• No results found

Benchmarking of Optimization Modules for Two Wind Farm Design Software Tools

N/A
N/A
Protected

Academic year: 2021

Share "Benchmarking of Optimization Modules for Two Wind Farm Design Software Tools"

Copied!
117
0
0

Loading.... (view fulltext now)

Full text

(1)

 

Benchmarking of Optimization

Modules for Two Wind Farm Design

Software Tools

Eftun YILMAZ

Submitted to the Office of Graduate Studies of Gotland University

in partial fulfillment of the requirements for the degree of MSc Wind Power Project Management, Master Thesis 15 ECTS

Supervisor: Assoc. Prof. Bahri Uzunoglu Examiner: Prof. Jens N. Sørensen

Master of Science Program in Wind Power Project Management, Department of Wind Energy,

Gotland University Cramérgatan 3

621 57 Visby, Sweden

(2)

 

ABSTRACT

Optimization of wind farm layout is an expensive and complex task involving several engineering challenges. The layout of any wind farm directly impacts profitability and return of investment. Several software optimization modules in line with wind farm design tools in industry is currently attempting to place the turbines in locations with good wind resources while adhering to the constraints of a defined objective function. Assessment of these software tools needs to be performed clearly for assessing different tools in wind farm layout design process. However, there is still not a clear demonstration of benchmarking and comparison of these software tools even for simple test cases. This work compares two different optimization software namely openWind and WindPRO commercial software tools mutually.

Keywords: wind energy, optimization, wind farm layout, wind turbine placement, efficiency

(3)

 

ACKNOWLEDGEMENT

I would like to acknowledge my supervisor Associate Professor Bahri Uzunoglu for his supervision and guidance throughout this thesis. I would also like to thank Gotland University for the great support and facility provided during the study period.

My parents deserve infinite thanks for giving me moral values and always providing me with the warmest support in their most difficult times ever in this passing year. I would like to mention my special thanks to my family members, my grandmothers, my aunt, my uncle, and my nephews, especially Hilmi Yılmaz, Meral Yılmaz, Bengi Yılmaz and Seda Bengi for their efforts to pursue this degree. I gratefully thank my friends even we have not been side by side during a whole year, Türkan, Dilek, Zeliha and Münteha. I also would like to express my thanks to my classmates for experiencing one year study period together.

Finally and most specifically, I owe my deepest gratitude to my best friend, helpmate, my spouse who has been sharing my stressful, desperate and the most enjoying times ever in my life besides encouraging me since the beginning of this experience by heart, Müfit. Without their encouragement, I would not have finished this degree.

Visby-Ankara Eftun Yılmaz, 2012

(4)

 

TABLE OF CONTENTS

ABSTRACT ...i 

ACKNOWLEDGEMENT ... ii 

TABLE OF CONTENTS ... iii 

LIST OF FIGURES ...vi 

LIST OF TABLES ... x  CHAPTER 1 ... 1  1.1 Introduction ... 1  1.2 Question Formulation ... 2 1.3 Scientific Methodology ... 2  1.4 Thesis Outline ... 2 

1.4.1 Chapter 2 Optimization Algorithms and Wakes Analysis ... 2 

1.4.2 Chapter 3 Methodology, Software Tools and Optimization Modules... 2 

1.4.3 Chapter 4 and 5 Experimental Tasks with Numerical Results, Discussion and Conclusion ... 3 

CHAPTER 2 ... 4 

Optimization Algorithms and Wakes Analysis ... 4 

2.1 Wind Park Layout Design Using Optimization Algorithms ... 4 

2.1.1 Overview to Objective Function... 5 

2.2 Heuristic Optimization Methods - Greedy Approach... 6 

2.2.1 Greedy Algorithm in Simple Cases ... 7 

2.3 Wakes Analysis ... 7 

2.3.1 Wake Models in WindPRO and openWind ... 9 

2.3.1.1. N.O Jensen (Original Park Model) ... 10 

CHAPTER 3 ... 12 

Software Tools, Methodology and Optimization Module Working Principles ... 12 

3.1 Software Tools and Optimization Modules ... 12 

3.1.1 WasP Utility Software ... 12 

3.1.2 WindPRO Utility Software ... 13 

3.1.2.1 WindPRO Optimize Module ... 14 

3.1.2.2 Optimization Algorithms in WindPRO ... 15 

3.1.2.3 Calculation Methods for Optimization in WindPRO ... 18 

3.1.2.3.1 Regular Pattern Model A ... 18 

3.1.2.3.1.1 Automatic Way ... 18 

3.1.2.3.2 Random Pattern Model B ... 19 

3.1.2.3.2.1 Automatic Optimization ... 19 

3.1.2.3.2.2 Wind turbine generator options ... 20 

3.1.2.3.2.3 Optimize Options ... 21 

3.1.2.4 Methodology of Simulation for WindPRO in line with WasP ... 23 

3.1.2.4.1 Project Properties ... 23 

(5)

  3.1.2.4.1.3 Background Map ... 24  3.1.2.4.1.4 Terrain Map ... 26  3.1.2.4.1.4.1 Line Object ... 26  3.1.2.4.1.4.1 Topography Concept ... 27  3.1.2.4.1.4.1.1 Height Contours ... 27  3.1.2.4.1.4.1.2 Roughness Lines ... 28 

3.1.3 AWS openWind Utility Software ... 29 

3.1.3.1 Optimization Algorithm in openWind ... 29 

3.1.3.2 Methodology of Simulation for openWind in line with WasP ... 33 

3.1.3.1.1 Globals ... 33 

3.1.3.1.2 Turbine Type ... 33 

3.1.3.1.3 Layers ... 34 

3.1.3.1.4 Loading Wind Resource Grid File ... 36 

3.1.3.1.5 Search Order Property ... 37 

3.1.3.1.6 Energy Capture ... 39 

3.1.3.1.7 Optimize Options ... 39 

3.2 Methodology ... 41 

3.2.1 Definitions of File Types ... 41 

3.2.1.1 File Formats ... 41 

3.2.1.2 Other Relevant File Formats ... 43 

3.2.2 Establishing Files within WindPRO in line with WasP ... 43 

3.2.2.1 Setting up Map Projection and Datum with Terrain Model ... 43 

3.2.2.1.1 Roughness ... 45 

3.2.2.1.2 Orography ... 45 

3.2.2.1.3 The Challenges of Creating the Map File ... 46 

3.2.2.1.4 TIN ... 47 

3.2.3 Wind Data Analysis ... 51 

3.2.3.1 Setting up Wind Statistics Data in WasP in line with WasP Climate Analyst 2.052  3.2.3.2 Wind Speed Distribution in WindPRO ... 55 

3.2.3.3 The Challenges for Selecting Wind Statistics ... 56 

3.2.4 Wind Resource Grid ... 60 

3.2.4.1 Setting up Wind Resource Grid ... 61 

3.2.5 Wind Turbine Generator and Hub Height ... 64 

CHAPTER 4 ... 66 

Numerical Results and Benchmarking of Optimization ... 66 

4.1 Optimization Constraints and Parameters used in Analysis ... 66 

4.1.1 The Anticipated Optimal Task Results from Literature Research ... 67 

4.1.2 Case A Rectangular Area ... 68 

4.1.2.1 WindPRO - Random Pattern Model B - Automatic Optimization ... 69 

4.1.2.2 Regular Pattern Model A- Automatic Optimization... 71 

4.1.2.3 openWind-Automatic optimization ... 74 

(6)

 

4.1.3 Case A. Rectangular Area ... 77 

4.1.3.1 WindPRO- Random Pattern Model B - Automatic optimization ... 77 

4.1.3.2 Regular Pattern Model A- Automatic optimization ... 79 

4.1.3.3 openWind-Automatic optimization ... 80 

4.1.3.4 Regular Pattern - Automatic optimization ... 82 

4.2.4 Case B Quadratic Area ... 82 

4.2.4.1 WindPRO- Random Pattern Model B - Automatic optimization ... 83 

4.2.4.2 Regular Pattern Model A- Automatic optimization ... 85 

4.2.4.3 openWind-Automatic optimization ... 87 

4.2.4.4 Regular Pattern - Automatic optimization ... 90 

CHAPTER 5. ... ..93 

Discussion and Conclusion ... 93 

References ... 99 

APPENDICES ... 101 

Appendix A ... 101 

Wind Energy ... 101 

A.1 Numerical Modeling of the Wind ... 102 

A.1.1 Linear Models ... 102 

Appendix B ... 103 

Coordinate Systems Overview ... 103 

B.1 Basic Coordinate Systems ... 103 

B.2 Global Coordinate Systems ... 103 

B.3 Map Projections ... 104 

B.4 Geodetic Datums ... 105   

(7)

 

LIST OF FIGURES

Figure 2.1 Summation sample of Greedy algorithm……… ... …….7

Figure 2.2 In this picture placement of total 15 number of turbines placed regarding to the rule of thumb with a typical pattern was shown. The turbines in white dots are placed representatively 7 diameters (between 5-9) apart in the prevailing wind direction, and 4 (between 3-5) diameters apart in the direction perpendicular to the prevailing winds the same as indicated before [23]……….8

Figure 3.1 Main tab for both algorithms; Random and Regular...…...15

Figure 3.2 Placement of wind turbines in optimum locations. (a) Optimize heuristic algorithm makes the decision of best choice at that time at each step without regarding future consequences. Next it places another wind turbine in best of still available locations. After that it tries moving the first wind turbine to find a better total solution. However the objective value of solution is 7 rather than it should be optimal as shown in 3.2(b)……..16

Figure 3.3 WindPRO Optimize module properties (a) Flow chart of Optimize (b) Optimized farm diagram with random pattern………...17

Figure 3.4 The main tab for regular pattern model A and various park design objects…. 18 Figure 3.5 Define calculation window shows random optimize pattern i.e. wind turbine generator type choice [7] ...…...20

Figure 3.6 Optimize calculation options window under random pattern (a) “Allow auto creation of new wind turbine generators” does not checked. Also, wind turbine generator type was selected via the “Browse wind turbine generator s” button which we have chosen Enercon 40.3 with specific hub height 55m in this case. (b) “Allow auto creation of new wind turbine generators” option was checked then the first and third options were tried under that flap [7]………..….21

Figure 3.7 Optimize options……… …..22

Figure 3.8 The term "Project Properties" is a generic term covering all information regarding to site description, coordinate system, maps and addresses [8]… 24 Figure 3.9 Coordinate system; datum and time zone selection………24

Figure 3.10 (a) Site selection (b) and map screen shot from project properties flap within zone……….24

Figure 3.11 Terrain model from left top to right bottom within height contours, TIN triangles, elevation and slope of the areas. [30]……….. 28

Figure 3.12 Line object window with purpose “height contour option” [author]. ………. 29

Figure 3.12 (a) Link to line objects option was selected (b) Line object window with purpose “roughness lines” option ………..……….……….….. 29

Figure 3.13 Greedy algorithm of Optimizer in openWind ………...……….….32

Figure 3.14 An example for global values settings window...…...33

(8)

 

Figure 3.16 (a) Tree-view hierarchy (b) Interpretation flap under layer properties……...35

Figure 3.17 (a) Single Point *.wrg under met mast layer dialog window (b) Format of wind frequency data as*.tab and selected coordinates of single point for met mast data created as *.rsf file first……….. ….37

Figure 3.18 A sample for layer hierarchy showing a “Site Layer” and several elevation layers………..…..38

Figure 3.19 Showing rearranged layer hierarchy [9]………...38

Figure 3.20 Energy capture settings………..39

Figure 3.21 Optimizer options window with relative criteria………... 40

Figure 3.22 WasP Coordinate system settings (a) Map projection settings (b) Spatial view from map window………...45

Figure 3.23 First map file was created in zone 33 with UTM WGS 84 datum 10x10 km2 according to WasPdale *.map like in WasP sample files………..……...47

Figure 3.24 TIN flap under line object screenshot from WindPRO...…...48

Figure 3.25 *.map file formats (a) The first map exported from WindPRO (b) The simpler format trial by comparing with exported *.wpo file format.(The rows represents Easting and Northing coordinates with 6 and 7 digits respectively.) ………...49

Figure 3.26 Spatial view of map file in zone 34 with UTM WGS 84 datum system with no data grid………50

Figure 3.27 Randomly created quadratic map in WindPRO with line object………..51

Figure 3.28 Four wind turbine generator implemented onto the terrain (a) The map created due to first data format in WindPRO with quadratic design and its screen shot in WindPRO (b) The second map file which was created manually and modeled on WasP with quadratic design and its screen shot after import it into WindPRO……….52

Figure 3.29 Time series data for one year period “1990-01-01; 00:00 –1990-31-12; 21:00” ………...54

Figure 3.30 One year wind climate data created with time series………...54

Figure 3.31 Observed wind climate grid with fitted Weibull distribution from west…….55

Figure 3.32 Anemometer’s location with elevation above surface level value of 0 m...56

Figure 3.33 Standard roughness lengths and classes in terms of height variations in wind atlas window………...56

Figure 3.34 Site data window shows wind statistics selection……….57

Figure 3.35 Variety of wind statistics data on the map regarding to countries……... ….57

Figure 3.36 (a) Met mast location within latitude and longitude coordinates (b) Undesired wind atlas generation selected from for R-class 1………58

Figure 3.37 A sample window from Meteo object data with (a) Weibull table and (b) data setup flap in detail………59

Figure 3.38 (a) Observed Mean Wind Climate data which was imported to WasP inside workspace hierarchy (b) Resource grid statistics………...60

(9)

 

Figure 3.39 (a) Weibull settings and power densities according to constant (18m/s) wind

speed case in WasP observed wind climate and (b) WindPRO Meteo object..61

Figure 3.40 (a) An example of *.rsf file and (b) its explanation column by column [8]… 63 Figure 3.41 The resource grid window settings shows that the height of met mast………..……….63

Figure 3.42 Statistical parameters show the implemented data………. 64

Figure 3.43 Workspace hierarchies for all processes………..64

Figure 3.44 A configuration of resource grid parameters with the grid spatial view with boundary and node coordinates………...65

Figure 3.45 Wind turbine generator settings with (a) power curve and (b) other properties. ………66

Figure 3.46 openWind (a) turbine specifications and (b) modified power curve……….. 67

Figure 4.1 Results according to Mustakerov’s and Elkinton’s papers [1,4]; (a) optimized farm layouts for simple farm design (b) wind turbine placement figure regarding to predominant wind direction from north; Nx and Ny are the number of cells in the grid (c) description of coordinate system [1,4]……… 70

Figure 4.2 Task A.1.1 Wind turbines placement toward the long side of rectangular wind park shape…..……… 71

Figure 4.3 Task A.1.2 Wind turbines placement toward the long side of rectangular wind park shape, grid step shows resolution 20m……….. 72

Figure 4.4 First row for 20m and second row for 40m resolution cases………. 73

Figure 4.5 Grid step shows resolution 40 m……… .74

Figure 4.6 Park design object minimum distance criteria [1]……… 74

Figure 4.8 The optimizer tries to pack many turbines (20) even there is 33.4 % decrease in the efficiency of the wind park……… 75

Figure 4.7 The green turbines represents positions of initially defined turbines which were determined by park design object before………. 75

Figure 4.9 Optimal layout of 6 turbines in total where the wind direction is from north as shown in openWind figures with blue triangle. ………76

Figure 4.8 The optimizer tries to pack many turbines (20) even there is 33.4 % decrease in the efficiency of the wind park………. 76

Figure 4.10 (a) Optimization progress optimal layout for 6 turbines in total where the wind direction is from north as shown in openWind figures with blue triangle (b) Progress during energy capturing………. 76

Figure 4.11 Optimal layout for 3 turbines in total where the wind direction is from north as shown in openWind figures with blue triangle……….78

(10)

 

Figure 4.13 Optimal layout for 10 turbines in total……….. 80 Figure 4.14 Optimal layout results regarding to 40m and second one for 20m resolution

cases respectively……….. 80 Figure 4.15 (a)Park design object minimum distance criteria (b) Optimal layout for green

turbines(14) within the limits determined in park design object at (a)..…81 Figure 4.16 The optimizer placed more turbines (20) and the efficiency of the wind park

decrease 1.1% according to (14) turbines case……….82 Figure 4.17 Optimal layouts for 7 turbines in total where the wind direction is from north as shown in figure with blue triangle………... 83 Figure 4.18 Optimal layout for 10 turbines in total where the wind direction is from north as shown in figure with blue triangle……… 83 Figure 4.19 Optimal layout for 14 turbines in total where the wind direction is from north as shown in figure with blue triangle………. 85 Figure 4.21 In total 22 turbines scattered in quadratic area within 20m resolution……… 87 Figure 4.22 Optimal layout results auto fill, 40m and 20m resolution cases respectively...87 Figure 4.23 (a) Park design object due to minimum distance criteria (b)The green

turbines(14) which represents geometry of initially defined turbines within the limits determined in park design object……… 88 Figure 4.24 The optimizer tries to pack many turbines (99) in possible limits of optimizer

controller and the efficiency of the wind park decrease to 57, 3% ………. 91 Figure 4.25 Automatically optimized turbines after optimization within 518 iterations

where the wind direction is from north as shown in figure with blue triangle. ………92 Figure 4.26 Turbine coordinates flap shows the distance between closest turbine locations

which are overlapped in this case where the wind direction is from north as shown in figure with blue triangle……… 92 Figure 4.27 Automatically optimized turbines within 501 iterations for 20m resolution case where the wind direction is from north as shown in figure with blue triangle…………..93 Figure 4.28 Initially defined 14 turbines………...94 Figure 4.29 Optimal layout of 14 turbines………94 Figure A.1 Atmospheric circulation of air. The arrows between the latitude lines indicate the direction of surface winds. The closed circulation or convection shown on the right indicates the vertical flow of air [14]………106 Figure B.1 Global Coordinate Systems (a) Geographic and (b) Cartesian coordinates [12] Figure B.2 UTM system [7]………...107 Figure B.3 Reference ellipsoids……….108 Figure B.4 Different datums………...…………108

(11)

 

LIST OF TABLES

Table 3.1 WasP terrain map file format definitions [8]………. 47 Table 4.1 In total 20x20, 7x2 and 5x4 total number of turbines were applied respectively in

quadratic area……… 89 Table 4.2 Case A.1, tasks A.1.1, A.1.2 and A.1.3……….94 Table 4.3 Case A.2, tasks A.2.1, A.2.2 and A.2.3……….95

Table 4.4 Case B.1, tasks B.1.1, B.1.2 and B.1.3 where Nx and Ny represents the easting and northing coordinates………95 Table 5.1 Optimal task analysis results in terms of, cost function, park efficiency, power production and total number of turbines (a) Case A.1 (b) Case A.2 (c).Case B……..98

(12)

 

CHAPTER 1

1.1 Introduction

The main challenge for designing a wind park is to maximize the energy capture within the given restrictions. There is a common desire to build tighter wind farms to optimize land utilization, so determination of wake losses for very close turbine spacing is utmost importance for optimization [1].

Development of cost effective wind parks concerns choice of the turbines type, number and their placement in the wind park area. Accordingly, selection of the total number and type of the turbines for sitting depends on wind conditions, terrain, investments costs, power output, park efficiency, geographical locations and others [1]. The placement of wind turbines in the layout is influenced by the turbine size, wind direction, wake interactions between wind turbines and land availability. Such as for a typical uniform wind direction indicates equal distances between turbines in rows and columns of approximately 5 rotor diameters [5]. However, for the case of predominant wind direction, Grady et al. [5] recommended turbines spacing as 8 to 12 rotor diameters apart in the downwind direction, and 1.5 to 3 rotor diameters apart in the crosswind direction [1]. The main goal of intended study is to identify the differences in turbine configurations of optimal layouts in two wind farm design software tools in comparison to available literature results. For the design of wind park layout, energy optimization objectives were implemented in algorithms of openWind and WindPRO software tools of industry. The optimization modules of these software tools do not include separate cost estimation to the best of our knowledge. The solution of optimal task in these software tools is implemented solely on the overall number of turbines. The main part of the total cost/year for the entire wind park includes operation and maintenance costs with turbine investment costs. However, operation and maintenance is neglected in this study in-line with the limitations of the compared software tools.

The purpose of this work concerns evaluation of various optimizers working principle individually in each program in terms of layout differences under the same conditions. The wind climate data based on fixed wind speed time series were created to generate a similar wind distribution throughout the entire site for a single direction and single wind speed. So that, the wind set up with same boundary layer was implemented during the run. However there were differences in basic fit of the distributions to the fixed wind speed climate data since the distribution fit in both software tools were not designed for this data which might have impacted the results. To sum up, the scope of this work is to design a controlled

(13)

 

experiment for benchmarking two different software to each other namely OpenWind and WindPRO.

1.2 Question Formulation

What is the difference between optimization modules for two wind farm design software tools in terms of wind turbine configurations regarding to similar constraints and parameters?

1.3 Scientific Methodology

 Theoretical Part covers theory about wind farm configurations and optimization algorithms based on literature study.

 Simulation Part includes the work based on data created in various file formats for different software tools.

 Experimental and Analysis Part presents the selection of data and then the simulation results and covers the comparison between optimization results with both resource grid models.

1.4 Thesis Outline

In this section, each chapter will be briefly described in terms of content, aim, importance and basis for this study.

1.4.1 Chapter 2 Optimization Algorithms and Wakes Analysis

In this chapter the basic theory behind optimization will be described. Furthermore, the relation between wind farm micrositing and wake effect will be stated briefly.

1.4.2 Chapter 3 Methodology, Software Tools and Optimization

Modules

Here, software tools and optimization modules are described regarding to their working principles. WasP 10 is used for the preparation part; EMD WindPRO 2.7 and AWS openWind community edition are used in the simulation part. Methodology describes the

(14)

 

simulation process in terms of input files, test tasks and output. The input is the wind data and terrain file prepared in WasP utility software. The experimental tests are described in terms of constraints for the tasks’ cases that are how the simulation is conducted in openWind and WindPRO. Afterwards, the results are reviewed.

1.4.3 Chapter 4 and 5 Experimental Tasks with Numerical

Results, Discussion and Conclusion

In the tasks each simulation case will be analyzed and described. The results display the simulated cases. The results are reviewed in context theory in discussion and summarized in conclusion section.

(15)

 

CHAPTER 2

Theory of Optimization Algorithms and Wakes Analysis

This chapter presents the theory behind optimization algorithms and the wake effect during micrositing of the wind turbines in a wind farm layout.

2.1 Wind Park Layout Design Using Optimization Algorithms

In general, wind farm layout optimization has been a subject of research for almost 15 years. Although there are many factors for positioning of turbines in onshore, simple and flat terrains like offshore wind farm sites have fewer types of problems while configuring the design of the farm. The mostly discussed optimization algorithms which were utilized in offshore turbine micrositing are classified as, genetic algorithms, greedy heuristic methodology, swarm optimization, gradient search, simulated annealing and evolutionary algorithms in terms of boundaries of the problem, such as minimization of cost functions and maximization of energy production regarding to wake models [4]. Furthermore, the optimization problem of the wind turbine distribution at a given site must consider interaction of the wind turbines sizes plus direction and intensity of the wind.

One needs necessary algorithms with advanced software tools to develop reasonable cases in sufficient time periods. Such as for many existing computers, solving millions of possibilities for simple cases are very time consuming. For example, if each cell in the farm grid can have two possible states that contain a turbine or does not contain a turbine then even for a small 10 x 10 grid, there are 2100≈1030 possible cases to evaluate [25]. Therefore, various kinds of algorithms have been prepared for solving these possibilities with the wind farm performance evaluation and for the optimization procedure. The reason why mostly from these algorithms; the greedy heuristic, genetic and combinatorial algorithms; that they have been selected for implementation, because they are well suited to the problems for micrositing of wind farms and they are easy to implement within objective functions.

Similar to the current project Grady S.A. [2] et al. and Wagner M. [5] et al. proposed various optimization schema based on genetic, heuristic and combinatorial approaches by considering conditions as coming wind within uniform unidirectional and uniform with variable direction. In that researches the investment cost and the total power extracted were

(16)

 

the variables optimized. For instance, in study of Elkinton C. [4] et al. different objective functions for genetic and greedy heuristic algorithms to utilize in wind farm micrositing were implemented and evaluated through design simulations.

On the other hand, in the study of Mustakerov, I. & Borissova, D. [1] combinatorial model designed for wind park area shape, size, orientation and different requirements and restrictions for defining wind turbines type, number and placement is proposed.

In this part of this study, commonly implemented objective functions for various optimization cases for simple wind farms were mentioned. Additionally, the algorithm implemented in software is predicted to be based [3, 4, 25] on heuristic (greedy approach) algorithm which will be mentioned later.

2.1.1 Overview to Objective Function

The choice of an appropriate objective function is critical to solve an optimization problem. The quantity was chosen to optimize could be different in various optimization types. Mostly it was done by developing an objective function which is tried to be maximized or minimized. The simplest solution was to either maximize the farm energy production or minimize the cost of energy. Appropriate candidates for optimization objectives include;

• Maximization of energy production; this approach has been used for several commercially available optimization software packages. It delivers the most energy to the grid, but does not account for the cost of that energy.

• Maximization of profit; this approach yields the greatest benefit to the farm owner and investors. The disadvantage of this approach is that, it requires knowledge of the value of the electricity on the grid and the inclusion of tax incentives, renewable energy credits, and any other economic incentives.

• Minimization of the cost per unit energy; this approach is a balance between maximum energy and minimum cost.

These models are now utilized as part of the objective function for each wind farm design that the optimization algorithm considers [2]. A widely accepted criterion for wind park design is the minimum cost per unit energy produced ratio that is calculated in this study. So that optimization proceed based on following objective function [1],

[Cost] P

(17)

 

Where cost is described above equation (2.1) and P is total power output from all of the N turbines in the entire wind farm. Power is called as “Park Yield” which includes the wake losses in energy calculation reports of WindPRO and openWind.

According to Mosetti et al., the investment cost of the wind turbines is modeled in such a manner that only the number of turbines need to be considered in calculating the total cost. Mosetti et al. assumed that the non-dimensionalized cost/year of a single turbine is one with a maximum reduction in cost of 1/3 for each additional turbine, provided a large number of them are installed. Consequently, it can be assumed that the total cost/year for the entire wind park can be expressed as in following equation (2.2) [4],

Costs= 2/3+ 1/3exp (-0.00174N^2)] (2.2)

The intended experimental study considers the control parameters as constant wind conditions and a constant wind turbine type. Additionally, the solution of optimal task was implemented, by neglecting the operation and maintenance costs which poses drawbacks in terms of the total investments, solely on the overall number of turbines which is the main part of capital cost requirement.

2.2 Heuristic Optimization Methods - Greedy Approach

A greedy algorithm is an algorithm that follows the way of heuristic technique to find an approximate global optimum by making the locally optimal choice at each stage. Both genetic and greedy heuristic algorithms are further evaluated through the use of design simulations.

These algorithms are working on partial solutions and recursion. First one solution is constructed at a time, either incomplete solution to the original problem or a subset of the original problem’s search space with a particular property hopefully shared by the real solution. Then complete solution to a reduced problem by decomposing the original problem into simpler problems and solving these problems are done. Last step is that trying to combine the partial solutions into a solution to the original problem. In other words, call to the same procedure to solve the problem for the smaller size sub input(s).

It is a easy to implement algorithm that constructs the solution step by step. At each step the value for one decision variable is assigned by making the best available decision. Heuristic is needed for making the decision of the best at each step of optimization regarding future consequences. The best ‘profit’ is chosen at every step which is so called the algorithm i.e. ‘Greedy’. So, one cannot expect the greedy algorithm to obtain the overall optimum. In other words, the algorithm makes an assumption that choosing a local optimum at each step, one will end up at a global optimum. As far as explained the algorithm iteratively makes one greedy choice after another, reducing each given problem

(18)

  into is fa algor possi

2.2.1

The will softw In fa of st samp expla As it other greed choo conta

2.3

The p sever direc insta they are in a smaller o aster than o rithms can ible solution

1 Greedy

basic idea be useful f ware tools (o act, this heur teps; finding ple is a sim ained in foll t is indicate r examples dy algorithm ose 12 inste ains 99 [10]

Wakes A

placement o ral factors w ction, wake alled in a w are position nstalled. Co ne. In other other optim fail to prod n [26].

Algorith

behind gre for understa openWind) ristic need n g an optim mple version lowing exam Figure 2.1 ed in Fig. 2 available as m will choo ead of 3 at ].

Analysis

of wind turb which have effects and ind farm, m ned. In othe onversely, i r words, a g mization me duce the bes

m in Simp

edy algorith anding how in later cha not to find a al solution n of this pro mple, 1 Summatio 2.1, it is a s well. With ose what ap the second

s

bines on the e to be taken d availability most of them er words, w if we are in greedy algo ethods. So st optimal s

ple Cases

hms was ex w the optim apter. a best soluti typically r oblem whic on sample o summation h a goal of r ppears to be d step, and e wind park n into acco y of the land m are impac wake effect i nstalling few orithm never that, for m solution and

s

xplained wi mizer works

ion but term equires unr ch covers a f Greedy al n example f reaching th e the optima d will not r k site (i.e. w unt such as d. For insta cted by at l increases w w turbines, r reconsider many other d may even ith some sa s for wind minates in a reasonably m part of the gorithm [35 for greedy a e largest-su al immediat reach the be wind park la s the numbe ance, if there least one w

ith the num it is possib rs its choice r problems, n produce th amples belo farm layou a reasonable many steps solution. It 5]. approach; t um, at each te choice, s est solution ayout) is aff er of turbin e are many wake regardl mber of turb ble that only

es and it , greedy he worst ow. This ut design e number s. Below t will be there are step, the so it will n, which fected by es, wind turbines less how ines that y few or

(19)

  none perpe By k produ or w Indee slow thum diam cross wind manu anoth conte are e Park The energ wind exist direc This place wind Figur e of them a endicular to keeping the uced, wind ind park. A ed, there is wed down, in mb, turbines meters apart swind direc d rose, the ufacturers o her. Typica ent is calcul expressed th Model) for separation gy behind th d park depen t some reco ctions and ro work assum ement issue d turbine lay re 2.2 In th rule o placed are affected o the wind d used land turbines sh A wind turbi s a wake be n comparis s in wind p in the prev tion as indi e Weibull or developer ally, the arra

lated throug hrough the J r both Wind distance be he neighbor nds on the t ommendatio otor diamet med a contr e which wil yout is not v his picture p of thumb w d representa by wake, direction [6] area at a m hould be put ine always ehind the tu on with the parks are us vailing wind cated in Fig distribution rs can calcu ay loss will gh WindPR Jensen/Risø dPRO and o etween the ring turbine terrain, the w ons for the t

ers sizes. ol experime ll be explai valid for this

placement o ith a typica atively 7 di for instanc ]. minimum w ut together in exposes to urbine that e wind arriv sually space d direction, g. 2.2. With n and the ulate the ene

l be somew RO using the ø single wak openWind. turbines d es [5]. The s wind direct turbines sep

ent with con ined in Ch s study. f total 15 n al pattern w iameters (b ce, if they a while increa n a group, w a wind sha is a path o ving in fron ed between , and betwe h knowledge roughness ergy loss du where aroun e Wind Atla ke model th depends on spacing of a tion and spe paration dis nstant wind apter 4. So number of tu was shown. between 5-9 are aligned asing the am which is ca ade in the d of wind is nt of the tu n approxima een 3 and 5 e of the win in the d ue to wind tu nd 5 % [21] as method [ hat is (N.O. the needed a number of eed on the tu stances depe direction fo o, a standar urbines plac The turbine 9) apart in t in one row mount of el alled as a wi downwind d quite turbu urbine. As a ately 5 and 5 diameters nd turbine r different di urbines shad ]. The wind 8, 18]. Arra Jensen i.e. d recovery f wind turb urbines’ siz ending on t or illustratio rd configura ced regardin es in white the prevaili w that is lectricity ind farm direction. ulent and a rule of d 9 rotor apart in rotor, the rections, ding one d energy ay losses Original of wind ines in a ze. There the wind on of the ation for ng to the dots are ng wind

(20)

 

direction, and 4 (between 3-5) diameters apart in the direction perpendicular to the prevailing winds the same as indicated before [23].

 

The software optimization modules in line with wind farm design tools in industry attempts to efficiently place the turbines in locations with highest winds while adhering to the constraints and optimizing stated objectives functions. As indicated before, the objective function considers different aims in line with the layout design. One of them is the profitability of the wind park design by minimization of the investments while maximize the wind park electrical power generation. Thus, the solution of the optimization task will be a compromise between the overall wind park investments costs and wind turbines energy output. Both depend on the type of wind turbines, on their number and placement within park area. On the other hand, the number of the wind turbines depends on the size of the wind park and wind direction.

In this study, the proposed results of experimental tests are analyzed in terms of, overall number of turbines, turbine positioning, number of iterations, layout differences, park efficiency and cost analysis. Therefore, the tasks solutions’ results describe the difference of optimal wind park layout designs for two software tools. Wind park layout configuration variations on the wind park power output for each program were evaluated individually in next chapters.

2.3.1 Wake Models in WindPRO and openWind

Wind flow calculations of a given wind farm is performed by many commercial software tools and empirically determine the installation positions of turbines based on the flow field. The flow field usually does not include the influence of turbines on the deflection of the original air flow that is wake effects. However, as the wake effects are complicated and strongly coupled, they play a crucial role in wind farm micrositing as indicated above. So, the solutions were empirically done by ignoring the wind profiles.

Current industry standard procedure for predicting the wake loss within wind farms includes using two of the three wake models implemented in WindPRO and openWind. These two models are the N.O. Jensen Park model and the Eddy Viscosity model by Ainslie [24]. A lesser known model developed by G.C. Larsen is the third wake model also available in WindPRO. The N.O Jensen [6] model is chosen for this experimental study. Therefore, this section describes this semi-empirical combination method in order to calculate the combined effect of multiple wakes [24].

(21)

 

2.3.1

A sim a win facili mom deter descr Figur At th down dista turbi are c affec

1.1. N.O J

mple physic nd farm was itate the op mentum and rmine the w ribe the win

re 2.3 (a) S he turbine, t nstream, the ance, x, as s ine is a func characteristi cting the pow

Jensen (O

cal model fo s developed ptimization d only take wind speed nd speed do Schematic o mean wind downstream the entrainm software to the wake ha e radius of shown in Fi ction of loca ics defining wer extracte

Original P

or the calcu d by [Jensen of turbine es into acco immediatel wnstream o (a) 0 1 (b) of wake mo d speed, “a m of the tur ment consta ools [7]. as a radius e the wake, r ig.2.3 (a) ab al wind spee g the local ed are hub h

Park Mod

ulation of th n, 1983] and e layout. T ount wind ly behind t of the turbin ] odel where a” is the ax rbine, r1 is t ant which is equal to the r1, increases bove. The p ed. Directio wind field height, roto

el)

he energy ou d refined by This model speed loss the rotor fo ne; (b) Equati xial inducti the downstr s so called w turbine rad s linearly, p power extra on, intensity d. Particular r diameter, utput from m y [Katic et a is based o ses by usin llowing exp ion (2.3) sh on factor, “ ream rotor r wake decay dius, rr. As th proportional acted from t y, and probab r aspects o and thrust c multiple tur al., 1986] in on conserv ng Betz the pression de hows that; u “x” is the radius, and y constant ut he wake pro l to the dow the wind by ability of occ of the wind coefficient [ rbines in order to vation of eory. To erived to (2.3) u0 is the distance alpha is tilized in opagates wnstream y a wind currence d turbine [5].

(22)

  Figur Ther the a tests, re 2.4 Sele open calcu reby, the po analysis of c , uncertainti (a) ected Wake nWind (b) ulation defi ositioning of configuratio ies can be ig e Parameter Wake dec initions win f turbines is ons. Additio gnored. (c) rs in Softw cay constan ndow within s mainly aff onally, in lin ware tools ( nt window n model para ffected by th ne with the (b) (a) Wake m in openW ameters. he wakes w constraints model selec Wind (c) W while it is cr implemente ction for WindPRO ucial for ed in the

(23)

 

CHAPTER 3

Software Tools, Optimization Module Working

Principles and Methodology

In this study, simplest test cases were chosen to visualize two dimensional cases. It should be noted, that the turbines and parameters are just a sample used to illustrate the comparison between two software optimization modules. Real design problems would include much more design items leading to large scale optimization tasks formulations which are not the scope of this work.

3.1 Software Tools and Optimization Modules

 

Different wind flow modeling software tools are used in this study for the simulation process. These are, WindPRO, WasP and lastly openWind. WindPRO uses WAsP calculations and an internal WAsP engine included in the WindPRO installation. It is important to be aware that the WAsP default parameter settings are a property of the WAsP software [7]. On the other hand, openWind calls WasP for the calculations, too. Also the same file formats that were established in WasP are used for both softwares. The analysis of the optimization module algorithms for WindPRO and openWind will be explained under the headings below.

3.1.1 WasP Utility Software

WAsP, Wind Atlas Analysis and Application Program, is the wind simulation software based on the linear model. The software was initially developed in the Risø National Laboratory in Roskilde, Denmark [8]. WAsP is today developed and distributed by the wind energy division at Risø DTU and has more than 2900 users worldwide, and can be used for various purposes such as;

• Wind farm annual energy production and efficiency estimations, • Wind resource mapping,

(24)

 

Although, there is a similarity between WasP and the openWind, WasP is differentiated from other software tools that were implemented in this study, with its working principle. One should consider the workspace hierarchy that starts with 'WAsP project 1'. To work with WAsP, you need to add new tasks to the project that is the modeling tool under the workspace area. These items are arranged in a hierarchy and they are called hierarchy members or just 'members', for short. The workspace root is always at the very top of the hierarchy. All of the members of the workspace are children of the workspace root.

The additional tools inside WasP are Wasp Climate Analyst and Wasp Map Editor. These are used for the creation of wind distribution and map files respectively. The WAsP Map Editor tool is used for formatting, editing, checking and transferring the WAsP map files. Additionally, the WAsP Climate Analyst is used for constructing the Observed Mean Wind Climate files needed by WAsP and the Observed Extreme Wind Climate files needed by WAsP Engineering [8].

3.1.2 WindPRO Utility Software

WindPRO is the software tool used for designing and planning for single wind turbines and wind farms. The software has been developed by EMD International A/S in Aalborg, Denmark [7]. The software has a modular structure which means that only the needed modules have to be chosen and be paid for. The software can be used for various purposes such as;

• Digitalizing maps concerns information about orographical and topographical effects.

• Wind farm annual energy production and efficiency estimations • Simple energy estimations for single wind turbines and wind farms, • Noise and shadows generation calculations,

• Photomontages of wind farm layouts, • Optimization of a wind park.

(25)

 

3.1.2.1 WindPRO Optimize Module

WindPRO Optimize module optimizes the power output from a wind farm by adjusting the layout of turbines. The software requires the wind field to be known (it uses WAsP) if the terrain is at all complex. For simple terrains where the wind field is assumed to be uniform, for example offshore, the software only requires a wind speed distribution like in this study. The user specifies the project area and can specify required generating capacity or number of turbines to use and a minimum turbine separation. The software then places turbines as follows. The first turbine set a point where the wind resource is greatest. Subsequent turbines set in the next best locations considering the park efficiency. Wake losses are considered and incorporated. Overall distance between turbines can be manually updated. Turbines can either be laid out in grid like rows with set spacing or arbitrarily, subject to the constraints during the layout optimization. Currently in WindPRO the optimization combines the layout constraints in terms of spacing and land use with the wind resource available. This process continues until all wind turbine generators have been placed. The program has been structured in a way that allows a compromise between the level of optimization and the computational time. Optimization in WindPRO consists of four elements [7].

• Site constraints in the form of a wind turbine generator area object where available land, exclusion zones and spacing requirements are defined.

• A wind resource map either imported (created in WasP) or created with the RESOURCE module.

• Optimization algorithms, which depending on requirements, will offer a suitable optimization result.

• A dedicated optimization window where results are updated and the history is tracked.

The optimization module algorithms cover the following situations that are shown in below Fig.3.1;

(26)

  Rand place Regu the w

3.1.2

Acco heuri heuri the lo An e to mo dom patter ed into the t ular pattern wind farm m F

2.2 Optim

ording to th istic placem istic algorith ocal optima essential dra ove one win

rn: It is use terrain in an n: It is used must obey ge Figure 3.1 M

mization A

he literature ment algorit hm is being al solution n awback of th nd turbine a ed when geo n order with d for offshor eometrical c

Main tab for

Algorithm

e work the thm (simila g unable to f nor is it prov he algorithm at a time as ometry is n h locations d re wind farm constraints. both algori

ms in Win

predicted ar to the g find some o viding a me m is that it c illustrated i not a concer distinctly wi ms or simila ithms; Rand

dPRO

algorithm t reedy algor of the soluti asure of the cannot run in Fig.3.2. rn. wind tur indier than o ar uncompli

dom and Reg

that softwar rithm) [3]. ons and the e quality of fast because rbine genera other. icated terrai gular re Optimize The proble erefore neith the solution e, it is only ators are in where e uses a em with her fined n.[3]

allowed

(27)

  Figur The follo 1 2 3 re 3.2 Plac algo rega still a be it sh principle of ws; 1. Optimize calculate a wind t wind tur requirem which yi 2. It now producti various turbines 3. This con turbines cement of orithm make arding futur available lo etter total so hould be opt f Optimize e utilizes a ed wind reso turbine in t rbine in an ments and in ields the ma tries movin on of the t locations a are moved ntinues, wit are placed (a) wind turbi es the decis re conseque ocations. A olution. How timal as sho explanation a grid of p ource map, the best of ny of best n each test c aximum com ng the firs two. Once and position one at a tim th the poss or the algor ines in opt sion of best ences. Next After that it t wever the o own in 3.2(b n is comple possible loc within the these locat remaining calculating mbined prod st wind tur these two ned in the me to try and sibility to f rithm canno timum loca t choice at it places an tries movin objective va b). ementary wi cations, pre project site tions. There g locations, the array lo duction. rbine to po are set, a t best found d find a bett fix the plac ot find space ations. (a) that time at nother wind g the first w alue of solut

ith the flow

eferably co e area, and s eafter it trie obeying m osses, and c ossibly ach hird wind t d, then the ter total pro

ements unt e to introduc (b) Optimize h at each step d turbine in wind turbin tion is 7 rat w chart pres oordinates f starts off by es placing a minimum d chooses the hieve a bet turbine is t two existin oduction.

til all desir ce yet anoth heuristic without n best of e to find ther than sented as from the y placing a second distances location tter total tested in ng wind red wind her one.

(28)

  Figurre 3.3 Wi Opti sequ ndPRO Op imized farm uence of ord ptimize mo m diagram dering by th (a) (b) odule prop with rando he optimize erties (a) om pattern module.) Flow chart (Red arrow t of Optim ws are show mize (b) wing the

(29)

 

3.1.

In th

3.1.2

This desig possi one c of pr geom gene can b 3.1.2 After patte park the p insta newl regul geom F

.2.3 Calc

is study, tw

2.3.1 Reg

pattern all gn. Here, th ible as men can graphic redefined w metry with o rators can b be hold in a 2.3.1.1 Auto r establishm ern can be s efficiency percentage o ance 95% (m ly created l lar pattern, metry of win Figure 3.4 T

culation

wo methods

ular Patte

lows user w he park des ntioned belo cally move, wind turbine optimization be realized o new layer omatic Way ment of the p selected un can be limi of park effic max. 5% a layouts wil the optim nd turbine g

The main tab

Method

for WindPR

ern Mode

when strict ign object ow. Manual rotate and e generator n area objec or can be up [7]. y park design nder Optimi ited. The m ciency. It c array losses) l be neglec ization pro generators in b for regular

ds for Op

RO optimize

el A

geometrica is used. Bo optimizatio change dist rs inside th ct. Accordin pdated and n object, fro ize. Accord maximum ar an be chose ), which ca cted unless ocess differs

nto the layo

r pattern mo

ptimizati

er was used al requirem oth manual on means th tances whil he wind farm ng to the pr a number o om the defin ding to the rray losses en to ensure an be a des they meet s from “ran out, as show odel A and

ion in W

d as explaine ments are m and automa hat with the

e the progra m area wh roject constr of new wind ne calculatio configuratio will decrea e minimum sign criterio t this effici ndom patte wn in Fig.3.4 various park

WindPRO

ed below.

met with the atic optimiz park design am keeps a hich was de traints, wind d turbine ge on window on of the f ase while in m park effici on. In other iency criter ern” with th 4. rk design ob

O

e terrain zation is n object, a number efined in d turbine enerators ; regular farm, the ncreasing iency for r words, rion. For he fixed bjects

(30)

 

3.1.2.3.2 Random Pattern Model B

As it was explained in theoretical part before, this pattern was chosen when geometry is not a concern. Wind turbine generators are placed automatically to the positions distinctly windier than the other in the layout. A sample window for calculation is shown in Fig.3.5.

3.1.2.3.2.1 Automatic Optimization

The automatic optimization of the layout is done regarding to energy production within specified areas in terrain. Many constraints or assumptions can be specified under optimization area object. Before the Optimizer will start a wind resource map (*.rsf file) at a reasonable high resolution should be calculated. In fact, the height of the *.rsf file should be consistent with the hub height of selected turbines for ignoring extrapolation The optimization area object defines the areas for siting the wind turbine generators [7]. The limits of these areas can be digitized directly by using import coordinate files by defining the limits with *.dxf problems. Alternatively meteo object can be used for the calculations. AutoCad/Autodesk) format can save it as from WasP Map Editor Tool.

For each partial area of optimization area individual requirements can be set regarding [31];

 Number of wind turbine generators (min and max).  Total installed capacity (min and max).

 Minimum distance between two wind turbine generators grid cell size.

As mentioned before park design object allows users to design a wind farm layout with a strict geometrical regular pattern layout. The wind turbine generators inside the optimization areas are included.

For the selection of the wind resources, there are two options as it is indicated above Fig.3.4. It can be given by either a wind resource map file (*.rsf) or meteo object, by holding the wind distribution that is expected to cover the whole area of interest. The latter option is common for offshore projects unlike the onshore orography and roughness typically will result in varying wind distribution over the area.

One can choose whether the Optimizer should move already defined wind turbine generators inside the area to the best positions or let Optimizer create the new wind turbine generators. For this, optimizer needs a reasonable resolution value to create the desired number of wind turbine generators or the total capacity required if there is sufficient space. Otherwise it places as many turbines as possible. Unlike in model A, there is no alternative

(31)

  selec insid wind fram Figur For M detai 3.1.2 New If yo you treate    Final outsi in th

ction for res de a geomet dow the spa mework in te re 3.5 Defi gen Model B, tw il below; 2.3.2.2 Win wind turbi ou have alre can decide ed as shown Delete th Keep the wind tur their opt Keep the a project lly you can ide the calcu he optimizat solution bec trical frame acing param erms of row ine calculat nerator type wo different d turbine g ine generato ady created e how alrea n in Fig.3.6 hem and cre em and lock rbine gener timal positio

em, but mov t with differ n decide ho ulation area tion by mov cause that m e with spec meters whic , column, an tion window e choice [7] t options en generator o ors can be c d wind turbi ady created below; eate new wi k them to th ators are re ons). ve them to t rent wind tu ow to handl a (the grid sp ving them in model uses cific distanc ch represen ngle of side w shows ra . nable altern options created ine generato new wind ind turbine g heir present equired at f their optim urbine gener le already d pan of the * nside the * s already de ces. Just on nt the resolu e and base o andom optim ative select

ors (and thes d turbine ge generators f positions (t fixed positi ized positio rators types defined win *.rsf file). Y .rsf definiti efined wind ne can spec ution for th of the park d mize patter ion of turbi se are prese enerators on from scratch to be used in ons while n ons (this is t and sizes). nd turbine You can cho on area or t

d turbine ge cify from o hat task wit design objec rn i.e. wind ines as expl ent in visibl n the map h n cases whe new are ins the way to o generators oose to inclu to delete th enerators optimizer thin this ct. d turbine lained in e layers) shall be ere some serted at optimize that fall ude them hem. The

(32)

  probl outsi is use Figur 3.1.2 In op energ pack whic locat lem is that ide the *.rsf ed, one can

re 3.6 Opti crea gene whic (b) “ then 2.3.2.3 Opti ptimizer to gy layout a k as many tu ch is not a c tion with th t, it is not f definition choose me imize calcu ation of new erator type ch we have “Allow auto n the first an imize Optio place the and full ene urbines due cost effectiv e highest en possible fo area since t teo object. ulation optio w wind turbi was selecte chosen Ene o creation o nd third opti ons turbines th ergy layout e to the defi ve solution. nergy resou or the prog there is no w (a) (b) ons window ine generato ed via the “ ercon 40.3 of new wind ions were tr hree option . The first fined constra Due to sec urce accordi gram to inc wind data to ) ) w under ran ors” does no “Browse wi with specifi d turbine ge ried under th s are availa option was aints onto t cond option, ing to the w clude wind o work on. U ndom patter ot checked. nd turbine fic hub heigh

enerators” o hat flap [7]. able. These s not selecte the optimiza , WindPRO wind resourc turbine ge Unless the

ern (a) “All . Also, wind generator s ght 55m in t option was . e are auto ed often be ation area a O finds the a ce maps and enerators *.rsf file low auto d turbine s” button his case. checked fill, fast ecause it available available d place a

(33)

  turbi to pl defin wake turbi is rep accor Also unde their Num locke optim is fix follo their Num with posit wind solut Park the n calcu preci and 1 Layo layou ine. This me ace the turb ned grid. In e loss caus ine the first peated every rding to bes , when the er this wind explanation mber of win ed at a cer mizing a ver xed to 20 wing the pr positions in mber of win the above tion with re d turbine ge tion will be k direction number of c ulations. So ision for ea 1 m/s [7]. outs tested uts will be ethod does bines which n addition t ed by the turbine is r y time new st park effic option for dow. The ab ns were giv nd turbine rtain stage ry large win for instanc rocess and o n the layout nd turbine process 21, egard to the enerators af achieved w and wind calculations o that with t ch step afte before sto tested dur not take in h is fast and o the fast l predefined eplaced tak turbines are ciency. Figure optimizatio bove Fig.4.3 ven below [7 e generator for a total nd farm, the ce, it mean optimization t. generators , wind turb e total wind fter locking when the cos

speed step

have been the angle a er, for instan

p in full op ring optimi nto account d uses eithe layout, full turbine pla king into acc e placed. Th 3.7 Optimiz on selected 35 shows th 7]; rs to lock: l number o e process ca ns 20 wind n stops and s per run a ine generat d farm prod g however, st of layout p: This deter carried out and direction

nce, for the

ptimization ization for the park ef r grid of th energy opt acement. A count the w he turbines ze options. there is ano hat the prop

Layouts fo of wind tur an take ages d turbine g d 20 wind tu after lockin tor will be i duction. On it does not is taken und rmines the t. Each opti n determina e last park c n. This allow each new fficiency wh e wind reso timizes take After it has ake of the s previously other sub-o perties inclu or large win rbines. The s. Therefore generators h urbine gene ng the first inserted on ne can incre t mean that der consider level of pre mization re ation it sho calculation, ws users to wind turbin hen selectin ource map o kes into acc

placed the second turbi placed are optimize opt uded in this nd farms h e reason is e, when the have been erators are lo t ones: Acc the best re ease the nu t a more op eration. ecision rega equires a nu ould show a in step of o specify ho ne generato ng where or a user ount the e second ine. This replaced tions tab tab plus has to be s, if one e number selected ocked to ordingly emaining umber of ptimized arding to umber of a highest 1 degree ow many or to be

(34)

  inclu layou produ full p layou possi steps the la

3.1.

Wa

This with proce

3.1.2

Ther insid mark proje impo Figur uded. Befor uts inside o uctions con park calcula uts, the calc ibilities but s because ha ayouts comp

.2.4 Met

asP

part descri explanation ess, from pr

2.4.1 Proj

re are some de WindPRO king an app ect propertie orted throug re 3.8 The reg re each opti optimizatio nsidering a s ation and th culations wi t slow dow aving accur plexity.

hodolog

ibes how th ns below pr reparation to

ject Prope

properties O, the first s proximate s es with rele gh “Backgro e term "Pro garding to s imization ru n area und simple array he best tota ill run faste wn the proce ate results s

y of Sim

he project i rovide an o o simulation

erties

should be d step for crea site location evant coordi ound Maps” oject Prope ite descripti un the prog der a wind y loss mode al result is er while an ess. This st so that incre

mulation

is executed overview ov n. defined befo ating a proje n on globe dinate system ”. erties" is a ion, coordin gram determ d distributio el. Each of t chosen. Wh increase wi tudy does n ease in num

for Win

before and ver the logic

ore one start ect starts wi , see Fig.3 m and attac a generic te nate system mines the sp on that give these layout hen one red ill be more not conside mber of steps

dPRO in

d after the cal sequenc ts to perform ith “Project .8. Second hing the ma erm coveri , maps and pecified nu ve the high ts are then u duce the nu accurate w er much nu s, hand in ha

n line wi

simulation. ce of the sim m a project t Properties” step is set ap which sh

ing all info addresses [ umber of est total used in a umber of ith more umber of and with

ith

Figures mulation t. So that ” flap by tting the hould be ormation 8].

(35)

  3.1.2 The p be se proje appro appe cente The mete syste the d surfa datum WGS the G with If on zone prope locat imple world 3.1.2 For t effec Map refer 2.4.1.1 Coor preliminary elected from ect site wa oximated a ar on the m er properly Universal ers and ther em indicates datum refer ace of the g ms are used S 84 and ED Global Posi UTM coord F ne created a will autom erties of tim tion. It’s im emented. T d geodetic s 2.4.1.2 Back this work t cts which is s”, link to th renced “wo rdinate Sys y site coordi m project ex as selected ccording to map as an or for determin Transverse refore very s how much rs to the se globe into a d in differe D50. These tioning Sys dinates syst Figure 3.9 C a new proje matically be me zone can mportant to The project system 1984 kground M the chosen s the objec he maps an orld file fo stem inates are en xplorer with d independ o this site l range crossh

ning the rig Mercator ( well suited h the globe d t of expand a flat map, ent parts of e datums ar stem (GPS) tem for this

Coordinate s ect from th e provided, n be modifi select the located in 4 datum. Map wind farm ctive for ma d prelimina ormat” whi ntered here. h approxima dently in a ocation. Th hair and can ght location (UTM) sys d for WAsP differs from ded algorith often refer f the world, re the refere ) satellites. case [29]. system; datu he Project E , based on ed by the u system wh Zone 24 w m site was aking simp ary site coor ich is two

. There is an ated site coo a specific he site coor n be adjuste of it on the stem is a C P map files m being exac hms which rred to as th , but many ences used The preferr um and time Explorer a d the country user based o hich match with UTM W selected by le terrain. I rdinates can files, an n option for ordinates. H zone and rdinates (sit d. It is impo globe. Cartesian co [28]. The d ctly round. A is used to he Ellipsoid countries a for position red datum, e zone selec default coo y and the l on the optio es the sele WGS 84 w y neglecting In the tab n be entered image file r projects w However in t coordinate te center) w ortant to set oordinate sy datum of th As indicate change the d. Several are changin ns determin WGS84 is ction. ordinate sys location. O onal country ected map t which name g the topog sheet “Bac d. GEO refer and a co which can this case es were will later t the site ystem in he UTM ed before e curved different ng to the ned from selected stem and therwise y and the that was d as the graphical ckground rs to geo oordinate

(36)

  speci point map proje For t wher coord accur the c as po and t to se whic whic to de After proje *.bm coord map The math the s scale was calcu on it manu point ification fil ting out the is linked to ect site as a the calibrati re coordina dinate sets a racy of the coordinates ossible. In W the map is p elect the po ch ensures th ch is used sh etermine th r saving the ect. The map mi file has dinates on t as shown in map’s sca hematically same units; e can then b selected as ulated by di t to the sele ually with r ts must be u e, like *.jpg location of o a Google E virtual offs ion of the m ates can be as far away definition [ should crea WindPRO t placed with oints in a re he calibratio hould be de heir exact lo e map it’s ad p is saved a been defin the map. Th n Fig.3.10 ( ale is repr as the ratio centimeters be represent s big scale ividing the ected site’s reference sy upper left, u g and *.jwg f the format Earth image hore site in map, the po e read or o y from each 7]. A set of ate a triangu these real co h a high acc egular sequ on for the m efined. The ocation. Th dded to the as a *.bmi f ned and lin he site cent a). resented by o between a s) that it rep ed as a frac map 1/10.0 map length real length ystem UTM upper right a (a g files or *. tted map fil e as *.jpg fi n Equator. ositions are obtained. It other as po f maps is us ular area on oordinates i curacy on th uence which map. When coordinates he coordinat list of map file which h nked to the ter should b y the den a unit distan presents in ction with 1 000 in orde h which is m h on Google M WGS84.

and then low

a) tif and *.tw e(s). In othe le which re marked by t is crucial ossible in or ed as the ba each map a is related to he world m h should be marking the s of the win te system i s where Wi holds the coo

project, o be located i

nominator o nce on the m the real are

in the num er to get a d measured w e Earth soft To be more wer right co wf files can er words, th presents the y clicking o l to place rder to be ab asis for this and this area o pixel coor ap. Actually e done man e first point nd farms we s usually p indPRO can ordinate inf ne can def inside the se of a fracti map and the a depicted merator [1]. H detailed vie with ruler wi ftware. The e precise, th orner like in be attached he attachme e backgroun on them on the three ble to maxim study. Spec a should be rdinates on y WindPRO nually and c t, coordinate ere inserted printed on t n use it in th formation. O fine the sit

elected bac ion. It is e actual dist on the map Here, the m ew of the s ith centime map was i he first sele n Fig. 3.10 (b d just by nt of the nd of the the map defining mize the cifically, e as large the map O leaded carefully e system in order the map. he actual Once the te center ckground defined tance (in p. A map map scale site. It is ter scale imported ection of b).

(37)

  Figur zone 3.1.2 This basis 3.1.2 The rough resul user shado basis inves eleva withi Digit discr find t Trian point coord the d Calcu here re 3.10 (a) 24. 2.4.1.3 Terr section des s software fo 2.4.1.3.1 Lin WindPRO hness lines, lts. The line has to find ow flickeri s for comp stigates the ation was ta in line objec tal Height rete points o the z values ngular Irre ts which is dinates at a digitized lin ulation tim does not ne Site selecti rain Map scribes how for the simul

ne Object line objec , which are e object als d and read ng, noise i uter genera e conditions aken as 0.00 ct are; Model (DH on the surfa s for points egular Netw s the basis any point. T nes, as they e depends eed a consid (b ion (b) and w the terrain lation in bo ct gives a u often the m so contribut Z-coordina mpact and ated 3D lan s above se 0 in all case HM): Norm ace. An inte in between work (TIN) of the co The TIN wil are only n on the size derable amo b) map screen n file (*.ma oth program unique visu most importa tes to a sig ates on trad visual imp ndscape mo ea level so es. The cruc

mally used erpolation ro n the table va N): The tria ontour lines ll usually on needed with e of the TIN ount of calc n shot from ap) is create . ual control ant inputs fo gnificant red ditional map pacts. Heigh odels used that z coo ial terms re for a table outine such alues. angle mode s and make nly be calcu hin the area N radius. H culation. Ab m project pro ed in WasP over the or the map c duction in w ps before ca ht contour l for visuali ordinates w lated with h e of (x, y, as the TIN l establishe es it possib ulated for a where the However the bove all, TIN

operties fla which is u height con creation and workload w alculating, lines also f ization. Th which indic height conto z) values model is n ed from the ble to calc a selected se objects are e data like N is analyze ap within used as a ntours or d energy when the such as, form the is thesis cates the our lines defining eeded to e digital ulate Z-ection of e placed. we used ed under

Figure

Figure 4.1  Results according to Mustakerov’s and Elkinton’s papers [1,4]; (a) optimized  farm layouts for simple farm design (b) wind turbine placement figure  regarding to predominant wind direction from north;  N x  and N y  are the number
Table 4.4 Case B.1, tasks B.1.1, B.1.2 and B.1.3 where Nx and Ny represents the easting  and northing coordinates
Figure B.1 Global Coordinate Systems (a) Geographic and (b) Cartesian coordinates [12]

References

Related documents

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

The profitability of one or several investments can be calculated for a single turbine, while budget planning is assisted through calculations of the critical mass of Horns Rev

When estimating the electricity generation with measurement series performed on heights other than the hub height, the simulations using the forest module in WindSim yielded

Moreover, Figure 33 shows that placing the platform to a dominant wind direction by analyzing yearly data might create energy losses in the system, since the wind direction

In this second model, the inputs that were wind speed forecasts are replaced with inputs of wind speed observations for train- ing, and then switched back to NWP forecasts, the

The finite element model was verified with the analytical method. This model can be used to predict the load distribution in the bearing and determine the contact conditions.

The distribution of the spectral noise level around the turbine farm suggests that the noise originates from individual wind turbines closest to the measurement location rather

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