Universidad Carlos III de Madrid

Aerodynamics 2
Bachelor in Aerospace Engineering

Report 1: Performance analysis of a finite

Aspect Ratio Wing

Chaoyi Lin Yang 100428553

Rodrigo Martín Ortiz 100432541
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

1 Introduction
One of the fundamental aspects of aerodynamics is the characterization of lift generation on
wing surfaces and its dependence on angle of attack. In this report, we explore lift generation
and its dependence on the angle of attack using a combination of computational techniques, th
and wind tunnel experiments.
The Vortex Lattice Method stands as a well-established approach for assessing the steady
aerodynamic performance of wings. This method discretizes the wing into a grid of panels
and utilizes the principles of potential flow theory to compute lift and other aerodynamic
characteristics. By employing the Vortex Lattice Method, we aim to numerically evaluate the lift
distribution on a specific wing geometry and its sensitivity to variations in the angle of attack.
Complementing our numerical analysis, we conducted wind tunnel experiments to empirically
validate our findings. The wind tunnel testing provides experimental means to quantify the
effects of angle of attack on lift generation, offering a valuable benchmark for comparison with
computational results.
The central objectives of this study encompass the determination of the lift distribution along
the span of the wing and the evaluation of the total lift generated by the wing. Additionally,
we seek to compare our computational results with the wind tunnel experiments, which helps
validate the numerical approach and identify any potential discrepancies between theory and
reality.
The report unfolds with an exploration of the methods employed in our project, offering
justifications for our choices, followed by a detailed presentation of results, and concludes with a
discussion that addresses the specific questions raised in this study.

1.1 Wing geometry

The wing geometry under study in this report is described by Figure 1 and Table 1. The
wing has no geometric or aerodynamic twist.

Figure 1: Wing Planform Geometry

1
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

ctip /croot 0.3

b/croot 3.5
ctip /croot 1.2
Λ 32◦
AR 6.01

Table 1: Wing Planform Geometry

2 Methods
2.1 Wind tunnel experiment
A model wing with a span of 12 cm was created and placed into a set-up bench. Some wholes
were drilled in the wing holder, such that the wing can be set to different angles of attack. The
wing and the set-up bench are placed in a scale, which is then put in the measuring point of the
wind tunnel. Then, the weight difference is measured for different angles of attack, taking special
care of making sure the measurement is made once the flow around the wing has stabilized.
Having these settings, the following measures are got on the wind tunnel in three different
angles of attack, where for each angles it is written down the maximum and minimum value to
have a mean value of it:

Velocity [m/s] Mass [gr]

AoA [degrees]
Max Min Max Min
5 9.91 9.85 5 3.8
10 9.83 9.54 8 7.9
15 9.76 9.68 10.7 10.5

Table 2: Experimental measurements

Now, in order to get an estimation of CL (α) with the experimental data, few things need to
be consider:
1. Equating lift to the weight of the airplane
2. The computation on every formula, some treatment in the data for the velocity and mass
are need to be consider:
xmax + xmin xmax − xmin
x̄ = and ∆x =
2 2

3. And the equation for the lift coefficient are the following:
2m̄g
C¯L =
ρŪ 2 S

4. Also, since it is a experimental data, uncertainty also needs to be considered due to

vibration of the machine, the flow surrounding the experiment, etc. So the ∆CL needs to
be computed as following:

∂CL ∂CL 2g 4m̄g

∆CL = ∆m + ∆U = 2
∆m + ∆U
∂m ∂U ρŪ S ρŪ 3 S

2
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

kg
5. In the previous formula, density of the air used is ρ = 1.293 and the wing area is
m
b2
S= = 23, 96cm2
AR
With all the necessary equations, now it is proceed to do the computation of those values
which gives the following results:

α Ū ∆U m̄ ∆m C¯L ∆CL
5º 9.88 0.03 4.4 0.6 0.285 0.0406
10º 9.6856 0.145 7.95 0.05 0.536 0.01942
15º 9.72 0.04 10.6 0.1 0.709 0.0125

Table 3: Resulting data

Now, having these results, the slope of the two variables can be obtained by applying a linear
regression resulting in the following figure:

Figure 2: Graph of CL vs AoA

dCL
Where the slope in this case is = 0.04246; notice that this value of slope is for AoA
dα
when it is in degrees, and so in the following sections, the computation of the slope need to be
consider this condition or transform it for AoA in radians.

2.2 3D panel method in XFLR5

XFLR5 is a software tool for the analysis, in this case, of a wing. One of the numerical
method to get the aerodynamics loads, its coefficients or the pressure distribution is the 3D
panel method. This numerical method solves the attached flow around the lifting surface using
potential flow assumption; where the geometry of the of the lifting surface is replaced by panels
of singularities that are solutions to the Laplace’s equation.
For the analysis of the wing, the number of panels is crucial to get an accurate lift coefficient
graph, the higher the number of panels the more precise; however due to the size of the wing, an
excessive number of panels XFLR5 could stuck and unable to perform the numerical method,

3
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

and the improvement on the precision of the values is minimum.

That said, the number of panels selected are 20, 40, 80 and 160 panels respectively and the
results are the following:

Figure 3: Graph of CL vs AoA of XFLR5 results

And, by applying the slope formula:

dCL CL2 − CL1
=
dα α2 − α1
With the following results:

dCL
Number of panels
dα
20 0.0771
40 0.0759
80 0.0761
160 0.0761

Table 4: Slope of each number of panels

Which it can be seen how the precision to get the slope is nearly negligible when the panels
increased from 80 to 160 panels.

2.3 Vortex Lattice Method

The following section describes the thought process and challenges during our particular
implementation of the Vortex Lattice Method.
The Vortex Lattice Method (VLM) is a fundamental computational technique in aerodynamics
used for analyzing and predicting the aerodynamic performance of wings and aircraft. It offers a
pragmatic means to assess lift, drag, and other aerodynamic parameters, primarily in subsonic
and low-speed flows.

4
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

However, VLM shows some underlying limitations. It is most applicable in subsonic, incom-
pressible flow regimes and may not capture all complex flow phenomena accurately. Moreover,
the effect of the thickness of the airfoil profile is not accounted. In addition, VLM is part of
potential flow theory, which has no ability to compute viscous drag due to its assumptions.
Therefore, VLM is often used as an initial design and analysis tool, with more advanced methods
required for detailed analysis.
A brief summary of the theoretical background on the Vortex Lattice Method feels necessary
in order to properly explain the process.
Under potential flow theory, the velocity vector field can be expressed as:
⃗ = V⃗∞ + ∇ϕ
V (1)

The perturbation velocity of a given panel can be expressed as the sum of the contributions of
each of the panels.
N
X
∇ϕi = ωij Γj (2)
j=1

Where ωij is the corresponding term from the Aerodynamic Coefficient Matrix, whose contents
store the geometrical relations between panels i and j. By applying flow tangency at the panels
(Neumann BC), a system of equations for the circulation of each of the panels can be obtained:

[AIC][Γ] = [b] (3)

Aerodynamic properties can be obtained once the circulation for all panels is known.

2.3.1 Panel Discretization

At the core of the VLM is the concept of discretizing the aerodynamic surface into a grid of
panels. Each panel represents a finite section of the geometry and serves as a computational
entity in the method. Panels are characterized by control points, which act as evaluation points
for vortex computations, and vertices that define the panel’s shape.
Regarding the implementation for this wing geometry, the team faced a challenge: in order
for all panels to be quadrilateral, the kink y-section needs to be a boundary between two panels.
The final solution was to perform to separate meshes of the right wing, one from wing-root to
kink, one from kink to wing-tip. Then, the x and y coordinates of the vertices and control points
are assembled into a single array. Then, apply symmetry to incorporate the left wing into the
mesh.
Some comments shall be performed regarding the choice of the horseshoe vortex and evaluation
point locations within the panels. Following the convention described in [1] (which comes from
results based on Thin Airfoil Theory), the producer spanwise vortex filament should be located
at c/4 of the panel chord, whereas the evaluation/control points are located at 3c/4.
The following figure contains the position of the vertices of the horseshoe panels (dots) and the
collocation points (crosses). The presented mesh was performed using N x = 5, N yroot−kink = 6
and N ykink−tip = 8.

5
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

Figure 4: Wing Mesh

Figure 5: VLM Model Visualization

Another key consideration is where to place the panels in the z direction. When considering
lifting surface methods, two choices are possible:
• z = 0 surface (VLM1)
• Camber surface (VLM2)
In the case of a wing NACA 0012, both options are equal (symmetric profile). During the
developing stages of the algorithm, the team decided it was worth the effort to generalize the
code to allow placing the panels on a z =
̸ 0 camber surface. Even though it may be outside of

6
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

the scope of the homework, some guidelines and comments for the general implementation are
included in the report.

2.3.2 Vortex Representation

In the VLM, aerodynamic effects are approximated using horseshoe vortices. Each panel is
associated with one or more of these vortices, symbolizing the local flow interactions. The vortices
induce velocity at the control points of neighboring panels, allowing the method to approximate
the complex interplay of aerodynamic forces. This is achieved through the computation of the
Aerodynamic Influence Coefficient Matrix (AIC).

2.3.3 Influence Coefficients and obtaining the Circulation Distribution

A key aspect of the VLM is the determination of influence coefficients. These coefficients
quantify the effect of one panel’s vortices on the velocity field at the control points of other
panels. Influence coefficients are obtained through mathematical formulations that consider
the geometric relationships between panels. The component aij of the Aerodynamic Coefficient
Matrix represents the downwash velocity that panel j produces on panel i. The velocity produced
by a horseshoe vortex on a point in space can be obtained from Biot-Savart law. However, a
function vring.m was provided, which performs these calculations, so that was implemented for
simplicity.
Note that the aerodynamic influence of a panel on another is only a function of their
geometrical relationship. This implies that influence is calculated using a unitary circulation
value for all the vortex elements during the execution of vring.m.
In order to fully define the system of equations, the independent terms [b] need to be expressed.
These correspond to the normal free stream velocity at each panel. Assuming no side-slip and
small angles of attack:
bi = −U∞ [1, 0, α] ∗ n⃗i (4)
Regarding the implementation of the mesh on the z = 0 surface, all independent terms become
bi = −U∞ ∗ α.
Finally, the linear system of equations can be solved, which yields the appropriate circulation
each panel needs in order to satisfy boundary conditions.

2.3.4 Lift coefficient calculation

Recall that the circulation of the panels is only a function of the free-stream velocity and the
angle of attack. In order to obtain a suitable model for the whole aircraft velocity and angle
of attack envelopes, U∞ and α values were set to one in the code. As a result, the obtained
circulation values from the linear system of equations are (Ki are constants):

Γi = Ki U∞ α (5)
The obtained circulation distribution of the wing under study is: The lift distribution is found
by use of Kutta-Joukowski Theorem:
Li = ρU∞ Γi (6)

The total lift is obtained by integrating the lift values on each panel over the wing surface.
The sum of the circulations of every column of panels (x-direction) is calculated, so that the
result is a circulation distribution as a function of the wingspan only. Then:

7
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

Ny
X
L = ρU∞ Γ(yi )∆y (7)
i=1

Lastly, the lift coefficient can be obtained by:

L
CL = (8)
qS

2.3.5 Results of VLM method

Figure 6: Wing Circulation distribution Γ(x, y)

8
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

Figure 7: Wing Circulation distribution Γ(y)

The results capture some interesting phenomena. Notice the valley in y = 0. This is a deeply
understood effect of the Vortex Lattice Method in swept wings, arising from the fact that the
vortex lines are intersecting at an angle. The effect of the orientation of these vortex lines causes
the lift distribution to decrease at the intersection point.

2.4 Analytical expression from the LLT for an elliptical distribution

The Prandtl lifting line theory is a mathematical model which predicts the lift distribution
by avoiding the computation on the surfaces. Instead, it will combine the Biot-Savart law to
explain the behavior of semi-infinite line vortexes, replacing the surface to the elementary flow,
in this case the vortex and the Kutta-Joukowski theorem, which relates the circulation of the
vortex to the lift.
dCL
With that premise, the analytical expression for for an elliptical distribution (τ̃ = 0 and
dα
dcL
= 2π) is the following:
dα
dCL dcL/dα
= dcL/dα
dα
1+
πAR
dCL
So, at the end, having all the data to compute the slope which is: = 4.7140. Notice that
dα
this value in the slope is for different AoA in radians, so in order to have it in degrees, it is
π dCL
necessary to perform a change which by multiplying to the value, giving = 0.0823 as
180 dα
value.

9
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

Figure 8: CL vs α for an elliptical lift distribution

3 Results of different methods

Having discussing the different method in the report, the slope of the lift coefficient on the
different method are the following:

Exp XFLR5 (160 panels) VLM LLT

dCL/dα 0.04246 0.0761 0.0764 0.0823

Table 5: Slope of the lift coefficient in the incompressible regime

Graphically, it shows the following:

Figure 9: Comparison of the different methods

10
Report 1: Performance analysis of a finite Aspect Ratio Wing Report I

4 Conclusion
In the figure 9 it can be seen how close the numerical methods VLM and 3D panel they are,
that is, both methods are sufficiently accurate for the wing analysis. This is due to the use of
NACA 0012 airfoil. If the wing were to contain a non-symmetric profile, the difference between
VLM and Panel Method would be much greater. Regarding the number of panels used in the
computation, it can be seen that when the panels increase, the lift coefficient slope also increases.
On the other hand, using the analytical expression of LLT, there is quite a difference with
respect to the numerical methods, because lift distribution of the wing is not elliptical as it can
be seen in figure 6, in which the lift distribution is similar as the circulation distribution from
the relation previously discussed. In addition, LLT is known to be more accurate as the aspect
ratio of the wing increases. As for the studied wing, the AR (6.01) is not large.
The only result which is really different from other methods is the experimental one, which
lift coefficient slope is almost is half of the rest of the methods; this difference could be because
on the errors made during the experiment such as; manual measure of the velocity, vibration of
compressor that create the airflow, imperfections in the wing model, accuracy of the weighing
scales, etc. Since a lot of errors could have been made during the experiment, it can be concluded
that the reliability in order to estimate the aerodynamics performances is very low or even it
cannot rely on this experiment, especially in the conditions that it was performed.
Even though the experiment was not carried out in a sufficiently controlled environment, one
needs to keep in mind all the limitations of the numerical methods that were discussed earlier,
like non-viscous flow, neglecting the effect of thickness (VLM), small angles of attack assumption,
etc... All in all, the fact that the numerical and theoretical methods yield larger lift coefficient
slopes than the experiment, seems to be in accordance with all the conservative assumptions
that were required in order to create the models.

5 Bibliography

References
[1] John D. Anderson. Fundamentals of Aerodynamics. McGraw-Hill, New York, NY, 2nd edition,
1991.

11