AESF 5310

Advanced Aerodynamics

Lecture 8: 5 Oct 2022

Gulfstream G600 Instructor: Prof Larry Li 1
Why has Boeing decided to use folding
wingtips on the 777X?

2
 Today’s agenda [AS5.1−5.3]
• Flow over 3D wings (i.e. wings of finite span)
• Wing terminology
• Wing taper and twist
• Airfoils vs wings: 2D vs 3D
• Wingtip/trailing vortices → Downwash velocity → Induced drag
• Modelling INC flow over 3D wings
• Prandtl’s lifting line theory (LLT)
• Lift and induced drag for untwisted elliptical wings

L8 3
 Wing terminology
Airfoil = 2D cross-section of a 3D wing (∞ wingspan)
Wing = 3D wing with a finite wingspan

Wing area (𝑺)

Wing root
(closest to fuselage, bears
highest bending loads)

Wingtip
(farthest from fuselage, shape
affects induced drag, here
L8 4
A380 has wingtip fences)
 Wing terminology

For the Boeing fans…

L8 5
 Common wing planforms

Straight Elliptical Tapered

Swept back Variable sweep Tail-less delta

L8 6
 Wing taper

𝑐𝑟𝑜𝑜𝑡 = Root chord

𝑐𝑡𝑖𝑝 = Tip chord

𝑐𝑡𝑖𝑝
Taper ratio: 𝜆 =
𝑐𝑟𝑜𝑜𝑡 𝑏
𝑐𝑟𝑜𝑜𝑡

𝑐𝑡𝑖𝑝
L8 7
 Wing taper: Coordinate transform
To transform the linear coordinate 𝑦 to the polar
𝑏
coordinate 𝜃 (which is easier to integrate), we define: 𝑦 = − 2 cos 𝜃
Chord length distribution in 𝜃: 𝑐 𝜃 = 𝑐𝑟𝑜𝑜𝑡 − (𝑐𝑟𝑜𝑜𝑡 − 𝑐𝑡𝑖𝑝 ) cos 𝜃
𝜋 𝜋 𝜋
For example, when 𝜃 = 10 : 𝑐 = 5 − 5 − 1 cos 10 ≈ 1.196
10

𝜋
𝜃=
10
𝜃=0 𝜃=𝜋

𝑐𝑟𝑜𝑜𝑡 = 5

𝑐𝑡𝑖𝑝 = 1
𝑐𝜃= 𝜋 ≈ 1.196
𝑏 10 𝑏
L8 𝑦=− 𝑦= 8
2 2
 Mean aerodynamic chord
Find by graphical method:
𝑐𝑡𝑖𝑝
𝑐𝑟𝑜𝑜𝑡

𝑐𝑟𝑜𝑜𝑡
𝑐ҧ
𝑐𝑡𝑖𝑝

𝑐𝑡𝑖𝑝
Find by equation:
𝑐𝑟𝑜𝑜𝑡
𝑐𝑡𝑖𝑝
Recall that: 𝜆 =
𝑐𝑟𝑜𝑜𝑡

2 𝜆2 + 𝜆 + 1
𝑐ҧ = 𝑐𝑟𝑜𝑜𝑡
3 𝜆+1

L8 9
 Wing twist: Two main types

Generally Not
preferred common

Geometric twist → Orientation of the chord line (i.e. 𝛼) varies across

the wingspan, but the airfoil section remains constant.

L8
Aerodynamic twist → Airfoil section (i.e. 𝛼𝐿=0 ) varies across the wingspan, 10
but 𝛼 remains constant.
 Wing twist: Coordinate transform
To transform the linear coordinate 𝑦 to the polar
𝑏
coordinate 𝜃 (which is easier to integrate), we define: 𝑦 = − 2 cos 𝜃 measured at wing root
𝜋
𝛼 distribution in 𝜃: 𝛼 𝜃 = (𝛼 − 𝛼𝑤𝑎𝑠ℎ𝑜𝑢𝑡 cos 𝜃 ) × 180
𝜋 𝜋 𝜋 𝜋
For example, when 𝜃 = 10 : 𝛼 = (𝛼 − 𝛼𝑤𝑎𝑠ℎ𝑜𝑢𝑡 cos 10 ) × 180
10

𝜋
𝜃=
10
𝜃=0 𝜃=𝜋

𝑏 𝑏
L8 𝑦=− 𝑦= 11
2 2
 Wing twist: Washout

(Washout angle = 3o)

L8 12
 Why use washout?
Wing root: Higher
AOA → Stalls first

In nearly all aircraft, you want the wing root to

stall first. This is to ensure that, at high AOAs,
there is still attached flow over the outboard
sections of the wings, making the aircraft more Wing tip: Lower AOA → Stalls later →
stable and more resistant to entering a spin. Keeps flow attached over outboard control
Root stall also ensures some aileron effectiveness surfaces (e.g. ailerons) → Retain roll control
during the stall, giving you greater control of the for stability, even if wing root has stalled
aircraft, especially if the stall occurs in a banked turn.
If your aircraft did not have washout, in most cases the entire wing would stall at once, or
worse, during a turn when you have aileron deflection in, the wing tip could stall first. If that
happened, your aircraft could roll aggressively and possibly enter a spin.

Typical washout is 2−6o. For the F/A-18 Hornet, the washout is 4o, which means 13
that if the wing root is at 10o AOA, the wing tip is at only 6o AOA.
 Why use washout?

L8 14
 What about wash-in? Is it ever used?
Yes, but very rarely…

Boeing 747-8 Grumman X-29

Wash-out to promote root stall and to Wash-in to compensate for the
reduce the bending moment at the additional root stalling promoted
L8 wing root (thus reducing weight) by the forward sweep 15
 Boeing 747-8 undergoing stall testing

L8 [Link] 16
 Airfoils vs wings: 2D vs 3D
Airfoil → Infinite span → 2D Wing → Finite span → 3D
Extends infinitely in
the spanwise direction

Extends infinitely in
the spanwise direction

Low 𝑷
Low 𝑷

High 𝑷 High 𝑷
Kutta
condition Wing-tip vortex

For an airfoil, lift depends on a physical For a wing, lift depends on the same thing
barrier (the airfoil itself) to maintain a but with added physics → Around the
pressure difference above and below. wingtips, high 𝑃 fluid below is free to ‘roll
L8 17
up’ towards low 𝑃 fluid above.
 Airfoils vs wings: 2D vs 3D
Airfoil → Infinite span → 2D Wing → Finite span → 3D
Extends infinitely in
Topic of L5−7
the spanwise direction Topic of L8−10

Thin Airfoil Theory Lifting Line Theory

TAT LLT
Extends infinitely in
the spanwise direction

Low 𝑷

High 𝑷
Low 𝑷

High 𝑷
Kutta
condition Wing-tip vortex

For an airfoil, lift depends on a physical For a wing, lift depends on the same thing
barrier (the airfoil itself) to maintain a but with added physics → Around the
pressure difference above and below. wingtips, high 𝑃 fluid below is free to ‘roll
L8 18
up’ towards low 𝑃 fluid above.
 Flow physics near the wingtip: Wingtip vortices
These are a pair of counter-rotating vortices
Airflow below wing is Airflow above wing is
that develop from the wingtips.
diverted towards wingtip diverted towards wing root

They propagate
backwards behind
the airplane, and
may be visible
because of
condensation.

They pose a safety

concern for small
aircraft following
These terms are often large aircraft
used interchangeably (min. separation for
takeoff/landing) but
Wingtip vortices are eventually
dissipated by the action of viscosity.
Low pressure
These wingtip vortices cause a spanwise
variation in pressure and, hence, in lift.
L8 19
High pressure
 A380 landing at Zurich airport

L8 [Link] 20
 A380 wake turbulence testing

L8 [Link] 21
 Wingtip/trailing vortices → Downwash velocity → Induced drag

Motion in trailing vortices produces a downwash velocity 𝑤 between the 2 wingtips:

This downwash velocity tends to push

the free-stream velocity vector downwards,
reducing the effective angle of attack:

−𝑤
𝛼𝑖 = tan−1 > 0 (𝑤 < 0 if downwards)
𝑉∞
𝛼𝑒𝑓𝑓 = 𝛼 − 𝛼𝑖 < 𝛼 so we expect less lift…

…but that’s not the

worst part of this story…
L8 22
 Wingtip/trailing vortices → Downwash velocity → Induced drag
…we also get more drag → Induced drag: the ‘cost’ of generating lift
The local lift vector is always ⊥ to the local relative wind (i.e. the local free-stream).
Downwash (𝑤) pushes the local free-stream down, thus tilting the local lift vector back
such that a horizontal component develops in the direction opposing the airplane’s
motion. This horizontal component of the tilted local lift vector is called induced drag.

Induced drag is a purely inviscid phenomenon:

No shear stresses → No skin friction
No viscosity
No BL → No BL sep. → No pressure drag
-BUT … still get finite drag due to induced drag.
-This does not violate d’Alembert’s paradox,
which applies only to 2D (not 3D) bodies.

L8 23
 How do we calculate lift and induced drag for 3D wings?

ANSWER: Use Prandtl’s lifting line theory (LLT) → 𝐶𝐿 and 𝐶𝐷𝑖 [AS5.3]

Before exploring LLT, let’s consider how a finite wingspan affects the spanwise
distribution of lift:
If the circulation distribution
Γ(𝑦) is known, we can sub it
into here and then integrate
across the wingspan to get
the total lift of the wing:
𝑏/2
𝐿 = න 𝐿′ 𝑦 𝑑𝑦
Ludwig Prandtl −𝑏/2

The effect of wingtip vortices is strongest at their origin (i.e. at the wingtips, where
𝑦 = ± 𝑏Τ2) and is weakest at the wing root (𝑦 = 0). This leads to zero lift at the
wingtips (due to pressure equalization there) and maximum lift at the wing root. 24
 How do we calculate lift and induced drag for 3D wings?

Real wing →

LLT →

L8 25
 Lifting line theory (LLT) [AS5.3]
LLT can be thought of as an extension of TAT, but LLT accounts for both (1) the flow
over the wing and (2) the effect of wingtip vortices. It does this by replacing the wing
with a superposition of an infinite number of horseshoe vortices along the lifting line:

The bound vortex is

assumed to sit at ≈ 𝑐/4
→ No longer have LE or
TE → Just a `lifting line’ Γ 𝑦 is the circulation
distribution along the
lifting line

L8 26
 LLT: Formal development [AS5.3]
Given the superposition of a continuous distribution of infinitesimal contributions, the total
downwash velocity induced at position 𝑦0 (along the lifting line) by the entire trailing vortex
sheet is found by integrating [AE5.14] – i.e. the velocity induced by a segment of the trailing
vortex sheet – across the wingspan to give:
𝑑Γൗ 𝑑𝑦
−1 𝑏/2 𝑑𝑦
𝑤 𝑦0 = ‫׬‬ [AE5.15]
4𝜋 −𝑏/2 𝑦0 −𝑦

1
Note that 4𝜋 here is exactly half the value for an infinite line vortex. The induced angle of
attack corresponding to this particular 𝑤(𝑦0 ) can be written as [see AF5.6]:

−𝑤(𝑦0 ) −𝑤(𝑦0 )
𝛼𝑖 𝑦0 = tan−1 𝑉∞
≈ 𝑉∞
with 𝑤 < 0 pointing downwards [AE5.16-17]

If the circulation distribution Γ(𝑦) is known across the wingspan (−𝑏Τ2 ≤ 𝑦 ≤ 𝑏Τ2), we can
determine the total lift and induced drag:

𝑏/2 𝑏/2
𝐿 = ‫׬‬−𝑏/2 𝐿′ 𝑦 𝑑𝑦 = ‫׬‬−𝑏/2 𝜌𝑉∞ Γ 𝑦 𝑑𝑦 (via Kutta−Joukowski’s theorem)
𝑏/2 𝑏/2
𝐷𝑖 = ‫׬‬−𝑏/2 𝐿′ 𝑦 sin 𝛼𝑖 𝑦 𝑑𝑦 ≈ ‫׬‬−𝑏/2 𝜌𝑉∞ Γ 𝑦 𝛼𝑖 𝑦 𝑑𝑦 (assuming small 𝛼𝑖 )

27
Note that 𝛼𝑖 𝑦 is also determined by Γ 𝑦 via [AE5.15], which is for the downwash velocity.
 LLT: Roadmap
Lifting line theory (LLT)
This lecture Next lecture
Untwisted Generic wing
elliptical wing

Know wing geometry

and flow conditions

VH2 Streamline
Find 𝐴𝑛
Cessna 172

Get Γ(𝜃)

Get 𝐶𝐿 and 𝐶𝐷𝑖

L8 28
 LLT: Elliptical lift distribution [AS5.3.1]
To demonstrate the applicability of LLT, we will use one of the simplest smooth functional
forms for Γ 𝑦 that satisfies the constraints expected of a real wing:
Γ0 = max circulation at 𝑦 = 0
Γ 𝑦 = 0 = max at wing root
Γ 𝑦 = ±𝑏/2 = 0 at wing tips

This is an elliptical lift distribution, which

we get by letting:

𝑦 2
Γ 𝑦 = Γ0 1 −( )
𝑏/2

Question: What are the aerodynamic

properties of a wing with such an elliptical
lift distribution?

Things we’re interested in: Lift coefficient (𝐶𝐿 ), downwash velocity (𝑤), induced drag
coefficient (𝐶𝐷𝑖 ), etc….and how these quantities are related to the geometry of the wing.
29
 LLT: Elliptical lift distribution
For an elliptical lift distribution
For lift, we use KJ to get: (see previous slide)

𝑏/2 𝑏/2 4𝑦 2
𝐿= 𝜌𝑉∞ ‫׬‬−𝑏/2 Γ(𝑦) 𝑑𝑦 = 𝜌𝑉∞ Γ0 ‫׬‬−𝑏/2 1− 𝑑𝑦 [AE5.37]
𝑏2

−𝑏
Using a change of variables, we let 𝑦 = ( 2 ) cos 𝜃:

𝜋𝑏 𝑏𝜋
𝐿 = 𝜌𝑉∞ Γ0 ‫׬‬0 sin2 𝜃 𝑑𝜃 = 𝜌𝑉∞ Γ0 [AE5.38]
2 4

Usually we aim for a targeted amount of lift, so let’s express Γ0 in terms of 𝐿 or 𝐶𝐿 :

1 𝑏𝜋
𝐿 = 𝜌𝑉∞2 𝐶𝐿 𝑆 = 𝜌𝑉∞ Γ0
2 4

This yields:
2𝑉∞ 𝑆𝐶𝐿
Γ0 = [AE5.40]
𝑏𝜋

L8
Spitfire 30
where 𝑆 is the planform area of the wing (WW2)
 LLT: Elliptical lift distribution
𝑦
Sub’ing Γ 𝑦 = Γ0 1 − (𝑏/2)2 into [AE5.15] gives (see AS5.3.1):

−Γ0
𝑤 𝑦 = <0 [AE5.35]
2𝑏

which shows that, for an elliptical lift distribution, the downwash velocity 𝑤 is constant over
the entire wingspan! This 𝑦 independence also extends to 𝛼𝑖 :
−𝑤 Γ0
𝛼𝑖 = = [AE5.36]
𝑉∞ 2𝑏𝑉∞

To get this in terms of lift, we substitute [AE5.40] into [AE5.36]:

Spitfire
𝐶𝐿 𝑆 𝐶𝐿
𝛼𝑖 = = [AE5.42] (WW2)
𝜋 𝑏2 𝜋 𝐴𝑅

where the aspect ratio (AR) of the wing is defined as:

𝑏2 𝑏
𝐴𝑅 ≡ For a rectangular wing, 𝐴𝑅 = 𝑏/𝑐 𝑐
𝑆

In the limit of 2D airfoils, LLT behaves as expected: 𝑏 → ∞, 𝐴𝑅 → ∞, 𝛼𝑖 → 0 , 𝐷𝑖 → 0 31

 LLT: Elliptical lift distribution
The induced drag is just the horizontal component of the section lift vector 𝐿′(𝑦), assuming
small 𝛼𝑖 𝑦 such that sin 𝛼𝑖 𝑦 ≈ 𝛼𝑖 𝑦 :
𝑏/2
𝐷𝑖 = ‫׬‬−𝑏/2 𝐿′(𝑦) 𝛼𝑖 𝑦 𝑑𝑦

The induced drag coefficient is therefore:

𝐷𝑖 1 𝑏/2
𝐶𝐷𝑖 = 1 =1 ‫׬‬ 𝜌𝑉∞ Γ(𝑦) 𝛼𝑖 𝑦 𝑑𝑦 [AE5.42a]
2𝑆
𝜌𝑉∞ 𝜌𝑉 2 𝑆 −𝑏/2
∞
2 2

For elliptical wings, because 𝛼𝑖 𝑦 is independent of 𝑦, the math is simple. We get:

Γ20 𝑏𝜋 2𝑉∞ 𝑆𝐶𝐿

𝐶𝐷𝑖 = 2 4 → Sub for Γ0 = [AE5.40]
𝑏𝑆𝑉∞ 𝑏𝜋

𝑆 𝐶𝐿2 𝐶𝐿2
𝐶𝐷𝑖 = =  Note 𝐶𝐷𝑖 ∝ 𝐶𝐿2 [AE5.43]
𝑏2 𝜋 𝜋 𝐴𝑅
Induced drag ≈ 25% of total drag during
Please remember: these results cruise, but increases during takeoff and
are for the elliptical case only! landing when 𝐶𝐿 is large.
L8 32
 LLT: Elliptical lift distribution
The induced drag is just the horizontal component of the section lift vector 𝐿′(𝑦), assuming
small 𝛼Question:
𝑖 𝑦 suchHow sin 𝛼𝑖 of
that much 𝑦 the≈𝛼 𝑖 𝑦 drag
total : of an airplane is
induced drag? A generic answer to this question is
𝑏/2
shown in the𝐷bar 𝑖 = chart
‫׬‬ in𝐿′(𝑦) 𝛼𝑖 𝑦Here,
[AF5.27]. 𝑑𝑦the induced
−𝑏/2
drag relative to parasite drag is shown for typical cruise
and take-off
The induced conditionsisfor
drag coefficient a generic subsonic jet. The
therefore:
parasite drag is the sum of (i) the skin friction drag and
(ii) the pressure 𝐷𝑖 to flow1 separation 𝑏/2
𝐶𝐷𝑖 drag
= 1 due 2
= 1 2 ‫׬‬−𝑏/2 𝜌𝑉associated
∞ Γ(𝑦) 𝛼𝑖 𝑦 𝑑𝑦 [AE5.42a]
with the entire airplane, 𝜌𝑉 𝑆 𝜌𝑉 𝑆
2 ∞including 2 ∞its wing.
In contrast to induced drag, which does not require
For elliptical wings,
viscosity, because
parasite drag𝛼does
𝑖 𝑦 require
is independent 𝑦, the
viscosityofand is math is simple. We get:
therefore not predicted by LLT (which is an inviscid2𝑉 𝑆𝐶
Γ20 𝑏𝜋 ∞ 𝐿
theory)...more𝐶𝐷on 𝑖
= this later…
2 → Sub for Γ0 = [AE5.40]
𝑏𝑆𝑉∞ 4 𝑏𝜋

𝑆 𝐶𝐿2 𝐶𝐿2
𝐶𝐷𝑖 = =  Note 𝐶𝐷𝑖 ∝ 𝐶𝐿2 [AE5.43]
𝑏2 𝜋 𝜋 𝐴𝑅
Induced drag ≈ 25% of total drag during
Please remember: these results cruise, but increases during takeoff and
are for the elliptical case only! landing when 𝐶𝐿 is large.
L8 33
 Elliptical wing Spitfire
VH2 Streamline (WW2)
For untwisted wings, both 𝛼 and 𝛼𝐿=0 are independent of 𝑦 → No variation across wingspan.
For an elliptical lift distribution, 𝛼𝑖 is also independent of 𝑦.
So if all 3 angles are independent of 𝑦, then…
…the section lift coefficient
𝐶𝑙 𝑦 = 2𝜋(𝛼 − 𝛼𝑖 − 𝛼𝐿=0 )
is also independent of 𝑦
independent of 𝑦
1
From the definition of the lift per unit span: 𝐿′ 𝑦 = 𝜌𝑉∞ Γ 𝑦 = 2 𝜌𝑉∞2 𝐶𝑙 𝑐(𝑦)

∴ Γ 𝑦 ∝ 𝑐(𝑦)
Consequently, to produce an elliptical lift distribution (like that in our analysis), we need an
elliptical wing planform (i.e. an elliptical chord distribution) → Elliptical wing!
In HW3, you will see that an elliptical lift distribution produces the lowest induced drag for a
given lift and AR. So then why don’t all aircraft have elliptical wings?

Elliptical planforms are expensive to make:

Rectangular planforms are cheaper but their lift distribution is not optimal:

A good trade-off is the tapered planform:

L8 34