Ecole Centrale de Nantes

Hydrodynamics
Final Project Report

LIFTING LINE THEORY

Youssef El Iskandarani
youssef.el-iskandarani@eleves.ec-nantes.fr

El Mehdi Es-sabar
El-Mehdi.Es-sabar@eleves.ec-nantes.fr

15/04/2022
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

1. Introduction
1.1 Lifting Line Theory - Prandtl Equation
The Prandtl lifting line theory is a mathematical model allowing the calculation of the performances
of a wing through based on its geometry. Analyzing a three-dimensional wing can prove itself to
be highly complex. Its discretization into multiple cross sections, referred to as airfoils, is therefore
an advantageous method. However, evaluating the performances of each airfoil independently
of the others, and adding all the contributions up does not allow finding correct results for the
complete wing. This is due to the fact that the performance of each airfoil is affected by that
of the neighboring ones. The Prandtl lifting line theory allows a more accurate two-dimensional
approach that results in the circulation distribution along the span of the wing. As seen in figure
1, this circulation actually represents the strength of the vortex. The material contour formed
around the lifting section is sufficiently large to surround the wing at its initial stage, and after
the formation of the velocity profile. As the circulation on this contour is 0, the wing creates a
starting vortex that is equal and opposite to the circulation on the wing. The constant circulation
on the wing is then always verified. The Prandlt mathematical model bases itself on the concept
of circulation and applies the Kutta-Joukowski theorem. The performances of the wing can be
calculated starting from the results stated beforehand.

Figure 1: Illustration of lifting line theory

Modeling lift distribution with circulation

In this work, the form of the Prandlt equation that will be used is shown as equation 1. This
form allows calculating the circulation distribution along a wing’s span, from which we can derive
the according performances.
Z +L !
1 dΓ(η)
Γ(y) = k∞ (y) λ(y) U∞ α(y) + (1)
4π U∞ −L (η − y)

1.2 Scope of the work

The work presented hereafter will tackle the coding of the lifting line theory and its casting into a
mathematical model that can be solved using MATLAB. This model will then be used to compare
the performances of three different wings. Therefore, our team will use equation (1) to evaluate
the circulation distribution along the span of the three different setups and come up with the lift
and drag coefficients that will allow us to quantify each wing’s performance for different aspect
ratios and angles of attack. Our code will then be evaluated by analyzing its convergence, and
comparing our numerical results with theoretical ones.

1
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

2. Methods
2.1 Implementation of the Model
The first step in implementing our model is to discretize the Prandtl equation. This will be done
using N points of discretization that allow traveling in the wing’s span direction, i.e. yi ∈ [−L, L]
with i ∈ {1, ..., N }. We also assume control points with coordinates η ∈ {1, ..., N − 1} located at
the center of each discretization sub-element. Our boundary conditions are such that Γ(y = −L) =
Γ(y = L) = 0. The discretized equation is therefore written in the form:
  
N −1 N −1
1 X Γ j
X Γj
Γi = π.li .U∞ α +  −  (2)
4πU∞ j=2 ηj−1 − yi j=2
η j − yi

The equation can therefore be set into a matricial system of the form:

[Aij ] . [Γi ] = {b} (3)

By developing equation 2, we get:

  
N −1 N −1
li
X Γ j
X Γj
Γi = π.α.li .U∞ +   − 
4 j=2 ηj−1 − yi j=2 j
η − yi

At this stage, we can differentiate two separate cases:

1. Case I: i = j

When the two indices are equal, we can factorize by Γi and write the equation as follows:
  
li 1 1
Γi 1 − − = π.α.li .U∞
4 ηj−1 − yi η j − yi

2. Case II: i ̸= j

When the two indices are not equal, as we expand the equation, we find that it can be cast
as follows:
  
li 1 1
Γi − Γj − = π.α.li .U∞
4 ηj−1 − yi ηj − yi

With those results in our hands, we can fill the matrix [Aij ] of our matricial system, equation (3).
The diagonal values will be the coefficient of Γi for the first case i = j. The other terms will be filled
with the coefficient of Γj from the second case where i ̸= j. The vector {b} is a constant value one
with the value π.α.li .U∞ . In order to satisfy our boundary condition Γ(y = −L) = Γ(y = L) = 0
, we set the first and last terms of our vector equal {b} to 0 as well as the first and last rows and
columns of our matrix [Aij ]. The final expression of our system is then:
 
1 0 0 ... 0
0 A22 A23     

0 A32 A33

 0 0
  Γ2   b 
 .. ..

. .   ..   .. 
    
  .  .  = . (4)
     
  ΓN −1   b 
 AN −2,N −1 0
  0 0
 AN −1,N −2 AN −1,N −1 0
0 0 0 1

Our matrices will then be implemented using a nested loop as follows:

2
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

A=zeros(N,N); %Matrix initialized with boundary condition partially satisfied

b=zeros(N,1); %Vector with satisfied boundary condition

A(1,1)=1;
A(N,N)=1; %Missing boundary conditions

for i=2:N-1
b(i)=pi*l(i)*ang*U;
for j=2:N-1
if i==j
A(i,j)= 1 - (l(i)/4)*( (1/(g(j-1)-y(i))) - (1/(g(j)-y(i))));
else
A(i,j)= - (l(i)/4)*( (1/(g(j-1)-y(i))) - (1/(g(j)-y(i))));
end
end
end

2.2 Generation of results

After generating our complete matricial system, we solve the system by writing Circ=A\b. The
linear system is now solved. We can then compute the lift and drag coefficients as shown in
equations 5, 6, and 7.
Z +L
2
Cz = Γ(y) dy (5)
S.U∞ −L
Z +L
2
Cx = − 2
w(y)Γ(y) dy (6)
S.U∞ −L
Z +L
dΓ(η)
w(y) = (7)
−L 4π(η − y)
The first equation is implemented using a trapezoidal rule between our Γ vector of values and our
y vector. To use equation 6, we use the expression of the induced velocity w(y) shown in equation
7, and implement it using a simple summation over the successive differences in our circulation
vector. The piece of code for the induced velocity and the two coefficients is illustrated below:

%Lift Coefficient Calculation

Cz = (2/(U*S))*trapz(y,Circ);

%Induced velocity
w=zeros(N,1);
for i=1:N
for j=1:N-1
w(i,1)=w(i,1)+(Circ(j+1)-Circ(j))/(4*pi*(g(j)-y(i)));
end
end

%Drag Coefficient Calculation

prod=diag(Circ)*w;
Cx=(-2/(S*U^2))*trapz(y,prod);

Our code is now ready to calculate the lift and drag coefficients for each wing depending on
the aspect ratio and the angle of attack.

3
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

3. Results
This section of our report will put forward the results generated by our code. Those results drop
into three main categories:
• Performance comparison: between the three wings through numerical results of the lift coef-
ficient and the performance coefficient defined by C
Cd .
L

• Convergence analysis: for the circulation along with the lift and drag coefficients for the three
wings and different aspect ratio.
• Error analysis: of calculations related to the elliptic wing versus the number of discretization
elements and the aspect ratio of the wing.

3.1 Performance Analysis

This part of the results section will discuss the numerical results of our code, with a comparison
between the three types of wings.

3.1.1 Lift Coefficient Comparison

This section illustrates a comparison of the lift coefficient CL between the three wings versus the
angle of attack for aspect ratios of 12, 16, 20, and 24 as shown in figures 2 to 5.

Figure 2: Wings comparison Figure 3: Wings comparison

Lift coefficient VS. angle of attack Lift coefficient VS. angle of attack
Aspect ratio = 12 Aspect ratio = 16

Figure 4: Wings comparison Figure 5: Wings comparison

Lift coefficient VS. angle of attack Lift coefficient VS. angle of attack
Aspect ratio = 20 Aspect ratio = 24

4
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

From figures 2 to 5, we can first notice that the lift coefficient increases linearly with the angle
of attack. For a rectangular wing with an aspect ratio of 12, this coefficient increases from around
0.1 at 1° angle of attack to 1.4 for 15°. For all aspect ratios, the elliptic wing generates the highest
lift, followed by the triangular and the rectangular ones. For a maximum angle of attack of 15° and
an aspect ratio of 20, lift coefficients of approximately 1.5, 1.49, and 1.48 are produced respectively
by the elliptic, triangular, and rectangular wings. Moreover, for all aspect ratios, the coefficients
are close to each other. Finally, all three types of wings produce more lift with an increasing aspect
ratio that implies an increase of the surface area of the wing. For a triangular wing at a 15° angle
of attack, this value increases from around 1.35 at an aspect ratio of 12 to 1.49 at 24.

3.1.2 Performance Ratio Comparison

This section illustrates the comparison of performances between the three wings based on the per-
formance ratio defined as C Cd . Figures 6 to 11 show respectively the evolution of the performance
L

ratio with respect to the angle of attack for each wing, for aspect ratios of 4, 8, 12, 16, 20, and 24.

Figure 6: Wings comparison Figure 7: Wings comparison

Performance ratio VS. angle of attack Performance ratio VS. angle of attack
Aspect ratio = 4 Aspect ratio = 8

Figure 8: Wings comparison Figure 9: Wings comparison

Performance ratio VS. angle of attack Performance ratio VS. angle of attack
Aspect ratio = 12 Aspect ratio = 16

5
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

Figure 10: Wings comparison Figure 11: Wings comparison

Performance ratio VS. angle of attack Performance ratio VS. angle of attack
Aspect ratio = 20 Aspect ratio = 24

First, the performance ratio increases linearly with an increasing angle of attack. Second,the
triangular wing presents the highest performance ratio followed by the rectangular then the elliptic
ones for all aspect ratios except for the first one (λ = 4) where the elliptic wing presents slightly
higher values than the rectangular one. Third, and for the first aspect ratio, the performances of
the three wings are very similar with a percent difference that increases with an increasing aspect
ratio. Finally, we can put forward the fact that the performance of all three wings decrease when
we increase the aspect ratio.

3.2 Convergence Analysis

This part of the results section puts forward the convergence analysis of our code. This process
was done to evaluate our calculations of the convergence along with those of the lift and drag
coefficients. The tests were done with respect to the number of discretization elements used in our
code, for different values of aspect ratios and depending on the wing type. Figure 12 illustrates
how increasing the number of discretization elements makes our code more accurate.

Figure 12: Comparison of Circulation results for N=10 and N=85

Circulation VS. Span coordinate

6
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

In figure 12, the circulation is plotted along the span of the wing with two different values of N.
The result shows clearly how a small discretization step with N=10 makes the results inaccurate.
The circulation distribution is however smoother and thus more realistic for a finer discretization
step with N=85.

3.2.1 Circulation
Effect of wing type on the convergence
This section of the work exposes the convergence results of our code for the calculations of the
circulation. The convergence will be discussed based on the wing type and the aspect ratio. First,
we discuss how our code converges for different wing types. Figures 13, 14, and 15 show the con-
vergence curves of the mean circulation over our profile for the different wings.

Figure 13: Convergence analysis

Mean circulation over profile VS. number of discretization elements
Rectangular wing

Figure 14: Convergence analysis

Mean circulation over profile VS. number of discretization elements
Triangular wing

7
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

Figure 15: Convergence analysis

Mean circulation over profile VS. number of discretization elements
Elliptic wing

As can be seen in the figures, our code converges at the same approximate values of discretization
elements for the circular and rectangular wings (N=90), while it converges for the rectangular and
triangular ones at about the same mean circulation over the profile (circulation = 150). Our code
converges the best for the triangular wing with the least number of discretization elements (N=85).

Effect of aspect ratio on the convergence

As to evaluate the effect of the aspect ratio on the convergence of our code, we analyzed the change
in convergence for the same setup with three different aspect ratios; 4, 8, and 16 as shown in figure
16. The result is that number of discretization elements required for convergence slightly decreases
for an increase in the aspect ratio.

Figure 16: Convergence analysis

Circulation VS. number of discretized elements
Aspect ratios = {4,8,16}

8
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

3.2.2 Lift and Drag Coefficients

Similarly to the circulation convergence analysis, this section puts forward the convergence results
of our code for the calculations of the lift and drag coefficients. The convergence will here also be
discussed based on the wing type and the aspect ratio.

Effect of wing type on the convergence

First, we discuss how our code converges for different wing types. Figures 17, 18, and 19 show the
convergence curves of the lift and drag coefficients for the different wings.

Figure 17: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Rectangular wing

Figure 18: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Triangular wing

9
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

Figure 19: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Elliptic wing

Lift is directly proportional to the surface area of the wing and its respective force is perpen-
dicular to the motion of the fluid. This hydrodynamic force is generated by a suction zone formed
on the top area of the profile. Due to the viscosity of the fluid, friction and turbulence along with
other phenomena are induced. The fluid’s resistance to the motion also generates drag. This factor
is very important in an aerodynamic study where both lift and drag depend on the displaced mass
of fluid. Figures 17, 18, and 19 illustrate the required number of discretization elements for our
code to converge for both the lift and drag coefficients. Convergence happens for the rectangular
wing for both quantities at N=65, and for the triangular one N=58 for the drag coefficient and
at N=65 for the lift one. For the elliptic wing, convergence happens at two very different values;
N=17 for the drag coefficient and N=40 for the lift one. The conclusions we can draw at this stage
are that our code converges the fastest for the drag coefficient. As for wing types, it is the easiest
for it to converge for the elliptic one, followed by the triangular then the rectangular ones which
nearly share the same convergence values.

Effect of aspect ratio on convergence

For the lift and drag coefficients as well, we want to evaluate how a different aspect ratio affects
the convergence of our code. Figures 20 to 23 show the convergence curves of the lift and drag
coefficients for the same setup at aspect ratios of 12, 16, 20, and 24.

Figure 20: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Aspect ratio = 12

10
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

Figure 21: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Aspect ratio = 16

Figure 22: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Aspect ratio = 20

Figure 23: Convergence analysis

Lift and Drag coefficients VS. number of discretization elements
Aspect ratio = 24

11
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

The results depicted in the figures above show that when the aspect ratio increases, the code
converges at a higher number of discretization elements for the drag coefficient and at a lower one
for the lift coefficient. For the first case, the code converges at N=53 for an aspect ratio of 12,
and at N=73 for a ratio of 24. For the second case, convergence happens respectively at N=75
and N=56 for ratios of 12 and 24. The figures also show that for a higher aspect ratio, the plots
of the lift and drag coefficients get closer to each other. The first coefficient is proportional to the
aspect ratio while the second is inversely proportional to it. This means that when the aspect ratio
increases the lift coefficient starts to increase while the drag coefficient decreases causing the lines
to get closer to each other.

3.3 Error Analysis

This part of the report will put forward the evaluation of our code based on a comparison between
numerical results and theoretical ones. For the elliptic wing, we know that the lift and drag
coefficients can be expressed as follows:
dCz 2k∞
= (8)
dα 1 + 2k∞
πλ

Cz2
Cx = (9)
πλ

After implementing equations 8 and 9, we calculated the percent errors in our numerical cal-
culations as function of the number of discretization elements and of the aspect ratio of the wing.
Figures 24 and 25 respectively show the former error evolution patterns for both the lift and drag
coefficients.

Figure 24: Error analysis

Percent error in lift and drag numerical calculations VS. Number of discretization elements
Elliptic wing

From figure 24, one can observe the exponential decrease of the percent error with an increas-
ing number of discretization elements at the same aspect ratio. This quantity reaches a nearly
zero value at around 13 elements for the drag coefficient, and at 35 for the lift one. Our code
therefore converges to the exact results. At low values of N, our code is inaccurate with a percent
error greater than 10% for N < 5 for Cd , and than 15% for CL . The error in calculating the lift
coefficient also exhibits a strange behavior for N < 3, where it spikes to 40% and exponentially
decreases afterwards. Finally, our code shows satisfactory results as it produces numerical results
that are sufficiently accurate for number of discretization elements greater than 50.

12
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

The error produced by our code was also evaluated for different values of he aspect ratio. Figure
25 illustrates the evolution of this quantity with respect to the aspect ratio of the elliptic wing.

Figure 25: Error analysis

Percent error in lift and drag numerical calculations VS. Aspect ratio
Elliptic wing

From figure 25, we can notice that for a fixed value of N, the percent error in the calculation of
the lift coefficient decreases with higher values of the aspect ratios, while it increases for the drag
one. The code still produces acceptable errors for aspect ratios that are not very large; 1.5% for
the lift coefficient and 2.3% for the drag one for an aspect ratio of 25. We can therefore conclude
that for the lift coefficient calculations, it is easier for our code to converge for higher values of the
aspect ratio while the inverse happens for the drag coefficient.

Our code then produces results that are in satisfactory accordance with the theoretical ones
for not very large numbers of discretization elements, and intermediate values of the aspect ratio.

13
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

4. Discussion
In this work, the Prandlt equation was numerically implemented and used to compare the per-
formances of three different wings. This mathematical model indeed proved its usefulness as it
allowed the study of the wings with their geometry as only input. Our code exhibited satisfact-
ory convergence results, as it converged for the circulation calculations at discretization steps no
smaller than 1001
of our wing length, and at ones no smaller than 75 1
for the calculations of the lift
and drag coefficients. Those discretization steps remain very correct and attainable with even low
computing resources. Our study also revealed that for an increase of the wing’s aspect ratio, our
code converges faster for the circulation and lift coefficient calculations, while the inverse happens
for the drag coefficient. In overall, our code exhibits a very satisfactory behavior which proves the
utility of the Prandtl equation, and its suitability to numerical analysis studies.

In addition to a fast convergence, our code produced very accurate results; for the elliptic wing,
the percent error reached nearly zero values at only 13 discretization elements for the calculations
of the drag coefficient, and values around 2% at 35 ones for those of the lift coefficient. This result
in indeed very satisfactory. It also shows that even though our code does not fully converge, it still
produces sufficiently accurate results. When increasing the aspect ratio, our code converges faster
for the lift coefficient calculations but slower for the drag one. Here again, the balance between
the two convergence trends has to be studied in order to find the most suitable aspect ratio for our
overall results.

Our code allowed us a performance comparison of three wings; a rectangular, triangular and
an elliptic one. This comparison was done based on two quantities; the lift coefficient CL and
the performance ratio defined as C Cd for several values of the aspect ratio in order to quantify its
L

influence on the respective performances. Our study revealed that the wing that produced most
lift is the elliptic one, followed by the triangular then the rectangular ones. For an increase in the
aspect ratio, all three of the wings produced more lift as their surface area accordingly increased.
However, no final conclusions can be drawn at this stage as the lift coefficient by itself does not
allow the full evaluation of the wing’s performance. In fact, the corresponding ratio revealed that
the elliptic wing was the least performant of the three even though it produces the highest lift.
The triangular wing is the one that generated the highest performance ratio. Moreover, all three
types of wings produce more lift with an increasing aspect ratio that implies an increase of the
surface area of the wing. Their performance however decreases as the ratio gets higher and higher.

Those results are not fully coherent with the reality; it is known that the elliptic wings are the
most efficient ones as they generate the least drag (aerocorner.com). An increase in the aspect ratio
would also generally mean a higher performance where lift increases higher than the overall drag
and a decrease in induced drag is also present. This shows that our code should be manipulated
in a very careful way and that its results have to be judiciously interpreted. Therefore, we cannot
draw solid conclusions from all of its results.

At this stage, the comparison performed with the code of the Prandtl equation allows drawing
conclusions about the application field of each wing type. It is important to mention that those
results may not be in full accordance with the reality, they only allow us a basis upon which we can
evaluate the general quality of the results that the theory and the code produce. Therefore, and ac-
cording to the results, the triangular wing would likely be used in commercial airplanes for example
where efficiency and fuel consumption are critical variables. The elliptic wing would instead be
used for applications where high lift is needed such as for marine propellers, while the rectangular
wing offers a compromise between the two others, being the simplest one to manufacture.

14
El Iskandarani, Es-sabar Final Project - Lifting Line Theory

5. References
aerocorner.com - About the Author Editorial Team and E. Team, “9 types of aircraft wings in
depth,” Aero Corner, 09-Dec-2020. [Online]. Available: https://aerocorner.com/blog/types-of-
aircraft-wings/

15