Aerospace Science and Technology 78 (2018) 241–247

Contents lists available at ScienceDirect

Aerospace Science and Technology

www.elsevier.com/locate/aescte

Short communication

An improved geometric parameter airfoil parameterization method

Lu Xiaoqiang a , Huang Jun a , Song Lei a,∗ , Li Jing b
a
School of Aeronautic Science and Engineering, Beihang University, XueYuan Road No. 37, HaiDian District, Beijing, 100191, PR China
b
Research and Development Center, China Academy of Launch Vehicle Technology, NanDaHong Men Road No. 1, DongGaoDi, FengTai District, Beijing, 100076,
PR China

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

Article history: In the process of airfoil optimization, it is required to represent an airfoil with parameters, and the goal is
Received 22 March 2017 to represent arbitrary airfoils with less parameters. In this paper, a new airfoil parameterization method
Received in revised form 3 February 2018 is proposed, called the IGP method, which realized camber-thickness decoupling so that camber and
Accepted 16 April 2018
thickness could be constructed respectively with fewer parameters compared to the previous methods.
Available online 20 April 2018
Also the IGP method is featured with clear physical meaning and consecution of parameter domain. The
Keywords: mathematical model is introduced. With this camber-thickness decoupling method, the definition and the
Airfoil parameterization method domain of the control parameters was determined. To validate the feasibility, the most used airfoils were
IGP method fitted and reconstructed by this method. Then according to the results of geometric and aerodynamic
Camber-thickness decoupling comparative analysis between original airfoils and fitted airfoils, the precision of the IGP method could
Physical meaning meet the requirement of airfoil optimization.
Feasibility validation © 2018 The Authors. Published by Elsevier Masson SAS. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction of the conceptual design phase, constructive method clearly has a

greater advantage.
In the past, there are many classic constructive methods during
In the design process of an aircraft, aerodynamic optimization
the airfoil construction. Among them, The PARSEC method uses 11
is throughout the conceptual design and the detailed design. Airfoil
physical parameters to describe the airfoil [4]; The orthogonal ba-
parameterization methods, namely expressing an airfoil by several
sis function method (OBF method)uses orthogonal polynomial to
parameters, is fundamental for aerodynamic optimization. The rea-
describe the upper and lower surfaces of the airfoil, and the air-
sons are twofold: on one hand, airfoil parameterization methods
determine whether the design space (the search range of opti- foil shape is determined by the five coefficients of the upper and
mal design) could cover the alternative airfoil library; on the other lower surfaces of the airfoil [5]; Class-Shape function Transforma-
hand, airfoil parameterization methods also have an important in- tion method (CST method) is defined by Bernstein polynomials and
fluence on the nonlinearity and continuity of the optimization generally uses 11 component shape parameters to determine the
problem in mathematics aspect. shape of the airfoil [6–9].
Airfoil parameterization methods can be categorized as either There are three issues to be aware of in the optimization pro-
constructive or deformative: deformative methods take an existing cess of the airfoil by using the constructive method.
airfoil then deform it to create the new shape; constructive meth-
ods represent an airfoil shape based purely on a series of param- 1) In the optimization process, the amount of computation in-
eters specified [1]. For a particular shape of the airfoil, the defor- creases exponentially, due to the growth of the number of
mative method could obtain more precise fitting effect compared variables. Under the premise that the design space could cover
with the constructive method [2,3]. However, when the alterna- the alternative airfoil library, the less the number of variables,
tive airfoil library is large, the constructive method can use fewer the higher the computational efficiency of the optimization
control parameters to describe more airfoils. As an airfoil parame- process.
terization method applied in an initial aircraft shape optimization 2) In the optimization process, the continuity of the design space
should be ensured. For a curve defined by the polynomial
function, the degenerate state may appear at specific parame-
ter combinations. In this case, the degenerate state means that
*
Corresponding author.
the curve generated by the function cannot be used as an air-
E-mail addresses: [email protected] (X. Lu), [email protected] (J. Huang),
[email protected] (L. Song), [email protected] (J. Li). foil.

https://doi.org/10.1016/j.ast.2018.04.025
1270-9638/© 2018 The Authors. Published by Elsevier Masson SAS. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
242 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247

Nomenclature

b XC camber line curvature on the location of maximum x abscissa (chord location)

camber xC camber line abscissa
C maximum camber xl lower surface abscissa
c1 , c2 coefficients of camber-line-abscissa parameter equa- xu upper surface abscissa
tion yC camber line ordinate
c3 , c4 coefficients of camber-line-ordinate parameter equa- yl lower surface ordinate
tion
yu upper surface ordinate
cov covariance
y ori original airfoil ordinate
k control parameter of camber-line parameter equations
y f it fitted airfoil ordinate
kC k value on the location of maximum camber
P a new reference value for plotting (instead of R 2 )
αT E angle between camber line and chord line on trailing
R2 fitting correlation coefficient edge
T maximum thickness βT E trailing edge boat-tail angle
t thickness βT E relative quantity of β T E
t 1 , ..., t 5 coefficients of thickness equation ρ0 leading edge radius
XC chordwise location of maximum camber ρ0 relative quantity of ρ0
XT chordwise location of maximum thickness σ variance

3) During the computation on the basis of thin airfoil theory, the problem and speed up the optimization process. Based on that, the
camber of airfoil is the only one to be considered. If the appro- IGP method reduces the number of control parameters in order to
priate airfoil parameterization method is applied to generate further increase the computational efficiency in the process of op-
the camber and the thickness distribution functions of the air- timization.
foil respectively, only the camber is needed to be optimized
in the design process, which could reduce the computational 2.1. Parameterization expression of airfoil curves
complexity and speed up the method optimization process.
In order to consider the camber and the thickness separately, it
Therefore, an improved geometric parameter airfoil parameteri- is necessary to determine the basis functions of both the camber
zation method (the IGP method) is presented. The IGP method, as and the thickness.
a constructive method, requires no need for the basic airfoil. In the To avoid the appearance of the airfoil degenerate state, based
IGP method, the camber is expressed based on the Bézier polyno- on the fitting study of airfoil by various basis functions, the Bézier
mial, and the thickness is expressed by the polynomial basis func-
curve is selected to describe the camber line.
tion. Besides the decoupling of the camber and the thickness, the

IGP method is also featured with clear physical meaning and fewer xC = 3c 1 k(1 − k)2 + 3c 2 (1 − k)k2 + k3
control parameters compared with other methods. In addition, the (1)
y C = 3c 3 k(1 − k)2 + 3c 4 (1 − k)k2
control parameters of the IGP method could also be directly related
to the corresponding airfoil shape parameters which are commonly Among Eqn. (1), c 1 , c 2 are the horizontal coordinates of the two
used in the general aerodynamic theory. control points of the cubic Bézier curves, and c 3 , c 4 are the vertical
In this paper, the part of method establishment, as the be- coordinates of the two control points of the cubic Bézier curves.
ginning part, defined the curve function parameters, geometric k is an independent parameter, whose range is [0, 1].
parameters, control parameters and the relations between them. Then, enlightened from the basis function of the thickness
Then by geometry fitting validation and aerodynamic validation of curve NACA “four-digit” airfoil series, the thickness expression is
the 2199 airfoils in the airfoil library, the domain of the 8 con- determined.
trol parameters was determined and the continuity of the domain
above were validated to ensure the feasibility of the IGP method. t = t 1 x0.5 + t 2 x + t 3 x2 + t 4 x3 + t 5 x4 (2)
In the end, the fitting of some typical airfoil was analyzed, and the
applicable scope of the method was discussed. Based on Eqn. (1) and Eqn. (2), the airfoil expression is deter-
mined as Eqn. (3) and Eqn. (4) below.
2. Method establishment The upper surface of an airfoil:

xu = xc
In the conceptual design phase, during aerodynamic analysis (3)
based on the potential flow theory, it is possible to use the thin y u = y C + 12 t (xc )
airfoil theory to simplify the calculation. The thin airfoil theory as- The lower surface of an airfoil:
sumes that for the ideal incompressible flow of the airfoil, if the 
angle of attack, thickness and camber are small, then the effect xl = xc
(4)
of the three can be considered separately. The lift characteristic yl = y C − 12 t (xc )
of small-thickness airfoil is determined by its camber, rather than
its thickness [10]. Under the premise above, the IGP method, by In summary, according to Eqns. (1)–(4), 9 curve function pa-
decoupling the camber and the thickness, could split the aerody- rameters are needed to describe and construct an airfoil. As for
namic optimization problem into two independent problem: the the standard airfoil considered in this paper, trailing edge thick-
camber optimization and the thickness optimization. Even if the ness is 0, then
number of control parameters did not change, the IGP method
could also help reduce the design space, simplify the optimization t (1) = 0 (5)
 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247 243

In the aspect of thickness

1) Maximum thickness equals T .

t( X T ) = T (12)
2) Chord location of maximum thickness equals X T .

Fig. 1. Illustration of geometric parameters. t( X T ) = 0 (13)

3) Trailing edge boat-tail angle equals β T E .
Namely, link Eqn. (2) and Eqn. (6)
t  (1) βT E
t1 + t2 + t3 + t4 + t5 = 0 (6) − = tan (14)
2 2
Due to Eqn. (6), only 4 parameters are required while express- 4) Leading edge radius equals ρ0 .
ing the thickness. Then only 8 parameters are required to express  
 t  
the entire airfoil.  = 1 (15)
 3  ρ
(1 + t  2 ) 2 0
2.2. Relationship between geometric parameters and airfoil expressions
5) Trailing edge thickness equals 0.
According to the formula derivation above, by plugging the
8 curve function parameters into Eqns. (1)–(4), the independent t (1) = 0 (16)
camber line, thickness line, upper and lower surface of an airfoil Finally, after the expansion of Eqns. (7)–(16), the equations ob-
could be derived. However, there is no obvious physical meaning tained are as follows:
with the 8 curve function parameter, and it is hard to do the qual- Camber backstepping equation set:
itative aerodynamic analysis and define the domain. Therefore, 8 ⎧
geometric parameters below were defined. The relations between ⎪
⎪ 3c (3k2 − 4kc + 1) + 3c 4 (−3kc2 + 2kc ) = 0
⎪ 3 c
⎪
curve function parameters and geometric parameters were derived ⎪
⎪
⎪ 3c 3kc (1 − kc ) + 3c 4 (1 − kc )kc = C
2 2
in order to prepare for the subsequently proposing the control pa- ⎨
rameters and the domain of them. 3c 1 kc (1 − kc )2 + 3c 2 (1 − kc )kc2 + kc3 = X C (17)
⎪
⎪
IGP method is a constructive method, which 8 geometric pa- ⎪ c4
⎪ 1−c2 = tan αT E
⎪
⎪
rameters are used to describe an airfoil, namely: maximum camber ⎪
⎩
(C ), chordwise location of maximum camber ( X C ), angle between
| 6c3 (3kc −2)+6c4 (−3kc +2) 2 2 | = b X C
(6c 1 (3kc −2)+6c 2 (−3kc +2)+3kc )
camber line and chord line on trailing edge (α T E ), camber line cur-
Thickness backstepping equation set:
vature on the location of maximum camber (b X C ), maximum thick-
⎧
ness (T ), chordwise location of maximum thickness ( X T ), trailing ⎪ t 1 X T0.5 + t 2 X T + t 3 X T2 + t 4 X T3 + t 5 X T4 = T
⎪
⎪
edge boat-tail angle (β T E ) and leading edge radius (ρ0 ). The geo- ⎪
⎪ − 0 .5
⎨ 0.5t 1 X T + t 2 + 2t 3 X T + 3t 4 X T + 4t 5 X T = 0
2 3
metric parameters are illustrated in Fig. 1. ⎪
β
In the case that the geometric parameters of the airfoil are 0.25t 1 + 0.5t 2 + t 3 + 1.5t 4 + 2t 5 = − tan 2T E (18)
⎪
⎪ √
known, to derive the specific expression of the airfoil, eight geo- ⎪
⎪ t = 2ρ0
⎪
⎪ 1
metric parameters are used as constraints to solve the eight pa- ⎩
rameters. Since the number of constraints is the same as that of t1 + t2 + t3 + t4 + t5 = 0
unknowns, the equation set has a unique solution. Therefore, two equation sets were derived to build the relations
In the aspect of camber between curve function parameters and geometric parameters. The
1) Independent parameter k at maximum camber equals k C . two groups of parameters were used to introduce the definition
Because the camber equation is a parameter equation, the in- and the domain of the control parameters in the next part.
dependent parameter kc at the maximum curvature is introduced.
 2.3. Definition and domain of control parameters
∂ y C 
=0 (7)
∂ k k=kC After the fitting and reconstruction of 2199 airfoils (more
hereof later) in the airfoil library, each airfoil could be represented
2) Maximum camber equals C . by several parameters. The airfoil library could be represented by
a parameter space, in which each parameter has its own domain.
y C (kc ) = C (8)
Then it was found that the domains of both curve function param-
3) Chord location of maximum camber equals X C . eters and geometric parameters were not continuous. In the actual
construction of the airfoil, the discontinuous domain means that
xC (kc ) = X C (9) a point where “the shape of corresponding curve is too strange”
4) Angle between camber line and chord line on trailing edge (strange point) exists in the domain. The discontinuous domain
equals α T E . would affect the drawing, aerodynamic calculation and other steps
in the subsequent optimization process, and might interrupt the
y C (1) optimization process. For example, if ρ0 was much fewer than T ,
− = tan αT E (10)
xC (1) the leading edge of the airfoil would be gourd-shape. Based on
that, it could be surmised that the ratio of ρ0 to T might change
5) Camber line curvature on the location of maximum camber
in a small range. Therefore, a new group of parameters is need to
equals b X C .
be introduced. After screening and mathematical transformation of
  
 y C (kc )  the curve function parameters and geometric parameters, the con-
 
 x (k )2  = b X C (11)
trol parameters were derived, and the “strange point” was put at
C c
244 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247

the edge of the domain. Ultimately, the control parameter domains

of the commonly used airfoil was derived, as Eq. (19) and Eq. (20).
⎧
⎪
⎪ c ∈ [0.010, 0.960]
⎨ 1
c 2 ∈ [0.020, 0.970]
(19)
⎪
⎪ c 3 ∈ [−0.074, 0.247]
⎩
c 4 ∈ [−0.102, 0.206]
⎧
⎪
⎪ X T ∈ [0.2002, 0.4813]
⎨
T ∈ [0.0246, 0.3227]
(20)
⎪ ρ0 ∈ [0.1750, 1.4944]
⎪
⎩
βT E ∈ [0.1452, 4.8724]
Among them, ρ0 and βT E are the dimensionless quantities of
ρ0 and βT E
ρ0
ρ0 = (21)
( XTT )2
βT E
βT E = (22)
arctan 1−TX Fig. 2. Histogram of fitting results.
T

Therefore, (c 1 , c 2 , c 3 , c 4 , X T , T , ρ0 , β T E ) is a feasible represen- In order to verify the coverage of the airfoil library, all the 2199
tation method of the 8-dimensional design space proposed in airfoils in Profili V2.21 airfoil library are fitted and aerodynamic ver-
this paper. The control parameters has advantages in optimization, ified. The airfoil library contains most airfoils in the UIUC Airfoil
while the geometric parameters is propitious of qualitative aerody- Data Site [11], including most of the widely used airfoil, such as
namic analysis. supercritical airfoils, NACA airfoils and man-powered aircraft air-
foils. Besides, the airfoil library includes 33 airfoils which are diffi-
3. Feasibility verification of the IGP method cult to use on common aircraft wings, such as the airfoil COANDA,
whose lower curve at the leading edge is lower convex; the air-
During the process of proposing any new airfoil parameteriza- foil BE6457E and EPPLER377, whose thickness at the trailing edge
tion method, both the forward problem and the inverse problem plummet; the boomerang-shaped JED-EJ75 and SARATOU; and the
need to be answered. The forward problem is whether the rea- gourd-shaped HT05. For the convenience of narration in this paper,
sonable shape airfoil could be obtained continuously while using the airfoil library whose above 33 airfoils are removed is called the
airfoil parameterization method to construct the airfoil by modify- common airfoil group (A total of 2166 airfoils, accounts for 98.5%
ing the control parameters in the domain. The forward problem is of the total airfoil library).
about the robustness of the airfoil parameterization method in the
The basis function of the IGP method is used to fit the airfoil,
optimization process. The inverse problem is whether the method
and the correlation coefficient R 2 of the fitting curve and the orig-
could accurately depict most airfoils that have been used, so that
inal curve is used to express the fitting precision.
the commonly used airfoil library could be well covered in the op-
timization process. The inverse problem is about the universality cov( y ori , y fit )
of the airfoil parameterization method in the optimization process. R2 = (23)
σ yori · σ yfit
Firstly, the forward problem was answered.
Among them, y ori is the ordinate of the original airfoil, y f it is
In order to validate the robustness of the method, the manual
the ordinate of the fitted airfoil (when the abscissa is 0, 0.01, 0.02,
verification method was used. In the 8-dimensional space com-
posed of the 8 control parameter domains described above, 10
. . . , 0.9, 1). σ is variance, cov is covariance.
points were taken evenly within the domain corresponding to Since the correlation coefficient R 2 is a value near 1, and the
each dimension. For example, as for X T with the domain [0.2002, relationship between the value itself and the fitting degree is non-
0.4813], the 10 points were [0.2002,0.2283,0.2564,0.2845,0.3126, linear. In order to intuitively express the fitting degree, the corre-
0.3408, 0.3689, 0.3970, 0.4251, 0.4532, 0.4813]. This forms a sam- lation coefficient R 2 is processed, and the new reference value P is
ple space including 108 sample points. Based on that, 108 airfoils used to plot. A small P value represents a good fitting degree.
were depicted corresponding to each sample point. Then each air- 
foil was manually checked one by one, in order to verify the con- P = 10 log10 1 − R 2 (24)
tinuity and the robustness. Since the “strange point” which may
Fig. 2 shows the result of fitting, the abscissa indicates the P
cause the unreasonable shape in the parameterization method is
value. The left ordinate and the histogram represent the frequency
on the edge of each dimension, the density of the selected point
corresponding to the P value. The right ordinate and the curve
can verify the continuity and robustness of the method in the air-
indicate the cumulative relative frequency corresponding to the P
foil optimization.
value.
Secondly, the inverse problem was answered. As shown in Fig. 2, IGP method has a good fitting degree for the
The universality of the commonly used airfoil library is vali- whole airfoil library. In a total of 2199 airfoils, the fitting precision
dated from two aspects: geometric fitting verification and aero- R 2 of 2140 airfoils reached to 0.999 (namely P = −30). The fitting
dynamic verification. Among them, the fitting verification is to precision R 2 of all 2199 airfoils reach 0.99 (namely P = −20).
determine whether the fitting airfoil is similar to the original wing Among the 59 airfoils whose fitting precision R 2 between 0.99
shape. The aerodynamic verification is to judge whether the fitted and 0.999, 29 airfoils drop in the common airfoil group, accounting
airfoil and the original airfoil have similar aerodynamic perfor- for 1.32% of the common airfoil group. After observing the shape
mance. of these airfoils, it is indicated that this method has less fitting
 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247 245

Table 1
Aerodynamic calculation conditions.

Variables Values
Reynolds number 5000000
Ncrit 11.0
Minimum angle of attack 0◦
Maximum angle of attack 5◦

Fig. 4. Histogram of drag coefficient absolute error.

Fig. 3. Histogram of lift coefficient absolute error.

degree of the airfoils which have special function like Griffith 30%
thick symmetrical suction airfoil [12].
Next, aerodynamic verification is done to the fitted airfoil. For
this purpose, XFOIL [13] is used to calculate the aerodynamic char-
acteristics of the original airfoil and the fitted airfoil. The calcula-
tion conditions are shown in Table 1.
In XFOIL, “Ncrit” is a user-specified parameter, which is the log
of the amplification factor of the most-amplified frequency which
triggers the transition. A suitable value of this parameter depends
on the ambient disturbance level in which the airfoil operates,
and mimics the effect of such disturbances on transition [13]. The
“Ncrit” value is 11.0 indicates that the situation is “clean wind tun- Fig. 5. Histogram of moment coefficient absolute error.
nel”.
According to this condition, the lift coefficient, the drag coeffi-
cient and the moment coefficient of the most common airfoils are Table 2
Low-speed wind tunnel dynamometer check
calculated by XFOIL. The aerodynamic analysis of 319 airfoils in the
precision criteria.
common airfoil groups were failed to execution, such as BE3259B,
BE8457E and AH-7-47-6, which is of too small thickness on the Item Qualified criteria

bottom half of airfoil or strange bulge on the lower leading edge σC L 0.0040
of airfoil. The difference between the coefficients of the original σC D 0.0005
σC m 0.0012
airfoil and the fitted airfoil is statistically analyzed. The absolute
error of the lift coefficient, the drag coefficient and the moment
coefficient is plotted as histogram, shown in Figs. 3–5. Among the
figures, the raised abscissa represents the absolute error of a coef- That is, the effect of the airfoil fitting error can be considered
ficient (for a single airfoil, the value is the maximum absolute error to be less than the degree that can be detected in the wind tunnel
at each angle of attack). The left ordinate and the histogram rep- test.
resent the frequency corresponding to the absolute error. The right In the aerodynamic analysis of all airfoils, the airfoil that
ordinate and the curve represent the cumulative relative frequency reaches the maximum error is airfoil FX S 03-182, which achieves
corresponding to the absolute error. the maximum absolute error at the 5◦ angle of attack. The max-
The average absolute error of the lift coefficient is 9.57E−3; imum absolute error of the lift coefficient is 9.71E−2; the maxi-
the average absolute error of the drag coefficient is 2.61E−4 and mum absolute error of the drag coefficient is 5.11E−3; the max-
the average absolute error of the moment coefficient is 2.65E−3, imum absolute error of the moment coefficient is 2.35E−2. Since
which are close to the corresponding value of “qualified criteria” the fitted airfoil has a higher lift coefficient and lower drag coeffi-
in GJB1061-91 “High-speed Wind Tunnel and Low-speed Wind cient, it can be speculated that the error might be due to the fact
Tunnel Dynamometer Check Precision Criteria”, as shown in Ta- that the number of points used to depict the airfoil is few in the
ble 2 [14]. airfoil library.
246 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247

Fig. 6. Example of NASA SC(2) airfoil series fitting.

3) The IGP method is a constructive method without the need of

basic airfoil. In the IGP method, the control parameters could
also be directly related to the corresponding airfoil shape pa-
rameters which are commonly used in the general aerody-
namic theory. Therefore, the modification of airfoil can be
directly guided by general aerodynamic theory, such as in-
creasing the leading edge radius to improve the airfoil stall
characteristics and increasing the camber to reduce zero-lift
angle of attack.

As for the aspect of supercritical airfoil fitting, the IGP method

has a good precision. In the shape and aerodynamic data compar-
ison verification process between the original airfoil and the fitted
airfoil, the fitting effects of most supercritical airfoils are at the av-
erage level among the airfoil library, such as RAE2822 and NASA
LANLEY.
For NASA SC(2) airfoil series, the trailing edge thickness is not
Fig. 7. Example of extreme points. 0, so the series does not meet the constraint of IGP method. There-
fore, the relative error of the lift coefficient is 6.58% for which the
Table 3 fitting of trailing edge has less precision. As shown in Fig. 6, the
Comparison of each method’s number of parameters. left part shows geometric fitting and the right part show the pres-
Methods Number of parameters sure distribution based on the fitting.
PARSEC method 11 For the DFVLR airfoil series, because of its three extreme points
OBF method 10 (as shown in Fig. 7, point A, B, and C) on camber line, the cubic
CST method 11 Bézier curve has less fitting effect, which leads to the deviation of
IGP method 8 the whole fitting. The aerodynamic analysis shows that the relative
error of the lift coefficient is 9.49%. As shown in Fig. 8, the left
4. Features and applicability discussion part shows geometric fitting and the right part show the pressure
distribution based on the fitting.
IGP method has the following characteristics: In summary, the IGP method can be applied effectively in the
design process of most airfoils. In the case of high-precision air-
1) In the IGP method, the number of control parameters is less foil optimization, this method can be used to obtain a preliminary
than the three constructive methods in Table 3. Compared optimized airfoil, and the deformative method can be applied to
with the PARSEC geometric parameter method, which has the modify and obtain a precise airfoil based on the preliminary opti-
same clear physical meaning, the three parameters are re- mized airfoil.
duced, and the size of the airfoil design space can be reduced
by the geometric progression, which accelerates the comput- 5. Conclusion
ing speed of the airfoil optimization.
2) In the airfoil construction process of IGP method, the cam- In this paper, to meet the requirement of rapid airfoil optimiza-
ber and the thickness are constructed respectively. The explicit tion, a method based on the respective construction of camber and
separation of camber and thickness is the main difference thickness was proposed, called IGP method. About the method, the
from the past efforts. For the aerodynamic optimization based fitting of the existing airfoils and the range of the parameters were
on thin airfoil theory, the IGP method only needs 4 parame- studied. The following conclusions were obtained:
ters to construct the airfoil camber, while the other description
method requires at least 10 parameters due to the coupling of 1) The IGP method uses 8 control parameters to describe the air-
camber and thickness. foil. For most of the airfoils, IGP method can achieve a high
 X. Lu et al. / Aerospace Science and Technology 78 (2018) 241–247 247

Fig. 8. Example of DFVLR airfoil series fitting.

precision level both in the aspect of geometric shape fitting References

and aerodynamic calculation results.
2) The IGP method realizes that camber and thickness are ex- [1] D.A. Masters, N.J. Taylor, T.C.S. Rendall, Review of Aerofoil Parameterisation
Methods for Aerodynamic Shape Optimisation, AIAA paper 2015-0761, 2015.
pressed separately, which can better serve the optimization [2] Hsiao-Yuan Wu, Shuchi Yang, Feng Liu, Comparison of Three Geometric Rep-
of airfoil camber design based on thin airfoil theory; the ge- resentations of Airfoils for Aerodynamic Optimization, AIAA paper 2003-4095,
ometric parameters are used as the control variables, which 2003.
has clear physical meaning and is more intuitive. Moreover, [3] Zhong Xiaoping, Ding Jifeng, Robust airfoil optimization with multi-objective
estimation of distribution algorithm, Chin. J. Aeronaut. 21 (4) (2008) 289–295.
compared with the traditional airfoil construction method, the [4] Helmut Sobieczky, Parametric Airfoils and Wings, Notes on Numerical Fluid
number of design variables in IGP method is less, which re- Mechanics, vol. 68, 1998, pp. 71–88.
duces the compute of airfoil optimization. [5] G.M. Robinson, A.J. Keane, Concise orthogonal representation, J. Aircr. 38 (3)
3) In the case of higher-precision airfoil optimization, the IGP (2001) 580–583.
[6] Brenda M. Kulfan, John E. Bussoletti, Fundamental Parametric Geometry Repre-
method could be used to obtain a preliminary optimized re- sentations for Aircraft Component Shapes, AIAA paper 2006-6948, 2006.
sults. As for the follow-up design step, a deformative method [7] Kulfan Brenda, A Universal Parametric Geometry Representation Method –
could be applied to modify and obtain a precise airfoil based “CST”, AIAA paper 2007-62, 2007.
[8] Brenda Kulfan, Recent Extensions and Applications of the “CST” Universal Para-
on the preliminary optimized airfoil.
metric Geometry Representation Method, AIAA paper 2007-7709, 2007.
[9] B.M. Kulfan, Universal parametric geometry representation, J. Aircr. 45 (1)
Conflict of interest statement (2008) 142–158.
[10] Ira H. Abbott, Albert E. Von Doenhoff, Theory of Wing Sections, Dover Publica-
tions Inc., New York, 1959, Section 4.2.
There is no conflict of interest. [11] UIUC Airfoil Data Site, http://m-selig.ae.illinois.edu/ads/coord_database.html.
(Accessed 19 February 2015).
[12] Sydney Goldstein, Low-drag and suction airfoils, J. Aeronaut. Sci. 15 (4) (1948)
Appendix A. Supplementary material 189–214.
[13] M. Drela, XFOIL 6.94 User Guide, MIT Department of Aeronautics and Astro-
nautics, 2001.
[14] GJB1061-91, Requirement for Force-Test Precision of High and Low Speed Wind
Supplementary material related to this article can be found on-
Tunnels, 1991 (in Chinese).
line at https://doi.org/10.1016/j.ast.2018.04.025.