Renewable Energy 54 (2013) 166e172

Contents lists available at SciVerse ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Developement and verification of a performance based optimal design software

for wind turbine blades
Bumsuk Kim a, Woojune Kim a, Sanglae Lee a, Sungyoul Bae a, Youngho Lee b, *
a
Green and Industrial Technology Center, Korean Register, 23-7 Jang-dong Yuseong-gu Daejeon, Republic of Korea
b
Division of Mechanical & Energy System Engineering, Korea Maritime University, 1Dongsam-dong Youngdo-gu, Busan, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: In this research, we developed software for designing the optimum shape of multi-MW wind turbine
Received 26 January 2012 blades and analyzing the performance, and it features aerodynamic shape design, performance analysis,
Accepted 13 August 2012 pitchetorque analysis and shape optimization for wind turbine blades. In order to verify the accuracy of
Available online 7 September 2012
the performance analysis results of the software developed in this research, we chose the 5 MW blade,
designed by NREL, as the comparison model and compared with the analysis results of well known
Keywords:
commercial software (GH-Bladed). The calculated performance analysis results of GH-Bladed and our
Wind turbine blade
software were consistent in all values of CP in all l ranges. Also, to confirm applicability of the optimum
Aerodynamic design
Performance analysis
design module, the optimum design of the new 5 MW blade was performed using the initial design data
Optimal design of the comparison model and found that solidity was smaller in our design even though it produced the
same output and efficiency. Through optimization of blade design, efficiency increased by 1% while the
thrust coefficient decreased by 7.5%.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction It is required to take into account the external environmental

Most of the leading wind turbine manufacturers consider blades
as their key components of wind turbine system and have
concentrated their efforts on developing their own blade design
and increasing the supply through in-house production facility. As
world wind energy market grows continuously, a number of
independent blade manufacturers are emerged recently, especially
in China. However, wind blade market is still dominated by the
leading wind turbine system manufacturers and a giant indepen-
dent blade supplier, LM Wind-power [1], and their overall share in
world blade market is estimated around 80 percent. Even though
the total amount of blade supply has increased due to the new
blade manufacturers, only a few of them have ability to produce
large blades bigger than 3 MW class. Therefore, supply shortage of
large blades is predicted as global offshore wind-power market
expands, and blade could be a bottle neck component for large
wind turbine system. In Korea, some blade manufacturers have
succeeded in producing 2e3 MW class prototype blades by
collaborating with foreign engineering companies. So as to enter
the market successfully, securing the fundamental technology on
design and production of multi-MW class blades is indispensable.
* Corresponding author.
E-mail address: pivlab@naver.com (Y. Lee).
conditions for a wind blade design that can be occurred during the
design life time. Therefore, design evaluations by aero-elastic
simulation, CFD analysis, and FEM analysis should be conducted
interactively to secure safety and reliability of blades and to satisfy
the requirements of international standards. So as to optimize
blade shape for the purpose of enhancing aerodynamic perfor-
mance and reducing thrust force, both of aerodynamic and struc-
tural designs should be considered. From the perspective of
aerodynamic design, thrust force, power performance, aero-
dynamic efficiency, and AEP (Annual Energy Production) are the
important, and from the perspective of structural design, composite
material lay-up, mass, stiffness, buckling stability, and ultimate and
fatigue loads are concerned. As of now, software that is capable of
creating optimum aerodynamic design and analysis for Multi-MW
class wind blades at a time has not been developed. Excel-based
BOT (Blade Optimization Tool) developed by ECN in Netherlands
is the only design tool that is comparable to such purpose [2].
Because BOT is not capable of designing initial shapes from a blank
state and can only optimize aerodynamic shapes of existing blades,
it cannot be considered as a complete aerodynamic design tool.
Therefore, the goal of this research is to develop integrated user-
friendly software which can make blade initial shape, power and
load analysis, aerodynamic optimum design, and pitch and torque
control strategy.

0960-1481/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.renene.2012.08.029
 B. Kim et al. / Renewable Energy 54 (2013) 166e172 167

Nomenclature T thrust force

am local axial flow induction factor
Cl,tipfoil lift coefficient of a tip airfoil a0m local tangential flow induction factor
CD drag coefficient aT critical axial induction factor
CL lift coefficient cm local chord length
CP,expected expected aerodynamic power coefficient fm local tip loss factor
CT thrust coefficient atipfoil angle of attack at maximum lift to drag ratio of tip
CT1 empirical thrust coefficient when a ¼ 1 airfoil
Dhub hub diameter h mechanical efficiency of drive train
Drotor rotor diameter qlimit the upper limit of twist angle
Gratio gear ratio qm local twist angle
Irange,max the upper limit of inverter range ldesign design tip speed ratio
Irange,min the lower limit of inverter range ɸm local inflow angle
N number of blade r air density
Prated rated power Ublade,min the lowest blade rotational speed
Q torque Ublade,max the upper limit of blade rotational speed
Rblade radius Ublade,rated rated blade rotating speed
Rem local Reynolds number

2. Developement of an optimum BLADE design tool Urated;gen  Irange;min

Ublade;min ¼ (4)
Gratio
2.1. Blade Design Module(BDM)
Urated;gen  Irange;max
At the baseline design stage, the design parameters presented in Ublade;min ¼ (5)
Gratio
Table 1 such as desired power output, rated wind speed, assumed
efficiency, maximum twist angle limitation, tip speed restriction, After confirming initial design parameters, baseline blade shape
inverter range, rated speed of generator, and gear ratio need to be is generated by calculating chord length and twist angle at every
set. Next, based on input parameters from Table 1, diameter of the calculation point set by the designer. These points can be defined
blade Drotor, rated speed Ublade,rated, design tip speed ratio ldesign, either on equal intervals or on unequal intervals as in Fig. 1. The
min/max values of rotor speed for producing power Ublade,min and calculation is performed according to Eqs. (6)e(10), and the
Ublade,max can be calculated according to Eqs. (1)e(5), but desig- detailed process is like below [3].
nating a specific diameter is also possible if blade length is pre-
determined. 1) Calculate tip loss factor, fm,tip
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1 1
0 2
8Prated  ðldesign mÞ
Drotor ¼ (1) B @ N2 ð1mÞ=m 1 þ ð1aÞ2 A C
hCP;estimated rpVrated 3
B C
2 B C
ftip;m ¼ cos1 Be C (6)
p B
@
C
A
 
Vtip;rated
Ublade;rated ¼ (2)
Drotor orDuser =2
2) Calculate axial flow induction factor, am
!
ðDrotor orDuser Þ=2  Ublade;rated
ldesign ¼ (3) 1 1 1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vdesign am ¼ þ fm  1  fm þ fm2 (7)
3 3 3
3) Check final convergence value after iterative calculations for
Table 1
fm,tip and am
Initial design parameters. 4) Calculate tangential flow induction factor, a0m

Parameter Unit Parameter Unit !

Shape parameters Operation parameters am
am 1 
N e Vtip,rated m/s ftip;m
Prated kW R m a0m ¼ (8)
Vdesign m/s Urated,gen rpm l2design m2
Vrated m/s Irange,min %
CP,estimated e Irange,max %
h e Gratio e 5) Calculate chord length, cm
atipfoil Calculated parameters
CL,tipfoil e R m
r kg/m3 Ublade,rated rpm
2
Dhub m ldesign e 2p 4ldesign m2 a0m R
Ublade,min cm ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2ffi (9)
Nldesign CL;tipfoil  2  
Drotor m rpm
n m2/s Ublade,max rpm 1  am þ ldesign m 1 þ a0m
qlim 
168 B. Kim et al. / Renewable Energy 54 (2013) 166e172
Fig. 1. Unequally spaced calculation section arrangement.
6) Calculate twist angle, qm
qm ¼ fm  atipfoil
After this procedure, chord length and twist angle distribution
along blade spanwise direction can be obtained, and it is easily
found that the values of the chord length cm are unrealistically high
in the inboard region. In the aerodynamic aspect, the root region
does not contribute to aerodynamic torque generation at all due to
its circular shape, therefore, the chord length around the blade root
must be reduced to practical ranges for the manufacturing and
economical purpose. However, blade root is exposed to high
bending moments which could cause aero-structural problem.
Thus, the result of additional structural analysis must be considered
to determine proper chord length around blade root to secure
structural stability.
Most of blade torque is generated at the spanwise points within
65%e95% ranges from the root, thus chord lengths in those regions
should be maintained during chord linearizing process. Generally,
five to seven types of airfoils are used in blade design, and the
designer can create additional elements between existing
airfoils(Table 2). The chord lengths and twist angles for the
newly-created elements are updated by interpolating nearby
values. Later in the performance analysis stage, aerodynamic
properties (CL, CD) of the airfoils are required for all elements.
Securing aerodynamic properties that reflect the exact Reynolds
number at each calculation elements is important to predict power
performance and thrust force precisely. Therefore, the software
calculates Reynolds numbers at all elements by using the Eqs. (11)e
(13) and the designer can check them through the GUI. The final
shape design variables cm and qm can be plotted shown in Fig. 2, and
can also be exported to Excel files. Finally, all numerical data and
graphs obtained from Blade Design Module can be easily verified
automatically through the report-printing function.
Vrot ¼ r  Ublade;rated
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vlocal ¼ 2 þ V2
Vrot design
 
cm  Vlocal
Rem ¼
n
Table 2
Arrangement of 2-D airfoils.
Section no. Position [m]
1 0.0272
2 0.0928
3 0.1574
4 0.2326
5 0.3177
6 0.4153
7 0.5305
8 0.6585
9 1.0000
2.2. Performance analysis module(PAM)
(10) Performance analysis is needed in order to find out whether the
aerodynamic design of blade obtained from BDM (Blade Design
Module) meets the design objectives. Generally, the analysis is
performed for torque, power output, efficiency, and thrust force
using the BEM (Blade Element Momentum) theory. The results of
the analysis, which depend on how accurately the flow induction
factor is calculated, are obtained by the following iterative calcu-
lation process. Flow induction factors am and a0m are assumed to be
zero as initial guess at step 1, and calculated results after step 2e6
are checked whether they are converged or not. If so, it proceeds to
step 8, and otherwise it goes back to step 2 and flow induction
factors are calculated again with updated initial values. Perfor-
mance analysis should be conducted independently at all calcula-
tion points in blade spanwise direction that were determined in the
blade design stage. Local torque and thrust force at each element
are calculated by using local aerodynamic properties and the
overall torque and thrust force can be yielded by integrating local
values along blade spanwise direction. Hence, securing reliable
aerodynamic data at each calculation point is indispensable to
improve the accuracy of performance analysis.
If the rotor is assumed to a simple actuator disk as in Fig. 3, the
force due to the pressure difference is equal to the force generated
by change of momentum across the disk. Then wake velocity Uwake
can be induced as (12a)UN. However, if the axial flow induction
factor of the blade is predicted to be over 0.5, Uwake obtains a value
of 0 or less at step 8, and then the momentum theory is not valid
any more. In that condition, the calculated value of the thrust
coefficient significantly differs from the experimental results as
(11)
(12)
(13)
Airfoil
Cylinder
Cylinder
DU-401
DU-350
DU-300
DU-250
DU-212
NACA64618
NACA64615
Fig. 2. Example of calculated chord length and twist angle.
 B. Kim et al. / Renewable Energy 54 (2013) 166e172 169

6) Calculate the flow induction factors (Eqs. (19)e(21))

g1
am ¼ (19)
1 þ g1

g2
a0m ¼ (20)
1  g2

 
Ncm CL cosfm þ CD sinfm
g1 ¼ H (21)
2pr 4floss;m sin2 fm

 
Ncm CL sinfm  CD cosfm
g2 ¼
Fig. 3. Flow model of momentum theory. 2pr 4floss;m sinfm cosfm

for a  aT, H ¼ 1.0;

shown in Fig. 4 and it is called ‘Momentum Theory Breakdown’. To
compensate it, the value of am must be corrected by empirical
correlation such as Eq. (22) if the calculated axial flow induction 4að1  aÞ
for a>aT ; H ¼ pffiffiffiffiffiffiffiffi 
factor at a local element exceeds a specific value, am  aT [4,5]. CT1  4 CT1  1 ð1  aÞ

1) Initialize am, a0m ¼ 0

2) Compute the inflow angle (Eq. (14)) 7) Check the convergences of flow induction factor. If the changes
exceed certain tolerance level, go to step 2.
1  am 8) Update flow induction factors considering the effects of
tanfm ¼   (14) momentum theory breakdown (Eq. (22))
lm 1 þ a0m
1 pffiffiffiffiffiffiffiffi
aT ¼ 1  CT1 (22)
3) Compute the angle of the attack (Eq. (15)) 2
 
am ¼ fm  b þ qset þ qm (15) 9) Compute local and overall thrust force (Eq. (23))

1  
4) Compute the tip and hub loss coefficients (Eqs. (16) and (17)) dT ¼ rNW 2 CL cosfm þ CD sinfm cm dr (23)
2
 N ðRrÞ 
2 
10) Compute local and overall torque (Eq. (24))
ftip;m ¼ acos e 2 rjsinfm j (16)
p
1  
 NðrRhub Þ  dQ ¼ rNW 2 r CL sinfm  CD cosfm cm dr (24)
2  2
fhub;m ¼ acos e 2 rjsinfm j (17)
p
11) Compute overall power output (Eq. (25))
5) Compute the solidity (Eq. (18))
dPaero ¼ Ublade;rated  dQ (25)
Ncm
sm ¼ (18)
2pr Thorough the above procedure, all values can be plotted and
saved as an Excel file. Also, all numeric data and graphs that are
generated at performance analysis stage can easily be verified
automatically through report-printing function.

2.3. Pitch and torque control module(PTM)

Torque control map provides the information to determine rotor

Fig. 4. Example of momentum theory breakdown.
speed of maximum efficiency for a specific wind speed at below
rated condition. The calculation points for torque control map are
determined by a combination of the blade rotating speed and the
wind speed. The number of the calculation points for rotating
speed is determined by DU, which divides the initial Ublade,mine
Ublade,max in Table 1, while the number of calculation point for wind
speed is depending on DV, which divides the interval Vcut-ine
Vcut-out. Finally, all calculation points needed for calculating the
torque control map are shown in Fig. 5. When the calculation point
170 B. Kim et al. / Renewable Energy 54 (2013) 166e172
Fig. 5. Definition of calculation points.
is determined after combining the wind speed and the rotating
speed, values for the torque, power output, efficiency, and the
thrust force along actual operating line can be calculated according
to PAM in Section 2.2, and the torque control map as shown in Fig. 6
is obtained.
At above the rated wind speed range, the optimum pitch angle
of the blade to regulate power production is calculated by
a combination of the changes in the wind speed and the pitch angle
while rotating speed is kept constant. After the wind speed range of
Velec, ratedeVcut-out is divided by, DV, the blade pitch angle for each
given wind speed that produces rated power output by changing
the blade pitch angle from 0 to 90 in 1 increments. Subsequently,
PAM can calculate all aerodynamic properties including torque,
power output, efficiency, thrust force, and the pitch angle at all
wind speeds. During the analysis of PAM, the change in the pitch
angle of the blade can be realized by changing the qm in Eq. (15).
The blade designer must consider the inhibition of unnecessary
instantaneous peak thrust force that occurs in a narrow wind speed
interval near the rated wind speed and it can be reduced by
controlling pitch angle of the blade [6]. This manipulation is called
thrust clipping or peak shaving and the software developed in this
research can limit the maximum thrust force as desired level in
Fig. 6. Example of torque control map.
user-friendly environment as in Fig. 7. After thrust clipping is
applied, performance analysis must be carried out again to check
the effects of thrust force restriction and to obtain new pitch
control map.
2.4. Blade optimization module(BOM)
The initial blade shape that had completed by BDMePAMe
PTM, the aerodynamic optimum design process can be conduct-
ed in order to maximize power and efficiency. Several restricting
options such as maximum chord length and twist angle limitation,
and axial thrust force ratio comparing to initial blade can be
chosen by designer. Research on optimization of the aerodynamic
design of the blade is usually done by BEM, which uses the flow
induction factor as suggested by Glauert [7]. To optimize aero-
dynamic shape at a designated TSR, the objective function is the
local torque at each calculation points, and chord length and twist
angle are defined as independent variables. The maximum overall
aerodynamic efficiency can be obtained by collecting the best
combination of chord length and twist angle which shows the
highest local torque at each calculation points. An increment of
chord length and twist angle for optimizing process can be
adjusted by designer, and the smaller increment could lead better
results.
Maalawi optimized the 100 kW class blade rotors with code
length and twist angle variation and Lanzafame predicted the
performance of the NREL Phase VI based on the results of the
experiment on S809 airfoil [8,9]. However, the blade models used
for optimization in the two researches were applied a single airfoil,
which is different from the contemporary large blades, which
applies five to seven types of airfoil in order.
The BEM theory calculates the overall performance by inte-
grating local torques along spanwise direction of the blade. During
this process, mass and momentum interactions in blade spanwise
direction among the elements are ignored. Therefore, it can be
thought that the power performance can be maximized when
every local torque has its maximum value. Thus, the final goal of
blade aerodynamic shape optimization is to find optimal combi-
nations of chord lengths and twist angles that secure the maximum
performance at every local element. Minimum and maximum
values for search range of the chord length and the twist angle are
defined by the designer, as shown in Fig. 8.
Optimum aerodynamic design is carried out at a certain TSR
(tip speed ratio) and the calculating range and cases should be
Fig. 7. Definition of thrust clipping rate.
 B. Kim et al. / Renewable Energy 54 (2013) 166e172 171

Fig. 9. Aerodynamic coefficients of the airfoils.

3. Optimal design and verification of the calculated results

3.1. Blade model for verification and validation

In order to assess the accuracy of the performance analysis

Fig. 8. Examples on the final optimum blade geometry.
chosen properly to reduce calculation time. Additionally, the
designer can choose the objectives of optimization such as securing
high performance, minimizing thrust force, and limiting the range
for chord length and twist angles by comparing with the initial
blade models.
Since various types of airfoils are applied to large blades and
optimizing calculation is conducted respectively for local elements,
the distributions of chord lengths and the twist angles do not show
continuous tendency after the initial optimization, especially for
airfoil transition region [10]. To solve this problem, an option to
limit the change rate of the chord length and the twist angle of the
neighboring elements is available and more than five orders of
curve fitting for chord length and twist angle distribution is
indispensable to make blade shape smooth globally.
module of the software developed in this research, we chose the
5 MW blade, developed by NREL, as the reference model and
compared the performance analysis results to those of GH-Bladed,
conventional system load analysis software. The specification for
reference 5 MW model is listed in Table 3 [11]. Because the results
of the performance analysis performed by NREL were obtained
through aero-elastic analysis after comprehensive modeling of the
5 MW wind turbine system, they cannot be compared directly to
results from performance analysis module of this software, which
only analyzes blades. Hence, in order to obtain static power curve of
the blade for comparing the results under same conditions, shape
information of 5 MW blade and aerodynamic data were used as
proposed by NREL in Table 3 and Fig. 9. With this information,
aerodynamic performance analysis under normal conditions in GH-
Bladed is carried out. Also, in order to verify the optimum shape
design module of the software developed in this research, the
specifications for the reference model Drotor ¼ 126 m, l ¼ 7.8,

Table 3
Aerodynamic properties of verification model.

Section Position [m] Chord [m] Twist [ ] Airfoil

1 2.8667 3.542 13.308 Cylinder.1
2 5.6000 3.854 13.308 Cylinder.1
3 8.3333 4.167 13.308 Cylinder.2
4 11.7500 4.557 13.308 DU40_A17
5 15.8500 4.652 11.48 DU35_A17
6 19.9500 4.458 10.162 DU35_A17
7 24.0500 4.249 9.011 DU30_A17
8 28.1500 4.007 7.795 DU25_A17
9 32.2500 3.748 6.544 DU21_A17
10 36.3500 3.502 5.361 DU21_A17
11 40.4500 3.256 4.188 NACA64A17
12 44.5500 3.010 3.125 NACA64A17
13 48.6500 2.764 2.319 NACA64A17
14 52.7500 2.518 1.526 NACA64A17
15 56.1667 2.313 0.863 NACA64A17
16 58.9000 2.086 0.370 NACA64A17
17 61.6333 1.419 0.106 NACA64A17
Fig. 10. Comparison of aerodynamic power coefficients.
172 B. Kim et al. / Renewable Energy 54 (2013) 166e172
Fig. 11. Comparison of chord and twist angle distribution.
Vrated ¼ 11.4 m/s, Gratio ¼ 1:97 were introduced as input parameters
for a new 5 MW blade design, which were used to compare the
distribution of chord lengths and the twist angles with the refer-
ence model.
3.2. Comparison of design and analysis results
The performance analysis results obtained from the software in
this research are shown in Fig. 10. The results were identical to the
results predicted by GH-Bladed and was shown to be CP,max ¼ 0.49 at
l ¼ 7.8. Fig. 11 shows the results of comparison between the chord
length change in the reference model and the newly-designed
5 MW blade, for the purpose of verifying the validity of design
application of BDM and BOM. Even though it has the same diameter
with the reference model, it generated 5 MW of power at same rated
wind speed and achieved the maximum CP of 0.49 at l ¼ 7.8 even
though it has smaller chord length distribution globally. Also, even
though the new model can produce the same power with same
efficiency like the reference model, it has lower solidity and is thus
considered to be a better model in terms of mass and aerodynamic
loads. Table 4 lists the final results of the performance analysis of
Table 4
Performance analysis results.
CP CT CP,max
l ¼ 7.8 l ¼ 7.8 e
Baseline 0.4798 0.8635 0.4822 at l ¼ 7.5
Optimized 0.4900 0.7976 0.4904 at l ¼ 7.95
Smoothed 0.4889 0.7989 0.4903 at l ¼ 8
5 MW blade. Baseline efficiency of the blade increased approxi-
mately by 1%, while the thrust coefficient decreased by about 7.5%.
4. Conclusion
In this research, a unique software for designing optimum shape
and analyzing aerodynamic performance of multi-MW wind
turbine blades has been developed, and the verification of its
performance analysis module and the aerodynamic shape design
module is conducted. The performance analysis results between
GH-Bladed and our software were identical in the entire region. The
maximum CP of 0.49 is obtained at l ¼ 7.8. To check the validity of
the blade optimizing module, the optimum design of the new
5 MW blade was conducted using the same specification for
reference model as input parameters. It is found that a model with
smaller solidity than reference model can be designed and it
produced the same output and efficiency. Through optimization of
blade design, efficiency increased by 1% while the thrust coefficient
decreased by 7.5%. If the optimum aerodynamic design method is
applied to a shape of well-designed blade, significant improve-
ments in power and efficiency cannot be expected. However, it can
achieve outstanding results if it is used when design conditions
have changed or used to decrease the thrust force and mass of blade
under the same conditions.
References
[1] BTM Consult. International wind energy developmentesupply chain assess-
ment 2010e2013. BTM Consult ApS; Jan. 2010.
[2] Bot ETG, Ceyhan O. Blade optimization tool user manual; 2009. ECN-E-09-092.
[3] Kim BS, Kim WJ, Bae SY, Han SG, and Kim ME. “A study on the aerodynamic
design of a 3MW wind turbine blade: part I”, Proc. KWEA 2009; Nov. 2009.
[4] Hansen Martin, OL. Aerodynamics of wind turbines. 2nd ed. London: Earth-
scan; 2008.
[5] Tony B, David S, Nick J, Ervin B. Wind energy handbook. WILEY; 2006.
[6] Kooijman HJT, Lindenberg C, Winkelaar D, van der Hooft EL. Aero-elastic
modeling of the DOWEC 6 MW pre-design in PHATAS; 2003. DOWEC-F1W2-
HJK-01-046/9.
[7] Glauert H. The elements of aerofoil theory and airscrew theory. Cambridge
University Press; 1959.
[8] Maalawi K, Badawy M. A direct method for evaluating performance of hori-
zontal axis wind turbines. Renewable and Sustainable Energy Reviews 2001:
175e95.
[9] Lanzafame R, Messina M. Fluid dynamics wind turbine design: critical anal-
ysis, optimization and application of BEM theory. Renewable Energy 2007;32:
2291e305.
[10] Kim BS, Kim WJ, Bae SY, Han SG and Kim ME. “A study on the aerodynamic
design of a 3MW wind turbine blade: part III”, Proc. KWEA 2009; Nov. 2009.
[11] Jonkman J, Butterfield S, Musial W, Scott G. Definition of a 5-MW reference
wind turbine for offshore system development; 2009. NREL/TP-500-38060.