You are on page 1of 19

Wind Tunnel Tests on a

Model Cessna

Akash Trivedi
CID: 00637814
Personal Tutor: Dr. P. Robinson
Submission Date: 02/05/2011

In this experiment, wind tunnel tests were carried out on a scale model of a Cessna 172 at two
different speeds. The lift and drag forces were obtained, several characteristic graphs plotted,
induced drag deduced and stalling speed inferred for a full scale model. It was concluded that
testing models in a wind tunnel is a very important and fundamental method of determining
flow regimes and aircraft performance.
Contents
1. Introduction Page 3

2. Objectives Page 3

3. Theory Page 3

4. Apparatus Page 4

5. Experimental Procedure Page 5

6. Results Page 6

7. Discussion ______ Page 10

8. Conclusion Page 13

9. References Page 14

10. Appendices Page 15

2|Page
1. Introduction

Ever since their conception in 1871 by Frank Wenham [1], wind tunnels have proved to be a
fundamental experimental tool in the field of aeronautics. The famous experiments by
Osbourne Reynolds would not have been possible without the use of a wind tunnel.

In this experiment, a closed loop wind tunnel will be used to obtain measurements for the lift
and drag of a small model Cessna 172 at two different speeds and over a range of incidences.

2. Objectives

The aims of this experiment are to:

• Derive the lift and drag forces at two speeds for a range of incidences.
• Plot graphs of certain non dimensional quantities characteristic of the flow around the
model aircraft, and then discuss the forms of the curves obtained.
• Apply wind tunnel data to a full scale aircraft to determine its stall speed.

3. Theory

There are many forces that act on an aircraft in flight, but they can be summarised into four
categories: lift, drag, thrust and weight. In this lab, the lift and drag forces will be focused
upon.

1 2
The lift force is given by L = u SCL , where  is the density, u is the true air speed, S is
2
the wing area and C L is the lift coefficient that is a function of the angle of attack.

1 2
The drag force is given by D = u SCD , where C D is the drag coefficient which is
2
dependent on:

• Wave drag – a drag force that exists above a critical Mach number. In this lab, the
wave drag is non existent due to the low speeds of the wind tunnel.
• Friction drag – depends on the air viscosity and accounts for most of the lift
independent drag. It is often referred to as
the zero-lift drag.
• Profile drag – drag arising due to pressure.
• Lift-dependent drag – is a function of angle
of attack, the wing aspect ratio and is also
influenced by geometrical factors.

As the wing of an aircraft is of a finite length, due to


pressure differences, a vortex is created at the wing
tip and this produces a downwash, w, over the wing
surface. As the net lift acts perpendicular to the
resultant of the free stream velocity, v , and the
Figure 1: Illustrating induced drag.
downward induced velocity, w , it is tilted

3|Page
w
backwards through an angle  . This in turn induced a backwards force, Di , which
v
contributes a certain fraction of the lift coefficient to the drag coefficient. However the
downwash itself is proportional to the lift coefficient and so the induced drag coefficient C D ,i
is proportional to C L2 . (Figure 1 illustrates this theory of induced drag).

kCL2
In this lab, C D = C D , 0 +where C D , 0 is the drag coefficient at zero lift, k is a measure of
AR
the wing’s aerodynamic efficiency and AR refers to the aspect ratio of the wing.

For these forces to be modelled using a wind tunnel, the conditions in the wind tunnel must
be similar to those on the real aircraft. It is not good enough to have a perfect scale model of
the other; this just satisfies geometric similarity. The flows must also be dynamically similar
and this involves a non dimensional parameter called the Reynolds number.

The Reynolds number can be thought of as the ratio of inertial forces to viscous forces, such
uD
that the Reynolds number is given by Re = where D is the characteristic length (the

wing root chord in this lab) and  is the dynamic viscosity.

When the Reynolds number (along with Mach number and incidence) is the same for the
model as it is for the aircraft, it is considered dynamically similar. Only then will the lift and
drag coefficients be properly modelled in the wind tunnel.

4. Apparatus

The wind tunnel that was used


was a closed circuit type in which
air is circulated continuously.
Figure 2 depicts how the fan
blows the air around the turning
vanes and through the contraction
into the test section, where the
flow in relatively uniform due to
the presence of gauzes and
honeycomb shaped grates. These
help to reduce turbulent
conditions and vortices. The flow
of air is then returned back to the
fan which only has to accelerate
the flow so that the losses due to
friction at the wall are overcome.
This means that this type of wind Figure 2: Schematic of a closed circuit wind tunnel. [2]
tunnel is desirable due to its low
operating costs.

4|Page
A 1/20th scale model of a Cessna 172 is mounted on a pivot atop a streamlined strut, such that
movement of a wire attached to the tail, allows the incidence of the model to be varied from
-2.0o to 13.0o. Wool tufts are attached to the model to allow simple flow visualisation.

The required forces on the model are measured by three calibrated strain gauges; two on the
model itself and one the tail wire. An average of two gauges provides the lift readings and the
other gives the drag. The dynamic pressure (in mm of water) is measured via a pitot static
tube coupled to a digital manometer. Finally, a temperature and total pressure probe is used to
measure the respective quantities.

Corrections must be made to the above readings to ensure dynamic similarity. These are
made in a spreadsheet and the raw data is given in appendices. The smaller the model is, the
greater the corrections that need to be made and hence the greater the errors. However, if the
model was made bigger, then it would be too big for the wind tunnel and the wings would be
too close to the wall and so the aerodynamic forces would be incorrectly measured.

5. Experimental Procedure

The procedure for this lab is as follows.

1. With the model incidence at 0o and the wind speed at zero, the readings on the three
strain gauge channels are zeroed.
2. The wind tunnel is first set up to 75% power.
3. Readings are taken for all three strain gauges at model incidences from -2.0o to 13.0o
and consequently tabulated in a spreadsheet.
4. Steps 1 to 3 are repeated at 50% power.

Possible errors:

• The incidences read on the meter for the model may not be the actual incidence for
the model in the tunnel.
• Corrective errors in the spreadsheet provided, due to greater variations than thought in
the tunnel compared to free air.
• These errors lead to incorrect measurements of lift and drag forces, as the air flow in
the tunnel is misrepresentative of free air flow.

5|Page
6. Results

75% power:

1.2

0.8

0.6
Lift Coefficient

0.4

0.2

0
-5 0 5 10 15
-0.2

-0.4

-0.6
Incidence (deg)
Figure 3: CL against incidence at 75% power.

Zero incidence corresponds closely to zero lift as is evident from Figure 3; the x and y
intercept are very close to the origin. There appears to be a linear response at low incidences.
Below around 1o of incidence, there is a negative lift coefficient hence a downward force on
the model. Following a linear increase in lift coefficient with incidence, there is a point
(around 12o of incidence) after which an increase in incidence does not result in an increase
of lift. The coordinates of the peak of Figure 3 correspond to the stalling incidence and lift
coefficient. Increasing the incidence further results in a drop in lift.

0.3

0.25

0.2
Drag Coefficient

0.15

0.1

0.05

0
-4 -2 0 2 4 6 8 10 12 14 16
Incidence (deg)

Figure 4: CD against incidence at 75% power.

6|Page
At low incidences, lift dependent drag is negligible but contribution is made from the other
two drag forms; profile and friction drag. This is as the model is pointed downwards slightly
and a negative lift force is generated. In the mid-section of Figure 4, the drag coefficient is
the lowest with a minimum of 0.108. This occurs as the effect of profile drag decreases but
the lift dependent drag increases. However, as incidence increases up to the stalling incidence
and further, there is a large contribution to the drag coefficient from lift dependent drag and
profile drag as the model is pointing up at a large angle.

0.3

0.25

0.2
Drag Coefficient

0.15

0.1

0.05

0
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Lift Coefficient
Figure 5: CD against CL at 75% power.

Figure 5 is often referred to as the drag polar; theoretically a parabola with its axis on the
zero lift axis (CL=0) and its vertex is CD,0. CD,0 is not the minimum drag coefficient in the
experimental curve, because with most aircraft designs, the zero lift incidence is a small
negative value. This indicates that the aircraft is pitched down. [3] Here, CD,min = 0.108 and
CL,min drag = 0.346.

Figure 5 has a similar appearance to Figure 4 due to the initial linear response of CL to
incidence. The difference is only noticeable at the highest recorded incidences. This is
because CL-incidence response becomes non linear and CL decreases. Hence, on Figure 5, the
curve appears to bend backwards.

7|Page
Figure 6: Drag-to-lift ratio against CL at 75% power.

Figure 6 has a discontinuity at CL = 0 as the lift would be zero at this point and thus the drag-
to-lift ratio would be infinite. This means that when approaching CL = 0 from a negative lift
coefficient, as the lift would be approaching zero, the drag-to-lift ratio tends to negative
infinity. However, when approaching CL = 0 from a positive lift coefficient, the lift force still
approaches zero but the drag-to-lift ratio tends to positive infinity.

This means that the larger the magnitude of the lift coefficient, the smaller the drag-to-lift
ratio would be as the graph in Figure 6 tends to zero. Therefore, a large lift coefficient infers
a large lift-to-drag ratio.

50% power:

1.2

0.8

0.6
Lift Coefficient

0.4

0.2

0
-5 0 5 10 15
-0.2

-0.4

-0.6
Incidence (deg)

Figure 7: CL against incidence at 50% power.

The form of the curve in Figure 7 is very similar to that of Figure 3. The difference is that
when power of the fan is dropped from 75% to 50%, the effect is that flow speed is reduced.
This means that the maximum lift coefficient has dropped 5.1%.

8|Page
0.3

0.25

0.2
Drag Coefficient
0.15

0.1

0.05

0
-5 0 5 10 15
Incidence (deg)

Figure 8: CD against incidence at 50% power.

Figure 8 is very similar to Figure 4 in terms of the shape of the curve. The response of drag
coefficient to incidence is largely the same. However, the minimum value for the drag
coefficient has increased by 13.9%.

0.3

0.25

0.2
Drag Coefficient

0.15

0.1

0.05

0
-0.4 0.1 0.6 1.1
Lift Coefficient

Figure 9: CD against CL at 50% power.

Once again, the form of the drag polar for 50% power shown in Figure 9 is very similar to the
drag polar at 75% power shown in Figure 5. At 50% power, CD,min = 0.123 and CL,min drag =
0.364. This corresponds to a 13.9% increase in CD,min and a 5.2% increase in CL,min drag when
the power is reduced from 75% to 50%.

9|Page
Figure 10: Drag-to-lift ratio against CL at 50% power.
Figure 10 is very similar in appearance to Figure 6. This is as the same argument applies; as
the lift coefficient approaches zero from either side of the origin, the drag-to-lift ratio
increases heavily as it tends to infinity. This again implies that as the magnitude of the lift
coefficient decreases, there is a greater amount of drag forces than there are lift forces.
Conversely, a large magnitude for the lift coefficient indicates a low drag-to-lift ratio hence a
large lift-to-drag ratio.

7. Discussion

Test speeds:

The speed in the test section can be measured by first calculating the pressure difference
between two static tappings in the contraction walls. Calibration for this experiment is such
that the dynamic pressure q  is given by q = 1.1( p1 − p2 ) where p1  p2 .
1 2q 
As q  =   u 2 , the speed is then given by u  = .
2 
q  varies slightly depending on the incidence so the speed at each incidence is worked out
and then the average is taken for the test speed.

• For 75% power: Test speed = 27.2 m/s.


• For 50% power: Test speed = 17.9 m/s.

Reynolds numbers:

uD
The Reynolds number is given by Re = .  is obtained from a Fortin barometer to be

1.166 kg/m3, the speed differs in accordance to the power as illustrated above, the
characteristic length is the wing root chord of the model, 80 mm and based on the operating
temperature, the dynamic viscosity is found to be 1.8272 x10-5 kg/(ms) [4]. Using this
information, the Reynolds number for each test speed can be determined.

• For 75% power: Re = 139000.


• For 50% power: Re = 91400.

10 | P a g e
Stalling speed of a full scale aircraft:

1 2 2L
Rearranging L = u SCL for the speed gives u = .
2 SCL
Assuming lift coefficient is at its maximum for 75% power and that at the point of stalling the
2mg
lift force is balanced by the weight: u = .
SC L ,max

This gives a stalling speed of the full scale aircraft to be 31.8 m/s. However, the actual
stalling speed is given to be 26 m/s. This means that the calculation based on data obtained
from the experiment gives a 22.3% higher stalling speed than it should.

As the weight is considered constant, it has to be argued that the denominator of the speed
expression must be higher in order to give a lower result. The wing area is also constant and
according to the ISA, it cannot be argued that the air density increases from ground level to
where the aircraft is flying, therefore it is only reasonable that the recorded value for CL,max in
this experiment should be higher in order to achieve a more accurate answer with a lower
percentage error.

For a NACA 2412 aerofoil that a Cessna 172 employs [5], the CL,max was determined to be
around 1.6 [6]. If this is the case, then the stalling speed is in fact 25.98 m/s, very close to the
given stalling speed, with negligible percentage difference.

A possible reason for this could be the fact that the model may be too large for the wind
tunnel. This would mean that the flow around the wingtips may actually be adversely affected
by the boundary layer of the tunnel walls. By inducing turbulent flow regimes in these
affected areas, the lift forces would be determined to be lower than should be. This would
result in a lower measured maximum lift coefficient.

Induced drag:

If a graph is plotted of CD against CL2, analogies can be made to the y = mx + c graph. This is
kC 2
as the equation for the drag polar is C D = C D , 0 + L . C D , 0 would correspond to the y
AR
k
intercept and the term, the gradient. The induced drag coefficient can be calculated by
AR
multiplying the gradient of the linear portion of the CD against CL2 graphs with the square of
the maximum lift coefficient. Thus the induced drag force can be deduced.

11 | P a g e
For 75% power:

• From theory: working at CL,max and assuming an elliptic distribution of lift and where
C2 1
k = 1, CD,i = L = 0.0483. As Di = u 2 SCD,i , the induced drag force can be worked
AR 2
out by using an average of the dynamic pressures (432.13 Pa) and the gross wing area
(0.038 m2) of the model. This results in an induced drag force of 0.793 N.

0.3

0.25

0.2
Drag Coefficient

0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8 1 1.2
Square of Lift Coefficient

Figure 11: CD against CL2 at 75% power.

• From graph in Figure 11: working at CL,max and using the 4th to 7th points inclusive,
CD,i = 0.0521. Using the same method as above, the induced drag force is 0.855 N.

The induced drag obtained from the experimental data is 7.9% higher than the theoretical
value. This means that the assumptions made for the theoretical value were good enough for a
fair approximation.

12 | P a g e
For 50% power:

• From theory: As above, working at CL,max and assuming an elliptic distribution of lift
and where k = 1, CD,i = 0.0434. This results in an induced drag force of 0.307 N.

0.3

0.25

0.2
Drag Coefficient

0.15

0.1

0.05

0
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
Square of Lift Coefficient

Figure 12: CD against CL2 at 50% power.

• From graph in Figure 12: working at CL,max and using the 4th to 7th points inclusive,
CD,i = 0.0456. This infers an induced drag force of 0.323 N.

The induced drag inferred from the experimental data is 4.94% higher than the theoretical
value. As before, this means that the assumptions made were valid for a decent
approximation.

Wool tuft observations

Figure 13: Cessna 172 model at 2o incidence.

13 | P a g e
At 2o incidence, CL is relatively low and so the induced drag is also low. This is evident from
the wool tufts (Figure 13) as only the tufts at the wingtip are disturbed by the vortices, with
the rest of the tufts along the wing appearing parallel to the flow. This indicates a small
downwash and thus a relatively low induced drag component.

Figure 14: Cessna 172 model at 10o incidence.

At 10o incidence, the downwash would be larger as the lift generated is higher (as the CL is
higher). This is conveyed by the erratic motion of the wool tufts (Figure 14). There is a larger
disturbance at the wing tip corresponding to larger vortices, and hence larger downwash and
therefore a higher induced drag. The tufts along the rest of the wingspan also appear to be
more affected by the higher incidence than those in Figure 13.

8. Conclusion

In this experiment alone, by analysing strain gauge responses as the incidence changed, it was
possible to obtain information regarding lift and drag forces, their non dimensional
coefficients, stalling incidence, stalling speed of a full scale model, induced drag forces and
much more. This illustrates the importance of utilising a wind tunnel to perform test on a
model. The tests can be used to infer many different results for the aircraft performance of a
full scale aircraft.

14 | P a g e
9. References

[1] National Aeronautics and Space Administration (2009) Whirling Arms and the First Wind
Tunnels [Online] Available from: http://www.grc.nasa.gov/WWW/k12/WindTunnel/
history.html [Accessed: 29/03/11]

[2] National Aeronautics and Space Administration (2009) Closed return wind tunnel
[Online: Image] Available from: http://www.grc.nasa.gov/WWW/K-12/airplane/tuncret.html
[Accessed: 30/03/11]

[3] Anderson, J. D., Jr. (2005) Introduction to Flight. International Edition. Singapore,
McGraw-Hill.

[4] University of Waterloo (1997) Fluid properties calculator [Online] Available from:
http://www.mhtl.uwaterloo.ca/old/onlinetools/airprop/airprop.html [Accessed: 03/04/11]

[5] McIver, J. (2003) Cessna Skyhawk: Performance Assessment [Online: pdf] Temporal
Images (Original source: Jane’s all the World’s Aircraft) Available from:
http://www.temporal.com.au/c172.pdf [Accessed: 03/04/11]

[6] Seetharam, H. C.; Rodgers, E. J.; Wentz, W. H., Jr. (1977) Experimental Studies of Flow
Separation of the NACA 2412 Airfoil at Low Speeds. [Online: pdf] National Aeronautics and
Space Administration, Langley Research Center. Available from: http://ntrs.nasa.gov/archive/
nasa/casi.ntrs.nasa.gov/ 19950002355_1995102355.pdf [Accessed: 03/04/11]

15 | P a g e
10. Appendices

Appendix 1: Data for 75% power

dynamic free-air incidence free-air


pressure (q) induced
drag
correction
INCIDENCE LIFT LIFT DRAG(1) PRESSURE L1 D1 PRESSURE Incidence (corr) Cd (tunnel)
(deg) (2) (3) (mm.H20) (Pa)
NO WIND -1 -57 -8 N/A 0.00
-2 -140 -108 310 53.74 -190 318 432.15 -2.329466209 0.001900119
0 -6 -104 280 53.74 -52 288 432.15 -0.102724571 0.000184717
2 150 -120 240 53.34 88 248 428.93 2.128859869 0.000290666
4 264 -101 230 53.8 221 238 432.63 4.344860528 0.002081833
6 408 -95 240 53.88 371 248 433.27 6.58751147 0.006042152
8 515 -90 256 53.86 483 264 433.11 8.768826841 0.010347037
10 600 -70 283 53.58 588 291 430.86 10.94275857 0.01555822
12 670 -58 358 53.95 670 366 433.84 13.06472259 0.019844121
13 630 -40 543 53.75 648 551 432.23 14.02128744 0.018258072

16 | P a g e
Efficiencies
Velocity Lift (N) Drag (N) Cl Cd D/L L/D Cl^2
(m/s)
0.00
27.22 -5.42665 2.3962280 -0.33046 0.147819 -0.44157 -2.26466 0.109202
27.22 -1.69198 2.1549533 -0.10303 0.131411 -1.27363 -0.78516 0.010616
27.12 2.106656 1.838829064 0.129248 0.113107 0.87287 1.14565 0.016705
27.24 5.686548 1.747379804 0.345898 0.10837 0.30728 3.25433 0.119646
27.26 9.702126 1.802968309 0.589279 0.115549 0.18583 5.38120 0.34725
27.25 12.69164 1.908129622 0.77114 0.126284 0.15035 6.65135 0.594657
27.18 15.48197 2.09623352 0.945595 0.14359 0.13540 7.38561 0.894151
27.28 17.6056 2.645312563 1.067926 0.180304 0.15025 6.65540 1.140467
27.23 16.82478 4.028125811 1.024361 0.263507 0.23942 4.17683 1.049314

17 | P a g e
Appendix 2: Data for 50% power

dynamic free-air incidence free-air


pressure (q) induced
drag
correction
INCIDENCE LIFT LIFT DRAG PRESSURE L1 D1 PRESSURE Incidence (corr) Cd (tunnel)
(deg) (2) (3) 1 (mm.H20) (Pa)
NO WIND -1 -57 -8 N/A 0.00
-2 -74 -78 138 22.83 -94 146 183.59 -2.382021243 0.002554665
0 -3 -76 121 23 -21 129 184.95 -0.098783081 0.000170814
2 62 -77 110 22.96 43 118 184.63 2.14674513 0.000376952
4 118 -76 108 23.08 100 116 185.60 4.362537104 0.00230072
6 165 -73 107 23.08 150 115 185.60 6.55238785 0.005341302
8 219 -73 113 23.21 204 121 186.64 8.752160188 0.009903293
10 258 -66 130 23.32 250 138 187.53 10.91889756 0.014780636
12 275 -55 180 23.48 278 188 188.81 13.00991193 0.017853605
13 252 -52 235 23.51 258 243 189.05 13.92627498 0.015018923

18 | P a g e
Efficiencies
Velocity Lift (N) Drag (N) Cl Cd D/L L/D Cl^2
(m/s)
0.00
17.74 -2.67311 1.1010125 -0.38317 0.160377 -0.41188 -2.42786 0.14682
17.81 -0.69636 0.9649491 -0.09908 0.137467 -1.38571 -0.72165 0.009817
17.79 1.032663 0.874783665 0.147187 0.125061 0.84711 1.18048 0.021664
17.84 2.56455 0.852641633 0.363628 0.123197 0.33247 3.00777 0.132225
17.84 3.907535 0.838846213 0.55405 0.124281 0.21467 4.65823 0.306971
17.89 5.350673 0.876761109 0.754423 0.133523 0.16386 6.10277 0.569155
17.93 6.56778 0.997745305 0.921663 0.154795 0.15192 6.58262 0.849462
17.99 7.267827 1.367177105 1.012951 0.208403 0.18811 5.31594 1.026069
18.01 6.674451 1.779989282 0.929062 0.262788 0.26669 3.74971 0.863157

19 | P a g e

You might also like