Professional Documents
Culture Documents
This study is based on initial testing accomplished to demonstrate the use of wing twist
for longitudinal (pitch) control in a joined-wing aircraft configuration. The forces and
moments required for pitch control were experimentally determined in a Gottingen wind
tunnel. Pressure measurements were also reported. The aeroelastic response will be
investigated in follow-on testing.
Nomenclature
AFIT
b
c
CL
CL(roll)
CM
CLt
CMo
CMcg
CN
CAD
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
angle of attack
angle of elevator deflection
angle of twist, pitch (with subscript)
viscosity, aeroelastic mass
air density
Air Force Institute of Technology
wing semi-chord, wing semi-span (with subscript)
mean aerodynamic chord
coefficient of lift with respect to angle of attack
rolling coefficient of moment
pitching coefficient of moment
coefficient of lift of the tail
coefficient of moment at zero lift
coefficient of moment at the center of gravity
yawing coefficient of moment
computer-aided design
The views expressed in this article are those of the author and do not reflect the official policy or position of the Air Force, Department of Defense or the U.S. Government.
Doctoral Candidate, AFIT/ENY, 2950 Hobson Way, WPAFB, OH 45433, AIAA Student Member.
Associate Professor of Aerospace Engineering, AFIT/ENY, 2950 Hobson Way, WPAFB, OH 45433, AIAA
Associate Fellow.
3
Professor of Aeronautical Engineering, Academia da Fora Area, 2710 Sintra, Portugal, AIAA Member
4
Associate Professor of Department of Mechanical Engineering, University of Victoria, BC Canada V8W 3P6, and
AIAA Member.
5
Senior Aerospace Engineer, AFRL/VASA, 2210 8th Street, Bldg 146, WPAFB, OH 45433, AIAA Associate
Fellow.
2
1
American Institute of Aeronautics and Astronautics
CFD
D
Fx
Fz
FEM
L
M
Mx
My
Mz
MAC
OML
Re
q
p
Smod
V
Subscripts
w
m
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 subjected
to 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.
(a)
(b)
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
2
American Institute of Aeronautics and Astronautics
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 joinedwing configuration. Corneille and Franke tested several configurations of joined-wings6 in the wind tunnel at the
Air Force Institute of Technology (AFIT).
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 and
drag (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 previously
mentioned. Therefore, special attention will be paid to changes in lift due to the aft wing, with various twist angles.
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 is
required of the aft wing, rather than a control surface deflection, to provide longitudinal control of the aircraft. This
CL
can be accomplished by determining the lift curve slope for the neutral aircraft,
e =0
3
American Institute of Aeronautics and Astronautics
For a conventional aircraft, the lift curve slope of the tail remains constant but shifts to the left as the
elevator, e , is deflected downward (positive by convention). Plotting CLt vs. e at a constant angle of attack, the
CLt
curve would be linear for most conventional aircraft. This stability derivative, , is a measure of elevator
effectiveness. 9 For the joined wing, elevator deflection, e is replaced with aft wing twist, . 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 as
aft 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
CMcg
CMcg
Original CM0
New CM0
=0
()
(+)
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 and
velocity 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
bm 1
(1.1)
=
bw 38
The air density ratio is fixed by the standard day altitude of the wind tunnel and the mission profile,
541ft
50 Kft
2040.9 lb ft 2
243.6 lb ft 2
= 8.3781
(1.2)
The velocity ratio is fixed by the maximum viable speed of the wind tunnel and the mission profile,
Vm 50 m s
=
= 0.2825
Vw 177 m s
(1.3)
2
3
m b m b m
=
=1
=
M
w mstr
2
3
b w b w
4
American Institute of Aeronautics and Astronautics
(1.4)
2
where is the aeroelastic mass ratio mstr is the structural mass per unit length, b 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:
M m m bm
=
M w w bw
III. Approach
A.
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 1.2l0.8l2m. To avoid turbulence, the usable test area must maintain uniform
flow velocity (less than 0.8% in pressure variation), limiting the testable area to 1.1l0.6l1.4m. A six-degree-offreedom 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 50
m/s, respectively. At these lower speeds, however, it was anticipated that the pressure sensors would be unable to
record 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
Re =
Vc
(1.5)
The values in Table 1 illustrate that the Reynolds number of the wind tunnel testing and the full-scale aircraft are
different by two orders of magnitude. As mentioned in the Theoretical Formulation, Section II, the Reynolds
number effects are likely insignificant and the pressure data will be used to help account for discrepancies.
5
American Institute of Aeronautics and Astronautics
Table 1. Reynolds number comparison for wind tunnel and full-scale aircraft
Wind tunnel testing
Vi (m/s)
c (m)
(Pa-s)
MAC (m)
(kg/m3)
Re
50
0.097
1.82e-5
0.097
1.19
3.23e5
118
3.68
1.82e-5
3.68
1.19
2.89e7
Full-scale aircraft
ingress/egress
177
3.68
1.82e-5
3.68
1.19
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-span
model be mounted vertically (Figure 4). The right half-span was chosen arbitrarily. The following equations were
used to transform the forces
(1.6)
L = Fx sin + Fz cos
D = Fx cos + Fz sin
(1.7)
where Fx and Fz were the components of the resultant pressure force acting on the vehicle measured by the Schenck
balance and is the angle-of-attack.
y
z
x -
Figure 4. Wing tunnel model with nominal aft wing.
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 tunnel
models were fabricated.
7
American Institute of Aeronautics and Astronautics
Figure 8. 0.6m half-span model with nominal, 15 down and 15 up (front to back) twisted aft wings.
B.
CL ( roll ) =
Mx
S mod
2
, CM =
My
Smod
2
, CN =
Mz
q
Smod
2
(1.8)
Testing was accomplished in the Portuguese Air Force Academys 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 described
in 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.
8
American Institute of Aeronautics and Astronautics
Fore Wing
Port number*
Aft Wing
position (mm)
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
20
9/10
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-ofattack. 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)
Nominal
+15
-15
Velocity
(m/s)
20, 30, 40, 50
20, 30, 40, 50
20, 30, 40, 50
AOA
(deg)
-15 to +15
-15 to +15
-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 interesting
to 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 -15 to +15 in 1 increments.
9
American Institute of Aeronautics and Astronautics
16
Figure 10. 16 degrees of offset between the front and aft wings.
10
American Institute of Aeronautics and Astronautics
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.
11
American Institute of Aeronautics and Astronautics
Figure 13. Wing twist effectiveness for nominal aft wing configuration adjusted for the zero lift angle- ofattack.
Figure 14. Experimental to analytical CMy 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 CMy from the experimental results (Figure 14).
( )
Lc . p .
1q S
2
L
d
d
(
c )m ( q S ) ( c )w
m
w
c. p.
1
2
(1.9)
= ( CL )m ( dc )m ( CL ) w ( dc ) w
where Lc. p. is the lift at the center of pressure, c is the chord of the forward wing (subscripts w and m represent the
full-scale and sub-scale, respectively) and d is the moment arm used to calculate the pitching moment, CMy , of the
12
American Institute of Aeronautics and Astronautics
vehicle to include both wings. Based on the aforementioned exercise of aligning the full-scale (analytical) and subscale wind tunnel results, ( CL )m = ( CL )w . Thus, (1.9) can be simplified
CMy = ( CL ) ( dc )m ( dc ) w
(1.10)
d corrected = d w + d w
(1.11)
d corrected d
=
c w c m
(1.12)
A correction factor moment arm, d , can be applied to equate the analytical to experimental pitching moment
coefficient, ( Cmy ) w ,
such that
CMy = ( CL ) ( dc )m
Applying (1.12),
((
dcorrected
c
) ( ))
d
c
CMy = ( CL ) ( cd ) w
Solving for the correction factor moment arm using the experimental data at -1.5 degrees angle-of-attack,
Cmy
d w = ( c ) w
CL
= ( 3.68m ) 0.5985
0.3028
(1.13)
(1.14)
(1.15)
= 7.2737 m
The correction is to center of pressure apparently which is apparently an inaccuracy in MSC/NASTRAN. This is
evidence 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 twistdown 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 17
and 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-down
configuration. 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.
13
American Institute of Aeronautics and Astronautics
Figure 16. Lift curves for twist up, down and nominal configurations.
Figure 17. Lift-drag ratio for twist up, down and nominal configurations.
14
American Institute of Aeronautics and Astronautics
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-ofattack at most velocities (Figure 20 and Figure 21).
15
American Institute of Aeronautics and Astronautics
The drastic change in pressure between 5 and 10 angle-of-attack (top plots in Figure 23) may account for the
dip in CMy 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.
17
American Institute of Aeronautics and Astronautics
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 pitching
moment 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 addressing
full-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 and
especially the superb model building skills of Capt. Jos 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.
1
Wolkovitch, J., The Joined Wing: An Overview, Journal of Aircraft, 23:161-178 (March 1986).
Roberts, 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 Overview
3
Kroo, 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).
4
Smith, 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 A
Joined-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.
7
Pope, A. , Wind Tunnel Testing, 1954.
8
Bisplinghoff, R., Ashley, H., and Halfman, R., Aeroelasticity, 1955.
9
Anderson, J., Introduction to Flight, 2nd Ed.
10
Roberts, 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.
12
Discussions with Dr. Raj Nangia, Nangia Aero Research Associates , 28 November 2006.
2
18
American Institute of Aeronautics and Astronautics