Acta Polvtechnica Vol. 43 No. 312003

Aerodynamic Design and Experimental
Investigation of the Sailplane Wing Tip

Devices
P. Anderle, L. Smrdek, F. N. Coton

This paper dcscribes an experirnental set-up for the inaestigation of wing tip d,evices \ydoPr! as part of a stud} into thc aeloci'ty and

oorti;ty dtstrlOutions in thi flow fuH behind, winglex, using hot-wire anemomztry. In this study, ffirt was foatsed, on g!'irylnS !-gre:Lter
wdarsnndi.ng of wlnt hopitnt in the region whire the wiiglet joins the wing.The measuremcnts were performed in the H_a(:l'2-Page

aind tmnel ;f de Department of Aerospaie Engineering at the Uniaersity of Gln"sgou. In order to carry..out mzasurements wilh tlu hot-wire

anemmutry s4stem, a neu traairst *cihonis* ,as dcsigned and manufactured. This traverse mcchanism was integrated wilh thp otlur test

instrumchtati.on to 6eate a complete mcasurement chain. The complete system atlowsfully autornated hotwire measuremcnts to be ma'dz oaer

a def.ned area using prograrnmable test parameters.

KEworfu : winglet, hot-wire anemomctry, traaerse mzchanism, uorticity.

I Notation
cDx vorticity component in YZ plane

b ving span

V1, Vy velocitY comPonents

U. free sream velocity

Z downstream distance

2 Introduction
In the late 1960s, designers began experimenting with

wing-tip geometries using'small' vertical extensions to reduce

the formation of tip vortices. The winglet concept actually

dates back to 1897, when Frederick l-anchester took out a

patent on the idea, incorporating it into some of his wing

theories []. His wing had nuo 'capping planes' at the end of
it, which became knor,vn in the 1920s as 'end plates', when

Prandd extended his basic lifting line concept [2].

The real break+hrough with winglets was made by Whit-

comb. In 1976, Richand Whitcomb, a NASA aerodynamicist,

published a paper that compared a wing with a winglet and

the same wingwith a simple extension to increase its span [3]'
Whitcomb showed that winglets reduced drag by about

20 percent and increased the wing lift-drag ratio by approxi-
mately 9 percent. Induced drag represents 30-40 Percent
of the total drag of a transPort plane in cruise condition, so

the induced drag reduction has a significant effect on fuel

consumption. Whitcomb began a focused investigation into

winglet ierodynamics and tested several designs in the wind

tunnels at the NASA Langley Research Center' The first
industrial application of the winglet concept was in general

aviation businessjets. Iior instance, it is claimed that a winglet
on a Boeing-747 could significandy reduce fuel burn on

long-range flights. Research into the effect of winglets on {irst
genlrationjet transport wings showed that they can produce

reasonable drag reduction in high lift conditions [4].
Wnglets are now being incorporated into most new com-

mercial and military transportjets. Since the 1980s, the most

modern high performance sailplanes also have small vertical
wing tip extensions.

The first sailplanes to have winglets were the ASW-20F8
GEMINI and NIMBUS 3. Sets of Whitcomb style winglets
were fitted to these wings in the late 1980s. Flight tests carried
out on these aircraft demonstrated the effect of winglets on

high aspect ratio wings [5].

3 How do winglets work
The primary effect of the winglet is to control the cross

flow in the tip region of the wing in such a way as to reduce

induced drag by displacing the vortices ounvard. The air flow-

ing over the winglet, due to the presence of the tip vortex'
strikes the winglet at an angle of attack. The winglet, like any

wing, produces lift which, in this case, has a component in the

forward direction. Thus, the winglet produces thrust (Ftg. l).
This thrust component alters the cross flow at the wing tip

g#.r.i=r\ff
ilutwasn

Fig. l: Forward thrust comPonent development



Acta Polytechnica Vol. 43 No. 3/2009

and recovers some of the performance that would be lost
through the additional drag caused by the inoease in weued
area.

4 Winglets for sailplanes
Theory and experience have shown that the most emcient

sailplane wing is one that is very long and slender. Having
a high aspect ratio wing is one way of cutting wing-tip losses.
In essence, the longer wing has the same tip losses but those
energy losses will affect a lesser proportion of the total wing.
In otherwords, the lift is distributed over the longerwingspan
and the trailing vorticity is spread out, dissipating less energy.

Fnrm aconstruction point ofview, a longwing is prone to
fle:r and has to be strengthened; this adds weight. The winglet
provides ttre effect of an increased aspect ratio without ex-
tending the wing-span and so does not increase the wing root
bending as much as an actual span er(tension would. The
moment arm of the lift from a span extension is approxi-
mately one-half of the wing semi-span, whereas the moment
arm of the winglet lift is 'mughly only one-half of the vertical
winglet span. This small increase does not overload the wing
or significantly alter the standand operating limitations. The
addition of wingles on sailplane wings also improves the
maximum lifVdrag coeffrcient for some l5 m spanJimited
FAI sailplane dasses [6].

The induced dragcoefficient is proportional to the square
of the lift coefficient hence the reduction in drag also im-
proves climbing capability [7], [8]. This improvemenr can
be used when sailplanes circle in thermal bubbles, the main
source of power to stay aloft [9]. Achieving a maximum
cross country speed during sailplane competitions is anorher
important consideration. Hence, the design of the wing-
lets must involve the compromise of maximizing the low
speed improvement without sacrificing high-speed perfor-
mance [0].

The winglet added to an ASW-19 clearly showed that for
some speeds the friction drag could orceed the induced drag
reduction provided by winglets [ll]. A correctly desigrred
winglet can, howeve4, be reasonably effective as illustrated in
a study using the ASW-20 sailplane [12].

5 Wind tunnel models of winglets
The wind tunnel models used in the experiments were

real wing tips taken from the wind of a SMCZ sailplane.
The models werre mounted vertically on a base plate that

was secured to a rail track mechanism. This mechanism al-
lowed the model to be moved backwards and forwards in the
wind tunnel working section to change the distance between
the model and the hot-wire measuremenr plane. In addition,
the base plate was designed to allow the incidence of the
model to be changed. During the o<perimental programme
four kinds of wing tip were investigated; a wing without
a winglet and then three 304C2 sailplane wing tips of dif-
ferent design. The key parzrmerers that defined the winglet
designs are shown in Fig. 2, and are the wingler airfioil, sweip-
back, cant angle, nuist distribution and the ratio of the winglit
root chord to the winglet tip chord (taper).

4

"plstt*

&L
Fig. 2: Key Design parameters of winglets

6 Wind tunnel
The Handley Page wind tunnel is an armospheric low-

-speed wind tunnel with a closed return circuit equipped with
a rectangular testing-section of dimensions 2.15 m by 1.6 rn
and length 3.38 m (FiS. 3).The corners have 650 mm fillets
that house lamps to provide lighting. Visual access to the
working section is provided by 0,84 square metres of plate
glass and acrylic windows that permit the model to be viewed
from many angles. Several venting slots in the tunnel walls at
the test section exit maintain near atmospheric static pres-
sure. The nozzle placed in front of the testing section has a
contraction ratio of l:4. The power supply is an electric motor
that drives a fan 2.3 m in diameter to provide the airflow in
the wind tunnel. The tunnel can reach speeds up to 60m/s.

D&|ml.iE otfheu,lndt wl m{€cdm lml
A"A lgd)x l90O
&a ?,200r,2(fi
C.C 3000 r 4m0
tlD 3000x4000
E€ 1600 x 2'150FF 1600x2150
GG dbn€t€.2300

Fig. 3: Handley Page wind Tunnel, University of Glasgow

7 Measurement procedure
The measurement chain consists of an x-wire sensol a TSI

IEA 300 constant temperaturc anemometry system, a per-
sonal computel an SMOCI transmitter and a traversing
mechanism, as shown in Hg. 4. This figure also shows the
model mounting arrangement described prwiously. The
X-wire sensor is connected to two channels of the IFA 300 by
coaxial cables. The IEA 300,hardware converts the acquired
signals fiom the sensors and transmits them to the controlling
computer via a BNC adapter block and data acquisition card.
The IEA 300 sofnvare installed on rhe computer then pro-
cesses the recorded data. Once the data has been recorded for
an entire X traverse, a master program, written in lABView,
sends a new instruction ro the SMOCI unit. The SMOCI con-
verts the instruction into the signal needed by the stepper
motors to move the traverse to its next Y position. When the
traverse mechanism reaches this new position, another signal
is sent to it and it begins its X traverse. The x-wire sensor con-
tinually samples data during this traverse and sends the signal
to the computer. This process continues until measurements
have been made over the entire measurement grid.



Acta Polvtechnica Vol. 43 No. 3/2003

Investigation Grids

lFA300 CTASystem
Eack Panel

Triggering

BNC
Adapter

Coryrputsr

I

I

rjyt
Step motors

Fig. 4: Wind tunnel testing configuration

8 Traverse rnechanism

The computer controlled Faverse mechanism for probe

positioning and data acquisitionwas specifically designed and

manufactured for the Present project. The traverse is a mo-

torized two-component mechanism that can move the probe

to any point within a 850 mmx930 mm grid. In the present

series of tests the traverse was mounted behind the test

models such that measurements could be made in planes

perpendicular to the onset flow. The location of the measure-

-..tt plutt. with respect to the test model could be varied

using the model mounting tracks described previously. The

way in which the traverse is assembled is shown in Ftg. 5

below, and Frg. 6. shows the assembled traverse in the wind

tunnel together with a winglet model'

The horizontal motion is provided by a large, off-the-shelf
linear slide driven by a stePPer motor. Vertical movement is

provided by a purpose-built traverse mechanism based on a
precision ball screw, which is positioned in front of a linear

slide mounted on an aluminium box-section suPPort. The

carriage ofthe ball screw is connected to the carriage ofthe
linear slide and so, when the ball scrcw is driven by a stePPer

moto6 the carriage moves up and down. The incremental

resolution of the linear motion is 0.03 mm,

X-wire Sensor

RS-232-C
Dightal Contol
Line

Fig. 5: Traverse mechanism



Acta Polytechnica Vol. 43 No. glZO0S

Fig. 6: Wingler models (l:l) placed in the Handley page wind tunnel

9 Hot wire anemometry
Measur€ments of the magnitude and associated direction

of the time-dependent velocities behind the winglet models
were obtained using a DANTEC 55P61 cross-wire probe con-
nected to a TSI IEA-300 three<hannel constant temperature
anemometer system. The sensorwires on the probe are 5 mm
diameter platinum plated tungpten wires with a lengttr/diam-
eter ratio of 250, which form a measuringvolume of approxi-
mately 0.8 mm in diarneter and 0.5 mm in height. The wires
are oriented perpendicular to each othe4, corresponding to
45 degrees from the free stream direction, which gives the
best angular resolution. An additional temperature probe was
used to correct the anemometer output voltages for any varia-
tion in ambient flow temperature. For probe calibration, an
open jet vertical wind tunnel with a maximum operating
velocity of 43 m/s was used. A support allowed the sensors of
the X-wire probe to be rotated by 30 degs in the plane of the
sensors. Variation of the flow velocity and yaw angle then
enabled the coefficients of the effective velociw method to
be determined. Fig. ? shows how the hot-wire anemomerry
system was integrated into the overall measuremenr sysrem.

l0 Investigation grid
X-travel:

One step of the stepper motor provides 0.03125 mm
of linear motion, and the time required for each step is
0.2980 sec. The size of the investigation grid in the x-direc-
tionwas 850 mm (FiS. 8) or 27200 steps of the stepper moror.
The time required to traverse this distance was, therefore,
8.1055 sec.

At a sampling rate of 2000 Hz, the sampling rime of
8.1055 seconds gives a total of 162ll investigation points
per line. To allow for turbulence in the flow, the data is aver-
aged in blocks of 48 measurement points corresponding to
approximately 2.5 mm of motion. Thus, 339 averaged data
values are collected during each X traverse.

0 sec 8.1055 sec
Xmax Travel

0 step

Start Point

.. X step1l
steps

9
o

o
E

o
{t
o

Y

O Auxiliary Pulse

Fig. 8; Investigation grid

Y-travel:

The length of the investigation grid in the Y-direction was
set at 600 mm. This length was divided into steps of 5 mm,
giving a total of 12l measurement points in the Y direction.

The total time taken to traverse the entire
850 mmx600 mm grid was just over 20 minutes, and a total
of 41019 measurements were obtained for each grid. The
scheme of this process is presented in Frg. 8.Fig. 7: Flow chart ofdata acquisition

6



Acta Polytechnica Vol. 43 No. 312003

I I Experimental results
The flow field behind the wing tip models was measured

in three planes (Z/b = 0 .2,7,h = l,Z,lb = 2) for angles of attack

c = (-3, 0, 3, 6, 9) degrees. For all tests the free stream velocity

U- was set at 33 m/s (l18.8 km/h) which corresponded to a

Reynolds number of Re = 0.8x 106 based on the mean chord
of the main wing. Data analysis and graphical presentation
were carried out using Tecplot software in the form ofvector
plots of velocity distribution and contour line plots of the

vorticity component ol* defined in equation I and Frg. 9'

Frgs. lG-13 present example measured data for four wing
tip configurations in the plane Zb = I when the wing was at
angle of attack cr=3 degrees and the fiee stream velocity

was U-=33 m/s. These figures illustrate the effectiveness of
the system in capturing the differences in the flow structures
behind the different winglet configurations. For example,
wing tip A (Iig. l0), which is basically a standard wing tip with
a small vertical extension that projects downwards from the

lower surface of the wing, produces an almost classical tip vor-
tex structure where the vorticity is concentrated in a single

structure that rolls up slightly inboard of the wing tip. Also
visible in this figure is the vorticity in the wake of the main
wing. Frg. ll, on the other hand, shows the velocity and
vorticity distributions behind a much larger winglet exten-
sion. In this case, there are two clear vortex structures; one at

the wingleVwing junction and the other at the tip of the

winglet. The effect of sweeping this type of winglet back can

be observed in lig. 12, where measurements on a similar
winglet with higher sweepback are presented. In this figure,

ETH
g@M
EilTI

Fig. l0: Wing tip model A

E@H
ct@tE6

the vorticity associated with the wingletAving junction does

not appear as a single well defined vortex but rather as a more

spread-out region of vorticity. The winglet tip vortex is also
less distinct, but it should be remembered that the tip of the
winglet will be closer to the measurement plane in this case

and so the roll-up may not be as complete. Iinally, the effect
of a small upward swept winglet is shown in Ftg. 13. In
this case, the vorticity is well distributed and follows the curva-

ture of the winglet. It should be noted that these cases are

presented merely as an example of the capability of the

measuring system. The evolution of the vortex structures
downstream of these wing tips is very complex and cannot

be inferred from observation of the behaviour in a single
measurement section.

avY )Vx
uv =- --^0x0y (l)

Fig. 9: Vorticity ox component

Fig. I l: Wing tip model B



Acta Polytechnica Vol. 43 No. gl200g

@EA

M$A
Fig. l2: Wing tip model C

HEES
Et@
MEE

Fig. l3: Wing tip model D

12 Conclusion
Experimental instrumentation for investigating of the

flow field behind a winglet model has been designed, manu-
factured and setup. A newly designed traverse mechanism has
made it possible for measurements to be made at high spatial
resolution within a large enough area for full size wing tip
models to be tested. Results have been presented to demon-
strate this capability. An additional feature of the system is
that the time required for testing is relatively short.

Acknowledgement
The authors wish to acknowledge the support of the

academic and technical staff of the University of Glasgow,
Aerospace Engineering Department. Special thanks go to the
technical stafffrom the Spencer Street Laboratory especially,
to Mr. R. Gilmou4, who helped with the experimental set-up,
and to Mr. T Smedley, Mr. A. Erwin, Mr. J. Kitching and
Mr. A Fraser for their help in manufacturing and installing
the equipment.

Thanks also go to Mr. Fotmesil and to Hph. for manufac-
turing the winglet models.

Mr P Anderle was supported by a studentship provided by
BAE Systems.

8

References
tl] Lanchester, F.W.: Aerodynamits. London: Constable &

Co, 1907.

121 Prandtl, L:Thcory of Lifiing Surfarcs. NACATN9, 1920.

t3l Whitcomb, R. T.: u{ Daign Approath and Setected. Wind,-
-Tunnel Result at High Subsonh Spud fm Wing-T:ip
Mounted, Winglzts. NASA'lN D-8260, July I 976.

I4l Montoya, L. C., Jacobs, P. F., Flechner, S. G.: Effect of
Winglzts sn a Fi,rst-Gmnation Jet Tronsport Wing, II i- Prei-
sure and, Spanwise Load Dnttuihutions for a Senispan. Modpl. at
Mach0.J. NASATN D-8478,June 1977.

t5l Marsden, D. l.: Winglets far Sailplanzs.Technical Soaring
Vol. 15, No. 4, October 1991.

t6l Thorsen, O. R.: Tluoretical and Experimmtal Annlysis of
the Winglzx Designcd for thc High-Pnformance Sailplanc
ASW-27. Thesis, TU Delft, March 1999.

t7l Nicks, O. W: Wing Extensi,onfm Improaing Ctimb Perfor-
manu. AIAA-83-2556, October 1983.

t8l Nics, O. W. : I Physital Vbw of Wing Anodyna;rnbs. Techni-
cal Soaring, Vol. 17, No. 4, October 1993.

l9l Kiceniuk, T: Dyamit Soaring and Sailplanc Enngetic.
Free flight-vol libre, Dec{an 2001.

[0] Smith, S.: Do Winglets Work. Frc:e flight-vol libre,
Aug/Sep 1997.



I

Acta Polytechnica Vol. 43 No. 312003

[1] Masalc P.: Winglet Det@ fu Seilllsnps. Free flighwol
libre, ApriVlt4tay 1992.

[12] Crosby, C. P., Ashman, P., Terblanche, H.: Fu'll-Scal"e

In-Fliglrt hesswe Measurenents on a Winglct Fitted n an

AS-W 2|.Technical Soaring, Vol. 20, No. 3,July 1996.

Ing. Pavel Anderle
e-mail: anderle@aerospace.fsik.cvut.cz

Aerospace Engineering Department

Czech Technical University in Prague
I(arlovo ndm. l3
l2l 35 Prague 2,Czech Republic

Dr. Ladislav Smrtek
e-mail : ladislav@aero. gla.ac.uk
Dr. Frank N. Coton
e-mail: frank@aero. gla.ac.uk

Aerospace Engineering Department

James Watt Building
Universiry of Glasgow
Glasgow Gl2 8QQ
Scotland, United Kingdom


	Scan3
	Scan4
	Scan5
	Scan6
	Scan7
	Scan8
	Scan9