Professional Documents
Culture Documents
TD GS AIAA Paper1 Final
TD GS AIAA Paper1 Final
net/publication/306033282
CITATIONS READS
5 2,559
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Geoffrey Spedding on 14 August 2017.
The tube-and-wing aircraft configuration has remained unchanged since the start of
commercial aviation. The state of minimum induced drag is comprised of an elliptically
loaded wing, which has a constant downwash profile. Interrupting this wing with a central,
payload-carrying body costs downwash, but is essential in fulfilling practical flight
objectives. Initial drag and lift estimates, streamwise velocity profiles, as well as sectional
and quasi-3D parasite drag estimates have been produced using force balance and Particle-
Imaging-Velocimetry (PIV) measurement techniques on a novel aircraft configuration,
called the Gull-Wing Configuration. The results here indicate it is possible to restore an
estimated 59% of the lost downwash while incurring no measurable increase in drag because
of induced drag reductions, despite increases in parasite drag.
Nomenclature
AR = wing aspect ratio
b = wingspan [cm]
c = chord length [cm]
d = fuselage diameter [cm]
d’ = distance from observation plane centerline to GW model centerline [cm]
D = drag force [N]
D’ = drag force per unit span [N/m]
Di = induced drag [N]
e = Oswald efficiency factor
k = induced drag factor
l = body length [cm]
lo = distance from wing chord-line to model body centerline [cm]
L = lift force [N]
L/D = lift to drag ratio
q = dynamic pressure [Pa]
Q = root mean square (RMS) of velocity fluctuations [m/s]
Rec = chord-based Reynolds number
S = wing planform area [ m2 ]
U = nominal mean freestream streamwise velocity [m/s]
Uref = empirical mean streamwise velocity [m/s]
(u,v,w) = streamwise, spanwise, and cross-stream velocity components [m/s]
wi = induced velocity, or downwash, due to wing lift [m/s]
(x,y,z) = streamwise, spanwise, and cross-stream coordinates [m]
α = wing set angle [˚]
= angle between model aft-body and model fore-body centerline [˚]
t = time between laser pulses [μs]
1
Graduate Research Assistant, Department of Aerospace and Mechanical Engineering, tylerdav@usc.edu, Student
Member AIAA.
2
Professor, Department of Aerospace and Mechanical Engineering, geoff@usc.edu, Member AIAA.
1
American Institute of Aeronautics and Astronautics
ε = fractional increase in induced drag for non-elliptically loaded wing
ν = kinematic viscosity [m2/s]
= density of air [kg/m3]
θ = angle between freestream flow and model centerline [˚]
Θ = momentum thickness [cm]
Cd = sectional (2D) drag coefficient
CD = total 3D drag coefficient
Cd,p = 2D parasite drag coefficient
CD,i = 3D induced drag coefficient
CD,p = quasi-3D parasite drag coefficient
Cl = sectional or 2D lift coefficient
CL = 3D lift coefficient
I. Introduction
T hough the optimum circulation distribution on a planar wing has an elliptic spanwise variation, dropping
gradually from its maximum at the centerplane (y = 0), the Current Dominant Configuration (CDC), which is
almost universally adopted in passenger and cargo transports, does not have such a circulation distribution [1,2].
The wing in the CDC is interrupted at some point by a central fuselage, and though there are a number of careful
designs merging the wing into the body to avoid discontinuous jumps, the basic design of the CDC is unaltered. The
idea that there could be aerodynamically more efficient designs has led to repeated configuration studies and much
work on certain aspects, such as blended wing bodies (BWB) [3].
Based on a reconsideration of some of the guiding principles and possibilities should passive pitch stability be
achieved through wing shape change alone, it has been proposed [4] that a new aircraft configuration with a shorter,
wider body and a control surface attached to a deflected aft-body can change the circulation about an entire wing-
body assembly so as to restore the uniform downwash profile (and, by implication, the elliptical loading
distribution). Because the aft control surface changes the location of the rear stagnation point, it is called a Kutta-
Edge (KE). Through initial Particle Imaging Velocimetry (PIV) measurements, Ref. [4] showed that this novel
aircraft configuration can be more aerodynamically efficient than the CDC, while still maintaining a central and
dedicated payload-carrying body. Perhaps not coincidentally, this independent design resembles that of a sea-bird,
so it is termed the Gull-Wing (GW) configuration.
Though these flow-field measurements were encouraging, direct and confirmatory force balance measurements
were not possible due to the weight of the model. The primary focus of this paper is on making direct drag and lift
measurements of a lightweight GW configuration aircraft as functions of only two geometric parameters.
Instantaneous and time-averaged velocity fields are also measured using PIV, where the momentum thickness in the
wake of the GW model is calculated, leading to sectional and quasi-3D parasite drag estimates. Force balance and
PIV measurement techniques are combined to quantify the aerodynamic performance benefits, and determine likely
sources of drag and lift variation by breaking drag into induced and parasite components.
B. Gull-Wing Model
The GW model is made from two sections, the fore- and aft-bodies, which can be arranged with only two
adjustable parameters, the aft-body deflection angle, δ, and the angle between the freestream flow and the GW body
centerline, θ. The aft-body is deflected at five different angles, δ = 0⁰, 2⁰, 4⁰, 6⁰, and 8⁰, using five acrylic wedges.
Molding clay is applied around the circumference of the wedge to smooth the transition from the fore-body to the
aft-body.
2
American Institute of Aeronautics and Astronautics
KE and NKE denote arrangements with and without the Kutta-Edge tail plate, respectively. The difference can
be seen in Fig. 2(b) and (c). The GW model is 28.55 cm in length in the NKE arrangement with a maximum body
diameter d = 6 cm. The body fineness ratio l/d = 4.75. The GW model uses a NACA 0012 airfoil to make up its
rectangular, untwisted wing. The wingspan b = 50.0 cm, and the chord-length c = 7.5 cm. The wing is fixed at a set
angle of attack α = 5.0⁰ ± 0.5⁰, and the chord-line of the wing is displaced from the centerline of the body by lo =
1.87 cm.
A solid steel rod, 0.7 cm in diameter, traverses 80% of the wingspan and is fixed inside the wing at the ¼-chord.
This rod protrudes out of one wing-tip and is inserted into a sting (Fig. 1). The model is mounted at the wing-tip,
instead of the belly of the GW model to minimize disturbances to the wing-fuselage-junction. This shrouded sting
projects through the wind tunnel floor onto a force balance.
The ideas behind the GW configuration are essentially inviscid, or Re-independent, and so wind tunnel tests at
lower Re must be carefully justified and controlled. Boundary-layer trips are placed on both body and wings to
avoid Re-dependent variations due to flow separation and possible reattachment. In particular, junction flows may
be adversely affected by growing and/or separating boundary layers at Re c = 105. If the benefits of the GW
configuration can be measured at Rec = 105, there is little reason to believe the same behavior would not be present
at larger Rec.
GW Prototype
x z
y Laser Sheet Envelope
Sting
Force Balance
Rotary Table
Figure 1. Dryden Wind Tunnel test section showing the GW model,
force balance, laser sheet, and camera.
1 (1)
q U 2 .
2
Depending on the precise air temperature, the product qS is approximately 10 N, where S is the wing planform area,
S = bc. The drag and lift coefficients for a finite wing are
D (2)
CD
qS
L (3)
CL .
qS
3
American Institute of Aeronautics and Astronautics
From Ref. [5], at α = 5˚ the NACA 0012 has a Cl = 0.55 and a Cd = 0.0075 at Rec = 3 106 . The smallest expected
non-zero force will be D = qSCd = 75 mN, when L = 5.5 N.
α
δ
θ
U
(a) FB AW AB KE
KE NKE
(b) (c)
Figure 2. (a) side view of the GW model, showing θ and δ. FB = fore-body, AW = acrylic
wedge, and AB = aft-body. (b) GW model in the KE arrangement (c) GW model in the
NKE arrangement.
Using the direct drag and lift measurements, CD and CL can be estimated. The induced cross-stream velocity, or
downwash, wi(z), can be estimated from
C L AR (4)
wi .
U
4
American Institute of Aeronautics and Astronautics
CL 2 (6)
C D,i .
eAR
CL is the total lift coefficient and e is the Oswald efficiency factor, defined as
1 (7)
e .
1 kAR
ε is the fractional increase in induced drag beyond the optimum, elliptically loaded wing and k is found empirically
from the slope of CD vs. CL2 [7-9]. AR is the aspect ratio defined as AR = b2/S. AR = 6.67 for the wing used in the
GW model. CD,p is the parasite drag coefficient, which includes all forms of drag independent of lift generation [8].
Force balance measurements yield an estimate of CD and CL directly.
The parasite drag coefficient may also be measured indirectly through the wake momentum thickness. However,
on an aircraft model with a number of separate lift and drag generating surfaces, a net lift of zero does not guarantee
that the parts themselves generate zero lift, and so does not guarantee that their induced drag will be zero either,
except in the curve fitting sense of simple equations such as Eq. (5). Thus, a direct CD,p comparison between force
balance and PIV measurements cannot be made.
b
a
z
x
const
c
d
Figure 3. Comparison of two different flow regions used for deriving the
momentum thickness and estimating Cd,p.
the integral form of the Navier-Stokes equations and Conservation of Mass through the streamtube (a-b-c-d) in Fig.
3, one may write
d u u
2 (8)
D' U 2
b
2 2
U U
dz ,
for a two-dimensional, non-turbulent, steady flow with no residual pressure gradients. Equation (8) can be expressed
in terms of the momentum thickness,
d u u
(9)
2
Θ 2 2 dz ,
b U
U
5
American Institute of Aeronautics and Astronautics
as
D' U 2 . (10)
D' (11)
Cd ,
qS
and substituting Eqs. (9), (1), and (10) into Eq. (11), with S = bc yields
2 (12)
C d,p .
c
u2(z) is measured in the (x,z) plane, at multiple spanwise (y) locations. Thus, a 3D wake defect volume was in
effect acquired to estimate a quasi-3D parasite drag coefficient, CD,p. One method of estimating CD,p from Cd,p is
adding up Cd,p found at each spanwise location, multiplying by the increment used for changing the spanwise
location, and then dividing by the total span over which the measurements were made. From the GW model
symmetry, the quasi-3D parasite drag coefficient for the entire GW model is
b
2 (13)
2
C D,p
b
C d,p, y 0 2 ( y )dy .
1
2. Equipment
The (u,w) velocity components in 2D planes (x,z) were measured as shown in Fig. 4. The PIV system used a
Nd:YAG laser, and a LaVision Imager Pro X 2M camera (1600 1200 14 bit pix) with a 70:210 mm focal length
Nikon lens. The laser-sheet generating optics and the camera reside on independent 3-axis traverses, controlled
using one stepper motor per axis. To build a wake defect volume, the camera and optics traversed the spanwise (y)
direction in unison.
6
American Institute of Aeronautics and Astronautics
u* U B ( z) u( z) , (14)
and is substituted for u2 in Eq. (9). u*(z) are normalized by Uref, an empirical mean freestream velocity in the x
direction, estimated from the upper left-hand corner of the time-averaged velocity field in a sub-window far from the
model.
DPIV
sampling
locations
Body Midline
x
(y/b = 0)
y/b = -0.1
(b)
Figure 4. (a) observation volume and (b) spanwise locations for the PIV measurements.
III. Results
7
American Institute of Aeronautics and Astronautics
0.036
0.035
0.034
0.033
CD
0.032
0.031
0.03
KE
NKE
0.029
0 2 4 6 8
δ [⁰]
(a)
0.032
0.0315
0.031
0.0305
0.03
CD 0.0295
0.029
0.0285
0.028 KE
NKE
0.0275
0 2 4 6 8
δ [⁰]
(b)
Figure 5. Total CD(δ) at θ = 0˚ in plot (a) and θ = -2˚ in plot (b). The error bars come from maximum
differences in any series of experiments at the same conditions.
8
American Institute of Aeronautics and Astronautics
0.37
0.35
0.33
0.31
CL
0.29
0.27
KE
NKE
0.25
0 2 4 6 8
δ [⁰]
Figure 6. CL( at θ = 0⁰.
Equation (4) relates wi and CL and one may compare CL with values expected (and measured in these
experiments) for a finite wing alone at the same Rec. CL in Fig. 6 is lower than for a clean wing because a large
central fuselage disrupts the circulation distribution, and the changes in CL may also be expressed as changes in wi,
perhaps a more direct way of envisioning the role of the KE.
The lost downwash fraction (compared with a clean wing) for ˚ (wi = -0.27 m/s) is 34%, and the lost
downwash for ˚ (wi = -0.35 m/s) is 14%, in the KE arrangement. So, 59%, or (34-14)/34, of the lost downwash
is restored when ˚, where total drag increases by 1.6%. Without the KE, only 27% of the lost downwash can be
restored at best (˚), where the drag increase is 6%. Figure 7 shows these changes, where Δwi is the percentage
of downwash lost as compared with the NACA 0012 wing only case, and ΔCD is the increase in drag above the
˚, NKE case. Figure 7 shows how the deflected KE can restore the central downwash, while incurring little or
negative increase in drag.
L/D(is shown in Fig. 8. Using only the average values, the optimum aft-body deflection angle for KE is
˚, and the optimum for NKE has a broader plateau between ˚ and ˚. There is a significant difference
between the base (˚) deflection and the estimated optimal values. The increases in L/D at the optimum for the
KE arrangement and the NKE arrangement are 18% and 7%, respectively. The shape of the L/D curves is strongly
influenced by the CL curves because the CL values are an order of magnitude larger than CD. Figure 8 also shows
that it is the combination of the aft-body deflection and KE that provide aerodynamic performance benefits. Notice
the higher average L/D for ˚ and ˚ in the NKE arrangement as compared with the KE arrangement,
presumably because the extra friction drag incurred by the tail area first needs to be paid.
9
American Institute of Aeronautics and Astronautics
40 KE NKE
35
30
25
Δwi 20
[%]
15
δ = 0⁰
10 δ = 2⁰
δ = 4⁰
5 δ = 6⁰
δ = 8⁰
0
-7 -5 -3 -1 1 3 5 7
ΔCD [%]
Figure 7. Δwi(ΔCD) for all tested δ, in the KE and NKE arrangements at θ = 0˚.
10.05
9.75
9.45
L/D 9.15
8.85
8.55 KE
NKE
8.25
0 2 4 6 8
δ [˚]
Figure 8. L/D(at θ = 0˚.
10
American Institute of Aeronautics and Astronautics
0.1
0.09
0.08
0.07
0.06
Θ/c 0.05
0.04
0.03
0.02
0.01
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
x/c
Figure 9. Θ/c(x/c) for 2˚, NKE at y/b = 0 and θ = -4˚.
0.4 0.4
0.2 0.2
0 0
-0.2 -0.2
z/c
z/c
-0.4 -0.4
-0.6 -0.6
-0.8 -0.8
-1 -1
0.7 0.8 0.9 1 1.1 -0.1 0 0.1 0.2 0.3 0.4
u u$
Uref Uref
(a) (b)
Figure 10. (a) u(z) is the solid line with square symbols, and UB(z) is the dashed line. (b) u*(z) is the solid
line with square symbols. These data are from y/b = 0, x/c = 0.088, and δ = 6˚, KE. y/b = 0 and θ = -4˚.
11
American Institute of Aeronautics and Astronautics
momentum defect for -0.02 ≤ y/b ≤ 0.02, or the middle 4% of the model, decreasing the step size Δy would help
greatly in decreasing ΔCD,p.
The quasi-3D parasite drag coefficient CD,p(δ) is shown in Fig. 14. On average, ΔCD,p are 10% of the average
CD,p values for all δ. In Fig. 14, the KE increases CD,p for all non-zero δ. This is expected as the KE may be thought
of as a control surface whose primary purpose is to decrease induced drag. The force balance data in Fig. 14, shown
with the square symbols, show the total drag coefficient, CD. The difference between FB and PIV is equal to CD,i.
0.4
0.2
-0.2
z/c
-0.4
-0.6
-0.8
-1
-0.1 0 0.1 0.2 0.3 0.4
$
u
Uref
(a)
0.4
0.2
-0.2
z/c
-0.4
-0.6
-0.8
-1
-0.1 0 0.1 0.2 0.3 0.4
$
u
Uref
(b)
Figure 11. Time-averaged, mean outer-flow subtracted velocity profiles for δ = 2˚ (a), and δ = 6˚ (b). The
solid line represents the KE arrangement, the dotted line represents the NKE arrangement, circle symbols
are for δ = 2˚, and triangle symbols are for δ = 6˚. y/b = 0, x/c = 0.088, and θ = -3˚.
12
American Institute of Aeronautics and Astronautics
0.4
0.2
-0.2
z/c
-0.4
-0.6
-0.8
-1
-0.1 0 0.1 0.2 0.3 0.4
$
u
Uref
Figure 12. Time-averaged, baseline-corrected streamwise velocity profiles for δ = 0˚ (solid black line,
square symbols), δ = 2˚ (dashed green line, circle symbols), and δ = 6˚ (dotted blue line, triangle
symbols), KE. y/b = 0, x/c = 0.088, and θ = -3˚.
0.09
δ=0˚, KE
0.08
δ=2˚, KE
0.07 δ=6˚, KE
NACA0009
0.06
0.05
Cd,p 0.04
0.03
0.02
0.01
0
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6
y/b
Figure 13. Cd,p (y/b). Cd,p = 0.0106 for a NACA0009 at Rec =105 is shown with the dotted line. (θ = -3˚).
Symbol sizes scale with uncertainty.
13
American Institute of Aeronautics and Astronautics
FB, KE
0.034 FB, NKE
PIV, KE
CD PIV, NKE
0.031
CD,p0.029
0.026
0.024
0.021
0 2 4 6 8
δ [˚]
Figure 14. CD,p(δ) from PIV measurements, and CD(δ) from force balance (FB) measurements. Both the KE
and NKE arrangements are shown.
Acknowledgments
This work has been supported through internal grants at USC. T.W. Davis dedicates this paper to B.K. Dovey.
References
1Munk, M. M., “The Minimum Induced Drag of Aerofoils,” NACA TR-121, 1923.
2Jones, R.T., Wing Theory, Princeton Univ. Press, Princeton, NJ, 1980, Chap. 7, pp. 105-128.
3Liebeck, R.H., “Design of the BWB Subsonic Transport”, AIAA-2002-0002, 40th AIAA Aerospace Sciences Meeting and
trailing edge,” Journal of Aircraft, Vol. 49, No. 5, 2012, pp. 1279-1289.
5Abbott, I.H., and Von Doenhoff, A.E., Theory of Wing Sections, Dover ed., McGraw-Hill Book Company, Inc., New York,
14
American Institute of Aeronautics and Astronautics
6Zabat, M., Farascaroli, S., Browand, F., Nestlerode, M., and Baez, J. “Drag Measurements on a Platoon of Vehicles.” California
Partners for Advanced Transit and Highways(PATH) Program, Institute of Transportation Studies, University of California
Berkeley, 1994, p. 18-31.
7Yechout, T.R., Introduction to Aircraft and Flight Mechanics, AIAA Education Series, Reston, 2003, Chap. 1, pp. 50-51.
8McCormick, B.W., Aerodynamics, Aeronautics, and Flight Mechanics, 2nd Ed., Wiley-India, United Kingdom, 1995, Chap. 4,
p. 165.
9Spedding, G. R., and McArthur, J., “Span Efficiencies of Wings at Low Reynolds Numbers,” Journal of Aircraft, Vol. 47, 2010,
749–763. doi:10.1007/s00348-008-0597-y
11Spedding, G. R., and Rignot, E.J.M., “Performance Analysis and application of Grid Interpolation Techniques for Fluid Flows,”
p. 115.
15
American Institute of Aeronautics and Astronautics