Jan 05

Section 6 – Initial Sizing & Analysis Techniques
1
Section 6(iv)
Initial Sizing & Analysis
Techniques
PD340 TRADE STUDY AND FINAL CONFIGURATION SELECTION
(WILLIAMS FJ44-2 ENGINES)
20500
20600
20700
20800
20900
21000
21100
21200
21300
21400
21500
21600
21700
21800
21900
22000
310 315 320 325 330 335 340 345
Reference Wing Area (sq.ft)
M
a
x
i
m
u
m

T
a
k
e
O
f
f

G
r
o
s
s

W
e
i
g
h
t

(
l
b
)
W/S
TTC
V
opt
Range 1
V
S
BFL
FEASIBLE SOLUTION
λ=0.40
λ=0.35
λ=0.30
b=54 ft
b=54 ft
b=50 ft
b=58 ft
Range 2
W/S=65 lb/sq.ft (317 kg/m
2
)
VS=90 kts @ MLW
BFL=3900 ft (1189 m)
TTC=18 min.
Vopt=375 KTAS or M0.65 @ FL 350
Range 1=700 nm (232 lb/PAX) & 850 nm (200 lb/PAX)
Range 2=800 nm (232 lb/PAX) & 950 nm (200 lb/PAX)
Aerodynamic Prediction, Devices &
Setting Requirements
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
2
Tier II Low-speed & High-speed
Aerodynamic Prediction
 The importance of predicting low-speed and high-speed
aerodynamic qualities of aircraft cannot be understated
 Vehicular definition relates to an initial appreciation of how the flight
envelope will look
 It is one of the integral components in formulating airplane operational
performance attributes
 Prediction of low-speed and high-speed aerodynamic attributes
covers the following categories
 Low-speed aerodynamics
 Clean wing lift characteristics and maximum lift
 Maximum lift generated by trailing and leading edge high-lift devices
 High-speed aerodynamics
 Zero-lift drag
 Vortex-induced drag at subsonic speeds
 3D effects, trim and ancillary drag contributors
 Total incremental drag due to OEI condition
 Compressibility or wave drag due to volume and lift
 Aerodynamic impact of winglets
 Buffeting qualities
J
a
n

0
5
S
e
c
t
i
o
n

6

I
n
i
t
i
a
l

S
i
z
i
n
g

&

A
n
a
l
y
s
i
s

T
e
c
h
n
i
q
u
e
s
C
o
p
y
r
i
g
h
t

2
0
0
5
b
y
A
s
k
i
n
T
.

I
s
i
k
v
e
r
e
n

A
l
l

R
i
g
h
t
s

R
e
s
e
r
v
e
d
3
T
i
e
r

I
I

L
o
w
-
s
p
e
e
d

&

H
i
g
h
-
s
p
e
e
d

A
e
r
o
d
y
n
a
m
i
c

P
r
e
d
i
c
t
i
o
n

(
c
o
n
t
.
)
Primary and secondary control surfaces and forces on an airplane
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
4
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
 C
Lmax
is the maximum lift coefficient the wing can generate
 C
Lmax
is dependent upon
 Wing sweep
 Wing aspect ratio
 Wing thickness-to-chord
 Flapping span and flap deflection angle
 High-lift device configuration
 In conceptual design, C
Lmax
is often predicted by inspecting other
aircraft of similar configurations; as a general rule
 Empirical methods are well suited to giving results with an adequate
level of accuracy for conventional aircraft configurations and technology
levels
 The primary goals are for highest (L/D)
TO
and (D/L)
LD
 Predictions should not exceed approximately C
Lmax
= 3.50 unless
suitable justification has been established
 Parametric analysis techniques can be utilised to confirm the validity of
prediction results
α α, angle of attack , angle of attack
C C
L L
, Lift Coefficient , Lift Coefficient
C C
Lmax Lmax
clean clean
C C
Lmax Lmax
landing landing
C C
Lmax Lmax
takeoff takeoff
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
5
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
 An expedient method to establish clean wing C
Lmax
and lift-curve
geometry
 First identify the 3D C

using the Vortex-Lattice method; closed-form
Helmbold method is good enough as well
 Predict the zero-lift angle-of-attack; can read off 2D test data results as
an initial guess; non-linear lift is predicted to commence at α
oL
+ 10°
 Use the algorithm C
Lmax
= 14 dC
L
/dα to estimate the maximum lift
coefficient for 1g stall
α
oL
L
i
f
t

C
o
e
f
f
i
c
i
e
n
t
,

C
L
Angle of Attack, α (deg.)
43 − 2AR
ref
3
4° x dC
L

dC
L

∆α = 10°
Vortex-Lattice Calculations
Empirical Algorithm
1
2
3
4
α
stall
Predicting the lift characteristics of a clean finite wing
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
6
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
Examples showing distinction between 1g and minimum
aerodynamic stall definitions
R
e
f
:

S
o
m
e

A
s
p
e
c
t
s

o
f

A
i
r
c
r
a
f
t

D
e
s
i
g
n

a
n
d

A
i
r
c
r
a
f
t

O
p
e
r
a
t
i
o
n
O
b
e
r
t
,

1
9
9
6
Ref: AGARD CP-102
F-28 Mk 4000
Boeing 747
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
7
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
Note the reference configuration
Use fractional change theory
to predict the ∆C
Lmax
of
alternative layouts
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
8
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
 Lift-to-drag ratio during takeoff manoeuvers
 Instantaneous OEI climb gradient at V
2
speed can be predicted using
the parametric correlation below
 Increasing the incremental lift with high-lift devices has a tendency of
reducing the available lift-to-drag ratio, hence, is detrimental to climb
Ref: Delft University Press
Synthesis of Subsonic Airplane Design
Torenbeek, 1982
Method to estimate lift-to-drag ratio of design candidates with
high-lift devices deployed
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
9
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
Ref: 1981-6 – No. 91
L’Aeronautique et L’Astronautique, 1981
Details of wing planform, airfoil section and twist distribution
geometry for A310 transport
J
a
n

0
5
S
e
c
t
i
o
n

6

I
n
i
t
i
a
l

S
i
z
i
n
g

&

A
n
a
l
y
s
i
s

T
e
c
h
n
i
q
u
e
s
C
o
p
y
r
i
g
h
t

2
0
0
5
b
y
A
s
k
i
n
T
.

I
s
i
k
v
e
r
e
n

A
l
l

R
i
g
h
t
s

R
e
s
e
r
v
e
d
1
0
T
i
e
r

I
I

L
o
w
-
s
p
e
e
d

&

H
i
g
h
-
s
p
e
e
d

A
e
r
o
d
y
n
a
m
i
c

P
r
e
d
i
c
t
i
o
n

(
c
o
n
t
.
)
Itemized breakdown of total drag and physical explanation of origins
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
11
 Predicting zero-lift drag
 Basis is modified Eckert’s equation for skin friction incorporating a
 Mixed (laminar) flow adjustment can be incorporated thereafter
 Component build-up method is used to generate reference condition
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
0.002000
0.002500
0.003000
0.003500
0.004000
0.004500
0.005000
0 5000 10000 15000 20000 25000 30000 35000 40000
Vehicle Wetted Area (sq.ft)
V
e
h
i
c
u
l
a
r

E
q
u
i
v
a
l
e
n
t

S
k
i
n

F
r
i
c
t
i
o
n

C
o
e
f
f
i
c
i
e
n
t

(
-
)
Unacceptably
Excessive
or Active Methods
Mean Line
Large Regionals & Large Business Jets
Small Regionals & Small Business Jets
Narrow-bodies
Wide-bodies
( ) | | | |
d
2
b
R act
f
M c 1 N log
A
c
+ η
=
equiv. sand roughness,
pressure & interference
Mach number
Survey of wetted areas and equivalent skin friction coefficients
Reynolds
number
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
12
δ
R
L
R
l
vt
T
op
D
wm
y
eng
y
eng
 Predicting vortex-induced drag
 Obert’s empirical method is suitable for subsonic analysis (M>0.4)
 Reduction in dC
D
/dC
L
2
due to slot-effect needs to be modeled as well
 Incremental drag due to 3D effects and ancillary drag contributors
 Most common method is form factors that account for
 3D effects
 Ancillary interference
 Excrescences
 Trim (goal should be keep it small)
 These values are computed based on thickness-chord ratios of the
wing, horizontal and vertical tails, and, the fineness ratios of the
fuselage, nacelle and other appendages
 OEI asymmetric drag estimation
 Windmilling drag estimated using “imaginary cut-off Reynolds number”
 It is an imaginary skin roughness (l/k) independent of engine size
 Assuming this roughness level an equivalent skin friction is computed using
the Prandtl-Schlichting form of Eckert’s equation
 Drag due to asymmetry is then based
on equilibrium of moments
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
007 . 0
AR
05 . 1
C d
C d
clean
2
L
D
+
π
=
|
|
.
|

\
|
vortex-induced
drag factor
ref. aspect ratio
| |
w
R
vt
eng
op wm wm
DOEI
S q
tan
l
y
T D D
C
δ + +
= ∆
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
13
 Predicting wave drag
 Difference in zero-lift drag coefficient between the fastest Mach number
(less than M =1.0) & Critical Mach is defined as transonic wave drag
 Can produce reasonable initial estimate of Critical Mach using modified
Korn’s equation
 Empirical exponential equation is then utilised to model the geometric
increase in drag within the drag rise and divergence regimes
 Supersonic wave drag accounts for contributions due to volume
displaced by the vehicle as well as lift distribution
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
( )
M
cos
c t
cos
C
10
1
M
cos
1
M
Qchd
m
2 / 3
Qchd
2
L
REF
Qchd
CR
∆ −
¦
)
¦

¹
¦
¹
¦
´
¦

Λ

|
|
.
|

\
|
Λ

Λ
=
ref. wing quarter chord sweep
airfoil technology operating lift coefficient
margin to divergence Mach
mean wing thickness
Mach number
C
D
Constant C
L
Constant C
L
increasing C
L
M
CR
M
DD
∆C
D
= 0.0020
J
a
n

0
5
S
e
c
t
i
o
n

6

I
n
i
t
i
a
l

S
i
z
i
n
g

&

A
n
a
l
y
s
i
s

T
e
c
h
n
i
q
u
e
s
C
o
p
y
r
i
g
h
t

2
0
0
5
b
y
A
s
k
i
n
T
.

I
s
i
k
v
e
r
e
n

A
l
l

R
i
g
h
t
s

R
e
s
e
r
v
e
d
1
4
T
i
e
r

I
I

L
o
w
-
s
p
e
e
d

&

H
i
g
h
-
s
p
e
e
d

A
e
r
o
d
y
n
a
m
i
c

P
r
e
d
i
c
t
i
o
n

(
c
o
n
t
.
)
Suggested target design and off-design characteristics
Ref: Some Aspects of Aircraft Design
and Aircraft Operation
Obert, 1996
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
15
 Buffet Envelop
 It is an additional en route limitation to the aircraft flight envelop
 Defines an upper threshold of flight level after an appreciation of climb
and cruise specific excess power residuals, and, maximum cabin
pressure differential are considered
 Buffeting is characterized by
 Breaks in C
L
-α, c
m
-α or c
x
-α curves and emergence of pressure divergence
on any of the lifting surfaces or fuselage
 The derivation of these boundaries are commonly performed using
extrapolated wind tunnel data to full-scale and subsequently verified with
flight testing
 Initial prediction methods can become mathematically quite extensive which
do not easily lend themselves to simplification
 In reaching and surpassing the threshold for buffeting the aircraft must
permit full controllability
 This means flow separation on a swept wing at high Mach number should not
initiate too far outboard to prevent strong roll or pitch-up tendencies
 Airworthiness rules stipulate cruise flight has to be limited to lift
coefficients where n = 1.30 can be reached without encountering buffet
 Free from buffet within the operationally expected envelop is desirable
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
Explanation of buffeting envelop for transport aircraft
Ref: AIAA 88-2043
The Integration of CFD and
Experiment: An Industry Viewpoint
Ref: AIAA-2002-0002
Design of the Blended-Wing-Body
Subsonic Transport
Liebeck, 2002
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
16
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
Buffet boundary for MD80 transport
Predicted and flight test derived buffet boundary for L-1011
R
e
f
:

S
o
m
e

A
s
p
e
c
t
s

o
f

A
i
r
c
r
a
f
t

D
e
s
i
g
n

a
n
d

A
i
r
c
r
a
f
t

O
p
e
r
a
t
i
o
n
O
b
e
r
t
,

1
9
9
6
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
17
Aerodynamic Devices
 These are appendages that either enhance performance or fix
problems, i.e. either lead to successful operation and/or
certificated airworthiness
 Winglets
 With greater emphasis being placed on improving aircraft cruise
efficiency winglet devices offer the most attractive drag reduction
 Another reason for selecting winglets is the aesthetic appeal
 There are two categories
 The conventional winglet; AR=1.5
 Blended winglets typified by a high aspect ratio (AR=3.5) and integrated by
way of pronounced filleted transition geometry between the wing and winglet
structures
 Benefits of winglets can be itemized as follows
 Decreased fuel burn and increased payload range attributes – achieved
through an aerodynamic performance improvement, i.e. net vehicular drag
reduction
 Higher cruise altitude and OEI drift-down ceiling –
due to a net vehicular drag reduction enabling a
greater amount of specific excess power at given
altitude and speed
 Improved takeoff performance – higher effective
OEI lift-to-drag and therefore higher second
segment climb gradient for given reference speed;
allows for higher TOGWs
 Reduced engine maintenance – the option of
retaining the original takeoff performance levels
prior to installation of winglets promotes a reduced
thrust concept
 Lower airport noise levels –
exploiting the reduced thrust concept
 Vortex Generators
 Flow over a lifting surface may tend to separate prematurely leading to
stall, diminished control authority, greater drag or even noise
 The separation can be either chordwise or spanwise
 Separation can occur at low-speed or high-speed (transonic flow)
BBJ with Aviation
Partners’ winglet
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
18
 To correct this situation a series of vortex generators or vortilons near
the wing or control surface leading edge are usually installed
 These energize the airflow over the surface and thereby assist in delaying the
onset of flow separation
 This is a common solution to imperfections like poor manufacturing tolerance
 Thin plates attached to engine nacelles or along the forward portion of
the fuselage body are called strakes – also shed vortices to energize
local flow or even correct directional stability at high angles of attack
 Not a desirable solution
 Can be avoided for the wing if thoughtful consideration is given to wing
thickness, section contour distribution and washout
 Measure of insufficient upfront work done on a new design if artificial devices
are employed to fix problems during flight testing
 Perpetual strides in CFD capabilities will have a tendency to minimise use of
vortex generators, or, at least establish a rationale that employing them is the
best compromise
Aerodynamic Devices (cont.)
Examples of vortex generators for high-speed (GV left) and low-
speed (Legacy right)
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
19
Aerodynamic Devices (cont.)
 Stall Strips
 Are spanwise strips added to the wing
leading edge to ensure stall begins at
that location first
 They provide more docile (acceptable)
stall characteristics
 It is an effective method to ensure
proper stall progression, however,
high-speed drag penalty
 Do not require these when leading edge high-lift device is used
 Wing Fences
 Act as barriers to deter cross-flow, thereby possible separation which
 High-speed drag penalty
 Ventral Fins
 Are surfaces that protrude from the
underside of the aft fuselage in an
inverted “V” configuration
 They improve stall protection by
scooping up air under the tail helping
to push the nose down at high alpha
 Another benefit is enhanced
directional stability at
sideslip and higher angles
of attack
 Ancillary benefit
 Can avoid the need of a
stability augmentation
system through inherent
improved directional stability
at high Mach numbers and
altitudes, and, increased
Dutch-roll damping
 Generates drag through greater wetted area and interference
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
20
Setting Requirements for Low-speed &
High-speed Aerodynamics
 Whenever an initial technical assessment is undertaken a
preliminary list of wing aerodynamic design requirements needs to
be generated
 Primary considerations include
 Aircraft performance and handling
 Aircraft certification
 The list constitutes a roadmap and is formulated by collaborative efforts
between conceptual design, aerodynamics and operational
performance functions
 The most important component is the wing design
 It is an iterative process and requires input from all three groups mentioned
above
 Issues concerning design philosophy generate fundamental questions about
how the goals are to be achieved
 Requisite number of development wings
 Requisite number of production wings (if a family concept)
 Scope of trade-off analysis and declaration of optimisation parameters
 Low-speed requirements and targets that need to be defined are
 All speed targets are with respect to 1-g stall concept
 Max expected L/D for each flap and/or slat angle
 Expected L/Ds at 1.13V
S
and 1.23V
S
for respective takeoff and landing
configurations
 Stable L/D versus CL at 1.13V
S
and 1.23V
S
and V
FE
 Expected C
D
at
 V
2
(1.13Vs) for each permissible takeoff flap configuration
 Mid-AUW, typical descent speed (e.g. 250 KCAS) in the clean configuration
(idle power)
 V
APP
(1.3V
S
), in the clean configuration
 V
REF
(1.23V
S
) in the landing configuration
 Alpha = 0.0 in ground effect for each takeoff flap configuration
 Expected C
Lmax
for each flap and/or slat angle assuming both clean and
with icing contamination
 Number of unprotected (anti-ice or de-ice) slat panels should be taken into
account
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
21
Setting Requirements for Low-speed &
High-speed Aerodynamics (cont.)
 Small Runback Ice behind ribs and edges of protected slat panels
 Double-Horn (3 in.) shapes on winglet (if applicable), wing-body fillet and
landing lights
 Takeoff ice on all forward facing aerodynamic surfaces including protected
slat panels should not result in stall speed increase of more than 3 KCAS
 Landing ice on all forward facing aerodynamic surfaces including protected
slat panels should not result in stall speed increase of more than 5 KCAS
 Delayed Turn-on ice on all slat panels should not advance stall onset ahead
of stall warning (Plus 1 sec., if applicable)
 Expected C
Lmax
in the landing configuration
 Expected C
LMU
(in ground effect at aircraft tip-back geometry limit minus
1° is approximately C
Lshaker
in free air) with no wing tip separation
 Special relationships and guidelines gathered through experience are
 C
Lmax
lowest takeoff flap > C
Lmax
landing / 1.21
 C
Lmax
clean > C
Lmax
landing / 1.50
 No significant lift loss due to residual de-icing fluids in aerodynamic critical
zones during lift off in ground effect
 Acceptable stall characteristics, uncontaminated and with icing assumptions
 Number of unprotected (anti-ice or de-ice) slat panels should be taken into account
 Small Runback Ice behind ribs and edges of protected slat panels
 Double-Horn (3 in.) shapes on winglet (if applicable), wing-body fillet and landing
lights
 Takeoff ice on all forward facing aerodynamic surfaces including protected slat panels
 Delayed Turn-on ice on all slat panels
 Double-Horn (1.5 in.) ice on all slat panels
 No winglet separation up to V
2
– 5 KCAS for all takeoff flaps
 No significant buffeting up to VFE for all flap and/or slat configurations
 Wing Stall Progression
 Should be preceded by trailing edge separation and/or buffeting of the inboard/mid-
wing
 Onset should not be defined by leading edge separation
 Should initiate on the inboard/mid-wing at the trailing edges
 For underwing podded engines, flow over the wing behind the nacelles should remain
attached and be adequately energised up to higher angles of attack
 Outboard wing leading edge should be adequately protected to higher angles of
attack with no significant losses in roll control effectiveness
 Approach and Landing Phase
 Pitch attitudes of 0-2° at V
REF
in the landing configuration
 Pitch attitudes at touch-down (V
REF
– 10 KCAS at 50 ft), in the landing configuration,
is less than the aircraft tip-back geometry limit by at least 2°
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
22
Setting Requirements for Low-speed &
High-speed Aerodynamics (cont.)
 Pitch attitudes of less than 4° at V
APP
in the clean configuration
 No abrupt changes in pitch stability with increasing alpha up to maximum
alpha
 Dihedral stability for all low speed configurations
 Wing tip, flaps, underwing podded engine ground clearances up to 10° in roll,
geometry limit in pitch or combination of both
 High-speed requirements and targets that need to be defined are
 Expected maximum M*L/D at design cruise speed
 Expected L/D at
 High AUW, maximum climb speed, initial cruise altitude
 Mid-AUW, typical climb speed, intermediary cruise altitude
αMAX
Range
No unnacceptable handling characteristics up to
αMAX
(roll-off, sudden pitch-up, severe buffetting, etc.)
Performance Requirements @ Shaker
CL
α
CL
MAX (no ice)
CL
Shaker (no ice)
Manoeuvre
Margin
2
0
1
0
Refer ence Speed
CL
REF
3 % or 5% Mar gin
No Ice
With Ice
Definition of target CL-α characteristics; note stick-pusher needs
to be accounted for aft-fuselage mounted engine configuration
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
23
Setting Requirements for Low-speed &
High-speed Aerodynamics (cont.)
 Expected C
D
at
 High AUW, typical climb speed, initial cruise altitude
 Mid-AUW, maximum climb speed, intermediary cruise altitude
 High AUW, typical cruise speed, initial cruise altitude
 High AUW, maximum cruise, initial cruise altitude
 M
DD
number at mid-AUW and initial cruise altitude
 Buffet boundaries margin of 1.4 g at
 High AUW, intermediate speed, initial cruise altitude
 Mid-AUW, M
MO
, intermediary cruise altitude
 Special relationships and guidelines gathered through experience are
 Speed stability (slope of L/D versus C
L
) assured at low AUW, M
FC
/V
FC
kink
(thrust lapse rate included)
 C
D
always increases with Mach and C
L
particularly for intermediate to high
speeds
 Shock waves strength and movement should not be abrupt with increasing
Mach up to M
MO
or alpha (C
L
) up to 1.5g
 Typical aircraft pitch angles during cruise
 Should not exceed +1.5-2.0° for most cases within the typical operations envelop
 Good design practise to ensure +0° for all operations
 Stable dihedral and weathercock characteristics up to M
MO
/V
MO
FC
/V
FC
 No aileron aerodynamic reversal up to M
FC
/V
FC
Jan 05
Section 6 – Initial Sizing & Analysis Techniques
24
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)
 Young, A.D., ”The Aerodynamic Characteristics of Flaps”, Aeronautical
Research Council Reports and Memoranda, Ministry of Supply, United
Kingdom, 1953
 “Aerodynamics”, Jet Transport Performance Methods, D6-1420, Seventh
Edition, Boeing Flight Operations Engineering, May 1989
 Obert, E., “Forty Years of High-Lift R&D – An Aircraft Manufacturer’s
Experience”, AGARD DCP 505, September, 1993
 Obert, E., “The Aerodynamic Development of the Fokker 100”, ICAS-88-
6.1.2, 1988
 Schaufele, R.D., Ebeling, A.W., “Aerodynamic Design of the DC-9 Wing and
High-Lift System”, Douglas Aircraft Div., McDonnell Douglas Corp., AIAA
Paper No. 670846, 1967, pp 2575-2583
 Shevell, R.S., “Aerodynamic Bugs: Can CFD Spray Them Away?”, AIAA-85-
4067, AIAA 3
rd
Applied Aerodynamics Conference, October 1985
 “Getting a Lift Out of Winglets”, Business and Commercial Aviation,
February 1998, pp. 56-65
 Dees, P., Stowell, M., “737-800 Winglet Integration”, SAE Paper 2001-01-
2989, 2001 World Aviation Congress, September 2001
 Isikveren, A.T., “Quasi-analytical Modeling and Optimization Techniques for
Transport Aircraft Design”, Section 7, “Predicting Low-Speed and High-
Speed Aerodynamic Attributes”, Report 2002-13, Royal Institute of
Technology (KTH), Ph.D. Thesis, Department of Aeronautics, Sweden, 2002
2001-01-2989
737-800 Winglet Integration
Paul Dees
Boeing Commercial Airplanes
Michael Stowell
Aviation Partners Boeing
ABSTRACT
A joint venture called Aviation Partners Boeing
successfully integrated winglets into the Next-Generation
737-800 by retaining performance improvements with
minimal weight penalty on the existing 737 wing design.
Program challenges included developing both retrofit
and production configurations using a common winglet
design, causing minimal impact on all customers, and
causing minimal disruption to the 737 production
process. Winglet benefits along with improved
performance include reduced engine wear and
enhanced visual appeal.
INTRODUCTION
The 737-800 wing was originally designed and certified
without winglets. The flight testing of winglets for the
Boeing Business Jet (BBJ) indicated the expected gains
in aerodynamic efficiency were real, as also were
increases in flight loads. The technical challenge then
wing design, keeping the improved aerodynamic
efficiency with minimal structural weight penalty and
minimal systems changes. The program challenge then
was how to integrate winglets into both existing fleet
aircraft and into new production aircraft. Another
program challenge was how to minimize cost of the flight
test and certification effort of several distinct wing
configurations, preferably using a common winglet
design. To meet these challenges, a joint venture called
Aviation Partners Boeing (APB) was formed between
The Boeing Company and Aviation Partners, Inc. where
the patented blended winglet technology (Reference 1)
was developed. Boeing has primary responsibility for
production winglets and APB has primary responsibility
for retrofit winglets on in-service airplanes.

AVIATION PARTNERS BOEING BACKGROUND
Aviation Partners Boeing is a limited liability corporation
owned by The Boeing Company (Boeing) and the
principals of Aviation Partners Incorporated (API). API’s
primary business is the application of performance
improvement technology to business jets. The joint
venture company was formed after Boeing Business
Jets contracted API to design and certify winglets on the
737-700 IGW business jet. The purpose of the joint
venture is to create a mechanism for an exchange of
data between API and Boeing with the goal of improving
the performance of Boeing products in production and in
Winglet technology for applications on current aircraft in
production as well future airplane programs. The joint
data, which will facilitate design and certification efforts
in the retrofit market.
WINGLET BENEFITS

Figure 1 - Blended winglet on 737-800

The addition of 8 foot tall Blended Winglets to the 737-
800 (see Figure 1) increases the aerodynamic efficiency.
For a given amount of lift, drag is reduced.
Direct economic benefits to the airlines include
combinations of these items (not all are available
simultaneously):
• Decreased fuel burn
• Improved take-off performance
• Reduced engine maintenance
• Lower airport noise levels
Figure 2 shows the flight-test derived winglet block fuel
burn improvement, which increases with cruise range. It
is based on an average of eastbound and westbound
missions and is common to both retrofit and production
winglets.
Figure 2 – Winglet block fuel burn improvement
Other less tangible benefits include high-tech visual
appearance and airline passenger appeal
(environmentally friendly).

Figure 3 – Blended winglet construction
Figure 3 shows the 737-800 Blended Winglet
construction. The winglet is approximately 70%
graphite-epoxy by weight.
RETROFIT WINGLETS
APB has primary responsibility for the retrofit (post
delivery and in service) winglet installations. In the
aircraft retrofit environment many of the challenges to
install winglets on the airplane are different compared to
the production modifications.
Many of the aerodynamic driven changes to the 737-800
are the same for the retrofit and production versions.
Changes common between the 737-800 retrofit and
production aircraft with winglets are:
• Winglet
• Stabilizer Trim settings
• Auto-throttle
• Flight Management Computer (FMC) data
Figure 4 – Retrofit winglet aircraft modifications
Most of the structural changes required differ between
the 737-800 retrofit and production aircraft. Figure 4
shows the primary retrofit changes and Figure 5
illustrates the structural modifications required for the
737-800 winglet retrofit.

Figure 5 – Retrofit wing modifications
Adding winglets increased both the wing dynamic and
static flight loads significantly. An economically viable
retrofit program minimizes the recurring costs of the
installation. This is difficult because the retrofit
modification is limited by existing parameters in the basic
airplane. For example, increasing skin thickness may be
the most efficient means of increasing the wing bending
strength, however skin replacement is not cost effective
for retrofit. For the Retrofit 737-800 the wing strength
was increased by the addition of straps and angles to
the stringers located inside the wing-box as shown in
Figure 5. Modification to the wing was minimized by the
development of a Speed-brake Load Alleviation System.
This system changes the angle of the in-flight speed-
Wing service life goals were achieved by reworking
existing fasteners in the lower wing skin. The fasteners
were removed and replaced with interference fit, special
fasteners for fatigue life improvement.
The increased pitch inertia at the wingtips by the addition
of winglets aggravated critical flutter modes. A reduction
in the low altitude operating speed was avoided by
adding 90 pounds of ballast per wing in the outboard
leading edge. Also, replacement of the removable outer
2 bay skin panels improved flutter tip modes.
PRODUCTION WINGLETS
Boeing has primary responsibility for the in-line
production winglet installations. The winglets are built
within Boeing to the same drawings as the APB retrofit
winglets.
Figure 6 – Production winglet installation modifications
The retrofit configuration used a load-alleviation system
to handle the increased flight loads. The production
winglet installation met the challenge by carefully
designing minimal additional bending and torsional
stiffness into the wing. The structural provisions were
designed to minimize weight impact on customers who
chose not to purchase the optional winglets. They were
also designed to minimize the impact of winglets on the
Boeing production facilities, especially final assembly.
Flutter considerations drove a significant effort to control
wing torsional stiffness and winglet weight and center of
gravity. Systems changes were also required to support
the addition of winglets. An overview of the required
changes for the production winglet installation is shown
in Figure 6.
The wing structural changes are shown in Figure 7. The
primary changes were upper and lower skin panel gage
changes and stringer gage changes over the outboard
2/3 of the wing. To minimize the weight penalty for
customers who do not choose winglets, these changes
stop at rib 25, and the configuration is known as “partial
provisions”. Partial provisions also include new ribs 25
through 27 with additional strength as needed. As with
the retrofit, some specific fastener locations are cold
worked to meet fatigue requirements. Some minor
strengthening is required in the center wing.
Figure 7 – Production winglet structural changes
The customers that choose winglets have new upper
and lower outboard skin panels from ribs 25 to 27 and
75 pounds of flutter ballast per wing that is required to
meet the flutter certification requirements of being flutter
–free at 15% greater airspeeds then Mdive/Vdive. It
would have been possible to trade flutter ballast weight
for greater increases in wing skin panel thickness, but
that was rejected as it would have penalized customers
not choosing winglets.
As with the retrofit, an absolute seal is installed to
prohibit any flammable fuel vapors from the inboard wing
from reaching any potential ignition sources in the
winglet.
Since the winglets improve cruise performance, a new –
800 winglet model engine database (MEDB) for the flight
mission computer (FMC) is required and is selected via
pin select. Likewise, a new Autothrottle is used with
winglets and includes a winglet setting via dipswitch.
These system changes are common with the retrofit
installation. All of the position and navigation lighting is
on the winglet, as with the retrofit configuration (Figure
8). The aft position light installation is in a low drag
streamlined fairing on the inboard portion of the winglet.
The early production winglets have a small light shield
inboard of the forward anti collision lights to prevent
strobe flashing from entering the cockpit. The new –6
stall management yaw damper (SMYD) accommodates
the shield’s impact on stick shaker speeds and is pin-
selectable. A retrofittable lighting product improvement
is in development to eliminate the light shield.
Figure 8 – Production winglet lighting
Another small systems change is required due to the
winglet aerodynamics altering the stabilizer trim angles.
This manifests itself as updated stabilizer trim switch
locations and a winglet “greenband” light plate in the
cockpit. Autothrottle, FMC, SMYD, and stabilizer
greenband changes are shown in Figure 9.
Figure 9 – Systems changes
FLIGHT TEST AND CERTIFICATION
Five different 737 aircraft were flight tested from 1998 to
2001 to validate and to certify the winglet installations. A
summary of these flight test programs is shown in Figure
10. Boeing and APB held joint flight test programs
wherever possible to minimize cost and share data.
Prototype winglet performance and loads were flown in
1998 and 1999 on the YC001 (737-800) and YG001
(737-700 BBJ) airplanes. The BBJ winglet installation
was certified on YG032 in 2000. It is similar but not
identical to the –800 retrofit winglet installation, which
was certified using YC020 flight test data. An example
of cooperation between Boeing and APB is the use of
YC020 flutter flight test data to correlate with Boeing
computational methods in support of the production
winglet flutter certification. This allowed a reduction in
YC714 flight test hours by avoiding additional flutter flight
testing.
Figure 10 – Flight test summary
APB worked with assistance from BCA to achieve
certification for the retrofit installation with the FAA and
JAA and obtained the Supplemental Type Certificate
(STC) in May, 2001. Certification of the Boeing
production installation, with assistance from APB,
occurred also in May and was done by Program Letter of
Definition (PLOD).
AIRLINE OPERATIONS
The first flight with certified 737-800 winglets was by
Hapag Lloyd on May 8th, 2001.
Initial production winglet customers included South
African Airways through GATX, Air Berlin, ILFC, and
American Trans Air. Initial retrofit winglet customers
included Hapag-Lloyd as launch customer and Air Berlin.
POTENTIAL FUTURE PROGRAMS
APB believes a tremendous interest in winglets exists in
the passenger and freighter market place. Current
committed 737 retrofit programs beyond the 737-800 are
the 737-700 and the 737-300. Figure 12 details the
status of all the 737 winglet programs.
Figure 12 – 737 Winglet program status

CONCLUSIONS
1. APB blended winglets were successfully integrated
and certified onto the Boeing 737-800, both as
retrofit and production installations.
2. Properly integrated winglets provide substantial
value to their operators.
3. The expected winglet performance benefit was
maintained with minimal weight penalty despite
impact on flutter were required for a successful
program.
5. A joint development and flight test program was an
important ingredient to support the certification
efforts.
6. A common design approach for both retrofit and
production winglet installations provides maximum
fleet commonality for the winglet customers.

REFERENCE
Gratzer, Louis B., “Blended Winglet”, US Patent
5,348,253, granted September 20, 1994.

CONTACT
Retrofit winglet sales information is available from Tom
VanDerHoeven at 1-800-winglets.
Production winglet sales information is available from
James Wilkinson at (206) 766-1380.

LOW-SPEED & HIGH-SPEED AERODYNAMICS

75
7 Predicting Low-Speed and High-Speed Aerodynamic
Attributes

The importance of predicting low-speed and high-speed aerodynamic qualities of
aircraft cannot be understated. The implication to vehicular definition relates to an initial
appreciation of how the flight envelope will look as well as being one of the integral
components in formulating the aeroplane’s operational performance attributes. The main
aim is to develop methodologies where the designer has an ability to approach the design
solution in a more sophisticated manner; not only in terms of departing from the usual
more simplified approach premise, but an account of the impact a technological decision
makes to the end result. These two primary goals must also be tempered by an appreciation
for reduction in the analysis complexity. This is surmised as being achievable by first of all
soliciting the designer’s philosophical requirements and translating this notion into single
all-encompassing algorithms that provide visibility to the designer. Secondly, the
methodologies must be impervious to stoppage when key information required on the part
of the designer is found to be lacking.

7.1 Low-Speed Aerodynamics: Lift
To consistently support design studies of not only quite complex conventional
planforms (with multiple cranks, dihedral, etc.), but also of more exotic layouts such as
multi-surface and non planar wings, it was recognised the algorithm to compute maximum
lift attributes adhere to a quasi-analytical philosophy. This task can be achieved by
concurrent utilisation of dedicated software to quantify the fundamental parameter of clean
wing lift-curve slope with well-established empirical methodologies.

7.1.1 Clean Wing Lift Attributes and Maximum Lift
The clean wing maximum lift can be computed for any original multi-surface or non-
planar planform geometric definition using a three-dimensional Vortex-Lattice Method
93

(VLM), which calculates aerodynamic properties of multi-wing designs that are swept
(symmetric or otherwise skewed), tapered, cambered, twisted and cranked with dihedral.
Unlike what is offered by classical VLM approaches, one particular approach models the
wake coming off the trailing edge of every lifting surface as flexible and changing shape
according to the flight state considered. With a distorting wake, non-linear effects such as
the interaction of multiple surfaces can be simulated more consistently. The source of the
basic theory for the VLM with flexible wake is cited as Moran
94
, and an exemplar of
software embodying these principles is one authored by Melin
95
. Succinctly, the classical
“horse-shoe” arrangement of other VLM programs has been replaced with a “vortex-sling”
arrangement. It basically works in the same way as the “horse-shoe” procedure with the
exception that the legs of the shoe are flexible and consist of seven (instead of three)
vortices of equal strength. Since the primary assumption of any VLM is linearity, two seed
computations are conducted for the lifting surface system at angles of attack (AoA or α)
where collinearity is likely as depicted in Figure 23 and labelled as Step 1; two such
candidates are suggested as α = 0° and +4°.
Following the protocol mapped out in Figure 23, the next step is to identify the zero-
lift AoA (α
oL
); this is found by extrapolating the lift-curve slope (dC
L
/dα) back to the point
at which C
L
= 0. The slope dC
L
/dα itself is quantified by comparing the computed VLM
lift at the two seed AoA VLM calculations. Wing lift carry-over into the fuselage body can
be accounted for by factoring the original (wing only) dC
L
/dα with a calibrated variation of
CONCEPTUAL AIRCRAFT DESIGN METHODS

76
α
oL
L
i
f
t

C
o
e
f
f
i
c
i
e
n
t
,

C
L
Angle of Attack, α (deg.)
43 − 2AR
ref
3
4° x dC
L

dC
L

∆α = 10°
Vortex-Lattice Calculations
Empirical Algorithm
1
2
3
4
α
stall

Figure 23. Predicting the lift characteristics of a clean finite wing using quasi-analytical
techniques (1-g stall concept shown).

a method given by Pitts et al
96

wing
L
vehicle
L
d
dC
d
dC
α
ξ =
α
(134)

where

gross
2
h
wing
L gross
net h
S
d
d dC 2 S
S
b
d
1
α
π
+
|
.
|

\
|
ς + = ξ (135)

is related to the fuselage external maximum width (d
h
), the net or exposed wing planform
area (S
net
) and the gross wing planform area (S
gross
). The parameter ς is a calibration
constant and was derived to equal 3.2. As a final point, Pitts et al stipulates that the use of
Eqn. (135) is only applicable for wing-body configurations not violating the constraint of
d
h
/ b < 0.2.
From known data
3,97-101
, Step 3 involves an AoA increment of ∆α = 10° to yield an
estimate of the cessation of the linear portion of the curve (usually around α = 8°) or the
beginning of non-linear lift leading eventually to stall. The final step involves adding 4°
times the vehicular dC
L
/dα to the now corrected C
L
computed for Point 3 in Figure 23 to
predict the clean wing C
Lmax
adhering to a 1-g stall concept, or, simply given as

( )
vehicle
L
regs max L
d
dC
064 . 0 1 14 C
α
Φ + =
o
(136)

LOW-SPEED & HIGH-SPEED AERODYNAMICS

77
When s = 1, the impulse function, Φ
regs
= Φ(s,1), introduces a multiplier derived from
information presented by Obert
3
, otherwise is zero for s < 1. An appropriate parameter
value is invoked in accordance with the analysis being conducted, i.e. under the premise of
a power-off 1-g stall concept (s = 0), or, the minimum speed in a stall manoeuvre in
accordance with FARs (s = 1) respectively.
If the value is of interest, the corresponding AoA for stall (α
stall
) can be estimated as
well. A suggested empirically derived method based on the same data
3,97-101
quoted earlier.
Working off the equivalent reference wing aspect ratio as the only independent variable for
analysis, α
stall
is found by incrementing the AoA at Point 3 shown in Figure 23 by (43 -
2AR
ref
) / 3, or alternatively put, by combining all the steps detailed above can be simplified

3
AR 2 73
oL stall

+ α = α (137)

Eqn. (137) is taken to be applicable for the 1-g stall concept only. Since the AoA for
stall will differ between the 1-g stall break and minimum speed in a stall manoeuvre, it is
suggested that Eqn. (137) be incremented by an additional ∆α ≈ 1.0° to model the
minimum speed (FARs) in stall manoeuvre AoA.

7.1.2 Maximum Lift Generated by Trailing and Leading Edge
High-Lift Devices
High-lift produced by flap and slat deflection is estimated based on methods presented
by Young
102
. This reference uses empirical correlation from assorted accumulated data and
predicts with adequate accuracy the aerodynamic characteristics of high lift devices. The
methods are not explained in great detail here; however, the salient features will be
appropriately noted. A similar and more detailed working account may be found in a
design review done by Pazmany
103
and Isikveren et al
104
.
Making allowances for effective chord, flap incidence and part span, the increment
due to the presence of any trailing edge flap is given by

) ( f ) 1 c c ( C
) 6 ( F
) AR ( F
) c c ( C C
W max L L flaps L
Λ

− ′ + ′ ′ ∆ = ∆ (138)

where (c´/c) is the effective chord ratio; F(AR) is the function relating the vehicular
dC
L
/dα and the aspect ratio, and this is standardised to an AR = 6.0; C
LmaxW
is the
maximum clean wing lift attainable, f (Λ) is a correction to the lift increment for a swept
wing, and

| |
| |
∆ ′ = +
− − +
C c c c c
b b b b b b b b
L f f
f f f f
λ λ β λ λ β
λ λ λ λ
1 1 2 1 1 2 22 22
3 22 3 21 3 12 3 11
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )

(139)

λ
1
(c
f
/c) is a function of effective chords, λ
2
(β) is a function of the flap angle and is
determined from experimental data (varies from one flap to another). The subscript 22
denotes the influence of an auxiliary flap or vane if applicable. The operation [λ
3
(b
fx2
/b) -
CONCEPTUAL AIRCRAFT DESIGN METHODS

78
λ
3
(b
fx1
/b)] is a part span correction factor, and, x = 2 and 1 define the outboard and inboard
(due to a central cut-out) ends respectively.
The first task is to take Eqn. (138), it’s coupled constituent Eqn. (139), and introduce
not only the fixed functional values related to design intent supplied by Young, but a
parameter to account for the stall concept adopted per chosen airworthiness regulations.
Additionally, by incorporating supplementary simplifications for sake of brevity, i.e. linear
sensitivity to AR, an all-purpose fixed quantity for effective chord, introduction of a
continuous functional form for the f(Λ) correction parameter, the final algorithm describing
change in lift due to trailing edge device deflection is proposed here as

( )
Qchd
3
flap flap geo
fowl dslot
TE
flaps L
cos 1 b 3 AR k
20
5 20
C Λ − β |
.
|

\
| Φ + Φ +
= ∆
(140)

The two design related impulse functions, Φ
dslot
= Φ(s,1) and Φ
fowl
= Φ(s,1), represent
the relative increase in lift compared to the default single-slotted flap prediction assuming
double slotted of Douglas type and Fowler flapping arrangements respectively. The
constant k
geo
is equal to 2.183 x 10
-3
and is universally applicable for all (chord extending)
flaps considered. The flap deflection angle in degrees is denoted by β
flap
with b
flap
defining
the part-span flap including fuselage carry-through, expressed as fraction of total reference
wingspan.
A series of fixed flap settings corresponding with deflection optima based on
experimental results given in literature
1,3-5,39
for given high-lift device types have been pre-
selected for field calculations. Single slotted flaps tentatively have pre-designated
deflection optima of 7
o
, 15
o
and 35
o
for intermediate takeoff, maximum takeoff and landing
configurations respectively. For double slotted flaps of Douglas type, initial guesses for
optimal flap deflections have been assumed to be approximately 10
o
, 20
o
for intermediate
and maximum takeoff, and 45
o
for landing. Congruous with the double slotted premise, the
Fowler assumes 10
o
, 20
o
and 45
o
for intermediate takeoff, maximum takeoff and landing
configurations respectively. Although optimal flap deflection is dependent upon a given
vehicular configuration and ambient conditions in which the aircraft operates, these
selected values were found to be very close to actual deflections used on contemporary
aircraft and hence adopted for simplicity. Regardless of this directive, the algorithm used to
determine C
Lmax
given above permits an opportunity to truly optimise flap setting for the
operational performance scenario considered; providing an extension is made to allow
cubic interpolation of C
Lmax
for the given intermediary flap setting.
These trailing edge high-lift devices may also be complemented by the introduction of
leading edge slats. Occasions where a slat lift increment is desired, a tentative maximum
deflection of 20
o
is assumed based on experimentation and actual examples
64,97,105
. The
increment in lift due to slat is only introduced for maximum lift prediction, i.e. maximum
optimal flap deflection usually pertaining to landing configuration. Furthermore, an upper
permissible boundary of C
Lmax
= 3.50 which is universally applicable to all devices has
been artificially set in keeping with conclusions drawn from surveys presented by Obert
3
.
Young
102
suggests a rather simplified expression relating lift increment due to slat to the
slat wing chord fraction. In the end, a more consistent approach exhibiting functional
similarity with Eqn. (140) was chosen to be a more accurate model

Qchd
3
flap geo
LE
flaps L
cos b AR k C Λ = ∆ (141)
LOW-SPEED & HIGH-SPEED AERODYNAMICS

79
where all other parameters retain the previously given definitions, except for k
geo
, now
taken to be 0.0470, and b
flap
is the slat part-span fraction.
To complete the entire prediction exercise, a trimmed lift coefficient needs to be
produced. As outlined by McCormick
34
a complete treatment involves augmenting
untrimmed vehicular lift coefficient according to the relative distance between vehicular
centre of gravity (x
cg
) and aerodynamic centre (x
ac
) locations, and then incrementing
contributions due to generated moment coefficient about the aerodynamic centre and the
moments created because of increase in drag due to trim. Such an approach requires a
detailed array of information; to simplify matters, sufficient accuracy can be achieved by
dropping the terms dependent upon moment coefficient and increase in drag.

( )

− + =
ac cg
t
max L trim L
x x
l
MAC
1 C C (142)

Many aircraft manufacturers adopt the simplified functional form given by Eqn. (142)
in their respective aerodynamic data handbooks. Default values for the non-dimensional
relative MAC distance (x
cg
– x
ac
) can be assumed as -0.05 for aft-fuselage mounted
vehicles, otherwise equal to approximately -0.15 for all other configurations.

7.1.3 Establishing the Accuracy of Clean Wing and High-Lift Prediction
Once each of the analytical and empirical constituents is combined to form the final
algorithm, a wide-ranging analysis has shown predictions are relatively consistent with
actual aircraft lift data. Using a generic supercritical profile as a basis for this investigation,
namely the MS(1)-0313, Figure 24 elucidates this by demonstrating a typical bandwidth of

-0.3
-0.2
-0.1
0
0.1
0.2
0.3
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
Vehicle Actual C
Lmax
(-)
E
r
r
o
r
,

ε
.

i
n

P
r
e
d
i
c
t
e
d

C
L
m
a
x

(
-
)
TE (or LE) Flaps Neutral
Max TE (or LE) Flaps
ε = +10%
ε = +5%
ε = -5%
ε = -10%

Figure 24. Prediction accuracy of algorithm to compute C
Lmax
using quasi-analytical
techniques. High-lift device set to neutral and maximum deflection shown.
CONCEPTUAL AIRCRAFT DESIGN METHODS

80
error (ε = predicted – actual) with respect to manufacturer quoted values falls within a ±5%
splay. More saliently, the study indicates there exists a good likelihood maximum lift
predictions will not exceed an error of around ε = ±0.15 irrespective of flap deflection.
The benchmarking data comprised either known aerodynamic performance or was
derived from vehicular stalling speeds. The aircraft used for this validation exercise were:
Boeing BBJ1
76
; Bombardier Aerospace Learjet 45
78
, Learjet 60
106
, Challenger CL-604
51
,
Global Express
64
, CRJ200
79
, CRJ700
80
and CRJ900
81
; Cessna Citation Excel
82
; Dassault
Aviation Falcon 2000
107
and Falcon 900
53
; Embraer ERJ 135
108
, ERJ 140
109
, ERJ 145
84
;
Fokker Aircraft Fokker 70
110
and Fokker 100
111
; Gulfstream Aerospace GIV-SP
89
and GV-
SP
90
; PD340-2 19 PAX regional jet conceptual design study
112
; and, Saab Aerospace Saab
340
113
and Saab 2000
114
. Note that all aircraft assuming maximum flap deflection data
points are displayed in Figure 24; data pertaining to neutral flap deflection is shown where
the original manufacturer information was available.

7.2 Zero-Lift Drag Estimation - The Equivalent Length Method
A common method for determining the zero-lift drag (C
Do
) of aircraft components is
an assumption that the constituent’s friction drag is equivalent to a flat plate having the
same wetted area and characteristic length. In this way, a very preliminary assessment of
the complete vehicular zero-lift drag estimation may be accomplished by summation of
these individual components. By creating a hybrid approach where the component build-up
method is benchmarked against a standardised closed form expression, economy of effort
can be achieved without incurring excessive degradation in predictive powers. A tool for
estimating zero-lift drag is the friction coefficient equation based on experimentation done
by Eckert
115
, which accounts for fully turbulent flow and compressibility effects. By
assuming an appropriate reference condition of Mach number and flight level, the
component build-up method may be employed and a characteristic equivalent length for
the entire vehicle can be derived from its equivalent skin friction coefficient - a quantity
commonly used for aircraft comparison exercises. This equivalent characteristic length
may in turn be reintroduced into Eckert’s equation and solved for any other Mach number
and flight level combinations the aeroplane encounters.

7.2.1 Derivation of The Equivalent Characteristic Length Method
Assuming the boundary layer is fully turbulent and accommodating effects due to
compressibility on skin friction, the friction coefficient (c
f turb
) according to Eckert based
on wetted area is given by

( ) ( )
d
2 b
R
turb f
M c 1 N log
A
c
+
= (143)

where M is the instantaneous Mach number, constants A = 0.455, b = 2.58, c = 0.144 and d
= 0.58 are coefficients of proportionality derived by Eckert, and, the Reynolds number
(N
R
) in atmospheric flight at given speed and flight level can be expressed as

b
sls sls
sls
R
l V N
u u
σ
u
ρ
= (144)

The identity ρ
sls
/u
sls
is approximately equal to approximately 6.9x10
4
s/m
2
, and l
b
is
any specified representative length of the body.
LOW-SPEED & HIGH-SPEED AERODYNAMICS

81
The results obtained by an approximate turbulent theory such as the one given by Eq.
(143) assumes a smooth adiabatic flat plate. In actual flight conditions, typical values of
skin friction exceed the predicted value significantly. This circumstance does not
necessarily invalidate the use of Eckert’s equation, but rather, raises the requirement of
additional adjustments to reflect actual physical observations. The first correction calls for
account of an equivalent sand roughness. The traditional method utilises the concept of a
cut-off Reynolds number
4
, which is determined using the characteristic length and skin
roughness derived from a table of values presented for different surfaces. Other sizable
contributions to the final value of skin friction includes dissimilar boundary layer
development and velocity profiles between streamlined shapes and the flat plate analogy,
and, pressure effects due to frontal area. Instead of relying on a sequence of discretised
computations, the aim here is to formulate a single-step prediction procedure for skin
friction coefficient that can incorporate these adjustments.
Examination of Eq. (143) reveals the theoretical turbulent skin friction coefficient is
primarily a function of Reynolds number with a supplementary account of compressibility
effects. In view of this situation, any adjustment that takes into account actual-flight
corrections should be expressed as being proportional to Reynolds number, or,
algebraically incorporated into the (log N
R
)
b
term. With this idea in mind, Eq. (143) would

( ) | | | |
d
2 b
R act
f
M c 1 N log
A
c
+ η
= (145)

where the parameter η
act
= 1 produces a skin friction result synonymous with Eckert’s
original theory, otherwise, for values η
act
≠ 1 constitutes an additional correction to
represent equivalent sand roughness, pressure and interference effects. Based on an
elaborate amount of experimentation done in wind tunnel and flight-testing, Poisson-
Quinton
116
was able to quantify the difference between actual values of skin friction and
theoretical turbulent friction assuming a smooth adiabatic flat plate. The results showed a
simple linear proportionality between c
f
and c
f turb
, namely,

turb f act f
c c τ = (146)

By initially equating Eq. (145) with a factorised Eq. (146) using the binomial
construct, solving for the constant of proportionality, τ
act
, and then re-arranging the interim
result such that η
act
becomes the subject, the Reynolds number adjustment parameter
becomes

( )
R
b 1
act
N log 1
act
10
− τ

= η (147)

Assuming an actual flight Reynolds number of around 20 x 10
6
where τ
act
was found
to equal approximately 1.45 as cited in Poisson-Quinton’s results
116
, produces a correction
of η
act
= 0.105, which would then be introduced into the modified Eckert’s equation given
by Eq. (145). The Reynolds correction coefficient of η
act
= 0.105 can be thought of as a
“mean curve” adjustment, representative of conventional technology/manufacturing
CONCEPTUAL AIRCRAFT DESIGN METHODS

82
levels
‡‡
, and therefore has been presented as the basis for establishing predictions at the
very initial design stage. Consideration must also be given to the fact a practical lower
limit of τ
act
= 1.30 (or potential C
Do
reduction of up to 10% from the mean curve) has been
derived when analysing some narrow bodies and larger aircraft types from data supplied by
Obert
3
, and this factor is in turn synonymous with a Reynolds correction coefficient of η
act

= 0.197.
Eq. (143) represents a condition where fully turbulent flow exists. It would be prudent
to give scope in accommodating mixed laminar and turbulent flow, hence permit the
designer to set a minimum goal of what proportion laminar flow shall occur over the
characteristic length of the body constituent in question. Since an algorithm to quantify a
realistic turbulent skin friction coefficient has been established with Eq. (143), this can be
used as a basis to formulate an extension such that a realistic skin friction assuming mixed
flow is produced. Working off a basic assumption that momentum thickness at given
transition point is synonymous for both laminar and turbulent flows (see Figure 25), the
final skin friction can be produced by summing the friction coefficients for partly laminar
and turbulent flow
2
.

l
b

Figure 25. The premise of mixed laminar and turbulent flow used to derive an
augmented realistic skin friction coefficient
2
.

Matching the momentum thickness of the laminar and fully turbulent boundary layer
at transition point T gives

x c x c
turb f T lam f
∆ = (148)

where c
f lam
is the skin friction coefficient for laminar flow, x
T
is the point along the body
characteristic length where flow transition occurs and ∆x is a distance ahead of the
transition point where fictitiously the onset of fully turbulent flow takes place. It can be
shown
34
the total flat plate friction coefficient for a mixed laminar and turbulent flow is
calculated from

( )
lam f turb f
b
T
turb f f
c c
l
x
c c − − = (149)

In this equation, c
f turb
is computed assuming a Reynolds number based on a body
characteristic length starting from the fictitious onset of turbulent flow to the end of the

‡‡
The aircraft surface can have many irregularities. These include gaps and steps, protruding flush rivet
heads, and, surface waviness due to airframe construction, dynamic distortion and cabin pressurisation.
LOW-SPEED & HIGH-SPEED AERODYNAMICS

83
body, and, c
f lam
is calculated based on the entire length of assumed laminar flow, or
distance x
T
. Substitution of Eq. (148) into Eq. (149) can produce an alternate form

turb f
b
T
f
c
l
x x
1 c
|
|
.
|

\
| ∆ +
− = (150)

Since c
f turb
also depends on ∆x, an iterative procedure is required to solve for ∆x in
Eq. (150). A valid form of simplification is in order here. Introducing a presumption the
fictitious distance ∆x consistently exhibits linear proportionality with x
T
for low to mid-
range values of ∆x / l
b
, scope can be given to dispense with the transcendental nature of
Eq. (150), hence permit a reduction in complexity. Investigations found that for x
T
/ l
b

values less than approximately 0.40, the total skin friction coefficient for mixed laminar
and turbulent flow can alternatively be expressed as

turb f
b
T
mf f
c
l
x
1 c
|
|
.
|

\
|
χ − = (151)

The constant of proportionality, χ
mf
, assists in ascertaining what proportion of the
completely turbulent flow premise imparts an influence on the mixed flow result.
Experimentation has found a useful value for this parameter is approximately χ
mf
= 0.74
for all x
T
/ l
b
< 0.40. The upper boundary of assumed laminar flow fraction is a reasonable
one for design prediction purposes since an example of the most successful flight testing of
combined passive and active laminar flow control technology achieved laminar flow up to
30% of wing chord
117
. In addition, experimentation conducted in a more operationally
pragmatic sense commonly produces transition at 15% wing chord
117
.
The component build-up method for zero-lift drag at given Mach number and flight
level is given as

W
I
1 i
i
wet
i
f
h , M
Do
S
S c
C

=
= (152)

where the product
i
wet
i
f
S c is the drag area of each component i. By choosing an appropriate
reference condition of Mach number and altitude
§§
, an equivalent skin friction coefficient
representative of the entire vehicle can be produced with the congruent relation

∑ ∑
= =
ε

I
1 i
i
wet
i
f
I
1 i
i
wet f
S c S c (153)

The parameter
ε
f
c is the equivalent skin friction for the sum of all constituent wetted
areas produced using the equivalent flat plate analogy representing the entire aeroplane. It

§§
The reference condition for Mach and flight level is open to the designer’s willingness to trade larger errors
in low speed for more accurate high-speed zero-lift drag or visa versa. Experimentation has found that a
speed near the final vehicle MRC or LRC at an altitude 4000 ft lower than the intended certified ceiling are
good reference conditions for a balanced error distribution.
CONCEPTUAL AIRCRAFT DESIGN METHODS

84
is now proposed that this notion of equivalence can be extended to quantify a characteristic
length as well. Since the entire vehicle has been replaced by the flat plate premise with a
corresponding value for
ε
f
c , by rearranging Eckert’s equation, Eq. (143) can be solved for
an equivalent characteristic length (l
ε
) given by the identity

| |
V
10
l
sls sls
sls
M c 1
c
A b / d
2
b / 1
f
u u
σ
u
ρ
=

ε
+
|
|
.
|

\
|
ε
(154)

Reintroducing this relation to Eckert’s equation, and assuming the error in N
R
due to a
now fixed equivalent characteristic length (i.e. independent of Mach number or flight level
effects) is small, a general zero-lift drag equation, designated hereon as the Equivalent
Characteristic Length Method (ECLM), which accounts for all variations of Mach number
and flight level can be given approximately as

| |
W
wet
d
2
b
sls sls
sls
h , M
Do
S
S
M c 1 l V log
A
C
+

)

¹
¹
´
¦
u u
σ
u
ρ

ε
(155)

For a detailed analytical treatment of en route performance, drag is an integral
parameter and has the primary requirement of being differentiable with respect to the
airspeed V for all cases. Eq. (155) appears to be in a form that is quite complex, and more
poignantly, not configured for a more in-depth calculus treatment. It was identified that this
problem may be avoided via the use of logarithmic differentiation. By utilising the relation
x = e
ln x
, Eq. (155) can be alternatively expressed as

| |
¦
)
¦

¹
¦
¹
¦
´
¦
|
|
.
|

\
|
θ
+ −

)

¹
¹
´
¦
u u
σ
u
ρ
− =
ε
ε
2
sls
2
sls sls
sls
b
f
a
V
c 1 ln d l V ln ln b exp 10 ln A c (156)

which is in a form ready for differentiation albeit the complexity has not been reduced.

7.2.2 Gauging the Robustness of the Equivalent Characteristic Length Method
An interesting question is to what extent the equivalent characteristic length
assumption is compatible to the exact component build-up method, and, more importantly
what is the upper threshold of relative errors the designer may expect. In an effort to
theoretically gauge the magnitude of inherent errors produced by this approach, the ECLM
expression was reconfigured as an error function with respect to the exact component
build-up method. The most expedient way to observe this would be the comparison of
resultant equivalent skin friction errors analytically and do so for a range of contemporary
regional transport and business jet Reynolds number regimes based on complete vehicular
characteristic lengths. If Eckert’s general equation is partitioned into Reynolds number and
compressibility dependent constituents, in conjunction with some algebraic manipulation,
Eq. (143) then becomes

LOW-SPEED & HIGH-SPEED AERODYNAMICS

85

b
2
b
2
1
f
l log
1
c

ϖ
+ ϖ
ϖ
= (157)

where the compressibility term is described by

| |
d
2
1
M c 1
A
+
= ϖ (158)

and the Reynolds number dependent constituent is defined as

u u
σ
u
ρ
= ϖ V log
sls sls
sls
2
(159)

-10.0%
-5.0%
0.0%
5.0%
10.0%
0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5
Reynolds Number Based on Vehicular Characteristic Length (x10
6
)
R
e
l
a
t
i
v
e

E
r
r
o
r

o
f

V
e
h
i
c
u
l
a
r

Z
e
r
o
-
L
i
f
t

D
r
a
g

(
-
)

+40%
+20%
+30%
+10%
-10%
-40%
-30%
-20%
0%
+70%
+60%
+50%
Error in l
ε
Error in l
ε

Figure 26. Resilience of ECLM accuracy for a given error in vehicular characteristic
length and en route Reynolds number based on vehicular characteristic length.

Now, by introducing the notion of error factor defined as the ratio of the fixed
vehicular characteristic length quantity derived from a reference Mach and flight level to
the exact value of vehicular characteristic length, or ε
l
= l
ε
/l
exact
, the relative error of an
equivalent characteristic length assumption can be gauged by considering deviations from
the exact value of
exact
f
c through a fractional comparison

CONCEPTUAL AIRCRAFT DESIGN METHODS

86

b
R
l
f
f
N log
log
1
c
c
exact

ε

ε
+ = (160)

Figure 26 (previous page) shows the variation of resultant prediction error compared
to the exact vehicular equivalent skin friction of zero-lift drag with Reynolds number based
on vehicular characteristic length whilst assuming various errors in the ε
l
ratio. To put
Reynolds number based on vehicular characteristic length into context, small business jets
typically operate at around N
R
= 10
6
, regional aircraft and larger business jets between N
R

= 1.5 x 10
6
and 2.0 x 10
6
, and larger regional and narrow-body aircraft from N
R
= 3.0 x 10
6

and higher. For a typical en route Reynolds number of 1.5 x 10
6
based on vehicular
characteristic length for regional transports, an error of -24% in l
ε
compared to l
exact

corresponds to a +5% overestimation of equivalent skin friction or total zero-lift drag.
Conversely, for the same Reynolds number, a -5% underestimation of zero-lift drag is
tolerated by a +33% error in equivalent characteristic length from the exact value. This
result demonstrates the resilience of ECLM.

7.3 Vortex-Induced Drag at Subsonic Speeds
Many methods exist in quantifying this phenomenon and the most simplest of them is
the Oswald Span Efficiency Method which assumes the vortex-induced drag coefficient of
three dimensional wings with an elliptical lift distribution equals the square of the lift
coefficient divided by the product of the aspect ratio and π. Additional drag produced by
non-elliptical lift distributions is made by using the Oswald Span Efficiency Factor (e),
which effectively reduces the aspect ratio. The vortex-induced drag factor
35
is given as

e AR
1
C d
C d
2
L
D
π
=
|
|
.
|

\
|
(161)

Numerous estimation methods for e have been developed but they mostly tend to
produce optimistically high values compared values of real aircraft. Obert
3
offers an
empirically derived equation for the vortex-induced drag factor applicable for Mach
numbers greater than about 0.40, based on actual aircraft regardless of power plant
installation, assuming typical centre of gravity locales, inclusion of wing twist effects, and
compressibility effects neglected.

007 . 0
AR
05 . 1
C d
C d
clean
2
L
D
+
π
=
|
|
.
|

\
|
(162)

Eq. (162) does not appear to account for the distinction of power plant installation
philosophy, i.e. clean wing, underwing podded or on-wing nacelle configurations, and the
direct impact this has on span loading distribution. As an exercise, Eq. (162) was compared
to Eq. (161) and Oswald span efficiency factor solved for a variety known e values of
equipment with different power plant installation philosophies not covered by the
statistical survey. Interestingly, the continuous functional form offered by Obert seemed to
match the values for these known examples with an adequate degree of accuracy. This
leads the author to believe a correlation between aspect ratio and power plant installation
LOW-SPEED & HIGH-SPEED AERODYNAMICS

87
philosophy must exist, hence Obert’s regression analysis inherently accounted for this
association.
For field performance where Mach numbers typically range between 0.15-0.25, it is
apparent that a change in vortex-induced drag factor will take place due to a change in the
span-wise lift distribution due to flaps extending and deflecting
118
. Literature demonstrates
this variation is proportional to wing geometry, non-ellipticity of the span-wise lift
distribution of the basic wing, the effect of flap cut-out and lift carry over by the fuselage
1
.
Concurrent to this circumstance, there is also an additional physical effect that needs to be
addressed. There is a reduction in the vortex-induced drag factor with increasing flap
deflection, or alternatively, as flap deflection is increased, a reduction in the vortex-
induced drag for given C
L
occurs. This is attributable to an increasing benefit generated by
the slot-effect at greater deflections and amounts to a measure of boundary layer control
thus preventing separation. In order to acknowledge these known phenomena, the
implication is an incremental change in the vortex induced-drag factor needs to be
introduced to Eq. (163). Such a model is proposed here to be

007 . 0 000487 . 0
AR
271 . 0 05 . 1
C d
C d
C d
C d
C d
C d
C d
C d
flap
flap
2
L
D
flaps
2
L
D
clean
2
L
D
2
L
D
+ β −
π
Φ +
=
|
|
.
|

\
|
|
|
.
|

\
|
∆ +
|
|
.
|

\
|
=
|
|
.
|

\
|
(163)

The impulse function, Φ
flap
= Φ(β
flap
,1), invokes a correction to the vortex-induced
drag factor to signify an irregularity in the lift distribution due to deployment of high-lift
devices. Studies comparing the vortex-induced drag estimate generated using Eq. (163) to
low-speed drag polars of the Saab 340
113
and Saab 2000
114
aircraft found the correlation to
be quite adequate. This means the maintenance of sufficient accuracy can be expected
using the one algorithm in predicting the vortex-induced drag regardless of flaps neutral or
extended.

7.4 Three Dimensional Effects and Ancillary Drag Contributors
Five form factors that account for three-dimensional effects, ancillary interference, and
excrescences are reviewed here. These values are computed based on thickness-chord
ratios of the wing, horizontal and vertical tails, and, the fineness ratios of the fuselage and
nacelle. All of the form factors itemised below were derived from original expressions
developed for GASP
39
and subsequently modified to suit known data more appropriately.
The modified wing form factor reads as

|
.
|

\
|
+
|
.
|

\
|
+ = ϕ
4
m m
wing
c
t
240
c
t
4 2 421 . 0 (164)

with the horizontal tail surface re-defined to be

( )

|
.
|

\
|
+
|
.
|

\
|
+ ξ − + = ϕ
4
m m
ht htail
c
t
240
c
t
4 2 893 . 0 1 1 . 0 1 (165)

CONCEPTUAL AIRCRAFT DESIGN METHODS

88
The ξ
ht
parameter represents horizontal tail placement non-dimensionalised by d
v
with
respect to the vertical tail tip and FRP water-line. Similarly with the wing, the vertical tail
form factor was amended to read as

|
.
|

\
|
+
|
.
|

\
|
+ = ϕ
4
m m
vtail
c
t
240
c
t
4 2 5 . 0 (166)

The fuselage form factor is predicated by body slenderness ratio. Assuming a
streamlined fuselage without a blunt nose

3
fuse
v
v
fuse
fuse
l
d
60
d
l
0025 . 0 1
|
|
.
|

\
|
+
|
|
.
|

\
|
+ = ϕ (167)

and finally, the nacelle form factor is based on the premise of slenderness as well

|
|
.
|

\
|
+ = ϕ
nac
nac
nac
l
d
35 . 0 1 17 . 1 (168)

The prediction of a low-speed drag polar for field performance requires account of
contributions due to extended undercarriage and high-lift devices. In the absence of
detailed undercarriage sizing, the drag due to extension of undercarriage can be quantified
with adequate accuracy using statistical correlation from known data. Based on
information gleaned from McCormick
34
, a useful linear regression equation was derived to
be

( ) 294 . 0 W 10 x 85 . 2
S
1
C
TO
5
W
LG D
+ = ∆

(169)

The total aircraft drag of a configuration geared for field operation is also affected by a
profile drag contribution from extended flaps and slats. Assuming a given trailing edge
(and/or accompanying leading edge) high-lift device has been deployed, an approximation
for the incremental drag is suggested as

( )
flap
2
flap
f
flap
2
flap
4
W
flap Do
cos
b b
MAC
c
c
0416 . 0 005339 . 0 10 x 268 . 5
S
1
C β
|
|
.
|

\
|
|
.
|

\
|
+ β − β = ∆

(170)

and typical values for the relative flap chord fraction of c
f
/ c = 0.26 and c
f
/ c = 0.15 for
trailing edge and leading edge devices respectively are suggested as initial estimates.

7.5 Total Incremental Drag due to One Engine Inoperative
Condition
The One Engine Inoperative (OEI) condition appears to be mostly disregarded in
conceptual design literature. It is usually classified as a preliminary design problem
1,4

because yawing and rolling considerations become rather complex in nature since these
LOW-SPEED & HIGH-SPEED AERODYNAMICS

89
must be trimmed out by primarily the rudder and then aileron. Drag due to engine wind-
milling, airframe sideslip, incremental changes in normal force vortex-induced and profile
drag from control surface deflection, asymmetric slipstream effects and lift distribution
reconfiguration producing independent vortex-induced contributions all combine to
complicate matters. By examining the exact approach, a number of valid simplifications
may be incorporated in order to reduce the scope of detailed information required whilst
retaining strong predictive powers and objective function sensitivity with respect to the
design variables. Studies have shown that many of these constituent contributors can be
neglected with the exception of vortex-induced and profile drag generated by rudder
deflection.

7.5.1 The General One Engine Inoperative Drag Constituent
If one considers the OEI asymmetric condition, studies have shown that many of these
constituent contributors can be neglected with the exception of vortex-induced and profile
drag generated by rudder deflection. Figure 27 demonstrates the pertinent forces and
moments once this simplification is introduced. By assuming the vertical tail utilises a
symmetric profile and all rudder deflections during asymmetric flight will be below stall,
equilibrium is achieved via,

δ
R
L
R
l
vt
T
op
D
wm
y
eng
y
eng

Figure 27. Simplifications of forces and geometric considerations during the asymmetric
thrust condition.

( )
op wm eng vt R
T D y l L + = (171)

where y
eng
is the moment arm from fuselage centre line to the critical and windmilling
engines, D
wm
is the drag produced by the wind-milling engine, T
op
is the instantaneous
available thrust produced by the critical engine at instantaneous velocity V and l
vt
is the
vertical tail moment arm. The instantaneous lift (L
R
) generated by the flapped vertical tail
is
34

CONCEPTUAL AIRCRAFT DESIGN METHODS

90

R vt , L vt R
C S q L δ η τ =
α
(172)

with q denoting the dynamic pressure, S
vt
the vertical tail reference area, τ a flap
effectiveness factor, and η, a correction which accounts for the effects of viscosity.
The functions τ and η are generally derived empirically since they behave in a non-
linear fashion with chord fraction (c
f
/ c) and rudder deflection angle (δ
R
). If a
simplification is sought, McCormick
34
demonstrates that thin airfoil theory can be utilised
to produce adequate results but the functions are still represented by dependent variables.
Assuming a typical c
f
/ c value for the flapped vertical tail of around 0.3, an estimate of τ =
0.66 may be derived using Weissinger’s approximation
34***
. Furthermore, McCormick
shows at an upper deflection of δ
R
= 30°, a value of η = 0.74 would be appropriate. By
assuming some level of conservatism for smaller deflections, an overall flap effectiveness
of τη = 0.49 applicable to the complete range of angles would result. This figure can be
substantiated against Torenbeek’s
1
presentation of overall effectiveness factors derived
from experimental data for plain flaps.
The lift-curve slope characteristics (C
Lα.vt
) can be estimated by the Helmbold
equation
34
based on an approximate lifting surface theory with the effects of sweep

vt,Qchd
) accounted for by a first order cosine relation given by Torenbeek
1
.

Qchd , vt
2
vt
vt
vt , l vt , L
cos
AR 4 2
AR
C C Λ
|
|
.
|

\
|
+ +
=
α α
(173)

Assuming a thin airfoil, section lift-curve slope of C
lα.vt
= 0.110 per deg. (2π per rad)
is given theoretically, however, it was found an average of 0.088 per deg. (5.04 per rad)
taken from Abbott and Von Doenhoff
69
yields more realistic predictions. Thus, from linear
thin airfoil and lifting surface theory, the rudder deflection required for equilibrium of the
OEI asymmetric condition is given by

( )
vt vt , L vt
op wm eng
R
l C S q
T D y
η τ
+
= δ
α
(174)

From this basis, the possibility of accounting for the influence of minimum control
speed limitations on field length and initial climb performance can be introduced at the
conceptual level, and, methods to predict these quantities with respect to operational
performance will be addressed in the takeoff field performance discussion of this report.
Since the geometric characteristics for equilibrium of asymmetric thrust has been
quantified, the next step should be an appreciation of to what extent performance shall be
degraded. By summing the forces and moments in Figure 27, and equating these to
represent contributions of vortex-induced and profile drag due to rudder deflection, the
total incremental drag contribution produced by an OEI asymmetric condition (∆C
DOEI
) is
approximated by
39

| |
w
R
vt
eng
op wm wm
DOEI
S q
tan
l
y
T D D
C
δ + +
= ∆ (175)

***
Two point vortices represent the airfoil and this function is dependent upon c
f
/ c ratio.
LOW-SPEED & HIGH-SPEED AERODYNAMICS

91
This relation is not only applicable for low speed field performance, it can also be
utilised for climb out analysis as well; specifically in relation to OEI maximum attainable
flight level and drift-down net level off height proficiency trade studies at ISA and more
importantly off-ISA conditions.

7.5.2 Drag Generated by Windmilling Engine
For multi-engine aircraft with engines not buried in the fuselage, the OEI performance
will be influenced by additional drag due to a windmilling engine during the equilibrium
condition of asymmetric flight. Torenbeek
1
proposes a conceptual method to estimate the
magnitude of the drag increment by considering this quantity to be a function of engine
frontal area, bypass ratio and internal configuration. Unfortunately, the method is rather
esoteric because the procedure employs the momentum theorem, which requires an
estimation of mean flow velocity in the nozzle exit together with the windmilling mass
flow. Typical values for the ratio of these speeds are offered but they are specific to engine
type thus not allowing for a continuous function concept. As an alternative, a more
simplified approach is proposed which assumes the windmill drag component can be
accounted for by representing it as an equivalent flat plate problem with an associated skin
friction value which is imaginary and independent of Reynolds number variation or
associative compressibility effects.
The notion of a “cut-off Reynolds number” can be useful in helping to quantify the
drag produced by a windmilling engine in this respect. Raymer
4
discusses the merits of
employing a cut-off Reynolds number parameter to account for expected higher skin
friction coefficients in conventional zero-lift drag estimation when the surface of a body is
relatively rough. By comparing the ratio of characteristic length and a skin-roughness value
(l/k), the cut-off Reynolds number (N
R cut-off
) is then determined by

b
off cut R
k
l
a N
|
.
|

\
|
=

(176)

where a and b are constants of proportionality and N
R cut-off
varies monotonically with l/k
for subsonic speeds.
Assuming a windmilling engine is essentially the nacelle but influenced by some
degree of imaginary roughness on the body in this condition, i.e. analogous to an internal
drag contribution, then a pre-designated cut-off Reynolds number would be independent of
Mach number variation for subsonic flight
†††
and atmospheric conditions because as
indicated by Raymer, there is a strong correlation to relative roughness alone. It is evident
that the internal drag generated is related to maximum static engine thrust potential, which
also may be postulated to be a function of engine size. This would mean the imaginary
value for k would increase proportionately with nacelle physical dimensions, therefore, the
imaginary relative roughness can be taken as approximately constant. In view of this, the
imaginary cut-off Reynolds number can be considered independent of nacelle size or
characteristic length as well. When the imaginary cut-off Reynolds number is quantified
empirically and substituted into Eckert’s equation for skin friction given by Eq. (143),
neglecting compressibility effects and hence adopting the Prandtl-Schlichting form, the
equivalent flat plate skin friction that simulates an imaginary roughness condition for a
windmilling engine would be given as

†††
OEI flight regime is considered predominately as a subsonic problem. This premise may not hold true for
extended range and in some instances driftdown operations.
CONCEPTUAL AIRCRAFT DESIGN METHODS

92

| |
b
wm
off cut R
wm
f
N log
A
c

= (177)

where A is equal to 0.455 and b = 2.58. Once the windmilling engine representative skin
friction is quantified, an incremental contribution to drag in the OEI asymmetric condition
is therefore given by

W
nac , wet
wm
f wm
D
S
S c
C = ∆ (178)

In order to derive the value for imaginary skin friction, known windmill drag
properties for the BAe 146-200
119
were used and the results were tested against other
installed aircraft engines. The suggested values for conceptual analysis were found to be
wm
off cut R
N

= 9.3x10
4
, or corresponding imaginary skin friction of
wm
f
c = 0.007274. Using this
information in conjunction with the nacelle wetted area estimation methodology described
previously, predictions of
wm
D
C ∆ were computed and subsequently compared to known
windmill drag data for both the Williams International FJ44-2A
120
small turbofan rated at
10.2 kN (2400 lb.f) and the CFM56-7B26
121
engine rated at 118 kN (26400 lb.f) used on
the B737-800 narrow-body commercial transport. Figure 28 demonstrates the level of
accuracy generated using the imaginary skin friction method; the results were found to be
quite encouraging, more so due to the fact the nacelle wetted area was not calibrated to any
known data before computing the final result.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Mach number (-)
W
i
n
d
m
i
l
l

D
r
a
g

(
N
)
Williams FJ44-2 Actual Data
Williams FJ44-2 Prediction
CFM56-7B28 Actual Data
CFM56-7B28 Prediction

Figure 28. Benchmarking predicted windmilling drag using the imaginary skin friction
method against actual engine windmilling data; ISA, sea level conditions.
LOW-SPEED & HIGH-SPEED AERODYNAMICS

93
Additionally, it is recommended that the drag contribution for turboprop engines using
this method should be obtained by factoring the equivalent turbofan result by 3. This
accounts for a simulated by-pass ratio increase due to the presence of propeller or larger
fan diameter contribution for given maximum static thrust rating or nacelle size. One
should recall the method is based upon the generic pitot nacelle, therefore, the actual
nacelle wetted area for a turboprop (or even S-duct and straights ducts) power plant
installation must be disregarded and the generic pitot introduced into the prediction
process. Inspection of the Saab 2000’s one engine inoperative drag assuming a propeller in
the auto-feathered condition
114
produced an estimation error of –2.8%.

7.6 Compressibility or Wave Drag
Compressibility is a drag increment caused by an increase in free stream Mach number
above a critical point where locally accelerated speeds increase sufficiently to reach Mach
numbers of unity and above. The free stream Mach number at which this first occurs is
called the critical Mach number, denoted here as M
CR
, and can be thought of as the lower
limit of the transonic flow regime. Steadily increasing values of free stream Mach number
above M
CR
are characterised by regions of supersonic flow terminated by normal shock
waves shifting aft and increasing in strength. The formation of shocks in the transonic flow
condition affects the drag up to the drag divergent Mach number (M
DD
), thereafter the drag
rise rate increases substantially as shown in Figure 29.

Mach number
C
D
Constant C
L
Constant C
L
increasing C
L
M
CR
M
DD
∆C
D
= 0.0020

Figure 29. Definitions for the transonic mixed flow regime and indication of speed
thresholds for certain drag escalation attributes.

The definition of what particular Mach number constitutes M
DD
is open to several
options. The most common is the Boeing definition where M
DD
is the speed at which an
incremental increase in viscous drag influenced by drag rise is equal to 20 counts (or ∆C
DD
DD
is a function of lift coefficient since shock formation and
strength directly relates to increases in airflow velocity. Typically, for an initial analysis
the drag rise is graphically estimated using a few rules of thumb rather than a more
comprehensive appreciation of the dependence of M
DD
on parameters like instantaneous
operating lift coefficient, quarter chord sweep, mean wing thickness ratio and type of
airfoil geometry employed as exemplified by Raymer
4
. Notwithstanding, this assumption
CONCEPTUAL AIRCRAFT DESIGN METHODS

94
does not necessarily invalidate this first order estimate’s predictive powers - but a wide
range of information exists which aid in predicting compressibility drag characteristics for
given set of design parameters adequately. Therefore, it would be deemed prudent in
attempting to derive a closed form expression that describes the mixed flow regime
simultaneously neglecting highly non-linear terms but having a stronger basis to set more
realistic goals.

7.6.1 Derivation of the Incremental Drag due to Compressibility
described by many different empirical models. Torenbeek
122
offers a variation of Korn’s
equation
123
to quantify the limits of wing section performance for given vehicle wing
thickness, sweep and typical operating lift coefficient envelope

( )
Qchd
m
2 / 3
Qchd
2
L
Qchd DD REF
cos
c t
cos
C
10
1
cos M M
Λ
+
|
|
.
|

\
|
Λ
+ Λ = (179)

where M
REF
is a wing section technology factor. Torenbeek suggests values of M
REF
=
0.935 for supercritical aft loaded, and, M
REF
= 0.87 for conventional peaky sections. Here,
a modification of this premise with an empirical fit more akin to the actual performance
produced by contemporary regional and business jet vehicles is proposed: a customary
technology factor of M
REF
= 0.850 for supercritical aft loaded sections is suggested as a
more pragmatic value with an occasional upper limit not exceeding M
REF
= M0.90.
By rearranging Torenbeek’s version of the modified Korn’s equation so that M
DD
is
the subject

( )

Λ

|
|
.
|

\
|
Λ

Λ
=
Qchd
m
2 / 3
Qchd
2
L
REF
Qchd
DD
cos
c t
cos
C
10
1
M
cos
1
M (180)

Torenbeek
1
offers an arbitrary mathematical representation of the condition where
drag rise is terminated (at speed M
DD
) and an increased drag rise rate begins

n
DD
DD Dcomp
M
M M
1 C C

|
.
|

\
|

+ ∆ = ∆ (181)

where the symbols n = 2.5 and ∆M = 0.05 are given by Torenbeek; they have no physical
significance but are derived from experimental data. As a consequence, Eq. (181)
implicitly relates M
CR
to M
DD
as

M M M
CR DD
∆ + = (182)

This information can be used in conjunction with Eq. (180), thereby, allowing the
definition of M
CR
to be given as

LOW-SPEED & HIGH-SPEED AERODYNAMICS

95

( )
M
cos
c t
cos
C
10
1
M
cos
1
M
Qchd
m
2 / 3
Qchd
2
L
REF
Qchd
CR
∆ −
¦
)
¦
`
¹
¦
¹
¦
´
¦

Λ

|
|
.
|

\
|
Λ

Λ
= (183)

Now, by introducing the concept of an impulse function or approximate unit step that
is critical Mach number dependent, i.e. Φ
Mcr
= Φ(M,M
CR
), and incorporating Eq. (182) and
Eq. (183) into Eq. (181) yields

n
CR
M DD M Dcomp
1
M
M M
1 C C
CR CR

|
.
|

\
|

Φ + ∆ Φ = ∆ (184)

which then leads to a closed form expression for the total compressibility drag contribution
including the concept of initial and supplementary drag rise and the final equation
conforms to the presupposed condition of differentiability with respect to airspeed V.

7.6.2 Quantifying Wave Drag due to Volume and Lift
As expounded by Torenbeek
124
, the wave drag of wings and slender bodies is
frequently related to the theoretical minimum wave drag of pointed optimum bodies. Even
though the implicit assumption involves smooth bodies in inviscid flow, by utilising
linearised theory as Mach number tends to unity from below
125
, the relative merits of
differing configurations can be compared as a guide to drag-rise behaviour. These optimum
bodies can be represented by the von Karman ogive, Sears-Haack or Adams optimum
either in isolation, as composite area distributions in pairs, or even all three in consort.
Correcting for deviations from the optimum by a factor, K
o
, as stipulated by Kuchemann
126

and introducing an empirical wave drag efficiency factor
4
, η
opt
, representing the ratio
between actual wave drag and that of the optimum body, the wave drag due to volume for
given body volume V
b

4
b
2
b
W
o opt
M
Dcomp
l
V
S
128
K C
sREF
π
η = ∆ (185)

The product η
opt
K
o
can be estimated from values quoted by Raymer
4
and
Torenbeek
124
. This is accomplished by initially choosing a reference Mach number that is
slightly faster than sonic speed; and one suggested reference is M
sREF
= 1.05. Raymer and
Torenbeek indicate a combined factor of approximately 2.5 for η
opt
K
o
Although this value reflects supersonic designs displaying a relatively poor volume
distribution, analysis of actual subsonic aircraft (even those catering to M
MO
speeds up to
M0.90) found η
opt
K
o
= 4.0 is more appropriate. Once the wave drag due to volume has
been quantified for the reference condition, the next step is to build the wave drag model
according to the operating parameters dictated by given flight conditions. Taking the
logarithm on both sides of Eq. (181) and solving for the exponent n produces

CONCEPTUAL AIRCRAFT DESIGN METHODS

96

|
.
|

\
|

|
|
|
.
|

\
|

=
M
M M
log
C
C
log
n
CR sREF
DD
M
Dcomp
sREF
(186)

Eq. (186) is then substituted back into Eq. (181), hence, an estimate of the wave drag
due to volume can be computed dynamically for an instantaneous operating lift coefficient,
or alternatively, for a given critical Mach number premise.
The drag due to lift of surfaces at supersonic speeds (∆C
Di wave
) with streamwise and
spanwise elliptical pressure load distributions is quantified by Jones’
127
classical though
not universally accepted relation describing the lower bound

cbox
2
L W wave Di
r C K C λ
π
β
= ∆ (187)

where the working dimensions are shown in Figure 30. K
W
is a deviation from the
theoretical minimum and recommended as being equal to 1.25
124
, β = (M
2
- 1)
1/2
, r = S
W
/
(b l
W
) is a shape parameter, and, the so-called corrected box ratio is defined as λ
cbox
= β b /
(2 l
W
). The fundamental assumption here is that the fuselage nose and tail do not contribute
to lift.

Figure 30. Definition of working parameters to compute drag due to lift in supersonic
flight
124
.

7.7 Quantifying the Aerodynamic Impact of Winglets
With greater emphasis being placed on improving aircraft cruise efficiency winglet
devices appear to offer the most attractive combination of drag reduction and aesthetic
appeal. It is therefore not surprising many existing aircraft types have been outfitted with
winglets as part of an overall enhancement package and many kits are offered to retrofit in-
service aircraft. The conventional winglet (AR
WL
≅ 1.5) approach is now being replaced by
LOW-SPEED & HIGH-SPEED AERODYNAMICS

97
so-called blended winglets
128
; typified by a high aspect ratio (AR
WL
≅ 3.5) and integrated
by way of pronounced filleted transition geometry between the wing and winglet
structures.
Regardless of the design philosophy, the known benefits of winglets can be itemised
as follows:

• Decreased fuel burn and increased payload range attributes – achieved through an
aerodynamic performance improvement, i.e. net vehicular drag reduction;
• Higher cruise altitude and OEI driftdown ceiling – due to a net vehicular drag
reduction enabling a greater amount of specific excess power at given altitude and
speed;
• Improved takeoff performance – higher effective OEI lift-to-drag and therefore
higher second segment climb gradient for given reference speed; allows for higher
TOGWs;
• Reduced engine maintenance – the option of retaining the original takeoff
performance levels prior to installation of winglets promotes a reduced thrust
concept;
• Lower airport noise levels – exploiting the reduced thrust concept.

All of these enhancements may not necessarily come to fruition concurrently; the
designer should expect a combination of a few at best. Nonetheless, it is evident if
appropriately designed and integrated with the main wing, the devices will translate into a
some sort of a direct economic benefit for the operator.
Many examples of winglet performance prediction and design optimisation is
available in literature
34,70,71,129-131
. In general, winglet configurations are analysed using the
VLM to establish optimal planform attributes, cant, camber and twist in achieving
maximum reduction of vortex-induced during cruise. A non-planar, three-dimensional
potential flow panel method is subsequently employed to evaluate the configuration under
takeoff and landing operating conditions, thereby gauging the possibility of adverse low
speed characteristics. The revised span load is examined referenced to the ultimate wake in
a Trefftz plane analysis in order to determine the induced drag and bending moment
distribution. The final step in the design cycle is to weigh the economic feasibility of
adding winglets to the aircraft; ideally, a revised (and re-optimised) performance estimate
would entail consideration of the change in aerodynamic qualities and the change in
aircraft empty weight. To alleviate the need for excessive effort, only a relative drag is
quoted at given operating lift coefficient and change in OWE due to a wing bending
moment increase. For conceptual design studies, one suggestion is to adopt these
percentages and empirically adjust the design prediction accordingly with no due regard
given to winglet design variable sensitivity. Unfortunately, as one would intuitively expect
this approach is susceptible to inconsistencies. Therefore, a requirement now arises for a
quasi-analytical method to quantify the change in vehicular drag due to winglets.

7.7.1 Quantifying the Drag Reduction of Winglet Devices
As depicted in Figure 31, by summing the forces in the direction of freestream and
adhering to the sense convention indicated, the total force of the local system can be
quantified to be

α − α =
ind WL ind WL x
sin L cos D F (188)
CONCEPTUAL AIRCRAFT DESIGN METHODS

98
where α
inc
is the winglet representative incidence and α
ind
is the spanwise induced angle of
attack instantaneously generated by the wing.

α
ind
α
inc
Forward
Inboard
Wing
L
WL
D
WL
α
ind
V

L
W
L

s
i
n

α
i
n
d
D
WL
cos α
ind
+
α
eff

Figure 31. Resolving local lift and drag forces generated by the winglet into the direction
of freestream.

A key requirement is to now formulate a semi-empirical expression for α
ind
.
Fundamentally, the trailing vortex shed at each wing-tip induces not only a downward
velocity in the region of the wing itself, but the circulatory motion also generates induced
velocities in a spanwise direction. When the freestream velocity is vectorially added to the
spanwise induced velocity component in plan-view, the resulting vector produces an angle
of attack α
ind
. Prandtl’s lifting-line theory stipulates the downward induced angle of attack
generated by finite wings is proportional to the operating C
L
and inversely proportional to
the wing AR
132
. Working off this premise and introducing a coefficient of proportionality

ind
) to represent a scaling factor between the downward and spanwise induced velocities
towards the tips, then

AR
C
L
ind ind
π
η = α (189)

A suggested scaling factor η
ind
= 7.2 was empirically derived from flow visualization
133
.
Since the true goal is to quantify a relative vehicular drag, Eq. (188) should be
examined in the non-dimensionalised form, i.e. divided by qS
W
. To this end, the
instantaneous lift coefficient produced by the winglet (C
L WL
) is given by

( )
oL eff WL L WL L
C C α − α =
α
(190)

LOW-SPEED & HIGH-SPEED AERODYNAMICS

99
where the effective angle of attack α
eff
is found by taking the difference between α
ind
and
the winglet representative incidence angle α
inc
. An account of the zero lift angle of attack
α
oL
is assumed here to be approximately α = –3° for 3% cambered aerofoil sections
commonly used for winglet devices. The lift-curve slope characteristics (C
Lα WL
) can be
estimated using the Helmbold equation
34
modified for compressibility correction as given
by Eq. (88).
The total winglet drag is determined by summing the winglet zero-lift (C
Do WL
) and
vortex-induced (C
Di WL
) drag. The C
Do WL
contribution is derived using the component
build-up method with an adjustment for interference as outlined earlier. It is highlighted
that the incremental zero-lift drag due to presence of winglets must be considered in
isolation from the vehicular characteristic length and the ECLM drag prediction algorithm.
In this context, the winglet device is taken to be an add-on to an existing vehicle wing
planform, and therefore, is not deemed to be a constituent in deriving the vehicular
characteristic length.
Upon substitution of Eq. (189) and Eq. (190) into Eq. (188) now expressed in an
equivalent non-dimensionalised form, the total incremental drag due to presence of
winglets is determined by summing the resolved local winglet lift and drag force
components, the change in drag due to compressibility if the winglet pre-empts the wing in
generating super-velocities, and a reduction in the wing vortex-induced drag. Recognising
an adjustment required to conform to the reference wing convention

( )
( )
Di comp D M ind WL L ind WL Di ind WL Do
W
WL
Di comp D ind WL i ind WL o ind WL
W
D
C C sin C cos C cos C
S
S
2
C C cos D cos D sin L
qS
2
C
CR
∆ + ∆ Φ + α − α + α =
∆ + ∆ + α + α + α − = ∆

(191)

with the ∆C
D comp
component considered to be greater than zero if the winglet M
CR
has
been exceeded by the freestream Mach number. By virtue of attaching winglets to the tips
of a wing, one would expect an alteration to the spanwise lift distribution and the trailing
vortex system downstream since circulation along the wingspan changes accordingly. In an
attempt to quantify the relative reduction in vortex-induced drag due to presence of
winglets (∆C
Di
) in the flow field, a useful basis is to refer to the fractional change in the
vortex-induced drag factor used to augment the original ∆C
Di
denoted by the subscript
“orig”

orig
Di 2
L
D
Di
C
dC
dC
C O = ∆ (192)

The fractional change operator for the vortex-induced drag factor, namely
OdC
D
/dC
L
2
, is quantified by comparing the original wing planform and an equivalent
wing planfom with winglets canted as some angle Γ
WL
off the vertical. By incorporating
the vortex-induced drag factor derived by Obert
3
and given by Eq. (162) the operator
becomes

CONCEPTUAL AIRCRAFT DESIGN METHODS

100

( )
( )
|
.
|

\
| π
+

=

=
orig rev
rev orig
orig
2
L D
2
L D
2
L
D
AR
150
1 AR
AR AR
dC dC
dC dC
dC
dC
O (193)
where for winglet span (or height) and root chord of h
WL
and c
WL
respectively the revised
aspect ratio (AR
rev
) is defined purely on the basis of geometry

( )
( )
WL WL WL R WL W
2
WL WL
rev
tan 1 c h S
tan h 2 b
AR
Γ λ + +
Γ +
= (194)

7.7.2 Proficiency of Drag Reduction due to Winglet Prediction
Figure 32 shows a comparison of the calculated and actual improvement in block fuel
for a Boeing B737-800 narrow-body transport. Actual data derived from flight-testing was
71
. With regards to the exercise of
predicting a change in block fuel due to presence of winglets, a calibrated drag model
assuming no wing tip device was created from information generated by Boeing
121
and
subsequently contrasted against an assumption of winglets installed. Some precision is lost
for short-range missions, i.e. 500 nm and less, which are characterised by lower operating
lift coefficients (C
L
< 0.5); nonetheless, the agreement for the B737-800 appears to be
mostly a good one.

0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
0 500 1000 1500 2000 2500 3000
Stage Length (nm)
B
l
o
c
k

F
u
e
l

I
m
p
r
o
v
e
m
e
n
t

(
-
)
737-800 Winglet Actual
737-800 Winglet Predicted
Brochure OEW
Typical Mission Rules
LRC Mach

Figure 32. Comparison between flight-test derived
71
and predicted improvement in block
fuel for B737-800 commercial transport.

One undesirable feature of this method is the fact ∆C
Di
approaches zero with
decreasing Γ
WL
; this analytical sensitivity does not parallel the winglet parametric study
results presented by Ishimitsu
70
. The problem can be allayed by stipulating an accepted
LOW-SPEED & HIGH-SPEED AERODYNAMICS

101
design protocol of winglet integration not violating a minimum cant angle or lower
threshold of Γ
WL
. Cant angles less than approximately 15° are not permissible because it is
indicative of a less pronounced rate change improvement in C
Di
with respect to wing root
bending moment, and of equal importance, it gives less scope to provide aerodynamic
interference relief between wing and winglet and is detrimental in delaying the formation
of shock waves on the winglet upper surface.

7.8 Validation of the Total Aerodynamic Drag Model
The aerodynamic performance characteristics of known contemporary aircraft were
available to validate the predictive powers of the methods discussed – henceforth referred
to as the Combined Drag Model (CDM). Figure 33, Figure 34, Figure 35 and Figure 36
show the agreement between predictions using CDM and flight test drag polars for the
Saab Aerospace AB Saab 2000
114
, Bombardier Aerospace Learjet 60
134
, Bombardier
Aerospace Global Express
135
and Boeing B737-800
121
respectively. Each chart indicates
two zones of prediction effectiveness: “Infrequent Excursions” alludes to operating points
within the certified aircraft flight envelope that are seldom impinged during typical
operation, i.e. very low and very high operating lift coefficients, whereas, the inner
boundary labelled “Core Predictions” are points that will always need to be considered
during the course of examining the viability of a design candidate from an operational
performance perspective. By virtue of conducting a validation exercise that encompasses
aircraft of varying size, mission role and even power plant installation philosophy, the
results indicate there exists a good likelihood that CDM will produce predictions well
within ±10%, and it is discernable that core predictions will stay within an acceptable ±5%
error bandwidth.

0.30
0.35
0.40
0.45
0.50
0.55
0.60
-10.0% -5.0% 0.0% 5.0% 10.0%
Prediction Error for Total Drag
M
a
c
h

n
u
m
b
e
r

(
-
)
Core Predictions
Infrequent Excursions
M
mo
Boundary
LRC Speed

Figure 33. CDM prediction effectiveness inspected for the Saab 2000 high-speed
turboprop regional transport.
CONCEPTUAL AIRCRAFT DESIGN METHODS

102

0.60
0.65
0.70
0.75
0.80
-10.0% -5.0% 0.0% 5.0% 10.0%
Prediction Error for Total Drag
M
a
c
h

n
u
m
b
e
r

(
-
)
M
mo
Boundary
Core Predictions
Infrequent Excursions
LRC Speed

Figure 34. CDM prediction effectiveness inspected for the Learjet 60 midsize turbofan

0.60
0.65
0.70
0.75
0.80
0.85
0.90
-10.0% -5.0% 0.0% 5.0% 10.0%
Prediction Error for Total Drag
M
a
c
h

n
u
m
b
e
r

(
-
)
Core Predictions
Infrequent Excursions
M
mo
Boundary
LRC Speed

Figure 35. CDM prediction effectiveness inspected for the Global Express ultra long
LOW-SPEED & HIGH-SPEED AERODYNAMICS

103

0.60
0.65
0.70
0.75
0.80
-10.0% -5.0% 0.0% 5.0% 10.0%
Prediction Error for Total Drag
M
a
c
h

n
u
m
b
e
r

(
-
)
Core Predictions
Infrequent Excursions
M
mo
Boundary
LRC Speed

Figure 36. CDM prediction effectiveness inspected for the B737-800 narrow-body
commercial transport; note that τ
act
= 1.30 used in generating the reference
condition.

CONCEPTUAL AIRCRAFT DESIGN METHODS

104

intentionally blank

LOW-SPEED & HIGH-SPEED AERODYNAMICS

105

Jan 05
Section 6 – Initial Sizing & Analysis Techniques
25
Tier II Low-speed & High-speed
Aerodynamic Prediction (cont.)

Tier II Low-speed & High-speed Aerodynamic Prediction
The importance of predicting low-speed and high-speed aerodynamic qualities of aircraft cannot be understated
Vehicular definition relates to an initial appreciation of how the flight envelope will look It is one of the integral components in formulating airplane operational performance attributes

Prediction of low-speed and high-speed aerodynamic attributes covers the following categories
Low-speed aerodynamics
Clean wing lift characteristics and maximum lift Maximum lift generated by trailing and leading edge high-lift devices

High-speed aerodynamics
Zero-lift drag Vortex-induced drag at subsonic speeds 3D effects, trim and ancillary drag contributors Total incremental drag due to OEI condition Compressibility or wave drag due to volume and lift Aerodynamic impact of winglets Buffeting qualities

Jan 05

Section 6 – Initial Sizing & Analysis Techniques

2

Tier II Low-speed & High-speed Aerodynamic Prediction (cont.)

Jan 05

Section 6 – Initial Sizing & Analysis Techniques

Primary and secondary control surfaces and forces on an airplane
3

Tier II Low-speed & High-speed Aerodynamic Prediction (cont.)
CLmax is the maximum lift coefficient the wing can generate

CL, Lift Coefficient

CLmax landing CLmax takeoff CLmax clean

α, angle of attack
CLmax is dependent upon
Wing sweep Wing aspect ratio Wing thickness-to-chord Flapping span and flap deflection angle High-lift device configuration

In conceptual design, CLmax is often predicted by inspecting other aircraft of similar configurations; as a general rule
Empirical methods are well suited to giving results with an adequate level of accuracy for conventional aircraft configurations and technology levels The primary goals are for highest (L/D)TO and (D/L)LD Predictions should not exceed approximately CLmax = 3.50 unless suitable justification has been established Parametric analysis techniques can be utilised to confirm the validity of prediction results
Jan 05

Section 6 – Initial Sizing & Analysis Techniques

4

Tier II Low-speed & High-speed Aerodynamic Prediction (cont.)
An expedient method to establish clean wing CLmax and lift-curve geometry
First identify the 3D CLα using the Vortex-Lattice method; closed-form Helmbold method is good enough as well Predict the zero-lift angle-of-attack; can read off 2D test data results as an initial guess; non-linear lift is predicted to commence at αoL + 10° Use the algorithm CLmax = 14 dCL/dα to estimate the maximum lift coefficient for 1g stall

Lift Coefficient, CL

4 4° x dCL dα 3 αstall 43 − 2ARref 3 1

dCL dα

Vortex-Lattice Calculations Empirical Algorithm

αoL

2 ∆α = 10° Angle of Attack, α (deg.)

Predicting the lift characteristics of a clean finite wing
Section 6 – Initial Sizing & Analysis Techniques 5

Jan 05

Tier II Low-speed & High-speed Aerodynamic Prediction (cont. 1996 6 .) Ref: AGARD CP-102 Boeing 747 F-28 Mk 4000 Examples showing distinction between 1g and minimum aerodynamic stall definitions Jan 05 Copyright  2005 by Askin T. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques Ref: Some Aspects of Aircraft Design and Aircraft Operation Obert.

Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 7 .Tier II Low-speed & High-speed Aerodynamic Prediction (cont.) Note the reference configuration Use fractional change theory to predict the ∆CLmax of alternative layouts Jan 05 Copyright  2005 by Askin T.

Tier II Low-speed & High-speed Aerodynamic Prediction (cont. Isikveren All Rights Reserved 8 . 1982 Method to estimate lift-to-drag ratio of design candidates with high-lift devices deployed Section 6 – Initial Sizing & Analysis Techniques Jan 05 Copyright  2005 by Askin T. is detrimental to climb Ref: Delft University Press Synthesis of Subsonic Airplane Design Torenbeek. hence.) Lift-to-drag ratio during takeoff manoeuvers Instantaneous OEI climb gradient at V2 speed can be predicted using the parametric correlation below Increasing the incremental lift with high-lift devices has a tendency of reducing the available lift-to-drag ratio.

Isikveren All Rights Reserved 9 .) Ref: 1981-6 – No. 1981 Details of wing planform. airfoil section and twist distribution geometry for A310 transport Section 6 – Initial Sizing & Analysis Techniques Jan 05 Copyright  2005 by Askin T. 91 L’Aeronautique et L’Astronautique.Tier II Low-speed & High-speed Aerodynamic Prediction (cont.

) Itemized breakdown of total drag and physical explanation of origins Jan 05 Copyright  2005 by Askin T. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 10 .Tier II Low-speed & High-speed Aerodynamic Prediction (cont.

003500 Mean Line 0.002500 Advanced Passive or Active Methods 0. pressure & interference Mixed (laminar) flow adjustment can be incorporated thereafter Component build-up method is used to generate reference condition 0.005000 Small Regionals & Small Business Jets Vehicular Equivalent Skin Friction Coefficient (-) 0.ft) Survey of wetted areas and equivalent skin friction coefficients Jan 05 Copyright  2005 by Askin T.) Predicting zero-lift drag Basis is modified Eckert’s equation for skin friction incorporating a Reynolds number adjustment parameter cf = [log(ηactNR )]b [1 + cM2 ]d Reynolds number Mach number A equiv.Tier II Low-speed & High-speed Aerodynamic Prediction (cont. sand roughness.002000 0 5000 10000 15000 20000 25000 30000 35000 40000 Vehicle Wetted Area (sq.004500 Large Regionals & Large Business Jets Narrow-bodies 0.004000 Unacceptably Excessive Wide-bodies 0.003000 0. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 11 .

007  dC 2  πAR  L  clean vortex-induced drag factor ref. nacelle and other appendages OEI asymmetric drag estimation Windmilling drag estimated using “imaginary cut-off Reynolds number” It is an imaginary skin roughness (l/k) independent of engine size Assuming this roughness level an equivalent skin friction is computed using the Prandtl-Schlichting form of Eckert’s equation Drag due to asymmetry is then based on equilibrium of moments Dwm LR yeng yeng D wm + D wm + Top ∆CDOEI = q Sw [ ] y eng l vt tan δR Top lvt δR Jan 05 Copyright  2005 by Askin T.4)  dCD  1. aspect ratio Reduction in dCD/dCL2 due to slot-effect needs to be modeled as well Incremental drag due to 3D effects and ancillary drag contributors Most common method is form factors that account for 3D effects Ancillary interference Excrescences Trim (goal should be keep it small) These values are computed based on thickness-chord ratios of the wing.05   = + 0. horizontal and vertical tails.) Predicting vortex-induced drag Obert’s empirical method is suitable for subsonic analysis (M>0. and. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 12 .Tier II Low-speed & High-speed Aerodynamic Prediction (cont. the fineness ratios of the fuselage.

) Predicting wave drag Difference in zero-lift drag coefficient between the fastest Mach number (less than M =1. wing quarter chord sweep margin to divergence Mach Empirical exponential equation is then utilised to model the geometric increase in drag within the drag rise and divergence regimes Supersonic wave drag accounts for contributions due to volume displaced by the vehicle as well as lift distribution Jan 05 Copyright  2005 by Askin T.Tier II Low-speed & High-speed Aerodynamic Prediction (cont.0020 MCR MDD Constant CL Mach number Can produce reasonable initial estimate of Critical Mach using modified Korn’s equation airfoil technology operating lift coefficient mean wing thickness MCR  1  =  cosΛ Qchd  3/2   (t c )m   − ∆M CL 1    −  MREF −  cos Λ Qchd   10  cos 2 Λ Qchd       ref. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 13 .0) & Critical Mach is defined as transonic wave drag CD increasing CL Constant CL ∆CD = 0.

Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques Ref: Some Aspects of Aircraft Design and Aircraft Operation Obert. 1996 14 .) Suggested target design and off-design characteristics Jan 05 Copyright  2005 by Askin T.Tier II Low-speed & High-speed Aerodynamic Prediction (cont.

30 can be reached without encountering buffet Free from buffet within the operationally expected envelop is desirable Jan 05 Section 6 – Initial Sizing & Analysis Techniques 15 Copyright  2005 by Askin T.Tier II Low-speed & High-speed Aerodynamic Prediction (cont. 2002 Ref: AIAA 88-2043 The Integration of CFD and Experiment: An Industry Viewpoint Bengelink. and. maximum cabin pressure differential are considered Buffeting is characterized by Breaks in CL-α. cm-α or cx-α curves and emergence of pressure divergence on any of the lifting surfaces or fuselage Ref: AIAA-2002-0002 Design of the Blended-Wing-Body Subsonic Transport Liebeck. 1988 Explanation of buffeting envelop for transport aircraft The derivation of these boundaries are commonly performed using extrapolated wind tunnel data to full-scale and subsequently verified with flight testing Initial prediction methods can become mathematically quite extensive which do not easily lend themselves to simplification In reaching and surpassing the threshold for buffeting the aircraft must permit full controllability This means flow separation on a swept wing at high Mach number should not initiate too far outboard to prevent strong roll or pitch-up tendencies Airworthiness rules stipulate cruise flight has to be limited to lift coefficients where n = 1. Isikveren All Rights Reserved .) Buffet Envelop It is an additional en route limitation to the aircraft flight envelop Defines an upper threshold of flight level after an appreciation of climb and cruise specific excess power residuals.

) Predicted and flight test derived buffet boundary for L-1011 Buffet boundary for MD80 transport Jan 05 Copyright  2005 by Askin T.Tier II Low-speed & High-speed Aerodynamic Prediction (cont. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques Ref: Some Aspects of Aircraft Design and Aircraft Operation Obert. 1996 16 .

e. AR=1.e.5 Blended winglets typified by a high aspect ratio (AR=3. allows for higher TOGWs Reduced engine maintenance – the option of retaining the original takeoff performance levels prior to installation of winglets promotes a reduced thrust concept Lower airport noise levels – BBJ with Aviation exploiting the reduced thrust concept Vortex Generators Partners’ winglet Flow over a lifting surface may tend to separate prematurely leading to stall. greater drag or even noise The separation can be either chordwise or spanwise Separation can occur at low-speed or high-speed (transonic flow) Jan 05 Copyright  2005 by Askin T. i. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 17 .5) and integrated by way of pronounced filleted transition geometry between the wing and winglet structures Benefits of winglets can be itemized as follows Decreased fuel burn and increased payload range attributes – achieved through an aerodynamic performance improvement. net vehicular drag reduction Higher cruise altitude and OEI drift-down ceiling – due to a net vehicular drag reduction enabling a greater amount of specific excess power at given altitude and speed Improved takeoff performance – higher effective OEI lift-to-drag and therefore higher second segment climb gradient for given reference speed. i. diminished control authority. either lead to successful operation and/or certificated airworthiness Winglets With greater emphasis being placed on improving aircraft cruise efficiency winglet devices offer the most attractive drag reduction Another reason for selecting winglets is the aesthetic appeal There are two categories The conventional winglet.Aerodynamic Devices These are appendages that either enhance performance or fix problems.

at least establish a rationale that employing them is the best compromise Jan 05 Copyright  2005 by Askin T. or.Aerodynamic Devices (cont.) Examples of vortex generators for high-speed (GV left) and lowspeed (Legacy right) To correct this situation a series of vortex generators or vortilons near the wing or control surface leading edge are usually installed These energize the airflow over the surface and thereby assist in delaying the onset of flow separation This is a common solution to imperfections like poor manufacturing tolerance Thin plates attached to engine nacelles or along the forward portion of the fuselage body are called strakes – also shed vortices to energize local flow or even correct directional stability at high angles of attack Not a desirable solution Can be avoided for the wing if thoughtful consideration is given to wing thickness. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 18 . section contour distribution and washout Measure of insufficient upfront work done on a new design if artificial devices are employed to fix problems during flight testing Perpetual strides in CFD capabilities will have a tendency to minimise use of vortex generators.

may also lead to unacceptable high-speed drag penalty Do not require these when leading edge high-lift device is used Wing Fences Act as barriers to deter cross-flow.Aerodynamic Devices (cont. and. thereby possible separation which could lead to tip stall High-speed drag penalty Ventral Fins Are surfaces that protrude from the underside of the aft fuselage in an inverted “V” configuration They improve stall protection by scooping up air under the tail helping to push the nose down at high alpha Another benefit is enhanced directional stability at sideslip and higher angles of attack Ancillary benefit Can avoid the need of a stability augmentation system through inherent improved directional stability at high Mach numbers and altitudes. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 19 .) Stall Strips Are spanwise strips added to the wing leading edge to ensure stall begins at that location first They provide more docile (acceptable) stall characteristics It is an effective method to ensure proper stall progression. increased Dutch-roll damping Generates drag through greater wetted area and interference Jan 05 Copyright  2005 by Askin T. however.

23VS) in the landing configuration Alpha = 0.Setting Requirements for Low-speed & High-speed Aerodynamics Whenever an initial technical assessment is undertaken a preliminary list of wing aerodynamic design requirements needs to be generated Primary considerations include Aircraft performance and handling Aircraft certification The list constitutes a roadmap and is formulated by collaborative efforts between conceptual design.13VS and 1. Isikveren All Rights Reserved . aerodynamics and operational performance functions The most important component is the wing design It is an iterative process and requires input from all three groups mentioned above Issues concerning design philosophy generate fundamental questions about how the goals are to be achieved Requisite number of development wings Requisite number of production wings (if a family concept) Scope of trade-off analysis and declaration of optimisation parameters Low-speed requirements and targets that need to be defined are All speed targets are with respect to 1-g stall concept Max expected L/D for each flap and/or slat angle Expected L/Ds at 1.0 in ground effect for each takeoff flap configuration Expected CLmax for each flap and/or slat angle assuming both clean and with icing contamination Number of unprotected (anti-ice or de-ice) slat panels should be taken into account Jan 05 Section 6 – Initial Sizing & Analysis Techniques 20 Copyright  2005 by Askin T.13Vs) for each permissible takeoff flap configuration Mid-AUW.23VS and VFE Expected CD at V2 (1.3VS). in the clean configuration VREF (1.13VS and 1. 250 KCAS) in the clean configuration (idle power) VAPP (1. typical descent speed (e.23VS for respective takeoff and landing configurations Stable L/D versus CL at 1.g.

underwing podded engine ground clearances up to 10° in roll.Setting Requirements for Low-speed & High-speed Aerodynamics (cont. flaps. maximum climb speed. sudden pitch-up. severe buffetting. geometry limit in pitch or combination of both High-speed requirements and targets that need to be defined are Expected maximum M*L/D at design cruise speed Expected L/D at High AUW. intermediary cruise altitude Jan 05 Section 6 – Initial Sizing & Analysis Techniques 22 Copyright  2005 by Askin T.) Performance Requirements @ Shaker CL No unnacceptable handling characteristics up to αMAX (roll-off. Isikveren All Rights Reserved . initial cruise altitude Mid-AUW. etc. note stick-pusher needs to be accounted for aft-fuselage mounted engine configuration Pitch attitudes of less than 4° at VAPP in the clean configuration No abrupt changes in pitch stability with increasing alpha up to maximum alpha Dihedral stability for all low speed configurations Wing tip.) CL MAX (n o ice) No Ice CL Shaker 3 % or 5% Mar gin (no ice) With Ice CLREF 20 Manoeuvre Margin Refer enc e Speed αMAX Range 10 α Definition of target CL-α characteristics. typical climb speed.

initial cruise altitude Mid-AUW. maximum cruise. initial cruise altitude MDD number at mid-AUW and initial cruise altitude Buffet boundaries margin of 1. initial cruise altitude Mid-AUW.0° for most cases within the typical operations envelop Good design practise to ensure +0° for all operations Wing loading to ensure passenger comfort and operational efficiency Stable dihedral and weathercock characteristics up to MMO/VMO Gradual degradation in stability derivatives up to MFC/VFC No aileron aerodynamic reversal up to MFC/VFC Jan 05 Copyright  2005 by Askin T.4 g at High AUW.) Expected CD at High AUW. maximum climb speed.5-2. MMO. typical cruise speed. MFC/VFC kink (thrust lapse rate included) CD always increases with Mach and CL particularly for intermediate to high speeds Shock waves strength and movement should not be abrupt with increasing Mach up to MMO or alpha (CL) up to 1. intermediary cruise altitude High AUW.Setting Requirements for Low-speed & High-speed Aerodynamics (cont. typical climb speed. intermediary cruise altitude Special relationships and guidelines gathered through experience are Speed stability (slope of L/D versus CL) assured at low AUW. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 23 . initial cruise altitude High AUW. intermediate speed.5g Typical aircraft pitch angles during cruise Should not exceed +1.

. E. Thesis. “Aerodynamic Design of the DC-9 Wing and High-Lift System”. SAE Paper 2001-012989. ICAS-886. P. Isikveren All Rights Reserved Section 6 – Initial Sizing & Analysis Techniques 24 . September. AIAA 3rd Applied Aerodynamics Conference... R. Department of Aeronautics. 1993 Obert. AIAA-854067. AGARD DCP 505. United Kingdom..2. pp 2575-2583 Shevell. McDonnell Douglas Corp. ”The Aerodynamic Characteristics of Flaps”. E.D. Douglas Aircraft Div. Ebeling.D. 1967. pp.1. “Quasi-analytical Modeling and Optimization Techniques for Transport Aircraft Design”.T. Stowell.) Additional Reading Young. 1988 Schaufele. 56-65 Dees. M. Jet Transport Performance Methods. 670846. R... Boeing Flight Operations Engineering. AIAA Paper No..Tier II Low-speed & High-speed Aerodynamic Prediction (cont. February 1998. May 1989 Obert. Aeronautical Research Council Reports and Memoranda.S.. 2001 World Aviation Congress. October 1985 “Getting a Lift Out of Winglets”. Business and Commercial Aviation. A.W. “Forty Years of High-Lift R&D – An Aircraft Manufacturer’s Experience”. Section 7. Ph.. Ministry of Supply. “737-800 Winglet Integration”. Royal Institute of Technology (KTH).. A. “Predicting Low-Speed and HighSpeed Aerodynamic Attributes”. “Aerodynamic Bugs: Can CFD Spray Them Away?”. A. 1953 “Aerodynamics”. Report 2002-13. Seventh Edition. September 2001 Isikveren.D.. D6-1420. Sweden. “The Aerodynamic Development of the Fokker 100”. 2002 Jan 05 Copyright  2005 by Askin T.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

drag is reduced. which increases with cruise range. Figure 3 – Blended winglet construction Figure 5 – Retrofit wing modifications Figure 3 shows the 737-800 Blended Winglet construction. Many of the aerodynamic driven changes to the 737-800 are the same for the retrofit and production versions. It is based on an average of eastbound and westbound missions and is common to both retrofit and production winglets.The addition of 8 foot tall Blended Winglets to the 737800 (see Figure 1) increases the aerodynamic efficiency. For a given amount of lift. Adding winglets increased both the wing dynamic and static flight loads significantly. Figure 4 – Retrofit winglet aircraft modifications Most of the structural changes required differ between the 737-800 retrofit and production aircraft. Figure 4 shows the primary retrofit changes and Figure 5 illustrates the structural modifications required for the 737-800 winglet retrofit. Changes common between the 737-800 retrofit and production aircraft with winglets are: • • • • Winglet Stabilizer Trim settings Auto-throttle Flight Management Computer (FMC) data Figure 2 shows the flight-test derived winglet block fuel burn improvement. The winglet is approximately 70% graphite-epoxy by weight. In the aircraft retrofit environment many of the challenges to install winglets on the airplane are different compared to the production modifications. Direct economic benefits to the airlines include combinations of these items (not all are available simultaneously): • • • • • Decreased fuel burn Increased payload-range Improved take-off performance Reduced engine maintenance Lower airport noise levels RETROFIT WINGLETS APB has primary responsibility for the retrofit (post delivery and in service) winglet installations. An economically viable . Figure 2 – Winglet block fuel burn improvement Other less tangible benefits include high-tech visual appearance and airline passenger appeal (environmentally friendly).

ILFC. Prototype winglet performance and loads were flown in 1998 and 1999 on the YC001 (737-800) and YG001 (737-700 BBJ) airplanes. SMYD. Autothrottle. FMC. and American Trans Air. Initial production winglet customers included South African Airways through GATX.Figure 10 – Flight test summary Figure 8 – Production winglet lighting Another small systems change is required due to the winglet aerodynamics altering the stabilizer trim angles. This manifests itself as updated stabilizer trim switch locations and a winglet “greenband” light plate in the cockpit. Current committed 737 retrofit programs beyond the 737-800 are the 737-700 and the 737-300. It is similar but not identical to the –800 retrofit winglet installation. APB worked with assistance from BCA to achieve certification for the retrofit installation with the FAA and JAA and obtained the Supplemental Type Certificate (STC) in May. An example of cooperation between Boeing and APB is the use of YC020 flutter flight test data to correlate with Boeing computational methods in support of the production winglet flutter certification. which was certified using YC020 flight test data. AIRLINE OPERATIONS The first flight with certified 737-800 winglets was by Hapag Lloyd on May 8th. 2001. Certification of the Boeing production installation. Initial retrofit winglet customers included Hapag-Lloyd as launch customer and Air Berlin. Boeing and APB held joint flight test programs wherever possible to minimize cost and share data. Air Berlin. 2001. Figure 12 details the status of all the 737 winglet programs. occurred also in May and was done by Program Letter of Definition (PLOD). The BBJ winglet installation was certified on YG032 in 2000. Figure 9 – Systems changes FLIGHT TEST AND CERTIFICATION Five different 737 aircraft were flight tested from 1998 to 2001 to validate and to certify the winglet installations. A summary of these flight test programs is shown in Figure 10. Figure 12 – 737 Winglet program status . POTENTIAL FUTURE PROGRAMS APB believes a tremendous interest in winglets exists in the passenger and freighter market place. This allowed a reduction in YC714 flight test hours by avoiding additional flutter flight testing. and stabilizer greenband changes are shown in Figure 9. with assistance from APB.

6. Properly integrated winglets provide substantial value to their operators. 3. APB blended winglets were successfully integrated and certified onto the Boeing 737-800. Proper treatment of additional winglet loads and their impact on flutter were required for a successful program. US Patent 5. Louis B.348. 1994.. CONTACT Retrofit winglet sales information is available from Tom VanDerHoeven at 1-800-winglets. granted September 20. REFERENCE Gratzer. A joint development and flight test program was an important ingredient to support the certification efforts. Production winglet sales information is available from James Wilkinson at (206) 766-1380. “Blended Winglet”. The expected winglet performance benefit was maintained with minimal weight penalty despite increased wing loads.253.CONCLUSIONS 1. 4. . A common design approach for both retrofit and production winglet installations provides maximum fleet commonality for the winglet customers. 2. both as retrofit and production installations. 5.

1. It basically works in the same way as the “horse-shoe” procedure with the exception that the legs of the shoe are flexible and consist of seven (instead of three) vortices of equal strength. not only in terms of departing from the usual more simplified approach premise. two such candidates are suggested as α = 0° and +4°. and an exemplar of software embodying these principles is one authored by Melin95. cambered. The implication to vehicular definition relates to an initial appreciation of how the flight envelope will look as well as being one of the integral components in formulating the aeroplane’s operational performance attributes. twisted and cranked with dihedral. The main aim is to develop methodologies where the designer has an ability to approach the design solution in a more sophisticated manner. it was recognised the algorithm to compute maximum lift attributes adhere to a quasi-analytical philosophy. Secondly. this is found by extrapolating the lift-curve slope (dCL/dα) back to the point at which CL = 0. Unlike what is offered by classical VLM approaches.LOW-SPEED & HIGH-SPEED AERODYNAMICS 75 7 Predicting Low-Speed and High-Speed Aerodynamic Attributes The importance of predicting low-speed and high-speed aerodynamic qualities of aircraft cannot be understated. two seed computations are conducted for the lifting surface system at angles of attack (AoA or α) where collinearity is likely as depicted in Figure 23 and labelled as Step 1.1 Clean Wing Lift Attributes and Maximum Lift The clean wing maximum lift can be computed for any original multi-surface or nonplanar planform geometric definition using a three-dimensional Vortex-Lattice Method93 (VLM).). 7. but an account of the impact a technological decision makes to the end result. non-linear effects such as the interaction of multiple surfaces can be simulated more consistently. which calculates aerodynamic properties of multi-wing designs that are swept (symmetric or otherwise skewed). one particular approach models the wake coming off the trailing edge of every lifting surface as flexible and changing shape according to the flight state considered.1 Low-Speed Aerodynamics: Lift To consistently support design studies of not only quite complex conventional planforms (with multiple cranks. This is surmised as being achievable by first of all soliciting the designer’s philosophical requirements and translating this notion into single all-encompassing algorithms that provide visibility to the designer. With a distorting wake. These two primary goals must also be tempered by an appreciation for reduction in the analysis complexity. but also of more exotic layouts such as multi-surface and non planar wings. etc. Wing lift carry-over into the fuselage body can be accounted for by factoring the original (wing only) dCL/dα with a calibrated variation of . Succinctly. Following the protocol mapped out in Figure 23. Since the primary assumption of any VLM is linearity. dihedral. 7. This task can be achieved by concurrent utilisation of dedicated software to quantify the fundamental parameter of clean wing lift-curve slope with well-established empirical methodologies. The slope dCL/dα itself is quantified by comparing the computed VLM lift at the two seed AoA VLM calculations. the classical “horse-shoe” arrangement of other VLM programs has been replaced with a “vortex-sling” arrangement. the methodologies must be impervious to stoppage when key information required on the part of the designer is found to be lacking. tapered. The source of the basic theory for the VLM with flexible wake is cited as Moran94. the next step is to identify the zerolift AoA (αoL).

The final step involves adding 4° times the vehicular dCL/dα to the now corrected CL computed for Point 3 in Figure 23 to predict the clean wing CLmax adhering to a 1-g stall concept. the net or exposed wing planform area (Snet) and the gross wing planform area (Sgross). Predicting the lift characteristics of a clean finite wing using quasi-analytical techniques (1-g stall concept shown). The parameter ς is a calibration constant and was derived to equal 3.2. (135) is only applicable for wing-body configurations not violating the constraint of dh / b < 0.76 CONCEPTUAL AIRCRAFT DESIGN METHODS Lift Coefficient.) Figure 23. α (deg.064 Φ regs ) dC L dα (136) vehicle . As a final point.97-101. or. Step 3 involves an AoA increment of ∆α = 10° to yield an estimate of the cessation of the linear portion of the curve (usually around α = 8°) or the beginning of non-linear lift leading eventually to stall. CL 4 4° x dCL dα 3 αstall 43 − 2ARref 3 1 dCL dα Vortex-Lattice Calculations Empirical Algorithm αoL 2 ∆α = 10° Angle of Attack. simply given as C L max = 14o (1 + 0. a method given by Pitts et al96 dC L dα where d  ξ = 1 + ς h b  d2 π  Snet h +  Sgross 2 dC L dα wing Sgross  (135) =ξ vehicle dC L dα (134) wing is related to the fuselage external maximum width (dh).2. From known data3. Pitts et al stipulates that the use of Eqn.

αstall is found by incrementing the AoA at Point 3 shown in Figure 23 by (43 2ARref) / 3. The subscript 22 denotes the influence of an auxiliary flap or vane if applicable. CLmaxW is the maximum clean wing lift attainable. Since the AoA for stall will differ between the 1-g stall break and minimum speed in a stall manoeuvre. λ2(β) is a function of the flap angle and is determined from experimental data (varies from one flap to another). If the value is of interest. introduces a multiplier derived from information presented by Obert3. A similar and more detailed working account may be found in a design review done by Pazmany103 and Isikveren et al104.2 Maximum Lift Generated by Trailing and Leading Edge High-Lift Devices High-lift produced by flap and slat deflection is estimated based on methods presented by Young102. f (Λ) is a correction to the lift increment for a swept wing. The operation [λ3(bfx2/b) - . Making allowances for effective chord. An appropriate parameter value is invoked in accordance with the analysis being conducted. the impulse function.1.0. 7. under the premise of a power-off 1-g stall concept (s = 0).e. it is suggested that Eqn. F(AR) is the function relating the vehicular dCL/dα and the aspect ratio. however. otherwise is zero for s < 1.1). or. by combining all the steps detailed above can be simplified to read α stall = α oL + 73 − 2 AR 3 (137) Eqn.LOW-SPEED & HIGH-SPEED AERODYNAMICS 77 When s = 1. Φregs = Φ(s. This reference uses empirical correlation from assorted accumulated data and predicts with adequate accuracy the aerodynamic characteristics of high lift devices. (137) be incremented by an additional ∆α ≈ 1. A suggested empirically derived method based on the same data3. and this is standardised to an AR = 6. (137) is taken to be applicable for the 1-g stall concept only. i. Working off the equivalent reference wing aspect ratio as the only independent variable for analysis. the increment due to the presence of any trailing edge flap is given by   F(AR ) ∆C Lflaps = ∆C′L (c′ c) + C L max W (c′ c − 1) f (Λ ) F(6)   (138) where (c´/c) is the effective chord ratio. the salient features will be appropriately noted. or alternatively put.0° to model the minimum speed (FARs) in stall manoeuvre AoA. the corresponding AoA for stall (αstall) can be estimated as well. The methods are not explained in great detail here. the minimum speed in a stall manoeuvre in accordance with FARs (s = 1) respectively. and ∆C ′ = λ 1 (c f 1 c) λ 2 (β1 ) + λ 1 (c f 2 c) λ 22 (β 22 ) L [ [λ ( b 3 ] f 22 b) − λ 3 ( b f 21 b) − λ 3 ( b f 12 b) + λ 3 ( b f 11 b) (139) ] λ1(cf/c) is a function of effective chords. flap incidence and part span.97-101 quoted earlier.

x = 2 and 1 define the outboard and inboard (due to a central cut-out) ends respectively. the Fowler assumes 10o. a more consistent approach exhibiting functional similarity with Eqn.183 x 10-3 and is universally applicable for all (chord extending) flaps considered. an all-purpose fixed quantity for effective chord. but a parameter to account for the stall concept adopted per chosen airworthiness regulations. introduction of a continuous functional form for the f(Λ) correction parameter.50 which is universally applicable to all devices has been artificially set in keeping with conclusions drawn from surveys presented by Obert3. linear sensitivity to AR. A series of fixed flap settings corresponding with deflection optima based on experimental results given in literature1.78 CONCEPTUAL AIRCRAFT DESIGN METHODS λ3(bfx1/b)] is a part span correction factor. 20o and 45o for intermediate takeoff. i. maximum optimal flap deflection usually pertaining to landing configuration.e. it’s coupled constituent Eqn. a tentative maximum deflection of 20o is assumed based on experimentation and actual examples64.1). The constant kgeo is equal to 2. and introduce not only the fixed functional values related to design intent supplied by Young. initial guesses for optimal flap deflections have been assumed to be approximately 10o. Φdslot = Φ(s.39 for given high-lift device types have been preselected for field calculations. Congruous with the double slotted premise. i. maximum takeoff and landing configurations respectively. Regardless of this directive. Occasions where a slat lift increment is desired.97. these selected values were found to be very close to actual deflections used on contemporary aircraft and hence adopted for simplicity.105. providing an extension is made to allow cubic interpolation of CLmax for the given intermediary flap setting. For double slotted flaps of Douglas type. These trailing edge high-lift devices may also be complemented by the introduction of leading edge slats. Furthermore. represent the relative increase in lift compared to the default single-slotted flap prediction assuming double slotted of Douglas type and Fowler flapping arrangements respectively. expressed as fraction of total reference wingspan. In the end. (139). the final algorithm describing change in lift due to trailing edge device deflection is proposed here as ∆C Lflaps  20 + Φ dslot + 5Φ fowl  3 =  k geo β flap AR (3 b flap − 1) cos Λ Qchd 20   (140) The two design related impulse functions. The flap deflection angle in degrees is denoted by βflap with bflap defining the part-span flap including fuselage carry-through. 20o for intermediate and maximum takeoff. Young102 suggests a rather simplified expression relating lift increment due to slat to the slat wing chord fraction. the algorithm used to determine CLmax given above permits an opportunity to truly optimise flap setting for the operational performance scenario considered. Additionally. Although optimal flap deflection is dependent upon a given vehicular configuration and ambient conditions in which the aircraft operates. 15o and 35o for intermediate takeoff. by incorporating supplementary simplifications for sake of brevity. Single slotted flaps tentatively have pre-designated deflection optima of 7o.1) and Φfowl = Φ(s. The increment in lift due to slat is only introduced for maximum lift prediction. and 45o for landing. The first task is to take Eqn. (140) was chosen to be a more accurate model ∆C Lflaps LE TE = k geo AR b flap cos 3 Λ Qchd (141) .3-5. an upper permissible boundary of CLmax = 3.e. (138). maximum takeoff and landing configurations respectively. and.

2 1.3 TE (or LE) Flaps Neutral Max TE (or LE) Flaps ε = +10% 0.0470.8 3 Vehicle Actual CLmax (-) Figure 24.3 Establishing the Accuracy of Clean Wing and High-Lift Prediction Once each of the analytical and empirical constituents is combined to form the final algorithm. otherwise equal to approximately -0. and bflap is the slat part-span fraction.2 Error. namely the MS(1)-0313.8 2 2.2 ε = -10% -0. As outlined by McCormick34 a complete treatment involves augmenting untrimmed vehicular lift coefficient according to the relative distance between vehicular centre of gravity (xcg) and aerodynamic centre (xac) locations. To complete the entire prediction exercise. Prediction accuracy of algorithm to compute CLmax using quasi-analytical techniques. and then incrementing contributions due to generated moment coefficient about the aerodynamic centre and the moments created because of increase in drag due to trim.4 1. a trimmed lift coefficient needs to be produced.2 2. 7. Figure 24 elucidates this by demonstrating a typical bandwidth of 0. to simplify matters.  MAC (x cg − x ac ) C Ltrim = C L max 1 +  lt   (142) Many aircraft manufacturers adopt the simplified functional form given by Eqn.15 for all other configurations. Such an approach requires a detailed array of information. now taken to be 0.1 ε = +5% 0 -0. High-lift device set to neutral and maximum deflection shown. sufficient accuracy can be achieved by dropping the terms dependent upon moment coefficient and increase in drag.3 1 1. .1. in Predicted CLmax (-) 0. (142) in their respective aerodynamic data handbooks. ε.6 1.4 2. Using a generic supercritical profile as a basis for this investigation.1 ε = -5% -0.LOW-SPEED & HIGH-SPEED AERODYNAMICS 79 where all other parameters retain the previously given definitions. a wide-ranging analysis has shown predictions are relatively consistent with actual aircraft lift data. except for kgeo.6 2. Default values for the non-dimensional relative MAC distance (xcg – xac) can be assumed as -0.05 for aft-fuselage mounted vehicles.

The aircraft used for this validation exercise were: Boeing BBJ176. More saliently. Saab Aerospace Saab 340113 and Saab 2000114. PD340-2 19 PAX regional jet conceptual design study112. By creating a hybrid approach where the component build-up method is benchmarked against a standardised closed form expression. Note that all aircraft assuming maximum flap deflection data points are displayed in Figure 24. Embraer ERJ 135108. the friction coefficient (cf turb) according to Eckert based on wetted area is given by c f turb = A (log N R ) b (1 + cM ) 2 d (143) where M is the instantaneous Mach number.1 Derivation of The Equivalent Characteristic Length Method Assuming the boundary layer is fully turbulent and accommodating effects due to compressibility on skin friction. Fokker Aircraft Fokker 70110 and Fokker 100111. b = 2.2.58. which accounts for fully turbulent flow and compressibility effects. economy of effort can be achieved without incurring excessive degradation in predictive powers.a quantity commonly used for aircraft comparison exercises.The Equivalent Length Method A common method for determining the zero-lift drag (CDo) of aircraft components is an assumption that the constituent’s friction drag is equivalent to a flat plate having the same wetted area and characteristic length.2 Zero-Lift Drag Estimation . a very preliminary assessment of the complete vehicular zero-lift drag estimation may be accomplished by summation of these individual components. the component build-up method may be employed and a characteristic equivalent length for the entire vehicle can be derived from its equivalent skin friction coefficient . CRJ20079.455. ERJ 140109.80 CONCEPTUAL AIRCRAFT DESIGN METHODS error (ε = predicted – actual) with respect to manufacturer quoted values falls within a ±5% splay.144 and d = 0. 7. Dassault Aviation Falcon 2000107 and Falcon 90053. data pertaining to neutral flap deflection is shown where the original manufacturer information was available. constants A = 0. the study indicates there exists a good likelihood maximum lift predictions will not exceed an error of around ε = ±0. The benchmarking data comprised either known aerodynamic performance or was derived from vehicular stalling speeds. ERJ 14584. This equivalent characteristic length may in turn be reintroduced into Eckert’s equation and solved for any other Mach number and flight level combinations the aeroplane encounters. and. Bombardier Aerospace Learjet 4578. 7. Cessna Citation Excel82. A tool for estimating zero-lift drag is the friction coefficient equation based on experimentation done by Eckert115. Global Express64. Challenger CL-60451. the Reynolds number (NR) in atmospheric flight at given speed and flight level can be expressed as NR = ρ sls σ V lb µ sls µ µ sls (144) The identity ρsls/µsls is approximately equal to approximately 6. and lb is any specified representative length of the body.15 irrespective of flap deflection. Gulfstream Aerospace GIV-SP89 and GVSP90.9x104 s/m2. In this way.58 are coefficients of proportionality derived by Eckert. . and. c = 0. Learjet 60106. By assuming an appropriate reference condition of Mach number and flight level. CRJ70080 and CRJ90081.

It would be prudent to give scope in accommodating mixed laminar and turbulent flow. It can be shown34 the total flat plate friction coefficient for a mixed laminar and turbulent flow is calculated from c f = c f turb − xT (cf turb − c f lam ) lb (149) In this equation. surface waviness due to airframe construction. this can be used as a basis to formulate an extension such that a realistic skin friction assuming mixed flow is produced. Matching the momentum thickness of the laminar and fully turbulent boundary layer at transition point T gives c f lam x T = c f turb ∆x (148) where cf lam is the skin friction coefficient for laminar flow. The premise of mixed laminar and turbulent flow used to derive an augmented realistic skin friction coefficient2. . These include gaps and steps. (143). xT is the point along the body characteristic length where flow transition occurs and ∆x is a distance ahead of the transition point where fictitiously the onset of fully turbulent flow takes place. cf turb is computed assuming a Reynolds number based on a body characteristic length starting from the fictitious onset of turbulent flow to the end of the ‡‡ The aircraft surface can have many irregularities. Working off a basic assumption that momentum thickness at given transition point is synonymous for both laminar and turbulent flows (see Figure 25).197. dynamic distortion and cabin pressurisation. and. (143) represents a condition where fully turbulent flow exists. Eq. and this factor is in turn synonymous with a Reynolds correction coefficient of ηact = 0. the final skin friction can be produced by summing the friction coefficients for partly laminar and turbulent flow2. Consideration must also be given to the fact a practical lower limit of τact = 1. Since an algorithm to quantify a realistic turbulent skin friction coefficient has been established with Eq. and therefore has been presented as the basis for establishing predictions at the very initial design stage. hence permit the designer to set a minimum goal of what proportion laminar flow shall occur over the characteristic length of the body constituent in question.30 (or potential CDo reduction of up to 10% from the mean curve) has been derived when analysing some narrow bodies and larger aircraft types from data supplied by Obert3.82 CONCEPTUAL AIRCRAFT DESIGN METHODS levels‡‡. lb Figure 25. protruding flush rivet heads.

§§ . experimentation conducted in a more operationally pragmatic sense commonly produces transition at 15% wing chord117. χmf. cf lam is calculated based on the entire length of assumed laminar flow.40. A valid form of simplification is in order here. Experimentation has found that a speed near the final vehicle MRC or LRC at an altitude 4000 ft lower than the intended certified ceiling are good reference conditions for a balanced error distribution. In addition.74 for all xT / lb < 0. (150). The upper boundary of assumed laminar flow fraction is a reasonable one for design prediction purposes since an example of the most successful flight testing of combined passive and active laminar flow control technology achieved laminar flow up to 30% of wing chord117. Investigations found that for xT / lb values less than approximately 0.h = ∑c S i =1 i f I i wet SW (152) where the product c if Siwet is the drag area of each component i. an iterative procedure is required to solve for ∆x in Eq. By choosing an appropriate reference condition of Mach number and altitude§§.LOW-SPEED & HIGH-SPEED AERODYNAMICS 83 body. (149) can produce an alternate form  x + ∆x   c f turb c f = 1 − T   lb   (150) Since cf turb also depends on ∆x. assists in ascertaining what proportion of the completely turbulent flow premise imparts an influence on the mixed flow result. Introducing a presumption the fictitious distance ∆x consistently exhibits linear proportionality with xT for low to midrange values of ∆x / lb. hence permit a reduction in complexity. Experimentation has found a useful value for this parameter is approximately χmf = 0. scope can be given to dispense with the transcendental nature of Eq. the total skin friction coefficient for mixed laminar and turbulent flow can alternatively be expressed as  x  c f = 1 − χ mf T  c f turb  lb    (151) The constant of proportionality.40. an equivalent skin friction coefficient representative of the entire vehicle can be produced with the congruent relation c ε ∑ Siwet ≡ ∑ c if Siwet f i =1 i =1 I I (153) The parameter cε is the equivalent skin friction for the sum of all constituent wetted f areas produced using the equivalent flat plate analogy representing the entire aeroplane. (148) into Eq. It The reference condition for Mach and flight level is open to the designer’s willingness to trade larger errors in low speed for more accurate high-speed zero-lift drag or visa versa. The component build-up method for zero-lift drag at given Mach number and flight level is given as C Do M . and. Substitution of Eq. or distance xT. (150).

84 CONCEPTUAL AIRCRAFT DESIGN METHODS is now proposed that this notion of equivalence can be extended to quantify a characteristic length as well. If Eckert’s general equation is partitioned into Reynolds number and compressibility dependent constituents. By utilising the relation x = eln x. Since the entire vehicle has been replaced by the flat plate premise with a corresponding value for cε . more importantly what is the upper threshold of relative errors the designer may expect. It was identified that this problem may be avoided via the use of logarithmic differentiation. drag is an integral parameter and has the primary requirement of being differentiable with respect to the airspeed V for all cases. (155) can be alternatively expressed as    ρ sls σ   V 2    1 + c 2   c = A[ln10] exp− bln ln V l ε  − dln  a sls θ       µ sls µ µ sls   ε f b (156) which is in a form ready for differentiation albeit the complexity has not been reduced. in conjunction with some algebraic manipulation. by rearranging Eckert’s equation. (155) appears to be in a form that is quite complex. The most expedient way to observe this would be the comparison of resultant equivalent skin friction errors analytically and do so for a range of contemporary regional transport and business jet Reynolds number regimes based on complete vehicular characteristic lengths. Eq. (143) then becomes . not configured for a more in-depth calculus treatment.2 Gauging the Robustness of the Equivalent Characteristic Length Method An interesting question is to what extent the equivalent characteristic length assumption is compatible to the exact component build-up method. Eq.h [ ] d Swet SW (155) For a detailed analytical treatment of en route performance. (143) can be solved for f an equivalent characteristic length (lε) given by the identity A   cε  f     1/ b lε = 10 ρsls σ V µ sls µ µ sls [1+cM ] 2 −d / b (154) Reintroducing this relation to Eckert’s equation. and assuming the error in NR due to a now fixed equivalent characteristic length (i. designated hereon as the Equivalent Characteristic Length Method (ECLM).2. the ECLM expression was reconfigured as an error function with respect to the exact component build-up method. which accounts for all variations of Mach number and flight level can be given approximately as C Do ≅ A   ρsls σ  V l ε   1 + cM 2 log    µ sls µ µ sls b M . Eq. In an effort to theoretically gauge the magnitude of inherent errors produced by this approach. and more poignantly. a general zero-lift drag equation. 7. independent of Mach number or flight level effects) is small. and. Eq.e.

0% (159) Relative Error of Vehicular Zero-Lift Drag (-) -40% -30% 5.0% 0.5 7. or εl = lε/lexact.0% -20% -10% Error in lε 0.0% 0% +10% +20% Error in lε -5. by introducing the notion of error factor defined as the ratio of the fixed vehicular characteristic length quantity derived from a reference Mach and flight level to the exact value of vehicular characteristic length. Resilience of ECLM accuracy for a given error in vehicular characteristic length and en route Reynolds number based on vehicular characteristic length. Now.0% +30% +40% +50% +60% +70% -10.5 6.5 2. the relative error of an equivalent characteristic length assumption can be gauged by considering deviations from the exact value of c fexact through a fractional comparison .LOW-SPEED & HIGH-SPEED AERODYNAMICS 85 cf = ϖ1  log l b  ϖ 2 1 +  ϖ2   b (157) where the compressibility term is described by ϖ1 = [1 + cM ] A 2 d (158) and the Reynolds number dependent constituent is defined as ρ  σ ϖ 2 = log sls V  µ sls µ µ sls  10.5 Reynolds Number Based on Vehicular Characteristic Length (x106) Figure 26.5 4.5 5.5 1.5 3.

and compressibility effects neglected. For a typical en route Reynolds number of 1. To put Reynolds number based on vehicular characteristic length into context.0 x 106. Interestingly. Additional drag produced by non-elliptical lift distributions is made by using the Oswald Span Efficiency Factor (e).86 CONCEPTUAL AIRCRAFT DESIGN METHODS cε f c fexact  log ε l  = 1 +   log N R  −b (160) Figure 26 (previous page) shows the variation of resultant prediction error compared to the exact vehicular equivalent skin friction of zero-lift drag with Reynolds number based on vehicular characteristic length whilst assuming various errors in the εl ratio. This leads the author to believe a correlation between aspect ratio and power plant installation . small business jets typically operate at around NR = 106.e. Obert3 offers an empirically derived equation for the vortex-induced drag factor applicable for Mach numbers greater than about 0. an error of -24% in lε compared to lexact corresponds to a +5% overestimation of equivalent skin friction or total zero-lift drag.007  dC   L  clean πAR (162) Eq. regional aircraft and larger business jets between NR = 1.3 Vortex-Induced Drag at Subsonic Speeds Many methods exist in quantifying this phenomenon and the most simplest of them is the Oswald Span Efficiency Method which assumes the vortex-induced drag coefficient of three dimensional wings with an elliptical lift distribution equals the square of the lift coefficient divided by the product of the aspect ratio and π. i.0 x 106 and higher. assuming typical centre of gravity locales. based on actual aircraft regardless of power plant installation. (162) does not appear to account for the distinction of power plant installation philosophy. and the direct impact this has on span loading distribution. The vortex-induced drag factor35 is given as  dC D  1  2 =  dC  πARe  L (161) Numerous estimation methods for e have been developed but they mostly tend to produce optimistically high values compared values of real aircraft. (162) was compared to Eq. As an exercise. clean wing.5 x 106 based on vehicular characteristic length for regional transports. (161) and Oswald span efficiency factor solved for a variety known e values of equipment with different power plant installation philosophies not covered by the statistical survey. Eq. 7.  dC D  1.40. and larger regional and narrow-body aircraft from NR = 3. underwing podded or on-wing nacelle configurations. This result demonstrates the resilience of ECLM. inclusion of wing twist effects. for the same Reynolds number. Conversely. the continuous functional form offered by Obert seemed to match the values for these known examples with an adequate degree of accuracy. a -5% underestimation of zero-lift drag is tolerated by a +33% error in equivalent characteristic length from the exact value. which effectively reduces the aspect ratio.5 x 106 and 2.05  2 = + 0.

LOW-SPEED & HIGH-SPEED AERODYNAMICS 87 philosophy must exist. In order to acknowledge these known phenomena.1). Such a model is proposed here to be  dC   dC D   dC D   2 = 2  + ∆ D   dC 2   dC   dC   L  flaps  L   L  clean  dC D  1. invokes a correction to the vortex-induced drag factor to signify an irregularity in the lift distribution due to deployment of high-lift devices. or alternatively.15-0. This means the maintenance of sufficient accuracy can be expected using the one algorithm in predicting the vortex-induced drag regardless of flaps neutral or extended. The modified wing form factor reads as 4  t t  ϕ wing = 0. These values are computed based on thickness-chord ratios of the wing. (163) to low-speed drag polars of the Saab 340113 and Saab 2000114 aircraft found the correlation to be quite adequate. Concurrent to this circumstance. All of the form factors itemised below were derived from original expressions developed for GASP39 and subsequently modified to suit known data more appropriately. and.1 (1 − 0.4 Three Dimensional Effects and Ancillary Drag Contributors Five form factors that account for three-dimensional effects. There is a reduction in the vortex-induced drag factor with increasing flap deflection. Studies comparing the vortex-induced drag estimate generated using Eq.893ξ ht ) 2 + 4   + 240     c m  c m     (165) . non-ellipticity of the span-wise lift distribution of the basic wing. the effect of flap cut-out and lift carry over by the fuselage1. the fineness ratios of the fuselage and nacelle.25. hence Obert’s regression analysis inherently accounted for this association. Φflap = Φ(βflap. a reduction in the vortexinduced drag for given CL occurs. This is attributable to an increasing benefit generated by the slot-effect at greater deflections and amounts to a measure of boundary layer control thus preventing separation.421 2 + 4   + 240     c m  c m     (164) with the horizontal tail surface re-defined to be ϕ htail 4  t t  = 1 + 0.007  dC  πAR  L (163) The impulse function. For field performance where Mach numbers typically range between 0. and excrescences are reviewed here.05 + 0.271 Φ flap  2 = − 0. 7. it is apparent that a change in vortex-induced drag factor will take place due to a change in the span-wise lift distribution due to flaps extending and deflecting118. (163). horizontal and vertical tails. as flap deflection is increased. the implication is an incremental change in the vortex induced-drag factor needs to be introduced to Eq. there is also an additional physical effect that needs to be addressed. ancillary interference.000487 β flap + 0. Literature demonstrates this variation is proportional to wing geometry.

26 and cf / c = 0.5 Total Incremental Drag due to One Engine Inoperative Condition The One Engine Inoperative (OEI) condition appears to be mostly disregarded in conceptual design literature. Assuming a streamlined fuselage without a blunt nose ϕfuse d  l  = 1 + 0. Assuming a given trailing edge (and/or accompanying leading edge) high-lift device has been deployed. the vertical tail form factor was amended to read as ϕ vtail 4  t t  = 0. Based on information gleaned from McCormick34.0025  fuse  + 60  v  l  d   fuse   v  3 (167) and finally. the drag due to extension of undercarriage can be quantified with adequate accuracy using statistical correlation from known data. Similarly with the wing.0416  f    bb  SW  c   flap  (170) ( ) and typical values for the relative flap chord fraction of cf / c = 0. an approximation for the incremental drag is suggested as ∆C Do flap = 1  c   MAC  2  cos 2 β flap 5. It is usually classified as a preliminary design problem1.268 x 10 −4 β flap − 0.15 for trailing edge and leading edge devices respectively are suggested as initial estimates.5 2 + 4   + 240     c m  c m     (166) The fuselage form factor is predicated by body slenderness ratio.35  nac  l   nac   (168) The prediction of a low-speed drag polar for field performance requires account of contributions due to extended undercarriage and high-lift devices.88 CONCEPTUAL AIRCRAFT DESIGN METHODS The ξht parameter represents horizontal tail placement non-dimensionalised by dv with respect to the vertical tail tip and FRP water-line. the nacelle form factor is based on the premise of slenderness as well   d  ϕ nac = 1.17 1 + 0. In the absence of detailed undercarriage sizing.85 x 10 −5 WTO + 0.294 SW ( ) (169) The total aircraft drag of a configuration geared for field operation is also affected by a profile drag contribution from extended flaps and slats.005339 β flap + 0.4 because yawing and rolling considerations become rather complex in nature since these . a useful linear regression equation was derived to be ∆C D LG = 1 2. 7.

7. equilibrium is achieved via. Drag due to engine windmilling. incremental changes in normal force vortex-induced and profile drag from control surface deflection. a number of valid simplifications may be incorporated in order to reduce the scope of detailed information required whilst retaining strong predictive powers and objective function sensitivity with respect to the design variables. Dwm yeng yeng lvt LR δR Top Figure 27. airframe sideslip. L R l vt = y eng (D wm + Top ) (171) where yeng is the moment arm from fuselage centre line to the critical and windmilling engines.LOW-SPEED & HIGH-SPEED AERODYNAMICS 89 must be trimmed out by primarily the rudder and then aileron. asymmetric slipstream effects and lift distribution reconfiguration producing independent vortex-induced contributions all combine to complicate matters. Studies have shown that many of these constituent contributors can be neglected with the exception of vortex-induced and profile drag generated by rudder deflection. By assuming the vertical tail utilises a symmetric profile and all rudder deflections during asymmetric flight will be below stall. By examining the exact approach. Figure 27 demonstrates the pertinent forces and moments once this simplification is introduced. Top is the instantaneous available thrust produced by the critical engine at instantaneous velocity V and lvt is the vertical tail moment arm. Simplifications of forces and geometric considerations during the asymmetric thrust condition. The instantaneous lift (LR) generated by the flapped vertical tail is34 . Dwm is the drag produced by the wind-milling engine.5. studies have shown that many of these constituent contributors can be neglected with the exception of vortex-induced and profile drag generated by rudder deflection.1 The General One Engine Inoperative Drag Constituent If one considers the OEI asymmetric condition.

If a simplification is sought. it was found an average of 0.66 may be derived using Weissinger’s approximation34***.vt) can be estimated by the Helmbold equation34 based on an approximate lifting surface theory with the effects of sweep (Λvt. By summing the forces and moments in Figure 27.088 per deg. . however. the possibility of accounting for the influence of minimum control speed limitations on field length and initial climb performance can be introduced at the conceptual level. τ a flap effectiveness factor.   AR vt  cosΛ C Lα . McCormick34 demonstrates that thin airfoil theory can be utilised to produce adequate results but the functions are still represented by dependent variables. from linear thin airfoil and lifting surface theory. the total incremental drag contribution produced by an OEI asymmetric condition (∆CDOEI) is approximated by39 y eng D wm + D wm + Top tan δ R l vt (175) ∆C DOEI = q Sw [ ] *** Two point vortices represent the airfoil and this function is dependent upon cf / c ratio. and. an estimate of τ = 0.110 per deg. Since the geometric characteristics for equilibrium of asymmetric thrust has been quantified. Thus. This figure can be substantiated against Torenbeek’s1 presentation of overall effectiveness factors derived from experimental data for plain flaps. the rudder deflection required for equilibrium of the OEI asymmetric condition is given by δR = y eng (D wm + Top ) q S vt C Lα . Furthermore.vt τ η l vt (174) From this basis. By assuming some level of conservatism for smaller deflections.04 per rad) taken from Abbott and Von Doenhoff69 yields more realistic predictions.Qchd) accounted for by a first order cosine relation given by Torenbeek1. McCormick shows at an upper deflection of δR = 30°. (5. (2π per rad) is given theoretically. a value of η = 0. an overall flap effectiveness of τη = 0.vt = 0.90 CONCEPTUAL AIRCRAFT DESIGN METHODS L R = q S vt C Lα .3. a correction which accounts for the effects of viscosity. Assuming a typical cf / c value for the flapped vertical tail of around 0.vt  vt .vt τ η δ R (172) with q denoting the dynamic pressure. The functions τ and η are generally derived empirically since they behave in a nonlinear fashion with chord fraction (cf / c) and rudder deflection angle (δR).74 would be appropriate.vt = C lα . methods to predict these quantities with respect to operational performance will be addressed in the takeoff field performance discussion of this report. the next step should be an appreciation of to what extent performance shall be degraded.Qchd  2 + 4 + AR 2  vt   (173) Assuming a thin airfoil.49 applicable to the complete range of angles would result. and equating these to represent contributions of vortex-induced and profile drag due to rudder deflection. The lift-curve slope characteristics (CLα. Svt the vertical tail reference area. section lift-curve slope of Clα. and η.

the equivalent flat plate skin friction that simulates an imaginary roughness condition for a windmilling engine would be given as OEI flight regime is considered predominately as a subsonic problem. then a pre-designated cut-off Reynolds number would be independent of Mach number variation for subsonic flight††† and atmospheric conditions because as indicated by Raymer. there is a strong correlation to relative roughness alone. As an alternative. Typical values for the ratio of these speeds are offered but they are specific to engine type thus not allowing for a continuous function concept. bypass ratio and internal configuration. ††† . neglecting compressibility effects and hence adopting the Prandtl-Schlichting form. In view of this. The notion of a “cut-off Reynolds number” can be useful in helping to quantify the drag produced by a windmilling engine in this respect. which also may be postulated to be a function of engine size. which requires an estimation of mean flow velocity in the nozzle exit together with the windmilling mass flow. Torenbeek1 proposes a conceptual method to estimate the magnitude of the drag increment by considering this quantity to be a function of engine frontal area.2 Drag Generated by Windmilling Engine For multi-engine aircraft with engines not buried in the fuselage. Unfortunately. Raymer4 discusses the merits of employing a cut-off Reynolds number parameter to account for expected higher skin friction coefficients in conventional zero-lift drag estimation when the surface of a body is relatively rough.e. the cut-off Reynolds number (NR cut-off) is then determined by N R cut −off l =a  k b (176) where a and b are constants of proportionality and NR cut-off varies monotonically with l/k for subsonic speeds. the OEI performance will be influenced by additional drag due to a windmilling engine during the equilibrium condition of asymmetric flight. It is evident that the internal drag generated is related to maximum static engine thrust potential. Assuming a windmilling engine is essentially the nacelle but influenced by some degree of imaginary roughness on the body in this condition. 7. analogous to an internal drag contribution. This would mean the imaginary value for k would increase proportionately with nacelle physical dimensions. the method is rather esoteric because the procedure employs the momentum theorem. By comparing the ratio of characteristic length and a skin-roughness value (l/k). This premise may not hold true for extended range and in some instances driftdown operations. therefore. the imaginary cut-off Reynolds number can be considered independent of nacelle size or characteristic length as well.LOW-SPEED & HIGH-SPEED AERODYNAMICS 91 This relation is not only applicable for low speed field performance. i. a more simplified approach is proposed which assumes the windmill drag component can be accounted for by representing it as an equivalent flat plate problem with an associated skin friction value which is imaginary and independent of Reynolds number variation or associative compressibility effects. specifically in relation to OEI maximum attainable flight level and drift-down net level off height proficiency trade studies at ISA and more importantly off-ISA conditions.5. the imaginary relative roughness can be taken as approximately constant. (143). it can also be utilised for climb out analysis as well. When the imaginary cut-off Reynolds number is quantified empirically and substituted into Eckert’s equation for skin friction given by Eq.

the results were found to be quite encouraging.007274.7 Mach number (-) Figure 28. more so due to the fact the nacelle wetted area was not calibrated to any known data before computing the final result.5 0.f) used on the B737-800 narrow-body commercial transport. sea level conditions. The suggested values for conceptual analysis were found to be wm N R cut −off = 9.4 0. Figure 28 demonstrates the level of accuracy generated using the imaginary skin friction method. .3x104. ISA.1 0. Williams FJ44-2 Actual Data Williams FJ44-2 Prediction CFM56-7B28 Actual Data CFM56-7B28 Prediction Windmill Drag (N) 0 0.f) and the CFM56-7B26121 engine rated at 118 kN (26400 lb.92 CONCEPTUAL AIRCRAFT DESIGN METHODS c fwm = [logN A b wm R cut −off ] (177) where A is equal to 0. known windmill drag properties for the BAe 146-200119 were used and the results were tested against other installed aircraft engines.2 kN (2400 lb.3 0. Once the windmilling engine representative skin friction is quantified.2 0.455 and b = 2. or corresponding imaginary skin friction of c fwm = 0. Using this information in conjunction with the nacelle wetted area estimation methodology described wm previously.58. an incremental contribution to drag in the OEI asymmetric condition is therefore given by ∆C wm D = c fwm S wet .nac SW (178) In order to derive the value for imaginary skin friction. predictions of ∆C D were computed and subsequently compared to known windmill drag data for both the Williams International FJ44-2A120 small turbofan rated at 10.6 0. Benchmarking predicted windmilling drag using the imaginary skin friction method against actual engine windmilling data.

This accounts for a simulated by-pass ratio increase due to the presence of propeller or larger fan diameter contribution for given maximum static thrust rating or nacelle size. Steadily increasing values of free stream Mach number above MCR are characterised by regions of supersonic flow terminated by normal shock waves shifting aft and increasing in strength. for an initial analysis the drag rise is graphically estimated using a few rules of thumb rather than a more comprehensive appreciation of the dependence of MDD on parameters like instantaneous operating lift coefficient. denoted here as MCR. MDD is a function of lift coefficient since shock formation and strength directly relates to increases in airflow velocity. 7. therefore.0020 MCR MDD Constant CL Mach number Figure 29. The formation of shocks in the transonic flow condition affects the drag up to the drag divergent Mach number (MDD). quarter chord sweep.6 Compressibility or Wave Drag Compressibility is a drag increment caused by an increase in free stream Mach number above a critical point where locally accelerated speeds increase sufficiently to reach Mach numbers of unity and above. thereafter the drag rise rate increases substantially as shown in Figure 29. The most common is the Boeing definition where MDD is the speed at which an incremental increase in viscous drag influenced by drag rise is equal to 20 counts (or ∆CDD = 0. the actual nacelle wetted area for a turboprop (or even S-duct and straights ducts) power plant installation must be disregarded and the generic pitot introduced into the prediction process. Notwithstanding. The definition of what particular Mach number constitutes MDD is open to several options. this assumption . One should recall the method is based upon the generic pitot nacelle. Definitions for the transonic mixed flow regime and indication of speed thresholds for certain drag escalation attributes. Typically. Additionally. it is recommended that the drag contribution for turboprop engines using this method should be obtained by factoring the equivalent turbofan result by 3. The free stream Mach number at which this first occurs is called the critical Mach number.0020). and can be thought of as the lower limit of the transonic flow regime.LOW-SPEED & HIGH-SPEED AERODYNAMICS 93 Additionally. Inspection of the Saab 2000’s one engine inoperative drag assuming a propeller in the auto-feathered condition114 produced an estimation error of –2. CD increasing CL Constant CL ∆CD = 0. mean wing thickness ratio and type of airfoil geometry employed as exemplified by Raymer4.8%.

Therefore. (180).but a wide range of information exists which aid in predicting compressibility drag characteristics for given set of design parameters adequately. a modification of this premise with an empirical fit more akin to the actual performance produced by contemporary regional and business jet vehicles is proposed: a customary technology factor of MREF = 0.87 for conventional peaky sections. (181) implicitly relates MCR to MDD as M DD = M CR + ∆M (182) This information can be used in conjunction with Eq. 7. As a consequence.05 are given by Torenbeek. MREF = 0. By rearranging Torenbeek’s version of the modified Korn’s equation so that MDD is the subject 3/ 2   (t c )m  CL 1 1   − M REF −   = cos Λ Qchd  10  cos 2 Λ Qchd  cosΛ Qchd      M DD (180) Torenbeek1 offers an arbitrary mathematical representation of the condition where drag rise is terminated (at speed MDD) and an increased drag rise rate begins ∆C Dcomp   M − M DD  = ∆C DD 1 +     ∆M  n (181) where the symbols n = 2.90. it would be deemed prudent in attempting to derive a closed form expression that describes the mixed flow regime simultaneously neglecting highly non-linear terms but having a stronger basis to set more realistic goals.850 for supercritical aft loaded sections is suggested as a more pragmatic value with an occasional upper limit not exceeding MREF = M0.5 and ∆M = 0.94 CONCEPTUAL AIRCRAFT DESIGN METHODS does not necessarily invalidate this first order estimate’s predictive powers . Here.6. Torenbeek suggests values of MREF = 0. they have no physical significance but are derived from experimental data. Eq. thereby. Torenbeek122 offers a variation of Korn’s equation123 to quantify the limits of wing section performance for given vehicle wing thickness. and. sweep and typical operating lift coefficient envelope M REF = M DD cos Λ Qchd  CL 1   +   cos 2 Λ  10  Qchd  3/ 2 + (t c )m cos Λ Qchd (179) where MREF is a wing section technology factor. allowing the definition of MCR to be given as .1 Derivation of the Incremental Drag due to Compressibility Much of what is known about this flow regime are largely experimental hence are described by many different empirical models.935 for supercritical aft loaded.

by utilising linearised theory as Mach number tends to unity from below125.0 is more appropriate. Raymer and Torenbeek indicate a combined factor of approximately 2. Correcting for deviations from the optimum by a factor. i.05. by introducing the concept of an impulse function or approximate unit step that is critical Mach number dependent. Even though the implicit assumption involves smooth bodies in inviscid flow. or even all three in consort.e.MCR). (181) yields   M − M CR  ∆C Dcomp = Φ M CR ∆C DD 1 + Φ M CR  − 1  ∆M   n (184) which then leads to a closed form expression for the total compressibility drag contribution including the concept of initial and supplementary drag rise and the final equation conforms to the presupposed condition of differentiability with respect to airspeed V. representing the ratio between actual wave drag and that of the optimum body. as stipulated by Kuchemann126 and introducing an empirical wave drag efficiency factor4. Sears-Haack or Adams optimum either in isolation. Although this value reflects supersonic designs displaying a relatively poor volume distribution. the next step is to build the wave drag model according to the operating parameters dictated by given flight conditions. and one suggested reference is MsREF = 1. the relative merits of differing configurations can be compared as a guide to drag-rise behaviour. the wave drag due to volume for given body volume Vb reads as ∆C Dcomp M sREF = ηopt K o 128 Vb2 π SW l 4 b (185) The product ηopt Ko can be estimated from values quoted by Raymer4 and Torenbeek124. ηopt. (182) and Eq. as composite area distributions in pairs. and incorporating Eq. Ko. ΦMcr = Φ(M. analysis of actual subsonic aircraft (even those catering to MMO speeds up to M0. These optimum bodies can be represented by the von Karman ogive. (183) into Eq. 7.5 for ηopt Ko is adequate. This is accomplished by initially choosing a reference Mach number that is slightly faster than sonic speed. the wave drag of wings and slender bodies is frequently related to the theoretical minimum wave drag of pointed optimum bodies.90) found ηopt Ko = 4. (181) and solving for the exponent n produces .2 Quantifying Wave Drag due to Volume and Lift As expounded by Torenbeek124.LOW-SPEED & HIGH-SPEED AERODYNAMICS 95 M CR  1  =  cosΛ Qchd  3/ 2   (t c )m   CL 1   − M REF −    − ∆M   cos 2 Λ  10  cos Λ Qchd    Qchd    (183) Now. Taking the logarithm on both sides of Eq.6. Once the wave drag due to volume has been quantified for the reference condition.

the so-called corrected box ratio is defined as λcbox = β b / (2 lW). or alternatively. r = SW / (b lW) is a shape parameter. 7.7 Quantifying the Aerodynamic Impact of Winglets With greater emphasis being placed on improving aircraft cruise efficiency winglet devices appear to offer the most attractive combination of drag reduction and aesthetic appeal. hence. KW is a deviation from the theoretical minimum and recommended as being equal to 1. an estimate of the wave drag due to volume can be computed dynamically for an instantaneous operating lift coefficient. The conventional winglet (ARWL ≅ 1. The fundamental assumption here is that the fuselage nose and tail do not contribute to lift. β = (M2 . It is therefore not surprising many existing aircraft types have been outfitted with winglets as part of an overall enhancement package and many kits are offered to retrofit inservice aircraft. for a given critical Mach number premise. and. Figure 30. Definition of working parameters to compute drag due to lift in supersonic flight124.5) approach is now being replaced by .1)1/2. (181). The drag due to lift of surfaces at supersonic speeds (∆CDi wave) with streamwise and spanwise elliptical pressure load distributions is quantified by Jones’127 classical though not universally accepted relation describing the lower bound ∆C Di wave = K W β 2 C L r λ cbox π (187) where the working dimensions are shown in Figure 30.25124. (186) is then substituted back into Eq.96 CONCEPTUAL AIRCRAFT DESIGN METHODS  ∆C Dcomp  M sREF   log   ∆C DD     n= M sREF − M CR   log   ∆M   (186) Eq.

it is evident if appropriately designed and integrated with the main wing. Unfortunately. typified by a high aspect ratio (ARWL ≅ 3. i. Nonetheless.e. the known benefits of winglets can be itemised as follows: • • • • • Decreased fuel burn and increased payload range attributes – achieved through an aerodynamic performance improvement. a requirement now arises for a quasi-analytical method to quantify the change in vehicular drag due to winglets.LOW-SPEED & HIGH-SPEED AERODYNAMICS 97 so-called blended winglets128. Lower airport noise levels – exploiting the reduced thrust concept. as one would intuitively expect this approach is susceptible to inconsistencies. camber and twist in achieving maximum reduction of vortex-induced during cruise. by summing the forces in the direction of freestream and adhering to the sense convention indicated. The final step in the design cycle is to weigh the economic feasibility of adding winglets to the aircraft. Regardless of the design philosophy. allows for higher TOGWs. Higher cruise altitude and OEI driftdown ceiling – due to a net vehicular drag reduction enabling a greater amount of specific excess power at given altitude and speed. Improved takeoff performance – higher effective OEI lift-to-drag and therefore higher second segment climb gradient for given reference speed. the devices will translate into a some sort of a direct economic benefit for the operator. the designer should expect a combination of a few at best. For conceptual design studies. The revised span load is examined referenced to the ultimate wake in a Trefftz plane analysis in order to determine the induced drag and bending moment distribution. All of these enhancements may not necessarily come to fruition concurrently. To alleviate the need for excessive effort. Therefore. a revised (and re-optimised) performance estimate would entail consideration of the change in aerodynamic qualities and the change in aircraft empty weight.71. In general. one suggestion is to adopt these percentages and empirically adjust the design prediction accordingly with no due regard given to winglet design variable sensitivity. 7.1 Quantifying the Drag Reduction of Winglet Devices As depicted in Figure 31.70.129-131.5) and integrated by way of pronounced filleted transition geometry between the wing and winglet structures. net vehicular drag reduction. Reduced engine maintenance – the option of retaining the original takeoff performance levels prior to installation of winglets promotes a reduced thrust concept. A non-planar. Many examples of winglet performance prediction and design optimisation is available in literature34. thereby gauging the possibility of adverse low speed characteristics. only a relative drag is quoted at given operating lift coefficient and change in OWE due to a wing bending moment increase.7. cant. the total force of the local system can be quantified to be ∑F x = D WL cos α ind − L WL sin α ind (188) . ideally. three-dimensional potential flow panel method is subsequently employed to evaluate the configuration under takeoff and landing operating conditions. winglet configurations are analysed using the VLM to establish optimal planform attributes.

the instantaneous lift coefficient produced by the winglet (CL WL) is given by C L WL = C Lα WL (α eff − α oL ) (190) .e. Eq. Fundamentally. To this end. Prandtl’s lifting-line theory stipulates the downward induced angle of attack generated by finite wings is proportional to the operating CL and inversely proportional to the wing AR132. divided by qSW. the resulting vector produces an angle of attack αind. (188) should be examined in the non-dimensionalised form. but the circulatory motion also generates induced velocities in a spanwise direction. A key requirement is to now formulate a semi-empirical expression for αind. V∞ αinc αind LWL LWL sin αind αeff αind Forward DWL DWL cos αind Inboard Wing + Figure 31. When the freestream velocity is vectorially added to the spanwise induced velocity component in plan-view.98 CONCEPTUAL AIRCRAFT DESIGN METHODS where αinc is the winglet representative incidence and αind is the spanwise induced angle of attack instantaneously generated by the wing. Resolving local lift and drag forces generated by the winglet into the direction of freestream. Working off this premise and introducing a coefficient of proportionality (ηind) to represent a scaling factor between the downward and spanwise induced velocities towards the tips. i. then α ind = ηind CL π AR (189) A suggested scaling factor ηind = 7. Since the true goal is to quantify a relative vehicular drag. the trailing vortex shed at each wing-tip induces not only a downward velocity in the region of the wing itself.2 was empirically derived from flow visualization experiments undertaken by Head133.

the total incremental drag due to presence of winglets is determined by summing the resolved local winglet lift and drag force components. (88).LOW-SPEED & HIGH-SPEED AERODYNAMICS 99 where the effective angle of attack αeff is found by taking the difference between αind and the winglet representative incidence angle αinc. In an attempt to quantify the relative reduction in vortex-induced drag due to presence of winglets (∆CDi) in the flow field. (188) now expressed in an equivalent non-dimensionalised form. The total winglet drag is determined by summing the winglet zero-lift (CDo WL) and vortex-induced (CDi WL) drag. the change in drag due to compressibility if the winglet pre-empts the wing in generating super-velocities. Upon substitution of Eq. is not deemed to be a constituent in deriving the vehicular characteristic length. namely OdCD/dCL2. The lift-curve slope characteristics (CLα WL) can be estimated using the Helmbold equation34 modified for compressibility correction as given by Eq. one would expect an alteration to the spanwise lift distribution and the trailing vortex system downstream since circulation along the wingspan changes accordingly. and a reduction in the wing vortex-induced drag. Recognising an adjustment required to conform to the reference wing convention ∆C D = 2 (− L WL sin α ind + D o WL cos α ind + Di WL cos α ind ) + ∆C D comp + ∆C Di qS W S WL (C Do WL cos α ind + C Di WL cos α ind − C L WL sin α ind ) + Φ MCR ∆C D comp + ∆C Di SW (191) =2 with the ∆CD comp component considered to be greater than zero if the winglet MCR has been exceeded by the freestream Mach number. An account of the zero lift angle of attack αoL is assumed here to be approximately α = –3° for 3% cambered aerofoil sections commonly used for winglet devices. (189) and Eq. is quantified by comparing the original wing planform and an equivalent wing planfom with winglets canted as some angle ΓWL off the vertical. By incorporating the vortex-induced drag factor derived by Obert3 and given by Eq. In this context. It is highlighted that the incremental zero-lift drag due to presence of winglets must be considered in isolation from the vehicular characteristic length and the ECLM drag prediction algorithm. (190) into Eq. the winglet device is taken to be an add-on to an existing vehicle wing planform. By virtue of attaching winglets to the tips of a wing. and therefore. a useful basis is to refer to the fractional change in the vortex-induced drag factor used to augment the original ∆CDi denoted by the subscript “orig” dC ∆C Di = O D C Di orig (192) dC 2 L The fractional change operator for the vortex-induced drag factor. (162) the operator becomes . The CDo WL contribution is derived using the component build-up method with an adjustment for interference as outlined earlier.

5.0% 2. One undesirable feature of this method is the fact ∆CDi approaches zero with decreasing ΓWL. i. The problem can be allayed by stipulating an accepted . the agreement for the B737-800 appears to be mostly a good one. this analytical sensitivity does not parallel the winglet parametric study results presented by Ishimitsu70.0% 737-800 Winglet Actual 737-800 Winglet Predicted 4. 500 nm and less. Comparison between flight-test derived71 and predicted improvement in block fuel for B737-800 commercial transport. Actual data derived from flight-testing was taken from results published by Dees and Stowell71.0% Brochure OEW Typical Mission Rules LRC Mach 0.2 Proficiency of Drag Reduction due to Winglet Prediction Figure 32 shows a comparison of the calculated and actual improvement in block fuel for a Boeing B737-800 narrow-body transport.7.5). Some precision is lost for short-range missions. With regards to the exercise of predicting a change in block fuel due to presence of winglets. a calibrated drag model assuming no wing tip device was created from information generated by Boeing121 and subsequently contrasted against an assumption of winglets installed.e.0% 1. which are characterised by lower operating lift coefficients (CL < 0.0% 0 500 1000 1500 2000 2500 3000 Stage Length (nm) Figure 32.100 CONCEPTUAL AIRCRAFT DESIGN METHODS AR orig − AR rev (193) π   AR rev 1 + AR orig   150  where for winglet span (or height) and root chord of hWL and cWL respectively the revised aspect ratio (ARrev) is defined purely on the basis of geometry O ∆ dC D dC 2 dC D L = = dC 2 dC D dC 2 orig L L ( ( ) ) AR rev = (b + 2 h WL tan ΓWL )2 S W + h WL c R WL (1 + λ WL ) tan ΓWL (194) 7.0% Block Fuel Improvement (-) 3. nonetheless.

35 0. the results indicate there exists a good likelihood that CDM will produce predictions well within ±10%.e. the inner boundary labelled “Core Predictions” are points that will always need to be considered during the course of examining the viability of a design candidate from an operational performance perspective. Each chart indicates two zones of prediction effectiveness: “Infrequent Excursions” alludes to operating points within the certified aircraft flight envelope that are seldom impinged during typical operation. mission role and even power plant installation philosophy.0% Prediction Error for Total Drag Figure 33. . Figure 35 and Figure 36 show the agreement between predictions using CDM and flight test drag polars for the Saab Aerospace AB Saab 2000114. 7. and it is discernable that core predictions will stay within an acceptable ±5% error bandwidth. very low and very high operating lift coefficients.55 Mach number (-) 0. whereas.8 Validation of the Total Aerodynamic Drag Model The aerodynamic performance characteristics of known contemporary aircraft were available to validate the predictive powers of the methods discussed – henceforth referred to as the Combined Drag Model (CDM).50 LRC Speed 0.0% 5. i. CDM prediction effectiveness inspected for the Saab 2000 high-speed turboprop regional transport. Figure 34. Bombardier Aerospace Global Express135 and Boeing B737-800121 respectively.60 0.30 -10. By virtue of conducting a validation exercise that encompasses aircraft of varying size.40 Infrequent Excursions 0. Cant angles less than approximately 15° are not permissible because it is indicative of a less pronounced rate change improvement in CDi with respect to wing root bending moment.0% 0. it gives less scope to provide aerodynamic interference relief between wing and winglet and is detrimental in delaying the formation of shock waves on the winglet upper surface.0% 10.LOW-SPEED & HIGH-SPEED AERODYNAMICS 101 design protocol of winglet integration not violating a minimum cant angle or lower threshold of ΓWL. Bombardier Aerospace Learjet 60134. Mmo Boundary 0. and of equal importance. Figure 33.45 Core Predictions 0.0% -5.

CDM prediction effectiveness inspected for the Learjet 60 midsize turbofan business aircraft.75 Mach number (-) LRC Speed 0.70 Infrequent Excursions 0.0% Prediction Error for Total Drag Figure 35.0% 5.0% 0.0% -5.60 -10.85 Mach number (-) 0. .0% 10.0% 0.70 0. CDM prediction effectiveness inspected for the Global Express ultra long range turbofan business aircraft.102 CONCEPTUAL AIRCRAFT DESIGN METHODS Mmo Boundary 0.65 Infrequent Excursions 0.0% 10.0% -5.75 0.80 Core Predictions 0.60 -10.90 Mmo Boundary Core Predictions 0. 0.0% Prediction Error for Total Drag Figure 34.65 0.80 LRC Speed 0.0% 5.

CDM prediction effectiveness inspected for the B737-800 narrow-body commercial transport.75 0. .80 LRC Speed Core Predictions Mach number (-) 0.0% 5.0% Prediction Error for Total Drag Figure 36.0% 10.0% 0.65 0.70 Infrequent Excursions 0.LOW-SPEED & HIGH-SPEED AERODYNAMICS 103 Mmo Boundary 0.60 -10.30 used in generating the reference condition. note that τact = 1.0% -5.

104 CONCEPTUAL AIRCRAFT DESIGN METHODS intentionally blank .

LOW-SPEED & HIGH-SPEED AERODYNAMICS 105 .