Aeroelastic Analysis of a Flexible Wing Wind Tunnel Model with

Variable Camber Continuous Trailing Edge Flap Design

Nhan Nguyen ⇤
NASA Ames Research Center, Moffett Field, CA 94035
Eric Ting †
Stinger Ghaffarian Technologies Inc., Moffett Field, CA 94035
Sonia Lebofsky ‡
Stinger Ghaffarian Technologies Inc., Moffett Field, CA 94035

This paper presents data analysis of a flexible wing wind tunnel model with a variable camber continuous trailing
edge flap (VCCTEF) design for drag minimization tested at the University of Washington Aeronautical Laboratory
(UWAL). The wind tunnel test was designed to explore the relative merit of the VCCTEF concept for improved cruise
efficiency through the use of low-cost aeroelastic model test techniques. The flexible wing model is a 10%-scale model
of a typical transport wing and is constructed of woven fabric composites and foam core. The wing structural stiffness
in bending is tailored to be half of the stiffness of a Boeing 757-era transport wing while the torsional stiffness is about
the same. This stiffness reduction results in a wing tip deflection of about 10% of the wing semi-span. The VCCTEF
is a multi-segment flap design having three chordwise camber segments and five spanwise flap sections for a total of
15 individual flap elements. The three chordwise camber segments can be positioned appropriately to create a desired
trailing edge camber. Elastomeric material is used to cover the gaps in between the spanwise flap sections, thereby
creating a continuous trailing edge. Wind tunnel data analysis conducted previously shows that the VCCTEF can
achieve a drag reduction of up to 6.31% and an improvement in the lift-to-drag ratio (L/D) of up to 4.85%. A method for
estimating the bending and torsional stiffnesses of the flexible wing UWAL wind tunnel model from static load test data is
presented. The resulting estimation indicates that the stiffness of the flexible wing is significantly stiffer in torsion than in
bending by as much as 3 to 1. The lift prediction for the flexible wing is computed by a coupled aerodynamic-structural
model. The coupled model is developed by coupling a conceptual aerodynamic tool Vorlax with a finite-element model
of the flexible wing via an automated geometry deformation tool. Based on the comparison of the lift curve slope, the
lift prediction for the rigid wing is in good agreement with the estimated lift coefficients derived from the wind tunnel
test data. Due to the movement of the VCCTEF during the wind tunnel test, uncertainty in the lift prediction due to
the indicated variations of the VCCTEF deflection is studied. The results show a significant spread in the lift prediction
which contradicts the consistency in the aerodynamic measurements, thus suggesting that the indicated variations as
measured by the VICON system may not be reliable. The lift prediction of the flexible wing agrees very well with the
measured lift curve for the baseline configuration. The computed bending deflection and wash-out twist of the flexible
wing also match reasonably well with the aeroelastic deflection measurements. The results demonstrate the validity of
the aerodynamic-structural tool for use to analyze aerodynamic performance of flexible wings.

I. Introduction
The aircraft industry has been responding to the need for energy-efficient aircraft by redesigning airframes to be aerodynam-
ically efficient, employing light-weight materials for aircraft structures and incorporating more energy-efficient aircraft engines.
Reducing airframe operational empty weight (OEW) using advanced composite materials is one of the major considerations
for improving energy efficiency. Modern light-weight materials can provide less structural rigidity while maintaining sufficient
load-carrying capacity. As structural flexibility increases, aeroelastic interactions with aerodynamic forces and moments can
alter aircraft aerodynamics significantly, thereby potentially degrading aerodynamic efficiency.
Under the Fundamental Aeronautics Program in NASA Aeronautics Research Mission Directorate, the Fixed Wing project
is conducting multidisciplinary research to investigate advanced concepts and technologies for future aircraft systems. A NASA
study entitled “Elastically Shaped Future Air Vehicle Concept” was conducted in 20101, 2 to examine new concepts that can
⇤ NASA Ames Research Center, Research Scientist, AIAA Associate Fellow, nhan.t.nguyen@nasa.gov
† Stinger Ghaffarian Technologies Inc., NASA Ames Research Center, Research Engineer, eric.b.ting@nasa.gov
‡ Stinger Ghaffarian Technologies Inc., NASA Ames Research Center, Research Engineer, sonia.lebofsky@nasa.gov

1 of 28

American Institute of Aeronautics and Astronautics

enable active control of wing aeroelasticity to achieve drag reduction. This study showed that highly flexible wing aerodynamic
surfaces can be elastically shaped in-flight by active control of wing twist and vertical deflection in order to optimize the local
angles of attack of wing sections to improve aerodynamic efficiency through drag reduction during cruise and enhanced lift
performance during take-off and landing.
The study shows that active aeroelastic wing shaping control can have a potential drag reduction benefit. Conventional flap
and slat devices inherently generate drag as they increase lift. The study shows that conventional flap and slat systems are not
aerodynamically efficient for use in active aeroelastic wing shaping control for drag reduction. A new flap concept, referred
to as Variable Camber Continuous Trailing Edge Flap (VCCTEF) system, was conceived by NASA to address this need.1
Initial study results indicate that, for some applications, the VCCTEF system may offer a potential pay-off in drag reduction
that could provide significant fuel savings. In order to realize the potential benefit of drag reduction by active span-load and
aeroelastic wing shaping control while meeting all other performance requirements, the approach for high lift devices needs to
be considered as part of the wing shaping control strategy.
NASA and Boeing are currently conducting a joint study to develop the VCCTEF further under the research element Active
Aeroelastic Shape Control (AASC) within the Fixed Wing project.3, 4 This study is built upon the development of the VCCTEF
system for NASA Generic Transport Model (GTM) which is essentially based on the Boeing 757 airframe,5 employing light-
weight shaped memory alloy (SMA) technology for actuation and three separate chordwise segments shaped to provide a
variable camber to the flap. This cambered flap has potential for drag reduction as compared to a conventional straight, plain
flap. The flap is also made up of individual 2-foot spanwise sections which enable different flap setting at each flap spanwise
position. This results in the ability to control the wing twist shape as a function of span, resulting in a change to the wing
twist to establish the best lift-to-drag ratio (L/D) at any aircraft gross weight or mission segment. Wing twist on traditional
commercial transport designs is dictated by the aeroelastic deflection of a fixed “jig twist” shape applied at manufacture. The
design of this jig twist is set for one cruise configuration, usually for a 50% fuel loading or mid-point on the gross weight
schedule. The VCCTEF offers different wing twist settings, hence different spanwise loadings, for each gross weight condition
and also different settings for climb, cruise and descent, a major factor in obtaining best L/D conditions.
The second feature of VCCTEF is a continuous trailing edge flap. The individual 2-foot spanwise flap sections are connected
with a flexible covering, so no breaks can occur in the flap planforms, thus reducing drag by eliminating these breaks in the
flap continuity which otherwise would generate vorticity that results in a drag increase and also contributes to airframe noise.
This continuous trailing edge flap design combined with the flap camber result in lower drag increase during flap deflections.
In addition, it also offers a potential noise reduction benefit.

Fig. 1 - Wing Configured with the Variable Camber Continuous Trailing Edge Flap
The VCCTEF is divided into 14 sections attached to the outer wing and 3 sections attached to the inner wing, as shown in
Fig. 1.4 Each 24-inch section has three camber flap segments that can be individually commanded to form a variable camber

2 of 28

American Institute of Aeronautics and Astronautics

trailing edge, as shown in Fig. 2. These camber flaps are joined to the next section by a flexible and supported material (shown
in blue) installed with the same shape as the camber and thus providing continuous flaps throughout the wing span with no drag
producing gaps.

Fig. 2 - Variable Camber Flap

Using the camber positioning, a full-span, low-drag, high-lift configuration can be activated that has no drag producing gaps
and a low flap noise signature. This is shown in Fig. 3. To further augment lift, a slotted flap configuration is formed by an air
passage between the wing and the inner flap that serves to improve airflow over the flap and keep the flow attached. This air
passage appears only when the flaps are extended in the high lift configuration.

Fig. 3 - Cruise and High Lift VCCTEF Configurations

Figure 4 illustrates the GTM equipped with the VCCTEF for wing shaping control. By actively shaping the wing aero-
dynamic surface using the VCCTEF, optimal aerodynamic performance could potentially be realized at any point in the flight
envelope. The VCCTEF relies on two mechanisms to improve aerodynamic performance: 1) wing twist optimization for flexi-
ble wing design, and 2) variable camber and continuous trailing edge for improved aerodynamics. This fixed-wing technology
may be referred to as Performance Adaptive Aeroelastic Wing (PAAW) technology.

Fig. 4 - GTM with VCCTEF

3 of 28

American Institute of Aeronautics and Astronautics

 II. Wind Tunnel Model of Flexible Wing with VCCTEF
For exploratory assessment of the aerodynamic potential of the VCCTEF concept, a 10%-scale aeroelastic model of a
softened Boeing 757-based GTM wing was constructed for a wind tunnel experimental investigation in the University of
Washington Aeronautical Laboratory (UWAL) in August of 2013.6 The semi-span of the model is 5.6075 ft, as shown in Fig.
5. The model is constructed of woven fabric composites skin and extruded polystyrene foam core. The composite laminates
and extruded polystyrene foam core are structurally tailored to attain half of the bending stiffness of the scaled baseline GTM
wing stiffness while keeping torsional stiffness about the same. This tailored stiffness is to achieve a 10% wing tip deflection.
The VCCTEF parts are fabricated by 3D printing. The flap segments are mechanically interlocking aerodynamic surfaces in
the chordwise direction and mate with silicone elastomer material between spanwise flap sections, as shown in Fig. 6. The flap
segments are hinged at three chordwise locations and are designed to be fully adjustable. The dimensions of the VCCTEF are
shown in Fig. 7.

Fig. 5 - UWAL Wind Tunnel Model with VCCTEF (Courtesy of University of Washington Aeronautical Laboratory)

Fig. 6 - VCCTEF Construction (Courtesy of University of Washington Aeronautical Laboratory)

4 of 28

American Institute of Aeronautics and Astronautics

 Fig. 7 - VCCTEF Dimension (Courtesy of University of Washington Aeronautical Laboratory)

Figure 8 shows an exploded view of the UWAL wind tunnel model of the flexible wing mated to a center body fairing
attached to an external floor-mounted balance.

Fig. 8 - Exploded View of Wind Tunnel Model (Courtesy of University of Washington Aeronautical Laboratory)

The UWAL aeroelastic wind tunnel model was built with different stiffness than the scaled GTM wing and its jig shape was
not optimized for best aerodynamic performance at test conditions. The same jig shape as that of the GTM wing was used. The
wash-out twist is therefore non-optimal for the model when it operates at the design lift coefficient of 0.51. A CFD optimization
was conducted prior to the test to identify an optimal jig twist. However, this optimized jig twist was not incorporated into the
final model fabrication due to programmatic issues.
The relevant model scaling information is given in Table 1.

5 of 28

American Institute of Aeronautics and Astronautics

 Full-Scale Semi-Span Model
M• 0.797 0.1162
CL 0.51 0.51
h, ft 36,000 0
q• , psf 211.09 20.00
S/2, ft2 975.5 9.638
c̄, ft 16.6417 1.5963
b/2, ft 62.4167 6.1262

Table 1 - Model Parameters

The wind tunnel test is designed to be an exploratory, proof-of-concept study. The objective of the wind tunnel experiment
is to explore the relative merit of the VCCTEF design as a drag reduction control device, and the ability to simulate the problem
in a relatively low-cost test. Lift, drag, side force, pitching moment, yawing moment, and rolling moment were recorded from
the external floor-mounted balance. In addition, aeroelastic deflections of the flexible wing model were also measured by a
VICON motion tracking system. The VICON system measured the 3D displacement of the wind tunnel model at 54 points on
the model.
To ensure that the wind tunnel model has correct aeroelastic properties, static load tests and frequency measurements
were conducted. A detail 3D NASTRAN model was constructed by UWAL for comparison with the measurements.6 The
3D NASTRAN results demonstrate an excellent agreement with the static load test data and measured frequencies. A stick
NASTRAN model was also constructed by UWAL to match the deflection information from the 3D NASTRAN model. In
addition, a NASTRAN Doublet Lattice flutter analysis was performed by UWAL to ensure that the wind tunnel is flutter-free in
the wind tunnel.6 Flutter speeds were determined to be well above the test section airspeed of 39.54 m/sec.

III. Wind Tunnel Test

The test was conducted at a nominal dynamic pressure of 20 psf. Off-condition data and additional runs at 10, 15, 25 and
30 psf were also collected for some cases. The nominal test section airspeed was Mach 0.1162. Figure 14 is a photograph of
the flexible wing wind tunnel model in the UWAL test section.

Fig. 9 - Flexible Wing Wind Tunnel Model in UWAL Test Section (Courtesy of University of Washington Aeronautical
Laboratory)

The wind tunnel model was tested with a total of 13 VCCTEF configurations ranging from zero to full deflection. These
VCCTEF configurations are designated as:

6 of 28

American Institute of Aeronautics and Astronautics

 • FLAP0 - baseline zero deflection
• FLAP1 - full deflection for all flap sections
• FLAP2 - varying from a maximum deflection at the inboard and outboard flaps to a minimum deflection at the mid-span
flap

• FLAP3 - varying from a minimum deflection at the inboard and outboard flaps to a maximum deflection at the mid-span
flap
• FLAP4 - varying monotonically from a maximum deflection at the inboard flap to zero deflection at the outboard flap
• FLAP5 - varying monotonically from zero deflection at the inboard flap to a maximum deflection at the outboard flap

• FLAP6 - similar to FLAP4 configuration but with a smaller deflection

• FLAP7 - varying monotonically from a maximum positive deflection at the inboard flap to a negative deflection at the
outboard flap
• FLAP8 - rigid-body deflection with the two outer camber segments at zero relative deflection

• FLAP9 - deflection of the trailing edge camber segments

• FLAP10 - intermediate deflection
• FLAP11 - full negative deflection

• FLAP12 - FLAP6 configuration plus a gurney flap

Figure 10 illustrates some of these flap configurations.

Fig. 10 - VCCTEF Flap Configurations (Courtesy of University of Washington Aeronautical Laboratory)

The test run matrix for the VCCTEF is shown in Table 2.

7 of 28

American Institute of Aeronautics and Astronautics

 FLAP 0 1 2 3 4 5 6 7 8 9 10 11 12
q• = 0 20 psf, a = 0o 18 25 32 42 55 73 79 86 93
87
q• = 0 20 psf, a = 1o 19 26 33 38 43 49 56 64 74 80 94
88
q• = 0 20 psf, a = 1.5o 27 34
q• = 0 20 psf, a = 2o 20 28 35 39 44 50 57 65 75 81 89 95
q• = 0 20 psf, a = 2.5o 29 36
q• = 0 20 psf, a = 3o 30 40 45 51 58 66 76 82 90 96
q• = 0 20 psf, a = 4o 22 41 46 52 59 67 77 83 91 97
q• = 0 20 psf, a = 4.5o 47 53
q• = 0 20 psf, a = 5o 23 60, 61 68 78 84 92 98
q• = 0 20 psf, a = 5.5o 62
q• = 0 20 psf, a = 6o 99
q• = 0 20 psf, a = 6.5o 100
q• = 10 psf, a = 2o 10.5o 117 121
q• = 15 psf, a = 2o 8.5o 118 122
104 106 109
q• = 20 psf, a = 2o 7o 111 102 113
116 107 123
q• = 25 psf, a = 2o 5.5o 119 124
q• = 30 psf, a = 2o 4.5o 120 125

Table 2 - Test UW2087 Run Matrix

The VCCTEF segment positions for these configurations are shown in Table 3. The following convention is used. A flap
section is a portion of the VCCTEF along the spanwise direction. There are 5 flap sections numbered from 1 at the inboard to
5 at the outboard. Each flap section is comprised of three camber segments labeled as A for the innermost camber segment,
B for the middle camber segment, and C for the trailing edge camber segment. This is shown in Fig. 11. The position angle
is denoted by (a/b/c) where a, b, and c are flap positions in degrees relative to the forward flap segment. For segment A, the
position angle is with respect to the fixed wing portion.

Fig. 11 - VCCTEF Flap Notation

8 of 28

American Institute of Aeronautics and Astronautics

 FLAP Run Section 1 Section 2 Section 3 Section 4 Section 5
1 25 (5/4/9) (6/7/9) (9/7/10) (7/8/10) (6/9/9)
1* 106 (1/7/11) (3/9/7) (5/7/8) (5/9/7) (5/10/10)
1** 107 (1/7/11) (3/9/7) (5/7/8) (5/9/7) (5/10/10)
32 (3/2/10) (4/5/4) (5/3/4) (4/6/5) (5/10/10)
2
36 (2/3/9) (4/5/4) (5/3/5) (4/6/4) (4/10/10)
38 (2/3/5) (5/6/5) (6/6/11) (5/6/5) (4/6/5)
3
41 (2/3/5) (4/6/5) (4/6/11) (4/6/5) (3/6/5)
42 (2/5/9) (3/6/3) (4/1/6) (2/5/2) (0/0/0)
4
47 (1/4/9) (3/6/2) (4/1/6) (2/5/1) (0/0/0)
49 (0/0/0) (2/5/1) (4/2/7) (3/5/6) (3/8/9)
5 51 (0/0/0) (2/5/1) (4/2/7) (3/5/6) (3/8/9)
53 (0/0/0) (2/5/1) (4/2/7) (3/5/5) (3/8/9)
55 (1/1/1) (0/2/0) (0/0/2) (0/0/0.5) (0/0/0)
61 (1/1/1) (0/2/0) (0/0/2) (0/0/0.5) (0/0/0)
6
62 (1/1/1) (0/2/0) (0/0/2) (0/0/0.5) (0/0/0)
111 (1/1/1) (0/1/1) (0/0/2) (0/0/0) (0/0/0)
7 64 (3/3/4) (2/3/2) (0/0/3) (0/0/0) (0/0/ 2)
7* 109 (3/4/3) (2/5/1) (2/0/2) (0/ 1/0) (0/0/ 2)
73 (4/0/0) (4/0/0) (5/0/0) (5/0/0) (4/0/0)
8
78 (3/0/0) (4/0/0) (5/0/0) (4/0/0) (4/0/0)
79 (0/0/6) (0/0/6) (0/0/8) (0/0/6) (0/0/6)
9 81 (0/0/6) (0/0/6) (0/0/8) (0/0/6) (0/0/6)
84 (0/0/6) (0/0/6) (0/0/8) (0/0/6) (0/0/6)
86 (2/3/2) (3/3/1) (4/2/5) (2/5/2) (2/5/3)
10 91 (2/3/2) (3/3/1) (4/2/5) (2/5/2) (2/5/3)
92 (1/3/2) (3/3/1) (4/2/5) (2/5/2) (2/5/2)
93 ( 1/0/ 3) (0/0/ 6) ( 1/ 4/0) ( 2/0/ 5) ( 4/0/ 4)
11
100 ( 1/0/ 4) (0/0/ 5) ( 1/ 3/0) ( 2/0/ 5) ( 3/0/ 4)

Table 3 - VCCTEF Deflections

IV. Aeroelastic Effects on Lift and Drag

The aeroelastic deflections of a flexible swept back wind tunnel model contribute significantly to the aerodynamic perfor-
mance due to the wash-out effect resulting from wing bending and twist. Additionally, the VCCTEF deflections also contribute
to angle of attack changes. The aeroelastic angle of attack of a wing section can be expressed as7
N
dW (ȳ) ∂ ac
ac (y) = a ai (y) g (ȳ) cos L Q (ȳ) cos L sin L + Â di (yh ) cos Lh (1)
d ȳ i=1 ∂ di

where a is the geometric angle of attack of the wing section about the pitch axis y, ai is the induced angle of attack due to the
downwash about the pitch axis y, g is the wing pre-twist angle about the elastic axis ȳ = y/ cos L (positive nose down), Q is the
wing torsional twist about the elastic axis ȳ (positive nose down), W is wing vertical bending along the elastic axis (positive
upward), L is the sweep angle of the elastic axis, di is the absolute deflection of the i-th flap segment of the VCCTEF about the
hinge axis yh which has a sweep angle of Lh , and ∂ ac /∂ di is the angle of attack sensitivity or camber control derivative due to
the VCCTEF flap deflection.
The following convention is used. A quantity expressed with the independent variable y is one with respect to the streamwise
direction, e.g., c (y) is the sectional chord in the streamwise direction. A quantity expressed with the independent variable ȳ is
one with respect to the elastic axis; e.g., Q (ȳ) is the wing twist about the elastic axis. The quantity di (yh ) is the flap deflection
about the hinge axis.
In general, the aeroelastic deflections are functions of the angle of attack and the flap deflection. Therefore, they can be

9 of 28

American Institute of Aeronautics and Astronautics

expressed as
N
∂ Q (ȳ) ∂ Q (ȳ)
Q (ȳ) = Q0 (ȳ) + a+Â di (2)
∂a i=1 ∂ di
N
∂W (ȳ) ∂W (ȳ)
W (ȳ) = W0 (ȳ) + a+Â di (3)
∂a i=1 ∂ di

where Q0 and W0 are the aeroelastic deflections at zero angle of attack, and the partial derivatives are the sensitivities of the
aeroelastic deflections with respect to the angle of attack and the flap deflection.
Therefore, the expression for the aeroelastic angle of attack of a wing section can be written as
 ✓ ◆
∂ Q (ȳ) ∂ dW (ȳ) dW0 (ȳ)
ac (y) = a 1 cos L sin L ai (y) g (ȳ) cos L Q0 (ȳ) cos L sin L
∂a ∂a d ȳ d ȳ
N  ✓ ◆
∂ ac ∂ Q (ȳ) ∂ dW (ȳ)
+Â cos Lh cos L sin L d (yh ) (4)
i=1 ∂ di ∂ di ∂ di d ȳ

It can be seen that the aeroelastic deflections can cause the desired sectional angle of attack to be non-optimal. The effect
of the adaptive aeroelastic wing shaping control by the VCCTEF is captured in the last term. The term ∂∂adc is the rigid camber
⇣ ⌘
control to compensate for the non-optimal sectional angle of attack. The two terms ∂∂ Q d and ∂ d
∂ dW
d ȳ are the aeroelastic wing
shaping control by leveraging wing flexibility to change the wash-out twist of a wing in order to achieve improved aerodynamic
performance. Thus, the effect of adaptive aeroelastic wing shaping control is to optimize the span load at any operating point
inside a given flight envelope.
The camber control derivative ∂∂adic can be directly estimated from thin airfoil potential flow theory by evaluating the follow-
ing integral transform with the potential kernel function f (q ) = cos q 1 as8
qi+1 qi+1
∂a 1 1
ˆ ˆ
= f (q ) dq = (cos q 1) dq (5)
∂ di p qi p qi

where
0
0 c
x = (1 cos q ) (6)
2
0
0 0 c
c xi = (1 cos qi ) (7)
2
0 0
and c is the airfoil chord and xi is the flap hinge position of the i-th flap segment measured normal to the hinge axis from the
trailing edge and is given by
0 0
xi = (n + 1 i) c f (8)
0
where c f is the flap chord of a camber segment measured normal to the hinge axis.
0 0 0 0 0
So the first hinge position is at x1 = Nc f , the last hinge position is at xN = c f , and the trailing edge position is at xN+1 = 0.
This integral is evaluated as

1(
p c⇤i+1
∂ ac cos c⇤ ) 1 c⇤2
= (9)
∂ di p
c⇤i

where 0
xi
c⇤ = 1 2 0 (10)
c
The sectional lift coefficient is expressed as
" #
N
dW (ȳ) ∂ ac
cL (y) = cL0 (y) + cLa (y) a (y) g (ȳ) cos L Q (ȳ) cos L sin L + Â di (yh ) cos Lh (11)
d ȳ i=1 ∂ di

where cL0 is the sectional lift coefficient at zero angle of attack and cLa is the sectional lift curve slope, both of which account
for the induced angle of attack.

10 of 28

American Institute of Aeronautics and Astronautics

 The sectional lift coefficient parameters cL0 and cLa vary along the wing span and can be estimated from a three-dimensional
aerodynamic analysis using a panel code or a CFD (Computational Fluid Dynamics) code.
Using aerodynamic strip theory, the total wing lift coefficient can be computed by integrating the local lift coefficient as
follows:
1 b/2
ˆ
CL = cL (y) c (y) dy (12)
S b/2
where c is the local chord length in the streamwise direction, and S is the wing reference area.
The lift coefficient can also be expressed in the standard form
CL = CL⇤0 +CL⇤a a +CL⇤d d + DCLe (13)
where CL⇤0 is the rigid lift coefficient at zero angle of attack, CL⇤a is the rigid lift curve slope of finite-aspect ratio wing, CL⇤d is
h i>
the vector of rigid lift coefficient sensitivities due to the VCCTEF deflection, d = d1 . . . dN is the VCCTEF deflection
vector, and DCLe is the incremental lift coefficients due to wing flexibility.
These quantities may be evaluated as
1 b/2 ⇥ ⇤
ˆ
CL⇤0 = cL0 (y) cLa g (y) cos L c (y) dy (14)
S b/2
1 b/2
ˆ
⇤
CLa = cL (y) c (y) dy (15)
S b/2 a
1 b/2 ∂ ac
ˆ
⇤
CLd = cL (y) cos Lh c (y) dy (16)
S b/2 a ∂d

1 b/2 dW (ȳ)
ˆ
DCLe = cLa (y) Q (ȳ) cos L sin L c (y) dy (17)
S b/2 d ȳ
h i
where ∂∂adc = ∂∂ ad c . . . ∂∂ dac .
1 N
The aeroelastic deflections of a wing can be modeled by the following equations:
⇢ ⇣ 0 ⌘2 ∂ Q 2 h i
∂ 0∂ W e
GJ + EB1 g EB2 g = cm + c m d c + c L q• cos2 Lc (18)
∂ ȳ ∂ ȳ ∂ ȳ2 ac d
cos L
✓ ◆
∂2 ∂ 2W 0 ∂Q
EI EB2 g = cL q• cos Lc (19)
∂ ȳ 2 ∂ ȳ 2 ∂ ȳ
where e is the aerodynamic center offset measured perpendicular to the elastic axis, cmac is the sectional pitching moment
coefficient about the aerodynamic center defined in the streamwise direction, cmd is the vector of pitching moment coefficient
sensitivities about the aerodynamic center defined in the streamwise direction for the VCCTEF, and B1 and B2 are the bending-
torsion coupling constants which are generally small for slightly twisted wings and therefore are usually neglected.
Using thin airfoil potential flow theory, the pitching moment coefficient sensitivity about the aerodynamic center, which is
taken to be at the quarter-chord location, of the i-th flap segment can be estimated using a kernel function g (q ) = cos 2q cos q
as
cLa qi+1
ˆ
cmd = (cos 2q cos q ) dq (20)
i 4p qi
where cLa is the sectional lift curve slope from Eq. (11).
This results in a theoretical expression as
p c⇤i+1
cL (1 + c⇤ ) 1 c⇤2
cmd = a (21)
i 4p
c⇤i

The Kutta-Joukowski theorem states that, for a two-dimensional airfoil, the lift force per unit length is L = r•V• G, where
G is the circulation.8 Then, from Eq. (11), the lift circulation is defined as
1
G (y) = V• cL (y) c (y)
2 ( " #)
N
1 dW (ȳ) ∂ ac
= V• cL0 (y) + cLa (y) a g (ȳ) cos L Q (ȳ) cos L sin L + Â di (yh ) cos Lh c (y) (22)
2 d ȳ i=1 ∂ di

11 of 28

American Institute of Aeronautics and Astronautics

 The lift circulation shows that the aeroelastic deflections can influence the spanwise lift distribution, hence induced drag.
As a wing becomes more flexible, aeroelastic deflections at off-design cruise conditions can cause induced drag penalty. By
controlling the aeroelastic deflections using the VCCTEF, the spanwise lift distribution can be tailored for off-design aerody-
namic performance. It should be noted that the VCCTEF can provide the same potential benefit for a stiff and, in the limit, even
rigid wing design since the VCCTEF can directly alter the spanwise lift distribution via camber control. To achieve optimal
off-design aerodynamic performance, a multi-disciplinary optimization can be conducted to determine the optimal VCCTEF
deflections at functions of flight conditions inside a flight envelope. As a matter of fact, the very terminology of “design”
and “off-design” flight conditions may need to be revised, since with the VCCTEF it might be possible to approach optimal
lift-to-drag ratios at many flight conditions.
For a flexible swept back wing, the bending slope dW d ȳ which is normally positive causes a local wing wash-out twist which
decreases the sectional angle of attack. As a result, the lift coefficient generally decreases at a given aircraft angle of attack
when compared to the lift coefficient for a rigid wing. Therefore, to achieve the same lift coefficient, an increase in the angle of
attack is required to compensate for the aeroelastic effect.
To model the aeroelastic effects, various modeling approaches are used. A standard approach is to use a finite-element
model coupled with an aerodynamic solver such as a panel code or a CFD code. To use the finite-element method, the structural
properties of a wing need to be established accurately. This information is then used to predict the aeroelastic deflection. Even
though such a method is considered to be standard, the modeling accuracy can still be subject to high uncertainty. Stiffness
uncertainty as well as aerodynamic prediction uncertainty can result in inaccurate results.
One approach to reducing modeling uncertainty in the data analysis using strictly experimental data is the Galerkin’s
method. The Galerkin’s method uses a series approximation of mode shapes using the analytical mode shapes of a simpler
structure with the same boundary conditions. For a cantilever wing, the mode shapes can be approximated by a series of an-
alytical mode shapes of a uniform cantilever beam.9 Since no structural information about the stiffness is required and the
aeroelastic deflection measurements are available, the Galerkin’s method is well suited for use in the data analysis. The re-
sulting formulation can add insight into the various physical effects involved in the wind tunnel test. A standard finite-element
aerodynamic panel solution is presented in the paper by Precup, Mor, and Livne.6 In addition, a preliminary static aeroelas-
tic analysis was also conducted by Rodriguez, Aftosmis, Nemec, and Smith using Cart3D inviscid CFD code coupled with
a structural beam model.10 This paper presents an aeroelastic study of the flexible wing UWAL wind tunnel model using a
finite-element modeling method coupled to a vortex-lattice aerodynamic solver developed by Nguyen, Ting, et al.11, 12
The general solution of the coupled bending-torsion aeroelasticity can be expressed as

Q (ȳ) = Y (ȳ) q (23)

W (ȳ) = F (ȳ) w (24)
h i h i h i>
where Y (ȳ) = y1 (ȳ) . . . yn (ȳ) and F (ȳ) = f1 (ȳ) . . . fn (ȳ) are mode shape functions, and q = q1 . . . qn
h i>
and w = w1 . . . wn are generalized displacements which can be solved from
" # " # 1 (" # " # " # )
q Ksq q + Kaq q Kaq w Fq0 Fqa Fqd
= + a+ d
w Kawq Ksww + Kaww Fw0 Fwa Fw0
" # " # " #
q0 ∂ q ∂ q
= + a + q• d (25)
w0 ∂a w ∂d w

Note that the stiffness matrices Kaq q , Kaq w , Kawq , and Kaww are dependent on q• . For low dynamic pressure well below the
divergence dynamic pressure, the stiffness matrix inversion can be approximated as
✓ ◆ ✓ ◆ 1
Ka 1 Ka Ka
Ks + q• = Ks 1 q• Ks 1 Ks 1 I + q• Ks 1 (26)
q• q• q•
Ka Ka Ka
⇡ Ks 1 q• Ks 1 Ks 1 + q2• Ks 1 Ks 1 Ks 1 (27)
q• q• q•

where qK•a is the aerodynamic generalized influence coefficients matrix which is independent of q• .
Thus, the aeroelastic deflections are obtained as functions of the angle of attack and the flap deflections as
⇥ ⇤
Q (ȳ) = Y (ȳ) A0q + Aaq a + Adq d q• + B0q + Baq a + Bdq d q2• + C0q +Caq a +Cdq d q3• (28)
⇥ ⇤
W (ȳ) = F (ȳ) A0w + Aaw a + Adw d q• + B0w + Baw a + Bdw d q2• + C0w +Caw a +Cdw d q3• (29)

12 of 28

American Institute of Aeronautics and Astronautics

 Thus, the lift coefficient of a flexible wing is expressed as

CL = CL⇤ + (a0 + aa a + ad d ) q• + (b0 + ba a + bd d ) q2• + (c0 + ca a + cd d ) q3• (30)

where CL⇤ is the rigid-wing lift coefficient and the remaining terms are due to the effect of aeroelasticity and
b/2 
1 dF (ȳ)
ˆ
a(0,a,d ) = cLa (y) Y (ȳ) A(0,a,d )q cos L A(0,a,d )w sin L c (y) dy (31)
S b/2 d ȳ
b/2 
1 dF (ȳ)
ˆ
b(0,a,d ) = cLa (y) Y (ȳ) B(0,a,d )q cos L B(0,a,d )w sin L c (y) dy (32)
S b/2 d ȳ
b/2 
1 dF (ȳ)
ˆ
c(0,a,d ) = cLa (y) Y (ȳ)C(0,a,d )q cos L C(0,a,d )w sin L c (y) dy (33)
S b/2 d ȳ
⇥ ⇤
Note that these integral are evaluated in the wing pitch axis y 2 0, b2 .
Equation (30) can be used to estimate the⇣ rigid wing⌘ lift coefficients. Note that if a swept back wing experiences a nose-
down twist (Y (ȳ) > 0) and upward bending dF( ȳ)
d ȳ > 0 , then the terms a(0,a,d ) are negative since they are due to the positive
terms A(0,a,d )q and A(0,a,d )w which come from Ks 1 which is always positive.
Note that when q• = 0, CL = CL⇤ . This suggests that one can estimate the rigid-wing lift coefficient by analyzing the lift
coefficient variation with the dynamic pressure.
If the dynamic pressure is held constant and if aeroelastic deflection measurements are available, then this information can
also be used to estimate the rigid-wing lift coefficient. In this case, the effective lift coefficient parameters are expressed as
b/2 
1 dW0 (ȳ)
ˆ
CL0 = CL⇤0 cLa (y) Q0 (ȳ) cos L + sin L c (y) dy (34)
S b/2 d ȳ
 ✓ ◆
1 b/2 ∂ Q (ȳ) ∂ dW (ȳ)
ˆ
CLa = CL⇤a cL (y) cos L + sin L c (y) dy (35)
S b/2 a ∂a ∂a d ȳ
 ✓ ◆
1 b/2 ∂ Q (ȳ) ∂ dW (ȳ)
ˆ
CLd = CL⇤d cL (y) cos L + sin L c (y) dy (36)
S b/2 a ∂d ∂d d ȳ

Since Q and dW d ȳ are generally positive for a wing with positive lift, the effect of aeroelasticity on a swept back wing is to
reduce lift and lift coefficient sensitivity due to a flap deflection.

V. Summary of Wind Tunnel Experimental Data

There was a total of 125 useful test runs. Test data consist of force, moment, and aeroelastic deflection measurements, and
were collected during a-sweep and q• -sweep runs. For clarity, a is the angle of attack of the flexible wing wind tunnel model,
which is defined as the angle between incoming air and the root attachment of the wing. The force and moment measurements
were provided by the external balance. The aeroelastic deflection measurements were provided by the VICON motion tracking
system. The nominal dynamic pressure is 20 psf. The full report of the wind tunnel test data analysis is published by Nguyen
et al.13
For summary, only test data for FLAP0 configuration is shown. The lift curve and drag polar for the baseline FLAP0
configuration are plotted in Figs. 12 and 13. The data come from 20 different runs. As can be seen, the data show high degree
of repeatability and correlation.

13 of 28

American Institute of Aeronautics and Astronautics

 Fig. 12 - FLAP0 CL vs. a Fig. 13 - FLAP0 CL vs. CD

Figures 14 and 15 are the plots of L/D for FLAP0, FLAP1, FLAP7, and FLAP11 configurations. The baseline FLAP0
configuration has a L/D value of about 16.1 at CL = 0.51. FLAP1 configuration has the highest L/D at the same CL as compared
to all the other VCCTEF configurations. Its L/D is about 17.2 which is about a 6.8% improvement. FLAP11 configuration is
not aerodynamically efficient since it is configured as a lift-reduction device with negative VCCTEF deflections.

Fig. 14 - L/D vs. a Fig. 15 - L/D vs. CL

Table 4 presents the summary of the drag comparison for all the VCCTEF configurations. As can be seen, FLAP1 con-
figuration is the most aerodynamically efficient, achieving a maximum of 6.31% drag reduction at the design cruise CL . All
VCCTEF configurations except FLAP11 and FLAP12 achieve varying degree of drag reduction. In terms of the absolute values
of L/D, the maximum increase in L/D is 4.85% due to FLAP1 configuration.

14 of 28

American Institute of Aeronautics and Astronautics

 FLAP CD @ CL = 0.51 (Counts) CD Reduction (Counts) % Drag Reduction L/D Max % L/D Increase
0 317 0 0 16.4182 0
1 303 14 4.42 17.2151 4.85
1* 298 19 5.99 17.1337 4.36
1** 297 20 6.31 17.1553 4.49
2 301 16 5.05 17.0012 3.55
3 300 17 5.36 17.1285 4.33
4 308 9 2.84 16.8104 2.39
5 304 13 4.10 16.9709 3.37
6 313 4 1.26 16.6054 1.14
7 313 4 1.26 16.6340 1.31
7* 312 5 1.58 16.5125 0.57
8 311 6 1.89 16.7022 1.73
9 305 12 3.79 16.9840 3.45
10 307 10 3.15 16.9420 3.19
11 331 -14 -4.42 15.7418 -4.12
12 326 -9 -2.84 16.1385 -1.70

Table 4 - Summary of Drag Reduction an L/D Improvements of VCCTEF (1 Drag Count = 0.0001)

Figure 16 shows the lift coefficient sensitivity to the dynamic pressure. As thy dynamic pressure increases, the lift coefficient
decreases. By extrapolation of the data to q• = 0, one can estimate the rigid wing lift coefficient. Both linear and cubic
polynomial extrapolation were used to correct the flexible wing lift curves to obtain the rigid wing lift curves, as shown in Fig.
17. The cubic polynomial method is supposed to yield more accurate estimates according to Eq. (30).

Fig. 16 - FLAP0 Cubic and Linear Variations of CL with q• Fig. 17 - FLAP0 Runs 18 - 20, 22 - 24, 70 - 72, 104 CL of
Flexible and Rigid Wing

The effect of aeroelasticity on the lift coefficient is also analyzed. The measured wing bending deflection, bending slope,
and torsional deflection about the elastic axis of FLAP0 configuration are shown in Fig. 18, 19, and 20.

15 of 28

American Institute of Aeronautics and Astronautics

 Fig. 18 - FLAP0 Bending Deflection Fig. 19 - FLAP0 Bending Slope about Pitch Axis

Fig. 20 - FLAP0 Torsional Deflection

The rigid wing lift coefficient can be estimated from the static bending and torsional deflections of the wing. The wing
aeroelastic deflection results in an effective change in the angle of attack, which is computed as13
b/2 
1 dW (ȳ)
ˆ
Da = c⇤La (y) Q (ȳ) cos L + w sin L c (y) dy (37)
CL⇤a S b/2 d ȳ

where c⇤La (y) is the sectional lift curve slope for the rigid wing.
This effective change in the angle of attack is a function of the angle of attack as

∂ Da
Da = Da0 + a (38)
∂a
By making an assumption that c⇤La (y) = CL⇤a , the rigid wing lift coefficient can be computed from the flexible wing lift
coefficient as13
CLa Da
CL⇤ = CL (39)
1 + ∂∂Da
a
Figure 21 shows the rigid wing lift coefficients estimated by the aeroelastic deflection correction method and the two q• -
correction methods as compared to the flexible wing lift coefficient.13 The aeroelastic deflection correction method yields the
highest estimated rigid-wing lift coefficient. Since there are some differences in the three estimated rigid wing lift coefficients,
an average rigid wing lift coefficient is computed and is shown in Fig. 21.

16 of 28

American Institute of Aeronautics and Astronautics

 Both the aeroelastic deflection correction method and the q• -correction method exhibit some degree of uncertainty. For
the aeroelastic deflection correction method, the assumption of the sectional lift curve slope being the same as the wing lift
curve slope is made. This assumption needs to be verified by aerodynamic analysis using CFD. Also, the aeroelastic deflection
measurements have uncertainty associated with the accuracy of the VICON system. For the q• -correction method, data extrap-
olation can result in less robust estimation if there are not sufficient data to establish the confidence. Figure 16 shows a good
correlation in the lift coefficient as a function of the dynamic pressure that is consistent with the aeroelastic analysis. Thus, this
establishes a degree of confidence in the q• -correction method. Also, depending on the type of regressors such as the cubic and
linear functions, the results can differ. Nonetheless, in the estimation, both the linear and cubic functions yield similar results.

Fig. 21 - FLAP0 Runs 18 - 20, 22 - 24, 70 - 72, 104 CL of Flexible and Rigid Wing

VI. Aeroelastic Deflection Measurements and Wing Stiffness Estimation

Fig. 22 - Wind Tunnel Model with VICON Dots in UWAL Test Section (Courtesy of University of Washington)

Aeroelastic deflection measurements were taken with the VICON motion tracking system. The VICON system uses 54
optical targets, called dots, to create a three-dimensional displacement of the wind tunnel model as it is loaded. Figure 22 is
a photograph of the wind tunnel model with the VICON dots. The displacement measurements at these dots are reduced to
the spanwise out-of-plane deflections and twists at six spanwise locations, referred to as Chord 1, Chord 2, Chord 3, Chord

17 of 28

American Institute of Aeronautics and Astronautics

4, Chord 5, and Chord 6, as shown in Fig. 23. The deflection measurement locations are referenced to the origin of the wind
tunnel model which starts at the intersection of the wing trailing edge and the center body.

Fig. 23 - VICON Target and Spanwise Deflection Measurement Locations

A series of static load and modal analysis tests were performed prior to the wind tunnel test runs. The static load tests were
conducted by applied concentrated loads at two wing span stations y = 60.75 inches and y = 25.75 inches and concentrated
torques at two wing span stations y = 60.75 inches and y = 35.25 inches. The deflections and twist about the wing pitch axis
y at the six measurement stations were measured by the VICON systems. Figures 24 to 27 are the plots of the deflections and
twists measured from the static load tests. The twist measurements at station y = 35.25 inches were not shown since the data
were taken with only one torque input of 95 in-lb and exhibit a considerable scatter. Figure 27 shows that the static load test in
torsion is not in pure torsion as the measured deflections under the applied torque are not zero.

Static Load Test Force Input at Station y = 60.75 in Static Load Test Force Input at Station y = 25.75 in
6 1.4
Chord 1 Chord 1
Chord 2 Chord 2
Chord 3 1.2 Chord 3
5
Chord 4 Chord 4
Chord 5 Chord 5
Chord 6 1 Chord 6
4
Deflection (in)

Deflection (in)

0.8
3
0.6

2
0.4

1
0.2

0 0
2 4 6 8 10 12 14 16 5 10 15 20 25 30 35 40
Force (lb) Force (lb)

Fig. 24 - Static Deflection Measurements for Force Input @ Fig. 25 - Static Deflection Measurements for Force Input @
y = 60.75 inches y = 25.75 inches

18 of 28

American Institute of Aeronautics and Astronautics

 Static Load Test Torque Input at Station y = 60.75 in Static Load Test Torque Input at Station y = 60.75 in
2 0.45
Chord 1 Chord 1
1.8 Chord 2 0.4 Chord 2
Chord 3 Chord 3
1.6 Chord 4 Chord 4
Chord 5 0.35 Chord 5
1.4 Chord 6 Chord 6
0.3

Deflection (in)
1.2
Twist (deg)

0.25
1
0.2
0.8
0.15
0.6
0.1
0.4

0.2 0.05

0 0
15 20 25 30 35 15 20 25 30 35
Torque (in−lb) Torque (in−lb)

Fig. 26 - Static Twist Measurements for Torque Input @ Fig. 27 - Static Deflection Measurements for Torque Input
y = 60.75 inches @ y = 60.75 inches

The data from the static load and modal analysis tests were used to validate a 3D NASTRAN static and modal analysis
models built by UWAL.6 A stick NASTRAN static model was then constructed by UWAL to provide the equivalent bending
and torsion stiffnesses of the wing. The stick NASTRAN model represents the wing as a series of beam sections at varying
sweep angles to account for changes in the wing stiffness across the wing span, as shown in Fig. 28. Figures 29 and 30 are
the plots of the bending and torsion stiffnesses for the UWAL stick NASTRAN model. The wing is significant stiffer in torsion
than in bending as can be seen from Figs. 29 and 30. The torsional stiffness at the wing tip is observed to be very large. This
could be due to structural reinforcements and flap support structures at the wing tip.
Both the 3D NASTRAN and stick NASTRAN models were provided to NASA Ames Research Center by UWAL. In
addition, UWAL also provided the static deflections computed by the 3D NASTRAN model. NASA Ames Research Center
later ran the UWAL stick NASTRAN model to obtain the static deflections. Both the static deflections computed by the stick
NASTRAN model and 3D NASTRAN were then compared against the static deflection measurements, as shown in Figs.
31 and 32. The static deflections computed by the 3D NASTRAN are in very good agreement with the static deflection
measurements. However, the static deflections computed by the stick NASTRAN model have poor agreement with the static
deflection measurements for the applied force at station y = 25.75 inches. The agreement for the applied force at the other
station is reasonably good. This is a result of the UWAL stick model generated, as a preliminary model, with more weight in
the stiffness fitting placed on outboard wing behavior of static aeroelastic and dynamic modal importance, including capturing
local “jumps” in stiffness in areas of backbone structure, stiff edges, and hard attachment points, as well as geometrical breaks
in the wing (see Figure 8).

UWAL Stick NASTRAN Model Coordinates

6

5.5

4.5

3.5
z (ft)

2.5

1.5

2 3 4 5 6 7 8
x (ft)

Fig. 28 - Nodal Coordinates of UWAL Stick NASTRAN Model (z = 0 from Tunnel Floor)

19 of 28

American Institute of Aeronautics and Astronautics

 4
x 10 UWAL Stick NASTRAN Bending Stiffness 5
x 10 UWAL Stick NASTRAN Torsional Stiffness
2.5 3.5

3
2

2.5

1.5

GJ (lb−ft2)
EI (lb−ft2)

1.5
1

0.5
0.5

0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
y (ft) y (ft)

Fig. 29 - UWAL Stick NASTRAN Bending Stiffness Fig. 30 - UWAL Stick Model Torsional Stiffness

Static Load Test Force Input at Station y = 60.75 in Static Load Test Force Input at Station y = 25.75 in
10 1.4
Chord 3 Measured Chord 3 Measured
9 Chord 4 Measured Chord 4 Measured
Chord 5 Measured 1.2 Chord 5 Measured
8 Chord 6 Measured Chord 6 Measured
Chord 3 3D Chord 3 3D
7 Chord 4 3D 1 Chord 4 3D
Chord 5 3D Chord 5 3D
Chord 6 3D Chord 6 3D
Deflection (in)

Deflection (in)

6
Chord 3 Stick 0.8 Chord 3 Stick
5 Chord 4 Stick Chord 4 Stick
Chord 5 Stick Chord 5 Stick
Chord 6 Stick 0.6 Chord 6 Stick
4

3 0.4

2
0.2
1

0 0
2 4 6 8 10 12 14 16 5 10 15 20 25 30 35 40
Force (lb) Force (lb)

Fig. 31 - NASTRAN Static Deflections for Force Input @ Fig. 32 - NASTRAN Static Deflections for Force Input @
y = 60.75 inches y = 25.75 inches
In order to work with a stick model that would capture static behavior well along inboard as well as outboard sections of
the wing and that would not have the stiffness discontinuities of Fig. 30, a smooth continuous stiffness stick model was sought.
The procedure used was based on an optimization problem solution that would minimize displacement errors for the loaded
wing between the new stick model and the detailed 3D NASTRAN model.
The static deflection is computed using the Galerkin’s method by assuming the following solution:

Q (ȳ) = Y (ȳ) q (40)

W (ȳ) = F (ȳ) w (41)
h i> h i>
where q = q1 . . . qn and w = w1 . . . wn are vectors of generalized coordinates, and
h i h i
Y (ȳ) = y1 (ȳ) . . . yn (ȳ) and F (ȳ) = f1 (ȳ) . . . fn (ȳ) are shape functions selected to be the natural mode
shapes of a uniform cantilever beam which are given by
p (2i 1) p ȳ
yi (ȳ) = 2 sin (42)
2L
cosh (bi L) + cos (bi L)
fi (ȳ) = cosh (bi ȳ) cos (bi ȳ) [sinh (bi ȳ) sin (bi ȳ)] (43)
sinh (bi L) + sin (bi L)

20 of 28

American Institute of Aeronautics and Astronautics

where bi L are the eigenvalues corresponding to their natural mode shapes of beam bending.
Then the bending and torsional deflections are computed as

q = Kq 1 Fq (44)
1
w = Kw Fw (45)

where Kq and Kw are the torsional and bending generalized stiffnesses, respectively, and Fq and Fw are the torsional and bending
generalized forces, respectively, which are defined as
ˆ L
dY> (ȳ) dY (ȳ)
Kq = GJ d ȳ (46)
0 d ȳ d ȳ
ˆ L
d 2 F> (ȳ) d 2 F (ȳ)
Kw = EI d ȳ (47)
0 d ȳ2 d ȳ2
Fq = Y> (ȳM ) M (48)
Fw = F (ȳP ) P>
(49)

where M and P are the concentrated applied torque at ȳ = ȳM and applied force at ȳ = ȳP .
For the estimation of the wing stiffness, the static deflections are all normalized to a unit value at the Chord 6 station.
Thus, the applied forces and applied torque are computed to correspond to these normalized static deflections. Furthermore,
the torsional deflections about the elastic axis ȳ are computed from the twist measurements are with respect to the wing pitch
axis y and the bending deflection measurements according to
ji
Qi = Wi tan L (50)
cos L
where ji is the measured twist about the wing pitch axis y at the Chord i station.
A linear regression is performed to compute the normalized static deflections and applied forces and torque as follows:
✓ ◆
DQ
Q̄i = M̄ + DQi (51)
DM i
✓ ◆
DW
W̄i = P̄ + DWi (52)
DP i
1 DQ6
M̄ = DQ (53)
DM i
1 DW6
P̄ = DW
(54)
DP 6

DQ
where DM i
, DW
DP i , DWi and DQi are obtained from the linear regression of the static deflection measurements as functions of
the applied forces and torque.
The optimization is set up to minimize two cost functions to estimate the bending and torsional stiffnesses as follows:
6 6
2 2
JW = Â Ŵi W̄i + Â Ŵi W̄i (55)
i=1 P̄@y=60.75 i=1 P̄@y=25.75
6
2
JQ = Â Q̂i Q̄i (56)
i=1 M̄@y=60.75

A cubic polynomial shape function is chosen to be a design shape function. This shape function is applied as a multiplier
to the baseline UWAL bending stiffness in the stick NASTRAN model. Thus, the design variables are the four polynomial
coefficients of the shape function. A gradient search is conducted to identify the optimal shape functions that minimize the
cost functions for the bending and torsional deflections. Because of the strange behavior in the UWAL torsional stiffness as
represented in the stick NASTRAN model, the optimization does not converge properly. As a result, it is decided that the
baseline torsional stiffness in the optimization assumes the same spatial distribution as the bending stiffness. The optimization
then converges to the bending and torsional stiffnesses, as shown in Figs. 33 and 34. The ratio of the torsional stiffness to
the bending stiffness is about 3:1, which indicates that the UWAL wind tunnel model is significantly stiffer in torsion than in
bending. This is consistent with the aeroelastic deflection measurements. The optimization also shows that the optimization
results depend on the initialization. Consequently, variations in the estimated stiffness are expected.

21 of 28

American Institute of Aeronautics and Astronautics

 4 5
x 10 x 10
7 2.5

6
2

1.5

GJ (lb−ft )
EI (lb−ft2)

2
3
1

0.5
1

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
y (ft) y (ft)

Fig. 33 - Optimized Bending Stiffness of UWAL Flexible Fig. 34 - Optimized Torsional Stiffness of UWAL Flexible
Wing Wind Tunnel Model Wing Wind Tunnel Model

Estimated Deflection for 2.68 lb Force Applied at Station y = 60.75 in Estimated Deflection for 28.96 lb Applied Force at Station y = 25.75 in
1 1.4
Measured Normalized Measured Normalized
0.9 Computed Computed
1.2
0.8

0.7 1

0.6
0.8
W (in)

W (in)

0.5
0.6
0.4

0.3 0.4

0.2
0.2
0.1

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
y (ft) y (ft)

Fig. 35 - Computed vs. Measured Normalized Static Fig. 36 - Computed vs. Measured Normalized Static
Deflections for Force Input @ y = 60.75 inches Deflections for Force Input @ y = 25.75 inches

Estimated Elastic Axis Twist for 41.21 in−lb Torque Applied at Station y = 60.75 in
1.2
Measured Normalized
Computed
1

0.8

0.6
Θ (deg)

0.4

0.2

−0.2
0 1 2 3 4 5 6
y (ft)

Fig. 37 - Computed vs. Measured Normalized Static Torsional Deflections for Torque Input @ y = 60.75 inches

22 of 28

American Institute of Aeronautics and Astronautics

 Figures 35 to 37 present the comparisons between the computed static deflections against the normalized static deflections
computed directly from the measured static deflections. The agreements between the computed and the measured normalized
static deflections are excellent
Using the optimized stiffnesses, a coupled aerodynamic-structural modeling approach is conducted using the finite-element
method coupled with the vortex-lattice method to predict the aerodynamic performance of the UWAL flexible wing wind tunnel
model.

VII. Coupled Aerodynamic-Structural Modeling

To establish the aerodynamic analysis, a vortex-lattice aerodynamic solver, Vorlax, is used to compute the aerodynamic
characteristics of the UWAL flexible wing wind tunnel model. Vorlax is an aerodynamic performance prediction code, devel-
oped by Miranda et al,17 that provides a rapid method for estimating force and moment coefficients as well as stability and
control derivatives of an aerodynamic configuration. An aerodynamic configuration is constructed within Vorlax by a series of
lifting panels that are formed by spanwise and chordwise locations of vortices based on the locations of the computing stations
specified by the user. Vorlax computes the vehicle aerodynamics in both the longitudinal and lateral directions independently.
The longitudinal and lateral computational results are then combined to produce overall aerodynamic characteristics of an aero-
dynamic configuration at any arbitrary angles of attack and sideslip. It is based on the vortex-lattice method. Vorview is a
graphical user interface front end that allows users to create a model and execute Vorlax interactively or in a batch mode.
Because the vortex-lattice method is based on potential flow theory, it is an inviscid code that can only compute inviscid drag
due to lift. It cannot predict viscous or wave drag. The compressibility effect on the aerodynamic coefficients is corrected by the
Prandtl-Glauert compressibility correction. For lift prediction at low angle of attack, Vorlax can provide reasonably accurate
prediction. Figure 38 shows a comparison between the lift coefficient predicted by Vorlax and the measured lift coefficient
of a sub-scale 5.5% wind tunnel model of the GTM tested in NASA Langley Research Center’s 14-Foot-By-22-Foot Wind
Tunnel.15 The agreement between the Vorlax lift prediction and test data is excellent.

Lift Coefficient of GTM

1
o
VORLAX β = 0
VORLAX β = 20o
14−Ft x 22−Ft β = 0o
14−Ft x 22−Ft β = 20o

0.5
L
C

−0.5
−10 −5 0 5 10 15
α (deg)

Fig. 38 - Lift Curve for NASA GTM

The UWAL flexible wing wind tunnel constructed in Vorlax is shown in Fig. 39. The model has the wing and fuselage
components. The wing component is defined by a series of wing sections in the directions parallel to the aircraft axis and parallel
to the edges of the VCCTEF. The fuselage component is defined by a series of closed sections in the y z plane of the aircraft.
A geometry tool, written in MATLAB, provides the ability to model the VCCTEF by deforming the trailing edge portion of the
wing panels on which the VCCTEF is defined. The geometry tool creates the rotation of each of the three camber segments
by a specified flap deflection. The trailing edge portion of the wing section that includes the VCCTEF thus is modified by the
rotation of the three camber segments. Furthermore, the geometry tool also provides the ability to deform the wing geometry on
the flight in all three directions: flapwise bending, chordwise bending, and twist. The deformation capability of the geometry
tool allows a finite-element model to couple directly to Vorview for coupled aerodynamic-structural modeling.16

23 of 28

American Institute of Aeronautics and Astronautics

 Fig. 39 - UWAL Wind Tunnel Model in Vorview

The UWAL model is set up with the panels sub-divided into 2000 sub-polygons. The flat plate option is set to off to properly
model the camber lifting surfaces of the model. Vorlax execution in the batch mode is controlled via a MATLAB script that
handles the interfaces with the geometry tool and a finite-element code, also written in MATLAB. The finite-element code is
fully capable of performing static and dynamic aeroelasticity using three dimensional beam elements to model wing structures.
The detail development of the aeroelastic finite-element modeling is described in a paper by Nguyen, Ting, et al.
Vorlax computes aerodynamic forces and moments which are then transferred to the finite-element model (FEM). The finite-
element code then computes the wing aeroelastic deflections which are then passed to the geometry tool. The Vorlax geometry
file .hrm is then directly modified by the geometry tool which changes the geometry to model the aeroelastic deflections. The
new geometry file is then read by Vorview which calls Vorlax to execute the computation. This iterative process continues until
a convergence criterion is met.
The lift and drag characteristics of the rigid wing reference geometry from which the wind tunnel model is constructed are
first computed by Vorlax without structural coupling. Figure 40 is the lift curve of the rigid baseline FLAP0 configuration.
For comparison, the rigid wing lift curves of the baseline FLAP0 configuration computed by Euler CFD code Cart3D10 and
estimated from the UWAL wind tunnel test data by the dynamic pressure and the aeroelastic deflection correction methods13
are also plotted. It can be seen that Vorlax under-predicts lift for the rigid wing UWAL wind tunnel model by some amount.
The lift prediction of FLAP0 configuration by Vorlax is still reasonably in good agreement with the dynamic pressure corrected
wind tunnel test data.
FLAP0 Rigid Wing CL
0.8
Vorlax
Cart3D
0.7 UWAL FLAP0 q∞−Corrected CL
UWAL FLAP0 Deflection−Corrected CL
0.6

0.5
CL

0.4

0.3

0.2

0.1
0 1 2 3 4 5 6
α (deg)

Fig. 40 - Rigid Wing FLAP0 Lift Curves

24 of 28

American Institute of Aeronautics and Astronautics

 The VCCTEF segment angles during the test were set using stiff aluminum tape and measured carefully before and after
each wind-on run to identify any drift or deviation from the desired angles. Aluminum tape hinge stiffness was measured, and,
given the low loads on the VCCTEF segments, it was expected that static aeroelastic deviation from the pre-set angles during
wind-on conditions would be negligible. While there were VICON reflective dots on all VCCTEF segments, the size of the dots
and the small distances chordwise between the dots made VICON measurements of the VCCTEF segment angles unreliable.
In follow-up tests of a high-lift VCCTEF wing configuration, the VCCTEF angle setting design was modified to remove any
uncertainty in the VCCTEF segment angles.17 Nevertheless, studies were carried out to assess the influence of some uncertainty
in the VCCTEF segment angle settings on the test results for the cruise configuration.6 For the present study, uncertainty based
on the VCCTEF measured angles was nevertheless used as an upper bound on the uncertainties expected. Table 5 summarizes
the VCCTEF deflection measurements for the baseline FLAP0 configuration. These measurements indicate variations in the
VCCTEF deflection with a mean of about 1 degree.

Run 1A 1B 1C 2A 2B 2C 3A 3B 3C 4A 4B 4C 5A 5B 5C
18 -0.1 0.0 -0.3 0.4 -1.1 2.1 -0.2 -1.1 2.6 -1.9 2.9 -2.1 0.6 -0.9 1.5
19 0.3 0.0 0.2 0.0 0.4 0.8 1.0 -2.4 4.7 -0.1 1.2 -1.2 -0.3 1.4 -0.1
20 0.7 0.1 0.3 0.4 -0.2 1.7 0.5 -0.9 3.4 0.4 1.1 -0.7 0.2 1.1 -0.8
21 0.0 0.1 0.2 -0.1 0.2 0.1 -0.2 -0.7 3.4 -0.7 0.9 -0.5 0.6 0.4 1.2
22 0.5 -0.3 0.4 0.4 -0.6 0.7 -0.5 -0.1 2.2 -0.8 0.6 -0.1 1.6 -1.5 3.7
23 1.7 0.5 1.1 1.6 0.4 2.3 1.6 -1.0 5.4 2.4 0.2 2.2 2.7 -1.6 3.8
24 1.9 0.6 1.2 1.9 0.1 3.3 2.5 -1.2 5.0 2.1 1.8 0.1 1.7 -0.7 4.1
69 0.3 0.2 -0.5 0.7 -0.7 0.8 0.4 -0.4 1.1 -0.6 1.1 -0.9 1.3 -0.8 -1.5
70 1.0 0.0 0.9 0.7 0.8 0.4 1.0 -1.2 3.9 0.0 1.8 -0.9 1.7 -2.1 1.7
71 1.5 0.5 0.7 1.7 0.3 1.7 1.2 0.6 1.5 1.4 0.6 1.4 1.5 0.3 -0.3
72 2.2 0.9 0.6 3.0 0.0 2.6 1.8 1.7 0.4 2.8 1.4 0.5 2.1 -0.8 1.8
Mean 0.91 0.24 0.44 0.97 -0.04 1.50 0.83 -0.61 3.05 0.45 1.24 -0.20 1.25 -0.47 1.37

Table 5 - VCCTEF Deflection Measured by VICON System for FLAP0 Configuration

The lift prediction taking into account of the uncertainty in the indicated VCCTEF deflections is shown in Fig. 41. It is
interesting to note that the variations in the indicated VCCTEF deflection are uni-directional in that they result in a higher lift
prediction than the baseline lift coefficient for the reference FLAP0 VCCTEF deflection. The estimated rigid wing lift curves
fall within the uncertainty band of the lift prediction. It is observed that the lift prediction for the mean indicated VCCTEF
deflection agrees very well with the rigid wing lift estimated by the aeroelastic deflection correction method. The spread in the
lift prediction based on the indicated VCCTEF deflections is rather large. The experimental lift data, however, shows a high
degree of correlation. This suggests that the actual variations in the VCCTEF deflection may be smaller than as indicated by
the VICON measurements. Without the knowledge of the true VCCTEF deflections, it may be difficult to assess the accuracy
of the tool for lift prediction. Given that any uncertainty in the VCCTEF deflection will result in a constant offset in the lift
coefficient, therefore this suggests that the lift curve slope may be a better indication for comparing the lift prediction. The
predicted rigid wing lift curve slope is calculated to be CL⇤a = 4.4401 for the baseline FLAP0 configuration which agrees well
with the estimated rigid wing lift curve slopes of CL⇤a = 4.5533 and CL⇤a = 4.7488 by the dynamic pressure and aeroelastic
deflection correction methods, respectively. The difference ranges from 2.5% to 6.5%.

25 of 28

American Institute of Aeronautics and Astronautics

 FLAP0 Rigid Wing C w/ VCCTEF Variations
L
0.8
Reference VCCTEF
UWAL FLAP0 q∞−Corrected CL
0.7 UWAL FLAP0 Deflection−Corrected CL
Mean Measured VCCTEF
0.6 Measured VCCTEF Variations

0.5

CL
0.4

0.3

0.2

0.1
0 1 2 3 4 5 6
α (deg)

Fig. 41 - Rigid Wing FLAP0 Lift Curves for Variations in VCCTEF Deflection

The lift prediction of the flexible wing UWAL wind tunnel model is computed next using the coupled Vorlax-FEM. Figure
42 shows the lift prediction of the flexible wing UWAL wind tunnel model by Vorlax for the reference FLAP0 VCCTEF and
the mean VCCTEF deflections. The lift curve for the mean VCCTEF deflection agrees very well with the measured lift curve in
the wind tunnel. The predicted lift curve slopes of CLa = 4.0229 and CLa = 4.0267 for the reference FLAP0 VCCTEF and the
mean VCCTEF deflections, respectively, are in excellent agreement with the measured lift curve slope of CLa = 4.1482 with
errors of about 3%.
FLAP0 Flexible Wing CL
0.65
UWAL FLAP0
0.6 Reference VCCTEF
Mean Measured VCCTEF
0.55

0.5

0.45
EI (lb−ft2)

0.4

0.35

0.3

0.25

0.2

0 1 2 3 4 5 6
y (ft)

Fig. 42 - Flexible Wing FLAP0 Lift Curves

Figures 43 shows the bending deflections computed for the reference geometry of FLAP0 configuration with and without
the mean VCCTEF deflection as compared to the measured bending deflection in the UWAL for an angle of attack of 6o . The
computed bending deflections in general agree well with the measured bending deflection. The bending deflection at the wing
tip for the baseline FLAP0 configuration matches the measured value. The wing tip bending deflection at an angle of attack
of 6o is about 11% of the wing semi-span. This is similar to the flexibility of modern transport wings. Figure 44 is the plot of
the wash-out twist about the wing pitch axis due to the wash-out effect of wing bending and the torsional deflection about the
elastic axis. The aeroelastic deflection measurements in UWAL indicate that the wash-out twist is primarily due to the bending
deflection, while torsional deflection contributes very little to the wash-out twist because the wing is significantly stiffer in
torsion than in bending. The computed wash-out twist for the baseline FLAP0 configuration shows a good agreement with the
measured wash-out twist at the wing tip, but exhibits a greater discrepancy along the wing span by as much as 0.7o . It should be
noted that the wash-out twist measurements by the VICON are considered to be less accurate than the wing bending deflection
measurements.

26 of 28

American Institute of Aeronautics and Astronautics

 o o
FLAP0 Bending Deflection at α = 6 FLAP0 Bending Wash−Out Twist (Positive Nose−Down) atα = 6
9 4
UWAL FLAP0
8 Reference VCCTEF
Mean Measured VCCTEF 3.5

7
3

dW/dysinΛ (deg)
2.5
5
W (in)

2
4
1.5
3

1
2

0.5 UWAL FLAP0

1 Reference VCCTEF
Mean Measured VCCTEF
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
y (in) y (in)

Fig. 43 - Flexible Wing FLAP0 Bending Deflection for Fig. 44 - Flexible Wing FLAP0 Bending Wash-Out Twist
a = 6o (Positive Nose-Down) about Wing Pitch Axis for a = 6o

VIII. Conclusion
This paper presents an aeroelastic analysis of a flexible wing wind tunnel model with a variable camber continuous trailing
edge flap (VCCTEF) design for drag minimization tested at the University of Washington Aeronautical Laboratory (UWAL).
The wind tunnel test was designed to explore the relative merit of the VCCTEF concept for improved cruise efficiency. The
VCCTEF is a multi-segment flap design having three chordwise camber segments and five spanwise flap sections for a total
of 15 flap segments. Elastomeric material is used to cover the gaps in between the spanwise flap sections, thereby creating
a continuous trailing edge. The flexible wing model is a 10%-scale model of a typical transport wing. The wing structural
stiffness in bending is tailored to achieve a wing tip deflection of about 10% of the wing semi-span. A method for estimating the
bending and torsional stiffnesses of the flexible wing UWAL wind tunnel model from static load test data has been developed.
The resulting estimation indicates that the stiffness of the flexible wing is significantly stiffer in torsion than in bending by as
much as 3 to 1. Variations in the estimated stiffness can result depending on the initialization of the optimization. This can
result in uncertainty which is not included in the analysis. The lift prediction for the flexible wing is computed by a coupled
aerodynamic-structural model. The coupled model is developed by coupling a conceptual aerodynamic tool Vorlax with a
finite-element model of the flexible wing via an automated geometry deformation tool. Based on the comparison of the lift
curve slope, the lift prediction for the rigid wing is in good agreement with the estimated lift coefficients derived from the wind
tunnel test data. Due to the movement of the VCCTEF during the wind tunnel test, uncertainty in the lift prediction due to
the indicated variations of the VCCTEF deflection is studied. The results show a significant spread in the lift prediction which
contradicts the consistency in the aerodynamic measurements, thus suggesting that the indicated variations as measured by the
VICON system may not be reliable. The lift prediction of the flexible wing agrees well with the measured lift curve for the
baseline configuration. The computed bending deflection and wash-out twist of the flexible wing also match reasonably well
with the aeroelastic deflection measurements. The results demonstrate the validity of the aerodynamic-structural tool for use to
analyze aerodynamic performance of flexible wings.

Acknowledgment
The authors would like to thank the Fixed Wing / Advanced Air Transport Technology Project under the Fundamental
Aeronautics Program of NASA Aeronautics Research Mission Directorate (ARMD) for funding support of this work. The
authors also would like to acknowledge Boeing Research and Technology and University of Washington for their collaboration
with NASA under NASA contract NNL12AD09T entitled "Development of Variable Camber Continuous Trailing Edge Flap
System for B757 Configured with a More Flexible Wing."

References
1 Nguyen, N., “Elastically Shaped Future Air Vehicle Concept,” NASA Innovation Fund Award 2010 Report, October 2010, Submitted to NASA Innova-

tive Partnerships Program.

2 Nguyen, N., Trinh, K., Reynolds, K., Kless, J., Aftosmis, M., Urnes, J., and Ippolito, C., “Elastically Shaped Wing Optimization and Aircraft Concept

27 of 28

American Institute of Aeronautics and Astronautics

for Improved Cruise Efficiency,” AIAA Aerospace Sciences Meeting, AIAA-2013-0141, Grapevine, TX, January 2013.
3 Boeing Report No. 2012X0015, “Development of Variable Camber Continuous Trailing Edge Flap System,” October 4, 2012.
4 Urnes, J., Nguyen, N., Ippolito, C., Totah, J., Trinh, K., and Ting, E., “A Mission Adaptive Variable Camber Flap Control System to Optimize High Lift

and Cruise Lift to Drag Ratios of Future N+3 Transport Aircraft,” AIAA Aerospace Sciences Meeting, AIAA-2013-0214, Grapevine, TX, January 2013.
5 Jordan, T. L., Langford, W. M., Belcastro, C. M., Foster, J. M., Shah, G. H., Howland, G., and Kidd, R., “Development of a Dynamically Scaled Generic

Transport Model Testbed for Flight Research Experiments,” AUVSI Unmanned Unlimited, Arlington, VA, 2004.
6 Precup, N., Mor, M., and Livne, E., “Design, Construction, and Tests of an Aeroelastic Wind Tunnel Model of a Variable Camber Continuous Trailing

Edge Flap (VCCTEF) Concept Wing,” 32nd AIAA Applied Aerodynamics Conference, 2014-2442, Atlanta, GA, June 2014, doi: 10.2514/6.2014-2442.
7 Bisplinghoff, R. L, Ashley, H., and Halfman, R. L, Aeroelasticity, Addison-Wesley, 1955, pp. 478 - 480.
8 Anderson, J. D., Fundamentals of Aerodynamics, McGraw-Hill, 2001, pp. 289 - 310.
9 Hodges, D. and Pierce, G., Introduction to Structural Dynamics and Aeroelasticity, Cambridge University Press, 2002.
10 Rodriguez, D., Aftosmis, M., Nemec, M., and Smith, S., “Static Aeroelastic Analysis with an Inviscid Cartesian Method,” AIAA Science and Technol-

ogy Forum, AIAA-2014-0836, National Harbor, MD, January 2014.

11 Nguyen, N., Ting, E., Nguyen, D., Dao, T., and Trinh, K., “Coupled Vortex-Lattice Flight Dynamic Model with Aeroelastic Finite-Element Model of

Flexible Wing Transport Aircraft with Variable Camber Continuous Trailing Edge Flap for Drag Reduction,” AIAA Atmospheric Flight Mechanics, AIAA-
2013-4746, Boston, MA, August 2013.
12 Ting, E., Nguyen, N., and Trinh, K., “Static Aeroelastic and Longitudinal Trim Model of Flexible Wing Aircraft Using Finite-Element Vortex-Lattice

Coupled Solution,” 55th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, AIAA-2014-0837, National Harbor, MD,
January 2014.
13 Nguyen, N., Precup, N., Urnes, J., Nelson, C., Lebofsky, S., Ting, E., and Livne, E., “Experimental Investigation of a Flexible Wing with a Variable

Camber Continuous Trailing Edge Flap Design,” 32nd AIAA Applied Aerodynamics Conference, AIAA 2014-2441, Atlanta, GA, June 2014.
14 Miranda, L.R., Elliot, R.D., and Baker, W.M., “A Generalized Vortex Lattice Method for Subsonic and Supersonic Flow Applications,” NASA CR-2865,

1977.
15 Nguyen, N. T., Nelson, A., and Pulliam, T., “Damage Adaptive Control System Research Report,” Internal NASA Report, April 2006.
16 Ting, E., Nguyen, N., and Lebofsky, S., “Static Aeroelastic Modeling of a Sub-Scale Wind Tunnel Model with Novel Flap Concept,” 56th

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, AIAA-2015-1407, Kissimmee, FL, January 2015.
17 Precup, N., Mor, M., and Livne, E., "The Design, Construction, and Tests of a Concept Aeroelastic Wind Tunnel Model of a High-Lift Variable Camber

Continuous Trailing Edge Flap (HL-VCCTEF) Wing Configuration", 56th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials
Conference, AIAA 2015-1406, Kissimmee, FL, January 2015, doi: 10.2514/6.2015-1406.

28 of 28

American Institute of Aeronautics and Astronautics