Course Notes – Slender Wings in

Incompressible flow
BS. in Aerospace Engineering, Aerodynamics II by O. Flores

v1.0
IMPORTANT: This document is a summary of the most important concepts and mathematical derivations
of the materials of the module Slender Wings of the course Aerodynamics II that are not covered in the course’s
bibliography (“Low speed Aerodynamics”, by J. Katz & A. Plotkin). This document does not attempt to
cover all the material discussed in class. It is merely a guide to help you in the study of thise module.

0.1 Introduction
Slender wings have very small aspect ratios AR = b2 /S ≪ 1, where b and S are the wing’s span and planform
area, respectively. They are important from a technological point of view, mainly because two factors. First,
they are more manoeuvrable than large AR wings. Second, they are more convenient for supersonic flight than
larger AR wings. Indeed, typical examples of aerospace vehicles with slender wings include acrobatic airplanes
and fighters. From the point of view of the potential aerodynamic theory, taking the limit of AR ≪ 1 in the
lifting surface equations allows for a dramatic simplification of the equations for the angle of attack and lift
problem, providing relatively simple solutions for the distributions of singularities (i.e., doublets) on the wing,
and for the computation of the corresponding aerodynamic forces and moments.
The structure of this chapter is as follows. First, a qualitative assessment of the simplifications provided
by the slender wing limit (AR ≪ 1) is presented in section 0.2. Then, the mathematical simplification of the
potential theory problem is provided in section 0.3. Solutions for the resulting system of equations are presented
in section 0.4, followed by the corresponding calculations of forces and moments (§0.5). The chapter closes with
a brief discussion of the physical significance and limitations of the Slender Wing Theory (§0.6).

0.2 Simplifying the Lifting Surface Theory for AR ≪ 1.

As discussed in class and in chapter 4 of Katz’s book [?], the equations for the Lifting Surface Theory (LST)
for the angle of attack and camber problem are
" #
u′ (x0 , y0 , 0+ )
 
x − x0
Z
′ + dηc 1
w (x, y, 0 ) = U∞ −α + = 1+ p dx0 dy0 . (1)
dx 2π P F (y − y0 )2 (x − x0 )2 + (y − y0 )2

In this equation, u′ and w′ are the perturbation velocities, α is the angle of attack of the wing, ηc (x, y) is the
camber distribution of the wing, and the integral is performed over the planform (PF) of the wing only. These
equations can be used to solve the potential flow around thin wings, with α ≪ 1, ηc ≪ c and ηt ≪ c.
From a physical point of view, the integral in equation (1) can interpreted as the contribution to the vertical
velocity at point (x, y, 0+ ) from a distribution of infinitesimal horseshoe vortices (see figure 1a). The heads
of these horseshoe vortices are placed on the wing’s surface, with their legs parallel to the x-axis, extending
downstream in the z = 0 plane towards x → ∞. The circulation per unit length of these differential horseshoe
vortices is γy (x, y) = 2u′ (x, y, 0+ ), and their heads correspond to the distribution of point vortices that were
used in Aerodynamics I to model 2D airfoils.
When we consider a large AR wing (as in figure 1b), the heads of all these differential horseshoe vortices
become clustered together. This justifies the approximation of Prandtl’s Lifting Line Theory, where all heads
Aerodynamics II v1.0

are grouped together in a single vortex line (i.e., the bound circulation of the wing, Γ(y)), while the legs of
these differential horseshoe vortices results in the vortex sheet that conforms the wake, and that is responsible
for the downwash on the wing.
We can also consider the opposite limit, when AR ≪ 1 and the wing is very slender. This limit is shown
in figure 1(c), where we can see that the heads of the horseshoe vortices become separated from each other.
Indeed, when we focus on any given x section (like the section indicated in the figure in blue), we can observe
that:
A) The vertical velocity at that section, w′ (x, y, 0+ ), will depend mostly on the circulation of the legs crossing
the blue line, with little contribution from the heads. All the legs crossing a given x-section seem to be
infinitely long, extending to very large distances (compared to the local span of the wing) both upstream
and downstream.
B) The circulation of the legs crossing section x only depends on the circulation of the horseshoe vortices
located upstream of x.

(a) z (b) z
y
y

γy (x,y)
Γ(y)

x x

(c) (d)
z z
y
y
x = cte
xLE(y)
Γ̃(x , y)

(x, y)
x b(x)
x

Figure 1: Sketch of the infinitesimal horseshoe vortices on (a) a general finite AR wing as in the Lifting Surface
Method (eq. 1), (b) a large aspect ratio wing as in the Lifting Line Theory, (c,d) a slender wing.

O. Flores 2 July 2021

Aerodynamics II v1.0

0.3 The equations for Slender Wing Theory

From a mathematical point of view, we can derive the equations for the slender wing theory by taking the limit
of AR ≪ 1 in the equation of the Lifting Surface Method, eq. (1). If ∆x, ∆y and ∆z are the characteristic
distance between points in the wing along the streamwise, spanwise and vertical directions, the limit of slender
wings implies
∂ ∂ ∂
AR ≪ 1 → ∆x ≫ ∆y, ∆z → ≪ , . (2)
∂x ∂z ∂y
As a consequence, the linearized equation of conservation of mass becomes

∂2ϕ ∂2ϕ ∂2ϕ ∂2ϕ ∂2ϕ

+ 2 + 2 =0 → + 2 = 0, (3)
∂x2 ∂y ∂z ∂y 2 ∂z

and the velocity potential ϕ(x, y, z) depends only on the cross-flow at section x, consistently with the qualitative
observation A in §0.2.
Noting that in equation (1) the term ∆x ∼ (x − x0 ) is much larger than ∆y ∼ (y − y0 ), the term inside the
square brackets simplifies to
" # " #   
x − x0 x − x0 x − x0 0 if x < x0 ,
1+ p ≈ 1+ p = 1+ = (4)
2
(x − x0 ) + (y − y0 ) 2 (x − x0 )2 |x − x0 | 2 if x > x0 .

This implies that, consistently with the qualitative observation B in §0.2, the area integral in equation (1) does
no longer extend to the whole planform of the wing, but it is restricted to the region of the wing upstream of
the point where we are computing w′ : from the leading edge of the wing, xLE (y), to section x (i.e., the blue
crossed area in figure 1d). Using (4) on (1) we obtain
b(x)/2 x
2u′ (x0 , y0 , 0+ )
Z Z
1
w′ (x, y, 0+ ) = dx0 dy0 , (5)
2π −b(x)/2 xLE (y0 ) (y − y0 )2

where b(x) is the span of the wing at section x, and we have assumed that the trailing edge of the wing is a
straight line with x = c, and that b(x) is a monotonic function of x (more details in section 0.6).
In order to further simplify equation (5), we define Γ̃(x, y) as the integral of the circulation of all the horseshoe
vortex heads upstream of point (x, y), as sketched in figure 1(d). Hence,
Z x Z x
Γ̃(x, y) = γy (x0 , y)dx0 = 2u′ (x0 , y, 0+ )dx0 . (6)
xLE (y) xLE (y)

Substituting into (5) we obtain

Z b(x)/2
1 Γ̃(x, y0 )
w′ (x, y, 0+ ) = dy0 . (7)
2π −b(x)/2 (y − y0 )2

This last expression can be integrated by parts,

Z
′ + dy0
w (x, y, 0 ) = u dv, with u = Γ̃(x, y0 ) and dv = , (8)
(y − y0 )2

to obtain
y0 =b(x)/2 Z b(x)/2
1 1 ∂ Γ̃(x, y0 ) dy0
w′ (x, y, 0+ ) = Γ̃(x, y0 ) − . (9)
2π y0 =−b(x)/2 2π −b(x)/2 ∂y0 (y − y0 )

O. Flores 3 July 2021

Aerodynamics II v1.0

The first term in (9) is zero, since y = ±b(x)/2 yields x = xLE , and then Γ̃ = 0 as per equation (6). Finally, we
obtain
−1 b(x)/2 ∂ Γ̃(x, y0 ) dy0
  Z
′ + ∂ηc
w (x, y, 0 ) = U∞ −α + = . (10)
∂x 2π −b(x)/2 ∂y0 y − y0
This expression relates the distribution of circulation Γ̃(x, y) with the wing’s camber and angle of attack. To
compute the aerodynamic loads, we need cp (x, y, 0+ ), which is obtained from Γ̃ as

−2u′ (x, y, 0± ) ∓1 ∂ Γ̃
cp (x, y, 0± ) = = , (11)
U∞ U∞ ∂x
where the ± accounts for both the upper and lower surface of the wing.
From a physical point of view, equation (10) implies that the vertical velocity on a point (x, y, 0+ ) on a
slender wing corresponds to the velocity produced by an array of infinitely long vortex filaments parallel to the
x-axis, with a circulation per unit length given by ∂ Γ̃/∂y.

0.4 Solving the equations for Slender Wing Theory.

In order to solve (10), we are going to use the following trigonometric change of variable
−b(x) b(x)
y= cos θ; y = −b(x)/2 ↔ θ = 0; y = b(x)/2 ↔ θ = π; dy = sin θdθ, (12)
2 2
and we are going to look for solutions for Γ̃ in the form of a series of sines,
∞
X
Γ̃(x, θ) = Γ̃n (x) sin(nθ). (13)
n=1

This solutions are analogous to the solutions that we obtained for Γ(y) in the Lifting Line Theory in Aerody-
namics I, but now the coefficients of the series of sines are functions of x, i.e. Γ̃n (x). Note also that the total
circulation at a given spanwise sectino of the wing is Γ(y) = Γ̃(c, y) (i.e., Γ̃ evaluated at the trailing edge of the
wing, x = c).
Introducing the trigonometric change of variable (12) into (10), and noting that
∞
∂ Γ̃(x, y0 ) ∂ Γ̃(x, θ0 ) X
dy0 = dθ0 = dΓ̃ = nΓ̃n (x) cos(nθ0 )dθ0 (14)
∂y0 ∂θ0 n=1

we obtain
 
′ + ∂ηc
w (x, θ, 0 ) = U∞ −α + =
∂x
∞
πX ∞
−1 −1 X 2nΓ̃n (x) π cos(nθ0 )dθ0
Z Z
dθ0
= nΓ̃n (x) cos(nθ0 ) = = ... (15)
2π 0 n=1
b(x)/2(cos θ0 − cos θ) 2π n=1 b(x) 0 cos θ0 − cos θ

  ∞
′ + ∂ηc X −nΓ̃n (x) sin(nθ)
w (x, θ, 0 ) = U∞ −α + = . (16)
∂x n=1
b(x) sin θ

Equation (16) can be used to relate the distribution of circulation on the wing (i.e., the functions Γ̃n (x))
with the wing’s planform, camber and angle of attack (i.e., b(x), ηc (x) and α, respectively). There are two
possible strategies to compute the functions Γ̃n (x) given b(x), ηc (x) and α:

O. Flores 4 July 2021

Aerodynamics II v1.0

Option #1: When only a few Γ̃n (x) are needed (i.e., for n = p only), we can multiply both sides of equation
(16) by sin θ sin(pθ) and integrate in θ between 0 and π. Taking into account that
Z π 
0 if p ̸= n,
sin(nθ) sin(pθ)dθ = π (17)
0 2 if p = n,

we obtain a closed expression for

π  
−2U∞ b(x)
Z
∂ηc
Γ̃p (x) = −α + sin θ sin(pθ)dθ. (18)
pπ 0 ∂x

Option #2: When we need all coefficients Γ̃n (x) for n = 1 . . . ∞, it is usually more convenient to multiply
both sides of equation (16) by sin θ, and try to express ∂ηc /∂x sin θ as a series of sines of θ. Then, it is
possible to identify one-by-one the terms of the sine-series on the left hand side and right hand side of the
equation.
Finally, it is very important to note that the trigonometric variable θ is a function of x and y. That implies
that the derivative of the camber ∂ηc /∂x = ∂ηc (x, y)/∂x must be computed when ηc = f (x, y), i.e. before
applying the trigonometric change of variable (12) to obtain a function of (x, θ). Otherwise, we need to use the
chain rule and take into account that θ(x).

0.5 Computing Forces and Moments

In order to compute forces and moments on slender wings, it is possible to use

−2u′ (x, y, 0± ) ∓1 ∂ Γ̃
cp (x, y, 0± ) = = , (19)
U∞ U∞ ∂y
to compute the pressure distribution on the surface of the wing, and integrate it to obtain the wing’s lift (L)
and the pitching moment with respect to the y-axis (My,LE ). However, this implies undoing the trigonometric
change of variable to express Γ̃ as a function of (x, y), differentiating with respect to x this distribution of
circulation, and then integrating in x and y to obtain forces and moments.
It is possible to compute L more directly from Γ̃(x, θ), making use of the Kutta-Joukowski theorem of
Aerodynamics I. Remember that
Z b(c)/2 Z b(c)/2
′ ′
L (y) = ρU∞ Γ(y) → L = L (y)dy = ρU∞ Γ̃(c, y)dy, (20)
−b(c)/2 −b(c)/2

where we have taken into account that Γ(y) = Γ̃(c, y). Substituting the solution (13) into this expression, and
operating with the trigonometric change of variable we obtain
Z π ∞  Z π 
b(c) b(c) X π
L= ρU∞ Γ̃(c, θ) sin θdθ = ρU∞ Γ̃n (c) sin(nθ) sin θdθ = ρU∞ b(c)Γ̃1 (c). (21)
0 2 2 n=1 0 4

This equation provides a closed expression for the total lift of the wing as a function of Γ̃1 (c) and the maximum
span of the wing, b(c), where the former is only a function of the geometry of the wing at the trailing edge.
In order to compute the pitching moment of the wing, it is convenient to compute first L(x), which is the
lift produced by the wing from the leading edge up to section x,
Z b(x)/2
L(x) = ρU∞ Γ̃(x, y)dy. (22)
−b(x)/2

O. Flores 5 July 2021

Aerodynamics II v1.0

Substituting c by x in equations (20) and (21) we obtain the final expression for L(x), namely
Z b(x)/2
π
L(x) = ρU∞ Γ̃(x, y)dy = · · · = ρU∞ b(x)Γ̃1 (x). (23)
−b(x)/2 4

From there, it is possible to derivate with respect to x to obtain dL(x)/dx, which represent the lift per unit
chord generated by an x-section of the slender wing. Multiplying by the distance to the y-axis, and integrating
in x, it is possible to obtain an expression for the pitching moment of the wing,
Z c Z c
dL(x) −π d
My,LE = −x dx = ρU∞ x [b(x)Γ̃1 (x)]dx. (24)
0 dx 4 0 dx

Note that the pitching moment of the slender wing depends on the chordwise distribution of Γ1 (x) only.
Finally, in order to compute the roll moment of the wing we can operate with the Kutta-Joukowski formula
as we did in the Lifting Line Theory in Aerodynamics I to obtain
Z b(c)/2 Z b(c)/2
Mx = yL′ (y)dy = yρU∞ Γ̃(c, y)dy =
−b(c)/2 −b(c)/2
π ∞
−b(c) −π
Z X b(c)
= cos θρU∞ Γ̃n (c) sin(nθ) sin θdθ = · · · = ρU∞ b(c)2 Γ̃2 (c). (25)
0 2 n=1
2 16

Observe that in this case, the roll moment depends on the value of the second harmonic at the trailing edge,
Γ̃2 (x). The rest of the harmonics, n ≥ 3, only have contributions to the induced drag of the slender wing (see
section 8.4 in Katz’s book [?]).

0.6 Discussion
Solution for planar wings
Planar wings are defined as wings with ∂ηc /∂x = f (x), with no dependency on the spanwise coordinate of the
wing. For these wings, equation (16) reads
∞
X −nΓ̃n (x) sin(nθ)
U∞ (−α + f (x)) = , (26)
n=1
b(x) sin θ

where the left hand side of the equation is a function of x only, with no dependency on θ. Then, using option
#2 in §0.4, we can multiply both sides of the equation by sin θ and identify the coefficients of the series of sines
on both sides of the equation to obtain
 
∂ηc −Γ̃1 (x)
U∞ −α + = , Γ̃n (x) = 0 for n ≥ 2. (27)
∂x b(x)

As a consequence, the lift of planar wings can be expressed as

 
π 2 ∂ηc
L = ρU∞ b(c)2 α − , (28)
4 ∂x x=c

and the formula for the lift coefficient as

 
2L π ∂ηc
CL = 2 S
= AR α − . (29)
ρU∞ 2 ∂x x=c

O. Flores 6 July 2021

Aerodynamics II v1.0

101

100

dCL/dα
10-1
SW —
LLT —
10-2 -2
10 10-1 100 101
AR

Figure 2: Comparison of dCL /dα for Slender Wing Theory and Lifting Line Theory for τ̃ = 0 (blue, elliptical
wing planform) and τ̃ = 0.1 (cyan, trapezoidal planform).

The expression for the lift coefficient of planar slender wings allows emphasizing the differences with the
expressions obtained for large AR wings in Aerodynamics I (i.e, LLT). In particular, the slope of the lift
coefficient of slender wings π/2AR. It does not depend on the shape of the wing, only on its AR (remember
that in the LLT the dCL /dα depends on both AR and the geometry of the planform of the wing). On the other
hand, the dCL /dα of slender wings is lower than the values predicted by the LLT for AR < 1, even if we use the
same value of the AR on both cases. This can be observed in figure 2, were the numerical values for dCL /dα
from Slender Wing Theory (SW) and Lifting Line Theory (LLT) are compared.

dCL π dCL 2π
= AR, = (1 − τ̃ ). (30)
dα SW 2 dα LLT 1 + 2/AR

Dependency of aerodynamic forces and moments on the geometry of the slender wing.
Note that, as in the Lifting Line Theory for large AR wings, the lift of the wing only depends on the first
harmonic of the series of sines of Γ̃ (i.e., the part of the lift distribution that is elliptic) and the roll moment
depends on the second harmonic. What is more surprising is that in slender wings these contributions from
Γ̃1 (c) and Γ̃2 (c) are functions of the angle of attack and the local geometry of the trailing edge only. This can
be observed in equation (18) for p = 1 and 2, where Γ̃p (c) becomes a function of b(c) and ∂ηc /∂x evaluated at
x = c only. The rest of the geometry of the wing (i.e., b(x) and ∂ηc /∂x for x < c) does not have any effect on
the total lift of the wing, L, or in the roll moment, Mx . Note also that the same is not true for the lift and roll
2 2
moment coefficients,
Rc CL = 2L/ρU∞ S and CM x = 2L/ρU∞ Sb(c), which depend also on the total surface of the
wing, S = 0 b(x)dx. In other words, the shape and camber distribution of the slender wing on x < c has no
effect on the total lift and roll moment, which only depend on the angle of attack and the local geometry of the
wing at the trailing edge.
The same is not true for the pitching moment, which depends on the distribution of camber of the whole
slender wing.
Finally, note that the lift (and the lift coefficient) are linear functions of the angle of attack and the maximum
camber of the wing: L is proportional to Γ̃1 (c), which is proportional to both α and ηc .

O. Flores 7 July 2021

Aerodynamics II v1.0

Limitations of the Slender Wing Theory.

Obviously, the Slender Wing Theory is based in a potential approximation of the flow around the wing, in the
limit of a thin wing. As such, the solution is limited to high Reynolds numbers and Froude numbers, Prandtl
of order one (i.e., gases like air), small angles of attack and small values of the camber and thickness of the
wing (when compared to the chord of the wing). For a description of non-potential effects on slender wings, the
reader is referred to section 5.6 of Andresson’s book [?].
The solution developed in the previous sections has been computed for the incompressible regime. However,
the fact that the slender wing approximation implies that ∆x ≪ ∆y, ∆z results in a simplification of the Laplace
equation, in which only the cross-flow components remain. This simplification appears naturally as well in the
linearlized equations for compressible potential flow, resulting in a negligible effect of subsonic Mach numbers on
the lift coefficient of slender wings. This can be demonstrated by applying the Prandtl-Glauert compressibility
correction to the solution of the Slender Wing Theory.

(a) (b)
y y
bmax bmax
b(x)

b(x)
(x,y) xmax (x,y)
x x

Figure 3: Regions of integration for slender wings with non-monotonic b(x).

Finally, as mentioned in §0.3, the equations for the Slender Wing Theory assume that b(x) is a monotonic
function of x, so that db(x)/dx ≥ 0 and bmax = b(c). This assumption is necessary to write the area integral as
it appears in equation (5), namely
Z b(x)/2 Z x
dx0 dy0 . (31)
−b(x)/2 xLE (y0 )

This allows the apearence of Γ̃(x, y) in equation 7, which is central to the analytical method developed in §0.3
and §0.4.
Figure 3(a) shows that this definition of the area integral fails for wings with non-monotonic b(x), since it
leaves outside of the integrated area the regions with span y ∈ [b(x)/2, bmax /2] and y ∈ [−bmax /2, −b(x)/2].
As a consequence, equation (5) does not take into account the velocity induced by the vortex lines drawn with
dashed lines in figure 3.
While the Theory of Slender Wings as developed in sections 0.3 and 0.4 is not applicable for wings with
non-monotonic b(x) and arbitrary camber, there is a caveat that allows the application of the theory for flat
wings (i.e., with ηc = 0). Let’s consider the wing sketched in figure 3(b) with ηc = 0 and α ̸= 0, were bmax is
the maximum span of the wing, achieved at section x = xmax . For x < xmax , the general theory works, and we
can compute Γ̃(x, y) = αU∞ b(x) sin θ and the vertical velocity at x = xmax , w′ (xmax , y, 0+ ). Since it is a planar
wing, only the first harmonic is present, and its intensity grows with b(x) for x < xmax .
Lets consider a section of the wing further downstream, at x = x1 :

O. Flores 8 July 2021

Aerodynamics II v1.0

1. At x = x1 , the circulation of the vortex lines (i.e., ∂ Γ̃/∂y) with |y| > b(x1 )/2 is the same as at x = xmax ,
since these vortex lines (dashed lines in figure 3b) are at the wake of the wing, and there are no horseshoe
vortex heads connected to them. This implies that Γ̃(xmax , y) = Γ̃(x1 , y) for |y| > b(x1 )/2.

2. On the surface of the wing (i.e., |y| < b(x1 )/2), the vertical velocity w′ has to be the same at x1 and at
xmax , since at both locations it is given by the boundary condition w′ = −U∞ α.
Together, (1) and (2) imply that ∂ Γ̃/∂y is the same on the whole span of the wing at xmax and at x1 , and so
are Γ̃ and w′ . In other words, the distribution of circulation on the legs of the horseshoes changes from x = 0
to xmax , but stays constant downstream of that point for the flat (i.e., α = 0) wing sketched in figure 3(b). As
a consequence, in order to compute the lift of the wing we can use the distribution of circulation at xmax , since
it is not going to change downstream. This yields
Z bmax /2
π 2 2
L= ρU∞ ]Γ̃(xmax , y)dy = · · · = ρU b α. (32)
−bmax /2 4 ∞ max

For the definition of the lift coefficient it is necessary to take into account the whole area of the wing, from
x = 0 to x = c:
2 π 2 2 π
CL = 2 S 4
ρU∞ bmax α = ARα. (33)
ρU∞ 2
Moreover, these formulas also applies to flat (i.e., ηc = 0) wings with a smooth trailing edge. Assume that
we have the wing shown in figure 4(a). We can always approach that geometry as a series of jumps in y with
size d → 0 (as in figure 4b), with the arguments provided for a single jump (figure 4c) being valid also for
each of the infinitesimal jumps. As a consequence, dΓ̃(x, y)/dy of the flat wing in figure 4 remains constant for
x ≥ xmax , and formulas (32) and (33) are also valid.

(a) (b) (c)

y y y
d

x x x

Figure 4: Approximation of a smooth trailing edge as a series of jumps, in the limit of d → 0. Only the right
hand side of the wing (i.e., y > 0) is plotted.

O. Flores 9 July 2021

Bibliography

[1] J.D. Anderson. Fundamentals of aerodynamics. McGraw-Hill, 2011.

[2] J. Katz & A. Plotkin. Low speed Aerodynamics. Cambridge U. Press, 2001.

10