Aerodynamic Design of Wind

Turbine Blades Utilising

Nonconventional Control Systems

I KADE WIRATAMA

A Thesis Submitted in Partial Fulfilment of the Requirements of the

University of Northumbria at Newcastle for the Degree of
Doctor of Philosophy

A research undertaken in the

School of Computing, Engineering and
Information Sciences

November 2012
Declaration

I declare that the work contained in this thesis has not been submitted for any other

award and that it is all my own work. I also confirm that this work fully acknowledges

opinions, ideas and contributions from the work of others.

Name I Kade Wiratama

Signature

Date 20 November 2012

i
Abstract
As a result of the significant growth of wind turbines in size, blade load control has
become the main challenge for large wind turbines. Many advanced techniques have
been investigated aiming at developing control devices to ease blade loading. Individual
pitch control system, adaptive blades, trailing edge microtabs, morphing aerofoils,
ailerons, trailing edge flaps, and telescopic blades are among these techniques. Most of
the above advanced technologies are currently implemented in, or are under
investigation to be utilised, for blade load alleviation. The present study aims at
investigating the potential benefits of these advanced techniques in enhancing the
energy capture capabilities rather than blade load alleviation. To achieve this goal the
research is carried out in three directions: (i) development of a simulation software tool
suitable for wind turbines utilising nonconventional control systems, (ii) development of
a blade design optimisation tool capable of optimising the topology of blades equipped
with nonconventional control systems, and (iii) carrying out design optimisation case
studies with the objective of power extraction enhancement towards investigating the
feasibility of advanced technologies, initially developed for load alleviation of large
blades, for power extraction enhancement. Three nonconventional control systems,
namely, microtab, trailing edge flap and telescopic blades are investigated. A software
tool, AWTSim, is especially developed for aerodynamic simulation of wind turbines
utilising blades equipped with microtabs and trailing edge flap as well as telescopic
blades. As part of the aerodynamic simulation of these wind turbines, the control system
must be also simulated. The simulation of the control system is carried out via solving
an optimisation problem which gives the best value for the controlling parameter at each
wind turbine run condition. Developing a genetic algorithm optimisation tool which is
especially designed for wind turbine blades and integrating it with AWTSim, a design
optimisation tool for blades equipped with nonconventional control system is
constructed. The design optimisation tool, AWTSimD, is employed to carry out design
case studies. The results of design case studies reveal that for constant speed rotors,
optimised telescopic blades are more effective than flaps and microtabs in power
enhancement. However, in comparison with flap and microtabs, telescopic blades have
two disadvantages: (i) complexity in telescopic mechanism and the added weight and
(ii) increased blade loading. It is also shown that flaps are more efficient than
microtabs, and that the location and the size of flaps are key parameters in design. It is

ii
also shown that optimisation of the blade pretwist has a significant influence on the
energy extraction enhancement. That is, to gain the maximum benefit of installing flaps
and microtabs on blades, the baseline blades must be redesigned.

iii
Acknowledgements

First of all I would like to thank the Indonesian Government for the financial support

and the opportunity given to me to pursue my PhD study at Northumbria University,

United Kingdom.

I would like to express my deepest gratitude to my principal supervisor Dr Alireza

Maheri for his great support and guidance to make this project a success. Without his

support, this work would have never been completed. I also wish to thank Dr John Tan

and Dr Askin Isikveren for their kind assistance, supportive advice and encouragements.

Finally, a million thanks are dedicated to my wife Bhineka for her patience, love and

care, as well as to my beloved kids, Wika and Meivi. Their support has been source of

my energy and strength giving me spirit to keep forward through many years of the

exhausted research. This work is dedicated to them. I also very thankful to all of my

relatives for their moral support and motivation, especially to my mother and father who

passed away in the first year of my PhD. May my achievement bring pride and joy to

them.

iv
Table of Contents
Declaration ......................................................................................................................... i
Abstract ............................................................................................................................. ii
Acknowledgements .......................................................................................................... iv
List of Figures ................................................................................................................viii
List of Tables..................................................................................................................xiii
List of Algorithms .......................................................................................................... xiv
Nomenclature .................................................................................................................. xv
1 Introduction ............................................................................................................... 1
1.1 Structure of the Thesis ........................................................................................ 2
1.2 Background ........................................................................................................ 2
1.2.1 Conventional Aero-Mechanical Load and Power Control Systems ........... 2
1.2.2 Advanced Aero-Mechanical Methods for Controlling Blade Load ............ 8
1.2.3 State-of-the-Art in Design and Optimisation of Wind Turbine Blades .... 11
1.3 The Overall Aim and Objectives of the Present Research ............................... 13
2 Aerodynamics of Wind Turbines – Blade Element Momentum Theory ................ 14
2.1 Introduction ...................................................................................................... 15
2.2 BEMT Method ................................................................................................. 16
2.2.1 Wind, Induced and Relative Velocity Fields ............................................ 16
2.2.2 Momentum Theory.................................................................................... 22
2.2.2.1 Thrust and Torque Coefficients in Terms of Induction Factors ........ 23
2.2.2.2 Effect of Tip and Hub Losses on Thrust and Torque Coefficients .... 26
2.2.2.3 Prandtl Tip and Hub Loss Factors ..................................................... 28
2.2.3 Blade Element Analysis ............................................................................ 29
2.2.3.1 Thrust and Torque Coefficients via Blade Element Force Analysis . 30
2.2.4 Blade Element Momentum Theory, BEMT .............................................. 31
2.3 BEMT Corrections ........................................................................................... 32
2.3.1 Angle of Attack Corrections ..................................................................... 33
2.3.2 Effect of Ground Shear ............................................................................. 34
2.4 The software, AWTSim-Advanced Wind Turbine Simulation ........................ 35
2.4.1 Blade Discretisation .................................................................................. 36
2.4.2 BEMT Calculator ...................................................................................... 38
2.5 Validation ......................................................................................................... 40
2.6 Summary .......................................................................................................... 42
3 Aerodynamic Performance Analysis of Wind Turbines Utilising Aerodynamic
Control Systems .............................................................................................................. 43
3.1 Introduction ...................................................................................................... 44
3.2 Power and Blade Load Control ........................................................................ 44
3.2.1 Power and Blade Load Control Systems via Blade .................................. 45
v
 3.2.2 Power and Blade Load Control Systems via Rotor ................................... 46
3.2.3 Active Control versus Passive Control ..................................................... 47
3.3 Simulation of Controller ................................................................................... 47
3.4 Aerodynamic Performance of Stall Regulated Wind Turbines ........................ 54
3.4.1 Control Simulation of Variable Speed Stall Regulated Wind Turbines ... 54
3.5 Aerodynamic Performance of Pitch Controlled Wind Turbines ...................... 59
3.5.1 Control Simulation of Constant Speed Pitch Controlled Rotors............... 59
3.5.2 Control Simulation of Variable Speed Pitch Controlled Rotors ............... 62
3.6 Aerodynamic Performance of Wind Turbines with Blades Utilising Flap ...... 67
3.6.1 Modifications Applicable to the Aerodynamic Performance Calculator .. 67
3.6.2 Control Simulation of Constant Speed Rotors with Blades Utilising Flap
70
3.6.3 Control Simulation of Variable Speed Rotors with Blades Utilising Flap 73
3.7 Aerodynamic Performance of Wind Turbines with Telescopic Blades ........... 76
3.7.1 Modifications Applicable to the Aerodynamic Performance Calculator .. 77
3.7.2 Control Simulation for Constant Speed Rotors with Telescopic Blades .. 79
3.7.3 Control Simulation for Variable Speed Rotors with Telescopic Blades ... 81
3.8 Aerodynamic Performance of Wind Turbines with Blades Utilising Microtabs
84
3.8.1 Modifications Applicable to the Aerodynamic Performance Calculator .. 85
3.8.2 Control Simulation of Constant Speed Rotors with Blades Equipped with
Microtabs ................................................................................................................ 87
3.9 A Preliminary Comparison of Different Types of Control Systems ................ 91
3.10 Summary ....................................................................................................... 95
4 AWTSimD, an Optimisation Tool for Wind Turbine Blades Equipped with
Nonconventional Aerodynamic Control Systems ........................................................... 97
4.1 Introduction ...................................................................................................... 98
4.2 Aerodynamic Design of Wind Turbine Blades ................................................ 98
4.2.1 Classification of Design Variables ............................................................ 98
4.2.2 Direct versus Inverse Design Methods ..................................................... 99
4.2.3 Inverse Design ......................................................................................... 100
4.2.4 Performance of Inverse Design ............................................................... 101
4.3 Optimal Aerodynamic Design of Wind Turbine Blades ................................ 104
4.4 Genetic Algorithm Optimisation Method....................................................... 105
4.4.1 Chromosome Representation .................................................................. 106
4.4.2 Initial Population Generation .................................................................. 106
4.4.2.1 Randomly Generated ....................................................................... 107
4.4.2.2 Perturbation of the Baseline Design ................................................ 108
4.4.3 Crossover ................................................................................................ 109

vi
 4.4.3.1 Arithmetic Crossover ....................................................................... 109
4.4.3.2 Geometric Crossover ....................................................................... 110
4.4.4 Mutation .................................................................................................. 113
4.4.5 Constraint Handling ................................................................................ 113
4.4.6 Regeneration ........................................................................................... 113
4.4.7 Termination ............................................................................................. 114
4.5 AWTSimD, Advanced Wind Turbine Simulation and Design ...................... 114
4.6 A Case Study: Blade Optimisation for Modified Pitch Controlled AWT-27
Wind Turbine ............................................................................................................ 117
4.7 Summary ........................................................................................................ 119
5 Design Optimisation of Wind Turbine Blades Equipped with Nonconventional
Aerodynamic Control Systems ..................................................................................... 120
5.1 Introduction .................................................................................................... 121
5.2 Potentials of Trailing Edge Flaps in Power Extraction Enhancement: Constant
Speed Rotors ............................................................................................................. 121
5.3 Potentials of Trailing Edge Flaps in Power Extraction Enhancement: Variable
Speed Rotors ............................................................................................................. 127
5.4 Potentials of Telescopic Blades in Power Extraction Enhancement .............. 132
5.5 Potentials of Microtabs in Power Extraction Enhancement ........................... 137
5.6 Summary ........................................................................................................ 139
6 Summary and Conclusion ..................................................................................... 140
6.1 Summary of Work, Achievements and Contribution ..................................... 141
6.2 Critical Appraisal and Future Work ............................................................... 144
References ................................................................................................................. 145
Appendix A-Sample Simulation and Design Input Files .................................................. 1
Appendix B-Microtab Data ............................................................................................... 1
Appendix C-Design Optimisation Results ........................................................................ 1

vii
List of Figures

Figure 1.1- Power curve for different power control methods

Figure 1.2-Flow kinematics diagram of stall control
Figure 1.3-Stall regulation
Figure 1.4-Pitch control to enhance/regulate rotor mechanical power
Figure 1.5- Flow kinematics diagram of pitch control
Figure 1.6- Conventional pitch control
Figure 1.7- Active stall pitch control
Figure 1.8-Size and power increase of commercial wind turbines over time (Robert,
2011)
Figure 1.9- Schematic of load control systems
Figure 2.1-Rotor and blade systems of coordinates
Figure 2.2-Relative velocity in aerofoil plane
Figure 2.3-Relative velocity in aerofoil plane; zero yaw
Figure 2.4-Pressure and velocity variations
Figure 2.5-Annulus control volume; Linear momentum balance
Figure 2.6-Annulus control volume; Angular momentum balance between disk and Far
Wake
Figure 2.7-Annulus control volume; Angular momentum balance
Figure 2.8-Relative velocity in aerofoil plane; zero yaw; Wilson-Walker model
Figure 2.9-Discretisation of blade into nseg segments
Figure 2.10-Blade element force analysis on a typical segment
Figure 2.11-Discretising rotor disk into nsec sectors (here, nsec  3 ) and the azimuth angle
associated to each sector
Figure 2.12-AWTSim versus WTPerf (Buhl, 2004)-Power curves at different pitch
angles
Figure 2.13-AWTSim versus WTPerf (Buhl, 2004)-Thrust force at different pitch angles
Figure 3.1-Different control systems affecting blade performance
Figure 3.2- Flow kinematic diagram at span location r
Figure 3.3- Three possible behaviour of objective P with respect to controlling
parameters qi
Figure 3.4-Diffrent distinct behaviours of P   curve
Figure 3.5-Iteration points in hill-climbing method of Algorithm (3.1)
Figure 3.6-Number of iterations against wind speed using hill-climbing method of
Algorithm (3.1)
Figure 3.7-Results of simulation of variable-speed stall-regulated AWT-27
Figure 3.8-Diffrent distinct behaviours of P  pitch curve
viii
Figure 3.9- Results of simulation of constant-speed pitch-controlled AWT-27
Figure 3.10-Rotor power versus rotor speed and blade pitch angle at various wind
speeds
Figure 3.11- Iteration points in pattern search method of Algorithm (3.2)
Figure 3.12- Number of iterations against wind speed using pattern search method of
Algorithm (3.2)
Figure 3.13- Results of simulation of variable-speed pitch-controlled AWT-27
Figure 3.14-(a) Parameters defining location and size of flap, (b) Flap deployment angle
sign convention
Figure 3.15- C L variation against angle of attack,  and flap deployment angle,  F
Figure 3.16- C D variation against angle of attack,  and flap deployment angle,  F
Figure 3.17-Power-flap deployment ( P   F ) curves at different wind speeds
Figure 3.18- Results of simulation of constant-speed flap-controlled AWT-27
Figure 3.19-Rotor power versus rotor speed and flap deployment angle at various wind
speeds
Figure 3.20- Result of simulation of variable-speed flap-controlled AWT-27
Figure 3.21-Comparison of power curves for variable speed flap controlled and variable
speed stall regulated wind turbines
Figure 3.22-Telescopic blade definition
Figure 3.23-Telescopic blade chord and pretwist distribution ( r  r / RT , c  c / RT ).
* *

Figure 3.24- Result of simulation of constant-speed telescopic blade AWT-27

Figure 3.25- Torque produced by each blade segment at wind speed 5m/s against the
radial location
Figure 3.26- Result of simulation of variable-speed telescopic blade AWT-27
Figure 3.27-Parameters defining location, size and actuation height of microtabs
Figure 3.28- C L variation against angle of attack, microtab deployed downward
Figure 3.29- C D variation against angle of attack, microtab deployed downward
Figure 3.30- C L variation against angle of attack, microtab deployed upward
Figure 3.31- C D variation against angle of attack, microtab deployed upward
Figure 3.32- Results of simulation of constant-speed AWT-27 equipped with microtabs
Figure 3.33- Results of simulation of constant-speed AWT-27 equipped with microtabs-
The states of the microtabs
Figure 3.34- Comparison of the power curves of different types of constant speed wind
turbines
Figure 3.35- Comparison of the power curves of different types of variable speed wind
turbines
Figure 3.36- Comparison of the power coefficient of different types of constant speed
wind turbines
ix
Figure 3.37- Comparison of the power coefficient of different types of variable speed
wind turbines
Figure 3.38- Comparison of the blade root bending moment of different types of
constant speed wind turbines
Figure 3.39- Comparison of the blade root bending moment of different types of
variable speed wind turbines
Figure 3.40- Comparison of rotor speed variation for different types of variable speed
wind turbines
Figure 4.1- (a) Inverse and (b) Direct Design
Figure 4.2-Redesigned chord of AWT-27 blade for variable speed rotor using inverse
method
Figure 4.3- Redesigned pretwist of AWT-27 blade for variable speed rotor using inverse
method
Figure 4.4-Power curve: Variable speed AWT-27 with original and redesigned blades
Figure 4.5-Power coefficient: Variable speed AWT-27 with original and redesigned
blades
Figure 4.6-Search-based design (Direct design optimisation)
Figure 4.7-Wind turbine blade chromosome
Figure 4.8- Randomly generated distribution for a typical real-number distributed
design variable
Figure 4.9- Initial population generation using perturbation of an initial design
candidate: Blade pretwist
Figure 4.10- Initial population generation using perturbation of an initial design
candidate: Blade chord
Figure 4.11-Arithmetic (weighted average) crossover
Figure 4.12-Geometric crossover
Figure 4.13- Pretwist formation of a child blade; child is formed based on the left
segment of parent 1and the right segment of parent 2 (Maheri, 2012)
Figure 4.14-Optimised blade pretwist for modified pitch controlled AWT-27
Figure 4.15-Optimum power coefficient of modified pitch controlled AWT-27
Figure 4.16-Search history for optimisation of the pretwist for modified AWT-27
constant speed pitch controlled blade
Figure 5.1-Flap 60-65%-constant speed rotor
Figure 5.2-Effect of flap size and location on power enhancement-constant speed rotor
Figure 5.3-Percent increase in the average power versus flap size and location-constant
speed rotor
Figure 5.4-Percent improvement in the average power due to blade optimisation-
constant speed rotor
Figure 5.5-Effect of flap size on the power extraction enhancement-constant speed rotor
Figure 5.6-Power produced by constant speed rotors utilising blades equipped with flap
Figure 5.7-Power coefficient of constant speed rotors utilising blades equipped with flap
x
Figure 5.8-Maximum flap bending moment in blades equipped with flap-constant speed
rotor
Figure 5.9-Flap 60-65%-variable speed rotor
Figure 5.10-Effect of flap size and location on power enhancement-variable speed rotor
Figure 5.11-Percent increase in the average power versus flap size and location-variable
speed rotor
Figure 5.12-Percent improvement in the average power due to blade optimisation-
variable speed rotor
Figure 5.13-Effect of flap size on the power extraction enhancement-variable speed
rotor
Figure 5.14-Power produced by variable speed rotors utilising blades equipped with flap
Figure 5.15-Power coefficient of variable speed rotors utilising blades equipped with
flap
Figure 5.16-Maximum flap bending moment in blades equipped with flap-variable
speed rotor
Figure 5.17-Optimised telescopic blade-constant speed rotor
Figure 5.18-Power enhancement via utilising telescopic blades-constant speed rotor
Figure 5.19-Optimised telescopic blade-variable speed rotor
Figure 5.20-Power enhancement via utilising telescopic blades-variable speed rotor
Figure 5.21-Microtab on optimised blades-power enhancement
Figure 5.22-Microtab on optimised blades
Figure A1-Sample simulation file for constant speed rotor with flap control
Figure A2-Sample simulation file for variable speed rotor with flap control
Figure A3- Sample simulation file for constant speed rotor with telescopic blades
Figure A4-Sample simulation file for variable speed rotor with telescopic blades
Figure A5- Sample simulation file for constant speed rotor with pitch control
Figure A6-Sample simulation file for variable speed rotor with pitch control
Figure A7- Sample simulation file for stall regulated constant speed
Figure A8-Sample simulation file for constant speed rotor with microtab
Figure A9-Sample design optimisation file for variable speed rotor with telescopic
blades
Figure C1-Flap 60-65%- constant speed
Figure C2-Flap 65-70%- constant speed
Figure C3-Flap 70-75%- constant speed
Figure C4-Flap 75-80%- constant speed
Figure C5-Flap 80-85%- constant speed
Figure C6-Flap 85-90%- constant speed
Figure C7-Flap 90-95%- constant speed
Figure C8-Flap 60-70%- constant speed
xi
Figure C9-Flap 60-75%- constant speed
Figure C10-Flap 60-80%- constant speed
Figure C11-Flap 60-85%- constant speed
Figure C12-Flap 60-65%- variable speed
Figure C13-Flap 60-70%- variable speed
Figure C14-Flap 70-75%- variable speed
Figure C15-Flap 75-80%- variable speed
Figure C16-Flap 80-85%- variable speed
Figure C17-Flap 85-90%- variable speed
Figure C18-Flap 90-95%- variable speed
Figure C19-Flap 60-70%- variable speed
Figure C20-Flap 60-75%- variable speed
Figure C21-Flap 60-80%- variable speed
Figure C22-Flap 60-85%- variable speed

xii
List of Tables
Table 2.1-Blade data file-Part 1
Table 2.2-Blade data file-Part 2
Table 2.3-Discretised blade with nseg  20
Table 2.4-AWT-27 wind turbine
Table 3.1-Number of independent controlling parameters
Table 3.2-Simulation of constant-speed wind turbines with different types of blades
Table 3.3-Simulation of variable-speed wind turbines with different types of blades
Table 3.4-Baseline blade data
Table 3.5-Data at the end of Step 2 of Algorithm (3.5)
Table 3.6-Telescopic blade data file (at the end of Step 3 of Algorithm (3.5))
Table 4.1- Results of the simulation of variable speed AWT-27 with original and
redesigned blades using inverse method
Table 5.1- Examined flap lengths and flap locations (all values in % of R )
Table 5.2- Examined telescopic ranges (all values in % of R )
Table C1-Lift and drag coefficients for different microtab actuation heights on the lower
surface
Table C2-Lift and drag coefficients for different microtab actuation heights on the upper
surface
Table C3-Changes in lift and drag coefficients for different microtab actuation heights
on the lower surface
Table C4-Changes in lift and drag coefficients for different microtab actuation heights
on the upper surface

xiii
List of Algorithms

Algorithm 2.1-Blade discretisation

Algorithm 2.2-BEMT calculator
Algorithm 3.1-Hill climbing search for finding optimum rotor speed for variable speed
stall regulated WT
Algorithm 3.2-Pattern search method for finding optimum rotor speed and pitch angle
for variable speed pitch controlled wind turbines
Algorithm 3.3-Reading C L ( ,  F ) and CD ( ,  F ) from 3-D tabulated data
Algorithm 3.4-Modified BEMT calculator for blades utilising flap
Algorithm 3.5-Modification of the data file of the baseline blade for a telescopic blade
Algorithm 3.6- Hill climbing search for finding optimum rotor radius ( RT ) for constant
speed telescopic blade wind turbines
Algorithm 3.7-Pattern search method for finding optimum rotor speed and rotor radius
for variable speed telescopic blade
Algorithm 3.8-Modified BEMT calculator for blades equipped with microtabs
Algorithm 3.9-GA for finding optimum state of microtabs
Algorithm 4.1- AWTSimD: Optimisation of wind turbine blades utilising active control
systems

xiv
Nomenclature
A0 , A1 , A2 Coefficient in Wilson-Walker state model
Ae Blade element (segment) area
a Axial induction factor
ae Boundary of heavy and light loading states
a' Rotational induction factor
B Number of blades
B0 , B1 , B2 Coefficient in Classical brake state model
C1 Child 1
C, c Blade chord
CD Drag coefficient
CL Lift coefficient
CM Pitching moment coefficient, Torque coefficient
CM 0
Torque coefficient at zero drag
Cp
Power coefficient
CT Thrust coefficient
CT0
Thrust coefficient at zero drag
D Drag per unit span length, Rotor Diameter
dF Width of flap
d MT Microtabs distance from leading edge
F Loss factor
f Function
fn Normal force on the blade cross section
fit Fitness
hhub Height above ground hub
hMT Microtabs actuation height
L Lift per unit span length
M Rotor torque, Blade bending moment
MT Microtab
ngen
No. of generation
N MT Number of microtab
n pop
Population size
nco Number of crossover
ndp
Number of design points
nMute
Number of mutation
nsec
Number sector

xv
nseg
Number segment
P1 Parent 1
P2 Parent 2
P Rotor mechanical power
pc Probably crossover
pm Probably mutation
pitch Pitch angle
q Controlling parameters
R Rotor radius, Rayleigh PDF
Rhub Hub radius
RT Blade telescopic radius
r Radial distance from the rotor centre
s MT Microtab length
T Rotor thrust
t max Aerofoil maximum thickness
V Flow velocity
Vav Site average wind speed
VAF Stream-wise velocity in the far wake
Vd Disk velocity
Vhub Wind speed at hub
Vrel
Relative velocity
Vw
Wind velocity
V Free stream velocity

V Wind velocity field

VI Induced velocity field

Vrel
Relative velocity field

VW Upstream mean velocity field
Vˆ Wind turbulence
w Normalised distribution force
x0 x-coordinate of the blade pretwist axis
xco
Fraction of total crossover
xip
Fraction of initial population
z Height above ground
z0
Roughness length
 Angle of attack
 opt
Angle of attack optimum
 Blade twist angle

xvi
0 Blade pretwist
e Elastic torsional deformation in the blade
 Blade cone angle
F Flap deployment angle
 c Correction applicable to angle of attack
C L Change in lift coefficient
C D Change in drag coefficient
 Tolerance
 Rotor yaw angle
 Speed ratio
r Local speed ratio
 Rotor speed
 Inflow angle
0 Inflow angle at zero rotational induction
 Blade element azimuth angle
 Air density
r Local solidity ratio

Subscripts
av Average
c Constraint, Crossover
co Crossover
dp design point
F Force, Flap
FB Flap-wise bending
FW Far wake
F, e End of length flaps location
F, s Start of length flaps location
gen Generation
hub Hub
ip Initial population
l Lower limit
M Bending moment
MT Microtab
MT, e End of length microtab location
MT, s Start of length microtab location
m Mutation
o Initial point
opt Optimum
pop Population
rated Rated
ref Reference
rel Relative
xvii
s Span-wise
Secondary
T Telescopic
T,e Telescopic part fully deployed
T, s Fix part of telescopic
T, tip Tip
u Upper limit

Superscripts
 Normalised, dimensionless

Systems of coordinates
nt  s Blade, Figure 2.1
x yz Blade and global coordinate, Figure 2.1
x  r  Disk (Rotor), Figure 2.1

Abbreviations
AOA Angle of Attack
AR Aspect Ratio
AWT Aerodynamic Wind Turbine
AWTSim Aerodynamic Wind Turbine Simulation
AWTSimD Aerodynamic Wind Turbine Simulation Design
BEMT Blade Element Momentum Theory
CFD Computational Fluid Dynamics
CV Control Volume
GA Genetic Algorithm
HAWT Horizontal Axis Wind Turbine
NREL National Renewable Energy Laboratories
PDF Probability Distribution (Density) Function
SNL Sandia National Laboratory

xviii
1 Introduction

1
1.1 Structure of the Thesis

Chapter 1 of this thesis is dedicated to the background of the conventional and

nonconventional aero-mechanical control systems, state of the art of blade design
optimisation and the aim and the objectives of this project. Chapter 2 elaborates on
blade element momentum theory (BEMT), the theory behind the software tool
AWTSim developed for aerodynamic analysis of wind turbines. The pseudo codes of
the core modules of the software tool are also given in this chapter. Using a stall-
regulated constant speed test wind turbine, at the end of this chapter the performance of
this software is validated against an accredited analysis tool. Further necessary
enhancements to AWTSim, making it capable of simulating wind turbines with
nonconventional control systems, is discussed and explained in Chapter 3. For each type
of wind turbines, its controlling system is simulated by solving an optimisation
problem. Enhanced AWTSim is capable of simulating both constant and variable speed
wind turbines with various nonconventional controlling systems. Chapter 4 details the
genetic algorithm optimisation method developed for design optimisation of wind
turbine blades equipped with nonconventional control systems. In Chapter 5 design
optimisation case studies are carried out for wind turbine blades equipped with
microtabs and trailing edge flaps as well as telescopic blades. The results of these
design case studies have been assessed towards a thorough investigation of their
capabilities in enhancing power capture capabilities. Chapter 6 summarises the research
carried out, the results obtained, conclusions, critical appraisal of the work and
suggested future work.

1.2 Background

1.2.1 Conventional Aero-Mechanical Load and Power Control Systems

Wind turbines are designed to produce maximum power at the most probable wind
speed. At high wind speeds, the generated power by a wind turbine far exceeds the
generator capacity. To protect wind turbine operation at high wind speeds it is needed to
limit the generated power otherwise wind turbine will be overloading its rotors,
mechanical power train, as well as its electrical generator leading to failure during
operation.

2
Pitch control and stall regulation are the most popular power control systems both based
on controlling the flow angle of attack. Figure (1.1) shows power curve for these two
control systems. According to this figure the ideal situation is for the turbine to be able
to produce as much power as possible from the wind up to the rated power of the
generator, then limits the power production at the rated value.

P
No Control
Rated
Power Pitch Controlled

Stall Regulated

V
Cut-in Velocity Cut-out Velocity
Figure 1.1- Power curve for different power control methods

Stall regulation is mechanically the most simple controlling strategy. In stall regulated
wind turbines the blades have been designed to stall in high winds without any pitching
control. The rotor is built with the blades fixed on the hub therefore it is rather simple in
construction and the pitch of the blades are adjusted only once when the wind turbine is
erected. In order to achieve stall-regulation at appropriate wind speeds, the wind turbine
blades operate closer to stall and result in lower aerodynamic efficiency below rated
power. This does not give a perfectly flat power curve above the rated wind speed.

A stall regulated wind turbine is normally operated at an almost constant rotor speed 
and thus the angle of attack  increases as the wind speed increases. Figure (1.2) shows
the flow diagram of stall regulation at a typical radial location r . According to this
figure, as the wind speed VW increases the inflow angle  and consequently the angle of

attack  increases. Increasing the angle of attack beyond a certain limit, called stall

3
angle of attack  s , causes significant drop in lift coefficient and consequently rotor
power (see Figure (1.3)).

2
1 2
Vrel 1

r Vrel

Vw2 Vw1

Figure 1.2-Flow kinematics diagram of stall control

Increase
in AoA
Coefficient of Lift CL

Stall Angle

Angle of Attack (AoA)

Figure 1.3-Stall regulation

In order for the turbine to stall rather than accelerate in high winds, the rotor speed must
be restrained. In a constant speed turbine the rotor speed is restrained by the generator,
which is governed by the network frequency, as long as the torque remains below the
pull-out torque. In a variable speed turbine, the speed is maintained by ensuring that the
generator torque is varied to match the aerodynamic torque. A variable-speed turbine
offers the possibility to slow the rotor down in high winds in order to bring it into stall.
This means that the turbine can operate further from the stall point in low winds,
resulting in higher aerodynamic efficiency. However, this strategy means that when a

4
gust hits the turbine, the load torque not only has to rise to match the wind torque but
also has to increase further in order to slow the rotor down into stall.

Using this type of stall regulating requires carefully designed rotor blade geometry and
carefully selected rotor speed to ensure that at higher wind speed, the flow does indeed
separate so that an increase in generated power can effectively be prevented.

Pitch control is the most common means of controlling the aerodynamic power
generated by the turbine rotor and also has a major effect on the aerodynamic loads
generated by wind turbine. The most effective way of influencing the aerodynamic
angle of attack and thus power generated is by mechanical adjustment of the blade pitch
angle. Pitch control can be used to regulate power generated by decreasing the power to
rated power or increasing the power through changes in the blade aerodynamic as
shown in Figure (1.4). In this figure the blue curve represents the power curve without
pitch control and the red curve represents the power curve when a pitch control system
is employed. By pitching the blade around its axis, the blade pitch angle changes the
angle of attack and aerodynamic forces. Pitching influences the power generated and
load occurred in the rotor blade and is therefore suitable for both power and load
control.

Flow kinematics of a pitch control system is shown in Figure (1.5). In this figure  , the

blade twist is a combination of the pretwist  0 and blade pitch angle pitch .

Rated
Power

V
Cut-in Velocity Cut-out Velocity
Figure 1.4-Pitch control to enhance/regulate rotor mechanical power
5
 2
1 2
1
r Vrel 

Vw

Figure 1.5- Flow kinematics diagram of pitch control

In practice, power control through blade pitch control can be achieved by two methods,
conventional pitch and active stall. In conventional approach, shown in Figure (1.6),
pitch to stall increases the angle of attack and the lift coefficient (and consequently the
rotor mechanical power). Pitching the blade in the direction of the feather position not
only reduces the driving force but all forces at the rotor blade and also the resulting
stress. By pitching to feather the quasi loads from mean aerodynamic force are reduced
at higher wind conditions and during storm therefore when a dangerous operating state
occurs (e.g. over-speeding or emergency stop), the blade pitch has to bring the rotor
blade to the feather position immediately consequently it will reduce power generated
and load generated on the rotor blade.

Working area for

conventional pitch

C L  0
CL

Twist to
Stall

Stall Angle

Angle of Attack (AoA)

Figure 1.6- Conventional pitch control

6
 Working area for
active stall

CL C L  0

Twist to
Stall

Stall Angle

Angle of Attack (AoA)

Figure 1.7- Active stall pitch control

In active stall scenario, as shown in Figure (1.7), pitch to stall increases the angle of
attack but reduces the lift coefficient (and consequently the rotor mechanical power).

Using mechanical energy for the blade pitch is more suitable at smaller wind turbine for
rated power less than 100kW. Hydraulic blade pitch systems are normally used for wind
turbines with rated power in the range of 300 kW to the multi-MW. Electrical blade
pitch system is also common to most of the wind turbines types. Many manufactures
trust in this solution for the pitch systems especially for larger wind turbines for rated
power above 500 kW (Robert, 2011).

Mechanisms that adjust blade pitch angle in response to the thrust loading were also
quite popular in the early days of the modern wind energy push. Approaches and
objectives were quite varied. Cheney (1978) regulated power with a centrifugally loaded
mass on an elastic arm. Bottrell (1981) had a system for cyclically adjusting pitch for
load balancing. Currin (1981) had a system for passively adjusting pitch for both power
and load control. Hohenemser (1981) studied alleviating yaw loads with cyclic pitch
adjustments. Corbet (1992) evaluated the use of all available blade loads to effect pitch
changes that would regulate the power output of a turbine, aiming at a flat power curve
in high winds. They reported that perfect regulation was very difficult to achieve, and
that even less than perfect regulation was a challenge. These approaches also depend on
quite substantial blade rotations to achieve perfect regulation.

7
1.2.2 Advanced Aero-Mechanical Methods for Controlling Blade Load
The size of rotor wind turbine has been steady increased over the past decade. Recently,
rotors of more than 120 m diameter are in prototype and now, commercial wind turbines
are available with capacities up to 3500kW (Herbert, 2007). The rapid increase in size
and capacity of commercially manufactured wind turbines between the years 1982 and
2009 is illustrated in Figure (1.8). According to this figure, within a very short time the
increase in wind turbine size have been remarkable. In 2010, for example, the largest
commercial wind turbine had a capacity of 7.5 MW and a diameter of 126 meters
(Robert, 2011).

Figure 1.8-Size and power increase of commercial wind turbines over time (Robert,
2011).

As a result of the significant growth of wind turbines in size, blade load control has
become the main challenge for large wind turbines (Nijssen 2006 and Johnson 2008).
According to Barlas (2010), many advanced techniques have been investigated aiming
at developing control devices to ease blade loading. Individual pitch control system,
adaptive blades, trailing edge microtabs, morphing aerofoils, ailerons, trailing edge
flaps, and telescopic blades are among these techniques.

Though, collective pitch control systems were primarily developed to limit the rotor
mechanical power at a rated value and to optimise the energy capture below that value
(Bossanyi 2000, Wright 2002, Van der Hooft 2003), recently, individual pitch control
systems have been successfully developed and utilised to alleviate low frequency
fluctuating loads by pitching the blades individually (Caselitz 1997, Bossanyi 2003, van
8
Engelen 2003, Larsen 2005, Lovera 2003). The concept of individual pitch control was
first introduced for helicopter rotor blades (Johnson 1982). Still some disadvantages are
evident, especially for the large scale application for wind turbine blades (Lovera 2003).
The response time for individual pitch control systems is not fast enough for high
frequency load fluctuation. Moreover, actuation of massive large blades requires
significant actuation force and energy.

Also, adopting from the helicopter blade technology, blade twist control based on
passive control system is a relatively new field in the wind turbine industry. This
approach, known as adaptive or smart blades, employs the blade itself as the controller
to sense the wind velocity or rotor speed variations and adjust its aerodynamic
characteristics to affect the wind turbine performance. Earlier work was carried out on
the project at Reading University by Karaolis (1989) and Kooijiman (1996) and then
progressed by other investigators. These blades are made of anisotropic composite
materials and change their shapes in response to the variations in wind turbine operating
conditions. It has been shown that these blades potentially can be used for both blade
load alleviation and enhancing energy capture capabilities (Lobitz, 2001, Maheri 2006,
Maheri et al 2006, Maheri et al 2007a, Maheri et al 2007b, Maheri et al 2007c, Maheri
and Isikveren 2009a and Maheri and Isikveren 2010) .

A different kind of aerodynamic device proposed for load alleviation is microtabs

(Baker and Mayda 2005, Chow 2007 and van Dam et al 2001, 2002, 2005a, 2005b and
2007). Microtabs are small aerodynamic control surfaces with deployment height of
order of magnitude of 0.01 of local chord, installed close to the trailing edge of the
blade.

Morphing blades, a concept adopted from aircraft morphing wings, has also the
potential to improve the system performance over the wind turbine operational envelope
(Stuart 1997, Farhan 2008, Barlas 2010). The morphing concept includes a wide
spectrum of shape adaptations such as variation in camber, twist, span and plan form
area. Camber control is a type of morphing aerofoils and an effective way of controlling
the aerodynamic forces by directly changing the shape of the aerofoil. This action has
direct effects on the force distribution on the blade, so it can be used for active load
alleviation purposes (Farhan 2008, Maheri 2009b).

9
Aileron, a concept borrowed from aerospace industry, is another active device, which
has been used for aerodynamic breaking in the past. Results of a recent research on
ailerons via simulating the behaviour of a wind turbine in turbulent wind indicates that
aileron load control can assist in power regulation and reduce root flap bending
moments during a step-gust and turbulent wind situation (Migliore 1995, Stuart 1996,
Enenkl 2002).

The concept of trailing edge flap follows the same principle as aileron, but by deflecting
the trailing edge portion of the aerofoil, to change the aerodynamic characteristics of the
blade in high-wind conditions and turbulent wind (Troldborg and Buhl 2005, Andersen
2006).

Compact trailing edge flaps made of smart materials is another concept under
investigation for load alleviation. Results of a recent research demonstrate large
reduction (50 -90%) in vibratory load (Barlas 2010).

Recently, the concept of variable length blades has been also proposed as a means of
controlling the load and increasing the energy yield of the turbine. Telescopic blades
retract/extend in response to the variations in wind speed (DOE 2005, GE Wind Energy
2006, Pasupulapati 2005, Shrama 2007).

Most of the above advanced technologies are currently implemented in, or are under
investigation to be utilised, for blade load alleviation, as the main challenge for design
and manufacturing of larger wind turbines. The present research is focused on
investigating the application of some these advanced techniques, namely, microtabs,
trailing edge flaps and telescopic blades in enhancing the energy capture capabilities
rather than blade load alleviation.

Load/power control systems can be divided in two methods: passive and active control.
Figure (1.9) summarises nonconventional load/power control systems.

10
 Advanced Techniques

Passive Control System Active Control System

 Bend Twist Adaptive

Blade Rotor Blade
 Stretch Twist Adaptive
Blade

 Camber Control
 Variable Diameter Rotor
(morphing)
(Telescopic Blade)
 Aileron
 Trailing Edge Flap
 Microtabs
 Active Twist
 Individual Pitch Control

Figure 1.9- Schematic of load control systems

1.2.3 State-of-the-Art in Design and Optimisation of Wind Turbine Blades

Alternative design approaches is another field of research in wind turbines, which
recently has become under attention. Integrated design, using computational intelligence
techniques in design and optimisation, and modification-based design are examples of
this line research. Seki (1996) investigated a method to obtain the optimum blade shape
for the rotor of several hundred kW horizontal axis wind turbines. Jureczko (2005)
developed a computer program package that would enable optimisation of wind turbine
blades with regard to a number of criteria. When designing a wind turbine, the goal is to
attain the highest possible power output under specified atmospheric conditions. From
the technical point of view, this depends on the shape of the blade. The change of the
shape of blade is one of the methods to modify stiffness and stability, but it may
influence aerodynamic efficiency of wind turbine. Other method to change dynamic and
mechanical properties of wind turbine is modifying the composite material, which the
blade is made of. The problem of determining the optimal shape of blade and
determining the optimal composite material is a complex one, as the mathematical
description of aerodynamic load is complex and a number of constraints and objectives
have to be satisfied. These considerations have prompted to take up the problem of the
multi-criteria optimum design of wind turbine blades. Mendes (2006) developed a

11
method to obtain optimal chord and twist distributions in wind turbine blades by using
genetic algorithms. The method was based on the use of basic pieces of both
aerodynamic techniques for power prediction in wind turbines and also in optimisation
tools. The optimisation of chord and pretwist distributions were computed to maximise
the mean expected power depending on the Weibull wind distribution at a specific site
because in wind power systems optimisation is highly site dependent (Jonkman 2003).
To optimise chord and twist distributions BEMT was used (Tangler 2002, Jonkman
2003, Tangler 2004). The BEMT is shown to give good accuracy with respect to
computational cost. In most of the optimisation procedures the main computational load
is the repetitive goal function evaluation. Therefore, they have implemented a BEM
procedure that provided high quality predictions in the linear and in the near stall zones.
Also, this procedure was efficient in the use of computational resources.

Ashwill (2007) at Sandia National Laboratories (SNL) developed advanced concepts

that enable the utilisation of longer blades that weigh less. These blades are more
efficient structurally and aerodynamically. Several of these concepts have been
incorporated into subscale prototype blades showing that the concept of flatbacks
aerofoils has a structural advantage in the in-board region of the blade, especially
towards the root.

More efficient blade designs via integrated design (structure and aerodynamics
considered simultaneously) also investigated and reported by Ashwill (2007). Maheri et
al (2006) developed a simple algorithm to modify an ordinary blade to a bend-twist
adaptive one. Through their algorithm the rotor radius, pretwist and chord distributions
of the original blade are modified and the optimum value of the induced twist due to
elastic coupling will be predicted. Modifications to the original blade was decided in a
manner that limit the output power at its rated value and improve the average power of
the wind turbine, while trying to reduce the negative effects on the blade size and
loading. Maheri and Isikveren (2008) developed an integrated design approach by
proposing an alternative approach for design of adaptive blades. The concept of
Variable State Design Parameters (VSDP) was proposed and investigated. It was
explained how VSDP can be employed to convert the traditional integrated design
process of intrinsically smart aero-structures, to a decoupled and hence computationally
efficient design process. Through a design case study they showed the practicality and
efficiency of the new approach. Traditional methods of design of aerodynamic surfaces
12
have been not efficient when applied to design of intrinsically smart aero-structures.
Introducing the induced deformation as a Variable State Design Parameter decouples
the analysis of these structures therefore the aerodynamic and structural design of these
structures can be carried out separately. Their design case study shows that adopting
design methodology based on VSDP is enables to design wind turbine smart blades
efficiently without any structural analysis involved in the aerodynamic design phase.
Using the concept of VSDP they continued by developing a design tool for adaptive
blades (Maheri and Isikveren 2009a).

1.3 The Overall Aim and Objectives of the Present Research

The overall aim of this research is to investigate the potentials of nonconventional
control systems, which have been initially developed for load control in energy capture
capability enhancement. To achieve this goal the following objectives have been
defined:
1. To develop the required analysis and simulation tools capable of simulating wind
turbines utilising advanced technologies developed for load alleviation of large
blades.
2. To develop suitable design optimisation methods and tool for design optimisation of
blades equipped with nonconventional control systems.
3. To carry out design optimisation case studies towards investigation of the feasibility
of advanced technologies, initially developed for load alleviation of large blades, for
power extraction enhancement.

Project has been delivered around three axes: simulation software tool development,
design optimisation methodology and tool development and case studies.

13
2 Aerodynamics of Wind
Turbines – Blade Element
Momentum Theory

14
2.1 Introduction
Various methods can be used in order to identify the blade performance and flow
characteristics. These methods can be classified mainly as BEMT-based methods and
computational fluid dynamics -based methods (Snel 1998, Crespo 1999, Vermeer 2003,
Snel 2003, Sanderse 2011).

Computational fluid dynamics (CFD) methods provide more accurate result of analysis
compared to BEMT. However, to acquire reliable result from computational method, a
vast amount of computational grids are required and advanced turbulence model needs
to be applied (Snel 2003, Sanderse 2009). High computational time for turbine wake
calculation makes CFD based models less practical in engineering use, particularly as
the evaluator module of blade design tools. CFD methods are very useful for
understanding the aerodynamic characteristics of rotor blades but it consumes too much
time and resources thus it is generally applied at the final performance evaluation stage
after all the design process is completed (Madsen,1996). On the other hand, BEMT is a
simple, yet efficient method for aerodynamic analysis of rotors. BEMT-based design
and analysis codes are somewhat of an industry standard. The evaluator module of
almost all wind turbine blade design tools are based on this theory (Moriarty, 2005).

The accuracy of the BEMT model is in general reasonable for a wind turbine in normal
operating conditions. A verification study (Madsen,2004) comprising the most common
aero-elastic codes in Europe showed a typical difference of 5%-10% between measured
and simulated mean blade loads and 5%-20% difference in dynamic blade loads.
Madsen investigated both uniformly and non-uniformly loaded actuator disks and the
effect of turbulent mixing to show the validity of the BEM theory (Madsen, 1996). It
was found that BEMT, with the application of a tip correction, gives a good correlation
with the CFD results.

The rest of this Chapter is structured as follows. In Section 2.2 the classical BEMT is
explained in details. Some corrections applicable to classical BEMT towards removing
some of the shortcomings and enhancing the accuracy of the results are explained in
Section 2.3. Section 2.4 and 2.5 detail the software tool developed based on BEMT and
its validation against WTPerf (Buhl, 2004), an accredited code by the National
Renewable Energy Laboratories, NREL, USA.

15
2.2 BEMT Method
BEMT combines two methods: Blade Element or Strip Theory and Momentum or
Actuator Disk Theory which is used to outline the governing equations for the
aerodynamic design and power prediction of a wind turbine rotor (Leishman, 2000).
Momentum theory analyses the momentum balance on a rotating annular stream tube
passing through a turbine and blade-element theory examines the forces generated by
the aerofoil lift and drag coefficients at various sections along the blade. The blade-
element theory assumes that the blade can be analysed as a number of independent
element in span direction. Combining these theories gives a series of equations that can
be solved iteratively.

BEMT postulates the effects of the presence and the rotation of the rotor on the flow
field around the rotor by introducing and calculating the field of the induced velocities.
This evaluation is based on an iterative algorithm in which the induced velocities are
initially assumed and then will be re-calculated by iteration.

The basis of BEMT is well established, but there are some differences among final
BEMT models because several strategies are used to solve the non linear equations
involved and also because many corrections are proposed to increase the precision of
predictions.

2.2.1 Wind, Induced and Relative Velocity Fields

Presence of a wind turbine acting as a partial barrier in the flow field and rotation of its
blades cause changes to the wind velocity field as approaches the wind turbine. These
changes which are in the form of a reduction in the wind speed and a rotation in the
flow field are known as “induced velocity field”. Wind speed retardation depends on the
amount of the extracted energy (wind turbine loading), while the induced
circumferential velocity depends on the wind turbine angular velocity. Flow in the plane
of a wind turbine rotor, referred as rotor disk, can be considered as a combination of the
 
upstream mean velocity field, VW and the induced velocity field, VI (Maheri, 2006).

  
V  VW  VI (2.1)

16
For a moving aerodynamic surface, the aerodynamic performance depends on the
“relative velocity field”. In case of a wind turbine blade the relative velocity at a general
point P located on the blade, as seen by the blade, is defined as:

  
Vrel  V  VP (2.2)

 
in which, V is the flow field given by Equation (2.1) and VP is the velocity of point P
on the blade.

In order to find the relative velocity, first the following systems of coordinates are
defined.

 Cartesian system of coordinates x  y  z

 Disk (rotor) system of coordinates x  r  , in which x is the rotor axis, r  is

the rotor plane, r is the radial coordinate limited between zero and rotor radius
0  r  R and  is the azimuth angle, measured from horizontal 3 O’clock-
clockwise.
 Blade system of coordinates n  t  s , in which n is the axis normal to the blade,
t  s is the blade plane, n  t is the aerofoil plane and s is the blade span-wise
coordinate.

These two systems of coordinates are shown in Figure (2.1). In this figure  stands for
the cone angle,  is the yaw angle and  is the rotor angular velocity. Here, it is
assumed that wind velocity has no vertical components.

17
 Rotor disk
r  (front Rotor disk r 
view) (side view)
z

n
z
Rotor axis,
x
y


n
r
t s r


s

(a) (b)

y
Rotor disk r 
(top view)

t x


VW
n
n
r


s

(c) (d)
Figure 2.1-Rotor and blade systems of coordinates

Using the n  t  s system of coordinates, wind velocity VW can be expressed as:

VWt  VW cos  sin   sin  cos  sin (2.3.a)

VWs  VW cos  sin   sin  cos  cos  (2.3.b)

18
VWn  VW cos  cos   sin  sin  cos  (2.3.c)

Equations (2.3.a) through (2.3.c) for special case of zero yaw angle can be rewritten as:

VWt  VW sin  sin (2.4.a)

VWs  VW sin  (2.4.b)

VWn  VW cos  (2.4.c)

Cone angle  is usually very small and therefore sin   cos  , but yaw angle  can be
large enough such that sin  and cos  have the same order of magnitude. In case of
nonzero yaw angle, neglecting terms including sin  in comparison with the terms
including cos  , Equations (2.3.a) through (2.3.c) can be rewritten as:

VWt  VW sin  cos  sin (2.5.a)

VWs  VW sin  cos  cos (2.5.b)

VWn  VW cos  cos  (2.5.c)


Induced velocity field, V I has two components of V Ix and V I . V Ix is directed opposite

to the axial component of the wind velocity and V I is due to induced angular velocity
in the opposite direction of the rotor angular velocity. The components of induced
velocity field, V Ix and V I can be related to the axial component of wind speed and the
tangential velocity of a general point on the blade located at a radial location r through
the following equations:

a  VIx VWx (2.6)

19
a   VI r (2.7)

where a and a  are the axial and rotational induction factors respectively. Using
induction factors and knowing that the induced velocity components are in the opposite
direction of the axial wind and blade rotation, one can write the induced velocity vector
at a general point in x  r  system of coordinates in terms of axial and rotational
induction factors and then transform it to n  t  s system of coordinates as:


VI  aVWx iˆ  raeˆ (2.8)


VI  aVW cos  cos eˆn  raeˆt (2.9)

The velocity of point P at radial location r on a blade rotating at an angular speed of

 is given by:


VP  reˆt (2.10)

  
Substituting for VW from Equation (2.5), VI from Equation (2.9) and VP from Equation
(2.10) into Equations (2.1) and (2.2), and combining Equations (2.1) and (2.2) leads to:


Vrel  VW cos  cos  (1  a)eˆn
 VW sin  cos  sin  r(1  a )eˆt (2.11)
 VW sin  cos  cos eˆs

Figure (2.2) shows the relative velocity in the plane of blade aerofoil. Inflow angle and
normalised in-plane relative velocity can be derived from this figure as

cos  cos  (1  a)
tan   (2.12)
r (1  a )  sin  cos  sin

Vrel | n t cos  cos  (1  a)

 (2.13)
VW sin 

20
where local velocity ratio,  r is defined as

r
r  (2.14)
VW

 t
Vrel | n t
r(1  a)  VW sin  cos  sin
n
VW cos  cos  (1  a)

Figure 2.2-Relative velocity in aerofoil plane

In the case of zero yaw angle,   0 , Equations (2.11), (2.12) and (2.13) can be rewritten
as:

Vrel  VW cos  (1  a)eˆn  r(1  a)eˆt (2.15)

cos  (1  a)
tan   (2.16)
r (1  a )

Vrel | n t cos  (1  a)
 (2.17)
VW sin 

and the velocity diagram changes to Figure (2.3).

Vrel  Vrel | nt

 t
r(1  a )

n
VW cos  (1  a)

Figure 2.3-Relative velocity in aerofoil plane; zero yaw

21
It should be noted that the cone angle  is very small (about 6-7 degrees). In most of
published papers and text books, the effect of cone angle on the flow kinematics is
completely neglected ( cos   1 , sin   0 and eˆr  eˆs ).

2.2.2 Momentum Theory

Momentum theory applied to the wind turbine aerodynamic is based on three basic
assumptions:
1. Axisymmetric flow
2. Steady flow
3. Frictionless flow

Figure (2.4) shows an axisymmetric flow through a wind turbine and typical qualitative
variations of pressure, velocity and rotation between upstream of a wind turbine and far
wake behind it.

Wind turbine rotor disk

P , VW

Stream tube

Rotor axis, x

Pressure field

P

VW Wind velocity
field

∞ Disk Far wake

Figure 2.4-Pressure and velocity variations

22
Now, flow through the rotor disk can be easily found by superposing the x-components
of the upstream wind and the induced velocities.

Vd  VWx  VIx  VW (1  a) (2.18)

As wind approaches a wind turbine it slows down. This retardation is a direct result of
the presence of the wind turbine acting as a partial barrier in the flow field. Moreover,
rotation of blades imposes a rotation in the velocity field. Further reduction in wind
speed behind the rotor disk is partly due to extracting energy from wind and partly due
to energy losses due to rotation caused by rotating blades. Since in upstream no energy
is extracted from the wind yet, a reduction in air velocity causes an increase in pressure.
This is equivalent to conversion of energy from kinetic to potential form.

2.2.2.1 Thrust and Torque Coefficients in Terms of Induction Factors

Applying the x-component of linear momentum equation to the annulus control volume
shown in Figure (2.5), gives thrust force as dTrotor  dQ(V  VFW ) ,where the volume

flow rate dQ  Vd dAdisk , V  VW and Vd  VW (1  a) and therefore

dTrotor  VW (1  a)(VW  VFW )dAdisk (2.19)

CV

dQV dT dQVFW

Figure 2.5-Annulus control volume; Linear momentum balance

Applying the energy equation for the same control volume gives the turbine power.

23
dP  0.5Vd dAdisk (VW2  VFW
2
) (2.20)

Turbine power can also be obtained by multiplying the thrust force and flow velocity at
the disk.

dP  dTVd (2.21)

Combining Equations (2.19), (2.20) and (2.21) concludes

VFW  (1  2a)VW (2.22)

By substituting VFW back into Equation (2.19), dTrotor is determined in terms of the wind
velocity at the upstream and the axial induction factor.

dTrotor  2a(1  a) VW2 dAdisk (2.23)

Thrust coefficient by definition is

dT
CT  , (2.24)
1 2 VW2 dAdisk

therefore as a result of Momentum Theory it becomes:

CT  4a(1  a) (2.25)

To determine torque coefficient by the Momentum Theory one can start from applying
the angular momentum equation about the x-axis for the control volume shown in
Figure (2.6) to find a relation between the rotation in far wake and circumferential
velocity at disk as r  
2
FW  
 r 2 disk  2rr / 2disk  2rVdisk

. Since the

circumferential velocity Vdisk is only due to induction, one can substitute V   VI from

Equation (2.7) in the above equation to find r 2 FW .

24
r  
2
FW  2r 2 a (2.26)

CV

dQ(r 2 ) disk dQ(r 2 ) FW

Figure 2.6-Annulus control volume; Angular momentum balance between disk and Far
Wake

Applying angular momentum equation about x–axis for the control volume shown in
Figure (2.7), the applied torque on the rotor will be determined.

 
dM x  dQ r 2 FW  
 r 2  (2.27)

Combining Equations (2.26) and (2.27) gives the rotor torque as

dM x  2Vw (1  a)ar 2 dAdisk (2.28)

Torque coefficient is defined as

dM x
CM  , (2.29)
1 VW2 rdAdisk
2

and finally, as a result of the Momentum Theory it becomes

CM  4r a(1  a) (2.30)

25
 CV

dQ(r 2 )  dQ(r 2 ) FW
dM

Figure 2.7-Annulus control volume; Angular momentum balance

2.2.2.2 Effect of Tip and Hub Losses on Thrust and Torque Coefficients
In momentum theory, the axisymmetric flow is the basic assumption, which holds if the
turbine rotor has an infinite number of blades with zero chord length. In the case of a
real turbine with a finite number of blades, the induced velocity on the blades is
different from the mean induced velocity in the flow annulus and therefore
circumferential symmetry does not hold. The non-uniformity of the induced flow field
makes the actual local CT and C M to be smaller than the expected values by the optimum
actuator disk theory. The departure of the induced velocity, CT and C M from their
momentum theory values is more significant near the tip and root of the blade. These
deviations from the uniform induced velocity flow field are called tip and hub losses.
The overall loss factor, F is defined as

F  Ftip Fhub (2.31)

in which Ftip is unity at inboard parts of the blade and takes smaller values near the tip of

the blade and Fhub is unity at outboard parts of the blade and takes smaller values near
the hub of the blade.

Most of industrial approved and commercial software such as WTPerf (Buhl, 2004) and
AeroDyn (Moriarty, 2005) use Wilson-Walker model, in which loss factor F is directly

26
applied to the disk velocity, Vd  FVW (1  a) and the difference between the free stream

velocity and far wake velocity is defined as VW  VFW  2aVW . With the above
assumptions thrust and torque coefficients can be calculated as:

CT  4Fa (1  a) (2.32)

C M  4r aF (1  a) (2.33)

For this model the relative velocity diagram becomes as shown in Figure (2.8) and the
normalised relative velocity becomes as given by Equation (2.34).

Vrel | n t (1  a) F cos 
 (2.34)
VW sin 

Vrel  Vrel | nt

 t
r(1  a )

n
VW cos F (1  a)

Figure 2.8-Relative velocity in aerofoil plane; zero yaw; Wilson-Walker model

Momentum theory predicts a parabolic variation for thrust coefficient with a maximum
value of 1 at a  0.5 , while the experimental data show that CT keeps increasing for

a  0.5 . For small axial induction factors, 0  a  ac  0.4 , known as light loading

state, predicted thrust coefficient by the momentum theory is in a good agreement with
the experimental data. In the case of heavy loading state, where a  ac , predicted CT

departs dramatically from its actual value. In the extreme loading situation, a  1 , wind
turbine acts as a drag driven device with a thrust coefficient of CT  (CT , Drag ) max  2

rather than CT  0 as predicted by Equation (2.25). Extrapolating Equation (2.25), with

a maximum value of CT  2 at a  1 , predicts reasonable values for CT . Separating light
and heavy loading state, Equation (2.32) can be re-written as

27
CT  4Fa (1  a) if a  ac (2.35.a)

CT  B0 a 2  B1a  B2 if a  ac (2.35.b)

where

2
B0   4F (2.35.c)
(1  ac ) 2

 4a c
B1   4F (2.35.d)
(1  ac ) 2

4a c  2
B2  2  (2.35.e)
(1  ac ) 2

2.2.2.3 Prandtl Tip and Hub Loss Factors

Among some theories for estimating the tip and hub loss factors, Prandtl theory is
simple and efficient and also gives acceptable results (Moriarty, 2005). In Prandtl theory
tip and hub loss factors are defined as

cos 1 exp(  f tip )

2
Ftip  if f tip  7 (2.36.a)


Ftip  1 if f tip  7 (2.36.b)

where

B( R  r )
f tip  (2.36.c)
2r sin 

and

2
Fhub  cos 1 exp(  f hub ) if f hub  7 (2.37.a)


28
Fhub  1 if f hub  7 (2.37.b)

where

B(r  Rhub )
f hub  (2.37.c)
2 Rhub sin 

In the above equations, B is the number of blades, R and Rhub are the rotor and hub

radius and  is the inflow angle which can be obtained from Equation (2.12) or
Equation (2.16).

2.2.3 Blade Element Analysis

Blade element (strip) theory is based on the main assumption: there is no span-wise
pressure gradient and therefore each segment can be analysed independently as a two-
dimensional aerofoil. In practice, a three dimensional blade is divided into a number of
segments and each segment is analysed as a two dimensional aerofoil. Flow kinematics
around each segment (Figure (2.9)) is given by Figure (2.8).

Segment i

n seg
1 2

Rhub
ri

Figure 2.9-Discretisation of blade into nseg segments

29
2.2.3.1 Thrust and Torque Coefficients via Blade Element Force Analysis
Figure (2.10) shows a blade segment (element) subjected to the aerodynamic forces in
the same system of coordinates as introduced in Figure (2.1). Assuming 2-dimentional
flow on the aerofoil and neglecting radial forces on the blade ( dFs  0 ), thrust force on

the element can be obtained as dT  dFn or

dT  dL cos   dD sin  (2.38)

Lift and drag coefficients are defined as

dL
CL 
12 V 
2
(2.39)
rel dAe
n t

dD
CD 
12 V 2
(2.40)
rel dAe
n t

dFt dL

dFn
n

 dD

Figure 2.10-Blade element force analysis on a typical segment

where Vrel nt is the relative velocity in the n  t plane (see Figures (2.2) and (2.3)) and

dAe  cdr is the element area. Combining Equations (2.38), (2.39) and (2.40) gives
thrust force on a blade element as

2
  C
dT  1 c Vrel2 n t L cos   C D sin  dr (2.41)

and for a turbine with B blades it becomes

30
dT  1 Bc Vrel2
2
  C
n t L cos   C D sin  dr (2.42)

Using Equations (2.24) and (2.17), thrust coefficient can be written as

 r cos 2  (1  a) 2 C L cos   C D sin  

CT  (2.43)
sin 2 

where  r , local solidity ratio, is defined as

Bc
r  (2.44)
2r

Aerodynamic forces on the blade element also produce a torque about the rotor axis
equal to dM x  rdFt (Figure (2.10)). Recalling Equations (2.39) and (2.40), for a

turbine with B blades the generated torque about the rotor axis can be expressed as

dM x  1 Bc Vrel2
2
  Cn t L sin   C D cos  rdr (2.45)

Inserting the above result into the definition of the torque coefficient C M , Equation
(2.29), yields to

 r cos 2  (1  a) 2 C L sin   C D cos  

CM  (2.46)
sin 2 

2.2.4 Blade Element Momentum Theory, BEMT

Equating thrust and torque coefficients obtained from the blade element force analysis
(with the assumption of zero drag force) and those obtained from momentum theory is
the base of the BEMT. Neglecting drag force in Equations (2.43) and (2.46), thrust and
torque coefficients will become

 r cos 2  (1  a) 2 C L cos 
CT0  (2.47)
sin 2 

31
  r cos 2  (1  a) 2 C L sin 
CM0  (2.48)
sin 2 

Combining Equations (2.47) and (2.48) with Equations (2.35) and (2.33) gives:

CT0
1 1
a F if a  0.4  CT0  0.96F (2.49.a)
2

 B1  B12  4 B0 ( B2  CT0 )
a if a  0.4  CT0  0.96F (2.49.b)
2 B0

a tan 
a  (2.50)
r cos 

Equations (2.16)/(2.12), (2.31), (2.47), (2.49), (2.50) and two set of tabulated data for
C L and C D can be solved to find a , a  , F ,  , C L , C D and CT0 . Knowing a ,  , C L and

C D one can use Equations (2.49) and (2.50) to calculate T and CT and Equations (2.45)

and (2.46) to find M and C M . Having rotor torque M , turbine mechanical power, P can
be easily calculated by

P   dM  M (2.51)

and the power coefficient C P can be determined from the following equation.

P P
CP   (2.52)
1 V Arotor
3 1 VW3 R 2
2 W 2

2.3 BEMT Corrections

BEMT is based on three main assumptions: steady flow, infinite number of blades and
axisymmetric flow. These assumptions make BEMT bounded within many limitations.
In practice most of these limitations can be removed by (i) applying some corrections,
32
(ii) averaging, and (iii) employing further assumptions to the original concepts. For
example, the ground shear and yaw error contradict the basic assumption of
axisymmetric flow. Dividing rotor disk area into a number of sectors and averaging the
results is a means to include the effect of ground shear. By employing tip and hub loss
correction factors, the effect of finite number of blades can be incorporated in the
original BEMT. In case of an unsteady flow, assuming that equilibrium in the wake is
maintained at each time step (Snel, 1995), BEMT can be used for determination of the
dynamic flow conditions at the rotor disc at each time step.

The accuracy of predicted aerodynamic loads on blades using BEMT strongly depends
on the accuracy of the lift and drag coefficients used as well as the validity of its
fundamental assumptions. Assuming that the fundamental assumptions are valid, or
alternatively violation of the validity of assumptions has been compensated by applying
corrections, using the wind tunnel experimental data for lift and drag coefficients makes
BEMT somewhat more accurate when compared to CFD methods. Models based on
BEMT are sensitive to the aerodynamic characteristics of the aerofoils used in the blade
(Tangler, 2002, 2004). Therefore, having a reliable model for predicting the post stall
aerodynamic coefficients and considering the effect of stall delay due to blade rotation
(Du, 1998) are also necessary for acceptable results. Sanderse (2011) and Snel (1998,
2003) give a comprehensive review of the BEMT and the corrections applicable to
remove some of its limitations.

2.3.1 Angle of Attack Corrections

Lift and drag coefficients are functions of the angle of attack and Reynolds number.
Angle of attack is in turn a function of the velocity field and the blade geometry and can
be expressed as

     0  pitch   c (2.53)

In the above equation  is the inflow angle, (Equation (2.16)) and  0 stand for pretwist,

pitch is the blade pitch angle and  c refers to cascade correction. The cascade

correction to the angle of attack has two components

33
 c  1   2 (2.54)

where  1 accounts for the effect of finite aerofoil thickness and  2 accounts for the
effect of finite aerofoil width (Spera 1994).

B cos  0 Aa
 1  (2.55)
2rc

1  1 (1  a)r (1  a)r 
 2  tan  tan 1  (2.56)
4 (1  2a ) R R 

 0 is the inflow angle prior to rotational induction, ( a   0 in Figure (2.3)) and Aa is the
aerofoil cross section area, normally taken as Aa  0.68ct max , where t max is the
maximum thickness of the aerofoil.

2.3.2 Effect of Ground Shear

The wind speed profile trends to a lower speed as the altitude move closer to the ground
level and this is designated as wind shear. As can be seen from Figure (2.11) the wind
speed at certain height (datum of azimuth angle) above ground can be predicated as a
function of height above ground z and the roughness length z 0 . The wind speed caused
by effect wind shear is given by

 z  h 
V  Vhub ln   / ln  hub  (2.57)
 z0   z0 
in which

z  hhub  r sin (2.58)

34
 3
Rotor disk

Wind speed
profile
Blade at
position
2 Sector 3 Datum of
Sector 1 azimuth
 1  0 angle

hhub
Sector 2 +
r

2

Ground

Figure 2.11-Discretising rotor disk into nsec sectors (here, nsec  3 ) and the azimuth angle
associated to each sector

2.4 The software, AWTSim-Advanced Wind Turbine Simulation

Based on the theory explained in Sections 2.2 and 2.3 a software tool has been
developed for the analysis of horizontal axis wind turbines. This software includes two
primary modules for blade discretisation and BEMT calculations, and some secondary
modules for calculating the annual average power and blade external/internal loading. It
also includes modules required for analysis of wind turbines utilising unconventional
blades which are explained in details in Chapter 3.

In order to analyse the aerodynamic performance of constant speed stall regulated wind
turbine, AWTSim requires five sets of inputs:

1. Blade geometry and topology data. These data include span-wise distribution of
chord c(r ) , pretwist  0 (r ) , aerofoil AF (r ) and aerofoil maximum thickness tmax (r )

35
 as well as rotor radius R (or diameter D ), hub radius Rhub and blade pitch angle

pitch .

2. Blade aerodynamic data. For each aerofoil used in the blade tabulated   C L and
  C D data are required.

3. Rotor characteristics: rotor speed  , cone angle  , number of blades B and hub
height hhub .

4. Wind turbine operating data: Wind speed at hub height VW and yaw angle  .

5. Site data: Site average wind speed Vav and probability distribution function, and

ground surface roughness length z 0 .

2.4.1 Blade Discretisation

Algorithm (2.1) describes steps for discretisation of a wind turbine blade as required for
BEMT. This algorithm requires blade topology data file as well as the number of
segments and sectors. Tables (2.1) and (2.2) show a typical set of the input data defining
the topology of a blade. Table (2.3) shows the results of discretisation based on
Algorithm (2.1). In this algorithm the normalised chord, radial coordinate and aerofoil
maximum thickness are defined as c*  c / R , r*  r / R and *
t max  t max / c ,

respectively.

36
Algorithm 2.1-Blade discretisation

Given:
 n seg and nsec
 *
Blade data file { R , Rhub , c * (r * ) ,  0 (r * ) , AF (r * ) and t max
*
(r * ) }

Step 1- Calculate length of each blade segment r  1 Rhub nseg

* *
 
Step 2- Find the centre of each blade segment:
2.1. r1*  Rhub
*
 0.5r *
2.2. For i  2 : nseg , do: ri*  ri*1  r *
Step 3- Find blade chord and pretwist at the centre of each segment:
3.1. For i  1 : nseg , do: calculate c * (ri* ) and  0 (ri* ) by linear interpolation within
the blade data file.
Step 4- Find blade aerofoil index and aerofoil thickness over each segment:
Step 5- Calculate the arc length of each rotor disk sector   360 / nsec
Step 6- Find blade position (azimuth angle) for each rotor disk sector:
6.1.  1  0 (horizontal-3 O’clock)
6.2. For j  2 : nsec , do:  j   j 1  

Table 2.1-Blade data file-Part 1

r* c * (r * )  0 (r * ) Table 2.2-Blade data file-Part 2
0.086 0.053 6.270 Starts Ends
AF *
0.109 0.056 6.100 t max @ @
index *
0.155 0.063 5.764 r r*
0.223 0.070 5.470 1 0.24 0.086 0.269
0.269 0.075 5.233
2 0.24 0.269 0.360
0.315 0.081 4.996
0.360 0.082 4.602 3 0.24 0.360 0.452
0.406 0.083 4.208 4 0.24 0.452 0.543
0.452 0.081 3.689 5 0.21 0.543 0.634
0.497 0.079 3.172 6 0.21 0.634 0.726
0.543 0.076 2.628
0.589 0.074 2.086
7 0.21 0.726 0.806
0.634 0.071 1.601 8 0.21 0.806 0.863
0.680 0.068 1.117 9 0.21 0.863 0.963
0.726 0.064 0.770 10 0.21 0.963 1.000
0.772 0.060 0.424
0.806 0.056 0.273
0.840 0.052 0.122
0.863 0.049 0.099
0.886 0.047 0.076
0.918 0.042 0.048
0.936 0.039 0.041
0.954 0.036 0.033
0.963 0.035 0.028
0.973 0.033 0.023
0.986 0.030 0.012
0.995 0.029 0.006
1.000 0.028 0.000

37
 Table 2.3-Discretised blade with nseg  20
segment AF
r* c * (r * )  0 (r * ) *
t max
# index
1 0.109 0.056 6.100 1 0.24
2 0.155 0.063 5.764 1 0.24
3 0.200 0.068 5.568 1 0.24
4 0.246 0.073 5.351 1 0.24
5 0.292 0.078 5.114 2 0.24
6 0.337 0.081 4.799 2 0.24
7 0.383 0.083 4.405 3 0.24
8 0.429 0.082 3.949 3 0.24
9 0.475 0.080 3.431 4 0.24
10 0.520 0.078 2.900 4 0.24
11 0.566 0.075 2.357 5 0.21
12 0.612 0.072 1.843 5 0.21
13 0.657 0.069 1.359 6 0.21
14 0.703 0.066 0.944 6 0.21
15 0.749 0.062 0.597 7 0.21
16 0.794 0.057 0.323 7 0.21
17 0.840 0.052 0.122 8 0.21
18 0.886 0.047 0.076 9 0.21
19 0.931 0.040 0.043 9 0.21
20 0.977 0.032 0.019 10 0.21

2.4.2 BEMT Calculator

Algorithm (2.2) shows the steps of calculating the extracted mechanical power by a
stall-regulated wind turbine based on the BEMT method explained in Sections 2.2 and
2.3. This algorithm, requires a discretised blade (as shown in Table (2.3)), rotor
characteristics and lift and drag tables corresponding to the aerofoils used in the blade.

38
Algorithm 2.2-BEMT calculator
Given:
  a ,  a , D, b,  , hhub, z0 , ,  , Vhub ,, pitch 
 n seg
 * * *

, nsec , ri , ci ,  0,i , AFi , t max,i ; i  1 : nseg and  j ; j  1 : nsec  (discretised
blade)
 C L   and CD   tabulated data for all aerofoils used in the blade
Step 1- Initialise C L,i , C D,i ,  i ,  i ,Vrel ,i  0 ; i  1 : nseg .
Step 2- Dimensionalise r  Rr * and ri  Rri* , ci  Rc i* , t max,i  ci t max,
*
i ; i  1 : n seg (

R  0.5D ).
Step 3- For each azimuth angle  j ; j  1 : nsec ,do:
3.1. Find wind shear field:
3.1.1. zi  hhub  ri sin j ; i  1 : nseg
3.1.2. S i , j  ln zi z 0  ln hhub z 0  ; i  1 : nseg
3.2. Find wind speed at the centre of each blade segment: Vi , j  VhubS i , j
Step 4- For j  1 : nsec , do:
4.1. For i  1 : nseg , do:
4.1.1. Calculate local speed ratio ri  ri  Vi , j and local solidity ratio
 r  Bci 2ri
i

4.1.2. Initialise a  1 / 3 ; a  0 ; converged  false

4.1.3. While ( converged  false ) do:
4.1.3.1. Calculate inflow angle
 Vi , j cos  cos  (1  a) 
 i  tan 1  

r
 i (1  a )  Vi, j sin  sin  j cos  
4.1.3.2. Calculate tip and hub losses F  Ftip Fhub .

cos 1 exp(  f tip ); otherwise:

2
If f tip  B( R  r ) 2r sin   7 : Ftip 

Ftip  1 .
2
If f hub  B(r  Rhub ) 2Rhub sin   7 : Fhub  cos 1 exp(  f hub )

;otherwise: Fhub  1.

1 cos F (1  a )

4.1.3.3. Calculate inflow angle at zero drag:  0  tan  

  r ,i 
4.1.3.4. Calculate angle of attack  i   i   0,i  pitch   c ;
B cos  0 Aa 1  1 (1  a)ri (1  a)ri 
 c   tan  tan 1 ;
2ri ci 4 (1  2a ) R R 
Aa  0.68ci t max,i
4.1.3.5. Use  i read off life and drag coefficients from tables: C L ,i
and CD ,i

39
Algorithm2.2-BEMT calculator-Continue
4.1.3.6. Calculate thrust coefficient at zero lift:
 r ,i cos 2  (1  a) 2 C L,i cos  i
CT0 
sin 2  i
4.1.3.7. Calculate anew . If CT0  0.96F :

anew   B1  B12  4B0 ( B2  CT0 ) 2 B0 , otherwise 
 
anew  1  1  CT0 / F 2 ; ( B0  2 (1  ac ) 2  4F ;
B1   4ac (1  ac )  4F ; B2  2  (4ac  2) (1  ac ) 2 ; ac  0.4 ).
2

4.1.3.8.   anew tan  i r ,i cos 

Calculate anew
4.1.3.9. If  a  anew   a  a  anew
   a  : converged  true ; Else:
a  anew , a  anew
 .
4.1.4. Calculate
Vrel ,i  V
i, j cos  cos  (1  a)  Vi , j sin  cos  sin j  ri (1  a) 
2 2

1
4.1.5. Calculate M i  Vrel2 ,i Bc i C L,i sin  i  C D,i cos  i ri r cos 
2
nseg
4.2. Calculate Pj   M i
i 1
nsec
1
Step 5- Calculate P 
nsec
P
j 1
j

2.5 Validation
AWTSim has been validated against the latest version of WTPerf (Buhl, 2005) using
the test wind turbine AWT-27. WTPerf is an accredited code developed and in use in
the National Renewable Energy Laboratories, NREL, USA. AWT-27 is a 2-bladed wind
turbine, one of the very few wind turbines with known specifications to public. AWT-
27 is taken as the case for study all through this project. Table (2.4) gives AWT-27
specifications. In order to compare the results obtained by AWTSim with those of
WTPerf, the input data files of a typical test run of WTPerf (obtained from the NREL
website) have been downloaded and used to generate the input data files of Tables (2.1)
and (2.2) as required by AWTSim. The design pitch angle for AWT-27 is 1.2° to stall (-
1.2). However, in order to compare the results with available results, pitch angles 0, 1
and 2 degrees to stall are considered for simulation. Figures (2.12) and (2.13) show the
power curves and thrust curves for this wind turbine obtained by WTPerf and AWTSim.

40
It can be observed that the difference between the predicted mechanical power and
thrust by two codes is very small. The reason for this difference can be explained as
follows. The blade aerodynamic loading and therefore the rotor mechanical power are
very sensitive to the accuracy of the predicted angle of attack. The more the number of
segments, more accurately the angle of attack is calculated. WTPerf is using 17 unequal
segments in this run while AWTSim has divided the blade into 20 equal segments.

Table 2.4-AWT-27 wind turbine

Rotor diameter 27.514 m
Hub height 30 m
Cone angle 7°
No. of blades 2
Aerofoils S800 Series
Original constant rotor speed 53.3 rpm
Rated power 300 kW
Design blade pitch angle 1.2° (to stall)
Hub radius 1.184 m
Yaw angle 0°
Tilt angle 0°

350
Rotor Mechanical Power (kW)

300

250

200

150

100 WTPerf (Pitch = 2) AWTSim (Pitch = 2)

WTPerf (Pitch = 1) AWTSim (Pitch = 1)
50
WTPerf (Pitch = 0) AWTSim (Pitch = 0)
0
5 10 15 20 25
Wind Speed (m/s)

Figure 2.12-AWTSim versus WTPerf (Buhl, 2004)-Power curves at different pitch

angles

41
 40

35

Thrust Force (kN)

30

25

20 WTPerf (Pitch = 2) AWTSim (Pitch = 2)

WTPerf (Pitch = 1) AWTSim (Pitch = 1)
15 WTPerf (Pitch = 0) AWTSim (Pitch = 0)

10
5 10 15 20 25
Wind Speed (m/s)

Figure 2.13-AWTSim versus WTPerf (Buhl, 2004)-Thrust force at different pitch

angles

2.6 Summary

In this chapter the theory behind AWTSim, a software tool developed for the
aerodynamic analysis of stall-regulated wind turbines was explained. AWTSim is based
on BEMT method. The main modules of AWTSim, the blade discretiser and the BEMT
calculator, were also explained. At the end of the chapter, it was shown that AWTSim
produces accurate results while compared with the accredited code WTPerf.

To this end, AWTSim is only capable of simulating constant speed stall-regulated wind
turbines which do not utilise any active control systems. In Chapter 3, the modifications
required making AWTSim applicable for variable speed rotors and blades with active
control systems are explained.

42
3 Aerodynamic
Performance Analysis of
Wind Turbines Utilising
Aerodynamic Control
Systems

43
3.1 Introduction
In Chapter 2, the theory behind the aerodynamic performance analysis tool was
explained. It was also shown that how the software tool can be used for simulation of a
constant-speed stall-regulated wind turbine. In this chapter, first power and load control
systems will be classified based on their effect and types and then the controller
simulation method is explained. It is followed by details on further development of the
software tool developed in Chapter 2 towards making it suitable for analysis of both
constant speed and variable speed rotors utilising blades with variable pitch, flap,
microtab and variable length blades. For each type a case study is run to evaluate the
performance of the software. Finally, a preliminary comparison of different types of
control systems is carried out and reported.

3.2 Power and Blade Load Control

Wind turbine rotor mechanical power and blade aerodynamic loading depend on wind
speed. Wind speed is a stochastic parameter which can be represented by

V  V  Vˆ (3.1)

in which, V is the instantaneous wind speed, V is the mean value and Vˆ is the
turbulence. Wind turbulence spectrum includes variations with time scales from a
fraction of seconds to several seconds.

Wind turbines utilise power and load control systems in order to:
1. increase extracted wind power at low wind speeds,
2. regulate the rotor power at its rated value rated value preventing the generator from
overloading at high wind speeds,
3. alleviate the steady and quasi-steady aerodynamic loads on blades (e.g. due to a
gradual rise in wind speed, or variations of aerodynamic load as the blade passes
through wind shear for wind turbines with lower rotor speeds), and
4. damp and alleviate the fluctuating behaviour of aerodynamic loads, mainly
produced by wind turbulence and ground shear for wind turbines with high rotor
speeds.

While the first two functions above are very important for all wind turbines irrespective
of their type and size, the third and fourth functions of a control system are of prime
44
importance in larger wind turbines, in which the long blades experience massive loads.
Since the present study is mainly focused on improving the energy capture capability of
smaller wind turbines, the first two functions are of prime interest here.

3.2.1 Power and Blade Load Control Systems via Blade

Power and blade load control can be carried out either through devices installed on
blades (or blade itself), or via mechanism affecting the rotor as a whole.

Figure (3.1) shows different conventional and nonconventional power and load control
mechanisms affecting the blade performance. Some of these control systems respond
only to wind variations with large time scales, while some other have shorter response
time and therefore can be used for controlling the effect of wind variations with smaller
time-scales.

Power/Blade Load
Control Systems
Affecting Blade
Performance

Blade Cross-section/
Blade Span Blade Twist
Aerofoil Topology

Elastic
Pitch Angle
Twist

Telescopic Morphing Adaptive

Microtab Flap Individual Conventional
Blades Aerofoils Blades

Figure 3.1-Different control systems affecting blade performance

According to this figure, control systems can be categorised based on the controlling
parameters affecting the blade span, blade cross-section (aerofoil) and blade twist.

Figure (3.2) shows a two-dimensional flow kinematic diagram around the blade at span
location r .

45
 Inflow t
Angle of
attack angle 
 Twist angle
Tangential   0  pitch   e
velocity Relative
f (r) velocity Vrel

n
Axial
velocity
f (VW )

Figure 3.2- Flow kinematic diagram at span location r

All control systems above, besides the telescopic blades, change the wind turbine
performance via imposing a change in the angle of attack  . The angle of attack is
related to inflow angle  and the blade twist angle (elastic twist  e , pretwist  0 and

pitch angle pitch ) via Equation (3.2).

     e   0  pitch (3.2)

Conventional and individual pitch control systems use the blade pitch angle pitch as the
controlling parameter. Adaptive blades control the performance via elastic twist angle
 e produced as a result of elastic coupling in the material of the blade. Stall-regulated
rotors rely on the influence of the inflow angle  (as a non-independent controlling
parameter) on the angle of attack and consequently the rotor output power.

3.2.2 Power and Blade Load Control Systems via Rotor

Referring to Equations (2.11), (2.12), (2.45) and (2.51) evidently yaw angle  and rotor
speed  have influence on both rotor mechanical power and blade loading and
therefore can be used as controlling parameters. Another parameter (not discussed in
Chapter 2) is the rotor tilt angle and that can be also used for affecting the flow
kinematics around the entire of the rotor. Both yaw and tilt control systems need heavy
mechanical systems to change the direction of the rotor. This makes these systems
highly sluggish and therefore unsuitable for accurate power and load control.
Controlling power through yaw or tilt angle also produce various complications such as
46
unbalance rotor loading and consequently increasing fatigue loading on blades and other
components. Both systems are rarely used in modern wind turbines. On the other hand
rotor speed can be easily controlled via changing the electrical load (generator output).
Variable speed rotors are very popular, in particular for smaller wind turbines.

3.2.3 Active Control versus Passive Control

In view of the above discussion the control systems can be also categorised as either
active or passive. In an active control, the controlling parameter is independent of the
wind turbine operating condition (state) and is adjusted through commands from
controller. It provides a full control on rotor power and/or blade loading. These
controllers influence the flow kinematics around a portion of the blade (i.e. flap,
microtab, morphing aerofoil, telescopic blades), the entire of the blade (i.e. conventional
and individual pitch control systems), or the entire of the rotor (i.e. yaw, tilt and rotor
speed).

In passive control, the controlling parameter depends on wind turbine operating

condition. In fact, no separate controller is in place and the blade itself acts as a
controller. A change in wind turbine operating condition (e.g. wind speed) affects the
flow kinematics around the entire of the blade either via changes in inflow angle (i.e. in
case of stall-regulated blades), or via changes in both inflow angle and blade elastic
twist (i.e. adaptive blades). A change in wind turbine operating condition leads to a
partial control on rotor power and/or blade loading

In case of adaptive and telescopic blades as well as blades utilising flaps, microtabs and
morphing aerofoils, corrections are required to be applied to the baseline blade topology
and/or geometry and/or aerodynamic characteristics.

3.3 Simulation of Controller

The aerodynamic performance of a wind turbine depends on the characteristics of the
control system in place (type, response time, controlling parameter, controllable
parameter, etc) as well as the wind turbine rotor characteristics (e.g. blade topology and
size, number of blades and rotor angular speed) and the operating condition (e.g. mean
wind speed at hub elevation, wind direction and turbulence level). Therefore, to be able
to simulate a wind turbine, the behaviour of the control system is also to be simulated
along with the aerodynamic behaviour of the wind turbine. This, however, is not
47
practical as it requires having the control system designed prior to the blade. An
alternative solution to this is to assume that the controller is capable of delivering the
expected functions perfectly. This implies that the controlling parameter is always
adjusted at its best possible value which leads to the best (goal) performance. Adapting
this approach, the optimum (best possible) controlling parameter, which optimises the
performance measure(s) can be found via solving an optimisation problem.

In case of variable speed rotors, the rotor speed is also required to be known. Two
approaches can be adopted to find the rotor speed. In the first approach, the rotor speed
is found such that the tip speed ratio is retained constant at its design value, where the
power coefficient C P has its maximum value. This approach serves well for
conventional blades. In the second approach, the rotor speed is found along with the
main controlling parameter (if there is any) via an optimisation procedure. For example,
in case of a variable speed pitch-controlled wind turbine, both rotor speed and the blade
pitch angle are to be found via solving an optimisation problem, with the controllable
parameter as variables to be found. In this case, control laws may also apply. For
example: the rotor power is controlled via changing rotor speed for    c , and via

changing the pitch angle for    c (  c is a constraint on the rotor speed); or: the blade
load is controlled via changing the pitch angle only, while the rotor power is controlled
via changing both rotor speed and pitch angle. The advantage of the second approach is
that it does not require the design tip speed ratio (in fact, it can be used to find the
design tip speed ratio), and that it is a general method applicable to all types of control
systems and allows implementing control laws.

In this project, all of controlling parameters (including rotor speed) are obtained via
solving an optimisation problem, which can formulated as follows:

max Pqi ; i  {1,2,..., nq } (3.3.a)

subject to

P  Prated (3.3.b)

qi ,l  qi  qi ,u ; i  {1,2,..., nq } (3.3.c)
48
In Equation (3.3), P is the rotor mechanical power at a given wind speed, qi stands for

the i-th controlling parameter limited to the interval [qi ,l , qi ,u ] . Number of independent

controlling parameters, nq depends on the type of the blade and the rotor (constant speed

or variable speed). Table (3.1) shows the number of controlling parameters for the
blades studied in this project.

Table 3.1-Number of independent controlling parameters

Number of independent controlling parameters
Constant speed rotor Variable speed rotor
Conventional blades-stall regulated 0 (passive control) 1
Conventional blades-pitch controlled 1 2
Blades utilising flap 1 2
Telescopic blades 1 2
Blades utilising microtab N MT (No. of microtabs) N MT  1
Adaptive blades 0 (passive control) 1

Optimisation methods can be classified in many different ways, including:

 Gradient-based versus derivative free methods. Gradient based methods needs an
explicit form of objective function correlating the objective to variables. Derivative-
free methods, on the other hand, do not require a known objective function and can
be applied to problems in which the objective or constraints are calculated through
numerical analysis (e.g. CFD, finite element analysis, BEMT)
 Global versus local search methods. Global search methods find the global optimum
irrespective of the number of local optima, whilst local search methods only find
local optima in the neighbourhood of the search initial point (or points).
 Constrained versus unconstrained methods. Some optimisation methods are
applicable to unconstrained problems only, while some other methods can handle
both equality and inequality constraints.

In order to select the best optimisation method for optimisation problem of Equation
(3.3), we need to identify the type of this problem first. Obviously, this is a constrained
problem and therefore we need an optimisation method capable of dealing with
constrained problems. As explained in Chapter 2, the rotor power can be calculated only
via a numerical iterative procedure (Algorithm (2.2)-BEMT Calculation). Since there is
49
no explicit correlation between the power and controlling parameters, gradient-based
optimisation methods cannot be used for this optimisation problem. This leaves us with
derivative free optimisation methods. Figure (3.3) shows three possible behaviour of
objective P with respect to a typical controlling parameter qi . Figure (3.3.a) shows a
case where there is only one optimum. In the second case, Figure (3.3.b), there are
multiple local optimum points. In the third case, Figure (3.3.c), there are multiple local
optima and multiple global optima. In view of this figure, a global optimisation method
should be employed to solve this problem as employing a local optimisation method
will lead to a local optimum point, unless the initial point is located within interval Q
(see Figure (3.3)).

Since this is a constrained problem with no known explicit correlation between the
objective and variables, and that the solution space may have multiple local/global
optimum points, meta-heuristic optimisation methods such as genetic algorithm (GA)
seem to be the natural choice for solving this problem. Meta-heuristic methods are
derivative-free global optimisation methods applicable to constrained problems. These
methods, however, need significant computational time to find the global optimum.

Hill-climbing and pattern search methods, on the other hand, find a local optimum in
the neighbourhood of the initial point. If the initial point is selected wisely (within the
interval Q ), a hill-climbing search finds the global optima. Hill-climbing and pattern
search methods are very efficient for problems with small number of variables.
Referring to Table (3.1), it can be observed that except one case the number of
independent controlling parameters is limited to 2, making hill-climbing and pattern
search alternative solutions subject to the condition of starting with a suitable initial
point.

Tables (3.2) and (3.3) summarise different types of blades which have been investigated
in this project. Depending on the type of the control system in use and the
corresponding controlling parameter(s) different optimisation methods are employed.
The rest of this chapter elaborates on these optimisation methods and the corrections
which are required to be applied to the BEMT calculator and the baseline blade
topology/ geometry/ aerodynamic characteristics.

50
 P

Q
Prated

(a)

qi
q i ,l q i ,u

P
Q

Prated

(b)

qi
q i ,l q i ,u

P
Q

Prated

(c)

qi
q i ,l q i ,u

Local optimum Global optimum

Figure 3.3- Three possible behaviour of objective P with respect to controlling
parameters qi .

51
 Table 3.2-Simulation of constant-speed wind turbines with different types of blades
Optimisation Corrections
Type of Corrections
method for applicable to
Blade type Independent controlling control on applicable to
finding blade
{parameters defining blade} parameters power or blade topology/
controlling aerodynamic
steady load geometry
parameters characteristics
Conventional blade-stall regulated Partial-
N/A N/A N/A N/A
R, c,  0 , AF , t max  Passive
Conventional blade-pitch controlled pitch angle
Full-Active Hill climbing N/A N/A
R, c,  0 , AF , t max  pitch s  pitch  pitch e
Blade utilising flap Flap deployment angle
R, c,  0 , AF , t max , RF ,s , RF ,e , d F   F , s   F   F ,e Full-Active Hill climbing N/A C L , C D 
Telescopic blade Blade deployment length
R, c,  R, c,  0 , AF , t max 
0 , AF , t max , RT , s , RT ,e  RT ,s  RT  RT ,e Full-Active Hill climbing N/A

Blade utilising microtab N MT microtab states N/A C L , C D 

R, c,  0 , AF , t max , RMT ,s , RMT ,e , N MT  mti   1,0,1
Full-Active GA search

52
 Table 3.3-Simulation of variable-speed wind turbines with different types of blades
Optimisation Corrections
Type of Corrections
method for applicable to
Blade type Independent controlling control on applicable to
finding blade
{parameters defining blade} parameters power or blade topology/
controlling aerodynamic
steady load geometry
parameters characteristics
Conventional blade-stall regulated
rotor speed  s     e Full-Active Hill climbing N/A N/A
R, c,  0 , AF , t max 
rotor speed and pitch angle
Conventional blade-pitch controlled
 s    e Full-Active Pattern search N/A N/A
R, c,  0 , AF , t max 
pitch s  pitch  pitch e
rotor speed and flap
Blade utilising flap deployment angle
R, c,  0 , AF , t max , RF ,s , RF ,e , d F   s    e Full-Active Pattern search N/A C L , C D 
 F , s   F   F ,e
rotor speed and blade
Telescopic blade deployment length
R, c,  R, c,  0 , AF , t max 
0 , AF , t max , RT , s , RT ,e 
 s    e Full-Active Pattern search N/A
RT ,s  RT  RT ,e
rotor speed  s     e
Blade utilising microtab N/A C L , C D 
R, c,  0 , AF , t max , RMT ,s , RMT ,e , N MT  and N MT microtab states Full-Active GA search
mti   1,0,1

53
3.4 Aerodynamic Performance of Stall Regulated Wind Turbines
The same code as explained in Chapter 2 without any further modifications serves well
for simulation of constant speed stall regulated wind turbines. However, in case of
variable speed stall regulated wind turbines, the rotor speed is a controlling parameter
and needs to be determined.

3.4.1 Control Simulation of Variable Speed Stall Regulated Wind Turbines

Simulating AWT-27 wind turbine for various rotor speeds, the effect of rotor speed on
the rotor mechanical power can be observed. Figure (3.4) shows four distinct behaviours
of rotor power with respect to rotor speed at different wind speeds:
Case 1- VW  8m / s : One feasible local/global optimum ( Pmax  Prated  300kW )

Case 2- VW  11m / s : One feasible local/global optimum at a boundary ( Pmax  Prated )

Case 3- VW  13m / s : Infeasible optimum at boundary; one feasible optimum (

Pmax  Prated )

Case 4- VW  25m / s : Multiple feasible local optima; global optima ( Pmax  Prated )

It can be seen that in all cases, if the initial point in a hill-climbing algorithm is taken as
the upper limit  0   u with a negative search direction (from right to left), the
algorithm will find the feasible global optima without getting trapped in a local optima
(if there is one).

500000
450000
Rotor Mechanical Power (W)

400000
350000
300000 V=8m/s
250000 V=11m/s
200000 V=13m/s
150000 V=25m/s
100000
50000
0
30 35 40 45 50 55 60 65
Rotor Speed (rpm)

Figure 3.4-Diffrent distinct behaviours of P   curve

54
Starting with    u as the initial point and searching from right to left, the hill-
climbing Algorithm (3.1) finds the global solution for optimisation problem (3.3) for all
four behaviours of P   , as explained below:
 Case (1): The search continues from right to left until the optimum point is passed
and the power at new iteration ( k  1 ) becomes less than the power at the previous
iteration P k 1  P k . At this point the search algorithm goes back to the previous
point ( k ) and recalculates P k 1 by a new step size: step new  step old / 2 . This
processes repeats until the step size becomes less than or equal to a predefined
tolerance ( step    ).

 Case (2): The first iteration leads to P k 1  P k and therefore the search algorithm
goes back to the initial point and recalculates P k 1 by a new step size:
step new  step old / 2 . The second iteration also leads to P k 1  P k and consequently
restarting from initial point with a smaller step size. This processes repeats until the
step size becomes less than or equal to a predefined tolerance ( step    ).

 Cases (3) and (4): The first iteration leads to P k 1  P k but since P k 1  Prated the
search algorithm continues in the same direction with the same step size until
P k 1  Prated . A this point the algorithm goes back to the previous point and

recalculates P k 1 by a smaller step: step new  step old / 2 . This processes repeats until

either the step size becomes less than or equal to a predefined tolerance ( step    )

or power lies between the tolerated margin P k 1  Prated   P .

Algorithm 3.1-Hill climbing search for finding optimum rotor speed for variable speed
stall regulated WT
Given:
 step ,   ,  l ,  u ,  P
 All parameters required for running Algorithm 2.2 except 
Step 1- Initialise:  0   u
Step 2- Start with    0 and simulate wind turbine to find P ( Algorithm 2.2);
P0  P .
Step 3- While step       P  Prated   P do:
3.1.    0  step  simulate wind turbine to find P
3.2. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
 0   0  step  , P0  P ; Else: Half the step: step  0.5step

55
To show the performance of this algorithm, AWT-27 wind turbine is simulated as a
variable speed wind turbine with the following search parameters:
 Rotor speed limits: l  30rpm and  u  65rpm

 Initial step size: step  u  l  / 10  3.5rpm

 Power tolerance:  P  0.01Prated  3kW

 Rotor speed tolerance:    0.1rpm

Figure (3.5) shows the iteration points when using the hill-climbing method of
Algorithm (3.1) for four typical wind speeds. Figure (3.6) shows the number of
iterations for each wind speeds. It should be noted that an exhaustive search with a grid
of size of    0.1rpm requires u  l     1  351 points (iterations). This is 35
times more than the average number of iterations when using the hill-climbing
algorithm.
500000
Rotor Mechanical Power (W)

400000
V=8m/s
300000 V=11m/s
V=13m/s
200000
V=25m/s

100000

0
30 35 40 45 50 55 60 65
Rotor Speed (rpm)

Figure 3.5-Iteration points in hill-climbing method of Algorithm (3.1)

18
16
14
No. of Iterations

12
10
8
6
4
2
0
5 10 15 20 25
Wind Speed (m/s)
Figure 3.6-Number of iterations against wind speed using hill-climbing method of
Algorithm (3.1)

Figure (3.7) shows the results of this simulation.

56
 350 65

Rotor Mechanical Power (kW)

300 60

Rotor Speed (rpm)

250 55

200 50

150 45

100 40

50 35

0 30
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(a) (b)

0.50 45
0.45 40
0.40 35
Power Coefficient

0.35
30
Thrust (kN)

0.30
25
0.25
20
0.20
15
0.15
0.10 10

0.05 5
0.00 0
5 10 15 20 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(c) (d)

9 200
180
Root Bending Moment (kNm)

8
160
140
Tip Speed Ratio

7
120
6 100
80
5
60
40
4
20
3 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)

Figure 3.7-Results of simulation of variable-speed stall-regulated AWT-27

57
Figure (3.7.a) shows that the power curve reaches its rated value (300kW) at about a
wind speed of 12 m/s and remains constant (within the tolerated margin of Prated   P ) at
higher wind speeds. This confirms that the control simulation of Algorithm (3.1)
successfully finds the rotor speed which is required to maintain the power at its rated
value.

The power coefficient curve of Figure (3.7.c), as expected, is horizontal at lower wind
speeds having its highest value achievable by this wind turbine. The power coefficient
decreases as the power remains constant at higher wind speeds.

The rotor speed as shown in Figure (3.7.b), as expected, increases with wind speed at
lower wind speeds until it reaches its upper bound (here 65 rpm). The concave shape of
the second part of the curve is due to the fact that the baseline wind turbine is stall-
regulated and its original power curve is below the rated power at high wind speeds (see
Figure (2.12)), hence the rotor speed increases at higher wind speeds to shift the power
curve up to its rated value.

The form of the tip speed ratio curve, as shown in Figure (3.7.e) is a very interesting
result. It can be observed that the tip speed ratio is not constant at low wind as it is
normally expected. As explained earlier in Section (3.3), two approaches can be adopted
to find the rotor speed. In the traditional approach, the rotor speed is found such that the
tip speed ratio is retained constant at its design value, while in the second approach, as
proposed and used in this study, the rotor speed is found via an optimisation procedure.
This figure confirms that the second approach is superior to the first one, as a constant
tip speed ratio will lead to a fluctuating power coefficient which is obviously not
optimum.

Finally, comparing the rotor thrust force, shown in Figure (3.7.d), and the root bending
moment curves, shown in Figure (3.7.f), it can be observed that, as expected, these
curves have almost identical forms.

58
3.5 Aerodynamic Performance of Pitch Controlled Wind Turbines
The BEMT-Calculator of Algorithm (2.2), without any modifications, can be used for
simulation of pitch controlled wind turbines. In case of constant speed rotor, blade pitch
angle and in case of variable speed rotor blade pitch angle and rotor speed are the
controlling parameters that need to be determined by solving optimisation problem of
Equation (3.3).

3.5.1 Control Simulation of Constant Speed Pitch Controlled Rotors

Simulating AWT-27 wind turbine for various blade pitch angle, the effect of blade pitch
angle on the rotor mechanical power can be observed. Figure (3.8) shows five distinct
behaviours of rotor power with respect to blade pitch angle at different wind speeds:
Case 1- VW  10m / s : One feasible optimum ( Pmax  Prated )

Case 2- VW  14m / s : Multiple feasible global optimum points ( Pmax  Prated )

Case 3- VW  15m / s : One infeasible local optimum; one feasible global optimum (

Pmax  Prated )

Case 4- VW  19m / s : One infeasible local optimum at boundary; one feasible global

optimum ( Pmax  Prated )

Case 5- VW  24m / s : Multiple local optima (feasible and infeasible); one feasible

global optima ( Pmax  Prated )

600000
V=10m/s
Rotor Mechanical Power (W)

500000 V=14m/s
V=15m/s
400000
V=19m/s
300000 V=24m/s

200000

100000

0
-5 -4 -3 -2 -1 0 1 2 3 4 5
Pitch Angle (deg)

Figure 3.8-Diffrent distinct behaviours of P  pitch curve

It can be seen that in all cases, if the initial point in a hill-climbing algorithm is taken as
the upper limit pitch 0  pitch u with a search direction from right to left, the algorithm

59
will find the feasible global optima without getting trapped in a local optima (if there is
one). In Case 2 with multiple global optimum points, since the search direction is from
right to left, optimum point closer to the upper limit pitchu is selected as the solution.

Replacing  with pitch in Algorithm (3.1), this algorithm can be used to find the
optimum pitch angle at each wind speed for a constant speed pitch controlled wind
turbine.

To show the performance of this algorithm, AWT-27 wind turbine is simulated as a

pitch controlled wind turbine with the following search parameters:
 Pitch limits: pitch l  5 and pitch u  5

 Initial step: step   pitch u  pitch l  / 10  1

 Power tolerance:  P  0.01Prated  3kW

 Pitch tolerance:  pitch  0.1



Figure (3.9) shows the results of this simulation. Referring to this figure, it can be
observed that the power curve reaches its rated value at about a wind speed of 12 m/s
and remains constant at higher wind speeds. This confirms that replacing  by pitch in
Algorithm (3.1), this algorithm successfully finds the suitable blade pitch angle to retain
the rotor power at its rated value at higher wind speeds. The power coefficient, as
shown in Figure (3.9.c), reaches a maximum value of 0.45 which is higher than the
maximum power coefficient of the previous simulation shown in Figure (3.7.c) (about
0.42). This confirms that this algorithm also successfully finds the optimum blade pitch
angle which increases the rotor power at lower wind speeds.

60
 350 3.5
3.0

Rotor Mechanical Power (kW)

300
2.5
250
2.0
200 1.5

Pitch (deg)
150 1.0
0.5
100 Wind Speed (m/s)
0.0
50 5 15 25
-0.5
0 -1.0
5 15 25
Wind Speed (m/s) -1.5
(a) (b)

0.50 40
0.45
35
0.40
30
Power Coefficient

0.35
Thrust (kN)

0.30 25
0.25
0.20 20

0.15 15
0.10
10
0.05
0.00 5
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(c) (d)

16 180

160
Root Bending Moment (kNm)

14
140
12
Tip Speed Ratio

120
10
100
8
80
6
60
4 40

2 20
5 10 15 20 25 5 10 15 20 25
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)

Figure 3.9- Results of simulation of constant-speed pitch-controlled AWT-27

61
3.5.2 Control Simulation of Variable Speed Pitch Controlled Rotors
For this type of wind turbine, the rotor speed and blade pitch angle are the controlling
parameters to be determined via solving optimisation problem of Equation (3.3).
Simulating AWT-27 wind turbine over a 2-dimensional grid on rotor speed and blade
pitch angle, the effect of these parameters on the rotor mechanical power can be
observed. Figure (3.10) shows four typical distinct behaviours for rotor power with
respect to blade pitch angle and rotor speed at different wind speeds. To produce these
figures the following data was used:
 Pitch limits: pitch l  5 and pitch u  5

 Rotor speed limits: l  30rpm and  u  65rpm

 Grid size in pitch-direction: 1

 Grid size in rotor speed-direction: 3.5rpm

Figure 3.10-Rotor power versus rotor speed and blade pitch angle at various wind
speeds

62
It can be seen that for all cases a pattern search method can be used to find the global
optima (located inside the inner contour) if point  pitch u ,  u  is taken as the initial
point. Algorithm (3.2) details the pattern search used for this purpose. The search
algorithm terminates when either of the termination criteria: step pitch   pitch ,

step     and P  Prated   P is satisfied.

Algorithm 3.2-Pattern search method for finding optimum rotor speed and pitch angle
for variable speed pitch controlled wind turbines
Given:
 step pitch ,  pitch , pitch l , pitch u , step ,   ,  l ,  u ,  P
 All parameters required for running Algorithm 2.2 except  and pitch
Step 1- Initialise:  pitch 0 ,  0    pitch u , u 
Step 2- Start with  pitch 0 , 0  and simulate wind turbine to find P ( Algorithm 2.2);
P0  P .
 
Step 3- While step pitch   pitch  step       P  Prated   P do:
3.1. Initialise move1  0 , move2  0
3.2. Move in  -direction: For  pitch 0 ,  0  step   simulate wind turbine to
find P
3.3. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
move1  1 ,  0   0  step  , P0  P
3.4. Move in pitch -direction: For  pitch 0  step pitch ,  0  simulate wind turbine
to find P
3.5. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
move2  1 , pitch 0  pitch 0  step pitch , P0  P
3.6. If move1  0  move2  0 then half the steps: step pitch  0.5step pitch ,
step  0.5step

To show the performance of this algorithm, AWT-27 wind turbine is simulated as a

variable speed pitch controlled wind turbine with the following search parameters:
 Pitch: pitch l  5 , pitch u  5 , step pitch   pitch u  pitch l  / 10  1 ,  pitch  0.1
 

 Rotor speed:  l  30rpm ,  u  65rpm , step   u  l  / 10  3.5rpm ,

   0.1rpm
 Power tolerance:  P  0.01Prated  3kW
Figure (3.11) shows the iteration points when using the pattern search method of
Algorithm (3.2) for four wind speeds shown in Figure (3.10). Red markers in Figure
(3.11) ( VW  24m / s ) correspond to infeasible solutions for which P  Prated (1   P ) .
63
Figure (3.12) shows the number of iterations for each wind speeds. It should be noted
that an exhaustive search with a grid of size of    0.1rpm and  pitch  0.1 requires

35451( u  l     1  351 by  pitch u  pitch l  /  pitch  1  101 ) simulation points.

This is 2100 times more than the average number of iterations when using the pattern
search algorithm.

Figure 3.11- Iteration points in pattern search method of Algorithm (3.2)

35
30
No. of Iterations

25
20
15
10
5
0
5 10 15 20 25
Wind Speed (m/s)

Figure 3.12- Number of iterations against wind speed using pattern search method of
Algorithm (3.2)
64
 350 70

65

Rotor Mechanical Power (kW)

300
60
250

Rotor Speed (rpm)

55
200
50
150
45
100
40
50 35
0 30
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(a) (b)

0.50 4
0.45
0.40 3
Blade Pitch Angle (deg)
Power Coefficient

0.35
0.30
2
0.25
0.20
1
0.15
0.10
0.05 0
5 15 25
0.00
Wind Speed (m/s)
5 15 25 -1
Wind Speed (m/s)
(c) (d)

10 40

9 35

8 30
Tip Speed Ratio

7 25
Thrust (kN)

6 20

5 15

4 10

3 5

2 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)

Figure 3.13- Results of simulation of variable-speed pitch-controlled AWT-27

65
Figure (3.13) shows the results of this simulation. Referring to Figure (3.13.c), it can be
observed that the power coefficient curve, as expected, is horizontal at lower wind
speeds and decreases as the power remains constant at higher wind speeds. The rotor
speed shown in Figure (3.13.b), as expected, increases with wind speed at lower wind
speeds until it reaches its upper bound (here 65 rpm). The second part of the rotor speed
curve at higher wind speeds indicates that the rotor speed does not vary significantly as
the wind speed changes. This behaviour is different from that of the variable speed stall
regulated (see Figure (3.7.b)). This is mainly due to having two controls (rotor speed
and blade pitch angle) and that rotor power is more sensitive to the blade pitch angle
rather than the rotor speed in this region.

The form of the tip speed ratio curve shown in Figure (3.13.e) is also different from that
of variable speed stall regulated wind turbine. It can be observed that the tip speed ratio
reduces from about 9 for wind speeds up to 10m/s to about 8 at wind speed of about 12
m/s, while the power coefficient remains constant for wind speeds up to about 12 m/s.
This figure confirms that the traditional method of finding rotor speed, in which the
rotor speed is found such that the tip speed ratio is retained constant at its design value,
is not valid for variable speed rotors which utilise an additional control system.

66
3.6 Aerodynamic Performance of Wind Turbines with Blades Utilising Flap
In case of blades utilising flap, in addition to the parameters defining the topology and
geometry of the blade R, c,  0 , AF , tmax  , three more parameters are also required to
define the location and size of the flap:
 Inboard radial location RF ,s

 Outboard radial location RF ,e

dF
 Width of the flap as a fraction of the chord length at the centre of the flap d F*  .
cF
These parameters are shown in Figure (3.14).

cF Flaps
dF

Rhub
RF , s
R F ,e

R
(a)

cF

dF
Chord line
F

(b)
Figure 3.14-(a) Parameters defining location and size of flap, (b) Flap deployment angle
sign convention

3.6.1 Modifications Applicable to the Aerodynamic Performance Calculator

A deployed flap changes lift and drag coefficients. These changes can be presented as:

C L | F  C L | F C L | F 0 (3.4)

CD | F  CD | F CD | F 0 (3.5)

67
in which, respectively, C L | F and C D | F are changes in lift and drag coefficients due

to a deployment of  F , C L | F and C D | F are the lift and drag coefficients at that

deployment angle, and CL | F 0 and C D | F 0 are the original lift and drag coefficients

(as there is no flap or no flap deployment  F  0 ). All these parameters depend on the
aerofoil angle of attack  as well. Figures (3.15) and (3.16) show two dimensional
(infinite length) C L and C D as functions of angle of attack,  and flap deployment

angle,  F . These results are obtained for aerofoil S808 with a flap width of 10% of

chord length ( d F*  d F c F  0.1 ) by running software XFOIL (Drela 1987, 1988,

1989a, 1989b, 1989c, 1990, 1998, 2008) as reported by Macquart (2012).

Figure 3.15- C L variation against angle of attack,  and flap deployment angle,  F

68
Figure 3.16- C D variation against angle of attack,  and flap deployment angle,  F

Since tabulated data C L ( ,  F ) and C D ( ,  F ) are three dimensional, reading off

data from these tables requires interpolations in two dimensions as explained in
Algorithm (3.3).

Algorithm 3.3-Reading C L ( ,  F ) and C D ( ,  F ) from 3-D tabulated data

Given:
 Aerofoils angle of attack  and flap deployment angle  F
 C L ( ,  F ) and C D ( ,  F ) tabulated data

Step 1- Find  F and  F , the closest flap angles in the data file to  F and  and   ,


the closest angles of attack to  .

Step 2- Using linear interpolation between C L (  ,  F ) and C L (  ,  F ) find
C L (  ,  F )
Step 3- Using linear interpolation between C L (  ,  F ) and C L (  ,  F ) find
CL (  ,  F )
Step 4- Using linear interpolation between C L (  ,  F ) and CL (  ,  F ) find
C L ( ,  F )
Step 5- Repeat Step 2 to 4 for C D

69
In order to analyse the aerodynamic performance of a wind turbine with blades utilising
flaps, the BEMT calculator of Algorithm (2.2) requires some modification as shown by
red boxes in Algorithm (3.4).

Algorithm 3.4-Modified BEMT calculator for blades utilising flap

Given:
 …
 
 RF* ,s , RF* ,e , d F* and  F

Step 5- …
Step 6- Dimensionalise … R F , s , R F ,e .
…
4.1.3.5 Use  i read off life and drag coefficients from tables: C L ,i and CD ,i
4.1.3.6 If RF ,s  ri  RF ,e : Using  F and  i read off lift and drag coefficient
corrections ( C L and C D ) from the table corresponding to d F*
(Algorithm 3.3); C L,i  C L,i  C L and C D,i  C D,i  C D
4.1.3.7 Calculate thrust coefficient at zero lift:
 r ,i cos 2  (1  a) 2 C L,i cos  i
CT0 
sin 2  i
…

3.6.2 Control Simulation of Constant Speed Rotors with Blades Utilising Flap
Here, the flap deployment angle is the controlling parameter to be found by solving the
optimisation problem of Equation (3.3). In order to observe the effect of the flap
deployment angles on the rotor mechanical power, it is assumed that the blades of the
baseline AWT-27 wind turbine are utilised by flaps with relative depth of d F*  0.1

extended from RF* ,s  RF , s / R  0.6 to RF ,e  RF ,e / R  0.65 . Simulating the wind

*

turbine for various flap deployment angles  20    F  20  , the rotor power at

different wind speeds is obtained and shown in Figure (3.17). According to this figure,
one can observe that at different wind speeds the P   F curves have one optimum point
only. It can be seen that in all cases, if the initial point in a hill-climbing algorithm is
taken as the upper limit  F ,0   F ,u with a search direction from right to left, the

algorithm will find the feasible global optima. Hence, a similar hill-climbing method as
in Algorithm (3.1), in which  is replaced by  F , can be used to find the optimum flap
deployment angle at each wind speed. In order to examine the performance of this
algorithm, using an initial step   F ,u   F ,l  / 10  4 , a power tolerance

70
 P  0.01Prated  3kW and a flap deployment angle tolerance of    0.1 simulation of
F

AWT-27 with blades utilising flaps is carried out. In this simulation flap deployment
angle is limited to the lower and upper bound  F ,l  20  and  F ,u  20  respectively.

Figure (3.18) shows the results of this simulation.

350000

300000 V=7m/s
Rotor Power (W)

250000 V=9m/s
200000 V=12m/s
150000 V=15m/s

100000 V=18m/s

50000

0
-30 -20 -10 0 10 20 30
Flap Deployment Angle (degree)

Figure 3.17-Power-flap deployment ( P   F ) curves at different wind speeds

Referring to Figure (3.18.a), it can be observed that the power curve at high wind speeds
does not remain horizontal. In fact it is very similar to that of the baseline constant
speed stall regulated wind turbine (see Figure 2.12). This indicated that the flap used in
this simulation is not aerodynamically efficient enough. This is in agreement with the
P   F curves shown in Figure (3.17) showing that rotor power has little sensitivity to

the variation of flap deployment angle  F .

71
 300 15

Rotor Mechanical Power (kW)

250 10

Flap Deployment Angle (deg)

5
200
0
150 5 15 25
-5
100
-10
50
-15

0 -20
5 15 25 Wind Speed (m/s)
Wind Speed (m/s) -25

(a) (b)

0.50 40
0.45
0.40
30
Power Coefficient

0.35
Thrust (kN)

0.30
0.25 20
0.20
0.15
10
0.10
0.05
0.00 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(c) (d)

16 200
Root Bending Moment (kNm)

14
150
12

10
Tip Speed Ratio

100
8

6 50
4

2 0
5 10 15 20 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(e) (f)
Figure 3.18- Results of simulation of constant-speed flap-controlled AWT-27

72
3.6.3 Control Simulation of Variable Speed Rotors with Blades Utilising Flap
Similar to the case of variable-speed pitch-controlled blades, here first an exhaustive
search is carried out to observe the behaviour of the output power as a function of the
controlling parameters rotor speed and flap deployment angle. Figure (3.19) shows the
results for four typical wind speeds. To generate this figure the following data was used:
flap extended between RF* ,s  0.6 and RF* ,e  0.65 of blade span with a relative depth of

d F*  0.1, flap deployment angle lower and upper limits  F ,l  20  and  F ,u  20  ;

rotor speed lower and upper limits  l  30rpm and  u  65rpm ; 4  grid size in  F -

direction and 3.5rpm in rotor speed-direction.

Comparing Figure (3.19) to Figure (3.10) (variable speed pitch controlled) one can
observed that the contours in Figure (3.19) are very close to vertical lines, showing that
the power is more sensitive to the rotor speed rather than the flap deployment angle.
Nevertheless, the same pattern search method as in Algorithm (3.2) can be used for
finding the optimum rotor speed and flap angle when pitch is replaced with  F .

Figure 3.19-Rotor power versus rotor speed and flap deployment angle at various wind
speeds
73
 350 70
65

Rotor Mechanical Power (kW)

300
60
250

Rotor Speed (rpm)

55
200 50
150 45
40
100
35
50
30
0 25
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(a) (b)

0.50 20
0.45 15
Flap Deployment Angle (deg)
0.40 10
Power Coefficient

0.35
5
0.30
0
0.25 5 15 25
-5
0.20
0.15 -10

0.10 -15
0.05 -20
0.00 -25
5 15 25 Wind Speed (m/s)
Wind Speed (m/s)

(c) (d)

9 45
40
8
35
Tip Speed Ratio

7 30
Thrust (kN)

25
6
20
5 15
10
4
5
3 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(e) (f)
Figure 3.20- Result of simulation of variable-speed flap-controlled AWT-27
74
Simulation of AWT-27 wind turbine as a variable speed flap controlled wind turbine
with the following search parameters leads to the results shown in Figure (3.20).
 Flap:  F ,l  20  ,  F ,u  20  , step F   F ,u   F ,l  / 10  4  ,   F  0.1

 Rotor speed:  l  30rpm ,  u  65rpm , step   u  l  / 10  3.5rpm ,

   0.1rpm
 Power tolerance:  P  0.01Prated  3kW

Comparing the power coefficient curve of this wind turbine shown in Figure (3.20.a)
(with a maximum value of about 0.427) with that of variable speed pitch controlled
shown in Figure (3.13.a) (with a maximum value of about 0.45), it can be observed that
pitch control is generally more efficient than flap control.

Figure (3.21) compares the power curve of the variable speed flap controlled wind
turbine with variable speed stall regulated wind turbine. These two wind turbines are the
same except the former is also equipped with flaps. One can observe that the power
curves are almost identical (slight difference in the horizontal section is due to using a
tolerance of  P  3kW for the power rather than different performances). This leads one
to conclude that once flaps are used in conjunction with another controlling system,
such as rotor speed, the accompanied controlling system dominates the control process.
In other words, there is no advantage of using flaps on variable speed wind turbines for
the purpose of power enhancement.

400
Rotor Mechanical Power (kW)

350

300

250
Variable Speed Flap Controlled
200
Variable Speed Stall Regulated
150

100

50

0
5 10 15 20 25
Wind Speed (m/s)
Figure 3.21-Comparison of power curves for variable speed flap controlled and
variable speed stall regulated wind turbines

75
3.7 Aerodynamic Performance of Wind Turbines with Telescopic Blades
Figure (3.22) shows the concept of a telescopic blade. A telescopic blade has a fixed
part attached to the hub and an extendable part. Rotor radius R is variable. It is limited
T

between R the span of the fixed part of the blade, Figure (3.22.b), and R when the
T, s T,e

telescopic part is fully deployed, Figure (3.22.c). While the fixed part of the blade, like
conventional wind turbine blades, can have variable chord length and pretwist
distributions and can be made of several aerofoils, the telescopic part is not pretwisted;
it has a constant chord length ( cT ),) and is made of one single aerofoil which is the
same as the aerofoil of the fixed part at span location r  R .
T, s

c(r ) cT  c( RT ,s ) (a)

Rhub
R
T, s
R
T

R  R R
T, s T T, e

c(r ) cT  c( RT ,s ) (b)
Fully contracted

Rhub
R R
T T, s

c(r ) cT  c( RT ,s )

(c)
Rhub Fully deployed
R
T, s

R R
T T, e

Figure 3.22-Telescopic blade definition

76
3.7.1 Modifications Applicable to the Aerodynamic Performance Calculator
In order to simulation a wind turbine with telescopic blades it is required to apply some
modifications to the data file of the baseline blade. The telescopic blade data file is the
same as the data file of the baseline blade for r  RT , s . For r  RT ,s (the telescopic

section) the aerofoil is AF RT ,s  and the chord length and pretwist distributions remain

constant with values cRT , s  and  0 RT ,s  , respectively. It should be noted that the

topology of the blade is defined against span location normalised rotor radius ( R for
baseline blade and RT for telescopic). Therefore, the span location normalised by the

rotor radius R , r *  r / R in the baseline data file should be converted to r *  r / RT in

the telescopic blade file. Algorithm (3.5) details steps for making telescopic blade data
file based on a baseline blade data file.

Algorithm 3.5-Modification of the data file of the baseline blade for a telescopic blade
Given:
 Original blade data file { R , Rhub
*
, c * (r * ) ,  0 (r * ) , AF (r * ) and t max
*
(r * ) }
 RT* ,s and RT* ; ( RT* ,s  RT*  RT* ,e )
Step 1- Find c ( RT , s ) ,  0 ( RT , s ) , AF ( RT* ,s ) and t max
* * * *
( RT* , s )
Step 2- In the original blade file, remove all data for r  RT , s and replace them with
* *

values calculated in Step 1 at r  RT , s and r  RT ,e :

* * * *

r c(r * ) t max (r * )
r*  c * (r * )   0 (r * ) AF (r * ) *
t max (r * ) 
R R c
… …. … … …
R *
T ,s
*
c (R ) *
T ,s  0 (R ) *
T ,s AF ( RT* ,s ) t *
max ( RT* , s )

RT* c * ( RT* ,s )  0 ( RT* , s ) AF ( RT* ,s ) *

t max ( RT* , s )

Step 3- Normalise the radial location in the new table by RT  RT* R :

r c(r * ) t max (r * )
r*  c (r ) 
* *
 0 (r * ) AF (r * ) *
t max (r * ) 
RT RT c
… …. … … …
R R
RT* , s c * ( RT* , s )  0 ( RT* , s ) AF ( RT* ,s ) *
t max ( RT* , s )
RT RT
R R
RT* 1 c * ( RT* , s )  0 ( RT* , s ) AF ( RT* ,s ) *
t max ( RT* , s )
RT RT

RT
Step 4- Rhub  Rhub
* *
and R  RT ; save telescopic blade data file { R , Rhub
*
, c * (r * ) ,
R
 0 (r * ) , AF (r * ) and t max
*
(r * ) }

77
Table (3.4) (repeated Table (2.1)) contains the first part of the blade data file of AWT-
27 wind turbine blade as the baseline. Tables (3.5) and (3.6) show the results of Steps 2
and 3 of Algorithm (3.5) applied to this blade for RT* , s  0.7 and RT*  1.1 . Normalised

chord distribution and pretwist distribution are shown in Figure (3.23).

Table 3.4-Baseline blade data Table 3.5-Data at the end of Table 3.6-Telescopic blade
Step 2 of Algorithm (3.5) data file (at the end of Step 3
of Algorithm (3.5))

r* c * (r * )  0 (r * ) r* c * (r * )  0 (r * ) r* c * (r * )  0 (r * )
0.086 0.053 6.270 0.086 0.053 6.270 0.078 0.048 6.270
0.109 0.056 6.100 0.109 0.056 6.100 0.099 0.051 6.100
0.155 0.063 5.764 0.155 0.063 5.764 0.141 0.057 5.764
0.223 0.070 5.470 0.223 0.070 5.470 0.203 0.064 5.470
0.269 0.075 5.233 0.269 0.075 5.233 0.244 0.068 5.233
0.315 0.081 4.996 0.315 0.081 4.996 0.286 0.073 4.996
0.360 0.082 4.602 0.360 0.082 4.602 0.328 0.075 4.602
0.406 0.083 4.208 0.406 0.083 4.208 0.369 0.076 4.208
0.452 0.081 3.689 0.452 0.081 3.689 0.411 0.074 3.689
0.497 0.079 3.172 0.497 0.079 3.172 0.452 0.072 3.172
0.543 0.076 2.628 0.543 0.076 2.628 0.494 0.070 2.628
0.589 0.074 2.086 0.589 0.074 2.086 0.535 0.067 2.086
0.634 0.071 1.601 0.634 0.071 1.601 0.577 0.064 1.601
0.680 0.068 1.117 0.680 0.068 1.117 0.618 0.062 1.117
0.726 0.064 0.770 0.700 0.066 0.966 0.636 0.060 0.966
0.772 0.060 0.424 1.100 0.066 0.966 1.000 0.060 0.966
0.806 0.056 0.273 r *  r / R, c *  c / R r *  r / RT , c *  c / RT
0.840 0.052 0.122
0.863 0.049 0.099
0.886 0.047 0.076
0.918 0.042 0.048
0.936 0.039 0.041
0.954 0.036 0.033
0.963 0.035 0.028
0.973 0.033 0.023
0.986 0.030 0.012
0.995 0.029 0.006
1.000 0.028 0.000
r *  r / R, c *  c / R
7 0.08
Normalised chord length c*

6 0.07

5 0.06
Pretwist (degrees)

0.05
4 Pretwist (deg)
0.04
3 c* (-)
0.03
2 0.02
1 0.01
0 0
0 0.2 0.4 0.6 0.8 1
Normalised span location r* (-)
Figure 3.23-Telescopic blade chord and pretwist distribution ( r *  r / RT , c *  c / RT ).
78
3.7.2 Control Simulation for Constant Speed Rotors with Telescopic Blades
Here, the rotor radius RT is the only controlling parameter. The same hill climbing
search method as explained in Algorithm (3.1), can be used to find the optimum value
of RT . For each examined RT within the search, the blade topology changes. Hence it is

required to modify the blade data file and discretise the blade for each examined RT .
Algorithm (3.6) details the modified hill climbing method for control simulation of
constant speed wind turbines with telescopic blades.

Algorithm 3.6- Hill climbing search for finding optimum rotor radius ( RT ) for constant
speed telescopic blade wind turbines
Given:
 step R* ,  R* , RT ,s , RT* ,e ,  P
*
T T

 All parameters required for running Algorithm 2.2

Step 1- Initialise: RT ,0  RT ,e
* *

Step 2- Start with RT  RT ,0

* *

2.1. produce telescopic blade data file (Algorithm 3.5)

2.2. discretise the blade (Algorithm 2.1)
2.3. simulate wind turbine to find P ( Algorithm 2.2); P0  P .
 
Step 3- While step R*   R*  P  Prated   P do:
T T

3.1. RT*  RT* ,o  step R*

T

3.1.1. produce telescopic blade data file (Algorithm 3.5).

3.1.2. discretise the blade (Algorithm 2.1)
3.1.3. simulate wind turbine to find P ( Algorithm 2.2)
3.2. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
RT* ,0  RT* ,0  step R* , P0  P ; Else: Half the step: step R*  0.5step R*
T T T

Figure (3.24) shows the results of simulating AWT-27 wind turbine with telescopic
blades using the following data: step R*  0.1 ,  R*  0.001 , RT , s  0.7 , RT ,e  1.1 and
* *
T T

 P  3kW .

Referring to Figure (3.24.b), it can be observed that, as expected, at higher wind speeds
(14m/s and above) the blade is contracted to maintain the rotor power at its rated value
by reducing the rotor area. Interestingly, at wind speeds of 5 and 6 m/s the telescopic
blade is not fully extended as opposed to what is expected. The reason for this can be
explained by referring to Figure (3.25).

79
 350 1.2

Rotor Mechanical Power (kW)

300
1.1

Telescopic Radius (-)

250
1.0
200

150 0.9
100
0.8
50

0 0.7
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(a) (b)

0.50 50
0.45
0.40 40
Power Coefficient

0.35
Thrust (kN)

0.30 30
0.25
0.20 20
0.15
0.10 10
0.05
0.00 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(c) (d)

14 250
Root Bending Moment (kNm)

12
200
Tip Speed Ratio

10
150
8
100
6

50
4

2 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)

(e) (f)
Figure 3.24- Result of simulation of constant-speed telescopic blade AWT-27
80
In Figure (3.25) the torque produced by each blade segment at wind speed 5m/s is
plotted against the radial location of that segment. It can be seen that at rotor span
locations above 0.73R, the produced torque by each segment is negative. This is due to
the fact that the telescopic part of the blade is not pretwisted and consequently the flow
kinematics is not optimised (also, see Equation (2.45)).

100

-100
dM(Nm)

-200

-300

-400

-500
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
RT*

Figure 3.25- Torque produced by each blade segment at wind speed 5m/s against the
radial location

Comparing the thrust curve of Figure (3.24.d) with the thrust curves of Figures (3.7.d),
(3.9.d), (3.13.f), (3.18.d) and (3.20.f), it can be observed that the maximum thrust force
for wind turbines utilising telescopic blades is higher than other types of wind turbines.
This is due to utilising larger blades.

3.7.3 Control Simulation for Variable Speed Rotors with Telescopic Blades
Having two controlling parameters, namely, the rotor radius RT and the rotor speed  ,
the same pattern search method as explained in Algorithm (3.2), can be used to find the
optimum values of RT and  . Similar to constant speed telescopic blades for each

examined RT within the search, the blade data file should be created and the blade
should be discretised. Algorithm (3.7) details the modified pattern search method for
control simulation of variable speed wind turbines with telescopic blades.

Figure (3.26) shows the results of simulating AWT-27 wind turbine with telescopic
blades and a variable speed rotor using the following data: step R*  0.1,  R*  0.001 ,
T T

81
RT* ,s  0.7 , RT* ,e  1.1 ,  l  30rpm ,  u  65rpm , step   3.5rpm ,    0.1rpm and

 P  3kW .

Comparing the variation of the telescopic blade radius of variable speed (Figure
(3.26.d)) with that of constant speed (Figure (3.24.b)), it can be observed that while for
the case of constant speed, the span of the telescopic blade is gradually decreases with
wind speed, in case of variable speed the span is fluctuating. This is due to the fact that
the blade span becomes the dominant controlling parameter because the rotor
mechanical power is more sensitive to the rotor diameter ( P  D 3 ) rather than rotor
speed ( P   ).

Algorithm 3.7-Pattern search method for finding optimum rotor speed and rotor radius
for variable speed telescopic blade
Given:
 step R* ,  R* , RT ,s , RT* ,e , step ,   ,  l ,  u ,  P
*
T T

 All parameters required for running Algorithm 2.2 except 

  
Step 1- Initialise: RT ,0 ,  0  RT ,0 ,  u
* *

Step 2- Start with R , 
*
T ,0 0

2.1. produce telescopic blade data file (Algorithm 3.5)

2.2. discretise the blade (Algorithm 2.1)
2.3. simulate wind turbine to find P ( Algorithm 2.2); P0  P .
 
Step 3- While step R*   R*  step       P  Prated   P do:
T T

3.1. Initialise move1  0 , move2  0

 
3.2. Move in  -direction: For RT ,0 ,  0  step  simulate wind turbine to find
*

P
3.3. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
move1  1 ,  0   0  step  , P0  P

3.4. Move in RT* -direction: For RT* ,0  step R* ,  0
T

3.4.1. produce telescopic blade data file (Algorithm 3.5)
3.4.2. discretise the blade (Algorithm 2.1)
3.4.3. simulate wind turbine to find P ( Algorithm 2.2)
3.5. If P  P0  P  Prated   P  P0  P  Prated   P  P0  P0  Prated  , then
move2  1 , RT* ,0  RT* ,0  step R , P0  P
*
T

3.6. If move1  0  move2  0 then half the steps: step R*  0.5step R* ,

T T

step  0.5step

82
 350 55

Rotor Mechanical Power (kW)

300
50

Rotor Speed (rpm)

250

200 45

150 40
100
35
50

0 30
5 10 15 20 25 5 10 15 20 25
Wind Speed (m/s) Wind Speed (m/s)

(a) (b)

0.45 1.10
0.40
1.08
0.35
Telescopic Radius (-)

1.06
Power Coefficient

0.30
0.25 1.04

0.20 1.02
0.15
1.00
0.10
0.98
0.05
0.00 0.96
5 10 15 20 25 5 10 15 20 25
Wind Speed (m/s) Wind Speed (m/s)

(c) (d)

9 50

8
40
7
Tip Speed Ratio

Thrust (kN)

30
6

5 20
4
10
3

2 0
5 10 15 20 25 5 10 15 20 25
Wind Speed (m/s) Wind Speed (m/s)

(e) (f)
Figure 3.26- Result of simulation of variable-speed telescopic blade AWT-27
83
3.8 Aerodynamic Performance of Wind Turbines with Blades Utilising
Microtabs

In addition to the parameters defining the topology and geometry of the blade
R, c,  0 , AF , t max , the following parameters are also required to define a string of
microtabs:
 Inboard radial location of string of microtabs RMT , s

 Outboard radial location of string of microtabs RMT ,e

d MT
 *
Microtab distance from leading edge d MT  , in which cMT is the chord length at
cMT
the centre of microtab
 Microtab length sMT
hMT
 *
Microtab actuation height hMT 
cMT
These parameters are shown in Figure (3.27).

Microtabs

sMT
Rhub
RMT , s
RMT ,e

cMT

hMT

d MT

Figure 3.27-Parameters defining location, size and actuation height of microtabs

84
3.8.1 Modifications Applicable to the Aerodynamic Performance Calculator
A microtab can have three states, namely, deployed upward (on the suction side of the
aerofoil), deployed downward (on the pressure side of the aerofoil), and neutral (not
deployed). These states are coded by -1, +1 and 0 respectively. A deployed microtab
changes lift and drag coefficients. These changes can be presented as:

C L | MT  C L | MT C L | MT 0 ; MT   1,1 (3.6)

C D | MT  C D | MT C D | MT 0 ; MT   1,1 (3.7)

in which, respectively, C L | MT and C D | MT are changes in lift and drag coefficients

due to deployment of a microtab upward ( MT  1 ) or downward ( MT  1 ), C L | MT
and C D | MT are the actual lift and drag coefficients due to the presence of the microtab

angle, and C L | MT 0 and C D | MT 0 are the original lift and drag coefficients (as there is

no microtab or the microtab is in neutral state MT  0 ). All these parameters depend on

the aerofoil angle of attack  as well. Figures (3.28) through (3.31) show C L and

C D as functions of angle of attack,  , microtab distance from leading edge d MT

*
, and
*
microtab actuation height hMT . These results are obtained for aerofoil S808 by CFD
analysis as reported by Hella (2012).

0.50
0.45
0.40
0.35
0.30
∆CL

0.25 S808, MT=+1, d*=95%, h*=1.1%

0.20 S808, MT=+1, d*=95%, h*=2.2%
0.15 S808, MT=+1, d*=95%, h*=3.3%
0.10
0.05
0.00
0 2 4 6 8 10 12 14 16
AoA (deg)

Figure 3.28- C L variation against angle of attack, microtab deployed downward

85
 0.020

S808, MT=+1, d*=95%, h*=1.1%

0.015 S808, MT=+1, d*=95%, h*=2.2%
S808, MT=+1, d*=95%, h*=3.3%
∆CD 0.010

0.005

0.000

-0.005

-0.010
0 2 4 6 8 10 12 14 16
AoA (deg)

Figure 3.29- C D variation against angle of attack, microtab deployed downward

0.2
S808, MT=-1, d*=80%, h*=1.1%
0.1 S808, MT=-1, d*=80%, h*=2.2%
0 S808, MT=-1, d*=80%, h*=3.3%

-0.1
∆CL

-0.2
-0.3
-0.4
-0.5
-0.6
0 2 4 6 8 10 12 14 16
AoA (deg)

Figure 3.30- C L variation against angle of attack, microtab deployed upward

0.030
0.025
0.020
0.015
0.010
∆CD

0.005
0.000
S808, MT=-1, d*=80%, h*=1.1%
-0.005 S808, MT=-1, d*=80%, h*=2.2%
-0.010 S808, MT=-1, d*=80%, h*=3.3%
-0.015
0 2 4 6 8 10 12 14 16
AoA (deg)

Figure 3.31- C D variation against angle of attack, microtab deployed upward

86
In order to analyse the aerodynamic performance of a wind turbine with blades utilising
microtabs, the BEMT calculator of Algorithm (2.2) requires some modification as
shown by red boxes in Algorithm (3.8). Moreover, in discretisation of the blade
(Algorithm (2.1)), the number of segments ( n seg ) should be selected such that segment

length ( r ) is equal to the length of microtabs ( s MT ). By doing this the number of

microtabs ( N MT ) distributed between the inboard and outboard radial locations RMT , s

and RMT ,e will b the same as the number of segments between these two span locations.

Algorithm 3.8-Modified BEMT calculator for blades equipped with microtabs

Given:
 …
*
 * * *

 RMT ,s , RMT ,e , d MT , hMT , s MT and the state of each microtab MT j   1, ,0,1 ;
j  1,2,..., N MT
Step 1- …
Step 2- Dimensionalise … RMT , s , RMT ,e .
…
4.1.3.5 Use  i read off lift and drag coefficients from tables: C L ,i and CD ,i
4.1.3.6 If RF ,s  ri  RF ,e : Using MT j and  i read off life and drag coefficient
corrections ( C L and C D ) from the table corresponding to d MT
*
and
*
hMT ; C L,i  C L,i  C L and C D,i  C D,i  C D
4.1.3.7 Calculate thrust coefficient at zero lift:
 r ,i cos 2  (1  a) 2 C L,i cos  i
CT 
0
sin 2  i
…

3.8.2 Control Simulation of Constant Speed Rotors with Blades Equipped with
Microtabs
Here, each of N MT microtabs distributed between RMT , s and RMT ,e acts as a controlling

device and the state of each microtab  1,0,1 becomes a controlling parameter which
needs to be determined via solving the optimisation problem of Equation (3.3). In other
words, we are dealing with N MT controlling parameters. Since the length of each

microtab ( s MT ) is much smaller than the length of the portion of the blade equipped
with microtabs, the number of microtabs can be large. This makes exhaustive search
impractical as the total number of examined points in an exhaustive search is 3 N (3 is MT

the number microtab states). For example, for the case of N MT  20 Algorithm (3.8) is

required to run about 3.5  10 9 times for each single wind speed. An N MT -dimensional
87
pattern search is also highly inefficient. Genetic algorithms (GA), on the other hand,
perform efficiently compared to other methods when dealing with large number of
variables.

Algorithm (3.9) details the GA used for finding the optimum state of each microtab. In
this algorithm, n pop , n gen , p c and p m are, respectively, number of population, number of

generations, probability of crossover and probability of mutation. In this algorithm the

fitness is defined as the rotor mechanical power ( fit  P ), parent selections for both
crossover and mutation is a random process, and the constraint handling is based on
rejection strategy (rejecting individuals contradicting the constraint P  Prated   P ).

Using this GA with the GA parameters n pop  20 , n gen  50 , pc  0.3 and pm  0.2 ,

AWT-27 wind turbine with blades equipped with microtabs between RMT ,s  0.6 and
*

,e  0.9 (where the blade is made of aerofoil S808) is simulated. Using 20segments
*
RMT

for discretising the blade, a string of 7 microtabs will be in the range of RMT ,s  0.6 to
*

,e  0.9 . The microtabs used for this simulation are located at d MT  80% and
* *
RMT
*
d MT  95% of the chord from the leading edge on upper and lower surface respectively,
*
with an actuation height of hMT  3.3% of the chord length.

Results of this simulation are shown in Figures (3.32) and (3.33). The power curve of
Figure (3.32) shows that although microtab can be used to regulate the power (the
horizontal section of the curve wind speeds between 14 to 21 m/s), comparing to other
controlling systems this device is not efficient enough to keep the curve horizontal at
lower or higher wind speeds (e.g. compared to variable speed pitch controlled wind
turbine of Figure (3.13), for which the power curve remains horizontal over the wind
speeds of 13 to 25 m/s). Figure (3.33) shows the optimum state of the string of
microtabs at each wind speed.

Referring to the power curve of Figure (3.32.a) showing that microtabs are not efficient
enough to keep the power curve horizontal, and recalling the discussion at the end of
Section 3.4.3, one can conclude that microtabs, like flaps, when used in conjunction
with another controlling system (e.g. rotor speed), the accompanied controlling system

88
dominates the control process. That is, there is no advantage in using microtabs on
variable speed wind turbines.

Algorithm 3.9-GA for finding optimum state of microtabs

Given:
 n pop , n gen , p c and p m
 P
 N MT
 All parameters required for running Algorithm (3.8) except the state of microtabs
MT j   1,0,1 ; j  1,2,..., N MT
Step 1- Initialise: MTi  0 , i  1,2,..., nseg
Step 2- Prdocude an initial poulation of n pop individuals:
2.1. i pop  0
2.2. While i pop  n pop do:
2.2.1. Genrate a string of size N MT with values randomly selected from
 1,0,1, representing the states of microtabs.
2.2.2. Using Algorithm (3.8), calculate rotor mechanical power P for this
individual; If P  Prated   P : i pop  i pop  1 , add this individual to the
population; Else: Reject this individual as infeasible solution
Step 3- For i gen  1 to ngen
3.1. Crossover (co): For ico  1 : pc n pop
3.1.1. Randomly select two individuals (parents)
3.1.2. Randomly select a cut point ( I cut ), an integer number between 1 and
N MT .
3.1.3. Creat a new child with microtab states similar to those of the first
parent for microtabs 1 to I cut and microtab states the same as those of
the second parent for microtabs I cut  1 to N MT
3.1.4. Using Algorithm (3.8), calculate rotor mechanical power P for the
produced child; If P  Prated   P reject this child as infeasible solution;
otherwise add this child to the population
3.2. Mutation: For imute  1 : pm n pop
3.2.1. Randomly select an individual (parent)
3.2.2. Randomly select a microtab ( I mute ), an integer number between 1 and
N MT .
3.2.3. Change the current value to a different value, randomly taken from
 1,0,1to create a new child
3.2.4. Using Algorithm (3.8), calculate rotor mechanical power P for the
produced child; If P  Prated   P reject this child as infeasible solution;
otherwise add this child to the population
3.3. Regeneration: Keep the first n pop individuals with heighest fitness ( P ) and
discard the rest of individuals.

89
 350 16

Rotor Mechanical Power (kW)

300 14

250 12

Tip Speed Ratio

200 10

150 8

100 6

50 4

0 2
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(a) (b)

0.50 40
0.45 35
0.40
30
Power Coefficient

0.35
25
Thrust (kN)

0.30
0.25 20
0.20 15
0.15
10
0.10
0.05 5

0.00 0
5 15 25 5 15 25
Wind Speed (m/s) Wind Speed (m/s)
(c) (d)
Figure 3.32- Results of simulation of constant-speed AWT-27 equipped with microtabs

Microtab 1 Microtab 2 Microtab 3 Microtab 4

Microtab State

Microtab 5 Microtab 6 Microtab 7

-1
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Wind Speed (m/s)
Figure 3.33- Results of simulation of constant-speed AWT-27 equipped with microtabs-
The states of the microtabs

90
3.9 A Preliminary Comparison of Different Types of Control Systems

The comparison of the power, power coefficient, bending moment at the root of the
blade and the rotor speed for different types of wind turbines simulated in this chapter
are shown in Figures (3.34) to (3.40).

350

300
Rotor Mechanical Power (kW)

250

200

150 Constant Speed Telescopic Blade

Constant Speed Pitch Controlled
100 Constant Speed Flap Controlled
Constant Speed Microtab
50 Constant Speed Stall Regulated

0
5 10 15 20 25
Wind Speed (m/s)
Figure 3.34- Comparison of the power curves of different types of constant speed wind
turbines

350

300
Rotor Mechanical Power (kW)

250

200

150
Variable Speed Telescopic Blade

100 Variable Speed Pitch Controlled

Variable Speed Flap Controlled
50 Variable Speed Stall Regulated

0
5 10 15 20 25
Wind Speed (m/s)
Figure 3.35- Comparison of the power curves of different types of variable speed wind
turbines

91
 0.45

0.40

0.35
Power Coefficient (-)

0.30

0.25

0.20

0.15 Constant Speed Telescopic Blade

Constant Speed Pitch Controlled
0.10
Constant Speed Flap Controlled
0.05 Constant Speed Microtab
Constant Speed Stall Regulated
0.00
5 10 15 20 25
Wind Speed (m/s)

Figure 3.36- Comparison of the power coefficient of different types of constant speed
wind turbines

0.45

0.40

0.35
Power Coefficient (-)

0.30

0.25

0.20

0.15 Variable Speed Telescopic Blade

0.10 Variable Speed Pitch Controlled

Variable Speed Flap Controlled
0.05
Variable Speed Stall Regulated
0.00
5 10 15 20 25
Wind Speed (m/s)

Figure 3.37- Comparison of the power coefficient of different types of variable speed
wind turbines

92
 240
220
200

Blade Root Bending Moment (kNm)

180
160
140
120
100 Constant Speed Telescopic Blade
80 Constant Speed Pitch Controlled
60 Constant Speed Flap Controlled
Constant Speed Microtab
40
Constant Speed Stall Regulated
20
0
5 10 15 20 25
Wind Speed (m/s)

Figure 3.38- Comparison of the blade root bending moment of different types of
constant speed wind turbines

240
220
200
Blade Root Bending Moment (kNm)

180
160
140
120
100
Variable Speed Telescopic Blade
80
Variable Speed Pitch Controlled
60
Variable Speed Flap Controlled
40 Variable Speed Stall Regulated
20
0
5 10 15 20 25
Wind Speed (m/s)

Figure 3.39- Comparison of the blade root bending moment of different types of
variable speed wind turbines

93
 65

60

55
Rotor Speed (rpm)

50

45
Variable Speed Telescopic Blade
40 Variable Speed Pitch Controlled
Variable Speed Flap Controlled
35
Variable Speed Stall Regulated

30
5 10 15 20 25
Wind Speed (m/s)

Figure 3.40- Comparison of rotor speed variation for different types of variable speed
wind turbines

With reference to these figures the following conclusions can be drawn.

 Using telescopic blades enhances the power capture capability significantly at lower
wind speeds for both constant speed and variable speed rotors (Figures (3.34) and
(3.35)).
 Telescopic blades provide a full (Figure (3.34)) and smooth (Figure (3.24.b))
control.
 Pitch control is the most efficient control system having the highest power
coefficient in both constant and variable speed rotors (Figures (3.36) and (3.37)).
 While variable speed stall regulated, variable speed pitch controlled and variable
speed flap controlled wind turbines tend to operate at maximum possible rotor speed
at some points, the variable speed telescopic wind turbines operate at lower rotor
speeds (Figure (3.40)).
 The bending moment at the root of the blade increases significantly by using
telescopic blades (Figures (3.38) and (3.39)).
 Although microtab is not as efficient as flap or pitch control systems, it increases the
load on blades significantly when used for power enhancement (Figures (3.38)).
 Pitch control system produces minimal blade root bending moment for both constant
speed and variable speed rotors, particularly variable speed (Figures (3.38) and
(3.39)).

94
 Microtab and flap have been initially developed for load alleviation purposes. These
controlling devices, however, can be used to regulate and enhance the rotor
mechanical power to some extent (Figure (3.34)).
 Microtab and flap slightly improve the rotor power extraction efficiency (power
coefficient) (Figure (3.36)).
 Flaps, when used in conjunction with another controlling system such as rotor
speed, the accompanied controlling system dominates the control process (Figure
(3.21)). This conclusion can be extended to microtabs by observing similar effect of
both controlling systems on the power curve (Figure (3.34)).

It should be noted that since none of the above wind turbines (except the baseline
constant speed stall regulated) has been optimised to operate optimally, some of these
conclusions might not be valid for optimally designed blades. Chapter 4 elaborates on
the optimisation methodology for each wind turbine blade utilising the above
controlling systems.

3.10 Summary
In this chapter different types of aerodynamic control systems for constant and variable
speed rotors were explained and classified. It was also explained that assuming the
controller performs perfectly, solving an optimisation problem leads to quantifying the
controlling parameters making simulation of wind turbines utilising control systems
possible. Three methods of optimisation, namely, hill climbing, pattern search and
genetic algorithm were developed and used to solve the optimisation problem. The
performance of the hill climbing and pattern search methods was evaluated and
reported.

The wind turbine simulator developed in Chapter 2 is only capable of simulating

constant speed stall regulated wind turbines which do not utilise any active control
systems. In this chapter the capabilities of this software was expanded via implementing
necessary modifications for simulating the following eight different types of wind
turbines.
1. Variable speed stall regulated rotors
2. Constant speed pitch controlled rotors
3. Variable speed pitch controlled rotors
4. Constant speed flap controlled rotors
95
5. Variable speed flap controlled rotors
6. Constant speed rotor with telescopic blades
7. Variable speed rotors with telescopic blades
8. Constant speed rotors with blades equipped with microtab

For each type a case study was carried out to demonstrate the capability and the
performance of the developed wind turbine simulator. It should be noted that, none of
the wind turbine simulators reported in open literature is capable of simulating wind
turbines of types 4 to 8 above. Moreover, they use inefficient exhaustive search methods
for simulating wind turbines of type 2 and 3, and wind turbine type 1 is simulated by
assuming that the tip speed ratio must be kept constant at its design value, an
assumption challenged in this Chapter.

Through a preliminary study, the performance of different types of wind turbines was
compared.

It was found that:

 Telescopic blades provide a full and smooth control and enhance the power capture
capability significantly. However, this enhancement comes with the price of a
significant increase in bending moment at the root of blade.
 Pitch control is the most efficient control system having the highest power
coefficient and minimal blade root bending moment for both constant speed and
variable speed rotors.
 While variable speed stall regulated, variable speed pitch controlled and variable
speed flap controlled wind turbines tend to operate at maximum possible rotor speed
at some points, the variable speed telescopic wind turbines operate at lower rotor
speeds.
 Microtab and flap have been initially developed for load alleviation purposes. These
controlling devices, however, can be used to regulate and enhance the rotor
mechanical power to some extent. Microtab is not as efficient as flap control
systems.

96
4 AWTSimD, an
Optimisation Tool for
Wind Turbine Blades
Equipped with
Nonconventional
Aerodynamic Control
Systems

97
4.1 Introduction
In this chapter, first design phases of a wind turbine blade are explained. It is followed
by explaining and classifying different design variables involved in each design phase.
Focusing on the aerodynamic design phase of wind turbine blade, two design methods
applicable to the aerodynamic design of wind turbine blades are explained, followed by
formulation of the design problem in the standard format of an optimisation problem. In
Section 4.3, the developed genetic algorithm optimisation method is explained in
details. Section 4.4 details the implementation of the optimisation tool and the wind
turbine simulation tool developed in Chapter 3.

4.2 Aerodynamic Design of Wind Turbine Blades

Wind turbine blades are designed in three consecutive phases, namely, conceptual,
aerodynamic and structural. Parameters defining a blade are as follows:
 Span (rotor radius)
 Aerofoil distribution
 Chord length distribution
 Pretwist distribution
 Material and structural features (including shell material and thickness distribution,
number and location of webs)

The rotor size (diameter) is a design parameter which is obtained at the conceptual
design phase of wind turbines alongside with the number of blades, hub height,
generator type (constant speed or variable speed), generator size (wind turbine rated
power), type of the control system, rated wind speed and rated rotor speed. Once the
rotor diameter is decided the span of the blade is fixed.

In the aerodynamic phase, the aerofoil, chord length and pretwist distributions are
obtained. Material and structural features of the blade are designed in the structural
design phase.

4.2.1 Classification of Design Variables

Design variables defining wind turbine blades can be classified into singular and
distributed groups:
 Singular: design variable is defined with one single value such as rotor radius and
the number of webs
98
 Distributed: design variables are distributed along the span of the blade, such as
chord length, pretwist, aerofoil index, material, shell thickness, web thickness. In a
design process each of these design variables is presented by ndp values at ndp

span locations. Here, ndp is called the number of design points (not to be confused

with design variables). For example, using ndp  5 design points, the blade chord

length distribution is presented as c(r ) : c1 , c2 , c3 , c4 , c5 @r1 , r2 , r3 , r4 , r5 , where c i

stands for the chord length corresponding to the span location ri . For design
variables with sharp variations more design points are required to capture the best
distribution.

Design variables can be also classified based on their types as follows:

 Real number, such as rotor radius, chord length, pretwist, shell thickness, web
thickness and location
 Integer number, such as number of webs and blades
 Indexed or coded, such as aerofoil index (e.g. S814, NACA6430, …) and material
type (e.g. glass fibre, carbon fibre, …)

4.2.2 Direct versus Inverse Design Methods

In a direct design approach, first the designer selects the design parameters and then
evaluates the design candidate based on a series of assessment criteria, including
constraints. A direct design method is iterative. Selection of design variables and
evaluation of design candidate repeats until all evaluation criteria are satisfied. This
method is the common practice and dominant in most of engineering design problems.

In cases for which the number of explicit analytical equations governing the problem is
the same as the number of design variables, an inverse design method can be adopted.
In an inverse design method, the designer sets some target values and then solve the
equations for the design variables leading to those target values. This method can be
applied to very simple cases with simple equations and small number of design
variables. Figure (4.1) compares these two design methods.

99
 Start Start

Select Design Variables

Set target qualities

Analysis (e.g. calculate power curve,

power coefficient, thrust and flap
Find design variables bending moment)

End
Evaluate against constraint

False
Feasible?

True

(a) (b) End

Figure 4.1- (a) Inverse and (b) Direct Design

4.2.3 Inverse Design

Although the aerodynamics of wind turbine blades is very complicated, an inverse
design method can be applied for chord and pretwist design of blades installed on
variable speed rotors. The inverse design method determines directly the blade
geometry that will achieve certain desired aerodynamic performance. Normally, the
geometry is then modified to account for structural and other considerations.

Having decided the rotor size, and hence the blade span, in conceptual design phase, the
aerofoil, pretwist and chord distributions are to be determined in blade aerodynamic
design phase. Referring the BEMT of Chapter 2, for a variable speed rotor with constant
R
tip speed ratio   , the maximum extracted energy from wind corresponds to the
V
following values of the axial and rotational induction factors (Burton, 2001):

a  1/ 3 (4.1)

2
a' 
92  2 (4.2)

100
where,   r / R is the normalised span location, R is the rotor radius. Substituting
these values in the BEMT equations the optimum chord and pretwist distributions can
be found as (Burton, 2001):

16R
c (4.3)
2
4 2 2 2 
9 BCL ,opt    1  2 2 
9  9  

2
tan   (4.4)
 2 
3 1  2 2 
 3  

 0     opt (4.5)

In the above equations, C L ,opt is the lift coefficient at optimum angle of attack  opt , the

optimum angle of attack is the angle of attack at which C L / C D is maximum,  stands

for inflow angle,  0 is the pretwist and c is the chord length.

The chord length obtained from Equation (4.3) is normally linearised for blade
manufacturing considerations. Often, the chord is linearised to produce a simpler blade
that is easier to manufacture. A straight line drawn through the 70% and 90% span point
not only simplifies the plan-form of blade design but also removes a lot of material
close to the root (Burton 2001 and Drew 2011).

4.2.4 Performance of Inverse Design

Although the inverse design method above is based on some optimality condition
(maximum energy extraction for case of a  1 / 3 and a'  2 / 92  2 ), in practice these
conditions will not completely satisfy. This affects the performance of this method in
design of blades. As a case study, the blade of AWT-27 is redesigned for a variable
speed rotor, assuming the same rotor radius (R=13.75m) but with aerofoil S809 (
 opt  14.2  ) all through the blade. Results are shown in Figures (4.2) and (4.3). Figure

(4.2) shows both chord distribution obtained by Equation (4.3) and its linearised form.

101
 0.25

0.20

Chord (m) 0.15

0.10

0.05
Redesigned AWT27 blade
Linearised redesigned AWT27 blade
0.00
0 0.2 0.4 0.6 0.8 1
Span Location (r/R)

Figure 4.2-Redesigned chord of AWT-27 blade for variable speed rotor using inverse
method

30

25
Pre-Twist (deg)

20

15

10

0
0 0.2 0.4 0.6 0.8 1
Span Location (r/R)
Figure 4.3- Redesigned pretwist of AWT-27 blade for variable speed rotor using inverse
method

It should be noted that the original blade is not designed based on an inverse method,
but has been designed using a direct search-based method as explained in the next
section. Figures (4.4) and (4.5) show the results of simulation of a variable speed AWT-
27 wind turbine once using the design blade using inverse method (results shown in
Figures (4.2) and (4.3)), and once using its original blade. Table (4.1) summarises the
results of the simulation of variable speed AWT-27 with original and redesigned blades.
It is evident that the inverse design method does not produce optimum solutions. The
blade designed using the inverse method produces less power while subjected to higher
loads.

102
 350

Rotor Mechanical Power (kW)

300

250

200
Original Blade AWT 27
150
Simplified Design Blade
100

50

0
5 10 15 20 25
Wind Speed (m/s)
Figure 4.4-Power curve: Variable speed AWT-27 with original and redesigned blades

0.45
0.40
0.35
Power Coefficient

0.30
0.25
0.20
0.15
0.10
Original Blade AWT 27
0.05
Simplified Design Blade
0.00
5 10 15 20 25
Wind Speed (m/s)
Figure 4.5-Power coefficient: Variable speed AWT-27 with original and redesigned
blades

In calculating the annual average power a site average wind speed of 5.7 m/s with a
Rayleigh PDF is used.

Table 4.1- Results of the simulation of variable speed AWT-27 with original and
redesigned blades using inverse method
Average Max Max Root
Maximum Max Power
Power Thrust Bending Moment
Power (kW) Coefficient
(kW) (kN) (kNm)

Original Blade
49.6 302.7 0.426 41.6 187.2
AWT-27
Simplified Design
45.2 301.9 0.383 53.4 234.6
Blade AWT-27

103
4.3 Optimal Aerodynamic Design of Wind Turbine Blades

As explained in previous section, inverse methods of blade aerodynamic design are

typically based on some optimality assumptions imposed on the blade angle of attack
distribution and/or tip speed ratio and/or axial and rotational induction factors. These
methods normally provide closed formed design equations for pretwist and chord
distributions. These methods are easy to use and give acceptable but not optimal
results. Assuming blade has a fixed topology and aerodynamic characteristics, these
design methods theoretically give the optimal topology of the blade. These design
methods, however, fail in optimal design of blades equipped with controlling
aerodynamic devices which actively affect the aerodynamic characteristics of blade. For
such cases optimal aerodynamic design is achievable by employing a direct search-
based design method, as shown in Figure (4.6).

Start

Select Design Variables

Analysis (e.g. calculate power curve,

power coefficient, thrust and flap
bending moment)

Evaluate against constraints

False
Feasible?

True

False
Optimum?

True

End

Figure 4.6-Search-based design (Direct design optimisation)

In a search-based design the objective of the optimisation is normally to maximise the

annual energy production rather than, for example, maximising the power coefficient at
a certain wind speed. The annual energy production of a wind turbine is influenced by
104
the wind characteristics of the site at which the unit is installed as well as the wind
turbine capability of generating power. Wind density probability function is a model
giving information on the magnitude and likelihood of wind in a site. Power curve, on
the other hand, provides information on the capability of a wind turbine in producing
power at various wind speeds. Annual average power, Pav , is defined as:

V0

Pav   P(V ) R(V )dV (4.6)

Vi

where, R(V ) is wind speed probability density function (PDF), P is the wind turbine

power and Vi and Vo are the cut-in and cut-out velocities, respectively. In this study, a
Rayleigh PDF represented by:

2 V 
2
  V 
2

R(V )    exp     (4.7)
V 4  Vav   4  V av  

has been used to calculate average power. Parameter Vav stands for the site average wind
speed.

The optimisation problem, therefore, can be summarised as

maximise Pav
subject to main constraint

P  Prated (4.8)

and possibly other constraints on, for example, blade maximum bending moment,
weight of the blade, etc.

4.4 Genetic Algorithm Optimisation Method

Genetic algorithm is a well known optimisation technique applicable to all kind of
optimisation problems, including constrained/unconstrained, linear/nonlinear,
real/integer/mixed value, concave/convex and continuous/discrete domains. Many text

105
books can be found on the fundamentals and application of genetic algorithm. For
instance, see Holland (1975), Michalewicz (1992) and Baeck (2010).

For the optimisation problem at hand, solutions (also called individuals and design
candidates) are wind turbine blade. The following sections elaborate on the developed
GA with the following order: chromosome representation of solutions (wind turbine
blades), initial population generation, crossover operator, mutation operator, fitness
definition, parent selection, regeneration, constraint handling and termination criteria.

4.4.1 Chromosome Representation

A real number encoding is used. Depending on the number of design points ( ndp )

considered for distributed design variables and the number of design variables included
in the optimisation problem, the maximum length of the chromosome (for three
distributed and one singular design variables) is 3ndp  1as shown in Figure (4.7)

ndp chord values ndp pretwist values span

c1 c2 c3 … c nd p  01  0 2  0 3 …  0 n R dp

Figure 4.7-Wind turbine blade chromosome

In the developed optimisation code in MATLAB, the user sets ndp and selects the

design variables to be included in the optimisation problem from the set of { c,  0 , R }.

For each design variable selected for optimisation a realistic range is also required to be
set. For those design parameters not selected for optimisation, a fixed value or
distribution is considered. For example, to optimise the AWT-27 blade for pretwist, the
pretwist is selected as design variable while the original chord and rotor radius are used.
Other required parameters are ndp and the lower and upper limits for the pretwist:

 0,l ,  0,u

4.4.2 Initial Population Generation

The initial population in most of GAs is generated randomly. A random initial
population generation method can generate both feasible and infeasible solutions. In
highly constrained problems and problems in which the constraints are very sensitive to

106
the design variables, generating feasible initial population can be very time consuming
due to high number of failed attempts.

In our case here, the output power (constraint of Equation (4.8)), is highly sensitive to
the blade pretwist and chord length distribution as well as the rotor radius. Particularly,
random generation of initial population becomes very time consuming when all these
design variables are included in the optimisation process. To overcome this problem,
two approaches have been adopted as explained below.

4.4.2.1 Randomly Generated

Depending on whether the design variable included in the optimisation is singular or
distributed, each individual in initial population is generated as follows:

For each singular design variable included in the optimisation, a random number
between the identified limits is assigned to that design variable. The limits are set based
on realistic values.

For each distributed design variable included in the optimisation there are ndp design

points. Each design point of each distributed design variable is assigned a random value
between the limits. It is possible to implement some heuristics when generating these
random numbers to reduce the chance of producing infeasible and unrealistic solutions.
For example, chord has a decreasing trend from hub to tip and pretwist has a concave
trend. Generally, five possible trends that can be selected for each real number
distributed design variable are shown in Figure (4.8).
Typical Distributed Design
Variable

0 0.2 0.4 0.6 0.8 1

Normalised Span Location (r/R)
Random trend Increasing trend Decreasing trend
Concave trend Convext trend

Figure 4.8- Randomly generated distribution for a typical real-number distributed

design variable

107
4.4.2.2 Perturbation of the Baseline Design
In this method, a new design candidate is produced by a random deviation of an initial
design rather than a randomly generated blade from scratch. This method is applicable
when the design variable is distributed (i.e. pretwist and chord). In this method, one of
the design points is randomly selected and the rest have the same value as the baseline
design. Figures (4.9) and (4.10) illustrate this method, assuming the design variables
are pretwist and chord and the baseline is AWT-27 wind turbine blade. In this example
the number of design points, n dp , is 6. A random number between 1 and 6 has been

selected (here 5). For design point number 5, the chord and pretwist are assigned new
“randomly” selected values (between identified limits).

7
6
5
Pretwist (degrees)

4
3
2 Baseline
1
Perturbation of the Baseline
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)
Figure 4.9- Initial population generation using perturbation of an initial design
candidate: Blade pretwist

0.1

0.08
Normalised Chord (c/R)

0.06

0.04
Baseline
0.02 Perturbation of the Baseline

0
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)
Figure 4.10- Initial population generation using perturbation of an initial design
candidate: Blade chord

108
In the developed software tool the initial population generation is a hybrid of both
approaches. That is, xip fraction of the initial population is produced using perturbation

method and the remaining (1- xip ) fraction is produced using random method.

Therefore, the following parameters are required to be set by the user:

 Population size, n pop

 Lower and upper limits for each design variable considered for optimisation
 Fraction of initial population to be generated based on perturbation method, xip

 In case of xip  1 , the trend of randomly generated distributed design variables:

t ip ,rand =1, 2, 3, 4, 5, respectively, for random, increasing, decreasing, concave

and convex.

4.4.3 Crossover
The purpose of the crossover operation is to exploit the search area via creating new
solutions from existing solutions in the current population. Two types of crossover has
been coded and implemented in the optimisation module. The first type is classical
arithmetic or weighted average crossover. The second type is the geometric crossover
recently proposed by Maheri et al (2012).

4.4.3.1 Arithmetic Crossover

In an arithmetic crossover, each gene in the chromosome of a child is a weighted
average of the corresponding genes of its parents as given by:

 child   parent  (1   ) parent

1 2
(4.9)

In the above equation  represents any gene of the chord, pretwist or rotor radius

section of the chromosome and   0,1 is a random number. Figure (4.11) illustrates
arithmetic crossover (curves show typical parents and child chord distribution).

109
 Hub
Tip

Parent 1

Child

Parent 2

Chord av Pretwist av R av
(a)
Child chromosome
0.1

0.09
Normalised Chord (c/R)

0.08

0.07

0.06 Chord Parent 1

Chord Parent 2
0.05
Chord Child
0.04
0 0.2 0.4 0.6 0.8 1
(b)
Normalised Span Location (r/R)

Figure 4.11-Arithmetic (weighted average) crossover

4.4.3.2 Geometric Crossover

This section explains the geometric crossover proposed by Maheri et al (2012). In a
geometric crossover, two parents are cut at a random span location and the child is
formed by swapping sections of parents as shown in Figure (4.12).

110
 Hub
Tip

Parent 1

Needs repair

Child

Parent 2

Chord Chord Pretwist Pretwist

R av
(P1) (P2) (P1) (P2)

Child chromosome
Figure 4.12-Geometric crossover

Having the radial coordinate normalised by span length ( r *  r / span ), the cut point is a

randomly selected design point rc* . The cut divides each parent blade into two parts. The
distributed design variables of the child blades are formed by those of the left and right
hand sides of each parent blade. A repair operation is also required to retain the
continuity of the distributed design variable (Maheri, 2012).

Figure (4.13) illustrates the process of forming a distributed design variable (here the
pretwist distribution  0 ) of a child from a pair of parents. The repaired pretwist is
obtained by multiplying the unrepaired pretwist by the left and right multipliers
M L (r * ) and M R (r * ) .

 
  0 (r * ) P M L (r * ) if 0  r *  rc*
   1

0 (r * ) C1 , R  (4.10)

 0 
  (r * ) M (r * ) if
P2 R r  r 1
c
* *

where, subscripts C1 , P1 , P2 and R stand for child 1, parent 1, parent 2 and repaired,

respectively. M L (r * ) and M R (r * ) are the left and right segments of a multiplier curve.

111
  0 ,c C1 , R
The multiplier curve for child 1 is a linear curve between 1 at r *  0 and at the
 0 ,c P1

 
0 ,c C1 , R
cut point; and at the cut point rc* and 1 at r *  1 as shown in Figure (4.13.c).
 
0 ,c P2

The pretwist at the cut point rc* is denoted by  0,c . The repaired pretwist at the cut point

is a combination of the left and right values proportional to the length of the left and
right segments respectively. That is, the repair process has less effect on the segment
with longer length.

10 10 Child 1 before repair

Pre-Twist (deg)

Pre-Twist (deg)
8 Parent 1 8
Child 1 after repair
6 6
4 Parent 2 4
2 (a) 2 (b)
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
cut point r* r*
Multiplier

(c)
0
0 0.2 0.4 0.6 0.8 1
r*

Figure 4.13- Pretwist formation of a child blade; child is formed based on the left
segment of parent 1and the right segment of parent 2 (Maheri, 2012).

The blade span of child 1, R , is the combination of those of parent blades in a weighting
sense (Maheri, 2012).

Rchild  rc* R parent1  (1  rc* ) R parent 2 (4.11)

Similar to initial population generation, in the developed optimisation module, the

crossover operation is a hybrid of both methods. That is, xCO fraction of the total

crossover operations in each generation ( nCO  n pop  pc ) is carried out using arithmetic

112
method and the remaining (1- xCO ) fraction is carried out using geometric method.
Therefore, the following parameters are required to be set by the user:
 Population size, n pop

 Probability of crossover, p c

 Fraction of total crossover to be carried out using arithmetic method, xCO

It should be noted that geometric crossover can be applied only if distributed design
variables are included in the optimisation.

In both arithmetic and geometric crossover, a roulette wheel based on the fitness is used
to select the parents.

The fitness is defined as the average power:

fitness  Pav (4.12)

4.4.4 Mutation
Mutation is a random operation. A random gene in the chromosome of a randomly
selected individual is selected for mutation. This gene will be replaced by a randomly
selected new value within the range of the corresponding design variable. In the
optimisation module, the following parameters are required to be set by the user:
 Population size, n pop

 Probability of mutation, p m

 Lower and upper limits for each design variable in the optimisation

4.4.5 Constraint Handling

After each crossover or mutation operator the feasibility of the offspring is checked. If
all constraints are satisfied, the offspring is added to the population. Infeasible solutions
will be discarded.

4.4.6 Regeneration
At the end of each generation, after nCO  n pop  pc crossover and nMute  n pop  pm

mutation, the population size increase from n pop to a maximum of n pop  nCO  nMute . At

113
this point, all individuals will be sorted based on their fitness. The first fittest n pop

individuals are passed to the next generation.

4.4.7 Termination
Genetic algorithm continues until the generation number reaches the set maximum
number of generations, n gen or when the maximum fitness in a generation becomes the

same as the average fitness (converged solution).

4.5 AWTSimD, Advanced Wind Turbine Simulation and Design

AWTSim developed in Chapters 2 and 3 now can be integrated with the GA
optimisation module as its evaluator to form the blade design optimisation tool
AWTSimD. That is, each produced design candidate at the stage of initial population
generation or as a result of crossover and mutation operations is evaluated using
AWTSim, as shown in Algorithm (4.1). AWTSim calculates the wind turbine
aerodynamic performance (blade and rotor loads and rotor mechanical power) between
cut-in and cut-out velocities and using a Rayleigh PDF and finds the average annual
power Pav at a given site average wind speed Vav .

In this algorithm, the data required for evaluation of each design candidate (blade)
depends on the type of the blade (telescopic, equipped with microtab, equipped with
flap, equipped with pitch, with no active control) under optimisation and the rotor speed
(constant or variable). The required data are as explained in Algorithm (2.2) and
Algorithms (3.1) through (3.9), except the design variables.

As explained in Step 1 of the algorithm, the initial population generation is carried out
in two parts, namely, random generation and perturbation of a baseline design.
Crossover is also performed in two parts (Steps 2.2.1 and 2.2.2), one arithmetic and one
geometric.

114
Algorithm 4.1- AWTSimD: Optimisation of wind turbine blades utilising active control
systems
Given:
 All data required for wind turbine simulation as identified in
o Algorithm 2.2 for constant speed stall-regulated
o Algorithm 3.1 for variable speed stall-regulated
o modified Algorithm 3.1 for constant speed pitch controlled (  in
Algorithm 3.1 replaced by pitch )
o Algorithm 3.2 for variable speed pitch controlled wind turbines
o Algorithms 3.3 and 3.4 as well as modified Algorithm 3.1 (in which  is
replaced by  F ) for constant speed flap controlled wind turbines
o Algorithms 3.3 and 3.4 as well as modified Algorithm 3.2 (in which
pitch is replaced by  F ) for variable speed flap controlled wind turbines
o Algorithms 3.5 and 3.6 for constant speed wind turbines with telescopic
blades
o Algorithms 3.5 and 3.7 for variable speed wind turbines with telescopic
blades
o Algorithms 3.8 and 3.9 for constant speed wind turbines with microtab-
equipped blades

except those blade topology data selected as design variables ( c,  0 , R )

 Selected design variables from the set of { c,  0 , R }, for each selected design
variable the upper and lower bounds
 ndp in case of selecting distributed parameters c and  0
 Site average wind speed Vav
 Cut-in and cut-out velocities ( Vi and Vo ) and wind speed increment V
 Constraints on maximum power Prated and allowable root bending moment
M root,max
 GA parameters: ngen , n pop , pc , pm , xip , xco and t ip ,rand in case of xip  1
 An initial design candidate in case of xip  0
Step 1- Initial population generation:
1.1. i pop  0
1.2. While i pop  n pop xip
1.2.1. Generate a design candidate using the method of perturbation of a
baseline design
1.2.2. Evaluate; If feasible: add to the population; i pop  i pop  1 .
1.3. While i pop  n pop
1.3.1. Randomly generate a design candidate
1.3.2. Evaluate; If feasible: add to the population; i pop  i pop  1 .
1.4. Construct the roulette wheel based on the fitness of individuals
1.5. Find the highest fitness in the initial population fit max and calculate the initial
population average fitness fit av
1.6. i gen  0

115
Algorithm 4.1- AWTSimD: Optimisation of wind turbine blades utilising active control
systems-continue
Step 2- Reproduction:While i gen  n gen    fit max  fit av 
2.1. For iCO  1 : n pop pc xCO
2.1.1. Using the roulette wheel select two parents
2.1.2. Perform arithmetic crossover to form a child
2.1.3. Evaluate; If feasible add to the population.
2.2. For iCO  n pop pc xCO  1 : n pop pc
2.2.1. Using the roulette wheel select two parents
2.2.2. Perform geometric crossover to form a child
2.2.3. Evaluate; If feasible add to the population.
2.3. For iMute  1 : n pop pm
2.3.1. Randomly select an individual
2.3.2. For selected individual, randomly select a gene
2.3.3. For selected gene, randomly change the value within the limits
2.3.4. Evaluate; If feasible add to the population.
2.4. Sort extended population based on the fitness
2.5. Trim the sorted extended population to the size of n pop
2.6. Find the highest fitness in the population fit max and calculate the population
average fitness fit av
2.7. Construct a new roulette wheel
2.8. i gen  i gen  1

Evaluate:
Step 1- Using suitable BEMT algorithm, find the power curve for the generated
design candidate
Step 2- If Pmax  Prated and other constraints are satisfied: feasible = true; calculate
average power and fitness: fitness  Pav

116
4.6 A Case Study: Blade Optimisation for Modified Pitch Controlled AWT-27
Wind Turbine

In Chapter 3, the blade of AWT-27 wind turbine was equipped with various active
control systems and was analysed for constant and variable speed rotors. Results of
modified constant speed AWT-7 with pitch control system were shown in Figure (3.9).
In these analyses the following search parameters for the control system were used:
 Pitch: pitch l  5 , pitch u  5 , step pitch   pitch u  pitch l  / 10  1 ,  pitch  0.1

 Power tolerance:  P  0.01Prated  3kW

Using the same set of search parameters as used in Chapter 3 for simulation of the pitch
control, together with the following design/optimisation parameters the optimisation
code AWTSimD is used to optimise the baseline blade of AWT-27 for constant speed
pitch controlled scenario.
 Design variable: pretwist  0

 Number of design points: ndp  5

 Constraints on maximum power Pmax  Prated  300kW

 GA parameters: n gen  40 , n pop  20 , pc  0.3 , pm  0.1 , xip  0.5 , xco  0.5 ,

t ip ,rand  3 (decreasing)

 Site average wind speed Vav  5.7m / s ; Cut-in and cut-out velocities Vi  5m / s ,

Vo  25m / s and wind speed increment V  1m / s

Results of design optimisation are shown in Figures (4.14) and (4.15). Figure (4.14)
shows the pretwist distribution of the optimised blade. As it can be seen the optimised
pretwist has a sharper slope compared to the initial pretwist. A sharper behaviour was
excepted as by adding a pitch control system, the blade pitch angle varies as blade tends
to enter deep stall due to its sharper pretwist variation.

Figure (4.15) compares the power coefficients of the baseline stall regulated, baseline
wind turbine with added pitch control system and optimised wind turbine with added
pitch control system. As expected, the power coefficient improves due to optimisation
of the blade pretwist. Figure (4.16) shows the optimisation search history. Rapid growth
of the population average fitness shown in this figure indicates the robustness of the

117
crossover operator used in the optimisation algorithm. On the other hand, gradual
improvement of the fitness of the best design candidate in the population to the latest
generations indicates the diversity of the population and the effectiveness of the
mutation operator.

10
Optimised Constant Speed Pitch Controlled
8 Baseline
Pretwist (degrees)

-2
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)

Figure 4.14-Optimised blade pretwist for modified pitch controlled AWT-27

0.50 Optimised Constant Speed Pitch Controlled

0.45 Constant Speed Pitch Controlled
0.40 Baseline
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)

Figure 4.15-Optimum power coefficient of modified pitch controlled AWT-27

48800

48600

48400
Fitness

48200 Max Fitness

Average Fitness
48000

47800
0 5 10 15 20 25 30 35 40
Generation
Figure 4.16-Search history for optimisation of the pretwist for modified AWT-27
constant speed pitch controlled blade

118
4.7 Summary
In this chapter first three phases of design of wind turbine blades were explained.
Elaborating on the direct and inverse design methods, an inverse design method for
design of variable speed wind turbine blades, taken from literature, was explained.
Applying the inverse design method to redesign the blade of AWT-27, it was shown
that the inverse design methods do not perform well in optimisation of blades topology.
Adopting a direct, also called search-base, design optimisation method, the aerodynamic
design of blades was formulated as a standard optimisation problem in which the
objective is to maximise the annual average power subject to constraints on maximum
power and blade loading.

The second part of this chapter details the genetic algorithm-based optimisation module
with some advanced features designed particularly for wind turbine blade optimisation
application.

The final part of this chapter elaborates on integration of AWTSim simulation tool and
the GA optimisation module towards creating AWTSimD, a design optimisation tool
for wind turbine blades equipped with nonconventional control systems. The
performance of the tool was demonstrated by performing a design optimisation case
study.

119
5 Design Optimisation of
Wind Turbine Blades
Equipped with
Nonconventional
Aerodynamic Control
Systems

120
5.1 Introduction
In Chapter 3, Section 3.9 (A preliminary comparison of different types of control
systems) different control systems were compared against each other and some
preliminary conclusions were drawn. It was mentioned that since none of the blades had
been optimised to operate optimally, some of the conclusions might not be valid for
optimally designed blades. In this chapter, employing the optimisation tool AWTSimD,
the potentials of flap trailing edge, microtab and telescopic blades in enhancing energy
capture capability of blades are investigated. In all cases only the pretwist is considered
for optimisation to keep the optimised blade structurally as close as the baseline blade,
making comparison possible. In the case of trailing edge flaps and telescopic blades
both constant speed and variable speed rotors are investigated. In the case of microtab,
only constant speed rotor is considered. Sample simulation and design optimisation
input files are given in Appendix A.

5.2 Potentials of Trailing Edge Flaps in Power Extraction Enhancement:

Constant Speed Rotors
In design optimisation of blades equipped with trailing edge flap three parameters,
namely, blade pretwist, flap length and flap location, are considered as design variables.
Referring to Figure (3.14), parameter RF ,e  RF ,s  stands for the flap length and

0.5RF ,e  RF ,s  , the radial location of the centre of the flap, represents the flap location.

In all cases the width of the flap d F is considered as 10% of the local chord C F (
 
d F  0.1C F ), where C F  chord @ r  0.5 RF ,e  RF , s . In simulating the flap control

system the following data are used:  F , l  20 ,  F , u  20  , step  F  4 ,   F  0.1

 

and  P  3kW .

Table (5.1) shows the examined flap lengths and flap locations for this study. For each
case shown in this table, using AWTSimD, the pretwist is optimised. For optimisation
module the following data were used:
 Design variable: pretwist  0

 Number of design points: ndp  5

 Constraints on maximum power Pmax  Prated  300kW

 GA parameters: n gen  40 , n pop  20 , pc  0.3 , pm  0.1 , xip  0.5 , xco  0.5 ,

t ip ,rand  3 (decreasing)
121
 Site average wind speed Vav  5.7m / s ; cut-in and cut-out velocities Vi  5m / s ,

Vo  25m / s and wind speed increment V  1m / s

Table 5.1- Examined flap lengths and flap locations (all values in % of R )
Flap location Flap length
0.5RF ,e  RF ,s 
Case RF , s R F ,e
RF ,e  RF ,s
1 60 65 62.5 5
2 65 70 67.5 5
3 70 75 72.5 5
4 75 80 77.5 5
5 80 85 82.5 5
6 85 90 87.5 5
7 90 95 92.5 5
8 60 70 65 10
9 60 75 67.5 15
10 60 80 70 20
11 60 85 72.5 25

Figure (5.1) shows the results for Case 1. The pretwist of the baseline AWT-27 and the
optimised pretwist are shown in Figure (5.1.a). Figures (5.1.b) through (5.1.d) show the
performance of wind turbine for three cases: (i) original AWT-27 without installing
flap, (ii) original AWT-27 blades equipped with flap (with original pretwist), and (iii)
optimised AWT-27 blades equipped with flap. Comparing the results of the optimised
blade with original blade, it can be observed that: the blade equipped with flap produces
more energy (Figures (5.1.b) and (5.1.c)) without increasing blade loading (Figure
(5.1.d)). Appendix C contains the results for other cases.

The amount of enhancement in the average power due to equipping blades with flap for
each case of Table (5.1) is shown in Figures (5.2) and (5.3). The calculations for the
average power are based on a site average wind speed of 5.7m/s and Rayleigh
probability density function. In this figure, results are shown for original blades without
flap, original blades with flap and optimised blades with flap.

Using the data shown in Figure (5.3), share of installing flap and optimisation in the
power enhancement are shown separately in Figure (5.4). Figure (5.5) shows the effect
of the flap size on the power enhancement.

122
 10
9
8
Baseline
7

Pretwist (degrees)
6 Optimised (Flap 60-65%)
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)
(a)
350000
300000
Rotor Power (W)

250000
200000
150000
Baseline
100000
Flap 60-65%
50000
Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Power Coefficient (-)

0.40
Flap 60-65%
0.35
0.30 Flap 60-65%-Optimised
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
180000
160000
Max Flap Bending Moment

140000
120000
100000
80000
(Nm)

60000 Baseline
40000 Flap 60-65%
20000 Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure 5.1-Flap 60-65%-constant speed rotor
123
 50
Optimised Blade with Flap
49 Original Blade with Flap
Original Blade No Flap

Average Power (KW)

48

47

46

45

44

43

Flap Location (% of span)

Figure 5.2-Effect of flap size and location on power enhancement-constant speed rotor

14 Optimised Blade with Flap

% Increase in Average Power

12 Original Blade with Flap

10
8
(KW)

6
4
2
0

Flap Location (% of span)

Figure 5.3-Percent increase in the average power versus flap size and location-constant
speed rotor

14
% Increase in Average Power

Contribution of Flap
12
Contribution of Optimisation
10
(KW)

8
6
4
2
0

Flap Location (% of span)

Figure 5.4-Percent improvement in the average power due to blade optimisation-

constant speed rotor

124
 14

% Increase in Average
12

Power (KW)
10
8
6
4
2
0
0 5 10 15 20 25
Flap Size (% of span)
Figure 5.5-Effect of flap size on the power extraction enhancement-constant speed
rotor

Figures (5.6) through (5.8) show the power curves, power coefficient and maximum
flap bending moment for all 11 examined cases for which the blade has been optimised.

350000
Rotor Mechanical Power

300000
250000 Baseline
200000 60-65%
(Watt)

65-70%
150000 70-75%
100000 75-80%
80-85%
50000 85-90%
90-95%
0
5 10 15 20 25
Wind Speed (m/s) (a)
350000
Rotor Mechanical Power

300000
250000
200000 Baseline
(Watt)

60-65%
150000 60-70%
100000 60-75%
60-80%
50000 60-85%
0
5 10 15 20 25
Wind Speed (m/s) (b)

Figure 5.6-Power produced by constant speed rotors utilising blades equipped with flap

125
 0.50 Baseline
60-65%
0.40 65-70%
70-75%

Power Coefficient (-)

0.30 75-80%
80-85%
0.20 85-90%
90-95%
0.10

0.00
5 10 15 20 25
Wind Speed (m/s) (a)
0.50
Baseline
Power Coefficient (-)

0.40 60-65%
60-70%
0.30 60-75%
60-80%
0.20 60-85%

0.10

0.00
5 10 15 20 25
Wind Speed (m/s) (b)
Figure 5.7-Power coefficient of constant speed rotors utilising blades equipped with flap

200000
Max Flap Bending

150000
Moment (Nm)

Baseline
60-65%
100000 65-70%
70-75%
75-80%
50000 80-85%
85-90%
90-95%
0
5 10 15 20 25
Wind Speed (m/s) (a)
200000
Max Flap Bending

150000
Moment (Nm)

Baseline
100000
60-65%
60-70%
50000 60-75%
60-80%
60-85%
0
5 10 15 20 25
Wind Speed (m/s) (b)
Figure 5.8- Maximum flap bending moment in blades equipped with flap-constant speed
rotor

126
According to the above figures the following conclusion can be drawn:
 Adding flap without optimisation improve the power extraction capability as high as
of about 5% for the case of flap located between 60-85% of span. However,
optimisation of the blade is required to obtain the highest power improvement
(Figure (5.4)). Improvement as a result of optimisation can be as high as 7% (for
case of 60-80%). The overall improvement can be reached as high as 12%.
 Location of flap is a key parameter influencing the amount of improvement in the
power extraction (Figures (5.2.a) and (5.3.a)). The best location for placing a flap is
at about 70% of the blade span from the root of the blade.
 The size of the flap has also significant effect on the amount of enhancement in the
average power. This effect, however, reduces dramatically as the size increases
(Figure (5.5)).

5.3 Potentials of Trailing Edge Flaps in Power Extraction Enhancement:

Variable Speed Rotors

Using the same set of data, blades of a variable speed AWT-27 are optimised for all 11
cases of Table (5.1). For these optimisation case studies it is assumed that  l  30rpm ,

 u  65rpm , step   u  l  / 10  3.5rpm ,    0.1rpm , as used in Chapter3.

Figure (5.9) shows the results for Case 1. Appendix C contains the results for other
cases. Similar to Section 5.2, here for the variable speed rotor, the average power and
the enhancement in the average power for all 11 cases of Table (5.1) is calculated and
shown in Figures (5.10) through (5.13). Figures (5.14) through (5.16) show the power
curves, power coefficient and maximum flap bending moment for all 11 examined cases
for which the blade has been optimised.

127
 10
9
8
Baseline

Pretwist (degrees)
7
6 Optimised (Flap 60-65%)
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 60-65%
Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 60-65%
0.40 Flap 60-65%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
200000
180000
Max Flap Bending Moment

160000
140000
120000
100000
(Nm)

80000
60000
Baseline
40000 Flap 60-65%
20000 Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure 5.9-Flap 60-65%-variable speed rotor
128
 53
52

Average Power (KW)

51
50
49
48
47
46
45
44
43

Flap Location (% of span)

Optimised Blade with Flap
Original Blade with Flap
Original Blade No Flap
Figure 5.10-Effect of flap size and location on power enhancement-variable speed rotor

20
% Increase in Average Power

18
16
14
12
10
(KW)

8
6
4
2
0

Flap Location (% of span)

Optimised Blade with Flap Original Blade with Flap

Figure 5.11-Percent increase in the average power versus flap size and location-variable
speed rotor

20
% Increase in Average Power

18
16
14
12
10
(KW)

8
6
4
2
0

Flap Location (% of span)

Contribution of Flap Contribution of Optimisation

Figure 5.12-Percent improvement in the average power due to blade optimisation-

variable speed rotor
129
 20
18

% Increase in Average
16
14

Power (KW)
12
10
8
6
4
2
0
0 5 10 15 20 25
Flap Size (% of span)
Figure 5.13-Effect of flap size on the power extraction enhancement-variable speed
rotor

350000
Rotor Mechanical Power

300000
250000 Baseline
60-65%
200000
(Watt)

65-70%
150000 70-75%
75-80%
100000 80-85%
50000 85-90%
90-95%
0
5 10 15 20 25
Wind Speed (m/s) (a)
350000
Rotor Mechanical Power

300000
250000
Baseline
200000
(Watt)

60-65%
150000 60-70%
60-75%
100000
60-80%
50000 60-85%
0
5 10 15 20 25
Wind Speed (m/s) (b)

Figure 5.14-Power produced by variable speed rotors utilising blades equipped with flap

130
 0.50 Baseline
60-65%
65-70%
0.40
70-75%
75-80%
0.30

Power Coefficient (-)

80-85%
85-90%
0.20 90-95%

0.10

0.00
5 10 15 20 25
Wind Speed (m/s) (a)
0.50
Baseline
Power Coefficient (-)

0.40 60-65%
60-70%
0.30 60-75%
60-80%
0.20 60-85%

0.10

0.00
5 10 15 20 25
Wind Speed (m/s) (b)
Figure 5.15-Power coefficient of variable speed rotors utilising blades equipped with
flap

200000
Max Flap Bending

150000
Moment (Nm)

Baseline
60-65%
100000 65-70%
70-75%
75-80%
50000 80-85%
85-90%
90-95%
0
5 10 15 20 25
Wind Speed (m/s) (a)
200000
Max Flap Bending

150000
Moment (Nm)

100000 Baseline
60-65%
60-70%
50000 60-75%
60-80%
60-85%
0
5 10 15 20 25
Wind Speed (m/s) (b)
Figure 5.16-Maximum flap bending moment in blades equipped with flap-variable
speed rotor
131
According to the above figures the following conclusion can be drawn:
 Comparing the results of the optimised blade with original blade, it can be observed
that using flaps on variable speed rotors can have a twofold effect. Firstly, blades
equipped with flap produce more energy (Figure (5.14) and (5.15)). Secondly, for
some cases, flap causes a reduction of the blade loading (see V=25 m/s in Figure
(5.16)).
 Adding flap without optimisation improve the power extraction capability as high as
of about 14% for case of flap located between 60-75% of span (Figure (5.11)). In
contrary to the constant speed rotor, the effect of optimisation is less and is limited
to only 6% (Figure (5.12)). The overall improvement can be reached as high as
18%.
 In contrary to constant speed rotors, neither the location of flap nor the size of the
flap affects the performance significantly. This is mainly due to having rotor speed
as the dominant controlling system in place(Figures (5.11) and (5.13)).

5.4 Potentials of Telescopic Blades in Power Extraction Enhancement

Following the same approach as taken in Sections 5.2 and 5.3, for two cases of Table
(5.2) the telescopic blades are optimised and compared with the original blades and
telescopic blades without optimisation. All simulation, design and optimisation
parameters are as given in Sections 5.2 and 5.3. Results are shown in Figures (5.17) and
(5.18) for constant speed rotors and Figures (5.19) and (5.20) for variable speed rotors.

Table 5.2- Examined telescopic ranges (all values in % of R )

Case RT , s RT , e

1 80 105
2 80 110

132
 350000

Rotor Mechanical Power

300000
250000
200000

(Watt)
150000
100000
50000
0
5 10 15 20 25
Wind Speed (m/s) (a)
0.50
0.45
0.40
0.35
0.30
Power Coefficient (-)

0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (b)
300000
Max Flap Bending Moment (Nm)

250000

200000

150000

100000

50000

0
5 10 15 20 25
Wind Speed (m/s)
Baseline
Constant Speed (80%-105%)
Constant Speed Optimised (80%-105%)
Constant Speed (80%-110%)
Constant Speed Optimised (80%-110%) (c)
Figure 5.17-Optimised telescopic blade-constant speed rotor

133
 52.00
49.74

Average Power (kW)

50.00
48.00 46.49
45.00 45.32
46.00
43.86
44.00
42.00
40.00
Baseline Telescopic Optimised Telescopic Optimised
(80%-105%) Telescopic (80%-110%) Telescopic
(80%-105%) (80%-110%)
(a)
13.4
14.0
% Increase in Average

12.0
10.0
8.0 6.0
Power

6.0
2.6 3.3
4.0
2.0
0.0
Telescopic Optimised Telescopic Optimised
(80%-105%) Telescopic (80%-110%) Telescopic
(80%-105%) (80%-110%)
(b)
15
Average Power
% Increase in

3.3
10
(KW)

5 2.6 10.1
3.4
0
80%-105% 80%-110%
Telescopic Range(% of span)
Contribution of Telescopic Effect Contribution of Optimisation
(c)
Figure 5.18-Power enhancement via utilising telescopic blades-constant speed rotor

In view of Figures (5.17) and (5.18) the following conclusions can be drawn:

 For constant speed rotors, as expected, utilising telescopic blades improves the
amount of the power extraction (Figure (5.17.a)). This amount slightly increases
when the telescopic part of the blade is allowed to expand more (2.6% versus 3.3%
in Figure (5.18.b)). This difference become significant when the blades are
optimised (6% versus 13.4% in Figure (5.18.b)).
 For constant speed rotors, the power coefficient improves only for higher wind
speeds above 10 m/s (Figure (5.17.b)). Recalling that the power coefficient is the
ratio of the extracted power by the rotor and the available wind power over the rotor
area, increase in power coefficient in higher wind speeds is partly due to higher
extracted power and partly due to blade contraction. On the other hand, the power
coefficient at lower wind speeds is significantly less than the baseline AWT-27 wind
turbines. Drop in the power coefficient in low winds, despite extracting more power,
is due to blade extension and consequently an increase in the rotor area.
134
 The maximum flap bending moment in the blades significantly increases with the
maximum length of the telescopic blades (Figure (5.17.c)). Optimisation of the
blade pretwist reduces the added blade loading significantly.
 Similar to blades equipped with flap, for constant speed rotors the effect of blade
optimisation on the power enhancement is significant. The power enhancement is
mainly due to the optimisation (Figure (5.18.c)).
350000
Rotor Mechanical Power

300000
250000
200000
(Watt)

150000
100000
50000
0
5 10 15 20 25
Wind Speed (m/s) (a)
0.50
0.45
0.40
0.35
0.30
Power Coefficient (-)

0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (b)
350000
Max Flap Bending Moment (Nm)

300000
250000
200000
150000
100000
50000
0
5 10 15 20 25
Wind Speed (m/s)
Baseline
Variable Speed (80%-105%)
Variable Speed Optimised (80%-105%)
Variable Speed (80%-110%)
Variable Speed Optimised (80%-110%) (c)
Figure 5.19-Optimised telescopic blade-variable speed rotor
135
 60.00 56.75

Average Power (kW)

55.00 52.65 52.59

50.00 48.25

45.00 43.86

40.00
Baseline Telescopic Optimised Telescopic Optimised
(80%-105%) Telescopic (80%-110%) Telescopic
(80%-105%) (80%-110%)
(a)
30.0 29.4
% Increase in Average

25.0 20.0 19.9

20.0
15.0
Power

10.0
10.0
5.0
0.0
Telescopic Optimised Telescopic Optimised
(80%-105%) Telescopic (80%-110%) Telescopic
(80%-105%) (80%-110%)
(b)
30
Average Power

25
% Increase in

20 19.9
15 10
(KW)

10
5 10 9.5
0
80%-105% 80%-110%
Telescopic Range(% of span)
Contribution of Telescopic Effect Contribution of Optimisation (c)

Figure 5.20-Power enhancement via utilising telescopic blades-variable speed rotor

According to Figures (5.19) and (5.20) the following conclusions can be drawn:

 For variable speed rotors, utilising telescopic blades improves the amount of the
power extraction (Figure (5.19.a)). This amount increases as the telescopic part of
the blade is allowed to expand more (10% versus 19.9% in Figure (5.20.b)). This
difference remains significant when the blades are optimised (20% versus 29.4% in
Figure (5.18.b)).
 For variable speed rotors, the power coefficient improves for both low wind and
high wind regions. This is mainly due to having two controlling parameters (blade
span and rotor speed).
 The maximum flap bending moment in the blades significantly increases with the
maximum length of the telescopic blades (Figure (5.19.c)). Optimum blades have
significantly less maximum flap bending moments.

136
5.5 Potentials of Microtabs in Power Extraction Enhancement
In Section 3.8 it was shown that unlike other controlling devices, microtab has little
effect on the power extraction enhancement when installed on the blade. In this section,

, s  0.6 to RMT ,e  0.9 , d MT  80%

* * *
adopting the same case study as in Section 3.8 ( RMT
*
and d MT  95% of the chord from the leading edge on the upper and lower surface
*
respectively, with an actuation height of hMT  3.3% of the chord length), the blades are
also optimised for pretwist. Results are shown in Figures (5.21) and (5.22).

According to Figures (5.21), (5.22.b) and (5.22.c), it is evident that pretwist

optimisation has significant influence on the extracted power. The baseline blades
extract more or less the same amount of power, with or without microtab, unless the
pretwist is optimised.

5.0
4.2
% Increase in Average

4.0

3.0
Power

2.0

1.0
0.1
0.0
Baseline Blade with MT Optimised Blade with MT

Figure 5.21-Microtab on optimised blades-power enhancement

137
 10
9
8
7 Baseline

Pretwist (degrees)
6 Optimised Blade
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
Baseline
150000 Baseline blade with microtabs
Optimisedblade with microtabs
100000
50000
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
0.40 Baseline blade with microtabs
Optimisedblade with microtabs
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
180000
160000
Max Flap Bending Moment

140000
120000
100000 Baseline
(Nm)

80000 Baseline blade with microtabs

60000 Optimisedblade with microtabs
40000
20000
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure 5.22-Microtab on optimised blades
138
5.6 Summary
In this chapter, employing the optimisation tool AWTSimD, the potentials of flap
trailing edge, microtab and telescopic blades in enhancing energy capture capability of
blades were investigated. In all cases only the pretwist was considered for optimisation
keeping the optimised blade structurally as close as possible to the baseline blade. In
case of trailing edge flaps the effect of size and location of the flap was also
investigated. In all cases the amount of enhancement in power production was broken
down into two parts, namely, contribution of the controlling device (flap, microtab and
telescopic blade) and the contribution of the optimisation. It was shown that
optimisation of the blade plays a major role in enhancing the power capture capability
for constant speed rotors.

139
6 Summary and Conclusion

140
6.1 Summary of Work, Achievements and Contribution

To meet the overall goal of this research: “investigation of the potentials of

nonconventional control systems which have been initially developed for load control in
energy capture capability enhancement”, a design optimisation code, capable of
simulating wind turbines with constant and variable speed rotors utilising telescopic
blades as well as blades equipped with microtab and trailing edge flaps was developed.

The aerodynamic analysis module of the code is based on the well-established method
of BEMT. A software tool, called AWTSim, was developed for aerodynamic analysis of
wind turbines. Using a stall-regulated constant speed test wind turbine, the performance
of AWTSim was validated against WTPerf, a simulation code accredited by NREL.
Further necessary enhancements to AWTSim were applied to make it capable of
simulating wind turbines with nonconventional control systems.

Analysis of the controlling system is based on a realistic assumption: the actual

controllers perform perfectly in finding and adjusting the controlling parameters to their
best (optimum) values. This assumption was then used to transform “simulation of the
controlling system” to “solving an optimisation problem”. It was shown that this
method performs well with sufficient accuracy and robustness as required for the
aerodynamic analysis of nonconventional blades. In solving the corresponding
optimisation problem, three methods of optimisation, namely, hill climbing, pattern
search and genetic algorithm were tailored and used to solve the optimisation problem.
The performance of the hill climbing and pattern search methods was evaluated and
reported.

AWTSim is a unique simulation tool capable of simulating the following types of wind
turbines:

1. Constant speed stall regulated rotors

2. Variable speed stall regulated rotors
3. Constant speed pitch controlled rotors
4. Variable speed pitch controlled rotors
5. Constant speed flap controlled rotors
6. Variable speed flap controlled rotors
141
7. Constant speed rotor with telescopic blades
8. Variable speed rotors with telescopic blades
9. Constant speed rotors with blades equipped with microtab

For each type a case study was carried out to demonstrate the capability and the
performance of the developed wind turbine simulator. It should be noted that, none of
the wind turbine simulators reported in open literature is capable of simulating wind
turbines of types 5 to 9 above.

For the optimiser module of the code, a robust genetic algorithm, with some advanced
features such as geometric crossover and initial population generation based on
perturbation of an existing design candidate, was developed. Integration simulation tool
AWTSim into the optimiser module constructs the design optimisation code
AWTSimD.

Employing AWTSim and AWTSimD and investigating three nonconventional control

systems, microtab, trailing edge flap and telescopic blades led to the following results.

Trailing edge flaps

 Adding flap without optimisation improves the power extraction capability as high
as of about 5% and 14% for constant speed and variable speed rotors, respectively.
Significant further improvement as a result of optimisation can be achieved (up to
7%) for constant speed rotors, while the effect of optimisation is less and limited to
only 6% for variable speed rotors. The overall improvement in the produced power
can be reached as high as 12% and 18% for constant speed and variable speed rotors
respectively.
 For constant speed rotors, the location of flap is a key parameter influencing the
amount of improvement in the power extraction. The best location for placing a flap
is at about 70% of the blade span from the rotor centre. The size of the flap has also
significant effect on the amount of enhancement in the average power. This effect,
however, reduces dramatically as the size increases. For variable speed rotors,
neither the location of flap nor its size affects the performance significantly.
 Flaps, when used in conjunction with another controlling system such as rotor
speed, the accompanied controlling system dominates the control process.

142
Telescopic blades
 Telescopic blades provide a full and smooth power control.
 Utilising telescopic blades in both constant and variable speed rotors improves the
amount of the power extraction. Power extraction enhancement is higher in variable
speed rotors. However, this enhancement comes with the price of a significant
increase in bending moment at the root of blade.
 In constant speed rotors, the improvement in the power extraction slightly increases
with the fully extended length of the blade (2.6% for permissible extended length of
105% versus 3.3% for permissible extended length of 110% of the baseline rotor
radius). This difference becomes significant when the blades are optimised (6%
versus 13.4% respectively). For variable speed rotors, the enhancement in the
produced power is more significant and sensitive to the permissible extended length
(10% for permissible extended length of 105% versus 19.9% for permissible
extended length of 110%). This difference remains significant when the blades are
optimised (20% versus 29.4%).
 For constant speed rotors, the power coefficient improves only for higher wind
speeds, while at lower wind speeds it is significantly less than the baseline wind
turbines. Drop in the power coefficient in low winds, despite extracting more power,
is due to blade extension and consequently increase in the rotor area. In case of
variable speed rotors, the power coefficient improves for both low wind and high
wind regions. This is mainly due to having both blade span and rotor speed as
controlling parameters.
 In both constant and variable speed rotors, the maximum flap bending moment in
the blades significantly increases with the maximum length of the telescopic blades,
unless the blades are optimised.

Microtabs:
 The baseline blades extract more or less the same amount of power, with or without
microtab, unless the pretwist is optimised.
 Microtabs on optimised blades can improve the produced power by up to 4%.

Also, it was shown that pitch control is the most efficient control system having the
highest power coefficient and minimal blade root bending moment for both constant
speed and variable speed rotors. For constant speed rotors, optimised telescopic blades
are more effective than flaps in power enhancement. However, in comparison with flap,
143
telescopic blades have two disadvantages: (i) complexity in telescopic mechanism and
the added weight and (ii) increased blade loading.

These results, together with the tools developed as part of this project can be used in
design of more efficient wind turbines.

6.2 Critical Appraisal and Future Work

Optimisation of blades equipped with microtabs is highly time-consuming. Each

optimisation run takes 20 hours on university work stations. This is due to the fact that
each design candidate evaluation as part of the blade optimisation includes a full
optimisation process by itself for the purpose of controller simulation (to obtain the best
microtab states). As a result of this the effect of the location of microtabs was not
investigated in this study.

In investigating the effect of microtabs, only deployment height of 3.3% was studied.
This deployment height has the maximum effect on the lift and drag coefficients.

In investigating the effect of flaps, only the case of flap width of 10% chord and
deployment angle limited to -20 to +20 degrees was investigated. This was mainly due
to lack of available data for other flap widths.

A continuation to the presented research, after addressing above shortcomings, can be

followed by:
 Design of blades from scratch with rotor radius, chord and aerofoil distributions as
design variables in addition to the pretwist.
 Investigating the full potentials of new devices will be completed by performing
cost analysis for each device. Cost analysis should include initial, operating and
maintenance cost.

144
References

145
Ashwill, T. & Laird, D. (2007) Concepts to facilitate very large blades. In: Proceedings
of ASME/AIAA Wind Energy Symposium, Reno, NV. 2007.

Andersen P. B., Gaunaa . M., et al. (2006) Load alleviation on wind turbine blades
using variable airfoil geometry. In: Proceedings of European Wind Energy Conference
and Exhibition2006, Athens.

Baeck, T., Fogel, D. B. & Michalewicz, Z. (2010) Handbook of Evolutionary

Computation. Taylor & Francis.

Baker, J. P., Standish, K. J. & van Dam, C. P. (2005) Two-dimensional wind tunnel and
computational investigation of a microtab modified S809 airfoil, AIAA 2005-1186. In:
Proceedings of the 43rd AIAA/ASME, Reno, NV, USA.

Barlas, T. & Van Kuik, G. M. (2010) Review of state of the art in smart rotor control
research for wind turbines. Aerospace Sciences, 46, 1-27.

Bossanyi, E. A. (2000) The design of closed loop controllers for wind turbines. Wind
Energy, 3.

Bossanyi, E. A. (2003) Individual blade pitch control for load reduction. Wind Energy,
6, 119-128.

Bottrell, G. W. (1981) Passive cyclic pitch control for horizontal axis wind turbines. In:
Proceedings of Wind Turbine Dynamics, NASA Conference Pub. 2185,DOE Pub.
CONF-810226, Cleveland, OH.

BTM Consult ApS, (2005) Ten Year Review of the International Wind Power Industry
1995-2004.

Burton, T., Sharpe, D., Jenkins, N. & Bossanyi, E. Wind energy handbook. John Wiley
& Son Inc.

Buhl, T. & Gaunaa, M., et al. (2005) Potential load reduction using airfoils with
variable trailing edge geometry. Journal of Solar Energy Engineering, 127, 503-516.
146
Buhl, Jr., M. L.(2004) NWTC Design Codes, 2004. WT_Perf Version 3.1. [Online].
Available from: http://wind.nrel.gov/ designcodes/ simulators/wtperf/ [Accessed 28
Aug. 2012].

Buhl, Jr, M. L. (2005) A new empirical relationship between thrust coefficient and
induction factor for the turbulent windmill state. National Renewable Energy
Laboratory, Technical report NREL/TP-500-36834.

Currin, H. (1981) North wind 4kW ‘Passive Control System Design, In: Proceedings
Wind Turbine Dynamics, NASA Pub. 2185, DOE Pub. CONF-810226, Cleveland, OH.

Cheney, M. C. & Speirings, P. S. M. (1978) Self regulating composite bearing less

wind turbine. Solar Energy, 20.

Caselitz, P., Kleinkauf, W., Kruger, T., Petschenka, J., Reichardt, M. & Storzel, K.
(1997) Reduction of fatigue loads on wind energy converters by advanced control
methods. In: Proceedings of the European Wind Energy Conference, Dublin, pp.555–
558.

Chow, R. & van Dam, C.P. (2007) Computational investigations of deploying load
control microtabs on a wind turbine airfoil. AIAA-2007-1018. In: Proceedings of the
45th AIAA/ASME, Reno, NV, USA.

Crespo, A., Hernández, J. & Frandsen, S. (1999) Survey of modelling methods for wind
turbine wakes and wind farms. Wind Energy, vol.2, 1–24.

Corbet, D. C. & Morgan, C. A. (1992) Report on the passive control of horizontal axis
wind turbines. Garrad Hassan and Partners, Bristol, UK, ETSU WN 6043.

DOE. (2005) Variable length wind turbine blade. DE-FG36-03GO13171.

Hella, D. (2012) Producing reliable aerodynamic lookup tables for microtabs installed
on wind turbine blades made of S800 series aerofoils. Northumbria University, MSc
Dissertation.

147
Drela, M. & Giles, M. B.(1987) Viscous-inviscid analysis of transonic and low
Reynolds Number airfoils. AIAA Journal, 25(10), 1347-1355.

Drela, M. (1988) Low-Reynolds Number airfoils design for the MIT daedalus
prototype: a case study. Journal of Aircraft, 25(8), 724-732.

Drela, M. (1989) XFOIL: An analysis and design system for low Reynolds Number
airfoils. In: Proceedings Conference on Low Reynolds Number Airfoil Aerodynamics,
University of Notre Dame.

Drela, M. (1989) Low Reynolds Number aerodynamics. T.J. Mueller (Editor),

[Lecture] Lecture Notes in Engineering #54, Springer Verlag, 1989.

Drela, M. (1989) Integral boundary layer formulation for blunt trailing edges. AIAA-89-
2166.

Drela, M. (1990) Elements of airfoil design methodology, applied computational

aerodynamics. In: Henne, P. (Editor) AIAA Progress in Aeronautics and Astronautics,
125.

Drela, M. (1998) Pros and Cons of Airfoil Optimization, Chapter in Frontiers of

Computational Fluid Dynamics. Caughey, D. A., Hafez, M. M. Eds. World Scientific.

Drela, M. (2008) XFOIL Subsonic Airfoil Development System, 2008. [Online]

Available from: http://web.mit.edu/drela/Public/web/xfoil/ [Accessed 28 August 2011].

Du, Z. & Selig, M. S. (1998) A 3-D stall-delay model for horizontal axis wind turbine
performance prediction. In: Proceedings of the 1998 ASME Wind Energy Symposium,
Reno, NV ,pp. 9–19.

Enenkl, B., Klopper, V., et al., (2002) Full scale rotor with piezoelectric actuated blade
flaps. 28th European Rotorcraft Forum.

Farhan, G. & Phuriwat, A. I. (2008) Skin design studies for variable camber morphing
airfoils. IOPP’S Journal.
148
GE Wind Energy LLC (2006) Advanced wind turbine program next generation turbine
development project, NREL laboratory, Report NREL/SR-5000-38752.

Global Wind Energy Council (2011) Global wind statistic 2010.

Herbert, G. M. J., Iniyan, S., Sreevalsan, E. & Rajapandian, S. (2007) A review of wind
turbine technologies. Renewable and Sustainable Energy Reviews,11.

Hohenemser, K. H. & Swift, A. H. P. (1981) Dynamics of an experimental two bladed

horizontal axis wind turbine with blade cyclic pitch variation. NASA, Report NASA
82N23716.

Holland, J. H. (1975) Adaptation in Natural and Artificial Systems . MIT Press,

Cambridge, MA.

Johnson, S. J., van Dame, C.P. & Berg, D.E. (2008) Active load control techniques for
wind turbines. SANDIA, Report SAND2008-4809.

Jureczko, M., Palwak, M. & Męźyk, A. (2005) Optimisation of wind turbine blade.
Materials Processing Technology, 167, 463-471.

Johnson, W. (1982) Self-Tuning Regulators for Multicyclic Control of Helicopter

Vibration, NASA, Report NASA.

Jonkman, J. M. (2003) Modelling of the UAE wind turbine for refinement of FAST AD.
National Renewable Energy Laboratory, Technical Report NREL/TP-500-34755.

Karaolis, N. M. (1989) The design of fibre reinforced composite blades for passive and
active wind turbine aerodynamic control. Dissertation University of Reading.

Kooijiman, H. J. T. (1996) Smart rotor, feasibility study for a passive pitch regulation
of a wind turbine rotor blade. ECN, Report Petten ECN-I-96-0011 ECN.

Larsen, T., Madson, H. & Thomson, K. (2005) Active load reduction using individual
pitch, based on local blade flow measurements. Wind Energy, 8, 67–80.
149
Leishman, G. (2000) Principles of helicopter aerodynamic. Cambridge Univ. Press,
Cambridge.

Lobitz, D. W., Veers, P. S., Laino, D. J., Migliore, P. G. & Bir, G. (2001) The use of
twist-coupled blades to enhance the performance of horizontal axis wind turbine.
Sandia National Laboratories, Report SAND2001-1003.

Lovera, Colaneri, M. P., Malpica, C. & Celi, R. (2003) Closed-loop stability analysis of
HHC and IBC, with application to a hinge less rotor helicopter. In: Proceedings 29th
European Rotorcraft Forum.

Macquart, T. (2012) Flap data generation using XFOIL. Internal Report, Northumbria
University.

Madsen, H. A. (1996) A CFD analysis for the actuator disc flow compared with
momentum theory results. In: Proceedings 10th IEA Symp. Aerodyn. Wind Turbines,
Tech. Univ. Denmark, Lyngby, Denmark, pp. 109–24.

Madsen, H. A. & Rasmussen, F. (2004) A near wake model for trailing vorticity
compared with the blade element momentum theory. Wind Energy, 7, 325-34.

Maheri, A. (2006) Aero-structure simulation and aerodynamic design of wind turbines

utilising adaptive blades. University of West England-Bristol, PhD Thesis.

Maheri, A., Noroozi, S., Toomer, C. & Vinney, J. (2006a) Single step versus coupled
aero-structure simulation of a wind turbine with bend-twist adaptive blades. European
Wind Energy Conference EWEC 2006, 27 February-2 March 2006,Athens, Greece.

Maheri, A., Noroozi, S., Toomer, C. & Vinney, J. (2006b) A Simple algorithm to
modify an ordinary wind turbine blade to an adaptive one. European Wind Energy
Conference EWEC 2006, 27 February-2 March 2006,Athens, Greece.

Maheri, A., Noroozi, S. & Vinney, J. (2007a) Application of combined analytical/FEA

coupled-aero-structure simulation in design of wind turbine adaptive blades. Renewable
Energy, 32(12), 2011-2018.
150
Maheri, A., Noroozi, S. & Vinney, J. (2007b) Decoupled design of wind turbine
adaptive blades. Renewable Energy, 32(10), 1753-1767.

Maheri, A. & Isikveren, A. T. (2008) Variable-state design parameters in design of

aero-structures made of intrinsically smart materials. High Performance Structures and
Materials IV. Editors de Wilde, W. P. & Brebbia, C.A., WIT Press, pp. 421-430.

Maheri, A. & Isikveren, A. T.(2008) Facilitating meta-design techniques for multi-

disciplinary conceptual design. Adaptive Computing in Design and Manufacture,
Bristol,UK.

Maheri, A. & Isikveren, A.T.(2009a) Design of wind turbine passive smart blades.
European Wind Energy Conference EWEC 2009, Marseille, France.

Maheri, A. & Isikveren, A. T. (2009b) A parallel genetic algorithm for optimal

synthesis of a morphing aerofoil. In: Topping, B.H.V. & Ivanyi, P. (Editors) In :
Proceedings of the First International Conference on Parallel, Distributed and Grid
Computing for Engineering. Civil-Comp Press, Stirlingshire, UK, Paper 9.

Maheri, A. & Isikveren, A. T. (2010) Performance prediction of wind turbines utilising

passive smart blades: approaches and evaluation. Wind Energy, 3, 255-265.

Maheri, A., Macquart, T., Safari, D. & Maheri, M. R. (2012) Phenotype Building
Blocks and Geometric Crossover in Structural Optimisation. In: Topping, B.H.V.
(Editors) In: Proceedings of the Eighth International Conference on Engineering
Computational Technology. Civil-Comp Press, Stirlingshire, UK, Paper 5.

Mayda, E. A., van Dam, C. P., & Nakafuji, D. Y. (2005) Computational investigation of
finite width microtabs for aerodynamic load control. AIAA-2005-1185. In: Proceedings
of the 43th AIAA/ASME, Reno, NV.

Mēndez, J. & Greiner, D. (2006) Wind blade chord and twist angle optimization using
genetic algorithms. In: Proceedings of the Fifth International Conference on
Engineering Computational Technology.
151
Migliore, P. G., Quandt, G. A., et al. (1995) Wind turbine trailing edge aerodynamic
brakes. National Renewable Energy Laboratory ,Technical Report NREL/TP-441-
7805.

Michalewicz, Z. (1992) Genetic Algorithms + Data Structures = Evolution Program .

Springer Verlag.

Moriarty, P. J., & Hansen, A. C. (2005) AeroDyn theory manual. National Renewable
Energy Laboratory, Technical report NREL/TP-500-3688.

Nijssen, R. P. L. (2006) Fatigue life prediction and strength degradation of wind turbine
rotor blade composites. Knowledge Centre Wind Turbine Materials and Constructions
(KC-WMC), Netherland.

Nakafuji, D. T. Y., van Dam, C. P., Michel, J. & Morrison, P. (2002) Load control for
wind turbines - a non-traditional microtab approach. AIAA2002-0054.In: Proceedings
of the 40th AIAA/ASME, Reno, NV, USA.

Pasupulati, S. V., Wallace, J. & Dawson, M. (2005) Variable Length Blades Wind
Turbine. Energy Unlimited Inc.

Robert, G. & Jochen, T. (2011) Wind power plants fundamentals, design, construction
and operation.Springer.

Sanderse, B., van der Pijl, S. P. & Koren, B. (2011) Review of computational fluid
dynamics for wind turbine wake aerodynamics. Wind Energy, 14, 799-819.

Sanderse, B. (2009) Aerodynamics of wind turbine wakes Literature review. ECN,

Report ECN-E-09-016.

Seki, K., Shimizu, Y. & Zhu, K. (1996) A design strategy for the improvement of an
existing 300kW WTGs rotor blade. Renewable Energy, 9, 858-86.

Shrama, R. N. & Madawala, U. (2007) The concepts of a smart wind turbine system.
In: Proceeding of 16th Australian Fluid Mechanics Conference, Australia.
152
Snel, H. & Schepers, J. G. (1995) Joint investigation of dynamic inflow effects and
implementation of an engineering method, ECN, Report Petten ECN-C-94-107.

Snel, H. (1998) Review of the present status of rotor aerodynamics, Wind Energy, 1,
46–69.

Snel, H. (2003) Review of aerodynamics for wind turbines, Wind Energy, 6, 203–211.

Stuart, J. G., Wright, A. D. & Butterfield, C. P. (1996) Considerations for an integrated

wind turbine controls capability at the national wind technology centre: an aileron
control case study for power regulation and load mitigation. National Renewable
Energy Laboratory, Technical Report NREL/TP-440-21335.

Spera, D. A. (1994) Wind Turbine Technology, Fundamental Concepts of Wind Turbine

Engineering. American Society of Mechanical Engineers.

Stuart, J. G., Wright, A. D., et al. (1997) Wind turbine control systems: dynamic model
development using system identification and the fast structural dynamics code. In:
Proceedings 35th AIAA/ASME.

Tangler, J. L. (2002) The nebulous art of using wind-tunnel airfoil data for predicting
rotor performance. National Renewable Energy Laboratory, Report NREL/CP-500-
31243.

Tangler, J. L. (2002) The nebulous art of using wind tunnel aerofoil data for predicting
rotor performance. Wind Energy, 5,245–257.

Tangler, J. L. & Kocurek, J. D. (2004) Wind turbine post-stall airfoil performance

characteristics guidelines for blade element momentum methods. National Renewable
Energy Laboratory, Report NREL/CP-500-36900.

Troldborg, N. (2005) Computational study of the risø-b1-18 airfoil with a hinged flap
providing variable trailing edge geometry. Wind Engineering, 29(2), 89-113.

153
Van der Hooft, E. L., Schaak, P. & Van Engelen, T. G. (2003) Wind turbine control
algorithms. ECN, Report ECN-C-03-111.

Van Engelen, T. & Van der Hooft, E. (2003) Individual pitch control inventory. ECN,
Report ECN-C–03-138.

Vermeer, L. J., Sørensen, J. N. & Crespo, A. (2003) Wind turbine wake aerodynamics.
Progress in Aerospace Sciences,39, 467–510.

Wind Turbine Dynamics, (1985) NASA Pub. 2185, DOE Pub. CONF-810226,
Cleveland, OH.

Wright, A. & Barlas, M. (2002) Design of State-Space-Based Control Algorithms for

Wind Turbine Speed Regulation. In : Proceeding of the 2002 ASME Wind Energy
Symposium, Reno, Nevada. pp. 299-309.

Yen, D. T., van Dam, C. P., Smith, R. L. & Collins, S. D. (2001) Active load control
for wind turbine blades using MEM translational tabs, AIAA-2001-0031.In:
Proceedings of the 39th AIAA/ASME, Reno , NV , USA.

154
Appendix A-Sample
Simulation and Design
Input Files

A-1
%====================
wttype=3;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm=53.3;
pitch=-1.2;
flap_s=-20;flap_i=.1;flap_e=20;
rflap1=.6;rflap2=.65;
r_telescop=1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;
Figure A1-Sample simulation file for constant speed rotor with flap control

%====================
wttype=3;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm_s=30;rpm_i=.1;rpm_e=65;
pitch=-1.2;
flap_s=-20;flap_i=.1;flap_e=20;
rflap1=.6;rflap2=.65;
r_telescop=1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A2-Sample simulation file for variable speed rotor with flap control

A-2
%====================
wttype=4;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm=53.3;
pitch=-1.2;
flap=0;rflap1=0;rflap2=0;
rtelescop1=.7;rtelescop_i=.001;rtelescop2=1.1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A3- Sample simulation file for constant speed rotor with telescopic blades

%====================
wttype=10;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm_s=30;rpm_i=.1;rpm_e=65;
pitch=-1.2;
flap=0;rflap1=0;rflap2=0;
rtelescop1=.7;rtelescop_i=.001;rtelescop2=1.1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A4-Sample simulation file for variable speed rotor with telescopic blades

A-3
%====================
wttype=2;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm=53.3;
pitch_s=-5;pitch_i=0.05;pitch_e=5;
flap=0;rflap1=0;rflap2=0;
r_telescop=1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A5- Sample simulation file for constant speed rotor with pitch control

%====================
wttype=8;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm_s=30;rpm_i=.1;rpm_e=65;
pitch_s=-5;pitch_i=0.05;pitch_e=5;
flap=0;rflap1=0;rflap2=0;
r_telescop=1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A6-Sample simulation file for variable speed rotor with pitch control

A-4
%====================
wttype=1;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm=53.3;
pitch=-1.2;
flap=0;rflap1=0;rflap2=0;
r_telescop=1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A7- Sample simulation file for stall regulated constant speed rotor

%====================
wttype=5;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm=53.3;
pitch=-1.2;
flap=0;rflap1=0;rflap2=0;
r_telescop=1;
rmt1=.6;rmt2=.9;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

Figure A8-Sample simulation file for constant speed rotor with microtab

A-5
%====================
wttype=10;
%====================
vw_s=5;
vw_i=1;
vw_e=25;
%====================
rpm_s=30;rpm_i=.1;rpm_e=65;
pitch=-1.2;
flap=0;rflap1=0;rflap2=0;
rtelescop1=.7;rtelescop_i=.001;rtelescop2=1.1;
rmt1=0;rmt2=0;
%====================
yaw=0;
b=2;
d=27.514;
cone=7;
h=30;
prated=300000;
tolprated=.01;
z0=0.025;
vav=5.7;
dens=1.225;
ip=5;
%====================
sim_results_1_par='vw,rpm,pitch,tsr,p,cp,thrust,mroot_maxvw,beta(nseg),flap
,r_telescop,{azimuth_mroot}';
sim_results_2_par='pav,pmax,cpmax,thrustmax,mrootmax';

nseg=20;nsec=1;rx=0.5;

x_par={'pretwist'};
x_type=[1]; %1-3: distributed real/integer/indexed, 4-6:singular
real/integer/indexed
x_npoint=[5];
x_limit_batch={[-3,10]};
x_dist_limit={[rhub_r,1]};
x_co=[1]; % fraction of total CO in the form of arithmetic/integer
arithmatic CO (the rest geometric/exchange)
x_pattern=[3]; %1:random, 2:increasing, 3:decresing,4:concave, 5:convex
%------------------------
parent_select=.5; %fraction of total selection randomly (the rest roulette-
wheel selection)
inipop_dev=0; %fraction of total initial population generated based on
deviation from baseline (the rest generated randomly)
%------------------------
popsize=20;
popsize_u=20;
pc=0.3;
pm=0.1;
ngen=40;
%------------------------
mutemethod=1;
comethod=1;
fitscalemethod=1;
termmethod=1;
%------------------------

fitness=pav;
const=[pmax,0,305000];

Figure A9-Sample design optimisation file for variable speed rotor with telescopic
blades

A-6
Appendix B-Microtab Data

B-1
 The lift and drag coefficients obtained using CFD analysis with ANSYS package
(Hella, 2012).

Table C1-Lift and drag coefficients for different microtab actuation heights on the lower
surface
S808 Lower TL =95%, S808 Lower TL S808 Lower TL=95%,
S808 Aerofoil
TH=1.1% =95%, TH=2.2% TH=3.3%
 CL CD CL CD CL CD CL CD
0 3.36E-01 1.17E-02 4.40E-01 1.41E-02 6.05E-01 1.70E-02 7.18E-01 2.01E-02
2 5.58E-01 1.29E-02 6.61E-01 1.56E-02 8.34E-01 1.88E-02 9.52E-01 2.27E-02
4 7.59E-01 1.59E-02 8.81E-01 1.76E-02 1.059 2.14E-02 1.1776 2.54E-02
6 9.47E-01 1.92E-02 1.0707 2.02E-02 1.2586 2.42E-02 1.3818 2.88E-02
8 1.1116 0.023195 1.234 2.39E-02 1.435 2.88E-02 1.5522 3.32E-02
10 1.259 0.028135 1.3437 2.83E-02 1.564 3.41E-02 1.688 3.91E-02
12 1.3596 0.035096 1.44 3.47E-02 1.6349 4.13E-02 1.7597 4.68E-02
14 1.3982 0.046187 1.4665 4.39E-02 1.6434 5.21E-02 1.7561 5.87E-02
16 1.3583 0.069745 1.4274 6.21E-02 1.5648 7.53E-02 1.6466 8.61E-02
18 1.13E+00 1.19E-01 1.2187 0.10036 1.2949 1.27E-01

Table C2-Lift and drag coefficients for different microtab actuation heights on the upper
surface
S808 Upper TL =80%, S808 Upper TL=80%, S808 Upper TL =80%,
S808 Aerofoil
TH=1.1% TH=2.2% TH=3.3%
 CL CD CL CD CL CD CL CD
0 3.36E-01 1.17E-02 9.61E-02 2.25E-02 -8.09E-02 3.00E-02 -2.16E-01 3.71E-02
2 5.58E-01 1.29E-02 3.19E-01 2.32E-02 1.52E-01 3.05E-02 1.96E-02 3.76E-02
4 7.59E-01 1.59E-02 5.40E-01 2.45E-02 3.81E-01 3.15E-02 2.53E-01 3.85E-02
6 9.47E-01 1.92E-02 7.50E-01 2.63E-02 6.03E-01 3.31E-02 4.79E-01 3.97E-02
8 1.1116 0.023195 9.48E-01 2.89E-02 8.15E-01 3.52E-02 6.98E-01 4.16E-02
10 1.259 0.028135 1.1263 3.23E-02 1.012 3.81E-02 9.04E-01 4.41E-02
12 1.3596 0.035096 1.281 3.69E-02 1.1883 4.20E-02 1.127 4.85E-02
14 1.3982 0.046187 1.399 4.38E-02 1.334 4.75E-02 1.2605 5.19E-02
16 1.3583 0.069745 1.4288 5.99E-02 1.4352 5.93E-02 1.4106 6.03E-02

B-2
Table C3-Changes in lift and drag coefficients for different microtab actuation heights
on the lower surface
S808 Lower TL =95%, S808 Lower TL =95%, S808 Lower TL =95%,
TH=1.1% TH=2.2% TH=3.3%
 C L C D C L C D C L C D
0 1.04E-01 2.41E-03 2.69E-01 5.31E-03 3.82E-01 8.40E-03
2 1.03E-01 2.66E-03 2.76E-01 5.91E-03 3.95E-01 9.77E-03
4 1.22E-01 1.67E-03 3.00E-01 5.48E-03 4.19E-01 9.49E-03
6 1.24E-01 1.01E-03 3.12E-01 4.94E-03 4.35E-01 9.59E-03
8 1.22E-01 7.05E-04 3.23E-01 5.63E-03 4.41E-01 9.97E-03
10 8.47E-02 1.76E-04 3.05E-01 5.99E-03 4.29E-01 1.09E-02
12 8.04E-02 -3.98E-04 2.75E-01 6.20E-03 4.00E-01 1.17E-02
14 6.83E-02 -2.29E-03 2.45E-01 5.90E-03 3.58E-01 1.25E-02
16 6.91E-02 -7.68E-03 2.07E-01 5.58E-03 2.88E-01 1.64E-02
18 9.16E-02 -1.90E-02 1.68E-01 7.17E-03 -1.19E-01

Table C4-Changes in lift and drag coefficients for different microtab actuation heights
on the upper surface
S808 Upper TL =80%, S808 Upper TL =80%, S808 Upper TL =80%,
TH=1.1% TH=2.2% TH=3.3%
 C L C D C L C D C L C D
0 -2.40E-01 1.08E-02 -4.17E-01 1.83E-02 -5.52E-01 2.54E-02
2 -2.39E-01 1.03E-02 -4.06E-01 1.76E-02 -5.38E-01 2.47E-02
4 -2.19E-01 8.57E-03 -3.78E-01 1.56E-02 -5.06E-01 2.26E-02
6 -1.96E-01 7.09E-03 -3.44E-01 1.38E-02 -4.68E-01 2.05E-02
8 -1.64E-01 5.68E-03 -2.97E-01 1.20E-02 -4.13E-01 1.84E-02
10 -1.33E-01 4.14E-03 -2.47E-01 9.96E-03 -3.55E-01 1.59E-02
12 -7.86E-02 1.82E-03 -1.71E-01 6.86E-03 -2.33E-01 1.34E-02
14 8.00E-04 -2.40E-03 -6.42E-02 1.32E-03 -1.38E-01 5.76E-03
16 7.05E-02 -9.86E-03 7.69E-02 -1.05E-02 5.23E-02 -9.44E-03

B-3
Appendix C-Design
Optimisation Results

C-1
 10
9
8
Baseline
7

Pretwist (degrees)
6 Optimised (Flap 60-65%)
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)
(a)
350000
300000
Rotor Power (W)

250000
200000
150000
Baseline
100000
Flap 60-65%
50000
Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Power Coefficient (-)

0.40
Flap 60-65%
0.35
0.30 Flap 60-65%-Optimised
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
180000
160000
Max Flap Bending Moment

140000
120000
100000
80000
(Nm)

60000 Baseline
40000 Flap 60-65%
20000 Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C1-Flap 60-65%- constant speed
C-2
 9
8 Baseline
7 Optimised (Flap 65-70%)

Pretwist (degrees)
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 65-70%
Flap 65-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 65-70%
0.40 Flap 65-70%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
200000
180000
Max Flap Bending Moment

160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
20000 Flap 65-70%
Flap 65-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C2-Flap 65-70%- constant speed
C-3
 10
9 Baseline
8 Optimised (Flap 70-75%)

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
50000 Flap 70-75%
Flap 70-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
0.40 Flap 70-75%
Flap 70-75%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000 Baseline
40000 Flap 70-75%
20000 Flap 70-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C3-Flap 70-75%- constant speed
C-4
 10
9 Baseline
8 Optimised (Flap 75-80%)

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
50000 Flap 75-80%
Flap 75-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 75-80%
0.40 Flap 75-80%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
Flap 75-80%
20000 Flap 75-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C4-Flap 75-80%- constant speed
C-5
 8
Baseline
7
Optimised (Flap 80-85%)
6

Pretwist (degrees)
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Power Coefficient (-)

200000
150000
100000 Baseline
50000 Flap 80-85%
Flap 80-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 80-85%
0.40 Flap 80-85%-Optimised
Rotor Power (W)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
Flap 80-85%
20000
Flap 80-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C5-Flap 80-85%- constant speed
C-6
 9
8 Baseline
7 Optimised (Flap 85-90%)

Pretwist (degrees)
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
50000 Flap 85-90%
Flap 85-90%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 85-90%
0.40 Flap 85-90%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
180000
Max Flap Bending Moment

160000
140000
120000
100000
(Nm)

80000
60000
40000 Baseline
Flap 85-90%
20000 Flap 85-90%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C6-Flap 85-90%- constant speed
C-7
 9
Baseline
8
7 Optimised (Flap 90-95%)

Pretwist (degrees)
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 90-95%
Flap 90-95%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 90-95%
0.40 Flap 90-95%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
Flap 90-95%
20000 Flap 90-95%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C7-Flap 90-95%- constant speed
C-8
 10
9 Baseline
8

Pretwist (degrees)
7 Optimised (Flap 60-70%)
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R)
(a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 60-70%
Flap 60-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 60-70%
0.40 Flap 60-70%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
Flap 60-70%
20000 Flap 60-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C8-Flap 60-70%- constant speed
C-9
 10
9 Baseline
8
Optimised (Flap 60-75%)

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 60-75%
Flap 60-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 60-75%
0.40
Flap 60-75%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000 Baseline
40000 Flap 60-75%
20000 Flap 60-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C9-Flap 60-75%- constant speed
C-10
 10
9 Baseline
8 Optimised (Flap 60-80%)

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
Flap 60-80%
50000
Flap 60-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Power Coefficient (-)

Flap 60-80%
0.40 Flap 60-80%-Optimised
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
Baseline
40000 Flap 60-80%
20000 Flap 60-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C10-Flap 60-80%- constant speed
C-11
 10
9 Baseline
8 Optimised (Flap 60-85%)

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 60-85%
Flap 60-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 60-85%
0.40 Flap 60-85%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000
Baseline
50000 Flap 60-85%
Flap 60-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C11-Flap 60-85%- constant speed
C-12
 10
9
8
Baseline

Pretwist (degrees)
7
6 Optimised (Flap 60-65%)
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 60-65%
Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 60-65%
0.40 Flap 60-65%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
200000
180000
Max Flap Bending Moment

160000
140000
120000
100000
(Nm)

80000
60000
Baseline
40000 Flap 60-65%
20000 Flap 60-65%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C12-Flap 60-65%- variable speed
C-13
 10
9 Baseline
Optimised (Flap 65-70%)
8
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 65-70%
Flap 65-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 65-70%
0.40 Flap 65-70%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
200000
180000
Max Flap Bending Moment

160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
20000 Flap 65-70%
Flap 65-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C13-Flap 60-70%- variable speed
C-14
 10
9 Baseline
Optimised (Flap 70-75%)
8
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 70-75%
Flap 70-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
0.40 Flap 70-75%
Flap 70-75%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
Baseline
40000
Flap 70-75%
20000 Flap 70-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C14-Flap 70-75%- variable speed
C-15
 10
Baseline
9 Optimised (Flap 75-80%)
8
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
50000 Flap 75-80%
Flap 75-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 75-80%
0.40 Flap 75-80%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000

50000 Baseline
Flap 75-80%
Flap 75-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C15-Flap 75-80%- variable speed
C-16
 10
9 Baseline
8 Optimised (Flap 80-85%)
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
50000 Flap 80-85%
Flap 80-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 80-85%
0.40 Flap 80-85%-Optimised
0.35
Power Coefficient (-)

0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000

50000 Baseline
Flap 80-85%
Flap 80-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C16-Flap 80-85%- variable speed
C-17
 10 Baseline
9 Optimised (Flap 85-90%)
8

Pretwist (degrees)
7
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1

Normalised Span Location (r/R) (a)

350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 85-90%
Flap 85-90%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 85-90%
0.40 Flap 85-90%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000
Baseline
50000 Flap 85-90%
Flap 85-90%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C17-Flap 85-90%- variable speed
C-18
 10 Baseline
9 Optimised (Flap 90-95%)
8
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 90-95%
Flap 90-95%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 90-95%
0.40 Flap 90-95%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000

50000 Baseline
Flap 90-95%
Flap 90-95%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C18-Flap 90-95%- variable speed
C-19
 10
9 Baseline
8 Optimised (Flap 60-70%)
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 60-70%
Flap 60-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 60-70%
0.40 Flap 60-70%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
40000 Baseline
Flap 60-70%
20000 Flap 60-70%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C19-Flap 60-70%- variable speed
C-20
 10
9 Baseline
8 Optimised (Flap 60-75%)
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000
Baseline
50000 Flap 60-75%
Flap 60-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Flap 60-75%
0.40
Flap 60-75%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000
Baseline
50000 Flap 60-75%
Flap 60-75%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C20-Flap 60-75%- variable speed
C-21
 10
9 Baseline
8
7 Optimised (Flap 60-80%)

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
250000
Rotor Power (W)

200000
150000
100000 Baseline
Flap 60-80%
50000
Flap 60-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(b)
0.50
0.45 Baseline
Power Coefficient (-)

Flap 60-80%
0.40 Flap 60-80%-Optimised
0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s)
(c)
200000
Max Flap Bending Moment

180000
160000
140000
120000
(Nm)

100000
80000
60000
Baseline
40000 Flap 60-80%
20000 Flap 60-80%-Optimised
0
5 10 15 20 25
Wind Speed (m/s)
(d)
Figure C21-Flap 60-80%- variable speed
C-22
 10
9 Baseline
8 Optimised (Flap 60-85%)
7

Pretwist (degrees)
6
5
4
3
2
1
0
-1
-2
-3
0 0.2 0.4 0.6 0.8 1
Normalised Span Location (r/R) (a)
350000
300000
Rotor Power (W)

250000
200000
150000
100000
Baseline
50000 Flap 60-85%
Flap 60-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (b)
0.50
0.45 Baseline
Flap 60-85%
0.40 Flap 60-85%-Optimised
Power Coefficient (-)

0.35
0.30
0.25
0.20
0.15
0.10
0.05
0.00
5 10 15 20 25
Wind Speed (m/s) (c)
250000
Max Flap Bending Moment

200000

150000
(Nm)

100000
Baseline
50000 Flap 60-85%
Flap 60-85%-Optimised
0
5 10 15 20 25
Wind Speed (m/s) (d)
Figure C22-Flap 60-85%- variable speed
C-23