Delft University of Technology

Experimental Simulations
AE4115

Group 15: Post-Test Report

MSc Aerospace Engineering

Aerodynamics Track

Authors:
Mohith Mohan Kanala (5306515)
Nikhil Joseph Jose (5321913)
Vassilis Karanikolas (5386128)
Vinicius Santos Costa (4473779)

April 19, 2021

 CONTENTS

Contents
1 Introduction 1
1.1 Scaling parameters and key variables . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Scaling Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Performance Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Methodology 3
2.1 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Test setup limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Test matrix design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3.1 Advance ratio (J) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.2 Freestream velocity (V∞ ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.3 Angle of attack (α) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.4 Propeller configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.5 Elevator deflection (δe ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.6 Other aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Test matrix limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Corrections 8
3.1 Blockage Corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.1 Solid Body blockage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.1.2 Wake blockage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.1.3 Propeller slipstream blockage . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1.4 Total blockage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2 Lift Interference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Downwash Correction at the Tailplane . . . . . . . . . . . . . . . . . . . . . . . . 13
3.4 Lift and drag correction for propeller thrust . . . . . . . . . . . . . . . . . . . . . 14
3.5 Implementation of Corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.6 Corrected Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 Results 16
4.1 Longitudinal Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.2 Control Effectiveness and Stability Margin . . . . . . . . . . . . . . . . . . . . . . 17
4.3 Reynolds number effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.1 Re effect from pressure taps . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.4 Acoustic Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.4.1 Background noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.4.2 Prop-off and Prop-on comparison. . . . . . . . . . . . . . . . . . . . . . . . 22
4.4.3 Effect of J . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.4.4 Comparisson between configurations . . . . . . . . . . . . . . . . . . . . . 23

5 Conclusion 23

Appendices i

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 i

 CONTENTS

A Full test matrix i

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 ii

 LIST OF TABLES

List of Figures
1 Schematic drawing of the LTT wind tunnel. Image taken from [1]. . . . . . . . . . 4
2 Schematic drawing aircraft model used during this experiment. Taken from [1].
Refer to this document for the exact dimensions of the model. . . . . . . . . . . . 4
3 Graphs for determining factors K1 , K3 and τ1 , in order to calculate the solid block-
age factor for every component of the model. [2] . . . . . . . . . . . . . . . . . . . 9
4 Drag polar graph for drag analysis for a lifting wing. [2] . . . . . . . . . . . . . . . 11
5 Lift Interference Correction [2] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
6 Comparison of Corrected and Uncorrected values for Tail Off measurements at
V∞ = 40 m/s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7 Comparison of Corrected and Uncorrected values for Prop-On measurements at
V∞ = 40 m/s and J = 1.85 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
8 Longitudinal Stability for different Propeller Rotation Configurations . . . . . . . 17
9 Elevator Effectiveness at V∞ = 40 m/s and J = 1.6 . . . . . . . . . . . . . . . . . 17
10 Elevator Effectiveness at V∞ = 40 m/s and J = 1.85 . . . . . . . . . . . . . . . . . 17
11 Elevator Effectiveness at V∞ = 40 m/s and J = 2.1 . . . . . . . . . . . . . . . . . 18
12 (a) Lift coefficient, (b) Drag coefficient and (c) Pitching moment coefficient, for
two values of free-stream velocity and 3 advance ratios. Propeller configuration:
Outboard-up. Elevator deflection: 8 deg . . . . . . . . . . . . . . . . . . . . . . . 19
13 Re effects on the pressure tap on the side of the fuselage . . . . . . . . . . . . . . 20
14 Background noise generated by the six microphones for the drag measurement sets
at different elevator [Link] in microphone 3 is applicable to all micro-
phones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
15 Results for mic 2 and mic 6 comparing different values J using for the same test
setup using the BPF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
16 Mic 6 results between different configuration at 1.6 J and 0 degrees elevator deflec-
tion. Left plots are at 40 m/s, right plots at 30 m/s. Top plots give results at 7◦ α
while the bottom plots give those . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

List of Tables
1 Component volume, shape factors τ1 , K1 or K3 , and solid blockage factor for every
component of the model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Wake blockage factors wb [%] for wing, tail and total for two values of velocities . 12
3 The Lift Interference Corrections . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Correction to pitching moment due to downwash at the tailplane. . . . . . . . . . 14
5 Full test matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 iii

 1 INTRODUCTION

1 Introduction
Wind tunnel experiments have been used to determine aerodynamic characteristics of aircraft since the first powered
flight by the Wright brothers. Their small home made wind tunnel proved to be a key ingredient in their success of
maintaining powered flight. Nowadays, wind tunnels come in different shapes and sizes each intended for different
aerodynamic measurements. They remain a very important tool to gauge aerodynamic characteristics, and offer
many benefits over their alternatives. Firstly, wind tunnel experiments are cheaper and safer as they use small
scale models than performing experiments with full scale models. Along with this they offer many benefits over
CFD simulations, which are very limited by the available computational power and model complexity. Although
there are many advantages of wind tunnel experimentation, there are also some limitations and disadvantages. The
limitations mainly arise due to two factors, namely similarity and scaling.

Similarity and scaling deal with the question on how to get the conditions of the wind tunnel experiment to be as
representative as possible of the conditions during free flight. Due to the fact that the model used in wind tunnels is
often smaller than life sized models, forces and flow phenomena behave differently than in free flight. Furthermore
the flow conditions in the wind tunnel themselves often differ from those of "real" flight, due to for example the pres-
ence of wind tunnel walls. In order to ensure that wind tunnel experiments give accurate and representative results,
these effects have to be taken into account, something that has been done to analyse wind tunnel data in this report.

This report is the post-test report of the AE4115 "Experimental simulations" course, where a set of 4 MSC aerody-
namics students performed a wind tunnel experiment on a aircraft model with tail mounted propellers. The goal of
this wind tunnel experiment was to primarily "quantify the effects of the propeller rotation configuration on aircraft
performance in terms of elevator effectiveness and control power, longitudinal stability and trimmed lift-over-drag
ratios." For this measurement challenge, measurements were performed using three different propeller configura-
tions, namely inboard-up and outboard-up counter rotating propellers and co-rotating propellers. To analyse the
effects of the propeller configurations on these aspects, force and moment measurements of the model were taken
using an external balance (see section 2 for a detailed test setup). In order to compare the results of the model to
a full scale aircraft all the forces and moments were non-dimensionalized into coefficients. The relevant set of key
variables and scaling parameters used during this experiment can be found in Figure 8.

The longitudinal stability is assessed using the pitching moment coefficient of the aircraft with the different pro-
peller configurations and variations of the elevator deflection angle as this is the most important parameter of the
longitudinal stability. The elevator effectiveness in the different configurations and control power can be assessed by
looking at how much lift is generated by the tail under the different conditions (of varying propeller configuration
and elevator deflection). To determine the lift over drag ratios in trimmed conditions, a c.o.m. is defined on the
aircraft model. Lift, pitching moment and drag curves are then setup using the measurement data to obtain the
required elevator deflection and lift over drag ratios to ensure aircraft trim. More information on what points were
chosen to asses these values and the reasoning for those points can be found in section 2.

Secondarily the goal of the experiment was to asses the effect of the boundary corrections and Reynolds number on
the performance parameters. The first is done by analysing the differences between the corrected and uncorrected
data in section 3. All the corrections that were used and their reasoning are furthermore listed here. For the effect
of the Reynolds number all measurements were repeated at two different wind velocities, namely 30 and 40 m/s
to see how this affects the measured coefficients and other data. The Reynolds number discussion can be found in
Figure 8. Finally the experiment assesses the effect of the operating conditions on propeller noise emission, by use
of 6 body mounted microphones. The discussion regarding this can be found in Figure 8.

1.1 Scaling parameters and key variables

Below the relevant performance indicators are identified. Subsequently, by the use of dimensional analysis in
Figure 8, we obtain a reduced set of dimensionless governing parameters which enable scaling. The method used

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 1

 1 INTRODUCTION

for dimensional analysis is the Buckingham’s Pi-theorem. These ratios are the similarity parameters that we will
define later on. The parameters that are relevant to the experiment are:
1. Pitching moment M 7. Dynamic pressure q 13. Angle of attack α
2. Lift FL 8. Dynamic viscosity µ 14. Elevator deflection angle δe
3. Drag FD 9. Wing surface area S 15. Speed of sound a
4. Propeller thrust T 10. Mean aerodynamic chord c̄ 16. Pressure fluctuation prms
5. Freestream velocity V∞ 11. Propeller rotation rate n 17. Noise Frequency f
6. Air density ρ 12. Propeller diameter D 18. Number of propeller blades NB

1.1.1 Scaling Parameters

First, the base units have to be identified. For mechanical phenomena, the base units are expressed based on the
SI system, using the unit of length (L) m, unit of mass (M ) kg and the unit of time (T ) s. The relevant variables
can then be defined as given below
• M = M L2 T −2 • V∞ = LT −1 • µ = M L−1 T −1 • n, f = T −1
• FL , FD = M LT −2 • ρ = M L−3 • S = L2 • D=L
• T = M LT −2 • q = M L−1 T −2 • c̄ = L • a = LT −1

Using the FL , FD , T and M as the dependent parameter, a set of key dimensionless variables can be defined as
follows below. Alongside this other scaling parameters are defined.
1. Pitching moment coefficient: Using the parameters ρ, V∞ , S and c̄, the pitching moment coefficient Cm
is derived as seen in Equation 1. This is the main parameter to determine the longitudinal stability.
2. Lift and Drag scaling: By using the parameters ρ, V∞ , S and c̄, the lift and drag coefficients CL and CD
can be derived as seen in Equation 2 and Equation 3. Taking the ratios of these two furthermore give the
Lift-over-drag ratio as seen in Equation 4.
M FL FD FL
Cm = 1 2 (1) CL = 1 2 (2) CD = 1 2 (3) (4)
2 ρV ∞ Sc̄ 2 ρV ∞ S 2 ρV ∞ S FD

3. Rotor and thrust scaling In a similar manner as for the lift and drag coefficients, by using the parameters
ρ, V∞ and propeller area πD2 /4 instead of the wing surface area, the dimensionless thrust coefficient TC is
derived as can be seen in Equation 5. Another method to non-dimensionalize the thrust can be done by using
the "Isolated rotor thrust coefficient". This is scaled using the parameters ρ, D and n as seen in Equation 6.
Both parameters can be used as means of comparing the thrust between configurations. In order to have
similar rotor conditions for different air-speeds the advance ratio can be defined using freestream velocity V∞
and the propeller tip speed n · D as seen in Equation 7. This parameter is kept constant between different
freestream velocities in order to maintain similarities between test conditions.
T T
TC = 1 (5) CT = (6) J=
V∞
(7)
ρV 2 πD 2
2 ∞ 4 ρn2 D4 nD

4. Reynolds and Mach numbers: The Reynolds and Mach numbers are two dimensionless parameters used
to gauge similarity between flow conditions during the wind tunnel compared to free flight. Due to the model
being scaled down it is often impossible to achieve similarity in both of these, something that should be taken
into account when scaling the results from wind tunnel data to free flight. The Reynolds number can be seen
as the ratio between inertial and viscous forces, while the Mach number is the ratio between V∞ and a. The
equations for both can be seen in Equation 8 and Equation 9 respectively.
ρV∞ c̄ V∞
Re = (8) Ma = (9)
µ a

5. Noise generation scaling:The main noise generating mechanism of interest to compare the propeller noise
emissions between operating conditions is the thrust. In order to relate the sound level and the thrust, the
pressure ratio is defined as the ratio of the acoustic pressure to the pressure jump over the propeller disc. To
express it, the parameters prms , T and D are used are used in Equation 10. The frequency at which the noise
is generated can furthermore be scaled using the "blade pass frequency (BPF)". This allows for comparison
between propellers rotating at different frequencies. The BPF obtained using the parameters NB , n and f
AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 2
 2 METHODOLOGY

prms D2 f
Π= (10) BP F = (11)
T nNB

1.1.2 Performance Parameters

Using the previously obtained dimensionless scaling parameters further performance parameters can be defined.
These will be used to used to analyse the results and achieve the measurement challenge, alongside the scaling
parameters.
1. Longitudinal Stability: In order to quantify the longitudinal stability of an aircraft, the change in pitching
moment coefficient is measured with respect to the change in angle of attack. This is expressed as:
∂Cm
Cmα = (12)
∂α
2. Elevator effectiveness and control power: To measure the effectiveness and control power of the aircraft’s
elevators, the changes in lift coefficient and pitching moment coefficient are measured with respect to the
change in elevator deflection angle. These alongside with the Cmα can then be used to determine the trimmed
lift over drag ratios.
∂CL ∂Cm
CLδ = (13) Cmδ = (14)
∂δe ∂δe

3. Propeller Noise: Along the pressure ratio expressed in Equation 10, the noise generated by the propellers
can be expressed in terms of the the Sound Pressure Level, calculated as:
prms
SP L = 20 log10 (15)
pref

where pref = 2 × 10−5 Pa. While it is a non-dimensional parameter (expressed in decibels), it cannot be used
as a scaling parameter because the noise is scaled with respect to the average threshold of human hearing and
is unrelated to the physical phenomenon that generates noise. Nonetheless, SPL measurements taken under
different conditions of flight can be used to understand the relative change in noise produced when switching
from one configuration to another.

2 Methodology
This section discusses methodology that was used to perform the experiment. First the experimental setup that
was used in the Low Turbulence Tunnel of the TU delft is given. After this the limitations of this test setup are
briefly discusses. Then the chosen test conditions and test matrix are discussed. Finally a discussion is presented
on the limitations of the chosen test matrix.

2.1 Experimental setup

All measurements were performed using TU Delft’s "Low Turbulence Tunnel" (LTT). The LTT is a closed-return,
closed-wall test section wind tunnel featuring an octagonal 1.80m by 1.25m test section, with 300 mm x 300mm
diagonal sections at the corners. Due to a large contraction ratio the turbulence in the tunnel is under 0.1%
between 2 and 5000 Hz. This provides excellent free stream flow conditions. The LTT is capable of providing a
free stream velocity up to 120 m/s, however due to model limitations the maximum allowed freestream velocity
for this experiment was limited to 40 m/s. The test section itself is slightly divergent, removing the need for any
buoyancy corrections. A disadvantage of this tunnel is that acoustic treatment is not applied, which allows for
acoustic reflections by the wind tunnel walls. A schematic of the wind tunnel can be seen in Figure 1.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 3

 2 METHODOLOGY

Figure 1: Schematic drawing of the LTT wind tunnel. Image taken from [1].

The model that was used for this experiment is an aircraft model with a fuselage, main wing, removable tailplane
and integrated nacelle into the tail-plane with a removable propeller. The model furthermore features attachable
flaps to the main wing which can simulate aelerons and plain flaps on the horizontal- and vertical-tailplane to
simulate elevators and rudders respectively. Transition strips are placed on the upper and lower side of the main
wing and tailplane to force boundary layer transition. Finally the propeller is a 6 bladed propeller that can be set
in any configuration. A schematic overview of the model can be seen in Figure 2. For the exact model dimensions
see [1].

Figure 2: Schematic drawing aircraft model used during this experiment. Taken from [1]. Refer to this document
for the exact dimensions of the model.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 4

 2 METHODOLOGY

The model is mounted upside down in the wind tunnel, attached to an external balance by three support struts.
Two struts are attached to the main wing while a final support strut is attached to the rear end of the fuselage.
The balance is located on a turntable which allows setting a sideslip angle on the model. Furthermore the angle
of attack of the model can be set by moving the aft support strut. The balance itself is a 6 part balance that can
be used to measure the aerodynamic loads and moments in all 3 directions. By taking wind-off zero measurements
the non aerodynamic loading on the balance can be extracted from the wind on measurements. The propellers
themselves are however not instrumented, therefore propeller force measurements are obtained from the external
balance data.

Alongside the external balance six static pressure taps and six microphones are mounted on the rear end of the
fuselage. These are mounted at the same positions on opposite sides of the fuselage. The static pressure taps
are mounted on the starboard side of the fuselage, while the microphones are flush mounted on the port side of
the fuselage. The pressure taps measure the static pressure simultaneously using an electronic pressure scanner,
along with the static and total pressure upstream of the test section. The microphone are capable of recording a
maximum sound pressure level of 130.5 db in a frequency range between 10 to 15 kHz.

2.2 Test setup limitations

The test setup of the model gives rise to many limitations. A set of limitations is given below. Not mentioned are
the limitations of the test setup that are corrected for. These can be found in section 3.
1. Acoustics measurement limitations: Due to the wind tunnel not being acoustically treated, the quality
and scalability of the obtained microphone data is significantly lowers, due to acoustic reflections. Furthermore
noise sources have a lot of directivity. The acoustic reflections disrupt that directivity. Limitations also arise
due to the measurement instruments themselves. In order to properly measure and determine the directivity
of a noise generating mechanism, a microphone array is required, to determine the source of a noise using
beamforming. However, only six microphones are available which are mounted all mounted on the fuselage.
This only allows limited analysis of the directivity in one axis. Furthermore, the microphones are flush
mounted on the fuselage. These microphones are subject to pressure fluctuations due to the boundary layer
of the fuselage, which gives a strong source of acoustic noise, lowering the SNR.
2. Re and Mach similarity: Due to the limitations of the wind tunnel and the test model neither Re nor
Mach similarity can be fully achieved. At the maximum allowed 40 m/s the mach number might be similar
to that of low speed (e.g. landing or take off) conditions during free flight. This does not allow for analysis
for different flight conditions with higher Mach numbers. Furthermore due to the fact that conditions in the
wind tunnel are the same as the atmospheric conditions outside of it, and due to the scaling of the model the
Reynolds number of the experiment is much lower than what it would be in reality. This effect is somewhat
accounted for by forcing boundary layer transition using the transition strips on the model. However the
boundary layer is relatively much bigger due to the lower Re. Furthermore other effects that scale with Re
will behave differently in free flight compared to the wind tunnel conditions.

2.3 Test matrix design

Many approaches can be taken to achieve the measurement challenge described in section 1. The design of the test
matrix was mostly limited by the available time to performed the measurement, which was 3 hours. Alongside this
changing the propeller configuration and control surface deflections takes a significant amount of time and therefore
can not be done too often. In order to use the available time as efficiently as possible an approach was chosen that
allows for a full 2-factor (with the exception of the J and δe , which are taken as 3-factor) factorial experimental
design over the linear lift curve region. It was chosen to investigate only the linear region as quantifying the full
effects in the non-linear region would require a lot more data, for which there is no time. Furthermore the flow
Re number has a big impact on the flow phenomena in the nonlinear region. Due to it not being possible to have
similarity between tunnel and free flight Re number, the results would not be representative.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 5

 2 METHODOLOGY

The full test matrix can be seen in Appendix A, along with a shortened version and a summary of the parameters
that were varied. The parameters that were varied during the experiment were α, J, V∞ , δe and the propeller config.
Roll, yaw and rudder deflections were not varied as the measurement challenge at hand is only to investigate the
longitudinal stability properties, and due to the limited time these variables are less important than the first 5.
Follow up studies might want to account for the effect of these variables on the stability under the different propeller
configurations. Furthermore the aeleron deflection was not varied. Since the propeller is mounted on the tailplane,
the effect of the aeleron in the different propeller configurations is assumed to be small. A follow up study can
investigate whether this is really the case. In following "sub-subsections" the reasons why each measurement point
is chosen will be given.

2.3.1 Advance ratio (J)

For the advance ratio 3 points were chosen to discern between the effects of the propeller configurations. The effect
was assumed to linearly vary in strength with J, making it so that 2 points would be sufficient. However, due to
the uncertainty of this assumption a third point was chosen. The reasoning for each point is given below.
• J = 1.6 is taken due to the fact it is the lowest value possible (highest thrust) for 40 m/s with the current
test setup. For this point, the effect should be most pronounced.
• Steps of 0.25 are chosen to have the last point (2.1) be on the point of maximum efficiency of the propeller.
• J is kept constant between 30 m/s and 40 m/s for comparison. Constant J was taken instead of constant
RPS since J is a scaling parameter. This can be used to compare results between freestream velocities.

2.3.2 Freestream velocity (V∞ )

For the V∞ the conditions that were deemed to be most important was to have velocities that are possible during
flight (for Mach similarity) and to have two points to study Re effects. This lead to the following reasoning.
• 40 m/s was chosen as this is the highest available V∞ , giving some Mach similarity, and as close as possible
Re number. This is however still very far off what the Re number would be during free flight.
• 30 m/s is taken as a secondary point to study Re effects. This gave enough difference in Re is enough to
quantify some effects, although since the measurements were taken exclusively in the linear region the effects
are small. Values lower than 30 m/s were deemed to be too low compared to free flight. While 30 m/s is
likely still possible although likely only during approach and take-off. Furthermore values between 30 m/s
and 40 m/s were considered too close in terms of Re numbers to quantify any effects.

2.3.3 Angle of attack (α)

As mentioned before, only the linear lift region was investigated. The reasoning for each point follows below.
• The linear lift curve region was found to be between -4◦ and 9-10◦ using tail-off reference data.
• At α higher than 8◦ , the propeller thrust gets affected significantly by the wake of the wing, thereby increasing
thrust. Below this value, other effects due to the wind tunnel boundaries cancel out this increase in thrust.
Therefore a potential maximum of 8 was taken. However, due to the uncertainty of what the exact effect
on the tail and thrust would be and due to the effect possibly being stronger depending on the propeller
configuration we took the upper limit at 7 degrees to ensure linearity.
• A second point as far away as possible from the first point is most optimal due to the reduction in error in
the resulting curve (as deviations at the two end points lead to less overall error in the curve). Therefore,
it was chosen to take -2◦ as the second point. At around -3◦ , the wing has zero lift. Keeping the wing lift
consistent at a positive value to ensure that the tail always has a downwash was deemed to be useful to
minimize possible errors occurring due to one point having an upwash and the other a downwash.
• These points allow the assumption that the thrust is unaffected by the main wing between -2◦ and 7◦ α.

2.3.4 Propeller configuration

It was chosen to only investigate the two counter rotating configurations due to the aspects mentioned below.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 6

 2 METHODOLOGY

• The effect of co-rotating propellers on the longitudinal stability was reasoned to lie somewhere between that
of the two counter rotating configurations. It is expected that the biggest difference between configurations
arises due to the propeller slipstream affecting the local angle of attack at the elevator. Since a co-rotating
configuration would have an increase in local angle of attack on the elevator (due to inboard up) on one side
and a decrease in the other (due to outboard up), measuring both counter rotating configurations also gives
an approximation of what the co-rotating effect is. One point at 0◦ δe deflection with co-rotating propellers
was taken to verify this assumption.
• The co-rotating configuration induces asymmetric loading, which would be hard to account for as asymmetric
control surface deflections would be needed. This would affect the lift over drag curves in a way that could
not be reproduced as the effect of these deflections would need to be taken into account. Since only the
longitudinal stability is of interest for our measurement challenge, it is best to avoid any asymmetric loading.
• As mentioned before, the time available is very limiting. If the co-rotating configuration were to be imple-
mented, more changes in elevator deflection would be needed, which in turn would again cost more time.
Excluding it makes more time available to study the other variables.

2.3.5 Elevator deflection (δe )

For the elevator deflection it was chosen to study 3 angles, namely 8◦ , 0◦ and -8◦ . Initially, the idea was to only
study 2 angles. However, to produce the trim curve, negative elevator deflection angles might be necessary, where
the negative angles might not follow the same linear relation as the positive angles. To verify this, one negative
and one positive angle were chosen along with the neutral state. Below the reasoning for each point can be found:
• It was deemed that δe is the most important parameter (along with propeller configuration and J) for the
longitudinal stability of the aircraft. This is due to the fact that the tail acts as a stabilizer for the moments
generated by the wing. Since the propellers are placed on the tail, there is a strong interaction between the
overall pitching moment of the aircraft and the thrust settings (power and configuration). Thus, varying the
δe to see how the thrust settings affect the pitching moment was of utmost importance.
• In order to reduce non linear effects the elevator deflection was chosen such that there will likely be a linear
variation between angle of attack and deflection. Finding information on the angles at which this assumption
does not hold true proved to be very challenging, especially since the tail airfoil is not symmetric. Instead
data for another aircraft, the Fokker-27 Friendship [3], to estimate the range of the linear region. From this
data, an elevator deflection between 10 and -15 degrees, for a tail angle of attack range between -5 and 8
degrees returned a linear relationship between the normal tail force and tail angle of attack. Since the exact
local tail angle of attack is not known for all test cases, the assumption was made that a maximum elevator
deflection of 8 still gave a linear relation.
• The negative deflection was chosen at the same angle as the positive deflection to give a better comparison.
An elevator deflection of 0 degrees was be taken as intermediate value to interpolate between the -8 and 8
degrees. This is used as a "neutral" value for the comparisons.
• The main trade off was whether it is worth spending more time switching the elevator deflection versus
obtaining other measurement points elsewhere. It was deemed to be important to analyze at least 3 elevator
deflections to better understand the effect of propeller configuration on the longitudinal stability. In the ideal
case more elevator deflections would have been better due to the aforementioned uncertainties from the lack
of elevator deflection data. However this would have been too time consuming.

2.3.6 Other aspects

The measurement points mentioned above were all that could be measured in the 3 hour time span. The original
test matrix was designed using conservative time estimations. Originally, measurement set 4 and drag sets 2 and
3 were not planned. However due to the experiment progressing faster than originally planned these points were
added to fill the 3 hour measurement time span. Each measurement was taken over a span of 15 seconds after V∞
conditions had stabilized around the desired value. This gives enough time for the flow to fully settle, and reduces
bias by measuring over 15 seconds.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 7

 3 CORRECTIONS

2.4 Test matrix limitations

Along with the limitations of performing wind tunnel experiments mentioned in ??, and the time constraint, the
chosen test matrix design also gives some limitations. First of all the biggest limitation is that only one measure-
ment per measurement point is taken. This choice was made due to the limited time available along with the fact
that it was deemed it to be more important to have more measurements to quantify the linear domain. One double
measurement was taken by accident, namely DPN 58 and 56. These gave almost identical results. If there were
to be two measurements per point there would not be enough time to perform all the desired measurements. The
trade off for this test matrix instead of one which allowed multiple measurements per points was made for this
reason. This gives the risk that some measurements are outliers. A measure that was taken against this was by
taking the two angles of attack as far apart as possible. Any errors in either affect the CL − α curve slope less than
if the points were taken close together.

Secondly a major limitation of the current setup is that only the linear region is analyzed. The effects in the non-
linear region are likely to be equally as, if not more, important than the effects in the linear region. As mentioned
before, this was chosen in the trade-off between linear and non-linear due to the time constraints. A better choice
might have been to only look at the non-linear region and simulate the linear region using CFD to obtain the full
flight domain. However since the main goal of the course is to learn and understand how to perform wind tunnel
experiments, and due to the fact that performing a CFD for the full linear range would be very time consuming
and outside of the scope of this course, this approach was not taken.

Some measures have however been taken to reduce the bias in the measurements due to measurement order. To
do so, every measurement set is randomized by hand. The only similarity between measurement sets is that only
two velocity changes are used (once from 40 m/s to 30 m/s and back) in order to minimize time. Furthermore δe
is constant in a measurement set. All other aspects about the order are completely random. This does not fully
eliminate any bias from being present due to measurement order but it does reduce the overall error.

3 Corrections
The lab experiment was performed in TU Delft’s Low-Turbulence Tunnel (LTT) in n closed, octagonal wind tunnel,
bounded by walls, which cause a non-physical increase of boundary layers along them, compared to free-flight
conditions [2]. As the experiment was executed in three dimensions, two sets of walls and their mutual interference
has to be accounted for. Consequently, as the performance parameters of the experiment are substantially influenced
by this effect, they have to be corrected before they can be used to draw valuable results. The boundary corrections
that have to be applied for this case are:

1. Blockage corrections (Solid body, wake, propeller slipstream)

2. Lift interference
3. Downwash correction at the tailplane

3.1 Blockage Corrections

3.1.1 Solid Body blockage

The total blockage effect, that has to be corrected for, is caused by the presence of the model inside a constrained
test section. The volume that the model takes up, decreases the effective area for the wind flow. Hence, by the
continuity conservation and Bernoulli’s equation, the flow velocity is increased in the vicinity of the model [2]. The
solid blockage effect is a function of the size and thickness of the model, thickness distribution and is independent
of the camber [2]. According to Herriot [2] “The model can be represented by a source-sink distribution and is
contained in the tunnel walls by an infinite distribution of images”. The effect of images is summed up to provide
a simpler equation for the solid blockage factor, for every component i [2].
K1 τ1 Vwing
sbw = (16)
C 3/2

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 8

 3 CORRECTIONS

K1 is the body shape factor for each component and can be derived from Figure 3a or the NACA Report 995 [4]
for a known thickness ratio of the airfoil t/c. The factor τ1 depends on the tunnel breadth-to-width ratio (B/H)
and the model span-to-tunnel width ratio (2b/B) and can be derived from Figure 3b or the NACA Report 995 [4].
Vwing [m3 ] is the volume of the model and C [m2 ] the cross-sectional area of the tunnel.
Similarly, for bodies of revolution, like fuselages and nacelles, the respective solid blockage factor is calulated
as:
K3 τ1 Vf uselage
sbf = (17)
C 3/2
K3 is the body shape factor of the fuselage and can be derived from Figure 3a or the NACA Report 995 [4] for a
known diameter-to-length ratio d/l. Similarly, the τ1 factor is calculated from Figure 3b or the NACA Report 995
[4], for bodies of revolution b/B = 0.

(a) Body shape factors K1 and K3 (b) Factor τ1

Figure 3: Graphs for determining factors K1 , K3 and τ1 , in order to calculate the solid blockage factor for every
component of the model. [2]

Subsequently, the model is divided into its sub-components; fuselage, main wing, horizontal and vertical
tailplane, and support struts. After every factor is computed separately, superposition will be applied and a
total solid blockage factor will be derived as:

sbtot = sbf + sbmw + sbht + sbvt + sbss (18)

• Support struts: The aft support strut and the wing support struts are supporting the whole structure and
they are not situated at the horizontal centerline of the tunnel. To compute their blockage corrections, they
are considered as half-models, and the current tunnel is mirrored with the upper wall as the symmetry axis.
This way, the support struts are transformed into full models in a tunnel with breadth 2H. In addition, it
is assumed that the wing support struts are at the vertical centerline of the mirrored tunnel, because the
measurements of the velocity increment associated with the support gear in the absence of the model were
needed [5], but they were not available.
Furthermore, since the geometry of the support struts is complicated, they are assumed to be elliptical rods,
with thickness ratio 0.18. With this information, their shape factor K1 was derived from NACA Report 995
[4]. This simplification deemed acceptable, as a more detailed calculation would be very time consuming,
with minimal gain in accuracy. Their blockage factor is calculated from the Equation 16.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 9

 3 CORRECTIONS

• Fuselage: The fuselage of the model is considered as a body of revolution with negligible geometric span to
tunnel breadth ratio b/B ≈ 0. Its factors K3 and τ1 are derived from Figure 3a and Figure 3b respectively,
or from NACA Report 995 [4]. In the end, Equation 17 is applied to find the solid blockage factor for this
component.

• Main wing: The airfoil of the main wing is the cambered airfoil DU 96-150, for which there is no already
calculated values of its shape factor K1 . To analytically calculate this factor, the following equation is used
[4]:
π 3/2 Λ 1 V
K1 = , with κ1 = (19)
16 t/c κ1 2sct
In Equation 19, s [m] is the wing half-span, c [m] is the mean-aerodynamic chord, t [m] is the wing’s thickness
at m.a.c., V [m3 ] is the volume of the wing, and Λ is a factor calculated for the non-cambered DU 96-150
airfoil as [4]: v "  2 #  
16 1 y u
Z uh  x i dy x
Λ= t 1 − CP 1+ d (20)
π 0 c c dx c

where (x/c, y/c) are the chord-wise coordinates of the non-cambered airfoil, dx dy
is the slope of the surface
of the base profile at x/c, and CP is the base profile pressure coefficient for an incompressible flow with an
angle of attack α = 0. The integral of the Equation 20 is calculated numerically with the trapezoidal rule.
In the end, Equation 16 is applied to find the solid blockage factor for this component.

• Horizontal tailplane: The airfoil of the horizontal tailplane is a symmetrical 64-Series NACA airfoil. Its
shape factors K1 and τ1 are derived from Figure 3a and Figure 3b respectively, or from NACA Report 995
[4]. In the end, Equation 16 is applied to find the solid blockage factor for this component.

• Vertical tailplane: Similarly with the horizontal tailplane, for the vertical tailplane, with a symmetrical
NACA 0015 airfoil, its shape factors K1 and τ1 are derived from Figure 3a and Figure 3b respectively, or
from NACA Report 995 [4]. However, since it is situated on the vertical centerline, its ratio of its span to
the tunnel height is used to find τ1 . In the end, Equation 16 is applied to find the solid blockage factor for
this component.

A cumulative table that contains the geometrical factors K1 /K3 and τ1 , along with the computed solid blockage
factor for every component, can be seen in Table 1.

Table 1: Component volume, shape factors τ1 , K1 or K3 , and solid blockage factor for every component
of the model.

Component V [dm3 ] τ1 [−] K1,3 [−] sb [%]

Aft support strut 0.45 0.881 1.046 0.004914
Wing support struts 1.76 0.856 1.046 0.037528
Fuselage 16.06 0.863 0.923 0.429625
Main wing 3.02 0.885 1.255 0.112734
Horizontal tailplane 2.45 0.865 1.019 0.072410
Vertical tailplane 0.35 0.864 1.035 0.010643
Total solid blockage 0.667853

3.1.2 Wake blockage

An additional effect of the presence of the walls that bound the wind tunnel section, is that they constrain the
expansion of the wake downstream of the model, as it would expand in a free-flight condition. Consequently,
the flow velocity inside the wake decreases, while the freestream velocity increases, so as to ensure continuity
conservation. Due to the velocity difference inside and outside the wake, a negative pressure gradient in the axial

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 10

 3 CORRECTIONS

direction is developed, and in the measured data the drag appears to be larger [2]. To account for that phenomenon,
the “Maskell’s Method” is applied and a wake blockage factor is calculated, according to the equation:
S 5S
wb = CD,0 + CD,s (21)
4C 4C
In Equation 21, S [m] is the wing span, C [m] is the cross-sectional area of the tunnel, CD,0 is the drag for zero
lift, which can be calculated from the drag polars in a CD − CL2 graph, as it is shown in Figure 4. The drag
coefficient due to separated flow CD,s is included in Equation 21, when it is CD,s > 0, which signifies that the flow
is separated. For an attached flow the second term of Equation 21 can be neglected.

CD,s = CD,u − (CD,0 + CD,i ) > 0 , for separated flow (22)

where CD,u is the uncorrected drag and CD,i is the lift-induced drag, calculated by the equation:
2
CL,u
CD,i = (23)
π0 AR
where CL,u is the uncorrected lift, AR is the wing aspect ratio and 0 the Oswald’s wing efficiency.
More specifically, CD,0 can be calculated from

Figure 4: Drag polar graph for drag analysis for a lifting wing. [2]

To account for all wings that contribute to the wake, two wake blockage factors are to be calculated, for the
main wing and the horizontal tailplane. It is assumed that the contribution of the vertical tailplane to the wake
created by the tail is negligible, since the sideslip angle β was zero during the whole experiment.

• Wing: Initially, the provided model-off data are corrected for the solid blockage of the support struts.
Subsequently, the data with the model without its tailplanes are corrected for solid blockage and the corrected
model-off coefficients are subtracted. Therefore, the corrected coefficients CD and CL2 can be plotted, similar
to Figure 4.
As it can be seen in the test matrix used, in Appendix A, the free-stream velocity V∞ was varied between
30 m/s and 40 m/s for this experiment, two drag polar plots are used to determine two values of CD,0 .
Furthermore, for angles of attack −2 deg and 7 deg, as were the values for this experiment, the flow appeared
to be attached, as CD,s < 0 and the drag polar was linear for these angles. For the main wing, the wake
blockage factor can be now calculated from Equation 21.

• Tail: To capture the effect of the horizontal tail wing on the wake, the measured data for disabled propeller
are first, corrected for solid blockage and afterwards, graphed in a drag polar plot, in a similar fashion to the
tail-off data. The zero-lift drag coefficient CD,0 is derived and from Equation 21, the wake blockage factor is
computed for attached flow.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 11

 3 CORRECTIONS

Table 2: Wake blockage factors wb [%] for wing, tail and total for two values of velocities

Free-stream velocity V∞ [m/s] 30 40

Wake blockage due to wing [%] 0.104138 0.098746
Wake blockage due to tail [%] 0.019289 0.018158
Total wake blockage [%] 0.123427 0.116904

3.1.3 Propeller slipstream blockage

When the propellers, which are installed at the horizontal tailplane are operating, they add momentum to the
flow and that, will cause the flow velocity inside the propeller slipstream to increase. To preserve flow continuity,
the flow velocity outside the slipstream will decrease, causing an “inverse blockage effect”, as a positive pressure
gradient in the axial direction is developed. Therefore, due to propeller slipstream blockage, the measured velocity
is overestimated and thus, a negative factor is needed to correct the parameter. From the “Actuator Disk Theory”,
Glauert [6] proposes a slipstream blockage factor:

TC SP
ss = − √ (24)
2 1 + 2 TC C
πD 2
where TC [−] is the thrust coefficient obtained at each measurement point from Equation 5, SP = 2 4 P [m2 ] is
the area of the 2 propellers that generate the thrust and C [m2 ] the tunnel’s cross-sectional area.
The thrust generated by the propellers can be estimated as the difference between the thrust from the propeller-
on and the propeller-off data. In this experiment, in order to be as accurate as possible for the thrust estimation,
for every elevator deflection angle δe ∈ [−8, 0, 8] deg used to measure the performance for every propeller-on
configuration, the respective data with propeller-off configuration were also measured, as it can be verified in the
test matrix in Appendix A.
Tof f − Ton
T = (25)
cos α

3.1.4 Total blockage

The total blockage factor tot is the sum of the solid blockage, the wake blockage and propeller slipstream blockage
factors. The latter is included in the sum, only in the propeller-on configurations, where the propellers generate
thrust.
tot = sb + wb + ss (26)

The total blockage corrections will be applied on the measured velocity Vunc , dynamic pressure qunc , raw lift
coefficient CDunc , drag coefficient CDunc and pitching moment coefficient CMc/4,unc , according to the equations:

q = qunc (1 + tot )2 (28) CD = CDunc (1 + tot )−2 (30)

V = Vunc (1 + tot ) (27) CL = CLunc (1 + tot )−2 (29) CMc/4 = CMc/4,unc (1 + tot )−2 (31)

3.2 Lift Interference

The walls of the wind tunnel impose conditions on the flow such as; no normal flow at the boundaries and increase
the lift and hence moment generated by the airfoil at certain angles of attack. This results in a perception that the
airfoil has greater camber than it would have in free flight, and prompts corrective action. The correction made,
accounting for the effect of the tunnel boundaries on the flow streamlines of the lifting airfoil, is called the lift
interference correction.
The rectification is done by using potential flow theory and modelling the airfoil as a point vortex. The wing
is approximated as a horseshoe vortex system with a bound vortex line at the quarter-chord of the airfoil. This
image system with vortices of alternating signs lined up till infinity, ensures that there is no flow normal to the
wall.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 12

 3 CORRECTIONS

(a) Vortex Span (b) Values for δ (c) Values for τ2

Figure 5: Lift Interference Correction [2]

The corrections are to be applied to the angle of attack ’∆α’, drag coefficient of the wing ’∆CDW ’, pitching
moment coefficient about the quarter chord point ’∆CM0.25c ’. They are defined as follows [7]:

S
∆α = ∆αuw + ∆αsc = δ ∆CLW [1 + τ2 (0.5c)] (32)
C
S 2
∆CDW = δ C (33)
C LW
1
∆CM0.25c = ∆αsc CLα (34)
8
Where the τ2 factor represents the effect that forces the streamlines to straighten unnaturally, related to the
upwash at the quarter-chord point.
Let us take a look at the procedure used to calculate effective vortex span be , δ, and ∆αuw according to [7].
The first step is to find the effective vortex span be . We know that be = b+b2 , using Figure 5a we find the ratio b
v bv

where b is known. From be we find k, assuming an octagonal jet using Figure 5b we get δ. The final step, after
calculating lt using the relation lt = 0.5c̄, we can use Figure 5c to calculate the τ2 factor.

Table 3: The Lift Interference Corrections

Velocity(m/s) [A.O.A] ∆α ∆CDW ∆CM0.25c

30 [-2] 0.000563618 0.000046657 0.000004823
40 [-2] 0.000563374 0.000046617 0.000004795
30 [+7] 0.005588020 0.004586310 0.000004823
40 [+7] 0.005558896 0.004538628 0.000004795

3.3 Downwash Correction at the Tailplane

Due to the image vortex system being applied, the change in the circulation at the horizontal tail due to upwash
at the wing is significant. Mirror vortices can reduce the angle of attack on the tail, creating the illusion that the
aircraft is more stable than it actually is. To correct this, the change in this angle is calculated as:

S
∆αt = ∆αuw + ∆αsc,t = δ CL [1 + τ2 (lt )] (35)
C W
Here, lt is the tail arm, i.e. the distance between the quarter-chord of the wing and the quarter-chord of the
horizontal tail. The change in the pitching moment about the tail is then given by:
 
δCM0.25c
(∆CM0.25c )t = ∆αt (36)
δαt

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 13

 3 CORRECTIONS

To calculate this, we would ideally need to vary the angle of attack of the horizontal tail and measure the
corresponding change in pitching moment about the centre of gravity. Since this cannot be done, it is estimated
as [2]:
   
δCM0.25c St lt qt dCLt
=− (37)
δαt S · (MAC) q dαt
where
  
qt dCLt 0.1AR
= · 0.8 (38)
q dαt ARt + 2
In these equations, St [m2 ] is the area of the horizontal tail and ARt is its aspect ratio. The changes in pitching
moment about the tail have been found to be as follows:

Table 4: Correction to pitching moment due to downwash at the tailplane.

V∞ [m/s] α [◦ ] (∆CM0.25c )t
30 -2 0.002338
30 7 0.019519
40 -2 0.002337
40 7 0.019418

3.4 Lift and drag correction for propeller thrust

In order to begin implementing the appropriate corrections to the uncorrected aerodynamic coefficients, the phe-
nomenon of “super-lift” caused by the propellers slipstream wetting the horizontal tailplane surface, has to be
accounted for and removed from the uncorrected lift and drag coefficient [8]. For that reason, the Eckert correc-
tion, as mentioned in the ICAS2006 paper of Eckert et. al [9], is applied to correct the measured CL and CD and
remove the effects of thrust. The following simplifications have to be adressed [9]:

• For the propeller jet speed at the c/4-line of the wetted wing area the over-speed ratio is about 2.
• A contraction of the jet over the wing and a possibly different size of the wetted area at the upper and the
lower wing surface will not be taken into account.
• Swirl effects changing the local angle of attack at the wing leading edge behind the propeller will be included
in a shape factor k.

The thrust-free aerodynamic coefficients (CT = 0) are written as a function of the thrust-influenced coefficients
(CT ) as:
   ! 12 −1
CLCT =0 = CLCT qq∞

q∞ DP
n=2
ci 
p
1 + CTi + 1
E
X
CL2
, with = 1 + k CTi p  (39)
C
 DCT =0
 = C DCT +  CT
qE 2
 q E b i=1
c̄ 2 1 + CTi
πAR ( q∞ ) −1

3.5 Implementation of Corrections

The procedure followed to correct the measured data for this experiment is described.
1. The aerodynamic coefficients of lift CL and drag CD have to be corrected for the influence of the thrust
created by the propellers. The thrust coefficient TC is calculated from the Equation 5 for an estimation
of thrust T according to the Equation 25. For the two propellers at the horizontal tailplane, the Eckert
correction [9] is applied, as it is shown in Equation 39, to remove the effects of thrust from the coefficients
CL and CD .
2. At these corrected coefficients CL and CD , along with the dynamic pressure and the pitching moment coef-
ficient at quarter-chord CMc/4 , the total blockage correction is implemented, according to Equations 27 - 31.
For the prop-off data, the correction due to propeller slipstream blockage from Equation 24 is not applied.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 14

 3 CORRECTIONS

3. The Lift interference corrections are applied on the angle of attack, drag coefficient and pitching moment
coefficient for every velocity and angle of attack configurations. Furthermore, for the pitching moment
coefficient, the correction due to downwash at the tailplane is applied.

3.6 Corrected Values

In Figure 6a, it can be seen that the trends of the uncorrected and corrected data are similar in nature, with a
positive slope as expected. The linear trend of both the graphs can be attributed to the fact, that the ’α’ chosen
lie in the linear flow region, far away from any non-linear behaviour such as stall. The error between the graphs is
minimal and uniform with ’α’, hence the inferences made for the uncorrected data also hold true for the corrected
data. The uncorrected data is relatively higher in magnitude, compared to the corrected data.
In Figure 6b, the comparison is done for the pitching moment coefficient. The data in both cases can be seen
to have a positive slope,indicating that the tail off configuration is unstable in nature. This in accordance to
conventional wisdom, as the primary purpose of a tail is to provide stability. The lowest and highest points of error
between the sets of data are at α = 7 and -2 respectively. The error is not uniform in this scenario and is seen
to decrease with an increase in the ’α’. In this case the corrected data is higher in magnitude, compared to the
uncorrected data.

(a) CL - α (b) CMp - α

Figure 6: Comparison of Corrected and Uncorrected values for Tail Off measurements at V∞ = 40 m/s

In Figure 7a, it can be seen that the trends of the uncorrected and corrected data are similar in nature, with a
positive slope as expected. The linear trend of both the graphs can be attributed to the fact, that the ’α’ chosen
lie in the linear flow region, far away from any non-linear behaviour such as stall. The lowest and highest points
of error between the sets of data are at α = -2 and +7 respectively. The error is not uniform in this scenario
and is seen to increase with an increase in the ’α’. It can be seen that the uncorrected data is relatively higher in
magnitude compared to the corrected data.
In Figure 7b, the comparison is done for the pitching moment coefficient. The data in both cases can be seen
to have a negative slope,indicating that the particular configuration chosen is stable in nature. The lowest and
highest points of error between the sets of data are at α = -2 and +7 respectively. The error is not uniform in this
scenario and is seen to increase with an increase in the ’α’. The uncorrected data is lower in magnitude compared
to the corrected data.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 15

 4 RESULTS

(a) CL - α (b) CMp - α

Figure 7: Comparison of Corrected and Uncorrected values for Prop-On measurements at V∞ = 40 m/s and
J = 1.85

4 Results
This section gives the results obtained from the wind tunnel experiment.

4.1 Longitudinal Stability

In Figure 8 the results for the longitudinal stability are shown for a δe of 0 using different values of J. The first
thing that can be seen for this is that the assumption that was used during the test matrix design regarding the
co-rotating configuration was correct; The result of the co-rotating configuration is seemingly in the middle of
the results from the outboard up and inboard up configuration. Furthermore with increasing angle of attack the
tailplane gives an decreasing CMP which corresponds to a nose down moment. With increasing J the lines start to
converge, which is expected as at higher values of J the thrust is lower therefore the difference of the configurations
is also lower. The lines for the different configurations area rotated and shifted version of the prop-off line. The lines
of all configurations are steeper than that of the prop-off configuration, which means that the tailplane has a bigger
impact on the longitudinal stability with the propellers. This makes sense as the propellers increase the effective
velocity at the tailplane, which in turn increase the lift (and drag) of the tail. This can be seen as a stabilizing effect.

The other elevator deflection angles (δe ) show similar results. However the lines are shifted up or down due
to the increase/decrease in tail lift by the δe . An increase in δe causes the CMP to increase while a decrease in
δe does the opposite. These plots are therefore not shown here, as the same conclusions can be obtained from them.

In terms of longitudinal stability the inboard-up configuration offers an overall higher pitching moment from the
tail at the quarter chord point. The reason for this is that the inboard-up config offers an increase in local angle of
attack to the tail due to the propeller spin, while the outboard-up decreases local angle of attack. Since the c.g. of
a full scale aircraft will be somewhere behind the quarter chord of the main wing the CM from the tail at the c.g.
location will be lower, as its closer to the tail. Therefore although the CMP values at quarter chord are negative
for all linear domain angles of attack for the inboard up configuration at 0 δe , they will likely be positive for the
"true" c.g. position. This will mean that a lower δe will be needed in the inboard-up config to trim the aircraft
than in the outboard-up config., which is benefitial in terms of drag and control power required. Therefore for the
linear region the inboard-up configuration offers the best longitudinal performance.

However in the non-linear region the story might be different as the inboard-up configuration will lead to an earlier
stall of the tail due to an increased local angle of attack. In this region the Outboard-up configuration will provide
more longitudinal stability, as the tail will take longer to stall. The co-rotating configuration has the benefits and
disadvantages of both configurations. It however leads to asymmetry of lift which can cause undesireable rolling

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 16

 4 RESULTS

moments. For the design of a plane in such configuration these factors must be taken into account in a trade-off
to chose the optimal configuration over the entire flight envelop.

(a) J = 1.6 (b) J = 1.85 (c) J = 2.1

Figure 8: Longitudinal Stability for different Propeller Rotation Configurations

4.2 Control Effectiveness and Stability Margin

(a) α = −2◦ (b) α = 7◦

Figure 9: Elevator Effectiveness at V∞ = 40 m/s and J = 1.6

(a) α = −2◦ (b) α = 7◦

Figure 10: Elevator Effectiveness at V∞ = 40 m/s and J = 1.85

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 17

 4 RESULTS

(a) α = −2◦ (b) α = 7◦

Figure 11: Elevator Effectiveness at V∞ = 40 m/s and J = 2.1

4.3 Reynolds number effects

The Reynolds number, given in Equation 8, represents the ratio of inertial forces to viscous forces in a flowfield.
For the experiment in the low speed wind tunnel the matching of the exact free flight conditions is unattainable,
due to the speed limitations of the wind tunnel. As a consequence, the Reynolds number in the experiment will be
significantly lower than that encountered in actual flight conditions. This variation of the Reynolds number causes
the “Reynolds number effects” to the performance parameters of the experiment. These effects are “phenomena
that are related to the change in flow development with Reynolds number for a particular configuration in free air.”
More specifically, some Reynolds number effects, pertinent to the experiment are:

• The location of the point, where the boundary • Flow separation point at the leading edge and the
layer transitions from laminar to turbulent. trailing edge.
• Boundary layer thickness • Shock wave-boundary layer interaction.
• Three-dimensional transition mechanisms (e.g. • Vortex separation
cross-flow instabilities) • Change in effective thickness of the trailing edge.

As the Reynolds number variation has a direct effect on the location of the transition of the boundary layer,
its thickness under a shock wave, and the thickness change of the trailing edge, these are called “direct Reynolds
number effects”. These direct effects can lead to the “indirect Reynolds number effects”, which can be the position
and strength of a shock and the overall circulation and pressure distribution.
A rather significant effect of the Reynolds number is the location of transition of the boundary layer from
laminar to turbulent, as it affects the drag of the aircraft. To account for that effect, transition is forced on various
locations of the testing model (wing, horizontal and vertical tailplane, fuselage, and nacelles). The forced transition
is accomplished by installing narrow strips of carborundum grains at given locations. More specifically, on the wing,
trip strips are positioned at 5% of the local chord on the upper side and 10% of the local chord on the lower side,
and on the tailplanes, at 5% of the chord on both upper and lower sides. The tripping mechanisms contribute
to the triggering of transition of the boundary layer without further increasing the boundary layer thickness and
without generating streamwise vortices that disturb the boundary layer.
For this experiment, the free-stream velocity in the wind tunnel was varied between 30 m/s and 40 m/s for all
propeller and elevator deflection configurations, as it can be seen in the test matrix in Appendix A. These values
correspond to Reynolds numbers Re30 ≈ 3.4 · 105 and Re40 ≈ 4.5 · 105 , respectively. The corrected coefficients
CL (Figure 12a), CD (Figure 12b) and CMp,c/4 (Figure 12c) are plotted w.r.t. the angle of attack that is varied
between −2 and 7 deg, for every measurement set. In the following Figure 12, the measured and corrected data
for outboard-up propeller configuration and elevator deflection δe = 8 deg are chosen to be plotted, since all
configurations exhibited the same behaviour.
As it is shown in Figure 12a and Figure 12c, the values of lift and pitching moment for 30 and 40 m/s correspond
almost entirely and it can be reasoned that the change of Re has minimal influence on these coefficients. On the

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 18

 4 RESULTS

other hand, from Figure 12b it can be observed that the increase of the free-stream velocity and consequently the
Reynolds number, causes a substantial decrease in the measured drag.

(a) CL

(b) CD

(c) CMp,c/4

Figure 12: (a) Lift coefficient, (b) Drag coefficient and (c) Pitching moment coefficient, for two values of free-stream
velocity and 3 advance ratios. Propeller configuration: Outboard-up. Elevator deflection: 8 deg

4.3.1 Re effect from pressure taps

The pressure taps also give some insight into Re effects. In Figure 13 the results can be seen for the pressure taps.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 19

 4 RESULTS

Figure 13: Re effects on the pressure tap on the side of the fuselage

Here it can be seen that there is a difference between the results for 30 m/s and 40 m/s. The Cp was normalized
using the corrected dynamic pressures

4.4 Acoustic Results

The acoustic results obtained from the wind tunnel experiments are given here. First the background noise is
discussed by use of the drag measurement data. After

4.4.1 Background noise

In order to analyse the effect of propeller configurations on noise emission an understanding of the background
noise present in the wind tunnel must be made. The results of the microphone data during the drag measurements
(in prop-off conditions) are shown in Figure 8. Here all elevator deflection results are shown for 40 m/s and the
two angles of attack. This data has not been corrected for scaling due to the fact it is merely used to analyze the
noise qualitatively.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 20

 4 RESULTS

Figure 14: Background noise generated by the six microphones for the drag measurement sets at different elevator
[Link] in microphone 3 is applicable to all microphones

From this a couple results are evident. First of all, the measurements at −2◦ angle of attack have a lower back-
ground noise than those at 7◦ . This is a repeating pattern that is also found for the prop-on measurements for
all configurations. The likely explanation is that at lower angles of attack there is less flow separation, which is a
significant source of broadband noise. Secondly the difference in noise between elevator deflections is very low. In
general the noise generated by 8◦ seems to be slightly lower than that of −8◦ . The reason for this is likely that
−8◦ perturbs the flow more than the 8◦ case, generating more turbulence. Only the sixth microphone, along with
the first below 1000 Hz, shows a significant difference between an elevator deflection of 8◦ and −8◦ . The difference
is however only visible for the 7◦ case in both mic 1 and 6. The reason why this can mostly be seen by mic 6 is
due to the fact that it is the closest mic to the tailplane. Furthermore it is only visible at 7◦ due to the flow being
less turbulent at −2◦ . The difference in noise might therefore be due to different turbulence being produce at the
tailplane between elevator deflections.

Finally for microphones 1 to 5 there are steep peaks below 1000 Hz which have almost the same amplitude for
both angles of attack and for all elevator deflections. The source of these peaks are very unclear. Upon further
investigations into the drag measurements at 30 m/s these peaks stay consistent in amplitude. The peaks present
in microphone 1 do however somewhat vary with the different freestream velocities. It is therefore unlike these
are generated by any flow phenomena, as they show no variance with V∞ , α and δe . Since the peaks in 1 do vary,
they are likely a mix of aerodynamic phenomena and ht phenomena creating the peaks in mics 2 to 5. These are
likely generated by either electronic noise, microphone self noise or some other measurement device. However these
peaks are not present in any of the prop-on data, and therefore can be ignored. The comparison was also made
between 30 and 40 m/s. Both showed a similar pattern in terms of noise, but the 30 m/s case had overall lower
noise levels (minus the aforementioned peaks which stayed the same).

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 21

 4 RESULTS

4.4.2 Prop-off and Prop-on comparison.

Next up the prop-off and prop-on configurations were compared to determine the effect of the presence of the
propeller on the noise. Notable is that for mics 1 and 6 the broadband noise is only slightly higher than that
observed in the drag measurements, while for the other mics this often has a more than 20 db difference. Looking
at the pressure tap, tap 1 which coincides in location with mic 1 seem to have a laminar boundary layer which
transitions to turbulent in subsequent taps. The laminar boundary layer may be less affected by the presence of
the propeller in terms of noise production than the other taps. This however does not explain why mic 6 has the
same broadband level.

The same comparison as was performed for the prop-off cases with regards to varying elevator deflections was also
done for the same propeller configurations. This yielded similar results, namely that in general the noise generated
by −8◦ δe was slightly higher than that generated by the other δe values. This pattern was consistent for the
broadband noise. In the many peaks (see below) present in the data this pattern was in general not present. This
result was the case for all propeller configurations. It was therefore decided to only look at one δe to compare
the differences between propeller configuration, namely the 0δe as then the co-rotating configuration can also be
compared.

4.4.3 Effect of J

To compare the different configurations and different values of J the sound frequency was non-dimensionalized
using the BPF. Upon plotting different J of the same and different propeller configurations together an interesting
pattern was found. The expectation was that at the BPF and the first 1-2 multiples of this frequency, peaks could
be seen. However peaks were found at the values of the shaft frequency and multiples thereof. Every multiple of
the shaft frequency has a corresponding peak. Figure 8 shows this phenomena for one of the mics; This pattern
was repeated for all the mics except for mic 6, which can be seen in Figure 8. The cause of this is likely not
aerodynamic in nature, and may be caused by various other reasons. The fact that this pattern cannot be seen in
mic 6, but is present for all other mics raises even more questions. A possibility is that this noise is highly suspect
to directivity and therefore only reached mics 1 to 5. Regardless this result means that the noise generated by
the BPF cannot be analysed for the first 5 mics. This is due to the fact that this noise is generated at an integer
multiple of the shaft frequency, thus the separation can’t be made between this and the shaft "resonance". It was
also found that these peaks had very little difference in height at lower frequencies between different, suggesting
this effect to be electronic in noise rather than mechanical. For the sixth mic however, the peaks could only be
seen for values of J of 1.6, and sometimes for 1.85. For J= 2.1 and most of 1.85 the peaks at the BPF were very
close to the broadband noise, having only a few db in difference. Therefore the analysis between configurations is
only performed for J = 1.6.

Figure 15: Results for mic 2 and mic 6 comparing different values J using for the same test setup using the BPF.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 22

 REFERENCES

4.4.4 Comparisson between configurations

Using the Π term for the pressure ratio defined earlier, the noise levels for the different configurations were compared,
as seen in Figure 16.

Figure 16: Mic 6 results between different configuration at 1.6 J and 0 degrees elevator deflection. Left plots are
at 40 m/s, right plots at 30 m/s. Top plots give results at 7◦ α while the bottom plots give those

From this a few aspects are noticeable. In general sound values at −2◦ α are lower than those at 7◦ α. This is
likely a result of the more turbulent flow at higher angle of attack. Furthermore Π values for the Co-rotating and
outboard up configuration are very close together for the 7◦ α angle of attack. This is due to the fact that they
spin in the same direction facing towards the mic. For the 30 m/s however there is some difference between them,
unlike for the 40 m/s. Finally the second harmonic of the BPF can be seen in some of the plots. For others however
the SNR ratio is too low to notice it. This result however doesn’t tell much about how the noise scales into a full
size model, as it is only a snapshot of the noise in one direction, facing towards the fuselage. In order to properly
determine the noise levels and difference between configurations a more in depth analysis is required in a more
representative testing environment, as described in the methodology section.

5 Conclusion

References
T. Sinnige and G. Eitelberg. Lab-exercise manual ae4115 experimental simulations. Technical report, Technical
University of Delft (TU Delft).

Jewel B. Barlow, William H. Rae, and Alan Pope. Low-Speed Wind Tunnel Testing. Wiley, Reading, Mas-
sachusetts, 1999. ISBN 978-0-471-55774-6.

Anonymous. -27 staartvlakmodel 1:9. eigenschappen van het horizontale staartvlak. Technical report, N.L.L.,
October 1955. N.L.L. Rapport A 1389.

John G. Herriot. Report 995 - blockage corrections for three-dimensional flow closed-throat wind tunnels, with
consideration of the effect of compressibility. Technical report, National Advisory Committee for Aeronautics,
1950.

H. C. Garner, E. W. E. Rogers, W. E. A. Acum, and E. C. Maskell. Subsonic wind tunnel wall corrections.
Technical report, North Atlantic Treaty Organization, October 1966. AGARD-AG-109.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 23

 REFERENCES

H. Glauert. Wind tunnel interference on wings, bodies and airscrews. Report 1566, Marine and Transport
Technology, 1933.

G. Eitelberg and T. Sinnige. Ae 4115: Experimental simulations, 2021.

Ekin Orer. Propeller slipstream correction for wind tunnel applications. Master thesis, Delft University of
Technology, sep 2020.

D. Eckert, G. H. Hegen, and W. Kuehn. Dnw’s method to correct for support and wall interference effects on low
speed measurements with a large propeller powered transport aircraft model. Optimage Ltd.,;, 2006.

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 24

 A FULL TEST MATRIX

Appendices
A Full test matrix

Measurement Propeller DPN V∞ RPS J α δe

Type Configuration [m/s] [−] [−] [deg] [deg]
1 0 0 0 -2 -8
Blank Outboard up
2 0 0 0 7 -8
3 40 93.7 2.10 -2 -8
4 40 106.4 1.85 7 -8
5 40 93.7 2.10 7 -8
6 40 123.0 1.60 -2 -8
7 30 70.3 2.10 7 -8
8 30 92.3 1.60 -2 -8
Meas. Set 1 Outboard up
9 30 70.3 2.10 -2 -8
10 30 79.8 1.85 7 -8
11 30 92.3 1.60 7 -8
12 30 79.8 1.85 -2 -8
13 40 106.4 1.85 -2 -8
14 40 123.0 1.60 7 -8
δe change 1 Outboard up 0 0 0 7 8
15 40 93.7 2.10 7 8
16 40 123.0 1.60 7 8
17 40 93.7 2.10 -2 8
18 30 70.3 2.10 7 8
19 30 70.3 2.10 -2 8
20 30 79.8 1.85 -2 8
Meas. Set 2 Outboard up
21 30 79.8 1.85 7 8
22 30 92.3 1.60 7 8
23 30 92.3 1.60 -2 8
24 40 106.4 1.85 -2 8
25 40 123.0 1.60 -2 8
26 40 106.4 1.85 7 8
δe change 2 Outboard up 0 0 0 7 0
27 40 123.0 1.60 7 0
28 40 123.0 1.60 -2 0
29 40 106.4 1.85 7 0
30 40 93.7 2.10 7 0
31 30 70.3 2.10 7 0
32 30 79.8 1.85 -2 0
Meas. Set 3 Outboard up
33 30 92.3 1.60 -2 0
34 30 70.3 2.10 -2 0
35 30 92.3 1.60 7 0
36 30 79.8 1.85 7 0
37 40 106.4 1.85 -2 0
38 40 93.7 2.10 -2 0
39 0 0 0 -2 0
Blank Outboard up
40 0 0 0 7 0
Prop change No prop 0
41 0 0 0 -2 0
Blank No prop

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 i

 A FULL TEST MATRIX

Measurement Propeller DPN V∞ RPS J α δe

Type Configuration [m/s] [−] [−] [deg] [deg]
42 0 0 0 7 0
43 30 0 0 7 0
44 30 0 0 -2 0
Drag meas No prop
45 40 0 0 -2 0
46 40 0 0 7 0
Prop. change Co-rotating 0
47 0 0 0 -2 0
Blank Co-rotating
48 0 0 0 7 0
49 40 93.7 2.10 7 0
50 40 106.4 1.85 7 0
51 40 93.7 2.10 -2 0
52 40 123.0 1.60 -2 0
53 30 70.3 2.10 -2 0
54 30 92.3 1.60 -2 0
Meas. Set 4 Co-rotating 55 30 70.3 2.10 7 0
56 30 79.8 1.85 7 0
57 30 92.3 1.60 7 0
58 30 79.8 1.85 7 0
59 40 106.4 1.85 -2 0
60 40 123.0 1.60 7 0
61 30 79.8 1.85 -2 0
Prop change Inboard up 0 0 0 0 0
62 0 0 0 7 0
Blank Inboard up
63 0 0 0 -2 0
64 40 93.7 2.10 -2 0
65 40 106.4 1.85 7 0
66 40 123.0 1.60 7 0
67 40 123.0 1.60 -2 0
68 40 106.4 1.85 -2 0
69 40 93.7 2.10 7 0
Meas. Set 5 Inboard up
70 30 92.3 1.60 -2 0
71 30 79.8 1.85 -2 0
72 30 70.3 2.10 7 0
73 30 79.8 1.85 7 0
74 30 92.3 1.60 7 0
75 30 70.3 2.10 -2 0
δe change 3 Inboard up 0 0 0 -2 8
76 40 106.4 1.85 7 8
77 40 106.4 1.85 -2 8
78 30 79.8 1.85 -2 8
79 30 92.3 1.60 7 8
80 30 70.3 2.10 -2 8
81 30 92.3 1.60 -2 8
Meas. Set 6 Inboard up
82 30 79.8 1.85 7 8
83 30 70.3 2.10 7 8
84 40 123.0 1.60 -2 8
85 40 93.7 2.10 -2 8
86 40 123.0 1.60 7 8

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 ii

 A FULL TEST MATRIX

Measurement Propeller DPN V∞ RPS J α δe

Type Configuration [m/s] [−] [−] [deg] [deg]
87 40 93.7 2.10 7 8
δe change 4 Inboard up 0 0 0 -2 8
88 40 123.0 1.60 -2 -8
89 40 93.7 2.10 -2 -8
90 40 106.4 1.85 7 -8
91 30 70.3 2.10 7 -8
92 30 79.8 1.85 -2 -8
93 30 70.3 2.10 -2 -8
Meas. Set 7 Inboard up
94 30 92.3 1.60 -2 -8
95 30 92.3 1.60 7 -8
96 30 79.8 1.85 7 -8
97 40 93.7 2.10 7 -8
98 40 106.4 1.85 -2 -8
99 40 123.0 1.60 7 -8
100 0 0 0 7 -8
Blank Inboard up
101 0 0 0 -2 -8
Prop change No prop. -8
102 0 0 0 -2 -8
Blank No prop.
103 0 0 0 7 -8
104 30 0 0 7 -8
105 30 0 0 -2 -8
Drag meas. 2 No prop.
106 40 0 0 -2 -8
107 40 0 0 7 -8
δe change 5 No prop. 0 0 0 7 8
108 0 0 0 -2 8
Blank No prop.
109 0 0 0 7 8
110 30 0 0 7 8
111 30 0 0 -2 8
Drag meas. 3 No prop.
112 40 0 0 -2 8
113 40 0 0 7 8

Table 5: Full test matrix

AE4115 - Experimental Simulations: Post-Test Report 2021 - Group 15 iii