Dokument 5

Passive and active flow control solutions for wind
turbine blades
vorgelegt von
M.Sc.
Georgios Pechlivanoglou
aus Athen (Griechenland)
von der Fakultät V - Verkehrs- und Maschinensysteme

der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften

- Dr.-Ing. -
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr.-Ing. W. Nitsche
Berichter: Prof. Dr.-Ing. C.O. Paschereit
Berichter: Dr.-Ing. C.N. Nayeri
Tag der wissenschaftliche Aussprache: 21. Februar 2012
Berlin 2013
D 83
Preface
This document gives a detailed summary of all the research efforts and major results
during the course of my PhD dissertation. The purpose of the research work entitled
“Smart Rotorblades For The Wind Turbines Of The Future” was to investigate the
possibilities of Aerodynamic Flow Control on wind turbine blades. Since this topic is
very innovative and relatively new to the wind energy world, the current work functions
also as a very detailed overview of the current state of research.
The reader is initially acquainted with the basic theory of airfoil flows and horizon-
tal axis wind turbine aerodynamics. After the basic theory follows the description of
the simulation techniques used during this research. The simulation codes used for
airfoil and wind turbine analysis are explained in further detail and the specific mod-
els used within them are briefly discussed. Along with the aforementioned theoretical
background information the reader is provided with extensive information on the ex-
perimental setups used during the current work. This section is aimed as a guide for
future relevant research as well as an educational part.
The basic theory is followed by a very extensive literature research and preliminary
investigation section. In this section several flow control solutions are examined and
rated in various categories. This section offers a detailed and rational way of categoriz-
ing the performance of the flow control solutions. It is aimed as a good starting point
for any further attempt in the same general field.
The last part of the thesis includes the presentation of the detailed experimental and
numerical investigations performed by the author. These investigations involve several
flow control solutions and are aimed at the investigation of their effect on wind turbine
performance and dynamics. The result of these investigations leads to the definition
of a “smart rotorblade” proposal. This rotorblade forms a general proposal regarding
Preface III
the way future rotorblades could be designed and function. It is by no means the only
possible solution to the problem rather a feasible approach.
It is the hope of the author that the current thesis will be a useful guide for students,
researchers and practicing wind turbine engineers who want to investigate flow control
solutions for wind turbines.
Georgios Pechlivanoglou
13.12.2012
Abstract
The current thesis describes the research work of the author in the field of wind turbine
aerodynamics and especially passive and active flow control solutions for wind turbine
applications. The basic theory description is accompanied by a thorough literature
review where many different flow control solutions are briefly analyzed. Special efforts
have been made to include a wide range of technical fields, apart from aerodynamic
performance, in the analysis process and thus try to provide a comprehensive study.
The best performing solutions are then thoroughly analyzed by means of flow simula-
tions (CFD), wind tunnel tests as well as aeroelastic simulations. Through this analysis
it is possible to estimate the merits of such solutions for wind turbine applications. A
final chapter with a fictive wind turbine "smart" rotorbade, equipped with flow con-
trol solutions is finally presented and discussed in this thesis. The aim of the author
is that this work will be able to function as a base for further research of other flow
control solutions...as well as for the development of new flow control concepts, which
will overcome current drawbacks.
Zusammenfassung
Die vorliegende Arbeit beschreibt die Forschungsaktivitäten des Autors auf dem Feld
der Windkraftanlagen-Aerodynamik, mit besonderem Augenmerk auf der passiven und
aktiven Strömungsbeeinflussung. Die Beschreibung der theoretischen Grundlagen wird
von einer detaillierten Literaturrecherche begleitet, in welcher viele verschiedene Meth-
oden zur Strömungsbeeinflussung analysiert werden. Dabei wird besonders darauf
geachtet, neben den aerodynamischen Aspekten auch technische Gesichtspunkte zu
berücksichtigen, um eine aussagekräftige Vergleichbarkeit der Ansätze zu erhalten. Die
erfolgversprechendsten Lösungen werden im Anschluss mittels Strömungssimulationen
(CFD), Windkanaltests und aeroelastischen Simulationen genauer untersucht. Mit
Hilfe dieser Untersuchungen ist es möglich den Nutzen strömungsbeeinflussender Maß-
nahmen für den Einsatz an Windkraftanlagen abzuschätzen. Abschließend wird im
letzten Kapitel ein fiktives, mit Elementen zur Strömungsbeeinflussung ausgestanztes
"smartes" Rotorblatt vorgestellt und diskutiert. Das Ziel des Autors ist es, dass diese
Arbeit als Grundlage für weitergehende Forschungsarbeiten an bestehenden Elementen
zur Strömungsbeeinflussung dient und Anstöße für die Entwicklung neuer Konzepte
liefert, welche die Nachteile der jetzigen Lösungen überwinden.
This work is dedicated to Petros, Telemachos and Myrto.
Without you, life would be without happiness!
Nomenclature
AoA Angle of attack
W incident velocity
cl lift coefficient
cd drag coefficient
cm moment coefficient
U∞ inflow velocity
r local blade element radius
a0 tangential induction factor
a axial induction factor
ve,op in-plane velocity due to structural deflection
ve,ip out-of-plane velocity due to structural deflection
D blade element drag force
L blade element lift force
pn load normal to rotor plane
pt load tangential to rotor plane
BEM blade element momentum theory
GDW generalized dynamic wake model
c chord length
B number of blades
T rotor thrust
Q rotor torque
F force, Prandtl’s tip-loss factor
Ct rotor thrust coefficient
DF N blade element normal force
DF T blade element tangential force
P MA blade element pitching moment
α angle of attack
Nomenclature VIII
Ω blade angular velocity

φ inflow angle
θ local pitch angle
θp blade pitch angle
β local blade twist
ρ air density
γ yaw angle
χ wake skew angle
ψ blade azimuthal angle
Contents
Preface II
Abstract IV
Zusammenfassung V
VI
Nomenclature VII
I Overview 1
1 Introduction 2
1.1 Wind Turbine State of the Art . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Current Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Thesis Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Research Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
II Basic Theory 8
2 Airfoil Aerodynamics 9
2.1 Airfoil Design Parameters . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Airfoil Performance Characteristics . . . . . . . . . . . . . . . . . . . . 11
3 HAWT Aerodynamics 14
3.1 Rotating Blade Aerodynamics . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Unsteady Inflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Aeroelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Contents X
4 Airfoil Simulations 20
4.1 Panel Method Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2 Finite Volume Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.1 Grid Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2.2 Flow Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5 Wind Turbine Simulations 26

5.1 Steady State HAWT Simulation . . . . . . . . . . . . . . . . . . . . . . 26
5.1.1 Prandt’s Tip and Root Loss Correction . . . . . . . . . . . . . . 28
5.1.2 Glauert Correction . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.1.3 Snel’s correction . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.1.4 360 Degree polar curve extrapolation . . . . . . . . . . . . . . . 30
5.2 Dynamic HAWT Simulation . . . . . . . . . . . . . . . . . . . . . . . . 31
5.3 Dynamic Stall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.4 Inflow Turbulence Modeling . . . . . . . . . . . . . . . . . . . . . . . . 36
5.5 Generalized Dynamic Wake Modeling . . . . . . . . . . . . . . . . . . . 37
5.6 Wind Turbine Structural Model . . . . . . . . . . . . . . . . . . . . . . 38
5.7 BEM with Active Elements . . . . . . . . . . . . . . . . . . . . . . . . 40
III Experimental Setup 42
6 Wind Tunnel Experiments 43

6.1 Wind Tunnel Description . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.1.1 Static Measurements . . . . . . . . . . . . . . . . . . . . . . . . 47
6.1.2 Dynamic Measurements . . . . . . . . . . . . . . . . . . . . . . 48
IV Initial Flow Control Solution Selection 51
7 Methodology 52
8 Passive Flow Control 55

8.1 Vortex Generators and Vortilons . . . . . . . . . . . . . . . . . . . . . . 55
8.2 Fixed Leading Edge Slat . . . . . . . . . . . . . . . . . . . . . . . . . . 58
8.3 Flow Vane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Contents XI
8.4 L.E. Protuberances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
9 Active Flow Control 68

9.1 Gurney Flap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
9.2 Rigid T.E. Flap (Aileron) . . . . . . . . . . . . . . . . . . . . . . . . . 71
9.3 Flexible T.E. Flap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
9.4 Flexible L.E. Flap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
9.5 Split Flaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
9.6 Stall Ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
9.7 Spoilers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
9.7.1 Inclined Spoiler . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
9.7.2 Vertical Spoiler . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
9.8 BL Suction and Blowing . . . . . . . . . . . . . . . . . . . . . . . . . . 93
9.8.1 Blow-Type Flow Control . . . . . . . . . . . . . . . . . . . . . . 93
9.8.2 Suction-Type Flow Control . . . . . . . . . . . . . . . . . . . . 97
9.9 Zero Mass Flux and Synthetic Jet Actuators . . . . . . . . . . . . . . . 101
9.10 Plasma Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
V Experimental and Numerical Investigations 112
10 Passive Flow Control Solutions 113

10.1 Vortex Generators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
10.1.1 Wind Tunnel Measurements . . . . . . . . . . . . . . . . . . . . 114
10.1.2 Wind Turbine Simulations . . . . . . . . . . . . . . . . . . . . . 114
10.2 Fixed Leading Edge Slat . . . . . . . . . . . . . . . . . . . . . . . . . . 118
10.2.1 CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 118
10.3 Flow Vane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Contents XII
11 Active Flow Control Solutions 135

11.1 Flexible Trailing Edge Flap . . . . . . . . . . . . . . . . . . . . . . . . 135
11.1.1 CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 135
11.1.3 Wind Turbine Aeroelastic Simulations . . . . . . . . . . . . . . 141
11.1.3.1 General Simulation Description . . . . . . . . . . . . . 141
11.1.3.2 Parametric Study . . . . . . . . . . . . . . . . . . . . . 146
11.1.3.3 AFC element controller performance . . . . . . . . . . 149
11.2 Flexible Leading Edge Flap . . . . . . . . . . . . . . . . . . . . . . . . 154
11.2.1 CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 154
11.3 Active Gurney Flap (Mini-Flap) . . . . . . . . . . . . . . . . . . . . . . 160
11.3.1 CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 160
11.3.2 Wind Tunnel Measurements (Steady State) . . . . . . . . . . . 161
11.3.3 Wind Tunnel Measurements (Unsteady) . . . . . . . . . . . . . 161
11.4 Stall Ribs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
11.4.1 CFD Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 168
11.5 Plasma Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.5.1.1 Experimental Setup . . . . . . . . . . . . . . . . . . . 176
11.5.1.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . 181
VI Blade Design 185
12 ”Smart Blade” Design Characteristics 186

12.1 AFC and PFC Solution Selection . . . . . . . . . . . . . . . . . . . . . 186
12.2 Actuator Mechanisms, Blade Integration and Control . . . . . . . . . . 189
12.2.1 Actuators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
12.2.2 Blade Integration . . . . . . . . . . . . . . . . . . . . . . . . . . 189
12.2.3 Control and Sensors . . . . . . . . . . . . . . . . . . . . . . . . 190
Contents XIII
VII Thesis Conclusions 193
13 Discussion On The Overall Conclusions 194

13.1 Research Outcome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
13.1.1 Experimental Equipment . . . . . . . . . . . . . . . . . . . . . . 194
13.1.2 Software Development . . . . . . . . . . . . . . . . . . . . . . . 194
13.2 Scientific Publications . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
13.3 Relevant Patents and Patent Applications . . . . . . . . . . . . . . . . 196
13.4 Feasibility of ”Smart Blades” . . . . . . . . . . . . . . . . . . . . . . . . 197
13.5 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
13.5.1 Airfoil Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
13.5.2 Dynamic Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
13.5.3 Aeroelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
14 Acknowledgments 200
List of Figures
1.1 Cut view of a modern HAWT with brief description of its main compo-
nents [152]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 Basic airfoil design and definition parameters [73]. . . . . . . . . . . . 9

2.2 Polar curves for NACA63XX airfoil series of various airfoil thickness
levels. [26]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Pressure distribution along the chord (Cp / xy ) and Normalized velocity
Q
distribution along the chord ( Vinf / xy ) for an AH93W174 airfoil at 9o AoA 13
3.1 Flow velocity vectors acting on a wind turbine rotor section. . . . . . . 14

3.2 Distribution of aerodynamic forces along the rotor blades. . . . . . . . 15
3.3 Crossflow pattern on a wind turbine blade. Visualization based on the
work of Soerensen at al. [148] . . . . . . . . . . . . . . . . . . . . . . . 17
3.4 Schematic representation of a non uniform inflow field. Such stochastic
turbulent fields are typically met by wind turbine rotors [Source:Wikipedia.org].
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.1 Comparison of the measured performance polar curve of DU 97-W-300

airfoil and the computed curves with CFD (OpenFOAMTM and XFOIL).
The experimental polar was measured at the large wind tunnel of H.F.I
TU Berlin. The CFD curves present simulation results with various
turbulence models and grid densities. . . . . . . . . . . . . . . . . . . 24
4.2 Comparison of the measured performance polar curve of DU 97-W-300
airfoil and the computed curves with RFOIL [164]. . . . . . . . . . . . 25
List of Figures XV
5.1 Blade representation with independent airfoil segments according to the

blade element theory. The velocity vectors are computed for each seg-
ment through the blade element momentum method. The AoA along the
blade changes due to blade rotation. . . . . . . . . . . . . . . . . . . . 27
5.2 Extrapolated 360o Cl curve of the DU 00-W-350 Mod2 wind turbine
airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.3 Extrapolated 360o Cd curve of the DU 00-W-350 Mod2 wind turbine
airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.4 Comparative graph of a static behavior of an airfoil (static polar) with
the dynamic behavior of the same airfoil (dynamic stall loops) [118]. . 34
5.5 Wind turbine and coordinate system representation for YawDyn [99]. . 39
5.6 Rigid blade representation for YawDyn with ideal spring-hinge. [99]. . 40
6.1 View of the nozzle and test section of the HFI/TU Berlin Wind Tunnel. 43
6.2 Detailed diagram of the HFI/TU Berlin Wind Tunnel measurement sec-
tion [114]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
6.3 The airfoil sections of the four test wings used for wind tunnel measure-
ments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6.4 The 6-component force balance used for the current wind tunnel cam-
paign [114]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.5 Basic schematic of the multi-core software implementation. . . . . . . . 49
7.1 Schematic representations of all the elements included in the preliminary

flow control investigation. . . . . . . . . . . . . . . . . . . . . . . . . . 53
8.1 Vortilon and Vortex Generator configuration . . . . . . . . . . . . . . . 55

8.2 Vortilons shape and location according to Shevell [58] . . . . . . . . . . 57
8.3 Leading edge slat. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
8.4 Different combinations of flapped and un-flapped slotted airfoil studied
by Weick and Shortal [5, 177]. . . . . . . . . . . . . . . . . . . . . . . . 59
8.5 Schematic representation of the flow vane concept. . . . . . . . . . . . . 61
8.6 Leading Edge protuberances . . . . . . . . . . . . . . . . . . . . . . . . 63
8.7 The proposal of DLR involves small spheres attached on the leading edge
[83] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
List of Figures XVI
8.8 Sinusoidal leading edge protuberances based on the Humpback Whale

flipper [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
8.9 The sinusoidal protuberances create 3D vortical flow phenomena which
re-energize the boundary layer at the pressure recovery region this de-
laying stall [63]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
9.1 Active Gurney Flap (or Micro Tab) for load alleviation . . . . . . . . . 68
9.2 Schematic representation of the trailing edge flow field of a typical trail-
ing edge configuration (top) and a trailing edge equipped with a Gurney
Flap (bottom) [120]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
9.3 Plain rigid flap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
9.4 Flexible flap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
9.5 Comparison between the pressure distribution of an airfoil with a de-
flected rigid flap and a deflected flexible flap [146]. . . . . . . . . . . . 75
9.6 Wind tunnel test wing equipped with flexible flap tested at the HFI/TU
Berlin wind tunnel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
9.7 Flexible Leading Edge. . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
9.8 Simple Split Flap configuration (top) and Double Split Flap (SplitF lap−
Aileron) configuration (Bottom) . . . . . . . . . . . . . . . . . . . . . . 80
9.9 Various operational configurations of the double split flap mechanism [24]. 81
9.10 Leading Edge Elastic Stall Rib. . . . . . . . . . . . . . . . . . . . . . . 83
9.11 Elastic Stall Rib located at the suction side of an aircraft wind, operating
as lateral control element [13]. . . . . . . . . . . . . . . . . . . . . . . . 83
9.12 Active Inclined Spoiler . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
9.13 Wind tunnel investigation results with comparative results between the
“clean” DU96W180 profile and the profile with inclined spoiler located
at 50%c and deflected to various angles [140]. . . . . . . . . . . . . . . 88
9.14 Vertical Spoiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
9.15 Vertical spoiler (dive brake) configuration proposed by H. Wagner in
1939 [173]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
9.16 Vertical spoiler configuration implemented on sailplane wing [162]. . . . 91
9.17 Tangential Leading Edge Blowing . . . . . . . . . . . . . . . . . . . . . 93
9.18 Velocity distribution in the boundary layer directly behind the slit for
tangential blowing (A) and right after the blowing point (B) [142]. . . . 94
List of Figures XVII
9.19 Schematic comparison of the conventional airfoil performance (left) and

the performance of the counter-flow concept (right) [16]. . . . . . . . . 95
9.20 One of the first test applications of the suction-type flow control was
realized by NACA at the wings of this Lockheed P-80 aircraft [21]. . . . 97
9.21 The basic aerodynamic principle of the suction-type flow control. The
“exhausted” boundary layer is removed through the slot thus enabling
the formation of a new boundary layer downstream of the slot [79]. . . 98
9.22 Elliptic airfoil with suction-type flow control and flap which defines the
Kutta-Joukowsky condition [79]. . . . . . . . . . . . . . . . . . . . . . . 99
9.23 The mechanical structure of the wing equipped with suction-type flow
control. [21]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9.24 Schematic view of a Synthetic Jet equipped wing . . . . . . . . . . . . 101
9.25 Section view of an airfoil section with an oscillating membrane based
synthetic jet mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . 103
9.26 Schematic of the electro-magnetic Synthetic Jet actuator of Renault as
it was used at the prototype/exhibition vehicle Renault Altica [85]. . . 104
9.27 Schematic of the passive Synthetic Jet actuator of GE Wind [27]. . . . 104
9.28 Wing with Plasma Actuator. . . . . . . . . . . . . . . . . . . . . . . . . 106
9.29 Schematics and images of the different plasma actuator types. a)
Schematic of a DC corona discharge actuator, b) Visualization of the
effect of the actuator (a), c) AC Barrier discharge actuator, d) Visu-
alization of the effect of the actuator (c), e) Three electrode discharge
actuator, f) Visualization of the effect of the actuator (e), g) Wall jet
actuator with balanced electrode voltage, h) Wall jet actuator with im-
balanced electrode voltage[Source: [167]]. . . . . . . . . . . . . . . . . 108
10.1 Comparison between the baseline configuration and the various VG con-
figurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
10.2 Simulation results presenting the distribution of the Cl and the AoA
along the baseline blade and the blade with vortex generators. . . . . . 116
10.3 View of the unstructured grid used during the CFD simulations of the
fixed slat configurations [129]. . . . . . . . . . . . . . . . . . . . . . . . 119
10.4 Pressure contour of the DU97W300 airfoil with the NACA 22 fixed
slat[129] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
List of Figures XVIII
10.5 Comparative view of the “baseline” configuration simulation in compar-

ison to the configuration with a fixed leading edge slat[129] . . . . . . . 120
10.6 Curve of the lift coefficient for a configuration with a slat. Comparison
between experimental and numerical results[129] . . . . . . . . . . . . . 121
10.7 Wind tunnel test configuration comprising the DU97W300 main airfoil
and the NACA 22 fixed leading edge slat[129] . . . . . . . . . . . . . . . 122
10.8 Schematic representation of the slat position with respect to the DU97W300
main wings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
10.9 Comparison between the baseline configuration and the slat configuration[129].124
10.10Comparison between the rotor Cp curve of the baseline configuration
and the slat configuration for the DU97W300 airfoil. (Simulation results)125
10.11Comparative AoA and Cl distributions (simulation results) across a ro-
torblade with and without a fixed leading edge slat. . . . . . . . . . . . 126
10.12Comparative Cl and Cd curves for the DU96W180 and the DU97W300
measured at the wind tunnel of HFI TU Berlin. . . . . . . . . . . . . . 128
10.13Schematic representation of the flow vane positions over the DU96W180
and DU97W300 main wings. . . . . . . . . . . . . . . . . . . . . . . . . 129
10.14Experimentally measured Cl and CCdl curves for the DU96W180 test wing
equipped with the flow vane. . . . . . . . . . . . . . . . . . . . . . . . . 130
10.15Comparative Cl and Cl /Cd curves for the DU97W300 measured at the
wind tunnel of HFI TU Berlin. . . . . . . . . . . . . . . . . . . . . . . . 131
10.16Simulated power curves for the test turbine with and without a Flow
Vane at the blade root region. . . . . . . . . . . . . . . . . . . . . . . . 133
11.1 CFD simulation (RANS) of the reference airfoil at 5o AoA. The flow is
fully attached. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
11.2 CFD simulation of positive flap deflection at 5o AoA. The local separa-
tion at the suction side of the flexible flap is apparent. . . . . . . . . . 137
11.3 CFD simulation of negative flap deflection at 20o AoA. The suction side
of the flexible flap is not effective anymore since it is inside the separated
flow region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
11.4 The wind tunnel test wing with the flexible flap at full deflection position
[131]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
11.5 Cl and Cd curves for the reference wing (DU96W180) and the wing with
the flexible flap trailing edge section, measured in the wind tunnel[131]. 139
List of Figures XIX
11.6 Schematic representation of the DU96W180 airfoil and comparison with

the flexible flap configuration. Shape deviations exist but are kept to a
minimum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
11.7 Experimentally measured Cl and Cd curves of the wing with fully de-
flected flexible flap (positive deflection), with and without VGs[131]. . . 140
11.8 Experimentally measured Cl and Cd curves of the wing with fully de-
flected flexible flap (positive deflection), with and without VGs[131]. . . 140
11.9 Results of the baseline aeroelastic simulation with QBladeAE [179]. . . 143
11.10Turbulent wind speed (x-component) time series at 89m (hub height)
produced with TurbSim[94] and used for the aeroelastic simulations [179].144
11.11Extrapolated “composite” 360◦ Cl -polar. The polar is based on exper-
imental date and is extended by means of Montgomerie extrapolation
[178]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
11.12Blade with 9 equidistant outer sections equipped with flexible flaps [179]. 146
11.13Load reduction potential of a single flap with a length of 1.5m at different
radial positions along the blade [179]. . . . . . . . . . . . . . . . . . . . 148
11.14Flap-wise bending moment time series. Baseline simulation and the
13.5m long flap [179]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
11.15Load reduction for a 9m flexible flap configuration. Parametric investi-
gation for various flap deflection speeds and flap deflection ranges [179].
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
11.16Flap angle for 9m flap over simulation time [179]. . . . . . . . . . . . . 151
11.17The local blade element force DF N at the station AE# 3 for the baseline
configuration compared with the single PID controlled 13.5m flexible
flap and the multiple individually controlled flaps (optimization loop
controller) [179]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
11.18Root Bending Moment time series comparison (baseline configuration
and Flexible T.E. Flap)[179]. . . . . . . . . . . . . . . . . . . . . . . . . 152
11.19Extremes of the control authority of the Flexible T.E. Flap[179]. . . . . 152
11.20Simulation results of the Flexible T.E. Flap aeroelastic analysis[179]. . 153
11.21Cl curves for the NACA 63(3) 618 and the NACA 4415 airfoil (Re =
1.3 · 106 ) from XFOIL simulations. . . . . . . . . . . . . . . . . . . . . 154
11.22Cl curves for the NACA 63(3) 618 airfoil (Re = 1.3·106 ) from wind tunnel
measurements at the large wind tunnel of TU Berlin. . . . . . . . . . . 156
List of Figures XX
11.23Comparison of the wind tunnel measurements and the XFOIL simulation

results[130]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
11.24Time series of the root bending moment (with and without the Flexible
L.E. Flap) resulting from the aeroelastic simulations[179]. . . . . . . . . 158
11.25Summary of the aeroelastic simulation results of the Flexible L.E. Flap
Simulation. The limited control authority of this device is apparent at
the flap deflection plot[179]. . . . . . . . . . . . . . . . . . . . . . . . . 159
11.26XFOIL simulation results for the DU96W180 airfoil in baseline configu-
ration and equipped with Gurney and wedge-shaped Gurney flaps. . . . 160
11.27Various Gurney flap configurations. a)Standard L-profile, b) Gurney
flap and Splitter Plate, c) Wedge-shaped Gurney flap [130]. . . . . . . . 161
11.28Various Gurney flap configurations tested on a wind tunnel wing based
on the DU96W180 airfoil[130]. . . . . . . . . . . . . . . . . . . . . . . . 162
11.29Gurney flap performance on a wind tunnel wing based on the AH93W174
airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
11.30Gurney flap performance on a wind tunnel wing based on the DU97W300
airfoil. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
11.31Measured wind tunnel data of lift variation due to AoA variation with
(red curve) and without (blue curve) active flow control[130]. . . . . . . 165
11.32Simulation results for the extreme actuation range of the Active Gurney
Flap[179]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
11.33Root bending moment timeseries with and without Active Gurney
Flap[179]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
11.34Simulation results for the extreme actuation range of the Active Gurney
Flap[179]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
11.35Comparative outline of the “clean” DU96W180 profile with similar airfoil
equipped with stall rib located at 33%c. . . . . . . . . . . . . . . . . . 168
11.36Aerodynamic performance graphs for the DU96W180 airfoil with de-
ployed stall rib. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
11.37Aerodynamic performance graphs for the DU96W180 airfoil with and
without a deployed stall rib. . . . . . . . . . . . . . . . . . . . . . . . . 170
11.38Vorticity field for a “clean” NACA 4418 airfoil (top) and for the same
airfoil with deflected stall rib (bottom)[130]. . . . . . . . . . . . . . . . 171
List of Figures XXI
11.39A, B and C type stall bumps. All different shapes were tested on various
positions on the wing[130]. . . . . . . . . . . . . . . . . . . . . . . . . . 171
11.40Lift and drag behavior comparison between the baseline measurement
and A-Type stall ribs in different positions. . . . . . . . . . . . . . . . . 173
11.41Lift and drag behavior comparison between the baseline measurement
and all types of stall ribs mounted at the same chord-wise position[130]. 174
11.42Power curve (left) and bending moment over wind speed (right) com-
parative graph for the turbine with and without stall ribs. . . . . . . . 175
11.43Angle-of-Attack and lift coefficient (Cl ) distribution along the blade span
for 10m/s wind speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
11.44Basic scheme of the experimental setup [50] . . . . . . . . . . . . . . . 177
11.45View of the cylinder model with the plasma actuators in operation [50]. 179
11.46View of the cylinder model with the plasma actuators in operation [51]. 180
11.47The DU 97W300 airfoil geometry and positions of streamwise momen-
tum injection[51]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
11.48∆Cd and ∆Cl values for unsteady actuation at Re=85,000 [51, 50]. . . 181
11.49∆Cd and ∆Cl values for unsteady actuation at Re=136,500 [51, 50]. . . 182
11.50Smoke visualization of the test cylinder with inactive (left) and active
(right) plasma actuators. The effect of the plasma actuator on the flow
is evident [50]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
11.51Unsteady operation of plasma actuators at the DU97W300 test wing [50].183
11.52Smoke visualization of the test wing with inactive (top) and active (bot-
tom) plasma actuators [51]. . . . . . . . . . . . . . . . . . . . . . . . . 184
12.1 Schematic representation of the proposed “Smart Blade”. . . . . . . . . 187

12.2 Schematic representation of the modular flexible flap and spoiler actua-
tion principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
List of Tables
6.1 Table of specifications for the wind tunnel of H.F.I TU Berlin. . . . . . 44

6.2 Specifications of the wind tunnel test wings used for static and dynamic
measurements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
10.1 Blade design parameters and simulation parameters and results. Com-
parison of the performance of the “Baseline” wind turbine blades against
the blades fitted with vortex generators. . . . . . . . . . . . . . . . . . 117
the blades fitted with fixed leading edge slat at the root region. . . . . 123
the blades fitted with a Flow Vane at the root region. . . . . . . . . . . 134
11.1 Specifications of the parameter setup for the numerical simulations of

the flexible trailing edge flap with QFLR5 and OpenFOAMTM . . . . . . 136
11.2 Turbine parameters used for the simulation. . . . . . . . . . . . . . . . 141
11.3 Blade structural parameters used for the simulation. . . . . . . . . . . . 142
11.4 Load reduction for 1.5m flap at different radial positions. . . . . . . . . 147
11.5 Load reduction potential for various flap lengths and different radial
positions. The Active Element numbers represent positions as these
were defined at Fig11.12. . . . . . . . . . . . . . . . . . . . . . . . . . 147
11.6 General test matrix of the stall bump experimental investigations. . . . 172
12.1 The final AFC and PFC solution selection for the proposed “Smart
Blade” design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Part I
Overview
1 Introduction
Wind energy is becoming a significant contributor of the world’s electrical energy gen-
eration systems. Thousands of wind turbines are installed every year around the world
and feed their generated electrical power to local or interconnected electricity grids.
The development of technology, manufacturing processes and economies of scale leads
to the construction of larger and more powerful wind turbines able to harvest very high
amounts of wind power thus significantly reducing the cost per kWh of the generated
electricity. This leads to a higher competitiveness of wind energy against the market
competition of conventional or other renewable energy sources. It is expected that in
the future wind turbines will reach previously un-anticipated levels of installed capac-
ities in the range of 15MW/unit thus reducing even more their electricity generation
costs and being more cost effective than the cheapest conventional sources of energy.
1.1 Wind Turbine State of the Art

Current wind turbines have maximum rotor diameters in the range of 130m. Most of the
modern large wind turbines follow the three bladed horizontal axis concept (HAWT)
with or without a gearbox and a modern generator with partial or full frequency-
conversion. The nacelle of modern multi-megawatt wind turbines weighs more than
300 metric tons and is connected to the tower via a yaw system. This system is respon-
sible for the alignment of the turbine rotor with the incoming wind and it comprises a
roller or gliding bearing structure which allows the rotation of the nacelle against the
tower. The necessary yawing moment is provided by a plurality of powerful electrical
drives located inside the nacelle.
The rotor of the wind turbine comprises the hub and the blades and is connected to
the wind turbine drive-train via the main shaft. The hubs of wind turbine rotors are
very crucial components since they accomplish a multitude of tasks and requirements.
1 Introduction 3
Figure 1.1: Cut view of a modern HAWT with brief description of its main compo-
nents [152].
They are the components on which, wind turbine blades are mounted and therefore
they have to be able to withstand all the aerodynamic, structural and operational loads
that are generated from the nacelle - blade interaction. The blades are mounted on
the hubs in a way that they can be rotated along a pitch axis (i.e. pitching blades).
The rotation is accomplished by the installation of pitch bearings between the hub and
the blades. Each pitch bearing is usually equipped with internal or external gearing
which is in constant mesh with a pinion driven by a powerful pitch motor. In this
way it is possible to actively and controllably pitch each one of the blades of the rotor.
The electrical power for the pitch motors is supplied to the hub by means of a slip-ring
connector located at the nacelle and a series of cable lines which pass through the main
shaft of the wind turbine. Inside the hub there are three switch cabinets, three sets
of frequency converters and three battery back-up systems for off-grid pitching oper-
ations. These triple redundancy systems are required in order to ensure the proper
pitching operation of the turbine under any circumstance.
1.2 Current Problems

The motivation for the research efforts presented at this document has to do with
the limitations and drawbacks of the current wind turbine designs. One of the main
drawbacks of the existing blade design concept is the use of the pitch system for power
1 Introduction 4
and load regulation. Pitch systems are unable to cope with the highly dynamic and
non-uniform inflow which is interacts with the large wind turbine rotors. The blades
are simply too large to assume that a uniform pitch angle variation is sufficient to effec-
tively adjust the system to the varying inflow conditions. Furthermore the enormous
size and weight of modern blades as well as their inherent highly elastic nature put
strict limits on the pitching rate, which is limited to approx. 11◦ /s. The combination of
the full blade pitch with the low pitch angle variation heavily limits the responsiveness
and the agility of current wind turbine pitch systems.
Another motivating factor for the current research is that current wind turbines are
under-actuated systems with the only control input parameters being the wind speed
and wind direction at a single point and the torque and power of the drive train.
The wind-speed and wind-direction measurements are performed behind the rotor by
a single (often two) anemometer and flow vane. The wind speed and wind direction
measurements acquired by the anemometer are therefore “one step behind” the rotor
in terms of control and pitch actuation. Additionally, the highly disturbed flow regime
behind the rotor is not representative of the true incoming flow speed and direction.
To improve the control strategy of wind turbines, torque and power measurements are
also included in the overall control algorithm. However even in this case the system
suffers from several time delays, elastic responses and control lags that prevent the
creation of an accurate and pro-active control system.
The inability of current control systems and wind turbine component designs to
perform optimally leads to adverse operational situations that have negative effects
on the system reliability, component design specifications, robustness e.t.c. The main
practical results of the current wind turbine design and control system limitations can
be summarized as:
1. Yaw misalignment due to flow vane errors
2. Inability to compensate the non-uniform blade inflow
3. Inability to adjust to turbulent inflow
4. High fatigue loads
5. Strong aeroelastic coupling

1 Introduction 5
6. Inability to damp blade and turbine oscillations
7. Response delay of existing control systems
8. Increased electrical and mechanical complexity due to pitch system
9. High actuation energy requirements for pitch systems
10. Reliability issues (e.g. backup batteries, pitch gear lubrication, slip-ring mainte-
nance)
11. Design limitations due to extreme loads (e.g. hub & pitch bearing deformation)
All the aforementioned issues are regarded as “necessary evils” for the current wind
turbine designs. To tackle the most adverse of these problems (e.g. fatigue load
issues), wind turbine manufacturers usually end up in the establishment of high design
safety factors which usually ensure that the structures will not fail under fatigue or
extreme loads. This is only a temporary solution since it leads to higher product
costs thus higher cost per kWh levels. The current work aims at the investigation of
several technical solutions that could potentially solve these problems by introducing
innovative aerodynamic flow control elements on the blades.
1.3 Thesis Objectives

The aim of the current thesis is the structured, methodical and in-depth investigation
of a multitude of aerodynamic solutions in an attempt to identify the best suitable
ones to be implemented on a blade design concept. The goals that need to be achieved
by the proposed wind turbine blade design solution are:
1. High performance
2. Increased reliability
3. Better aerodynamic load management
4. Better power regulation
5. Reduced cost through innovative design concepts

1 Introduction 6
The final wind turbine blade design proposed at the end of this document is one
example of an advanced blade design solution for the next generation of multi-megawatt
wind turbines. It is by no means the only solution in this direction, but it is nevertheless
a good example of the form, functionality and potential of the future “smart blades”.
1.4 Research Methodology

The thesis includes at a very big spectrum of aerodynamic flow control applications
which are considered for use on wind turbine blades. The initial part of the research
focuses on the thorough understanding of wind turbine aerodynamics and the theoret-
ical and experimental investigation of several flow control solutions. An initial element
selection is performed based on extensive literature review, empirical estimations and
simple simulations. The aim of this initial investigation is the selection of the best
performing flow control solutions with regard to a multitude of multi-disciplinary pa-
rameters.
The best performing passive and active flow control solutions are selected for further
detailed investigations. These investigations include steady state CFD simulations of
airfoil sections equipped with flow control solutions, steady state and dynamic wind
tunnel tests and steady and unsteady Blade Element Momentum method (BEM) based
wind turbine simulations. From all the simulations and experiments it is possible to
identify the flow control solutions with the best overall performance. The active flow
control elements are additionally simulated with a custom active unsteady BEM code
coupled with a structural model. In this way, aeroelastic investigations of rotating
systems can be coupled with active flow control simulations.
A great challenge in the process of integration of active flow control solutions to wind
turbine blades is the complexity of the control strategies and algorithms. In order to
get more insight into this issue, dynamic wind tunnel tests were performed. The wind
tunnel test wing was equipped with active flow control solutions and various control
strategies were tested with extensive parametric investigations. The aim of these tests
was the definition of the best performing control strategy for AFC solutions on wind
turbines. The final implementation of the actual AFC control strategy on a wind tur-
bine (instead of a constant chord wind tunnel wing) is naturally a very complex task
1 Introduction 7
and it is out of the scope of the current work. However it is believed that the control
strategy investigation presented in this document can be the starting point for a more
detailed, in-depth investigation on advanced AFC control system strategies.
Finally, an actual wind turbine blade design concept is proposed based on all the
previous research results and experiences. This blade incorporates several passive and
active flow control solutions aiming at improving the overall blade performance and
allowing a better power and aeroelastic load management. The development of such
“smart blades” for the wind turbines of the future will hopefully allow the easier inte-
gration of wind energy in the world energy market by reducing the operational cost of
wind turbines as well as by improving their electrical power quality.
Part II
Basic Theory
2 Airfoil Aerodynamics
2.1 Airfoil Design Parameters

The airfoils are some of the most fundamental shapes in aerodynamics. They are two-
dimensional slender shapes that are designed for operation within fluid flows in order
to generate forces due to their interaction with the flow. Subsonic airfoils are generally
characterized by a rounded leading edge and a relatively sharp trailing edge. The geo-
metric characteristics of airfoils are crucial for their performance, therefore a universal
communication terminology is developed allowing the proper shape reproduction of
airfoil shapes based on some basic dimension characteristics.
Figure 2.1: Basic airfoil design and definition parameters [73].
The main characteristics of the airfoil shapes (Fig.2.1) are summarized bellow:
• Upper surface: Also known as “suction side” is the low pressure region of the
airfoil extending from the leading edge to the trailing edge.
• Bottom surface: Also known as “pressure side” is the high pressure region of the
airfoil extending from the leading edge to the trailing edge.
• Chord line: The straight line connecting the leading edge and the trailing edge
of the airfoil.
• Mean line: Also known as “camber line” is the line which extends from the leading
edge to the training edge of the airfoil and is at equal distance between suction
and pressure side.
• Leading edge radius: The radius of the imaginary circle that defines the curvature
of the airfoil leadig edge
• Trailing edge thickness: The thickness size of the airfoil trailing edge.
• Maximum thickness: The maximum distance between suction side and pressure
side.
• Point of max. thickness: The chord-wise location of the maximum airfoil thick-
ness point.
• Maximum camber: The maximum distance between the chord line and the cam-
ber line.
• Point of max. camber: The chord-wise location of the maximum airfoil camber
point.
All the airfoil design parameters mentioned above are measured in relative length
terms and more specifically in percentage of chord line length (%c). This allows the di-
mensionless description of the general airfoil shapes therefore the information echange
regarding various different airfoil shapes is handled without the complexity of length
units.
Apart from the basic airfoil parameters, what is also crucial for the complete de-
scription of an airfoil is the shape of the outer contour (i.e. the thickness distribution).
In order to communicate the airfoil contours, a normalized Cartesian system is used
where the airfoil is described with a multitude of Cartesian (x,y) coordinate points.
For this coordinate system, the foremost point of the leading edge is (0,0) and the aft
most point of the trailing edge is (1,0).
Airfoil shapes and airfoil performance comparison is very important in all aerospace
applications and it is of course crucial for wind turbine blade design. The establishment
of a reliable airfoil description and communication system is therefore crucial for the
performance comparison of different airfoil shapes.
2.2 Airfoil Performance Characteristics

Since the airfoil shape description system is established it is important to describe the
basic performance characteristics of airfoils since these are crucial for all the following
research steps presented at the current document. The design and simulation process
of wind turbine blades currently is heavily based on airfoil performance curves (see
also 5.1) therefore a brief but thorough explanation of the basic theory behind airfoil
performance will be attempted.
The principles of lift and drag of infinite span airfoil sections are very well estab-
lished and explained by the basic fluid mechanics’ literature (e.g. [5, 79, 70]). The airfoil
performance, computed or measured is described and communicated mostly through
non-dimensional coefficients which allow the performance comparison of various airfoils
of different absolute dimensions tested with different methods. The establishment of
the airfoil performance curves allows the selection of the proper airfoils for each wind
turbine blade design. The most useful performance curves for HAWT applications are
the curves of the lift, drag, moment coefficient and glide ratio over angle of attack
(Cl /AoA, Cd /AoA, Cm /AoA, CAoAl /Cd
). Figure 2.2 presents some examples of airfoil lift
and drag polar curves
The airfoils are also characterized by their pressure and velocity distribution curves
[79] (Fig. 2.3). The shape of the airfoil strongly affects the pressure distribution on
the airfoil surface. By properly adjusting the airfoil shape it is possible to fine-tune
the airfoil pressure distribution in order to adjust the airfoil performance.
There is a great amount of airfoil shapes available in the literature and each one
of these airfoils is characterized by its own performance curves. To select the proper
airfoils for a wind turbine blade design it is important to establish a set of boundary
Figure 2.2: Polar curves for NACA63XX airfoil series of various airfoil thickness
levels. [26].
conditions. These boundary conditions define the characteristics of of the used airfoils
and are set by the blade designer. Each designer can set different boundary conditions,
but the most common ones for HAWTs are [70]:
• Surface roughness tolerance [127]
• Stall initiating from the trailing edge
• No laminar separation bubbles
• Smooth stall (no rapid loss of lift at stall)
• Now noise generation
• High structural stiffness
• High glide ratio at the design point
• Clearly defined laminar-turbulent transition
The following chapters deal mostly with the specific characteristics of wind turbine
aerodynamics, but as it was already mentioned the treatment of the blade as a com-
ponent is often done through a discrete amount of individual airfoils [70].
Figure 2.3: Pressure distribution along the chord (Cp / xy ) and Normalized velocity
Q
distribution along the chord ( Vinf / xy ) for an AH93W174 airfoil at 9o
AoA
3 HAWT Aerodynamics
3.1 Rotating Blade Aerodynamics

The aerodynamics of HAWT blades are in the most part identical to the classical finite
wing aerodynamics met on fixed wing aircraft. However there are some differences
which introduce different effects and complexities and lead to the extremely complex
problem that is rotor aerodynamics and aeroelasticity. A brief introduction into the
aerodynamics of HAWTs is attempted at this part of the document.
Figure 3.1: Flow velocity vectors acting on a wind turbine rotor section.
A finite wing rotating around an axis (Fig. 3.1) produces a lift force (normal to the
inflow direction) and a drag force (parallel to the inflow direction)(Fig. 3.2). Depending
on the local angle of attack, both lift and drag components contribute to the rotor
thrust and torque. Naturally, wind turbines make use only of the torque in order to
generate electricity while the thrust is absorbed by the wind turbine structure as a
load. The angle of the inflow of each airfoil segment of a HAWT blade comprises a real
wind velocity component and a peripheral wind velocity component due to the rotor
rotation and the distance of the said airfoil segment from the rotation axis. This means
that along the blade the angle of attack constantly varies. For this reason the blades
are designed with a structural twist in order to compensate for this AoA variation.
Figure 3.2: Distribution of aerodynamic forces along the rotor blades.
Apart from the AoA variation along the blades of a HAWT rotor, there are also
several other phenomena that add to the complexity of the system. The centrifugal
and Coriolis forces acting on the boundary layers of the blades significantly vary the
aerodynamic performance of their inner region (root region) [70] and cause strong
crossflows. The “pumping effect” caused by the low pressure of the outboard region
of the blades’ suction side contributes even more to the 3D flows (Fig. 3.3) of the
inner region. Additionally, the pressure equalization between the pressure and the
suction side at the blade tip introduces another region of strong crossflow which is also
combined with the generation of a strong spiraling tip vortex structure [70].
3.2 Unsteady Inflow

Wind turbines operate within the boundary layer of the Earth, thus their inflow con-
ditions are far from uniform [70]. The inflow of wind turbines varies in time and space
due to the velocity variation in Earth’s shear layer and the stochastic atmospheric
turbulence (Fig. 3.4).
The wind shear is usually described as an exponent function (Eq. 3.1)
x ν
Vo (x) = Vo (H)( ) (3.1)
H
where the Vo is the inflow mean wind velocity in m/s at hub height, H is the
hub height in m and ν is the shape parameter of the exponent function. This shape
parameter is defined empirically for various ground locations. The purpose of the shape
parameter is the modification of the shape of the exponent function according to the
local surface roughness conditions at each location (e.g. low vegetation, forest, water
e.t.c).
Apart from the wind shear a very important characteristic of the incoming wind flow
is the atmospheric turbulence. Turbulence exists in various scales inside the wind field
and is a stochastic 3-dimensional effect which is very hard to measure and practically
impossible to predict. Turbulence is usually modeled with appropriate simplified math-
ematical models (see also section 5.4) or is numerically computed with finite volume
solvers and unsteady simulations (e.g. Large Eddy Simulation (LES)).
The simulation of the inflow field is of very high importance for the wind industry
since it is a decisive parameter for the whole wind turbine simulation process which
can highly affect the simulation outcome and overall uncertainty.
Figure 3.3: Crossflow pattern on a wind turbine blade. Visualization based on the
work of Soerensen at al. [148]
Figure 3.4: Schematic representation of a non uniform inflow field. Such

stochastic turbulent fields are typically met by wind turbine rotors
[Source:Wikipedia.org].
3.3 Aeroelasticity
Wind turbine blades are some of the largest aerodynamic structures currently built.
They are rotating through highly unsteady inflow fields, are built from elastic com-
posite materials and are mounted on top of high steel or concrete towers. The whole
HAWT therefore as a system is a sum of elastic components that are coupled together
and are subject to highly varying aerodynamic loads. Through the strong coupling of
unsteady aerodynamics with the elastic structural response of the individual compo-
nents the system complexity increases enormously [152].
Aeroelasticity is a crucial part of wind turbine simulations therefore all the mod-
ern simulation and wind turbine certification codes include aeroelastic models. The
treatment of the aeroelastic behavior of wind turbines and wind turbine blades ranges
from simplified analytical structural models to complex adaptive finite element method
(FEM) numerical solvers. The development of wind turbine aeroelastic simulation
codes is intensive and the need for the development of even larger and even more reli-
able and cost effective wind turbines adds more momentum to this development.
The following chapters describe in more detail how all the aforementioned technical
fields (i.e. aerodynamics, aeroelasticity and wind inflow) were treated in the current
research work.
4 Airfoil Simulations
4.1 Panel Method Code

Several tools are available for the simulation of the performance of airfoil sections. The
simulation tool selection strongly depends on the scope of the investigation and the
available computational and time resources. In this work, the Open Source 2D panel
method code XFOIL [46] was used as a basic airfoil simulation tool. XFOIL is a viscous-
inviscid panel method aerodynamic tool for airfoil performance design and simulation.
The basic theory of XFOIL is based on a simple linear-vorticity stream function invis-
cid panel method with an explicit Kutta condition closure equation system. In order to
achieve good compressible flow predictions a Karman-Tsien compressibility correction
model is incorporated in the code [47].
The viscous part of the simulation is achieved with the description of the boundary
layer and the wake via a two-equation lagged integral boundary layer equation. The
boundary layer equations are coupled with an en laminar-turbulent transition model.
Accurate transition modeling is generally a non-trivial task and most only recently
RANS CFD models have included a pure physical representation of transition. For
the airfoil development, traditionally empirical or semi-empirical models were and still
are the state of the art in the field of laminar-turbulent transition. The en model is
probably the most popular transition model in the field of airfoil aerodynamics [64]
as well as in the field of wind energy [70]. The en transition model influences the
amplification factor of the Tollmien-Schlichting waves of the flow and thus can only
simulate the phenomenon of "Natural Transition". In case of high inflow turbulence
or highly distributed 3D roughness effects, other transition effects occur (e.g. bypass
transition) and these cannot be predicted with the use of the en model [47, 46]. Wind
turbine applications include highly turbulent flows but in absence of a more complete
transition model with reasonable computational costs, the en model was and still is
the dominant transition model. Lately more advanced transition models have been
implemented in commercial and other popular aerodynamic simulation codes and their
use is expected to increase. One of these advanced transition models is the k − kl − ω
model proposed by Walters et al. for RANS solvers [174], which was published in 2008
and was implemented in the 2.1 Version of Open FOAMTM in December 2011. However
this version of OpenFOAMTM and the aforementioned transition model was not used
in the course of this thesis.
The generated lift of the airfoils is computed through the integration of the computed
pressure distribution and the drag is computed from the wake momentum thickness
downstream the airfoil. At this point it is worth noting that the wake trajectory shape
is assumed from the inviscid part of the computation thus introducing an error which
increases with the increase of AoA. Panel methods however are already limited to mod-
erate AoA simulation ranges, therefore the wake trajectory issue is not critical for the
airfoil computation accuracy at low AoA [47].
Overall, XFOIL is a powerful and flexible design and simulation tool that is used ex-
tensively from the wind turbine industry in order to analyze the low and medium AoA
performance of airfoils in very short time with satisfactory accuracy. XFOIL being a
panel method code faces all the inherent limitations of such codes, such as the fact that
it cannot handle large flow separation, unsteady inflow conditions e.t.c. In addition to
that, XFOIL is not able to model multi-element configurations thus limiting its scope
to simple single element airfoils.
4.2 Finite Volume Method

4.2.1 Grid Generation
For the current research effort, finite volume CFD simulations were performed as more
detailed investigations after the use of the two dimensional panel method simulation
code (XFOIL). For these simulations, many custom codes were used for the finite vol-
ume grid generation phase and the solver control phase. The following paragraphs
provide a brief but thorough description of the methods and tools used for the CFD
research.
To investigate the performance of several PFC and AFC solutions without exceeding
the computational power of a powerful desktop PC it was decided to perform only
quasi-2D CFD simulations. At these simulations the flow domain comprises a 2D
airfoil and an outer boundary which is linearly-extruded in order to acquire a finite
thickness. The initial grid generation process begins with the use of a custom code
which imports one (or more) airfoils of various sizes and in various positions. An outer
boundary is also defined by the user and imported in combination with the airfoils.
The Open Source grid generation tool Gmsh [28] is used for the automatic generation
of an unstructured grid consisting of triangular and quadrilateral elements. The grid
generation algorithm forms a fine mesh around the the airfoil(s) in order to accurately
resolve the boundary layer and achieve a low Y + value. The cell size progressively
expands towards the outer boundaries of the grid in order to keep the total number of
elements low enough for a relatively fast computation.
The two-dimensional grid generated by Gmsh is imported to enGrid [65] and is ex-
truded to a thickness of 100mm. The extruded grid takes the form of a finite volume
grid which can now be handled by finite volume CFD solvers. The boundary con-
ditions for the wall (airfoil), inlet, outlet and empty side planes are defined and the
grid with the boundary conditions are exported into OpenFOAMTM [123] solver format.
The grid quality is defined by several user-input parameters (cell skewness, expansion
ratio, cell aspect ratio e.t.c) which were analyzed and calibrated during the initial re-
search phases via several grid sensitivity simulations. After achieving grid-independent
solutions, the optimal grid-generation parameters were fixed and used for the rest of
the computations.
4.2.2 Flow Solver

For the numerical flow simulations, the Open Source solver library OpenFOAMTM was
used (version 1.6). This library includes a very large amount of solvers which allow the
user to perform a big variety of simulations. Due to the large amount of configurations
to be tested it was necessary to achieve time and resource economy for the simula-
tions. For this reason the RANS type solver (simpleFOAM) was selected for all the
steady state CFD simulations. This solver incorporates time averaging and ensemble
averaging to the incompressible N avier − Stokes equations thus finally solving the
Reynolds-Averaged N avier − Stokes equations (Eg. 4.1) [19].
θūi θūi θūi θp̄ θ

= 0ρ + ρu¯j =− + (τ¯ij − ρu0i u0j ) (4.1)
θxi θt θxj θxi θxj
The closure schemes selected for the RANS simulations were the one-equation
Spalart-Allmaras [6] turbulence model and the K − ω SST (Shear Stress Transport)
two-equation model [113]. The simpleFOAM algorithm was set up to solve the RANS
equations for the unstructured airfoil grids mentioned before. The computations were
set to perform 5000 iterations in order to assure that the computation is stabilized
and that the results converge properly. The simulations were executed without the
use of a laminar-turbulent transition model and the boundary layer was assumed to be
turbulent throughout the computational domain.
The inflow conditions were automatically varied after the end of a computation
and the next computation automatically started for a new inflow angle. This allowed
the automatic computation of a complete airfoil polar without user interference. The
computation of each angle of attack takes approx. 1,5hours at a 4 core, 3.2GHz desktop
PC with 6GB of RAM memory. The simulations were activated in sequence but were
not coupled, therefore no investigation on hysteresis effects was possible.
In order to validate the performance of the grid generation tool in combination
with the OpenF OAM TM solver performance, several validation cases were investigated.
Various airfoils were computed with OpenF OAM TM and were compared to simulations
with XFOIL and wind tunnel measurements performed at the Hermann Föttinger
Institute (TU Berlin).
The DU97W300 airfoil has a relatively high sensitivity to roughness and therefore the
comparison between various wind tunnel measurements or between CFD simulations
and experimental results is a complicated task. As Fig.4.2 shows the boundary layer
transition location plays a very significant role for the performance of the airfoil since
it does vary the stall angle and the maximum lift of the airfoil significantly. The airfoil
used for the wind tunnel measurements had a smooth surface but with a roughness
similar to that of the actual wind turbine rotorblades. Therefore it is not surprising
that XFOIL over-predicts the performance of this airfoil. The wind tunnel results are
much closer to the results of XFOIL for a fully turbulent boundary layer (Fig. 4.1).
Furthermore Fig.4.1 shows that a high number of wall cells are required in order to
capture the performance of the boundary layer without using a transition model or
wall functions. The best results were extracted with a surface representation with at
least 6000 wall cells, leading to a grid with a total of aprox. 0.7 · 106 elements. The
2 − equation Spalart Allmaras turbulence model generally outperfroms the k − ωSST
model for the simulations of this airfoil both in speed and accuracy (see also Fig. 4.1).
Cl
1.5
0.5
0
XFOIL
XFOIL Tripped at 5%c
-0.5 Experiment (TU Berlin)
S-A (9000 wall cell Grid)
S-A (6000 wall cell Grid)
-1 S-A (3000 wall cell Grid)
k-ω SST (6000 wall cell Grid)
6
Note: All curves at 1.5x10 Re
-1.5
-15 -10 -5 0 5 10 15 20 25
AoA [°]
Figure 4.1: Comparison of the measured performance polar curve of DU 97-W-300

airfoil and the computed curves with CFD (OpenFOAMTM and XFOIL).
The experimental polar was measured at the large wind tunnel of H.F.I
TU Berlin. The CFD curves present simulation results with various
turbulence models and grid densities.
Figure 4.2: Comparison of the measured performance polar curve of DU 97-W-300

airfoil and the computed curves with RFOIL [164].
5 Wind Turbine Simulations
5.1 Steady State HAWT Simulation

The most common tool to perform stationary wind turbine performance simulations
is the combination of the Blade Element Momentum Method (BEM) of Betz and
Glauert with the Blade Element Theory [70]. These theories date back to the early
1930s and are still being widely used and further developed together with other sim-
ulation tools. A very detailed description of the Blade Element Momentum Method
(BEM) and the Blade Element Theory for wind turbine applications was performed
by Hansen [70]. Therefore the following paragraphs do not describe the method in
extreme detail, but rather give an overview of the BEM model, its assumptions and
its limitations. Based on the work of Hansen [70] an Open Source BEM-based wind
turbine simulation software called QBlade was developed by D. Marten and the au-
thor [128]. The software development was funded by the Hermann Föttinger Insti-
tute of TU Berlin (Chair of Fluid Dynamics) and the author and is available at:
http://fd.tu-berlin.de/en/research/projects/wind-energy/qblade/
The blade element theory is based on the assumption that wind turbine blades can be di-
vided into discrete elements that act independently from each other and their performance
characteristics are identical to those of 2D airfoil sections. In this manner, the definition of
the inflow conditions (i.e. local AoA, local inflow velocity e.t.c) for each one of these elements
would represent a different state of operation for each one of the represented airfoil sections.
The integration of the aerodynamic forces of all the individual sections along the blade pro-
vides the performance characteristics of the blade as a whole.
The blade element momentum method is used in order to compute (or better estimate) the
local inflow conditions that are required by the blade element theory. The momentum method
assumes that the loss of momentum (or loss of pressure) along the rotor plane is caused by the
extraction of energy from the flow. This energy extraction is caused by the work produced
by the wind turbine rotor. The momentum theory defines discrete annuli which are formed
Figure 5.1: Blade representation with independent airfoil segments according to the
blade element theory. The velocity vectors are computed for each seg-
ment through the blade element momentum method. The AoA along
the blade changes due to blade rotation.
by the area between two consecutive airfoil sections, rotated around the blade rotation axis.
The momentum within each annulus is assumed to be constant, therefore the induced velocity
can be computed for each one of the blade sections [70]. The induced velocities along the
axis of the wind turbine rotor and wind turbine blade axis are represented with the axial and
tangential induction factor respectively (eq.5.1 and 5.2). These are defined as:
1
a= (5.1)
4 sin2 φ
σCn +1
and:
1
a0 = (5.2)
4 sin2 φ cos φ
σCt −1
The coupling of the blade element theory with the blade element momentum method is
necessary since the performance of the individual airfoil sections affects the induction factor
of the momentum theory. Furthermore the variation of the induction factor leads to inflow
condition variations which affect again the airfoil section performance. An iterative compu-
tation scheme is therefore required in order to solve the said set of equations and achieve a
fully converged solution.
The main assumptions and limitations of the BEM theory are summarized in the following
list and they remain valid for all the wind turbine simulation approaches based on this method.
• The inflow wind velocity is assumed to be constant and uniform, while in reality it is
fluctuating in the time and space domain.
• The shed vorticity is assumed to happen at the root and tip of the finite length wing
(i.e. wind turbine blade). This accounts for an energy loss which is computed through
empirical correction models.
• The rotational effects are accounted for only with respect to the velocity distribution
along the blade. All the other rotation-induced (e.g. Coriolis forces, centrifugal forces)
are ignored.
• The momentum balance assumes that the blades rotate aligned with the rotor plane.
Off-plane rotation and curved blades cannot be simulated with the standard BEM
algorithms.
• The airfoils (elements) are assumed to operate independent from each other.
• No interaction between blades is considered. The algorithm performs the simulation

for a single blade and then adjusts the result for the complete rotor.
• Only steady state aerodynamics are considered through the airfoil polar utilization
(Cl /AoA, Cd /AoA and Cm /AoA).
5.1.1 Prandt’s Tip and Root Loss Correction

Actual wind turbines have a finite amount of blades while the basic BEM model assumes an
infinite amount of blades. The difference in the vortex system in the wake between theory
and practice is corrected via the Prandtl’s Tip and Root correction [70]. This is an empirical
correction model which is based on the correction factor F (Eq.5.3 and eq.5.4).
2
F = cos−1 (e−f ) (5.3)
π
where:
BR−r
f= (5.4)
2 rsinφ
B is the number of wind turbine blades, R is the rotor radius, r is the local radius and φ is
the flow angle. Based on this correction, the axial and tangential inductions factor equations
(eq.5.1 and 5.2) are modified and take the following form (eq.5.5 and 5.6):
1
a= (5.5)
4F sin2 φ
σCn +1
and:
1
a0 = (5.6)
4F sin2 φ cos φ
σCt −1
Other blade tip and root corrections have also been developed and can also be applied to
the BEM [144]. One example is the correction factor proposed by Shen et al. [144] in order
to correct a numerical instability of Prandtl’s tip and root correction factors. This correction
was also implemented in the software QBlade and a thorough description of these correction
models can be found in the QBlade manual and in the relevant paper by Marten and Pechli-
vanoglou [128].
5.1.2 Glauert Correction

Another correction applied at the BEM calculations has to do with the loading level of the wind
turbine rotor disc. When the wind turbine rotor is highly loaded and the axial induction factor
becomes larger than approximately 0.4 (i.e. turbulent wake state), the standard momentum
equation breaks down and a correction is necessary. Glauert proposed such a correction which
is commonly used and is thoroughly described by Hansen [70]. For the turbine simulations
an updated version of this correction was used. This version was proposed by Buhl [25] and
was implemented in QBlade where the correction equation takes the form of eg. 5.7:

8 40 50
CT = + 4F + a+ − 4F a2 (5.7)
9 9 90
And when solved for the axial induction factor is:

p
18F − 20 − 3 CT (50 − 36F ) + 12F (3F − 4)
a= (5.8)
36F − 50
5.1.3 Snel’s correction

Another correction for the BEM which is commonly used and was also implemented in QBlade
is the correction proposed by Snel [147] for the Himmelskamp effect [74] of wind turbines. This
effect was discovered in the late 40’s by H. Himmelskamp and explains the stall delay that is
observed at the inner sections of HAWT blades. The strong action of the centrifugal forces
at this part of the blade and the generation of Coriolis forces accelerates the local boundary
layer thus reducing its thickness and delaying flow separation. In order to correct for this
effect, Snel proposed a pre-procesing modification of the lift polars for the inner airfoils of
wind turbine blades before the initiation of the BEM iterations. The corrected lift curves have
an extended linear lift part and a thus a higher Clmax value and a stall delay. The correction
is performed according to the eq. 5.9:
3.1λ2 c 2

dcl
CL,3D = CL,2D + g sin(a − a0 ) − CL,2D (5.9)
1 + λ2 r dalinear
Where g is the blending factor which ensures a smooth geometrical transition of the polar
curves and is computed according to eq. 5.10
g = 1, 0 < α < 30,

1 (5.10)
g= 2 (1 + cos (6α + 180)), 30 < α < 60,
g = 0, 60 < α < 360
5.1.4 360 Degree polar curve extrapolation

As mentioned before, the blade element theory uses two-dimensional airfoil performance data
in order to compute the overall performance of wind turbine blades. These curves are either
computed or measured in the wind tunnel and imported in the form of lift, drag and pitch
moment coefficient tables. Most of the wind tunnel measurements as well as many numerical
airfoil simulation codes (e.g. XFOIL) fail to compute the performance of airfoils at very high
AoA. During the analysis however of wind turbine blades often some blade sections operate
at high AoA (e.g. sections at the blade root region) [129]. The airfoil performance tables
therefore need to be extended for the complete 360o range of AoA. This is done with the use
of empirical extrapolation models based on flat plate experimental measurements and a few
360o airfoil wind tunnel measurements.
For the current research work and for the wind turbine simulation software QBlade which
was developed within the scope of this thesis, a hybrid airfoil polar extrapolation technique
was used. The initial approach is based on the extrapolation proposed by Viterna [171] which
assumes the deep-stall behavior of all the airfoils to be similar to that of a flat plate. This
extrapolation technique is universally used with acceptable accuracy. However modern wind
turbines use very thick airfoils (thickness higher than 30% of the chord length), which do
not perfom in any way similar to a flat plate. In order to correct this effect, an additional
empirical extrapolation methodology was incorporated. This methodology was developed by
Montgomerie [117] and was based on some actual 360o wind tunnel measurements of NACA
airfoils and several mathematical extrapolation and smoothing techniques.
Figure 5.2: Extrapolated 360o Cl curve of the DU 00-W-350 Mod2 wind turbine
airfoil.
The polar extrapolation technique of Montgomerie [117] allows the precise yet empirical
control of the extrapolation through a number of polar curve shape parameters. These pa-
rameters provide a finer and smoother extrapolation of the polar curves but also introduce
another error factor through the additional empirical correction into the blade element theory
computations [128]. Extensive details on these 360o extrapolation techniques and their effect
on wind turbine simulations can be found in the QBlade manual as well as the publication of
Marten and Pechlivanoglou [128].
5.2 Dynamic HAWT Simulation

The BEM simulation software (QBlade), developed in the course of the research work is able
to perform the complete wind turbine blade aerodynamic design as well as the steady state
Figure 5.3: Extrapolated 360o Cd curve of the DU 00-W-350 Mod2 wind turbine
airfoil.
performance simulation. This functionality is suitable for the investigation of various blade
designs as well as of performance-increasing PFC elements, but the software is not able to
simulate the dynamic behavior of the AFC solutions. To simulate the dynamic effects of
HAWT systems it is necessary to make use of an unsteady simulation code. Furthermore as
discussed before aeroelasticity is a very important parameter in wind turbine blade design.
The assumption therefore of perfectly rigid blades for the aerodynamic simulations would in-
troduce a significant error to the simulation results. For this reason an unsteady aerodynamic
wind turbine simulation code was selected and it was coupled with a medium complexity
structural model.
The selected aerodynamic code is AeroDyn developed by the researchers of NREL / NWTC
[118]. It is a conventional dynamic blade element momentum (BEM) code coupled with a
generalized dynamic wake (GDW) model for the computation of the blade inflow. The aero-
dynamic loads on the blade sections are computed with the classical blade element theory
(BET). The aerodynamic model of AeroDyn is also coupled with a dynamic stall model which
is responsible for the implementation of the dynamic stall parameters to the static polar curves
of the blade sections. The control of AeroDyn is performed via a modified version of QBlade
which handles the appropriate input communication.
AeroDyn was selected for the dynamic simulations required, not only because it is a high
quality validated and certified code, but also because of its Open Source license. The access
to the source code of AeroDyn was necessary to implement the appropriate modifications for
the simulation of AFC elements.
The aerodynamic code (AeroDyn) is coupled with a structural model which is also a prod-
uct of NREL / NWTC. This is the Open Source medium complexity structural wind turbine
code YawDyn [99]. YawDyn is also controlled through the modified version of QBlade.
To achieve a realistic wind turbine simulation with AFC elements it is necessary to use
an accurate wind inflow model. For this purpose the TurbSim atmospheric modeling code
of NREL / NWTC [94] was used. The TurbSim code seamlessly cooperates with AeroDyn
and is also controlled through the modified version of QBlade. TurbSim like all the other
codes of NREL / NWTC can be obtained from the “Solver Download” web-page of NWTC
(http : //wind.nrel.gov/designcodes/simulators/).
Similar to the pre-processing and the solver control, the post-processing of the dynamic
simulation outputs is handled within the modified version of QBlade. This allows the out-
puts to be imported into dynamic graphs and to be analyzed and compared to the “baseline”
simulation cases. The following paragraphs describe the function and theory of all these afore-
mentioned modules in further detail.
5.3 Dynamic Stall

As mentioned before when using the BEM method for the simulation of a wind turbine per-
formance, the airfoil section performance is extracted from static airfoil polar tables. The
use of static polars produces quite accurate results for rotating blades which do not undergo
large AoA oscillations near the Clmax region. The elasticity however of wind turbine blades
in combination with the unsteady inflow velocity lead to large AoA fluctuations especially
at the outer 30% of the blade span [118]. The oscilating AoA of the airfoil section of wind
turbine blades introduces dynamic hysterisis effects (dynamic stall) that cannot be modeled
or described by static polars. For this reason the use of dynamic stall models for BEM simu-
lations is considered being necessary.
The dynamic stall model most commonly used for wind turbine simulations is the Beddoes
& Leishman model [118] developed in the late ’80s. This model uses the ”Normal” and
”Chordwise” force coefficients (CN and CC respectively) which are computed by the existing
Figure 5.4: Comparative graph of a static behavior of an airfoil (static polar) with
the dynamic behavior of the same airfoil (dynamic stall loops) [118].
Cl and Cd values of the standard airfoil polars. The Beddoes-Leishman dynamic stall model
I ) and a circulatory one
uses the normal force coefficient as a non-circulatory component (CN
C ). There are given as:
(CN
C
∆CN = CN A φ C
α ∆α, (5.11)
I 4 I
∆CN = φ ∆α (5.12)
M α
where CN A is the normal force coefficient curve slope, M is the Mach number, φC
α is the
circulatory indicial function and φIα is the non-circulatory indicial function. The chordwise
force coefficient CC is based on the circulatory component of CN .
The dynamic stall model generates a hysterisis loop around the Clmax and post stall region
which essentially works as a “dynamic” extension of the static polars. The lift and drag
coefficient values of the static polars are dynamically exchanged with the updated values
computed by the dynamic stall model. The lift and drag coefficient are calculated with the
following equations:
CL = CN cos (α) + CC sin (α), (5.13)
CD = CN sin (α) − CC cos (α) + CD0 (5.14)
The Beddoes-Leishman model was initially developed for the helicopter industry and its
implementation for wind turbine simulations requires some slight modifications. In the current
work the dynamic stall model was used as implemented in the AeroDyn code of NREL [118].
The main modifications of the Beddoes-Leishman model found in AeroDyn are:
• Extension of the model’s effective AoA range.
• Modified method for the calculation of the effective flow separation point.
• Use of both CN and CC for the calculation of the effective flow separation point.
• Application of a saturation function which limits the impulsive contribution to the

pitching moment of rapid effects such as the “tower shadow effect”.
• Small model modification for a more accurate prediction of the dynamic stall hysteresis
in “deep stall” situations.
The modeling of dynamic stall effects is a very complex process which is especially hard to
validate in a 3D flow environment due to the lack of experimental data especially for wind
turbine applications. The aforementioned implementation of Beddoes-Leishman type models
might have smaller or greater discrepancies from reality depending on the simulation and
blade design parameters. Wind turbine applications have some characteristics that further
increase the complexity of modeling dynamic flow effects. The introduction of "Theodorsen’s
Theory" principles for the dynamic performance of airfoils, in combination to the extension of
the standard dynamic stall models for deep-stall and the dominant trailing edge separation are
some of these additional complexities. Current large rotorblades also experience extreme tip
deflections which introduce significant plunging effects on the outboard airfoil sections. These
effects also need to be implemented in the dynamic flow response models of the aeroelastic
simulation codes. Furthermore the introduction of passive and active flow control elements
requires adaptations and extensive validations of the existing dynamic stall models in order to
achieve realistic performance representations. Some wind turbine oriented implementations
of Beddoes-Leishman type dynamic stall models have already been proposed in order to cover
many of the aforementioned limitations of the initial helicopter-oriented models, such as the
proposal of Hansen et al. [71]. Furthermore additional implementations take into account
unsteady effects due to flexible flap deformations (e.g. [8]). It is however necessary to develop
and implement a much more general model that takes into account most of the physical
phenomena of dynamic stall in order to achieve accurate and reliable aeroelastic simulation
capability for practically any wind turbine blade design.
In this thesis the previously described and already implemented dynamic stall model of the
AeroDyn simulation code is used for all the simulations. The development of a new dynamic
stall model or the modification of the existing and validated aeroelastic code is out of the
scope of this thesis.
5.4 Inflow Turbulence Modeling

As previously mentioned, the steady state BEM simulations assume a uniform and constant
wind inflow. This is far from reality where the wind velocity varies in both space and time.
To simulate more realistic inflow conditions it is necessary to include time-varying wind inflow
information. These information are computed through mathematical turbulence models for
a large amount of intervals (time steps) and with a spatial distribution sufficient to fit the
whole wind turbine rotor. The wind inflow input data have the form of a four-dimensional
matrix and contain information about the three components of the wind velocity and their
time correlation. There are several models that generate wind inflow matrices but the most
commonly used is a combination (model overlay) of several sub-models.
The wind inflow simulation was performed with the TurbSim software of NREL / NWTC
[94]. TurbSim is an Open Source code which includes several spectral models that can be used
to calculate the incoming flow to the wind turbine. These spectral models are the following:
• SNLWIND-3D SMOOTH : For flows over flat homogeneous terrain.
• WF-UPW, WF-07D, WF-14D: For flows upwind of a multi-row wind farm and with 7
and 14 rotor diameter row-to-row spacing within the farm.
• IEC Kaimal
• von Karman NTMs
• NWTCUP : Highly turbulent condition simulation spectral model.
• GP-LLJ : For the simulation of the large vertical shears and coherent turbulence struc-
tures that are typical phenomena at the Great Plain regions.
TurbSim offers the additional possibility to superimpose randomized coherent turbulent

structures over the standard spectral models. These coherent structures are generated based
on Kelvin-Helmholz breakdown wave computations performed with Large Eddy Simulations
(LES) and Direct Numerical Simulations (DNS) by the researchers of the NWTC [94].
The functionality of TurbSim was integrated in the modified version of QBlade through
which the necessary input files are generated. The implementation of the wind inflow simula-
tion code at the same Graphical User Interface (GUI) platform allowed the efficient and fast
simulation work-flow.
5.5 Generalized Dynamic Wake Modeling

As discussed above, the typical wind turbine dynamic simulations use the combination of blade
element momentum method and blade element theory. The first is used for the calculation
of the inflow angle along the blade and the second resolves the aerodynamic performance of
the blade. In this work however and through the use of the AeroDyn code of NREL/NWTC
[118] an alternative to the blade element momentum method was utilized for a large part of
the simulations. This alternative inflow angle calculation method is the generalized dynamic
wake method (GDW). GDW was initially developed by Pitt and Peters [135] in 1981 for heli-
copter rotors and was further developed and extended for its implementation in AeroDyn [118].
The GDW is based on a potential flow solution for Laplace’s equation (Eq.5.17) through
which the equations for the pressure distributions in the rotor plane are created. The rotor is
assumed as an infinite number of slender blades (lifting lines) and through the Euler equations
(Eq.5.15) for inviscid and incompressible flow and the conservation of mass (Eq.5.16), the
solution results to the pressure distribution calculation (Laplace equation). The initial values
for these differential equations solved by the GDW (equation initialization) are provided by
the co-existing BEM method routine. The basic equations of the GDW method are:
θui θUi 1 θρ
+ U∞j =− , (5.15)
θt θxj ρ θxi
θui
= 0, (5.16)
θxi
∇2 p = 0 (5.17)
GDW offers some specific advantages over the traditional BEM method and that is the
reason why it is often preferred for the inflow angle calculation. The advantages of GDW over
BEM are:
• Dynamic wake effect modeling

• Tip loss modeling
• Skewed wake aerodynamics modeling
• Induced velocity calculation with 1st order differential equations solved non-iteratively
The aforementioned advantages of the GDW make it valuable as a replacement for the
traditional BEM method during wind turbine simulations. This method however also has
disadvantages which are often similar to those of the BEM method. One of its main disad-
vantage is the assumption of lightly loaded rotors and therefore assumes that the induced
flow velocities are small compared to the mean flow velocity. At low wind speed where the
rotor enters the turbulent wake state the GDW becomes numerically unstable. To eliminate
this instability issue, AeroDyn automatically switches to the traditional BEM method with
the subsequent Glauert correction when the mean wind speed is bellow 8m/s. Another dis-
advantage of the GDW is its inability to model the wake rotation donwstream of the wind
turbine rotor. Once again the traditional BEM method is used by AeroDyn to calculate the
tangential induction factor (Eq.5.2) which defines the wake rotation. Finally, similar to the
BEM method, the GDW cannot model the wake aerodynamics of rotors with large aeroelastic
deflections or large coning angles.
5.6 Wind Turbine Structural Model

The aeroelastic wind turbine blade simulations require the coupling of the aforementioned
rotor aerodynamic simulation model to a structural simulation model. From the available
structural models for wind turbine simulations the one which was selected for the simulations
presented in this document is the YawDyn of NREL/NWTC. This is an Open Source and
medium complexity wind turbine structural code used mostly for the preliminary aeroelastic
analysis of wind turbines1 . YawDyn was selected due to its relative simplicity and due to the
fact that it focuses largely on the simulation of the blade behavior rather than the simulation
of the whole wind turbine. This allows the user to easily perform comparisons between the
performance of different blades with PFC and AFC solutions without introducing too many
irrelevant wind turbine - specific parameters.
Principally YawDyn includes a structural model for a rigid wind turbine nacelle which can
be yawed around the vertical axis. The nacelle can be horizontal or assume a tilt angle (pos-
itive or negative). The nacelle yaw is defined by several dynamic parameters such as the yaw
1
Since the end of 2010 YawDyn is not supported or suggested for use. Instead, the more advanced
FAST structural model should be used.
rate, the yaw friction (damping), the yaw stiffness e.t.c. In this way it is possible to simulate
the yawing motion of a turbine and its effects on the load profiles of its blades.
Figure 5.5: Wind turbine and coordinate system representation for YawDyn [99].
The main shaft of the wind turbine is modeled as a rigid element. The rotor hub can be
modeled as a rigid element or as a teetered element2 . In case of a teeter rotor, the struc-
tural model requires the teeter spring parameters as well as the teeter damper parameters.
2
Only for 2-bladed teeter rotors.
Figure 5.6: Rigid blade representation for YawDyn with ideal spring-hinge. [99].
Furthermore the teeter limit angle needs to be defined. The simulations assumed a modern
three-bladed HAWT, therefore the teeter hub option was not used. Instead, a rigid hub was
assumed as well as a rigid hub-blade connection. The blades are represented in the model
with a rigid beam connected to the hub by an ideal spring-hinge. The blade pitch angle is
fixed to a user defined value and does not change during the simulation time. In this way the
structural model allows the simulation of dynamic blade deflection while at the same time the
blade is essentially treated as a rigid body. This greatly simplifies the simulation setup pa-
rameter requirements. Furthermore by keeping the complexity of the structural model to the
minimum, it is possible to investigate the actual aerodynamic effects of various AFC solutions.
5.7 BEM with Active Elements

The full aeroelastic simulation of a conventional wind turbine can be done with various simu-
lation codes. However the implementation of dynamically actuated flow control elements on
wind turbine blades introduces another level of complexity to the existing aeroelastic simu-
lation problem. It is therefore necessary to include the active flow control elements to the
aerodynamic performance simulation code. This would allow the proper simulation of the
aerodynamic response of the blade to the actuation of the flow control elements. Further-
more, the coupling of the aerodynamic solver (AeroDyn) with the structural model (YawDyn)
allows the proper simulation of the full aerodynamic and structural response of the turbine
to the AFC inputs.
To implement the ACF elements into the aerodynamic solver it is necessary to compute
and properly represent their aerodynamic effect variations. The aerodynamic representation
of the AFC elements has to be compatible with the existing aerodynamic model. To do
that, the principle of the airfoil lift and drag polar representation is maintained. Each air-
foil is registered with a “mother” polar curve that represents its steady state performance.
The representation of the various dynamic states of the AFC elements is registered in the
form of various “children” polar curves. These additional polars are connected with the main
(“mother”) polar curve and receive individual IDs which allow the definition of the AFC state
during the simulation.
The aerodynamic solver (AeroDyn) was accordingly modified to be able to use input data
for several polar curves for each blade section per calculation time step. A relatively simple
controller was also implemented inside the aerodynamic solver code. The task of the con-
troller is the selection of the most suitable “child” polar curve for the achievement of the
control goals. In this way the solver computes the aerodynamic performance of the blade and
then the controller decides the ideal state of the AFC element. This state is implemented
at the next time-step by shifting to the appropriate polar curve and the calculation is repeated.
Naturally the modified aerodynamic solver remains coupled with the structural aeroelas-
ticity model (YawDyn) as well as the inflow turbulence model (TurbSim). Therefore a quite
accurate simulation of the AFC elements’ aeroelastic performance is achieved. Through these
dynamic wind turbine simulations with AFC it is possible to define the most effective size
and position of each AFC element on the blade. Furthermore it is possible to compare the
aeroelastic performance of rotorblades with different flow control solutions.
The controller which defines the position of the AFC elements is one of the most crucial
components of the simulation. The development however of extremely advanced and complex
controllers for the aeroelastic simulation of AFC elements on HAWT blades is out of the
scope of this research work. It was therefore decided to develop two simpler controllers for
the actuation of the AFC elements within the aeroelastic simulations. One is based on a
simple optimization loop which acts on each active blade section individually and is able to
optimize the AFC element state. With this controller no communication is able between the
various elements and therefore the control parameters can only be local (element-specific) such
as the local lift coefficient. A slightly more advanced controller however was also developed
based on the PID controller principle. This controller is not anymore element-specific but its
control input comes from the full-blade performance parameters (e.g. flap-wise root bending
moment). The required computation effort is much smaller since the PID controller determines
the right AFC element state on-the-fly for each time step of the aeroelastic computation.
Part III
Experimental Setup
6 Wind Tunnel Experiments
6.1 Wind Tunnel Description

The wind tunnel test campaigns took place at the wind tunnel facility of TU Berlin (HFI)
(Fig. 6.1). This is a closed loop wind tunnel with a closed test section and a cross-section
√
A equal to 2 × 2 m2 . The length l of the test section is equal to 10m and the contraction
ratio is 6, 25 : 1. The wind tunnel fan is powered by a 500kW speed regulated DC motor.
The maximum free stream velocity Usmax achieved in the wind tunnel is 50 m/s and the
turbulence level of the free stream is below 0.5%.
Figure 6.1: View of the nozzle and test section of the HFI/TU Berlin Wind Tunnel.
For the steady state wind tunnel measurements four constant chord, zero twist quasi-2D
test wings were machined out of ObomodulanTM with a high precision CNC milling machine.
The chord of the test wings is 600mm and the span 1540mm. The constant airfoil sections of
the test wings are the DU96W180, the AH93W174, the DU97W300 and the N ACA633 618.
These airfoils are considered to be popular wind turbine airfoils and mostly represent airfoil
Figure 6.2: Detailed diagram of the HFI/TU Berlin Wind Tunnel measurement sec-
tion [114].
Table 6.1: Table of specifications for the wind tunnel of H.F.I TU Berlin.
Wind tunnel GroWiKa (TU Berlin)
Test section dimensions [m] 2m · 1.44m
Contraction Ratio 6.25 : 1
Reynolds number 1.3 · 106
Turbulence Intensity < 0.5%
AoA Range −8o to 25o
Wake blockage correction Yes
Solid blockage correction Yes
shapes that are typically found at the mid and outboard blade sections. Table 6.2 shows the
main characteristics of these test wings.
The DU96W180 airfoil was designed in 1996 at the Technical University of Delft (Nether-
lands) [165]. It was designed with the RFOIL code which is a modification of the XFOIL code
of Prof. M. Drela [47] and was measured at the wind tunnel of TU Delft. It is a relatively
thick airfoil (18%c thickness) which is derived from the DU95W180 airfoil and is designed
for the mid-span and outboard sections of wind turbine blades. Its characteristic is its low
camber, its smooth pressure recovery region and the cambered pressure side at the trailing
edge region (s-shaped trailing edge) which is responsible for a significant amount of the gen-
erated lift. The test wing used for the measurements during this work was designed by the
author in a way that allows the replacement of the rear 30% of its chord. A rigid rear segment
was manufactured after the exact coordinates of the DU96W180 airfoil and was used for the
Table 6.2: Specifications of the wind tunnel test wings used for static and dynamic
measurements.
Airfoil DU96W180 AH93W174 N ACA633 618 DU97W300
Chord [m] 0.6 0.6 0.6 0.6
Thickness [%c] 18%c 17.4%c 18%c 30%c
Span [m] 1544mm 1544mm 1544mm 1535mm
reference measurements. An alternative form-flexible rear segment was also manufactured for
the flexible flap wind tunnel tests.
The AH93W174 airfoil was developed in 1993 by D. Althaus at TU Stuttgart [7] and was
measured at the Laminar Wind Tunnel of TU Stuttgart. It is a relatively thick airfoil (17, 4%c
thickness) which is designed for the mid-span and outboard sections of wind turbine blades.
Its relatively high camber in combination with the moderately s-shaped trailing edge lead to
relatively high lift generation. The test wing used was precisely machined down to the exact
dimensions from a block of model-making material (ObomodulanTM ).
The N ACA633 − 618 airfoil is a part of the NACA 6-series airfoils developed in the 50s
by the National Advisory Committee for Aeronautics in the United States [5]. They were
developed as advancements to the older 4 and 5 digit airfoils. They are characterized by a
relatively moderate pressure distribution with smooth pressure recovery. They generate quite
high lift and have very low drag at a small operational range due to their extended laminar
BL region. N ACA633 − 618 has been a popular wind turbine airfoil and it is found at the
mid-span or outer region of many wind turbine blades (18%c thickness). It is (as all the
6-series airfoils) sensitive to surface roughness but it is selected for the current tests due to
its popularity with the wind turbine blade designers. The test wing was designed by the
author in such a way that it can have exchangeable leading and trailing edge modules. The
front and back 25%c of the test wing consists of detachable leading and trailing edge segments.
All the segments are precisely machined out of solid model-making material (ObomodulanTM ).
The DU97W300 was designed in 1997 at TU Delft [165] and and tested at the local wind
tunnel. It is one of the most popular examples of the DU wind turbine airfoil family. It
is a thick root region airfoil section (30%c thickness) and it is commonly used on modern
turbine blades. It is characterised by its moderate camber and flat pressure recovery region.
A large portion of its lift is generated by its strongly cambered pressure side of its trailing
edge (s-shaped trailing edge). The test wing for the current experiments was CNC machined
out of a solid piece of ObomodulanTM . It is equipped with special mounting mechanisms that
allow an auxiliary airfoil to be mounted in various positions.
NACA 633618
DU97W300
AH93W174
DU96W180
0 0,2 0,4 0,6 0,8 1
Figure 6.3: The airfoil sections of the four test wings used for wind tunnel measure-
ments.
The test wings are mounted on the 6-component force balance of the large wind tunnel
facilities of the Hermann Foettinger Institute (HFI) TU Berlin. The wind tunnel test section is
equipped with additional splitter walls to reduce the effects of the wind tunnel boundary layer
and the span-wise flows, due to blockage, to the measurement results. The test wings occupy
the complete span between the wind tunnel splitter walls (a=1.554m) and are attached to
the external 6-component balance fully de-coupled from the wind tunnel construction (Figure
6.2 and 6.4). The gap between the test wing and the splitter walls is reduced to 0.1mm by
means of sealing with thin tape. This is done to avoid pressure leaks from the pressure side
of the airfoil towards the suction side. The lift, drag and moment coefficients are measured at
1.3 · 106 Reynolds Number. The data acquisition for the forces measured by the 6-component
balance is done via a C-DAQ unit of National Instruments at a scanning sampling rate of
100Hz with an output rate of 2Hz after the noise filtering. The resolution of the balance is
approx. 0.01N and the maximum permissible load was 3kN.
6.1.1 Static Measurements

The airfoils were tested in various configurations which were compared with baseline mea-
surements via a comparison of static airfoil polar curves. These curves were measured by
pitching the airfoil around a pitching axis located at its quarter-chord position. The polars
were measured at an angle of attack (AoA) range of −15o to +25o with a 0.5o and 2sec step
between measurements. The polars were measured from −15o to +25o and directly back to
−15o in order to capture any lift hysteris effects.
Figure 6.4: The 6-component force balance used for the current wind tunnel cam-
paign [114].
The measured wind tunnel data were post-processed and corrected for solid blockage and
wake blockage. Solid blockage represents the cross-section reduction that a closed wind tunnel
experiences due to the existence of a solid body (test wing) inside it. Its effect is a function of
the test wing thickness, thickness distribution and size and it is independent from its camber
[136]. The current wind tunnel measurement data were corrected for solid blockage with
Thom’s correction [136] equation (eq.6.1):
K1 (V olwing )
sb = 3 (6.1)
C2
Where K1 is the test wing span factor V olwing is the volume of the test wing and C is
the area of the wind tunnel at the test section less the wind tunnel boundary displacement
thickness.
The wake blockage is caused by the flow-speed-up effect which occurs in closed test section
wind tunnels due to the existence of the low velocity wake behind the test wing. The com-
pensation of the velocity loss due to the wake (through Bernoulli’s principle) leads to a higher
velocity distribution over the test wing. Thom’s wake blockage correction equation (eq. 6.2)
[136] was used in order to adjust the wind tunnel measurements of the test campaigns.
∆V
wb = = τ cdu (6.2)
Vu
where
c
h
τ= (6.3)
4
and c is the chord length of the test wing, while h is the spacing between wake sources. A
detailed description of the correction methodology and parameters used for the measurements
at the large wind tunnel facilities of H.F.I/TU Berlin can be found at the PhD Thesis of R.
Meyer [114].
6.1.2 Dynamic Measurements

Dynamic measurements were performed in order to investigate the unsteady and transient
effects of AFC elements on pitching wings. Such tests are much closer to real-life condi-
tions experienced by wind turbine blades. The test wing for the dynamic measurements was
equipped with AFC elements and a custom measurement and control software was developed
for this purpose.
The measurements for the dynamic investigations took place in the wind tunnel facilities
of TU-Berlin (HFI). The main wind tunnel setup remains identical to the one used for the
static measurements. The test wing configuration is mounted between the wind tunnel splitter
plates (Fig. 6.2) on the de-coupled 6 component wind tunnel balance [114]. The forces are
measured based on the strain gauge principle. The angle of attack of the airfoil is varied
by a stepping motor driven by a CNC-controller. The rate of AoA variation is in the range
of 2o /sec and the worm-drive transmission system between stepping motor and wind tunnel
test wing assures the precise AoA variation without gear-backlash. The forces are measured
through the measurement of voltage signals from the 6 component balance. The voltage
signal measurement is done with a 24bit analog digital converter NI 9219 made by National
Instruments. The communication of the AD converter with the computer is done with a
custom software developed with the LabView visual programming language.
A desktop PC with a Quad-Core Processor is responsible for the communication with
Figure 6.5: Basic schematic of the multi-core software implementation.
sensors and actuators. The main tasks are the voltage data acquisition from the AD-converter,
the calculation of the corresponding forces and velocities, the computation of a proper control
signal based on the implemented AFC element actuation strategies and the transmission of
the actuation signal it to the servo micro-controller (Fig. 6.5). To ensure a stable operation
of the overall system the aforementioned tasks are performed on separated synchronized loops
running on individual processor cores (multi-core approach). Thereby data are passed from
one loop to the other with the use of global variables.
Inside the loops at each iteration the following tasks were accomplished:
Data Acquisition Loop:
• Acquisition of voltage data from the ADC
• Storage of the data in a global variable
Angle of Attack Loop:
• Communication with the CNC-Controller, in order to change the angle of attack of the
airfoil
Control Loop:
• Read voltage data from the global variable
• Compute actual lift coefficient

• Compute control signal, based on the implemented controller
• Send control signal to the servo controller
User Interface Loop:
• Visualization of data
• Handling of User Events

Part IV
Initial Flow Control Solution

Selection
7 Methodology
The preliminary selection process is aimed at filtering the big bulk of proposed aerodynamic
flow control elements. The aerodynamic flow control solutions are selected based on literature
research and a total of 14 elements is listed. The preliminary selection filters out the elements
that are clearly out of the scope or specifications of the current work thus maintaining only
the elements which worth further investigation. A second (detailed) selection process takes
place to finally select the best-performing elements which could be implemented on a complete
wind turbine blade structure.
The proposed elements are:
1. Vortex Generators and Vortilons
2. Fixed Leading Edge Slat
3. Flow Vane
4. L.E. Protuberances
5. Gurney Flap and Active Gurney Flap
6. Rigid T.E. Flap
7. Flexible T.E. Flap
8. Flexible L.E. Flap
9. Split Flaps
10. Stall Ribs
11. Spoilers
12. BL Suction and Blowing
13. Zero Mass Flux and Synthetic Jet Actuators

7 Methodology 53
Figure 7.1: Schematic representations of all the elements included in the preliminary
flow control investigation.
14. Plasma Actuators
The current investigation includes an analysis of each aerodynamic element based mostly
on literature or preliminary simulations and measurements performed by the author. Each
7 Methodology 54
aerodynamic element is briefly described and its roles are defined. Furthermore an estimation
is attempted for the performance of each element with respect to the following parameters:
• Operation in harsh environmental conditions (e.g. ice)
• Operation without being affected by lightning strikes
• Simple, robust and reliable structure
• Lowest possible complexity and cost
• Operation for extended periods (6 − 12 months) without maintenance
• Ability to withstand mechanical loads during transportation or even better implemen-

tation on-site during wind turbine erection
In order to investigate all the performance requirements described above, the investigation
focuses on five different technical fields.
• Aerodynamics
• Mechanical/Electrical Structure
• Blade Integration
• Operation/Reliability
• Development and Implementation Costs
The following paragraphs analyze briefly all the above mentioned criteria for the elements
under investigation.
8 Passive Flow Control
8.1 Vortex Generators and Vortilons

The Vortex Generators as well as the Vortilons (fig. 8.1) are passive flow control solutions
that work under the same general principle. Both devices generate vortices that propagate
downstream enhancing the mixing of the boundary layer with the free flow. In this way the
boundary layer is re-energized and is able to remain attached on the airfoil contour despite
the adverse pressure gradients that develop at the pressure recovery region of airfoils. Vortex
generators were developed first by Taylor in 1947 [161] and by Bruynes in 1951 [23] for United
Aircraft Corporation. In their simplest form consist of vertical thin plates mounted normal
to the wing/rotorblade surface at a small angle with respect to the local inflow angle. They
essentially generate simple trailing vortices (similar to that of wings are hihg AoA) which
propagate downstream. They were first tried on wind turbines by Boeing at the 2.5MW
M od − 2 wind turbine in 1982 [152].
Figure 8.1: Vortilon and Vortex Generator configuration
Vortilons were invented by R. Shevell during the development of the commercial aircraft
DC-9 at Douglas Aircraft Corp. The patent for this aerodynamic element was filed in 1966
[58] and since then vortilons have been frequently used on commercial and military aircrafts
all over the world.
Aerodynamics
The purpose of vortilons and VGs is the creation of strong vortices able to stabilize the flow
over the suction side of an airfoil thus delaying stall. Their main benefit is the combination
of strong flow stabilizing effects with small parasitic drag penalty. To achieve the creation
of vortices at high AoA, vortilons have to be positioned at the pressure side extending in
front of the leading edge of the airfoil in order to intersect with the stagnation streamline at
high angles of attack. The outward flow component found on swept wings (and rotors) in
combination with the location of the vortilon initiates the formation of a strong vortex at the
inboard side of the vortilon which re-energizes the boundary layer of the airfoil by free stream
- boundary layer mixing.
As stall control elements vortilons are very effective and they have the benefit of low drag
and non-interference with the airflow at low AoA.
Mechanical/Electrical Structure
The mechanical structure of vortilons and VGs is very simple since these elements are passive
and permanently fixed at the pressure side of the airfoil. The selection of design, manufac-
turing process and materials depends upon the scope and specific details of each project but
generally these parameters are well understood.
Blade Integration
The process of integrating vortilons or VGs on existing wind turbine blade structures is simple
since virtually no modification on the current blade production is needed. The elements can
be attached to the outer surface of the blades during production or even at the field as retrofit
solutions.
Figure 8.2: Vortilons shape and location according to Shevell [58]
Operation/Reliability
Since the elements are passive there are almost no issues regarding the reliability of the so-
lution. Considering also the fact that the aerodynamic and centrifugal loads on VGs and
vortilons are very small it is safe to assume that both operational characteristics and reliabil-
ity will be predictable and not problematic. The main reliability issue would probably be the
detachment due to bonding failure of the adhesive. However previous experience with such
bonds dictates that such an event can be avoided.
Costs
The production and integration costs of VGs and vortilons are very low due to the combina-
tion of passive element, simple structure and low aerodynamic and centrifugal load. In mass
scale the production cost of these elements can be dramatically reduced while the integration
cost in the field1 is also small.
8.2 Fixed Leading Edge Slat
Figure 8.3: Leading edge slat.
Aerodynamics
The slotted and multi-element airfoils (fig. 8.3)were initially introduced by Handley Page [124]
in Britain and Lachmann in Germany [79] but the first thorough and systematic study of these
elements was performed by Weick and Shortal in 1933 [177]. In their study Weick and Shortal
examined various combinations of slots and various slot locations aiming to identify the aero-
dynamically optimal configurations. For an un-flaped airfoil the most effective position for a
single slot is near the leading edge, while the effectiveness decreases when the slot is moved aft.
Multiple slots are relatively ineffective unless they include a slot near the leading edge [5, 177].
1
The integration of VGs and vortilons on the blade structures should be done in the field in order
to eliminate damages to the elements and the blade during blade handling.
Figure 8.4: Different combinations of flapped and un-flapped slotted airfoil studied
by Weick and Shortal [5, 177].
The general aerodynamic effect of the leading edge slot or the fixed slat is often misin-
terpreted as being similar to a flow duct configuration. In reality the operational principle
behind it is a multifaceted flow interaction problem. It involves the reduction of the pressure
gradient on the main airfoil in combination with the modification of the Kutta condition for
the fore part of the slotted airfoil (or the slat element) [79, 11]. This leads to higher lift values
at high AoA compared to the clean airfoil configuration as well as stall delay [177].
Drag penalties exist in the case of fixed slotted airfoils but they are not very high thus
maintaining a relatively high Cl/Cd ratio. Additionally in the case of wind turbines where
the operational AoA range is very small a slotted profile or a slat can be designed in such a
way that it has the minimum possible drag and the highest possible lift while the turbine is at
nominal operational conditions. Slats or slots can also significantly improve the performance of
the inner blade region, which is often stalled due to the limited structural twist of WT Blades.
In the case of passive fixed element the mechanical structure is reduced to the mounting ele-
ments of the slat on the existing blade. In case of a slotted airfoil the additional mechanical
structure is basically inexistent since there are no extra elements and the slot is just created
in the blade mold during the blade building process. In case of a fixed slat the mechanical
construction is slightly more complicated but still fairly simple. In the current study only the
passive slat solutions will be investigated since wind turbines operate in a quite narrow AoA
band where the said elements can be properly configured. Additionally, the loads imposed on
the slat structures are not really high during normal operation [90] or even during off design
operation.
Blade Integration
The fixed slats are easy to integrate in the blade structure since they only require simple
mounting points on conventional wind turbine blades. These elements could even be applied
on existing wind turbine blades in the field as retrofit solutions. The integration of slots on
wind turbine blade structure requires some modification at the production molds of the blades
but these changes can also be retrofitted to existing molds.
The integration of semi-active slats2 could also be designed in such a way that all the ac-
tive elements are located inside the slat structure requiring only the existence of a mounting
point on the wind turbine blade together with some sort of power supply (electric power, air
pressure e.t.c).
In the case of fixed slats or passive slots the operation of this configuration is not expected to
be problematic. Slats are not really prone to aeroelastic effects and the additional loads that
they impose on the blades can be calculated and handled by the blade structure. Naturally
reliability is high since these systems have no moving parts hence the main problem that they
could face would be damages due to erosion but these aspects already concern the existing
blade structures.
Costs
The cost of passive slat and slot elements is considered to be low due to the simplicity of the
solution. In the case of passive slotted airfoil the cost is almost zero since it only requires one-
time modifications to the mold structure. In the case of the slat however additional elements
have to be produced and placed on the blades on special mounting points therefore the cost
is higher relative to the slot solution. The effectiveness however of the slat configuration is
2
The slats are always deployed but are able to vary their AoA and possibly to translate relative to
the main airfoil
potentially higher.
8.3 Flow Vane
Figure 8.5: Schematic representation of the flow vane concept.
The Flow Vane concept (fig. 8.5) is a concept developed by the author. It is a relatively un-
explored solution which seems very promising as a means of power regulation and stall control
of wind turbines. The general configuration includes a conventional wind turbine blade which
is equipped with an additional aerodynamic profile of smaller chord over the main profile.
The distance between the profiles (gap) is approximately equal to the chord length of the
small upper profile.
The only examples of the state of the art which resemble this configuration are the biplanes
[119, 59, 143], the low speed canard aircrafts [34, 91] and the Ball-Bartoe Jetwing concept
[149, 95]. However while all of the aforementioned technical solutions bare resemblance to
the currently proposed Flow Vane, none of them is similar in realization and none of them is
applied to rotor wings.
Aerodynamics
The function of the upper profile (in its passive form) is to induce a down-wash at the trail-
ing edge region of the main profile and modify its circulation thus suppressing stall. This
would increase the effective AoA range of the main airfoil thus increasing the efficiency of the
inner part of conventional blades which currently operate in stall (or near stall) conditions
due to insufficient structural twist angles. Such a passive configuration would lead to higher
performance of the wind turbine blades thus improving the energy yield. The main variable
parameters of such a system would be similar to the performance parameters of biplane wings
and they would be [44, 119]:
• The Gap (The distance between the two airfoils)
• The Chord Lengths
• The Stagger (the horizontal positioning of the flow vane in relation to the main airfoil)
• The Decalage (the angle difference between the two chords)
• The Overhang (one half the difference in span of the blade and the flow vane)
The variation of one or more of these parameters would change significantly the aerody-
namic performance of the whole system.
The mechanical construction of the flow vane is fairly similar to the fixed slat construction. It
would be possible to manufacture such a flow vane element in a single-piece design or a modular
multiple-element concept. The flow vane element can be manufactured from composite, plastic
or metal material depending on the performance requirements and then design specifications.
The aerodynamic loading of the flow vane is low due to its position and size, therefore a stiff
load-carrying structure is not necessarily a requirement.
Blade Integration
The Flow Vane requires very few modification of the existing wind turbine blade structures,
since it can be easily attached to the blade surface during or after production. The required
modifications are simple and can be implemented in the current blade production process
with minimal modifications in the molds. In such a way the Flow Vane would basically be an
external element.
In terms of operation and reliability it is possible to do some good estimations regarding the
performance of the Flow Vane. The low aerodynamic loading and the passive (fixed element)
principle offer a high level or reliability to the final system. Like any other passive flow con-
trol element (i.e. fixed slat), the flow vane would require some minimal maintenance but this
would be limited to shape and structural inspections.
Costs
The cost of the Flow Vane is in the same range with the slat, even though slight modifications
of the current molds near the spar region would be required for the element mountings. Due
to the small chord and thickness of the flow vane, its manufacturing could be accomplished
in small molds and via automatized processes (e.g. injection molding), together with the
mounting struts thus reducing the the overall cost of the solution.
8.4 L.E. Protuberances
Figure 8.6: Leading Edge protuberances
The concept of leading edge protuberances (fig. 8.6) as a means of flow control is a con-
cept that exists in nature for thousands of years [63] and since the early 80’s it is extensively
investigated as a flow control means for aircrafts, helicopters and wind turbines. One of the
first applications of this concept was described by Boeing [145] in 1981 and included the
use of conical leading edge protuberances at the stabilator surfaces in order to increase its
effectiveness at high AoA. The same effect was also achieved by the multiple leading edge
protuberances described by Thompson [163] in 1982 mostly intended for applications with
swept wings. Another similar approach was proposed in 1993 by NASA [40] but in this case
the effect was the boundary layer re-laminarization via boundary layer protuberances. One of
the latest efforts in this field is represented by the proposal of DLR [84] in 2006 where small
sized spheres attached on the leading edge of an airfoil significantly delay stall of airfoils. One
of the first proposals for active protuberances was described by Schmidt [57] in 1998, where
the actuation of various protuberances would enable the active flow control over the airfoil.
Figure 8.7: The proposal of DLR involves small spheres attached on the leading
edge [83]
One of the most significant developments in this field however was initiated by the research
of Fish et al. in the mid 90’s [62]. During this research it was found that the irregular
leading edge protuberances on the flippers of the Humpback Whales were responsible for the
exceptional maneuverability of these mammals. These protuberances were responsible for the
prevention of stall on the flippers of the whales. The phenomenon was extensively investi-
gated and relevant patents were filled in the following years [175]. Extensive research efforts
followed since then, which might lead to significant design changes in existing airfoil structures.
Figure 8.8: Sinusoidal leading edge protuberances based on the Humpback Whale
flipper [2].
Aerodynamics
Of all the proposed variations, the design of leading edge protuberances based on the tuber-
cles of the Humpback Whale seems to be the prevailing aerodynamic solution in this field.
Miklosovic et al. [41] performed wind tunnel tests of 3D flipper models (with and without
tubercles) and noted significant stall delay, increased lift and reduced drag at high AoA for
the flipper with the tubercles. The increase in the stall angle during the aforementioned ex-
periments was in the range of 40%, while the increase in maximum lift was 6% while the drag
increased for a limited AoA range and decreased again at high AoA (reduced pressure drag due
to flow reatachement and reduction on the low pressure separation volume). Further research
results were provided by Levshin et al. [2] for the effect of amplitude and wavelength of the
sinusoidal leading edge airfoils by investigating 2D wings based on wind tunnel investigations
on a NACA 634 − 021 profile. The results showed that the increased amplitude of the protu-
berances reduced the maximum lift value of the airfoil while creating a very smooth stall and
post-stall region. The effect of the wavelength is not so critical and the reduced wavelength
mostly promotes a smoother stall behavior. The aerodynamic model created by Nierop et al.
[49] generally proves the previous experimental findings and provides an analytical method
which helps to explain the effect, but also to design applications with airfoils ecqupped with
leading edge protuberances.
The mechanical structure of the leading edge protuberances is generally simple and easy to
produce. Depending on the design and configuration of the solution there is a variation of
the mechanical complexity and from the production point of view, the sinusoidal leading edge
Figure 8.9: The sinusoidal protuberances create 3D vortical flow phenomena which
re-energize the boundary layer at the pressure recovery region this de-
laying stall [63].
appears to be the most complex of these solutions. In order to produce such a sinusoidal
leading edge (as an extra retrofit part or incorporated in the blade) it would be necessary
to build or modify a negative mold structure. The production however of such elements as
retrofit add-ons is generally also feasible. The existing production methods are capable of
achieving this design with high quality and relatively low cost. Since the purpose of the said
elements can be accomplished by passive elements with no or very limited drag penalties
during off-design operation, it is reasonable to focus the current investigation only to passive
elements.
Blade Integration
The attachment of the leading edge protuberances on existing blades or new blades in the
field3 would be a relatively easy process which could be performed with adhesives or mechani-
cal fixing elements. The elements would also be easy to replace in case of wear or damage and
since they would not be subjected to very high loads, their stiffness and subsequent weight
would not be high.
3
Except from the case of implementation of the protuberances at the blade mold structure, the
implementation of these elements on the blade should be done in the field in order to avoid
damages during blade handling process.
The operation of the leading edge protuberances according to the existing literature does not
cause aeroelastic effects or fluttering. Due to the passive nature of the configuration there
are also no control/automation issues therefore good reliability and long operational life is
expected. The main reliability issue that can be expected is de-bonding, mostly in the case
of retrofit solutions and in cases where the bonding material is exposed to UV radiation,
subjected to high deformation4 e.t.c.
Costs
The cost level of the leading edge protuberance solution in terms of production and integration
is generally similar to the cost of other relevant solutions such are the vortilons. These passive
elements have high potential for significant cost reduction if produced in large numbers. Con-
trary to vortilons however these elements are 3D morphed shapes following the overall blade
shape. Therefore their production would probably require some sort of large mold structure
thus slightly increasing the initial investment cost.
4
Contrary to the vortilon solution, the leading edge protuberances, could be implemented as large
leading edge “plugs” thus being forced to follow the bending of the main wind turbine blade
structure
9 Active Flow Control
9.1 Gurney Flap
Figure 9.1: Active Gurney Flap (or Micro Tab) for load alleviation
The principle of Gurney flaps and similarly the one of Micro Tabs (fig. 9.1) was probably
accidentally discovered (and forgotten) in the early days of aviation and later re-discovered
by Dan Gurney in the 60s [79]. The Gurney flap is a simple flat plate in the order of 1%
of the chord length, located perpendicular to the pressure or suction side of the airfoil at
the trailing edge. When properly sized, the Gurney flap increases the total lift of the airfoil
while reducing the drag [105]. The main role of this solution is performance increase and load
alleviation.
Aerodynamics
The initial systematic aerodynamic investigation of this element was performed by Liebeck
[105] in the 70s and it was found that improper sizing of a Gurney flap would lead to unfavor-
able drag increase without significant further lift benefits. Storms [157] found a lift increase
in the order of 13% for a Gurney flap size of 0.5%c with minimal or no drag penalties for low
and moderate Cl values. Mayda and van Dam [35, 167] based on the research of Bechert et
al. [48] investigated the effects of serrated and slit Gurney flaps (i.e. micro tabs) in order
to eliminate the 2D vortex shedding from the solid Gurney flaps which can cause vibration
and noise. The aforementioned also identified the optimal Gurney flap size as a persentage
of chord legth (which lies within the range predicted by the initial research of Liebeck [105])
but also the optimal spacing between the micro tabs.
Figure 9.2: Schematic representation of the trailing edge flow field of a typical trail-
ing edge configuration (top) and a trailing edge equipped with a Gurney
Flap (bottom) [120].
Additionally C.P. van Dam et al. [109] investigated the implementation of Micro Flap (i.e.
active Gurney Flap) and deployable Micro Tabs as means for load alleviation in wind tur-
bine blade structures. They found that both micro flaps and micro tabs are suitable for the
task of load alleviation mostly due to the fact that they can be rapidly deployed. The main
difference between these two configurations was the slight aerodynamic lag of the micro tabs
due to their position (more fore than the micro flap). The general behavior of these elements
however is similar to the behavior of the airfoils with divergent trailing edge [153] and only
the possibility of control adds functionality and versatility to the whole solution.
The actuating mechanism in the case of an active Gurney flap requires low actuation force
since the small size of the element leads to lower drag force on the Gurney Flap itself. Alter-
natively the implementation of sinking micro-tabs [109] could further simplify the actuation
process by implementing simple pneumatic actuation devices such as expanding tubes. Gen-
erally the actuating mechanisms for rigid flaps are simple and relatively cost efficient, however
the fact that the said devices can provide only load reduction and no significant power regu-
lation means that the integration of additional devices would be required thus demanding a
more complicated actuation and control system for the sum of the implemented AFC solutions.
Blade Integration
The integration of Gurney flaps and Micro tabs in the blade structure is a straight forward
process. These elements are very small in dimensions and their actuators are equally small,
therefore only minor changes need to be made at the current blade structures in order to
implement such devices. Especially in the case of Micro Flaps (i.e. active Gurney Flap), the
flap mounting point can easily be integrated at the trailing edge region of the blades and the
actuators could even be mounted externally without significant aerodynamic penalties for the
blade. The deployable Micro Tabs require some larger modifications but they can also be
integrated in replaceable modules thus reducing the integration complexity.
To achieve a significant load reduction during the operation of wind turbines a fast and reli-
able control and actuation system is needed. From the aerodynamic and mechanical point of
view Gurney flaps and Micro tabs are suitable for fast control and actuation.
Costs
The material cost of these elements is minimal and the cost of their actuators and power
units is also low, therefore the overall cost of the solution is relatively small. It has to be
noted once more though that these elements cannot achieve the complete control of the wind
turbine (contrary to larger flaps for example).
9.2 Rigid T.E. Flap (Aileron)
Figure 9.3: Plain rigid flap
The conventional rigid flap configuration (fig. 9.3) is one of the oldest concepts of airflow
control, therefore there is a vast amount of information available on the topic. Its use in
aviation is universal and it has been already tried in helicopter rotors [17, 60] and more
recently in wind turbine blades [138, 166, 30, 14, 167]. In both helicopter and wind turbine
cases there have been several research efforts that led to significant scientific results and
even prototypes, but no commercial product is yet available (as of 2011). The potential
however of this flow control solution is very high therefore the current topic is still under
heavy investigation. The main field of operation of such flap systems is power regulation and
the secondary is load alleviation.
Aerodynamics
The aerodynamic performance of rigid flaps of different design characteristics is generally
well understood. Since the 1930s and 1940s, detailed wind tunnel investigations were per-
formed in order to produce useful design data related to the operational characteristics of
T.E. flaps [5, 20]. Generally flaps vary the airfoil camber thus varying the lift coefficient of
the airfoil with a penalty for the airfoil drag [166]. Their effectiveness in wind turbine appli-
cations is considered to be high both related to power regulation as well as load alleviation
[138]. In addition to that, small flap deflections are able to overcome, up to a certain extend,
the early laminar-turbulent transition caused by the increased leading edge roughness of the
wind turbine blades due to contamination and erosion [150, 139]. The main aerodynamic
disadvantage of the rigid flap concept is the significant increase in drag which is caused at
deflection angles higher than 10o [5] due to flow separation at the low pressure side of the flap.
In terms of manufacturing the rigid flap is simple since, in its basic form, consists of a 2D airfoil
rotatably mounted on a hinge system on the main airfoil. The actuating mechanisms of the
flap elements can be either based on electromechanical drives, servomotors, hydraulic/pneu-
matic cylinders, pneumatic muscles [184] or even piezoelectric actuators [160]. Conventional
flaps are widely used in aerodynamic and hydrodynamic applications, therefore the actuating
mechanism technology is advanced and mature thus it is possible to estimate the performance
of most of these mechanisms. However the implementation of the mechanisms on wind turbine
structures introduces new performance requirements due to the operational characteristics of
wind turbines.
Taking the special operational parameters of wind turbines into consideration, the piezo-
electric solutions would be the most sensitive in lightning strikes. The mechanical complexity
as well as the existence of electric wiring would reduce the performance of the electromechan-
ical solutions. The generally unavoidable leakages of the hydraulic systems is also a potential
drawback, but their high reliability, the absence of electrical conductive wiring and high force
potential makes them an attractive solution. The pneumatic systems appear to be very at-
tractive, especially in the form of pneumatic muscles due to the relatively high force potential,
the elimination of the leakage problem1 , and the elimination of conductive wiring.
Blade Integration
The integration of rigid flaps in wind turbine blade structures is not very complicated in terms
of manufacturing. Several modifications in the mold structure are required in order to create
the necessary spaces which can incorporate the flaps. Additionally, the blade structure has
to be appropriately modified in order to ensure that the cut-outs for the flaps will not cause
stress concentration and potentially lead to local fatigue failure. Additional space2 has to be
1
Minor leakages do exist but they do not contaminate the system and the environment since the
working medium is air
2
Preferably protected from the elements as well as dust and ice.
arranged inside the blade shell in order to host the actuating mechanism. Finally, some type of
signal transmitting element (cable, air pressure tube, hydraulic tube) has to be implemented
in the blade structure.
One major issue regarding the operation of a wind turbine blade equipped with rigid flaps is
that of flap-wise bending. Conventional blades bend significantly due to the high bending mo-
ments and their considerable length. However the implementation of a rigid flap element with
a linear rotation axis would not tolerate high span wise deflections. One possible solution for
this problem is the reduction of the flap span up to the point that the local wind turbine blade
deflection is within the tolerances of the flap mechanism. Another possible solution would
be to implement a span-wise flexible material/design for the flap structure thus enabling the
differential flap deflection along the span of the flap.
The reliability of the flap solution strongly depends on the actuating mechanism but if a reli-
able and proven actuating concept would be selected then the reliability of the whole system
would most probably be high. The replace-ability of each individual flap module can be a
relatively simple task which can be performed at the field.
Costs
The development cost for flap integration in wind turbine blade structures would be relatively
low due to the considerable technical experience with flaps in aviation. The flap system it-
self however would add cost to the overall wind turbine blade cost due to added complexity,
increased amount of components, additional control devices e.t.c. The flaps have generally
large effect on wind turbine aerodynamics and achieve power regulation and as well as load
reduction, but they are expected to add considerably to the cost of the turbine. The main
chance of implementation of such systems would probably prove to be cost effective only in
the case of elimination of the current pitch system.
9.3 Flexible T.E. Flap

The idea of flexible flaps (fig. 11.4) and their extension which is the fully morphing wing
goes back to the beginning of aviation [87] and is related to the investigation of the morphing
Figure 9.4: Flexible flap
behavior of the bird wings. Since then, continuous efforts have been made in order to im-
plement the concepts found in birds to man-made aerodynamic structures. One of the first
efforts in this direction was by H.F. Parker [126] in 1919 who proposed a chord-wise flexible
wing with variable camber. Many more efforts in this direction took place the following years
[106, 134, 132] and many companies, universities and research institutes are currently active
in research and development in this field [167].
Aerodynamics
The primary motive for altering the airfoil geometry is to improve the airfoil efficiency during
off-design operation as well as to offer better control of the generated lift and drag. From the
aerodynamics’ point of view, flexible flaps increase the camber of the airfoil thus modifying
the Kutta condition for the flow and the circulation of the airfoil [87]. The feasibility of the
utilization of modern flexible flaps for the control of the airflow has been investigated exten-
sively by many aerospace companies for the implementation of such techniques in helicopter
rotors [159] as well as in aircraft wings [154, 155, 116, 104].
The implementation of flexible flaps in wind turbines is also a topic under extensive inves-
tigation. Almost all the efforts of wind industry in the field of aerodynamics are related to the
development of the so called ”Smart Blades”, which will be able to offer better performance,
reliability and control of wind turbines. From the big bulk of research projects on the topic
of wind turbine rotorblades, a great amount of research is focused on the flexible flap concept
[138]. Currently however most of the research efforts are focusing on the implementation of
flexible flaps for load alleviation during wind turbine operation [14, 168, 9] rather than rotor
Figure 9.5: Comparison between the pressure distribution of an airfoil with a de-
flected rigid flap and a deflected flexible flap [146].
stall control [30] or even more the complete wind turbine power regulation which is the ulti-
mate target of the current work.
The mechanical realization of the flexible flap can be technically achieved in various ways,
many of which have been already proposed in the course of time through publications, pro-
totypes and patent applications. From the bulk of relevant solutions the most extensively
developed and tested are:
• The multiple link design mechanism of DLR [116] comprising a number of hinged links
covered by a flexible “skin”.
• The belt rib concept developed by DLR and others [97, 31].
• The mechanical/hydraulic concept of EADS [132].
• The sliding linkage concept of SAAB [18].
• The serrated rib mechanism developed by Boeing [22].

• The SMA based system, developed by NASA and DARPA [92].
• The piezo-actuated flexible flaps developed by RIS and TU Delft [160, 158].
Figure 9.6: Wind tunnel test wing equipped with flexible flap tested at the HFI/TU
Berlin wind tunnel.
Blade Integration
The integration of Flexible Flap modules at the wind turbine blade structure is generally
similar to the integration of plain rigid flaps. However in the case of flexible flaps there is no
need for implementation of rotating shafts mounted at the sides of the flaps (like the ones
usually used for mounting plain flaps). Therefore it is possible to produce the flexible flaps in
modules which attach to the blade only via a single connecting surface. In this way a largely
conventional blade structure with continuous truncated trailing edge could be equipped with
plurality of flexible flap modules without significant modifications.
The successful operation of the flexible flap concept has been demonstrated during various
research projects in the past. In the field of wind energy recent investigations of Barlas et al.
[169] also prove the fact that flexible trailing edge mechanisms can be effectively used at least
for load alleviation purposes. The reliability and shape accuracy issues [93] though remain a
weak point of these systems, mostly due to the fact that the materials and actuators used are
mostly prototypes and not commercial products. The final mechanical solution however of
each approach as well as its control and operational characteristics will define the operational
and reliability standards.
Costs
To produce a precise cost calculation for the flexible flap solution is naturally impossible since
all the available examples of this solution are prototype structures intended for research pur-
poses, however it can be estimated that the development cost of such a solution is quite high
mostly due to the technical challenges related to actuator technologies and material develop-
ment. The implementation costs however as well as the serial production costs would be low
due to the simple modular construction possibilities as well as the small number of moving
parts.
9.4 Flexible L.E. Flap
Figure 9.7: Flexible Leading Edge.
The first efforts to produce airfoils with variable leading edge camber (fig. 9.7) or leading
edge hinged flaps date back to the early 1920s [72] and they generally coincide with the various
proposals for variable camber airfoils of that time [125, 77]. Even though aerodynamics and
aviation in general advanced significantly, the idea of leading edge flaps and deformable leading
edge structures remained. It was however mostly “realized” through research projects, patent
documents or prototype structures [55]. The examples of the various proposed concepts and
ideas are numerous and vary from semi hinged leading edge flaps with mechanical actuators
[37, 38, 111], to inflatable leading edge flaps [180] and even elastic deformable leading edge
with sliding internal shape mechanism [133]. Recently, even SMA (Smart Material Actuators)
were proposed for the deformation of the leading edge of airfoils [54] but the development of
such solutions seems to be currently in slow progress. In the wind energy industry more or
less similar concepts were proposed, such as the leading edge flap of Coleman [53] but until
now none of these aerodynamic devices has been used in small or large scale production on
wind turbine blades.
Aerodynamics
The general aerodynamic effect of the leading edge flaps and variable camber leading edge
devices is the increase of maximum lift and the delay of stall [5] due to camber variations
and modification of the relative position of the stagnation point relative the the leading edge
[76]. There are experimental investigations in this field but due to the fact that the slat
configurations are used almost universally, the leading edge flap and deformable leading edge
configurations are currently investigated to a smaller extend.
The actuating mechanisms for leading edge flap systems can be of various designs and working
principles. As it was mentioned at the beginning of the paragraph there are numerous patent
documents with different actuating configurations. From the operating principle point of
view, the leading edge flaps without hinged connections (deformable leading edge) and with
simple actuating mechanisms are the most favorable for wind turbine applications. The
most important characteristics of such an actuating mechanism should be the outer skin
shape accuracy and control, the simple, reliable and low cost construction and if possible the
elimination of electrical cables and complicated control electronics from the blade structure
for increased resistance against lightning strikes and the weather elements.
Currently there are no commercial solutions for such leading edge flap structures suitable
for direct implementation on wind turbine blade structures. It is however estimated that the
development of such solutions would be reasonable in terms of complication since the basic
principles of operation are well established and well understood.
Blade Integration
The main challenges for the integration of flexible leading edge systems in wind turbine blade
structures are:
• The creation of modular and replaceable flexible leading edge systems with variable
leading edge dimensions in order to be distributed along the blade leading edge region.
• The integration of the flexible leading edge systems in such a way that they do not
cause early boundary layer transition, which leads to lower performance.
• The design and production of the flexible leading edge systems in such a way that they
are able to withstand the continuous span-wise bending of the wind turbine blades.
Given the fact that the simple and reliable actuators discussed above are available, the flexible
leading edge solutions are particularly promising. The small deflections (−10o to +10o ) and
the simple actuation and control strategies required would further enhance reliability. The
large aerodynamic influence of the flexible leading edge and the simplicity of control increase
the functionality and the performance of the solution as a power regulation and also as a stall
control means.
Costs
The cost level of the flexible leading edge is generally comparable to the one of the flexible
flap solution since both the principle of operation and the actuating mechanisms fall in the
same general category.
9.5 Split Flaps

The split flap configuration (fig. 9.8) is one of the simplest and oldest [15] flap configurations
both aerodynamically and in terms of actuation. The aerodynamic behavior is comparable to
the behavior of the simple rigid flap and its effectiveness is derived from the large increase of
camber and in some cases (translatable split flap) from the effective increase in wing chord [5].
Aerodynamics
The Cl/Cd curve of an airfoil equipped with split flap is similar to that of an airfoil with a
plain flap at the low Cl region while at the high Cl region the split flap is superior to the
plain flap since it produces a lot more lift but not as much drag3 . Due to the fact that one
surface (usually the suction side) remains unchanged, the activation of the split flap creates
a divergent trailing edge which acts as a bluff body increasing drag [5]. This means that an
3
This is true only for small split-flap deflections.
Figure 9.8: Simple Split Flap configuration (top) and Double Split Flap
(SplitF lap − Aileron) configuration (Bottom)
airfoil equipped with a split flap would be able to operate in permanent high lift state at the
inner part of a wind turbine rotor, where drag is not of such high importance [153, 165, 80],
thus being able to regulate the lift of that part of the rotor. The implementation of the split
flap concept at the outer portion of a wind turbine rotor would achieve significant lift control as
well as drag control, thus sufficiently controlling the rotor thrust4 . For a negative lift control
(turbine emergency shut down) a double split flap [176] would be necessary, or an inverse
split flap where the flap is located at the suction side. The aerodynamic benefits of a suction
side aileron and a pressure side split flap (double split flap configuration) would definitely be
aerodynamically superior and enable full wind turbine rotor control [24]. Additionally the
simultaneous counter-deflection of both flaps would create a form of aerodynamic brake thus
achieving the complete deceleration of the rotor. The possibility of drag increase decoupled
from lift could also lead to an edge-wise damping control system. Such a system would be
beneficial for the blade and rotor in-plane load control.
4
The implementation of a double split flap system would enable the simultaneous yet controllable
deflection of both flaps in such a way that they significantly increase the rotor drag without
increasing the torsional loads and cyclic vibrations, thus achieving an efficient but smooth rotor
deceleration.
Figure 9.9: Various operational configurations of the double split flap mechanism
[24].
The mechanical structure of the split flap systems is generally similar to that of the plain flaps.
In the case of simple split flap and due to the one − side deflection of the flap, the mechanical
linkage of the flap is actually simpler than the one found at plain flap systems. The estimation
of the author however is that the most effective and versatile implementation of the split flap
concept would be via the double split flap configuration (suction side aileron and pressure
side split flap), therefore in this case the mechanical structure is more complicated than the
one of the single split flap. The need for differential movement of each flap would demand
two actuating mechanisms and roughly double number of components compared to the plain
flap configuration.
Blade Integration
The integration of the split flap and the double split flap system on the wind turbine blade
structure is almost identical to the integration of the plain rigid flap. The spatial requirements
for the double split flap would be slightly higher due to the fact that double mechanical
components and actuators are needed, but generally the differences are only in the detailed
implementation design issues. An integration optimization could be performed so that single
split flap systems are incorporated at the inner rotor section (e.g. the inner 50% of the blades),
offering mainly lift control while double split flaps could be implemented at the outermost
region of the blades (e.g last 20% of the span) in order to offer precise lift and drag control
as well as load alleviation possibilities.
The split flap concept has the potential to provide significant control possibilities that would
benefit the wind turbine behavior in various ways including lift(thrust) control, load allevi-
ation, drag control e.t.c. In addition to that the simple actuating mechanisms required and
the utilization of standard commercial parts increases the operational flexibility as well as the
reliability and availability of the whole system.
Costs
The system development cost of the split flap system is similar to that of the plain flap system,
even though more control strategies can be developed for the former. The integration and
production costs are relatively higher than the plain flap system, but probably still within the
same cost range.
9.6 Stall Ribs

The first known implementations of deformable/inflatable membranes as airflow spoilers (fig.
9.10) are the aircraft air brakes described at the US Patent documents of Hunter in 1943
[81] and Campbell in 1944 [32]. The inflatable membranes were located at the pressure side
of airfoils (or even the aircraft fuselage) and their main purpose was drag increase. In 1961
Barber [13] proposed an inflatable spoiler at the suction side of an aircraft wind which was
intended as a lateral control device.
The next implementation of a rib at the suction side of an airfoil was proposed by Vought
Aircraft Co. [98] together with a tangential blowing type airflow control in order to increase
Figure 9.10: Leading Edge Elastic Stall Rib.
Figure 9.11: Elastic Stall Rib located at the suction side of an aircraft wind, oper-
ating as lateral control element [13].
the lift and effective lift AoA range of jet aircrafts. When the flow control system was not
operating, the rib functioned as a “smooth spoiler” by introducing large pressure gradients to
the local flow thus stalling the wing at an early AoA. Finaly, the latest implementation of the
inflatable rib was presented in a patent document owned by the British Government which
referred to the implementation of such an inflatable device on the suction side of transonic
airfoils for both stall and shock control purposes [56].
The implementation of the inflatable rib in wind turbine blades was first proposed by
Holzem in 1990 [78] as a means of stalling the complete wind turbine rotor in order to avoid
over-speed. The proposed system could operate either as an open system with an open mem-
brane and vertical air blowing (combination of pneumatic spoiler and deformable membrane
concept) or as a closed system with air-tight deformable membrane.
Aerodynamics
The first investigations of the aerodynamic behavior of surface span-wise ribs were published
by Jacobs in 1934 [89] but were more focused on small scale protuberances which were more
relevant to the leading edge icing issues and leading edge shape deviation problems of these
times. Later investigations by H.L. Dryden and by H. Schlichting [142] defined with higher
accuracy the critical Reynolds Number where a specific 2D protuberance would trigger tran-
sition due to its destabilizing curvature [67] and showed that incompressible flows are much
more sensitive regarding this issue than compressible flows. For protuberances higher than
the local roughness height and until a height where the flow detaches immediately 5 , the
protuberance height does not influence significantly the behavior of the flow [142]. For low
Reynolds numbers however, 2D surface protuberances can lead to improved flow behavior and
stall resistance as it was shown by Jacob [3] who found that large scale roughness (or smooth
ribs) improve the stalling characteristics of airfoils.
Generally it can be safely stated that the inflatable ribs have significant aerodynamic poten-
tial as flow control elements, however thorough investigations are required in order to analyze
possible, yet unknown adverse effects due to the implementation of the inflatable stall ribs on
wind turbine blades. The stall rib concept, however is effectively a stall control concept there-
fore its non-linear aerodynamic response makes it suitable mostly for power regulation. To
achieve accurate and load alleviation it would require the separate activation of a multitude
of span-wise stall rib segments which is unfeasible (with respect to cost-effectiveness).
5
The absolute height the protuberance on an airfoil for an immediate flow detachment depends on
the location and shape of the protuberance.
One of the main benefits of the inflatable stall rib is the simplicity of the actuation mechanism
and its control strategy. The system can be divided into multiple independent stall rib units in
order to increase the actuation precision, the effectiveness and the redundancy of the overall
system. Small separate units driven by independent control air pressure valves would also
enable the rapid deployment of the elements with very small deployment lag.
Blade Integration
The integration of such small elements on wind turbine blade structures is generally a very
simple procedure with small changes in the blade production process. However due to the
aerodynamic issues related to early transition, the tolerances related to the stall rib parts have
to be very tight thus increasing the production costs. The integration on the other hand of the
air pressure lines and the control system of the device is a simple and low cost process. Due
to the sensitivity of the rubber membranes to UV radiation, nature’s elements, temperature
fluctuations e.t.c, the modular stall ribs have to be designed in such a way that they can be
easily and replaced without high O&M costs.
The main benefit of the stall rib concept is the simple control and operation which does not
require complicated sensors as well as its high resistance against lightning strikes since there
is no need for electrical cable installation on the blade (only air pressure plastic tubing).
The low cost of the devices combined with their high effectiveness is another major positive
factor together with their rapid but also gradual operation (compared with traditional spoiler
solutions) which reduces the adverse aeroelastic phenomena.
On the other hand the sensitivity of the membranes to UV radiation, temperature fluc-
tuations, weather conditions, ice e.t.c as well as the shape deviations caused by the aging
of the rubber membrane are some of the critical issues need to be extensively investigated
for the development of this solution. Additionally it is important to minimize the negative
performance of the blade due to early transition triggered by manufacturing quality problems,
installation problems, membrane aging or dirt cumulation on the stall rib mechanism.
Costs
The development, production and integration costs of the stall rib solution are low mostly
due to its simplicity and small size. The maintenance costs are slightly high mostly due to
the fact that the rubber membranes need to be replaced after some time of operation.
9.7 Spoilers
9.7.1 Inclined Spoiler
Figure 9.12: Active Inclined Spoiler
The utilization of inclined spoiler devices as lateral control or roll control devices for aircrafts
has been long (fig. 9.12). Since the early years of motorized flight, many airplanes used
spoilers during landings in combination with flaps or during normal flight as means of flight
control. Due to the popularity of the said technical solution extensive research has been
performed, mostly from aircraft manufacturers and NACA [61]. However, many of the results
of the aforementioned research reveal that the performance of the spoilers during wind tunnel
investigations varies significantly from the performance during real life conditions [112] due
to the complex aerodynamic and aeroelastic phenomena produced by the spoilers. Therefore
extensive simulations and real life experiments are ideally required for the proper design and
development of a spoiler system. In wind turbine applications the inclined spoilers would
primarily be a means of power regulation.
Aerodynamics
The spoilers’ performance varies largely depending on Reynolds number, spoiler design char-
acteristics, spoiler deflection angle and spoiler-airfoil interaction [112, 182]. Generally small
spoiler deflections lead to more stable flows with less aeroelastic effects but are also less effec-
tive in terms of lift and drag variation. On the other hand large spoiler deflections are very
effective in terms of lift and drag control but introduce very large disturbances in the flow
which lead to large instabilities, reversed flows behind the spoiler, separation bubbles at the
spoiler hinge line and spoiler-airfoil interaction phenomena [112, 182].
The aeroelastic behavior of spoilers is a demanding research topic mostly due to the var-
ious parameters which can alter the investigation results. Extensive experiments [182] and
numerical simulations [112, 96] have shown that special design and control strategies can be
implemented in order to significantly reduce the adverse aeroelastic effects of spoilers.
The addition of spoilers on wind turbine blades is not a new idea since several relevant
patents and old publications are in existence [1, 102]. Contrary to the crude, more conceptual
proposals of the past, modern proposals [66, 139, 156] in the field mostly focus on more refined
designs intended to reduce the lift of the blades and regulate the power of the turbine while
eliminating or significantly reducing the structural loads on the blades.
To investigate the general performance of inclined spoilers at various deflection angles and
in combination with dedicated wind turbine profiles, wind tunnel tests were performed at
the H.F.I TU Berlin wind tunnel with the DU96W180 (fig. 9.13) test wing and their results
were generally in agreement with the existing literature. During the wind tunnel tests it was
observed that a spoiler with a chord of 15%c with its hingeline located at 50%c and in a
deflected position of 30o is able to reduce the lift coefficient of the profile at the design AoA
to zero thus being a very effective element for power regulation of the wind turbine [140].
Another significant issue has to do with the effect of spoiler deployment to the structural
loads of wind turbine blades. From the wind tunnel tests it was found that the spoiler
deployment introduces large pitching moments which could significantly increase the torsional
load of the blade structure [140, 139]. For that reason it is beneficial to locate the spoiler near
the main spar of the blade (around 50%c) thus reducing the spoiler torque arm.
The actuation of inclined spoilers is a relatively straight-forward process with low technical
risk. The actuation can be accomplished with various principles utilizing mechanical, elec-
tromechanical, pneumatic or hydraulic actuators which are readily available as commercial
products. The implementation of hydraulic or pneumatic actuators could eliminate possi-
ble problems with lightning strikes. It would also be possible to implement fail safe (spring
loaded) spoiler actuation solutions.
The main technical considerations regarding the mechanisms of a spoiler solution have to
do with the performance of the spoiler hinge and actuation system under blade deformations
and nature’s elements (heavy rain, icing, dust e.t.c).
Figure 9.13: Wind tunnel investigation results with comparative results between the
“clean” DU96W180 profile and the profile with inclined spoiler located
at 50%c and deflected to various angles [140].
Blade Integration
The integration of an inclined spoiler system in a rotor blade structure is generally not a very
complicated modification. However the implementation of the spoiler near the middle of the
chord of the blade, to reduce the twist moments during spoiler actuation introduces some
extra structural difficulties. The weakening of this structurally crucial area, due to spoiler
openings, is not desirable. Furthermore it is important to assure that the actuators of the
spoiler are well protected and that rain and dust cannot enter the blade inner structure. The
“packaging” is less of a problem though since the blade thickness around 50%c is generally
sufficient for the implementation of all the necessary hardware.
The main consideration regarding the operation of wind turbine blades equipped with inclined
spoilers is related to adverse aeroelastic phenomena. NREL6 researchers performed extensive
tests in the past with pultruded blades equipped with spoiler configurations [110] and found
6
National Renewable Energy Laboratory
that the creation of a feasible wind turbine blade design with spoilers is possible but requires
extensive research in order to prevent the initiation of flutter during the operational envelope
of the wind turbine.
In terms of reliability, the simplicity of the necessary actuators and the large existing
knowledge base in the field of inclined spoilers generally reduces the development risks.
Costs
The production and operation costs of the Inclined Spoiler solution are generally comparable
or even lower to those of the Split Flap, the aeroelastic considerations however require large
efforts for simulation and field testing during the development of the solution, thus raising
the overall cost.
9.7.2 Vertical Spoiler
Figure 9.14: Vertical Spoiler
The implementation of vertical spoilers (fig. 9.14) on airfoil structures is an airflow control
technique used since the early days of aerodynamics and aviation. Some of the early vertical
spoiler designs, such as the one proposed in 1935 by J.G. Lee [103] comprised sliding vertical
spoilers or rotatably emerging and hinged spoilers [172]. Later designs were based on airfoil-
like structures which would rotate 90o in order to operate as vertical spoilers (dive brakes)
(fig. 9.15) [173].
Figure 9.15: Vertical spoiler (dive brake) configuration proposed by H. Wagner in

1939 [173].
Vertical spoilers are also being used extensively in sailplane wings as “air brakes” (fig. 9.16).
The spoilers are still emerging from the wing but via a simple hinge mechanism which emerges
the spoiler and translates it in the span wise direction [121].
Figure 9.16: Vertical spoiler configuration implemented on sailplane wing [162].
Aerodynamics
Due to their extensive use in aircrafts, the aerodynamic behavior of vertical spoilers has been
extensively investigated since the early ’30s. One of the first extensive investigations was
performed by Jacobs in 1934 and was published in the form of NACA Report 446 [88] and
NACA Report 449 [89] and showed that vertical spoilers of even very small height (in the range
of 10mm) were able to considerably reduce the lift produced by an airfoil. The performance
of these elements is optimum when their location is at the suction side of the airfoil and near
the leading edge region.
Different spoiler shape variations were investigated extensively by NACA and published
in the form of the Technical Note 801 [181] and it was found that the flat type vertical
spoiler without gap between spoiler and airfoil is the most effective spoiler shape in terms
of lift influence. However in-house wind tunnel tests performed by the author showed that
flat shaped vertical spoilers create aeroelastic instabilities due to vortex shedding. Such
instabilities can cause strong vibrations as well as high noise levels. When the spoiler height
exceeds a certain limit, the aerodynamic instabilities lead to flutter with possibly dangerous
structural effects.
In terms of performance it was found during the aforementioned wind tunnel tests that a
15mm Vertical Spoiler located at 32%c of the DU96W180 test wing at 7o AoA (the design AoA
for this profile) can reduce the lift coeffilient to zero. Therefore the aerodynamic performance
of these applications is sufficient to achieve the conditions to fully stop the wind turbine rotor.
The mechanical structure of the vertical spoiler is relatively simple and comprises small num-
ber of individual parts. The spoiler actuators can be electro-mechanic, pneumatic, hydraulic
e.t.c. The vertical spoiler mechanism can be of the hinged or sliding type or any other suitable
concept. The main technical considerations regarding the performance of the mechanisms and
the actuators of the vertical spoilers have to do with the resistance of these mechanisms to
corrosion, water/ice and dust.
Blade Integration
The integration of vertical spoiler systems into current wind turbine blade structures is a
relatively simple process if the vertical spoilers are of the hinged type and do not require high
depth modifications of the current blade molds. In this case however the actuating mechanisms
and the hinge configuration require a larger volume inside the blade. The implementation
of the sinking/sliding vertical spoiler configuration, comprises only a plurality of slots on
the surface of the blade. In this case all the actuating elements and control devices are
implemented inside the blade occupying its empty inner volume. The benefits of such a
solution are apparent, but critical reliability issues arise since the inner part of the blade
becomes accessible to the elements (rain, dust e.t.c.).
Another major issue to be considered is the influence of the vertical spoiler integration to
the aerodynamic performance of the blade. According to the literature [181] and the in-house
wind tunnel tests [140] the optimum location for the vertical spoiler configuration in terms
of aerodynamic performance and ease of integration is the location of maximum thickness.
However most of the wind turbine profiles maintain a laminar flow until 35−45%c (i.e. behind
the point of maximum thickness) during operation at the design AoA. Vertical spoiler systems
could trigger early transition thus increasing the airfoil drag.
The main operation and reliability issues of the vertical spoilers are related to the aerody-
namic phenomena which are discussed above. Early boundary layer transition reduces the
power performance of the turbine. Possible flutter effects during spoiler actuation significantly
increase the fatigue loads of the blades with adverse effects on the blade and wind turbine
reliability. Additionally vertical spoilers do not allow a precise airflow control, but rather
a rapid aerodynamic stall control. Therefore the control system of vertical spoilers is more
suitable for power regulation at high wind speeds. Furthermore, this system is also suitable
as an emergency aerodynamic rotor brake.
Costs
The cost of the vertical spoiler solution is generally low due to the ease of integration and
the simple actuation mechanism. The control system of vertical spoilers is simple and due
to the small size of the spoiler elements and the low actuation forces, the actuators are also
small in size and cost. The result of all these characteristics is a system with relatively
low development and implementation cost. The maintenance cost, assuming that reliable
actuation and blade integration solutions are implemented can also remain low. This makes
vertical spoilers attractive flow control elements for wind turbines.
9.8 BL Suction and Blowing

9.8.1 Blow-Type Flow Control
Figure 9.17: Tangential Leading Edge Blowing
The idea of boundary layer flow (fig. 9.17) control was initially investigated by L. Prandtl
[137] and was focused on the boundary layer suction technique (See also paragraph 9.8.2)
but many researchers since that time have investigated both the blow-type as well as the
suction-type flow control. The idea of injecting a high velocity air stream in order to vary the
flow characteristics of airfoils and wings is well established since the 20s and 30s. Numerous
design concepts have been proposed during the last 80 years and many of them have been
extensively investigated [115, 167].
Despite the complexity and high cost of this flow control solution, a number of aircrafts
in the past have been equipped with such systems in order to delay stall and reduce their
landing speed and required landing rolling length. The Naval version of McDonnell Douglas
F-4 “Phantom” and the Blackburn Buccaneer are two representative examples [151]
Aerodynamics
Injecting high energy air into the boundary layer produces an increase in the stalling angle
of attack and maximum lift coefficient by delaying boundary layer separation. Boundary
layer control by mass injection (blowing) prevents boundary layer separation by supplying
additional energy to the exhausted boundary layer. Therefore injecting a high velocity air mass
into the air stream essentially tangent to the wall surface of the airfoil reverses the boundary
layer friction deceleration and the boundary layer separation is delayed [151, 142, 79].
Figure 9.18: Velocity distribution in the boundary layer directly behind the slit for
tangential blowing (A) and right after the blowing point (B) [142].
The effectiveness of wings can be greatly improved using blow-type flow control. Further-
more if the intensity of the blown jet is high enough, lift augmentation is achieved (fig. 9.18).
In extreme cases even the lift predicted by potential theory can be surpassed (i.e. the “jet
flap effect”) due to the initiation of “super circulation” [142]. Stream-wise blowing however
requires large amounts of air and energy, reducing the overall benefits of the flow control so-
lution itself. A more recent and promising blow-type flow control concept is the counter flow
fluid injection which is able to exert high-authority control to global flows using low energy
modifications to critical flow regions. In this case the air blow slit is located at the pressure
side near the leading edge stagnation point location and the control air-flow is directed tan-
gentially to the surface but with a forward direction. During the operation of such a flow
control system two different effects are present. One effect (“Boundary Layer Enhancement”)
is caused by the increased turbulence levels away from the wall region that acts to transport
higher-energy outer flow into the wall region. The effect of “Virtual shaping effect” is also
utilized to aerodynamically thicken the airfoil at high AoA. Both these effects help to delay
flow separation [16].
Figure 9.19: Schematic comparison of the conventional airfoil performance (left)

and the performance of the counter-flow concept (right) [16].
The mechanical structure of the blow-type flow control system is definitely very complicated.
All the solutions proposed so far increase dramatically the complexity of the overall system
by introducing various flow generation components (e.g fans, bleed-air valves e.t.c) as well
as flow ducts, flow distribution systems, nozzles and of course control valves. The air flow
consumption of a blow-type flow control system determines the size and performance of many
system components, therefore the design optimization procedures are essential to reduce both
the complexity and cost. In addition to that the operation pattern of the blow-type flow con-
trol system (e.g. continuous operation, intermittent operation, emergency operation) strongly
influences the design characteristic as well as the overall system complexity.
Blade Integration
The integration of multiple cavities, ducts, plenum and nozzles in wind turbine blade struc-
tures is a task with significant complexity. It requires major redesign of the production mold
structures and possible modification of the complete production process. Additionally the for-
mation of slits at the outer, load carrying, surface of the blade structure affects significantly
the structural integrity of the blade thus requiring further structural reinforcements.
The operation of the blow-type flow control systems is generally straight-forward since the
usual control elements are one or more flow valves (analog or digital) which activate and
regulate the operation of the flow control system. To achieve operational flexibility for wind
turbine operation, it is assumed that the blade would be equipped with plurality of slits
and respective control valves thus forming multiple independent systems. For installation
and maintenance purposes the valves could be located near the blade root region. That
option could induce significant delay to the control of the whole system due to the long
ducts between the valve and the surface slit. The existence of long actuation delays could
significantly increase the response time of the system thus removing the potential of rapid load
alleviation. If the valves would be located near each slit, the operational conditions would
be highly improved (elimination of delay) but the valves would not be easily accessible while
their electrical connection systems (for the valve control) would be prone to damages due to
lightning strikes.
Costs
The cost of the blow-type flow control solution is generally the main reason for it being
so unattractive in both aerospace and wind energy applications. The sum of the design,
production, integration and operation costs is usually so high that the flow control benefit is
completely eliminated.
9.8.2 Suction-Type Flow Control

The principle of suction-type flow control was first introduced in 1904 by L. Prandtl [137] as
a means to prevent flow separation from the surface of a cylinder. Since then numerous re-
searchers worked intensively on the development of that concept for several applications (but
mostly for aerospace purposes). The main focus of research in this field was the development
of wings with suction-type flow control concepts which would enable the sustained formation
of extended laminar boundary layer regions. This significantly reduces the skin-friction drag
with significant benefits in fuel consumption in the case of aircraft applications. NACA and
many other research institutes worldwide started systematic research in this field (fig. 9.20)
from the early 30s and continued mostly until the mid-90s [21].
Figure 9.20: One of the first test applications of the suction-type flow control was
realized by NACA at the wings of this Lockheed P-80 aircraft [21].
Aerodynamics
Figure 9.21: The basic aerodynamic principle of the suction-type flow control. The
“exhausted” boundary layer is removed through the slot thus enabling
the formation of a new boundary layer downstream of the slot [79].
The boundary layer looses kinetic energy due to skin friction phenomena thus its “ex-
hausted” molecules are prone to separation. By introducing one or more slots on the solid
wall surface and applying suction it is possible to remove the low energy boundary layer and
replace it with a “fresh” high energy boundary layer from the free flow [142, 115, 5] (Fig. 9.21).
The result of the “energetic enhancement” of the boundary layer separation delay and forma-
tion of extended laminar boundary layer regions. This leads to reduced drag and increased
effective AoA range due to separation suppression. Suction-type flow control also enables the
effective operation of airfoils with thick or highly curved trailing edge since the circulation
can be adjusted by the boundary layer suction system and the rear stagnation point can be
defined by a small sized flap (Fig. 9.22). In this way airfoils with elliptic or even circular
cross sections can generate very high lift [79].
The implementation of a suction-type flow control on wind turbine blades could lead to
significant drag reduction in combination with improved performance of the inner part of the
blades which operates at high AoA. The effective stall control over the blade would allow for
more effective designs with high lift airfoils to be produced.
Despite the fact that the aerodynamic potential of the suction-type flow control is significant,
the mechanical realization has difficulties and practical disadvantages which induce great
Figure 9.22: Elliptic airfoil with suction-type flow control and flap which defines
the Kutta-Joukowsky condition [79].
obstacles toward’s its commercial success. The suction orifices need to be of very small size
to avoid parasitic influence to the flow(early laminar-turbulent transition). The extremely
small size of the suction orifices however (Fig.9.23) induces a high risk of blockage due to
contamination, water infusion or ice formation.
Furthermore the manufacturing process which allows the production of such small orifices
on the airfoil surface is usually slow and expensive since it is normally based on CNC electron-
beam drilling technology [79, 21]. Furthermore the surface of the wing cannot be coated since
that would completely block the suction orifices and has also to undergo thorough and frequent
cleaning to maintain its performance characteristics over a long time. The high energy demand
of this concept and the cost of the vacuum pump system reduces even more its benefit.
Blade Integration
The integration of the suction-type flow control components to conventional wind turbine
blades is a quite complicated task due to the specific restrictions of the said configurations.
Due to the need for small orifices it is necessary to add a metal (or plastic) outer layer on
the blade which will include the suction openings. This layer cannot be painted and has to
be connected with the rest of the suction components (suction ducts, vacuum pumps e.t.c.).
Additionally wind turbine blades need to incorporate all the necessary suction ducts in their
structure, as well as vacuum pumps near the root region (or even inside the rotor hub). All
these actions increase the production cost of the blades and require major redesign of the
blade production process, blade molds and production tools.
Figure 9.23: The mechanical structure of the wing equipped with suction-type flow
control. [21].
The main problems of the suction-type flow control are related to the ease of blockage due to
contamination, water and ice formation. In such an event the flow control system is unable to
operate and is also unable to return to normal operation without the assistance of a separate
system (anti-icing, anti-blocking e.t.c). Therefore the operational reliability of the system is
potentially low. A possible failure of the flow control system on a wind turbine blade would
create large differences in lift and drag between the blades of the same rotor thus inducing
potentially hazardous vibrations and loads.
Another major disadvantage of this solution is the operational energy demand which is
very high even compared to other active flow control solutions (e.g. blowing-type flow con-
trol). This fact in addition to the generally increased complexity, maintenance difficulty and
potentially low reliability makes the suction-type flow control less attractive as a technical
option.
Costs
All the aforementioned disadvantages have dramatic influence in the overall cost. The me-
chanical and structural disadvantages increase the blade production cost and affect the han-
dling and transportation costs as well. Mechanical complexity together with the reduced
time between maintenance inspections (as well as frequent cleaning) significantly increase the
maintenance costs.
9.9 Zero Mass Flux and Synthetic Jet Actuators
Figure 9.24: Schematic view of a Synthetic Jet equipped wing
The development of synthetic jet and Zero Mass Flux (ZMF) actuators (fig. 9.24) for flow
control is a direct development of the “traditional” suction-type and blow-type flow control
techniques and the existence of synthetic jets is mostly due to the inherent disadvantages
of aforementioned “traditional” techniques relative to energy consumption, control fluid con-
sumption and overall cost. The main characteristic of synthetic jets is the intermittent (or
quasi steady) operation and the utilization of the surrounding fluid (air) for the creation of
the “zero net mass flow” operation. Various synthetic jet systems have been developed so far
and even more are currently under development for various applications ranging from spot
cooling [82] to airflow control [167].
The research regarding the implementation of synthetic jets to commercial airplanes, heli-
copters and wind turbines is active and more findings are constantly being published, therefore
the current document only includes a limited amount of synthetic jet concept proposals and
their aerodynamic effects as a means of ranking the overall solution with respect to its func-
tionality in the wind turbine aerodynamic control field.
Aerodynamics
The application of periodic jets for flow modifications on airfoils has been investigated by
many researchers in the past [4] and it was found that there exists a significant potential for
controllable lift increase in addition to stall control via this method. The main principle of
operation of synthetic jets and zero mass flux methods is the excitation of flow instabilities by
means of low amplitude fluctuations [52]. The proper placement of the flow control synthetic
jets can lead to an efficient flow control solution based on the virtual aero-shaping of the airfoil
[108, 107]. In such a way the geometric characteristics of the airfoil can be virtually modified
via the operation of the synthetic jet actuators thus modifying the aerodynamic performance
of the airfoil. Even more significant benefits can be gained if the system operates in a way that
achieves a proper coupling of the synthetic jet actuation frequency and the reduced frequency
(F + ) of the airfoil [183].
Another solution currently under investigation is the utilization of synthetic jet actuators
at the trailing edge region (tangential blowing) as means of circulation control [29] thus in-
creasing the airfoil lift. One more relevant concept is the “pneumatic Gurney flap” [42] which
utilizes the high pressure air-sheets created by synthetic jet actuators at the trailing edge
region in order to modify the Kutta condition, leading to lift variations. Finally synthetic jets
are also used as “pneumatic vortex generators” leading to stall suppression when activated
[27]. The concept of “pneumatic vortex generators” is currently under investigation by Gen-
eral Electric Wind Energy as a possible means of wind turbine flow control [33].
Various actuation mechanisms have been proposed for synthetic jet actuators and each mech-
anism is based on slightly different principles of operation. The most common mechanisms
involve piezo-electric actuators which cause periodic oscillations to an elastic membrane which
vibrates inside an air chamber (Fig. 9.25). The oscillatory motion of the membrane leads
to continuous suction and blowing cycles, forming the zero mass flux. The piezo-actuated
synthetic jets can achieve very high oscillation frequencies and are quite cost effective and
robust. The formation of very strong jets is hard to achieve, due to the limited displacement
of the vibrating membranes.
Synthetic jet pulse
Oscillating Membrane
Figure 9.25: Section view of an airfoil section with an oscillating membrane based
synthetic jet mechanism.
Similar configurations can also be used with electro-magnetic actuators [85]. The main op-
erational difference with the piezo-actuators has to do with the fact that the electro-magnetic
actuators can achieve much higher oscillatory displacements, allowing the development of
much stronger jets. Their weight is higher compared to the piezo-actuators and their size is
usually also considerably larger, but their ability to operate in a very wide range of frequencies
is a significant benefit.
Another category of synthetic jet actuators includes the passive (or quasi-active) actuators
such as the resonant cavities. Most of the relevant designs utilize an air pressure stream
(from the flow or from air pressure equipment) as the main flow and either a secondary flow
or a resonance effect for the initiation of pulsing flow phenomena. One of the designs oriented
toward’s wind energy applications is the passive system designed by GE Wind in collaboration
with the University of Stuttgart [27] (Fig. 9.27). This system utilizes a secondary flow to
switch the direction of the main flow. Such elements have very low production cost and the
absence of moving or electric parts makes them very attractive for wind turbine applications.
Their inability however to work in large frequency range and the lack of precise control are
some of their aerodynamic disadvantages.
Blade Integration
Depending on the actuation technology used for the synthetic jet mechanism as well as its
placement the blade integration properties change. If synthetic jet actuators are located near
the maximum thickness region (e.g. pneumatic vortex generators) then the blade integration
is relatively simple due to the ample space available at this region. If the location however
is near the trailing edge (i.e. pneumatic Gurney flap) then integration becomes a critical
parameter of the solution. One important issue regarding wind turbine blade integration of
synthetic jet actuators is the structural consequences of the surface discontinuities caused by
Figure 9.26: Schematic of the electro-magnetic Synthetic Jet actuator of Renault

as it was used at the prototype/exhibition vehicle Renault Altica [85].
Figure 9.27: Schematic of the passive Synthetic Jet actuator of GE Wind [27].
the synthetic jet slots/holes. Ideally the slots/holes could be formed in the laminate structure
without (or at least with minimal) interruption of the laminate fibers. This however is difficult
to achieve in reality therefore the most probable solution would be the integration of additional
parts in the blade structure thus interrupting or significantly changing the laminate structure
of the blade. Such insert parts would be harder to integrate at the trailing edge region, but
in this case a complete replacement of the trailing edge with insert parts could be a feasible
solution.
The existence of synthetic jet mechanisms inside the blade structure increases the complex-
ity of the blade as well as the required maintenance effort. For this reason the flow control
mechanisms (individual or integrated in inserted units) should be removable and replaceable
in such a way that their maintenance and replacement is fast and cost effective.
The high number of active flow control elements naturally increases significantly the com-
plexity and the maintenance requirements of each wind turbine blade. A very important
parameter for the viability of the solution has to do with the required maintenance time and
effort. A system that allows fast inspection and repair or replacement of the flow control
units would have very little impact on the overall maintenance and reliability of wind turbine
blades.
Costs
The main benefit of the synthetic jet solutions compared to other fluidic flow control solu-
tions (suction-type, blow-type e.t.c) is the low energy demand and consequently the reduced
complexity. Both these factors strongly affect the overall cost of the solution, therefore the
synthetic jet systems in general are more cost effective than other competitive fluidic flow con-
trol systems. Furthermore among the synthetic jet systems, the passive systems are definitely
less costly but often less effective than their active counterparts.
9.10 Plasma Actuators

A recent development in the field of aerodynamic flow control is that of plasma actuators (fig.
9.28) which were first developed in the late 60s by Velkoff and Ketchman [69]. Only in recent
years however (mid 90’s until today) did plasma actuators undergo considerable research and
development, thus the first commercial applications are yet to come into realization.
Figure 9.28: Wing with Plasma Actuator.
The basic principle of plasma actuators is the creation of electric field between two electrodes
by applying high voltage between them. The electric field that is consequently created induces
a 2D wall jet close to the wall surface. This is possibly caused by the impact forces of
the colliding plasma ions and the air molecules and particles in the actuator region. The
plasma induced jet adds momentum to the boundary layer and triggers flow instabilities thus
modifying the boundary layer properties [141, 167]. The main parameters influencing the
behavior of the plasma actuators are:
• Electrode Shape
• Electrode Size
• Dielectric Gap
• Electrode Distances
• Voltage
• Frequency (for AC actuators)
• Waveform (for AC actuators)
• Nature of dielectric wall
• Free-stream flow velocity
• Humidity
Currently there are several types of plasma actuators under development and each type
has its advantages and disadvantages which are described in further detail in the following
paragraphs.
Aerodynamics
Plasma actuators can be used in various types of flow control and flow modification applica-
tions depending on their type and positioning. The use of plasma actuators in an intermittent
mode allows the excitation of Tollmien-Schlichting instabilities of laminar flows and thus trig-
ger transition and achieve stall delay. Other types of actuators such as the plasma wall jet
actuators are able to create plasma sheets, vertical or at angle with the wall surface thus
achieving effects similar to vortex generators [141, 167]. Shear flows can also be manipulated
by plasma actuators via triggering Kelvin-Helmholz instabilities [68].
The application of these principles in airfoil flow control is currently under extensive in-
vestigation and the results for low and medium Reynolds numbers are positive, while the
effectiveness of these solutions at high Reynolds numbers is significantly reduced [167]. With
respect to wind turbine applications, plasma actuators are under extensive research [51, 39]
and their applications in this field seem to be promising. Apart from the apparent application
in place of the popular passive vortex generator solution, there is also the possibility to utilize
them as means of drag and vorticity reduction at the blade root region. Recent experiments
by Pechlivanoglou and Eisele [51] have shown that the existence of plasma actuators could
reduce the drag due to the Karman vortex shedding behind a bluff body and at the same
time generate lift. Such a bluff body is the cylindric root of wind turbines blades where the
application of plasma actuators is currently investigated [51].
The main configurations of plasma actuators are [167, 36, 45](Fig. 9.29):
• DC Surface Corona Discharge
• AC Surface Dielectric Barier Discharge
• Sliding Discharge
• Wall Jet
The general mechanical structure of these elements is very simple since in all configurations
there are no moving parts and the whole structure comprises at least two electrodes and a
Figure 9.29: Schematics and images of the different plasma actuator types. a)
Schematic of a DC corona discharge actuator, b) Visualization of the
effect of the actuator (a), c) AC Barrier discharge actuator, d) Visu-
alization of the effect of the actuator (c), e) Three electrode discharge
actuator, f) Visualization of the effect of the actuator (e), g) Wall jet
actuator with balanced electrode voltage, h) Wall jet actuator with
imbalanced electrode voltage[Source: [167]].
dielectric material. The mechanical construction is very robust, easy to handle and simple to
install. The production of the plasma actuators is a simple industrial process. The physical
size of these flow control elements is very small thus making them easy to install and to
replace.
Blade Integration
The significant advantages of the Plasma Actuators in the field of mechanical structure (e.g.
small size and robust construction) lead to easy and effortless integration on wind turbine
blade structures. The integration process involves a simple adhesion step where the plasma
actuators in form of stripes are glued on the blade surface. The only elements that need to
be properly integrated are the power cables of the actuators7 . The overall power requirement
of such systems is very low.
The ease of integration of the plasma actuators is also feasible during production. Handling
and transportation does not pose a problem for the reliability of plasma actuators. The
implementation of the actuators on site, just before or even after the wind turbine erection
would also be feasible. This would allow easier handling of the blades and reduction (or
elimination) of the damages to the plasma actuators themselves.
Due to the limited implementations of the plasma actuator technology in current products
and aerodynamic applications the available information regarding the reliability and operation
characteristics of such applications is virtually non-existent. A known issue with the plasma
actuator configurations is their sensitivity toward’s Reynolds number variations and more
specifically, their effectiveness reduction with the increase of the free-stream velocity [167].
Pechlivanoglou and Eisele [51] investigated the operation of plasma actuators on wind turbine
airfoils at a wide range of Reynolds Numbers and found that the effect of all the investigated
plasma actuators vanished at Re ≥ 105 .
The energy conversion efficiency is also another operational parameter, which highly de-
pends on the actuators design and undergoes further investigation. Currently plasma actuator
systems have low efficiency and very high thermal losses. Additionally the effects of environ-
mental conditions need to be further investigated to examine the possibility of a further
implementation of plasma actuators on wind turbine blades. More specifically the plasma
actuators would be required to operate effectively and reliably under rain, hail, ice, tolerate
dust and contamination, as well as lightning strikes.
Costs
Due to the experimental nature of the plasma actuators it is relatively hard to accurately
estimate the production, integration and operation costs. Current figures however show that
7
The thickness of the power cables of the actuators is small due to the high voltage - low current
configuration.
due to the simple mechanical construction and the ease of implementation, the initial costs are
low. Additionally the simple power requirements (High Voltage DC current or High Voltage
AC) the electrical/electronic configuration is also expected to be simple and inexpensive.
The only consideration has to do with the operational and replacement cost of the plasma
actuators and of course the frequency of replacement.
Element selection matrix

n
cs atio
mi ics egr
ent yn
a
an I nt tion
m rod ech de era st M
Ele Ae M Bla Op Co SU
VGs & Vortilons 2.0 4.5 5.0 4.0 5.0 20.5
Slat 4.0 4.5 4.5 4.5 3.5 21.5
Flow Vane 4.5 4.5 5.0 3.5 3.0 20.5
L.E. protuberances 2.5 4.0 5.0 3.5 4.5 19.5
Gurney Flap 3.0 4.0 4.5 4.0 4.0 19.5
Rigid T.E. Flap 4.0 5.0 4.0 4.0 3.0 20.0
Flexible T.E. Flap 4.5 2.0 4.5 3.5 4.0 17.5
Flexible L.E. flap 4.5 3.0 2.0 4.5 4.0 18.0
Multi-element Flap 4.0 2 4.5 2.0 2.0 14.5
Split Flap 4.5 4.0 4.0 4.0 2.5 19.0
Stall Ribs 3.5 4.5 4.0 3.0 4.0 19.0
Spoilers 3.0 3.0 2.0 2.0 2.0 12.0
BL Suction & 4.0 2.0 2.0 2.0 1.0 11.0
Blowing
Zero Mass Flux 3.5 4.0 3.5 4.0 4.0 19.0
Plasma Actuators 3.0 5.0 4.5 2.0 3.0 17.5
Part V
Experimental and Numerical

Investigations
10 Passive Flow Control Solutions
The initial overall investigation of the various PFC and AFC elements concluded to an elab-
orate rating system with various importance (weight) factors for each performance category.
This ranking system allowed the author to make a rational selection of the best performing
solutions. The best (overall) AFC and PFC solutions were selected for further numerical and
experimental investigations. This part of the document presents the most important results
of these investigations.
The experimental and numerical investigations of the selected passive and active flow control
elements were focused on wind tunnel measurements. Supplementary CFD simulations were
also performed mostly for the investigation of the flow field. The wind tunnel results were used
for steady state and dynamic wind turbine simulations using the codes that were developed
during the course of this work (QBlade and QBladeAE). The following paragraphs present the
results of the experimental and numerical airfoil and wind turbine investigations. Since the
operational regime of wind turbines blades is often extending towards the post stall angles,
the investigations presented in this thesis tend to focus more on the moderate and high AoA
region.
10.1 Vortex Generators

Vortex generators are a common and well established passive flow control solution for wind
turbine blades. By generating strong chord-wise vortices they enhance the boundary layer
mixing with the outer flow thus re-energizing the BL and delaying flow separation. Several
forms of vortex generators are used on wind turbine blades with slight differences in perfor-
mance. For the current thesis work the vortex generators used were based on the design of
TU Delft [170]. They are triangular vertical fins with a fin angle of 18o , a height of 1.5%c
and a distance between fin-pairs of 8%c.
Note: Some parts of the current research have been published in the form of a conference
proceedings’ publication described at [75]
10.1.1 Wind Tunnel Measurements

The VGs were tested on the AH93W174 wind tunnel test wing. They were tested at several
chord-wise locations in order to establish a clear operational pattern. Figure 10.1 presents
the performance comparison between the various VG configurations and the Baseline (clean
wing) configuration.
The VGs delay the stall thus increase the the maximum lift of the airfoil by aprox. 30%.
The stall angle increases from aprox. 12o to aprox. 23o . The drag increases at low AoA due
to the higher skin friction drag generated by the vortex generators. On the other hand at
high AoA the absence of separation due to the existence of VGs reduces the pressure drag.
The overall effect of the implementation of vortex generators is visible at the Cl/Cd graph
(fig. 10.1). The maximum Cl/Cd value is reduced (lower AoA range) while the high AoA
Cl/Cd ratio is increased.
The variation of the chord-wise position of the VGs also affects the overall configuration
performance. The closer the VGs are located to the trailing edge, the higher the drag of
the configuration at high AoA. On the other hand when the VGs are located away from the
leading edge, their performance at high AoA is significantly reduced. Figure 10.1 shows the
optimal position of the vortex generators for this airfoil, which is around 20%c.
10.1.2 Wind Turbine Simulations

The measured airfoil performance curves were imported into QBlade and were extrapolated
to 360 degrees AoA according to the methodology described at section 5.1.4. They were
integrated to a generic wind turbine blade design. The generic blade is a part of a generic
wind turbine configuration which is used for all the simulations in order to establish a common
comparative basis. Since the wind tunnel tests with the vortex generators were conducted with
the AH93W174 test wing, the generic rotorblade was mainly based on this airfoil. From the
structural point of view such a wind turbine blade would probably not be feasible. However
for the purpose of simplification and since there is no difference in the main aerodynamic
effects, such a blade was used for the following simulations.
The wind turbine blade with the vortex generators was tested against the baseline blade
configuration and the difference in performance and static loads has been identified (Fig.
10.2). The design characteristics of the vortex generator simulation are presented at Tab.
10.1.
The results indicate that the introduction of VGs leads to a power increase in the range of 2%
with a subsequent increase on the maximum root bending moment of about 0.69%. Based on
Cl 2
1.5
0.5
Airfoil: AH93W174
Baseline
VGs @ 10%c
0 VGs @ 15%c
VGs @ 20%c
VGs @ 30%c
-0.5
-10 -5 0 5 10 15 20 25 30 35
AoA
Cl/Cd
40
30
20
10
0 Airfoil: AH93W174
Baseline
VG2 @ 10%c
-10 VG2 @ 15%c
VG2 @ 20%c
-20 VG2 @ 30%c
-10 -5 0 5 10 15 20 25 30 35
AoA
Figure 10.1: Comparison between the baseline configuration and the various VG
configurations
a Weibull distribution with k = 2 and A = 9, the annual energy yield increases by 1.55% with
the introduction of vortex generators. This result is in agreement with previous industrial
experience and given the low cost of this solution it proves its viability.
Figure 10.2: Simulation results presenting the distribution of the Cl and the AoA
along the baseline blade and the blade with vortex generators.
Table 10.1: Blade design parameters and simulation parameters and results. Com-
the blades fitted with vortex generators.
Simulation Input Parameter
WT Nom. Capacity 2.5 MW
Rotor Diameter 88.9 m
Blade Length 43.45 m
Max Structural Twist 12o
Design λ 7.5
Nominal r.p.m 15 s−1
Cut-In Wind Speed 3.5 m/s
Cut-Out Wind Speed 25 m/s
Weibuhl K Parameter 2
Weibuhl a Parameter 9
“Baseline” Blade Design Parameters
Airfoil Section Position
Root Circle 1.2 - 2.7 m
Transition Airfoil 2.7 - 7.7 m
DU00W401 7.7 - 9.2 m
DU97W300 9.2 - 12.2 m
AH93W174 12.2 - 44.45 m
“Vortex Generator” Blade Design Parameters
Root Circle 1.2 - 2.7 m
Transition Airfoil 2.7 - 7.7 m
DU00W401 7.7 - 9.2 m
DU97W300 9.2 - 12.2 m
AH93W174 + VGs 12.2 - 44.45 m
Simulation Result Comparison
Max. Root Bending Moment 0.69 % increase
Energy Yield 1.55 % increase
Max. Rotor Cp 0.513 % decrease
10.2 Fixed Leading Edge Slat

The use of a fixed leading edge slat has great potential for the aerodynamic performance
improvement of wind turbine blades. Its performance was analyzed and optimized by means
of extensive wind tunnel investigations. The results of the wind tunnel measurements were
incorporated to the steady state wind turbine simulation code (QBlade). Supplementary CFD
simulations were performed for the investigation of the configuration flow-field.
proceedings’ publication described at [129]
10.2.1 CFD Simulations

Several slat configurations were simulated with the OpenFOAMTM CFD solver. The grid used
for the simulation comprised 1.3 million finite volume elements (see also Section 4.2) and the
SIMPLE steady state RANS solver was used. Figure 10.3 presents a view of the unstructured
grid used for the fixed slat configurations.
The simulation results agree with the established multi-element airfoil theory [146]. The
existence of the leading edge slat removes the suction peak from the leading edge region of
the main airfoil (Fig. 10.4). The discrete shear layer generated by the slat is transported
downstream over the main airfoil but does not mix with the airfoil’s boundary layer (Fig.
10.5). Figure 10.5 also shows that the flow separation region of the “baseline” airfoil is much
larger compared to the airfoil with the slat at the same AoA (18o on both cases).
The results of the CFD simulations are qualitatively very close to reality and capture the
main flow phenomena related to the operation of the slat. The quantitative results however
even though close to the wind tunnel measurements, were not accurate enough to be used for
the extraction of polar curves for the subsequent wind turbine performance simulations. The
main reason for that seems to be the selection and fine tuning of turbulence models as well as
the representation of the complex multi-element aerodynamic phenomena by the unstructured
grid. Figure 10.6 presents a comparative graph of the Cl curve of a configuration with fixed
leading edge slat. The values computed with the numerical solver are close to the results
obtained during the wind tunnel measurement but they are not accurate enough for use as
blade design input parameters.

For the investigation of the slat performance an auxiliary test wing was machined out of
ObomodulanTM . The wing, based on the NACA 22 auxiliary airfoil [5], has a constant chord
of 150mm and can be mounted on the DU97W300 test wing (Fig. 10.7). The mounting
Figure 10.3: View of the unstructured grid used during the CFD simulations of the
fixed slat configurations [129].
Figure 10.4: Pressure contour of the DU97W300 airfoil with the NACA 22 fixed
slat[129] .
Figure 10.5: Comparative view of the “baseline” configuration simulation in com-

parison to the configuration with a fixed leading edge slat[129] .
mechanism of the slat allows the variation of the position of the slat. This is necessary for
the parametric investigation of the slat performance at various positions relative to the main
test wing.
The current document includes only the comparison of the baseline configuration with the
best performing fixed slat configuration (fig. 10.8). Both configurations were measured at
1.35·106 Reynolds number and the results were corrected for solid blockage and wake blockage
including a chord correction for the slat configuration (for the wind tunnel corrections see also
Section 6.1.1). Figure 10.9 presents the performance comparison between baseline and slat
configuration with respect to the lift coefficient (Cl ) and the airfoil efficiency ( CCdl ). The slat
configuration clearly outperforms the baseline configuration for moderate and high AoA. The
generated lift with the addition of the slat is higher for the slat configuration for AoA values
higher than 5o . The stall is delayed and the maximum Cl is increased by approx. 50%. During
the wind tunnel tests and due to the high test section blockage it was not possible to obtain
reliable experimental values for angles of attack higher than 23o . Up to this angle no clear
Clmax was obtained, it is therefore possible that the maximum lift value is higher than the
one registered at the current graph.
The efficiency (glide ratio) of the slat configuration is slightly lower than the “baseline” con-
figuration for low and moderate AoA. This is due to the higher overall drag due to additional
slat element. However the significant lift increase and stall delay at high AoA leads to reduced
2.5
Cl
1.5
0.5
0 DU97W300 + Slat
OpenFOAM
Wind Tunnel
-0.5
-5 0 5 10 15 20
AoA
Figure 10.6: Curve of the lift coefficient for a configuration with a slat. Comparison
between experimental and numerical results[129] .
drag (lower pressure drag) and higher lift thus higher overall efficiency. The glide ratio for
high AoA is significantly improved. Considering the fact that the inner region of wind turbine
blades usually operates at high AoA [73] the addition of a fixed leading edge flap would po-
tentially increase the overall performance of rotorblades and the energy yield of wind turbines.

To investigate the effect of the fixed leading edge slat on wind turbine blade performance,
steady state BEM simulations were performed with QBlade. A conventional wind turbine
blade design was compared with a modified blade design that incorporated a fixed leading
edge slat at its inner region. The main parameters of this comparative simulation are presented
in Tab. 10.2.
The addition of the fixed leading edge slat improved the performance of the inner blade
region. The maximum rotor Cp was increased by 1.5% while for lower wind speed the Cp
Figure 10.7: Wind tunnel test configuration comprising the DU97W300 main airfoil
and the NACA 22 fixed leading edge slat[129] .
increase was even higher (Fig. 10.10). The energy yield was therefore increased by 0.5%. On
the other hand, the higher generated lift leads to a slightly higher (1% increase) root bending
moment.
Figure 10.11 shows the lift and AoA distribution along the baseline blade and the blade
with the fixed leading edge slat. The leading edge slat significantly increases the lift generated
at the inner blade region. This lift increase modifies the momentum balance of several blade
annuli of the BEM equation. The resulting effect of this on the local axial induction factor
leads to an AoA reduction due to the existence of the fixed leading edge slat.
Overall the addition of a fixed leading edge slat has the potential to increase the performance
of wind turbine blades with low cost and low complexity.
Figure 10.8: Schematic representation of the slat position with respect to the
DU97W300 main wings.
the blades fitted with fixed leading edge slat at the root region.
Blade Length 37.5 m
Max Structural Twist 15.13o
Design λ 8
Cut-Out Wind Speed 20.0 m/s
Weibuhl k Parameter 2
Weibuhl A Parameter 8
Root Circle & Transition Airfoil 0 - 2.65 m
DU97W300 2.65 - 14.0 m
AH93W174 14.0 - 37.5 m
“Slat” Blade Design Parameters
Root Circle 0 - 2.65 m
DU97W300+NACA 22 Slat 2.65 - 14.0 m
AH93W174 14.0 - 37.5 m
Max. Root Bending Moment 1% increase
Energy Yield 1.5% increase
Max. Rotor Cp 1 % decrease
2.5
Cl
1.5
0.5
Main Airfoil: DU97W300

0
Slat Airfoil: NACA 22
Cl baseline
Cl Slat
-0.5
-5 0 5 10 15 20 25
AoA
30
Cl
25
20
15
10
0
Main Airfoil: DU97W300
Slat Airfoil: NACA 22
-5 Cl/Cd baseline
Cl/Cd Slat
-10
-5 0 5 10 15 20 25
AoA
Figure 10.9: Comparison between the baseline configuration and the slat
configuration[129].
0.6
Cp
ΔCp
Betz's Limit 0.3
0.5 Baseline blade
Blade with slat
0.4 Cp difference
0.2
0.3
0.2
0.1 0.1
-0.1
0
-0.2
2 4 6 8 10 12 14 16 18 20
V [m/s]
Figure 10.10: Comparison between the rotor Cp curve of the baseline configuration
and the slat configuration for the DU97W300 airfoil. (Simulation
results)
Figure 10.11: Comparative AoA and Cl distributions (simulation results) across a

rotorblade with and without a fixed leading edge slat.
10.3 Flow Vane

The flow vane concept was invented and developed by the author as an alternative to the
fixed leading edge slat solution. The principle of the solution is the use of a thin airfoil with
a chord smaller than 20% of the main airfoil chord over the main airfoil. The auxiliary airfoil
(flow vane) is placed at a distance over the main airfoil and generates a discrete shear layer
that prevents the boundary layer of the main airfoil from separating. The purpose of the flow
vane is not to generate additional lift but to prevent the flow from separating from the main
airfoil.
This passive flow control solution was investigated by means of two wind tunnel test cam-
paigns with two different test wings. For this reason no further CFD simulations were con-
sidered necessary.

The test wings used for the flow vane experiments were based on the DU96W180 and the
DU97W300 airfoil (Tab. 6.2). The DU96W180 wing was equipped with a full span flow
vane element which was constructed from sheet metal based on the camber-line of the NACA
22 airfoil. This was done in order to investigate the effect of the flow vane camber on the
performance of the configuration. The flow vane element had a chord length equal to 15% of
the main airfoil chord. The DU97W300 wing was equipped with a full span constant chord
flow vane element modeled after the NACA 22 auxiliary airfoil. The chord of this flow vane
element was equal to 20% of the chord of the main airfoil.
Figure 10.12 presents a comparison of the performance of the two test wings (DU96W180
and DU97W300) in “clean” configuration at 1.35 · 106 Reynolds number. The curves show
a very similar trend for the lift between the two test wings. The drag of the thicker airfoil
(DU97W300) is naturally higher due to its increased thickness.
The flow vane elements were tested in several positions and angles relative to the main
airfoil. For space economy however only 2 positions of the flow vane will be displayed in
the current document (Fig. 10.13). The addition of the flow vane on the DU96W180 test
wing significantly improves its lift performance throughout its operational angle range. At
moderate AoA the linear part of the lift curve is “shifted upwards”, a performance behavior
typical of the trailing edge flap configurations. At the same time the stall is delayed and
thus the Clm ax increases by aprox. 40% and the stall angle by aprox. 5o (Fig. 10.14). The
additional drag of the flow vane in combination with the pressure interaction between the
main airfoil and the flow vane element cause a drag increase. Consequently the efficiency
(glide ratio) of the configuration with the flow vane is reduced by aprox. 60%.
Cl
1.5
0.5
-0.5
(Re=1.4x106)
-1
DU96W180 Baseline
DU97W300 Baseline
-1.5
-25 -20 -15 -10 -5 0 5 10 15 20 25
AoA
Cd
0.4
(Re=1.4x106)
0.35 DU96W180 Baseline
DU97W300 Baseline
0.3
0.25
0.2
0.15
0.1
0.05
0
-25 -20 -15 -10 -5 0 5 10 15 20 25
AoA
Figure 10.12: Comparative Cl and Cd curves for the DU96W180 and the
DU97W300 measured at the wind tunnel of HFI TU Berlin.
Figure 10.13: Schematic representation of the flow vane positions over the
DU96W180 and DU97W300 main wings.
Since the flow vane is primarily a stall delay feature its most effective placement is near
the rotorblade root where the operational AoA is usually very high. The flow vane concept
was therefore tested on the DU97W300 test wing which consists of a typical blade-root airfoil
section. The wind tunnel measurements showed a significant shift of the lift curve toward
higher Cl values for moderate AoA. Furthermore the maximum lift coefficient was increased
by approx. 50% and the stall was delayed by aprox. 10o . The drag of the configurations
with flow vane was increased therefore the glide ratio was reduced (Fig. 10.15). The first flow
vane configuration showed significantly less drag compared to the second configuration. This
is due to the higher distance of the flow vane element from the suction side of the main airfoil
and the subsequent reduction of the pressure interaction between the two elements. The flow
vane configuration seems to have a consistent performance pattern when applied to various
airfoils.
Cl
2
1.5
0.5
-0.5
DU96W180 (Re=1.4x106)
Baseline
-1 Flow Vane Position 1
Flow Vane Position 2
-1.5
-20 -15 -10 -5 0 5 10 15 20 25
AoA
Cl/Cd
60
50
40
30
20
10
-10
DU96W300 (Re=1.4x106)
-20
Baseline
-40
-25 -20 -15 -10 -5 0 5 10 15 20 25
AoA
Figure 10.14: Experimentally measured Cl and CCdl curves for the DU96W180 test
wing equipped with the flow vane.
Cl
2.5
1.5
0.5
DU97W300 (Re=1.4x106)
Baseline
0
-0.5
-5 0 5 10 15 20 25
AoA
Cl/Cd
30
25
20
15
10
0
DU97W300 (Re=1.4x106)
Baseline
-10
-10 -5 0 5 10 15 20 25 30
AoA
Figure 10.15: Comparative Cl and Cl /Cd curves for the DU97W300 measured at
the wind tunnel of HFI TU Berlin.

In order to simulate the performance of the Flow Vane on wind turbine aerodynamics, a
37.5m blade was designed. The main airfoil sections for this demonstration blade consisted
of cylindrical airfoil for the root and the DU97W300 and DU96W180 airfoils for the rest of
the blade span. The 360o performance curves for the blade sections were extrapolated by
means of the Montgomerie extrapolation [117] based on the airfoil polars measured at the
wind tunnel which were presented above. The twist distribution of the blade was selected for
a TSR of 8.0 and a the chord distribution was done after Schmitz [73]. The baseline blade
configuration was compared with a retrofit blade were the inner 14m of the blade span were
equipped with a Flow Vane. For this part the measured performance curves of the airfoils
with the Flow Vane were used. Table 10.3 presents the design characteristics of the blades
and turbine used for the current simulation.
The simulation results revealed a slight decrease in power for the low wind regime and a
significant power increase at the Nominal Wind regime. The Cp of the turbine was slightly
reduced due to the additional drag of the Flow Vane element as was the maximum Root
Bending Moment due to the additional lift generated by the Flow Vane. However the overall
energy yield was increased by at least 2% for a Weibull distribution with "k" parameter equal
to 2 and an scale paremeter (A) equal to 6. Figure 10.16 presents the comparative power
curves as well as their relative difference.
ΔPower [W]
Power [W]
1.5x106 6.0x104
Baseline
FlowVane (Root Region)
ΔPower
PolyFit2
4.0x104
1.0x106 2.0x104
5.0x105 -2.0x104
-4.0x104
0 -6.0x104
0 5 10 15 20 25
Wind Speed [m/s]
Figure 10.16: Simulated power curves for the test turbine with and without a Flow
Vane at the blade root region.
the blades fitted with a Flow Vane at the root region.
Blade Length 37.5 m
Max Structural Twist 23o
Design λ 8
Cut-Out Wind Speed 20.0 m/s
Weibuhl k Parameter 2
Weibuhl A Parameter 6
Root Circle & Transition Airfoil 0 - 2.65 m
DU97W300 2.65 - 11.0 m
DU96W180 11.0 - 37.5 m
“Flow Vane” Blade Design Parameters
Root Circle 0 - 2.65 m
DU97W300+ Flow Vane 2.65 - 11.0 m
DU96W180+ Flow Vane 11.0 - 14.0 m
DU96W180 14.0 - 37.5 m
Max. Root Bending Moment 1.5% increase
Energy Yield 2% increase/decrease
Max. Rotor Cp 0.5 % decrease
11 Active Flow Control Solutions
11.1 Flexible Trailing Edge Flap

The flexible trailing edge flap is one of the most promising active flow control solutions. It has
been investigated in various forms from several researchers and its performance in conjunction
with wind turbine blades has been evaluated [9]. In the current thesis a new flexible flap
concept was investigated based on the use of a high deflection pneumatically actuated flexible
flap mechanism developed by the author. This concept is developed as a power regulation as
well as load alleviation solution and therefore the flexible flaps are expected to have very high
aerodynamic control authority.
proceedings’ publication described at [131, 130] and [179].
To investigate the general flow behavior with such high deflection flaps in terms of lift and
drag variations several simulations were performed. The initial simulations were performed
with the XFOIL panel method code of M. Drela [47] running under the XFLR5 platform.
The OpenFOAMTM CFD toolbox [123] was also used at a next step mostly for investigations
regarding the flow separation around the flap. Table 11.1 shows the general details of the
XFOIL and OpenFOAMTM simulations.
The results of the XFOIL simulations are in agreement with the general theory regarding the
significant control authority of high deflection flexible flaps. The CFD flow-field simulations
revealed significant flow separation at the flexible flap region during large deflections. The
following figures (Fig. 11.1, 11.2 and 11.3) present some of the CFD results where the flow
separation at the suction side of the deflected flexible flap is evident.
Table 11.1: Specifications of the parameter setup for the numerical simulations of
the flexible trailing edge flap with QFLR5 and OpenFOAMTM .
XFOIL / XFLR5 Specifications
Model 2D&3D Panel Method
Elements 2D 200 Panels
Elements 3D 1000 Panels
Transition Model en
Ncrit parameter 9
Reynolds Number 1.3 · 106
AoA Range −8o − 12o
OpenFOAM Specifications
Solver simpleFOAM
Turbulence model k − ωSST
Grid Type Unstructured (Hexa)
Grid shape C-Grid
BL Grid Prism layer
Grid cells 200.000
Y+ <1
Reynolds Number 1.3 · 106
AoA Range 0o − 20o
Figure 11.1: CFD simulation (RANS) of the reference airfoil at 5o AoA. The flow
is fully attached.
Figure 11.2: CFD simulation of positive flap deflection at 5o AoA. The local sepa-
ration at the suction side of the flexible flap is apparent.
Figure 11.3: CFD simulation of negative flap deflection at 20o AoA. The suction
side of the flexible flap is not effective anymore since it is inside the
separated flow region.

The test wing consists of the main (forward) section and an exchangeable rear section. For
the rear section two elements were produced. One was the reference rigid section and the
other was the active, flexible flap section. In this way both the reference and flow control
measurements can take place without the need for two different test wings. Figure 11.5 shows
measured Cl and Cd curves of the reference airfoil and the modified airfoil with the flexible
flap section in neutral position (flexible flap not deflected).
Figure 11.4: The wind tunnel test wing with the flexible flap at full deflection po-
sition [131].
The two curves fit very close to each-other with the main differences located at the low AoA
region and the general slope angle of the linear part of the lift curve. The slight differences in
performance and the slight drag increase for moderate AoA are attributed to the small shape
deviations (presented also in Fig. 11.6) and the slightly round trailing edge of the flexible flap.
Furthermore, the S-shaped trailing edge of the original DU96W180 airfoil was not reproduced
by the flexible flap segment.
The flexible flap positive deflection (downwards) leads to an increase of the lift which shifts
the Cl curve up. The drag is significantly increased due to the local flow separation at the
suction side of the trailing edge flap (Fig. 11.7). This flow separation is also captured by
the CFD simulations and is caused by the strong adverse pressure gradient due to the flap
deflection. In the wind tunnel this effect is verified with the extensive use of flow tufts which
identify the flow separation region. To quantify the contribution of this feature to the overall
drag, vortex generators (VGs) were placed at 60%c of the suction side. The addition of VGs
energizes the flow and prevents the flow separation at the suction side of the flap causing
Figure 11.5: Cl and Cd curves for the reference wing (DU96W180) and the wing
with the flexible flap trailing edge section, measured in the wind
tunnel[131].
Figure 11.6: Schematic representation of the DU96W180 airfoil and comparison

with the flexible flap configuration. Shape deviations exist but are
kept to a minimum.
significant drag reduction. The installation of VGs at the suction side of the test wing leads
also to the increase of lift due to the increased influence of the suction side to the overall
generated lift.
The negative flap deflection generally has the opposite results regarding the lift behavior of
the configuration. The lift is significantly reduced(Fig. 11.8) and the Cl curve is shifted down-
Figure 11.7: Experimentally measured Cl and Cd curves of the wing with fully
deflected flexible flap (positive deflection), with and without VGs[131].
Figure 11.8: Experimentally measured Cl and Cd curves of the wing with fully
deflected flexible flap (positive deflection), with and without VGs[131].
wards to such an extent that the Cl at Clmax angle of the reference measurement becomes
negative! The significant reduction in lift is accompanied with large drag increase at low AoA
which is progressively reduced. At high AoA (higher than 16o ) the drag of the configuration
with the upwards deflected flexible flap is lower than the reference measurement.
At the same time the lift follows an increasing trend with the increase of AoA and the stall is
significantly delayed in comparison to the reference measurement. This, inevitably increases
the Glide Ratio (Cl /Cd ). A Glide Ratio reduction however is desirable for rotor deceleration
and eventual rotor-stop. The concept of high deflection trailing edge flaps can therefore be
used for load alleviation and partial power regulation but it is not able to achieve complete
rotor deceleration and rotor-stop [131].
11.1.3 Wind Turbine Aeroelastic Simulations

11.1.3.1 General Simulation Description
The behavior of the flexible trailing edge flap was analyzed with the use of the custom aeroe-
lastic code QBladeAE [178] which is based on the YawDyn code of NREL/NWTC [99]. The
simulations were based on the blade of a fictive 2.5MW wind turbine with a rotor diameter
of 89m and a fixed rotational speed of 15 rpm. The blade was generated at QBladeAE using
typical wind turbine airfoils throughout its stations. The last 60% of the blade span consists
of a single airfoil (DU96W180) which is characterized by a standard polar curve as well as
additional “child” polars representing the upwards and downwards flap deflection. Table 11.2
presents the main characteristics of the turbine used for the aeroelastic simulations.
Table 11.2: Turbine parameters used for the simulation.

Parameter Symbol Value
Nominal power PN 2500kW
Rotor diameter D 89m
Blade length lblade 43.3m
Hub radius Rhub 1.2m
Rotor tilt angle τ 4◦
Blade precone angle PC 2◦
Power regulation - pitch
Rotational speed (fixed) n 15rpm
Tip speed ΩR 70m/s
Number of blades B 3
Nominal wind speed vN 13m/s
Cut-in wind speed vin 3.5m/s
Cut-out wind speed vout 25m/s
Hub height HH 89m
The specific characteristics of the YawDyn structural model require some additional param-
eters such as the blade mass, the flapping mass moment of inertia, the root spring constant.
These are defined according the guidelines of Weinzierl [178] and are presented in Tab. 11.3.
Table 11.3: Blade structural parameters used for the simulation.

Parameter Symbol Value
Blade mass mblade 9840kg
Distance to center of gravity cogx,blade 10.6m
Flap mass moment of inertia Jf lap 1.89 · 10 kgm2
6
The blade pitch angle is set to 5o for this simulation and the hub, tower and nacelle are
defined as rigid bodies at the YawDyn structural model. The Beddoes-Leishman dynamic
stall model was de-activated in order to avoid the uncertainties that could be introduced at
the dynamic stall model due to the flap deflection (polar switching). The wind field was
generated with QBladeAE with the use of the TurbSim code of NREL/NWETC [94] and is
based on an IEC Kaimal spectrum using a turbulence intensity of 15% and a mean wind
speed of 13m/s at 89m hub height. The shear layer was modeled with a power law exponent
of 0.2 and a surface roughness length of z0 = 0.03. These assumptions are according to the
Germanischer Lloyd wind turbine certification standard (2010 Edition par.4.2.3.1.2). Figure
11.10 shows a time-series of the wind speed at hub height for a 90sec simulation.
The simulation was executed and the main load parameters of the baseline wind turbine
were computed for an operation of 90 seconds. Figure 11.9 presents some of the calculated
performance parameters.
For the aeroelastic simulation with flexible flaps a “composite” polar curve was generated.
The main polar is the 360o polar of the DU96W180 airfoil based on the aforementioned wind
tunnel test results. The “child” polars are based also on the wind tunnel test measurements
but their lift and drag values converge to the main 360o polar for very high and very low AoA
angles. This follows the general assumption that the outer blade part which will be retrofitted
with the flexible flap does not operate at these conditions outside the AoA envelope that is
covered by the wind tunnel measurements. Figure 11.11 presents the “composite” polar curve
for the flexible flap configuration.
16
wind speed [m/s]
14
12
10
8
6
350
thrust [kN]
300
250
200
3000
power [kW]
2000
1000
bending moment [kNm]
3000
2000
1000
4
blade deflection [m]
1
0 10 20 30 40 50 60 70 80 90
time [s]
Figure 11.9: Results of the baseline aeroelastic simulation with QBladeAE [179].
15
14
13
wind speed [m/s]
12
11
10
7
0 10 20 30 40 50 60 70 80 90
time [s]
Figure 11.10: Turbulent wind speed (x-component) time series at 89m (hub height)
produced with TurbSim[94] and used for the aeroelastic simulations
[179].
2
neutral position
slight neg. deflection
full neg. deflection
1.5 slight pos. deflection
full pos. deflection
0.5
0
cl
-0.5
-1
-1.5
-2
-150 -100 -50 0 50 100 150
α
Figure 11.11: Extrapolated “composite” 360◦ Cl -polar. The polar is based on exper-
imental date and is extended by means of Montgomerie extrapolation
[178].
11.1.3.2 Parametric Study

In order to identify the performance characteristics of various flexible flap configurations, the
user-friendly environment of QBladeAE was utilized and several alternative blades with active
elements were generated. The parametric investigation focused on the flap length variation,
the flap deflection variation and finally the variation of the flap actuation speed. For the
comparison of the different control authority of the various simulated configurations, the
standard deviation σ is taken as indicator. It is defined as:
v
u n
u 1 X
σ= t (xi − x̄)2 (11.1)
n−1
i=1
The configuration performance is expressed as percentage of the standard deviation σ0 of the

baseline simulation:
σi
χ=1− (11.2)
σ0
For the parametric investigation that follows, the outer part of the blade is divided in nine
equidistant parts of 1.5m, (Fig.11.12).
DU-96-W-180
9 8 7 6 5 4 3 2 1
1.5m
43.25m
Figure 11.12: Blade with 9 equidistant outer sections equipped with flexible flaps
[179].
The parametric simulations involve a 1.5m flap section located at different span-wise po-
sitions. Each position was independently simulated and was compared against the “baseline”
simulation presented above. The load reduction χ on the out-of plane bending moment was
taken as the critical performance parameter (Tab.11.4). The highest reduction was achieved
for the flap at r/R = 92% of the blade length (Pos. AE# 2 and AE# 3). The reduction in
flap-wise bending moment was in the range of 14%.
Figure 11.13 shows the load reduction potential of the flexible flap in relation with its
spanwise position. Compared to the investigations by Anderson [10], the results are in good
agreement.
Apart from the spanwise position, the flap length is also a decisive design parameter. Several
Table 11.4: Load reduction for 1.5m flap at different radial positions.
AE Pos. σAF M B [kNm] χ [%]
Baseline 274.6 0.0
Pos. 1 239.6 12.7
Pos. 2 237.2 13.7
Pos. 3 237.4 13.5
Pos. 4 238.3 13.2
Pos. 5 239.7 12.7
Pos. 6 241.3 12.1
Pos. 7 242.9 11.6
Pos. 8 244.8 10.9
Pos. 9 246.8 10.1
active blade sections were simulated in order to identify the influence of the flap length on the
turbine behavior. The flap lengths that were modeled of 4.5m, 9m, 10.5m, 12m and 13.5m.
The performance of the various configurations was defined again as a function of the reduction
in the out-of plane bending moment (Tab. 11.5). The simulation results show that the load
reduction increases with the increase of flap length, but with a non-linear relation. Small
flap lengths achieve a load reduction around 28%, while long flaps achieve up to 50% load
reduction.
Table 11.5: Load reduction potential for various flap lengths and different radial
positions. The Active Element numbers represent positions as these
were defined at Fig11.12.
AE# DR [m] σAF M B [kNm] χ [%]
Default - 274.6 0.0
123 4.5 193.7 29.5
456 4.5 196.9 28.8
789 4.5 205.6 25.1
123456 9.0 152.8 44.4
456789 9.0 169.1 38.4
1234567 10.5 135.2 50.8
2345678 10.5 136.9 50.1
3456489 10.5 145.8 46.9
12345678 12.0 138.4 49.6
23456789 12.0 137.6 49.9
123456789 13.5 131.4 52.1
Figure 11.14 shows the flap-wise root bending moment time-series for the baseline simula-
14
reduction in out-of-plane bending moment χ [%]
13
12
11
10
9
65 70 75 80 85 90 95 100
r/R [%]
Figure 11.13: Load reduction potential of a single flap with a length of 1.5m at
different radial positions along the blade [179].
tion and the one for a 13.5m long flap, with a maximum load reduction of about 52%.
Other critical operational parameters for the performance of the flexible flap configuration
are the flap deflection speed as well as its effective range. The definition of a specific required
speed and range is very important for the further development of the mechanical and electrical
design of any active flow control element. The existing flexible flap concept developed by the
author uses pneumatic muscles in order to achieve the required flap deflection and thus its
deflection speed is in the range of 20◦/s with a flap deflection of about ±20◦ .
In order to identify the influence of the flexible flap speed and range to the overall wind
turbine load profile, several simulations were performed for various flap speeds and flap deflec-
tions. Figure 11.15 shows the influence of the flap speed and flap range on the load reduction
potential (flap-wise bending moment) of the blade. It is shown that the selected flexible flap
configuration is indeed an efficient design achieving high control authority and significant load
reduction potential. Figure 11.16 presents the flap angle of the existing configuration (20◦/s
speed and ±20◦ deflection).
2400
baseline
13.5m flap
out-of-plane bending moment blade1 [kNm] 2200
2000
1800
1600
1400
1200
1000
800
600
0 10 20 30 40 50 60 70 80 90
time [s]
Figure 11.14: Flap-wise bending moment time series. Baseline simulation and the
13.5m long flap [179].
11.1.3.3 AFC element controller performance

In order to quantitatively compare the performance of the two different controllers (PID and
optimization loop) of the AFC elements, a set of simulations was performed. This simulation
included a 13.5m flap on the standard blade (Tab. 11.2) and turbulent inflow field. The results
of the simulation are presented in Fig. 11.17 where it is clear that the PID controller based on
blade-specific control parameters performs better than the optimization loop controller with
section-specific control parameters.
The extensive parametric investigations which were presented in brief in this chapter were
performed in order to identify the performance trends of various versions of the flexible flap
configuration. In order however to compare the AFC solutions with each other it is important
to maintain a constant blade configuration and to vary only the “composite” polar curves in
order to simulate each AFC solution. For this purpose the parametric investigation is not
repeated for the other AFC solutions.
As previously mentioned, the optimal Flexible Trailing Edge Flap configuration achieved
a load reduction potential (Root Bending Moment Load) of 52%. Figure 11.18 graphically
shows the effect of a large (13.5m) flexible trailing edge flap on the load behavior of the test
turbine. The effective range of this configuration is presented in Fig. 11.19 where the turbine
performance is simulated with the flexible flap "locked" at its extreme positions.
efficient flap configuration
50
45
40
35
30
χ 25
20
15
10
5 40
0
g]
35
e
30
[d
25
e
ng
20
ra
15
r
10
to
0 5 5
a
10 15
tu
20 25 0
ac
actuator spee 30
d [deg/s]
s.
ab
Figure 11.15: Load reduction for a 9m flexible flap configuration. Parametric inves-
tigation for various flap deflection speeds and flap deflection ranges
[179].
Figure 11.20 presents the overall performance results of the current solution. It is apparent
that the flexible flap is very effective as a load alleviation element while it is able to effectively
control the loads without frequently reaching extreme deflections and without showing signs
of control saturation.
15
10
5
flap angle [deg]
-5
-10
-15
0 10 20 30 40 50 60 70 80 90
time [s]
Figure 11.16: Flap angle for 9m flap over simulation time [179].
5000
blade element normal force DFN of AE#3 [kNm]
baseline
PID: 13.5m flap
LOOP: 9x1.5m flap
4500
4000
3500
3000
2500
2000
1500
1000
0 10 20 30 40 50 60 70 80 90
time [s]
Figure 11.17: The local blade element force DF N at the station AE# 3 for the
baseline configuration compared with the single PID controlled 13.5m
flexible flap and the multiple individually controlled flaps (optimiza-
tion loop controller) [179].
2400
Baseline
13.5m Flexible TE Flap
2200
2000
Root bending moment [kNm]
1800
1600
1400
1200
1000
800
600
0 10 20 30 40 50 60 70 80 90
Time [s]
Figure 11.18: Root Bending Moment time series comparison (baseline configuration
and Flexible T.E. Flap)[179].
3000
2500
2000
Root bending moment [kNm]
1500
1000
500
-500 Baseline
Max. const. neg. deflection
Max. const. pos. deflection
-1000
0 10 20 30 40 50 60 70 80 90
Time [s]
Figure 11.19: Extremes of the control authority of the Flexible T.E. Flap[179].
16
Wind speed [m/ s] 14
12
10
6
300
250
Thrust [kN]
200
150
100
2500
2000
Power [kW]
1500
1000
500
2500
Bending moment [kNm]
2000
1500
1000
reduced load due to flap
500
20
15
Flap deflection [deg]
10
5
0
-5
-10
-15
-20
0 10 20 30 40 50 60 70 80 90
Time [s]
Figure 11.20: Simulation results of the Flexible T.E. Flap aeroelastic analysis[179].
11.2 Flexible Leading Edge Flap

The concept of Leading Edge (or nose) flaps exist since the early 1920s [76]. There is there-
fore adequate documentation and theoretical background which describes their performance
(e.g. [76] and [5]). There are however very limited (if any) systematic investigations of the
performance of Leading Edge flaps when deflected upwards. For this reason several simula-
tions were performed with XFOIL in order to investigate the behavior of leading edge flap
configurations implemented on various airfoil shapes. Figure 11.21 shows the lift polar curves
for various Leading Edge Flap deflections for a NACA 63(3) 618 and a NACA 4415 airfoil. The
general performance trend of the configuration remains similar among various airfoil shapes
and is characterized from stall delay and sharper stall for downwards L.E. flap deflection (L.E.
camber increase). The upwards L.E. flap deflection (L.E. camber reversal) leads to an early
loss of lift at the, normally, pre-stall region and a smooth gradual stall for higher AoA.
Cl
Cl
2 2
1.5 1.5
1 1
0.5 0.5
0 0
-0.5 -0.5
-1 -1
-1.5 -1.5
-20 -15 -10 -5 0 5 10 15 20 25 -20 -15 -10 -5 0 5 10 15 20 25
AoA AoA
Figure 11.21: Cl curves for the NACA 63(3) 618 and the NACA 4415 airfoil (Re =
1.3 · 106 ) from XFOIL simulations.
The negative L.E. camber (upwards deflected L.E. flap) causes an adverse pressure gradient
which leads to local separation near the leading edge. This flow separation region depend-
ing on the current AoA can remain with the form of a stable separation region or rapidly
expand to all the suction side leading to airfoil stall. This unsteady behavior has the poten-
tial to cause significant hysteresis loops as was observed by the author during wind tunnel
experiments.Further investigations at the wind tunnel are therefore necessary for the char-
acterization of the post stall and dynamic behavior of the configuration. A indication for
the performance characterization of the plain leading edge flap can be found at the work of
Hoerner [76] where such configurations are thoroughly discussed. There it is seen that low
camber or symmetric airfoils at low or negative AoA (essentially similar to a negative flap
deflection) experience flow separation and loss of lift with subsequent drag increase.

For the wind tunnel measurements of the flexible leading edge flap, the modular NACA
63(3) − 618 test wing was used. The “baseline” measurements were performed with the rigid
test wing. The flexible leading edge flap then replaced the “baseline” leading edge module.
The flexible flap mechanism was conceived, designed and built by the author (patent pending).
It consists of an outer flexible skin which is supported by a movable inner mechanism. The
said mechanism is re-positioned with the use of pneumatic actuators thus re-shaping the outer
skin and modifying the airfoil contour (leading edge flap deflection).
proceedings’ publication described at [130] and [179].
The results from the wind tunnel tests are presented in Fig. 11.22. The baseline wind tunnel
measurement results for this airfoil generally agree with previous wind tunnel measurements of
the same airfoil (e.g. Abbott [5]). The general behavior of the flap for downwards deflections
also follows the established theory since a Clmax increase and a stall delay is observed. The
slightly lower lift slope at the pre-stall region is attributed to the shape deviations at the
leading edge region caused by inaccuracies and deflections on the flexible leading edge outer
skin. The abrupt lift drop observed at 10o AoA is also caused by the surface irregularities of
the flexible outer skin of the leading edge flap which cause local separation. The installation
of VGs at 10%c prevents the local separation phenomena and leads to experimental results
that closely match the theory and the XFOIL simulations (as seen in Fig. 11.22).
In the case of upwards deflection, the results are in agreement with the XFOIL simulations.
The early smooth stall and the overall lower lift coefficients observed during the XFOIL
simulations are evident at the experimental results (Fig. 11.23). A rapid lift drop is observed
around 9o AoA and it is caused by the rapid stall of the suction side which is triggered by
the high adverse pressure gradient formed by the negative camber (upwards flap deflection).
When the flexible flap mechanism is is replaced by a rigid upwards deflected flap model the
said lift drop occurs at higher AoA and it is even more pronounced. This proves that the
rapid lift drop (rapid stall) is a consistent characteristic of the configuration and it is not
significantly affected by the presence of a flexible skin at the airfoil contour.
The unsteady nature of the flow when the leading edge flap is deflected upwards leads to a
significant lift hysteresis. This is not observed at the steady state simulations but it is evident
at the wind tunnel measurement where the lift polar is characterized by a hysteresis loop
(Fig. 11.22. This behavior is generally not desirable for wind turbine operation since it may
increase vibrations and blade fatigue loads.
Figure 11.22: Cl curves for the NACA 63(3) 618 airfoil (Re = 1.3 · 106 ) from wind
tunnel measurements at the large wind tunnel of TU Berlin.

In order to compare the aeroelastic behavior and the load reduction potential of different
AFC solutions it is necessary to maintain the main performance and system design variables
constant. The optimal parametric configurations were therefore identified during the para-
metric analysis of the Flexible T.E. Flap and they were maintained for the investigation of the
Flexible L.E. Flap. The L.E. Flap was simulated with a deflection range of ±10o and a flap
length of 13.5m. The flap rate was set to 20deg/s [122] and the flap followed the commands
of the PID controller targeting to maintaining a constant out-of-plane root bending moment
value. Figure 11.24 shows the overall performance of the Flexible L.E. Flap with respect to
its target parameter. Compared with the performance of the Flexible T.E. Flap at the same
simulation with the same inflow field, it is obvious that the Flexible L.E. Flap solution is
Figure 11.23: Comparison of the wind tunnel measurements and the XFOIL simu-
lation results[130].
under-performing. The main reason for that is the minimal lift variation due to L.E. flap
actuation at the pre-stall regime.
Figure 11.25 presents the overall performance of the wind turbine and the flap during the
aeroelastic simulation. It is worth noting that the flap often reaches its range limit without
being able to satisfy the target parameter of the controller. The overall therefore performance
of the AFC solution provides very little control authority at the current operational point (low
α) and is thus considered sub-optimal for the purpose of operational load alleviation. The
maximum load reduction potential of the Flexible L.E. Flap during the aeroelastic simulations
of the current research work was 7.9%. However the post stall characteristics of the up-wards
deflected Flexible L.E. Flap could have a beneficial effect as an active stall devices for power
regulation of wind turbines. Such an investigation is naturally out of the scope of the current
paper and will not be further investigated.
2600
Baseline
13.5m[Flexible[LE[Flap
2400
Root[bending[moment[[kNm]
2200
2000
1800
1600
1400
1200
1000
800
0 10 20 30 40 50 60 70 80 90
Time[[s]
Figure 11.24: Time series of the root bending moment (with and without the Flex-
ible L.E. Flap) resulting from the aeroelastic simulations[179].
16
Wind speed [m/ s] 14
12
10
6
300
250
Thrust [kN]
200
150
3000
2500
Power [kW]
2000
1500
1000
3000
2500
2000
1500
1000
flap saturation
12
10
8
6
4
2
0
-2
-4
-6
-8
-10
-12
0 10 20 30 40 50 60 70 80 90
Time [s]
Figure 11.25: Summary of the aeroelastic simulation results of the Flexible L.E.
Flap Simulation. The limited control authority of this device is ap-
parent at the flap deflection plot[179].
11.3 Active Gurney Flap (Mini-Flap)

The CFD simulation of Gurney Flaps has been performed extensively in the past (e.g. [48],
[109]). The performance characteristics of Gurney flaps are well established and experimen-
tally verified since the late 70s (e.g. [105], [157]). For the scope of the current thesis it
was not considered necessary to perform even more CFD simulations of airfoils with Gurney
flaps. Some XFOIL simulations were performed in order to investigate the behavior of Gur-
ney flaps on the performance of the DU96W180 airfoil, where relevant data did not exist.
The DU96W180 airfoil was simulated with a 1.5%c Gurney flap and a 1.5%c Wedge-shaped
gurney flap. Figure 11.26 presents some of the results of the XFOIL simulations. The general
performance of the configurations is in accordance with the theory. The addition of a Gurney
flap generally leads to a significant lift increase, an upwards shift of the whole lift polar.
proceedings’ publication described at [130] and [179].
Cl
Cd
2 0,09
DU96W180 @ Re=1.2x106
XFOIL Simulation
0,08
Baseline
Gurney Flap (1.5%c)
1,5
0,07 Wedge-shaped Flap
0,06
1
0,05
0,04
0,5
0,03
DU96W180 @ Re=1.2x106 0,02

0 XFOIL Simulation
Baseline (XFOIL)
Gurney Flap (1.5%c) 0,01
Wedge-shaped Flap
-0,5 0
-5 0 5 10 15 -5 0 5 10 15
AoA AoA
Figure 11.26: XFOIL simulation results for the DU96W180 airfoil in baseline config-
uration and equipped with Gurney and wedge-shaped Gurney flaps.
11.3.2 Wind Tunnel Measurements (Steady State)

Several Gurney flap variants were tested on the DU96W180 wind tunnel test wing. Various
Gurney flap sizes were evaluated in order to identify its optimal dimensions. Some of the
basic variants of Gurney flaps are shown in fig. 11.27. Furthermore, conventional Gurney flap
configurations with a height of 1.5%c were tested on the AH93W174 and the DU97W300 wind
tunnel test wings. The purpose of these tests was the identification of performance similarities
as well as the investigation of hysteresis effects caused by identical elements (Gurney flaps of
the same height) on different airfoil sections.
Figure 11.27: Various Gurney flap configurations. a)Standard L-profile, b) Gurney

flap and Splitter Plate, c) Wedge-shaped Gurney flap [130].
All the tested configurations have the same general lift performance characteristics. The
wedge shaped Gurney flap seems to have a slightly better overall performance compared to
the standard Gurney flap configuration and the Gurney flap equipped with a splitter plate
has a slightly reduced drag. A combination of a wedge-shaped Gurney flap with the splitter
plate would probably be the optimal shape. In terms of the element’s size, the wind tunnel
tests verified the previous results found in the literature ([48], [109]). A Gurney flap height
in the range of 1 − 1.5%c is found to be optimal with respect to lift and drag increase. The
spanwise serrations of the Gurney flap performed as expected [48], thus reducing the drag
with some lift loss penalty. Figure 11.28 presents the behavior of all the different Gurney flap
configurations tested on the DU96W180 airfoil. The performance of Gurney flaps on the other
airfoils (DU97W300 and AH93W174) follows similar trends and it is presented in Fig.11.29
and Fig. 11.30.
11.3.3 Wind Tunnel Measurements (Unsteady)

In order to experimentally investigate the dynamic performance of AFC solutions, the concept
of active Gurney flap was selected and was tested at the large wind tunnel of TU Berlin. The
AH93W174 wind tunnel test wing was selected and was fitted with a hinge mounted Gurney
flap (or mini-flap) with a height of 2%c. The flap in this configuaration is able to pivot and
change its angle relative to the airfoil chord. It is able to achieve very high deflection angles
Figure 11.28: Various Gurney flap configurations tested on a wind tunnel wing
based on the DU96W180 airfoil[130].
(Gurney flap principle), both towards the pressure and the suction side. The actuation of
the flap is achieved via a number of digitial servo-actuators controled by the a desktop PC.
The wind tunnel hardware and control software was modified in a way that is able to perform
dynamic measurements. The actuator that varies the AoA of the test wing is electronically de-
coupled from the digital 6-component force balance and an additional AoA sensor is mounted
in order to provide a continous AoA signal of the configuration.
The principle of the current wind tunnel experimental configuration is the simulation of the
wind variations and wind gusts found on actual wind turbine blades with continuous pitching
motion of the test wing. Such an AoA variation leads to lift and drag variations. The purpose
of the active Gurney flap would be to stabilize the generated lift to a desirable level. The
AoA variation during the dynamic AFC tests in the wind tunnel was extracted from dynamic
BEM and Multi-body Wind Turbine Simulation to match as much as possible the actual wind
turbine behavior. In this way the performance of the Active Gurney Flap system could be
more easily assessed for an actual wind turbine application.
During the dynamic investigations several control strategies were tested, starting from stan-
dard PID controllers with semi-empirical parameter tunning models, to DIC (Direct Inverse
Controllers) with Neural Network tunning strategies and pure self learning Neural Network
Figure 11.29: Gurney flap performance on a wind tunnel wing based on the
AH93W174 airfoil.
controllers. The results of the closed loop measurement using the manually tuned PID-
Controller showed a reduction potential for the dynamic lift loads in the range of 70% as well
as a stable controller behavior. The use of Neural Networks to define an inverse model of
the considered system was successful. The defined network was used as a Direct Inverse Con-
troller.The closed loop measurements with this DIC controller showed, that a load reduction
of 36.8% is possible with this configuration. Both the control signal however as well as the
flap deflection signal indicated a certain controller instability.The overall results of this phase
of the research showed that current control strategies have exhibit good results (Fig.11.31)
with respect to wind turbine applications for load alleviation [122]. At the same time however
it has to be noted that there is still a very large potential for improvement of these control
strategies, especially the Newral Network DIC control systems. Further information on the
current experimental investigation are available at Reference [122].

The active Gurney flap was tested in a dynamic, unsteady mode at the wind tunnel of TU
Berlin mounted on an Althaus AH93W174 airfoil [122] and showed a significant load allevia-
Figure 11.30: Gurney flap performance on a wind tunnel wing based on the
DU97W300 airfoil.
tion potential. The results from the static and dynamic measurements of the aforementioned
configuration were compared to current wind tunnel results of the DU96W180 airfoil equipped
with the same Gurney flap configuration (Fig. 11.28). For the purpose of comparison in the
current aeroelastic simulations the steady state performance curves are extracted from the
wind tunnel measurements of the DU96W180 airfoil. The active Gurney flap was modeled as
a single flap with a length of 13.5m. It was actuated with a rate of 180deg/s and a deflection
of ±90o [122]. Figure 11.32 shows the performance range of the Active Gurney Flap. The
simulations presented in this figure are performed with the gurney flap fixed at the +90o and
−90o position.
With the PID controller activated the Active Gurney Flap displayed a load reduction poten-
tial of 35.8% with respect to the root bending moment values (Fig. 11.33). The performance
of the active Gurney flap is better than that of the flexible leading edge flap but neverthe-
less its control authority is significantly smaller than that. The low actuation force however
and the simple manufacturing and blade integration are some of the strongest advantages of
this solutions. Figure 11.34 presents the overall results of the aeroelastic simulations with
the Active Gurney Flap. The "flap deflection" plot of Fig. 11.34 shows the activity of the
Active Gurney Flap which is constantly high. At some instances the flap reaches its control
Figure 11.31: Measured wind tunnel data of lift variation due to AoA variation with
(red curve) and without (blue curve) active flow control[130].
authority limit without achieving its target. The overall performance of the Gurney flap is
satisfactory and its dynamic response is adequate, but the limited control authority of such a
small trailing edge element is naturally an adverse performance parameter.
3000
2500
Root[bending[moment[[kNm]
2000
1500
1000
500
-500 Baseline
Max.[const.[neg.[deflection
Max.[const.[pos.[deflection
-1000
0 10 20 30 40 50 60 70 80 90
Time[[s]
Figure 11.32: Simulation results for the extreme actuation range of the Active Gur-
ney Flap[179].
2400
Baseline
13.5mgGurneygFlap
2200
Rootgbendinggmomentg[kNm]
2000
1800
1600
1400
1200
1000
800
600
0 10 20 30 40 50 60 70 80 90
Timeg[s]
Figure 11.33: Root bending moment timeseries with and without Active Gurney
Flap[179].
16
Wind speed [m/s] 14
12
10
300
250
Thrust [kN]
200
150
100
2500
2000
Power [kW]
1500
1000
500
2500
2000
1500
1000
500
90
60
30
0
-30
-60
-90
0 10 20 30 40 50 60 70 80 90
Time [s]
Figure 11.34: Simulation results for the extreme actuation range of the Active Gur-
ney Flap[179].
11.4 Stall Ribs

To investigate the aerodynamic effect of a stall rib located at the leading edge region of
an airfoil (DU96W180), preliminary XFOIL simulations were performed. The normal airfoil
performance was compared to that with the stall rib deployed (Fig. 11.35). The simulation
results showed that a stall rib of even small size is able to induce a local suction peak (Fig.
11.36) with considerable influence on the global airfoil aerodynamics. The suction peak caused
by the stall rip triggers boundary layer transition and depending on the chord-wise location
of the stall rib and the local AoA can also trigger flow separation (i.e. stall).
proceedings’ publication described at [130].
Figure 11.35: Comparative outline of the “clean” DU96W180 profile with similar
airfoil equipped with stall rib located at 33%c.
The 2D CFD simulations with XFOIL showed that the stall rib causes early stall (at approx.
6o AoA and a significant post stall lift drop (Fig. 11.37). However at approx. 9o AoA the lift
starts increasing again and fully recovers at 19o AoA where it reaches the value of the “clean”
airfoil configuration. This behavior is in accordance with the findings of older studies of "rake"
type spoilers in the past. Hoerner describes the performance of such a configuration which
exhibits the same characteristics[76]. The stall rib looses its effectiveness as soon as the flow
fully separates from the suction side of the airfoil. There is however a certain interaction of
the element with the partially separated flow until the airfoil exhibits full separation. The
characterization of this interaction is worth investigating, but it is out of the scope of the
current work.
The absolute control authority naturally varies with the size, the shape and the chord-wise
location of the stall ribs, however the overall of trend of their performance remains. The main
disadvantage of this configuration is that it is not able to regulate the power of the turbine
rotor due to limited control authority. Furthermore the stall ribs are not able to stop the
rotor since their high angle of attack performance is very limited.
Furthermore the integration of these elements on the external surface of the blades can
introduce surface irregularities which could lead to boundary layer transition when the stall
ribs are in the retracted position (i.e. normal operation). Such an effect would lead to blade
efficiency reduction.
Figure 11.36: Aerodynamic performance graphs for the DU96W180 airfoil with de-
ployed stall rib.
In addition to the experimental investigations, the stall rib concept was also investigated
numerically with OpenFOAMTM simulations. These simulations aimed mostly to investigate
the general flow-field and the separation region downstream of the rib. Figures 11.38 depict
Figure 11.37: Aerodynamic performance graphs for the DU96W180 airfoil with and
without a deployed stall rib.
the vorticity contour for a 1.3 · 106 Reynolds number flow around a NACA 4418 airfoil with
and without an A-Type stall rib.
Figure 11.38: Vorticity field for a “clean” NACA 4418 airfoil (top) and for the same
airfoil with deflected stall rib (bottom)[130].

The performance of the stall ribs was investigated in the wind tunnel where several rib shapes,
sizes and positions were investigated. The investigations were performed with the DU96W180
test wing and steady polar data were extracted. Three different types of stall rib shapes were
tested in various locations on the wing chord (suction side) in order to identify the general
performance parameters. Stall rib A represents a smooth convex shape with a height (hA )
of 10mm and a length (LA ) of 17mm. Stall rib B has a higher height-to-length ratio and is
asymmetric with a height (hB ) and a length (LB ) of 10mm. Stall rib C is a right and isosceles
triangle with its hypotenuse acting as a flow ramp. The height (hC ) and length (LC ) of this
stall rib configuration is 10mm. Figure 11.39 shows the different shapes of stall ribs tested
at the wind tunnel. Table 11.6 presents the pattern of the different stall rib wind tunnel test
configurations.
Figure 11.39: A, B and C type stall bumps. All different shapes were tested on
various positions on the wing[130].
Table 11.6: General test matrix of the stall bump experimental investigations.
Stall Bump Type Test Positions
A Type 10%c, 20%c, 50%c
B Type 10%c, 20%c, 50%c
C Type 10%c, 20%c, 50%c
A + B Type 10%c and 50%c
The results of the wind tunnel tests for different types of stall bumps and various positions
provide significant insight about their performance and effectiveness. Figure 11.40 shows the
lift and drag curves of the stall rib (A-Type) with respect to its chord-wise position. The
wind tunnel results show that the positioning of the stall rub near the leading edge delays
its effect (in terms of AoA) and reduces the maximum control authority at low AoA. On the
other hand the placement near the leading edge increases the control authority at higher AoA.
The optimal stall rib position for the DU96W180 airfoil in terms of overall control authority
is around 20 − 30%c (i.e. near the max. thickness location of the airfoil).
The experiments with different stall rib shapes tested at the same chord-wise position
showed that the most severe impact on lift and drag was caused by the C-Type ribs followed
by the B-Type and the A-Type (Fig. 11.41). This result was generally anticipated since
the stall ribs with sharp shape cause larger separation region downstream and prevent the
re-attachment of the boundary layer.
Figure 11.40: Lift and drag behavior comparison between the baseline measurement
and A-Type stall ribs in different positions.
Figure 11.41: Lift and drag behavior comparison between the baseline measure-
ment and all types of stall ribs mounted at the same chord-wise
position[130].

The performance of stall ribs on wind turbine blades was calculated by means of the QBlade
BEM simulation software. The simulation was based on a 2.5MW turbine with 44.5m blades.
The outer 2/3 of the blade span was modeled with the DU96W180 airfoil with and without the
stall ribs. The steady state performance of the turbine was compared and Fig. 11.42 presents
the main differences in performance. The aim of the simulation is to examine the potential of
stall ribs for power regulation. For this reason, the baseline turbine is pitch regulated while
the turbine equipped with stall ribs has a fixed pitch and deploys the stall ribs for power
control. As the graph 11.42 shows the stall ribs can partially control the power of the turbine
(low wind speed region) but are not able to control the power of the turbine for wind speeds
higher than 17m/s. Figure 11.43 presents the AoA and Cl distribution along the blade span.
The lower lift coefficient of the sections with deployed stall ribs lead to the modification of
the axial induction factor during the BEM calculation. This in turn leads to the modification
of the local AoA which is visible especially at the outboard sections of the blade.
Figure 11.42: Power curve (left) and bending moment over wind speed (right) com-
parative graph for the turbine with and without stall ribs.
Figure 11.43: Angle-of-Attack and lift coefficient (Cl ) distribution along the blade
span for 10m/s wind speed.
11.5 Plasma Actuators

11.5.1.1 Experimental Setup
For this experiment another experimental setup was used since the measurements were per-
formed at a different wind tunnel facility. The current experimental setup comprises an open
wind tunnel (Eiffel type) equipped with a radial fan (I), a cylindrical settling chamber (II),
a nozzle (III), a test section (IV) and a reflection-free diffuser (V) (Fig.11.44). The fan is
driven by a speed variable electric motor. The settling chamber is equipped with honeycomb
structures and fine grids in order to reduce the turbulent structures of the flow. The large
size of the settling chamber is required for the damping of the flow pulses generated by the
centrifugal fan.
The test wings are mounted on a three component balance located inside the test section.
The balance measures forces by utilizing strain gages [12].
proceedings’ publication described at [51].
Figure 11.44: Basic scheme of the experimental setup [50]
The voltage supply of the bridge and the voltage measurement of the bridge response is
realized by an analog-digital-converter NI 9237 made by National Instruments. With this
device 50000 samples are measured with a sample rate of 25kHz and the mean value is
computed. A balance calibration is conducted before each series of wind tunnel measurements.
The inflow velocity in the test section of the wind tunnel is measured using a Prandtl tube.
The measurement of the pressure difference is realized by a pressure transducer. It transfers
the pressure difference in a range of 0 to 100 Pa to a measurable voltage, which is captured
by a analog-digital-converter NI 9172 made by National Instruments.
For the plasma actuators a function generator is used to create a square wave with a
frequency of 8kH, a duty cycle of 50% and a peak to peak voltage of 5V. Feeding this square
wave signal to the high voltage generator results in a steady operation of the actuator and
thus in a steady momentum injection. To achieve a pulsation of the actuator and thus an
oscillating wall jet a modulation of this square wave is necessary. Another square wave with
a low frequency fpl and a duty cycle DC is generated with custom software and is fed via an
analog output module made by National Instrument to the function generator as an external
trigger signal. The resulting signal is fed to the high voltage generator and the amplified signal
is sent to the plasma actuator. During the force measurements the actuators were operated
with a peak to peak voltage of 15kV [50].
The plasma actuator experiments focused on models resembling the root region of a wind
turbine blade. In particular a circular PVC cylinder with 4mm wall thickness and an outer
diameter of 0.11m and a span of 0.27m was mounted on a Polyethylene shaft. It was equipped
with three electrodes. Figure 11.45 shows the general layout of this test model. The plasma
actuator electrodes consist of 70µm thick and 3mm wide aluminum tape. The electrodes 1
and 2 are exposed to the flow, whereas the outer edges are electrically sealed with 50µm thick
Kapton tape to avoid plasma generation in this region. The electrodes are placed with an
azimuthal distance of 8o between each other. With this configuration three different kinds
of momentum injections can be realized depending on which electrodes are supplied with
voltage.
For counter stream-wise momentum injection electrode 1 and the covered inner electrode
are supplied with voltage. To create a stream-wise wall jet, the electrode 2 and the covered
inner electrode is supplied with voltage. When all electrodes are supplied with high a.c.
voltage, a streamwise and a counter streamwise jet is generated. Thus both opposing jets
form a stagnation point in the middle of both exposed electrodes resulting in a radial eluded
jet (Fig.11.46).
Apart from the cylindrical model, a dedicated wind turbine airfoil was also used of the
creation of a constant chord wind tunnel wing. This wing was based on the DU97W300
airfoil with a constant cord of 0.15m and a span of 0.278m. It was build using a sandwich
construction of glass fiber reinforced plastic and DepronTM . The test wing was equipped
with two plasma actuators which could realize a stream-wise momentum injection. The first
plasma actuator was placed at 10%c on the suction side and a second actuator was placed at
20%c on the pressure side. Note that the actuator on the pressure side was mounted after
the measurements with only one plasma actuator on the suction side. Figure 11.47 shows the
contour of the DU97W300 and the positions of streamwise momentum injection.
The actuator design was similar to the linear actuator shown in figure 11.45. Therefore a
9mm wide electrode made of 70µm thick aluminum tape was glued onto the surface of the
wing followed by four layers of 50µm Kapton tape, which served as dielectric. The exposed
electrode made of 3mm wide and 70µm thick aluminum tape was glued with the trailing edge
at the corresponding momentum injection position. Smoke wire flow visualization is applied.
A thin ConstantanTM strand was mounted on a stand and placed at a small gap between test
section and nozzle. Oil droplets are applied to the strand with a small injector.
Figure 11.45: View of the cylinder model with the plasma actuators in operation
[50].
Figure 11.46: View of the cylinder model with the plasma actuators in operation
[51].
Figure 11.47: The DU 97W300 airfoil geometry and positions of streamwise mo-
mentum injection[51].
11.5.1.2 Results
The cylinder model was measured at Reynolds numbers of Re = 85, 000 and Re = 136, 500.
Baseline measurements with the “clean” configuration as well as with the inactive plasma
actuators were performed to consider the tripping effect of the electrodes. The cylinder was
also equipped with a 2D boundary layer trip of 2mm diameter to compare the results of active
flow control with plasma actuators and the results of passive flow control using a 2D-trip. The
measurements with the actuators operating were performed at a reduced frequency F + = 1
and a duty cycle of DC = 0.2. The azimuthal position of momentum injection was varied
step by step and for every position a measurement with switched off actuator was performed
immediately before and after the measurement with switched on actuators. The mean of these
values was removed from the data of switched on actuator in order to capture the lift increase
and the drag decrease according to the following equations.
∆cd = cd,on − 0.5 (cd,of f bef ore + cd,of f af ter )

(11.3)
∆cl = cl,on − 0.5 (cl,of f bef ore + cl,of f af ter )
The configuration was tested in various states, with the actuators being activated in various
frequencies and duty cycles. For brevity only the results of the unsteady simulations will be
presented in the following paragraphs.
Figure 11.48: ∆Cd and ∆Cl values for unsteady actuation at Re=85,000 [51, 50].
The initial wind tunnel measurements were conducted at Re = 85 · 103 (Fig. 11.48) and
the result was the significant lift enhancement with simultaneous drag reduction due to the
delay of separation at the downwind side of the cylinder. For azimuthal plasma actuator
positions between 60o and 88o the aforementioned effect is present regardless of the direction
of momentum injection from the actuator. The results are similar for the experiments with
Re = 136 · 103 and all the respective momentum injection directions. The activation of
the plasma actuators triggers Tollmien-Schlichting instabilities thus forcing laminar turbulent
transition and enhancing the mixing inside the boundary layer.
Figure 11.49: ∆Cd and ∆Cl values for unsteady actuation at Re=136,500 [51, 50].
As seen in the respective smoke visualization images (Fig. 11.46) the effect of the plasma
actuators is evident and it leads to consistent observations with respect to Reynolds number
variations as well as actuator performance parameter variations.
Figure 11.50: Smoke visualization of the test cylinder with inactive (left) and active
(right) plasma actuators. The effect of the plasma actuator on the
flow is evident [50].
The injection of streamwise momentum injection over the suction and pressure side of the
DU97W300 test wing leads to a significant lift increase (∆Cl ≈ +0.36) as well as a drag
reduction (∆Cd ≈ −0.11). These values were extracted at Re = 85 · 103 . When however the
free stream velocity reaches a higher value (Re = 180 · 103 ) the effect of the plasma actuator
completely disappears(Fig.11.51). The smoke visualization presented at Fig.11.52 shows the
effect of plasma actuators on the flowfield around the test wing at low Reynolds number flows.
It is apparent that the injection of momentum by the plasma actuator significantly reduces
separation thus increasing lift and reducing drag.
Figure 11.51: Unsteady operation of plasma actuators at the DU97W300 test wing
[50].
The overall outcome of all the experimental investigations presented above is that plasma
actuators are able to significantly influence the performance of airfoils (and consequently
wind turbine blades). The main requirement however of the plasma actuator concept is
that of a low Reynolds number flow. This crucial drawback of plasma actuators renders them
completely ineffective for large wind turbines. A possibility for utilization of plasma actuators
on wind turbines is their application on small Vertical Axis Wind Turbines (VAWTs). Another
possibility in the application of plasma actuators at the blades of small horizontal axis wind
turbines in order to assist at the low-wind speed regime and to allow low-wind speed start
up.
Due to the inability of plasma actuators to function properly when installed on medium
and large sized wind turbines, the wind turbine simulation results (steady and unsteady) were
omitted.
Figure 11.52: Smoke visualization of the test wing with inactive (top) and active
(bottom) plasma actuators [51].
Part VI
Blade Design
12 ”Smart Blade” Design
Characteristics
12.1 AFC and PFC Solution Selection

After the preliminary (Section IV) and detailed aerodynamic and aeroelastic investigations
(Section V) of several Passive and Active Flow Control Solutions, a better understanding of
the potential of each solution is achieved. Since no single flow control element was able to
simultaneously fulfill all the requirements set at the beginning of the research efforts (e.g. full
power regulation and load alleviation) a combination of elements should be used in order to
achieve full rotor control. Furthermore a selection of passive flow control solutions should also
be incorporated to provide higher and more stable performance and also to augment the AFC
elements.
An example therefore of a “Smart Blade” is proposed and described in this section of the
thesis. This is by no means the only possible configuration but one technical solution that,
according to the author, fulfills best the current specific requirements.
To select the final AFC and PFC solutions for the “Smart Blade” it is necessary to assign
a technical solution to one or more performance requirements. A technical solution therefore
can satisfy more than one requirements but may require the existence of another solution in
order to do so. Table 12.1 attempts to briefly describe the final AFC and PFC selection.
Table 12.1: The final AFC and PFC solution selection for the proposed “Smart
Blade” design.
Technical Solution Effect Span Position Chord Pos.
Flex. T.E. Flap Load management 70 − 90% depth:10%c
Flex. T.E. Flap Partial power reg. 40 − 90% depth:20%c
Spoiler units Stall control 20 − 70% @ 20%c
Spoiler units Rotor stopping 20 − 70% @ 20%c
L.E. Slat Stall delay 5 − 40% @ 20 − 30%c
VGs Stall delay 40 − 90% @ 20 − 30%c
er
oil
Sp
s p
or fla
rat ge
e ne
ed
xG g
r te ilin
Vo tra
le
e xib
Fl
n)
t
w
g le
do
in
d
le
tw
ca
(s
Sla
d e
la
12 ”Smart Blade” Design Characteristics
alB
t on
ti
en
sla
ge v
g ed on
C
din
ea
dl
Fixe
Figure 12.1: Schematic representation of the proposed “Smart Blade”.

187
The primary AFC element of the proposed blade design is the flexible trailing edge flap.
The purpose of this is to alleviate the peak operational dynamic loads during the turbine
operation and at the same time to achieve the task of partial power regulation. To reduce the
dynamic loads, the aeroelastic simulations[178] showed that a relatively small but fast moving
flap is sufficient. On the other hand to achieve a reasonable range of power regulation it is
necessary to incorporate a trailing edge flap with a depth of at least 20%c and large reflection
angles. High actuation speed is not as critical for power regulation. However if the flexible
flap system is to be combined for both load and power management then it should consist of
a large and fast flap system which covers 50 − 60% of the outer blade span(Fig. 12.1).
Since the overall principle of the current research work is to achieve advanced power reg-
ulation through innovative “smart” aerodynamic solutions, it is assumed that the proposed
“smart blade” does not have a pitch regulating system. Therefore it is securely fixed on the
rotor hub, similar to the blades of stall turbines. The flexible flap concept therefore is able to
regulate the power of the turbine as long as there is attached flow over its contour. At very
high wind speeds though the absence of a pitch system leads to a considerable increase of the
local angle of attack along the blade. This AoA increase leads to flow separation near the
trailing edge, which consequently cancels the effect of the flexible flap. Since stall regulation
is not an option for multi-MW wind turbines due to increased fatigue load increase, it is nec-
essary to prevent or at least delay flow separation. For this reason the proposed blade design
is equipped with vortex generators throughout the blade span with a flexible trailing edge flap
(Fig. 12.1). These are located around 20 − 30%c (depending on the airfoil characteristics)
and lead to a considerable extension of the operational range of the flexible flap.
The other operational region where the flexible trailing edge flap needs augmentation is the
rotor deceleration. Once again the AoA increases due to the reduction of the rotor rotational
frequency and the blade stalls. Especially in the case of an emergency stop the effort to
integrate a Flexible Flap + VG configuration able to bring the rotor to a halt is immense.
Instead, a series of spoilers are proposed along the span of the blade. Their chord wise
location should be at or near the main spar of the blade to reduce the dynamic and fatigue
loads to the blade during their operation. It is worth noting that the operation of spoilers
on wind turbine airfoils was tested by the author during a series of wind tunnel tests(see also
Ref.[140]) the results of which however are not included in the present documents for space
economy. Furthermore the relevant literature includes abundance of information on the static
and dynamic performance of spoilers (e.g. [182, 112, 96].
Since the turbine blade concept proposed by the author’s research work does not include
a blade pitch system, the addition of a round hub-connecting flange is also not necessary.
The investigation however of other blade root design concepts is out of the scope of the
current document. For this reason the root region of the proposed “smart blade” is of the
conventional type with the addition of the a series of flatback airfoils[86]. In order to improve
the aerodynamic performance of the entire inner region of the blade a fixed leading edge slat is
proposed. It can lead to considerable performance increase with low cost and low production
and integration effort[129]. It is proposed that the tip of the slat element is formed as a
winglet[79] in order to reduce the strength of its trailing vortex, which can be a significant
noise source.
12.2 Actuator Mechanisms, Blade Integration and

Control
12.2.1 Actuators
One of the decisive parameters regarding the use of an AFC solution on wind turbines blades
is the overall performance of its actuators. Cost, weight, complexity and reliability of the
AFC actuator play a very important role to the overall wind turbine performance. Through
the research process in the course of this thesis it was possible to highlight the most critical
challenges for the AFC actuator concepts and to estimate the performance of different concept
against these challenges. The overall conclusion is that pneumatic actuators are the most
desirable solution due to their low cost, wide and long industrial use and high tolerance
against adverse environmental conditions.
Among the existing pneumatic actuators the concept of the pneumatic muscle is especially
preferred. These innovative actuators achieve a very high actuation power density compared
to the conventional pneumatic cylinders while having very low cost and no moving parts
(monolithic actuators). They can be powered by low pressure (6-8bar) compressed air systems
with digital valve controllers. The pneumatic muscles are able to operate in pulse mode (on/off
operation) as well as in a progressive mode through digital proportional valve systems.
The connection of the actuators to the digital valves and the power supply units is done
through flexible plastic tubing thus reducing cost and weight over a high pressure metal tube
system. Furthermore the use of plastic tubing instead of metal tubing or cabling (i.e. electrical
actuators) solves the problem of lightning-strike sensitivity.
12.2.2 Blade Integration

Figure 12.2 shows a schematic representation of the pneumatic muscle system and its inte-
gration on the rotorblade. The pneumatic actuation in combination with the muscle concept
allows the actuation system to be fully encapsulated inside the flap unit which can then be
directly attached to the connection interface on the blade (Fig.12.2). Such an integration
method allows the simple installation of the AFC modules on-site and reduces the trans-
portation and crane lifting sensitivity of the blade. The blade production is not altered by
the addition of the selected AFC elements. Even existing production molds could be suitably
modified in order to allow the production of blades with provision for trailing edge mounted
flexible flap and suction side mounted spoiler.
The installation of the AFC modules could take place during wind turbine erection and
would require only the connection of the power (air pressure) supply to the AFC elements and
the attachment of the AFC modules on the blade (Fig. 12.2). In this way the transportation
and crane operation of blades would remain identical to the current practices. The AFC
elements would be installed after the connection of the blades on the wind turbine hub (rotor
assembly on the ground). The fully assembled rotor would then be lifted and positioned on
the assempled wind turbine nacelle.
12.2.3 Control and Sensors

Wind turbine control systems are currently relatively simple but are constantly becoming
more complex and more sophisticated. The addition of AFC elements on the blades represents
a great challenge but also a great opportunity for the control system development of wind
turbines. Current wind turbine control systems are generally under-controlled since they have
to cope with the fact that a very unstable input (atmospheric wind) has to be converted to an
output (electrical power) which is as stable as possible. This is currently done with a series of
sensors that are able to provide information only for the past state of the system. Both the
anemometer, the flow vane and the generator torque and load sensors used in modern turbines
are reacting after the impact of the wind flow to the rotor. A very promising alternative to
the current control strategies is the use of sensors that are able to give input to the controller
regarding the state of the wind flow before its impact to the rotorblades (e.g. LiDAR and
SODAR)[100].
The addition of fast responding AFC systems for power regulation but mostly for load
alleviation requires the use of responsive and efficient control strategies in combination with
innovative sensor concepts. For an effective load alleviation control strategy and due to the
stochastic nature of the atmospheric flows it is important to develop and implement sensor
systems that have a relatively high frequency and are able to, idealy, provide input for the
incoming flow before its impact on the blades. Currently the most prominent sensor solutions
for “smart blades” are based on the following general concepts:
• Strain Gauges: Installation of strain gages at the blade root. The measured bending
moment is used as a control input parameter for the AFC elements. The main time lag
for this sensor concept is caused by the distance of the flow control actuator (outboard
blade region) to the blade root (sensor position)[43].
• LiDAR: A forward facing LiDAR scans the field forward of the rotor plane and extracts
flow velocities through Doppler Effect analysis. The extracted velocity field in com-
bination with the rotor position determines the actuation of the AFC elements. This
method achieves a near-perfect control on the expense of considerable computing power
requirements[100].
• AoA Sensors on blades: Various types of sensors (e.g. Pitot tube, force sensor, ultrasonic
sensor) can be installed at least on one spanwise location of each blade (preferably at
the outboard region). The AoA or velocity readings of each sensor are processed and
translated to control commands for the AFC elements. Such a system achieves high
local performance but requires the installation of more than one sensor per blade in
order to achieve proper input for a large range of AFC elements along the blade span.
During the research investigations of the current work the “local AoA sensor” concept was
compared to the “root region strain sensor” concept. The results of this comparative anal-
ysis are presented at the Section 11.1.3. At this point it is worth noting that the sensor
requirements and the control system complexity is significantly reduced when the AFC ele-
ments perform Power Regulation. For this operation the existing control strategies seem to
be suitable to provide all the necessary system control parameters.
The addition of a single AFC element on each wind turbine blade increases the complexity of
the wind turbine control system significantly. The use however of active flow control elements
on blades achieves its full operational potential when more than one independent elements
are installed on each blade. Furthermore the task of the AFC system on a wind turbine is
to achieve a good balance between various and often conflicting parameters. In this case the
control system is called to analyze all the available input parameters and choose a suitable
control strategy able to satisfy all or most of the operation requirements. Such a control
system is very difficult to model and to realize. Simplified control strategies were developed
in order to drive the AFC elements. Their results were very promising but their performance
is still far from optimal.
Figure 12.2: Schematic representation of the modular flexible flap and spoiler ac-
tuation principle.
Part VII
Thesis Conclusions
13 Discussion On The Overall
Conclusions
13.1 Research Outcome

The current research work focused on the detailed investigation, simulation and experimental
analysis of various AFC solutions for the development of a new generation of wind turbine
blades. For the purpose of the current investigations several software tools and experimen-
tal setups were developed. The following list briefly sums up the software and hardware
developments, as well as the relevant publications and inventions (patent applications).
13.1.1 Experimental Equipment

For the experimental investigations the large wind tunnel facilities of H.T.I TU Berlin were
used. Four different test wings based on wind popular wind turbine airfoils were developed.
Most of these test wings were built following the modular construction concept. This allowed
the application and testing of various AFC elements(see also Sec.6.1).
A custom wind tunnel control software was developed in order to allow the dynamic wind
tunnel AoA variation with simultaneous force measurements in combination with control
signal transmission for the AFC elements. For this purpose a multi-core approach was followed
wich allowed the utilization of all 4-CPUs of a desktop PC for dedicated tasks.
13.1.2 Software Development

For the time efficient CFD simulation of various configurations of single and multi-element
airfoils in large AoA ranges it is necessary to achieve high degree of process automation.
For this reason a custom CFD tool was developed based on the OpenFOAMTM CFD toolbox
and several grid generation tools (GMSH and enGrid) (Sec.4.2). This software tool allowed
the automatic grid generation as well as the automatic AoA variation and data extraction
throughout several computations thus allowing the execution of many computations in a short
time.
For the transformation of the 2D (or better quasi-2D) CFD simulations and the wind tunnel
measurements to useful wind turbine blade performance data, a custom Blade Element Theory
simulation software was developed. This software tool (QBlade) was developed as a seamless
wind turbine blade design tool with full airfoil design and simulation capacity (XFOIL module)
as well as steady state wind turbine performance simulation (BEM module) (Sec.4.2).
The dynamic simulations of AFC elements on wind turbine rotors as well as the virtual
AFC control system research was performed with a custom aeroelasticity code based on the
previously mentioned steady state simulation platform (QBlade). This software tool makes
use of the user friendly QBlade platform and integrates a modified version of the open source
aeroelastic codes of NREL/NWTC (TurbSim, AeroDyn, YawDyn). The aeroelastic codes
are modified in a way that allows the simulation of AFC elements and includes various AFC
controllers. The final software is an extension of QBlade this is called QBlade Active Elements
(QBladeAE) (Sec.5.2).
Through the development of the various experimental and numerical tools it was possible
to evaluate efficiently several AFC solutions on dynamic rotating systems (i.e. wind turbine
rotors).
13.2 Scientific Publications

The results of individual phases of the work were presented and published in the form of
Scientific Papers. These were presented at several technical conferences and were published
at the conference proceedings.
• Active Aerodynamic Control of Wind Turbine Blades with High Deflection Flexi-
ble Flaps, G. Pechlivanoglou, J.Wagner, C.N. Nayeri, C.O. Paschereit, 48th AIAA
Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Ex-
position, 4 - 7 January 2010, Orlando, Florida, USA.
• The Effect of Distributed Roughness on the Power Performance of Wind Turbines, G.

Pechlivanoglou, C.N. Nayeri, C.O. Paschereit, ASME Turbo Expo 2010, 14 - 18 June
2010, Glasgow, Scotland
• Fixed Leading Edge Auxiliary Wing as a Performance Increasing Device for HAWT
Blades, G. Pechlivanoglou. C.N. Nayeri, C.O. Paschereit, DEWEK 2010, 17 - 18 Novem-
ber 2010, Bremen, Germany
• Flow Control Using Plasma Actuators at the Root Region of Wind Turbine Blades,
O. Eisele, G.Pechlivanoglou, C.N. Nayeri, C.O. Paschereit, DEWEK 2010, 17 - 18
November 2010, Bremen, Germany
• Integration of a wind turbine blade design tool in XFOIL/XFLR5, D. Marten, G.Pechlivanoglou,

C.N. Nayeri, C.O. Paschereit, DEWEK 2010, 17 - 18 November 2010, Bremen, Germany
• Performance optimization of wind turbine rotors with active flow control, G.Pechlivanoglou,
C.N. Nayeri, C.O. Paschereit, ASME Turbo Expo 2011, 6 - 10 June 2011, Vancouver,
Canada
• Experimental Investigation of Dynamic Load Control Strategies using Active Microflaps

on Wind Turbine Blades, O. Eisele, G.Pechlivanoglou, C.N. Nayeri, C.O. Paschereit,
EWEA 2011 14-17 March 2011, Brussels, Belgium
• Vortex Generators for wind turbine blades: A combined wind tunnel and wind turbine
parametric study, H. Mueller-Vahl, G.Pechlivanoglou, C.N. Nayeri, C.O. Paschereit,
ASME Turbo Expo 2012, 11 - 15 June 2012, Copenhagen, Denmark
• Performance Optimization of Wind Turbine Rotors with Active Flow Control - PART
2: Active Aeroelastic Simulations, G. Weinzierl, G.Pechlivanoglou, C.N. Nayeri, C.O.
Paschereit, ASME Turbo Expo 2012, 11 - 15 June 2012, Copenhagen, Denmark
13.3 Relevant Patents and Patent Applications

Among the various patents and patent applications of the author, the ones relevant to the
current research work are presented in the list bellow.
• DE102009038768 A1 Airfoil section for complex geometric shaped rotor blade of

wind turbine, has leading edge including flexible outer layer and extended with adjusting
unit to ensure adjustment of outer contour of leading edge
• DE102009051411 A1 Device for determining airflow at rotor blade of wind turbine,

has sensor unit measuring aerodynamic forces caused by airflow at mounting points at
rotor blade and eliminating centrifugal forces at rotor blade during rotation of rotor
blade
• DE102010046711 A1 Rotor blade for wind power plant, has main wing including
leading edge, profile surface and trailing edge, and auxiliary wing arranged at specific
distance from profile surface of main wing during operation
13.4 Feasibility of ”Smart Blades”

After the detailed analysis of several AFC solutions a ”smart blade” design proposal was
created. The proposed blade offers many advantages over the conventional blades used on
commercial wind turbines and effectively manages to eliminate the use of the conventional
pitch system for load alleviation and power regulation. It is however by no means the only
feasible example of a ”smart blade” configuration. The combination of AFC elements selected
for the proposed blade is the result of an overall compromise which aims at balancing all the
technical, aerodynamic, and economic aspects of the design.
It is worth noting that the actual sizes of the AFC elements as well as their performance is
based on measurements and simulations which cannot be validated on an actual wind turbine
since no turbines exist with AFC elements. A certain uncertainty is therefore anticipated and
the results have to be treated with the appropriate caution.
Overall however the current and other research efforts (e.g. [168, 167]) proves the feasibility
of ”smart blades” equipped with AFC systems that are able to replace the traditional pitch
systems. According to the author’s opinion such new blade design concepts will allow the
construction of even larger wind turbines and will consequently boost the large wind turbine
sector even more. The required research efforts in the field of blade manufacturing and control
system strategies for such ”smart blades” are high but the potential of such solutions fully
justifies the development costs.
13.5 Future Work

For the current research project, the author tried to analyze to a sufficient depth the function
and operation of various AFC solutions. Furthermore the best performing AFC solutions
were included in aeroelastic dynamic simulations in order to estimate their performance with
higher accuracy. The problem however of the design of an actual ”smart blade” with AFC
elements is very complex and requires much more interdisciplinary research which is out of
the scope of the current work. The following paragraphs aim to describe the required future
work for the successful continuation of the current project. The required future steps are
divided into categories in order to assist the reader and the future researcher to identify the
research topics that need to be addressed.
13.5.1 Airfoil Sections

All the current aerodynamic and aeroelastic simulation codes make use of the Blade Element
Theory for the computation of the aerodynamic forces on turbine blades. This theory however
has several limitations, especially when used in combination with AFC elements. Without
going to extreme detail, some of the problems of the Active BEM simulations are presented
bellow:
• AFC elements affect dynamic stall but are currently not included in the dynamic stall
models used within the BEM simulations (e.g. Beddoes-Leishman model)
• The BEM theory assumes shed vorticity at the root and the tip of the blades and applies
the well-known Prandtl Root and Tip corrections. The existence however of finite span
flaps, ailerons and spoilers lead to vorticity shedding away from the blade tip and root
region. Suitable corrections have to be developed for such effects.
• The concept of ”jumping polars” used at the current and other (e.g. [43]) active BEM
simulation code makes use of several tables that include steady state airfoil perfor-
mance data. This is naturally a simplification that introduces significant amounts of
inaccuracies especially for rapid AFC element deployment/actuation.
Further developments are therefore necessary in order to achieve reliable on-the-fly compu-
tation of the unsteady aerodynamic effect of wind turbine blades with AFC elements.
Another source of uncertainty regarding the airfoil section performance has to do with the
high AoA and deep stall behavior. The current 360o extrapolation models have significant
drawbacks which become even more significant with the addition of AFC systems. The use of
advanced CFD methods for the computation of the blade performance for the complete AoA
range is proposed. A combination of an Unsteady Panel Method aerodynamic code coupled
with and Integral Boundary Layer Equation solver could possibly be a viable solution to the
aforementioned problem, at least for moderate AoA performance regions.
13.5.2 Dynamic Effects

Most (if not all) the aeroelastic simulation codes incorporate a dynamic stall model which dy-
namically varies the lift and drag values of the airfoil sections according to their instantaneous
aerodynamic conditions. The most common dynamic stall model is the Beddoes-Leishman
model which was developed for helicopter rotors and it is slightly modified in order to account
for the deep stall effects found on wind turbines [8]. The integration of AFC elements requires
further adaptation of the B-L dynamic stall model or even the development of a completely
new model.
The use of an advanced on-the-fly unsteady aerodynamic code, as is proposed above, would
also help in the field of dynamic stall which could be possibly computed rather than modeled.
13.5.3 Aeroelasticity
The aeroelastic model used for the current research project (YawDyn) is a very crude model
which was selected especially for its simplicity. Its low requirements for input parameters
allowed the author to focus mostly on the operational differences of the simulated AFC el-
ements rather than the aeroelastic simulation details. The use of a more advanced code
however is advised for more detailed investigations. The use of the FAST code, which is the
more advanced open source aeroelastic code of NREL/NWTC is one possibility. One has to
note however that FAST has no inherent twist-bent coupling which could become a critical
omission when the trying to simulate the performance of flap or other AFC solutions that
significantly influence the pitching moment coefficient of wind turbine blade sections. For
this reason the modified CurveFAST aeroelastic code should be better suited for such active
element implementation [101]. Naturally an even more advanced multi-body aeroelastic code
would be able to capture the dynamics of the blades with AFC elements even better. How-
ever such codes are complicated and require a very large volume of input data (e.g. blade
spring-damper coefficients) in order to generate reasonable results.
14 Acknowledgments
The extend of the current research field is especially large as well as multidisciplinary. That
meant that the individual tasks of the current research project were highly complicated as
well as interconnected. Several approaches were followed (experimental, analytical, numerical
e.t.c) and custom codes were developed. All that would not have been possible without the
valuable contributions of a number of people.
I would like to acknowledge Prof. Dr.-Ing. C.O. Paschereit as well as Dr.-Ing. C.N. Nayeri.
Without their insight and trust the completion of this work would not have been possible.
I would also like to acknowledge SMART BLADE GmbH and especially of its CEO Dipl.-
Ing. U. Lang for his encouragement, his trust and motivation. The support and significant
technical contribution of Dipl.-Ing. J. Wagner of Tembra GmbH is also highly acknowledged
as well as the contribution of Dipl.-Ing. S. Wetzig.
On the various technical and scientific fields I would like to acknowledge the support and
significant contribution of the following people:
• Dipl.-Ing. S. Führ : Without his brilliant guidance, patient instruction and many tech-
nical contributions in the field of applied aerodynamics, wind tunnel testing and CFD
simulations, this work would not have been possible!!
• Dipl.-Ing. R. van Rennings: For his significant contribution to the project by creating
the automatic CFD grid generator and solver control code ("PolarFOAM").
• Dipl.-Ing. D. Marten: For his excellent work on the programming of the QBlade turbine
blade design and simulation software. His level of skill and motivation is exceptional
and it led to the creation of an extremely useful Open Source software from a vision
and a set of ideas that I had in my mind.
• Dipl.-Ing. O. Eisele: For his excellent work on the plasma actuator research project as
well as the development of the advanced control strategies for the wind tunnel experi-
ments with the active Gurney flap. I am especially glad that our cooperation extends
beyond the scope of the current project and I am glad for our friendship.
14 Acknowledgments 201
• Dipl.-Ing. G. Weinzierl : For his brilliant work with QBlade AE and his valuable input,
his critical discussions and overall high scientific and engineering contributions through-
out the thesis. I am happy to have worked and to be friends with him and I am glad
that I will get to work with him in the future..
• Dipl.-Ing. B. Dolan: For his contribution to QBlade AE and his valuable work with
the functionality of the program.
Furthermore I would like to thank Dipl.-Ing. H. Sader, Dipl. -Ing. A. Lagonikas and Dr.
P. Pechlivanoglou for their valuable discussions and contributions as well as for their patience
during the review of the current document. Special thanks are due to Dipl.-Ing. D. Choidas
who was my mentor, colleague and inspiring force throughout the last decade as well as to
Prof. Dr.-Ing. A. Nassikas for introducing me to the world of alternative energy sources.
I would also like to thank my parents for all their support throughout my academic and
professional career. Last but not least, I would like to thank my life’s partner Myrto Lamprea
for all her priceless support, love and warmth and most of all her patience and understanding
during the hardest part of the current research work (i.e. the compilation of all results and
the composition of the current document).
Bibliography
[1] Wind motor. Pat. No. US 2622686.
[2] D. Custodio C. Henoch H. Johari A. Levshin. Effects of leading edge protuberances on

airfoil performance. AIAA Fluid Dynamics Conference and Exhibit, page 16, 2006.
[3] J. Jacob A. Santhanakrishnan. Effect of regular surface perturbations on flow over an

airfoil. 35th AIAA Fluid Dynamics Conference and Exhibit, page 17, 2005.
[4] A. Darabi I. Wygnanski A. Seifert. Delay of airfoil stall by periodic excitation. Journal
of Aircraft, 33:4, 1996.
[5] I.H. Abbott and A.E. Von Doenhoff. Theory of wing sections. Dover Publications, 1958.
[6] S.R. Spallart S.A. Allmaras. A one equation turbulence model for aerodynamic flows.
AIAA, 1994.
[7] D. Althaus. Niedriggeschwindigkeitsprofile. Vieweg, 1996.
[8] Mac Gaunaa Peter Bj\Orn Andersen. A dynamic stall model for airfoils with deformable
trailing edges. Journal of Physics, 2007.
[9] Peter B. Andersen. Load alleviation on wind turbine blades using variable airfoil geom-
etry (2D and 3D study). MSc Thesis TU Denmark, page 111, 2005.
[10] Peter B. Anderson. Load alleviation on wind turbine blades using variable airfoil ge-
ometry (2d and 3d study), 2005.
[11] J.A. Axelson and G.L. Stevens. Investigation of a slat in several different positions on
an naca 64a010 airfoil for a wide range of subsonic mach numbers. NACA Technical
Note.3129, page 36, 1954.
[12] Marcel Bachman. Dielectric barrier discharge plasma actuation on low reynolds number
airfoils. Diploma Thesis, 2006.
[13] Inventor: E.H. Barber. Aircraft wing construction. Pat. No. US 3136501.
Bibliography 203
[14] T K Barlas and G A M van Kuik. State of the art and prospectives of smart rotor
control for wind turbines. Journal of Physics, 75:21, 2007.
[15] Inventor: G.E. Barnhart. Patent: Airfoil construction. Pat. No. US 2158686.
[16] G. Tillman S.S. Ochs J.S. Kearney B.E. Wake. Control of high-reynolds-number tur-
bulent boundary layer separation using counter-flow fluid injection. 3rd AIAA Flow
Control Conference, page 14, 2006.
[17] Oren Ben-Zeev and Inderjit Chopra. Advances in the development of an intelligent
helicopter rotor employing smart trailing-edge flaps. Smart materials and structures,
5:11–25, 1996.
[18] Inventor: P. Berry. Patent: Segmented flap with variable camber for aircraft wing. Pat.
No. US 6123297.
[19] J. Blazek. Computational fluid dynamics: Principles and applications. ELSEVIER,

2001.
[20] Lee E. Boddy. The high speed characteristics of several flaps and spoilers on the upper
surface of the horizontal stabilizer of a model of a radial engine pursuit airplane. NACA,
1946.
[21] A. L. Braslow. A history of suction-type laminar-flow control with emphasis on flight

research. NASA-Monographs in Aerospace History, 13:31, 1999.
[22] Inventor: S.T. Brown. Patent: Variable camber flap and seal. Pat. No. US 4427169.
[23] H. Bruynes. Fluid mixing device. U.S. Patent No. 2,558,816.
[24] Inventor: E. Bugatti. Patent: Aircraft. Pat. No. US 2279615.
[25] M.L. Buhl. A new empirical relationship between thrust coefficient and induction factor
for the turbulent windmill state. Technical Report, NREL/TR-500-36834, 2005.
[26] T. Burton, D. Sharpe, N. Jenkins, and E. Bossanyi. Wind Energy Handbook. Wiley
Interscience, 2001.
[27] E. Gharaibah G. Toplack A. Gupta W. Wuerz C. Cerretelli. Unsteady separation control

for wind turbine applications at full scale reynolds numbers. 47th AIAA Aerospace
Sciences Meeting, page 13, 2009.
[28] J.F Remacle C. Geuzaine. Gmsh manual: A finite element mesh generator with built-in
pre- and post-processing facilities, October 2010.
Bibliography 204
[29] S. Benjanirat L.N. Sankar C. Tongchitpackdee. Numerical studies of the effect of active
and passive circulation enhancement concepts on wind turbine performance. Journal of
Solar Energy Engineering, 128:12, 2006.
[30] Douglas S. Cairns, J.C. Blockey, and Jon Ehresman. Design and feasibility of active
control surfaces on wind turbine blade systems. AIAA Aerospace Sciences Meeting and
Exhibit, page 12, 2008.
[31] Stefan Anders Lucio Flavio Campanile. Aerodynamic and aeroelastic amplification in
adaptive belt-rib airfoils. Aerospace Science & Technology, 9:55–63, 2005.
[32] Inventor: R.G. Campbell. Aerodynamic brake. Pat. No. US 2400388.
[33] C. Cerretelli, E. Gharaibah, G. Toplack, A. Gupta, and W. Wuerz. Unsteady separation

control for wind turbine applications at full reynolds numbers. AIAA.
[34] J.R. Chambers and L.P. Yip. Aerodynamic characteristics of two general aviation canard
configurations at high angles of attack. AlAA 2nd Applied Aerodynamics Conference,
page 13, 1984.
[35] R. Chow and C. P. van Dam. Computational investigations of small deploying tabs and
flaps for aerodynamic load control. Journal of Physics, 75:11, 2007.
[36] T.E. McLaughlin R.D. VanDyken K.D. Kachner E.J. Jumper T.C. Corke C.L. Enloe.
Mechanisms and responses of a single dielectric barrier plasma. 41st AIAA Aerospace
Sciences Meeting and Exhibit, page 12, 2003.
[37] Inventor: J.B. Cole. Wind edge movable airfoil having variable camber. Pat. No. US
4650140.
[38] Inventor: J.B. Cole. Variable camber leading edge assembly for an airfoil. Pat. No. US
4706913.
[39] J. Cooney. Feasibility of plasma actuators for active flow control over wind turbine
blades. 47th AIAA Aerospace Sciences Meeting, page 11, 2009.
[40] Inventor: T.R. Creel. Boundary layer relaminarization device. Pat. No. US 5205519.
[41] M. M. Murray L. E. Howle F. E. Fish D. S. Miklosovic. Leading-edge tubercles delay

stall on humpback whale “megaptera novaeangliae” flippers. Physics of fluids, page 4,
2004.
Bibliography 205
[42] H. de Vries I. Cleine E. van Emden G.G.M. Zwart H. Stobbe A. Hirschberg H.W.M.
Hoeijmakers C.S. Boeije. Fluidic load control for winf turbine blades. 47th AIAA
Aerospace Sciences Meeting, page 8, 2009.
[43] Wilson D.G., Berg D.E., Lobitz D.W., and Zayas J.R. Optimized active aero dynamic
blade control for load alleviation on large wind turbines. AWEA WINDPOWER 2008
Proceedings, page 7, 2008.
[44] W.S. Diehl. relative loading on biplane wings of unequal chords. NACA Report No.
501, page 6, 1935.
[45] G.I. Font D. Edelstein D.M. Orlov. Characterization of discharge modes of plasma
actuators. AIAA Journal, page 6, 2008.
[46] M. Drela. "XFOIL: An analysis and design system for low reynolds number airfoils".
In Conference on Low Reynolds Number Airfoil Aerodynamics, 1989.
[47] M. Drela and M.B. Giles. "viscous-inviscid analysis of transonic and low reynolds
number airfoils". AIAA Journal, page 10, 1986.
[48] R. Meyer D.W. Bechert and W. Hage. Drag reduction of airfoils with miniflaps. what
can we learn from dragonflies. AIAA (Fluids 2000), page 30, 2000.
[49] S. Alben M.P. Brenner E.A. van Nierop. How bumps on whale flippers delay stall: An
aerodynamic model. Physical Review Letters, page 4, 2008.
[50] O. Eisele. Flow control using plasma actuators at the root region of wind turbine blades.
Master’s thesis, TU Berlin.
[51] O. Eisele, G. Pechlivanoglou, C.N. Nayeri, and C.O. Paschereit. Flow control using
plasma actuators and the root region of wind turbine blades. In DEWEK 2011. DEWI,
2010.
[52] Gad el Hak M. Flow Control: Passive, Active, and Reactive Flow Management. Cam-
bridge University Press, 2007.
[53] Inventor: C. Coleman et. al. Wind turbine rotor aileron. Pat. No. US 5320491.
[54] Inventor: E.J. Breitbach et. al. Profile edge of an aerodynamic profile. Pat. No. US
6076776.
[55] Inventor: K.G. Brady et. al. Airplane wing. Pat. No. US 2749060.
Bibliography 206
[56] Inventor: P.R. Ashill et. al. Airfoil with variable geometry expansion surface. Pat. No.
US 5433404.
[57] Inventor: R.N. Schmidt et al. Fluid flow control devices. Pat. No. US 5755408.
[58] Inventor: R.S. Shevell et. al. Stall control device for swept wings. Pat. No. US 3370810.
[59] Inventor: D.S. Fahrney. Slot foil aircraft wing. Pat. No. US 2172370.
[60] Daniel Feszty, Eric A. Gillies, and Marco Vezza. Alleviation of rotor blade dynamic
stall via trailing edge flap flow control). AIAA 41st Aerospace Sciences Meeting and
Exhibit, page 10, 2003.
[61] J. Fischel and M.F. Ivey. Collection of test data for lateral control with full span flaps.
NACA, 1948.
[62] F.E. Fish and M.J. Battle. Hydrodynamic design of the humpback whale flipper. Journal
of Morphology, 225:9, 1995.
[63] F.E. Fish and G.V. Lauder. Passive and active flow control by swimming fishes and
mammals. Annual Review of Fluid Mechanics, page 34, 2006.
[64] White F.M. Viscous Fluid Flow (3rd Edition). McGraw-Hill, 2005.
[65] EnGits GmbH. engrid - open-source mesh generation, December 2010.
[66] Bert Gollnick. Untersuchung der Moglichkeiten des Einsatzes von Stromungselementen
fur Windenergieanlagen. 2007.
[67] R.A. Granger. Fluid Mechanics. Dover Publications, 1995.
[68] Oertel H. Prandtl’s Essentials of Fluid Mechanics. Springer, 2004.
[69] J. Ketchman H. Velkoff. Effect of an electrostatic field on boundary layer transition.

AIAA Journal, 16:3, 1968.
[70] Martin O.L. Hansen. Aerodynamics of Wind Turbines (2nd Edition). Earthscan, 2008.
[71] M.H Hansen, M. Gaunaa, and H.A. Madsen. A beddoes-leishman type dynamic stall
model in state-space and indical formulations. RISOE Technical Report.
[72] Harris. Biplane with variable camber. ARC RM 677, 1921.
[73] E. Hau. Wind Turbines: Fundamentals, Technologies, Applications, Economics (2nd

Edition). Springer, 2006.
Bibliography 207
[74] H. Himmelskamp. Profiluntersuchungen an einem umlaufenden propeller. Mitteilungen

aus dem Max-Plank-Institut fuer Stroemungsaufgeben, 1950.
[75] H.Mueller-Vahl, G. Pechlivanoglou, C.N. Nayeri, and C.O. Paschereit. Vortex generators
for wind turbine blades: A combined wind tunnel and wind turbine parametric study.
In Proceedings of ASME IGTI Turbo Expo 2012 ASME/IGTI June 11 - 15, 2012,
Copenhagen, Denmark. ASME, 2012.
[76] S.F. Hoerner and H.V. Borst. Fluid-Dynamic Lift. L.A. Hoerner, 1985.
[77] Inventor: A.A. Holle. Aerofoil for aeroplanes. Pat. No. US 1379921.
[78] Inventor: A. Holzem. Wind turbine wing with a pneumatically actuated spoiler. Pat.
No. US 5106265.
[79] E.L. Houghton and P.W. Carpenter. Aerodynamics for engineering students. Elsevier
Publications, 2006.
[80] A. Hunter and B.E. Thomson. Turbulent separated flow over a rounded divergent
trailing edge. AIAA 45th Aerospace Sciences Meeting and Exhibit, page 11, 2007.
[81] Inventor: W.H. Hunter. Aerodynamic brake. Pat. No. US 2428936.
[82] Nuventix Inc. Synjet technology overview. http://www.nuventix.com/technology/,

March 2009.
[83] G. et al. Inventor: Dietz. Patent: Lifting surface with improved separation behavior
under strongly variable angle of attack. Owner: DLR, Pat. No. DE 10 2005 018 427
A1.
[84] W. Geissler Inventor: G. Dietz, H. Mai. Auftriebsflügel mit verbessertem ablöseverhal-

ten bei stark veränderlichem anstellwinkel. Pat. No. DE 102005018427.
[85] R. Thierry Inventor: P. Gillieron. Air flow control method for motor vehicle, involves
generating air flow through slits arranged on wall such that separation of air boundary
layer from outer surface of wall shift forwards in direction opposite to vehicle movement
direction. Pat. No. FR 2859160.
[86] K. J. Jackson. Innovative design approaches for large wind turbine blades. Wind Energy,
8:141–171, 2005.
[87] J.D. Jacob. On the fluid dynamics of adaptive airfoils. In 1998 ASME International
Mechanical Engineering Congress and Exposition, 1998.
Bibliography 208
[88] N. Jacobs. Airfoil section characteristics as affected by protuberances. NACA, Report

no. 446, 1934.
[89] N. Jacobs. Airfoil section characteristics as affected by protuberances of short span.

NACA, Report no. 449, 1934.
[90] J.A.Kelly and G.b. McCullough. Aerodynamic loads on a leading-edge flap and a
leading-edge slat on the naca 64a010 airfoil section. NACA Technical Note.3220,
page 34, 1954.
[91] N.R. Foster J.D. Martin. Exploiting the close-coupled canard. AIAA/AHS/ASEE
Aircraft Design, Systems and Operations Meeting, page 7, 1987.
[92] K. Appa C.A. Martin J.N. Kudva and A. Peter Jardine. Design, fabrication, and testing
of the darpa / wright lab "smart wing" wind tunnel model. AIAA, page 6.
[93] Donny P. Wang Jonathan D. Bartley-Cho and Jayanth N. Kudva. Shape estimation
of deforming structures. 42nd AIAA/ASMBASCE/AHS/ASC Structures, Structural
Dynamics, and Materials Conference and Exhibit, page 11.
[94] B.J. Jonkman. TurbSim user’s guide. NREL, Technical Report NREL/TP-500-46198,
2009.
[95] Inventor: O.E. Bartoe Jr. Patent: Air foil structure. Pat. No. US 3807663.
[96] J.H. Kim and O.H. Rho. Numerical simulation of the flowfield around an airfoil with
stationary and oscillating spoiler. AIAA, 1996.
[97] Inventor: S. Kota and J.A. Hetrick. Patent: Adaptive compliant wing and rotor system.
Pat. No. US 2006/0186269.
[98] Inventor: K.M. Krall. Airfoil construction. Pat. No. US 4296900.
[99] D.J. Laino and A.C. Hansen. YawDyn user’s guide. NREL, Technical Report, 2003.
[100] J. Laks, L. Pao, and A. Wright. "combined feed-forward/feedback control of wind

turbines to reduce blade flap bending moments". In Proceedings of 47th AIAA Aerospace
Sciences Meeting Including The New Horizons Forum and Aerospace Exposition, 5 - 8
January 2009, Orlando, Florida. AIAA, 2009.
[101] S. Larwood. Dynamic Analysis Tool Development for Advanced Geometry Wind Turbine
Blades. PhD Dissertation: University of Davis California, 2009.
Bibliography 209
[102] H. Lawson-Tancred. Wind turbine blades. Pat. No. US 4692095.
[103] J.G. Lee. Control device for airplanes. Pat. No. US 2164531.
[104] Victor R. Lessard. Low Speed Analysis of Mission Adaptive Flaps on a High Speed Civil
Transport Configuration. NASA, 1999.
[105] R.H. Liebeck. On the design of subsonic airfoils for high lift. AIAA 9th Fluid and
Plasma Dynamics Conference, page 24, 1979.
[106] Inventor: D.G. Lyon. Patent: Variable shaped airfoil. Pat. No. US 3179357.
[107] D. Pitt V. Kibens D. Parekh A. Glezer M. Amitay. Control of internal flow separation
using synthetic jet actuators. AIAA Paper 2000-0903, page 13, 2000.
[108] E. Chatlynne N. Rumigny A. Glezer M. Amitay. Virtual aero-shaping of a clark-y airfoil

using synthetic jet actuators. AIAA Paper 2001-0732, page 13, 2001.
[109] E.A. Mayda and C.P. van Dam. Computational investigation of finite width microtabs
for aerodynamic load control. AIAA, page 13, 2005.
[110] T. Olsen G. Quandt M.C. Cheney and P. Arcidiacono. Analysis and tests of pultruded
blades for wind turbine rotors. NREL Subcontractor Report, 1999.
[111] Inventor: M.E. McKinney. Variable camber leading edge airfoil system. Pat. No. US
4040579.
[112] H.C. Seetharam W.G. Kuhn M.D. Mack and J.T. Bright. Aerodynamics of spoiler
control devices. AIAA Aircraft Systems and Technology Meeting, 1979.
[113] F.R. Menter. Two equation eddy-viscosity turbulence models for engineering applica-
tions. AIAA Journal, 32, 1994.
[114] Robert K. J. Meyer. Experimentelle untersuchungen von rückstromklappen auf

tragflügeln zur beeinflussung von strömungsablösungen. PhD Thesis TU Berlin, page
125, 2000.
[115] R. Von Mises. Theory of Flight. Dover Publications, 1959.
[116] Hans Peter Monner. Realization of an optimized wing camber by using formvariable
flap structures. Aerospace Science & Technology, 5:445–455, 2001.
Bibliography 210
[117] B. Montgomerie. Methods for root effects, tip effects and extending the angle of attack
range to + − 100o with application to aerodynamics for blades on wind turbines and
propellers. FOI Swedish Defence Research Agency, Scientific Report FOI-R-1035-SE,
2004.
[118] P.J. Moriarty and A.C. Hansen. AeroDyn theory manual. NREL, Scientific Report
NREL/EL-500-36881, 2004.
[119] M.M. Munk. General biplane theory. NACA Report No. 151, page 47, 1923.
[120] D.H. Neuhart and O.C. Pendergraft. A water tunnel study of gurney flaps. NASA
Technical Memorandum 4071, 1988.
[121] H. Neuhierl. Stoer oder bremsklappe fuer modellflugzeuge. Pat. No. DE 7738361.
[122] G. Pechlivanoglou N. Nayeri O. Eisele and O. Paschereit. Experimental investigations

of dynamic load control strategies using micro tabs on wind turbine blades. EWEA,
page 10, 2011.
[123] OpenCFD. OpenfoamÂ :

R open source cfd, December 2010.
[124] Inventor: H. Page. Wing and similar member of aircraft. Pat. No. US 1353666.
[125] H.F. Parker. The parker variable camber wing. NACA Report No.77, page 40, 1920.
[126] Inventor: H.F. Parker. Patent: Variable camber rib for aeroplane wings. Pat. No. US
1341758.
[127] G. Pechlivanoglou, S. Fuehr, C.N. Nayeri, and C.O. Paschereit. The effect of distributed
roughness on the power performance of wind turbines. In Proceedings of ASME IGTI
Turbo Expo 2010 ASME/IGTI June 14 - 18, 2010, Glasgow, Scotland, UK. ASME,
2010.
[128] G. Pechlivanoglou, D. Marten, C.N. Nayeri, and C.O. Paschereit. Integration of a wind
turbine blade design tool in xfoil/xflr5. In DEWEK 2011. DEWI, 2010.
[129] G. Pechlivanoglou, C.N. Nayeri, and C.O. Paschereit. Fixed leading edge auxiliary wing
as a performance increasing device for hawt blades. In DEWEK 2011. DEWI, 2010.
[130] G. Pechlivanoglou, C.N. Nayeri, and C.O. Paschereit. Performance optimization of wind
turbine rotors with active flow control. In Proceedings of ASME IGTI Turbo Expo 2011
ASME/IGTI June 6 - 10, 2011, Vancouver, Canada. ASME, 2010.
Bibliography 211
[131] G. Pechlivanoglou, J. Wagner, C.N. Nayeri, and C.O. Paschereit. Active aerodynamic
control of wind turbine blades with high deflection flexible flaps. In 48th AIAA Aerospace
Sciences Meeting Including the New Horizons Forum and Aerospace Exposition. AIAA,
2010.
[132] Inventor: J. Perez. Patent: Mechanism for at least regionally adjusting the curvature
of airfoil wings. Pat. No. US 6644599.
[133] Inventor: D. Pierce. Fluid dynamic lift generating of control force generating structures.
Pat. No. US 3716209.
[134] Inventor: D. Pierce. Patent: Flexible aerofoils. Pat. No. US 4113210.
[135] D.M. Pitt and Peters D.A. Theoretical prediction of dynamic-inflow derivatives. Vertica,
5, 1981.
[136] Pope, Rae, and Barlow. Low Speed Wind Tunnel Testing. Wiley Interscience, 2006.
[137] L. Prandtl. Ueber fluessigkeitsbewegungen bei sehr kleiner reibung. International Math-
ematical Congress, Heidelberg, 3:7, 1904.
[138] Proceedings. The application of smart structures for large wind turbine rotor. In 50th
IEA Topical Expert Meeting. TU Delft, 2006.
[139] R. R. Ramsay, J. M. Janiszewska, and G. M. Gregorek. Wind tunnel testing of an s809

spoiler flap model (for wind turbines). AIAA Meeting papers, page 11, 1997.
[140] J. Rauch. Grundlegende Untersuchung zu aerodynamischen Konzepten der Lastregelung

und Leistungsregelung als Alternative zur Pitch-Regelung für Rotoren von Windkraftan-
lagen. Master Thesis, 2008.
[141] T.C. Corke H. Othman M.P. Patel S. Vasudevan T. Ng R.C. Nelson. A smart wind
turbine blade using distributed plasma actuators for improved performance. 46th AIAA
Aerospace Sciences Meeting and Exhibit, page 17, 2008.
[142] H. Schlichting and K. Gersten. Boundary Layer Theory. Springer Verlag, 1999.
[143] Inventor: B. Schlipf. Adjusting device for adjusting a high lift flap and airfoil wing
comprising such an adjusting device. Pat. No. WO 2008001336.
[144] W.Z. Shen, R. Mikkelsen, J.N. Sorensen, and C. Bak. Tip loss corrections for wind
turbine computations. Wind Energy, 2005.
Bibliography 212
[145] Inventor: A. Sigalla. Airplane all-moving airfoil with moment reducing apex. Pat. No.
US 4291853.
[146] A.M.O. Smith. High-lift aerodynamics. Journal of Aircraft, page 30, 1975.
[147] H. Snel, R. Houwking, G.J.W. van Bussel, and A. Bruining. Sectional prediction of
3d effects for stalled flow on rotating blades and comparison with measurements. In
Proceedings of EWEC. H.S. Stevens & Associates, 1993.
[148] N. N. Soerensen, J.A. Michelsen, and S. Schreck. Navier Stokes predictions of the NREL
Phase VI rotor in the NASA Ames 80 ft by 120 ft wind tunnel. Wind Energy, 5.
[149] U.P. Solies. In flight flow angle measurements on the ball-bartoe jetwing powered lift
aircraft. AIAA, page 11, 1990.
[150] D.M. Somers. Effect of Flap Deflection on Section Characteristics of S813 Airfoil (Pe-
riod of Performance: 1993 - 1994). NREL, 2005.
[151] J. D. Sorenson. Influence of boundary layer blowing on the low-speed aerodynamic per-
formance of a 45degree swept wing airplane. AIAA 10th Annual Meeting and Technical
Display, page 13, 1974.
[152] David A. Spera. Wind Turbine Technology (2nd Edition). ASME, 2009.
[153] K.J. Standish and C.P. van Dam. Analysis of blunt trailing edge airfoils. AIAA 41st
[154] E. Stanewsky. Aerodynamic benefits of adaptive wing technology. Aerospace Science &
Technology, 4:439–452, 2000.
[155] E. Stanewsky. Adaptive wing and flow control technology. Progress in aerospace science,
37:85, 2001.
[156] H. Stiesdal. Method for regulating a windmill and an apparatus for the use of said
method. Pat. No. US 2003/0091436 A1.
[157] B.L. Storms and C.S. Jang. Lift enhancement of an airfoil using a gurney flap and
vortex generators. AIAA 31st Aerospace Sciences Meeting and Exhibit, page 8, 1993.
[158] F.K. Straub. Design of a servo flap rotor for reduced control loads. Smart materials
and structures, IOP Publishing:68–75, 1996.
[159] F.K. Straub. A feasibility study of using smart materials for rotor control. Smart
materials and structures, IOP Publishing:1 – 10, 1996.
Bibliography 213
[160] H T Ngo F K Straub and D B Domzalski. Development of a piezoelectric actuator

for trailing edge flap control of full scale rotor blades. Smart materials and structures,
10:25–34, 2001.
[161] H. D. Taylor. The elimination of diffuser separation by vortex generators. United

Aircraft Corporation Technical Report R-4012-3.
[162] F. Thomas and J. Milgram. Fundamentals of Sailplane Design. College Park Press,
1999.
[163] Thompson. Counter-rotating vortices generator for an aircraft wing. Pat. No. US
4323209.
[164] W.A Timmer and A.P. Schaffarczyk. The effect of roughness at high reynolds numbers
on the performance of aerofoil DU 97-W-300Mod. Wind Energy, 7, 2004.
[165] W.A. Timmer and R.P.J.O.M van Rooij. Summary of the delft university wind turbine
dedicated airfoils. AIAA 41st Aerospace Sciences Meeting and Exhibit, page 11, 2003.
[166] Niels Troldborg. Computational Study of the Ris −B1 − 18 Airfoil Equipped with
Actively Controlled Trailing Edge Flaps. PhD thesis, TU Denmark, September 2004.
[167] C.P. van Dam D.E. Berg S.J. Johnson. Active load control techniques for wind turbines.
SANDIA Report, 66:132, 2008.
[168] J. W. van Wingerden, A. W. Hulskamp, B. Marrant T. Barlas, G. A. M. van Kuik,

D.-P. Molenaar, and M. Verhaegen. On the proof of concept of a “smart” wind turbine
rotor blade for load alleviation. Wind Energy, Wiley Interscience:16, 2008.
[169] J-W van Wingerden A. Hulskamp T. Barlas and G. van Kuik. Closed loop control wind
tunnel tests on an adaptive wind turbine blade for load reduction. AIAA 46th Aerospace
Sciences Meeting and Exhibit, page 11, 2008.
[170] C. M. Velte, M. O. L. Hansen K. E. Meyer, and P. Fuglsang. Evaluation of the per-

formance of vortex generators on the DU 91 − W 2 − 250 profile using stereoscopic piv.
IIISCI, 7.
[171] L.A. Viterna and R.D. Corrigan. Fixed pitch rotor performance of large hawts. In
Proceedings, workshop on large HAWTs, NASA CP-2230. DOE Publication, NASA
Lewis Research Center, 1981.
[172] H. Wagner. An Tragfluegeln angeordnete Abreissleisten. Pat. No. DE 727732.

Bibliography 214
[173] H. Wagner. Aircraft having its wings provided with disturbing bodies. Pat. No. US
2218128.
[174] D.K. Walters and D. Cokljat. A Three-Equation Eddy-Viscosity model for Reynolds
Averaged Navier Stokes simulations of transitional flow. Journal of Fluids Engineering,
130, 2008.
[175] Watts. Scalloped wing leading edge. Pat. No. US 6431498.
[176] F.E. Weick and C.J Wenzinger. Wind tunnel research comparing lateral control devices,
particularly at high angle of attack (upper surface ailerons on wings with split flaps).
NACA Report No.499, page 15, 1934.
[177] J.A. Weick. The effect of multiple fixed slots and a trailing-edge flap on the lift and
drag of a clark y airfoil. NACA Report No.427, page 6, 1933.
[178] G. Weinzierl. A BEM based simulation tool for wind turbine blades with active flow
control elements. Master’s thesis, TU Berlin.
[179] G. Weinzierl, G. Pechlivanoglou, C.N. Nayeri, and C.O. Paschereit. Performance op-
timization of wind turbine rotors with active flow control, part 2: Active aeroelastic
simulations. In Proceedings of ASME IGTI Turbo Expo 2012 ASME/IGTI June 11 -
15, 2012, Copenhagen, Denmark. ASME, 2012.
[180] Inventor: N.A. Weisend. Inflatable airfoil device. Pat. No. US 6443394.
[181] C. J. Wenzinger. Tests of round and flat spoilers on a tapered wing in the NACA 19foot
pressure wind tunnel. NACA, 1941.
[182] C. Ostowari W.H. Wentz and H.C. Seetharam. Effects of design variables on spoiler con-
trol effectiveness hinge moments and wake turbulence. AIAA 19th Aerospace Sciences
Meeting, 1981.
[183] I. Wygnanski. Some observations affecting the control of separation by periodic excita-
tion. AIAA Paper 2314-2000, 2000.
[184] N.T. Yerkes. Pneumatic artificial muscle activation for trailing edge flaps. AIAA 46th

Dokument 5

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dokument 5

Uploaded by

Copyright:

Available Formats

Passive and active flow control solutions for wind

von der Fakultät V - Verkehrs- und Maschinensysteme

Doktor der Ingenieurwissenschaften

Tag der wissenschaftliche Aussprache: 21. Februar 2012

Ω blade angular velocity

5 Wind Turbine Simulations 26

III Experimental Setup 42

6 Wind Tunnel Experiments 43

IV Initial Flow Control Solution Selection 51

8 Passive Flow Control 55

8.4 L.E. Protuberances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

9 Active Flow Control 68

V Experimental and Numerical Investigations 112

10 Passive Flow Control Solutions 113

11 Active Flow Control Solutions 135

VI Blade Design 185

12 ”Smart Blade” Design Characteristics 186

VII Thesis Conclusions 193

13 Discussion On The Overall Conclusions 194

2.1 Basic airfoil design and definition parameters [73]. . . . . . . . . . . . 9

3.1 Flow velocity vectors acting on a wind turbine rotor section. . . . . . . 14

4.1 Comparison of the measured performance polar curve of DU 97-W-300

5.1 Blade representation with independent airfoil segments according to the

7.1 Schematic representations of all the elements included in the preliminary

8.1 Vortilon and Vortex Generator configuration . . . . . . . . . . . . . . . 55

8.8 Sinusoidal leading edge protuberances based on the Humpback Whale

9.19 Schematic comparison of the conventional airfoil performance (left) and

10.5 Comparative view of the “baseline” configuration simulation in compar-

11.6 Schematic representation of the DU96W180 airfoil and comparison with

11.23Comparison of the wind tunnel measurements and the XFOIL simulation

12.1 Schematic representation of the proposed “Smart Blade”. . . . . . . . . 187

6.1 Table of specifications for the wind tunnel of H.F.I TU Berlin. . . . . . 44

11.1 Specifications of the parameter setup for the numerical simulations of

1.1 Wind Turbine State of the Art

1.2 Current Problems

1. Yaw misalignment due to flow vane errors

2. Inability to compensate the non-uniform blade inflow

3. Inability to adjust to turbulent inflow

4. High fatigue loads

5. Strong aeroelastic coupling

6. Inability to damp blade and turbine oscillations

7. Response delay of existing control systems

8. Increased electrical and mechanical complexity due to pitch system

9. High actuation energy requirements for pitch systems

1.3 Thesis Objectives

3. Better aerodynamic load management

4. Better power regulation

5. Reduced cost through innovative design concepts

1.4 Research Methodology

2.1 Airfoil Design Parameters

Figure 2.1: Basic airfoil design and definition parameters [73].

2.2 Airfoil Performance Characteristics

• Surface roughness tolerance [127]

• Stall initiating from the trailing edge

• No laminar separation bubbles

• Smooth stall (no rapid loss of lift at stall)

• Now noise generation

• High structural stiffness

• High glide ratio at the design point

• Clearly defined laminar-turbulent transition

3.1 Rotating Blade Aerodynamics

Figure 3.2: Distribution of aerodynamic forces along the rotor blades.

3.2 Unsteady Inflow