ABOUT WINGLETS

by Mark D. Maughmer

Over the past ten years, from initially being able to do little to improve
overall sailplane performance, winglets have developed to such an
extent that few gliders leave the factories without them. They are now a
familiar sight to nearly every soaring pilot. Few, however, really
understand what winglets do.

Understanding Drag

A winglet’s main purpose is to improve performance by reducing drag. To understand

how this is done, it is first necessary to understand the distinction between profile drag
and induced drag.

Profile drag is a consequence of the viscosity, or stickiness, of the air moving along the
surface of the airfoil, as well as due to pressure drag (pressure forces acting over the front
of a body not being balanced by those acting over its rear). As a wing moves through
viscous air, it pulls some of the air along with it, and leaves some of this air in motion.
Clearly, it takes energy to set air in motion. The transfer of this energy from the wing to
the air is profile drag.

Profile drag depends on, among other things, the amount of surface exposed to the air
(the wetted area), the shape of the airfoil, and its angle of attack. Profile drag is
proportional to the airspeed squared. Readers interested in a more thorough explanation
of these concepts are directed to refs. 1 and 2.

To measure an airfoil’s profile drag in a wind tunnel, a constant-chord wing section is

made to span the width of the wind-tunnel test section. In this way, the airflow is not free
to come around the wing tips. There is thus no flow in the spanwise direction -- the wing
section behaves as if it belonged to a wing of infinite span.

Induced drag is the drag that is a consequence of producing lift by a finite wing. If a
wing is producing lift, there must be higher pressure on the underside of the wing than on
the upper side. Thus, there is a flow around the wingtip from the high-pressure air on the
underside of the wing to the low-pressure air on the upper side (fig. 1). In other words,
there is spanwise flow on the finite wing that was not present on the infinite wing (fig.
2). This spanwise flow is felt all along the trailing edge as the flow leaving the upper
surface moves inward while that on the lower surface moves outward. As these opposing
flows meet at the trailing edge, they give rise to a swirling motion that, within a short
distance downstream, is concentrated into the two well-known tip vortices. Clearly, the
generation of tip vortices requires energy. The transfer of this energy from the wing to the
air is induced drag.
This process can be idealized as a “horseshoe” vortex system (fig. 3). As a consequence
of producing lift, “an equal and opposite reaction” must occur -- air must be given a
downward velocity, or downwash. With this downwash comes spanwise flow, tip
vortices, and induced drag. The goal is to minimize this drag by minimizing the amount
of energy used in producing the required downwash -- to reduce the energy that is
“wasted” in creating unnecessary spanwise flow and in the rolling up of the tip
vortices.

In observing the flowfield around the wing in Fig.2, it should be clear that the greater the
span, the less the tip effect is felt on the inboard portions of the wing. That is, the greater
the span, the more “two-dimensional like” will be the rest of the wing and, consequently,
the less its induced drag. As the span approaches infinity, the downwash and induced
drag approach zero. Likewise, if the wing is not producing lift, there will be no
downwash and thus no induced drag.

It is found that the induced drag is a function of the inverse of the square of the airspeed--
it is smallest at high speeds and increases as the aircraft slows down. It also depends on
the weight squared divided by the span squared, (W/b)2, how much weight each foot of
wing is asked to support. Thus, it increases with the square of the aircraft weight and
decreases with the inverse of the span squared.

Induced drag also depends on the wing design itself -- how efficiently it produces lift. As
a reference point, the most efficient planar wing (a wing with no dihedral or a winglet) is
one that has an elliptical loading (greatest at the root and decreasing toward the tip,
following the equation of an ellipse). Typical planar wings are slightly less efficient,
while non-planar geometries can be somewhat better than the elliptical case.

Controlling Induced Drag

It has been known for over a century that an endplate at the tip of a finite wing can reduce
spanwise flow and induced drag. Unfortunately, to be effective at this, the endplate must
be so large that the increase in skin friction drag due to excessive wetted area far
outweighs the reduction in induced drag.

A winglet provides a way to do better.3 Rather than being a simple “fence,” it carries an
aerodynamic load. The idea is to produce a flowfield that interacts with that of the main
wing to reduce the amount of spanwise flow. That is, the spanwise induced velocities
from the winglet oppose and thereby cancel those generated by the main wing.

This effect has been measured experimentally (Fig. 4). Here it is observed that the
spanwise flow has been largely eliminated by the presence of the winglet. In essence, the
winglet diffuses or spreads out the influence of the tip vortex such that the downwash,
and thereby the induced drag, is reduced. In this way, the winglet acts like an endplate in
reducing the spanwise flow but, by carrying the proper aerodynamic loading, it
accomplishes this with much less wetted area. Nevertheless, recalling the penalty of
profile drag with increasing airspeeds, the designer’s goal is to gain the most reduction in
induced drag for the smallest increase in profile drag.

The Winglet Design Process

My involvement began over a decade ago when I was asked by Peter Masak to help in
the design of winglets for the then-current crop of 15-meter racing sailplanes. Early
design procedures were based on the idea of a crossover point -- a breakeven airspeed
below which winglets improves performance by reducing induced drag and above which
their extra wetted area adds enough profile drag that performance is lower. Our first
successful winglets for sailplanes were guided by this notion. A trial-and-error approach
was employed that eventually led to some significant improvements.4 In 1989, one of
these designs was adopted by Schempp-Hirth as the “factory winglet” for the Ventus. In
retrospect, with the understanding that has come since, it seems that this process, while
systematic and logical, was accompanied with a great deal of luck. It now seems
somewhat remarkable that with the tool then at hand, we were able to come up with a
design that worked so well.

In spite of some success, I was somewhat frustrated by the lack of tools then available to
analyze or design winglets. Thus, along with a succession of excellent students, a
research effort was begun at Penn State to better this situation. In 1994, a collaborative
research arrangement with M&H Soaring (Monty Sullivan and Heinz Weissenbueller) in
Elmira, New York was begun. Their close proximity to Penn State, along with their
acceptance that it would not be a trivial matter to fabricate and flight test the number of
trials needed to develop and validate sound design methods, resulted in a fruitful and
enjoyable cooperation that continues still.

As our ability to predict the induced drag for a given wing geometry improved,5 so did
the ability to predict the effects on sailplane performance due to small changes in
geometry. With this, it became possible to design winglets that much more closely
achieved the intended results.

Some rules-of-thumb were established. First, whether it be with up-turned tips or

winglets, it can be beneficial to go “out-of-plane.” Second, while a great deal of work
has been directed toward achieving the minimum induced drag,6 our experience is that
driving a winglet toward this optimum penalizes the profile drag far more than it benefits
the reduction in induced drag. The design goal is clearly to minimize the overall drag,
not just the induced drag.

The cross-country performance of a sailplane can now be predicted with enough accuracy
to determine whether small changes in winglet geometry are beneficial or not. To do
this, straight flight and turning speed polars are calculated, including the influence of
variations in the spanwise lift distribution over the speed range, profile drag of the
aerodynamic surfaces as they depend on Reynolds number, flap deflections, and trim
drag.7 Optimum flap settings over the speed range are also computed. These results are
then used to predict average cross-country speeds in given weather conditions. After the
optimum bank angles are determined for a range of thermal strengths, sizes, and lift
profiles, a MacCready climb/glide analysis shows the average cross-country speed of the
glider as a function of thermal strength. So rather than design the winglet to simply not
hurt the lift-to-drag ratio below a certain airspeed, the winglet can be tailored to give the
best cross-country performance over a wide range of operating conditions.

Performance Gains: A Case Study

To see the performance increases that are possible with winglets, the predicted speed
polars for the Schempp-Hirth Discus 2, with and without winglets, ballasted and
unballasted, are shown in Fig. 5. Although gains are demonstrated, they are difficult to
assess using the polars shown. Thus, the data are replotted in terms of L/D verses
velocity in Fig. 6. In addition to demonstrating the gains from carrying water ballast at
higher cruising speeds, the benefit of winglets can now be seen. To get an even better
idea of the gains in L/D, in Fig.7 these data are again replotted in terms of the percentage
increase in L/D relative to the unballasted and ballasted glider without winglets. Note
how this winglet's crossover point occurs at airspeeds that are above the maximum
allowable -- there are no allowable flight conditions in this case for which the winglets
penalize performance. Although a slight overall gain could be achieved by tailoring the
winglet more for climb, this would result in relatively large penalties at high speeds.
While the percentage gain in L/D does not appear to be very great, it is significant that it
comes without any penalty at higher speeds.

The effect of winglets on the percentage change in average cross-country speed relative
to that of the baseline aircraft is presented in Fig. 8. The winglets improve the cross-
country performance for all the thermals considered, that is, for thermals having a 500’
radius and strengths of up to 12 kts. As expected, the performance gain is very
significant for weak thermals -- having winglets allows for some climb rate, whereas
without them it is minimal or zero. As the thermal strengths increase, the benefits
decrease; however, for this glider winglets do not hurt cross-country speed even for
average thermal strengths of more than 12 kts.

The point at which full water ballast becomes beneficial is indicated by the crossing of
the unballasted and ballasted curves at an average thermal strength of about 8 kts,
corresponding to a climb rate with full ballast that is predicted to be about 5.2 kts. In this
case, “full ballast” corresponds to a wingloading of 10.6 lbs/ft2 rather than the 9.0 lbs/ft2
allowed by U.S. Standard Class rules. As indicated, ballast causes a reduction in average
cross-country speed for average thermal strengths of less than 8 kts. For thermal
strengths greater than this, winglets improve the cross-country speed, but only by a half-
percent or so. The glider with winglets, however, can profitably carry ballast in slightly
weaker conditions than can the glider without winglets.
Other Issues

From the experience of designing winglets for a variety of sailplanes (as well as for a few
non-sailplane applications), it seems that all wings can be improved with winglets,
although the better the original wing from an induced drag standpoint, the smaller the
gain possible with winglets (and the more difficult is the design process).

It is sometimes heard that winglets were tried on a certain glider and did not work. What
this really says is that a particular poor design did not work. As an example of how
critical some of the design issues are, the effect of toe (incidence) angles on the Discus 2
winglet design is presented in Fig. 9. Obviously, a small deviation from the optimum can
cause the winglet to become a speed brake. Furthermore, each glider must have winglets
specifically designed for it -- rules of thumb can be disastrous. From personal
experience, there is no doubt that it is much easier to make a glider worse with winglets
than it is to make it better!

Winglets sometimes can fix problems of the original wing. For example, in the case of a
flapped glider, it is important that the flaps/ailerons extend to the wingtip. Otherwise,
when the flaps are deflected upward for high-speed cruise, the tips are loaded far more
than they should be. Although only a small portion of the wing is influenced, it can result
in a very significant induced drag increase. In these cases, cutting the tip back to the
aileron in order to mount the winglet has resulted in unexpected gains, especially at high
speeds.

By understanding of how winglets achieve induced drag reduction, it also becomes clear
how they can produce other performance and handling gains. In particular, it has been
found that winglets improve the flow in the tip region and thereby improve the
effectiveness of the ailerons. This is in part due to the local angle of attack in the vicinity
of the ailerons being reduced less by the reduced downwash velocities, and by the
reduction of spanwise flow helping to keep the ailerons effective. One of the benefits of
greater control effectiveness is that smaller aileron deflections are required for a given
rolling moment. This means less drag for a given roll rate and a higher maximum roll
rate. Likewise, woolen tufts attached to glider wings have shown that much of the flow
over the inside tip during turning flight is separated, which is nearly eliminated by the
presence of a winglet. Winglets also benefit safety -- ailerons remain effective much
deeper into a stall than before.

The improvement in handling qualities are very succinctly described by Werner Meuser,
the current 15-Meter Class World Champion, in a message sent to me by the Schempp-
Hirth factory describing his first impressions of the new Ventus 2ax winglets. “…..very
impressed by the handling change. He reported the glider got more gentle and harmless
at low speeds and felt very ‘clean’ close to stall speed., which seems to have decreased
remarkably. Even in steep circles, there was no tendency to stall or ‘misbehave’. Before
anything at the wing started to separate, he felt the tailplane couldn’t handle the angle of
attack anymore. Allover, very positive impressions….”
Conclusions

Although performance gains achieved with winglets are only a few percent at moderate
thermal strengths, such small differences can be important in determining the outcome of
many cross-country flights and contests. For example, in the 1999 U.S. Open Class
Nationals, just 68 points separated the first six places. This difference amounted to less
than 1.5% -- far less than the performance advantage that can be achieved using well-
designed winglets.

Since their shaky introduction many years ago, the acceptance of winglets is now
widespread. At the World Championships in Uvalde, Texas in 1991, of 105 competing
gliders, 19 used winglets. At the most recent championships in South Africa, essentially
every glider entered had winglets or some type of tip treatment.

Thus, after more than a decade of winglets being applied to sailplanes, it is clear that the
benefits are far-reaching. If properly designed such that the profile drag penalty is of
little consequence over the range of airspeeds at which the glider is flown, then there
seems to be no reason whatsoever not to take advantage of the performance and handling
qualities benefits that winglets offer.

Finally, although some of the spinning characteristics of gliders with winglets have been
explored, testing has not been extensive. The anecdotal evidence indicates that gliders
with winglets are more reluctant to spin, but once they do, the altitude required for
recovery is somewhat greater. Given that many glider fatalities are a consequence of
stall/spin accidents during approach, at altitudes from which safe recovery is not possible,
a question worth pondering is whether even the most basic training gliders might benefit
from the installation of winglets.

------------------------------------------------------------------------------------------------------------

References
1
Thomas, F., Fundamentals of Sailplane Design, Translated by Judah Milgram, College Park
Press, MD, 1999.
2
Falk, T.J. and Matteson, F. H., “Sailplane Aerodynamics,” American Soaring Handbook,
Soaring Society of America, 1971.
3
Whitcomb, R.T., “A Design Approach and Selected Wind-Tunnel Result at High Subsonic
Speed for Wing-Tip Mounted Winglets,” NASA TN D-8260, July 1976.
4
Masak, P.C., !Design of Winglets for Sailplanes," Soaring, June 1993, pp. 21-27.
5
Mortara, K.W. and Maughmer, M.D., “A Method for the Prediction of Induced Drag for Planar
and Non-Planar Wings,” AIAA Paper 93-3420, Aug. 1993.
6
Munk, M.M., “Minimum Induced Drag of Aerofoils,” NACA Technical Report No. 121, 1921.
7
Maughmer, M.D. and Kunz, P.J., !Sailplane Winglet Design," Technical Soaring, Vol.
XXll, No. 4, Oct. 1998, pp. 116-123.
Fig. 1 Higher pressure air on the wing lower surface flowing around
wingtip to upper surface.

Fig. 2 Spanwise flow on a finite wing - solid lines, upper surface;

dashed lines, lower surface.
 Fig. 3 Idealized “horseshoe” vortex system.

Fig. 4 Experimentally determined flowfield crossflow velocity vectors

behind model with and without winglets.6
 V (kts)
0 40 60 80 100 120
0

2
VS (kts)

4
Discus 2, 685 lbs
Discus 2 WL, 685 lbs
6 Discus 2, 1060 lbs
Discus 2 WL, 1060 lbs

Fig. 5 Predicted straight flight polars of unballasted and ballasted

Discus 2, with and without winglets.

50

40

L/D

30

20
Discus 2, 685 lbs
Discus 2 WL, 685 lbs
Discus 2, 1060 lbs
10 Discus 2 WL, 1060 lbs

0
40 60 80 100 120
V (kts)

Fig. 6 Comparison of predicted lift-to-drag ratios for unballasted

and ballasted Discus 2, with and without winglets.
 3

Discus 2 WL, 685 lbs

2 Discus 2 WL, 1160 lbs

∆ L//D %
1

0
40 60 80 100 120
V (kts)

-1

Fig. 7 Percentage gain in predicted lift-to-drag ratios due to

winglets for unballasted and ballasted Discus 2.

4 Baseline: Discus 2, 685 lbs

∆ VCC %

Discus 2 WL, 685 lbs

Discus 2, 1160 lbs
3 Discus 2 WL, 1160 lbs

0
2 4 6 8 10
Thermal Strength (kts)
-1

Fig. 8 Percentage gain in predicted average cross-country speed

due to winglets and ballast relative to unballasted Discus 2 (685 lbs)
without winglets.
 5

∆ VCC %
Baseline: Discus 2, 685 lbs

Winglet Toe Angles

3 (From zero-lift angle)
-3 deg
1 deg
2
5 deg

0
2 4 6 8 10

-1 Thermal Strength (kts)

Fig. 9 Percentage change in predicted average cross-country speed

as it depends on winglet toe angle for an unballasted Discus 2. Toe
angles are measured relative to the zero-lift angle of attack.
WINGLET DESIGN
FOR SAILPLANES
Peter Masak
IN THE ONGOING QUEST for higher perfor- the French manufacturer Centrair. The over- design an efficient pair of winglets for a Nim-
mance sailplanes, winglets have provided a riding concern repeatedly expressed by rac- bus III for the World championships in 1989
means for improving the performance with ing pilots was that the winglets, although they at Wiener Neustadt, Austria.
only a modest price per L/D point gain. Wing- were known to provide a significant gain at
lets act to reduce induced drag and act to low speed, would detract from performance Marsden had proposed using an unusual dou-
control the crossflow in the tip region of the at the high speed cruise condition, with a ble element winglet on the Nimbus III (emu-
wings in such a way as to improve the han- resulting net loss or perhaps no achieved gain lating the primary wing feathers of a soaring
dling characteristics at the same time. in overall performance. bird) which was inspired by a successful ver-
sion on Marsden’s DG–200. His experiments
By introducing a vertical cambered surface This concern is justified since winglets act to had shown that he was obtaining a significant
at the tip, the downwash field behind the wing reduce both induced drag and drag due to improvement in lift capability of a tip section
is spread horizontally by several inches. Since crossflow at the tip; however, at high speed fitted with winglets.
the induced drag is inversely proportional to neither of these effects are large and thus
the effective width of this downwash field, the there is some speed at which the overall sur- Experiments with dual winglets
winglet therefore acts to reduce induced drag face friction drag of the winglet exceeds the The initial promise of dual winglets on the
by displacing the vortices outward. Presum- induced/interference drag reduction provided Nimbus III tips did not prove out in either
ably the greatest effect would be obtained by by the winglet. The graphs below show this flight tests or wind tunnel tests. Although a
introducing a high lift large surface winglet effect with large winglets added to an ASW– gain in lift was measured, the interference
which would displace more air outward and 19 at Braunschweig. Clearly the key is to pro- drag of the two lifting surfaces caused the
alter the circulation pattern in a more signifi- vide a minimum drag surface which does not airflow across the rear winglet to be sepa-
cant way. However, the design of winglets stall at circling speeds. rated at even modest lift coefficients. This re-
involves the compromise of maximizing the sulted in the winglet not being effective at
low speed improvement without sacrificing Prompted by interest from Dr. David Marsden either high or low flight speeds. At speeds
high speed performance. Pilots will not fly at the University of Alberta, and my own suc- below 55 knots, the rear winglet would expe-
with winglets if they perceive any deteriora- cessful experience a decade ago with a home- rience massive separation (seen with tufts);
tion of high speed performance. built HP–18, the challenge was struck to and at speeds higher than that, the winglet

BACKGROUND
0
First use of winglets
Winglets for modern aircraft were first pro-
posed by Dr. Richard Whitcomb, at NASA 0.4
Langley in the mid–1970’s. At that time, wind with winglets
tunnel models and subsequent full size flight
without winglets
tests on a Boeing 707 commercial jetliner dem-
0.8
onstrated a significant reduction in total drag
sink rate – m/s

at high lift coefficients.

After the publication of the design philoso- 1.2

phy, numerous researchers in industry tack-
led winglet design with varying degrees of speed polars
success. Most tried to use potential flow meth- 1.6
ods for predicting tip inflow angles and sur-
face pressure distributions, however given the
nature of the flow field at the tip, this has lead 2.0
many investigators to the wrong conclusions.
60 80 100 120 km/h 140 160
Potential flow analysis seems to steer the de- 40
signer in the direction of excessively large
winglets, while experimental data suggests
that large winglets pay a greater–than–pre-
dicted penalty in high speed performance. 35
Since potential flow methods cannot accu- L/D
rately predict the vortex roll–up at the tip, or
the influence of secondary flows on the bound- 30
ary layer, these methods have not provided
the complete picture of the effect of winglets lift – drag polars
on performance. Also, potential flow methods 25
do not show the significant influence of the
effect of the fore–aft position of the winglets.
20
Experience with sailplanes
In sailplane racing circles, winglets were tried
and then dropped by a number of university Figure 2 — Influence of winglets on the performance of an ASW–19
flying groups (Darmstadt, Braunschweig), and

6 free flight 2/92

 Ratio of winglet root chord to sailplane tip chord winglet at low inflow angles (high speed =
It would seem that the winglet might ideally low coefficient of lift, Cl).
be designed as an extension of the wing, and
thus the optimum winglet would be a smooth Taper ratio
transition of the wing from horizontal to verti- The effect of taper ratio on inflow angles and
cal. Experiments suggest otherwise. the resulting optimum twist distribution was
analyzed theoretically by K.H. Horstmann in
friction drag due to the highly cambered air- If the root chord of the winglet is equal to the his PhD thesis. It was shown that as taper
foils was so high as to cause an overall loss. tip chord of the wing, then the inflow angle at ratio increases, the optimum twist distribution
the tip will be less than when the winglet is a for the winglet varies more linearly from root
Second Iteration smaller fraction of the tip chord. The result to tip. From a construction standpoint it is
The narrow tip chord of the Nimbus III (9 in) will be that at high speed, the inflow angle also easier and more accurate to build a wing-
forced an abnormally low chord for the dual may not be sufficient so as to prevent separa- let with a linear change in twist angle along
winglets (3–4 in). The resulting low Reynold’s tion of the airflow from the outer (lower) sur- the winglet span. This favours a winglet with a
number of the winglet elements probably con- face of the winglet. Since other considera- larger tip chord. We also want to try to maxi-
tributed to the separation problem and high tions require that a toe–out angle be set (about mize the tip chord so as to maximize the
drag. Thus it was evident that this design –3 degrees), it is desirable to allow some vor- Reynold’s number. Accordingly, a ratio of tip
could be improved by going back to the con- tex induced flow to wrap around the wingtip to root chord of 0.6 was selected.
ventional single element winglet. (An airfoil’s and provide a positive angle of attack for the
Reynold’s number is related to its size — all winglet at all flight speeds. Toe–out
else being equal, a small airfoil does not The determination of toe–out was based on
“work” as well as a large one. The Re of a For the various winglets fabricated, the fol- the simple consideration that we were trying
typical sailplane wing is 1,000,000. ed.) lowing ratios of root chord of the winglet to tip to maximize the speed at which no further
chord of the wing were used: benefit is gained from the winglet, and thus
DESIGN OPTIMIZATION select an angle of attack (α) setting for the
• DG–600 0.60 • Discus 0.70 winglet that will minimize the high speed drag.
Apart from the selection of a winglet airfoil, • Ventus 0.57 • Nimbus III 0.95
there were five key parameters that had to be • ASW–20 0.50 Considering the Cl–vs–α prediction for the
chosen to optimize the design: PSU–90–125 winglet airfoil, an angle of attack
The choice of the root chord of the winglet is of –3 degrees corresponds to a Cl of 0. Given
• Cant angle • Twist distribution also constrained by the nominal tip chord of the fact that even at high speed there is a
• Sweepback • Taper ratio the wing, and by considering Reynold’s num- small inflow component at the tip, the winglet
• Ratio of winglet root chord to sailplane ber effects. Too small a winglet chord can will actually be generating a slightly positive
tip chord result in extensive laminar separation and high lift, even with the –3 degree root toe–out. Cal-
drag. For the Nimbus III and Discus winglets, culations show that when the wing is operat-
Cant angle the small nominal tip chords force the winglet ing at a nominal lift coefficient of 1.0 (which
The selection of cant angle evolved from an geometry to be smaller than would be desir- corresponds to the circling lift coefficient), the
unusual consideration specific to sailplanes: able from a Reynold’s number consideration. lift coefficient of the winglet is 0.6 at the root
the narrow and highly flexible wings pro- and reduces to zero at the tip.
vide for a wingtip angle in flight which can Twist distribution
approach 30 degrees on some sailplanes The twist distribution on a winglet is normally WINGLET AIRFOIL
when flying with water ballast. A more com- selected so as to provide a uniform load dis-
mon angle for modern 15 metre ships is 7–12 tribution across the winglet span. Since the The winglet airfoil was designed with the fol-
degrees. inflow angle is higher at the base, the winglet lowing criteria in mind:
is twisted to higher angles of attack toward
On winglets that are nominally set to a cant the tip. This is opposite to the general design • to minimize drag at low Cl conditions
angle of 0 degrees (at right angles to the methodology for wings, which normally have • to design the winglet airfoil to be tolerant
wing), as the wing deflects, the winglet gen- washout (either geometric or aerodynamic) of low Re
erates a sideload in flight which has a com- so as to decrease the angle of attack towards • to maximize tolerance to negative α
ponent oriented downward. This is a self the tips.
defeating situation, since the winglet is gen- These design requirements are different than
erating additional drag by contributing to the The determination of optimum twist for our for a conventional sailplane airfoil. The result-
weight of the aircraft. Thus a more reason- winglets was made by iterating experimen- ing custom airfoil designed by Dr. Maughmer
able approach is to set the winglets at least tally. When flight tested, the first set of wing- and Mr. Selig of Pennsylvania State Univer-
at a cant angle on the ground of 0 degrees lets fabricated stalled at the root first with a sity is shown in the figure below. Dr. Maugh-
plus the in–flight local tip deflection angle. progressive stall developing upwards towards mer described the airfoil design philosophy
the winglet tip. By twisting the winglet to in- as follows:
Sweepback crease the angle of attack at the tip, the en-
The selection of the sweepback angle was tire surface of the winglet could be made to “The airfoil has the traditional undercamber
based on experimental observations. It was stall simultaneously. Two degrees of twist from removed from the lower surface trailing edge
first believed that the sweepback angle for root to tip proved to be optimum. area, which minimizes the tendency to form
the winglet should be equal to that for the detrimental laminar separation bubbles at low
main wing (0 degrees), however experience The second benefit of positive twist on the or negative angles of attack. At the price of a
proves otherwise. If a vertical winglet with no winglet is that the high speed performance is little Clmax, which isn’t important for a winglet
sweepback is built, it will be observed that enhanced — there is less likelihood of devel- anyway, the drag is lower than other sailplane
the root of the winglet will stall first and that oping separation on the outer surface of the airfoils everywhere up to Cl = 0.85, as well as
the tip will remain flying.

The optimum situation from an aerodynamic

standpoint is to have the aerodynamic load-
ing such that the entire winglet surface stalls
uniformly. This can be achieved by sweeping
back the winglet, which will increase the load-
ing on the tip. Because of the rapid variation
in angle of attack of the winglet as a function
of height, a large degree of sweepback is 0 50 percent of chord 100
required to load the tip correctly. For our wing-
lets, a 30 degree leading edge sweep angle PSU–90–125 winglet airfoil
was used to achieve this effect.

2/92 free flight 7

at negative Cl’s, so that sideslips and hori- FINAL DESIGN The dolphining performance is naturally im-
zontal gusts can be tolerated. The corners of proved with the winglets since they act to
the laminar bucket have been rounded to The final choice of design parameters is re- reduce induced drag while pulling positive
avoid unstable yawing moments that would flected in the design of the Ventus and ASW– ‘g’, and several pilots have perceived their
be generated otherwise if the sailplane yawed 20 winglets, which have been highly success- sailplanes to have improved glide perform-
to angles exceeding those corresponding to ful in competition. The ASW–20 winglet went ance even at high cruising speeds in strong
the sharp corners of the traditional Wortmann through two iterations and the Ventus, three, weather.
sailplane airfoils. Finally, the airfoil was de- before it was concluded that the design had
signed to avoid laminar separation bubbles reached a high level of refinement. Flight Test Data
down to Re = 350,000.” These positive results are confirmed by flight
FLIGHT TEST RESULTS tests based on three high tows with each sail-
WING AERODYNAMICS plane type which show the following perform-
Competition Results ance gains as measured by the two–glider
The change in the lift distribution of a wing The response of pilots flying with winglets in comparison technique.
with and without winglets is shown below. The competition has been very positive overall.
boundary condition at the wingtip of the main Certainly one of the measures of the success
wing no longer requires that the lift taper to of the design is the fact that pilots after a ASW–20 flight test data:
zero at the tip. The assumed lift distribution period of evaluation have chosen to fly with (pilots –Striedieck, Seymour)
for a wing with a winglet is assumed to termi- the winglets. At the 1991 World contest in
nate at an imaginary point equal to unfolding Uvalde, Texas, ten pilots chose to fly with our speed duration ∆ with ∆
the vertical winglet in the horizontal plane. As winglets – 8 Ventus, 1 ASW–20B, and 1 Nim- winglets ft/min
a result the outer portion of the wing carries a bus III. At the end of the contest, a Ventus 50 mi/h 5 min + 30 ft 6
higher load than it does without the winglet. flying with our winglets had won four of twelve 65 mi/h 5 min + 7 ft 1.5
Recent calculations on sailplanes with dou- contest days and on the fastest day of the 80 mi/h 2 min + 10 ft 5
ble trapezoidal planforms such as the ASW– contest, the top five places in the 15 metre 100 mi/h 2 min 0 0
20 or LS–6 suggest that this outer tip loading class went to sailplanes flying with our wing-
is more efficient from the standpoint of in- lets. Additionally the trophy for the highest Ventus flight test data:
duced drag. speed achieved overall went to Jan Anderson (pilots –Mockler, Masak)
of Denmark, flying a Ventus with our winglets
Secondly, the additional lift capability of the (his speed also exceeded the highest speed (knots) flap ∆ with
main wing means that the Clmax of the overall achieved in the Open Class). Two weeks prior, winglets
wing is increased and the sailplane’s circling at the 15 metre Nationals in Hobbs, New 40 dry, 53 wet +2 9.1 ft/min
performance will be enhanced. Mexico, Reinhard Schramme from Germany 50 dry, 66 wet 0 9.0 ft/min
established an unofficial record of sorts by 60 dry, 79 wet 0 9.8 ft/min
Structural Loading flying his Ventus–C around a closed course 84 dry, 110 wet –2 3.3 ft/min
One of the key advantages of winglets is that of greater than 500 km with an average speed
they provide a performance increase while of 171 km/h (he would have won were it not
only fractionally increasing the root bending for a photo penalty). Maximum performance gains
moment on the spar compared to a span ex- with Masak winglets
tension. Whereas the moment arm of a span Bruno Gantenbrink and Hermann Hajek of
extension is one–half the semi–span of the Germany chose to retrofit winglets to their sailplane winglet airfoil L/D gain
wing (about 7.5 metres), the moment arm of a Ventus–C’s and were delighted with the han- ASW–20 NASA Van Dam 2.1
winglet is only equal to approximately one– dling and performance qualities that they ob- Discus PSU–90–125 2.5
half the vertical span (0.3 m) plus the de- served. Mr. Hajek noted as a particular ad- Ventus PSU–90–125 3.5
flected wing elevation at the tip. For sailplanes vantage the improvement in his ability to main-
which are certified with tip extensions, one tain constant bank angle and speed with a
can be assured that the winglet will not over- full load of water. With winglets the effective CONCLUSIONS
load the wing and all standard operating limi- dihedral is increased and the sailplane can
tations will apply (Ventus, ASW–20, DG–600). be banked steeper while retaining control. The overall performance gains measured in
free flight on sailplanes retrofitted with wing-
lets are impressive and are supported by posi-
tive contest results. Handling qualities are im-
span loading with winglet proved in all cases, including improvement in
roll rate and roll authority at high lift condi-
additional lift from wing tions.
in presence of winglet

span loading without winglet The performance measurements have shown

a higher gain in performance than would oth-
erwise be predicted by conventional theory.
It is believed that major benefits are derived
from inhibiting the secondary flow that con-
taminates the boundary layer near the tip re-
gion. Prediction of this phenomenon requires
computational power out of my grasp, and
the present designs have been developed
Lift distribution on a wing with and without winglet via experimentation and in–flight testing.

By August 1991, there were over forty–five

sailplanes in the world flying with winglets
designed and fabricated by the author. No
moment arm negative reports or dangerous incidents (ie.
of winglet lift
flutter) of any kind have been reported. As a
result of the positive service experience,
Transport Canada have recently issued a sup-
plementary type certificate for flight with wing-
lets on the Ventus model, using JAR–22 as a
moment arm of lift from span extension basis for compliance. •

A bibliography is on page 13

8 free flight 2/92

 induced downwash is reduced. A well de-

Do winglets work? signed winglet is equivalent to about half

its height in span increase. At the same time,
the winglet adds much less additional struc-
tural load to the wing than a tip extension
does. Detailed studies of the combined struc-
tural and aerodynamic effects of winglets
on transport aircraft show that they are not
quite equal in overall performance to a sim-
Steve Smith ple span extension. Current conventional
from Pacific Soaring Council West Wind edge, not just at the tip. The distribution of wisdom states that winglets should only be
lift along the span of the wing determines used in cases where there is some limiting
ELL, in a word, yes. But don’t they how much vorticity is shed along the trail- constraint on wingspan. Applying these re-
W hurt high speed performance? Not
necessarily. It seems I’m often discussing
ing edge. It can be proven that for a planar
wing (no winglets), the induced drag is the
sults to sailplane design would indicate that
winglets should not be used on Open class
winglets with glider pilots. So I’d like to try smallest when the spanwise distribution of sailplanes, but should be used on 15 metre
to provide some technical framework for lift is shaped like an ellipse. This lift distri- and Standard class sailplanes.
understanding what winglets do. bution produces the vorticity distribution
with the minimum energy. In steady flight, What about high speed performance?
Sources of Drag First, in order to under- induced drag varies in proportion to the Looking at the figure, you can see that in-
stand winglets, you need to understand drag. square of the weight, and inversely with the duced drag becomes unimportant at high
Airplanes have three primary sources of square of the wingspan and velocity: speeds, whereas the parasite drag becomes
drag. The first source is often called para- dominant. A crossover point occurs where
Dinduced = kµ( W / bV ) 2
site drag or profile drag, and this has to do the induced drag benefit of the winglet is
with the skin friction created by airflow over If the aircraft is heavier, it needs more lift, outweighed by the increase in parasite drag.
the aircraft surface. The second source is and so produces more induced drag. If the
called induced drag, which is a result of lift is distributed over a longer wingspan, Here’s a realistic example. Suppose a wing-
generating lift with a finite wing span – let is installed that reduces the induced
an infinite wing would be nice, but it drag by 10% and adds 1% to the para-
won’t fit in your trailer! The third drag Typical drag curve for a sailplane in steady flight site drag. At the speed for best L/D,
source is caused by compressibility ef- where induced drag and parasite drag
fects on aircraft that fly nearly as fast as are equal, the net improvement would
the speed of sound, or faster. Except for 50 be 4.5% (.5 x .1 – .5 x .01 = .045). This
John McMaster’s Altostratus, we don’t amounts to about 6 ft/min for a typical
need to worry about compressibility 40 15 metre sailplane. At a speed of 1.73
drag. The primary effect of winglets is to drag (lbs) times the best L/D speed, parasite drag
reduce the induced drag. 30 is 90% of the total, and induced drag
only 10%. At this speed, the net im-
total icit
Parasite drag is naturally affected by the 20 ras
provement is almost zero (.1 x .1 – .9 x
amount of wetted surface area. It also stall pa .01 = .001). For a sailplane with a best
depends on whether the boundary layer 10 L/D speed of 60 knots, the theoretical
is laminar or turbulent – but that’s an- induced crossover speed for these winglets is 104
other story. For now, you need to know knots. Above this speed, these winglets
that parasite drag increases in propor- airspeed (kts) 30 60 90 120 degrade performance.
tion to the square of the airspeed. This
turns out to be sort of universal – most But overall cross-country performance
aerodynamic forces increase in proportion the trailing vorticity is spread out more as is a balance between the low and high speed
to the square of the velocity, because the well, dissipating less energy. If the aircraft performance. Classical MacCready theory
ability of the air to produce forces is related flies faster, it produces the same lift with indicates that 50% of the time is spent cruis-
to the kinetic energy in the flow: less angle of attack, less disturbance to the ing and 50% climbing. In this case, the
2 flow, and creates weaker vorticity in the break-even speed would occur where the
Dparasitic = kµV
trailing wake. disadvantage at high speed equals the ad-
Induced drag is a bit more complicated. A vantage at low speed. Because the actual
finite wing ends with a wingtip, where the For a given aircraft weight, the total drag is drag is much higher at cruise, we can’t com-
higher pressure air under the wing can leak the combination of the parasite drag and pare on a percentage basis. The compari-
around the end and fill the low pressure the induced drag. Looking at the above son must be made based on actual sink
area on top of the wing. This flow around diagram, you can see that a minimum drag rate. Since half the time is spent cruising,
the tip forms a vortex that trails off down- point occurs where the parasite drag and the break-even cruise speed occurs where
stream. The flow around the tip also re- the induced drag are equal. At lower speeds, the increased sink rate equals the reduced
duces the lift in the area near the tip by the parasite drag is small, but the induced sink rate at low speed. In other words, how
tending to equalize the low pressure above drag increases very fast. At higher speeds, fast do you need to fly so that the sink rate
the wing. The vortex contains energy in the parasite drag increases but induced drag with winglets is 6 ft/min greater than with-
form of the swirling flow velocity. We call becomes small. This trade-off between para- out winglets? For the example used here,
the force required to pull the wing along to site drag and induced drag is what makes this occurs at 2.3 x best L/D speed or 138
produce these tip vortices “induced drag”. the design of winglets interesting. knots. It’s pretty rare that your MacCready
The mechanism through which the wing directed speed to fly would be this fast!
“feels” the presence of the tip vortices is the How do winglets reduce induced drag?
downward velocity induced on the wing by Adding a winglet to a wing has a similar You might point out that as soaring condi-
the vortices. It is as if the wing is flying in a effect to adding wing span. By providing tions become stronger, the MacCready
self-generated region of sink. more length of trailing edge, the vorticity is model doesn’t apply: the fraction of time
spread out more for the same total lift, so spent circling becomes much smaller. But
This concept is very oversimplified – a more the energy loss is less. The detailed interac- that doesn’t necessarily mean that the time
realistic explanation requires a fair bit of tions between the wing and winglet are a spent flying slow (near best L/D) also be-
math and physics. What really happens is bit different than a simple span extension, comes small. Efficient use of cloud streets
that vorticity is shed all along the trailing but the effect is similar. In both cases, the still dictates flying slowly in good lift. So,

16 free flight 4/97

suppose you never fly slower than 70 knots. tip chord very small, but if the chord is too The impression is that the aircraft “grooves”
At this speed, the winglets improve your small, it will be prone to stall early. So, better in a turn. The increase in aileron ef-
sink rate by almost 4 ft/min. You would now you want to put a tip extension on the fectiveness and the reduction in adverse yaw
need to fly 118 knots in order for the wing- wing, and you still try to achieve that ellip- both come from the lift of the winglet when
let penalty to be 4 ft/min, negating the ben- tical lift distribution, but the tip chord must the aileron is deflected. When the aileron is
efit. About the only situation where soaring not get too small. So, you maintain more deflected, there is less “tip loss” of the added
speed is consistently high enough that wing- surface area and compensate by reducing lift. There is much less of an increase in the
lets would actually hurt overall is ridge run- the airfoil camber or twisting the wing tip vortex strength, again because the vor-
ning. Even in ridge soaring, there may be slightly to reduce the tip angle of attack. ticity is spread out along the longer trailing
long gaps to cross where the benefit of the edge, and the tip is further away. As a re-
winglets would offset any cruise penalty. The added wetted surface area increases sult, adverse yaw may be eliminated. For
the parasite drag. The second effect explains heavily ballasted sailplanes, the increased
Can the same argument be applied why winglets can have such a small chord control and safety offered by the winglets
to tip extensions? (and therefore smaller wetted area) without may be a big advantage, regardless of any
Well, that depends on the structural limi- stalling. As the sailplane slows down and improvement in glide performance.
tations on the sailplane. First of all, for the angle of attack increases to maintain
the same improvement in induced drag, a the lift equal to the weight, the tip exten- Other bad things about winglets
shorter span extension will be required sion experiences the same angle of attack One disadvantage that is not often discussed
(about half, right?) but the tip extension has increase, but a winglet does not. The flow is the reduction in flutter speed. Classical
more wetted area, so more parasite drag. angle experienced by the winglet is deter- flutter occurs when the natural frequency
This added area is needed to prevent the tip mined by the strength and distribution of in bending and the natural frequency in
extension from stalling at low speed. The the trailing vorticity, which is indirectly in- torsion get too close together. The torsion
reason winglets don’t need the same area fluenced by the increased angle of attack. frequency is always somewhat higher than
to prevent stalling will be explained later. The net result is that the effective increase the bending frequency. By adding weight
in angle of attack for the winglet is much above the plane of the wing, the torsional
Anyway, a tip extension equivalent to the less than the increase in angle of attack on moment of inertia is increased, which re-
winglet example might improve induced the wing. So, the lift doesn’t build up as fast duces the torsion frequency of the wing.
drag 11%, but add 2% in parasite drag. At on the winglet and the wing stalls first. In Of course, tip extensions also reduce flutter
the best L/D speed: .5 x .11 – .5 x .02 = practise, this effect is exploited to reduce speed. Both can be compensated for by
.045 (once again). But there is a crucial the wetted area of the winglet as much as clever addition of balance weights to the
assumption hidden in these examples. The possible to the point where, ideally, the wing, but this is a complex problem requir-
comparison is made at constant weight. If wing and winglet would stall at about the ing sophisticated analysis.
you install your tip extensions, are you al- same time.
lowed to ballast the sailplane to the same Conclusion I hope I’ve answered more
weight? If so, then the example is still valid. Other good things about winglets questions than I’ve raised. I’m happy to dis-
Now compare the performance of this tip Aside from the performance improvement cuss winglets in more detail with anyone,
extension at 1.73 times the best L/D speed, offered by winglets, there are other bene- feel free to contact me by email at
where parasite drag is 90% of the total, and fits. The most notable of these are the [email protected] ❖
we find: .1 x .11 – .9 x .02 = -.007. So, increase in dihedral, increase in aileron
now the tip extension that appeared to be effectiveness, and the reduction of adverse Steve Smith is a Senior Aerospace Engineer
equivalent at low speed degrades high speed yaw. The increase in effective dihedral im- at the NASA Ames Research Center. A full
performance 0.7% at the speed where the proves handling in thermals. There is less discussion of winglet design concepts can
winglets still provide a 0.1% benefit. One need for “top stick” to prevent a spiral dive. be found in “free flight” 2/92 p6.
way to explain this is to say that the tip
extension reduced the wing loading. What
is really happening is that the parasite drag
was increased for the same weight. What if
you must reduce the gross weight when
you install the tip extensions? In that case,
• safe and docile handling
the tip extensions hurt even more. This
also illustrates why high wing loading is
KR-03a “Krosno” • affordable price for clubs
• fantastic cockpit visibility
so important for Open class sailplanes. • advanced corrosion control
the trainer we’ve all been
waiting for and polyurethane finish
The results here depend on many assump- • robust metal structure has been
tions, but they do challenge the conven- fatigue tested to 18,000 hours
tional wisdom that winglets are not as good
as tip extensions. One major difference
between sailplanes and transport aircraft is
the range of speeds over which they per-
form. Transport aircraft adjust their cruising
altitude so that they cruise only slightly faster
than the best L/D speed, but sailplanes are
expected to perform well at almost twice
the best L/D speed.
3A
What about stall? I mentioned that tip KR
-0 Now flying
extensions are prone to tip stall, but wing- in Canada!
lets are not. Two effects come into play
here. First is that fact that as you scale down
an airfoil, the critical angle of attack for
stall is reduced. This is called a “Reynolds
number effect”. In essence, the basic char-
Ed Hollestelle, 2371 Dundas St. London ON N5V 1R4
(519) 461-1464 phone & fax SOLAIRE
acter of the flow is affected by the size of [email protected] CANADA
the wing. To achieve the desired elliptical
lift distribution, you would like to make the

4/97 free flight 17