Академический Документы
Профессиональный Документы
Культура Документы
Abstract— Airfoil optimization is an important subject for wind obtained by CST method is superior to the geometry obtained by
turbines in order to increase the flow efficiency along the blade PARSEC parameterization method for the specified flow
sections. The first important subject for airfoil shape conditions.
optimization is the mathematical description of airfoil or its
parametrical form. This subject directly effects computational TABLE OF CONTENTS
cost of the optimization process and general efficiency of the
airfoil. In this study, NACA 2411 airfoil has been optimized by a 1. INTRODUCTION ....................................................... 1
genetic algorithm coupled with an airfoil analysis software. 2. AIRFOIL SHAPE PARAMETERIZATION METHODS . 2
Geometry of the airfoil is represented by two different airfoil 3. COMPARISION AND ANALYSIS ............................... 3
shape parameterization techniques namely; PARSEC method
(parametric section) and CST method (class/shape function 4. RESULTS AND CONCLUSION .................................. 5
transformation). REFERENCES ............................................................... 6
The objective of this study is to find the best airfoil BIOGRAPHY ................................................................. 7
representation scheme which consumes less computational effort
and gives the best lift to drag ratio in a large design space for the 1. INTRODUCTION
ideal aerodynamic design optimization. In order to generate
different airfoil shapes and control the genetic algorithm, Matlab In the area of aerodynamic design, one of the typical problem is
subroutines were developed in accordance with different airfoil to find the optimum shape which is efficiently suited for a
parameterization schemes mentioned above. These airfoil shapes specified range of flow conditions. Airfoil optimization has a
are used as individuals for the genetic algorithm. A Matlab script
crucial role for wind turbines in order to increase the efficiency
was embedded into the code that calls the potential flow solver
software (XFOIL) to analyze the flow around the airfoils. Fitness along the blade sections. For wind turbines “optimum shape of
function of each individual is specified as “lift to drag ratio” an airfoil” means that it minimizes some costs like drag and
obtained by the flow analysis. pitching moment effectively while satisfying the given
performance constraints so that it extracts maximum energy
The aim of the optimization process is to find the unique airfoil
from the wind.
shape which gives the maximum of the lift to drag ratios in a
certain solution space. The flow is assumed to be inviscid and
uniform for the sake of simplicity. Mach number, Reynolds Generally, numerical shape optimization methods can be
number and design lift coefficient are chosen as 0.03, 350,000 and classified into two general approaches. First approach is direct
1, respectively. Tournament selection method is used to select the design approach and the second one is inverse design
individuals which have high fitness values for the next approach [1]. In the direct design approach, a reference airfoil
generation. The genetic operators; cross-over and mutation rates geometry is to be specified first. This airfoil geometry is
are chosen as 0.45 and 0.1 respectively. The code can be executed analyzed by a flow analysis code. Then, resulting flow field
until a pre-defined number of iterations or a certain convergence
criteria is obtained. In the study, population and generation
variables which are obtained by the analysis code are
numbers are chosen as 8 and 200 respectively. Fitness increment evaluated. When the airfoil geometry meets the design
with respect to generation is plotted in order to evaluate the requirements, the process is terminated otherwise the
results. geometry of airfoil is changed by a parameterization scheme
The results for each shape function are compared in terms of and analyzed again. This iterative process continues until the
sensitivity to the optimized geometry and computational cost. design requirements are met. Direct design approach is
Design spaces for each parameterization method were balanced presented in Fig.1.
by changing the design parameters so that the control areas on
the specified curves were similar. Hence, parameterization
schemes are compared with respect to the CPU time, the number
of scheme parameters and the best fitness values achieved by the
analysis code. The results have showed that the final geometry
Figure 2. Inverse airfoil design process [2] Figure 4. Airfoil geometries generated via PARSEC
method
In this study, direct design approach is used. Iterative process
is controlled by a genetic algorithm code to be able to CST Method
minimize computational cost.
In CST parameterization method, general form of
mathematical expression that represents airfoil geometry (Eq.
2. AIRFOIL SHAPE PARAMETERIZATION METHODS 2) is defined as product of “class function” and “shape
An important process of airfoil optimization is function”.
parameterization of the airfoil geometry. One important point
୶ ୶ ୶ ୶ ୶
for the airfoil shape parameterization is the number of design Ƀ ቀ ቁ ൌ ට ቂͳ െ ቃ σே
ୀ ܣ ቂ ቃ ቂ ቃ Ƀ ் (2)
ୡ ୡ ୡ ୡ ୡ
variables. Too many design variables give more control over
the airfoil shape whereas are higher in cost in terms of CPU
time. Some of the important parameterization techniques used In Eq. 2, the formula which is given in Eq. 3 gives the unique
in literature are discrete approach, Joukowski transformation, shape of the geometry between the round nose and sharp aft
splines or polynomial approximations of curves, Hicks-Henne end of the airfoil. This formula defines a well-behaved
shape functions, PARSEC and Kulfan’s shape/class functions analytic characteristic because the term √(x/c) which is the
method (CST) [3, 4]. source of nonanalytic characteristic of the general
mathematical expression for airfoils (Eq. 2) is not included
PARSEC Method [6].
PARSEC parameterization method is used for representing ୶ ୶
upper and lower surface curves of the airfoil. PARSEC ܵ ቀ ቁ ൌ σே
ୀ ܣ ቂ ቃ (3)
ୡ ୡ
2
In Eq. 2, the formula which is given in Eq. 4 is defined as If Bernstein polynomial is used as shape function, the unit
“class function”. shape function can be decomposed into further components. A
Bernstein polynomial of order N is composed of n+1 terms.
ேଵ ୶ ୶ ேଵ ୶ ேଶ These terms has the form which is given in Eq. 5.
ܥேଶ ቀ ቁൌቂ ቃ ቂͳ െ ቃ (4)
ୡ ୡ ୡ
ܵǤ ሺݔሻ ൌ ܭǡ ݔ ሺͳ െ ݔሻି (5)
In Eq. 2, N1 and N2 values are used as 0.5 and 1 respectively.
Different N1 and N2 values mathematically defines a variety ܭǡ value in Eq. 5 is the binomial coefficients defined as in
of basic class for geometric shapes. This basic class definition the Eq. 6.
via different combinations of exponents in the class function
Ǩ
gives different basic geometrical shapes like airfoils, wedge, ܭǡ ൌ (6)
ǨሺെሻǨ
cone, rectangle, ellipse etc. [6]. These geometric shapes
obtained by class function and unit shape function are given in
Figure 5. All unit shape function components can be scaled with proper
factors to define different airfoil shapes.
Here, unit shape function can be decomposed into two or more P0-Lower Leading Edge Radius, Rle-lo 0.009 0.007
component shape functions. If it is decomposed into two, the P2-Position of Upper Crest Point, Xup 0.32 0.37
first one ܵ1=[1−x/c] means to an airfoil with a round nose and
P3-Upper Crest Point, Yup 0.077 0.08
zero boat-tail angle. The second component ܵ2=[x/c] means an
airfoil with zero nose radius and a finite boat-tail angle. If the P4-Upper Crest Curvature, YXXup -0.65 -0.63
two component unit shape functions scale with a factor, P5-Position of Lower Crest Point, Xlo 0.1747 0.1749
different airfoil shapes can be obtained [6]. The two
components of unit shape function and scaling factor KR are P6-Lower Crest Point, Ylo -0.034 -0.032
shown in Figure 6. P7-Lower Crest Curvature, YXXlo 0.60 0.62
P10-Trailing Edge Direction Angle, αTE -4.90 -4.55
P11-Trailing Edge Wedge Angle, βTE 15.0 15.10
P9-Trailing Edge Thickness (TTE) and P8-Trailing Edge Offset
(Toff) are zero.
3
the control areas on the specified curves are similar. 10 design solution space. In Figure 7, PARSEC solution space and CST
variables for PARSEC method are used to be able to generate solution space are presented.
different airfoils whereas the number of design variables for
CST method is 8. Hence, this gives approximately 2.5E27
candidate solutions for each parameterization scheme. The
numbers in the range for each design variables for two
different parameterization scheme are also given in Table 2.
4
0.45 and 0.01 respectively. Population and generation numbers The original airfoil geometry gives maximum L/D value 73.2
are chosen as 8 and 200 respectively. Fitness increment with where drag coefficient is 0.01367 and angle of attack is 6.6°.
respect to generation is plotted in order to evaluate the results. The drag coefficient of optimized PARSEC geometry is
The code is executed until convergence criteria is reached. 0.01193 at 5.7° angle of attack for the specified design lift
coefficient. Also, the drag coefficient of optimized CST
4. RESULTS AND CONCLUSION geometry is 0.01145 at 5.8° angle of attack. In terms of
maximum L/D values obtained at the specified design lift
The convergence histories for 200 generations are illustrated in coefficient, it is 83.8 for PARSEC geometry and 87.3 for CST
Figure 9. Best individuals in the generations (shown in red) are geometry. Hence, it corresponds an increase 14.5% in L/D for
generally obtained in the L/D range between 75 and 80 bounds the PARSEC geometry and an increase 19.4% for the CST
for the PARSEC method (upper graph). Mean L/D values geometry with respect to the original NACA 2411 airfoil.
(blue) for each generation are in the middle between the bounds
of 70 and 75 for the same method. For the CST method (lower Some increment and decrement values that corresponds to
graph) however, it can be seen that general trend of best given “cl”, “cd” and “cl/cd” values are shown in the Table 4
individuals in the generations are obtained in the L/D range after the optimization process.
between 80 and 85 bounds and mean L/D trend for each
generation can be seen roughly close to the upper bounds Table 4. Change in aerodynamic performance coefficients
between 70 and 75.
Orginal
NACA Optimized Optimized inc./dec. inc./dec.
2411 (PARSEC) (CST) (PARSEC) (CST)
Comp.
- 651 771 - -
Time (s)
5
REFERENCES
[1] B. A. Gardner and M. S. Selig, “Airfoil design using a
genetic algorithm and an inverse method, 41st Aerospace
Sciences Meeting and Exhibit, AIAA, Reno, Nevada, 6-9
January 2003.
6
BIOGRAPHY
Erkan Orman is a lecturer in the
Department of Airframe and
Powerplant Maintenance at Faculty
of Aeronautics and Astronautics of
the Anadolu University in Eskişehir,
Turkey. He received his Master’s
degree in the area of wind turbine
performance, from Anadolu
University in 2010. He is currently
a registered student in PhD program after Master’s
degree in the Graduate School of Science of Anadolu
University. His research interests include low Reynolds
number flows, computational fluid dynamics, turbo
machines, wind turbines, unmanned aerial vehicles with
research focuses on shape optimization, geometry
parameterization and optimization algorithms.