...

Claessens-Thesis-2006.PDF

by user

on
Category: Documents
7

views

Report

Comments

Transcript

Claessens-Thesis-2006.PDF
Master of Science Thesis
The Design and Testing of Airfoils
for Application in Small Vertical
Axis Wind Turbines
M.C. Claessens
November 9, 2006
Ad
The Design and Testing of Airfoils
for Application in Small Vertical
Axis Wind Turbines
Master of Science Thesis
For obtaining the degree of Master of Science in Aerospace
Engineering at Delft University of Technology
M.C. Claessens
November 9, 2006
Faculty of Aerospace Engineering
·
Delft University of Technology
Delft University of Technology
c Aerospace Engineering, Delft University of Technology
Copyright All rights reserved.
DELFT UNIVERSITY OF TECHNOLOGY
DEPARTMENT OF
AERODYNAMICS
The undersigned hereby certify that they have read and recommend to the Faculty of
Aerospace Engineering for acceptance the thesis entitled “The Design and Testing
of Airfoils for Application in Small Vertical Axis Wind Turbines” by M.C.
Claessens in fulfillment of the requirements for the degree of Master of Science.
Dated: November 9, 2006
Supervisors:
Prof.Dr.Ir. P.G. Bakker
Ir. W.A. Timmer
Ir. C. Ferreira
Dr.Ir. S. Mertens
Preface
This report is the final work of my Master of Science thesis at Aerospace Engineering,
TU Delft. In this report the research that I have did during the 18 months of my
studies in the area of VAWT aerodynamics can be found.
I would like to thank my supervisor Nando Timmer for all his time, the people at the
LTT wind tunnel, Daniel Twigt for reviewing my report and everybody else who has
contributed to my report. Special thanks go out to my girlfriend and my family for all
their support.
Maarten Claessens
Delft, the 2nd of October, 2006
M.Sc. thesis
M.C. Claessens
vi
M.C. Claessens
Preface
M.Sc. thesis
Summary
In recent years more focus is put on the applications of wind turbines in the urban
environment. One of the ways to do this is using a turbine with a vertical axis (a
VAWT). This type of wind turbines is around for many centuries. The modern equivalent which is based on lift producing blades only exists for 30 years. In this period
airfoils for this application have been developed, but still much work can be done in
this field.
During this project a design process is developed with the purpose of improving the
NACA 0018 airfoil, which is commonly used in VAWT turbines. The aerodynamics
involved in VAWT are investigated to find the design goals for the airfoil characteristics.
Furthermore the currently used NACA 0018, which is used as the design reference, is
investigated.
The two main pillars of the design process are the RFOIL program and the VAWT
simulation program. RFOIL is a panel method based program with boundary layer
equations which can calculate the properties of 2D airfoils. RFOIL gives accurate
enough results to be a powerful design tool in the Reynolds number range from 300,000
to 700,000.
The VAWT simulation program, written in Matlab, calculates the performance of a
VAWT using 2D airfoil data. The final Matlab program allows to adjust the turbines
geometry, to chose from multiple airfoils and to set a dynamic stall model on or off.
As such 2D airfoil characteristic from RFOIL or wind tunnel tests can be inserted to
view the turbines performance with this airfoil.
M.Sc. thesis
M.C. Claessens
viii
Summary
Real life turbine
Flow conditions
2D airfoil shape
Design criteria
Airfoil design
Turbine
simulation
Performance
results
Airfoil data (RFOIL)
This design process resulted in a final design: the DU 06-W-200. The Du 06-W-200
airfoil is a laminar, 20% thick airfoil with 0.8% camber. The original NACA 0018
airfoil is a turbulent, symmetric airfoil with 18% thickness. To be able to compare
both airfoils wind tunnel measurements were performed in the LTT wind tunnel at the
Delft University of Technology. For the DU 06-W-200 airfoil the following conclusions
are made:
• The added thickness of 2% will add to the blade strength and this is reached
without decreasing the performance
• The added camber of 0.8% increases the performance with respect to a symmetric
airfoil
• The DU 06-W-200 performance equals the NACA 0018 for negative angles of
attack
• The DU 06-W-200 has a much higher CL,max for positive angles of attack, resulting in a wider drag bucket
• Deep stall occurs at higher angles of attack with a smaller drop in lift coefficient
• In contrast to the NACA 0018 the DU 06-W-200 does not have laminar separation
bubbles which extend over the trailing edge
• The increase in turbine performance at the operating tip speed ratio of λ = 3 is
8% in clean conditions and twice as much when dirty
M.C. Claessens
M.Sc. thesis
Table of Contents
Preface
v
Summary
vii
List of Figures
xiii
List of Tables
xvii
List of Symbols
xix
1 Introduction
1
1.1
1.2
Wind turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
VAWT basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
2
1.3
Project overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
2 VAWT aerodynamics
7
2.1
Basic aerodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.2
Flow conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1 Angle of attack . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
9
2.3
2.2.2
Deep stall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3
Dynamic stall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.4
Reynolds number . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.5
Laminar separation bubbles . . . . . . . . . . . . . . . . . . . . . . 17
2.2.6
Virtual camber
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Design criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
M.Sc. thesis
M.C. Claessens
x
Table of Contents
3 Simulation methods
3.1 Momentum theory based models . . . . . .
3.1.1 Single Streamtube Model . . . . . .
3.1.2 Multiple Streamtubes Model . . . .
3.1.3 Double actuator disc theory . . . . .
3.1.4 Double-Multiple Streamtubes Model
3.2 Non - momentum theory based models . . .
3.2.1 Vortex models . . . . . . . . . . . .
3.2.2 CFD models . . . . . . . . . . . . .
3.3 Dynamic stall models . . . . . . . . . . . .
3.3.1 Gormonts model . . . . . . . . . . .
3.3.2 Strickland et al. modification . . . .
3.3.3 Paraschivoiu et al. modification . .
3.3.4 Massé and Berg modification . . . .
3.4 Simulation model . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
21
21
22
23
24
24
27
27
27
28
28
29
31
31
32
4 RFOIL
4.1 Turbulent profiles . . . . . .
4.2 Laminar profiles . . . . . . .
4.2.1 NACA 64-418 . . . .
4.2.2 NLF-0416 . . . . . .
4.2.3 S824 . . . . . . . . .
4.3 Higher Reynolds numbers . .
4.4 Fixed transition at 5% chord
4.5 Conclusions . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
33
34
37
37
38
39
40
42
44
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
45
45
46
48
50
52
53
55
55
55
58
60
60
60
62
64
67
.
.
.
.
.
.
.
.
5 Airfoil design parameters
5.1 Design variables . . . . . . . .
5.1.1 Thickness . . . . . . .
5.1.2 Camber . . . . . . . .
5.1.3 Boundary layer tripping
5.1.4 Noise . . . . . . . . . .
5.1.5 Self starting turbines .
5.2 Turbulent profiles . . . . . . .
5.2.1 Thickness . . . . . . .
5.2.2 Camber . . . . . . . .
5.2.3 Simulation . . . . . . .
5.3 Laminar profiles . . . . . . . .
5.3.1 The NACA 6-series . .
Simulation . . . . . . .
5.3.2 NLF profiles . . . . . .
5.3.3 Experimental results . .
5.4 Conclusions . . . . . . . . . .
M.C. Claessens
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
M.Sc. thesis
Table of Contents
xi
6 Airfoil design
69
6.1
6.2
6.3
Thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Camber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Fine tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.4
Final design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.5
RFOIL comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.6
6.5.1 Free transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.5.2 Fixed transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Measurement comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.7
6.6.1 Free transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.6.2 Fixed transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7 Conclusions and recommendations
87
7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
8 Bibliography
91
A Airfoil coordinates
93
A.1 NACA 0018 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
A.2 DU 06-W-200 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
B NACA 0018 wind tunnel results
97
B.1 Test setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
B.2 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
B.2.1 Wake rake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
B.2.2 Balance system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
B.3 Free transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
B.4 With trip applied . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
B.5 Fixed transition at 5% chord . . . . . . . . . . . . . . . . . . . . . . . . . 102
B.6 Large angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
C DU 06-W-200 RFOIL data
105
C.1 RFOIL characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
C.2 Pressure distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
D Du 06-W-200 wind tunnel results
D.1 Free transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
D.2 Transition fixed at 5% chord . . . . . . . . . . . . . . . . . . . . . . . . .
D.3 Large angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
M.Sc. thesis
109
110
111
112
M.C. Claessens
xii
M.C. Claessens
Table of Contents
M.Sc. thesis
List of Figures
1.1
The two main designs of lift driven wind turbines . . . . . . . . . . . . . .
1
1.2
Flow diagram of the modeling of the turbine . . . . . . . . . . . . . . . . .
4
1.3
More detailed overview of the airfoil design . . . . . . . . . . . . . . . . .
5
2.1
2.2
2D cross section from the turbine . . . . . . . . . . . . . . . . . . . . . . .
Power efficiency results for a VAWT at Vinf = 10 m
s . . . . . . . . . . . . .
8
8
2.3
The angle of attack variation as function of θ at Vinf = 10 m
s . . . . . . . .
9
2.4
The deep stall characteristics of the NACA 0018 profile . . . . . . . . . . . 10
2.5
The deep stall characteristics of multiple profiles . . . . . . . . . . . . . . . 11
2.6
Visualization of dynamic stall at λ = 2.14 . . . . . . . . . . . . . . . . . . 12
2.7
Flow visualization at 4 different positions . . . . . . . . . . . . . . . . . . . 13
2.8
Schematic illustration of the dynamic stall for different λ . . . . . . . . . . 13
2.9
The Reynolds number variation at Vinf = 10 m/s . . . . . . . . . . . . . . 14
2.10 Reynolds effects on the lift curve of the NACA 0018 profile . . . . . . . . . 15
2.11 Reynolds effects on the maximum lift the NACA 00xx series profiles . . . . 15
2.12 Reynolds number influence on the Sandia 5 meter turbine . . . . . . . . . . 16
2.13 Reynolds number influence on the Sandia 2 meter test rotor . . . . . . . . 16
2.14 Laminar separation bubbles for the NACA 0018 . . . . . . . . . . . . . . . 17
2.15 The principal of virtual camber as a result of curvilinear flow . . . . . . . . 18
3.1
Overview of the development of the streamtube models . . . . . . . . . . . 22
3.2
Components of the local angle of attack . . . . . . . . . . . . . . . . . . . 22
3.3
3.4
3.5
2D schematic of the streamtube model . . . . . . . . . . . . . . . . . . . . 23
Schematic of the two actuator discs behind each other . . . . . . . . . . . 24
The Strickland dynamic stall model applied to the NACA 0018 . . . . . . . 30
4.1
NACA 0018 characteristics at Re=300,000 trip at 70% . . . . . . . . . . . 34
4.2
NACA 0018 characteristics at Re=500,000 trip at 80% . . . . . . . . . . . 35
M.Sc. thesis
M.C. Claessens
xiv
List of Figures
4.3
4.4
4.5
4.6
4.7
4.8
4.9
4.10
NACA 0018 characteristics at Re=700,000 trip at 80% . . . . . . . . .
NACA 64-418 characteristics at Re=280,000 . . . . . . . . . . . . . . .
NACA 64-418 characteristics at Re=700,000 . . . . . . . . . . . . . . .
NLF 0416 characteristics at Re=500,000 . . . . . . . . . . . . . . . . .
S824 characteristics at Re=720,000 . . . . . . . . . . . . . . . . . . .
NACA 4418 characteristics at Re=3,000,000 . . . . . . . . . . . . . . .
NACA 64-218 characteristics at Re=3,000,000 . . . . . . . . . . . . . .
Comparison between 5% trip RFOIL and measured data at Re=300,000
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
36
37
38
39
40
41
41
42
4.11 Comparison between 5% trip RFOIL and measured data at Re=700,000 . . 43
5.1
5.2
5.3
5.4
NACA characteristics with different thickness at Re≈250,000
Symmetric profiles with varying thickness at Re=150,000 . .
The Sandia 5m turbine with NACA 0015 and 0012 blades .
Variation of wind speed in the VAWT . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
47
47
48
49
5.5
5.6
5.7
5.8
5.9
NACA characteristics with different camber at Re≈250,000 . . .
SA7026 airfoil characteristics at Re=100,000 . . . . . . . . . .
SA7026 airfoil characteristics at Re=300,000 . . . . . . . . . .
The radiated aerodynamic noise from a VAWT . . . . . . . . .
Camber influence on the performance coefficient at Re=200,000
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
50
51
51
52
53
5.10
5.11
5.12
5.13
Influence of thickness on the turbine performance at Re=200,000 . . .
Calculated characteristics of different turbulent profiles at Re=300,000 .
Calculated characteristics of different turbulent profiles at Re=700,000 .
Calculated characteristics of different cambered profiles at Re=300,000
.
.
.
.
.
.
.
.
54
56
56
57
5.14
5.15
5.16
5.17
5.18
Calculated characteristics of different cambered profiles at Re=700,000
Turbulent profile characteristics for thickness variation . . . . . . . . . .
Turbulent profile characteristics for camber variation . . . . . . . . . .
Calculated characteristics of different laminar profiles at Re=300,000 . .
Calculated characteristics of different laminar profiles at Re=700,000 . .
.
.
.
.
.
.
.
.
.
.
57
58
59
61
61
5.19
5.20
5.21
5.22
Simulation of laminar profiles compared to the NACA 0018 . . . . . .
Characteristics for NLF airfoils with increasing camber at Re=300,000
Characteristics for NLF airfoils with increasing camber at Re=700,000
Simulation of NLF profiles compared to the NACA 0018 . . . . . . .
.
.
.
.
.
.
.
.
62
63
63
64
.
.
.
.
5.23 Comparison of the SNLA 0018 profile with the NACA 0018 . . . . . . . . . 65
5.24 The NACA 0018 and SNLA 0018 characteristics for Re=3 million . . . . . 65
5.25 Results of the Sandia 34m turbine blades at 34rpm. . . . . . . . . . . . . . 66
6.1
6.2
Variation in thickness at Re=500,000 . . . . . . . . . . . . . . . . . . . . . 70
Performance results for NLF profiles with different thickness . . . . . . . . 71
6.3
Variation in camber at Re=500,000
M.C. Claessens
. . . . . . . . . . . . . . . . . . . . . 72
M.Sc. thesis
List of Figures
xv
6.4
Performance results for NLF profiles with different camber . . . . . . . . . 72
6.5
Airfoil nose radius optimization . . . . . . . . . . . . . . . . . . . . . . . . 73
6.6
Removal of bump in the pressure distribution . . . . . . . . . . . . . . . . 74
6.7
Final shape of the DU 06-W-200 compared with the NACA 0018 . . . . . . 75
6.8
Comparison clean RFOIL and wind tunnel data for Re=300,000
. . . . . . 76
6.9
Comparison clean RFOIL and wind tunnel data for Re=500,000
. . . . . . 77
6.10 Comparison clean RFOIL and wind tunnel data for Re=700,000
. . . . . . 77
6.11 Comparison fixed transition RFOIL and measured data for Du 06-W-200 . . 78
6.12 Comparison clean wind tunnel data for Re=300,000 . . . . . . . . . . . . . 80
6.13 Comparison clean wind tunnel data for Re=500,000 . . . . . . . . . . . . . 80
6.14 Comparison clean wind tunnel data for Re=700,000 . . . . . . . . . . . . . 81
6.15 Comparison clean wind tunnel data for Re=1,000,000 . . . . . . . . . . . . 81
6.16 Comparison clean lift over drag data for Re=500,000 . . . . . . . . . . . . 82
6.17 Comparison clean lift over drag data for Re=700,000 . . . . . . . . . . . . 82
6.18 Comparison dirty wind tunnel data for Re=300,000 . . . . . . . . . . . . . 83
6.19 Comparison dirty wind tunnel data for Re=700,000 . . . . . . . . . . . . . 84
6.20 Turbine simulation results with the clean DU 06-W-200 . . . . . . . . . . . 85
6.21 Turbine simulation results with the dirty DU 06-W-200 . . . . . . . . . . . 86
B.1 The model placed inside the octagonal wind tunnel . . . . . . . . . . . . . 98
B.2 The wake rake with the total and static pressure tubes . . . . . . . . . . . 98
B.3 The manometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
B.4 Manometer detail of the total pressures in the wake . . . . . . . . . . . . . 99
B.5 Schematic of the test section . . . . . . . . . . . . . . . . . . . . . . . . . 99
B.6 Wind tunnel data for Re=150,000 - 500,000 . . . . . . . . . . . . . . . . . 100
B.7 Wind tunnel data for Re=500,000 - 1,000,000 . . . . . . . . . . . . . . . . 100
B.8 Wind tunnel data for Re=300,000 - 1,000,000 with trip at 80%
. . . . . . 101
B.9 Wind tunnel data for Re=300,000 and 700,000 with trip at 30% . . . . . . 101
B.10 Wind tunnel data for the fixed transition NACA 0018 at Re=300,000 . . . 102
B.11 Wind tunnel data for the fixed transition NACA 0018 at Re=700,000 . . . 102
B.12 NACA 0018 lift data for large angles at Re=300,000 . . . . . . . . . . . . 103
B.13 NACA 0018 drag data for large angles at Re=300,000
. . . . . . . . . . . 103
B.14 NACA 0018 lift data for large angles at Re=500,000 . . . . . . . . . . . . 104
B.15 NACA 0018 drag data for large angles at Re=500,000
. . . . . . . . . . . 104
C.1 RFOIL data for the DU 06-W-200 with free transition . . . . . . . . . . . . 105
C.2 RFOIL data for the DU 06-W-200 with trip at 40% up and 50% low . . . . 106
C.3 RFOIL data for the DU 06-W-200 with trip at 5% . . . . . . . . . . . . . . 106
M.Sc. thesis
M.C. Claessens
xvi
List of Figures
C.4 RFOIL data for the clean airfoil at α=0 . . . . . . . . . . . . . . . . . . . 107
C.5 RFOIL data for the clean airfoil at α=5 . . . . . . . . . . . . . . . . . . . 108
D.1 Characteristics of the clean DU 06-W-200 profile . . . . . . . . . . . . . . 110
D.2 Characteristics of the DU 06-W-200 profile with trip at 5% . . . . . . . . . 111
D.3 DU 06-W-200 lift data for large angles at Re=300,000 . . . . . . . . . . . 112
D.4 DU 06-W-200 drag data for large angles at Re=300,000 . . . . . . . . . . 112
D.5 DU 06-W-200 lift data for large angles at Re=500,000 . . . . . . . . . . . 113
D.6 DU 06-W-200 drag data for large angles at Re=500,000 . . . . . . . . . . 113
M.C. Claessens
M.Sc. thesis
List of Tables
1.1
The Turby geometry specifications . . . . . . . . . . . . . . . . . . . . . .
3
1.2
The Turby operating specifications . . . . . . . . . . . . . . . . . . . . . .
3
3.1
Gormont values for M1 , M2 and γmax . . . . . . . . . . . . . . . . . . . . 29
3.2
Strickland values for γ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
A.1 NACA 0018 coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
A.2 DU 06-W-200 coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
M.Sc. thesis
M.C. Claessens
xviii
M.C. Claessens
List of Tables
M.Sc. thesis
List of Symbols
Abbreviations
CFD
DMS
DMSV
HAWT
LDV
NACA
NASA
NLF
VAWT
Computational Fluid Dynamics
Double Multiple Streamtube model
Double Multiple Streamtube model with variable interference factors
Horizontal Axis Wind Turbine
Laser Doppler Velocimetry
National Advisory Committee for Aeronautics
National Aeronautics and Space Association
Natural Laminar Flow
Vertical Axis Wind Turbine
Greek Symbols
α
αB
αref
Δ
Ω
π
Blade element angle of attack
Effective blade angle of attack
Reference angle of attack
Difference
Rotational speed
Pi = 3.14159
ρ
Air density
θ
Angle of the blade element with respect to the start position
[◦ ]
[◦ ]
[◦ ]
[−]
rad s
[−]
kg
m3
[◦ ]
Latin Symbols
λ
ν
B
M.Sc. thesis
Tip speed ratio
Dynamic viscosity
Number of blades
[−]
kg
]
[ ms
[−]
M.C. Claessens
xx
c
CD
CL
CM
CN
CP
Cp
CT
D
F
M
N
R
r
Re
S
Sα̇
T
u
u
V
W
List of Symbols
Blade element chord
Drag coefficient
Lift coefficient
Moment coefficient
Normal force coefficient
Power coefficient
Pressure coefficient
Tangential force coefficient
Turbine diameter
Force
Mach number
Normal force
Maximum turbine blade radius
Local turbine blade radius
Reynolds number
Swept turbine area
Sign of the instantaneous rate of change of α
Tangential force
Upwind interference factor
Downwind interference factor
Airspeed
The local airspeed at the blade element
[m]
[−]
[−]
[−]
[−]
[−]
[−]
[−]
[m]
[N ]
[−]
[N ]
[m]
[m]
[−]
2
m
[−]
[N ]
[−]
[−]
m
s
m
s
Subscripts
D
e
inf
max
N
ref
ss
T
x
Disk
Equilibrium
Infinity
Maximum
In direction normal to the blade element
Reference
Steady stall
In direction tangential to the blade element
In direction of the x-axis
Superscripts
dyn
mod
M.C. Claessens
Dynamic
Modificated
M.Sc. thesis
Chapter 1
Introduction
1.1
Wind turbines
In the history of mankind wind energy has played an important role. Among others
wind energy was harnessed to grind grain. These so called wind mills are of the
horizontal axis design (HAWT) and were also used for pumping water and later for
sawing wood etc. Using multiple blades the wind energy generated by the atmosphere
is converted to kinetic energy inside the turbine. As more attention was put on the
environmental aspect of traditional (fossil) fuels the development of wind turbines for
generating electricity became more interesting.
(a) HAWT
(b) VAWT
Figure 1.1: The two main designs of lift driven wind turbines
In general there are two main categories of wind turbines: Horizontal Axis Wind
Turbines (HAWT) and Vertical Axis Wind Turbines (VAWT), see figure 1.1. Although
M.Sc. thesis
M.C. Claessens
2
Introduction
HAWT designs are widely used, they have the disadvantage that they have to be
positioned perpendicular to the wind direction. VAWT have the advantage that they
are independent of wind direction for their operations. This latter group is divided
again in two groups: lift driven VAWT (Darrieus) and drag driven VAWT (Savonius).
As the maximum possible efficiency of lift driven turbines is larger then for drag driven
turbines, the main attention today is focused on lift driven turbines. The first turbine
of this design was patented in 1931 by G.J.M. Darrieus.
1.2
VAWT basics
The concept of VAWT can have differently shaped blades. As the forces of the blades
can be large, the ideal blade has a Troposkien (nearly parabolic) shape with which the
centrifugal force is translated through the blade to the shaft. This type of blade is
mainly used in large turbines and prevents the blade from failing because of too large
rotational speeds. A large disadvantage is the decreasing radius near the top and the
bottom of the turbine. These parts experience only low rotational speeds and therefore
generate almost no power.
Another concept is the H-Darrieus or Musgrove VAWT. The blades are straight and
therefore the radius is equal over the total length of the blade, see figure 1.1(b). The
power is now generated over the complete length of the blade. In contrast to the
Troposkien shape blade extra strength is necessary to cope with the centrifugal forces.
The blades can be rotated slightly to disperse the moment forces on the axis over a
larger angle. The first prototypes of the H-Darrieus were developed in 1986.
The typical VAWT consists of the following parts:
• Supporting mast
• Rotational axis
• Supporting struts for the blades
• Blades
• Generator
• Converter
The blades of a VAWT have to develop lift and must have enough thickness to withstand the loads. To achieve this they have a certain shape, comparable to aircraft
wings. This shape determines how the wind energy is conversed to forces on the blade.
The goal of this study is to develop a new airfoil profile for an H-Darrieus vertical axis
wind turbine. In most of the existing turbines of this type standard profiles like the
NACA 0015 and NACA 0018 are used. These profiles were developed in the 1930’s by
the NACA as standard profile series for turbulent flow. Although these airfoils have
M.C. Claessens
M.Sc. thesis
1.2 VAWT basics
3
existed for a long time, not much experimental data is available. As the design involves
a relatively small turbine, the Reynolds numbers are small. For most existing airfoils
no data is available for these Reynolds numbers.
For this project the Turby is chosen as design case, a VAWT turbine developed in
cooperation with the TU Delft (see figure 1.1(b)). This is a relatively small turbine
designed to operate in urban environments on high buildings. The main characteristics
are given in tables 1.1 and 1.2 below.
Table 1.1: The Turby geometry specifications
Overall height
Weight (inc. blades)
Base flange Diameter
Bolt circle
Bolt holes
Rotor Diameter
Height
Rotor blades Number
Material
Weight (3 blades)
2890 mm
136 kg
250 mm
230 mm
6 x M10
1999 mm
2650 mm
3
composite
14 kg
Table 1.2: The Turby operating specifications
Cut-in wind speed
Rated wind speed
Cut-out wind speed
Survival wind speed
Rated rotational speed
Rated blade speed
Rated power at 14 m/s
M.Sc. thesis
4 m
s
14 m
s
14 m
s
55 m
s
120 - 400 rpm
42 m
s
2.5 kW
M.C. Claessens
4
1.3
Introduction
Project overview
The main goal is to design a new airfoil for VAWT application using the Turby as design
case. To be able to design this airfoil, the turbine has to be modeled. In figure 1.2 the
flow diagram of the model is given.
Real life turbine
Flow conditions
2D airfoil shape
Design criteria
Airfoil design
Turbine
simulation
Performance
results
Airfoil data (RFOIL)
Figure 1.2: Flow diagram of the modeling of the turbine
In chapter 2 the flow conditions which are encountered in VAWT aerodynamics are
discussed. These flow conditions determine for a large part the design criteria of the
new airfoil and are used in making a sufficiently accurate simulation program. The
simulation methods and the final simulation program are discussed in chapter 3. In
the rest if the chapters the design of the new airfoil will be discussed. In figure 1.3 this
process is shown in more detail.
During the development and testing of airfoils the NACA 0018 airfoil is used as reference. The RFOIL program is used to modify the blade shape and to calculate the
characteristics of the new shape. Although RFOIL is a strong design tool, it has its
limitations. The program is not able to predict all flow phenomena, resulting in limited
accuracy and the angular range is limited until the airfoil is stalled. The RFOIL program is validated using wind tunnel data of the NACA 0018 airfoil and other profiles
measured at the TU Delft. Starting from the NACA 0018 airfoil the RFOIL program
is then used to investigate other airfoils, make adjustments and calculate their characteristics. In chapter 5 these characteristics are compared and used in the simulation
program. The results of the simulation program and the airfoil characteristics are distilled into a general design of the airfoil. In chapter 6 this general design is fine tuned
until the optimum design is reached. The characteristics of this design are measured
in the wind tunnel and a comparison is made between this final design and the NACA
0018.
M.C. Claessens
M.Sc. thesis
M.Sc. thesis
airfoils (*)
Improve Compare the
Validate RFOIL data
Optimal
results
Final design
windtunnel tests
Final design
NACA 0018
windtunnel tests
Final evaluation
(*) The comparison is made usign the airfoils characteristics and pressure distributions and by
simulating the turbine using the 2D RFOIL data
New design
RFOIL data
Airfoil design
NACA 0018
RFOIL data
Original airfoil
NACA 0018
1.3 Project overview
5
Figure 1.3: More detailed overview of the airfoil design
M.C. Claessens
6
M.C. Claessens
Introduction
M.Sc. thesis
Chapter 2
VAWT aerodynamics
In this chapter the prevailing aerodynamic phenomena for the VAWT are discussed.
In the first section the basics involved with the aerodynamic analysis are given. The
flow conditions which are encountered by this turbine are given in the second section.
The flow conditions determine for a large part the design criteria of the airfoil that has
to be developed. These criteria are given in the last section. These phenomena also
have to be incorporated in the simulation model of the turbine. The way in which this
is done is discussed in chapter 3.
2.1
Basic aerodynamics
As the VAWT have a rotational axis perpendicular to the oncoming airflow, the aerodynamics involved are more complicated than of the more conventional HAWT. The
main benefit of this layout is the independence of wind direction. The main disadvantages are the high local angles of attack involved and the wake coming from the
blades in the upwind part and from the axis. If the turbine is represented in a two
dimensional way (see figure 2.1) these characteristics are more obvious.
The rotational speed can be varied by the turbines controller for a certain wind speed.
The rotational speed ω is therefore represented by the tip speed ratio λ. This parameter
gives the tip speed Rω as factor of the free stream velocity Vinf :
λ=
Rω
Vinf
(2.1)
The Reynolds number is a measure of the viscous behavior of air:
Re =
M.Sc. thesis
ρV c
ν
(2.2)
M.C. Claessens
8
VAWT aerodynamics
Downwind
Upwind
Vinf
Vinf
Rω
Rω
(a) 3D schematic
(b) 2D cross section
Figure 2.1: 2D cross section from the turbine
The performance of the turbine is given by the power coefficient CP . This coefficient
represents the produced energy of the turbine as part of the total wind energy passing
through the swept area of the turbine. This area equals the frontal area of the turbine
given by the height times the diameter. This coefficient is normally plotted against
the tip speed ratio λ at a certain Reynolds number, see figure 2.2. The tip speed ratio
and Reynolds number are in this case both dependent of Vinf .
Power coefficient for 3 NACA 0018 blades with trip at 70%
0.7
10 m/s
0.6
0.5
Cp
0.4
0.3
0.2
0.1
0
0
1
2
3
4
5
6
7
8
Tip speed ratio
Figure 2.2: Power efficiency results for a VAWT at Vinf = 10 m
s
M.C. Claessens
M.Sc. thesis
2.2 Flow conditions
2.2
9
Flow conditions
Below the prevailing flow conditions for a VAWT are described. These conditions are
important to set the design goals and the correct boundary conditions for the design.
The concluding design goals are given in the following section.
2.2.1
Angle of attack
One of the largest challenges of the VAWT is the wide range of angles of attack the
blades experiences. When the turbine starts from zero rotational speed, the blades
even experience back flow. With increasing rotational speed, the maximum angle of
attack decreases. See figure 2.3. The larger the rotational speed, the smaller the
influence of the free stream flow on the local speed W . The blade will be optimized
for the Turby VAWT, which operates at a tip speed ratio of 3. As can be seen in
figure 2.3 the maximum angle of attack in this situation equals 16.5◦ . The angle varies
with the turbine azimuth angle θ. As dynamic and exterior effects can increase this
angle, a larger angle of attack has to be taken into account in the design phase. In the
interval where the angle of attack is negative, the maximum angle of attack is lower.
The blades extract energy from the airflow at the upwind side, resulting in a lower air
speed and thus a lower angle of attack at the downwind side.
30
λ=2
λ=3
λ=4
20
α (◦ )
10
0
−10
−20
−30
−100
−50
0
50
100
θ (◦ )
150
200
250
300
Figure 2.3: The angle of attack variation as function of θ at Vinf = 10 m
s
M.Sc. thesis
M.C. Claessens
10
VAWT aerodynamics
2.2.2
Deep stall
If the angle of attack over a wing is increased, at some moment the airflow will separate.
The separation starts at the trailing edge of the airfoil and shifts forward with increasing
angle. If the angle is increased further the separation moves forward to the leading
edge. This phenomenon is called deep stall. At very low Reynolds numbers separation
can occur at the airfoils nose right away, but this is outside the scope of this research.
If the airfoil is in deep stall, this condition will be maintained for some time, even if the
angle is decreased again. This will cause a hysteresis loop, see figure 2.4. Here the deep
stall starts at α = 21◦ and the hysteresis loop ends at α = 13.5◦ . This phenomenon
has a strong negative influence on the performance of the blade, because in the loop
the lift is low and the drag remains high.
1.2
1.2
cl
cl
0.8
0.8
0.4
0.4
0.0
0.0
0.0
0.1
0.2
0.3
0.4
-20
-10
0
10
20
30
-0.4
-0.4
-0.8
-0.8
NACA 0018
6
Re = 0.5 x 10
-1.2
-1.2
cd
o
α( )
Figure 2.4: The deep stall characteristics of the NACA 0018 profile
The angle at which deep stall occurs dependents on the Reynolds number and the nose
radius. In the VAWT application of airfoils large angles of attack are encountered.
At the operating tip speed ratio this phenomenon therefore should be avoided or its
influence should be kept as small as possible. Figure 2.5 shows the results for deep
stall for different profiles measured at the TU Delft plotted against the parameter yc
at xc = 0.0125.
The negative values in figure 2.5 are the values for the lower side of the profile. All the
test models had a chord of 60 cm, [W.A. Timmer, 2003]. The resulting straight line
can be translated to the following relation between the thickness of the nose and the
M.C. Claessens
M.Sc. thesis
2.2 Flow conditions
11
Figure 2.5: The deep stall characteristics of multiple profiles
deep stall angle:
αdeep−stall = 1114
2.2.3
y c
x
=0.0125
c
(2.3)
Dynamic stall
Dynamic stall is a phenomenon that occurs at airfoils with rapid changing angle of
incidence. The resulting effect of this changing angle is a difference, a hysteresis, in
the lift, drag and moment characteristics between increasing and decreasing angle of
incidence. Dynamic stall is characterized by the shedding and passage of a vortex-like
disturbance over the low pressure surface of the lifting body. The main parameter of
influence is related to the airfoil motion and the boundary layer separation. The main
fields of research in dynamic stall are helicopter or fighter aircraft application. Some of
those methods are modified for wind turbine applications and also research specifically
in this field is performed.
The Darrieus type wind turbine is especially susceptible to dynamic stall, as the change
in angle of incidence is large, especially at low tips speed ratios. As the blades perform
a complete circle, the blades in the downstream part of the turbine are influenced by
the wake resulting from the upstream blades. A good understanding of dynamic stall
and the resulting wake is therefore important.
First visualization of the dynamic stall for the VAWT was done by
[Brochier et al., 1986].
Using a water channel, visualizations were made with
LDV and hydrogen bubbles at an Reynolds number of 10,000 and tip speed ratios
M.Sc. thesis
M.C. Claessens
12
VAWT aerodynamics
varying from 1 to 8 on a Darrieus turbine with two NACA 0018 blades. The results
for λ = 2.14 are shown in figures 2.6(a) and 2.6(b).
Trailing
edge vortex
Leading
edge vortex
Tower wake
(a) Flow visualization
(b) Schematic diagram
Figure 2.6: Visualization of dynamic stall at λ = 2.14, [Brochier et al., 1986]
The first vortex is formed at the leading edge of the airfoil. A second vortex, turning
in the opposite direction, originates form the trailing edge. Together they form the
characterizing doublet of two counter rotating vortices, which travels downstream to
meet the second blade. This is confirmed by measurements made in a water channel using PIV by [Fujisawa and Shibuya, 2001] on a one NACA 0018 blade Darrieus
turbine without central column. The flow visualization is showed in figure 2.7. The
formation of the vortices is clearly visible. In this case two pairs of stall vortices are
found. The first pair is formed at small blade angles and develops through the wake.
The second pair is formed at large blade angles and will follow the blade at the inner
side.
As the encountered angles of incidence are larger at lower tip speed ratios, the dynamic
stall is more present. As can be seen in figure 2.8 the dynamic stall will start earlier
and the vortices themselves are larger. The structure of the stall itself is independent
of the tip speed ratio. At higher ratios, above 4, the dynamic stall will become of
less importance. Although dynamic stall will increase the performance of the turbine,
it includes also large disadvantages. It will cause an increase in noise, aero-elastic
vibrations and blade fatigue.
Both studies show the strong asymmetry in the flow properties inside the turbine.
The blades pass through the wake in only a part of the cycle. In this part they will
experience highly turbulent flow.
M.C. Claessens
M.Sc. thesis
2.2 Flow conditions
13
Flow
0o
90o
45o
Figure 2.7: Flow visualization at four different positions, [Fujisawa and Shibuya, 2001]
(a) λ = 1
(b) λ = 2
Figure 2.8:
Schematic illustration of the
[Fujisawa and Shibuya, 2001]
M.Sc. thesis
dynamic
(c) λ = 3
stall for
different λ,
M.C. Claessens
14
2.2.4
VAWT aerodynamics
Reynolds number
A crucial factor for small turbine design is the low Reynolds range (< 1 million) in
which they operate (see figure 2.9). Most studies in aerodynamics are performed for
aircraft applications in which the Reynolds number lies above 3 million. It is very
difficult and often impossible to find the right data for airfoils in this low Reynolds
number range.
5
5
x 10
λ=2
λ=3
λ=4
4.5
4
3.5
Re
3
2.5
2
1.5
1
0.5
−100
−50
0
50
100
θ (◦ )
150
200
250
300
Figure 2.9: The Reynolds number variation at Vinf = 10 m/s
In figures 2.10 to 2.13 the effects of the Reynolds number on the aerodynamics of an
airfoil is shown. For the currently used NACA 0018 profile the maximum lift coefficient
and the stall angle of attack drastically decreases with a decreasing Reynolds number
(figure 2.10). These effects can be noticed for all turbulent NACA symmetric airfoils,
see figure 2.11. The difference in maximum lift coefficient between Re = 3 million
and Re = 0.3 million can be as much as 60%. The VAWT operates at low Reynolds
numbers and high angles of attack, therefore the negative Reynolds number effects on
the airfoils performance have to be taken into account when working with airfoil data
of higher Reynolds number.
From experiments performed on VAWT by [Sheldahl et al., 1980] the influence of the
Reynolds number is also shown. The chord Reynolds number was changed by altering
the rotational speed of the turbine. Also the wind speed was changed to view the
performance over the tip speed range, see figures 2.12 and 2.13.
M.C. Claessens
M.Sc. thesis
2.2 Flow conditions
15
1.4
1.2
1.0
Cl
0.8
0.6
0.4
0018, RE=62,000
0018, RE=125,000
0018, RE=250,000
0018, RE=1,250,000
0.2
0.0
0
5
10
15
20
25
30
α
Figure 2.10: Reynolds effects on the lift curve of the NACA 0018 profile,
[Jacobs and Sherman, 1937]
1.8
1.6
Clmax
1.4
1.2
1.0
NACA 0009
NACA 0012
NACA 0015
NACA 0018
0.8
0.6
1.0E+04
1.0E+05
1.0E+06
1.0E+07
Re
Figure 2.11: Reynolds effects on the maximum lift the NACA 00xx series profiles,
[Jacobs and Sherman, 1937]
M.Sc. thesis
M.C. Claessens
16
VAWT aerodynamics
0.5
Re = 250.000
Re = 280.000
Re = 300.000
0.4
Cp
0.3
0.2
0.1
0.0
0
1
2
3
4
5
6
7
8
9
Tip speed ratio
Figure 2.12:
Reynolds number influence on the Sandia 5 meter turbine,
[Sheldahl et al., 1980]
0.4
Re = 106.000
Re = 156.000
Re = 204.000
Re = 290.000
Cp
0.3
0.2
0.1
0.0
0
1
2
3
4
5
6
7
8
9
Tip speed ratio
Figure 2.13:
Reynolds number influence on the Sandia 2 meter test rotor,
[Paraschivoiu, 2002]
M.C. Claessens
M.Sc. thesis
2.2 Flow conditions
2.2.5
17
Laminar separation bubbles
In low Reynolds conditions a specific feature called laminar separation bubbles is often
present. The laminar boundary layer is no longer able to follow the contour of the
airfoil as a result of the adverse pressure gradient. At the same time the instabilities in
the boundary layer have not developed enough to cause the layer to become turbulent.
Therefore the laminar boundary separates. This flow can turn turbulent and reattach
itself to the airfoil again, forming the so called laminar separation bubble. In some cases
the bubble can extend over the trailing edge of the airfoil because the adverse gradient
is too large for the turbulent layer to reattach. The latter occurred in the case of the
NACA 0018 airfoil at lower Reynolds numbers, see figure 2.14. At higher Reynolds
numbers the appearance of separation bubbles shifts to higher angles of attack. The
most effective solution is to apply a turbulator, e.g. zigzag tape, at a position before
the laminar layer separates.
1.2
1.2
cl
NACA 0018
6
Re=0.5x10
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
Clean
Trip at 80%
-1.2
cd
-1.2
o
α( )
Figure 2.14: Laminar separation bubbles for the NACA 0018
M.Sc. thesis
M.C. Claessens
18
VAWT aerodynamics
2.2.6
Virtual camber
Research performed by [Migliore et al., 1980] shows that the aerodynamic characteristics of an airfoil differ between situations of curvilinear flow fields and rectilinear flow
fields. Due to the fact that the airfoil is rotating the symmetric airfoil behaves like an
airfoil in rectilinear flow with camber and with a virtual angle of incidence (see figure 2.15). The influence of the curvilinear flow field on the aerodynamic characteristics
depends very much on the blade chord to turbine radius ratio Rc . If this ratio becomes
larger, the influence of curvilinear flow increases as well. Virtual camber causes an upward shift of the lift curve and introduces an aerodynamic moment. Virtual incidence
causes the lift curve to shift to the left. The exact impact of these phenomena on the
airfoil performance in VAWT is not yet established.
αi
Figure 2.15: The principal of virtual camber as a result of curvilinear flow,
[Migliore et al., 1980]
M.C. Claessens
M.Sc. thesis
2.3 Design criteria
2.3
19
Design criteria
From the flow conditions in which the VAWT turbine operates, described in the previous section, certain airfoil design parameters can be established. Optimum values for
these parameters may be conflicting and compromises should be made. The design is
made using the NACA 0018 airfoil as reference. In chapter 5 more information about
the design of the new airfoil is given.
The parameters of interest are:
• Designed to perform well at Reynolds numbers between 150,000 and 700,000
• Low zero drag coefficient
• Increased width of the drag bucket to maintain performance over a larger range
of α
• Increased thickness to improve structural strength
• Smooth stall characteristics to reduce noise
• Large separation bubbles which extend over the trailing edge are a source of noise
Also deep stall characteristics are important with regard to airfoil design:
• Postpone deep stall to a larger angle of attack; a thicker airfoil will in general
also have deep stall at a higher angle of attack
• The deep stall hysteresis loop should be as small as possible. Thicker airfoils
have generally a larger drop in lift at deep stall
• The lift coefficient drop at deep stall should be as small as possible
Furthermore the use of cambered airfoils instead of symmetrical ones is investigated.
M.Sc. thesis
M.C. Claessens
20
M.C. Claessens
VAWT aerodynamics
M.Sc. thesis
Chapter 3
Simulation methods
The shape of the turbine blade has to be connected to the performance of the turbine.
A simulation program was made using 2D airfoil data to predict the performance of
the turbine. The way in which 2D airfoil data is generated will be discussed in the
following chapters. In this chapter the different methods in which the turbine can
be simulated and the way in which the 2D characteristics are implemented in these
methods is discussed.
Different methods are possible for simulating the VAWT, each with its own benefits
and drawbacks. The three main directions in modeling are: momentum based models,
vortex models and CFD modeling. Each of these methods has its own benefits and
drawbacks in terms of accuracy and complexity. In the sections below these methods
will be explained in more detail.
3.1
Momentum theory based models
Momentum based models are based on of the actuator disk theory, which is generally
used for rotor aerodynamics, adjusted for the VAWT. The basic model is called the
single streamtube model. This model was developed in two directions. The first one
was splitting the single streamtube in a number of parallel streamtubes, resulting in the
multiple streamtube model. The other direction was the double actuator disc analysis
for the VAWT. This model places two actuator disks is tandem formation, resulting
in two interference factors; one for the upwind side and one for the downwind side.
These models were eventually combined in the Double Multiple Streamtube (DMS)
model. One disadvantage of this model is that the interference factors are fixed for the
upwind and downwind side, as such they can not be adjusted for each streamtube. By
modifying the DMS model this is made possible. The only major improvement since
was the implementation of a dynamic stall model. There are many different dynamic
stall models, with different accuracy and complexity, see section 3.3.
M.Sc. thesis
M.C. Claessens
22
Simulation methods
Single Streamtube model
Multiple Streamtube model
Double Actuator Disc theory
Double Multiple Streamtube (DMS) model
DMS model with variable interference factors (DMSV)
Figure 3.1: Overview of the development of the streamtube models
The momentum based models are limited to smaller tip speed ratios and to a solidity
below 0.2, because the Glauert momentum theory is not valid outside this range. Also
quasi-steady flow through the turbine and constant streamwise velocity as function of
free stream velocity is assumed. The velocities normal to the free stream velocity are
neglected.
3.1.1
Single Streamtube Model
This model was first developed by Templin ([Templin, 1974]) for the VAWT. It is
based on the actuator disk theories applicable for propellers and is the most basic
model based on the momentum theory. The flow through the turbine is assumed to
have one constant velocity.
ωR
α
V cosθ
V sinθ
W
Figure 3.2: Components of the local angle of attack
The local angle of attack is given by:
α = arctan
cos(θ)
Rω
V − sin(θ)
(3.1)
The forces on the blade elements are based on available experimental 2D airfoil data:
M.C. Claessens
CN = CL cos(α) + CD sin(α)
(3.2)
CT = CL sin(α) − CD cos(α)
(3.3)
M.Sc. thesis
3.1 Momentum theory based models
23
dN = CN q c dz
(3.4)
dT = CT q c dz
(3.5)
The drag experienced by one element, or the force in the direction of the airflow, is:
dD = qc (CN sin(θ) − CT cos(θ))
(3.6)
The total drag for a turbine with N blades with chord C on the complete interval
(0 < θ < 2π) and (−H < z < H) results in:
Nc
D=
2π
3.1.2
H
z=−H
2π
q (CN sin(θ) − CT cos(θ)) dθdz
(3.7)
θ=0
Multiple Streamtubes Model
This model is developed by Strickland [Strickland, 1975] and is also based on the
momentum theory. The main improvement with respect to the single streamtube model
is that more streamtubes make different induced velocities possible (see figure 3.3).
Each streamtube has its own velocity, allowing a change in velocity depending on the
direction perpendicular to the freestream flow. The accuracy is dependent on the
number of streamtubes used. It gives good results for low tip speed ratios and low
solidity.
y
streamtube
Vinf
V
Δθ
V’
θ
x
Figure 3.3: 2D schematic of the streamtube model
The total span of the single streamtube is divided in multiple streamtubes using a fixed
angle Δθ as width.
Δθ =
M.Sc. thesis
2π
Nθ
(3.8)
M.C. Claessens
24
Simulation methods
For each of these streamtubes the momentum equations and blade elements have to be
calculated, resulting in N interference factors ui , i = 1 → N .
3.1.3
Double actuator disc theory
Vinf
Downwind zone
Equilibrium
velocity zone
Upwind zone
The main disadvantages of the previous models is the inability to make a distinction
between the upwind and downwind part of the turbine. To make this possible, two
actuator discs are placed behind each other, connected at the center of the turbine (see
figure 3.4).
V
Ve
V’
V’’
r
r
Figure 3.4: Schematic of the two actuator discs behind each other
The velocities are determined by two interference factors, u and u :
V = u Vinf
(3.9)
Ve = (2u − 1) Vinf
(3.10)
V = u Ve = u (2u − 1) Vinf
3.1.4
(3.11)
Double-Multiple Streamtubes Model
The
DMS
model
described
by
[Loth and McCoy, 1983]
and
[Paraschivoiu and Delclaux, 1983] combines the multiple streamtubes model with the
double actuator disc theory. This allows to model velocity variations in the direction
perpendicular to the freestream flow and between the upwind and downwind part
of the turbine. The previous models were not able to calculate the influence of the
upwind part on the downwind part. As a result non-symmetrical airfoils, which
depend on this difference, could not be simulated accurately. It is easily understood
that the wind velocities at the upwind part are larger than these at the downwind
part, because the blades have already extracted energy from the wind.
M.C. Claessens
M.Sc. thesis
3.1 Momentum theory based models
25
The flow is divided horizontally into tubes, each with the angular width of:
Δθ =
2π
Nθ
(3.12)
The streamwise force acting on one blade element is:
ΔFx = ΔFN cosθ + ΔFT sinθ
(3.13)
The tangential and normal force components ΔFN and ΔFT are linked to the airfoil
characteristics in the following way:
CN =
ΔFN
1
2
2 ρW cΔz
(3.14)
CT =
ΔFT
1
2
2 ρW cΔz
(3.15)
Resulting in the equation:
1
ΔFx = ρW 2 cΔz[CN cosθ + CT sinθ]
2
(3.16)
This is the force which is experienced by one blade element. Each element passes
through the stream tube of Δθ
2π and there are B number of blades, so the averaged force
equals:
ΔFx =
1
BΔθ
ρW 2
cΔz[CN cosθ + CT sinθ]
2
2π
(3.17)
The local velocity W is the resultant of the rotational speed ΩR and the airspeed at
the rotor V . These variables can be transformed to the single variable, tip speed ratio
λ:
λ=
W =V
Ωr
V
[λ − sinθ]2 + [cosθ]2
(3.18)
(3.19)
The local angle of attack depends on the tip speed ratio λ and the position of the
element θ:
cosθ
−1
(3.20)
α = tan
λ − sinθ
M.Sc. thesis
M.C. Claessens
26
Simulation methods
Upwind half − π2 ≤ θ ≤ π2 :
2
W =V
2
ωR
− sin θ
V
2
2
+ cos θ
cos θ
α = arctan ωR
V − sin θ
Downwind half
π
2
≤θ≤
(3.21)
(3.22)
3π
2 :
W
2
=V
2
ωR
− sin θ
V
2
2
+ cos θ
cos θ
α = arctan ωR
V − sin θ
(3.23)
(3.24)
The resulting force on the blade element can be calculated by first calculating CN and
CT and the resulting force in streamwise direction ΔFx with equation (3.17). Using
the momentum theory the streamwise force also can be written as:
ΔFx = 2ρSU (U∞ − U )
(3.25)
Filling in the calculated value of ΔFx the factor UU∞ is found, which equals the interference factor a. By starting with U = U∞ after one calculation the interference factor
a is found. The next iteration round the following airspeed is used:
U = aU∞
(3.26)
These iteration steps can be continued until the value of the interference factor reaches
the requested accuracy.
M.C. Claessens
M.Sc. thesis
3.2 Non - momentum theory based models
3.2
27
Non - momentum theory based models
Besides momentum based models different models are available for simulating vertical
axis wind turbines. The two most important are vortex models and CFD. For this
study these models were not applicable because they are too time consuming for the
design process. To give a complete overview of the available simulation models both
models will be explained in brief below.
3.2.1
Vortex models
Vortex models are based on vorticity equations. The blade element is replaced by a
lifting line which represents the flow field at distances more then one chord away from
the airfoil. The benefit is that pressure field values are not needed to obtain a velocity
field. In contrast to momentum based models this method is also applicable for VAWT
turbines with high solidity and at large tip speed ratios. Furthermore the momentum
based models are not capable of giving information of the wake structure near the
turbine, as velocity normal to the airflow is neglected. The main vortex models for
the VAWT turbine are the free-wake vortex model and fixed-wake momentum theory
(combination of vortex theory and momentum method).
As for our study both the solidity and tip speed ratios are not large and only the
overall performance of the turbine is needed, the vortex models offer no benefits over
the momentum based models. As the amount of work needed for the vortex models is
substantially more, this type of models is neglected.
3.2.2
CFD models
If even more accuracy and details are needed, Computational Fluid Dynamics (CFD)
offers the best solution in comparison to momentum and vortex models. Even unsteady
calculations are a possibility. This method uses a grid around a 3D model of the turbine
to calculate the complete airflow around it. The mesh size and computational models
determine the accuracy of the result. A lot of computer power and time is required
for CFD and the generation of the 3D model and mesh take a lot of time. Also
implementing any airfoil changes require a lot of time. As both the computational
time and airfoil changes are very important for the model used during this project,
CFD is not a feasible option.
M.Sc. thesis
M.C. Claessens
28
3.3
Simulation methods
Dynamic stall models
Dynamic stall is a phenomenon which occurs on airfoils with a rapidly changing angle
of incidence. The effect of this changing angle is a difference, a hysteresis, in the lift,
drag and moment characteristics between increasing and decreasing angle of incidence.
Dynamic stall is characterized by the shedding and passage of a vortex-like disturbance
over the low pressure surface of the lifting surface. The main parameters of influence
are related to the airfoil motion and the boundary layer separation. The main fields of
research in dynamic stall are concentrated on helicopter or fighter aircraft applications.
Some of those methods are modified for wind turbine applications and also research
specific for this field is performed. The models can be split in (semi-)empirical and theoretical models. For the implementation in the streamtube models the semi-empirical
methods are most useful, although these models are very crude.
3.3.1
Gormonts model
A model which empirically approaches the dynamic stall behavior for helicopter application was developed by [Gormont, 1973]. It models the hysteresis effect by defining a
reference angle of attack at which static airfoil data is taken. The difference between
the geometrical angle α and the reference angle αref determines the influence of the
dynamic stall.
αref = α − K1 Δα
(3.27)
The reference angle of attack is off course different for increasing and decreasing α:
K1 =
Δα =
1
−0.5
when α̇ ≥ 0
when α̇ < 0
γ1
γ1 Sc + γ2 (S − Sc )
(3.28)
when S ≤ Sc
when S > Sc
cα̇ S =
2W t
Sc = 0.06 + 1.5 0.06 −
c
γ2
for lift characteristic
2
γ1 =
0
for drag characteristic
M − M2
γ2 = γmax max 0, min 1,
M1 − M2
M.C. Claessens
(3.29)
(3.30)
(3.31)
(3.32)
(3.33)
M.Sc. thesis
3.3 Dynamic stall models
29
The resulting dynamic coefficients are given by:
CLdyn = CL (α0 ) + m(α + α0 )
(3.34)
dyn
= CD (αref )
CD
(3.35)
CL (αref ) − CL (α0 ) CL (αss ) − CL (α0 )
,
m = min
αref − α0
αss − α0
(3.36)
The values for M1 , M2 and γmax are dependent on whether the lift or drag is calculated
(see table 3.1).
Table 3.1: Gormont values for M1 , M2 and γmax
M1
M2
γmax
3.3.2
Lift characteristics
0.4+5.0(0.06-t/c)
0.9+2.5(0.06-t/c)
1.4+6.0(0.06-t/c)
Drag characteristics
0.2
0.7+2.5(0.06-t/c)
1.0+2.5(0.06-t/c)
Strickland et al. modification
The Gormont method hase been adjusted for VAWT application by [Strickland, 1975].
The following assumptions were made:
• Sc is equal to 0, because the thickness used in VAWT application is usually larger
then 12%
• The flow is incompressible, no Mach number dependency
• The dynamic stall model is only applied if α ≥ αss
• The lift curve curve slope and zero lift angle remain the same
• The dynamic stall only effects the angle at which stall occurs
αref
cα˙B 1/2
= αB − γ K1 Sα̇
2W (3.37)
K1 is again dependent on the sign of the instantaneous rate of change of the angle of
attack:
K1 = 0.75 + 0.25Sα̇
M.Sc. thesis
(3.38)
M.C. Claessens
30
Simulation methods
Table 3.2: Strickland values for γ
Lift
Drag
Moment
γ
1.4 − 6.0 (0.06 − t/c)
1.0 − 2.5 (0.06 − t/c)
1.0 − 2.5 (0.06 − t/c)
The final values of the lift, drag and moment coefficients are given by:
CL
CM
αB
=
αM − αB0
= CM (αM )
CL (αM )
(3.39)
(3.40)
CD = CD (αM )
(3.41)
1.5
1.5
1
1
0.5
0.5
Cl
Cl
See figures 3.5(a) and 3.5(b) for the results of the Strickland dynamic stall model when
applied to the NACA 0018 wind tunnel data.
0
0
−0.5
−0.5
−1
−1
Original data
Dynamics stall data
−1.5
0
0.05
0.1
0.15
Cd
0.2
(a) Dynamic stall drag
0.25
0.3
−1.5
−30
−20
−10
0
α
10
20
30
(b) Dynamic stall lift
Figure 3.5: The Strickland dynamic stall model applied to the NACA 0018
M.C. Claessens
M.Sc. thesis
3.3 Dynamic stall models
3.3.3
31
Paraschivoiu et al. modification
In the Darrieus type wind turbines there are regions with large scale turbulence, see
figure 2.6(b). Turbulence stabilizes the boundary layer of the airfoil. Dynamic stall
therefore occurs at higher angles of attack then at regions without large scale turbulence. The main modification that [Paraschivoiu, 2002] made to the Gormont method
modified by [Strickland, 1975] is limiting its application to a certain area of the wind
turbine at which the turbulence is low, between the azimuth angles of 135 and 15
degrees.
3.3.4
Massé and Berg modification
In practice it was found that the model over predicts the effects of dynamic stall at
large angles of attack. For the original Gormont model in section 3.3.1 this is not a
problem, as in helicopter applications these large angles are not encountered. To solve
this problem Massé as described by [Paraschivoiu, 2002] introduced a modification in
which linearly interpolates between the static lift and drag values and the dynamic
values. The influence decreases at larger angles until the static values are restored at
α = AM αss . In this modification the Mach number is still present, in contrast with
the modification by Strickland.
CLmod
CL +
=
mod
=
CD
AM αss −α
AM αss −αss
CD +
CLdyn − CL
CL
AM αss −α
AM αss −αss
dyn
− CD
CD
when α ≤ AM αss
when α > AM αss
when α ≤ AM αss
when α > AM αss
CD
(3.42)
(3.43)
The value of AM is empirically determined for VAWT by Berg as described by
[Paraschivoiu, 2002]. According to his research the best value for VAWT equals:
AM = 6
(3.44)
Also the static angle of stall αss is proposed to be the angle at which the lift curve no
longer follows the linear lift behavior.
M.Sc. thesis
M.C. Claessens
32
3.4
Simulation methods
Simulation model
In this chapter different methods of modeling VAWT are discussed. The goal of the
simulation program is the possibility to use 2D blade shape characteristics and calculate
their performance in a VAWT. Fast blade shape adjustment and short calculation time
are essential to incorporate the simulation program into the airfoil design process.
CFD and vortex models are too time consuming to use for this purpose. Therefore the
choice is made for momentum based models which are incorporated into the simulation
program, which is built in Matlab.
For this project it is important to have a simulation model which can differentiate
between the upwind and downwind side of the turbine. This enables the program to
take into account airfoils with camber. For the final simulation program the Double
Multiple Streamtubes model was chosen. As dynamic stall model the [Strickland, 1975]
modification to the Gormont model is chosen. As all these models are relatively crude,
the option is left to chose whether to enable or disable dynamic stall. To be able
to compare the different models, in the program it is possible to also choose for the
Multiple Streamtubes model. The program allows to swiftly implement adjusted blade
shapes by inserting the RFOIL data for these changed airfoils. Also the airfoils which
were subject to the design process are available for implementation in the model.
M.C. Claessens
M.Sc. thesis
Chapter 4
RFOIL
The simulation program discussed in the previous chapter uses 2D airfoil data to calculate the VAWT performance. The 2D data can be determined by wind tunnel testing,
only this is not a feasible option for the design process since wind tunnel testing and
the airfoil models are expensive. It is not economic to manufacture a model from
each design and analyze it in the tunnel. For the practical design process the RFOIL
program is used. This is based on the XFOIL program and uses the panel method in
combination with boundary layer equations to calculate 2D airfoil characteristics. It is
a fast design tool which offers the possibility to make changes in pressure distributions
and geometries. Different computational options, like CFD, are too time consuming
to be considered for use within the scope of this study.
To validate the RFOIL results, the data of four airfoils is available. These airfoils are
divided into two groups: laminar and turbulent. The following airfoils or airfoils of
the same family will be used later on in the design process. For the NACA 0018 data
is available for Re = 300,000, 500,000 and 700,000. For the NACA 64-418 no data
at Re = 500,000 is available and for the NLF-0416 profile only data at Re=500,000 is
present.
• NACA 0018 (turbulent)
• NACA 64-418 (laminar)
• NFL-0416 (laminar)
All these profiles are compared at lower Reynolds numbers (< 1,000,000). RFOIL has
problems to accurately predict the behavior of airfoils at these numbers, as laminar separation and reattachment play an important role. These aerodynamic phenomena are
difficult to predict with high accuracy by computational methods. At higher Reynolds
numbers the RFOIL performance is more accurate. Two examples at Re=3,000,000
are given in section 4.3.
M.Sc. thesis
M.C. Claessens
34
RFOIL
4.1
Turbulent profiles
For this study tests on the turbulent NACA 0018 were performed at the TU Delft.
Good data sets are not available in literature for the NACA 0018 profile. That is
why tests were performed on the profile as part of this study. An overview of the
test results can be found in Appendix B. The lift and drag of the NACA 0018 are
determined using a balance. As the balance measures the forces on the total model,
3D effects as tunnel wall interference are accounted for. The results for the drag are
corrected for these effects. The calculated RFOIL data is corrected for the wind tunnel
data by adjusting the critical amplification factor Ncrit . The amplification factor is a
measure of the growth of instabilities in the boundary layer. The critical value is the
value at which RFOIL decides the boundary layer turns turbulent. By adjusting this
value, the RFOIL data can be adjusted for the different turbulence levels at different
Reynold numbers of the wind tunnel. In the wind tunnel the turbulence level increases
with increasing Reynolds number, the same trend can be viewed in the used critical
amplification factor for RFOIL modeling.
In figure 4.1 it can be seen that the predicted RFOIL lift at Re = 300,000 is almost
equal to the measurements using Ncrit = 12. The RFOIL lift slope is equal to the
corrected lift slope of the measurements. The drag is in good agreement with wind
tunnel data (see figure 4.1).
1.2
1.2
cl
NACA 0018
6
Re = 0.3x10
cl
0.8
0.8
0.4
0.4
Meas. 2D corr.
RFOIL N=9
RFOIL N=12
0.0
0.000
0.0
0.015
cd
0.030
0
5
10
15
o
20
α( )
Figure 4.1: NACA 0018 characteristics at Re=300,000 trip at 70%
For Re = 500,000 the differences between measured and calculated lift slope become
larger. The lift slope of the RFOIL data is steeper than the measurements, see figM.C. Claessens
M.Sc. thesis
4.1 Turbulent profiles
35
ure 4.2. The lift is overpredicted until stall occurs. After α = 13 the RFOIL lift
decreases slightly, where the measured lift does not. RFOIL under predicts the lift
at high angles of attack, but the validity of RFOIL at such large angles of attack is
doubtful. The drag is underpredicted over the whole range, see figure 4.2. A drag
increase starts later, resulting in too optimistic L/D values at α ≥ 10. The value for
Ncrit is chosen equal to 12, as the results more closely resemble the measured data.
1.2
1.2
cl
NACA 0018
6
Re = 0.5x10
cl
0.8
0.8
0.4
0.4
Meas. 2D corr.
RFOIL N=9
RFOIL N=11
RFOIL N=12
0.0
0.000
0.0
0.015
cd
0.030
0
5
10
15
o
20
α( )
Figure 4.2: NACA 0018 characteristics at Re=500,000 trip at 80%
At Re = 700,000 the same phenomena are present as at Re = 500,000. In this case the
differences are even larger, resulting in a larger overprediction of L/D (see figure 4.3).
The drag bucket of the RFOIL data is again wider than that of the measured data.
Also here the RFOIL data using Ncrit = 12 is slightly more similar to the measured
data for larger angles. In this case however this gives problems with convergence after
α = 16. For smaller angles the drag coefficient for Ncrit = 9 is more accurate. As it
is important to be able to gather data at high angles of attack, a value of Ncrit = 9 is
preferred.
M.Sc. thesis
M.C. Claessens
36
RFOIL
1.2
1.2
cl
NACA 0018
6
Re = 0.7x10
cl
0.8
0.8
0.4
0.4
Meas. 2D corr.
RFOIL N=9
RFOIL N=10
RFOIL N=12
0.0
0.000
0.0
0.015
cd
0.030
0
5
10
15
o
20
α( )
Figure 4.3: NACA 0018 characteristics at Re=700,000 trip at 80%
M.C. Claessens
M.Sc. thesis
4.2 Laminar profiles
4.2
37
Laminar profiles
For the laminar airfoils two different types are reviewed, the NACA 6-series and the
NLF-series. The NACA series are developed after the turbulent NACA series. The
NLF profiles are more modern laminar profiles developed by NASA in the 1980’s.
Furthermore the S824 profile is reviewed as a symmetric laminar airfoil, although the
measurements on this profiles were not conducted at the TU Delft.
4.2.1
NACA 64-418
The result for the NACA 64-418 at Re = 280,000 using RFOIL is given in figure 4.4.
The experimental data is measured using pressure holes in the model. No corrections
for 3D effects are necessary. The lift data are almost the same using Ncrit = 12. Only
the maximum lift coefficient is lower for the RFOIL data. The drag is at low α’s is
slightly higher and the drag bucket at positive Cl is wider.
1.2
cl
NACA 64-418
Re = 0.3x106
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
10
20
-0.4
-0.4
Meas. Re=280,000 (30%)
RFOIL, Re=300,000 (30%)
-0.8
-0.8
cd
o
α( )
Figure 4.4: NACA 64-418 characteristics at Re=280,000
The result for simulating the NACA 64-418 at Re = 700,000 using RFOIL is given in
figure 4.5. Again the RFOIL results are very close to the measurements using Ncrit = 9.
For the negative Cl the airfoil stalls smoother and later. This also results in a wider
drag bucket. For positive Cl , however, the results are very good. Only the lift is
slightly lower for α ≥ 10.
M.Sc. thesis
M.C. Claessens
38
RFOIL
1.2
1.2
cl
NACA 64-418
Re = 0.7x106
cl
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
10
20
-0.4
-0.4
Meas. Re=700,000 (30%)
RFOIL, Re=700,000 (30%)
-0.8
-0.8
cd
o
α( )
Figure 4.5: NACA 64-418 characteristics at Re=700,000
4.2.2
NLF-0416
Unfortunately not much data is available for NLF profiles at low Reynold numbers.
Data is available for Re=500,000, measured at the DUT, faculty of Aerospace Engineering. The angle of attack ranges from -4◦ to 15◦ , see figure 4.6. The RFOIL lift
data has a flatter slope compared to the measured data. Also the maximum measured
lift coefficient is much larger. The same goes for the drag: the measured drag is larger
and the drag bucket is wider.
M.C. Claessens
M.Sc. thesis
4.2 Laminar profiles
39
1.6
1.6
NLF 0416
6
Re = 0.5x10
cl
cl
1.2
1.2
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-10
0
-0.4
-0.4
-0.8
cd
-0.8
10
20
Meas.
RFOIL N=9
RFOIL N=12
o
α( )
Figure 4.6: NLF 0416 characteristics at Re=500,000
4.2.3
S824
The symmetric S824 is also a laminar profile and was measured by [Maughmer, 1999].
In this case the RFOIL data is again very close to the measurements using Ncrit = 9.
Only now, instead of under predicting the lift, the lift is over predicted for α ≥ 10.
M.Sc. thesis
M.C. Claessens
40
RFOIL
1.2
1.2
cl
S824
6
Re = 0.72x10
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-10
0
10
20
-0.4
-0.4
Meas.
RFOIL N=9
-0.8
-0.8
cd
o
α( )
Figure 4.7: S824 characteristics at Re=720,000
4.3
Higher Reynolds numbers
As stated earlier, RFOIL has some difficulties to calculate airfoil behavior at lower
Reynolds numbers. To illustrate this the results at higher Re for one turbulent airfoil
(NACA 4418) and one laminar airfoil (NACA 64-218) will be discussed in this section.
For the turbulent profile RFOIL tends to underestimate the drag over the whole range,
see figure 4.8. The qualitative behavior is almost the same, only the lift slope of the
measured data from [Abbott and Doenhoff, 1949] is lower. As a results the RFOIL
data has a higher CL,max . So even at higher Reynolds numbers RFOIL has some
trouble predicting turbulent profiles very accurately. Still the RFOIL program is a fast
and qualitatively correct design tool for 2D airfoils.
For the laminar profiles RFOIL is an accurate design tool for predicting the drag over
the whole range, see figure 4.9. The behavior is almost the same, only at α > 8
again the lift slope of the measured data from [Abbott and Doenhoff, 1949] decreases.
RFOIL proves to be a powerful design tool for laminar profiles.
M.C. Claessens
M.Sc. thesis
4.3 Higher Reynolds numbers
41
1.4
cl
cl
1.4
1.0
1.0
0.6
0.6
0.2
0.2
-0.2
0
0.015
-0.6
0.03
0.045
-15
-5
-0.2
5
15
-0.6
NACA 4418
Re = 3.0x106
25
Abbott
RFOIL
-1.0
-1.0
o
α( )
cd
Figure 4.8: NACA 4418 characteristics at Re=3,000,000
1.4
1.4
cl
cl
1.0
1.0
0.6
0.6
0.2
0.2
-0.2
0
-0.6
-1.0
0.015
0.03
0.045
NACA64-218
Re = 3.0x106
-10
-0.2
0
10
20
-0.6
Abbott
RFOIL
-1.0
cd
o
α( )
Figure 4.9: NACA 64-218 characteristics at Re=3,000,000
M.Sc. thesis
M.C. Claessens
42
RFOIL
4.4
Fixed transition at 5% chord
If a turbine operates in the field the blades will eventually get dirty from particles
and insects in the air. As a consequence the boundary layer will turn turbulent at
the nose of the blade, which results in a turbulent boundary layer over the airfoil. To
mimic this situation in the wind tunnel, zigzag tape is applied at 5% chord of the
airfoil. The resulting turbulent boundary layer will imediately have some thickness as
a result of the thickness of the tape. In RFOIL it is only possible to apply a trip to the
boundary layer without physical height. To get comparable boundary layer thickness
at 5% chord the trip is applied further near the nose. RFOIL has problems at higher
angles of attack when the stagnation point at the pressure side moves past this trip
point. To avoid this from happening the trip is applied at 1% at the suction side and
5% at the pressure side of the airfoil. If negative angles are investigated, the trips are
applied the other way around.
To validate RFOIL performance for fixed transition airfoils, the RFOIL values are
compared to the measured values for the NACA 0018 at Re=300,000, see figure 4.10
and Re=700,000, see figure 4.11.
1.2
1.2
cl
NACA 0018
6
Re = 0.3x10
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
cd
10
20
30
Meas. trip at 5%
Meas. trip at 30%
RFOIL trip at 1%
o
α( )
Figure 4.10: Comparison between 5% trip RFOIL and measured data at Re=300,000
The results of RFOIL deviate substantially from the measured values. The wind tunnel
results with zigzag tape applied at 30% are also added. It is clear that the lift results
of RFOIL are almost the same as for these last wind tunnel meaurements. The drag
values for both Reynolds numbers are also closer to zigzag tape applied to 30% chord.
Therefore the RFOIL results for ”dirty airfoils” have to be viewed with some scepticism.
M.C. Claessens
M.Sc. thesis
4.4 Fixed transition at 5% chord
43
1.2
1.2
cl
NACA 0018
6
Re = 0.7x10
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
20
30
Meas. trip at 5%
Meas. trip at 30%
RFOIL trip at 1%
-1.2
cd
o
α( )
Figure 4.11: Comparison between 5% trip RFOIL and measured data at Re=700,000
The fixed transition airfoil charateristics will not be used in the airfoil comparisons.
M.Sc. thesis
M.C. Claessens
44
RFOIL
4.5
Conclusions
After the validation the following conclusions are drawn concerning the RFOIL results:
• For the turbulent profiles RFOIL predicts the behavior accurately for Re=300,000
to an angle of attack of 15◦ . For Re=500,000 and 700,000 RFOIL tends to
overestimate the maximum lift coefficient, at 700,000 even by 10%. Also the
calculated lift slope is too steep.
• For the laminar NACA 6-series profiles RFOIL underestimates the drag at low angles of attack and underestimates the maximum lift coefficient. For Re=500,000
the drag prediction is accurate but the maximum lift coefficient is under predicted. For Re=700,000 the drag is again under predicted, but the lift prediction
is accurate to α = 10◦ . The lift slope is in all cases accurate.
• For the laminar NLF profiles only a data set for Re=500,000 is available. In
this case RFOIL under predicts the drag, the lift slope and the maximum lift
coefficient.
• For fixed transition airfoils RFOIL is incapable of producing good results. The
RFOIL values for a tip at 5% chord are more similar to applying zigzag tape at
30% chord. This can be the results the thickness of the applied rougness which
can not be simulated by the RFOIL program
• The calculated RFOIL data is compared to the measured wind tunnel data. To
compensate for the low turbulence it the tunnel the critical amplification number
of 12 should be used at Reynolds number 300,000. At Reynolds number of 500,000
Ncrit is equal to 11 and at 700,000 it equals 9
The RFOIL program is a design tool which allows changes to the airfoils geometry
and the pressure distributions. It is a fast program and gives qualitatively good results
compared to wind tunnel data. The quantitative values show differences with measured
wind tunnel data. The differences are larger at lower Reynolds numbers than at higher
Reynolds numbers. These discrepancies should be kept in mind when designing the
new airfoil. As during this project the RFOIL data of one airfoil is compared to the
RFOIL data of another airfoil, complete similarity between RFOIL and wind tunnel
data is not necessary. However, to compare the results of the design process with the
NACA 0018 airfoil wind tunnel tests should be performed.
M.C. Claessens
M.Sc. thesis
Chapter 5
Airfoil design parameters
In the previous two chapters the design tools are introduced; the RFOIL program to
generate 2D airfoil data and the simulation program to use the 2D data to calculate the
VAWT performance. In chapter 2 the design goals for the new airfoil are constructed.
In this chapter the design tools are used to generate a general design of the new airfoil
that complies with the design goals. The fine tuning and validation of this model
will be performed in the next chapter. To comply with the design goals the following
parameters can be altered: thickness, camber, type of airfoil (laminar or turbulent)
and noise emission. Also the possibility to design an airfoil which can make the turbine
self-starting is investigated. First in section 5.1 a study of each of these parameters is
given.
As RFOIL is used, the new designs will be compared to the RFOIL values of the
current airfoil, the NACA 0018. As stated before, there are two main groups of airfoils
which may be applicable for VAWT use. These groups are: turbulent profiles (NACA
4-series) and laminar profiles (NACA 6-series and NLF profiles). First these groups
will be compared to choose which one provides the most potential for a good airfoil.
The comparisons will be done at the Reynolds numbers 300,000 and 700,000.
5.1
Design variables
In this section an overview will be given of the influence of thickness and camber on
the aerodynamic properties. Also the relevance of boundary layer tripping and noise
emissions are treated. Finally a short comment about design parameter influence on
the self starting properties of turbines is given.
M.Sc. thesis
M.C. Claessens
46
Airfoil design parameters
5.1.1
Thickness
The current airfoil used on the Turby is the NACA 0018. The thickness has been
chosen to give enough structural strength with respect to the loads on the blades. A
thicker airfoil of 20% can be an option.
The advantage of higher thickness:
• Increase of the drag bucket
• Increase of structural strength
The disadvantages:
• Higher drag coefficient at lower angles of attack
• Chance of ’overshooting’ the maximum, past a certain unknown point more thickness will result in a lower efficiency
The optimum of the thickness is difficult to find. In older VAWT less thick airfoils
were used of 12% or 15%. The 18% thick airfoils also produce very good results. The
question is how much the thickness can be increased without performance loss. From
figure 5.1 the increase of thickness from 9% to 15% results in a wider drag bucket.
But if the thickness is increased form 15% to 18% the drag bucket does not increase
anymore. Figure 5.2 shows for a lower Reynolds number the same results.
M.C. Claessens
M.Sc. thesis
5.1 Design variables
47
1.2
1.2
NACA 00xx
Re = 0.25x106
cl
cl
0.8
0.8
0.4
0.4
NACA 0009
NACA 0012
NACA 0015
NACA 0018
0.0
0.000
0.0
0.015
0.030
-0.4
-10
0
10
20
-0.4
cd
30
o
α( )
Figure 5.1:
NACA characteristics with different thickness at Re≈250,000,
[Jacobs and Sherman, 1937]
1.0
1.0
NACA 00xx
6
Re = 0.15x10
cl
cl
0.6
0.6
0.2
0.2
0.000
0.015
0.030
0.045
-10
-5
0
-0.2
-0.2
-0.6
-0.6
5
10
NACA 0009
NACA 0012
NACA 0033
-1.0
cd
-1.0
o
α( )
Figure 5.2: Symmetric NACA characteristics with varying thickness at Re=150,000,
[Althaus, 1980]
M.Sc. thesis
M.C. Claessens
48
Airfoil design parameters
If the airfoil reaches a certain thickness, more thickness will decrease instead of increase
the performance of the turbine. Most turbines are not tested with airfoils of varying
thickness, but the results of one test is given in figure 5.3. It is clear that the 15%
thick airfoil performs better then the 12% thick airfoil, with a CP,max increase of 44%.
0.5
3 NACA0015, Re=300.000
3 NACA0012, Re=400.000
0.4
Cp
0.3
0.2
0.1
0.0
0
1
2
3
4
5
6
7
8
9
Tip speed ratio
Figure 5.3:
The Sandia 5m turbine with NACA 0015 and 0012 blades,
[Sheldahl et al., 1980]
5.1.2
Camber
Until now in almost all vertical axis wind turbines only symmetrical airfoils have been
applied (mostly NACA 0015 and NACA 0018). From a physical point it is clear that
the velocity at the upwind side of the turbine will be higher than at the downwind
side, as the blade extracts energy from the air at the first passage. The lower airspeeds
result in lower angles of attack at the downwind side, as these two parameters are
directly related. The power extraction from the wind is a function of V 3 . Only a small
variation in wind speed gives a large difference in possible power extraction. Therefore
an optimal power extraction at the upwind side is preferable. This will have as negative
result that the velocity, and therefore the extracted power, at the downwind side will
decrease even more.
To increase the efficiency of the turbine camber can be added to the profile. In this
case the profile will be more efficient at the upwind side. Even more energy will be
extracted from the wind, and the angle of attack at the downwind side will be even
lower. This is very important, because the efficiency at the downwind side will be much
lower. With a lower negative maximal a.o.a. the drag bucket can be moved more easily
to higher lift coefficient values.
M.C. Claessens
M.Sc. thesis
5.1 Design variables
49
Upwind
Downwind
1
1
0.9
0.9
λ=2
0.8
0.8
0.7
λ=4
0.6
V /Vinf
V /Vinf
0.7
0.5
0.6
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
−90
−45
0
θ (◦ )
45
90
0
90
135
180
θ (◦ )
225
270
Figure 5.4: Variation of wind speed in the VAWT
The increase in camber results in a shift of the lift curve in upwards direction. The
stall angle and the maximum lift coefficient increase, see figure 5.5.
The lift over drag curve shifts to the right with increasing camber. The performance
at higher lift coefficients is increasing, but at negative lift coefficients the performance
decreases at the same time.
M.Sc. thesis
M.C. Claessens
50
Airfoil design parameters
1.4
cl
NACA xx12
Re = 0.25x106
1.4
cl
1.0
1.0
0.6
0.6
0.2
0.2
0.000
-0.2
-0.6
0.015
0.030
0
10
20
-0.2
with
different
30
NACA 0012
NACA 2412
NACA 4412
NACA 6412
-0.6
cd
Figure 5.5:
NACA characteristics
[Jacobs and Sherman, 1937]
5.1.3
-10
o
α( )
camber
at
Re≈250,000,
Boundary layer tripping
At low Reynolds numbers the laminar boundary layer can separate before it turns
turbulent. This causes the laminar separation bubbles to appear. These bubbles alter
the effective shape of the profile and influence the lift curves, see section 2.2.5. If the
boundary layer is made turbulent artificially before the starting point of these bubbles,
the formation of the bubbles is prevented. This tripping of the boundary layer has a
large influence on the performance at low Reynolds numbers.
In [Gopalarathnam et al., 2001] an extensive study is made on the effects of boundary
layer tripping. The drag is decreased in the high lift, low Reynolds number condition,
compared to the clean airfoils. This is at a cost in the low lift, high Reynolds number
condition. For lower Reynolds numbers, 100,000 and smaller, and situations with
thicker airfoils, tripping can improve the performance substantially. Research into the
SA7026 profile shows the effect of applying a fixed transition, see figures 5.6 and 5.7.
The effects are much more significant at low Reynolds numbers. At higher Reynolds
numbers the tripping of the boundary layer gives a slight shift in the lift over drag curve.
If the Reynolds number is high enough, the flow will become turbulent naturally before
laminar separation bubbles can appear.
M.C. Claessens
M.Sc. thesis
5.1 Design variables
51
0.030
0.025
Cd
0.020
0.015
0.010
Clean
Trip at 0.2c
Trip at 0.4c
0.005
0.000
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Cl
Figure 5.6: SA7026 airfoil characteristics at Re=100,000, [Gopalarathnam et al., 2001]
0.025
0.020
Cd
0.015
0.010
Clean
Trip at 0.2c
Trip at 0.4c
0.005
0.000
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Cl
Figure 5.7: SA7026 airfoil characteristics at Re=300,000, [Gopalarathnam et al., 2001]
M.Sc. thesis
M.C. Claessens
52
5.1.4
Airfoil design parameters
Noise
Not much research is performed on the noise emission of the VAWT. Form
[Iida et al., 2004] a the aerodynamic sound of a VAWT is numerically modeled using discreet vortex methods. The complicated wake structure can be captured using
this method. It shows that the VAWT produces less sound than a horizontal axis wind
turbine with the same power coefficient at normal operating speed. The simulated
sound against the tip speed ratio is given below in figure 5.8. The operating tip seed
ratio is 3, which results in a produced aerodynamic noise of 60 dB.
Figure 5.8: The radiated aerodynamic noise from a VAWT, [Iida et al., 2004]
M.C. Claessens
M.Sc. thesis
5.1 Design variables
5.1.5
53
Self starting turbines
In general a VAWT with fixed pitch blades is unable to start on itself. The main
problem for Darrieus turbines is the negative power coefficient at low tip speed ratios.
If the power coefficient is positive, the turbine is able to rotate independently and
produce power. If the coefficient is negative, the turbine needs extra power to be able
to rotate.
The region of negative CP is influenced by the blade camber, thickness
and turbine solidity.
The influence of these parameters was investigated by
[Kirke and Lazauskas, 1991]. By adding 4% camber the negative coefficients are completely avoided, see figure 5.9. Kirke and Lazauskas have also predicted that a turbine
with cambered NACA 4415 blades of 0.32 m chord should easily self-start in a 10 m/s
wind (Re = 200,000), unlike an otherwise identical turbine with symmetrical NACA
0015 blades. However the maximum efficiency of the turbine would be significantly
lower with cambered blades.
0.4
NACA 0015
NACA 4415
0.3
Cp
0.2
0.1
0.0
0
1
2
3
4
5
-0.1
Tip speed ratio
Figure 5.9: Camber influence on the performance coefficient at Re=200,000,
[Kirke and Lazauskas, 1991]
By increasing the thickness, the region with negative CP coefficients is drastically
decreased, see figure 5.10. [Kirke and Lazauskas, 1991] have predicted the performance
of a turbine at Re = 200,000 with symmetrical NACA blades of thickness ranging from
12% to 21% and found that increasing the blade thickness reduces the severity of
the dead band of negative torque between λ = 1 to 3 but does not eliminate it, i.e.
increased thickness is an improvement but does not in itself ensure self-starting.
The increase of solidity also diminishes the region of negative CP and even makes the
CP values completely positive. This is only applied at high solidity values of 0.6 and
higher. As this will result in very large and expensive blades, so this is not an option
in this research.
M.Sc. thesis
M.C. Claessens
54
Airfoil design parameters
0.4
NACA 0012
NACA 0015
NACA 0018
NACA 0021
0.3
Cp
0.2
0.1
0.0
0
1
2
3
4
5
-0.1
Tip speed ratio
Figure 5.10: Influence of airfoil thickness on the turbine performance at Re=200,000,
[Kirke and Lazauskas, 1991]
M.C. Claessens
M.Sc. thesis
5.2 Turbulent profiles
5.2
55
Turbulent profiles
Most vertical axis wind turbines built so far use airfoil profiles from the NACA 4-series
and most of them use symmetric airfoils. The turbulent NACA 4-series were developed
in the 1930’s. Also the NACA 0018 belongs to this group. The main characteristic of
the profiles is a large nose radius. This results in a high pressure peak followed by a
steep pressure slope. The flow will turn turbulent very fast, as the laminar boundary
layer can not follow this pressure rise. Because of the large nose radius the stall
occurs very gentle. The drag bucket is not a real bucket, but a slowly increasing line,
see figure 5.11. First will be investigated if the NACA 0018 profile can be adjusted by
changing thickness or adding camber in such a way that the performance for the VAWT
will be improved. The airfoils are constructed from a fixed thickness distribution:
±yt =
√
t 0.29690 x − 0.12600x − 0.35160x2 + 0.28430x3 − 0.10150x4
0.20
(5.1)
To this thickness distribution camber can be added in the form of a mean line:
m
(2px − x2 ) if y < ymax
p2
m
yc = 2 (1 − 2p) + 2px − x2 if y ≥ ymax
p
yc =
5.2.1
(5.2)
(5.3)
Thickness
The extra thickness causes the airfoil to have more drag at smaller α. This is caused
by the extra friction drag resulting from a larger surface. In some airfoil series a
larger thickness makes a wider drag bucket. For the turbulent profiles the drag bucket
becomes smaller and the maximum lift coefficient decreases as well, see figures 5.11
and 5.12.
5.2.2
Camber
Adding camber shifts the lift curve and therefore also the drag bucket up, see figures 5.13 and 5.14. As the VAWT operates at higher positive angles then lower angles
this could result in better performance.
M.Sc. thesis
M.C. Claessens
56
Airfoil design parameters
1.2
1.2
6
cl
Re = 0.3x10
cl
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
0018
0015
0020
1420
-1.2
-1.2
20
o
α( )
cd
Figure 5.11: Calculated characteristics of different turbulent profiles at Re=300,000
1.4
1.4
6
cl
Re = 0.7x10
cl
1.0
1.0
0.6
0.6
0.2
0.2
0.000
-0.2
0.015
0.030
-20
-10
-0.2
-0.6
-0.6
-1.0
-1.0
-1.4
-1.4
cd
0
10
20
0018
0015
0020
1420
o
α( )
Figure 5.12: Calculated characteristics of different turbulent profiles at Re=700,000
M.C. Claessens
M.Sc. thesis
5.2 Turbulent profiles
57
6
c1.2
l
cl
Re = 0.3x10
1.2
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
0018
1418
2418
4418
-1.2
-1.2
20
o
α( )
cd
Figure 5.13: Calculated characteristics of different cambered profiles at Re=300,000
c1.2
l
6
Re = 0.7x10
cl
1.2
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
cd
10
20
0018
1418
2418
4418
o
α( )
Figure 5.14: Calculated characteristics of different cambered profiles at Re=700,000
M.Sc. thesis
M.C. Claessens
58
5.2.3
Airfoil design parameters
Simulation
The influence of the changes in camber and thickness on the performance of the VAWT
is found by using the simulation. The results for changing the thickness can be found
in figure 5.15, and changing the camber in figure 5.16. The extra thickness results
in performance loss over the complete interval. A thickness decrease would improve
performance, but because of structural demands this is not an option.
0.55
Vinf = 10 m/s
NACA 0015
NACA 0018
NACA 0020
CP
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 5.15: Turbulent profile characteristics for thickness variation
If camber is added the performance also declines. This decline is not consistent over
the complete interval. If only 1% is added, the decline in CP,max is very little. At tip
speed ratios larger than 3.5 the performans is a little better. However this region is
not important, as the VAWT operates at a ratio of 3 and during start up the ratios
are even lower.
M.C. Claessens
M.Sc. thesis
5.2 Turbulent profiles
59
0.55
Vinf = 10 m/s
NACA 0018
NACA 1418
NACA 2418
NACA 4418
CP
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 5.16: Turbulent profile characteristics for camber variation
M.Sc. thesis
M.C. Claessens
60
5.3
Airfoil design parameters
Laminar profiles
As the turbine operates at low Reynolds numbers, the flow tends to be laminar over a
larger part of the airfoil. The amount of disturbance needed to make the flow transition
to turbulent is larger at lower Reynolds numbers. A logical choice would be to look
at laminar profiles. These profiles have a sharper nose and the point of maximum
thickness lies more aft. This enables a larger part with laminar flow and therefore
lower drag at small angles of attack. The stall behavior is more rigorous compared to
turbulent profiles. At stall the drag increases sharply, which results in a typical sharp
drag bucket, see figure 5.17. At higher angles of attack the drag will be higher. For
the comparison the NACA 6-series and the more modern NFL-series are used.
5.3.1
The NACA 6-series
If the turbine operates for a longer time in the field the blades get dirty and laminar
flow is prevented. The 6-series are known to perform better than the turbulent profiles
in laminar flow, but generally perform almost just as well when dirty. There is not
one thickness distribution and mean line in this series, so the airfoils can not simply
be scaled to the desired characteristics. However, some airfoils of this series have the
same camber line in common and have thickness distributions that are similar.
Increasing the thickness or introducing camber has predictable influence on the airfoil
characteristics. The laminar airfoils profit form lower drag in the range from [−9◦ ≤
α ≤ 9◦ ] for Re=300,000 and [−7◦ ≤ α ≤ 7◦ ] at Re=700,000, see figures 5.17 and 5.18.
At higher angles the turbulent profile clearly performs better.
Simulation
The characteristics of the laminar profiles are inserted into the simulation program.
The results are given in figure 5.19 below. The laminar counterpart of the NACA 0018
profile, the NACA 63018, performs worse over the complete interval. Especially at
lower tip speed ratios the laminar profile can not compete with the turbulent one.
This is caused by the fact that laminar profiles perform worse at higher angles of
attack. At higher tip speed ratios the maximum angle of attack decreases and the
performance of laminar profiles equals that of the turbulent. Changing the thickness
or introducing camber does not have a beneficial effect on this performance.
M.C. Claessens
M.Sc. thesis
5.3 Laminar profiles
c1.2
l
61
6
cl
Re = 0.3x10
1.2
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
10
20
-0.4
-0.4
0018
63018
63020
63118
63120
63418
-0.8
-0.8
-1.2
-1.2
o
α( )
cd
Figure 5.17: Calculated characteristics of different laminar profiles at Re=300,000
c1.2
l
6
cl
Re = 0.3x10
1.2
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
cd
10
20
0018
63018
63020
63118
63120
63418
o
α( )
Figure 5.18: Calculated characteristics of different laminar profiles at Re=700,000
M.Sc. thesis
M.C. Claessens
62
Airfoil design parameters
0.55
Vinf = 10 m/s
NACA 0018
NACA 63-018
NACA 63-118
NACA 63-020
NACA 63-120
CP
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 5.19: Simulation of laminar profiles compared to the NACA 0018
5.3.2
NLF profiles
Since 1980 the focus on airfoil designed shifted more to the laminar flow airfoils. It was
found that the laminar profiles when dirty can perform just as well as the turbulent
profiles in dirty conditions. But if the airfoils are clean, the laminar profiles perform
much better. New construction methods and materials like composites made more
smooth and accurate blades possible. This ensured that the laminar design can perform
as calculated in practice. The emphasis of NASA also focused on these natural-laminarflow (NLF) while retaining the high lift coefficients on turbulent NASA airfoils. Three
airfoils were designed: the NLF-0416 and the flapped NLF-0215F and NLF-0414F. The
laminar flow and the resulting low drag at small angles attack combined with good high
lift performance should give improvements in the VAWT case in comparison with the
NACA 0018.
The NLF-416 is used as the basis, which is transformed to a NLF-0018 profile by
increasing thickness and removing camber in RFOIL. Besides the symmetrical airfoil
also cambered airfoils with 0.5, 0.8 and 1 percent camber are investigated. In figure 5.20
the results for Re=300,000 are given. The NLF 0018 profile clearly performs better
then the NACA 0018. The NLF profiles have a much larger drag bucket then the
NACA 6-series. The 6-series does have lower drag at low angles of attack, but also a
small drag bucket. The NLF series have overcome this and produce lower drag at low
angles combined with a wider drag bucket. But if the Reynolds number is increased
to 700,000, see figure 5.21 the benefits are smaller. The NACA 0018 has in this case
a higher CL,max and drag bucket, although the NLF profile still performs better for
−10 < α < 10. The simulation program will have to decide which has the most effect
on the performance.
If the 2D data used in the simulation program the symmetrical NLF 0018 profile is
M.C. Claessens
M.Sc. thesis
5.3 Laminar profiles
1.2
cl
63
NLF 0x18
6
Re=0.3x10
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
-0.4
0
10
20
-0.4
-0.8
NACA 0018
NLF 0018
NLF 0(0.5)18
NLF 0(0.8)18
NLF 0118
-0.8
-1.2
-1.2
cd
o
α( )
Figure 5.20: Characteristics for NLF airfoils with increasing camber at Re=300,000
1.2
cl
NLF 0x18
6
Re=0.7x10
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-0.4
-10
0
10
20
-0.4
-0.8
-1.2
-20
-0.8
cd
-1.2
NACA 0018
NLF 0018
NLF 0(0.5)18
NLF 0(0.8)18
NLF 0118
o
α( )
Figure 5.21: Characteristics for NLF airfoils with increasing camber at Re=700,000
M.Sc. thesis
M.C. Claessens
64
Airfoil design parameters
more efficient at tip speed ratios of 3.5 and higher. The normal operating conditions
are however around λ = 3. If camber is added the maximum efficiency increases, but
the profile becomes more peaky. Efficiency is lost at lower and higher tip speed ratios,
see figure 5.22. If the thickness is increased the CP curve shifts to lower tip speed
ratio’s, which is very beneficial.
0.55
NACA-0018
NLF-0018
NLF-0(0.8)18
10 m/s
NLF-0(0.8)20
NLF-0118
NLF-0416
Cp
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 5.22: Simulation of NLF profiles compared to the NACA 0018
5.3.3
Experimental results
Unfortunately only one investigation into the usefulness of laminar profiles for VAWT
applications is performed see [Leclerce, 1997]. The developed airfoil had to comply
with the following goals:
1. Modest values of CL,max with sharp stall characteristics
2. Low zero lift drag coefficient
3. Wide drag bucket
4. Operating at Re = 1 ∗ 106 - 3 ∗ 106
5. Large laminar flow region (50%)
6. Symmetrical airfoil
The Sandia 34 meter VAWT, which is originally equipped with two NACA 0018/0021
blades was compared to the same turbine with two SNLA 0018/NACA 0021 profile
blades. The SNLA 0018 profile is a modified NACA 0018 profile to produce a larger
laminar flow region, see figure 5.23.
M.C. Claessens
M.Sc. thesis
5.3 Laminar profiles
65
0.15
0.10
y/c
0.05
0.00
0.0
0.5
-0.05
1.0
x/c
NACA0018
-0.10
SNLA 0018
-0.15
Figure 5.23: Comparison of the SNLA 0018 profile with the NACA 0018, [Berg, 1985]
If the characteristics of both airfoils are compared, the only benefit of the SNLA profile
is the lower drag coefficient at lower angles of attack, see figure 5.24. The drag bucket
is smaller and the maximum lift coefficient and stall angle are smaller then for the
NACA 0018 profile. The characteristics match those of the NACA 6-series.
cl
1.2
SNLA 0018
Re = 3.0x106
cl
1.2
0.8
0.8
0.4
0.4
NACA 0018
SNLA 0018
0.0
0.000
0.0
0.010
cd
0.020
0
10
20
o
α ( ) 30
Figure 5.24: The NACA 0018 and SNLA 0018 lift characteristics for Re=3 million,
[Berg, 1985]
The results for the 34 meter turbine, equipped with both profiles, are shown in figure 5.25. The turbine was operated at a constant rotational speed, the wind speed
was altered. At lower wind speeds the maximum angle of attack of the profiles is very
low. With increasing wind speed this angle is increased. The low drag coefficients at
M.Sc. thesis
M.C. Claessens
66
Airfoil design parameters
low angles of attack of the SNLA profile are only beneficial at very low airspeeds. At
higher speeds the SNLA profile is according to these experiments less efficient as the
turbulent NACA 4 series airfoil.
(a) SNLA profile
(b) Turbulent profile
Figure 5.25: Results of the Sandia 34m turbine blades at 34rpm. [Berg, 1985]
The higher efficiency of the SNLA profile at low air speeds does improve the self
starting qualities of the turbine. See section 5.1.5 for more information about the self
starting properties. According to [Leclerce, 1997] and [Berg, 1985] the dynamic stall
behavior of SNLA profiles are better than the NACA 4-series as a result of the steeper
stall characteristics. The vibratory loads are reduced, therefore increasing the lifetime
of the turbine and possibly the noise. It is also mentioned by [Berg, 1985] that the
NLF airfoils perform better then the NACA 00xx in the Re range from 1 to 5 million,
below 1 million it is the other way around. However no prove is given to support
this theory. Unfortunately not enough data is available to check the results with the
Matlab simulation program. The Troposkien shaped blades can not be calculated, the
program is only applicable to VAWT with straight blades.
M.C. Claessens
M.Sc. thesis
5.4 Conclusions
5.4
67
Conclusions
Form the comparisons made in this chapter the following conclusions are made:
• For turbulent profiles little improvement possible. A decrease of thickness to 15%
would be benificial, but from structural point of view this is not an option
• Large increase in camber does not have beneficial results
• More thickness is not beneficial for turbulent and laminar profiles
• NACA 6-series does decrease drag at low Cl, but the drag bucket is too narrow:
the total performance is less
• NLF profiles have low drag at low CL , but also good performance at higher CL
• Laminar NLF profiles are the most promising
The optimal results for an airfoil for VAWT applications has low drag at small angles
of attack combined with a wide drag bucket. Turbulent profiles generally have a wide
drag bucket and laminar profiles have low drag at low angles of attack. The NLF
laminar profiles combine these both characteristics and therefore prove to be the best
basis for the airfoil design.
M.Sc. thesis
M.C. Claessens
68
M.C. Claessens
Airfoil design parameters
M.Sc. thesis
Chapter 6
Airfoil design
In the previous chapter is was established that the laminar NLF profiles are the most
promising airfoil series for a new VAWT profile. In this chapter the NLF 0018 profile will be taken as basis to investigate thickness (section 6.1) and camber changes
(section 6.2). When these parameters are fixed, more detailed changes to the airfoil
geometry and pressure distribution are investigated in section 6.3, resulting in the final
design given in section 6.4.
The final design is measured in the wind tunnel. The RFOIL data is compared with
the measured data to see whether the real airfoil complies with the expected results in
section 6.5. Finally the design will be compared to the NACA 0018 profile in section 6.6
to see how and where the improvements are made.
M.Sc. thesis
M.C. Claessens
70
Airfoil design
6.1
Thickness
If the thickness is increased, the drag bucket becomes smaller. The minimum thickness
with respect to structural demands is 18%. If thickness can be added, it would be a
great benefit. This would result in more structural strength and effectively a larger
operating envelope. In figure 6.1 the results for NLF profiles with different thickness
is given. The decrease in drag bucket between 18% and 20% is only small. If the
thickness is increased further, the zero lift drag increases and the drag bucket becomes
smaller. Although this leads to a strong blade, the negative effects on performance will
be too high.
1.6
1.6
cl
1.2
NLF
6
Re = 0.5x10
cl
1.2
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
cd
10
20
NACA 0018
NLF 0(0.8)18
NLF 0(0.8)20
NLF 0(0.8)22
NLF 0(0.8)25
o
α( )
Figure 6.1: Variation in thickness at Re=500,000
From the performance graph (see figure 6.2) the results of increasing thickness are
given. The increase in thickness leads to a forward shift of the power curve. If the
thickness is increased from 18% to 20% the maximum powercoefficient remains the
same, but is reached at lower tip speed ratios. If the thickness is increased further,
the CP,max decreases and the forward shift of the curve is only small. The optimum is
a thickness of 20%, as the CP,max is at its highest value and the efficiency curve lies
more forward than for 18%.
M.C. Claessens
M.Sc. thesis
6.2 Camber
71
0.55
Vinf = 10 m/s
NACA-0018
NLF-0(0.8)18
NLF-0(0.8)20
NLF-0(0.8)22
NLF-0(0.8)25
CP
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 6.2: Performance results for NLF profiles with different thickness
6.2
Camber
In the previous chapter the concept of camber has been discussed. In this new airfoil
design it was important to implement some camber to use the fact that the wind speed
are higher at the upwind side of the turbine.
In figure 6.4 the simulation results for varying camber are given. The addition of
camber to the airfoil does improve the performance significantly. when increasing
camber from 0 to 0.5% and 0.8% the CP,max increase is relatively high. If the camber
is increased further to 1%, the CP,max does increase, but at the cost of a smaller
interval. The optimum percentage of camber is chosen to be 0,8%.
M.Sc. thesis
M.C. Claessens
72
Airfoil design
1.2
cl
NLF 0x18
6
Re=0.5x10
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
10
20
-0.4
-0.4
NACA 0018
NLF 0018
NLF 0(0.5)18
NLF 0(0.8)18
NLF 0118
-0.8
-0.8
-1.2
-1.2
o
α( )
cd
Figure 6.3: Variation in camber at Re=500,000
0.55
Vinf = 10 m/s
NACA-0018
NLF-0018
NLF-0(0.5)18
NLF-0(0.8)18
NLF-0118
NLF-0416
CP
0.50
0.45
0.40
2.0
2.5
3.0
3.5
4.0
4.5
5.0
Tip speed ratio
Figure 6.4: Performance results for NLF profiles with different camber
M.C. Claessens
M.Sc. thesis
6.3 Fine tuning
6.3
73
Fine tuning
In this stage the thickness and camber values are set. Improvements are still possible
by fine tuning the pressure distribution and geometry of the airfoil in more detail.
In figure 6.5 the result of a nose radius adjustment is shown. For revision x of the
airfoil the nose radius is increased slightly. The result is an increase in CL,max and the
drag bucket is substantially wider (see the highlighted areas).
1.2
cl
DU 06-W-200
Re=0.5x106
cl
1.2
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
NACA 0018 clean
DU rev. f clean
DU rev. x clean
-1.2
-1.2
cd
o
α( )
Figure 6.5: Airfoil nose radius optimization
In figure 6.6 the airfoil is modified for a “bump” in the pressure distribution which
is present in the left pictures. By smoothing the pressure distribution the bump is
removed, see the results at the right pictures.
Numerous adjustments to the airfoil result in the final design of the airfoil. This shape
is given in the next section.
M.Sc. thesis
M.C. Claessens
74
Airfoil design
(a) α = −6 before
(b) α = −6 after
(c) α = −7 before
(d) α = −7 after
Figure 6.6: Removal of bump in the pressure distribution
M.C. Claessens
M.Sc. thesis
6.4 Final design
6.4
75
Final design
In the previous sections the final design of the new airfoil was established. The name
of the airfoil is determined by the conventions of the Aerospace Engineering faculty:
DU 06-W-200
The “DU” stands for Delft University, the “06” for the year 2006, the “W” for the
Wind turbine application and the “200” for the thickness of 20,0%. In figure 6.7 the
shape of the DU 06-W-200 airfoil is given. The coordinates from the RFOIL and the
wind tunnel model are given in appendix A.2.
0.15
0.10
y/c
0.05
0.00
0.0
0.5
-0.05
1.0
x/c
-0.10
NACA 0018
-0.15
DU 06-W-200
Figure 6.7: Final shape of the DU 06-W-200 compared with the NACA 0018
M.Sc. thesis
M.C. Claessens
76
Airfoil design
6.5
RFOIL comparison
In this section the computational results from RFOIL will be compared to the measured
wind tunnel data. The RFOIL data for the DU 06-W-200 airfoil is given in appendix C
and the data of the wind tunnel tests is given in appendix D. First the free transition
data will be compared. Also a comparison is made for the fixed transition airfoil data,
only the RFOIL data in this case should be viewed with skepticism (see section 4.4).
6.5.1
Free transition
In figures 6.8 to 6.10 the results for the comparison between RFOIL and measurement
data for the clean DU 06-W-200 airfoil are given. The RFOIL CL,max for Re=300,000
and 500,000 are a slightly too low and the drag is underestimated for all Reynolds
numbers. The qualitative aspects of the characteristics match however. For negative
angles of attack the lift curves fit exactly, for the positive angles of attack the RFOIL
lift curve slope is slightly steeper. The graphs show that the RFOIL program indeed
has problems to predict the characteristics when the airfoil stalls. When stall begins,
RFOIL has is no longer able to cope with the aerodynamic phenomena involved. The
solution becomes more and more instable until RFOIL is incapable of giving a result.
However, before the airfoil stalls the RFOIL results are very accurate.
Du 06-W-200
Re = 0.3x106
1.2
CL
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
p03s
RFOIL
-1.2
-1.2
CD
o
α( )
Figure 6.8: Comparison clean RFOIL and wind tunnel data for Re=300,000
M.C. Claessens
M.Sc. thesis
6.5 RFOIL comparison
77
Du 06-W-200
Re = 0.5x106
1.2
1.2
cl
CL
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
p05s
RFOIL
-1.2
-1.2
o
α( )
CD
Figure 6.9: Comparison clean RFOIL and wind tunnel data for Re=500,000
1.2
Du 06-W-200
Re = 0.7x106
1.2
cl
CL
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
p07s
RFOIL
-1.2
-1.2
o
α( )
CD
Figure 6.10: Comparison clean RFOIL and wind tunnel data for Re=700,000
M.Sc. thesis
M.C. Claessens
78
Airfoil design
6.5.2
Fixed transition
As stated in chapter 3 the RFOIL data for fixed transition conditions is not very
reliable. In figure 6.11 a comparison between the RFOIL data for the DU 06-W-200
is compared to the measured data at Re=300,000 and 700,000. The zigzag tape in
the measurement is applied at 5% chord and for the RFOIL calculations the trip is
applied at 1%. The RFOIL data overpredicts the results of the DU 06-W-200. the
same results were visible for the NACA 0018. This data can be found in section 6.6.2.
The drag is much lower and the drag bucket wider than for the measured data. The
real characteristics of the airfoils in dirty conditions can only be determined using wind
tunnel tests.
1.2
1.2
cl
DU 06-W-200
dirty
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.010
0.020
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
cd
-1.2
10
20
Meas. 03
RFOIL 03
Meas. 07
RFOIL 07
o
α( )
Figure 6.11: Comparison fixed transition RFOIL and measured data for Du 06-W-200
M.C. Claessens
M.Sc. thesis
6.6 Measurement comparison
6.6
79
Measurement comparison
The goal is to improve the performance at positive angles by adding camber and keep
the performance at negative angles equal to the NACA 0018. Furthermore 2% extra
thickness is added to improve the strength of the blades. In this section the wind
tunnel results from the Du 06-W-200 will be compared to the NACA 0018 wind tunnel
data. The wind tunnel data for the DU 06-W-200 airfoil is given in appendix D and
the data of the NACA 0018 wind tunnel tests is given in appendix B.
6.6.1
Free transition
In figures 6.12 to 6.15 the results for the comparison of the clean NACA 0018 and
DU 06-W-200 airfoils are given. The measurement results for the clean airfoil match
the predictions made using RFOIL. The zero lift drag of the DU 06-W-200 is slightly
higher than for the NACA 0018 as a results of the extra thickness. The performance
for the negative angles is the same for Re=300,000 and 500,000. For Re=700,000 and
1,000,000 the drag bucket is slightly smaller for the negative angles, but the CL,max
remains as high as for the NACA 0018.
For positive angles of attack the drag bucket is wider as a result of the added camber.
The CL,max occurs at the same angle of attack as for the NACA 0018, only the values
are much higher.
A characteristic that is difficult to predict is the deep stall angle of attack. For the
Reynolds numbers from 300,000 to 700,000 these angles are much higher for the Du
06-W-200 than for the NACA 0018. For Re=1,000,000 no measurements were made
for the DU 06-W-200 because of possible balance overload.
To give an indication of the difference in efficiency of both airfoils the lift over drag
data is given, for Re=500,000 in figure 6.16 and for Re=700,000 in figure 6.17. The
DU 06-W-200 clearly has a higher lift over drag ratio for positive angles of attack. For
negative angles the efficiency of both airfoils is nearly the same.
M.Sc. thesis
M.C. Claessens
80
Airfoil design
Du 06-W-200
6
Re=0.3x10
1.2
cl
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NACA 0018
-1.2
-1.2
o
α( )
cd
Figure 6.12: Comparison clean wind tunnel data for Re=300,000
Du 06-W-200
6
Re=0.5x10
1.2
cl
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NACA 0018
-1.2
-1.2
cd
o
α( )
Figure 6.13: Comparison clean wind tunnel data for Re=500,000
M.C. Claessens
M.Sc. thesis
6.6 Measurement comparison
81
Du 06-W-200
6
Re=0.7x10
1.2
cl
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NACA 0018
-1.2
-1.2
o
α( )
cd
Figure 6.14: Comparison clean wind tunnel data for Re=700,000
Du 06-W-200
6
Re=1.0x10
1.2
cl
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NACA 0018
-1.2
-1.2
cd
o
α( )
Figure 6.15: Comparison clean wind tunnel data for Re=1,000,000
M.Sc. thesis
M.C. Claessens
82
Airfoil design
60.0
L/D
6
Re=0.5x10
40.0
20.0
0.0
-30
-20
-10
0
10
20
α
30
-20.0
-40.0
DU 06-W-200
NACA 0018
-60.0
Figure 6.16: Comparison clean lift over drag data for Re=500,000
6
60.0
Re=0.7x10
L/D
40.0
20.0
0.0
-30
-20
-10
0
10
20
α
30
-20.0
-40.0
DU 06-W-200
NACA 0018
-60.0
Figure 6.17: Comparison clean lift over drag data for Re=700,000
M.C. Claessens
M.Sc. thesis
6.6 Measurement comparison
6.6.2
83
Fixed transition
The fixed transition characteristics of both the NACA 0018 and Du 06-W-200 were
not correctly predicted by RFOIL. In this section the measured characteristics of both
airfoils are compared, see figures 6.18 and 6.19. The DU 06-W-200 is an airfoil which
is based on a large laminar region. If the airfoil is dirty, the flow over the complete
airfoil is turbulent. The “laminar” DU 06-W-200 is expected to have a larger decrease
in performance than the “turbulent” NACA 0018.
The zero lift drag of the Du 06-W-200 is higher than for the for the NACA 0018, which
is expected for the thicker DU 06-W-200. The DU 06-W-200 has a much higher CL,max
for both negative and positive angles of attack. As a results the drag bucket is much
wider. For Re=300,000 the deep stall angle is the same for both airfoils for positive
angles, but for Re=700,000 the deep stall angles are much higher than for the NACA
0018.
Du 06-W-200
Re = 0.3x106
c0.8
l
cl
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NACA 0018
-1.2
-1.2
cd
o
α( )
Figure 6.18: Comparison dirty wind tunnel data for Re=300,000
The decrease in performance for the DU 06-W-200 in comparison to the clean airfoil
is less than for the NACA 0018. The new design, although it is a laminar profile,
performs good in dirty configuration as well.
M.Sc. thesis
M.C. Claessens
84
Airfoil design
1.2
1.2
Du 06-W-200
Re = 0.7x106
cl
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
DU 06-W-200
NAC0018
-1.2
-1.2
cd
o
α( )
Figure 6.19: Comparison dirty wind tunnel data for Re=700,000
M.C. Claessens
M.Sc. thesis
6.7 Simulation results
6.7
85
Simulation results
The measured data for the NACA 0018 and the DU 06-W-200 was implemented into
the simulation program. The calculated performance difference between both airfoils
is visualized for free transition (figure 6.20) and for fixed transition (figure 6.21). The
shaded regions indicate where the DU 06-W-200 outperforms the NACA 0018 profile.
For the clean blades the increase in CP starts much earlier, resulting in more power
output at lower tip speed ratios. The CP,max is increased with 5% to 0.48. This value
of CP,max lies for the DU 06-W-200 exactly at λ = 3, the operating tip speed ratio. If
this value is compared to the CP value for NACA 0018 at λ = 3, the increase is even
8%.
0.6
DMSV
Vinf = 10 m/s
0.5
DU 06-W-200
NACA 0018
0.4
CP
0.3
0.2
0.1
0
0
1
2
3
4
5
6
7
-0.1
Tip speed ratio
Figure 6.20: Turbine simulation results with the clean DU 06-W-200
If the blades are dirty, the performance of both airfoils decreases. The NACA 0018
has a large region with negative CP , where during the start-up phase extra energy is
needed to let the turbine rotate. This problem is not present at the DU 06-W-200
airfoil. Furthermore the CP,max is increased with 4% to 0.33. If the values of both the
airfoils at λ = 3 are compared, it increases even 108% to 0.28.
The DU 06-W-200 is clearly an improvement with respect to the NACA 0018. The
original turbine had to stop at Vinf = 14 m
s to prevent the blades from failing. The
new profile is 2% thicker, resulting in a higher structural strength. This ensures that
the turbine can operate at higher wind speeds. Also the maximum power coefficient
is increased and the performance is better at lower tip speed ratios. This offers the
possibility of operating at lower rotational speeds at high wind speeds while still producing adequate power. With lower rotational speeds the forces on the blades will be
lower as well, eliminating the dangers of failure.
The NACA 0018 airfoil has a problem with laminar separation bubbles that extend
M.Sc. thesis
M.C. Claessens
86
Airfoil design
0.6
DMSV
Vinf = 10 m/s
0.5
Du 06-W-200 dirty
NACA0018 dirty
0.4
CP
0.3
0.2
0.1
0
0
1
2
3
4
5
6
7
-0.1
Tip speed ratio
Figure 6.21: Turbine simulation results with the dirty DU 06-W-200
over the trailing edge of the airfoil. The bubbles cause the blades to vibrate and are a
source of loud noise. As the turbines are destined to operate in urban environments,
noise emission should be kept as low as possible. The laminar separation bubbles can
be eliminated using a trip on the blades, only this was not economically feasible for
the manufacturer to implement on the Turby VAWT. The DU 06-W-200 does not have
this problem, resulting in a noise reduction without the need of applying a trip on the
blades.
M.C. Claessens
M.Sc. thesis
Chapter 7
Conclusions and recommendations
This report covers the design process of an airfoil dedicated to VAWT applications.
In most VAWT produced today standard NACA symmetric airfoils are used. These
airfoils were developed in the 1930’s and for many of them usable measurement data
is not available. During this project the current NACA 0018 is measured and a new
design, the DU 06-W-200, is developed to improve the turbines performance. The
conclusions are given in section 7.1. Recommendations for further research are given
in section 7.2.
7.1
Conclusions
Design goals The NACA 0018 airfoil which is currently used in the Turby VAWT is
taken as reference point for the new design. After reviewing the available literate on
VAWT aerodynamics the following design goals were set:
• Operate at Reynolds numbers between 150,000 and 700,000 over a wide range of
angles α
• Large width of the low drag bucket
• Cambered airfoil to improve the performance of the turbine
• A larger thickness than 18% to increase structural strength
• Start of deep stall at a large angle of attack with the minimum amount of drop
in lift coefficient
• No laminar separation bubbles which extend past the trailing edge, which results
in noise reduction
M.Sc. thesis
M.C. Claessens
88
Conclusions and recommendations
RFOIL program RFOIL is a computational program based on the panel method
combined with boundary layer equations. This method is capable to predict 2D airfoil
characteristics accurately for higher Reynolds numbers until the airfoil enters stall. The
RFOIL code has more problems to accurately predict the the airfoil characteristics at
Reynolds numbers between 300,000 and 700,000. The airfoil design process for this
project lies in this Reynolds number range. However RFOIL gives accurate enough
results to be a powerful design tool in this range. The calculated characteristics for
a dirty airfoil shows no real similarity with available wind tunnel data, therefore the
RFOIL program should be used for clean airfoil design only.
Simulation program The developed simulation program is based on the momentum
theory. To enable velocity variation perpendicular to the free stream flow and between
the upwind and downwind side, the Double Multiple Streamtube model with variable
interference factors is implemented. The final Matlab program allows to adjust the
turbines geometry, to choose from multiple airfoils and to implement a dynamic stall
model or not. As such 2D airfoil characteristic from RFOIL can be inserted to view
the turbines performance with this airfoil.
Design process These design tools together enable to design a new airfoil. The design
process consisted of an iteration process of adjusting the airfoils shape, calculating the
characteristics using RFOIL and calculating the performance results using the simulation program. The calculated RFOIL characteristics and the simulated performance
results are then compared to the original NACA 0018 profile and previous designs in
the design process. By adjusting the type of airfoil (laminar or turbulent), the thickness and camber an optimum design is found in the laminar NLF series. Using the
NLF 0018 as starting point, this design is optimized by again adjusting the camber
and thickness, but also by fine-tuning the geometry and the pressure distribution. This
finally resulted in the final design:
DU 06-W-200
Validation The Du 06-W-200 airfoil is a laminar, 20% thick airfoil with 0.8% camber.
The original NACA 0018 airfoil is a turbulent, symmetric airfoil with 18% thickness.
To be able to compare both airfoils wind tunnel measurements were performed in the
LTT wind tunnel at the Delft University of Technology. This measured 2D data is
also used in the simulation program. From this data the following conclusions can be
drawn:
• The added thickness of 2% will add to the blade strength and this is reached
without decreasing the performance
• The added camber of 0.8% increases the performance with respect to a symmetric
airfoil
M.C. Claessens
M.Sc. thesis
7.2 Recommendations
89
• The DU 06-W-200 performs equal to the NACA 0018 for negative angles of attack
• The DU 06-W-200 has a much higher CL,max for positive angles of attack, resulting in a wider drag bucket
• Deep stall occurs at higher angles of attack with a lower drop in lift coefficient
• In contrast with the NACA 0018 the DU 06-W-200 does not have laminar separation bubbles which extend over the trailing edge
• The increase in turbine performance at operating tip speed ratio is 8% in clean
conditions and 108% when dirty
7.2
Recommendations
Within the time frame of this project and within the facilities available certain issues
are not covered in this project. Below the recommendations for future research in
this subject are listed. These recommendations are mainly focused on improving the
modeling methods for VAWT.
CFD modeling Performing CFD simulations of a complete 3D VAWT. This gives the
possibility to visualize the flow phenomena and to get more detailed quantitative data
of the flow through the turbine. The influence of the blades at the upwind side on
the flow at the downwind side and the influence of the rotation axis can be quantified.
Optimization of the geometry at the tips of the blades is also possible using CFD.
VAWT wind tunnel testing Wind tunnel test on VAWT to get more exact data to
validate the simulations. In literature little applicable data on these turbines is available. Testing the turbine under controlled conditions offer the possibility to measure
the power curve of the turbine without drive train influences and free stream airflow
variations. This data can be used to validate and improve the momentum based simulation model and the CFD simulations.
M.Sc. thesis
M.C. Claessens
90
M.C. Claessens
Conclusions and recommendations
M.Sc. thesis
Chapter 8
Bibliography
[Abbott and Doenhoff, 1949] Abbott, I. and Doenhoff, A. V. (1949). Theory of Wing
Sections. Dover Publications, Inc., New York. [40]
[Althaus, 1980] Althaus, D. (1980). Profilpolaren fur den Modellflug: Windkanalmessungen an Profiles im Kritischen Reynoldszahlbereich. Neckar Verlag. [47]
[Berg, 1985] Berg, D. (1985). Structural design of the sanidia 34-meter vertical axis
wind turbine. Technical Report SAND84-1287, Sandia National Laboratories. [65,
66]
[Brochier et al., 1986] Brochier, G., Fraunié, P., Beguiér, C., and Paraschivoiu, I.
(1986). Water channel experiments of dynamic stall on darrieus wind turbine blades.
Journal of Propulsion, 2(5):445–449. [11, 12]
[Claessens, 2006a] Claessens, M. (2006a). Du 06-w-200 measurement report. Technical
report, TU Delft. [109]
[Claessens, 2006b] Claessens, M. (2006b). Naca 0018 measurement report. Technical
report, TU Delft. [97]
[Fujisawa and Shibuya, 2001] Fujisawa, N. and Shibuya, S. (2001). Observations of
dynamic stall on darrieus wind turbine blades. Journal of Wind Engineering and
Industrial Aerodynamics, 89:201–214. [12, 13]
[Gopalarathnam et al., 2001] Gopalarathnam, A., Broughton, B., McGranaham, B.,
and Selig, M. (2001). Design of low reynolds number airfoils with trip. 19th AIAA
Applied Aerodynamics Conference. [50, 51]
[Gormont, 1973] Gormont, R. (1973). A mathematical model of unsteady aerodynamics and radial flow for application to helicopter rotors. Technical report, Boeing Co.
Vertol Div. [28]
M.Sc. thesis
M.C. Claessens
92
Bibliography
[Iida et al., 2004] Iida, A., Mizun, A., and Fukudome, K. (2004). Numerical simulation
of aerodynamic noise radiated from vertical axis wind turbines. Technical report,
Kogakuin University Department of Mechanical Engineering. [52]
[Jacobs and Sherman, 1937] Jacobs, E. and Sherman, A. (1937). Airfoil characteristics
as affected by variations of the reynolds number. Technical report, N.A.C.A. [15,
47, 50]
[Kirke and Lazauskas, 1991] Kirke, B. and Lazauskas, L. (1991). Enhancing the performance if a vertical axis wind turbine using a simple variable pitch system. Wind
Engineering, 15(4):187–195. [53, 54]
[Leclerce, 1997] Leclerce, C. (1997). Why use natural laminar airfoil profiles in vertical
axis wind turbine applications? AIAA Meeting Papers on Disc, -(AIAA 97-0005).
[64, 66]
[Loth and McCoy, 1983] Loth, J. and McCoy, H. (1983). Optimization of darrieus turbines with an upwind and downwind momentun model. Journal of Energy, 7(4):313–
318. [24]
[Maughmer, 1999] Maughmer, M. (1999). Wind-tunnel test of the s824 airfoil. Technical report, Pennsylvania State University. [39]
[Migliore et al., 1980] Migliore, P., Wofle, P., and Fanucci, J. (1980). Flow curvature
effect on darrieus tubrine blade aerodynamics. Journal of Energy, 4(2):49–55. [18]
[Paraschivoiu, 2002] Paraschivoiu, I. (2002). Wind Turbine Design. Polytechnic International Press. [16, 31]
[Paraschivoiu and Delclaux, 1983] Paraschivoiu, I. and Delclaux, F. (1983). Double
multiple streamtube model with recent improvements. Journal of Energy, 7(3):250–
255. [24]
[Sheldahl et al., 1980] Sheldahl, R., Klimas, P., and Feltz, L. (1980). Aerodynamic
performance of a 5-meter-diameter darrieus turbine with extruded aluminum naca0015 blades. Technical Report SAND80-0179, Sandia National Laboratories. [14,
16, 48]
[Strickland, 1975] Strickland, J. (1975). The darrieus turbine: A performance prediction model using multiple streamtubes. Technical Report SAND75-0431, Sandia
Laboratories. [23, 29, 31, 32]
[Templin, 1974] Templin, R. (1974). The Elements of Airfoil and Airscrew Theory.
Cambridge University Press, second edition. [22]
[W.A. Timmer, 2003] W.A. Timmer, R. v. R. (2003). Summary of the delft university
wund turbine dedicated airfoils. 41st AIAA Aerospace Sciences Meeting and Exhibit,
pages 11–21. [10]
M.C. Claessens
M.Sc. thesis
Appendix A
Airfoil coordinates
In this appendix the coordinates of the NACA 0018 (section A.1) and Du 06-W-200
(section A.2) profiles are given. These are the coordinates used for the wind tunnel
models and for the program RFOIL.
M.Sc. thesis
M.C. Claessens
94
A.1
Airfoil coordinates
NACA 0018
x/c
1
0.9938
0.9837
0.9728
0.961
0.9484
0.9349
0.9207
0.9059
0.8906
0.8749
0.8588
0.8426
0.8261
0.8095
0.7929
0.7761
0.7594
0.7426
0.7257
0.7089
0.6921
0.6752
0.6584
0.6416
0.6249
0.6081
0.5914
0.5748
0.5582
0.5416
0.5251
0.5087
0.4923
0.476
0.4598
0.4437
0.4277
0.4117
0.3959
0.3802
0.3647
0.3493
0.334
0.3189
0.3041
0.2894
0.275
0.2608
0.2469
y/c
0.0019
0.0032
0.0053
0.0075
0.0099
0.0124
0.0151
0.0178
0.0206
0.0235
0.0263
0.0292
0.0321
0.0349
0.0378
0.0405
0.0433
0.0459
0.0486
0.0511
0.0537
0.0561
0.0585
0.0608
0.0631
0.0653
0.0674
0.0695
0.0715
0.0734
0.0752
0.0769
0.0786
0.0801
0.0816
0.0829
0.0842
0.0853
0.0864
0.0873
0.0881
0.0887
0.0893
0.0896
0.0899
0.09
0.09
0.0898
0.0895
0.089
x/c
0.2333
0.2201
0.2073
0.1949
0.1829
0.1714
0.1604
0.15
0.14
0.1306
0.1218
0.1134
0.1056
0.0982
0.0913
0.0848
0.0787
0.073
0.0677
0.0627
0.0579
0.0535
0.0493
0.0454
0.0417
0.0382
0.0349
0.0318
0.0289
0.0261
0.0235
0.0211
0.0188
0.0166
0.0146
0.0127
0.011
0.0094
0.0079
0.0065
0.0053
0.0042
0.0032
0.0024
0.0017
0.0011
0.0007
0.0003
0.0001
0
y/c
0.0884
0.0876
0.0867
0.0856
0.0844
0.0831
0.0817
0.0802
0.0786
0.0769
0.0752
0.0734
0.0716
0.0698
0.0679
0.0661
0.0642
0.0623
0.0605
0.0586
0.0567
0.0549
0.053
0.0512
0.0493
0.0475
0.0456
0.0437
0.0419
0.04
0.0382
0.0363
0.0344
0.0325
0.0306
0.0287
0.0267
0.0248
0.0228
0.0208
0.0189
0.0169
0.0149
0.0128
0.0108
0.0088
0.0068
0.0048
0.0029
0.001
x/c
0
0.0001
0.0003
0.0007
0.0011
0.0017
0.0024
0.0032
0.0042
0.0053
0.0065
0.0079
0.0094
0.011
0.0127
0.0146
0.0166
0.0188
0.0211
0.0235
0.0261
0.0289
0.0318
0.0349
0.0382
0.0417
0.0454
0.0493
0.0535
0.0579
0.0627
0.0677
0.073
0.0787
0.0848
0.0913
0.0982
0.1056
0.1134
0.1218
0.1306
0.14
0.15
0.1604
0.1714
0.1829
0.1949
0.2073
0.2201
0.2333
y/c
-0.001
-0.0029
-0.0048
-0.0068
-0.0088
-0.0108
-0.0128
-0.0149
-0.0169
-0.0189
-0.0208
-0.0228
-0.0248
-0.0267
-0.0287
-0.0306
-0.0325
-0.0344
-0.0363
-0.0382
-0.04
-0.0419
-0.0437
-0.0456
-0.0475
-0.0493
-0.0512
-0.053
-0.0549
-0.0567
-0.0586
-0.0605
-0.0623
-0.0642
-0.0661
-0.0679
-0.0698
-0.0716
-0.0734
-0.0752
-0.0769
-0.0786
-0.0802
-0.0817
-0.0831
-0.0844
-0.0856
-0.0867
-0.0876
-0.0884
x/c
0.2469
0.2608
0.275
0.2894
0.3041
0.3189
0.334
0.3493
0.3647
0.3802
0.3959
0.4117
0.4277
0.4437
0.4598
0.476
0.4923
0.5087
0.5251
0.5416
0.5582
0.5748
0.5914
0.6081
0.6249
0.6416
0.6584
0.6752
0.6921
0.7089
0.7257
0.7426
0.7594
0.7761
0.7929
0.8095
0.8261
0.8426
0.8588
0.8749
0.8906
0.9059
0.9207
0.9349
0.9484
0.961
0.9728
0.9837
0.9938
1
y/c
-0.089
-0.0895
-0.0898
-0.09
-0.09
-0.0899
-0.0896
-0.0893
-0.0887
-0.0881
-0.0873
-0.0864
-0.0853
-0.0842
-0.0829
-0.0816
-0.0801
-0.0786
-0.0769
-0.0752
-0.0734
-0.0715
-0.0695
-0.0674
-0.0653
-0.0631
-0.0608
-0.0585
-0.0561
-0.0537
-0.0511
-0.0486
-0.0459
-0.0433
-0.0405
-0.0378
-0.0349
-0.0321
-0.0292
-0.0263
-0.0235
-0.0206
-0.0178
-0.0151
-0.0124
-0.0099
-0.0075
-0.0053
-0.0032
-0.0019
Table A.1: NACA 0018 coordinates
M.C. Claessens
M.Sc. thesis
A.2 DU 06-W-200
A.2
95
DU 06-W-200
x/c
1.00003
0.995071
0.988745
0.980844
0.971354
0.960703
0.949361
0.937673
0.925781
0.913799
0.901782
0.889754
0.877727
0.865704
0.853697
0.841679
0.829686
0.81774
0.805771
0.793799
0.781854
0.769889
0.757924
0.746007
0.734077
0.722128
0.710225
0.698325
0.686453
0.674589
0.662687
0.650773
0.638859
0.626934
0.615006
0.603147
0.591292
0.57943
0.567547
0.555676
0.543792
0.531879
0.519984
0.508056
0.496115
0.484187
0.472242
0.460354
0.448446
0.436599
y/c
0.00033
0.000991
0.00179
0.002821
0.004134
0.005676
0.00738
0.009188
0.011064
0.012967
0.014892
0.016828
0.018779
0.020741
0.022719
0.024714
0.026717
0.028741
0.030796
0.032857
0.034939
0.037053
0.03918
0.041327
0.043511
0.045716
0.047936
0.050176
0.052417
0.054685
0.056949
0.059221
0.06148
0.063716
0.065934
0.068109
0.070229
0.072307
0.074343
0.076339
0.078282
0.080195
0.08206
0.083885
0.085679
0.087431
0.089147
0.090814
0.092426
0.09399
x/c
0.424787
0.413003
0.40129
0.389595
0.377963
0.366361
0.354814
0.343266
0.331726
0.320178
0.308593
0.297054
0.285489
0.273928
0.262419
0.25092
0.239441
0.228016
0.216637
0.205318
0.194052
0.182866
0.171792
0.160792
0.149938
0.13917
0.128563
0.118128
0.107876
0.097863
0.088085
0.078586
0.069402
0.060586
0.052175
0.044254
0.036871
0.030107
0.024057
0.018781
0.014332
0.010681
0.007756
0.00546
0.00372
0.002419
0.001446
0.000702
0.000259
0.000038
y/c
0.095466
0.096877
0.098182
0.099392
0.100483
0.101453
0.102282
0.102969
0.103514
0.103912
0.10418
0.104309
0.10429
0.104142
0.103851
0.103406
0.102815
0.102074
0.101172
0.100103
0.098861
0.097447
0.095845
0.094042
0.092038
0.089806
0.087366
0.084689
0.081772
0.07861
0.075175
0.071484
0.067515
0.063284
0.058794
0.054063
0.049124
0.044037
0.038898
0.033816
0.02893
0.024359
0.020195
0.016479
0.013151
0.010168
0.007489
0.005102
0.002932
0.000944
x/c
0.99997
0.994978
0.988627
0.980723
0.971321
0.960662
0.949424
0.93803
0.926586
0.91515
0.903719
0.892291
0.88086
0.869394
0.857897
0.846387
0.83486
0.823301
0.811752
0.800182
0.788585
0.77696
0.765344
0.753726
0.742078
0.730458
0.718795
0.707153
0.69543
0.683701
0.671981
0.660268
0.648597
0.63703
0.625585
0.614212
0.60288
0.591592
0.58032
0.569025
0.557686
0.546245
0.534692
0.523068
0.511354
0.499605
0.487811
0.475993
0.464176
0.452348
y/c
-0.001169
-0.001364
-0.001544
-0.001708
-0.001989
-0.002393
-0.002872
-0.00347
-0.004214
-0.005108
-0.00616
-0.007367
-0.008731
-0.010249
-0.011903
-0.013685
-0.015593
-0.017617
-0.019742
-0.021981
-0.024315
-0.02674
-0.029234
-0.031815
-0.034463
-0.03718
-0.039964
-0.042807
-0.045719
-0.048657
-0.051627
-0.05461
-0.057558
-0.060434
-0.063191
-0.065812
-0.068293
-0.07062
-0.072781
-0.074781
-0.07662
-0.078317
-0.079904
-0.081386
-0.08278
-0.084091
-0.085322
-0.08648
-0.087561
-0.088562
x/c
0.440534
0.428714
0.416892
0.405087
0.393277
0.381483
0.369702
0.357917
0.346164
0.334386
0.322646
0.310901
0.299179
0.287455
0.275771
0.264104
0.252445
0.240824
0.229267
0.217718
0.206246
0.194835
0.183459
0.172199
0.161017
0.149918
0.138941
0.128085
0.117402
0.106888
0.096591
0.08651
0.076717
0.067244
0.058159
0.049507
0.041417
0.03397
0.027288
0.021456
0.016534
0.012477
0.009224
0.006653
0.00463
0.003065
0.001862
0.00098
0.000385
0.000046
y/c
-0.089484
-0.090318
-0.091073
-0.091739
-0.092315
-0.092805
-0.093194
-0.093494
-0.093688
-0.093785
-0.093783
-0.093673
-0.093459
-0.093136
-0.092709
-0.09216
-0.091498
-0.090722
-0.089822
-0.088783
-0.087627
-0.08631
-0.084857
-0.083251
-0.081472
-0.07952
-0.077391
-0.075067
-0.072555
-0.069816
-0.066865
-0.063669
-0.060244
-0.056572
-0.052655
-0.048508
-0.044177
-0.039704
-0.035182
-0.03074
-0.026484
-0.022518
-0.018871
-0.015533
-0.012507
-0.009755
-0.007256
-0.004968
-0.002855
-0.000911
Table A.2: DU 06-W-200 coordinates
M.Sc. thesis
M.C. Claessens
96
M.C. Claessens
Airfoil coordinates
M.Sc. thesis
Appendix B
NACA 0018 wind tunnel results
In this appendix a summary of the results of the wind tunnel tests on the NACA 0018
profile is given. First the test setup is discussed in section B.1. A more thorough
report is available, [Claessens, 2006b]. In section B.3 the results for the airfoil in clean
configuration are given. The laminar separation bubbles present in clean configuration
were eliminated using zigzag tape. The results of this configuration can be found in
section B.4. The results for the airfoil with fixed transition, with zigzag tape applied at
5% chord, are given in section B.5. Finally an overview of the measurements performed
at angles of attack to 180◦ angle of attack cna be found in section B.6. As it is a
symmetrical airfoil, this results in the data for the complete rotation of the airfoil.
B.1
Test setup
The tunnel used is the Low Turbulence Tunnel (LTT) at the Technical University
of Delft. This wind tunnel is build in 1954 and has very little turbulence. This
makes it especially suited for making research at low Reynolds numbers on the NACA
0018 profile. The turbulence level is 0.02% at low speeds and 0.20% at maximum speed.
Test section: 1.25 m high, 1.8 m wide
Maximum airspeed: 120 m/s
M.Sc. thesis
M.C. Claessens
98
NACA 0018 wind tunnel results
Figure B.1: The model placed inside the octagonal wind tunnel
B.2
Model
The NACA 0018 model used is made of solid aluminum. The full length is 2.8 m, the
wing section itself is 1.8 m wide.
NACA 0018 profile
Length: 1.8 m
Chord: 0.25 m
B.2.1
Wake rake
The small wake rake is used for measurements in the wake of the model. It consists of
50 total pressure measurement tubes and 12 static pressure measurement tubes.
Figure B.2: The wake rake with the total and static pressure tubes
M.C. Claessens
M.Sc. thesis
B.2 Model
99
For the final measurements the manometer using fluids is used, see figures B.3 and B.4.
It uses lasers to read the height of the different tubes of which each one is connected to
one tube of the wake rake. This method is highly accurate; it has an error of 0.1 Pa.
But before the measurements begin a zero run has to be made, which can also have an
error of 0.1 Pa.
Figure B.4: Manometer detail of the
total pressures in the wake
Figure B.3: The manometer
B.2.2
Balance system
The model is attached on both sides to a yoke, see figure B.5. This yoke is a new system
which allows the angle of attack to be accurately controlled and ensures improved
balance measurement results. The previous system had to be controlled manually,
the new system is computer controlled. The balance can measure the forces in three
directions and the moments around the three axis. For this experiment only the forces
in two directions and the moments around one axes are used.
Balance
Bracket
Model
Figure B.5: Schematic of the test section
M.Sc. thesis
M.C. Claessens
100
B.3
NACA 0018 wind tunnel results
Free transition
1.2
1.2
cl
NACA 0018
clean
cl
0.8
0.8
0.4
0.4
0.0
0.0
0.000
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
30
Re=150,000
Re=300,000
Re=500,000
-1.2
-1.2
cd
o
α( )
Figure B.6: Wind tunnel data for Re=150,000 - 500,000
1.2
1.2
cl
NACA 0018
clean
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
-1.2
10
20
30
Re=500,000
Re=700,000
Re=1,000,000
cd
o
α( )
Figure B.7: Wind tunnel data for Re=500,000 - 1,000,000
M.C. Claessens
M.Sc. thesis
B.4 With trip applied
B.4
101
With trip applied
1.2
1.2
NACA 0018
with trip
cl
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
20
Re=300,000
Re=500,000
Re=700,000
Re=1,000,000
-1.2
cd
30
o
α( )
Figure B.8: Wind tunnel data for Re=300,000 - 1,000,000 with trip at 80%
1.2
1.2
NACA 0018
with trip
cl
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
cd
-1.2
10
20
30
Re=300,000 (70%)
Re=300,000 (30%)
Re=700,000 (80%)
Re=700,000 (30%)
o
α( )
Figure B.9: Wind tunnel data for Re=300,000 and 700,000 with trip at 30%
M.Sc. thesis
M.C. Claessens
102
B.5
NACA 0018 wind tunnel results
Fixed transition at 5% chord
1.2
1.2
cl
NACA 0018
Re = 0.3x106
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
10
20
Meas. trip at 5%
Meas. trip at 30%
RFOIL trip at 1%
-1.2
-1.2
30
o
α( )
cd
Figure B.10: Wind tunnel data for the fixed transition NACA 0018 at Re=300,000
1.2
1.2
cl
NACA 0018
Re = 0.7x106
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.015
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
20
30
Meas. trip at 5%
Meas. trip at 30%
RFOIL trip at 1%
-1.2
cd
o
α( )
Figure B.11: Wind tunnel data for the fixed transition NACA 0018 at Re=700,000
M.C. Claessens
M.Sc. thesis
B.6 Large angles
B.6
103
Large angles
1.2
Measured data at Re=300,000
1
Synthezised data from Re=150,000 and 500,000
0.8
0.6
0.4
Cl
0.2
0
0
20
40
60
80
100
120
140
160
180
-0.2
-0.4
-0.6
-0.8
-1
-1.2
alpha
Figure B.12: NACA 0018 lift data for large angles at Re=300,000
2
1.8
1.6
Cd, balance
1.4
1.2
1
0.8
0.6
0.4
Measured data at Re=300,000
0.2
Synthezised data from Re=150,000 and 500,000
0
0
20
40
60
80
100
120
140
160
180
alpha
Figure B.13: NACA 0018 drag data for large angles at Re=300,000
M.Sc. thesis
M.C. Claessens
104
NACA 0018 wind tunnel results
1.2
Measured data at Re=500,000
1
0.8
0.6
0.4
Cl
0.2
0
0
20
40
60
80
100
120
140
160
180
-0.2
-0.4
-0.6
-0.8
-1
-1.2
alpha
Figure B.14: NACA 0018 lift data for large angles at Re=500,000
2
1.8
1.6
Cd, balance
1.4
1.2
1
0.8
0.6
0.4
0.2
Measured data at Re=500,000
0
0
20
40
60
80
100
120
140
160
180
alpha
Figure B.15: NACA 0018 drag data for large angles at Re=500,000
M.C. Claessens
M.Sc. thesis
Appendix C
DU 06-W-200 RFOIL data
In this appendix the calculated data using the RFOIL program is given. In section C.1
the RFOIL characteristics of the Du 06-W-200 are given for Re=300,000, 500,000 and
700,000. In section C.2 some detailed pressure distribution information is given for
three angles of attack; α = -5, 0 and 5.
C.1
RFOIL characteristics
1.2
cl
DU 06-W-200
clean
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.010
0.020
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
cd
10
20
Re=300,000
Re=500,000
Re=700,000
-1.2
o
α( )
Figure C.1: RFOIL data for the DU 06-W-200 with free transition
M.Sc. thesis
M.C. Claessens
106
DU 06-W-200 RFOIL data
1.2
cl
DU 06-W-200
trip
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.010
0.020
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
Re=300,000
Re=500,000
Re=700,000
-1.2
cd
20
o
α( )
Figure C.2: RFOIL data for the DU 06-W-200 with trip at 40% up and 50% low
1.2
1.2
cl
DU 06-W-200
dirty
cl
0.8
0.8
0.4
0.4
0.0
0.000
0.0
0.010
0.020
0.030
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
cd
10
20
Re=300,000
Re=500,000
Re=700,000
-1.2
o
α( )
Figure C.3: RFOIL data for the DU 06-W-200 with trip at 5%
M.C. Claessens
M.Sc. thesis
C.2 Pressure distributions
C.2
107
Pressure distributions
(a) Cp
(b) Amplification factor
(c) DS
(d) DP
(e) Cf
(f) H
Figure C.4: RFOIL data for the clean airfoil at α=0
M.Sc. thesis
M.C. Claessens
108
DU 06-W-200 RFOIL data
(a) Cp
(b) Amplification factor
(c) DS
(d) DP
(e) Cf
(f) H
Figure C.5: RFOIL data for the clean airfoil at α=5
M.C. Claessens
M.Sc. thesis
Appendix D
Du 06-W-200 wind tunnel results
In this appendix an abstract of the measured wind tunnel data is given. The test
setup is the same as used for the NACA 0018 wind tunnel tests, see section B.1. For
more information the reader is referred to measurement report [Claessens, 2006a]. In
section D.1 the results for the airfoil in clean configuration are given. The results for
the airfoil with fixed transition, with zigzag tape applied at 5% chord, are given in
section D.2. Finally an overview of the measurements performed at angles of attack
to 78◦ angle of attack can be found in section D.3.
M.Sc. thesis
M.C. Claessens
110
D.1
Du 06-W-200 wind tunnel results
Free transition
Du 06-W-200
clean
1.2
cl
1.2
cl
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
-1.2
cd
20
30
p03s
p05s
p07s
p10s
o
α( )
Figure D.1: Characteristics of the clean DU 06-W-200 profile
M.C. Claessens
M.Sc. thesis
D.2 Transition fixed at 5% chord
D.2
111
Transition fixed at 5% chord
Du 06-W-200
Dirty
c1.2
l
cl
1.2
0.8
0.8
0.4
0.4
0.0
0.0
0
0.015
0.03
0.045
-30
-20
-10
0
-0.4
-0.4
-0.8
-0.8
-1.2
10
20
30
p03d
p07d
p03s
p07s
-1.2
o
α( )
cd
Figure D.2: Characteristics of the DU 06-W-200 profile with trip at 5%
M.Sc. thesis
M.C. Claessens
112
Du 06-W-200 wind tunnel results
D.3
Large angles
1.4
Cl
Du 06-W-200
6
Re=0.3x10
1.2
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
70
80
o
α( )
Figure D.3: DU 06-W-200 lift data for large angles at Re=300,000
2
Du 06-W-200
6
Re=0.3x10
Cd 1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
70
80
o
α( )
Figure D.4: DU 06-W-200 drag data for large angles at Re=300,000
M.C. Claessens
M.Sc. thesis
D.3 Large angles
113
1.4
Cl
Du 06-W-200
6
Re=0.5x10
1.2
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
70
80
o
α( )
Figure D.5: DU 06-W-200 lift data for large angles at Re=500,000
2
Du 06-W-200
6
Re=0.5x10
Cd 1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
50
60
70
80
o
α( )
Figure D.6: DU 06-W-200 drag data for large angles at Re=500,000
M.Sc. thesis
M.C. Claessens
Fly UP