aerodynamic optimization of an uav design

American Institute of Aeronautics and Astronautics

1

Aerodynamic Optimization of an UAV Design

Pedro J. Boschetti *

Universidad Simón Bolívar, Caracas, 89000-1080, Venezuela

Elsa M. Cárdenas †

Universidad Nacional Experimental Politécnica de la Fuerza Armada, Caracas, 1060, Venezuela

and

Andrea Amerio ‡

Universidad Simón Bolívar, Caracas, 89000-1080, Venezuela

The Maracaibo Lake, Venezuela, is an important petroleum extraction region and

besides it is a source of constant pollution. However, the early detection of the oil leakages

minimizes the environment impact. In 2003 an UAV for the special mission of patrolling that

region in search for oil leakages was designed. The purpose of this research is to optimize the

aerodynamic characteristics of the initial design. The general methodology was to evaluate

the drag coefficient and the lift coefficient of the design by theoretical, and experimental

ways, making modifications in critical parts, such as the landing gear and wing tips, and

later to evaluate whether these modifications actually improved the aerodynamic efficiency.

The experimental study consists of several tests in a small wind tunnel, using 1:20.2 scale

models. Polar curves of design and later modifications were traced, obtaining the

aerodynamic efficiency for cruise flight is better in the last optimized version than in the

original design. Finally, it is possible to conclude that with a few modifications over a design,

the aerodynamic performance of an aircraft can be changed and these may be studied using

simple tools.

Nomenclature

A = reference area

A1 = reference area 1

A2 = reference area 2

Af1 = lateral area of the fuselage up and down of the wing

Af2 = frontal area of the fuselage

c = camber of the airfoil

CD = drag coefficient of the airplane

CDcs = drag coefficient due to secondary components

CDf = drag coefficient of the fuselage

CDh = profile drag coefficient of the horizontal tail

CDi = induced drag coefficient

CDint = drag coefficient due to interaction

CDint cs = drag coefficient due to interaction between secondary components

CDp = profile drag coefficient of the airplane

CDpw = profile drag coefficient of the wing

CDv = profile drag coefficient of the vertical tail

CDw = drag coefficient of the wing

Cf = friction coefficient

* Graduate Research Student, Direction of Investigation, 1080, Valle de Sartenejas, AIAA Student Member.

† Instructor Professor, Department of Aeronautical Engineering, 1060 Av. La Estancia, Chuao.

‡ Assistant Professor, Department of Industrial Technology, 1080, Valle de Sartenejas.

AIAA 5th Aviation, Technology, Integration, and Operations Conference (ATIO) <br>26 - 28 September 2005, Arlington, Virginia

AIAA 2005-7399

Copyright © 2005 by the Authors. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


2

CL = lift coefficient

cl = airfoil section lift coefficient

CLw = wing lift coefficient

cl,α = airfoil section lift slope

CL/CD = aerodynamic efficiency

D = aerodynamic drag

df = frontal average diameter of fuselage

Di = induced drag

Dp = profile drag

lf = length of fuselage

RA = wing aspect ratio

Re = Reynolds number

Ref = fuselage Reynolds number

RT = wing taper ratio

Rew = wing Reynolds number

V = airspeed of the freestream

t = airfoil maximum thickness

Sh = horizontal stabilizers surface

Sv = vertical stabilizer surface

Sw = wing surface

α = geometric angle of attack relative to the freestream

ρ = air density

∆CD = drag coefficient difference

κD = platform contribution to the induced drag factor

Ω = total twist, geometric plus aerodynamic

Ωopt = optimum total twist to minimize induced drag

I. Introduction

INCE the second decade of the twentieth century, petroleum has been exploited in the Lake Maracaibo basin in

Venezuela. Over the years, the continuous oil leakages from offshore facilities, and transporting pipelines has

deteriorated the delicate ecosystem of this region. For this reason, Petroleos de Venezuela, S.A. carries out daily

manned helicopter flights over this area in search of possible petroleum leakages, in order to take early measures to

prevent disasters.

Looking for more economical alternatives with regard to operation and maintenance costs, which would

additionally permit the surveillance of this area during day and night, as well as through adverse climatic conditions

for manned aircrafts, in the year 2002 the design of a small monoplane unmanned aircraft, capable of accomplishing

this mission, was initiated as a joint project of the Universidad Nacional Experimental de la Fuerza Armada

(UNEFA,) and Universidad Simón Bolívar (USB). The preliminary design was finished in May 2003, using a design

methodology based on analytical and statistical estimates and calculations.1

The design of the Unmanned Aircraft for Ecological Conservation (ANCE, for its Spanish acronym,) has a

monoplane twin-boon pusher configuration. Powered by a twin-blade propeller of 0.915 m coupled to a linear

arranged two-piston motor, two strokes, air cooled, with 26 kW maximum power at 3500 rpm, and using 92 octane

gasoline. The estimated gross weight is 182 kg, with a payload of 40 kg which it is believed will consist of infrared

equipment capable of detecting pollutant substances on the water surface. The length of the airplane is 4.65 m; its

wingspan is 5.18 m, with a straight rectangular wing of 2.89 m2 of surface formed by a NACA 4415 lifting section.

Figure 1 shows the sketch of ANCE. It is expected that it will have a cruise speed of 46.77 m/s, with a stall speed of

28.66 m/s at sea level, and a service ceiling of 3600 m, and that it would be capable of remaining up to 11 hours in

flight. Due to the fact that the area where the plain is expected to operate is a lake, and that it is surrounded by level

ground, it is expected that it will be capable of taking off from roughly prepared airstrips of approximately 377 m,

and to land on approximately 428 m, by means of its fixed tricycle landing gear. See Ref. 2 for more complete

information.

As part of the design process of this UAV, after the first sketch was finished, it was deemed necessary to study

thoroughly the aerodynamic characteristics of the design using methods more in line with reality, seeking to improve

as far as possible these characteristics, in order to make the design aerodynamically more efficient. Therefore, the

purpose of this investigation is to optimize the aerodynamic characteristics for cruise flight of the initial design,

S

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


3

using analytical, experimental and numerical tools. This work shows step by step the optimization process and

methods used to determine the improvement degree of each one.

II. Aerodynamic Forces

It is a well known fact that an airplane in straight and leveled flight is subjected to four basic forces that make it

keep its balance. Initially there is the weight, originated by the attraction between the aircraft mass and the ground,

and it must be equal to lift in order to maintain balance at the vertical axis. Then, there is the thrust produced by the

engine or engines which in this case is equal to drag or the resistance to advancing, generated by the interaction

between the air and the vehicle in movement. At the same time, the resistance has a certain component derived from

lift, which makes the balance of an airplane a case for study. For this purpose it is necessary fully to understand each

one of these forces.

Lift and drag are considered aerodynamic forces because they are produced by the passing of the airplane

through the air mass and, even though the thrust may be generated by means that are characteristically aerodynamic

or aerothermodynamic, it is not considered an aerodynamic force belonging to the geometry of the aircraft.

III. Aerodynamic Drag

The aerodynamic drag in an airplane may be derived from the tangential actions of fluid reactions on the external

skin, called friction drag, and from the pressure component of the asymptotic velocity resulting from the actions

produced over the body, called pressure drag. This at the same time is divided in stream drag, wave drag, and

induced drag; the latter caused by the vortices emerging from the tips of the wing, and is a function of the lift.3

The sum of the friction drag, the stream drag, and the wave drag is called profile drag, which is not related to lift.

In conclusion, as shown in Eq. (1)4 drag is the sum of the profile drag and the induced drag

ip DDD += (1)

These forces may be converted to dimensionless when multiplied by 2/AρV2, A being a reference area which in

this case is the wing surface. In this way the Eq. (2) is obtained.

DiDpD CCC += (2)

The induced drag coefficient for a straight wing may be calculated Eq. (3),5,6

This shows the relation that the

induced drag has with the wing aspect ratio, the lift coefficient and the twists of the wing.

( )

22

12

+⋅

⋅⋅−⋅

⋅+

⋅=

T

,lL

A

D

A

LDi

R

cC

RR

CC

Ωπ

π

κ

π

α (3)

It is quite difficult to calculate exactly the profile drag due to the complex forms of aircrafts, the multiple

components they have and the different flow conditions to which they are subjected. The best option in most cases is

the aerodynamic testing in wind tunnels. This provides a great deal of information on aerodynamic performance in

relatively short time periods, in comparison with the computational fluid dynamic methods.7 However, these tests

involve energy and equipment maintenance costs which make them very expensive in most cases. Therefore, the

theoretical estimate of an aircraft drag, although inaccurate, is a good approximation in initial cases, and even more

when dealing with aircraft of geometric characteristics and flow conditions similar to other existing models.

IV. Aerodynamic Optimization Process

The aerodynamic optimization process consisted in improving efficiency as far as possible through drag

diminishing. For this purpose, the drag coefficient and the lift coefficient of the original design was determined by

theoretical and experimental means, which will be explained later.

Following an extensive bibliographical review, it was determined that for subsonic aircraft the viscid drag must

be diminished (in the subsonic case, equal to the profile drag) by controlling the laminar flow and the drag due to

lift.8

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


4

In order to reduce the viscous drag it is necessary to make emphasis on some aspects of the airplane, such as the

installation of the propulsion plant, air escape, installation of the landing gear, installation of antenna and other

external devices.9

With respect to the airplane’s powerplant, no changes were made in it. It is cooled with air, and has its air inlet

and, air exhaust at each side of its pistons. The careless placement of fairings might cause overheating, exhaustion

or, the stagnation of gasses produced by combustion. It remained with its pistons, air inlet and, air exhaust, subject

to the free stream of air. The recommendation in Ref. 9 was followed and, an elongated conical form spinner was

placed.

In order to reduce the drag produced by the landing gear it is covered with fairings. The supports are covered

with fairings with a section formed by NACA 0015 airfoil, and the upper portion of the pneumatic tires with a

section each one in the form of a drop.

With respect to the observation camera that is located in the lower part of the fuselage, it must have a 360

degrees visibility, making it impossible to place fairings around it. The antennae are located inside the fuselage

under the main cargo door, and they cause no problem with regard to the drag.

After making these modifications this model was called X-2, and the previous one X-1. In Fig. 2; both models

are shown from port side view in a comparative way.

The reduction of the induced drag has been a topic for study, and a great variety of mechanisms have been

created to reduce it.8 However, in the case of ANCE the best option is that suggested in Ref. 5, which consists in

making a rectangular wing to have the same efficiency of an elliptic wing by means of a simple geometric and/or

aerodynamic twist. One of its principal advantages is its simple construction, which does not significantly increase

the weight, nor adds relevant aerodynamic loads to the wing structure.

The optimum twist angle may be estimated by means of Eq. (4),6 giving as a result 7.38 degrees of twist with

respect to the mean aerodynamic chord.

( )

απΩ

,l

LT

c

CR

⋅

⋅+⋅=

12opt (4)

To apply this to the ANCE wing design, it was decided to add only aerodynamic twist, using the same airfoil of

the wing at the tip, but with flap up at 80% of the chord. This was made to avoid modifications in the projected

airplane structure,2 due to the fact that the wing rear spar is located at 80% of the chord. The twist only covers

stations 2.32 to 2.4 m of the midspan, thus avoiding a large variation in the total lift coefficient.

In order to determine its lift coefficient, at an angle of attack equal to the optimum twist angle necessary to

reduce the induced drag, the airfoil section was tested at the same Reynolds number of flight of the airplane (Re =

1388022), using the 4.1 version of the airfoil analysis computational code VisualFoil.10

This is a numerical tool used

to compute the lift coefficient, the drag coefficient, and the moment coefficient for NACA airfoil sections of four

and five digits, and its analysis is based on the vortex panel method for an ideal incompressible flow.

Then the airfoil with flap at 80% of the chord was tested, until determining one that had the same lift coefficient

as that resulting from the simple profile at Ωopt. The resulting flap deflection is 13.8 degrees upwards.

The modified design with the modified wing tips, beside the modifications of model X-2, was called X-3, and

Fig. 2 shows in port side view, in comparison with the other models. Figure 3 shows the wing tip three-dimensional

detail. Both design modifications were evaluated using theoretical and experimental ways.

V. Theoretical Estimation of the Aerodynamic Drag

In order to make an appropriate theoretical calculation of the aerodynamic drag, it is necessary to analyze its

causes. All the components of the airplane generate drag when tested separately, and the sum of these, plus the drag

caused by the interaction between components is the total drag.

In the case of the airplane under study, it may be said that the drag produced by the fuselage, the wing, the

horizontal and vertical stabilizers, plus the drag produced by secondary components, such as those of the landing

gear, the observation camera, and the engine radiator, and the drag produced by interaction, generate the total drag

of the airplane. The factors involved in the estimation of the drag coefficient of the airplane may be observed in Eq.

(5).

intDDcsDvDhDwDfD CCCCCCC +++++= (5)

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


5

The fuselage drag coefficient is calculated using Eq. (6), which is an approximation of experimental results

realized at high Reynolds numbers.11,12

⋅+⋅+⋅⋅=

2

2154300030f

f

f

f

f

f

Dfl

d

l

d.

d

l.C (6)

In the particular case where the fuselage is not of circular section, the frontal diameter average of the fuselage is

equal to 2π-1/2

Af21/2

. The reference area of the calculated coefficient is the frontal area, which is the reason why the

drag coefficient obtained must be multiplied by the expression 4-1

π df2 Sw

-1 to be used in the Eq. (5).

The drag coefficient of the wing corresponds to Eq. (7); this is only valid when the airplane is on level flight

where the lift generated by the vertical and horizontal stabilizers is insignificant, and all the induced drag is

produced by the wing.

DiDpwDw CCC += (7)

The profile drag coefficient of the wing and other surfaces, such as the horizontal and vertical stabilizers,

produced when the lift from these is zero, is estimated by adding the pressure drag coefficient produced by friction

on the surface and the pressure drag coefficient produced by the airfoil, as may be seen on Eq. (8).11

( ) ( )

⋅+⋅+⋅=

4602212

ct

ct.CC fDpw (8)

The estimated friction coefficient is equal to 1.33/Re0.5

for laminar regime,13

and 0.455/(log10 Re)2.58

for the

turbulent regime.14

The drag due to interaction is the result of the mutual interaction between the boundary layer of a surface and

that of the body in contact with this surface. According to previous studies, the drag produced by two adhered bodies

is far greater than the drag of these two bodies separately, inclusive from 30 to 55% of the smaller body drag.13

The

interaction drag is supposed to be a function of the thickness of the boundary layer on the contact walls and may be

estimated for the joints between walls and airfoil sections by means of the reference areas of the mentioned surfaces,

and the maximum thickness and chord of the airfoil section, as shown in Eq. (9).11,12

( ) ( )( ) 21

2

221

00030750

AA

t

ct

.ct.A,AC intD

+⋅

−= (9)

In the calculation of the interaction drag coefficient are involved, as may be seen from Eq. (10), the drag

coefficients due to interaction wing – fuselage, horizontal stabilizer – vertical stabilizers, and secondary components

– fuselage. The first two cases may be calculated by means of Eq. (9), because it involves airfoils sections and walls,

but in the case of secondary components, which consists of the adherence of bodies it is convenient to follow the

recommendation of Ref. 13, and suppose that it is 30% higher than the drag coefficient of each secondary

component.

( ) ( ) csintDvhintDfwintDintD CS,SCA,SCC ++= 1 (10)

However, according to previous experimental studies,15

the induced drag between walls and airfoils, such as the

one existing between the wing and the fuselage, varies depending on the relative vertical position of the wing with

respect to the fuselage.

VI. Theoretical and Numerical Estimation of Lift Coefficient

For this investigation, the lift coefficient estimate was made by means of the Prandtl’s Lift Line Theory,16

which

is quite effective for calculating the lift distribution and lift coefficient in rectangular wings with no swept or

dihedral.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


6

For this purpose a numerical code was created and programmed to obtain the lift coefficient at different angles of

attack, and at different Reynolds numbers. In the program, the solution for the equation for the spanwise lift

distribution is obtained using a Fourier’s series, based on the method developed by Glauert and Lotz,17

and the

global lift coefficient is calculated by simple integration. The program is capable of printing the lift coefficient, the

lift line slope, and to plot the lift distribution curve knowing only the wingspan, the root chord, the tip chord, the

geometric and aerodynamic twist, the airfoil section lift slope of each station for a determined Reynolds number, and

the number of spanwise stations.

The slope of the lift curve is obtained by means of the ratio cl/a, using data obtained through a computer code for

airfoil analysis, VisualFoil,10

at Reynolds flight number from 18 to -15 degrees with respect to the mean

aerodynamic chord..

VII. Wind – Tunnel Testing

Following the theoretical study of the lift coefficients and drag coefficient of each design modification it is

necessary to carry out aerodynamic tests in the wind tunnel.

The experiments were carried out between December 2003 and March 2005, using only one wind tunnel, model

Rollab SWT – 009, located within the facilities of the aeronautical engineering laboratory of the UNEFA, in

Maracay, Venezuela. The wind tunnel is a closed – circuit, closed throat, and unpressurized facility. It has a square

test section of 0.32 m x 0.32 m with transparent walls, and it was specially designed to test wings and small three-

dimensional models at low speeds.

The normal testing range is for airflow speed from 14 to 47 m/s, and it is determined by means of the pressure

differences between the test section and the downwash section. The wind tunnel has a three component TEM

balance capable of measuring lift, drag and moment.

The model support mechanism is capable of an angle of attack from 40 to -40 degrees, although these may vary

depending on the support being used. The models are fastened from below the test section on three struts.

Depending on the type of the model to be tested, it is usual to use extensions on these supports in order to avoid the

least interference between the supports and the model.

For the present study, a calibration of the tunnel was carried out before each run, in order to guarantee the

accuracy of the scale, the aerodynamic alignment angle, and the turbulence factor, which resulted in 1.38.

For this purpose a rectangular wing of 0.254 m span and 0.052 m of chord, formed by NACA 0015 airfoil

section was used. The symmetric characteristic of the airfoil section facilitates the obtainment of the aerodynamic

alignment angle.

The wing was tested in each run from 16 to -16 degrees, at different velocities in order to obtain lectures at

different Reynolds number values. After due considerations to the fact that this was a straight wing,18

the obtained

results were compared with those presented in the literature,19,20

showing discrepancies with the stall region values

reported in Ref. 20, but not with those reported in Ref. 19.

Based on the blueprints for model X-1, a fiberglass, reinforced with polyester resin wind tunnel model was

constructed, and painted in black. The test section size limited the scale, permitting a maximum wingspan of 0.256

m, which gave a scale of 1:20.2 for the models to be tested. The airfoil section used on the wing of the model is

NACA 4415, and the one used for the tail surface is NACA 0009 section, according to the design. The second scale

model, based on the blueprints of model X-2, was made with the negatives and same materials as scale test model

X-1, to which in the landing gear and spinner modifications were incorporated. Figures 4 and 5 show photographs of

the wind tunnel models X-1 and X-2, respectively. The dimensions of the two models are shown in table 1, being

compared with those of the full scale airplane.

Each model was tested held in an upside down position, at different angles of attack and at several speeds in

order to obtain data at different Reynolds numbers. Buoyancy, blockage, and tare-and-interferences corrections were

applied. See Ref. 21 for a more complete discussion.

VIII. Results and Discussion

A. Theoretical Analysis and Numerical Simulation

Before begin the theoretical calculations of the drag coefficient generated by models X-1, X-2, and X-3, the

calculation of the wing lift coefficient was carried out. For models X-1 and X-2 the wing is identical, while the wing

on model X-3 has a small variation at the tip. The deflection at the wing tip was made applying aerodynamic twist

between stations 2.32 and 2.4 (measured in meters) of the midspan, and for this it was necessary to altered the

original program code. In the case of the untwisted wing a data convergence of 10-6

was achieved, while for the

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


7

twisted wing was about 10 -4

, using a one hundred and one stations in both cases. The input parameters are shown in

table 2 and the results in Fig. 6 and table 3. Having the lift coefficient of the two proposed wings, it was possible to

calculate the induced drag coefficient at each angle of attack using Eq. (3).

Then the drag coefficient of each aircraft component was estimated using Eqs. (6) to (10), and the global drag

coefficient by using Eq. (5). The results corresponding to the three models are shown on table 4, for an angle of

attack equal to zero with respect to the fuselage, the drag coefficient as a function of angle of attack for the three

models are shown in Fig. 7, and the polar of the three models is plotted in Fig. 8. In this, the efficiency increase of

model X-3 with respect to model X-2 and of the latter with respect to X-1 can be appreciated. The global drag

coefficient variation, from model X-1 with respect to model X-3, is significant. The results show a drag decrease of

13.8% with respect to X-2, and of 16.6% with respect to X-3. The landing gear modifications generated 83.1% of

the drag coefficient variation, and the 16.9% to the decrease of the induced drag.

B. Wind Tunnel Testing

Based on all results obtained in the wing tunnel tests, the values of the uncertainties related with the final results

were calculated. For this purpose the possible variables that may interfere in final results were analyzed. The airflow

temperature, pressure differences, lift and drag variations were taken into consideration. Table 5 shows the average

deviation and standard deviation for lift coefficient, drag coefficient, and airflow speed.

The results obtained for the X-1 design tested model are summarized in table 6 for lift coefficient, and in table 7

for drag coefficient. On both, the angle of attack and the Reynolds numbers with respect to the wing and the

fuselage are reported. Figures 9, 10, and 11 show the lift coefficient curve, the drag coefficient curve, both as a

function of angle of attack, and the polar curve for each Reynolds number obtained, respectively. Tables 8 and 9

show the resultant data from the wing tunnel tests for lift coefficient and drag coefficient at different angles of

attack, and the Reynolds numbers for the ANCE X-2 wind tunnel model. Figures 12, 13, and 14 show the lift

coefficient curve and the drag coefficient curve as a function of angle of attack, and the polar curve for each

Reynolds number obtained, respectively.

Figures 15 and 16 show the lift coefficient curves as a function of Reynolds number for several angles of attack

of the models X-1 and X-2 respectively. For all the angles of attack these curves present a depression Rew between

8.4x104 and 8.7x10

4, and they have a similar behavior. For some major angles of attack the lift coefficient presents a

negative slope at the end of the curve.

Figures 17 and 18 show the drag coefficient curves as a function of Reynolds number for several angles of

attack. In these, a trend similar to rounded bodies can be observed, the curves fall progressively until reaching their

lowest points Ref between 4.2x105 and 4.4x10

5, a slight increase and, finally, a slight negative slope. This represents

the transition from laminar to turbulent flow condition, very usual in Reynold numbers similar to those studied. At

low angles of attack the lifting surfaces produce low lift values. At high angles of attack the slopes following the

depressions tend to be more pronounced than those observed at angles of attack close to zero.

C. Scale Effects Conversion

Due to the fact that the tests were not carried out at the flight Reynolds number, it is necessary to apply the scale

effects correction. For this purpose the data obtained at the wind tunnel for both test models, at the highest Reynolds

number, were used. With these the procedure described in Ref. 21 for the correction of scale effects over the drag

coefficient (extrapolation method), and over the lift coefficient curve (Jacob´s method) was carried out.

In order to correct the scale effects over the lift coefficient curve the results of Ref. 19 were considered, where

the slope of the lift curve of the NACA 4415 airfoil section is different for Reynolds numbers similar to those of

these study with respect to those of flight. In the curves shown in Fig. 9 and 12 the zero lift angle approximately

coincides with those shown in Ref 19 together with the slope of the lift curve which would result for a finite

rectangular wing.20

Although Ref. 19 does not offer details of the behavior of this airfoil for such low Reynolds

numbers. It may be observed that for Re higher than 8.3x104, the slope of the lift curve is equal to the corresponding

1.26x106, only for low lift coefficients. In order to extrapolate the lift coefficient curve, the slope of the lift curve

corresponding to the X-2 model data, for Rew equal to 1.2x105, with lift coefficients under 0.3 were taken. The

maximum lift coefficient and the stall data were taken from the data obtained through the numerical simulation of

the wing.

The data obtained from the wind tunnel tests with ANCE X-2 scale test model were used for determine the

features of the X-3 version. The only difference between both versions is the wing tip twist, which does not affect

the slope of the lift curve, as revealed by the numerical study which was carried out. However, there is a lift

coefficient variation between the original wing and the twisted wing equal to 0.0094. Thus, in order to determine the

lift curve of the ANCE X-3 in flight, the same slope of the lift curve of the X-2 version was taken, and each CL

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


8

obtained was diminished in 0.0094. The maximum lift coefficient and the stall data were taken from the numerical

study.

The drag coefficient curve of the ANCE X-3 was plotted by means of a variation on the extrapolation method,

adding the profile drag coefficient of the wing tunnel model ANCE X-2, the induced drag coefficient calculated with

the X-3 model lift coefficient.

Figures 19 and 20 show the lift coefficient and the drag coefficient as a function of angle of attack for models

ANCE X-1, X-2, and X-3. Figure 21 shows the polar curve of the three versions of the design. From this, the

increase of optimum efficiency and the cruise flight efficiency of the X-3 model with respect to X-2, and the latter

with respect to the X-1, can be appreciated.

Table 10 shows the comparison of the drag coefficient obtained with the wind tunnel testing data for cruise

flight. These results reveal that the improvement applied in the landing gear yield a ∆CD of -0.0028 (73.7% of ∆CD),

and the twists applied to the wing a ∆CD of -0.001 (26.1% of ∆CD). Between both the total drag coefficient variation

is of -0.0038. This means a decrease in the drag coefficient of 7.98% with respect to the initial one. Table 10 also

shows the aerodynamic efficiency for each version of the ANCE. X-2 model resulted to be 5.88% more efficient

than X-1, and X-3 model, 6.49%.

IX. Conclusion

A detailed theoretical analysis of aerodynamic resistance, a wide range of wind tunnel tests on scale models, and

a numerical code based on Prandtl’s Lift Line Theory, were fundamental tools in the ANCE aerodynamic drag

cleanup process. The modifications made on the landing gear, and the variation in local wing tip twist resulted in an

efficiency increase when applied to the ANCE design. These modifications produced a total efficiency increase of

16.6% when estimated by the theoretical method, and 6.49% based on experiment results, and a drag decrease of

0.0087 theoretically calculated, and of 0.0038 experimentally obtained.

Although the theoretical analysis is a good start for this type of studies, and its results show the approximate

performance of an aircraft in flight, this is not reliable due to the difference between the theoretical calculated data,

and the experimental data, obtained from wind tunnel testing.

The aerodynamic data generated at UNEFA facilities has been used to feed the database for the ANCE vehicle.

This database will be used to estimate the performance, and flight capacities of the airplane.

Acknowledgments

The authors wish to thanks the financial support of Direction of Investigation, Universidad Simón Bolívar,

Caracas, and FUNDACITE Aragua, Maracay, both in Venezuela. We also want to acknowledge to the Department

of Aeronautical Engineering of Universidad Nacional Experimental Politécnica de la Fuerza Armada, Maracay, for

allowing the use of the aerodynamic laboratory facilities.

References 1 Boschetti, P. J., and Cárdenas, E. M., “Diseño de un Avión No Tripulado de Conservación Ecológica,” Theses, Department

of Aeronautical Engineering, Universidad Nacional Experimental Politécnica de la Fuerza Armada, Maracay, Venezuela, 2003. 2 Cárdenas, E. M., Boschetti, P. J., Amerio, A., and Velásquez, C. D., “Design of an Unmanned Aerial Vehicle for Ecological

Conservation,” AIAA Paper 2005-7056, Sep. 2005. 3 Lausetti, A., “La Polare del Velivolo,” Esercizi di Mecánica del Volo, edited by Libreria Editrice Universitaria Levrotto &

Belle, Torino, Italy, 1975. pp. 10-17. 4 Vennard, J. and Street, R., “Fluid Flow about Immersed Objects,” Elementary Fluid Mechanics, 5th ed., edited by John

Wiley & Son, Inc., New York, 1976, pp. 645-692. 5 Phillips, W. F., “Lifting-Line Analysis for Twisted Wings and Washout-Optimized Wings,” AIAA Journal of Aircraft, Vol.

41, No. 1, 2004, pp. 128-136; also AIAA Paper 2003 – 0993, Jan. 2003. 6 Phillips, W. F., Fugal, S. R. and Spall, R., “Minimizing Induced Drag with Geometric and Aerodynamic Twist, CFD

Validation,” AIAA Paper 2005-1034, Jan. 2005. 7 Miller, C. G., "Development of X-33/X-34 Aerothermodynamic Data Bases: Lessons Learned and Future Enhancements,"

NATO-AVT Symposium, Paper RTO-MP-35, Oct. 1999. 8 Bushnell, B. M., “Aircraft Drag Reduction – a review,” Journal of Aerospace Engineer, Vol. 217, No 1, 2003, pp. 1-18. 9 Coe, P., “Review of Drag Cleanup Test in Langley Full – Scale Tunnel (from 1935 to 1945) Applicable to Current General

Aviation Airplanes,” NASA TN D-8206, Jun. 1976. 10 Visual Foil/Model Foil, Ver. 4.1, Hanley Innovations, Ocala, FL, 2001. 11 Hoerner, S. F. Résistance á L’avancement dans les Fluides, edited by Gauthier Villars Editeurs, Paris, France, 1965,

Chapter XIV.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


9

12 Zeidan, F., “Estudio Teórico-Practico de la Resistencia al Avance de una Aeronave,” Aerodinámica y Práctica Avanzada,

edited by Consejo de Publicaciones de la Universidad de los Andes, Mérida, Venezuela, 1995, pp. 89-96. 13 Von Mises, R., “Air Resistance or Parasite Drag,” Theory of Flight, edited by Dover Publications, Inc., New York, 1959,

pp. 95-111. 14 Rebuffet, P., “Phénomènes el principes généraux,” Aérodynamique Expérimentale, Tome 1, edited by Librairie

Polytechnique Béranger, Paris, France, 1966, pp. 102. 15 Prandtl, L., “Effects of Varying the Relative Vertical Position of Wing and Fuselage,” NACA TN-75, 1921. 16 Prandtl, L., “Applications of Modern Hydrodynamics to the Aeronautics,” NACA TR-116, 1921. 17 Peery, D. J., “Spanwise Air-Load Distribution,” Aircraft Structures, edited by McGraw Hill Book Company, New York,

1950, pp. 213-249. 18 Anderson, J. D., “Incompressible Flow over Finite Wings”, Fundamentals of Aerodynamics, 2nd ed., McGraw-Hill, Inc.,

New York, 2002, pp. 340-347. 19 Jacobs, E. N., and Sherman, A., “Airfoil Section Characteristics as Affected by Variations of the Reynolds Number,”

NACA R-586, 1937. 20 Sheldahl, R. and Klimas, P., “Aerodynamic Characteristics of Seven Symmetrical Airfoil Sections Through 180-Degree

Angle of Attack for use in Aerodynamic Analysis of Vertical Axis Wind Turbines,” Sandia National Laboratories, Albuquerque,

NM, SAND80-2114, Mar. 1981. 21 Rae, W. H., and Pope, A., Low-Speed Wind Tunnel Testing, 2nd ed, edited by Wiley Interscince Publication, New York,

1984, pp. 198-205, 419-424, 457-464.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


10

Table 1 Reference dimensions.

Dimension Full scale 1 to 20.2 scale

Surface, m2

2.899 0.007105

Wing span, m 5.18 0.256

Medium chord, m 0.604 0.030

Length, m 4.65 0.231

Table 2 Input data parameters to the code.

Wing Geometry X-1 and X-2 X-3

Effective wing span, m 4.8 4.8

Tip chord, m 0.604 0.604

Root chord, m 0.604 0.604

Angle of attack in the root, deg 6.37 6.37

Geometric and aerodynamic twisted, deg 0 7.38§

Airfoil Flow Conditions

Lift section Airfoil NACA 4415

Reynolds number 1388022

Compressible effects No

Transition on laminar separation

Table 3 Data obtained of wing lift coefficient at different angles of attack.

CLw α, deg

X-1, X-2 X-3

16 1.1095 1.1021

14 1.4252 1.4167

12 1.5359 1.5269

10 1.4384 1.4293

8 1.3246 1.3154

6 1.1403 1.131

4 0.9954 0.9861

2 0.8035 0.7941

0 0.6115 0.6021

-2 0.4195 0.4101

-4 0.2275 0.2181

-6 0.0355 0.0262

-8 -0.156 -0.166

Table 4 Summary of data obtained for drag theoretical analysis.

CDcs

CDpw

CDh

+

CDv

CDf CDint Camera Powerplant

Landing

gear

CDp CDi

(α=0) CD

X-1 0.0113 0.0036 0.0004 0.0052 0.0037 0.0061 0.0063 0.0366 0.0160 0.0525

X-2 0.0113 0.0036 0.0004 0.0035 0.0037 0.0061 0.0008 0.0293 0.0160 0.0453

X-3 0.0113 0.0036 0.0004 0.0035 0.0037 0.0061 0.0008 0.0293 0.0145 0.0438

∆CD 0 0 0 -0.0018 0 0 -0.0055 -0.0073 -0.0015 -0.0087

§ Only between wing stations 2.4 and 2.32 (measured in meters) of the midspan.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


11

Table 5 Estimated uncertainties.

Average Deviation Standard Deviation

CL 0.001914239 0.002707143

CD 0.000208864 0.000295378

V. m/s 0.03317083 0.04187857

Table 6 Experimental data obtained in the wind tunnel test of lift coefficient as function of angle of attack

and Reynolds Number. for X-1 model.

Rew 5.3E+04 6.1E+04 7.1E+04 7.9E+04 8.4E+04 8.9E+04 9.9E+04 1.1E+05

Ref 2.5E+05 2.9E+05 3.3E+05 3.7E+05 4.0E+05 4.2E+05 4.7E+05 5.0E+05

α CL

-0.70 0.1504 0.3030 0.2356 0.2491 0.1728 0.1993 0.3030 0.3630

3.57 0.4100 0.5078 0.4631 0.4360 0.3374 0.4320 0.5302 0.5228

5.70 0.5981 0.7000 0.5428 0.5562 0.5359 0.5274 0.6437 0.6400

7.79 0.6597 0.7925 0.6805 0.7207 0.6784 0.6549 0.7475 0.7029

10.95 0.8197 0.9421 0.7406 0.8081 0.8364 0.7049 0.8415 0.7785

12.95 0.8775 1.1641 0.8748 0.9147 0.9287 0.8287 0.9081 -

16.95 0.9921 1.0723 1.0078 0.9669 1.0198 0.9926 1.0410 -

Table 7 Experimental data obtained in the wind tunnel test of drag coefficient as function of angle of

attack and Reynolds Number. for X-1 model.

Rew 5.3E+04 6.1E+04 7.1E+04 7.9E+04 8.4E+04 8.9E+04 9.9E+04 1.1E+05

Ref 2.5E+05 2.9E+05 3.3E+05 3.7E+05 4.0E+05 4.2E+05 4.7E+05 5.0E+05

α CD

-0.70 0.0571 0.0492 0.0419 0.0377 0.0314 0.0296 0.0525 0.0492

3.57 0.0591 0.0583 0.0414 0.0487 0.0408 0.0380 0.0664 0.0551

5.70 0.0826 0.0690 0.0569 0.0613 0.0518 0.0479 0.0747 0.0689

7.79 0.0862 0.0892 0.0723 0.0854 0.0628 0.0578 0.0901 0.0764

10.95 0.1155 0.1318 0.1198 0.1127 0.0969 0.0976 0.1232 0.0999

12.95 0.1404 0.1890 0.1341 0.1471 0.1167 0.1152 0.1447 -

16.95 0.2075 0.2459 0.1770 0.1928 0.1660 0.1681 0.1875 -

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


12

Table 8 Experimental data obtained in the wind tunnel test of lift coefficient as function of angle of attack

and Reynolds Number. for X-2 model.

Rew 4.1E+04 5.5E+04 6.7E+04 7.7E+04 8.7E+04 9.4E+04 1.0E+05 1.1E+05 1.2E+05

Ref 1.9E+05 2.6E+05 3.1E+05 3.6E+05 4.1E+05 4.4E+05 4.8E+05 5.1E+05 5.4E+05

α CL

-9.23 -0.7621 -0.4313 -0.4649 -0.4317 -0.5186 -0.4043 -0.3982 -0.3073 -0.2391

-7.10 -0.4522 -0.5021 -0.4526 -0.4193 -0.4728 -0.2535 -0.2428 -0.2119 -0.0970

-4.96 -0.2904 -0.4068 -0.4402 -0.3239 -0.3266 -0.1303 -0.0873 0.0081 0.1008

-2.83 -0.2779 -0.0624 -0.1475 -0.1454 -0.1802 -0.0901 0.0445 0.1659 0.2245

-0.70 0.0331 0.0331 0.0331 0.1161 -0.0673 0.1438 0.1285 0.2199 0.2740

1.44 0.1948 0.2943 0.1577 0.2946 0.1459 0.2393 0.2602 0.3154 0.3772

3.57 0.3562 0.4727 0.2822 0.4315 0.2588 0.3901 0.3680 0.4522 0.4840

5.66 0.6587 0.4768 0.3984 0.5604 0.4638 0.5050 0.5392 0.6431 0.6179

7.76 0.6648 0.7319 0.6287 0.6082 0.5370 0.6220 0.6645 0.7115 0.7354

9.77 0.6573 0.9721 0.7327 0.7246 0.6626 0.7522 0.7044 0.8074 0.8570

11.79 0.9497 0.9677 0.7840 0.8857 0.7586 0.8305 0.8902 0.9477 0.9266

13.78 0.9416 0.9591 0.9440 0.9608 0.8506 0.8505 0.9774 1.0431 1.0293

15.52 0.9116 1.0939 1.0820 0.9312 0.8879 0.9040 1.0192 1.1377 1.0920

Table 9 Experimental data obtained in the wind tunnel test of drag coefficient as function of angle of

attack and Reynolds Number. for X-2 model.

Rew 4.1E+04 5.5E+04 6.7E+04 7.7E+04 8.7E+04 9.4E+04 1.0E+05 1.1E+05 1.2E+05

Ref 1.9E+05 2.6E+05 3.1E+05 3.6E+05 4.1E+05 4.4E+05 4.8E+05 5.1E+05 5.4E+05

α CD

-9.23 0.1324 0.048 0.0732 0.0723 0.0648 0.0538 0.0687 0.0662 0.0707

-7.10 0.0848 0.0765 0.0609 0.0621 0.0509 0.0415 0.0541 0.0529 0.0503

-4.96 0.0815 0.0611 0.0494 0.0527 0.0378 0.0342 0.0431 0.0421 0.0389

-2.83 0.0791 0.0586 0.0470 0.0503 0.0402 0.0317 0.0414 0.0412 0.0378

-0.70 0.0775 0.0571 0.0454 0.0427 0.0387 0.0342 0.0433 0.0457 0.0431

1.44 0.0986 0.0806 0.0447 0.0481 0.0502 0.0416 0.0496 0.0510 0.0475

3.57 0.1205 0.0808 0.0531 0.0604 0.0529 0.0499 0.0498 0.0603 0.0534

5.66 0.1211 0.1056 0.0782 0.0610 0.0632 0.0667 0.0643 0.0761 0.0675

7.76 0.1439 0.1067 0.0875 0.0742 0.0790 0.0759 0.0828 0.0893 0.0713

9.77 0.1440 0.1309 0.1040 0.0986 0.1035 0.0982 0.0934 0.1015 0.0904

11.79 0.1876 0.1433 0.1287 0.1170 0.1087 0.1166 0.1145 0.1199 0.1015

13.78 0.2090 0.1792 0.1367 0.1289 0.1376 0.1245 0.1317 0.1408 0.1255

15.52 0.2498 0.2249 0.1588 0.1567 0.1596 0.1423 0.1536 0.1596 0.1502

Table 10 Drag coefficient at cruise lift coefficient for the three versions of ANCE at flight Reynolds Number

X-1 X-2 X-3

CD (α=0) 0.0476 0.0448 0.0438

∆CD - -0.0028 -0.0038

CL/CD 12.406 13.181 13.268

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


13

Figure 1: Sketch of the initial version of ANCE.

Figure 2: Port side view of the three versions.

Figure 3: Wing tip detail of X-3 model.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


14

Figure 4: ANCE X-1 wind tunnel model.

Figure 5: ANCE X-2 wind tunnel model.

Figure 6: Wing lift coefficient as a function of angle

of attack estimated numerically.

Figure 7: Drag coefficient as a function of angle of

attack estimated theoretically, for each model.

Figure 8: Polar curves for each version.

Figure 9: Lift coefficient as a function of angle of

attack of ANCE X-1 wind tunnel model, at several

Reynolds Numbers.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


15

Figure 10: Drag coefficient as a function of angle

of attack of ANCE X-1 wind tunnel model, at several

Reynolds Numbers.

Figure 11: Polar curve of ANCE X-1 wind tunnel

model, at several Reynolds Numbers.

Figure 12: Lift coefficient as a function of angle of

attack of ANCE X-2 wind tunnel model, at several

Reynolds Numbers.


of attack of ANCE X-2 wind tunnel model, at several

Reynolds Numbers.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


16

Figure 14: Polar curve of ANCE X-2 wind tunnel

model, at several Reynolds Numbers.

Figure 15: Lift coefficient as a function of

Reynolds Number of ANCE X-1 wind tunnel model,

at several angles of attack.

Figure 16: Lift coefficient as a function of



Figure 17: Drag coefficient as a function of



Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399


17

Figure 18: Drag coefficient as a function of




of attack for ANCE X-1, X-2, and X-3.

Figure 20: Lift coefficient of ANCE X-1, X-2, and

X-3, as a function of angle of attack.

Figure 21: Polar curves of ANCE X-1, X-2, and

X-3.

Dow

nloa

ded

by I

ND

IAN

IN

STIT

UT

E O

F SC

IEN

CE

on

July

22,

201

4 | h

ttp://

arc.

aiaa

.org

| D

OI:

10.

2514

/6.2

005-

7399

aerodynamic optimization of an uav design

Documents