Electrical Energy Systems Department of Electrical Engineering Den Dolech 2, 5612 AZ Eindhoven P.O. Box 90159, 5600 RM Eindhoven The Netherlands www.tue.nl
Author:
M.A. Shoeb Student ID:
0827705
Integration of Wind Power into Electricity Markets
Supervisors:
prof. ir. W.L. Kling ir. ing. J.E.S. de Haan H. M. Lopes Ferreira MSc
By Md. Asaduzzaman Shoeb
Reference EES.13.A.0002
Date July 2013
The department of Electrical Engineering of the Eindhoven University of Technology disclaims any responsibility for the contents of this report
2
3 This thesis is the end product of a graduation project for the master’s program in Sustainable Energy Technology at Eindhoven University of Technology (TU/e). The research was performed at the research group Electrical Energy Systems (EES), which is part of the department of electrical engineering at Eindhoven University of Technology. This project was part of the EIT KIC Innoenergy Offwindtech project WP1. The research has been carried out from October 2012 to June 2013. Electrical Energy Systems Department of Electrical Engineering Den Dolech 2, 5612 AZ Eindhoven P.O. Box 513, 5600 MB Eindhoven The Netherlands www.tue.nl This report describes the activities conducted within this research project. The main conclusions of this research will be presented at the 12th International Workshop on Large-Scale Integration of Wind Power into Power Systems as well as on Transmission Networks for Offshore Wind Power Plants, London UK, 22-24 October 2013. The accompanying draft version of related conference contribution is entitled ’Market value of Wind Power’.
4
5
Acknowledgements First of all I would like to express heartiest gratitude to European Commission, KIC- Innoenergy and Global Sustainable Electricity Partnership for giving the opportunity and all sorts of support to pursue the prestigious Erasmus Mundus masters program Environomical Pathways for Sustainable Energy Systems (SELECT). This research project would not have been possible without the support and encouragement of many people. I take this opportunity to express gratitude to the people who have been instrumental in the successful completion of this project. First of all I would like to thank my supervisor, Prof. W.L. Kling, for giving the opportunity to conduct this research project. I would like to express deepest gratitude to my advisors, J.E.S. de Haan and H.M. Lopes Ferreira who were abundantly helpful and offered invaluable assistance, support and guidance. Special thanks to Jerom for being such a tremendous supportive and inspiring personality. I felt motivated and encouraged every time I attended his meeting. I am also thankful to Phuong Nguyen, B.M.J. Vonk (Bram) and Asare-Bediako (Ballard). I wish to express gratitude to my fellow students, Md. Nasimul Islam Maruf, Benozir Ahmed and Abu Niyam for sharing these two years journey of SELECT program with me. Without their company, life would be much more boring and difficult. I also like to thank other graduate students who shared the room with me during the graduation project, especially Jahidul Islam, for his unlimited source of information and sense of humor. I also would take this opportunity to express my love and gratitude to my beloved family for their understanding & endless love, throughout the duration of my studies. Finally, all praise to the almighty Allah (SWT), without His mercy nothing would have been possible.
6
7
Abstract Variability and unpredictability constraints of wind hinder the cost-efficient integration of wind power generation into the electricity markets. It is obvious that the forecast of wind power generation closer to real time is more accurate. This logical reasoning might lead to the hypothesis that wind power can be sold best in intra-day markets. However, it is certainly not evident that market prices at intra-day stage are equal or larger compared to prices at day ahead stage. Therefore, the actual market value of wind power depends on the accuracy of forecasting at day ahead and intra-day, and is determined by the related market prices at day ahead and intra-day moments. Within the framework of EIT KIC INNOENERGY Offwindtech project, a Market Value software tool is developed to assess the actual market value of wind power generation. Case studies of wind power plants can be investigated with respect to their geographical wind site and the respective power market it will be integrated in. The versatility of the tool enables to perform simulations with a diversity of objectives e.g. policy makers determine the requested subsidy / feed-in tariff for wind power generation. Investors can trace the optimal location of their planned wind power plant. Additionally, the effect of wind power generation conditions such as forecast techniques can also be compared. In this work, a case study is introduced to evaluate the potential market value of different wind sites (offshore/onshore) and of different power market concepts (day ahead/intra-day) in the Netherlands for a reference year 2011. The geographical distinction of wind sites is made between offshore, onshore close to shore (coastal), and inland onshore wind power generation. These three locations differ based on the combination of wind regime (large wind yield at offshore) and costs of wind farm installation (low costs at onshore), operation and maintenance. This assessment elaborates on the hypothesis that onshore wind power; traded closer to real time would be more beneficial, due to lower installation costs and smaller imbalance costs. The results show that the average revenue per MWh wind generation almost equals for all three locations. However, the yearly gross revenue in euro is approximately 20% higher for the offshore and coastal location compared to the onshore location. The data analysis reveals that the forecast of onshore wind power generation is more optimistic compared to the forecast of offshore and coastal wind generation. This will result in more negative imbalances for onshore wind generation, which are commonly come along with larger imbalance costs. This effect has the largest impact at day ahead markets, where at intra-day markets, closer to real time, this effect attenuate. The implementation of more accurate forecasting limits the expense of total imbalance costs. The results support the general hypothesis that wind farms located at the east coast of the Netherlands have approximately the wind regime as an offshore location, since the market value of the wind farms located at offshore and coastal sites are very similar. However, due to smaller installation costs, onshore wind power, close to shore (coastal), sold at intra-day markets is by far the most cost-efficient concept to implement wind power generation, not considering complementary assets within the generation portfolio of market parties.
8 A brief market analysis to commercially validate the software tool is also presented. The market is still very much new as wind farm evaluation tools from electricity market point of view are not frequently and freely available. Therefore, it has the potential to penetrate the market through execution of a complete business plan.
9
Table of Contents Acknowledgements ....................................................................................................5 Abstract ......................................................................................................................7 1 Introduction ........................................................................................................ 15 1.1 1.2 1.3 1.4
2
Project Objective ........................................................................................................................................ 15 Offwindtech project .................................................................................................................................. 15 Research Methodology ............................................................................................................................ 17 Thesis Layout ............................................................................................................................................... 17
Background ......................................................................................................... 19
2.1 Introduction ................................................................................................................................................. 19 2.2 Wind Energy................................................................................................................................................. 20 2.2.1 Energy from the Wind ..............................................................................................................................20 2.2.2 Wind Shear ....................................................................................................................................................21 2.2.3 Turbulence and Variability ....................................................................................................................23 2.2.4 Analytical Frequency Distribution of Wind ....................................................................................25 2.2.5 Power-Wind Speed characteristic.......................................................................................................27 2.3 Electricity Market....................................................................................................................................... 28 2.3.1 Electricity market structure ..................................................................................................................29 2.3.2 Market Architecture ..................................................................................................................................32 2.4 Wind Power Forecasting ......................................................................................................................... 34 2.4.1 Evaluation criteria of Wind Power Forecasting...........................................................................35 2.5 Summary and Conclusion ....................................................................................................................... 36
3
The Market Value Tool ........................................................................................ 37
3.1 Introduction ................................................................................................................................................. 37 3.2 Algorithm....................................................................................................................................................... 37 3.2.1 Wind speed from sensor height to hub height ...............................................................................38 3.2.2 Wind speed to wind power .....................................................................................................................38 3.2.3 Wind power and market price to market value ...........................................................................40 3.3 Application of the tool .............................................................................................................................. 41 3.4 Summary and Conclusion ....................................................................................................................... 42
4
Case Study .......................................................................................................... 43
4.1 Introduction ................................................................................................................................................. 43 4.2 Scenarios........................................................................................................................................................ 43 4.2.1 Geographical Locations ...........................................................................................................................43 4.3 Simulation ..................................................................................................................................................... 45 4.3.1 Market Prices ................................................................................................................................................46 4.3.2 Wind Power Generation ..........................................................................................................................47 4.3.3 Wind Speed Forecasting ..........................................................................................................................47 4.3.4 Power Imbalances ......................................................................................................................................48
10 4.4 4.5
5
Results ............................................................................................................................................................ 49 Summary and Conclusion ....................................................................................................................... 51
Validation of Results ........................................................................................... 53
5.1 Introduction ................................................................................................................................................. 53 5.2 Verification of Data.................................................................................................................................... 53 5.2.1 Wind speeds ...................................................................................................................................................54 5.2.2 Market Prices ................................................................................................................................................56 5.2.3 Power Imbalances ......................................................................................................................................57 5.3 Recommendations to reduce imbalance costs ............................................................................... 60 5.3.1 Energy Storage Technologies................................................................................................................61 5.4 Summary and Conclusion ....................................................................................................................... 64
6
Business Plan ...................................................................................................... 65
6.1 Introduction ................................................................................................................................................. 65 6.2 Market analysis ........................................................................................................................................... 65 6.2.1 Macro market ...............................................................................................................................................65 6.2.2 Micro market ................................................................................................................................................66 6.2.3 Porter five forces analysis .......................................................................................................................66 6.2.4 SWOT analysis ..............................................................................................................................................67 6.3 Business model canvas ............................................................................................................................ 68 6.3.1 Customer Segments....................................................................................................................................68 6.3.2 Value Proposition .......................................................................................................................................68 6.3.3 Channels ..........................................................................................................................................................68 6.3.4 Customer Relationship .............................................................................................................................68 6.3.5 Revenue Streams .........................................................................................................................................68 6.3.6 Key Resources ...............................................................................................................................................69 6.3.7 Key Activities .................................................................................................................................................69 6.3.8 Key Partnerships .........................................................................................................................................69 6.3.9 Cost Structure ...............................................................................................................................................69 6.4 Summary and Conclusion ....................................................................................................................... 69
7
Conclusions ......................................................................................................... 71
7.1 Conclusions................................................................................................................................................... 71 7.2 Recommendations ..................................................................................................................................... 72 7.2.1 Market Value Tool ......................................................................................................................................72 7.2.2 Storage possibility ......................................................................................................................................72 7.2.3 Graphical User Interface .........................................................................................................................73 7.2.4 Case study .......................................................................................................................................................73 7.2.5 Business Plan ................................................................................................................................................73
References ................................................................................................................ 74 Appendix .................................................................................................................. 81 Appendix A: Fatigue load ........................................................................................ 81 Appendix B: Wake effect of wind farm .................................................................... 82
11
Appendix C: Appendix D: Appendix E: Appendix F: Appendix G: Appendix H: Appendix I: Appendix J: Appendix K:
Key wind power forecasting methods found in literature ..................... 83 Demand and supply curve ................................................................... 85 Actors in Dutch electricity market ........................................................ 87 Graphical user interface of ‘Market Value’ tool .................................... 88 Weibull distributions of Wind Speed data sets .................................... 90 Imbalance cost.................................................................................... 93 Case study- Comparing two turbines .................................................... 95 Business model canvas ......................................................................... 99 Storage Technologies ........................................................................ 100
12
List of Figures FIGURE 1-1 RELATION AMONG THE FOUR TOOLS WITHIN THE OFFWINDTECH WP1 .......................................................... 16 FIGURE 1-2 SCHEMATIC DEPICTION OF THE TOOL ...................................................................................................................... 17 FIGURE 2-1 GLOBAL ANNUAL INSTALLED WIND CAPACITY 1996-2012 ............................................................................ 19 FIGURE 2-2 POWER COEFFICIENT CURVE WITH RESPECT TO TIP SPEED RATIO ..................................................................... 21 FIGURE 2-3 DECREASE IN WIND SPEED AS INFLUENCED BY VARIETIES OF TERRAIN ROUGHNESS ..................................... 23 FIGURE 2-4 SAMPLE WIND DATA SHOWING THE TURBULENCE ................................................................................................ 24 FIGURE 2-5 VAN DER HOVEN SPECTRUM ................................................................................................................................... 25 FIGURE 2-6 WEIBULL DISTRIBUTION FOR DIFFERENT K VALUES ............................................................................................ 26 FIGURE 2-7 TYPICAL POWER CURVE ............................................................................................................................................ 28 FIGURE 2-8 SIMPLE POWER SYSTEM NETWORK.......................................................................................................................... 29 FIGURE 2-9 COMPETITIVE ELECTRICITY MARKET....................................................................................................................... 30 FIGURE 2-10 ORGANIZATION OF ELECTRICITY SECTOR ............................................................................................................. 32 FIGURE 2-11 ELECTRICITY MARKET ARCHITECTURE ............................................................................................................... 33 FIGURE 2-12 POSITION OF GATE CLOSURE .................................................................................................................................. 34 FIGURE 3-1 INPUTS AND OUTPUTS OF MARKET VALUE TOOL ................................................................................................. 38 FIGURE 3-2 NORMALIZED STANDARD DEVIATION OF THE WIND DISTRIBUTION AT ANY GIVEN TIME ............................... 39 FIGURE 3-3 AGGREGATED POWER CURVE ................................................................................................................................... 40 FIGURE 3-4 TIME SERIES ENERGY CALCULATION........................................................................................................................ 40 FIGURE 3-5 MARKET VALUE CALCULATION ................................................................................................................................. 41 FIGURE 3-6 EXEMPLARILY DEPICTION OF ANNUAL PROFILE OF WIND FARM REVENUE ....................................................... 42 FIGURE 4-1 CASE STUDY SCENARIOS............................................................................................................................................ 43 FIGURE 4-2 TYPICAL WIND DIRECTION OF NETHERLANDS ....................................................................................................... 44 FIGURE 4-3 GEOGRAPHICAL LOCATIONS OF THREE WIND SITES UNDER CONSIDERATION ................................................... 45 FIGURE 4-4 ANNUAL PROFILE OF DAY AHEAD MARKET PRICE.................................................................................................. 46 FIGURE 4-5 YEARLY PROFILE OF WIND POWER GENERATION IN THE ONSHORE SITE ........................................................... 47 FIGURE 4-6 DURATION CURVE OF FORECAST ERROR FOR 24 HOURS UP TO 1 HOUR AHEAD FORECASTING ..................... 48 FIGURE 4-7 CLOSE VIEW OF THE VARIATIONS OF FORECAST ERROR ....................................................................................... 48 FIGURE 4-8 ANNUAL PROFILE OF WIND POWER IMBALANCES WITH DAY AHEAD FORECASTING ....................................... 49 FIGURE 5-1 INTRA DAY MARKET PRICE OF 3041 RANDOM HOURS OF DUTCH POWER MARKET IN 2011 ....................... 53 FIGURE 5-2 WEIBULL DISTRIBUTION OF REAL WIND SPEED IN ONSHORE SITE..................................................................... 54 FIGURE 5-3 WEIBULL DISTRIBUTION OF DAY AHEAD FORECASTED WIND SPEED IN OFFSHORE SITE ................................ 55 FIGURE 5-4 WEIBULL DISTRIBUTION OF DAY AHEAD FORECASTED WIND SPEED IN COASTAL SITE .................................. 55 FIGURE 5-5 DAY AHEAD MARKET PRICE OF DUTCH POWER MARKET IN 2011 ..................................................................... 56 FIGURE 5-6 3041 HOURS DATA OF DAY AHEAD MARKET PRICE .............................................................................................. 57 FIGURE 5-7 ANNUAL PROFILE OF POWER IMBALANCES AT ONSHORE SITE FOR 24 HOURS AHEAD WIND SPEED FORECASTING ......................................................................................................................................................................... 58 FIGURE 5-8 PROFILE OF 3041 HOURS OF POWER IMBALANCES AT ONSHORE SITE FOR 24 HOURS AHEAD WIND SPEED FORECASTING .......................................................................................................................................................................... 58 FIGURE 5-9 ANNUAL PROFILE OF POWER IMBALANCES AT ONSHORE SITE FOR 2 HOURS AHEAD WIND SPEED FORECASTING .......................................................................................................................................................................... 59 FIGURE 5-10 PROFILE OF 3041 HOURS OF POWER IMBALANCES AT ONSHORE SITE FOR 2 HOURS AHEAD WIND SPEED FORECASTING .......................................................................................................................................................................... 59 FIGURE 5-11 ANNUAL PROFILE OF IMBALANCE COSTS FOR DAY AHEAD FORECASTING AT THE COASTAL SITE................ 60
13 FIGURE 5-12 PROFILE OF 3041 HOURS OF IMBALANCE COSTS FOR DAY AHEAD FORECASTING AT THE COASTAL SITE . 60 FIGURE 6-1 PORTER'S FIVE FORCES .............................................................................................................................................. 66 FIGURE 6-2 SWOT MATRIX OF ‘SELECT WIND’ ..................................................................................................................... 67 FIGURE A-1 TYPICAL SN CURVE .................................................................................................................................................... 81 FIGURE B-2 WAKE EFFECT IN A WIND FARM ............................................................................................................................. 82
List of Tables TABLE 2-1 APPROXIMATE VALUES OF ROUGHNESS LENGTH FOR VARIOUS TYPES OF TERRAIN .......................................... 22 TABLE 4-1 COORDINATES OF THE WIND SITES ........................................................................................................................... 44 TABLE 4-2 RESULTS ....................................................................................................................................................................... 50 TABLE 4-3 GROSS VALUES IF TRADED IN DAY AHEAD MARKET IN 2011................................................................................ 51 TABLE 5-1 AVERAGE WIND SPEED ............................................................................................................................................... 54
Nomenclature TSO DSO DNO BRP BSP PTU ENTSO-E GCT NWP ME MSE RMSE MAE SDE APX WPP IEA EWEA
Transmission System Operator Distribution System Operator Distribution Network Operator Balance Responsible Party Balance Service Provider Programme Time Unit European Network of Transmission System Operators for Electricity Gate Closer Time Numerical Weather Prediction Mean Error Mean Square Error Root Mean Square Error Mean Absolute Error Standard Deviation of the Error Amsterdam Power Exchange Wind Power Producer International Energy Agency The European Wind Energy Association
14
15
1 Introduction 1.1
Project Objective
In Europe, an unprecedented evolution towards the deployment of renewable energy sources, fostered by the Kyoto Protocol and by the European Union objectives has been perceived. Among these renewable energy sources, one has been evolving in particular in the last two decades. This one is wind energy. Nowadays, wind-based energy generation is showing a good maturity and presents the potential to compete in the energy market along with the traditional energy sources. The fourth edition of the Global Wind Energy Outlook shows that wind power could supply up to 12% of global electricity by 2020, creating 1.4 million new jobs and reducing CO2 emissions by more than 1.5 billion tons per year, more than 5 times today’s level. By 2030, wind power could provide more than 20% of the global electricity supply. However, in order to effectively integrate large shares of wind power in electricity supply systems, a successful market integration of wind power is regarded essential. Due to variability and predictability constraints of wind power generation, Balance Responsible Parties experience difficulties to meet their energy schedules. The rather negative correlation between system load and wind power generation, create financial challenges experienced by Balance Responsible Parties. Several new market arrangements have been introduced in particular for market parties, to enable a more cost-efficient integration of wind power generation into power markets. Therefore, a pre-emptive assessment of the wind energy value is a need to realize this market integration. It is necessary to determine the enhanced, cost effective scenario to integrate wind energy into power markets. However, there is a lack of tools accessible in this area. Investigation performed in this area usually makes use of internal developed tools, which were precisely built for any certain case study. Furthermore, those tools are often not published or available. The aim of this project is to develop a model to evaluate the market value of wind energy for any particular location, to be connected to the electricity supply system. Both, energy harvested by wind turbines, as well as electricity markets have been emphasized in this project. This model is being developed as a part of the framework of EIT KIC INNOENERGY Offwindtech project by the name of ‘Market Value’ tool. The main objective of the project is to design a simple, but effective tool to demonstrate the value of energy generated by wind farms of any specific location, to be sold in any particular power market.
1.2
Offwindtech project
EIT KIC INNOENERGY Offwindtech project is intended to develop a general tool for the analysis of the offshore wind farm development including technology selection, wind farm layout optimization, grid integration and market analysis. Within this framework, four tools are developed to enable a complete evaluation of any specific wind farm. The concept of the project is characterized by the possible interlink between all four separate tools. This enables an overall
16 assessment of specific wind farms, with detailed results. However, all tools can be used independently. The four tools within the Offwindtech project are: Wind Farm layout Grid connection Economic analysis Market value The relations among the tools are schematically depicted in Figure 1-1.
Figure 1-1 Relation among the four tools within the Offwindtech WP1
A brief introduction of the other three tools is given belowWind Farm layout: This tool consists in computing the total power output and the wind speed of each wind turbine for any wind farm layout and any wind condition (wind direction and wind speed) and to evaluate the options regarding the wind farm connection mainly referred to the inter-array connection of the wind turbines and its connection to the offshore platform. IREC (Catalonia Institute for Energy Research) is responsible for this tool. Grid connection: The tool aims to provide transmission system topologies with low costs to connect multiple offshore wind farms to the main grid. The objective is to identify feasible solutions, which can be analyzed in more detail by grid planners. The tool provides a first cost estimation for the transmission system to ease the decision of the grid planner. The second objective is to have a simple and generic tool, which can be used by different parties. Therefore, the tool offers a user-friendly interface for data input. The planner can specify different wind farm properties as well as investment costs. This way, the planner has the flexibility to adapt the algorithm to different wind farm locations. The tool uses reliable cost functions by default in case the planner doesn’t have cost data. KUL (Katholieke Universiteit Leuven) is responsible for this tool. Cost analysis: The main objective of this tool is the comparison of different floating structures for offshore wind in deep water through technical-economic aspects. An economic assessment has been done with the different incomes and expenses along the project life. CAPEX and OPEX costs calculation are described taking into account all the cost contributors: floating structure type, material, wind turbine, electric infrastructure (submarine cable, offshore substation), and energy production. The outputs of the economic study are economic indicators used to
17 determine the profitability of the project such as Net Present Value (NPV), which is calculated for each type of floating structure in order to compare different technologies. This tool has been developed by Tecnalia (www.tecnalia.com).
1.3
Research Methodology
This research begins with a literature review of wind energy and electricity markets, which is used in subsequent segments of the research. As wind power has variability and limited predictability constraints, its market value is not only dependent on short-term power markets energy prices but also on the location and type of wind turbines. The power-wind speed curve is investigated with respect to different parameters like turbulence intensity, roughness length and turbine type. Afterward, aggregated wind speed-power curves are developed to represent a wind farm instead of a single turbine. Wind speed data is used to estimate the energy generation from the wind farm using the wind speed-power curve. Market prices are incorporated with the generated energy to calculate the revenue (market value) of wind farms. The approach is schematically presented in Figure 1-2. The algorithm converts certain ‘historical’ time series of wind speeds and certain ‘historical’ time series of energy prices to a time series market value of wind power. A case study of three locations based on wind regimes in the Netherlands is investigated using this tool. Finally, a business analysis is presented for the tool in order to develop the commercial version.
Figure 1-2 Schematic depiction of the tool
1.4
Thesis Layout
The outline of this thesis looks as follows. Chapter 2, describes the basic concepts of wind power which includes the kinetic energy extraction from the wind flow, distribution of wind speed and different factors affecting the power generation. The electricity market concepts are also discussed in this chapter. The algorithm of ‘Market Value’ tool and its applications are discussed in chapter 3. A case study is presented in chapter 4, where six scenarios of wind farms of different wind regimes and market conditions in the Netherlands are compared from the market value point of view for the year
18 2011. The data sets used for the calculations of the case study are incomplete. Therefore, the representativeness of the incomplete data set with respect to a complete data set is assessed in chapter 5. Some approaches are also discussed in chapter 5 to reduce imbalance costs, specially supported by energy storage possibilities. Chapter 6 discusses about the market analysis for the commercialization of the ‘Market Value’ and the business plan. Finally chapter 7 contains an overview of the conclusions of the results and discussions presented in previous chapters and also includes recommendations for future improvements.
19
2 Background 2.1
Introduction
Sun unevenly heats the surface of the Earth. It is this unevenly heat that causes the wind, which is the movement of air from an area of high pressure to an area of low pressure. Cooler air moves in to fill the void as hot air rises. Wind energy has been harnessed for hundreds of years. The use of horizontal axis windmill for pumping water and grain grinding started back in about 500-900 AD in Persia [1]. In modern age, wind turbines are used to generate electricity from wind energy. The first half of the 20th century saw the construction or conceptualization of a number of larger wind turbines, which substantially influenced the development of today’s wind turbine technology [2]. The recent growth of the wind energy sector has been most phenomenal among renewable sources. Wind technology is now established as a mainstream electricity source. Additionally, the penetration of wind power has been increasing for the last two decades and is still going on. The average growth of wind power is 22.7% per year for the last five years. Until now more than 285 GW of wind power has been globally installed, among those, 45GW new capacity was added in 20121. Figure 2-1 shows the global annual installed capacity of wind.
Figure 2-1 Global Annual Installed Wind Capacity 1996-2012 (Source: gwec.net)
Worldwide interest in wind energy increased during the oil crisis in the 70s and 80s of the last century. But, prior to the crisis, wind energy industry faced a challenge, as the oil price was going down and the cost of wind energy were significantly larger. Different countries put on some market regulations to promote wind energy. The technological progress in wind power industry is taking the cost down to make it competitive. Along with the higher investment cost, and variability and predictability constraints, it is a challenge to integrate wind power into power systems. Market prices of energy and wind generation are hardly correlated because wind power generation does not follow system load. It is quite a challenging job for Balance Responsible Parties2 to cost-efficiently integrate wind power into power markets due to variability and predictability constraints. Power markets with gate closure time close to real time; in the likes of intra day market (gate closure time two hours ahead) could be a more suitable choice for wind power integration considering the more accurate forecast. 1 2
BTM Consult (http://www.btm.dk/reports/world+market+update+2012) Performed the role that the supply of energy corresponds to the anticipated consumption
20 To estimate the energy generated by wind turbines several phenomena have to be considered which include specially the geographical condition of the site, type of turbines applied and wind forecasting techniques. This chapter elaborates on the general concepts of wind energy, parameters to be considered to estimate energy generation by wind turbine and electricity market.
2.2
Wind Energy
About 1% to 2% of the energy coming from the sun is converted into wind energy, which is almost 100 times more than the energy converted into biomass by all plants on earth. Furthermore, electricity captured and converted from only 1% of wind energy available would supply the whole present consumption of the world. It is estimated that wind resource technically available is 53,000 terawatt hours per year, which more than doubles the total world demand for electricity predicted for 20203. 2.2.1 Energy from the Wind Wind is a moving object with mass, carrying kinetic energy in an amount, which can be determined given by the equation 2.1. (2.1) Here E is the kinetic energy in Joule (J), m is the air mass in Kilogram (Kg) and v denotes wind speed in m/s. Equation 2.2 is used to calculate the mass of air flowing into the turbine per second, where ̇ , A and denote air mass flow rate, swept area of the turbine and air density respectively. ̇ (2.2) And therefore, the power in the wind reflected on the rotor disc (with a certain swept area) of wind turbines is calculated by inserting equation 2.1 into equation 2.2, which results in equation 2.3. (2.3) This is a simplified way to calculate the power available in the wind reflected on the rotor disc of a turbine. The power is a function of the third power of the wind speed. Therefore, if the wind speed is doubled, power in the wind increases by a factor of eight. From general hypothesis, wind turbines cannot extract all the energy available in the wind, as the wind speed cannot be zero beyond the wind turbine. This concept introduces a coefficient to indicate the performance of extracting power from the wind. The so-called power coefficient of performance of a wind turbine is a measure of how efficiently the wind turbine converts the energy in the wind into electricity. The maximum value of the power coefficient is limited by the Betz’s limit[3]. A German physicist Albert Betz concluded in 1919 that no wind turbine could convert more than 16/27 (59.3%) of the kinetic energy of the wind into mechanical energy turning a rotor. The theoretical maximum power efficiency of any design of wind turbine is 0.59 3
http://www.megawind.com/eng/ener.html
21 (i.e. no more than 59% of the energy carried by the wind can be extracted by a wind turbine). This is called the “power coefficient”, Cp. The Cp value depends on the aerodynamic performance of the blades and is a function of tip speed ratio4. As depicted in Figure 2-2, the tip speed ratio has an optimal value for a maximum power coefficient, deviation from this value results in reduction of power coefficient value. Besides, blade pitch5 angle also have influence on the power coefficient, which is discussed in [4]. The Betz’s limit, tip speed ratio and power coefficient are further discussed in [3], [5], [6] and [2]. In practice however, obtainable values of the power coefficient center around 45 percent. This value below the theoretical limit is caused by the inefficiencies and losses attributed to different configurations, rotor blades profiles, finite wings, friction, and turbine designs. With consideration of other factors for the entire wind turbine system - e.g. the gearbox, bearings, generator, e.g. - only 10-30% of the power of the wind is ever actually converted into electricity.
Figure 2-2 Power coefficient curve with respect to tip speed ratio[7]
2.2.2 Wind Shear In wind energy industry, wind shear is generally defined as the variation of wind speed with height above the ground level. It is essential to accurately determine wind speeds at turbine hub heights (generally ranging from 60 to 130 meters[8]) because the turbine’s potential electricity generation, and therefore the economic feasibility is determined by the wind speed at hub height. Even though wind speed data is usually not measured at wind turbine hub heights, still measured data can be used along with wind shear models. The sensor height wind data will be extrapolated to a wind speed data set at desired hub heights. In wind energy studies, two mathematical models or ‘laws’ have generally been used to model the vertical profile of wind speed or the wind shear. The first approach, the log law, is based on The ratio of the blade tip speed and the wind speed is called tip speed ratio. Blade pitch or simply pitch refers to turning the angle of attack of the blades of the wind turbine into or out of the wind to control the production of power. 4 5
22 a combination of theoretical and empirical research. The second approach, is the power law [2]. Both approaches are subject to uncertainty caused by the variable, complex nature of (wind) flows. A case study to show the comparison of these two laws is performed in [9]. Based on statistical analyses of prediction errors, insignificant differences can be found between these two laws. Some other case studies are analyzed in [10], [11]. ( ) ( )
(
)
(
)
(2.4)
The log law is based on principles of boundary layer flow and is given below as equation 2.4[2], where Z and Zr are the target and reference heights, respectively. V(Z) and V(Zr) are the wind speeds of target and reference height and Zo is the surface roughness length. The surface roughness length is a parameter, characterizes shear and is also the height above ground level where the wind speed is theoretically zero. It varies according to the terrain of the site. The typical values of surface roughness lengths are given in Table 2-1. Table 2-1 Approximate values of roughness length for various types of terrain[2]
Terrain Description Roughness length, Z0 (m) Very smooth, ice or mud 0.00001 Calm open sea 0.0002 0.0005 Blown sea Snow surface 0.003 Lawn grass 0.008 Few trees 0.01 Rough pasture 0.03 Fallow field 0.05 Crops 0.1 Many trees, hedges, few buildings 0.25 Forest and woodlands 0.5
Suburbs 1.5 Centers of cities with tall buildings 3 The effect of different terrain conditions on the roughness length for wind speed calculations is shown in Figure 2-3. It requires higher height to get the same wind speed for a place with bigger roughness length compared to the ones with smaller roughness length. Sometimes the log law is modified to consider the effective ground level at a site [2][9] .
23
Figure 2-3 Decrease in wind speed as influenced by varieties of terrain roughness [12]
The power law represents a simple model for the vertical wind speed profile; equation 2.5 [2] shows the empirically developed equation. Here, is the power law exponent and other variables are defined as before. ( ) ( )
(
)
(2.5)
Many studies showed that under certain conditions is equal to 1/7 [13], indicating a correspondence between wind profiles and flow over flat terrain. In practice, the exponent is a highly variable quantity. Even though, some researches indicates that the exponent as a function of roughness length and wind speed [13],[14]. A study showed that the calculated values of wind shear parameters are often inconsistent with the typically assumed values for the power law exponents and surface roughness lengths [9] and even for the flat sites power law does not represent the wind shear in [9]. For best representation of the wind shear at a site, analyzing the wind data is an important aspect of accurate prediction of hub height wind speeds rather than relying on tabulated values of wind shear parameters. 2.2.3 Turbulence and Variability Turbulence is the stochastic wind speed fluctuation imposed on the mean wind speed. It is generally accepted that variations in wind speed due to turbulence occurs with periods from less than a second to tens of minutes and that it has a random character. Figure 2-4 shows a typical pattern of wind fluctuation due to turbulence for a period of ten minutes. These fluctuations occur in all three directions: longitudinal (in the direction of the wind), lateral (perpendicular to the average wind), and vertical. Turbulent fluctuations in the flow need to be
24 quantified for wind energy applications. For example, turbine design considerations can include maximum load and fatigue (see Appendix A) prediction, structural excitations, control, system operation, and power quality [2]. Turbulence also has effect on the power generation [15], [16].
Figure 2-4 Sample wind data showing the turbulence[2]
The most basic measure of turbulence is the turbulence intensity. It is defined by the ratio of the standard deviation of the wind speed to the mean wind speed. In this calculation, both the mean and standard deviation are calculated over a time period longer than that of the turbulent fluctuations, but shorter than periods associated with other types of wind speed variations. The length of this time period is normally no more than an hour. The turbulence intensity, I is defined by equation 2.6 where 𝜎 is standard deviation and U is mean wind speed. 𝜎 (2.6) Turbulence intensity depends on a number of site-specific conditions like surface roughness and vertical gradient (large shear) in lower boundary layer and has a wide range of variation. Common values for turbulence intensity on land are 10% to 15%, which decreases with wind speed. This value on sea is generally as low as 5% but increases with wind speed because of higher waves and more surface length. The values of 20% and higher in free stream are considered as high and cause fatigue damage [17]. Effective turbulence in wind farm is higher than in free stream due to the wake effect (see Appendix B). The fluctuation of wind speed over a longer period of time (from hours to years) is represented by the term, variability and very important for wind resource assessments [17]. This variation could be inter-annual, annual or diurnal. The inter-annual variations in wind speed occur over a time period longer than one year. Estimation of inter annual variability of wind at a given site is important to estimate the mean wind speed of that site. This variation has a large effect on long-term energy generation from wind turbine. The annual variability is the significant variations in seasonal or monthly averaged wind speeds, which are very common over most of the places in the world. This variability reflects the season or months with maximum and minimum mean wind speed. Large wind variations can also occur on a diurnal or daily time
25 scale. This type of wind speed variation is due to differential heating of the earth’s surface during the daily radiation cycle. Daily variations in solar radiation are responsible for diurnal wind variations in temperate latitudes over relatively flat land areas. The largest diurnal changes generally occur in spring and summer, and the smallest in winter[2]. The variance of wind speed fluctuations for different time periods is depicted in Figure 2-5 by the Van der Hoven spectrum. The maxima for fluctuations occur with periods of four days, one day and one minute, while the periods of between 10 min to 2 hours have very low fluctuations. This pattern of fluctuation contains useful information to make decisions about the wind speedforecasting horizon.
Figure 2-5 Van der Hoven Spectrum (source: www.ecn.nl)
2.2.4 Analytical Frequency Distribution of Wind Statistical analyses can be used to determine the wind energy potential of a given site and to estimate the energy output from a wind turbine installed there[18][19]. If time series measured data are available at the desired location and height, there may be very minimal need for data analyses in terms of probability distributions and statistical techniques. On the other hand, if projection or comparison of measured data from one location to another is required, or when only an incomplete data set is available, then there are distinct advantages for the use of analytical representations for the probability distribution of wind speeds[2]. A probability distribution is a term that describes the likelihood that certain values of wind speed will occur within a certain time horizon. Probability distributions are typically characterized by a probability density function or a cumulative density function. In wind data analysis, two probability distributions most commonly are used- Rayleigh distribution and Weibull distribution.
26 Weibull distribution is based on two parameters while Rayleigh distribution uses only the mean wind speed. Thus Weibull distribution is better in terms of representing a wider variety of wind regimes. The ability to assume the characteristics of many types of distribution functions make Weibull the most commonly used distribution. Weibull probability density functions require two parameters: shape factor, K and scale factor A. Both of these parameters are functions of mean wind speed and variance. These two parameters are defined by the equation 2.7 and 2.8 where Vmean and σ2 are mean wind speed and variance respectively. The common values of shape factor are in between 1.4 to 3.5. For these values the argument of the Gamma function is in between 1.25 to 1.7 [17]. The detail of Gamma function is available in [20]. ( 𝜎
)
(
)
(
)
(2.7) (2.8)
The Weibull probability density function is given by the equation 2.9 ( )
( )
( ( ) )
(2.9)
Examples of a Weibull probability density function, for various values of K, are given in Figure 2-6. As shown, as the value of K increases, the curve has a sharper peak, indicating that there is less wind speed variation. Thus, large shape factor indicates less turbulence. On the other hand, larger scale factor implies higher probability of high wind speed. Higher wind speed above the rated value results stable rated power output. Therefore, higher value of shape and scale factor indicates a good wind site. Weibull distributions are used in this thesis to compare different sets of wind speed data.
Figure 2-6 Weibull distribution for different K values[2]
27 2.2.5 Power-Wind Speed characteristic Energy extraction from wind by a particular wind turbine at a variety of different wind speeds and regimes is described by the wind power curves. While these curves have different shapes, they are specific to a particular turbine and offer insights when choosing a wind turbine for an individual location [21]. A typical power curve of a large (capacity of MWs), three bladed, pitch controlled wind turbine is shown in Figure 2-7. The elaboration of characteristic features within a power curve is given below: Cut-in Wind Speed: At very low wind speeds, there is insufficient torque exerted from the wind by the turbine blades to efficiently rotate the wind turbine. Besides, electrical constraints disable wind turbines operating at low wind speeds. However, as the wind speed increases, the wind turbine will begin to rotate and generate electrical power. The speed at which the turbine first starts to rotate and generate power is called the cut-in wind speed. Rated wind Speed and Rated Power: The power output of wind turbine is proportional to the cube of the wind speed. Therefore, the electrical output power rises rapidly with the increase of wind speed above the cut-in speed as shown in Figure 2-7. However, typically somewhere between 10 and 17 meters per second, the power output reaches the limit that the electrical generator is capable of. This limit to the generator output is called the rated power output and the wind speed at which it is reached is called the rated wind speed. At higher wind speeds, the design of the turbine is arranged to limit the power to this maximum level and there is no further rise in the output power. This power regulation after the rated wind speed is done by speed control and by pitch- or stall- control mechanisms. With pitch control mechanisms, the blade angle with rotor hub is varied6, to reduce lift or increase drag on the blades. This control of aerodynamic power on rotor disc enables to maintain a stable power output at higher wind speeds. On the other hand, stall control is governed by the aerodynamic of the blades. The blades are designed in such a way that after the rated speed the power extraction tends to limit under the rated value. However, typically large turbines use the pitch control mechanism to keep the power at the constant level. Cut-out Wind Speed: As the speed increases above the rated wind speed, the forces on the turbine structure continue to rise and, at some point, there is a risk of damage to the rotor. As a result, a braking system is employed to bring the rotor to a standstill. This is called the cut-out speed and is usually around 25 meters per second.
6
Change in blade pitch angle, changes the value of power coefficient.
28
Figure 2-7 Typical Power Curve
Wind turbine power curves are sensitive to different site-specific parameters. In [15], [22–24] the effect of turbulence intensity and terrain on the power curve is described. For the case of a wind farm, where numbers of wind turbines are aggregated, the wind speed not only varies in time but also in space. Hence, the power generations from the individual turbines are not fully correlated. Therefore, for a wind farm, the power curve of one single turbine is not sufficient. Aggregated power curve representing all the turbines in the farm essential to determine. The more units and larger distance between the wind turbine units, the lower level of high frequency power fluctuations in aggregated power. In [25] the methodology is described to determine a multi-turbine power curve which is valid for every single wind turbine within the area. The aggregated power curve has smoothened cut-in and cut-off region. The power curve of a particular wind turbine is mostly free accessible.
2.3
Electricity Market
As depicted in Figure 2-8 electricity generated from power stations reach the consumers through transmission and distribution networks. In the beginning, during the late 1800s there was brutal and inefficient competition among power companies with overlapping distribution lines, competition for customers was fierce and cost of electricity was high. This problem inspired the introduction of natural monopoly and regulated electricity business. Then, the regulation was at the state level rather public [26]. But, as transmission technology developed, it brought new possibilities for trade and competition.
29
Figure 2-8 Simple power system network[27]
De-integration of electric industry became possible with the substantial development in the long distance transmission system. Generation could be split off to form a separate competitive market. While, the remaining parts of the utilities were left behind as regulated monopolies. By 1990, the general trend towards deregulation influenced many countries towards de-integrated electricity market. Many European countries have pioneered the creation of competitive markets since the initiation of market liberalization process in the EU in 1996 and 2003[28][29]. 2.3.1 Electricity market structure De-integration, unbundling or introducing competitive markets in all segments (i.e. generation, transmission, distribution) of the electricity industry is not practically feasible. The distribution and transmission of electrical power involve natural monopolies where competition would not work on the long run, due to the large fixed costs of the investment and dominant economies of scale. As shown in Figure 2-9, generation companies can compete in the wholesale market, where the retailers buy the energy according to the demand in the retail market. Afterwards, consumers buy the energy from the competitive retail market though large consumers often buy from the wholesale market. Transmission system operators (TSO) and distribution system operator (DSO) maintain transmission and distribution systems respectively. However, the balance between generation and consumption is very important for system operation close to the nominal frequency and respecting voltage limitations [30].
30
Figure 2-9 Competitive electricity market
The foremost actors in the electricity market mechanism are explained below: Transmission System Operator (TSO): A transmission system operator (TSO) is an operator that supports the transmission of electrical power from generation plants over the high voltage grid to regional or local electricity distribution operators. It has to be a non-commercial organization, neutral and independent with regard to the market members [31]. Secondly, the TSO is responsible for the instantaneous balance between generation and system load. The role of the TSO in a wholesale electricity market is to manage the security of the power system in real time and co-ordinate the supply of and demand for electricity, in a manner that avoids fluctuations in frequency or interruptions of supply. The System Operator also guarantees the provision of reserves that will allow for sudden contingencies. It is attained by defining the optimal combination of generating stations and reserve providers for each market trading period, instructing generators when and how much electricity to generate, and managing any contingent events that cause the balance between supply and demand to be disrupted. In addition to the roles of real-time dispatch of generation and managing security, the System Operator also carries out investigations and planning to ensure that supply can meet demand and system security can be maintained during future trading periods [32]. All the TSOs in the EU form an entity called The European Network of Transmission System Operators for Electricity (ENTSO-E). ENTSO-E are assigned to implement the new European legislation about cross border exchanges of electricity, ensure optimal management, cross border trade of electricity and coordination of operation and technical evolution of the European electricity transmission network. The functions and challenges of European TSOs are explained in [33]. Distribution System Operator (DSO): The distribution system consists of the electricity lines, cables and substations that connect the built environment and industry to the distribution grid.
31 The role of distribution system operator is to operate, maintain and develop an efficient electricity distribution system. Moreover, the DSO is also responsible to facilitate effective and well-functioning retail markets7. Effective retail markets give options to the customers, allow them to choose the best supplier and allow suppliers to offer options and services best tailored to customer needs [34]. The functions of DSO are elaborated in [35], [36]. Balance Responsible Party (BRP): A balance responsible party is the entity, having generation or load or both in their portfolio and is responsible to maintain the anticipated generation or load within a program time unit (PTU)8. In current markets, generators, consumers and electricity traders are usually represented by balance responsible parties (BRPs), which allow the transmission system operator (TSO) in a control area to keep the power balance. While the TSO is responsible for maintaining the power balance on a momentary base, the BRPs have the compulsion or the target to balance their position over a PTU. Therefore, a BRP is either penalized or rewarded based on its behavior over a period of time instead of on its momentary behavior [37]. However, different market areas in Europe use different PTU lengths. Due to these differences in PTU lengths, most cross-border trades in the European grid are conducted with hourly products [38]. Balance Service Provider (BSP): The TSO guarantees the provision of reserves that will allow for imbalances in the system. A balance service party is the reserve provider for each markettrading period and manages any contingent events that cause the imbalance. BSPs are instructed by the TSO, when and how much electricity to generate. Any BRP could be contracted by the TSO to as BSP. The actors of Dutch electricity market are listed in Appendix E. However, the price of electricity in a market is defined by demand and supply curves [26], [29] (see Appendix D). In a conventional top down system generation follows the demand to balance the system. But the introduction of intermittent renewable energy sources, forcing deviation from this approach due to their inability to follow the load. So, the bottom up system, where the loads are also contributing to balance the system, is coming into play. As depicted in Figure 2-10, both generation and load determine the performance of power balancing.
In Netherlands DSOs are not involved in the well functioning of retail markets. In fact there is no DSO, instead Netherlands has Distribution Network Operators, DNO. 8 As energy is a merchantable product while power is not, electricity trades are always in terms of energy per fixed time interval or a multiple of this. This unit of time is called the programme time unit (PTU)[38]. 7
32
Figure 2-10 Organization of electricity sector
2.3.2 Market Architecture The introduction of competition needs the design of specific market architecture(s) or market design(s) [39]. A market's architecture is a map of its component "submarkets". This map includes the type of each market and the linkages between them. The submarkets of a power market include the wholesale spot market, wholesale forward markets, and markets for ancillary services. "Market type" classifies markets as, for example, bilateral, private exchange, or pool [26]. Figure 2-11 shows different stages of market architecture. The long-term energy market utilizes a bilateral forward market that trades individualized forward contracts and centralized futures exchanges (or future markets) that trade standardized futures contracts. In bilateral markets buyers and sellers trade directly with each other and know who purchase their service. Such markets are extremely flexible as the trading parties can specify any contract terms they desire, but this flexibility comes at a price. Negotiating and writing contracts is expensive. Assessing the creditworthiness of one's counter party is also expensive and risky [26]. Nonetheless, an exchange provides security for traders by acting as the counter party to all trades, eliminating trader’s concerns over creditworthiness. Exchanges utilize auctions and are sometimes called auction markets where it is very easy to find a best price. An exchange has a number of advantages over a bilateral market. It can reduce trading costs, increase competition, and produce a publically observable price though it cannot provide the flexibility as forward contracts. In the short-run, forward energy can be traded in the day-ahead market (functioning 24 h before delivery) and in intraday markets (functioning within the day). Ancillary services (support the transmission of energy from resources to
loads, while maintaining a reliable
33 operation of the power system [40]) includes real-time power balancing markets which balance demand with supply in and near real time.
Figure 2-11 Electricity Market Architecture
The most liquid power market is the day-ahead market. As the traded energy correspond to specific time period or programme time unit (PTU), market signals given through this market allow to value electricity differently according to the delivery times during the day. Intraday markets allow trading electricity from day-ahead market closure to few hours before delivery [41]. The degree of centralization of day-ahead and intraday markets is the key element to distinguish different market designs[39]. These markets can be organized from a centralized auction (e.g. Spain) to a completely decentralized bilateral market (e.g. UK). Centralize market can optimize generation power scheduling by incorporating uncertainties and inter temporal and network constraints [42]. The temporal position of the “gate closure” is a very significant design parameter for intraday markets. For each PTU a BRP should mention its net trade to the TSO. If in real-time the actual exchange with other parties varies from the beforehand stated value, the BRP will be settled for its imbalances and is penalized because the TSO will deploy reserve energy from BSPs. At the Gate Closure Time (GCT), all parties should notify their TSO about their expected physical exchanges and options for bilateral trades cease. Predictions for load and generation tend to be more accurate as the prediction horizon decreases. Therefore, a delay of the GCT could be an effective means for countering the increased uncertainty and fluctuation on the supply side due to stochastic (renewable) generators such as wind generation [38][42][43]. When market participants are able to trade electricity near real-time, it indicates that better acquaintance concerning the actual generation is available to all the market participants [44]. Figure 2-12 shows position of gate closure with respect to day ahead and intraday market.
34
Figure 2-12 Position of gate closure[42]
The system operator manages the balancing market. It balances the system using balancing offers/bids and reserves. On the other hand, the system operator computes imbalances (measuring the actual injections/withdrawals of energy in real-time and comparing it with forward contract positions) and settles them using imbalance prices [42]. One of the most vital design parameter of balancing markets is the definition of imbalance prices. The distribution of balancing cost and incentives given to the market participants are determined based on the imbalance price. There are basically two types of designs: a dual-price design and a single price design [45]. Dual imbalances prices are usually computed using average prices and artificial penalties added to imbalance costs. Single price design is reputed to give more volatile signals since imbalance prices are computed using the proposed price of marginal offers or bids and this can change for each settlement period. Dual-price settlement system is used in several European countries [42]. However, many countries are improving market design in order to get balancing signal right [45].
2.4
Wind Power Forecasting
Wind power integration comes with various issues. One of these issues, perhaps the most important, is the uncertainty and variability associated with it. This is a concern because, in a power system, the total amount of electricity that is provided at each instant has to match a varying load from the electricity consumer. To achieve this in a cost effective way, power plants must be scheduled in advance according to increasing marginal operating costs [46]. Therefore, wind power forecasting is essential for effective electricity market design and power system operation. Moreover, Effective forecasting is also important for trading in the long-term and short-term market with minimum imbalance cost [47]. Thus, improvement of the performance of wind power forecasting tool has significant economic and technical impact as well on the system by increasing wind power penetration. Studies have been devoted to the improvements of wind forecasting techniques by number of researchers with wide experience in the field. For example, in [46], [48], [49] different wind power forecasting models have been reviewed. However, the forecasting of generated wind power is directly performed or it is calculated from the forecasted wind speed applying appropriate transformations. Wind forecasting systems have been classified according to time horizons or methodology. In [49] four categories of forecasting system are mentioned according to the time horizons: very short term, short term, medium term and long term. The prediction horizon is not strictly defined and may vary depending on the application of forecasting model [46], [50], [51]. According to the methodology and principle used for forecasting, the forecasting systems are classified into two categories- physical approach and statistical approach [46], [49]. Physical
35 approach is based on lower atmosphere or numerical weather prediction (NWP)9 using weather forecast data like temperature, pressure, surface roughness and obstacles. In general, wind speed obtained from the local meteorological service and transformed to the wind turbines at the wind farm is converted to wind power [52]. The statistical methods do not consider meteorological conditions, rather are based on vast amount of historical data. For example, the artificial intelligence approach belongs to statistical approach. The concept of artificial intelligence approach is to establish the relationship between input and output by artificial intelligence methods, instead of analytical method. The model, which is described in this way, is usually non-linear model. Many artificial intelligence methods are more excellent than the conventional methods and have good development prospect [48]. The combination of both approaches, a hybrid method is also used many researchers [46]. 2.4.1 Evaluation criteria of Wind Power Forecasting It is difficult to precisely forecast wind power; it has a character of the inherent uncertainty. Hence, it is crucial that wind power forecasting is accurately evaluated. It is very important to evaluate the prediction error measures on data that have not been used to build the prediction model. The evaluation methods on the uncertainty of wind power forecasting are listed as follows [53][54]: The Mean Error (ME): ̅̅̅
∑
(2.10)
Where, ̂ is the measured power at time t+k; ̂ is the power forecast for time t+k at time t; N is the number of prediction errors used for method evaluation. The Mean Square Error (MSE): ̅̅̅̅̅
∑
(2.11)
The Root Mean Square Error (RMSE): √
√
∑
Both systematic and random errors have impact on RMSE value.
9http://www.ncdc.noaa.gov/data-access/model-data/model-datasets/numerical-weather-
prediction
(2.12)
36 The Mean Absolute Error (MAE): (2.13)
∑ MAE value is also affected by both systematic and random errors. The standard Deviation of the error (SDE): √∑
̅ ]
[
(2.14)
SDE is only affected by random errors. The Mean Absolute Percentage Error (MAPE): ∑
∑
̂
(2.15)
It is crucial to establish a more precise and comprehensive error evaluation system. Different evaluation methods have different effects subject to the characteristic of wind power forecasting system. For example, the RMSE is more sensitive to the existence of erroneous data when compared to the MAE. Hence, MAE should be preferred as a main evaluation criterion, if there is doubt about the quality of the evaluation set, because it offers better robustness when tackled with large prediction errors. Existing wind power prediction methods (see Appendix C) are improving and the prediction accuracy continues to increase [49]. Even though, emphasized attention is required in further development of artificial intelligence, combining physical and statistical models, practical application of the models and universal error evaluation system.
2.5
Summary and Conclusion
This chapter consists of the literature review of wind energy and electricity markets, which is used in subsequent chapters of this thesis. As wind power has variability and limited predictability constraints, its market value is not only dependent on short-term power markets energy prices but also on the location and size of wind turbines. The power-wind speed curve is investigated with respect to different parameters like turbulence intensity, roughness length and turbine types. The wind power forecasting is also discussed in this chapter to have the background knowledge even though quantitative wind speed or power forecasting is beyond the scope of this thesis.
37
3 The Market Value Tool 3.1
Introduction
Following the EU Renewables Directive and the 20-20-20 targets [55], on one hand, wind power capacity in Europe has entered in a new large-scale development phase: this is not a marginal technology anymore [42]. On the other hand, the market integration of renewable generation is a challenge. A generation portfolio of merely wind energy cannot compete in current liberalized power markets because of the stochastic behavior of wind power generation. Moreover, the cost of wind power is higher compared to the conventional fossil fuel sources. Meanwhile, the integration of new market arrangements improves the market competitiveness of wind energy. Nowadays, energy can be traded more close to real-time and across borders. This enables wind farm operators to forecast more accurately their generation schedules. Trading across borders and closer to real time could lead to limited price volatility and provide more solid value to electricity. Therefore, in order to assess the actual value of wind generation in today’s electricity markets, a tool is developed to calculate the market value of wind energy within the framework of EIT KIC INNOENERGY Offwindtech. The ‘Market Value’ tool assesses the market value of wind power, which includes the revenue, imbalance cost and generated energy. This tool can be used, by providing a certain portfolio of wind turbines or wind farms and the wind characteristics of that location. Furthermore, the user selects in which power market the wind energy should be integrated. It enables the user to optimize conditions of the market for the study, enabling thus the analysis of many different scenarios. Therefore, a wide range of possibilities can be dissected. This tool enables to perform simplified research on the market value of wind power in general, or to perform a more precise, detailed analysis. This chapter elaborates the algorithm and the applications of ‘Market Value’ tool. A case study using this tool is presented in chapter 4 and chapter 5. A preliminary graphical user interface is given in Appendix F.
3.2
Algorithm
The tool is developed in MATLAB [56] environment. The inputs and outputs of the tool are schematically depicted in Figure 3-1.
38
(a)
(b)
Figure 3-1 Inputs and Outputs of Market Value tool
3.2.1 Wind speed from sensor height to hub height Wind speeds are measured with sensors mounted at weather stations at random heights, in general not equal to the hub height of wind turbines. Therefore, wind data needs to be up or downscaled to specific hub heights in relation to the desired wind turbine. Consequently, the wind data will be suitable for further calculation steps. Two laws are used to calculate the wind shear10- log law and power law. These theories are discussed in section 2.2.2. Based on statistical analyses of prediction errors, these two laws have trivial differences between them. Even though, power law methods sometimes show poor performance compared to log laws [9]. Log laws, demonstrated in equation 2.4, have been used in the ‘Market Value’ tool to estimate the raw wind speed data at rotor hub height. Particular care must be given in defining the appropriate roughness length, which depends on the terrain conditions. 3.2.2 Wind speed to wind power In order to transform the wind speed data set into a wind power data set, the power-wind speed characteristic of the used wind turbine needs to be known. The power-wind speed curve of a particular wind turbine is mostly documented on public accessible specification sheets. The exemplarily wind turbine of Figure 3-3 is a variable speed, pitch controlled 5 MW wind turbine. Due to the time horizon of hourly time steps in the “market value” tool, certain wind (turbine) singularities can be snubbed. Therefore, the rotor disc averaging effect and certain delays due to the large inertia of wind turbines do not need to be contemplated. The only translation to be made is the fact that large wind farms cannot be simulated based on one single local wind speed measurement. A power-wind speed curve of one single turbine is not sufficient. Due to the spatial distribution of the discrete wind turbine units (the distances between the units) in
The variation of wind speed with height above the ground level in practice is generally defined as wind shear 10
39 combination with the stochastic nature of the wind speed, the power outputs from the different units within the area are not essentially the same at the same time. The normalized standard deviation (relative to mean wind speed) of the distribution of the wind speeds at the individual wind turbine units at any specified time is a function of the dimension of the area and wind turbulence intensity as portrayed in Figure 3-2. Here, ‘I’ denotes the turbulence intensity.
Figure 3-2 Normalized standard deviation of the wind distribution at any given time[25]
Therefore a multi-turbine power curve is determined to represent the aggregation of all individual turbines within the area. Equation 3.1 is used to calculate aggregated multi-turbine power curve [25]. Where, is the jth element of single-turbine power curve and is the probability of spatial distribution11. ∑
(3.1)
This spatial distribution is calculated by identifying the standard deviation from Figure 3-2. The detail approach is described to calculate a multi-turbine power curve depending on the type of wind turbine and the specific wind site condition in [25]. This aggregated power curve takes charge for the cut-in and cut-off smoothing behavior of wind turbine aggregation as can be seen in Figure 3-3. Higher standard deviation results in smoother curves, which is depicted in Figure 3-3. However, the estimated annual energy production for a given wind speed distribution based on these curves should be equal. The aggregated power-wind speed curve can now represent fairly a wind farm.
(
11
√
)
Where, σ, μ are the standard deviation and mean respectively.
40
Figure 3-3 Aggregated Power Curve
This aggregated power curve is used to calculate the time series of the aggregated power generation in combination with the wind speed time series (with the same resolution as the original wind speed time series). These calculation steps are schematically illustrated in Figure 3-4. Nevertheless, users of the market value tool do not need to be familiar with these specifications and translations. Firstly, these variables are quantified in fact sheets, or in the manual of the provided data. Secondly, user of the tools can choose not to detail these phenomena and stick to general assumptions the tool provides, if requested.
Figure 3-4 Time series energy calculation
3.2.3 Wind power and market price to market value Electricity market prices are an input to the tool, which is used to calculate time series revenue by multiplying it with the time series power generation. The revenue of the per unit energy generation is calculated by dividing the sum of the revenue by the sum of energy generated during project time. The calculations are depicted in equation 3.2.
41
Revenue of the wind farm (Euro/h) = Market Price (Euro/MWh)*Energy Generation (MWh/h)
(3.2a)
Net Value or Yield of the wind farm (Euro/h)= Revenue (Euro/h) - Imbalance Cost (Euro/h)
(3.2b)
Estimated Subsidy= Cost of the Wind farm (Euro/MWh) - Value of the wind Farm (Euro/MWh)
(3.2c)
The deviation from the beforehand projected/contracted power due to unpredictability constraints, result in imbalance costs for the balance responsible party. Based on imbalance prices, imbalance costs are calculated. The subtraction of imbalance costs from the revenue provides the net revenue or value of the farm. Required subsidy is estimated by comparing the unit value and cost of the energy. Figure 3-5 depicted the steps schematically. However, the per unit cost of energy also depends on the interest rate, capacity factor etc. which are not considered here.
Figure 3-5 Market value calculation
3.3
Application of the tool
The flexibility of the market value tool enables to perform simulations with a variety of objectives. This tool provides the provision to compare the market value with respect to different turbine types, markets and wind regime. Various parties can benefit from the tool and conduct their own specific research. Figure 3-1(b) depicts the outputs of the market value tool. Possible target group of users of this tool are: Policy makers Regulators Market parties Research/Education Investors Industry
42 For instance, outputs of the tool help policy makers to optimize support schemes and estimate support needs. From the time series wind generation and revenue, wind farm operators determine when to cost efficiently schedule planned maintenance. For example, by utilizing figures like Figure 3-6, farm operators can find out the periods with comparatively small revenue to plan maintenance. For offshore wind farms, weather conditions are also a measure to consider maintenance. Wind power investors can optimally locate their planned wind farm to define appropriate business cases comparing the market value of different locations. Market parties use the tool to assess the cost-efficient maximization of wind power integration into power markets. A case study, investigated in chapter 4 and chapter 5 of this thesis is an appropriate illustration of this application, where wind farms are compared based on the specific sites and markets. Wind farm owners can compare different turbine types to select the optimum one for any site. Furthermore, the effect of wind generation conditions such as forecast techniques can be compared to assess the actual market value. Research institutes can conduct a diversity of research topics. Apart from the autonomous use of the market value tool, it can be used in combination with the other three related Offwindtech tools. There is also possibility to compare the value of wind power generation with other generation technologies like solar generation. The attractiveness between diverse technologies (wind, solar, hydro) within the roof of renewable generation can be determined.
Figure 3-6 Exemplarily depiction of annual profile of wind farm revenue in Euros
3.4
Summary and Conclusion
The ‘Market Value’ tool is developed as a part of the EIT KIC Innoenergy Offwindtech framework. This tool uses the time series wind speed and electricity market prices to calculate the time series revenues of any specific wind farm. It can be used for any location and any electricity market. It allows the user to optimize conditions of the market for the study, enabling thus the analysis of many different situations. Hence, a broad array of possibilities can be analyzed. This tool enables to perform simplified research on the market value of wind power in general, or to perform a more precise, detailed analysis. This tool has also the potential to directly compare wind power generation market performance with other renewable sources, e.g. solar energy. A case study is discussed in the following chapter, which elaborates on the applications of the tool.
43
4 Case Study 4.1
Introduction
In this chapter a case study is introduced to assess the market value of wind power located in the Netherlands for a reference year 2011 by using the ‘Market Value’ tool. The assessment is performed for qualitative comparison of different wind sites and market scenarios. The geographical distinction of wind sites is made between offshore, onshore close to shore (coastal), and inland onshore wind power generation. These three locations differ based on the combination of wind regime and costs of wind farm installation and maintenance. The generated power is sold in the Dutch power market, the energy exchange platform of APXEndex, for operating spot and futures markets for electricity. The imbalance costs due to predictability constraints of wind power are also taken into account.
4.2
Scenarios
Three different wind sites have been chosen based on their characteristic wind site experience. Offshore wind power generation represents high wind speeds. However, its installation and maintenance costs are significantly larger compared to onshore wind power generation. Nevertheless, onshore wind sites experience smaller energy yield. The third location is onshore close to shore (or coastal) with wind regime of approximately offshore but lower investment and maintenance cost (explained in section 4.2.1). The generated power is traded in the Dutch power market, the energy exchange platform of APX-Endex, where two scenarios are assessed, selling to the day ahead market and close to real time, the intraday market. Eventually, the distinction is made between 6 different scenarios, which is depicted in Figure 4-1.
Figure 4-1 Case Study Scenarios
4.2.1 Geographical Locations The geographical distinction of the three wind sites is based on the combination of wind regime and costs of wind farm installation. In the Netherlands, wind flows on annual basis mainly directed from southwest towards northeast [57]. Figure 4-2 shows the combined wind rose [58] for two places of Netherlands (one is northern12 and another is eastern13) based on ten years
12
Twenthe is a place in eastern Netherlands.
44 hourly wind data. This rose supports the statement about the trend of typical wind direction of Netherlands. The wind rose illustrates that, at least 40% of the time, wind is directed from south-west and in between. Moreover, as indicted by colors in the legend, high wind speeds above 12 meter per second are predominantly directed from this direction. Even though, a significant time of the year wind is also directed from east, it will have a large impact on the yearly energy generation of wind power. This is caused by the absolute small wind speeds, between 2 and 4 meter per second. Due to this phenomenon, it can be stated that the west coast of the Netherlands experiences almost the same wind condition as offshore. Therefore, this coastal area is supplied by valuable wind, unlike e.g. the north coast of the Netherlands, where the wind covered already a long distance over land.
Figure 4-2 Typical wind direction of Netherlands
The onshore site for the study is selected from the eastern part of Netherlands. The offshore site is in North Sea, few kilometers away from The Hague. A place close to Ijmuiden, on coast of North Sea is considered as the coastal site. The geographical locations of the sites are shown in Table 4-1and Figure 4-3. Table 4-1 Coordinates of the wind sites
Latitude Onshore Site Offshore Site Coastal Site
0
51.93 N 53.30 N 51.90 N
Longitude 6.590 E 3.390 E 4.140 E
Europlatform is an offshore structure used for weather data measurement. Wind data of both locations are taken from [88] 13
45
Figure 4-3 Geographical locations of three wind sites under consideration
4.3
Simulation
MATLAB version: R2011b (7.13.0.564), 64 bit (maci64) The wind sites selected for this case study are based on their wind regime and available wind speed data. However, site-specific parameters, like roughness length and turbulence intensity (to estimate standard deviation) are not provided however are required to be used in the ‘Market Value’ tool. Therefore, the approximated values for these parameters are considered supported by literature. Roughness lengths for different locations are estimated based on the Table 2-1. The onshore location is considered as plane field with few trees. Therefore, the related roughness length is determined to be 0.01 meter. Unlike the onshore site, the roughness length for the coastal site is considered a bit higher (0.03 meter), mainly because of small buildings and uneven terrain. The offshore site is considered as a calm sea, which suggests the roughness length of 0.0002 meter. It is also kept in consideration that deviation from actual roughness length of a particular location has effect on the yearly energy yield, since higher roughness factor (from the actual value) estimates higher wind speed. For the sake of comparisons, the standard deviation of wind speed in space of each location is considered of the same value. However, it is possible to estimate the standard deviation with the knowledge of turbulence intensity, average wind speed and area of the wind farm, using Figure 3-2.
46 It is important to define/estimate the investment costs and operation and maintenance costs to calculate the unit cost of wind energy. Investment costs depend on several parameters like distance to shore (for offshore sites), nominal power of each wind turbine, capacity factor, project life, distance between wind turbines, type of seabed (for offshore), inflation, interest rate, etc., which are very much specific for individual wind farm and location. The ‘Cost Analysis’ tool, another tool within the Offwindtech framework explicitly dealt with the cost of the wind farm. A well illustrated cost calculator for different technology and investment year to compare the investment cost is found in [59]. The levelised costs of electricity in this case study is assumed 60 euro/MWh, 80 euro/MWh and 48 euro/MWh for onshore, offshore and coastal location respectively. These assumptions are made based on the knowledge from [60], [59] and [61], which are very much generalized as the intention of this case study is to qualitatively assess of the value of wind energy for the different scenarios. 4.3.1 Market Prices Day ahead and intra day market prices are made available by APX Endex [62]. Figure 4-4 represents the annual profile of Dutch day ahead market prices of 2011. Day and night patterns as well as seasonal patterns are clearly perceptible in the graph and some peaks are also there. There are few points with extremely high values of price, which is not shown in this graph. These moments are caused by a significant mismatch between available generation and system load, e.g. a loss of a large generator with low marginal costs14.
Figure 4-4 Annual profile of day ahead market price
The data set of Dutch intra day market prices of 2011, made available by APX-Endex is incomplete. Among the 8760 hours of the year, 3041 random hours are available with intra day price. In order to use the data in an appropriate way, an assessment is performed to verify whether or not this incomplete data set can represent a complete year. The assessment shows that the incomplete data can be used, as is more elaborated in section 5.2. Intra-day market 14
The cost required to unit increment of power of a generating unit.
47 prices are commonly larger compared to day ahead market prices, however this difference is relatively small. The profile of day ahead and intra-day market prices is strongly correlated. 4.3.2 Wind Power Generation The yearly profile of wind power generation is calculated with the market value tool. Figure 4-5 depicts a yearly profile of wind power generation at an onshore site in 2011. The random behavior of wind power is perceived in this figure, a typical condition of wind energy. The related power generation for all scenarios in this work is calculated from 24 hours ahead forecasted, 2 hours ahead forecasted and real wind speed, further elaborated in the next chapter.
Figure 4-5 Yearly profile of wind power generation in the onshore site
4.3.3 Wind Speed Forecasting The actual wind speed of 2011 of the designated locations along the forecasted wind speed is made available based on the work of [63]. As expected, the wind forecast techniques are more accurate when the forecast is performed closer to real time. This is depicted in Figure 4-6, where from day ahead (D-24h) up to one hour ahead (D-1h) the duration curves of forecast errors are exhibited.
48
Figure 4-6 Duration curve of forecast error for 24 hours up to 1 hour ahead forecasting
In extreme conditions, the forecast error is twice as large for 24 hours ahead forecasting compared to 2 hours ahead forecasting. There is no significant variation in forecast errors between one hour to four hours ahead forecasting. This observation is a typical behavior of wind variations as can be concluded from the “van der Hoven Spectrum”, which is discussed in section 2.2.3. Figure 4-7 shows how significant these variations are.
Figure 4-7 Close view of the variations of forecast error
4.3.4 Power Imbalances Imbalance prices are acquired from the Dutch Transmission System Operator TenneT TSO B.V. [64]. The available imbalance price is subdivided into two parts. There is a separate price for upward regulation (when the system is short) and one for downward regulation (when the system is long). Due to the (conventional) fuel based generation portfolio in the Netherlands, upward regulation involves larger costs, compared to downward regulation, which occasionally has negative prices.
49 Negative imbalances occur when the generated power is below the earlier declared/forecasted value, or system load turns out to be larger as declared/forecasted. In these situations, the TSO needs to send a signal to reserve generation for upward regulation. Imbalance costs for negative imbalances are (based on historical data) always positive (BRP has to pay), as it requires additional fuel to ramp up (or even start a new generator) the reserve generations. Positive imbalances occur when the generated power is above the earlier declared/forecasted value, or system load turns out be smaller as declared/forecasted. In these situations, the imbalance cost could be either positive or negative. If there is adequate generation in the system, the BRP, responsible for the surplus of generation, may have to pay to another party to consume the surplus of energy (consequently positive imbalance costs). However, negative imbalance costs (BRP is ‘rewarded’) are also possible if there is the possibility to sell the surplus of energy15. As the PTU of the Netherlands is 15 minutes [38], the available imbalance costs are settled for every 15 minutes. To match with the other hourly data set, for the sake of calculation, the imbalance price data set is converted into hourly basis. The power imbalances are calculated by taking the differences from the real wind power generation from the forecasted values. Positive and negative imbalances are identified to compute the imbalance cost from the analogous imbalance price. Figure 4-8 illustrates the annual profile of wind power imbalances based on day ahead forecasting for the onshore site. The variation of wind power imbalances is very random without perceptible pattern as depicted in Figure 4-8.
Figure 4-8 Annual profile of wind power Imbalances with day ahead forecasting (onshore wind site)
4.4
Results
Based on the algorithm presented in section 3.2, revenues, imbalance costs and market values of the six scenarios of the case study are calculated which is shown in Table 4-2. The calculated revenue of wind generation in this case study is relatively high, compared to values from literature. This is caused due to theoretical approach of perfect circumstances, not considering 15
May be due to the failure of other generation
50 imperfections due to roughness levels and turbulence of the wind site, and maintenance and failures of wind turbines, etc. The revenue of wind generation traded in intraday market is, in relative terms, slightly higher than the value it could obtain from the day ahead market. This fact is due to the somewhat higher value of intra day market price compared to the day ahead price. The revenue per MWh wind generation is almost the same for all the locations. However, the yearly gross revenue in euro is approximately 20% higher for the offshore and coastal location compare to the onshore location, due to the wind intensive site, which is shown in Table 4-3. Table 4-2 Results [€/MWh]
Onshore Wind Farm Day Ahead Intra-day Market Market Costs [€/MWh] Revenue [€/MWh] Imbalance cost [€/MWh] Value [€/MWh]
Coastal Wind Farm Day Ahead Intra-day Market Market
Offshore Wind Farm Day Ahead Intra-day Market Market
60
60
48
48
80
80
54.57
58.46
54.57
58.16
54.49
57.74
12.29
4.85
10.20
3.95
9.56
3.64
42.28
53.61
44.57
54.20
45.00
54.09
The data analysis reveals that the forecast of onshore wind power generation is more optimistic compared to the forecast of offshore and coastal wind generation. This will result in more negative imbalances for onshore wind generation, which commonly come along with larger imbalance costs. This effect has the largest impact at the day ahead markets, where at intra-day markets, closer to real time, this effect attenuates. The implementation of more accurate forecasting limits the expense of total imbalance costs. As depicted in Table 4-2, the imbalance costs of wind generation are reduced significantly as one may expect when changing from day ahead closer to real time, towards the intra-day stage. At the intra-day stage, wind will not vary significantly anymore within 4 hours to 10 minutes, as can be concluded from the “van der Hoven Spectrum” in Figure 2-5. However, the imbalance costs are relatively higher for onshore location compared to the other two sites. This is based on the power–wind speed curve of wind turbines. At moments of larger wind speeds, wind turbines operate at rated power, where variations in wind speed (forecast inaccuracy) do not lead to power deviations. As depicted in Figure 5-2, Figure 5-3 and Figure 5-4, the probability of higher wind speeds is larger at offshore and coastal wind sites than the onshore site. Therefore, imperfections in wind forecast have a smaller impact for offshore and coastal wind generations, and consequently wind imbalance costs are smaller compared to onshore wind generation. Onshore wind generation does not have this offsetting of power imbalances. Table 4-3 shows the gross values of revenue and market value if the generated wind energy is traded in the day ahead market in 2011 since only for day ahead market complete data is available. The results in Table 4-2 and Table 4-3 support the general expectation stated earlier in
51 this chapter that the wind farm located at the east coast of Netherlands has approximately the wind regime as an offshore location. Even though the offshore wind farm in this case study has slightly better market value because of smaller imbalance costs. However, the installation costs are predominately high for offshore wind farm. Therefore, onshore wind generation close to shore (coastal) is most cost-efficient, to be sold at the intra-day market. Table 4-3 Gross values if traded in day ahead market in 2011
Energy Generated in 2011 [GWh] Day ahead Revenue in 2011 [M€] Day ahead Value in 2011 [M€]
4.5
Onshore Wind Farm
Coastal Wind Farm
Offshore Wind Farm
7.0400
9.2176
9.4116
0.3588
0.4796
0.4880
0.2790
0.3794
0.4376
Summary and Conclusion
The application of the market value tool presented in chapter 3 is elaborated in this chapter with the help of a case study in Netherlands. Three wind regime- and two electricity marketconditions are considered in this case study, which provides a total of six scenarios to be investigated. The case study is based on the available data of these locations and electricity markets. Some parameters like roughness length, standard deviations are assumed based on literature, though these parameters are very much location specific. The results obtained from the case study show that the wind energy generated by the coastal wind farm, traded in the intra day market is the most cost effective option.
52
53
5 Validation of Results 5.1
Introduction
In this chapter, a validation-assessment is executed to determine the representativeness of the incomplete data with respect to a complete year. This is performed for wind power generation, wind power imbalances and market prices of the whole year. Furthermore, several ways to reduce the imbalances (eventually imbalance costs) in wind power generation is also discussed in this chapter. The available data set of the Dutch intraday market prices used for the case study is incomplete with random 3041 hours instead of total 8760 hours of the year. As mentioned in chapter 4, the idea of this case study is to assess whether or not the incomplete data set can be used with representing results. Therefore, all the other available data sets (day ahead market price, wind power generation) are also assessed with 3041 hours corresponding to the available intraday prices and compared with a complete data set. Significant changes in average values would imply that with the incomplete data set, no veracious conclusions can be drawn. Figure 5-1 shows the available hours of intra day market prices. It is of impact not considering and verifying the values of the incomplete data set, only to assume it can represent a complete data set. This analysis shows how it would affect the qualitative analysis of the case study results.
Figure 5-1 Intra day market price of 3041 random hours of Dutch power market in 2011
5.2
Verification of Data
Wind speed data sets, day ahead market price data set, and power imbalances of the analogous 3041 hours with available intra day market price are compared with the data sets of all 8760 hours of the year, 2011. The imbalance cost of those 3041 hours is also scrutinized to define the representativeness with respect to the complete year.
54 5.2.1 Wind speeds For this case study, nine sets of wind speed data of 8760 hours are used (three for each location). As mentioned before, for the sake of comparison only the respectively 3041 hours of each data set are considered. Each incomplete wind speed data set is compared with the respective complete data set to assess the representativeness of it. Frequency distribution of each 18 (9 complete and 9 incomplete) wind speed data sets are obtained where shape factors are assumed based on the variance of the wind speed of each data set and scale factors and frequency distributions are calculated using Equation 2.7 and 2.9 respectively. Frequency distributions of both complete and incomplete data sets for onshore wind farm are almost similar while the average wind speeds differ by very small margin (shown in Table 5-1). It is to mention that the annual energy generation largely depends on the frequency distribution of wind speed rather than the mean wind speed [17]. Weibull distributions of complete and incomplete data sets of real wind speed in onshore site are presented in Figure 5-2. Table 5-1 Average wind speed (m/s)
Onshore Site 24 hours ahead forecasted wind speed 2 hours ahead forecasted wind speed Real wind speed Offshore site 24 hours ahead forecasted wind speed 2 hours ahead forecasted wind speed Real wind speed Coastal Site 24 hours ahead forecasted wind speed 2 hours ahead forecasted wind speed Real wind speed
8760 hours data
3041 hours data
8.45 8.08 8.12
8.39 8.18 8.20
10.00 10.28 10.32
10.16 10.22 10.26
10.25 10.42 10.44
10.45 10.48 12.51
Figure 5-2 Weibull Distribution of real wind speed in onshore site
55 In case of both offshore and coastal sites the, the frequency distributions of complete and incomplete wind speed data sets are very similar though incomplete data sets has slightly higher scale factors. The Weibull distributions of offshore and coastal day ahead forecasted wind speeds are depicted in Figure 5-3 and Figure 5-4 respectively. Weibull distributions of other wind speed data sets are given in Appendix G. It is worthwhile to mention that for all the wind speed data sets, the maximum frequency of wind speed is between the cut-in and rated wind speed. Therefore, it can be assumed that the incomplete data for all the wind sites in this case study roughly represent the wind data of whole year of 2011 for all three locations.
Figure 5-3 Weibull Distribution of day ahead forecasted wind speed in offshore site
Figure 5-4 Weibull Distribution of day ahead forecasted wind speed in coastal site
56 5.2.2 Market Prices The day ahead market prices made available by APX Endex have seasonal and daily variations. As depicted in Figure 5-5, market prices are higher during the noon and evening while July is the month with minimum price16 compared to other months of the year. Figure 5-6 shows the 3041 hours of the year, which are used for the calculations. Most of the hours situated within the second half of the year. Therefore, the incomplete data set of market prices is bit optimistic than the complete year data.
Figure 5-5 Day ahead market price of Dutch power market in 2011
This observation also supported by the average market price of the data sets. The average market price is 52.03 euro/MWh for a complete year while the value is 54.23 euro/MWh for the considered 3041 hours of the incomplete data set. This implies that the calculated values of yearly revenue per MWh would be slightly lower if the whole year is considered.
16
May be due to no heating demand
57
Figure 5-6 3041 hours data of day ahead market price
5.2.3 Power Imbalances Predictability constraints of wind speeds cause deviations of output power from the before mentioned values which lead to power imbalances. Figure 5-7 depicts the annual profile of wind power imbalances at the onshore location while the wind speed is forecasted 24 hours ahead. Power imbalances of the 3041 hours of the year corresponding to the available intra day market price, are depicted in Figure 5-8. As the selected hours mainly located in second half of the year where the power imbalances are relatively high. However, a fair distribution of the negative and positive imbalance hours keeps the average imbalance very close to the average imbalance of the whole year.
58
Figure 5-7 Annual profile of power imbalances at onshore site for 24 hours ahead wind speed forecasting
Figure 5-8 Profile of 3041 hours of power imbalances at onshore site for 24 hours ahead wind speed forecasting
In the case of wind speed forecasting 2 hours ahead at onshore, the average hourly power imbalances of the whole year almost equal the selected 3041 hours average power imbalance like the 24 hours ahead forecasting. Figure 5-9 and Figure 5-10 show the power imbalances for 2 hours ahead forecasting at the onshore location. It is noticeable that the overall trend of power imbalances of day ahead wind forecasting is higher compared to 2 hours ahead wind forecasting.
59
Figure 5-9 Annual profile of power imbalances at onshore site for 2 hours ahead wind speed forecasting
Figure 5-10 Profile of 3041 hours of power imbalances at onshore site for 2 hours ahead wind speed forecasting
Hourly wind imbalance costs are calculated from wind power imbalances and imbalance prices. Figure 5-11 shows the hourly wind imbalance costs of 2011 at the coastal site with day ahead forecasting, where Figure 5-12 shows the selected 3041 hours. It is evident from the figures that hours with higher imbalance cost mostly situated in the 2nd half of the year and the selected 3041 hours includes the hours with higher imbalance costs. Therefore, the average hourly imbalance costs are higher for the selected 3041 hours data compared to the whole year average. Day ahead forecast imbalance costs at the coastal site is exemplarily taken, since the same conclusion can be drawn for all the other sets of imbalance costs.
60
Figure 5-11 Annual profile of imbalance costs for day ahead forecasting at the coastal site
Figure 5-12 Profile of 3041 hours of imbalance costs for day ahead forecasting at the coastal site
Hence, the calculated imbalance costs per MWh in this case study is slightly pessimistic from the wind farm operator point of view. However, the day ahead market price data set of 3041 hours is found to be only slightly optimistic compared to the whole year. Therefore, it is assumed in this work that the calculated market value per MWh in day ahead market is rather close to the value if it is calculated for whole year.
5.3
Recommendations to reduce imbalance costs
Usually, it is very difficult to predict the output of wind farms precisely. Therefore, wind farms always come with power imbalances. As a stochastic generator with little or no control over its generation, market prices and imbalance prices spread are the key factors in the value of imbalance costs. Therefore, the value of wind energy significantly decreases with the imbalance cost as described in earlier chapters.
61 In The Netherlands, wind power producers (WPP) are completely accountable for their imbalances [65], as any other electricity producer. Each WPP is either directed by a BRP, or it can assign its accountability to a third party. As considered in the case study, WPPs can manage their own imbalances by trading in different short-term markets, such as the intraday market. BRPs often use other strategies such as managing imbalances within their own generation portfolio supported by controllable thermal units or bilateral trading. Some BRPs have flexible generation like gas generators and diesel generators to support the market integration of wind power. Load shedding is also a technique that some BRPs consider to maintain power balance within their portfolio. Electrical energy storage systems are another prospect to consider handling wind power imbalance costs. However, the cost of storage systems is an issue to optimize with respect to the imbalance costs. Suitable forecasting techniques are required (to be adopted) for the purpose of accurate forecasting. Input parameters to the forecaster are optimized based on the correlation with the desired output of the forecaster. Different wind power forecasting methods used in literature are highlighted in Appendix C. Evaluation methods of forecaster are discussed in section 2.4.1. In the following sub-section prospective storage technologies suitable/feasible for wind energy are briefly described. 5.3.1 Energy Storage Technologies ‘A logical solution to variability of output from wind turbines is to provide a means to store excess energy and to return that energy to the grid when the wind turbine output drops off’ [66]. This is an availability enhancement tool and provides spinning reserve and a firm source of supply from the generation side viewpoint. New electricity storage technologies have begun to appear with the competency to afford very fast response to load fluctuations. The first response capability helps to react during the power imbalances by wind power generators (due to the forecast errors). If the cost of energy storage system is such that the added benefits resulting from its installation including increased energy sales (less spillage), increased capacity payments and decreased imbalance costs amount to more than its operating and repayment costs, then it becomes a viable option. Electrical energy storage devices with the ability to store large amounts of energy for several hours or more could provide the necessary flexibility for smoothing of wind power. In this way, the possibilities to reduce imbalance cost can be improved. There is a growing research interest in using energy storage to increase the value of wind energy in electricity markets [67], [68]. However, the impact of market mechanisms, network constraints and forecasting accuracy of wind power must be further investigated to fully determine the advantages and limitations of energy storage for this purpose. Energy storage systems typically considered for wind applications include: pumped hydro, flywheels, batteries, super capacitors, compressed air, superconducting magnetic energy storage systems (SMES) and hydrogen [69]. In [70], possible electrical energy storage systems are explicitly described and compared.
62 5.3.1.1
Compressed air energy storage (CAES)
When power is abundant, it is used to power a large air compressor, which pushes pressurized air into an underground geologic storage structure. Later, when power demand is high, the stored air is released back up to the surface, where it is heated and rushes through turbines to generate electricity. Compressed air energy storage plants can re-generate as much as 80 percent of the electricity they take in. Its advantages are robustness and flexible size. The disadvantages are the initial capital costs, slow response, and limitations in terms of location. To overcome these drawbacks, the construction of storage tanks has been proposed, particularly for the deployment of small-scale CAES [70]. The power rating can be as high as 400 MW, while the maximum discharge can be 100 hours. Considering these properties, CAES might be a prospective technology to be used with wind energy. CAES has large storage capability as pumped hydro but with less geographical restrictions and lower cost. The integration of CAES technology with wind energy from European perspective is discussed in [71]. 5.3.1.2 NaS battery This is a high-temperature battery, with a working temperature in the region of 300 °C. Its ability of quick reversibility between charging and discharging makes it a potential option to integrate with wind energy. Its advantages includes: efficient operation, low maintenance, long life, and good scale production potential. At present, the world's largest is installed in Japan with a rating of 34 MW and 220 MWh [70]. 5.3.1.3 Lead-acid battery This is a mature technology, especially with the experience gathered from decades of use in the vehicle industry. The disadvantages are the poor low-temperature performance, low durability and environmental concerns due to the use of lead. Search shows that it can store maximum 50 MW for few hours [70]. 5.3.1.4 Nickel cadmium (Ni–Cd) battery Nickel cadmium battery is a mature technology initially used for consumer appliances. The advantages of Ni-Cd batteries includes: robustness to deep discharges, a long life cycle, temperature tolerance, and a higher energy density than lead-acid. In spite of its advantages, this technology is banned in Europe for consumer appliances due to the use of highly toxic cadmium. Other drawbacks of Ni-Cd battery are the costs, the need for advanced monitoring during charge and discharge, and the periodic need to perform a complete cycle. However, the technology has been deployed at the Golden Valley Electric Association, Alaska, providing 27 MW for 15 min or 46 MW for 5 min to stabilize the local power grid in the event of sudden loss of generation [72]. 5.3.1.5 Super Capacitors Super capacitors provide an outstanding capacitance compared with normal capacitors by making use of their particular structure. The main advantages of super capacitors are-
63 exceptional efficiency, the performance at low temperatures, no need for maintenance, immunity to deep discharges, speed of response, and extreme durability. However, it comes with high cost and low energy density. Super capacitors are considered as short-term storage devices suitable for wind energy applications[73]. 5.3.1.6 Hydrogen storage system (fuel cell) The surplus of electricity can be used in electrolysis to generate hydrogen. Hydrogen storage has such a potential that it is one of the few potential solutions theoretically able to store energy for long periods, even between winter and summer. Finally, the hydrogen is supplied to a fuel cell, which then uses it to produce electricity. Several types of fuel cells exist, with different levels of efficiency and cost. However, the other options, to produce electricity from the hydrogen are combined cycle power plant or CHP plant. Hydrogen can be mixed with natural gas to use for firing. Hydrogen energy storage could become an economically feasible alternative to integrate with wind energy if cost and performance goals of hydrogen technology are obtained [68]. 5.3.1.7 Superconducting magnetic energy storage (SMES) Superconducting Magnetic Energy Storage (SMES) is a novel technology that stores electricity from the grid within the magnetic field of a coil comprised of superconducting wire with nearzero loss of energy. These coils do not degrade with usage or time, so durability and reliability depend only on the auxiliary equipment, such as power converters. High efficiency and low maintenance is the main feature of SMES. Other advantages of SMES includes durability and reliability, short response times and no self-discharge. Very high cost and the impact of the magnetic field are the disadvantages of the technology. The power rating of SMES can be up to 10 MW [70]. 5.3.1.8 Pumped Thermal Energy Storage (PTES) This is a new technology developed by Saipem17. Two gravel bed reservoirs are linked in a closed loop circuit. At the loading phase, electrical energy is used to compress the gravel from the lowpressure reservoir towards the high-pressure reservoir. The gas flowing out of the reservoir is compressed, its temperature rises to 800°C. The hot gas is sent into the high-pressure reservoir, where its sensible heat is stored in the gravel bed. Along the loading phase, the gravel bed of the low-pressure tank is progressively cooled down from basically +400°C to -70°C, while the gravel bed contained in the high-pressure tank is heated from the ambient to +800°C. During the discharge period, the gas flows in closed loop in the opposite direction. The cycle is similar to a closed loop gas turbine. Pumped thermal energy storage technology is explicitly explained in [74]. The round trip efficiency of the process is about 70%.
17
www.saipem.com
64 Cost, efficiency and energy density of storage technologies are the parameters to be considered for the selection of the appropriate storage technology to integrate with wind energy. However, the topology of the storage system (e.g. central storage or individual storage) can be another constraint to be contemplated. In case of individual storage system for every wind turbine the capacity required is rather small compared to the central storage system (for whole wind farm). For example, the maximum imbalances calculated for a wind turbine in the case study discussed in this thesis is close to 2 MW. Flow batteries e.g. Lead-acid, Ni-Cd, NaS, Li-ion are the suitable storage options for this purpose. However, NaS, Li-ion and Ni-Cd are still very expensive and NiCd has poor efficiency as well. Lead-acid battery with its moderate cost and good efficiency could be the optimized option for storage in case of single wind turbine. On the other hand the maximum imbalance for the whole wind farm is around 115 MW. Pumped hydroelectric energy storage, compressed air energy storage are the probable technology for their higher power ratings (for CAES up to 400 MW and for PHES up to 5000 MW). For the studied case study pumped hydro is out of scope because of the plane geography of Netherlands. And for it is also economically viable with its higher capital cost to use pumped hydro storage for 115 MW. However, detail investigations are required to identify appropriate and optimized storage technology from the financial and technical point of view. A detail list of power ratings, capital costs, efficiencies, energy densities of different storage technologies are given in Appendix K.
5.4
Summary and Conclusion
The incomplete data sets are assessed with respect to the complete data set to observe how the incomplete data sets affect the qualitative analysis of the results. The incomplete data set of day ahead market prices is slightly optimistic while the incomplete data set of imbalance costs is also optimistic. Therefore, it can be concluded that the calculated results with the incomplete data sets roughly represent the complete data sets to provide viable conclusions. This chapter has also focused on how to reduce the imbalance costs. The specific parameters that have significant effect on the accuracy of the wind power forecasting are highlighted. The storage technologies feasible for wind energy integration are briefly mentioned. Some storage technologies are theoretically very attractive to integrate with wind e.g. PTES but yet to prove practically. Prominent storage technology like pumped hydro is not discussed here because of highly dependence on geographical condition and high capital cost. Flow batteries (Lead-acid) may be the most promising storage technologies for short-term storage device for single wind turbine. For central storage systems for a wind farm, compress air energy system may be the suitable option.
65
6 Business Plan 6.1
Introduction
Wind energy installations totaled 285 GW globally by the end of 2012 and the IEA’s New Policies Scenario suggests that total capacity would reach 587 GW by 2020. Even some highly ambitious predictions like GWEO Advance scenario suggests that with right policy support wind power could reach as high as 1100 GW by 2020, supplying between 11.7-12.6% of global electricity. However, wind power comes with the predictability and variability constraints. The poor correlation between wind power generation and electricity demand makes it more difficult to integrate wind power in to the electricity market. Variability of wind power disables to sell on full capacity wind power on peak moments and forecast mismatch of wind induces imbalance costs. Therefore, the importance of evaluating different potential wind energy sites and market conditions is getting higher priority for the sake of intelligent investment. There is a lack of tools available in this area. Research performed in this area usually makes use of internal developed tools, which were specifically built for that certain case study. Furthermore, those tools are not published or available. The ‘Market Value’ tool developed in this thesis has the potential to meet the demand of the industry for evaluating a wind site from market value perspective. This tool provides the possibility to perform a detailed study of the market value of wind power. Here, the user has freedom to adapt to its own needs and requests and to perform analyses in his field. This tool could also be beneficial for array of group including researchers, industry, market parties, policy makers, etc. However, an appropriate market analysis and business plan needs to be investigated before developing the suitable commercial version of the tool. There is the option to use the ‘Market Value’ tool along with three other tools of the Offwindtech project. However, the market analysis and business plan described in this chapter considers the ‘Market Value’ tool as an autonomous product. It is very important to have an appropriate name for the product. For the time being the proposed commercial name of the tool is ‘SELECT WIND’.
6.2
Market analysis
Understanding the attractiveness and dynamics of the focus market is the main goal of a market analysis. Initially, in this market analysis section potential markets have been selected based on the services provided by the ‘SELECT WIND’. This analysis will help to identify which market and customer segment has the highest attractiveness to start the venture. 6.2.1 Macro market The broad, macro-level market assessment is important to the entrepreneur to investigate the worthiness or readiness of the market for that particular product or service. Thus, reaching a clear conclusion about market attractiveness is critical. But this macro-level assessment – done at the 30,000-foot level, so to speak – is only half the market domain story. It is essential aerial exploration and a good look at the road ahead, but for the full picture observers are needed on the ground.
66 As the wind energy industry booming with the time, there are potential out there for the tools like ‘SELECT WIND’. This sort of tools still is not frequently available in the market. Hence, there is attractiveness in the macro market for this type of services. 6.2.2 Micro market No matter how large and fast growing a market may be, entering it in the face of other competition is likely to be difficult. Most successful entrepreneurs, rather than targeting the entire market, identify a much smaller segment of customers within the overall market. For the product in consideration, the micro market will include the researchers and wind farm investors. 6.2.3 Porter five forces analysis Porter's five forces is founded on the Structure-Conduct-Performance paradigm in industrial organizational economics. It has been functional to a varied array of complications, from helping businesses become more profitable to helping governments stabilize industries. As depicted in Figure 6-1, Porter's five forces include - three forces from 'horizontal' competition: the threat of substitute products or services, the threat of established rivals, and the threat of new entrants; and two forces from 'vertical' competition: the bargaining power of suppliers and the bargaining power of customers[75].
Figure 6-1 Porter's five forces
Bargaining power of suppliers: The suppliers for this business will be the software providers, computer providers. These supplies are widely available in the market. So, the bargaining power of the suppliers is very low. Bargaining power of buyers: The prospective users or buyers of ‘SELECT WIND’ will be the wind farm owners, investors and researchers. Some potential customer segments may have their own
67 tools. Therefore, it will be a challenge to provide additional services to convince the customers. So, the bargaining power of the buyers is very high. Threats of new entrants: Wind farm evaluation tool from market point of view is an emerging market. There is high threat of new entrants in this market. The big actors in wind energy sector may come into this market with new tools. Degree of Rivalry: Since, this niche market is still at very early stage, the degree of rivalry is low. Threat of substitute: Big companies may come up with complete solution for any wind farm project. Deviation from competitive electricity market will reduce the value of the tool. Therefore, the threat of substitute is high. 6.2.4 SWOT analysis The SWOT analysis is an enormously handy tool for understanding and evaluating all sorts of situations in a business venture. SWOT is an acronym for Strengths, Weaknesses, Opportunities, and Threats. The SWOT analysis of the ‘SELECT WIND’ is illustrated in Figure 6-2.
Internal origin
Helpful
Harmful
STRENGTHS
WEAKNESSES
Variety of Analysis options Easy access User friendly Cost effective
External origin
OPPORTUNITIES
Scope to improve the tool Profit margin will be good New market
Market is not verified Would be a small player in the market No direct marketing experiences Very simple mechanism
THREATS
Market demand could dry out Vulnerable to reactive attack by major competitors Technology lifetime is shorter
Figure 6-2 SWOT matrix of ‘SELECT WIND’
The key strength of the tool is the diversity of analysis options. Moreover, this tool will be made very easy accessible by Internet. It will also have user-friendly interface. Nonetheless, the market is not verified yet since this type of product is not very common. And considering the wind industry size, this product will be a small actor in the market. However, there is always scope to improve or adapt the tool according to the demand of the customers. As the variable cost is very small and moderate initial cost, the profit margin will
68 relatively good for the long run. Even though there will some factors, which can be considered as threats to the business. The market demand could dry out with saturated wind energy industry. Established players in wind industry may enter in this niche segment that will be great threat. Completely different technology and different idea may evolve to meet the demand of this market.
6.3
Business model canvas
A business model describes the rationale of how an organization creates, delivers, and captures value [76]. A business canvas from [76] is a convenient way to elaborate the business model of any product or service. The nine blocks of the canvas has been described to portrait the business model of the ‘SELECT WIND’. The Business Canvas is attached in Appendix J. 6.3.1 Customer Segments The Customer Segments Building Block outlines the different clusters of people or organizations a product or service aims to reach and serve. For the product in discussion the customer segment mainly includes researchers, policy maker, investor and wind farm operators. 6.3.2 Value Proposition The Value Propositions Building Block explains the bunch of services that create value for a precise Customer Segment. The ‘SELECT WIND’ tool provides variety of options to compare different scenarios of wind farm. Technical analysis e.g. energy generations as well as economic analysis e.g. revenues, market values are possible to perform by this tool. Moreover, free versions will be made available for students to use the tool for their research. The services, different users can acquire from the tool is briefly explained in section 3.3. 6.3.3 Channels The Channels Building Block describes how a business connects with and reaches its Customer Segments to deliver a Value Proposition. Different conferences will be attended to promote/show the tool. A website will be launched as well to promote the tool, where latest update and news about the will be available. 6.3.4 Customer Relationship The Customer Relationships Building Block describes the types of associations a business establishes with specific Customer Segments. A contact will be maintained with users and assistance will be provided in case of any problem. Customers will be encouraged to inform about any specific service they expect from the tool. 6.3.5 Revenue Streams The Revenue Streams Building Block represents the cash a business generates from each Customer Segment. The initial idea is to provide the primary version of the tool for free of cost. Later on an upgraded version will be developed. The interested customers will be able to buy to key to use the tool for commercial purposes. However, still there will be a free student version of the tool.
69 6.3.6 Key Resources The Key Resources Building Block describes
the most important assets required making a business model work. The key resource of this business is the skilled people who are developing/improving the tool. Relationships with partners like Innoenergy are also potential resources of this business. 6.3.7 Key Activities The Key Activities Building Block describes
the most important things a business must do
to make its business model work. Research will be continuing to improve the tool to meet the demand of the customers. Intellectual property or copyright will be claimed to secure the business. 6.3.8 Key Partnerships The Key Partnerships Building Block describes
the network of partners that make the business model work. The key partners of this business will be wind turbines manufacturers who can promote the tool to their customers. KIC Innoenergy will also be a partner who can provide logistic and administrative support. Moreover, KIC network will be utilized to promote the tool. EWEA (European Wind Energy Association) can also be considered as a partner. 6.3.9 Cost Structure The Cost Structure describes all costs incurred to operate a business model. The major cost for this business will be the registration cost of the software used to develop the tool e.g. MATLAB. There will be cost for applying for copyright or intellectual property.
6.4
Summary and Conclusion
The market is still very much new as wind farm evaluation tools from electricity market point of view are not frequently and freely available. The developed version of the ‘SELECT WIND” with user friendly and easy assessable interface can make a good impact in this niche market. The Porter’s five forces analysis indicates that the there are high risk of new entrants and being substituted by other services, eve though it has very low supplier bargaining power and diversity of services. The SWOT analysis also indicates that the strengths and opportunities are dominated by the threats and weaknesses. However, these analyses are done without proper market researches and without any experience of market research. The value created by various options of ‘SELECT WIND’, it has the potential to penetrate the market through the execution of perfect business plan.
70
71
7 Conclusions 7.1
Conclusions
Today, wind-based energy generation is showing a good maturity and presents the potential to compete in the energy market along with the traditional energy sources. Even though, in order to effectively integrate large shares of wind power in electricity supply systems, a successful market integration of wind power is regarded crucial. The integration of wind energy into electricity markets is challenging due to the variability and predictability constraints of wind generation. The rather negative correlation between system load and wind power generation creates financial defies experienced by balance responsible parties. Moreover, wind power is expensive and unlike the conventional generations, difficult to predict accurately. Hence, a preemptive assessment of the value of wind energy is a need to realize cost efficient market integration. The main objective of this work is to develop a tool to evaluate the market value of wind energy for any specific electricity market and any specific wind regime. This project was made within the framework of EIT KIC Innoenergy Offwindtech project. The software tool “Market Value” determines the performance of wind power in certain electricity markets where the relation between energy prices, imbalance costs and wind energy yield assemble to determine the actual value of wind power generation. The energy extractions from wind flow and electricity market concepts are investigated to acquire better insights before going into the developing phase of the algorithm. Wind regimes, wind turbine specifications and electricity market prices are the three significant pillars in developing the tool. This tool can provide a variety of outputs based on the demand of the users, which includes, but not limited to, imbalance costs, per unit revenues, per unit market values, per unit investment costs. It allows the user to enhance conditions of markets for the analysis, hence supporting the analysis of many different circumstances. Therefore, a broad array of possibilities can be analyzed. This tool allows to perform simplified research on the market value of wind power in general, or to perform a more precise, detailed analysis. This tool has also the potential to directly compare wind power market performance with other renewable sources e.g. solar energy. A Dutch case study of 2011 has been considered using the market value tool. Three locations based on wind regime (onshore, offshore, coastal) and two markets (day ahead market and intra day market) are selected. The scenarios are compared with respect to their market values, imbalance costs, revenues and costs. The revenues per MWh from the three wind sites are very similar. Even though the gross yearly energy production from the onshore wind farm is much lower than the other two wind farms (due to poor wind speed profile at onshore site). The forecasting of the onshore wind farm is more optimistic, which causes higher imbalance cost compare to the offshore and the coastal wind farm. Nonetheless, the forecasting is more accurate when it is done 2 hours ahead compared to the 24 hours ahead forecasting, as wind speed does not fluctuate significantly within one to four hours (supports van der Hoven spectrum). Therefore, the imbalance cost is higher for the day ahead markets than the intra day markets. The revenues generated from the intraday market are considerably higher than the
72 revenues generated from day ahead market for the three wind farms. This is mainly caused due to the higher prices at intra-day. Since the east coast of The Netherlands shows similar wind conditions as offshore, the results of coastal and offshore wind farm in this case study were very close. However, considering the higher investment cost of offshore wind farm, wind energy generated from the coastal wind farm is most cost effective to be sold in intra day market. Though, the coastal sites may be considered as unfeasible, executable or impractical for largescale wind farm due to practical/commercial reasons e.g. touristic issues, horizon values. Offshore sites have the potential for large-scale wind farm and better forecasting. The investment and maintenance costs of offshore wind farm are expected to reduce with experience and development of technology. In this case study the imbalance costs essentially affect the market values of the wind generations. However, the practical approach of integrating wind power generation into power systems relies on complementary assets. Balance responsible parties counterbalance the wind power imbalances with assets (e.g. conventional flexible generations, load shedding, storage systems) within the portfolio. The data set available for intra day market prices was only for 3041 hours instead 8760 hours of 2011. For the sake of comparisons only those 3041 hours of all other data sets were taken into account. The incomplete data sets are evaluated with respect to the complete data set to observe how the incomplete data sets affect the qualitative analysis of the results. The incomplete data set of day ahead market prices is slightly optimistic while the incomplete data set of imbalance costs is also optimistic. Therefore, the calculated results with the incomplete data sets roughly represent the complete data sets. The preliminary business analyses of the ‘Market Value’ tool reveal that there is potential market for this tool and it is possible to commercially launch this tool through the execution of a perfect business plan.
7.2
Recommendations
In this section some recommendations have been made for the further improvements of the ‘Market Value’ tool. 7.2.1 Market Value Tool The wind speed-power curve can be further improved, specially the region between cut-in and rated wind speed. The standard deviation used to calculate the aggregated power curve in this work is a function of turbulence intensity and dimensions of the wind farm. Therefore, turbulence intensity and dimensions of the wind farm can be used as input to the tool. 7.2.2 Storage possibility Storage possibility is an important option that can be introduced with the ‘Market Value’ tool. This option will add a big value to the tool from user point of view. Several storage technologies can be installed within the tool with their specifications (costs, power ratings, energy density, efficiency, etc.). The tool can suggest the suitable storage systems for any specific scenarios.
73 7.2.3 Graphical User Interface A very preliminary graphical user interface has been developed in this work. There is potential to improve the graphical user interface. The most important improvement can be the introduction of options to insert the time series data files. Power-wind speed curves for most commonly used wind turbines can be preinstalled in the tool to enable the users to select the desired wind turbine. The MATLAB GUI Builder performs slowly. Therefore other platforms with more sophisticated attributes to build a GUI and faster speed can be explored. 7.2.4 Case study In the case study the roughness length is assumed supported by literature. Nevertheless, the theoretical roughness length is often mismatched with actual values. Therefore, more accuracy in terms of results can be obtained using exact or closer to exact roughness lengths for any specific location. It can be computed if wind data of at least two different heights of any location is available. 7.2.5 Business Plan The targeted niche market for the commercial use of this tool can be further explored. Based on detail market analysis, a more elaborate and hands-on business plan can be developed.
74
References [1] [2] [3]
[4]
[5]
[6]
[7]
[8]
[9]
[10] [11]
[12]
[13] [14]
[15]
“Early History Through 1875.” [Online]. Available: http://www.telosnet.com/wind/early.html. [Accessed: 07-Jun-2013]. J. F. Manwell, J. G. McGowan, and A. L. Rogers, Wind Energy Explained: Theory, Design and Application (Google eBook). Jon Wiley and Sons Limited, 2010, p. 704. M. Ragheb and A. Ragheb, “Wind Turbines Theory-The Betz Equation and Optimal Rotor Tip Speed Ratio,” in Fundamental and Advanced Topics in Wind Power, vol. 1, no. 1, R. Carriveau, Ed. InTech, 2011, pp. 19–38. E. Muljadi and C. P. Butterfield, “Pitch-controlled variable-speed wind turbine generation,” Industry Applications, IEEE Transactions on, vol. 37, no. 1, pp. 240–246, 2001. K. Busawon, L. Dodson, and M. Jovanovic, “Estimation of the power coefficient in a wind conversion system,” in Decision and Control, 2005 and 2005 European Control Conference, 2005, pp. 3450–3455. G. Son, H. Lee, and J. Park, “Estimation of Wind Turbine Rotor Power Coefficient Using RMP Model,” in Industry Applications Society Annual Meeting, 2009. IAS 2009. IEEE, 2009, pp. 1–8. M. A. Chowdhury, N. Hosseinzadeh, and W. X. Shen, “Smoothing wind power fluctuations by fuzzy logic pitch angle controller,” Renewable Energy, vol. 38, no. 1, pp. 224–233, Feb. 2012. “Hub height.” [Online]. Available: http://www.wind-energy-the-facts.org/en/part-itechnology/chapter-3-wind-turbine-technology/technology-trends/hub-height.html. [Accessed: 06-Jun-2013]. M. L. Ray, A. L. Rogers, and J. G. McGowan, “Analysis of wind shear models and trends in different terrains,” University of Massachusetts, Department of Mechanical and Industrial Engineering, Renewable Energy Research Laboratory, 2006. S. Rehman and N. M. Al-Abbadi, “Wind shear coefficients and energy yield for Dhahran, Saudi Arabia,” Renewable energy, vol. 32, no. 5, pp. 738–749, 2007. G. Giovanni and S. Sauro, “Wind shear coefficients, roughness length and energy yield over coastal locations in Southern Italy,” Renewable Energy, vol. 36, no. 3, pp. 1081– 1094, 2011. Senate Department of Urban Developement And Environment, “Wind Conditions in Conurbations,” Berlin Digital Environmental Atlas. [Online]. Available: http://www.stadtentwicklung.berlin.de/umwelt/umweltatlas/ed403_01.htm. [Accessed: 27-May-2013]. R. N. Farrugia, “The wind shear exponent in a Mediterranean island climate,” Renewable Energy, vol. 28, no. 4, pp. 647–653, Apr. 2003. J. S. Irwin, “A theoretical variation of the wind profile power-law exponent as a function of surface roughness and stability,” Atmospheric Environment (1967), vol. 13, no. 1, pp. 191–194, Jan. 1979. K. Kaiser, W. Langreder, H. Hohlen, and J. Højstrup, “Turbulence correction for power curves,” in Wind Energy, Springer, 2007, pp. 159–162.
75 [16] [17] [18]
[19]
[20] [21]
[22]
[23]
[24]
[25] [26] [27] [28]
[29] [30] [31] [32]
A. Albers, “Turbulence and Shear Normalisation of Wind Turbine Power Curve,” development, vol. 3, p. 4, 1994. S. Gerard, “Course Material-5P425 Wind Energy.” Eindhoven University of Technology. C. G. Justus, “Winds and wind system performance,” Research supported by the National Science Foundation and Energy Research and Development Administration. Philadelphia, Pa., Franklin Institute Press, 1978. 120 p., vol. 1, 1978. J. S. Rohatgi, V. Nelson, and A. M. West Texas, Wind characteristics: an analysis for the generation of wind power, vol. 239. Alternative Energy Institute, West Texas A&M University, 1994. R. El Attar, Special functions and orthogonal polynomials, vol. 3. Lulu. Com, 2006. “The Environmental and Energy Resources Library - Wind Power Curves & Community cConsiderations.” [Online]. Available: http://www.eerl.org/index.php?P=FullRecord&ID=3130. [Accessed: 08-Jun-2013]. A. Tindal, C. Johnson, M. Leblanc, K. Harman, and E. Rareshide, “SITE-SPECIFIC ADJUSTMENTS TO WIND TURBINE POWER CURVES Wind Speed,” in AWEA WINDPOWER Conference, 2008, no. June, pp. 1–11. E. Rareshide, A. Tindal, C. Johnson, A. Graves, E. Simpson, J. Bleeg, T. Harris, and D. Schoborg, “Effects of complex wind regimes on turbine performance,” in Scientific Proceedings. American Wind Energy Association WINDPOWER Conference, 2009. S. Honhoff, “Power Curves--The effect of environmental conditions,” in Scientific proceedings. GE Wind, AWEA Wind Speed and Energy Workshop, Portland, OR (USA), 2007. P. Norgaard and H. Holttinen, “A multi-turbine power curve approach,” Nordic Wind Power Conference, no. March, pp. 1–2, 2004. S. Stoft, “Power system economics: designing markets for electricity, 2002.” John Wiley & SonsNew York. “You Have a Choice | DPL Energy.” [Online]. Available: http://www.dplenergy.com/about/you_have_a_choice. [Accessed: 04-Jun-2013]. E. Commission and others, “Directive 2003/54/EC of the European Parliament and of the Council of 26 June 2003 concerning common rules for the internal market in electricity and repealing Directive 96/92/EC,” Official Journal of the European Union, vol. 46, 2003. J. Frunt, “Smart Grid Operation through ICT-5P435.” Eindhoven University of Technology, Eindhoven. P. M. Anderson and A. A. Fouad, Power system control and stability. John Wiley & Sons, 2008. “Nord Pool Spot.” [Online]. Available: http://www.nordpoolspot.com/How-does-itwork/Transmission-system-operators-TSOs/. [Accessed: 08-Jun-2013]. “System Operator market operation service provider » Electricity Authority.” [Online]. Available: http://www.ea.govt.nz/industry/mo-service-providers/systemoperator-market-operation-service-provider/. [Accessed: 08-Jun-2013].
76 [33]
[34] [35]
[36]
[37]
[38]
[39] [40]
[41] [42]
[43] [44] [45]
[46] [47]
[48]
“Home - ENTSO-E - European Network of Transmission System Operators for Electricity.” [Online]. Available: https://www.entsoe.eu/home/. [Accessed: 08-Jun2013]. Working Group Distribution Customers and Operation, “The Role of Distribution System Operators (DSOs) as Information Hubs,” 2010. “Swissgrid - Distribution system operators (DSO).” [Online]. Available: http://www.swissgrid.ch/swissgrid/en/home/experts/dso.html. [Accessed: 08-Jun2013]. D. Waldemar, “The role of distribution system operator in the context of energy security – the case of Poland,” PRZEGLĄD ELEKTROTECHNICZNY (Electrical Review), 2011. J. Frunt, I. Lampropoulos, and W. L. Kling, “The impact of electricity market design on periodic network frequency excursions,” in Energy Market (EEM), 2011 8th International Conference on the European, 2011, pp. 550–555. J. Frunt, W. L. Kling, R. M. Hermans, F. A. Nobel, and W. W. de Boer, “Impact of design variables on balancing markets,” in Energy Market (EEM), 2010 7th International Conference on the European, 2010, pp. 1–6. R. Wilson, “Architecture of power markets,” Econometrica, vol. 70, no. 4, pp. 1299– 1340, 2002. NYISO, “Ancillary services manual.” [Online]. Available: http://www.nyiso.com/public/webdocs/documents/manuals/operations/ancserv.p df. [Accessed: 05-Jun-2013]. W. Ch., “Fluctuating Wind Energy and Adequate Market Design for European Power Systems,” Energy Policy, 2009. C. Hiroux and M. Saguan, “Large-scale wind power in European electricity markets: Time for revisiting support schemes and market designs?,” Energy Policy, vol. 38, no. 7, pp. 3135–3145, Jul. 2010. F. Musgens and K. Neuhoff, “Modelling dynamic constraints in electricity markets and the costs of uncertain wind output,” 2006. R. Barth, C. Weber, and D. J. Swider, “Distribution of costs induced by the integration of RES-E power,” Energy Policy, vol. 36, no. 8, pp. 3107–3115, 2008. L. Vandezande, L. Meeus, R. Belmans, M. Saguan, and J.-M. Glachant, “Well-functioning balancing markets: A prerequisite for wind power integration,” Energy Policy, vol. 38, no. 7, pp. 3146–3154, 2010. X. Wang, P. Guo, and X. Huang, “A Review of Wind Power Forecasting Models,” Energy Procedia, vol. 12, pp. 770–778, Jan. 2011. J. Matevosyan and L. Soder, “Minimization of imbalance cost trading wind power on the short-term power market,” IEEE Transactions on Power Systems, vol. 21, no. 3, pp. 1396–1404, 2006. M. Lei, L. Shiyan, J. Chuanwen, L. Hongling, and Z. Yan, “A review on the forecasting of wind speed and generated power,” Renewable and Sustainable Energy Reviews, vol. 13, no. 4, pp. 915–920, May 2009.
77 [49] [50]
[51]
[52]
[53] [54]
[55] [56] [57] [58]
[59] [60]
[61] [62] [63]
[64]
[65]
X. Zhao, S. Wang, and T. Li, “Review of Evaluation Criteria and Main Methods of Wind Power Forecasting,” Energy Procedia, vol. 12, pp. 761–769, Jan. 2011. A. M. Foley, P. G. Leahy, A. Marvuglia, and E. J. McKeogh, “Current methods and advances in forecasting of wind power generation,” Renewable Energy, vol. 37, no. 1, pp. 1–8, 2012. S. S. Soman, H. Zareipour, O. Malik, and P. Mandal, “A review of wind power and wind speed forecasting methods with different time horizons,” in North American Power Symposium (NAPS), 2010, 2010, pp. 1–8. M. Lange and U. Focken, “New developments in wind energy forecasting,” in Power and Energy Society General Meeting-Conversion and Delivery of Electrical Energy in the 21st Century, 2008 IEEE, 2008, pp. 1–8. D. Hou, E. Kalnay, and K. K. Droegemeier, “Objective verification of the SAMEX’98 ensemble forecasts,” Monthly Weather Review, vol. 129, no. 1, pp. 73–91, 2001. M. Lange, “On the uncertainty of wind power predictions-Analysis of the forecast accuracy and statistical distribution of errors,” Transactions of the ASME-N-Journal of Solar Energy Engineering, vol. 127, no. 2, pp. 177–184, 2005. “What do we want to achieve?” [Online]. Available: http://ec.europa.eu/energy/renewables/index_en.htm. [Accessed: 06-Jun-2013]. “MATLAB - The Language of Technical Computing.” [Online]. Available: http://www.mathworks.com/products/matlab/. [Accessed: 06-Jun-2013]. “KNMI - Daggegevens van het weer in Nederland.” [Online]. Available: http://www.knmi.nl/klimatologie/daggegevens/index.cgi. [Accessed: 14-Jun-2013]. “NRCS National Water and Climate Center - Climate Products - Climate Data - Wind Rose.” [Online]. Available: http://www.wcc.nrcs.usda.gov/climate/windrose.html. [Accessed: 13-Jun-2013]. “Economics | EWEA.” [Online]. Available: http://www.ewea.org/policyissues/economics/. [Accessed: 16-Jun-2013]. “The Economics of Wind Energy.” [Online]. Available: http://www.ewea.org/fileadmin/files/library/publications/reports/Economics_of_ Wind_Energy.pdf. [Accessed: 16-Jun-2013]. J. G. Slootweg, “Course Material-5P425 Wind Energy.” Eindhoven University of Technology, 2013. “APX | Power Spot Exchange – Making Markets Work.” [Online]. Available: http://www.apxgroup.com/. [Accessed: 14-Jun-2013]. T. Aigner and T. Gjengedal, “Modelling wind power production based on numerical prediction models and wind speed measurements,” in 17th Power Systems Computation Conference, Stockholm, 2011. “TenneT Export data.” [Online]. Available: http://www.tennet.org/english/operational_management/export_data.aspx. [Accessed: 15-Jun-2013]. J. P. Chaves-Ávila, R. A. Hakvoort, and A. Ramos, “Short-term strategies for Dutch wind power producers to reduce imbalance costs,” Energy Policy, vol. 52, no. null, pp. 573–582, Jan. 2013.
78 [66]
[67]
[68] [69]
[70]
[71]
[72]
[73] [74]
[75] [76] [77] [78]
[79]
[80] [81]
T. Hennessy and M. Kuntz, “The multiple benefits of integrating electricity storage with wind energy,” in Power Engineering Society General Meeting, 2005. IEEE, 2005, pp. 1952–1954. W. A. Amos, “Economic assessment of wind energy coupled with a reversible hydrogen fuel cell,” National Renewable Energy Laboratory, Golden, CO, Milestone Type P report, 2000. M. Korpås, R. Hildrum, and A. T. Holen, “Hydrogen energy storage for grid-connected wind farms,” in IASTED International Conference" Power and Energy Systems, 2001. E. Naswali, C. Alexander, H.-Y. Han, D. Naviaux, A. Bistrika, A. von Jouanne, A. Yokochi, and T. K. A. Brekken, “Supercapacitor energy storage for wind energy integration,” in Energy Conversion Congress and Exposition (ECCE), 2011 IEEE, 2011, pp. 298–305. H. L. Ferreira, R. Garde, G. Fulli, W. Kling, and J. P. Lopes, “Characterisation of electrical energy storage technologies,” Energy, vol. 53, no. null, pp. 288–298, May 2013. C. Bullough, C. Gatzen, C. Jakiel, M. Koller, A. Nowi, and S. Zunft, “Advanced adiabatic compressed air energy storage for the integration of wind energy,” in Proceedings of the European Wind Energy Conference, EWEC, 2004, vol. 22, p. 25. C. Naish, I. McCubbin, O. Edberg, and M. Harfoot, “Outlook of energy storage technologies,” European Parliament’s committee on Industry, Research and Energy (ITRE), Brussels, 2008. C. Abbey and G. Joos, “Supercapacitor energy storage for wind energy applications,” Industry Applications, IEEE Transactions on, vol. 43, no. 3, pp. 769–776, 2007. “Energy storage for offshore wind power.” [Online]. Available: http://www.remenergy.it/public/ml_attach/5 SAIPEM SA - Jacques Ruer.pdf. [Accessed: 28-Jun-2013]. N. Argyres and A. M. McGahan, “An interview with Michael Porter.,” The Academy of Management Executive, vol. 16, no. 2, pp. 43–52, 2002. A. Osterwalder and Y. Pigneur, Business model generation: a handbook for visionaries, game changers, and challengers. Wiley, 2010. A. Boone, “Simulation of short-term wind speed forecast errors using a multi-variate ARMA (1, 1) time-series model,” KTH, 2005. J. L. Torres, A. Garcia, M. De Blas, and A. De Francisco, “Forecast of hourly average wind speed with ARMA models in Navarre (Spain),” Solar Energy, vol. 79, no. 1, pp. 65–77, 2005. E. Cadenas and W. Rivera, “Wind speed forecasting in three different regions of Mexico, using a hybrid ARIMA--ANN model,” Renewable Energy, vol. 35, no. 12, pp. 2732–2738, 2010. R. G. Kavasseri and K. Seetharaman, “Day-ahead wind speed forecasting using< i> f-ARIMA models,” Renewable Energy, vol. 34, no. 5, pp. 1388–1393, 2009. G. Sideratos and N. Hatziargyriou, “Using radial basis neural networks to estimate wind power production,” in Power Engineering Society General Meeting, 2007. IEEE, 2007, pp. 1–7.
79 [82] [83]
[84]
[85] [86]
[87]
[88]
S. Li, “Wind power prediction using recurrent multilayer perceptron neural networks,” in Power Engineering Society General Meeting, 2003, IEEE, 2003, vol. 4. C. Potter, M. Ringrose, and M. Negnevitsky, “Short-term wind forecasting techniques for power generation,” in Australasian Universities Power Engineering Conference (AUPEC 2004), 2004, pp. 26–29. P. Pinson, L. E. A. Christensen, H. Madsen, P. E. Sørensen, M. H. Donovan, and L. E. Jensen, “Regime-switching modelling of the fluctuations of offshore wind generation,” Journal of Wind Engineering and Industrial Aerodynamics, vol. 96, no. 12, pp. 2327–2347, 2008. A. Kusiak, H. Zheng, and Z. Song, “Models for monitoring wind farm power,” Renewable Energy, vol. 34, no. 3, pp. 583–590, 2009. R. Jursa and K. Rohrig, “Short-term wind power forecasting using evolutionary algorithms for the automated specification of artificial intelligence models,” International Journal of Forecasting, vol. 24, no. 4, pp. 694–709, 2008. N. Amjady, F. Keynia, and H. Zareipour, “Wind power prediction by a new forecast engine composed of modified hybrid neural network and enhanced particle swarm optimization,” Sustainable Energy, IEEE Transactions on, vol. 2, no. 3, pp. 265–276, 2011. “KNMI - Klimaatdata en Advies - Potentiële wind - Download.” [Online]. Available: http://www.knmi.nl/klimatologie/onderzoeksgegevens/potentiele_wind/. [Accessed: 13-Jun-2013].
80
81
Appendix
Appendix A:
Fatigue load
Fatigue loads can be thought of as the loss of strength that a material experience when subjected to a cyclic stress history. The process often starts at the surface of a material where small imperfection such as scratches cause stress concentrations. As the crack size grows, stress concentrations grow accordingly, causing the cracks to grow further. This process eventually leads to rupture of the material. It is important to note that fatigue occurs even when the applied loads are far below the material's elastic limit. This implies that not only the extreme load values are important; the small-amplitude time history of the load affects component reliability due to fatigue damage. Due to the turbulent nature of wind, the structural components of a wind turbine are exposed to highly varying loads. The major reasons of varying load may be The operational state of a turbine varies constantly due to the ever-changing environmental conditions Heavy large wind turbine blades rotating in the earth gravity field experience periodic change of gravity load. Turbines usually feature slender structures with low eigenfrequencies, which can easily be excited by gusts or even by the operation itself They are often designed to endure 20 years of operation or even more, where they produce power for 80 to 90 % of the time. Therefore, fatigue damage is a major consideration when designing wind turbines. The dimensioning of practically all major components is at least partially driven by fatigue. In order to describe a material's ability to withstand cyclic stress histories, an SN curve is often used. An SN curve can be constructed by applying constant- amplitude cyclic stress to a test specimen and count the number of load cycles until rupture occurs. Repeating the test for a number of stress ranges allows the stress range to be plotted against the number of cycles that the material can withstand.
Figure A-1 Typical SN curve
82
Appendix B:
Wake effect of wind farm
The term “wake effect” originates from the wake behind a ship. In contrast to a ship-generated wake, however, a wind turbine wake is a long trail of turbulent wind escaping the turbine with weakened wind speed. For wind turbines, wake effect narrates to the wind speed dearth and reduced energy content wind possesses after leaving a specific utility-scale wind turbine. The kinetic energy in wind is converted into electricity by wind turbine. As wind flows through a turbine, the bulk of air downwind of the turbine has a lower wind speed and higher turbulence than wind in the free stream. The free stream is the air far upstream from a wind turbine that is traveling at its natural velocity and that has not yet been slowed down, deflected, or otherwise impacted by a wind turbine or other obstacle. Therefore, wind exiting a turbine contains less kinetic energy than does wind before passing through a turbine. This diminished, turbulent wind from an upwind turbine reduces the energy entering downwind turbines, thereby decreasing the downwind turbines’ overall energy output. It is important to consider the wake effect from neighboring wind farms and the possible impact of the wind farms that will be built in future.
Figure B-1 Wake effect in a Wind Farm
83
Appendix C: literature
Key wind power forecasting methods found in
The wind power or speed forecasting methods, which are commonly used in literatures, are briefly introduced belowI. Persistence Persistence wind or power forecasting assumes that the wind (speed and direction) or power at a certain future time will be the same as it is when the forecast is made. Persistence is obviously a very simple method and is mentioned here since it is used as a reference to evaluate the performance of advanced methods. An advanced method is worth implementing if it outperforms persistence. Wind, however, is somehow persistent in nature. Persistence is a difficult method to beat, especially on the short-term (1–6 hr). II. Random Time Series This method had been presented for eliminating diurnal non-stationary, and vectorized hourly wind speed was expressed as a vector auto regression (VAR) model. The results showed that the presented VAR model can yield satisfactory hourly wind speed forecast as long as 72 h ahead under normal weather conditions. The short-term wind speed forecast errors using a multi-variate ARMA (1, 1) time-series model is simulated in [77]. The ARMA (autoregressive moving average process) and persistence models is used in [78] to predict the hourly average wind speed up to 10 h in advance. Hybrid models consisting of Autoregressive Integrated Moving Average (ARIMA) models and Artificial Neural Network (ANN) models is developed in [79]. The use of fractional-ARIMA or f-ARIMA models is examined in [80], and forecast wind speeds on the day-ahead (24 h) and two-day-ahead (48 h) horizons. III. Neural Network Artificial neural networks comprise of small interconnected components, called neurons. Inputs of neurons are weighted and summed together, to be assessed by a transfer function, which is the input for the next neurons or the output of the network. The weights are set during a training stage or can be adapted on-line. Two wind power forecasting methodologies based on radial basis neural networks and fuzzy logic techniques is described in [81] to estimate the quality of the numerical weather predictions. Comparing their performance with the performance of the persistence method showed the effectiveness of both tools. In [82] a method is discussed to do time series prediction forecast of wind power generation using recurrent multilayer perceptron (RMLP) neural networks. The paper presented a four layer RMLP network and the extended Kalman filter based backpropagation through time algorithm was used to train the RMLP networks. IV. Adaptive Neural Fuzzy Inference System (ANFIS) Application of an Adaptive Neural Fuzzy Inference System (ANFIS) to forecasting a wind time series is introduced by Cameron Potter[83]. It concluded that ANFIS was a promising forecasting
84 technique. Fuzzy inference algorithm was used to interpolate the missing and invalid wind data in ANFIS model. V. Markov-Switching Autoregressive (MSAR) Model A methodology based on a Markov-switching autoregressive model with time-varying coefficients is presented in [84]. The quality of this methodology was demonstrated from the test case of two large offshore wind farms in Denmark. An advantage of the method was that one can easily derive full predictive densities along with the usually generated point forecasts. VI. Nearest Neighbor Search (NNS) Literature shows that the k-nearest neighbor model, combined with the principal component analysis approach [85], outperformed other models studied. Data mining and evolutionary computation were integrated for building the models for prediction and monitoring. VII. Evolutionary Algorithms (EA) In [86], this short-term prediction method based on neural networks and the nearest neighbor search. In comparison to the manually specified neural network model, the new method gets a reduction of the prediction error for the most wind farms. VIII. Particle Swarm Optimization (PSO) A new forecasting engine including a new enhanced particle swarm optimization component and a hybrid neural network is presented in [87]. The proposed wind power forecasting strategy was applied to real-life data from wind power producers in Alberta, Canada, and Oklahoma, USA. The presented numerical results demonstrated the efficiency of the proposed strategy
85
Appendix D:
Demand and supply curve
Demand curve depicts the demand of electricity with the price and on the other hand supply curve represents the electricity supply with respect to the price. The aggregation of demand and supply curve determines the market clearing price and market clearing volume. Figure D-1 depicts a simple aggregation of demand and supply curve.
Figure D-1 General demand and supply graph
The generic goal of power system’s economy is to maximize the social surplus, to maximize the consumption benefits while minimizing the production costs. The social surplus is maximum, when a competitive market is allowed to operate freely and the price is settled at the intersection of the supply and demand curves. The sum of net consumer’s surplus and producer’s surplus is defined as the social surplus. Social surplus is also called global surplus. Figure D-2 illustrates the consumer’s surplus and producer’s surplus. An aggregated bid curve, showing supply and demand from Amsterdam Power Exchange is depicted in Figure D-3.
86
Figure D-2 Social surplus
Figure D-3 Aggregated bid curves for supply and demand from Amsterdam Power Exchange [29]
87
Appendix E:
Actors in Dutch electricity market Regulator
Transmission System Operator (TSO) Amsterdam Power Exchange
Distribution Network Operators
Generation Companies
ENERGIEKAMER. It is part of the Netherlands competition authority. (www.nma.nl/regulering/energie). TenneT (www.tennet.org) APX-ENDEX is the energy exchange that operates spot and futures markets for electricity and natural gas in the Netherlands, the U.K. and Belgium. (www.apxendex.com)
Enexis (formerly Essent) Liander (formerly Nuon) Stedin (formerly Eneco) Delta NRE Obragas Cogas Intergas Rendo Westland NetH Nuon Essent Eneco Oxxio NEM Greenchoice Electrabel
88
Appendix F:
Graphical user interface of ‘Market Value’ tool
The graphical user interface is build based on the case study described in Chapter 5. Therefore, there are three input column for three different locations. The users have the flexibility to check the financial parameters e.g. Imbalance cost, per unit investment cost, market value and required subsidy by putting different inputs. The graphical interface is developed using MATLAB GUI Builder. Inputs: Investment and O&M cost in million: The user can enter the investment and O&M cost of the respective wind farm, which will be used to calculate per unit cost and subsidy. Roughness length of the terrain considered: As roughness length varies very randomly with the locations, therefore there is an option to enter the roughness length of the respective terrain. Standard deviation of the wind farm: This is another location dependent parameter, which the user can put by himself/herself. This value depends on the turbulence intensity and dimension of the wind farm. Select market: The user can choose the desired electricity market to trade. Day ahead or intraday market can be selected from the ‘Select Market’ dropdown menu. Outputs: Costs in Euro/MWh, Imbalance Costs in Euro/MWh, Market Value in Euro/MWh and Subsidy required in Euro/MWh. There are also options to view the graphs at the bottom of the interface. The main drawback of the tool is that still it is not capable to take the times series data like market price, wind speed from the users. And secondly it takes lot of time to run the program in MATLAB GUI Builder. A screen shot of the graphical user interface is presented in Figure F-1.
89
Figure F-1 Graphical user interface of SELECT WIND
90
Appendix G:
Weibull distributions of Wind Speed data sets
Weibull distributions measured for different wind speed data sets within the case study described in chapter 4 are depicted in the following figures in this section.
Figure G-1 Weibull Distribution of wind speed at offshore site
Figure G-2 Weibull distribution of 2 hours ahead forecasted wind speed at offshore site
91
Figure G-3 Weibull distribution of wind speed at coastal site
Figure G-4 Weibull distribution of 2 hours ahead forecasted wind speed at coastal site
Figure G-5 Weibull distribution of 24 hours ahead forecasted wind speed at onshore site
92
Figure G-6 Weibull distribution of 2 hours ahead forecasted wind speed at onshore site
93
Appendix H:
Imbalance cost
Annual profile of imbalance cost for 24 hours ahead and 2 hours ahead at onshore location for whole year and 3041 hours of the year depicted in the following figures in this section.
Figure H-1 Annual profile of imbalance cost for day ahead forecasting at onshore site
Figure H-2 Profile of 3041 hours imbalance cost for day ahead forecasting at onshore site
94
Figure H-3 Annual profile of imbalance cost for 2 hours ahead forecasting at onshore site
Figure H-4 Profile of 3041 hours imbalance cost for 2 hours ahead forecasting at onshore site
95
Appendix I:
Case study- Comparing two turbines
A case study is also performed using the ‘Market Value’ tool to compare to types of wind turbines. The case study specifications are given belowWind Turbines in consideration: V90-3MW and V126-3MW Power Market: Dutch day ahead market (market price was made available by APX-Endex) Wind Site: Ijmuiden (wind data of that location in 2010 is collected from www.knmi.nl) Number of Turbine: 60 Lifetime: 20 years The roughness length and standard deviation of the wind farm is estimated based on the knowledge from the literatures. The specifications of the turbines are given in the Table H-1. Table H-1 Specifications of the turbines
V90-3MW
V126-3MW
Cut-in Wind Speed (m/s)
3.5
3
Cut-out Wind Speed (m/s)
25
22.5
Wind Class
IEC IA and IEC IIA
IEC IIIA
Rotor Swept Area (m2)
6362
12469
Rotor Diameter (m)
90
126
The wind speed-power curves of the two turbines are depicted in the Figure H-1.
Figure H-1 Wind Speed-Power curve of V90-3MW and V126-3MW
Results: Even though the rated power of the two turbine is same but their annual energy production varies. This is very expected that the V126 would capture more energy with its large rotor swept
96 area. The yeraly profile power generated by the two turbines are depicted in Figure H-2 and Figure-3. The trend of power generateion is approximately same.
Figure H-2 Yearly profile of wind power generation by V90 turbine
It evident from the Figure H-2 and Figure H-3 that the V126 produces comparatively more powers during low wind speeds than the V90.
Figure H-3 Year profile of wind power generation by V126 turbine
Figure H-4 and Figure H-5 illustrate the annual profile of generated revenue by the two turbines for the year 2010 in Ijmuiden. However, the revenues are also largely dependent on the market price. These two graphs represent that the revenue generated by V90 during first six months of the year is very low unlike the V126. Even for the months with highest revenue e.g. October, December, there is lot more instances with very low revenue for V90 compared to the V126.
97
Figure H-4 Year profile of the revenue earned by the V90 turbine
Figure H-5 Yearly profile of revenue earned by the V126 turbine
Figure H-6 depicts the total revenue generated by each turbine in the year 2010. As expected, V126 generates almost double revenues than the V90. It is because V126 has lower cut-in wind speed, which enables it to extract energy from lower wind speeds and it reaches to the rated power output at lower wind speed. However, the V90 has higher cut-off wind speed but the location considered for the case study has seldom wind speeds higher than 20 m/s. Therefore, V126 is the more feasible wind turbine to be used than V90 for the wind site taken into consideration in this case study.
98
Figure H-6 Total revenue generated by each turbine in the year 2010
99
Appendix J:
Business model canvas
100
Appendix K:
Storage Technologies
Figure K-1 Comparison of different storage technologies[70]
101
Market Value of Wind Power J.E.S. de Haan
M.A. Shoeb
H.M. Lopes Ferreira
W.L. Kling
Eindhoven University of Technology Electrical Energy Systems Eindhoven, the Netherlands
[email protected]
Eindhoven University of Technology Electrical Energy Systems Eindhoven, the Netherlands
[email protected]
Eindhoven University of Technology Electrical Energy Systems Eindhoven, the Netherlands
[email protected]
Eindhoven University of Technology Electrical Energy Systems Eindhoven, the Netherlands
[email protected]
Abstract—Variability and predictability constraints of wind hinder the cost-efficient integration of wind power generation into power markets. Within the framework of EIT KIC INNOENERGY Offwindtech project, a ‘Market Value’ tool is developed. Here, the market value of wind power generation can be assessed with respect to its wind site and the respective power market it is integrated in. A case study is introduced to compare the potential market value of different wind sites (offshore/onshore) and of different power market concepts (day ahead/intra-day). It is found that the relative market value of wind power does not significantly differ between its diverse conditions. Nevertheless, when considering the costs of wind power generation, coastal wind power generation, to be sold in intra-day markets, has the largest potential to be costefficient. Future reduction of offshore installation and maintenance cost could further increase the market value/competition of offshore wind power. The difference of predictability accuracy between onshore and offshore wind power generation has negligible impact on the results Keywords-day ahead market; intra-day market; wind forecast; wind power generation
I. INTRODUCTION The integration of wind power generation into power markets requires a different approach with respect to energy trading. Due to variability and predictability constraints of wind power generation, Balance Responsible Parties experience difficulties to meet their energy schedules. Due to the inaccurate prediction and the rather negative correlation between system load and wind power generation, challenges are faced. As a result, the generation portfolio of market parties still relies mainly on conventional generation. Nonetheless, several new market arrangements have been introduced, to enable a more cost-efficiently integration of wind power generation into power markets. Two main examples are trading across borders and trading closer to real-time, at intra-day. Several studies have investigated the financial impact and the costs of integrating wind power generation e.g. [1]. This work aims at assessing the value, rather than the costs of wind power. It is somehow self-evident that the forecast of wind power generation closer to real time is more accurate. This logical reasoning is followed by the hypothesis that wind power should always be sold in intra-day markets. Though, market prices differ between day ahead and intra-day. It is
certainly not evident that market prices at intra-day stage are equal or larger compared to prices at day ahead stage. Therefore, the actual market value of wind power depends on the related market prices at day ahead and intra-day and is determined by the accuracy of forecast techniques at day ahead and intra-day. Supported with a case study, the market value of wind power is determined, based on geographical conditions of exemplarily chosen wind power plants and based on specific power market concepts it will be integrated in. The paper is structured as follows: the project of EIT KIC INNOENERGY Offwindtech and the applied algorithm for the ‘market value’ tool are introduced in section II. The case study is introduced in section III in combination with a data analysis of wind power generation and market prices of the year 2011. The algorithm of the tool is used to acquire results, presented in section IV. Finally, conclusions are drawn in section V. II. KIC INNOENERGY A. KIC Offwindtech Within the EIT KIC INNOENERGY Offwindtech project, four tools are developed. Their inter-relation is schematically shown in Fig. 1. The peculiar concept of the project is characterized by the possible interconnection of all 4 separate tools. This enables an overall assessment of specific wind power plants, with detailed results. However, all tools can be used independently. B. ‘Market value’ tool The ‘market value’ tool is developed for, e.g., policy makers, regulators, wind power plants operators, investors,
Figure 1. Relation of the four tools within the Offwindtech WP1 framework.
102 market parties and academics. Besides, the versatility of the tool enables to perform simulations with a diversity of objectives. Policy makers could determine the requested subsidy or feed in tariff for wind power generation. Investors can trace optimal locations of their planned wind power plants. Market parties use the tool to calculate their optimal market integration of wind power. The tool can also be used to investigate the effect, impact or performance of different forecast techniques. The ‘market value’ tool is developed in Matlab environment. Its inputs are schematically depicted in Fig. 2. C. Algorithm The algorithm of the market value tool is comprehensively explained in [2]. Here, wind speeds on sensor height are converted to wind speeds experienced at the related hub-height of wind turbines [3]. The wind speed time series is converted into wind power generation time series representing a wind power plant. Therefore, a multi turbine power curve approach of [4] is applied based on wind turbines with a power–wind speed characteristic of [5] and [6]. The calculated wind power generation time series in [MWh/h] is multiplied with market prices [EUR/MWh] to calculate the revenue of wind power generation [EUR/h]. Re venue of wind generation EUR / h
(1)
wind power generation MWh / h market prices EUR / MWh
To determine the financial yield of wind power generation, the time series imbalance costs [EUR/h] are deducted from the wind revenue [EUR/h]. Yield of wind power generation EUR / h
(2)
revenue of wind power generation EUR/ h wind imbalance cos ts EUR / h
Here the wind imbalance costs are a result of the multiplication of expected wind power generation – actual wind power generation in [MWh/h] with imbalance costs in [EUR/MWh]. Wind imbalance cos ts EUR / h
(3)
exp ected wind power generation MWh / h actual wind power generation MWh / h imbalance cos ts EUR / MWh
Figure 2. Schematic overview of the inputs and outputs of the ‘Market Value’ tool.
III. CASE STUDY In this section, a case study is introduced to assess the market value of wind power located in the Netherlands for a reference year 2011. The geographical distinction of wind sites will be made between offshore, coastal, and inland onshore wind power generation. These three locations differ based on the combination of wind regime and costs of wind power plant installation. The assessment will elaborate on the hypothesis that coastal wind power, traded closer to real time is most cost-efficient. The generated power is fictively sold in the Dutch power market, the energy exchange platform of APX-Endex, operating spot and futures markets for electricity. APXEndex made data available [7], and in [2] it is elaborated how this data was used. With respect to the full integration of large scale wind power into spot markets, it must be clarified that day ahead and intra-day markets are not (yet) mature and liquid to support this integration. It is expected that market prices would significantly be affected. The wind data set with its associated forecast data set was made available based on work of [8]. The related power imbalance costs in [EUR/MWh] were acquired from the Dutch Transmission System Operator TenneT TSO B.V. [9]. A. Cases The distinction is made between 6 different concepts as depicted in Fig.3. B. Wind site Three different wind sites have been chosen based on their characteristic wind site experience. Offshore wind power generation represents high wind speeds, however, its installation and maintenance costs are significant larger compared to onshore with smaller energy yield. Nevertheless, the main property of the coastal location is its wind regime with approximately offshore conditions, but with low installation costs. The statement that coastal installation, especially in the Dutch west coastline, has offshore wind conditions is deduced from the fact that in the Netherlands, wind flows on annual basis mainly directed from south-west towards north-east. This is geographically depicted in Fig. 4, with a reference to the three wind sites assessed in this work. The analysis is based on data from [10].
Figure 3. Schematic depiction of the 6 concepts of wind power plants within the case study.
103 E. Market prices The APX Endex market data of day ahead is depicted in Fig. 7. Day and night patterns as well as seasonal patterns are clearly noticeable. The correlation of market prices and wind power generation is weak.
Figure 4. Geographical depiction of the three wind sites and the main wind direction over the Netherlands.
C. Wind power generation The wind power generation data set of the onshore site is depicted in Fig. 5. The randomly behavior is noticeable. D. Wind forecast and power imbalances The wind forecast techniques used in this work become more accurate closer to real time. This effect is depicted in Fig. 6 where from day ahead (D-24h) up to one hour ahead (D-1h) the duration curves of forecast errors are shown. The related imbalances for the case study are derived using (3) and the results show that besides wind generation, also wind power imbalances are identified with certain patterns or trends, in order to anticipate on.
IV. RESULTS Based on the algorithm (section II) and the case study (section III) results are shown in Table 1. Generally, the calculated values of revenues of wind power generation are relatively high, compared to values from literature. This is caused by theoretical considerations with perfect circumstances, not considering imperfections such as turbulence of wind sites, or maintenance and failures of wind turbines, or taking into consideration other externalities. Secondly, based on analyses of market price data, it is found that intra-day market prices are commonly larger compared to day ahead, however, this difference is relatively small. Likewise, the profile of the day ahead and of the intra-day market prices is strongly correlated. Therefore, it can be concluded that the impact of inaccuracy of wind power forecast is the main cause of dissimilar results for the resulting economic benefits of wind power in day ahead and intra-day markets. Thirdly, the data analysis revealed that the forecast of inland onshore wind power generation is optimistic (more generation) compared to the forecast of offshore wind power generation. This will result in additional negative imbalances for onshore wind power generation which commonly come along with larger imbalance costs, due to the Dutch thermal generation oriented system. This effect has the largest impact at the day ahead market, and closer to real time at the intra-day market, this effect attenuates. Thus, the performance of a more accurate forecasting limits the expense of total imbalance costs.
Figure 5. Exemplarily depiction of an annual profile of onshore wind power generation.
In Table 1 it is listed that the wind power imbalance costs reduce significantly as expected when changing from day ahead stage towards the intra-day stage. This effect is similar for all three locations. The accuracy of wind prediction at the intra-day stage is related to the fact that wind will not vary significantly up to real time, as can be concluded from the “van der Hoven Spectrum” [3]. Therefore, forecast should become more accurate which is confirmed by the results of Table 1. The impact of wind inaccuracy is smallest for offshore wind generation.
Figure 6. Duration curve of wind forecast mismatches of day-24h up to day-24h.
Figure 7. Exemplarily depiction of an annual profile of day ahead market prices.
104 Primarily, this is due to the optimistic (more generation) forecast of onshore which leads to relatively lower offshore imbalance costs. Next, this is based on the power–wind speed curve of wind turbines. Hence, at moments of higher wind speeds, wind turbines operate at rated power, where variations in wind speed (leading to forecast inaccuracy) do not lead to power deviations. At offshore wind sites the probability of higher wind speeds is larger, due to smaller roughness factors. Therefore, imperfections in wind forecast have a smaller impact on offshore wind power generation, and consequently wind imbalance costs are smaller compared to onshore wind power generation. Even though offshore wind power generation is forecasted more accurately and therefore smaller imbalance costs are experienced, its installation and maintenance costs are higher. Therefore, in this case study, coastal wind power generation is still most cost-efficient, operating in the intraday market. Table 1. Results of the six concepts of wind power generation based on wind sites and power market. Offshore Onshore Coastal Wind Wind Wind Costs 80.0 60.0 48.0 [€/MWh] Day Ahead revenue 54.49 54.57 54.57 [€/MWh] Day Ahead imbalances 9.56 12.29 10.20 [€/MWh] Day Ahead value 45.00 42.28 44.57 [€/MWh] Intra-day revenue 57.74 58.46 58.16 [€/MWh] Intra-day imbalances 3.64 4.85 3.95 [€/MWh] Intra-day value 54.09 53.61 54.20 [€/MWh]
V. CONCLUSIONS The integration of wind energy into power markets is challenged by variability and predictability constraints of wind power generation. The variability of wind power restricts the desired sale of wind power on peak moments at rated power. Furthermore, forecast mismatches of wind induce imbalance costs. Therefore, in order to assess the actual market value of wind power, an algorithm is developed, within the framework of EIT KIC INNOENERGY Offwindtech project. This software tool “Market Value of Wind Power” determines the performance of wind power in certain power markets where the relation between energy prices, imbalance costs and wind energy yield are congregated to determine the actual value of wind power generation. Supported with a case study, the market value of different wind sites (offshore/onshore) and of different power market concepts (day ahead/intra-day) is assessed. In our example, the results show that coastal wind power generation is indeed the most cost-efficient concept, integrated in intra-day markets. This outcome is owed to low installation and maintenance costs in combination with a high energy yield wind site. Nevertheless, the concept of coastal wind power does not have the highest potential in the medium and long term, as it is rather unlikely that coastlines will be equipped with large scale wind power plants. Thus, due to the residential
and recreational value of coastal areas in The Netherlands, this area will most probably remain unexploited. A significant step is already made with the integration of intra-day power markets where closer to real time wind power can be traded and, evidently, lower imbalance costs are involved. Nevertheless, these spot markets are currently not (yet) mature and liquid to support the large scale integration of wind power generation. A second adequate step to increase the cost-efficient integration of wind power is under development, where installation and maintenance costs of offshore wind power generation are reduced applying new techniques and gaining experience in this field. Finally, this work’s approach is to display how wind power can be integrated into power markets as an independent power source. More practically, wind power is part of a generation portfolio relying on complementary assets. Therefore, wind power imbalances are respectable counterbalanced with assets within the generation portfolio of market parties. Nevertheless, with the increased share of wind power generation, it is plausible that wind power could solely be integrated into power markets as a cost-efficient, independent power source. VI. REFERENCES Holttinen, H., Meibom, P., Orths, A., Lange, B., O'Malley, M., Tande, J.O., Estanqueiro, A., Gomez, E., Söder, L., Strbac, G., Smith, J.C. van Hulle, F., 2009. “Impacts of large amounts of wind power on design and operation of power systems, results of IEA collaboration”. In: 8th International Workshop on LargeScale Integration of Wind Power into Power Systems as well as on Transmission Networks of Offshore Wind Farms, 14-15 October, 2009. [2] M.A. Shoeb “Integration of Wind Power into Electricity Markets”, Master thesis, Eindhoven University of Technology, EES.13.A0002, 2013. [3] Manwell, J.F., J.G. McGowan, and A.L. Rogers. Wind Energy Explained: Theory, Design and Application. John Wiley & Sons, Chichester, 2003. [4] Per Norgaard and Hannele Holttinen, “A Multi-Turbine Power Curve Approach” Nordic Wind Power Conference 2004, Chalmers University of Technology. [5] J.E.S. de Haan, J. Frunt, W.L. Kling, "Wind Turbines' Kinetic Energy Storage Potential for Frequency Support", European Wind Energy Conference EWEC 2011, Brussel, Belgium, 14-17 March, 2011. [6] J. G. Slootweg, W.H. De Haan, H. Polinder, W.L. Kling, “General model for Representing Variable Speed Wind Turbines in Power System Dynamics Simulations” IEEE Transactions on Power Systems, Vol. 18, NO.1, Feb. 2003. [7] Data made available by APX-ENDEX, Available online at: www.apxgroup.com Last accessed: Aug. 2013. [8] T. Aigner and T. Gjengedal, “Modelling Wind Power Production based on Numerical Prediction Models and Wind Speed Measurements”, in Power System Computational Conference 2011, Stockholm, 2011, pp. 1-7. [9] TenneT TSO B.V., “System & transmission data, Export data“ Available online at: http://www.tennet.org/english/operational_ management/export_data.aspx Last accessed: Aug. 2013. [10] Wind speed data , Available online at: http://www.knmi.nl/klimatologie/onderzoeksgegevens/potentiele_wi nd/ Last accessed: 2013. [1]
105
