all variates in Stor-Rotliden and Horns Rev 1 (see results in Appendix 8.1.2)

However, the correlations for Lyngsmose and Nørrekær Enge were not considered separately. Instead all turbines from Lyngsmose, Nørrekær Enge and Lillgrund were combined and correlations determined in order to see the connections of a larger population of same turbine type (Siemens SWT-2.3-93) while having different wind conditions. Correlations was also found for all turbine types combined in order to see the connection between wind conditions and the studied variates.

-33-

5 Results and discussion 5.1 Results of case study 5.1.1 Summary of collected data The sample consists of the following variates from Lillgrund: wind speed, standard deviation of wind speed, alarms, time loss, energy availability and time availability. The sample is representative for turbines of the same type but not for the whole population of Vattenfall’s turbines. 5.1.1.1

Wind conditions

Figure 8 shows the nacelle direction distribution in Lillgrund. The nacelle direction should correspond to the wind direction if the turbine is working properly. Hence, the winds are assumed to be coming from the same directions as the nacelle is directed. West, west-southwest, south-southwest and east are the most frequent directions.

SLG: Nacelle direction distribution 2012-2015 NNW 12% 10% NW 8% 6% WNW 4% 2% W 0%

N NNE NE ENE E

WSW

ESE SW

SE SSW

SSE S

Figure 8. Nacelle direction distribution average of all Lillgrund turbines, 2012-2015.

Figure 9 shows the average wind speed per turbine for the years 2008-2015. The average annual wind speed for each unit is under the IEC-class II (8.5 m/s). Therefore, the wind speed should not induce more wear on the turbines than designed for. The range of average wind speed is 6.25-7.73 m/s and the mean wind speed is 6.82 m/s. The wind speed seems to be higher for the turbines standing in the first row of the main wind directions.

-34-

SLG: WS average 2008-2015 7,20 7,10 7,31 7,34 7,13 7,34 7,56

6,54

6,94

6,87

6,89 7,10

6,95

7,18

7,12

7,38

6,74 6,27 6,64

6,54 6,40 6,61 6,65

6,75 7,06

7,73

6,91

6,65 6,50 6,31 6,57 6,80

6,66

6,25

6,26

6,34

6,42

6,67

6,36 6,52

6,65

6,94

6,65

6,56

6,76

6,84

6,89

7,58

Figure 9. Average wind speed per unit in Lillgrund (meters/second), 2008-2015.

Figure 10 shows the average turbulence intensity per turbine for the years 2008-2015. The average annual turbulence intensity is over the IEC-class A (16% turbulence intensity) for almost all turbines but is lower for some. Therefore, the turbulence intensity at the site may contribute to more wear of the turbine than designed for. The range of average turbulence intensity is 15-21% and the mean turbulence intensity is 18%. The turbines in the middle of the layout tend to have a bit higher turbulence intensity than the ones in the outer parts. The turbulence intensity seems to be lower for the turbines standing in the first row of the main wind directions. Appendix 8.3 and 8.4 shows the wind speed and turbulence intensity average for 20092015 and 2010-2015 which is used in correlations for selected alarms, blade vibration alarms, energy availability and time availability.

SLG: TI average 2008-2015 16% 17% 16% 17% 16% 16% 15%

18%

19% 19%

18% 19% 18% 17%

19% 18% 15%

18% 20% 20%

19% 20%

19% 19%

19% 18%

19%

18% 20% 21% 20% 19%

19%

20%

21%

21%

21%

20%

19%

20%

18%

18%

19%

19%

18%

18%

17%

15%

Figure 10. Average turbulence intensity per unit in Lillgrund (in percentage), 2008-2015.

-35-

5.1.1.2

Alarms

Figure 11 and Figure 12 shows the average annual amount of alarms per turbine in Lillgrund for the years 2008-2015. The range of amount of alarms is 77-168 alarms per turbine and year and the average value is 117 alarms per turbine and year. The average duration of an alarm is 4 hours, 20 minutes and 4 seconds. This means that a turbine has a downtime of around 21 days in one year in the Lillgrund park if the alarms do not appear at the same time.

Amount of alarms per year

SLG: Alarms/year 180 160 140 120 100 80 60 40 20 0

Unit Figure 11. Average annual alarms per unit in Lillgrund, 2008-2015.

SLG: Alarms/year 99 81 126 87 92 119 121

105

77 130

125 109 120 151

121 135 121

168 126 105

105 118 139 125

165 113

123

107 102

113 113 116

121

104

117

156

114

148

110 142

105

115

114

85

96

117

104

107

Figure 12. Average annual alarms per unit in Lillgrund, 2008-2015.

Figure 13 shows the amount of alarms per turbine every year since commission year 2007. One has to keep in mind that the turbines were not fully active whole 2007, the last turbines were installed in November. The first full year was in 2008. By looking at this graph one can see that the alarms are higher in amount in the beginning of the operation and get lower every year except for the years 2008, 2009, 2013 and 2015. This trend can be compared to the bathtub curve (see chapter 2.2.1).

-36-

SLG: Alarms/turbine Amount of alarms per unit

300 250 200 150 100 50 0 2007

2008

2009

2010

2011 Year

2012

2013

2014

2015

Figure 13. Total amount of alarms per unit in Lillgrund every year since commission year 2007.

Table 8 shows the ten most frequent errors in Lillgrund. Total amount of alarms from 2008-2015 is 44,867. Most of these alarms cannot be traced to their origin, for example “Dummy alarm” has no further information attached than it is an alarm that has been transferred from the old system. Therefore some of these were thereby excluded (see appendix 8.5) together with the alarms with known other cause than wind conditions, e.g. “Software update”, the remaining alarms were analysed called “selected alarms” presented in chapter 5.1.3. “Comm failure” and “Comm Failure” are false alarms and therefore not real alarms. Table 8. Most frequent errors of all turbines in Lillgrund 2008-2015.

Alarm name

Quantity 2008-2015

Dummy alarm - Converted from old system

6982

Comm failure

6493

Turbine in local operation

5414

Comm Failure

5190

PI: Old Data!

3496

Blade vibration, waiting

3391

Manual stop

1634

Manual stop, Operating on generator, Stopped due to internal error. Manual reset is 912 required Remote stop - Owner

862

Yaw converter error

724

-37-

5.1.1.3

Time loss

Figure 14 shows the time loss average in percentage per turbine of the years 2008-2015. The time loss average differs from 1.94%-4.91% and have an average value of 3.12%.

SLG: Time loss/year 6,00%

TIme loss (%)

5,00% 4,00% 3,00%

2,00% 1,00% 0,00%

Figure 14. Average annual time loss in Lillgrund, 2008-2015.

5.1.1.4

Energy and time availability

Figure 15 shows the energy availability of the turbines in Lillgrund as an average of the years 2009-2015. The energy availability ranges from 96.77-99.35% and its mean value is 98.56%.

Figure 15. Average annual energy availability per unit in Lillgrund, 2009-2015.

-38-

SLGH03

SLGG05

SLGG03

SLGF06

SLGF04

SLGF02

SLGE06

SLGE03

SLGE01

SLGD07

SLGD04

SLGD02

SLGC08

SLGC06

SLGC04

SLGC02

SLGB08

SLGB06

SLGB04

SLGB02

SLGA07

SLGA05

SLGA03

100,00% 99,50% 99,00% 98,50% 98,00% 97,50% 97,00% 96,50% 96,00% 95,50% 95,00%

SLGA01

Energy availability (%)

SLG: Energy availability/year

Figure 16 shows the time availability of the turbines in Lillgrund as an average of the years 2009-2015. The energy availability ranges from 96.04-98.94% and its mean value is 98.01%.

SLG: Time availability/year Time availability (%)

100,00% 99,00% 98,00% 97,00% 96,00%

95,00%

SLGH03

SLGG05

SLGG03

SLGF06

SLGF04

SLGF02

SLGE06

SLGE03

SLGE01

SLGD07

SLGD04

SLGD02

SLGC08

SLGC06

SLGC04

SLGC02

SLGB08

SLGB06

SLGB04

SLGB02

SLGA07

SLGA05

SLGA03

SLGA01

94,00%

Figure 16. Time availability per unit in Lillgrund, 2009-2015.

5.1.1.5

SAP data – repairs and replacements

Figure 17 and Figure 18 shows the repairs and replacements in Lillgrund. The amount of repairs per turbine is increasing each year while the replacements have an almost constant amount with an exception for year 2012. This may be to an increased wear on the turbine and therefore more components with a need for repair or that the technicians are getting better of reporting these issues in to the SAP system each year. Both converters and the electrical grid and distribution system have many repairs and replacements according to the data, where the converter has been replaced the most. Unfortunately, it will not be possible to draw conclusions from this since there is a lack of reporting from the early years of Lillgrund’s life time. In addition, one has to keep in mind that the every individual technician may do this different and the difference between sites may be significant.

SLG: Repairs Amount of repairs per unit

3 2,5 2 1,5 1 0,5 0 2008

2009

2011

2012 Year

Figure 17. Annual amount of repairs in Lillgrund.

-39-

2013

2014

2015

Amount of replacements per unit

SLG: Replacements 1,2 1 0,8 0,6 0,4 0,2 0 2008

2009

2011

2012 Year

2013

2014

2015

Figure 18. Annual amount of replacements in Lillgrund.

5.1.2 All alarms Figure 19 shows a scatter diagram of alarms and wind speed per unit, 2008-2015. The turbines are widely scattered and there is no certain connection between alarms and wind speed. The correlation coefficient was -0.038 which manifests the fact that alarms and wind speed average are not correlated.

Amount of alarms per unit and year

SLG: Alarms & WS 2008-2015 180 160 140 120 100 80 60 40 20 0

0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 19. Average annual alarms compared with average wind speed for each unit in Lillgrund, 2008-2015.

Figure 20 shows a scatter diagram of alarms and turbulence intensity per unit, 2008-2015. The turbines are widely scattered but it seems to be a slightly positive connection between alarms and turbulence intensity. The correlation coefficient was 0.209 which indicates that there is some correlation between alarms and turbulence intensity average but the scatter is too wide to draw a certain conclusion.

-40-

Amount of alarms per unit and year

SLG: Alarms & TI 180 160 140 120 100 80 60 40 20 0 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 20. Average annual alarms compared with turbulence intensity for each unit in Lillgrund, 2008-2015

5.1.3 Selected alarms Figure 21 displays the annual number of selected alarms per turbine from the years 2009-2015. The amount of alarms ranges between 16-56 alarms per turbine and the mean value is 32. There are fewer alarms in the outer parts of the layout while the turbines in the middle tend to have a bit more failures per unit. The average duration of a selected alarm is 7 hours, 52 minutes and 33 seconds.

SLG: Selected alarms/year 26 16 45 20 25 30 23

23

27

18 45

26 26 34 19

48 50 27

38 28

28 32 34 39

51 31

33

37 27 38 28 31

29 56 29

52 40 44

45 31

24

36

24

19

17

34

16

37

Figure 21. Average annual amount of selected alarms per unit in Lillgrund, 2009-2015.

Figure 22 shows a scatter diagram of selected alarms and wind speed per unit, 2009-2015. The turbines are widely scattered and the correlation coefficient was -0.190. The scatter is too wide to determine a certain correlation between the selected alarms and turbulence intensity average.

-41-

SLG: Selected alarms & WS 2009-2015 Amount of selected alarms per unit and year

60 50 40 30 20 10 0 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 22. Average annual selected alarms and average wind speed for each unit in Lillgrund, 2009-2015.

Figure 23 shows a scatter diagram of selected alarms and turbulence intensity per unit, 2009-2015. The turbines are widely scattered but there seems to be a slightly positive connection between alarms and turbulence intensity. The correlation coefficient was 0.237 which indicates that there is some correlation between alarms and turbulence intensity average but the scatter is too wide to draw a definite conclusion.

SLG: Selected alarms & TI 2009-2015 Amount of selected alarms per unit and year

60 50 40 30 20 10 0 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 23. Average annual selected alarms and average turbulence intensity for each unit in Lillgrund, 2009-2015.

5.1.4 Blade vibrations In Siemens turbines with the controller type WTC-3 blade and rotor vibrations are measured and if they are out of safe levels the turbine shuts down and sends an alarm to the surveillance centre either “Blade vibrations, waiting” or “Blade vibrations, stop”. If “Blade vibrations, waiting” occurs the turbine starts again automatically when the climate is stabilized. However, if “Blade vibrations, stop” occurs the turbine will need a manual restart. The blade vibrations considered in this study are “Blade vibrations, waiting”. Blade vibrations represented a large share of the total amount of alarms in Lillgrund, 8%. They were also likely to be correlated to the selected wind conditions. Therefore these alarms were studied deeper. An average blade vibration alarm is 49 minutes and 18 seconds.

-42-

Figure 24 shows a bar diagram of the amount of blade vibration alarms every month during 2010-2015. A common month of blade vibration alarms is March as well as January and February.

SLG: Blade vibration alarms per month 700

Amount of blade vibration alarms

600 500 400 300 200 100 0 1

2

3

4

5 6 7 8 Number of month

9

10

11

12

SLGH04 SLGH02 SLGG04 SLGG02 SLGF05 SLGF03 SLGE07 SLGE04 SLGE02 SLGD08 SLGD06 SLGD03 SLGD01 SLGC07 SLGC05 SLGC03 SLGC01 SLGB07 SLGB05 SLGB03 SLGB01 SLGA06 SLGA04 SLGA02

SLGH03 SLGG05 SLGG03 SLGF06 SLGF04 SLGF02 SLGE06 SLGE03 SLGE01 SLGD07 SLGD04 SLGD02 SLGC08 SLGC06 SLGC04 SLGC02 SLGB08 SLGB06 SLGB04 SLGB02 SLGA07 SLGA05 SLGA03 SLGA01

Figure 24. Blade vibrations per month, accumulated for the years 2010-2015 in Lillgrund.

Figure 25 shows the average annual number of blade vibrations that occurred during 2010-2015. The graph indicates that outer turbines in the layout tend to have less blade vibrations while the inner turbines have more blade vibration alarms. Also, they are less common in the main wind directions.

SLG: Blade vibration alarms/year 7 5 2

7

0 9 10 5

7

2

18

29

2 10 18 0

26

25 4

8

2

18

11

16 15 26

26 13

15

21 17

11

15

14 25

23 1

10

29 6

8

9

6

6

4

17

3

10

Figure 25. Average annual amount of blade vibration alarms per unit in Lillgrund, 2010-2015.

-43-

Figure 26 shows a scatter diagram of blade vibrations and wind speed per unit, 2010-2015. The turbines have a very wide spread in number of blade vibration alarms. The correlation coefficient is -0.329 but the scatter is too wide to draw a certain conclusion.

Amount of blade vibrations per unit and year

SLG: Blade vibrations & WS 35 30 25 20 15 10 5 0 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 26. Average annual amount of blade vibrations and average wind speed for each unit in Lillgrund, 2010-2015.

Figure 27 shows a scatter diagram of blade vibrations and turbulence intensity per unit, 2010-2015. The turbines are scattered but a positive connection between alarms and turbulence intensity can be seen in the graph. The correlation coefficient was 0.451 which indicates that there is some correlation between blade vibration alarms and turbulence intensity average but the scatter is too wide to draw a certain conclusion.

Amount of blade vibrations per unit and year

SLG: Blade vibrations & TI 35 30 25

20 15 10 5 0 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 27. Average annual amount of blade vibrations and average turbulence intensity for each unit in Lillgrund, 20102015.

Figure 28-Figure 30 shows the percentage of total average wind speed one hour, one day and one week before a blade vibration alarm occurs for the years 2010-2015. It is interesting to investigate this in order to see if there is a higher wind speed and/or turbulence intensity than the average wind speed and turbulence intensity at the park. This helps to understand which wind conditions actually set off an alarm and contribute to downtime. The graphs indicates that the most common wind speed an hour before is 7-8 m/s, 7-9 m/s a day before and around 6-7 m/s a week before. This is a bit higher in comparison to the overall wind speed -44-

distribution 2010-2015 (Figure 60). This contributes to a correlation between a higher wind speed and a higher amount of blade vibration alarms.

SLG: WS one hour before blade vibration alarm 25,00% 20,00% 15,00%

10,00% 5,00% 0,00% 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

Figure 28. Percentage of total average wind speed the hour before a blade vibration alarm in Lillgrund, 2010-2015.

SLG: WS one day before blade vibration alarm 14,00% 12,00% 10,00% 8,00% 6,00% 4,00% 2,00% 0,00% 1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17 18 19 20 21 22

Figure 29. Percentage of total average wind speed the day before a blade vibration alarm in Lillgrund, 2010-2015.

-45-

SLG: WS one week before blade vibration alarm 30,00% 25,00% 20,00% 15,00% 10,00% 5,00% 0,00% 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Figure 30. Percentage of total average wind speed the week before a blade vibration alarm in Lillgrund, 2010-2015.

Figure 31- Figure 33 shows the percentage of total average turbulence intensity the one hour, one day and one week before a blade vibration alarm occurs for the years 2010-2015. The graphs indicate that the most common turbulence intensity level an hour before is between 15-20%, as well as for the day and week before. This is a bit higher in comparison to the overall turbulence intensity distribution of 2010-2015 where the most common intensity is 10-15% (Figure 61). Consequently, a trend can be seen that the turbulence intensity one hour, one day and one week before a blade tends to be higher than usual and this contributes to that there is a positive correlation between high amounts of blade vibration alarms for higher turbulence intensity.

SLG: TI one hour before blade vibration alarm 60,00%

50,00%

40,00%

30,00%

20,00%

10,00%

0,00% 0,05

0,1

0,15

0,2

0,25

0,3

0,35

0,4

Figure 31. Percentage of total average turbulence intensity the hour before a blade vibration alarm in Lillgrund, 20102015.

-46-

SLG: TI one day before blade vibration alarm 45,00% 40,00% 35,00% 30,00% 25,00% 20,00% 15,00%

10,00% 5,00% 0,00% 0,05

0,1

0,15

0,2

0,25

0,3

0,35

0,4

0,45

0,5

Figure 32. Percentage of total average turbulence intensity the day before a blade vibration alarm in Lillgrund, 2010-2015.

SLG: TI one week before blade vibration alarm 45,00% 40,00% 35,00% 30,00% 25,00% 20,00% 15,00% 10,00% 5,00% 0,00% 0,05

0,1

0,15

0,2

0,25

0,3

0,35

0,4

Figure 33. Percentage of total average turbulence intensity the week before a blade vibration alarm in Lillgrund, 20102015.

The figures in appendix 8.1.2.1 shows that there is a strong correlation for the turbulence intensity an hour before a blade vibration alarm and amount of blade vibration alarms. This strengthens the assumption that there is a connection between high turbulence intensity and many blade vibration alarms.

-47-

5.1.5 Time loss Figure 34 displays a scatter diagram of time loss in percentage of total time of a year and average wind speed per unit, 2008-2015. The turbines are widely scattered but there is a slightly positive connection between time loss and wind speed. The correlation coefficient was 0.309 which indicates that there is some correlation between time loss and average wind speed. Nonetheless, the scatter is too wide to draw a certain conclusion.

SLG: Time loss & WS 2008-2015 6,00%

TIme loss (%)

5,00% 4,00% 3,00% 2,00% 1,00% 0,00% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 34. Annual average time loss in percent and average wind speed for each unit in Lillgrund, 2008-2015.

Figure 35 shows time loss in in percentage of total time of a year and average turbulence intensity per unit, 2008-2015. The turbines are widely scattered in amount of lost seconds and correlation coefficient is 0.2023. However, the scatter is too wide to draw a definite conclusion.

SLG: Time loss & TI 2008-2015 6,00%

TIme loss (%)

5,00% 4,00% 3,00% 2,00% 1,00% 0,00% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 35. Average annual time loss in percent and average turbulence intensity for each unit in Lillgrund, 2008-2015.

-48-

5.1.6 Energy availability and time availability Figure 36 shows a scatter diagram of energy availability and wind speed per unit, 2009-2015. The turbines have some differences in availability and the correlation coefficient is very low, -0.116. Consequently, no correlation can be seen between these parameters.

SLG: Energy availability & WS 2009-2015 Energy availability (%)

100,00% 99,50% 99,00% 98,50% 98,00% 97,50% 97,00% 96,50% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 36. Energy availability in percentage and average wind speed for each unit in Lillgrund, 2009-2015.

Figure 37 shows a scatter diagram of energy availability and turbulence intensity per unit, 2009-2015. The turbines are widely scattered and it seems like there is no connection between alarms and turbulence intensity. The correlation coefficient was 0.054 which contributes to the fact that energy availability and turbulence intensity are not correlated.

Energy availability per unit (%)

SLG: Energy availability & TI 2009-2015 100,00% 99,50%

99,00% 98,50% 98,00% 97,50% 97,00% 96,50% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 37. Average annual energy availability in percentage and average turbulence intensity for each unit in Lillgrund, 2009-2015.

Figure 38 shows time availability and average wind speed per unit, 2009-2015. The turbines have a spread of time available and wind speed. The correlation coefficient is -0.351 which indicates a low correlation between time availability and average wind speed. Nonetheless, the scatter is too wide to draw a definite conclusion.

-49-

Time availability (%)

SLG: Time availability & WS 2009-2015 100,00% 99,50% 99,00% 98,50% 98,00% 97,50% 97,00% 96,50% 96,00% 95,50% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 38. Time availability in percentage and average wind speed for each unit in Lillgrund, 2009-2015.

Figure 39 displays a scatter diagram of time availability and turbulence intensity per unit, 2009-2015. The turbines are widely scattered but it appears to be a slightly a positive connection between time availability and turbulence intensity. The correlation coefficient was 0.288 which indicates that there is some correlation between time availability and average turbulence intensity. However, the scatter is too wide to draw a certain conclusion.

Time availability (%)

SLG: Time availability & TI 2009-2015 100,00% 99,50% 99,00% 98,50% 98,00% 97,50% 97,00% 96,50% 96,00% 95,50% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 39. Time availability in percentage and average turbulence intensity for each unit in Lillgrund, 2009-2015.

-50-

5.2 Results of Siemens SWT-2.3-93 turbine 5.2.1 Lyngsmose Figure 40 shows the amount of alarms per turbine every year since commission year 2008 in Lyngsmose. One has to keep in mind that the turbines were not in full operation until February 2008. The first full year with producing turbines with was in 2009. By looking at this graph one can see that the alarms are slightly higher in amount in the beginning of the operation and then get lower. This trend does roughly follow the bathtub curve (see chapter 2.2.1 Failure curve).

Amount of alarms per unit

WLY: Alarms/turbine 200 180 160 140 120 100 80 60 40 20 0

2008

2009

2010

2011

2012

2013

2014

2015

Year Figure 40. Amount of alarms per unit every year in Lyngsmose since commission year 2008.

5.2.2 Nørrekær Enge Figure 41 shows the amount of alarms per turbine every year since commission year 2009 in Nørrekær Enge. One has to keep in mind that the turbines were not in full operation until August 2009. The first full year with producing turbines was in 2010. By looking at this graph one can see that the alarms are slightly higher in amount in the beginning of the operation and then get lower every year. This could almost apply to the bathtub curve (see chapter 2.2.1 Failure curve).

Amount of alarms per unit

WNE: Alarms/turbine 90 80 70 60 50 40 30 20 10 0 2009

2010

2011

2012 Year

2013

2014

Figure 41. Amount of alarms per unit every year in Nørrekær Enge since commission year 2009.

-51-

2015

5.2.3 Combination of Siemens SWT-2.3-93 turbines Figure 42 displays the correlation between amount of alarms per turbine and year and the average wind speed of each turbine for the sites Lillgrund, Lyngsmose and Nørrekær Enge. There seems to be a slightly negative correlation between the parameters.

Amount of alarms per unit and year

Alarms & WS 180 160 140 120 100 80 60 40 20 0

SLG WLY WNE

0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 42. Amount of annual alarms and average wind speed for each unit in SLG, WLY and WNE.

Figure 43 displays the correlation between number of alarms per turbine and year and the average turbulence intensity of each turbine for the sites Lillgrund, Lyngsmose and Nørrekær Enge. There seems to be a positive correlation between the parameters.

Number of alarms per year

Alarms & TI 180 160 140 120 100 80 60 40 20 0

SLG WLY

WNE

0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 43. Amount of annual alarms and average turbulence intensity for each unit in SLG, WLY and WNE.

-52-

Figure 44 displays the correlation between number of blade vibration alarms per turbine and year and the average wind speed of each turbine for the sites Lillgrund, Lyngsmose and Nørrekær Enge. There seems to be a slightly negative correlation between the parameters but they are widely scattered.

Numer of blade vibration alarms per year

Blade vibration alarms & WS 35

30 25 20

SLG

15

WLY

10

WNE

5 0 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 44. Amount of annual blade vibration alarms and average wind speed for each unit in SLG, WLY and WNE.

Figure 45 displays the correlation between number of blade vibration alarms per turbine and year and the average turbulence intensity of each turbine for the sites Lillgrund, Lyngsmose and Nørrekær Enge. There seems to be a positive correlation between the parameters.

Number of blade vibration alarms per year

Blade vibration alarms & TI 35 30 25 20

SLG

15

WLY

10

WNE

5 0 0%

5%

10% 15% Turbelence intensity (%)

20%

25%

Figure 45. Amount of annual blade vibration alarms and average turbulence intensity for each unit in SLG, WLY and WNE.

-53-

Figure 46 and Figure 47 shows the time loss and wind speed and turbulence intensity. The turbines are widely scattered and do not have any correlations.

Time loss & WS 8%

Time loss (%)

7% 6% 5%

4%

SLG

3%

WLY

2%

WNE

1% 0% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 46. Time loss and average wind speed for each unit in SLG, WLY and WNE.

Time loss & TI 8%

Time loss (%)

7% 6% 5% 4%

SLG

3%

WLY

2%

WNE

1%

0% 0%

5%

10% 15% Turbulence intensity (%)

20%

Figure 47. Time loss and average turbulence intensity for each unit in SLG, WLY and WNE.

-54-

25%

Figure 48 and Figure 49 shows the energy availability and wind speed and turbulence intensity. The turbines are scattered and do not have any clear correlations.

Energy availability & WS Energy availability (%)

100% 98% 96% 94%

92%

SLG

90%

WLY

88%

WNE

86% 84% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 48. Energy availability and average wind speed for each unit in SLG, WLY and WNE.

Energy availability & TI Energy availability (%)

100% 98% 96% 94% 92%

SLG

90%

WLY

88%

WNE

86%

84% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 49. Energy availability and average turbulence intensity for each unit in SLG, WLY and WNE.

-55-

Figure 50 and Figure 51 shows the time availability and wind speed and turbulence intensity. The turbines are widely scattered and there is no visible correlation.

Time availability & WS 100%

Time availbility (%)

99% 98% 97%

96%

SLG

95%

WLY

94%

WNE

93% 92% 0

1

2

3

4 5 Wind speed (m/s)

6

7

8

9

Figure 50. Time availability and average wind speed for each unit in SLG, WLY and WNE.

Time availability & TI 100%

Time availbility (%)

99% 98% 97% 96%

SLG

95%

WLY

94%

WNE

93%

92% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 51. Time availability and average turbulence intensity for each unit in SLG, WLY and WNE.

5.3 Results of Stor-Rotliden Figure 52 shows the amount of alarms per turbine every year since commission year 2010 in Stor-Rotliden. One has to keep in mind that the turbines were not in full operation in 2010. The first full year with producing turbines with was in 2011. By looking at this graph one can see that the alarms are around the same level but lowers a bit each year. This can almost be applied to the bathtub curve theory (see chapter 2.2.1 Failure curve).

-56-

SRL: Alarms/turbine Amount of alarms per unit

160 140 120 100 80 60 40 20 0 2010

2011

2012

2013

2014

2015

Year Figure 52. Amount of alarms per unit every year in Stor-Rotliden since commission year 2010.

The correlations of Stor-Rotliden are shown in figures in appendix 8.1.2.2. These figures show a weak trend of higher energy availability and time availability for higher turbulence intensity. Otherwise the turbines are too scattered to find any correlations.

5.4 Results of Horns Rev 1 Figure 53 shows the amount of alarms per turbine each year for the years with available data, the park was commissioned in 2002. Unfortunately, year 2002-2006 are missing. Since this lack of data the infant failure alarms levels are missing and it is hard to compare this with the bathtub curve (see chapter 2.2.1). Also, it could be that some alarms are missing for 2007 since they are so low in amount while the others have similar totals. One can see that the amount of alarms is higher year 2015 than the years before; this may be because the wear-out period has started. This would be an early start of the wear out period since the park only is 14 years old yet. To draw a certain conclusion one has to get data from the coming years as well.

WH1: Alarms/turbine Amount of alarms per unit

120 100 80 60 40 20 0 2007

2008

2009

2010

2011 Year

2012

2013

2014

2015

Figure 53. Amount of alarms per unit in Horns Rev 1, data missing 2002-2006.

The correlations of Horns Rev 1 are shown in figures in Appendix 8.1.2.3. By looking at these figures no strong correlation can be found since the turbines are too scattered. -57-

5.5 Results of all turbines combined Figure 54 and Figure 55 displays the time loss and wind speed and turbulence intensity for all turbines in the investigated parks. It is hard to find a correlation here because the turbines are scattered and the time loss seems to be around the same for all wind speeds and turbulence intensities.

Time loss (%)

Time loss & WS 18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

SLG WLY WNE SRL WH1 0

2

4

6 Wind speed (m/s)

8

10

12

Figure 54. Time loss and wind speed for all studied turbines.

Time loss (%)

Time loss & TI 18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

SLG WLY WNE SRL WH1 0%

5%

10% 15% Turbulence intensity (%)

Figure 55. Time loss and turbulence intensity for all studied turbines.

-58-

20%

25%

Figure 56 and Figure 57 shows the energy availability and wind speed and turbulence intensity for all turbines in the investigated parks. It is hard to find a correlation here because the turbines are scattered but a weak trend can be seen for a lower wind speed and higher turbulence there is higher availability.

Energy availability & WS Energy availability (%)

100%

95% SLG

90%

WLY 85%

WNE SRL

80%

WH1

75% 0

2

4

6 Wind speed (m/s)

8

10

12

Figure 56. Energy availability and wind speed for all studied turbines.

Energy availability & TI Energy availability (%)

100% 95% SLG

90%

WLY 85%

WNE SRL

80%

WH1

75% 0%

5%

10% 15% Turbulence intensity (%)

Figure 57. Energy availability and turbulence intensity for all studied turbines.

-59-

20%

25%

Figure 58 and Figure 59 displays the time availability and wind speed and turbulence intensity for all turbines in the investigated parks. There is no correlation for these parameters and the time availability seems independent of the wind conditions.

Time availability & WS 100%

Time availbility (%)

98% 96% 94%

SLG

92%

WLY

90%

WNE

88%

SRL

86%

WH1

84% 0

2

4

6 Wind speed (m/s)

8

10

12

Figure 58. Time availability and wind speed for all studied turbines.

Time availability & TI 100%

Time availbility (%)

98% 96% 94%

SLG

92%

WLY

90%

WNE

88%

SRL

86%

WH1

84% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 59. Energy availability and turbulence intensity for all studied turbines.

5.6 Summative discussion By looking at the graphs of the wind speed and turbulence intensity for the layout of Lillgrund (Figure 9 and Figure 10) one can see that wind speeds are higher and turbulence intensity are lower in the main wind directions (Figure 8). Also, there are lower wind speeds and higher turbulence intensities in the inner parts of the layout. The wind speed and turbulence is probably affected by wakes from the other turbines resulting in lower wind speed and higher turbulences after a turbine. For time loss (Figure 62), energy availability (Figure 63) and time availability (Figure 64) it is hard to tell if the availabilities have any connection to the place in the layout. Altogether, wakes seems to contribute to the wind conditions and amount of alarms but not any visible impact on time loss and the availabilities. Regarding all figures with alarms for the Siemens turbines high turbulence intensity seem to be correlated to a higher amount of alarms. Still, there is high energy availability for the same wind conditions. From this -60-

it can be said that alarms do not affect energy availability considerably. One explanation to this may be that when there is high turbulence it is also likely that it is low wind speed (this trend can be seen for in Figure 9 and Figure 10) and when the turbine suffered to an alarm it did not loose much energy production while it was stopped. Many of the scatter diagrams cannot be considered as linear relationships. When the relationship is nonlinear one has to be very careful to draw definite conclusions based on the correlation factor. Since no relations in this document seem to be truly linear one has to keep in mind that the correlation factors may be misleading in some cases. To see the relationship between amount of alarms in the beginning, operating period and end of a turbines lifetime is interesting in terms of when it starts to increase again after the infancy period. This has been illustrated in Figure 13, Figure 40, Figure 41, Figure 52 and Figure 53. For Horns Rev 1 (Figure 53) one can wonder whether the wear-out period already has started off or if the amount of alarms the 14th year are an exception compared to the coming years. It is interesting to see the trend the upcoming years for all the parks. The wind conditions one hour, one day and one week before a blade vibration alarm were investigated in order to see if these wind conditions differed from the average annual conditions. From this investigation it was seen that there is slightly higher wind speed and turbulence intensity before an alarm occurred. This is very interesting and contributes to the hypothesis that complex wind conditions actually induce more alarms than calmer climates.

5.7 Uncertainties There are always factors of uncertainty when performing studies. In this case one of them is the wind measurements from the anemometer. The anemometer is placed behind the rotor and will therefore measure values influenced by the rotor. It will probably reduce the wind speed and increase the turbulence intensity but the degree of influence is unknown. There is also the possibility of technical errors from wind measurements or in the data about nacelle direction, time loss, energy availability and time availability. There has been a lack of data for some time intervals leading to less data availability and less credibility of the data. In turn, this lack of data leads to weakened credibility of overall result since it is based on a smaller sample than it was planned for. Regarding selected alarms one have to keep in mind that by excluding some alarms a human uncertainty is introduced. Exclusion of alarms is affected by which knowledge the investigator has concerning alarms and its nature. Another factor is that the population may have been too little. A bigger population would have given a result with higher reliability. There can also be disturbing factors in the variates which may lead to an apparent correlation between the variates and wind conditions even though it does not exist, or vice versa. An additional uncertainty is that correlations may be random and true for only these cases but not true in general.

-61-

6 Conclusions By looking at the results a lower wind speed and a higher turbulence can be seen in the middle of the layout in Lillgrund. There is also higher wind speed and turbulence at the turbines in the outer rows of the main wind directions. For Siemens SWT-2.3-93 there are a higher amount of alarms and blade vibration alarms for higher turbulence intensities while there are no strong correlations for turbulence intensity and energy availability and time availability in Siemens SWT-2.3-93. This leads to the connection that alarms do not have a visible influence on the energy availability and time availability of Siemens SWT-2.3-93. All graphs containing alarms per year for the turbines seem to follow the bathtub curve trend but it is too early to conclude if certain wind conditions has affected the start of the wear-out period. This will be possible to tell in a few years when the parks are closer to the end of designed life time. The wind speed and turbulence intensity one hour, one day and one week before a blade vibration alarm was higher than the average wind speed and turbulence intensity. This contributes to the possibility that wind speed and turbulence intensity increases the risk of a blade vibration alarm. Unfortunately by looking at the correlations from the Vestas turbines there is hard to tell if a true correlation exits. Therefore, the Vestas turbines cannot contribute to a correlation between wind conditions and time loss, energy availability and time availability.

6.1 Future work From the conclusions it is recommended that the following matters are examined further: 







In order to get a stronger conclusion regarding correlations of wind conditions and the studied parameters a larger population of each turbine type should be examined. Since there were many alarms without information in the data of this study a larger population would increase the amount of data with useful information. This would give the result a higher credibility. To get a better perspective of when the wear-out period of the designed life time starts for a site with certain wind conditions a longer time period should be examined for each turbine type. This can only be done when more years of the designed lifetime has passed for the parks. There are not many parks yet which are close to the end of its designed life time. It would also be interesting to investigate shear. This may be a wind condition that contributes to wear on turbines. Measurements at different heights at the parks are necessary when studying shear. Contributions of uncertainties should be examined in order to see how much the results would differ if the uncertainties were reduced. Calculation of how much the rotor affects the measurements of the anemometer could give a rule of thumb of how the rotor reduces the wind speed and how much it increase the turbulence intensity.

6.2 Suggestions for Vattenfall In order to be able to calculate amount of failures, repairs and replacements it is necessary to have more data from the SAP system. It would then be possible to see all maintenance performed in each turbine and this would enable cost calculations. If every service of a turbine from the first day of operation until the last day were reported into SAP it would be possible to see exactly how many components that have been replaced or repaired and from this it would be possible to calculate the total cost of the turbines maintenance. This would increase the knowledge about when and which components fail in a turbine during its designed lifetime. -62-

Another suggestion concerns the extraction of data. During this project it was not possible to extract data about the turbines by oneself and instead requests have been sent to SCADA to help with this matter. In order to spare them extra work it would be good if there were handbooks of how to extract this data by oneself and also access to the database. In these handbooks it would be good to include definitions of the data which makes it easy to understand what the data can be used for.

-63-

7 References Ackerman, T., 2014. Wind Energy 4th lecture. [Sound Recording] (Royal Institute of Technology). Ackerman, T., 2014. Wind Energy 5th lecture. [Sound Recording] (Royal Institute of Technology). Alewine, K. & Chen, W., 2012. A review of Electrical Winding Failures in Wind Turbine Generators. IEEE Electrical Insulation Magazine, 28(4), pp. 392-397. Bergström, H. et al., 2013. Wind power in forests - Winds and effects on loads, Stockholm: Vindforsk. Blom, G. et al., 2005. Sannolikhetsteori och statistikteori med tillämpningar. 5th ed. Lund: Studentlitteratur. Chatfield, C., 1983. Statistics for technology. 3rd ed. London: Chapman & Hall. Chatfield, C. & Collins, A., 1986. Introduction to multivariate analysis. London: Chapman & Hall. Dahlberg, J.-Å. & Poppen, M., 1992. Fatigue loads on wind turbine blades in a wind farm, s.l.: FFA. Freudenreich, K., 2016. Blade vibrations & failures [Interview] (24 May 2016). Gipe, P., 2004. Wind power. 1st ed. London: James & James (Science publishers) Ltd. Global Wind Energy Council, 2016. Global statistics. [Online] Available at: http://www.gwec.net/wp-content/uploads/vip/GWEC-PRstats-2015_LR_corrected.pdf [Accessed 7 March 2016]. Horste, A. & El-Thalji, I., 2011. Växellådshaverier på landbaserade vindkraftverk, Stockholm: Elforsk. International Electrotechnical Commission, 2005. Wind turbines – Part 1: Design requirements, s.l.: s.n. Jakobsen, V. H., 2016. IEC wind class [Interview] (18 May 2016). Lindqvist, M. & Lundin, J., 2010. Spare Part Logistics and Optimization for Wind Turbines - Methods for CostEffective Supply and Storage, s.l.: Uppsala Univeristet. Manwell, J., McGowan, J. & Rogers, A., 2009. Wind energy explained - theory, design and application. 2nd ed. Chippenham, Wiltshire: John Wiley & Sons Ltd.. Ribrant, J. & Bertling, L., 2007. Survey of failures in wind power systems with focus on Swedish wind power plants during 1992-2005. Tampa, Florida, IEEE. Siemens, n.d. Wind Turbine SWT-2.3-93. [Online] Available at: http://www.energy.siemens.com/nl/en/renewable-energy/wind-power/platforms/g2platform/wind-turbine-swt-2-3-93.htm [Accessed 4 May 2016]. Svensk Vindenergi, 2016. Statistik om vindkraft. [Online] Available at: http://www.vindkraftsbranschen.se/wp-content/uploads/2016/02/Statistik-och-prognosvindkraft-20160218.pdf [Accessed 7 March 2016]. Tavner, P. et al., 2012. Study of weather and locations effects on wind turbine failute rates. Wind energy, 9 May, pp. 175-187. The Wind Power, 2016. SWT-2.3-93. [Online] Available at: http://www.thewindpower.net/turbine_en_22_siemens_2300.php [Accessed 3 June 2016]. The Wind Power, 2016. V80/2000. [Online] Available at: http://www.thewindpower.net/turbine_en_30_vestas_v80-2000.php [Accessed 27 May 2016].

-64-

The Wind Power, 2016. V90/1800. [Online] Available at: http://www.thewindpower.net/turbine_en_31_vestas_1800.php [Accessed 3 June 2016]. The Wind Power, 2016. V90/2000. [Online] Available at: http://www.thewindpower.net/turbine_en_32_vestas_2000.php [Accessed 3 June 2016]. Vattenfall AB, 2013. Så fungerar vindkraft. [Online] Available at: http://corporate.vattenfall.se/om-energi/el-och-varmeproduktion/vindkraft/sa-fungerarvindkraft/ [Accessed 8 March 2016]. Vattenfall AB, 2015. Vindkraft. [Online] Available at: http://corporate.vattenfall.se/om-oss/var-verksamhet/var-elproduktion/vindkraft/ [Accessed 7 March 2016]. Vattenfall, 2009. Lillgrund Vindkraftpark - Ett svenskt pilotprojekt inom havsbaserad vindkraft. Värnamo: Fält & Hässler. Vestas, n.d. V90-1.8/2.0 MW® at a Glance. [Online] Available at: https://www.vestas.com/en/products/turbines/v90-2_0_mw#!options-available [Accessed 27 May 2016]. Vilminko, A., 2016. Maintenance of turbines [Interview] (20 May 2016). Wizelius, T., 2007. Vindkraft i teori och praktik. 2:1 ed. s.l.:Studentlitteratur.

-65-

8 Appendices 8.1 Figures 8.1.1 Data summary of Lillgrund

SLG: WS distribution 2010-2015 12%

Distribution

10% 8% 6% 4% 2% 0% 1

2

3

4

5

6

7

8 9 10 11 Wind speed (m/s)

12

13

Figure 60. Wind speed distribution in Lillgrund, 2010-2015.

SLG: TI distribution 2010-2015 30% Distribution

25% 20% 15%

10% 5% 0%

Turbulence intensity (%) Figure 61. Turbulence intensity distribution in Lillgrund, 2010-2015.

-66-

14

15

16

17

SLG: Time loss/year 3% 2% 4% 4% 3% 3% 4%

2%

2% 3%

3% 3% 3%

3%

3% 4%

4% 2% 2%

3% 3% 3% 4%

4% 3%

3%

5%

3% 3% 3% 2%

4%

3%

3%

3%

4%

4%

3%

3% 3%

2%

5%

3%

2%

2%

4%

3%

3%

Figure 62. Average annual time loss in Lillgrund, 2008-2015.

SLG: Energy availability 99% 99% 98% 97% 99% 99% 98%

99%

99% 99%

99% 99% 99%

99%

98% 98% 99%

97% 99% 99%

98% 99% 98% 99%

99% 99%

97%

99% 99% 99% 99%

98%

98%

99%

98%

99%

97%

99%

99% 99% 99%

99%

97%

99%

99%

Figure 63. Average annual energy availability in Lillgrund, 2008-2015.

-67-

99% 99%

97%

SLG: Time availability 98% 99% 97% 97% 98% 98% 97%

99%

99% 98%

99% 99% 99%

98%

98% 97%

97% 99% 99%

98% 99% 98% 98%

98% 98%

98%

97%

98% 99% 98% 99%

97%

98%

98%

98%

98%

98%

98%

98% 98%

99%

96%

98%

98%

99%

97%

98%

98%

Figure 64. Average annual time availability in Lillgrund, 2008-2015.

SLG: Average WS hour before a blade vibration alarm 2010-2015 8,42 8,00 9,96 8,49 8,14 9,11 7,96

8,20

8,78 8,82

9,06 8,46 8,21 8,31

8,57 7,91 8,32

9,13 8,38 9,01

8,76 9,17 8,85 7,63

8,26

9,17

9,24

8,13 8,70 8,86 8,45 8,61

8,11

8,56

7,82

8,20

10,09

9,87

7,47 8,82

8,93

9,48

9,69

9,23

9,63

8,77

10,56

9,94

Figure 65. Average wind speed the hour before a blade vibration alarm in each unit, 2010-2015.

-68-

SLG: Average TI hour before a blade vibration alarm 2010-2015 19% 19% 17% 18% 18% 18% 20%

19%

18%

16% 19%

15% 19% 19%

19% 20%

15%

19%

18%

18%

19%

18% 20%

20%

19%

20%

17%

19% 19% 19% 20% 17%

19%

19%

20%

20%

16%

17%

20% 20%

19%

19%

18%

18%

17%

18%

16%

18%

Figure 66. Average turbulence intensity the hour before a blade vibration alarm in each unit in Lillgrund, 2010-2015.

8.1.2 Correlations 8.1.2.1

Lillgrund

SLG: Software updates & WS 12

Amount of alarms

10 8 6 4 2 0 0

1

2

3

4 5 Wind speed (m/s)

Figure 67. The amount of software update alarms and wind speed in Lillgrund.

-69-

6

7

8

9

SLG: Software updates & TI 12 Amount of alarms

10 8 6 4 2 0 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 68. The amount of software update alarms and turbulence intensity in Lillgrund.

Amount of blade vibration alarms per unit and year

SLG: Blade vibration alarms & WS one hour before, 2010-2015 35 30 25 20 15 10 5 0 0

2

4

6 Wind speed (m/s)

8

10

12

Figure 69. Amount of blade vibration alarms per year and the wind speed an hour before a blade vibration alarm for each unit in Lillgrund, 2010-2015.

-70-

Amount of blade vibration alarms per unit and year

SLG: Blade vibration alarms & TI one hour before, 2010-2015 35 30 25 20

15 10 5 0 0%

5%

10% 15% Wind speed (m/s)

20%

25%

Figure 70. Amount of blade vibration alarms per year and the turbulence intensity an hour before a blade vibration alarm for each unit in Lillgrund, 2010-2015.

8.1.2.2

Stor-Rotliden

Amount of alarms per unit and year

SRL: Alarms & WS 2012-2015 300 250 200 150 100 50 0 0

1

2

3

4 5 Wind speed (m/s)

6

7

Figure 71. Amount of alarms per year and wind speed for each unit in Stor-Rotliden, 2012-2015.

-71-

8

9

Amount of alarms per unit and year

SRL: Alarms & TI 2012-2015 300 250 200 150 100 50 0 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 72. Amount of alarms per year and turbulence intensity for each unit in Stor-Rotliden, 2012-2015.

SRL: Time loss & WS 2012-2015 6%

TIme loss (%)

5% 4%

3% 2% 1% 0% 0

1

2

3

4 5 Wind speed (m/s)

Figure 73. Time loss and wind speed for each unit in Stor-Rotliden, 2012-2015.

-72-

6

7

8

9

SRL: Time loss & TI 2012-2015 6%

TIme loss (%)

5% 4% 3% 2% 1% 0% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 74. Time loss and turbulence intensity for each unit in Stor-Rotliden, 2012-2015.

SRL: Energy availability & WS 2012-2015 Energy availability (%)

100% 99% 98%

97% 96% 95% 94% 0

1

2

3

4 5 Wind speed (m/s)

6

Figure 75. Energy availability and wind speed for each unit in Stor-Rotliden, 2012-2015.

-73-

7

8

9

SRL: Energy availability & TI 2012-2015 Energy availability (%)

100% 99% 98% 97% 96% 95% 94% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 76. Energy availability and turbulence intensity for each unit in Stor-Rotliden, 2012-2015.

SRL: Time availability & WS 2012-2015 Time availability (%)

100,00% 99,00% 98,00%

97,00% 96,00% 95,00% 94,00% 0

1

2

3

4 5 Wind speed (m/s)

6

Figure 77. Time availability and wind speed for each unit in Stor-Rotliden, 2012-2015.

-74-

7

8

9

SRL: Time availability & TI 2012-2015 Time availability (%)

100,00% 99,00% 98,00% 97,00% 96,00% 95,00% 94,00% 0%

5%

10% 15% Turbulence intensity (%)

20%

25%

Figure 78. Time availability and turbulence intensity for each unit in Stor-Rotliden, 2012-2015.

8.1.2.3

Horns Rev 1

Amount of alarms per unit and year

WH1: Alarms & WS 140 120 100 80 60 40 20 0 0

2

4

6 Wind speed (m/s)

Figure 79. Amount of alarms and wind speed for each unit in Horns Rev 1.

-75-

8

10

12

Amount of alarms per unit and year

WH1: Alarms & TI 140 120 100 80 60 40 20 0 0%

2%

4%

6% 8% Turbulence intensity (%)

10%

12%

14%

Figure 80. Amount of alarms and turbulence intensity for each unit in Horns Rev 1.

Time loss (%)

WH1: Time loss & WS 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 0

2

4

6 Wind speed (m/s)

Figure 81. Time loss and wind speed for each unit in Horns Rev 1.

-76-

8

10

12

Time loss (%)

WH1: Time loss & TI 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 0%

2%

4%

6% 8% Turbulence intensity (%)

10%

12%

14%

Figure 82. Time loss and turbulence intensity for each unit in Horns Rev 1.

Energy availability (%)

WH1: Energy availability & WS 100% 98% 96% 94% 92% 90% 88% 86% 84% 82% 80% 0

2

4

6 Wind speed (m/s)

Figure 83. Energy availability and wind speed for each unit in Horns Rev 1.

-77-

8

10

12

Energy availability (%)

WH1: Energy availability & TI 100% 98% 96% 94% 92% 90% 88% 86% 84% 82% 80% 0%

2%

4%

6% 8% Turbulence intensity (%)

10%

12%

14%

Figure 84. Energy availability and turbulence intensity for each unit in Horns Rev 1.

WH1: Time availability & WS Time availability (%)

100% 98% 96% 94%

92% 90% 88% 86% 84% 0

2

4

6 Wind speed (m/s)

Figure 85. Time availability and wind speed for each unit in Horns Rev 1.

-78-

8

10

12

WH1: Time availability & TI Time availability (%)

100% 98% 96% 94% 92% 90% 88% 86% 84% 0%

2%

4%

6% 8% Turbulence intensity (%)

10%

12%

14%

Figure 86. Time availability and turbulence intensity for each unit in Horns Rev 1.

8.2 Average wind speed and turbulence intensity in Lillgrund 2008-2015 Unit

WS average 20082015

TI average 20082015

SLGA01

6.94

18%

SLGA02

6.65

19%

SLGA03

6.56

19%

SLGA04

6.76

18%

SLGA05

6.84

18%

SLGA06

6.89

17%

SLGA07

7.58

15%

SLGB01

6.65

18%

SLGB02

6.52

20%

SLGB03

6.42

21%

SLGB04

6.67

20%

SLGB05

6.80

19%

SLGB06

6.91

19%

SLGB07

7.06

18%

SLGB08

7.73

15%

SLGC01

6.36

19%

SLGC02

6.26

21%

SLGC03

6.34

21%

-79-

SLGC04

6.57

20%

SLGC05

6.65

19%

SLGC06

6.75

19%

SLGC07

7.12

18%

SLGC08

7.38

17%

SLGD01

6.66

19%

SLGD02

6.25

20%

SLGD03

6.31

21%

SLGD04

6.61

19%

SLGD06

6.95

19%

SLGD07

7.18

18%

SLGD08

7.56

15%

SLGE01

6.65

18%

SLGE02

6.50

20%

SLGE03

6.40

20%

SLGE04

6.64

20%

SLGE06

7.10

19%

SLGE07

7.34

16%

SLGF02

6.54

19%

SLGF03

6.27

20%

SLGF04

6.87

19%

SLGF05

6.89

18%

SLGF06

7.13

16%

SLGG02

6.74

18%

SLGG03

6.54

19%

SLGG04

6.94

18%

SLGG05

7.34

17%

SLGH02

7.20

16%

SLGH03

7.10

17%

SLGH04

7.31

16%

-80-

8.3 Average wind speed and turbulence intensity in Lillgrund 2009-2015 Unit

Average WS Average TI 2009-2015 2009-2015

SLGA01

6.94

17.57%

SLGA02

6.65

18.69%

SLGA03

6.59

18.57%

SLGA04

6.73

17.63%

SLGA05

6.80

17.63%

SLGA06

6.89

16.62%

SLGA07

7.58

15.44%

SLGB01

6.64

18.25%

SLGB02

6.49

20.19%

SLGB03

6.39

20.71%

SLGB04

6.64

19.59%

SLGB05

6.75

19.37%

SLGB06

6.92

19.31%

SLGB07

7.05

18.41%

SLGB08

7.71

14.98%

SLGC01

6.37

18.91%

SLGC02

6.23

20.53%

SLGC03

6.31

21.03%

SLGC04

6.54

20.13%

SLGC05

6.62

19.20%

SLGC06

6.77

19.25%

SLGC07

7.10

18.16%

SLGC08

7.37

17.04%

SLGD01

6.65

18.71%

SLGD02

6.27

20.00%

SLGD03

6.28

20.62%

SLGD04

6.59

19.19%

SLGD06

6.96

18.82%

SLGD07

7.18

17.74%

SLGD08

7.57

14.66% -81-

SLGE01

6.66

18.19%

SLGE02

6.49

19.62%

SLGE03

6.40

20.04%

SLGE04

6.61

20.16%

SLGE06

7.06

18.51%

SLGE07

7.33

15.65%

SLGF02

6.51

19.06%

SLGF03

6.28

20.19%

SLGF04

6.84

19.24%

SLGF05

6.85

18.20%

SLGF06

7.07

15.96%

SLGG02

6.69

18.03%

SLGG03

6.49

19.03%

SLGG04

6.92

18.12%

SLGG05

7.29

16.72%

SLGH02 7.21

16.07%

SLGH03 7.07

16.83%

SLGH04 7.29

15.69%

8.4 Average wind speed and turbulence intensity in Lillgrund 2010-2015 Unit

Average WS Average TI 2010-2015 2010-2015

SLGA01

6.98

17.54%

SLGA02

6.67

18.55%

SLGA03

6.63

18.26%

SLGA04

6.72

17.55%

SLGA05

6.81

17.38%

SLGA06

6.94

16.18%

SLGA07

7.64

15.13%

SLGB01

6.67

17.96%

SLGB02

6.52

20.10%

SLGB03

6.42

20.60%

SLGB04

6.66

19.47%

-82-

SLGB05

6.76

19.21%

SLGB06

6.98

18.97%

SLGB07

7.13

18.27%

SLGB08

7.76

14.71%

SLGC01

6.41

18.70%

SLGC02

6.19

20.33%

SLGC03

6.34

21.00%

SLGC04

6.58

20.07%

SLGC05

6.63

18.83%

SLGC06

6.87

18.73%

SLGC07

7.15

17.96%

SLGC08

7.44

16.98%

SLGD01

6.67

18.63%

SLGD02

6.36

19.40%

SLGD03

6.30

20.43%

SLGD04

6.64

18.85%

SLGD06

7.05

18.39%

SLGD07

7.27

17.43%

SLGD08

7.65

14.29%

SLGE01

6.72

17.91%

SLGE02

6.54

19.28%

SLGE03

6.47

19.76%

SLGE04

6.66

20.30%

SLGE06

7.09

18.33%

SLGE07

7.39

15.40%

SLGF02

6.55

18.80%

SLGF03

6.35

19.87%

SLGF04

6.88

19.00%

SLGF05

6.89

18.04%

SLGF06

7.08

15.78%

SLGG02

6.71

17.84%

SLGG03

6.49

18.72%

SLGG04

6.99

17.79%

-83-

SLGG05

7.33

16.73%

SLGH02 7.29

15.76%

SLGH03 7.13

16.49%

SLGH04 7.37

15.39%

8.5 Excluded alarms Lillgrund Awaiting autoreset, Pitch lubrication Awaiting autoreset, Pitch lubrication, Fault status Awaiting autoreset, Pitch lubrication, Operating on generator Awaiting autoreset, Stopped, untwisting cables Awaiting autoreset, Stopped, untwisting cables, Operating on generator Cable untwisting procedure active, OK, Fault status Cable untwisting procedure active, Stopped, untwisting cables, Fault status Comm Failure Comm failure Comm failure: SLGA01-ErrorCode, SLGA01-Status, SLGA01-WTOperationState Comm failure: SLGA01-WTOperationState, SLGA01-ErrorCode, SLGA01-Status Comm failure: SLGA02-ErrorCode, SLGA02-Status, SLGA02-WTOperationState Comm failure: SLGA02-WTOperationState, SLGA02-ErrorCode, SLGA02-Status Comm failure: SLGA03-ErrorCode, SLGA03-Status, SLGA03-WTOperationState Comm failure: SLGA03-WTOperationState, SLGA03-ErrorCode, SLGA03-Status Comm failure: SLGA04-ErrorCode, SLGA04-Status, SLGA04-WTOperationState Comm failure: SLGA04-WTOperationState, SLGA04-ErrorCode, SLGA04-Status Comm failure: SLGA05-ErrorCode, SLGA05-Status, SLGA05-WTOperationState Comm failure: SLGA05-WTOperationState, SLGA05-ErrorCode, SLGA05-Status Comm failure: SLGA06-ErrorCode, SLGA06-Status, SLGA06-WTOperationState Comm failure: SLGA06-WTOperationState, SLGA06-ErrorCode, SLGA06-Status Comm failure: SLGA07-ErrorCode, SLGA07-Status, SLGA07-WTOperationState Comm failure: SLGA07-WTOperationState, SLGA07-ErrorCode, SLGA07-Status Comm failure: SLGB01-ErrorCode, SLGB01-Status, SLGB01-WTOperationState Comm failure: SLGB01-WTOperationState, SLGB01-ErrorCode, SLGB01-Status Comm failure: SLGB02-ErrorCode, SLGB02-Status, SLGB02-WTOperationState Comm failure: SLGB02-WTOperationState, SLGB02-ErrorCode, SLGB02-Status Comm failure: SLGB03-ErrorCode, SLGB03-Status, SLGB03-WTOperationState Comm failure: SLGB03-WTOperationState, SLGB03-ErrorCode, SLGB03-Status -84-

Comm failure: SLGB04-ErrorCode, SLGB04-Status, SLGB04-WTOperationState Comm failure: SLGB04-WTOperationState, SLGB04-ErrorCode, SLGB04-Status Comm failure: SLGB05-ErrorCode, SLGB05-Status, SLGB05-WTOperationState Comm failure: SLGB05-WTOperationState, SLGB05-ErrorCode, SLGB05-Status Comm failure: SLGB06-ErrorCode, SLGB06-Status, SLGB06-WTOperationState Comm failure: SLGB06-WTOperationState, SLGB06-ErrorCode, SLGB06-Status Comm failure: SLGB07-ErrorCode, SLGB07-Status, SLGB07-WTOperationState Comm failure: SLGB07-WTOperationState, SLGB07-ErrorCode, SLGB07-Status Comm failure: SLGB08-ErrorCode, SLGB08-Status, SLGB08-WTOperationState Comm failure: SLGB08-WTOperationState, SLGB08-ErrorCode, SLGB08-Status Comm failure: SLGC01-ErrorCode, SLGC01-Status, SLGC01-WTOperationState Comm failure: SLGC01-WTOperationState, SLGC01-ErrorCode, SLGC01-Status Comm failure: SLGC02-ErrorCode, SLGC02-Status, SLGC02-WTOperationState Comm failure: SLGC02-WTOperationState, SLGC02-ErrorCode, SLGC02-Status Comm failure: SLGC03-ErrorCode, SLGC03-Status, SLGC03-WTOperationState Comm failure: SLGC03-WTOperationState, SLGC03-ErrorCode, SLGC03-Status Comm failure: SLGC04-ErrorCode, SLGC04-Status, SLGC04-WTOperationState Comm failure: SLGC04-WTOperationState, SLGC04-ErrorCode, SLGC04-Status Comm failure: SLGC05-ErrorCode, SLGC05-Status, SLGC05-WTOperationState Comm failure: SLGC05-WTOperationState, SLGC05-ErrorCode, SLGC05-Status Comm failure: SLGC06-ErrorCode, SLGC06-Status, SLGC06-WTOperationState Comm failure: SLGC06-WTOperationState, SLGC06-ErrorCode, SLGC06-Status Comm failure: SLGC07-ErrorCode, SLGC07-Status, SLGC07-WTOperationState Comm failure: SLGC07-WTOperationState, SLGC07-ErrorCode, SLGC07-Status Comm failure: SLGC08-ErrorCode, SLGC08-Status, SLGC08-WTOperationState Comm failure: SLGC08-WTOperationState, SLGC08-ErrorCode, SLGC08-Status Comm failure: SLGD01-ErrorCode, SLGD01-Status, SLGD01-WTOperationState Comm failure: SLGD01-WTOperationState, SLGD01-ErrorCode, SLGD01-Status Comm failure: SLGD02-ErrorCode, SLGD02-Status, SLGD02-WTOperationState Comm failure: SLGD02-WTOperationState, SLGD02-ErrorCode, SLGD02-Status Comm failure: SLGD03-ErrorCode, SLGD03-Status, SLGD03-WTOperationState Comm failure: SLGD03-WTOperationState, SLGD03-ErrorCode, SLGD03-Status Comm failure: SLGD04-ErrorCode, SLGD04-Status, SLGD04-WTOperationState Comm failure: SLGD04-WTOperationState, SLGD04-ErrorCode, SLGD04-Status Comm failure: SLGD06-ErrorCode, SLGD06-Status, SLGD06-WTOperationState Comm failure: SLGD06-WTOperationState, SLGD06-ErrorCode, SLGD06-Status -85-

Comm failure: SLGD07-ErrorCode, SLGD07-Status, SLGD07-WTOperationState Comm failure: SLGD07-WTOperationState, SLGD07-ErrorCode, SLGD07-Status Comm failure: SLGD08-ErrorCode, SLGD08-Status, SLGD08-WTOperationState Comm failure: SLGD08-WTOperationState, SLGD08-ErrorCode, SLGD08-Status Comm failure: SLGE01-ErrorCode, SLGE01-Status, SLGE01-WTOperationState Comm failure: SLGE01-WTOperationState, SLGE01-ErrorCode, SLGE01-Status Comm failure: SLGE02-ErrorCode, SLGE02-Status, SLGE02-WTOperationState Comm failure: SLGE02-WTOperationState, SLGE02-ErrorCode, SLGE02-Status Comm failure: SLGE03-ErrorCode, SLGE03-Status, SLGE03-WTOperationState Comm failure: SLGE03-WTOperationState, SLGE03-ErrorCode, SLGE03-Status Comm failure: SLGE04-ErrorCode, SLGE04-Status, SLGE04-WTOperationState Comm failure: SLGE04-WTOperationState, SLGE04-ErrorCode, SLGE04-Status Comm failure: SLGE06-ErrorCode, SLGE06-Status, SLGE06-WTOperationState Comm failure: SLGE06-WTOperationState, SLGE06-ErrorCode, SLGE06-Status Comm failure: SLGE07-ErrorCode, SLGE07-Status, SLGE07-WTOperationState Comm failure: SLGE07-WTOperationState, SLGE07-ErrorCode, SLGE07-Status Comm failure: SLGF02-ErrorCode, SLGF02-Status, SLGF02-WTOperationState Comm failure: SLGF02-WTOperationState, SLGF02-ErrorCode, SLGF02-Status Comm failure: SLGF03-ErrorCode, SLGF03-Status, SLGF03-WTOperationState Comm failure: SLGF03-WTOperationState, SLGF03-ErrorCode, SLGF03-Status Comm failure: SLGF04-ErrorCode, SLGF04-Status, SLGF04-WTOperationState Comm failure: SLGF04-WTOperationState, SLGF04-ErrorCode, SLGF04-Status Comm failure: SLGF05-ErrorCode, SLGF05-Status, SLGF05-WTOperationState Comm failure: SLGF05-WTOperationState, SLGF05-ErrorCode, SLGF05-Status Comm failure: SLGF06-ErrorCode, SLGF06-Status, SLGF06-WTOperationState Comm failure: SLGF06-WTOperationState, SLGF06-ErrorCode, SLGF06-Status Comm failure: SLGG02-ErrorCode, SLGG02-Status, SLGG02-WTOperationState Comm failure: SLGG02-WTOperationState, SLGG02-ErrorCode, SLGG02-Status Comm failure: SLGG03-ErrorCode, SLGG03-Status, SLGG03-WTOperationState Comm failure: SLGG03-WTOperationState, SLGG03-ErrorCode, SLGG03-Status Comm failure: SLGG04-ErrorCode, SLGG04-Status, SLGG04-WTOperationState Comm failure: SLGG04-WTOperationState, SLGG04-ErrorCode, SLGG04-Status Comm failure: SLGG05-ErrorCode, SLGG05-Status, SLGG05-WTOperationState Comm failure: SLGG05-WTOperationState, SLGG05-ErrorCode, SLGG05-Status Comm failure: SLGH02-ErrorCode, SLGH02-Status, SLGH02-WTOperationState Comm failure: SLGH02-WTOperationState, SLGH02-ErrorCode, SLGH02-Status -86-

Comm failure: SLGH03-ErrorCode, SLGH03-Status, SLGH03-WTOperationState Comm failure: SLGH03-WTOperationState, SLGH03-ErrorCode, SLGH03-Status Comm failure: SLGH04-ErrorCode, SLGH04-Status, SLGH04-WTOperationState Comm failure: SLGH04-WTOperationState, SLGH04-ErrorCode, SLGH04-Status Customer / guest visit Dummy alarm - Converted from old system Lightning detektor activated Lightning detektor activated, Fault status, Stopped due to internal error. Manual reset is required Lightning detektor activated, Operating on generator, Stopped due to internal error. Manual reset is required Lightning detektor activated, Stopped due to internal error. Manual reset is required Manual idle stop Manual idle stop, Operating on generator, Stopped due to internal error. Manual reset is required Manual idle stop, Stopped due to internal error. Manual reset is required Manual stop Manual stop Manual stop, Fault status, Stopped due to internal error. Manual reset is required Manual stop, Operating on generator, Remote Stop Command is active Manual stop, Operating on generator, Stopped due to internal error. Manual reset is required Manual stop, Operating on generator, Stopped due to internal error. Manual reset is required Manual stop, Remote Stop Command is active Manual stop, Stopped due to internal error. Manual reset is required OK, Comm failure PI: Old Data! Remote stop - Owner Remote stop - Owner Remote stop - Owner, Operating on generator, Remote Stop Command is active Remote stop - Siemens Remote stop - Siemens Remote stop - Siemens, Operating on generator, Remote Stop Command is active Remote stop - Siemens, Remote Stop Command is active Remote Stop Command is active, Manual stop, Operating on generator Remote Stop Command is active, OK, Operating on generator Remote Stop Command is active, Remote stop - Owner Remote Stop Command is active, Remote stop - Owner, Operating on generator Remote Stop Command is active, Remote stop - Siemens Remote Stop Command is active, Stopped for SW update -87-

Remote Stop Command is active, Stopped for SW update, Operating on generator Remote stop Service, Operating on generator, Remote Stop Command is active Scheduled service work Software error Station is in Service Mode Stopped for SW update Stopped for SW update, Operating on generator, Remote Stop Command is active Stopped for SW update, Operating on generator, Stopped due to internal error. Manual reset is required Stopped for SW update, Remote Stop Command is active Turbine in local operation Turbine in local operation VPICalc: Stale PI Data!

-88-