• No results found

Investigating Wind Data and Configuration of Wind Turbines for a Turning Floating Platform

N/A
N/A
Protected

Academic year: 2021

Share "Investigating Wind Data and Configuration of Wind Turbines for a Turning Floating Platform"

Copied!
80
0
0

Loading.... (view fulltext now)

Full text

(1)

Investigating Wind Data and

Configuration of Wind Turbines for a

Turning Floating Platform

NURCAN SÖNMEZ

(2)
(3)

Acknowledgements

First of all I would like to thank Hexicon AB for giving this opportunity to me, especially I would like to thank to John Holm. Also, I would like to thank to Stefan Ivanell for accepting to be my supervisor and his continuous support to me all the way to the end. It was a very unique experience for me to work with Stefan Ivanell and Hexicon AB. Also, I really appreciate Soroor and Filipe for their help. They have been answering all my questions with patience, it was a very good experience to work with them as well. I would like to thank Stefan Söderberg for his time and consideration. It was a pleasure to meet with him.

Also I would like thank to my family, especially to Nazile, Ahmet, Sevim and Sevgiye for always believing in me and their support. Even though we were not in the same country, they always encouraged me for new challenges, I really appreciate it and I always feel very lucky to have a fantastic family.

(4)

Abstract

Wake interactions on a floating platform for offshore wind energy applications were in-vestigated. The study is performed in collaboration with Hexicon AB which has a patent family for innovative floating platforms, which are able to turn automatically. The Jensen model is used for wake effect calculations and the simulations were performed in MATLAB. The present study starts with wind speed and wind direction data analysis for the specific site that Hexicon AB plans to construct its first platform. Data analysis is followed by wake interaction studies for H4-24MW type Hexicon AB platform. Wake interaction simulations were performed for three different cases. Fixed turbine and platform, Nacelle yawing and fixed platform and Nacelle yawing and turned platform. Different cases were investigated in order to see wake interactions for different wind directions. Wind direction effect on wake interactions were performed between 90 and 90 with an increment of 10 . After having the simulation results for Nacelle yawing and turned platform case the results were compared with ANSYS - CFX simulations results. The results didn’t match exactly but they were very close, which is an indicator to the validity of the Jensen Model.

After finding out the possible behavior of wake interactions for different wind directions, power calculations were performed for the same three cases. In order to perform the power calculations the wake interactions for different wind directions were taken into account. In case of platform turning it was assumed that power losses were caused both by wake interactions and in case of thrusters activation. The losses that would be caused by different thrust forces on the turbine blades were not included.

The last study was performed to suggest different layouts. In the second case, Nacelle yawing and fixed platform, it was found out that nacelle yawing for most of the angles is not possible because it creates wake regions in front of the rotor area. It was decided to propose new turbine configurations on the platform which are tolerant to different nacelle yawing angles. The simulations were run without considering any constructions limitations, meaning that the availability of platform structure was not included.

The study is ended by performing some probabilistic results for platform turning behav-ior.

(5)

Contents

1 Background 1

1.1 The Aim of the Project . . . 1

2 Introduction 5 2.1 Offshore Wind Energy . . . 5

2.2 Floating Turbines . . . 6 2.3 Hexicon Technology . . . 9 2.4 Measurement Techniques . . . 12 2.4.1 LIDAR . . . 13 2.4.2 SODAR . . . 13 2.4.3 iSpin . . . 14

2.4.4 Measurement on the Hexicon Platform . . . 15

3 Wind Statistics 17 3.1 Selected Site . . . 18

3.2 Wind Speed Data Investigation . . . 18

3.2.1 Weibull Distribution . . . 18

3.2.2 Fitting Weibull Distribution to Wind Speed Data . . . 19

3.2.3 Wind Speed Data . . . 20

3.2.4 Seasonal Wind Speed Data . . . 21

3.3 Wind Direction Data Investigation . . . 22

3.3.1 Wind Direction Data . . . 23

3.3.2 Seasonal Wind Direction Data . . . 24

3.4 Results in Wind Statistics . . . 24

4 Hexicon Platform 27 5 Wake Effect Analysis and Power Calculations 29 5.1 Wake Effect . . . 29

5.1.1 The N. O. Jensen Model . . . 31

5.1.2 Wake Overlap . . . 32

5.2 The Jensen Model Applied to Hexicon Platform . . . 33

(6)

5.3.1 Wake Interaction in Fixed Nacelle and Platform . . . 35

5.3.2 Wake Interaction in Turbine Yawing . . . 40

5.3.3 The Jensen Model Comparison with CFD Simulations . . . 42

5.3.4 Wake Interaction in Turned Platform . . . 44

6 Power Calculations 45 6.1 Power Generation in Fixed Nacelle and Platform . . . 46

6.2 Power Generation in Nacelle Yawing . . . 47

6.3 Power Generation in Turned Platform . . . 48

7 Turbine Layout Simulations on Hexicon Platform 54

8 Results and Discussion 62

(7)

List of Figures

1 Offshore Support Structures . . . 6

2 Development of Offshore Support Structures . . . 7

3 Floating Offshore Concepts . . . 8

4 Floating Concepts and Installed Prototypes, from left to right; TLWT, WindFloat, TLB B, TLB X3, Hywind II, SWAY, Jacket, Monopile and the onshore. . . 9

5 Hexicon Offshore Platforms . . . 10

6 Stall and Pitch Control . . . 10

7 Yawing Examples . . . 12

8 Selected site close to Hanö, outside Karlshamn. South Baltic Sea . . . 18

9 Weibull Distribution Fitting . . . 20

10 Wind Speed Data at the Platform Location, 1981-1988 . . . 21

11 Seasonal Wind Speed Data . . . 22

12 Wind Direction Data at the Platform Location, 1981-1988 . . . 24

13 Seasonal Wind Direction Data . . . 25

14 Wind Energy Rose Map, 1981 . . . 26

15 Seasonal Wind Energy Rose and Corresponding Wind Direction Data . . . 27

16 Hexicon 4 x 6 MW Platform . . . 28

17 Flow Around an Idealized Turbine . . . 30

18 N. O. Jensen Wake Effect Model . . . 31

19 Comparison of the Jensen Model and Actuator Disk Theory for Wake Effect Ex-pansion on Hexicon Platform, k = 0.04 . . . 34

20 Wake Effect on Hexicon Platform k=0.075 . . . 35

21 Direction of the wind = 60 . . . 37

22 Direction of the wind = 50 . . . 37

23 Direction of the wind = 40 . . . 38

24 Direction of the wind = 30 , 20 , 10 . . . 38

25 Direction of the wind = 60 , 50 , 40 . . . 39

26 Direction of the wind and Nacelle = 2.5 , 3 , 3.5 - MATLAB . . . 41

27 Direction of the wind and Nacelle = 4 , 4.5 , 5 - MATLAB . . . 42

28 Direction of the wind and Nacelle = 3 , 4 , 5 - ANSYS CFX . . . 43

(8)

30 Single Turbine Power Output - Fixed Turbine, k = 0.04 and k = 0.075 . . . 47

31 Single Turbine Power Output - Nacelle Yawing, k = 0.04 and k = 0.075 . . . 47

32 Direction of the wind = 20 , 5 , 60 . . . 49

33 Platform Behavior . . . 50

34 Platform Angle Differences . . . 51

35 Total Power Output - Trust force control losses are not considered . . . 52

36 Total Power Output and Wind Speed Comparison . . . 53

37 Wake Expension for 5 Nacelle Yawing Tolerant Layout . . . 56

38 Wake Expension for 6 Nacelle Yawing Tolerant Layout . . . 57

39 Wake Expension for 7 Nacelle Yawing Tolerant Layout . . . 57

40 Wake Expansion for 8 Nacelle Yawing Tolerant Layouts . . . 58

41 Wake Expansion for 9 Nacelle Yawing Tolerant Layouts . . . 60

42 Wake Expansion for 10 Nacelle Yawing Tolerant Layouts . . . 60

(9)

List of Tables

1 Variable Weighting . . . 15

2 Measurement Techniques Weighting . . . 16

3 Variable Weighting Results . . . 16

4 Weibull Distribution Parameters . . . 19

5 Wind Speed Profile . . . 21

6 Seasonal Average Wind Speed . . . 21

7 Characteristics of Turbines and Hexicon Platform . . . 33

8 Cost of Turning the Platform . . . 51

9 Total Power Output Data . . . 52

(10)

1 Background

1.1

The Aim of the Project

Studies of wind phenomena have a long lasting history. In 1920, the German physicist Albert Betz proposed a theoretical maximum for the power output of a wind turbine; this is known as the Betz limit and his theory is still valid today [1]. Later in the century, in 1983, N.O. Jensen published his studies on wind generator interaction; this is commonly known as the Jensen wake effect model[2] and it is widely used in the industry as a first estimate to the wake effects on turbines.

Wind energy has come very far since Betz and Jensen studies, and taking into account the increased awareness of civilizations effect on earth, i.e. current global warming, it has become a far more studied topic; this has led to significant developments in the area, and attempts are being made at finding methods to more accurately model wind turbines, their interactions in farms, and on how to best model the location they are placed (Atmospheric Boundary Layer). Some of these developments include improvements of Jensen’s model, for example, the Larsen Model, Fradsen Model, as well as using the Navier-Stokes equations and to accurately predict the wake effect in the turbines. Nevertheless, the Jensen Model is still accepted, and it is one of the most widely used wake effect models today.

(11)

crucial to turn the platform at the right time.

The aim of the project is therefore to decide the behaviour of the platform based on the wind direction. This will be achieved by following the steps described below:

• Data Analysis

As already mentioned, the behaviour of wind is most of the time unpredictable, with regards to speed and direction. As such, the behaviour of each platform will be unique for a specific site. Hexicon plans to have a pilot plant close to the island Hanö, near Karl-shamn, Sweden; this site will be used as a case study for the rest of this work.

The wind speed and direction data for the specific site will be analysed to calculate expected power output, visualize the dominant wind direction for positioning, the direction change by time, and to display the expected power production for different directions, which is defined as energy rose.

• Wake Effect Study for Hexicon 4 x 6MW platform

As mentioned, Hexicon aim to develop state of the art floating foundations for offshore, and they have a number of concepts in the drawing board; this thesis will be based on their H4-24MW concept (See Section 4), and this is due to the fact that CFD studies on wake interaction and turbulence intensity have already been performed on it, and can thus be used as a base of comparison.

(12)

To study the effects of the wake to their full extent, three scenarios will be studied. In the first study it will be assumed that turbines and platform will be fixed, i.e. the wind direction will be changing (in the range between 90 and 90 (North to South)) while the turbines and platform will be maintained in the same position; this does not a scenario on the operational profile of the turbines as these have nacelle yawing. However, due to known problem of yaw misalignments, investigating this case would give an idea about the worst case scenario. In the second study the turbines will be allowed to yaw, while the platform is maintained stationary. In this case , meaning that turbines will be yawed from nacelle, and their interactions will be analysed; the platform will again be maintained at a fixed position. When the turbines will be yawed, wake will be expanded depending on the yawing angle, which was not considered in layout studies (See Section 4). However, since Hexicon concept is built on turning the platform , it is important to see the yawing limitations and decide the platform behaivor accordingly. At the end of this step it will possible to understand the angle at which the wakes from the first turbine column will start to interact with the ones in the second one. The last study will be performed to analyse the behaviour of the platform. In this case, the platform will be allowed to turn with the varying wind direction, thus having all turbines in the second column in the free wind. This turn will be automatic for changes in the wind direction up to ⌥90 , for the wind direction changes that are more than ⌥90 , thrusters will be activated in order to turn the platform. In order to turn the platform automatically pitch control will be used. In case of a wind direction change, turbine blades will be controlled by pitch angle, with this control less trust forces will appear in the blades so that different forces on the turbines will be balanced by automatic turn. Applying less thrust force to the turbine will cause losses in the power output. It is noteworthy that the losses that will be caused by pitch control won’t be included to this study.

(13)

dependent losses won’t be included to this study (See Section 2.3). Overall, it is believed that showing three cases would be a good base to comment on Hexicon platform concept. • Configuration of Wind Turbines

(14)

2 Introduction

2.1 Offshore Wind Energy

As mentioned earlier, theoretical studies on wind energy have been carried since the start of the last century. Despite of this, it is only in the past decade that wind energy has started to become a reality, and that it has gained a relevant portion on the energy market. The slow rate of adoption is mainly attributed to economic reasons, which this is largely due to the fact that power plants run on fossil fuels are still cheaper and easier to install. Nevertheless, this is slowly changing as governments around the globe are becoming more environmentally conscious, and/or need to meet international targets. One approach that has proved successful is the implementa-tion of feed-in-tariffs, which essentially consists of government guarantee of electricity purchase; this ensures that the electricity produced by the renewable energy source is bought despite of a possibly higher price, thus ensuring financial attractiveness for the investors, providing payback periods between 7 - 10 years. These incentives/targets boosted the technology developments within wind energy, which consequently lowered the overall costs of project developments [3]. For instance, these developments drove for the offshore, where the wind conditions are better, i.e. higher mean speed, and lower turbulence [4].

Comparing the installed capacities of onshore and offshore wind energy, 110.7GW versus 6.6GW respectively, it is clear that offshore wind energy is still in its early stages of development [5]. However, considering that the first offshore wind farm was installed in 1991, Vindeby Denmark, and after 22 years, in 2013 the capacity in 11 countries in Europe reached to 6, 562MW [6]. It is clear that it is progressing successfully; a report released in April 2013 by South Baltic Program and South Baltic OFF.E.R suggests that a total of 40GW installed capacity in offshore wind by

2020and 150GW in 2030 in the region [7, 6].

(15)

Figure 1: Offshore Support Structures [9]

Tripod and jacket foundations are applied in intermediate water depths. The main difference between concepts applied to shallow waters and transitional waters is the way the that they transfer the loads to the seabed.

Currently the share of commonly used support structures are as follows [6] ; • 76 % Monopolies • 12 % Gravity based • 5 % Jackets • 5 % Tripods • 2 % Tripiles • 0.08 % Floating • 0.08 % Experimental

The development of the technologies can be seen in Figure 2. According to EWEA, in 2013 the average water depth was 20 meters and the average distance to the shore was 30 km. There are also 2 prototypes that were installed with floating concept, which is in the developing stage. Hex-icon AB also developing floating turbine concept. Detailed information about Floating Turbines can be found in Section 2.2.

2.2 Floating Turbines

(16)

Figure 2: Development of Offshore Support Structures [10]

however there are few locations in the world where the wind conditions are favourable and where the depths are sufficiently shallow to allow for conventional bottom mounted foundations, as such, the development towards floating foundations will open up the whole offshore environment without being constricted to water depth; this is essentially the same approach the oil industry took when trying to find new sources of oil.

In addition to the improved wind conditions and theoretical limitless depth, floating founda-tions also have the benefit of eliminating the ’Not In My Back Yard’ (NIMBY) syndrome. With respect to the wind industry, NIMBY refers to the behavioural syndrome in which the population approves of the idea and the development of wind installations as long as it is not close to them i.e. their back yard. By introducing floating foundations, and therefore not being constrained to water depth, it is possible to place the turbines further from shore, which eliminates the NIMBY aspect of offshore wind. Also by placing the turbines further from shore (over 20km), the wind turbine manufacturers are given a larger degree of freedom in turbine design. For instance, there is less of a need to take noise into consideration, and as such, other tip speeds can be used that provide the turbine with a better performance [11].

(17)

• Spar Type Floating Support Structure Spar type platform concept uses a deeper draft in order to shift the center of gravity to below levels. However, having a large draft has negative economic implications [12].

• Tension Type Floating Support Structure Tension-leg concepts are supported by the taut mooring lines; this concept is resistant to the pitching motions caused by the waves. However, the high loads and fatigue on the mooring line should be taken in the consideration on the design of such concepts.

• Buoyancy Type Floating Support Structure Semi-submersible concepts have a small water plane area, which reduces wave-structure interactions, and the buoyancy created by a large displaced volume to both support and stabilise the structure.

Figure 3: Floating Offshore Concepts [13]

(18)

Figure 4: Floating Concepts and Installed Prototypes, from left to right; TLWT, WindFloat, TLB B, TLB X3, Hywind II, SWAY, Jacket, Monopile and the onshore. [11]

TLWT, WindFloat, TLB B, TLB X3, Hywind II, SWAY, Jacket, Monopile floating turbines were investigated. Different parameters were taken into account in the study, for instance, farm size, offshore distance and water depth. In the end of the study it was concluded that floating turbines may be produced at equal or lower levelised cost of energy than bottom fixed turbines[11].

2.3 Hexicon Technology

As mentioned earlier, Hexicon AB develops multi-turbine floaters whose patented key feature is its ability to turn towards the wind to ensure best performance with a given layout. Hexicon AB has four different concepts, i.e. turbine layouts, each with a different number of turbines (ranging from 3 to 6) and rated capacities ranging from 9MW to 36MW. Figure 5 presents the four different layouts.

Modern wind turbines are very complex machines and run according to a very specific opera-tional profile. The operaopera-tional profile is designed to both maximise energy capture, and minimise loads to extend turbine lifetime; this is established with control systems [14]. The control system dictates, amongst other parameters, the yaw of the turbine and the pitch of the blades. The yawing mechanism ensures that the turbine is always facing the wind, while the pitch of the blades controls aerodynamic performance, i.e. how much energy is ’absorbed’ from the wind; it is due to pitch control that the power curve of a modern wind turbines differs so much from older ones.

(19)

(a) H3-18MW (b) H4=24MW

(c) H4-24MW-D (d) H6-36MW

Figure 5: Hexicon Offshore Platforms

(20)

around 10m/s 12m/swhere it achieves its rated capacity, i.e. maximum power output. The rated capacity is then maintained by pitching the blades for higher wind speeds [16].

To maximize power output, as in the pitch control, another control mechanism applied to wind turbines, called as yaw control (can be also called as nacelle control or nacelle yawing). The idea behind nacelle yawing is to follow the upcoming wind direction, in other words, when the direction of the upcoming wind is changing, the nacelle yaws to face the wind. By facing the wind continuously, more energy is captured [17].

Nacelle yawing helps to optimize the power output by following the wind direction. However, if more than one turbine is taken into consideration, e.g. a wind farm, at certain wind directions there will be wake interactions between the turbines. Figure 7(a) shows two turbines with com-pletely undisturbed wind. However, for a radical change in wind direction, e.g. 90 represented in Figure 7(b), the same layout would have one turbine in the wake of the other. This is not desired as it affect the power production of the one downwind and can cause excessive fatigue loads, reducing the lifetime of the turbine; this is one of the reasons that turbine manufacturers have a recommended separation distance of at least 3D. Due to the close proximity of the tur-bines, wake interactions in Hexicon’s platforms will occur at distances much closer than what is recommended by the turbine manufacturers if the control system fails; this makes the platforms’ turning system such an important feature, and the understanding of when to turn such an im-portant parameter in the already complex control system.

As already mentioned, the platforms will turn against the wind to maximise power output for any given layout. The turning (yawing) is achieved by the use of thrusters, and the turbines’ own pitch, i.e. thrustforce; it has been hypothesised that by pitching the blades of each individual turbine to different extents, a thrust force difference would be created that could be sufficient to align the platform to the predominant wind direction. However, as the power losses correspond-ing to the different pitches are unknown, these will be excluded from this study. Regardcorrespond-ing the thrusters, it has been proposed that two will be used each rated at 4MW each, and can turn the platform 360 in 40 minutes; this will be considered in the power calculations.

(21)

(a) Yawing might not cause wake (b) Yawing might causes wake

Figure 7: Yawing Examples

the downwind turbines. The results for this will be presented and discussed in Section 5.3.

2.4 Measurement Techniques

This study aims to describe different types of measurement techniques and propose one of them for Hexicon platform. Wind measurement has a significant importance in wind farm projects. First to find out yearly wind speed profile so that the expected annual wind production can be calculated. As it is known, site specific wind measuring is one of the first things that is done in a wind project. At least 1 year wind measuring is required before continuing the wind projects. At this stage of the project, having precise measurements, would directly effect the expected annual wind production calculations. Second, wind measurement are used when the turbines start to operate as well. As mentioned before, wind turbines have control systems and the input for those control systems comes from the wind measurements. When a turbine starts to operate, by knowing the incoming wind speed the loads can be controlled or by knowing the wind direction the nacelle can be turned. In other words, having accurate measurements enable real life optimization for turbine control systems.

(22)

introducing the technologies, a weighting study is going to be performed in order to suggest the one that has the highest weighting of all.

2.4.1 LIDAR

Light Detection And Ranging, usually referred to as, LIDAR, is a measurement technique which uses the light energy backscattered from microscopic particulates transported by the wind. It has been used in a wide range of atmospheric studies, including wind measuring. LIDAR systems are able to sample the data every second.

The best advantage of LIDAR is its accuracy. It has been used for the industry for some years and its accuracy is proven however, the equipment is very expensive. Even though it is said by the LIDAR manufacturers that it is feasible to use LIDAR measuring technique, since it pays off with the accurate data that it feeds to control system, it is not preferable for its high cost. Since it requires high investment, some companies offer to rent them. For example, EMD and AWE. Renting the equipment can be a solution at the wind energy assessment step of a project, however, if it is required to have accurate measurements in the wind farm constantly, then renting a LIDAR wouldn’t be a solution. LIDAR is not only very accurate but also has a high mobility i.e. it can be moved easily in the project area.

The location of the measuring equipment is very vital. LIDAR, can be mounted either to the turbine and to the site. When it is fitted onto the turbine, it has two possible locations, nacelle or hub. However, mounting it to the nacelle is suggested for the simplicity [18]. De-tailed information about nacelle and hub mounted LIDAR comparison can be found in reference [18, 19].

2.4.2 SODAR

(23)

The most important advantages of SODAR is its price when it is compared with LIDAR. It is shown that SODAR has as accurate results as LIDAR, however, it is also shown that SO-DAR might have a poor performance when the atmosphere is thermally well mixed. Another performance problem might occur in a situation, when the humidity layer that the SODAR is measuring decreases to less than 40%. This phenomena results increased atmospheric absorption of the radiated and backscattered acoustic energy [20]. However, its price still makes it a desired technique. As it was said before, SODAR is accepted as a cheaper solution to LIDAR. However, it is only able to measure up to a height of 150m, which is lower than LIDAR. However, this shouldn’t be accepted as a disadvantage since the turbine hub heights hasn’t reached to 150m. It has been used for long periods in the industry, which makes it a mature product.

When it comes to it’s monitoring system, not a lot of research is found, however, product developers is confident about it’s performance in the monitoring, as well as measuring. As it has been mentioned before SODAR can be shown as the most mature product when it is compared with LIDAR and iSpin technologies[21].

2.4.3 iSpin

The latest innovation in wind measurement is the iSpin. It was developed by 2 professors at the Technical University of Denmark (DTU) and patented at 2004. The innovation that has been performed with iSpin is noteworthy.

(24)

2.4.4 Measurement on the Hexicon Platform

As mentioned in the beginning of the section, measurement techniques are used for a range of purposes, however, one has a significant importance for Hexicon, which is the control of the turbine and platform by using the input form measurement technique. This study focused on 3 different measurement techniques, LIDAR, SODAR and iSpin. It is suggested to use iSpin technology on the Hexicon platform for the following reasons.

In order to decide which solution would be the most suitable for a Hexicon platform a study is performed by weighting some defined variables. It was decided to compare Price, Results Accuracy, Mobility, Maturity and Technical Complexity. The importance of the variables were decided according to their importance for Hexicon platform concept. As expected Price is a very important variable, not only for the Hexicon as concept, but also for every project. Result Accuracy shares the same weighting as Price, as accurate measuring is a core variable that is required for Hexicon platform concept, as the platform is able to turn according to wind direction

change. Technical Complexity takes place as the 3rd important variable. Technical complexity

of a measuring system is important for floating concept, since the sites are not easily reachable and it is required to have the system which doesn’t require frequent maintenance. Maturity

takes the 4th place, since it gives a good insight for the proven technology. The last variable

is chosen as Mobility. In most of the wind projects, having a measurement technique that has high mobility is an advantage. However, in Hexicon platform concept, it is planned to have a measurement device in each platform, mobility has the lowest weighting. In Table 1, variables and their weightings are defined.

Table 1: Variable Weighting

Price Result Accuracy Technical Complexity Maturity Mobility

0.3 0.3 0.2 0.15 0.05

(25)

Table 2: Measurement Techniques Weighting

Price Result Accuracy Technical Complexity Maturity Mobility

LIDAR 0.2 0.33 0.3 0.3 0.25

SODAR 0.3 0.33 0.3 0.5 0.25

iSpin 0.5 0.33 0.4 0.2 0.5

After performing weighting calculation the results showed that iSpin is the most advantageous measurement technique for Hexicon platform. This is mostly caused becaused of it’s price. iSpin is the cheapest solution of all. Even though it hasn’t reached technological maturity, having a low price makes the iSpin one of the most important measurement technique.

Table 3: Variable Weighting Results

LIDAR 0.274

SODAR 0.336

(26)

3 Wind Statistics

Wind data assessment is one of the key points for successful wind energy projects. Data assess-ment gives the possibility to know the wind speed and wind direction profile of the site, which can then be used to determine the wind turbine and to estimate the power output, which could be used for financial calculations, for instance pay-back period, and dominant wind direction for turbine positioning. Data assessment is essential for this project as well; it is crucial to know the wind direction data for Hexicon, since the platform behavior will be decided according to wind direction data.

Wind data is provided by WeatherTech Scandinavia AB, a consultancy company focused on wind data analysis. Considering that wind data collection is a long and expensive process, there are models that are developed in order to predict the data for a given location. WeatherTech Scandinavia AB uses one of those model to predict wind speed and wind direction. To run the model, latitude and longitude of the specific location are needed. Latitude and longitude for this project is for the location of the first platform that Hexicon is going to install.

The most advantageous part of using modeled data, instead of using data from mat masts that are supplied by different organizations, can be said as the height of the measurement. The height of the mat masts are usually in between 10 50m , so usually measurements are taken in this range, however, it is known that the wind turbines that are in use today, have a hub height of 80m or more, especially in the offshore conditions hub height is higher. In the modeled data, the measurement height is 120m and the hub height of turbines which will be used for H4-24 MW platform is 105m. Having measurements close to hub height of the turbine will increase the reliability of the results. An important remark about the data is the period that it reflects. The modeled data is in years between 1981-1988, which might be considered as an old data, however, it is decided that using data from higher heights has a better reflection of the site than having the latest data.

(27)

tho the turbine. In this case a wind rose map is drawn and the direction is chosen accordingly. Moreover to this conventional method, wind direction data will be used by Hexicon incessantly, since the platform is not fixed. This requires a detailed study for wind direction, that will be performed in the Section 3.3.

3.1 Selected Site

Selected site plays one of the most important roles in a wind farm project. All the design, for instance farm design, wind turbine selection, speed data, direction data are done based on the project site. As it was mentioned before, the specific data that will be used in this thesis is decided according to the location of Hexicon’s first platform.

Figure 8: Selected site close to Hanö, outside Karlshamn. South Baltic Sea

3.2 Wind Speed Data Investigation

3.2.1 Weibull Distribution

(28)

Weibull Distribution is the most commonly used distribution for wind speed data. It is a con-tinuous distribution that has two parameters.

8 < :

f (v; ↵, ) = ↵⇣v⌘↵ 1e (v)↵, v > 0, ↵, > 0

0, v 0 (3.1)

The two parameters, ↵ as the shape parameter and as the scale parameter, are needed to be

estimated in order to analyze the wind speed behavior. The methods that are used to estimate Weibull distribution vary. The two most commonly used estimation methods are Maximum Like-lihood Method (MLM) and the Least Square Method. However, different studies have already proved that MLM is the most efficient estimation method for weibull distribution parameters [23, 24].

In this study, Weibull Distribution parameters will be estimated by using MATLAB Curve Fit-ting application which also uses Maximum Likelihood Method in the estimation. Details about shape and scale parameters estimation for the wind speed data can be found in Section 3.2.2. 3.2.2 Fitting Weibull Distribution to Wind Speed Data

In this part Weibull distribution is fitted to the wind speed data, which can be seen in Figure 9. Data fitting is a common procedure to run simulations. The idea behind distribution fitting is to fit the data to a distribution and generate random numbers according to that distribution, which supposed to be used in simulations and give an overview of the expected overcome. In particularly, for this project the necessity to generate random numbers occur with the need of platform behavior decision. Moreover, fitted data will have scale and shape parameters which are the parameters for Weibull Distribution. Those parameters are some of the inputs that are required by the software that Hexicon uses (ANSYS - CFX) to calculate power output. Given data can be used to generate random wind speeds and those speeds can be used to predict power generation for next year, for example. Weibull Distribution parameters can be found in Table 4.

Table 4: Weibull Distribution Parameters

Scale Parameter (c) 9.99 m/s

Shape Parameter (k) 1.96

(29)

This procedure is done only for wind speed data, for wind direction data another distribution should be fitted. Fitting two variable to two different distributions might create a problem, having in the mind that they might be related. In other words, random numbers will be generated for wind direction and wind speed, after generation, those variables are going to be combined, to predict the platform behavior. Since they are not generated together, combining them would not be a valid idea. Because of this reason, it is decided to use real data so that there wouldn’t be any doubts about the method that is going to be used.

0 5 10 15 20 25 30 0 0.02 0.04 0.06 0.08 0.1 0.12 Data Density U0 data Weibull Fit

Figure 9: Weibull Distribution Fitting

3.2.3 Wind Speed Data

As it was explained in the previous section, real data will be used for this study and it is crucial to see wind speed profile for specified location. In order to have an idea about the profile, a graph is drawn, to see the frequencies of given wind speeds. As it can be seen in Figure 10 6

m/s- 10 m/s range has the highest frequencies in the data, as it is expected the average velocity

(30)

Table 5: Wind Speed Profile

Avarage Wind Speed 8.8392 m/s

Max. Wind Speed 33.2000 m/s

Min. Wind Speed 0.3000 m/s

0 5 10 15 20 25 30 35 0% 1% 2% 3% 4% 5% 6% 7% 8% 9%

10% Hourly Wind Speed Data for Hexicon Prototype Location

Wind Speed [m/s]

Frequency

Figure 10: Wind Speed Data at the Platform Location, 1981-1988 3.2.4 Seasonal Wind Speed Data

Looking at the seasonal data has a significant importance in order to comment to data in a more accurate way, even though the data might give information for the yearly energy production, it is known that this energy production is going to fluctuate through out the year and by looking at the seasonal data, a more accurate picture about seasonal energy production can be seen. Season 1 is accepted as January to March, Season 2 is accepted as April to June, Season 3 is accepted from July to September and finally Season 4 is accepted as October to December. The average wind speed for each season can be seen in Table 6.

Table 6: Seasonal Average Wind Speed

Season Average Wind Speed

Season 1 10.4567 m/s

Season 2 7.8191 m/s

Season 3 7.0651 m/s

(31)

The seasonal data shows that, Season 1 and 4, have higher average wind speeds than all data’s average wind speed. Hence, the power output for these months should be expected higher than Season 2 and 3. According to the location of the platform, an advantage about the data should be pointed out. In other words, according to its cold climate, more energy is consumed in the winter in Sweden than the summer. Seasonal graphs show that, energy production will be higher in winter than summer which can be an added value for Hexicon platform in a sense that Hexicon platform can be used to meet the demand in winter.

0 5 10 15 20 25 30 35 0% 2% 4% 6% 8% 10%

12% Hourly Wind Speed Data for Hexicon Prototype Location (JAN − MARCH)

Wind Speed [m/s]

Frequency

(a) Average Wind Speed - Season 1

0 5 10 15 20 25 30 35 0% 2% 4% 6% 8% 10%

12% Hourly Wind Speed Data for Hexicon Prototype Location (APR − JUNE)

Wind Speed [m/s]

Frequency

(b) Average Wind Speed - Season 2

0 5 10 15 20 25 30 35 0% 2% 4% 6% 8% 10%

12% Hourly Wind Speed Data for Hexicon Prototype Location (JULY − SEP)

Wind Speed [m/s]

Frequency

(c) Average Wind Speed - Season 3

0 5 10 15 20 25 30 35 0% 2% 4% 6% 8% 10%

12% Hourly Wind Speed Data for Hexicon Prototype Location (OCT − DEC)

Wind Speed [m/s]

Frequency

(d) Average Wind Speed - Season 4

Figure 11: Seasonal Wind Speed Data

3.3 Wind Direction Data Investigation

(32)

inves-tigated for Hexicon. Hexicon’s platforms turn based on the changes in the wind direction in order to maximize the power generation from the platform, so that having wind direction information is precious.

Before going through details, an important point should be made for the coordinate system that is going to be used. A regular coordinate system accepts 0 as East and the degrees increase counterclockwise. The following corresponding angles are 90 North, 180 West and 270 South. The coordinate system that is followed by the wind industry differs from the regular coordinate system. In order to analyze the wind direction, wind rose maps are used. The wind wind rose map will be used to analyze the direction for this study as well. The corresponding angles for the wind rose map are as follows; 0 as North, 90 as East, 180 as South and 270 as West. Another use of direction data is for placing the platform. By looking at the wind rose, dominant wind direction can be seen and sitting can be done accordingly. Please find detailed information for wind direction profile is presented in the following section.

3.3.1 Wind Direction Data

Wind direction data was plotted to wind rose map. In the map, dominant directions and the wind speeds for related directions are shown. As it was shown in Figure 10 and Figure 11, the most frequent wind speeds are in between 5 10m/s and 10 15m/s range. At the same time

wind speeds between 15 20m/sshould be taken into account. An important part is, higher

(33)

be definitely added to power output. Simulations will show the behavior of the platform for different angles and their corresponding wind speeds in Section 5.3.

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 30 − 35 m/s

Figure 12: Wind Direction Data at the Platform Location, 1981-1988

3.3.2 Seasonal Wind Direction Data

In Section 3.2.4, the importance of seasonal data was mentioned. Seasonal data approach is applied to wind direction data too. Seasonal data shows that it is a possibility to capture more energy by turning the platform. For instance, as it can be seen in Figure 13(b) and Figure 13(c), which belong to Season 2 and 3, significant amount of energy can be produced from non-dominant direction. It can be commented as the dominant direction doesn’t vary a lot when seasonal direction data is compared to all direction data. However, wind speed frequencies for non-dominant direction is considerable, which should definitely be added to energy production. Detailed information about the Seasonal energy production can be found in Section 3.4.

3.4 Results in Wind Statistics

The wind speed and direction data was investigated in detail in previous section. As a result of wind speed and direction data investigation, it is important to see the combination of those two data. A wind energy rose, which means specifying the power output for corresponding direction, gives a better understanding for wind speed and direction data combination.

(34)

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 30 − 35 m/s

(a) Average Wind Direction-Season 1

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 m/s

(b) Average Wind Direction-Season 2

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 m/s

(c) Average Wind Direction-Season 3

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 30 − 35 m/s

(d) Average Wind Direction-Season 4

Figure 13: Seasonal Wind Direction Data

is applied for wind energy rose as well and real data for year 1981 is used. Results can be in seen in Figure 14.As the yearly data is used, Wind Energy Rose map have a similar shape as all wind direction data (Figure 12). The dominant wind direction is quite the same, even though this data is run only for year 1981.It obviously shows that every direction has contribution to yearly production, in particularly, even the fraction of occurrence is not that high, the opposite of dominant direction has capacity to produce energy between 20 - 25 MWh range. It is important to mention that wind speeds between 2 25m/s are included to wind energy rose since 2m/s is the cut-in speed and 25m/s is the cut-out speed for the turbine. Power output for the speeds that are not in this range, calculated as 0.

It is also important to see the seasonal distribution of energy. Again this study is done only for year 1981, as for the previous example.

(35)

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 MWh

Figure 14: Wind Energy Rose Map, 1981

(36)

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 MWh

(a) Wind Energy Rose-Season 1

2% 4% 6% 8% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 25 − 30 30 − 35 m/s (b) Wind Direction-Season 1 5% 10% 15% WEST EAST SOUTH NORTH 0 − 5 5 − 10 10 − 15 15 − 20 20 − 25 MWh

(c) Wind Energy Rose-Season 2

5% 10% 15% WEST EAST SOUTH NORTH 0 − 2 2 − 4 4 − 6 6 − 8 8 − 10 10 − 12 12 − 14 14 − 16 16 − 18 18 − 20 20 − 22 22 − 24 m/s (d) Wind Direction-Season 2 5% 10% 15% WEST EAST SOUTH NORTH 0 − 2 2 − 4 4 − 6 6 − 8 8 − 10 10 − 12 12 − 14 14 − 16 16 − 18 18 − 20 20 − 22 22 − 24 MWh

(e) Wind Energy Rose-Season 3

5% 10% 15% WEST EAST SOUTH NORTH 0 − 2 2 − 4 4 − 6 6 − 8 8 − 10 10 − 12 12 − 14 14 − 16 16 − 18 18 − 20 m/s (f) Wind Direction-Season 3 2% 4% 6% WEST EAST SOUTH NORTH 0 − 2 2 − 4 4 − 6 6 − 8 8 − 10 10 − 12 12 − 14 14 − 16 16 − 18 18 − 20 20 − 22 22 − 24 MWh

(g) Wind Energy Rose-Season 4

2% 4% 6% WEST EAST SOUTH NORTH 0 − 2 2 − 4 4 − 6 6 − 8 8 − 10 10 − 12 12 − 14 14 − 16 16 − 18 18 − 20 20 − 22 22 − 24 m/s (h) Wind Direction-Season 4

Figure 15: Seasonal Wind Energy Rose and Corresponding Wind Direction Data

4 Hexicon Platform

(37)

Figure 16: Hexicon 4 x 6 MW Platform

The turbines are sited to the following coordinates; the position of the 1st turbine is (0, 78),

the position of the 2nd turbine is (0, 78), the position of the 3rd turbine is (215, 253) and the

position of the 4th turbine is (215, 253). The proposed layout was based on the following

as-sumptions; it was decided to locate four turbines into two columns, the constant that decides the wake expansion, which will be explained in detail in Section 5.1.1, is accepted as 0.05. Based on wake expansion the distance between first and second column is decided, second column turbines were placed along the decided x axes, the specific location of the second column turbines were calculated based on the wake expansion constant 0.05 and 10% of the turbine diameter value were added between wake diameter and the second column turbines.

(38)

5 Wake Effect Analysis and Power Calculations

5.1 Wake Effect

Turbulence intensity and wake effect can be shown as the main factors of a wind farm that want to be eliminated. It is known that the turbulence intensity is created by the wind phenomenon and might increase in a wind farm, related to turbine locations. On the other hand, the reason that wake effect occurs can be explained by the energy conservation law. It is known that wind turbines convert the kinetic energy (wind speed) into mechanical energy. According to Equa-tion 5.1, after some amount of kinetic energy is converted to mechanical energy, the downstream wind speed is going to be decreased [26].

˙ m ✓U2 2 2 U2 1 2 ◆ = W˙ (5.1)

In addition to this, according to Bernoulli, mechanical energy per unit mass along a streamline is conserved which is described in Equation 5.2.

p

⇢+

V2

2 + gz (5.2)

As it can be seen from the equation, the theorem gives a relationship between pressure, velocity and gravity. When this equation is applied to wind turbines, potential energy from gravity can be neglected as elevation is not big for wind turbines. When gz term is neglected than the equation becomes a relationship between pressure and velocity, which should be balanced along a streamline [26]. It was just mentioned that wind speed is diminished downstream. Since there has to be balance between velocity and pressure and a diminished wind speed causes an increase in the downstream pressure. Figure 17 depicts the relationship between pressure and velocity in a turbine.

(39)

Figure 17: Flow Around an Idealized Turbine [27]

turbine) [28]. It is very important to locate the turbines on a region that is not under the wake region because it is known that wake effect could cause 20% to 45% of power coefficient loss in the wind farm [29].

Having this in mind, careful calculations should be performed. In order to analyze the wake effect there are different models available. The N.O Jensen model, Larsen model, Fradsen mod-els are the examples that are used to decide the wake effect level by calculating the wind speed deficit [30]. When the wind speed deficit for a certain distances is calculated, the distance be-tween turbines can be defined, generally it is suggested as 5 rotor diameter for onshore and 7 rotor diameter for offshore. The N.O Jensen model is the simplest and most widely accepted model. In this study wake effect calculations will be done by using the Jensen model. Details about the Jensen model can be found in Section 5.1.1.

(40)

5.1.1 The N. O. Jensen Model

The Jensen Model will be used in order to find the wake effect of the turbines. As it has been mentioned before, Hexicon AB has its own design for the platform layout and this thesis project is going to evaluate the design by performing Jensen wake model.

Figure 18: N. O. Jensen Wake Effect Model [32]

The Jensen model is a linear model and calculates the wake effect by formulating the wake diameter and the wind speed deficit. One of the most important parameters in the Jensen Model is Wake Decay Constant, k, can be seen in Figure 18 (depicted as ↵) and it depends on the turbine hub height and terrain roughness. (Equation 5.3)

k = 0.5

ln⇣zz0⌘ (5.3)

In onshore and offshore conditions the value of the k changes accordingly to the terrain rough-ness. The suggested offshore k value is 0.04 and this value is going to be used in the calculations [30]. At the same time, it is important to show how the wake behaves in worse cases as well. This study is going to demonstrate the wake expansion when k is 0.075, the suggested onshore

kvalue and it can be found in Section 5.2.

The wake diameter is described as;

Dw= D (1 + 2ks) (5.4)

(41)

defined as s = x/D . Since Hexicon design is going to be used, value will be used as the platform designed. Equation (5.5), define the wind speed deficit after the wind hits the turbine.

The importance of turbine thrust coefficient, ct, is noteworthy in wake effect.The reason that

wake occurs can be explained simply as follows; as the free wind approaches a turbine rotor, it

hits the blades, some amount of momentum is extracted by the blades, which is related to ctand

the rest of the wind continues, which has lower wind speed than the upcoming wind because,

wind turbine converted some amount of energy into mechanical energy. The reason behind ct’s

existence in Equation (5.5), which shows the velocity deficit behind a turbine, also shows ct

contribution to velocity deficit.

u = U1·  1 1 p 1 cT (1 + 2ks)2 (5.5) 5.1.2 Wake Overlap

Before analysing the outcomes of the study, wake overlap should be explained. In the Jensen Model [2], velocity deficit is formulated for the case in which the turbine is under full wake. However, in some wind farms, the layout is designed in a way that turbines might be affected from partial wakes. This point has a significant importance for Hexicon platform as well, since turbines are very close to each other and platform will be turning constantly. There is a high partial wake possibility for the turbines that are in the second column, which can be defined as wake overlap. There are different studies and suggested methods to calculate wake overlap. For example, Djerf proposed 4 overlap methods[33].

• Sum of Squares of velocity deficits • Energy Balance

• Geometric Sum • Linear Superposition

(42)

they propose another method that is more related to trigonometric relationships between inter-acting turbines [32]. In this study wake overlaps will be related to the overlap area, in other

words, ⇣Aoverlap

ARotor

will be added to velocity deficit formula (See Equation 5.6). The intersec-tions for different wake regions on a turbine rotor area will be calculated, the the corresponding

area will be assigned to different wind speeds. In case of overlap of the wake regions (1st and

2ndturbines effect the same region on a turbine rotor area) the average of the speeds will be taken.

u = U1·  1 1 p1 c T (1 + 2ks)2 · ✓A overlap ARotor ◆ (5.6)

5.2 The Jensen Model Applied to Hexicon Platform

On a Hexicon platform the turbines are not located downstream of each other i.e. they have free flowing wind at all times. When this is the case, it would be expected that first column turbines won’t create any wake regions on the second column turbines. However, this case can only be expected when the turbines are perpendicular to the incoming wind direction. In a case that the platform will be controlled to face the wind perpendicularly, no power losses will occur from the wake. Before starting to investigate the Hexicon platform behavior by applying the Jensen Model, it is noteworthy to talk about the characteristics of the turbine and the platform. The Jensen Model is implemented to MATLAB by using the turbine and platform characteristics that are shown in Table 7.

Table 7: Characteristics of Turbines and Hexicon Platform

Hub Height 105m

Rotor Diameter 138 m

Capacity of a Turbine 6MW

Cut-in Wind Speed 2 m/s

Cut-out Wind Speed 25 m/s

Dimensions of the platform 507x215 m

(43)

been mentioned before, Wake Decay Constant, k, has a significant importance in the Jensen Model and 0.04 is the suggested k value for offshore conditions (See Section 5.1.1). So by using this value the simulation was run to see how the turbines are affected by the wake. Figure 19(a) shows the wake interaction on Hexicon H4-24MW platform.

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(a) Jensen Model - MATLAB (b) Actuator Disk Theory - ANSYS CFX

Figure 19: Comparison of the Jensen Model and Actuator Disk Theory for Wake Effect Expansion on Hexicon Platform, k = 0.04

As it can be seen from Figure 19(a), incoming wind for second column turbines is not disturbed by wake when the k is selected as 0.04 and the rotor diameter as 138 m. It is important to com-pare the results with other available software in order to see the performance of the simulation. Hexicon AB runs related simulations on ANSYS CFX, which is set to use Actuator Disk Theory. The result for wake expansion on Hexicon platform by using k = 0.04 is shown in Figure 19(b). When simulation results from MATLAB and ANSYS CFX is compared (See Figure 19). It can be concluded that the results are the same, i.e. neither show any wake regions in the platform. In order to see the effect of k, a comparison can be made by running the simulations for different

kvalues. In general, for onshore conditions, k is suggested as 0.075, which is a higher constant

than suggested offshore k value, since the terrain roughness is higher onshore; this means wake diameter is bigger than offshore conditions. The experimental value of k is selected based on onshore conditions and the results can be seen in Figure 20.

(44)

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

Figure 20: Wake Effect on Hexicon Platform k=0.075

can be commented as wake interaction is observed in the wake region of the 3rdand 4th turbines,

however, since the upstream wind for 3rd and 4th is not in the wake region, wake doesn’t create

velocity deficit in the incoming wind. For the future studies, when it comes to design a wind farm by using Hexicon Platforms, this intersection has to be taken into account. However, for a single platform, even though the wake decay applied is higher than suggested,incoming wind is not disturbed by the wake regions. In conclusion, the layout that Hexicon AB suggested is tolerant to the worst case scenerio k value.

In the following section this study will be repeated for different wind directions, also known as attack angles, to see until which direction the layout is tolerant to wake effect when the nacelle is not yawed and platform is fixed.

5.3 Effect of Wind Direction in Wake Interactions

5.3.1 Wake Interaction in Fixed Nacelle and Platform

(45)

Before going to the reasoning behind these assumptions, it is important to mention the con-clusion of Section 2.4. Recently, the industry has developed new measurement techniques which are used to measure and monitor the wind speed and wind direction in a more accurate way. With the help of recent developments, for instance, LIDAR and iSpin, more energy is captured, so that the projects are able to pay off faster than before. With the help of developed measure-ment techniques control systems are used more accurately. Nevertheless this section is going to assume that those control systems are not applied to the turbines.

The reasoning behind these assumptions are to understand the wake interactions when there wouldn’t be a control system to follow the wind direction. It is also noteworthy to mention about yaw misalignment at this point. The modern control systems are able to react the wind direction change very quickly. However, the wind direction change is faster than nacelle yawing, so most of the operating times turbines cannot face the wind perpendicularly [34, 35]. Taking this into account, the following studies are going to be performed by assuming the Nacelle and the Platform fixed.

The Jensen model is applied to see the wake interaction when the wind direction changes in between 90 to 90 , North to South. It is accepted that turbines face the West and corre-sponding angle for West is 0 . The first angle that the turbine started to be affected is 60 .

When the wind blows from this direction the 4th turbine is almost fully under the wake of the

1st turbine. Almost 30% of the speed is diminished in the wake, which means the wind speed

that arrives to the 4th turbine decreases to 6.30m/s from 9m/s.

The wake keeps affecting turbines for 50 as well. For this direction 4th turbine is fully under

the wake. This angle might be commented as the most undesired degree as the 4th turbine is

not only under the wake of both turbines in the first column but also the wake is overlapped, which means there are 3 different wind speeds that the turbine faces, which can also be explained

by dividing the rotor area into three. The first area is under the effect of the 1st turbine, the

second area is under the effect of the 2nd turbine and the third area is under the effect of both

(46)

−100 0 100 200 300 400 500 600 −600 −400 −200 0 200 400 600 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

Figure 21: Direction of the wind = 60

After applying this approach it is found that the velocity in the 4th turbine when the wind

blows from 50 is 5.53m/s (the undisturbed wind speed is 9m/s), which can be commented for this layout, the velocity deficit is higher than fully wake region when an overlap region occurs.

Different wake effect regions for the 4th turbine can be seen in Figure 22.

−100 0 100 200 300 400 500 600 −600 −400 −200 0 200 400 600 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

Figure 22: Direction of the wind = 50

The next angle investigated was 40 . As it can be seen in Figure 23, 4th turbine is under fully

wake of 2ndturbine, where 1thturbine doesn’t effect the rotor area. When the wind blows at an

angle of 40 , the wind speed that the turbine receives is 5.51m/s which results 39% of velocity

deficit on the 4th turbine.

40 is the last degree that creates a fully wake region for the 4th turbine. The following wind

(47)

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4 a

Figure 23: Direction of the wind = 40

instance, as it can be seen in Figure 24(b) 14% of the turbine rotor area is under the wake which

is caused by 2nd turbine and this wake interaction causes a decrease of 1.26m/s in the wind

speed and resulted as 7.74m/s.

−100 0 100 200 300 400 500 600 −600 −400 −200 0 200 400 600 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(a) Direction of the wind = 30

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4 b

(b) Direction of the wind = 20

−100 0 100 200 300 400 500 600 −600 −400 −200 0 200 400 600 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(c) Direction of the wind = 10

(48)

As it was mentioned in the beginning of the section, this section assumes that nacelle control is not in use. In the existing systems, the control mechanisms, including nacelle control, are not only used to capture more energy but also to protect the turbine from high level loads and high turbulence intensity levels in order to maintain the predicted life time of the turbine. For

instance, Figure 23 may not be the general case because 4th turbine is under full wake. It is

important to mention that, in real life when this scenario is predicted, the turbine might be shutted down by control system in order not to cause any damage in the turbine. Having said that, platform turning depends on the balance in the thrust force, shutting down a turbine might create unexpected circumstances for platform behavior. Hence, some attention should be paid to this scenario and the control system should be designed to avoid its occurrence.

Lastly, as it can be seen in Figure 25, when symmetric angles are applied, the same velocity

deficit levels can be found out for the 3rd turbine.

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4 c

(a) Direction of the wind = 60

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(b) Direction of the wind = 50

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(c) Direction of the wind = 40

(49)

5.3.2 Wake Interaction in Turbine Yawing

The previous section gave an overview about wind direction change when the turbines nacelles’ and the platform are fixed. As it just mentioned, it is known that, turbines that are in use today have nacelle yawing, which means turning the turbine to a direction that the turbine will be facing wind perpendicularly. However, as it was explained in Section 2.3 nacelle yawing might create wake effect problems since the layout is not designed by considering the nacelle yawing. As it was also mentioned in Section 2.3, turbines on a Hexicon platform will be yawed, however; it is desired to set a degree in which turbines are not effected by wake. That degree will be used to define the platform behavior as well, up to that degree only nacelle yawing is going to be used and after that degree, platform is going to start to turn automatically.

Before giving the details about nacelle yawing, it is important to explain a modification that is done in wind direction data. It is known that wind direction data in wind rose map uses a different coordinate system than a regular coordinate system, meaning that 0 for normal co-ordinate points out the East, on the other hand, 0 in wind rose map points out the North. The modification was done based on wind rose map. In Section 3.3.1 the dominant wind di-rection is shown as West, which corresponds to 270 in wind rose map. In order to simplify the changes around 270 , the data is modified based on accepting 270 as 0 . In other words, in the following examples, ⌥x ’s are done for the angles that are around dominant wind direction. The simulations are done by following assumptions; all the turbines have a center and accord-ingly to the angle of upcoming wind, turbines are going to rotate by accepting their center as the rotating origin. In order to see the effect of nacelle yawing on Hexicon Platform, different simulations have been run. The suggested angle for having wake interaction according to nacelle yawing by Hexicon AB is 10 . In order to make sure that lower nacelle yawing degrees don’t cause any wake regions, it is decided to start simulations from 2.5 and continue simulations with an increment of 0.5.

(50)

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(a) Wind and Nacelle direction = 2.5

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(b) Wind and Nacelle direction = 3

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(c) Wind and Nacelle direction = 3.5

Figure 26: Direction of the wind and Nacelle = 2.5 , 3 , 3.5 - MATLAB

However, it is observed that at ⌥3.5 the wake interaction starts and the wake region increases for following degrees. In other words, the last degree that the power output is 100% for all individual turbines is observed as ⌥3.5 . The experiment continued for 4 , 4.5 and 5 , in order to see how big the nacelle yawing impact in the turbines. Results of the simulations can be seen in Figure 27. As it can be seen from Figure 27 the wake interaction is very low, meaning that 4 nacelle yawing causes 1.5% power loss, 4.5 nacelle yawing causes 4% power loss and 5 nacelle yawing causes 6% power loss.

(51)

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(a) Wind and Nacelle direction = 4

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(b) Wind and Nacelle direction = 4.5

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(c) Wind and Nacelle direction = 5

Figure 27: Direction of the wind and Nacelle = 4 , 4.5 , 5 - MATLAB

To sum up, the results showed that in order not to cause any wake interaction by yawing the nacelle, the proposed system will be allowing nacelle yawing up to ⌥3.5 , so that 100% power output can be expected.

5.3.3 The Jensen Model Comparison with CFD Simulations

As the wake diameter expansion simulations were compared with Hexicon’s simulation in Sec-tion 5.2, the results for nacelle yawing is going be compared as well. As it was menSec-tioned before, in the beginning of this study Hexicon had results for 0 and 10 incoming wind angle, where the wind was perpendicular to the turbines. Those simulations showed that nacelle yawing for

10 creates wake region on the 3rdturbine, which is in the second column. After the simulations

(52)

(a) Wind and Nacelle direction = 3 (b) Wind and Nacelle direction = 4

(c) Wind and Nacelle direction = 5

Figure 28: Direction of the wind and Nacelle = 3 , 4 , 5 - ANSYS CFX

After comparing wake diameter expansion for different wind directions, it was seen that at 4 ,

wake region starts to be occurred in the 3rd turbine. The percentage of the wind rotor area that

is under the wake is not known; however, it is known that the wind speed is 3.05% less in the wake region. Whereas wind speed deficit for 5 nacelle yawing is 5.52% .

(53)

5.3.4 Wake Interaction in Turned Platform

As it was explained before, the aim of this project is to decide the behavior of Hexicon platform accordingly to incoming wind direction. Results for two cases has already been shown in previous sections. When the turbines and platform were fixed, the incoming wind angle was changed from North to South. Undesired wake regions occurred in front of the turbines, which are located in

the 2nd column in the platform. These simulations followed by yawed nacelle and fixed platform.

In Section 5.3.2, it was demonstrated that after a certain degree nacelle yawing causes wake interactions. The Hexicon concept is developed in order not to have those wake regions which is made possible by turning the platform to a position that turbines will always face the free wind. Figure 29 is a demonstration of Hexicon concept. It can be seen from Figure 29 that when the Hexicon technology applied to platform, none of the turbines are going to be under the wake interaction for any wind direction.

Since none of the turbines are under the wake, it can be concluded that the power output will only be based on the wind speed for all the directions since the turbine will be turned ac-cordingly. As it was explained in Section 1.1, the platform is going to turn automatically. As it was mentioned before, one way of automatic will be based on pitch control. In other words, by having thrust force differences on the blade. Having those differences is going to create some power losses, since the blades will be pitched to control forces on the blade, meaning diminished forces on the blade. Those losses that will be created by turning the platform by pitch control are ignored in this study. In addition to this automatic turn, in case of dramatic changes in incoming wind angle, the wind will be followed by activating the thrusters. It was pointed out that thrusters will consume energy which will be included to this study.

Having these said, the control system has to be designed in a way that the platform would be turned for the cases that the platform turning energy consumption would be compensated. In other words, the energy that will be generated from turbines by turning the platform should be more than the energy that the thrusters consume for turning the platform.

(54)

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(a) Direction of the wind = 70

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(b) Direction of the wind = 40

−400 −300 −200 −100 0 100 200 300 400 −400 −300 −200 −100 0 100 200 300 400 Downstrean Distance [m] Rotor Positioning Turbine 1 Turbine 1 Turbine 2 Turbine 2 Turbine 3 Turbine 3 Turbine 4 Turbine 4

(c) Direction of the wind = 20

Figure 29: Direction of the wind = 70 , 40 , 20

6 Power Calculations

The wake diameter expansion and its effect on velocity were shown in Section 5.3. After observ-ing the characteristics of wake expansion on a turbine, it is important to see the wake expansion effect on power output as well. The power output is calculated as it is shown in Equation (6.1). The first power output results will be shown for Fixed Nacelle and Platform scenario. In order to perform the calculation, incoming wind speed is chosen as 9 m/s which is a close value to

average wind speed for the selected site. Corresponding cp value is used accordingly to 6MW

turbine characteristics and the air density, ⇢, is accepted as 1.225 kg/m3.

P (v) = 1

2⇢Av

3c

References

Related documents

Since the water surface is smooth compared to land surface, the atmospheric flow in the MABL experiences lower wind drag which results in higher wind velocity, lower

This study presents an alternative to rural exploitation, it investigates whether it is possible to produce sufficient wind power to satisfy urban demand within 20 000 meters of

The cost per assembly as function of production volume is seen in Figure 12 (Stewart &amp; Gilbert, 2014). Figure 12: Breakeven point for HPDC and DMLS. In 2015, Niklas Kretzschmar

Factors that, in several studies, have shown to be of importance for creating acceptance for a project in the local community are participation in the planning process, information

Entering cost data is done by entering cost per unit (i.e. component), which considers all the costs from in-house and outsourcing as well as the number of units. Two types of

In order to eval- uate the impact of the wind farm, average values of all three spectral moments (the radar reflectivity factor, absolute radial velocity, and spectrum width) of

Floating offshore wind turbines are usually divided into the substructure (the platform) that interacts with waves, current and wind, the turbine (tower and RNA) sensitive to wind

Figure 3.8 is presenting density measurements (upper graph) from the Cluster satellites as they encountered an event inside the magnetosheath. This example illustrates a case