Finite Wings

Kelvin’s Theorem

The total vorticity in a potential flow must be zero.

The total circulation of a given fluid element cannot change unless work is done on it. This leads to several conclusions. The “fluid
element” can include all the fluid in the whole flowfield of interest to us. So, we can say that if positive vorticity is generated
somewhere in the flow, there must be negative vorticity generated somewhere else to compensate for it. This is seen in the case of the
starting vortex

The starting vortex is a time-rate-of-change issue. The same arguments lead to what are known as the Helmholtz vortex theorems for
steady flow.

Helmholtz Vortex Theorems

1. The strength of a vortex filament is constant along its length. This leads to the next argument:

2. A vortex filament cannot begin or end in an irrotational flow; it must extend to the boundaries of the fluid element, or form a closed
loop.

Note the fact that vorticity can indeeed be generated (or absorbed) at boundaries of the potential flow. Such boundaries may be the
rotational core of a vortex, a sink, or a solid wall, or a boundary between two layers of fluid of different properties. Thus, vorticity can
be generated at density gradients in a fluid: for example, at the interface between a “cold front” and a “warm front”, or at the edge of a
plume of hot air rising (such as occurs above conference halls where particularly intense discussions or faculty meetings are
occurring).

From the above, we can infer the vortex system generated by a finite wing moving in an otherwise irrotational flow.

Any change in bound circulation along the wing span must be accompanied by the trailing of a vortex filament whose strength is equal
and opposite to the said change.
Thus, if the change in bound circulation is continuous, a “sheet” of vorticity, rather than discrete vortices, will be trailed all along the
span. The strength of this trailing vortex sheet at any spanwise station y0 is thus:

dΓ
γ(y0 ) = –
dy
y0

Although such a sheet of vorticity is trailed everywhere along the span (where the bound vorticity is changing), the sheet on either side
of midspan will roll up into one tip vortex. This can be seen by considering the rate of change of the bound circulation along the span,
and the sign of the velocity induced by each element of the sheet on its neighbors. The rate of change of the bound circulation is
generally sharpest near the tips, where circulation must drop to zero. It would not make much sense to design wings where this is not
so, for it would mean that the outboard portions of the wing were not of much use. So, the vortex sheet strength must be highest near
the tips, and hence this portion of the sheet will roll up first into a tight “tip vortex”. Other portions of the sheet, further inboard, will
eventually roll up into the tip vortex as well, but this may only happen several wing chord lengths downstream of the wing. Viewed
from far away (such as by AE’s who look up at the sound of every passing airplane), the vortex system will thus appear to consist of
two tip vortices, one from each wing tip. Now we know from the theorems above that this is not the entire vortex system. First, there
must be the “bound vortex” along the wing span itself. This, plus the vortex sheet in the near field of the wing, and the tip vortices,
form what looks like a “horseshoe” , though the wing being straight, the poor horse who wears such a shoe with square corners must
indeed be uncomfortable.

The vortex system must extend downstream to wherever the bound circulation was first set up: starting vortices must have been
created at the same time, and left there, spinning away madly forever. This completes the vortex system of the finite wing:

1. Bound circulation, varying along the span, and dropping to zero at each wingtip.
2. Inboard trailing vortex sheet.
3. Tip vortices, one from each wing tip, getting stronger downstream until all of the inboard vortex sheet has rolled up into the vortex
on each side.
4. Starting vortex, left behind where the bound circulation was first set up.

This is close to reality, with a few minor modifications:

a) The starting vortex is generally created as a continuous increase of vorticity over a long distance, unless the aircraft suddenly
changes angle of attack to create lift. After that, as angle of attack, flap settings etc. are changed, corresponding starting-vorticity
distributions are left behind.
b) These vortices don’t last forever: viscosity and disturbances in the air dissipate them, or they run into the ground (the runway) and
get dissipated.

Prandtl’s Lifting Line Model

As the 20th century was beginning, the Wright Brothers were busily studying the characteristics of airfoil sections in their own wind
tunnel, in order to develop propellers and wings for their own flight test. Other pioneers were jumping off cliffs with wings strapped to
their backs, or refining the only aspect of aerospace engineering that held REAL promise: lighter-than-air airships. Ludwig Prandtl’s
group was also doing experiments and constructing theoretical descriptions of aerodynamics in Germany. The simple model of the lift
and induced drag of wings in low-speed flight is called the Prandl Lifting Line model. It consists of the following:

a) A straight bound vortex Γ(y), placed along the quarter-chord line of the wing.
b) A trailing vortex distribution of strength γ(x,y) such that
∂γx ∂γy
=
∂y ∂x
(yes, there’s also the starting vortex, but we hope it has been left so far behind that the velocity induced by it is negligible, and hence
does not vary with time as the wing keeps flying forward)

The flow problem can be solved as a steady potential flow in the incompressible regime. We know that this flow is described by the
Laplace equation.

One boundary condition to be satisfied is that the disturbances caused by the wing must go to zero when one goes far enough away.
We’ll bear this in mind, but we are much more interested in the boundary condition to be satisfied in the near field:

The near-field boundary condition to be satisfied by the potential flow solution is:
∂φwing ∂φ wake
+ + U∞ α = 0
∂z ∂z

Here z is along the vertical axis, which is normal to the wing, since the wing is at a very small angle of attack.
The boundary condition says:
“The normal velocity due to the wing + wake + freestream must be zero at the wing surface.”
Otherwise flow must be going right through the wing, in which case it could not support much of a pressure difference.

wb + wi + U∞α = 0

Thus, we can satisfy the boundary condition at the 3/4-chord point. After all, this is a point on the wing, and so the boundary condition
should hold there. For reasons that we shall see, the 3/4-chord point is a convenient boundary-condition-satisfying station in aerody-
namics, when we have to consider the effects of vortices in the wake. Its certainly more convenient than the 1/4-chord point, where we
have placed the bound vortex itself, and we don’t want to get into the difficulties associated with being right in the center of the
vortex.

Consider one of the many horseshoe vortices which go into building up the vortex system of the wing. This is a vortex whose
spanwise component (its contribution to the bound vorticity on the wing), is located in the region -y0 < y <y0, and strength is ∆Γ0.

Downwash at (c/2,y) which is on the 3/4 chord line due to this element is (using the Biot-Savart expression):
– ∆Γ
∆w b = ---------- ( Cosβ1 – Cosβ 2 )
4πd
  
– ∆Γ  y 0 + y y 0 – y 
∆wb = -----------------  -------------------------------------------- + -------------------------------------------
c(y)
4π ----------   c + ( y + y )2  c + ( y – y )2
2 2
2  --
 2- 0 --
 2- 0 

. For large aspect ratio, we can neglect the square of c/2 in the square root terms without much error. Thus, to good approximation,
– ∆Γ
∆w b = -----------------------
c(y )
2π  ----------
2
. The induced velocity at (c/2, y) due to all such bound vortex elements is .
–Γ ( y )
w b = -----------------------
( y )
2π  c--------- -
2 

Downwash due to the trailing vortices is:

b  – d Γ ( y )
---  d y 0 
1
w i = ------ 2 ----------------------------- dy0
∫
4π –-----b- y – y0
2
. Summing up the normal velocity components due to the bound vortices, the trailing vortices and the angle of attack,

b  – d Γ ( y )
–Γ ( y ) ---  d y 0 
1
---------------------- – --------------- 2 ----------------------------- dy 0 + α = 0
πc ( y )U ∞ 4πU – b y – y ∫∞ ------ 0
2
. This is Prandtl’s Lifting Line equation, or the Fundamental Equation of Lifting Line Theory. It is an “integro-differential equation”.
The solution which we seek from it is the spanwise load distribution G(y). All it really says is:
αe = α – αi
, i.e, “The effective angle of attack at each spanwise section is the geometric angle of attack minus the angle of attack induced by the
wake”.

The effective angle of attack is that corresponding to the bound circulation. The geometric is the angle at which one sets the section,
including its zero-lift angle of attack if it is cambered, and the wing twist if any. The induced angle of attack is the angle formed as
inverse tangent of the ratio of induced velocity to freestream velocity.

Now in the above derivation we conveniently assumed that the section lift curve slope was 2π when we converted lift coefficient to
angle of attack. This restriction is not necessary. Also, we can write out the equation explicity including the zero-lift angle of attack.
b  – d Γ ( y )
– 2Γ ( y ) ---  d y 0 
1
---------------------------------- – --------------- 2 ----------------------------- dy 0 + α ( y ) – α L = 0 ( y ) = 0
m 0 ( y )c ( y )U ∞ 4πU –b y – y ∫ ∞ ------ 0
2

subject to Γ(y=+b/2)=0

Note: We could have derived the bound circulation contribution term as follows:

Section lift coefficient corresponding to the section effective angle of attack is:
L' ρU∞ Γ ( y )
cleff = --------- = ----------------------- = 2πα e
cq ∞ c ( y )q ∞
This strengthens the argument that if you find the downwash at the 3/4 chord point, but put your lifting line corresponding to the lift at
the 1/4-chord, things will all work out. This is used extensively in unsteady aerodynamics, where one has to figure out the induced
velocity due to the wake as a function of time, after each change in angle of attack.
 Glauert Solution

The Lifting Line equation can be solved using a series approach. The equation relates the geometric angle of attack, the induced angle
of attack, and the bound circulation. The induced angle of attack and the bound circulation depend on one another. The Glauert
approach is to assume a fairly general form for the bound circulation, and express the induced velocity in terms of this assumed form.
The assumed form is a series of sine and cosine functions of the spanwise location, with an unknown coefficient multiplying each such
function. Using this series form, the integrals and derivatives are carried out quite simply, with the coefficients left undetermined. The
lifting line equation is then applied at several locations along the span, and as many equations are obtained as there are unkown coeffi-
cients. The problem then reduces to solving a set of simultaneous algebraic equations for the unknown coefficients.

Once the coefficients are obtained, the full expression for the bound circulation is available. The induced velocities can also be calcu-
lated directly. From these, the lift and induced drag are obtained.

We begin the solution by using a transformation of coordinates, similar to what was used in thin airfoil theory.
b
y = --- Cosθ
2
. Assume that the unknown spanwise distribution of bound circulation is given by:
N
Γ ( θ ) = 2bU∞ ∑ An Sin( nθ )
1
, subject to the conditions that
Γ (0 ) = Γ( π) = 0
Substituting into the lifting-line equation gives:
N N Sin ( nθ0 )
4b
α ( θ0 ) = --------------------
m 0 c ( θ0 )
- ∑ A n Sin ( nθ0 ) + α l = 0 ( θ 0 ) + ∑ nA n ----------------------
Sinθ 0
1 1
. Note: this uses the integral . Satisfy the relation for the angle of attack, above, at N values of the spanwise coordinate q0. This gives
N equations for the N unknown coefficients An, n=1,2....N. Solve this set of equations. Then the wing lift coefficient is:
π
Cos ( nθ ) dθ Sin ( nθ 0 )
∫ ----------------------------------- ≡ π ----------------------
Cosθ – Cosθ 0 Sinθ 0
0
–b
------
2
2
C L = ----------
U ∞S
- ∫ Γ ( y )dy
–b
------
2
. Use the integral , for n =1, and is zero if n is not equal to 1. Then the wing lift coefficient becomes:
π
π
∫ Sin ( nθ )Sinθ dθ = ---
2
0

2
b
C L = A 1 π ----- = A 1 π ( AR )
S

Induced Drag
b
---
2
Di =
∫ L' ( y )Sin ( αi( y ) ) dy
b
– ---
2
. For small angles, we can use the approximation . Thus the wing induced drag coefficient is:
Sin ( α i ) ≈ α i
b
---
2
2
C D = -----------
i U∞ S ∫ Γ( y )αi ( y ) dy
b
– ---
2

N
2  A n 2
C D = π ( AR )A 1 1 +
i ∑ n  ------
 A 1
2
. Thus, where e is called the “spanwise efficiency factor”.
2 2
CL CL
C D = ---------------- ( 1 + δ ) = -------------------
i π ( AR ) π ( AR )e

Elliptical Lift Distribution

The lift-to-induced-drag ratio of the wing is: . We see that the maximum of the lift-to-induced-drag ratio occurs when all the
coefficients An are zero, for n greater than 1. Under this condition, the expression for the bound circulation becomes
L CL π ( AR )A 1
----- = --------- = ---------------------------------------------------------------
N -
Di CD  A  2
∑
2
n  -----n-
i
π ( AR )A 1 1 +
 A 1
2