Wind Tunnel Testing of a Twisted Wing for Longitudinal Control in a Joined-Wing Aircraft

8/18/2019 Wind Tunnel Testing of a Twisted Wing for Longitudinal Control in a Joined-Wing Aircraft

1/18


2/18

American Institute of Aeronautics and Astronautics

2

CFD = computational fluid dynamics

D = dragF x = X component of the resultant pressure force acting on the vehicleF z = Z component of the resultant pressure force acting on the vehicle

FEM = finite element model L = lift

M = mass primary quantity (with subscript) M x = X component of moment acting on the vehicle M y = Y component of the moment acting on the vehicle

M z = Z component of the moment acting on the vehicle

MAC = mean aerodynamic chordOML = outer mold line

Re = Reynolds numberq∞ = dynamic pressure

p = total pressureS mod = wing area for the modelV = velocitySubscripts

w = pertaining to the aircraft (wing)

m = pertaining to the wind tunnel model

I. Introduction

HE Joined-Wing Sensorcraft is unique in two major areas. First, it is an aircraft built around its sensors,

that is, the sensors are built into the composite skin. The second unique aspect of this design is provided

by the joined-wing configuration. In order to maximize the useable wing surface area for the sensors, and

avoid interference with the sensors, control surfaces should be minimized or eliminated. Minimizing the

control surfaces takes advantage of the Wolkovitch effect of the joined-wing configuration.1 Unlike a

conventional planform, wing bending acts in the plane connecting the fore and aft wings. To resist this

bending the wing box structure must maintain a forward spar as far forward and aft spar as far aft as

possible (see

Figure 1). Therefore, conventional control surfaces should be avoided, since there is insufficient chord

available. Additionally, eliminating control surfaces minimizes interference with the sensor array, resulting

in a more effective system. To achieve the longitudinal control objective the composite wing will be subjectedto flexible aft wing twist. Initial experiments were conducted to determine whether the twisted aft wing could

produce the forces and moments required for pitch control. A later study will employ wing twist and

measure the aeroelastic response.

Figure 1. (a) Tilted bending plane of a joined wing and (b) Optimum wing structures. 2

The joined-wing concept was first introduced in 1970s patents by Wolkovitch, who published an overview in

1986.1 More detailed aerodynamic and structural studies by Kroo, Gallman and Smith3 have confirmed the

Wolkovitch effect and defined some characteristics of joined-wing structures that are advantageous to the design.Later, Smith and Kroo continued their research along with Cliff and built a demonstrator joined wing aircraft.4 In

T

(a) (b)


3/18


3

addition to Smith and Kroo, Tyler, Schwabacher and Carter of the US Air Force Research Lab (AFRL) completed

complementary computational fluid dynamics (CFD) and wind tunnel examinations5 of an AFRL generated joined-

wing configuration. Corneille and Franke tested several configurations of joined-wings6 in the wind tunnel at the

Air Force Institute of Technology (AFIT).

II. Theoretical Formulation

A. Aerodynamic Wind Tunnel Testing Theory

Before testing a full-scale article in an uncontrolled environment, a scaled model can be used to gather valuabledata from wind tunnel testing under controlled conditions. Not only does this testing reduce cost and mitigate risk to

aircraft development, it allows the engineer to control many of the variables associated with flight testing an aircraft.

In this way, innovative designs and theories put to practice can be evaluated in a controlled environment with lessuncertainty.

An aerodynamic model that is tested under the same Reynolds and Mach numbers will have the same forces and

moments as the full-scale aircraft.7 It is generally accepted7 that incompressible effects can be neglected below 0.4

M. In addition, with Reynolds numbers above 4105, the oscillatory air forces associated with the Reynolds number

are relatively small.8 Thus, the flutter speed and frequency are relatively unaffected by Reynolds number disparities.

Also, above 1.5106 the boundary layer effects are predictable.7 In this study, Mach and Reynolds numbers could not be matched due to the low speed of the wind tunnel and

model size limitations. Although the Reynolds number effects are arguably insignificant, the pressure data,

discussed next, will help account for any effect that might be experienced.

Figure 2. Pressure at a point along the chord

In addition to directly measuring the forces and moments using a balance, pressure measurements were

collected in the wind tunnel. The pressure forces can be integrated chordwise along the surface of the airfoil to

determine the two-dimensional lift distribution of the airfoil at a specific spanwise location (Figure 2). Both lift anddrag (not to include drag due to shear forces) can be calculated from pressure distribution.7 Integrated static

pressures from pressure ports along the chord of the wing are used in this calculation. For a joined-wing aircraft, the

majority of the lift will come from the forward wing, since it is the largest. The lift produced by the aft wing,

however, is also of interest since this is the surface which will be manipulated to control pitch, as previouslymentioned. Therefore, special attention will be paid to changes in lift due to the aft wing, with various twist angles.

B. Conventional and Wing Twist Stability Derivatives

Conventional aircraft use an elevator or similar control surface on the horizontal tail to change the lift, and in

turn, creating a change in the coefficient of moment at zero angle of attack,0 M

C , and keep the aircraft trimmed at

different speeds. The range of required control surface deflection to provide longitudinal control of the aircraft must

also be determined for the joined-wing aircraft. In this case, however, we wish to determine how much twist isrequired of the aft wing, rather than a control surface deflection, to provide longitudinal control of the aircraft. This

can be accomplished by determining the lift curve slope for the neutral aircraft, 0e

LC

δ α =

∂∂ .


4/18


4

For a conventional aircraft, the lift curve slope of the tail remains constant but shifts to the left as the

elevator, eδ , is deflected downward (positive by convention). Plotting Lt C vs. eδ at a constant angle of attack, the

curve would be linear for most conventional aircraft. This stability derivative, Lt

C

α

∂∂ , is a measure of elevator

effectiveness. 9 For the joined wing, elevator deflection, δe is replaced with aft wing twist, θ. The moment curve can

then be used to measure how well the twist controls the pitch of the aircraft. The moment referred to includes all the

lifting surfaces (fore and aft wing). Like elevator deflection, where down is positive, positive will be designated asaft wing trailing edge down (Figure 3). Since the aft wing will be twisted, it is possible that this joined wing may

not have a constant McgC

α

∂

∂ for various angles of twist like a conventional wing.

Figure 3. Theoretical effect of aft wing twist on moment coefficient.

C. Scaling Laws

There are three primary ratios that must be considered for scaling the aeroelastic characteristics of the full-scale

joined wing. These ratios capture the critical parameters used in scaling the aeroelastic model and are based on physical limitations of a particular test set-up. These ratios include characteristic length ratio, air density ratio andvelocity ratio. The length ratio is established by the size of the full-scale vehicle compared to wind-tunnel

restrictions. Thus, it is defined as discussed in Section III, Approach, as

1

38

m

w

b

b= (1.1)

The air density ratio is fixed by the standard day altitude of the wind tunnel and the mission profile,

2541ft

250Kft

2040.98.3781

243.6

m

w

lb ft

lb ft

ρ ρ

ρ ρ

∞ ∞

∞ ∞

= = = (1.2)

The velocity ratio is fixed by the maximum viable speed of the wind tunnel and the mission profile,

500.2825

177

msm

msw

V

V = = (1.3)

Generally aeroelastic equations impose an aeroelastic mass ratio of unity,

2 3

2 3

1

str

m m m

w str

w w

m M

b b

m M

b b

πρ πρ μ

μ

πρ πρ

∞ ∞

∞ ∞

⎛ ⎞ ⎛ ⎞⎜ ⎟ ⎜ ⎟

⎛ ⎞ ⎝ ⎠ ⎝ ⎠= = =⎜ ⎟⎛ ⎞ ⎛ ⎞⎝ ⎠⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠

(1.4)

θ=0θ (+)

Original CM0

New CM0

α+ α-αe

CMcg

αa

θ(−)

ΔCmcg due to positive θ


5/18


5

where μ is the aeroelastic mass ratio str m is the structural mass per unit length,2

bπρ ∞ is the mass of a characteristic

volume of air above the wing and M is structural mass. Thus, since the air density and length ratios are fixed, (1.4)can be written in terms of a mass ratio7:

3

m m m

w w w

M b

M b

ρ

ρ ∞

⎛ ⎞ ⎛ ⎞= ⎜ ⎟ ⎜ ⎟

⎝ ⎠ ⎝ ⎠

III. Approach

A. Wind Tunnel Model Design and Research Requirements

i.Scaling Requirements

As mentioned in Section II. A, one of the primary considerations for scaling the aerodynamic characteristics of

the full-scale article is matching the Reynolds number. In addition, there are real-world limitations which must be

taken into account, such as the size and speed of the wind tunnel, the instrumentation used, and the ability to

produce an accurate model at the final scale factor.The scale of the test model was determined primarily by the physical constraints of the Gottingen wind tunnel at

the Portuguese Air Force Academy, where the tests were accomplished. The tunnel was used in an open test-section

configuration with a cross section of 1.20.82m. To avoid turbulence, the usable test area must maintain uniform

flow velocity (less than 0.8% in pressure variation), limiting the testable area to 1.10.61.4m. A six-degree-of-

freedom Schenck wind tunnel force balance was used to measure the forces and moments experienced by the model,

which dictated that the wing be mounted vertically. With these constraints, the wind tunnel model was limited to

0.6m for half-span.

ii.Design Requirements

The full-size Joined-Wing Sensorcraft was sized appropriately for a 0.6m half-span, resulting in a 1:38 scale

model. Recent experience with this wind tunnel has shown that this size model will remain outside the shear layer

induced by the wind tunnel at flow velocities up to 50 m/s. The test velocities were approximately 20, 30, 40 and 50m/s, respectively. At these lower speeds, however, it was anticipated that the pressure sensors would be unable torecord valid data. The usable pressure data test conditions were 30, 40 and 50 m/s. As mentioned above, the

highest velocity that will produce flow without disruption from the shear layer induced by the wind tunnel is 50 m/s.

Thus, this was the final speed.With the given constraints, it was not possible to match the Reynolds number in this wind tunnel. External to

this research, a computational fluid dynamics (CFD) model will be used to match the wind tunnel results. Once the

CFD is tuned to the wind tunnel results, the CFD will be used to match the Reynolds number of the full size aircraft.Once this is satisfactorily completed, it will be compared to the current research. To calculate Reynolds number the

following equation was used with forward wing mean aerodynamic chord (MAC) as the characteristic length

ReVc ρ

μ

∞= (1.5)

The values in Table 1 illustrate that the Reynolds number of the wind tunnel testing and the full-scale aircraft aredifferent by two orders of magnitude. As mentioned in the Theoretical Formulation, Section II, the Reynoldsnumber effects are likely insignificant and the pressure data will be used to help account for discrepancies.


6/18


6

Table 1. Reynolds number comparison for wind tunnel and full-scale aircraft

Wind tunnel testing Full-scale aircraft loiter Full-scale aircraft

ingress/egress

Vi (m/s) 50 118 177

c (m) 0.097 3.68 3.68

μ (Pa-s) 1.82e-5 1.82e-5 1.82e-5MAC (m) 0.097 3.68 3.68

ρ (kg/m3) 1.19 1.19 1.19

Re 3.23e5 2.89e7 4.33e7

iii.Measurement Requirements

In designing the wind tunnel model, the measurement requirements were taken into account. Lift, drag and side

force coefficients were measured by the Schenck wind tunnel force balance, which also dictated that the half-spanmodel be mounted vertically (Figure 4). The right half-span was chosen arbitrarily. The following equations were

used to transform the forces

sin cos x z L F F α α = + (1.6)

cos + sin x z D F F α α = − (1.7) where F x and F z were the components of the resultant pressure force acting on the vehicle measured by the Schenck

balance and α is the angle-of-attack.

Figure 4. Wing tunnel model with nominal aft wing.

iv.Twist Tailored Model Design

As mentioned previously, this study made use of an existing design, supplied as a finite-element model (FEM)

by the Sensorcraft program office for the purpose of mitigation risk to their program. The FEM was used to create

the shape of the wind tunnel model, both in the nominal configuration, and with ±15˚ of aft wing twist. Structural

modifications were made to the FEM to enable a twisted aft wing, and then the outer-mold line (OML) of thetwisted configurations was used to create the wind tunnel shapes.

The aft wing twist requirement was set at ±15°. This value was chosen based on calculations for pitch control onthe AFIT-designed wing10, although it is an otherwise arbitrary value for the purpose of this design. In order toreduce the force required to achieve wing twist, simulating a realizable actuator, the aft-wing FEM was modified in

two ways. First, a slit was made spanwise in the skin of the aft wing to allow for a more unrestricted twist (Figure

5). Then the ribs of the FEM were modified from a solid section to a 3-sided rod design to allow for the skin todeform without the restriction of the rib near the location of the slit (Figure 6). These modifications allowed for the

use of a 49,856 N (11,208 lbs) actuator, which is less than that used in a Boeing F-15 Eagle elevator actuator,

y

z

x-α


7/18


7

124,550 N (28,000 lbs).11 The wind tunnel model was not designed to dynamically twist, but to have a fixed twist

built into the rigid model. Therefore, to accomplish the required test objectives, three different wind tunnel model

configurations were required. The OML of the nominal wing (no twist), and resulting wing twisted up and down

15° were given to a computer-aided design (CAD) modeler to make detailed drawings from which the wind tunnelmodels were fabricated.

Figure 5. Aft wing with spanwise slit (white, not to scale).

Figure 6. Modified rib (top), one rib (bottom).

v.Wind Tunnel Model Fabrication

The wind tunnel models were fabricated from foam, balsa wood and fiberglass. Placement of the pressure ports

was also finalized during fabrication and was complicated by the small size of the 1:38 scale model. The leading-edge-down twist is illustrated in Figure 7. This depicts the FEM prediction of the OML of the wing twisted down

15°. Note that the primary area of twist occurs at the root, which is expected, given the location of the actuationforce is a coupled force at the aft wing-tail joint (the arrows indicate the approximate location of the forces in Figure

5.

Figure 7. Aft wing total transition with spanwise slit.

Due to the complexity of the design, it was decided to build just one model with a reconfigurable aft wing. The

nominal aft wing could be replaced by either of the twisted (15° up or down) aft wings (Figure 8).


8/18


8

Figure 8. 0.6m half-span model with nominal, 15 down and 15 up (front to back) twisted aft wings.

B. Wind Tunnel Testing

i.Test Equipment Description

A Schenck force balance was used to measure the forces and moments in the x, y, and z directions. The model

was mounted on a platform which sits on top of the Schenck scale. The platform has a disk cutout such that the

model can be rotated to new angles of attack without adjusting the wind tunnel. The coefficient of pitching moment,CM, coefficient of moment in roll, CL(roll), and yaw, C N from the mount position are shown in equation (1.8).

mod mod mod( )

2 2 2

, , y x z

L roll M N S S S

M M M C C C

q b q c q b∞ ∞ ∞

= = =

(1.8)

Testing was accomplished in the Portuguese Air Force Academy’s Gottingen wind tunnel. The closed circuit

horizontal tunnel was used in the open test-section configuration with a test section of 1.2x0.8x2m, a contraction

ratio of 1:5.53, and a test velocity range from 5 to 70 m/s. The test limitations particular to this study are describedin section i, Scaling Requirements.

Pressure measurements were taken from a single span location on each of the front and aft wings. The location

was chosen halfway between the joint and root on each wing to minimize flow interaction from another surface.

Figure 9 depicts the approximate cross-sectional locations of these ports, while Table 2 lists the exact locations with

respect to the leading edge of the wing.

Figure 9. Pressure port locations of fore and aft wings.


9/18


9

Table 2. Port position relative to the leading edge.

Port number * Fore Wing

position (mm)

Port number* Aft Wing

position (mm)

1/2 5 11/12 3

3/4 16.5 13/14 8.5

5/6 29 15/16 14.5

7/8 41 17/18 209/10 53

*Note: Odd port numbers are on the top of the wing and even are on the bottom.

Each of the 18 pressure ports were connected to two pressure transducers via NetScanner™ Model 9016

Ethernet Intelligent Pressure Scanners to acquire the 18 discrete measurements. The system was connected to a

National Instruments data acquisition system, which in-turn was connected to a personal computer. A personal

computer was used to run LabVIEW software and record the data.

ii.Test Procedures

Prior to operating the wind tunnel, the ambient values of the forces and moments were recorded at each angle-of-

attack. The force and moment measurements were taken until each was within 0.05 N or N-m, respectively. Once

this tare was recorded, the data acquisition systems were configured for the test run.For wind tunnel operation, the outside air pressure was recorded for use in determining the dynamic pressure of

the test run. The tunnel was turned on and adjusted to the speed required for the data capture. The wind tunnel air

flow temperature was monitored until the temperature was stable, indicating steady flow and readiness for test.

Once the desired angle-of-attack was set, two to three force and moment measurements were typically recorded

while pressure data was simultaneously collected. Then, the angle-of-attack was incremented by one degree and the process repeated. This process was repeated at each airspeed for each configuration, nominal and ±15° aft wing.

The test matrix is outlined in Table 3.

Table 3. Wind Tunnel Test Matrix

Aft Wing Twist

(deg)

Velocity

(m/s)

AOA

(deg)

Nominal 20, 30, 40, 50 -15 to +15

+15 20, 30, 40, 50 -15 to +15-15 20, 30, 40, 50 -15 to +15

iii.Test Set-up

Initial flow visualization testing was accomplished in an attempt to show stall characteristics. Although testing

was accomplished through a wide range of angles-of-attack, stall was not apparent. It would have been interestingto determine where stall occurs for a joined-wing configuration where the forward and aft wings are not inline, but

vertically offset. In this case, the vertical offset produced a joint angle (fore wing dihedral plus aft wing anhedral) of

over 16° (Figure 10).Each configuration, nominal, twist-up and twist-down was tested at varied conditions. Since the focus of this

study is on pitch control, changes in angle of attack were the primary focus. Testing was accomplished, for the most

part, at angles of attack from -15º to +15° in 1º increments.


10/18


10

Figure 10. 16 degrees of offset between the front and aft wings.

IV. Results and Discussion

As mentioned in the Approach, Section III, the wind tunnel test was completed to demonstrate the use of wing

twist for longitudinal (pitch) control in a joined-wing-aircraft configuration and validate models that are used inanalysis methods. The forces and moments required for pitch control and pressure measurements were recorded

from the experimental examination. The following discussion will focus on pitch control, but also highlight some

characteristics worthy of note based on these tests.

A. Analytical and Experimental Test Force and Moment Results

Most important to this portion of the study is to determine aerodynamic forces such that pitch control is

realizable. Not only does coefficient of pitching moment, MyC , come into play to demonstrate wing twist

effectiveness, but also features such as the usable range of alpha before separation occurs and possibly Reynolds

number effect.Since the data is normalized and put in coefficient form, the LC α curves are consistent for the various airspeeds in

the nominal aft wing configuration (Figure 11). The analytical results produced in MSC/NASTRAN from the FEM

were also reasonably consistent with the experimental data within the linear regime (Figure 11). However, since

camber and thickness were not modeled in MSC/NASTRAN, zero lift occurs at -2˚ angle-of-attack for theexperimental data and -3.5˚ angle-of-attack for the analytical data. If the lift is plotted such that zero lift occurs at

zero angle-of-attack, the data can be matched in the linear regime (Figure 12).

16º


11/18


11

Figure 11. Lift curves for nominal aft wing configuration.

Figure 12. Experimental and analytical lift adjusted for the zero lift angle- of- attack.

While the lift curves are fairly consistent, the wing twist effectiveness seems to be affected by the higher velocity

(Figure 13), possibly due to the difference in Reynolds number.


12/18


12

Figure 13. Wing twist effectiveness for nominal aft wing configuration adjusted for the zero lift angle- of-

attack.

Figure 14. Experimental to analytical MyC Δ for nominal aft wing configuration.

The analytical results produced in MSC/NASTRAN from the FEM after correction for zero lift angle-of-attack still

shows a MyC Δ from the experimental results (Figure 14).

( ) ( )( ) ( ) ( ) ( )( ) ( ) ( ) ( )

. . . .

1 12 2

c p c p

My My Mym w

L Ld d c cq S q S m w

m w

d d L Lc cm wm w

C C C

C C

∞ ∞

Δ = −

= −

= −

(1.9)

where . .c p L is the lift at the center of pressure, c is the chord of the forward wing (subscripts andw m represent the

full-scale and sub-scale, respectively) and d is the moment arm used to calculate the pitching moment, MyC , of the


13/18


13

vehicle to include both wings. Based on the aforementioned exercise of aligning the full-scale (analytical) and sub-

scale wind tunnel results, ( ) ( ) L Lm wC C = . Thus, (1.9) can be simplified

( ) ( ) ( )d d My L c cm wC C ⎡ ⎤Δ = −⎣ ⎦ (1.10)

A correction factor moment arm, d Δ , can be applied to equate the analytical to experimental pitching moment

coefficient, ( )my wC ,corrected w wd d d = + Δ (1.11)

such that

corrected

mw

d d

c c

⎛ ⎞ ⎛ ⎞= ⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠

(1.12)

Solving for wd and substituting, (1.10) becomes,

( ) ( ) ( ) ( )( )corrected d d d My L c c cm wC C Δ⎡ ⎤Δ = − −⎢ ⎥⎣ ⎦

(1.13)

Applying (1.12),

( ) ( )d My L c wC C ΔΔ = (1.14)

Solving for the correction factor moment arm using the experimental data at -1.5 degrees angle-of-attack,

( )

( ) 0.59850.30283.68m

7.2737 m

my

w w L

C d cC

−

ΔΔ =

=

= −

(1.15)

The correction is to center of pressure apparently which is apparently an inaccuracy in MSC/NASTRAN. This isevidence of the need for experimental results, since this correction factor can be applied to future analytical models.

The sign convention of the correction factor moment arm is consistent with the angle of attack.

Examination of force data revealed that the deflection of the twist-down aft-wing configuration made a

difference in the results (Figure 15 through Figure 19). In particular, the pitching moment coefficient of the twist-

down data are not as linear as the twist-up and nominal data (Figure 16). There is a slightly lower coefficient of lift

at higher angles of attack and the lift-drag ratio is much lower for the twist-down configuration data in Figures 17and 18, respectively. In Figures 19 and 20, the coefficient of drag is much higher for the twist-down configuration,

as well. Figure 19 reveals the drag at low angles of attack contributes the most to the drag polar in the twist-downconfiguration. A likely source of these results is the noticeable break modeled into the outer-mold line to allow

freedom of movement at the root joint (Figure 7). Possibly a shroud, planned for the later models, would decrease

the drag, especially noticeable for the twist-down configuration (Figure 18).

Figure 15. Twist effectiveness for twist up, down and nominal configurations.


14/18


14

Figure 16. Lift curves for twist up, down and nominal configurations.

Figure 17. Lift-drag ratio for twist up, down and nominal configurations.


15/18


15

Figure 18. Drag polar for twist up, down and nominal configurations.

Figure 19. High drag at low angles of attack in the twist-down configuration.

Further examination suggests that there is a breakdown in the flow due to separation at low angles-of-attack.12 It is

evident from the departure from the linear region in the axial force plot that this occurs below 5 degrees angle-of-

attack at most velocities (Figure 20 and Figure 21).


16/18


16

Figure 20. Separation onset at low angle-of-attack.

Figure 21. Separation onset delayed at higher Reynolds number.

The trends in the experimental force and moment data indicate that there is an extremely tight angle-of-attack

range in which the vehicle is not stalled or in a turbulence region. Thus, the airfoil design is critical to prevent such

an early stall. In addition, shrouding the joint is important to decrease the stall explicitly evident in the aft-wing

twist-down configuration. Also, the calibration for zero lift angle-of-attack and . .c pd found in the LC α and MyC curves

will be useful for the validity of the present doublet-lattice aerodynamics and for the next-generation design.

B. Analytical and Experimental Test Pressure Results

In addition to force and moment data, the pressure was measured at the mid-section of each of the wings. The

intent was to use it to help explain phenomena in the force and moment data that may not be fully understood. The

30 m/s case is presented here with a comparison to the analytical results produced in MSC/NASTRAN from theaerodynamic model (Figure 22).


17/18


17

The drastic change in pressure between 5 and 10 angle-of-attack (top plots in Figure 23) may account for the

“dip” in MyC between 5 and 10 degrees angle-of-attack (Figure 15).

Figure 22. Analytical and experimental pressure results with nominal aft wing configuration.

Figure 23. Experimental results for the nominal aft wing configuration.


18/18


18

The subject matter was part of a larger area which can be reduced to this problem statement:

Incorporate flexible twist for pitch control in the design of a high altitude long-endurance aircraft,

including consideration of nonlinear response, and experimentally validate feasibility.

This will be accomplished by determining aerodynamic forces such that pitch control is realizable, demonstrating

nonlinear response on an aeroelastically scaled model and demonstrating implementation of flexible pitch control on

a scaled model.

The first task was accomplished by means of wind tunnel experimentation. The results in Section IV B

demonstrate that pitch control is effected by aft-wing twist, which is represented by plotting coefficient of pitchingmoment versus angle of attack (Figure 16). It shows effective control by an increase or decrease in pitching moment

coefficient depending on the angle of aft-wing twist. This change in pitching moment coefficient is also in the

predicted direction.

The remaining tasks are to be accomplished in the near future and the subject of a follow-on paper addressingfull-scale model reduction, while maintaining nonlinear aeroelastic properties.

Acknowledgments

The authors would like to thank the Portuguese Air Force Academy for the use of their wind tunnel facilities andespecially the superb model building skills of Capt. José Costa and the technical and practical expertise of Dr.Madruga Matos. The authors also thank Mr. Jenner Richards from the University of Victoria for the drawing

expertise used to build the model.

1Wolkovitch, J., “The Joined Wing: An Overview,” Journal of Aircraft , 23:161-178 (March 1986).2Roberts, Ronald, Sensor-Craft Analytical Certification, MS Thesis, Graduate School of Engineering, Air Force

Institute of Technology (AETC), Wright-Patterson AFB, Ohio, March 2003. AFIT/GAE/ENY/03-06.

Wolkovitch, The Joined Wing: An Overview3Kroo, I.M. , Gallman, J.W. and Smith, S.C.. “Aerodynamic and Structural Studies of Joined-Wing Aircraft,”

Journal of Aircraft , 28(1): 74-81 (January 1991). 4Smith, S.C., Cliff, S.E., and Kroo, I.M., “The Design of a Joined-Wing Flight Demonstrator Aircraft”,AIAA/AHS/ASEE Aircraft Design, Systems and Operations Meeting, St. Louis, MO, September 1987.5 Tyler, C., Schwabacher, G., and Carter, D., “Comparison Of Computational and Experimental Studies For AJoined-Wing Aircraft”, AIAA-2002-0702, January 2002.

6 Corneille, J., and Franke, M.. “Wind Tunnel Tests of a Joined-Wing Missile Model”, AIAA Paper 2000-0938,

January 2000.7Pope, A. , Wind Tunnel Testing, 1954.8Bisplinghoff, R., Ashley, H., and Halfman, R., Aeroelasticity, 1955.9Anderson, J., Introduction to Flight, 2nd Ed.10Roberts, Ronald, Sensor-Craft Analytical Certification, MS Thesis, Graduate School of Engineering, Air Force

Institute of Technology (AETC), Wright-Patterson AFB, Ohio, March 2003. AFIT/GAE/ENY/03-06.11 Kimler, F. and Canfield, R., “Structural Design of Wing Twist for Pitch Control of Joined Wing SensorCraft”,AIAA 2006-7134, 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth,

Virginia, September 2006.12Discussions with Dr. Raj Nangia, Nangia Aero Research Associates , 28 November 2006.

Wind Tunnel Testing of a Twisted Wing for Longitudinal Control in a Joined-Wing Aircraft

Documents

Transcript of Wind Tunnel Testing of a Twisted Wing for Longitudinal Control in a Joined-Wing Aircraft