AIAA 2001-2406

THE DESIGN OF WINGLETS FOR HIGH-PERFORMANCE

SAILPLANES
Mark D. Maughmer*
The Pennsylvania State University
University Park, Pennsylvania 16802
Copyright 2001 by Mark D. Maughmer. Published by the
American Institute of Aeronautics and Astronautics, Inc., with
permission.

Abstract
Although theoretical tools for the design of
winglets for high-performance sailplanes were initially
of limited value, simple methods were used to design
winglets that gradually became accepted as benefiting
overall sailplane performance. To further these gains,
an improved methodology for winglet design has been
developed. This methodology incorporates a detailed
component drag buildup that includes the ability to
interpolate input airfoil drag and moment data across
operational lift coefficient, Reynolds number, and flapsetting ranges. Induced drag is initially predicted using
a relatively fast multi- lifting line method. In the final
stages of the design process, a full panel method,
including relaxed-wake modeling, is employed. The
drag predictions are used to compute speed polars for
both level and turning flight.
The predicted
performance is in good agreement with flight-test
results. The straight and turning flight speed polars are
then used to obtain cross-country performance over a
range of thermal strengths, sizes, and shapes. Example
design cases presented here demonstrate that winglets
can provide a small, but important, performance
advantage over much of the operating range for both
span limited and span unlimited high-performance
sailplanes.

Subscripts
W
wing
WL
winglet
WT
wing tip
Introduction
From initially being able to do little to improve
overall sailplane performance, winglets have developed
to such an extent over the past ten years that few gliders
now leave the manufacturers without them. This
change was brought about by the efforts of a number of
people to better understand how winglets work, to
develop theoretical methods to analyze performance,
and to develop design methods that allow the benefits
to be tailored such that gains in cross-country
performance are achieved over a wide range of soaring
conditions. The story of this development is an
interesting case study in engineering design, in which
trial and error, theoretical analysis, and flight testing all
contributed to the successful solution of a difficult
problem.
Although compared to other modern flight
vehicles, the high-performance sailplane appears to be
relatively simple, the design of such aircraft to
maximize average cross-country speeds in any given
weather situation is quite challenging.1 This is largely
due to the fact that in flying cross country, the sailplane
must be able to climb effectively in thermals at low
speeds, as well as being able to glide efficiently
between thermals at high speeds. Thus, a successful
design must balance the conflicting requirements of
climbing and cruising over a broad range of possible
soaring conditions.
For efficient climbing, a sailplane must be able to
circle and maneuver with a low sink rate in thermals
that can change dramatically in strength, size, and shape
from day to day, and possibly even over the duration of
a single flight. As this requires turning flight at low
speeds and high lift coefficients, the reduction of
induced drag is a major consideration in the design
process. Clearly, although it can penalize the efficiency
in cruising flight, the most straightforward method to
reduce induced drag is through the use of large spans.

Nomenclature
b
span
c
wing chord
cl
section lift coefficient
h
winglet height
CDp profile drag coefficient averaged over span
K
induced drag factor
S
planform area
V
airspeed
VCC average cross-country speed
VCR crossover velocity
VS
sink rate
W
weight

air density
______________
*

Associate Professor, Department of Aerospace Engineering,

Senior Member, AIAA.

mission profile, but not without overly penalizing the

cruise performance. Thus, it was with some skepticism
that efforts were taken to try to improve this situation.
The first task undertaken was a result of the
observation that a winglet does not operate exactly as
does a wing. For that reason, a winglet specific airfoil,
the PSU 90-125, was designed. As not a great deal was
known at this time about exactly how a sailplane
winglet should operate, this was a conservative design
that was intended to perform reasonably well over a
wide range of operating conditions.
From this point, a trial and error process began that
used flight-testing as the primary method of
determining the important design parameters. Although
vortex-lattice and panel methods were of some value
for gaining insight, they were of limited value in the
actual design process. Likewise, as the beneficial
influence of a winglet is due to it favorably altering the
flow field over the entire wing, meaningful wind-tunnel
experiments require the use of a full-span model.
Unless the wind tunnel has a very large test section,
however, at the large aspect ratios typical of sailplanes,
this would result in model chords that would likely
operate at subcritical Reynolds numbers. To address
these problems, methods of simulating full-scale flow
fields with truncated spans have been explored, but in
every case, the necessary compromises would produce
questionable results.6 For these reasons, the parameters
that were deemed the least important were set to what
seemed to be reasonable values, while the values of the
more critical ones were determined from flight test.
Using some of the results from earlier work on winglets
for transport and general aviation type aircraft,7-9 along
with simple calculations, the winglet height, planform
shape, and cant (dihedral) angle, as defined in Fig. 1,
were fixed. From this point, the goal was to establish
the spanwise load distribution on the winglet that would
interact in a reasonable way with the main wing and
thereby produce an overall drag reduction. As the basic
shape of this loading could be adjusted with twist or
sweep, the twist was set, again being guided by the
earlier work on winglets. In so doing, the geometry of
the winglet blade was fixed. For minimum induced
drag, if the planform is close to elliptical, then an
elliptical load distribution requires that the spanwise lift
coefficients be roughly constant. Thus, with the
planform set, flight tests were undertaken in which
woolen tufts were attached to the winglet and the load
distribution adjusted using sweep until the stall pattern
over the height of the winglet was uniform.
The last design parameter to be determined was the
winglet toe angle. To establish this quantity, the idea
employed was simply that there seemed to be little
benefit in having the winglet being able to carry a load
beyond that of the wing. This point was found by
adding tufts to the main wing and adjusting the winglet

Among the various classes of racing sailplanes,

however, only the Open Class allows unlimited span,
while all of the other FAI (Fdration Aronautique
Internationale) classes, Club, Standard, 15m Racing,
and 18m Classes, have a maximum allowable value.
The inter-thermal cruise, on the other hand, corresponds
to flight at high speeds and low lift coefficients such
that the reduction of profile drag dominates the design
process. This trade-off between climbing and cruising
is complicated further in that the optimum cruising
speeds vary with the soaring conditions and depend on
the achieved climb rate in thermals. Typically, the
optimum cruising speed, called the MacCready speedto-fly, is determined for a given sailplane and weather
conditions using an idealized climb/glide cycle.2 It is
found that in weak weather, in which it is more difficult
to regain altitude lost during cruise, the optimum
cruising speed is not too much faster than that
corresponding to the minimum glide angle (maximum
lift-to-drag ratio) of the glider. In strong weather, the
ability to achieve high climb rates dictates much faster
optimum cruising speeds.
Because of the requirement to cruise at speeds much
greater than that of the maximum lift-to-drag ratio, it is
just as important that a modern sailplane have a
reasonably flat speed polar, in which the sink rate does
not increase too fast with speed, as it is that the glider
have a high lift-to-drag ratio. In order to gain
somewhat greater flexibility in the matching of the
sailplane to soaring conditions on a particular day, most
of the competition classes allow the use of disposable
water ballast. In strong weather, the ballast is carried to
increase the wing loading such that the speed polar
shifts to higher airspeeds. The penalty in climb due to
carrying additional weight is more than offset by having
a higher lift-to-drag ratio at a given cruising speed. In
weak weather, the ballast is not carried or can be
dumped to regain the now needed better climbing
ability. Similar gains are achieved with flaps, which
are permitted in all of the FAI classes except Standard.
In climbing flight, the flaps are lowered to achieve
higher lift coefficients, while in cruise, they are
deflected upward to shift the airfoil low-drag region to
a lower lift-coefficient range. This upward deflection
also reduces the nose-down pitching moment of the
airfoil and thereby reduces trim drag.
The efforts at Penn State to develop winglets for
high-performance sailplanes began in the early 1980s
with a collaborative effort to design winglets for the
15m Class competition sailplanes of that era. Although
work had already been done in this area, in practice it
was found winglets provided little or no benefit to
overall sailplane performance.3-5 The widely held
belief at that time, essentially the same as that held
regarding winglets for transport-type aircraft, was that it
was thought possible to help the climb portion of the
2

Considered another way, the effect of the winglet is

to produce a vertical diffusion of the vorticity in tip
region. This diffusion process is also realized as an
expansion of the wake in the far field due to induced
velocities from the nonplanar components of the
winglet. The out of plane bound vortex on an upward
winglet induces horizontal velocities on the free wake
that cause a spanwise spreading of the wake field.
When referenced to the actual span, the resulting
efficiency can be greater than that of an elliptical
loading, emulating the effect of a span increase.16 It
should be noted that, while still beneficial, a winglet
downward oriented would produce a contraction of the
wake and is not as effective in reducing the induced
drag as is a winglet oriented upward.
The profile drag contribution of the winglet is more
straightforward than that of the induced drag. Any
addition of wetted area will carry with it an increment
in profile drag. Thus, adding winglets to aircraft causes
an increase in wetted area and a corresponding increase
in profile drag. The effect of the increased area is felt
primarily at higher speeds, as the profile drag
coefficient remains relatively constant while the drag
increases with the square of the velocity.
The
additional wetted surface area penalty due to a winglet
may be offset somewhat by removing some portion of
the original wing tip when mounting the winglet. The
large chords of the wing tip relative to the much smaller
chords of the winglet provide a substantial
compensation in wetted area, although the lower
Reynolds number due to the smaller winglet chords will
typically have somewhat larger profile drag
coefficients. This cutting back of the tips is particularly
effective in fixed-span classes. The total span is
maintained at the maximum allowable by using a
winglet dihedral angle of less than ninety degrees. In
this way, a winglet may be added with less increase in
wetted surface area than would occur if the tip of the
existing planform were simply extended and turned
upward.
It is also found that the amount of laminar flow over
the outer portions of the main wing is increased by the
presence of winglets.17, 18 Without winglets, the
transition line on the upper surface moves forward as
the tip is approached, whereas the imposed favorable
pressure gradient with winglets causes it to move aft.
Consequently, to some extent, winglets offset their own
increased wetted area penalty by decreasing profile
drag over a part of the main wing.
Simply stated, the design goal of a winglet is to
produce the most reduction in induced drag for the
smallest increase in profile drag. The induced drag
benefit of winglets is greatest at higher lift coefficients
and lower flight velocities, while the profile drag
penalty grows in magnitude as the lift coefficient
decreases and the velocity increases. With the benefit

toe angle until both wing and winglet stalled at roughly

the same time.
Although it took a period of time and some contest
successes before they began to be accepted, the result of
the process was the first generally successful winglets
that benefited overall cross-country performance over a
wide range of thermal sizes and strengths.10 In
retrospect, with the understanding that has come since,
it seems that this process, while systematic and logical,
was accompanied with a certain amount of luck.
Even though the strategy employed resulted in a
successful design, it was somewhat frustrating that the
then available tools were not of much use in the
development of these winglets. For this reason, a
research program was undertaken in order to develop
tools and a methodology for winglet design.6, 11-14
These efforts began with the design of a new airfoil.
Using the lessons learned and the improved
understanding of how winglets operate, the PSU 94-097
airfoil was designed with much less conservatism than
did its predecessor. In addition, theoretical tools were
developed and validated with flight-test measurements.
These methods are now quite reliable and the winglets
designed using them generally meet their design goals
without modification. Designs have been developed for
a number of sailplanes. Those for the Schempp-Hirth
Ventus 2, shown in Fig. 2, and the Schleicher ASW-27,
detailed in Fig. 3, are typical of these designs and
installations.
Technical Discussion
One of the consequences of producing lift on a
finite wing is the generation of spanwise flow. In
particular, the pressure gradients caused by the lower
pressures on the upper surface relative to the higher
pressures on the lower surface lead to inward spanwise
flow on the upper surface and outward spanwise flow
on the lower. At the trailing edge, the merging of these
two flows having different directions generates the
vorticity that is shed from a finite wing and is the origin
of induced drag. It has been known for over a century
that an endplate at the tip of a finite wing can reduce the
spanwise flow and thereby reduce the induced drag.
Unfortunately, to be effective, the endplate must be so
large that the increase in wetted area drag far outweighs
any drag reduction. A winglet, rather than being a
simple fence which limits the spanwise flow, carries an
aerodynamic load that produces a flow field that
interacts with that of the main wing thereby reduces the
amount of spanwise flow.15 In essence, the winglet
diffuses or spreads out the influence of the tip vortex
such that the downwash and, in turn the induced drag,
are 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.
3

winglet must operate is not as wide as that required for

a typical wing. Thus, an airfoil designed specifically
for a winglet can have somewhat lower drag than an
airfoil designed for, say, a small unmanned air vehicle.
One important goal for the winglet airfoil design is
to avoid poor section performance at low flight
velocities. As the principle benefit of a winglet is in
climb, stalling of the winglet in these conditions would
certainly result in an overall loss in performance. Thus,
the section must generate the maximum lift coefficients
required by the winglet as the aircraft approaches stall.
Likewise, low-drag performance over the entire
operating range is important, but it must be considered
in conjunction with the other constraints. As the profile
drag increases with velocity squared, excessive section
drag coefficients at low lift coefficients would severely
impact aircraft performance at higher flight speeds.
This consideration drives the lower lift coefficient
portion of the airfoil drag polar. The degree to which
these considerations influence the overall performance
is again difficult to ascertain without considering the
entire flight profile of the sailplane. The determination
of the amount of gain needed at low-speed to offset a
loss at high speed requires a relatively accurate method
of performance evaluation.

and penalty being at different points in the flight

regime, the optimization of the winglet geometry
becomes complicated and ultimately requires an
effective means of evaluating the changes in
performance due to winglets over the entire flight
envelope of the glider.
Winglet Geometry Issues
In the course of designing a winglet, a number of
design variables must be considered. For fixing the
geometry, the most important include determination of
the airfoil section, chord distribution, height, twist,
sweep, and toe angle. Because so many variables are
involved, the design problem is difficult. It is further
complicated by the operational profile of a sailplane,
which combines a low-speed, high-lift coefficient climb
phase with a high-speed, low-lift coefficient cruise
phase, both of relatively equal importance.
Airfoil Considerations
As in most airfoil design efforts, the goal of the
winglet airfoil design is to generate the lift required
with the lowest possible drag. In the case of the winglet
airfoil, the operational low-drag region for the winglet
should correspond to that of the wing. Likewise, in
low-speed flight the winglet should not stall before the
main wing.
The relationship between the winglet lift coefficient
and that of the main wing is unique for every
wing/winglet combination, and ideally, every
combination would have a specifically designed winglet
airfoil. In most cases, however, such an effort is not
warranted by the small gain in performance that would
result. It should also be noted that the information
needed to guide the airfoil design depends on the details
of the winglet geometry, which in turn, is driven by the
aerodynamic characteristics of the airfoil. Thus, the
winglet/airfoil design process is iterative, and the result
is the product of a number of design iterations.
Consequently, in addition to the need for an accurate
airfoil design method, the need is also clear for an
accurate method of assessing the impact of winglet
design details on the overall sailplane performance.
The attainment of the desired design goals for the
winglet is made more difficult by the narrow chords
and resulting low Reynolds numbers. This situation
establishes a trade-off between trying to reduce the
wetted area increase by using small chords against that
of high profile drag coefficients due to the low
Reynolds numbers. The small chords of the winglet
dictate an airfoil that operates efficiently at Reynolds
numbers in the range of 7.0x104 to 1.0x106. At such
low values, laminar separation bubbles and the associated increases in profile drag are an important
concern. Fortunately, this problem is helped somewhat
in that the range of lift coefficients over which the

Chord Distribution and Height

The most suitable winglet chord distribution is
determined by a number of conflicting factors. Most
important, the winglet must be able to generate the
spanwise loading needed to produce the favorable interaction with the induced velocity field of the main wing.
At low flight velocities, very small winglet chords can
require lift coefficients greater than the airfoil can
produce. This, of course, causes the winglet to be
ineffective and can result in excessive drag due to the
winglet stalling. Winglet chords that are too large, on
the other hand, can also lead to poor performance in
that high loading on the winglet can excessively load
the tip region of the main wing and lower its planform
efficiency. In extreme cases, this can cause the
outboard sections of the main wing to stall prematurely.
To avoid this situation, the winglet would have to be
inefficiently under-loaded with the larger chords doing
little but increasing the wetted area and the profile drag.
This trade-off is further complicated by wanting small
chords to minimize the added wetted area against not
having chords so small as to result in high drag due to
low Reynolds numbers. Although this depends on the
airfoil under consideration, an appropriate airfoil can
operate at quite low Reynolds numbers before the
penalty due to an increased profile drag coefficient
offset the drag reduction due to less area. This breakeven point is, in fact, that at which halving the
Reynolds number causes the profile drag coefficient to
double. For most cases, the planform shape can be set
4

due to the winglets is zero. Below this speed, winglets

are beneficial, while above this speed they are detrimental. Thus, the crossover point is the flight speed at
which the benefit in induced drag due to winglets is
equal to the profile drag penalty and occurs when

without concern for the increased profile drag

coefficient due to Reynolds number effects.
Although not so critical, once the basic chord
dimension has been determined, the spanwise chord
distribution should be such that the loading on the
winglet is near elliptical and the induced drag on the
winglet itself will be minimized. Once the chord
distribution has been established, the winglet height is
determined by the trade-off between the induced drag
benefit and the wetted area penalty.

DPROFILE + DINDUCED = 0
This simple expression indicates that the more the
induced drag can be reduced for a given increase in
profile drag, the higher will be the crossover point and
the more effective the winglet.
After making a number of simplifying assumptions
and expressing the total drag change due to the addition
of a winglet in terms of the appropriate quantities, the
derivative with respect to the winglet height can be set
to zero and solved for the crossover velocity, VCR, to
obtain

Twist, Sweep, and Toe Angle

After sizing the chord distribution and height by
considering the required loading, profile drag and
Reynolds number constraints, the winglet load
distribution can be tailored further by spanwise twist
and planform sweep. Increasing the sweep has the
same effect on the load distribution as does adding
wash-in along the winglet. Thus, the problem is
simplified if one, say twist, is fixed and the sweep is
tailored to achieve the best overall performance. The
only concern to the designer is that too much sweep can
introduce cross-flow instabilities that will cause the
boundary layer to transition earlier than would
otherwise be the case. Although there is not much
information on this subject at the Reynolds numbers of
interest here, it is known that the instability is reduced
as the Reynolds number decreases. Consequently, as
has been verified in wind-tunnel tests on winglet
geometries, this should not be a problem provided that
sweep angles are not in excess of, say, thirty-five or
forty degrees.
After the planform has been determined, the toe
angle at which the winglet should be mounted must be
determined. This angle controls the overall loading on
the winglet, as well as its overall effect on the load
distribution of the main wing. Since the angle of attack
of the winglet is a function of the lift coefficient of the
wing, most likely the toe angle setting will only be truly
optimal for one flight condition. Nevertheless, the
determination of this angle to yield the best possible
performance over the entire flight envelope is perhaps
the most critical element of the design process.

VCR =

const

h C Dp,WL

W
b

where the constant depends on how the winglet area

increases and the overall induced drag decreases with
winglet height, h.
The lower the profile drag
coefficient of the added winglet area, CDp,WL,, and the
greater the span loading, the higher will be the velocity
of the crossover point.
The understanding of the crossover point allows it
to be controlled through the geometry of the winglet.
In the early stages of using this simple idea, the
crossover point was set to be higher than any
anticipated cruising speeds for both the unballasted and
ballasted cases. The use of this formula resulted in
winglets that generally improved performance and,
although based on a simple theory, was as good as the
not so good ability to predict the changes in induced
drag due to changes in winglet geometry.
As the ability to predict the induced drag for a given
wing geometry improved,11 the crossover point method
was modified to take advantage of this. The expression
equating the change in profile drag with the change in
induced drag can be written more explicitly in terms of
parameters describing the winglet geometry and the
resulting aerodynamic influences as

The Winglet Design Process

Past Methodologies
Several approaches to winglet design have been
utilized at Penn State.6, 11-14 All of these methodologies
have attempted to quantify, in one way or another, the
tradeoff between the profile drag penalty and the
induced drag benefit. Prior to the current approach, all
other efforts made use of what has been termed the
crossover point on the sailplane speed polar. This point
corresponds to the airspeed at which the flight polar of
the base aircraft and the aircraft with winglets intersect,
or equivalently where the percent change in sink rate

(S C Dp )WL - (S C Dp )WT +

4W 2

2 VCR4

K2
K

- 21
b2
b1
2

= 0

where the terms having the WT subscript correspond

to those for any area near the wingtip that is removed to
mount the winglet, the subscript 1 to the original
wing, and 2 to the one modified with the winglet.
5

The weight of the sailplane, W, in this simple

expression is considered to be unchanged by the
wingtip modification. For fixed span classes, of course,
b1 = b2. If allowed, however, it is desirable to increase
the span up to the point at which it hurts the profile drag
more than it helps the induced drag. Thus, the problem
for the winglet designer is to minimize the first term of
this equation, the profile drag increase, while
maximizing the other two, the drag area removed and
the induced drag reduction. The induced drag factor,
K2, should be made as small as possible. Likewise, the
net area increase should be minimized, as should the
profile drag coefficient of any added area. While this
expression does not capture the details of winglet
design, it does capture the essence of the task.
Using either the closed form relation presented
earlier, or some computational method of predicting the
aircraft speed polars, the crossover velocity is adjusted,
primarily using the toe angle and twist distribution, to
allow the winglet to benefit performance over some part
of the operational flight speed range. Shifting the
crossover speed not only affects the speed range over
which a benefit is achieved, but it also effects the
magnitude of that benefit across the chosen range.
Shifting the crossover to higher velocities reduces the
performance gains due to the winglet at lower speeds,
while shifting the crossover to lower velocities achieves
a much larger drag reduction, but only over a small
portion of the flight polar.
A number of winglets were designed, fabricated,
and flight tested using these methods, and while based
on simple ideas, they helped to establish the following
rules of thumb. First, whether it be with up-turned tips
or winglets, it is beneficial for the design to be out-ofplane. Second, while a great deal of work has been
directed toward determining the optimum geometries
for minimum induced drag,19, 20 experience has shown
that pushing too far toward this optimum penalizes the
profile drag far more than can be offset by the resulting
induced drag reduction.14 The design goal is clearly to
minimize the overall drag, not just one component of it.
For example, the optimum loading for minimum
induced drag must be continuous across the juncture
between the main wing and the winglet. This would
require that the chords at this point be the same, or that
the lift coefficient at the root of the winglet be
proportionally greater than that of the wingtip. Either
way, the profile drag is considerably greater than that
achieved with current designs. Up to now, it has been
found that most of the potential induced drag benefit is
achieved by designing or modifying the wing planform
to be non-planar. Once this is done, the efforts of the
designer are most rewarded by working toward
minimizing the profile drag.

The broad nature of the sailplane mission profile

greatly complicates the choice of an optimum crossover
speed. In weak conditions, gains at low velocities in
climb will offset a loss in cruise performance.
Conversely, in strong conditions, not penalizing the
high-speed cruise will be the most important to overall
cross-country performance. While it is an effective
method of predicting the change in aircraft performance
due to the addition of winglets, and it does ensure some
benefit, the use of the crossover point idea generally
will not produce the best design.
An optimal
configuration cannot be determined without specifically
taking into account the impact of the winglets on the
average cross-country speed. To do this, a fast,
accurate prediction of the sailplane performance has
been developed and combined with a thermal model,
allowing the calculation of MacCready average crosscountry speeds for specific weather conditions and
aircraft configurations.13, 14 The average speed is then
used to determine the suitability of a design. This
approach allows the entire flight profile to be taken into
account in the design and yields a simple result
encompassing the broad range of contributing factors.
While MacCready theory is often used to assess
overall performance, these efforts generally lack the
ability to accurately and rapidly account for small
changes in an aircraft configuration. The simplifications typically used, such as parabolic flight polars
and approximated airfoil characteristics, introduce
errors that are on the same order as the changes brought
on by winglets. While useful for exploring trends and
the basic characteristics of winglets, these methods are
generally not accurate enough for design.
Prediction of Sailplane Performance
The calculation of sailplane performance forms the
major component of the winglet design problem. As
already stated, the performance evaluation must have
sufficient resolution to discern the effect of winglets.
As these effects are relatively small, errors or inconsistencies in other portions of the calculation may
overshadow them.
The accuracy necessary for
successfully undertaking design activities such as
winglets is obtained through the use of a performance
program that has been developed to predict the straight
and turning flight polars of sailplanes.13, 14 To achieve
the accuracy required, this program accounts for the
effects of airfoil selection, trim drag, static margin,
fuselage drag, flap geometry, and flap deflection
scheduling. The most important element of the method
is the analysis of the wing planform aerodynamics.
Essential to the accuracy of the analysis method is
the interpolation of two-dimensional airfoil data. Wing
profile drag is such a large portion of the overall drag
that small errors in its determination can eclipse the
effects of winglets. In addition to the requirement of

Present Design Approach

having accurate profile drag data, this necessitates

interpolation of airfoil drag and moment data over the
operational ranges of lift coefficient, Reynolds number,
and flap deflection.
The other essential component for predicting the
planform aerodynamics is the determination of the span
efficiency and lift distribution. The lift distribution
directly affects the wing profile drag, and the planform
efficiency dictates the induced drag of the wing. As
this is where the benefit of the winglet is quantified, an
accurate method of determining these two items is of
critical importance.
In the present approach, use is made of both a multilifting line method and a three-dimensional liftingsurface panel code. The multi-lifting line method used
here, which has been integrated directly into the
performance program, divides the several chordwise
lifting lines into segments, each segment having a
parabolic distribution of vorticity.3 This produces a
continuous sheet of vorticity that is shed into the wake.
The method allows the spanwise lift distribution and
induced drag of non-planar wing geometries to be
predicted with reasonable accuracy and much less
computational effort than required by a threedimensional panel method. Although not accounting
for the consequences of thickness and a free wake, as
the panel method used here is able to do, the multilifting line procedure is able to quantify the effects of
winglets. For initial design iterations, the increased
speed of the multi-lifting line method more than offsets
the small loss in accuracy.
The use of the multi-lifting line program and the
interpolation of airfoil characteristics allow the
performance program to produce accurate straight and
turning flight polars for any aircraft configuration. The
predicted performance for an unflapped sailplane, the
Discus, is presented along with flight-test data21 in Fig.
4. The predicted performance compares very well with
the measured results. A similar comparison for a
flapped sailplane, the ASW-22B, is presented in Fig. 5.
The agreement for the individual flap settings is
generally good, although there is some disagreement for
the high-speed, negative flap deflections. At high
speeds, however, small measurement errors have a
large effect, and the differences between the predicted
and measured points are not as large as the scatter
between some of the measured points.
Similar
comparisons over a wide range of sailplane types have
demonstrated that the method is able to resolve small
enough differences between configurations to be of
value in the winglet design effort.
For the final detailed design of the winglet, use is
made of a panel method program that takes free-wake
effects into account.11 For the calculation of induced
drag, this program makes use of the Kutta-Joukowsky
theorem in the near field.22 This eliminates some of the

problems associated with attempting to account for

wake relaxation in the far field using a Trefftz plane
analysis. While the differences in results between a
relaxed wake and a fixed wake analysis are generally
small, these differences can be significant in
determining the final winglet toe and twist angles.6
Analysis of Cross-Country Performance
With straight and turning flight polars available,
analysis of crossover speeds is possible, but as mentioned previously, a more rigorous means of evaluating
designs is desirable. This task is accomplished by a
program that calculates the MacCready average crosscountry speeds for a given configuration using the
straight and turning flight polars generated by the
performance program.13, 14
The thermal model used in this analysis has a
distribution of lift that varies parabolically with thermal
radius. Thus, the thermal profile is defined by the
strength of the lift at the core and the radius. Clearly,
the thermal profile has a significant impact on the
cross-country performance of a sailplane, and the most
useful performance index would be the result of some
particular mix of thermal strengths and profiles.1
Nevertheless, the use of a single, representative thermal
profile, as is done here, greatly simplifies the
interpretation of the results while still yielding a
meaningful comparison between different sailplane
configurations.
To obtain the optimal climb rate for a particular
configuration, the thermal profile is superimposed over
the predicted turning polars. The straight flight polar is
then searched for the inter-thermal cruise speed to
optimize the MacCready cross-country speed. The
result is a trade-off of climbing and cruise performance,
properly weighted to account for the variations in
soaring conditions over which the sailplane might be
operated.
Cross-Country Performance Gains
A Restricted Span Design Example
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. 6.
Although gains are demonstrated, they are difficult to
assess in the figures shown. Thus, the data are replotted
in terms of lift-to-drag ratio verses velocity in Fig. 7. In
addition to demonstrating the gains in carrying water
ballast at higher cruising speeds, the winglets are seen
to increase the lift-to-drag ratio over a significant
portion of the operating range. To get an even better
idea of the gains in lift-to-drag ratio, these data are
again replotted in terms of the percentage increase in
lift-to-drag ratio relative the same sailplane without
winglets. These results are presented in Fig. 8. It

aspect ratio wing can be so far from elliptical, the

increase in span efficiency due to properly designed
winglet can yield a greater reduction in induced drag
than does a comparable span increase. In addition, by
reducing the spanwise flow at the wingtip, the winglet
allows the tip region to operate better at high lift
coefficients. This can result in improved turning
performance and handling qualities.
For a span extension to achieve the best result, it is
important that the chord distribution be continuous at
the junction between the wing and the extension.
Otherwise, the abrupt change in the loading that would
result causes a significant penalty in induced drag.
Such a discontinuity between the wing and a winglet,
on the other hand, is not a problem. For the same
increase in load perimeter (spar length), the winglet can
have significantly less area and thereby a lower increase
in profile drag than does a span extension.
To demonstrate the benefit of winglets on an
unrestricted-span sailplane, the percentage increase in
average cross-country speed for an ASW-22B due to
pure span extensions compared to that due to a partial
span extension plus a winglet is presented in Fig. 10. In
this case, the area increases and the load perimeters for
both are comparable. In fact, in spite of having less
span, the extensions with winglets using less area but a
slightly longer load perimeter, demonstrate a small but
definite performance advantage over the glider with
pure span extensions. This example also indicates that
work remains to be done in finding the best tip
treatment for unlimited span gliders, and that the
potential exists for additional improvement.

should be noted that this winglet is such that the

crossover points occur at airspeeds that are above the
maximum allowable. While not true for all gliders, in
this case there are no flight conditions for which the
winglets penalize performance. While the percentage
gain in lift-to-drag ratio does not appear to be very
great, it is important that it is achieved without any
penalty at higher speeds.
As has been noted, although the gain in lift-to-drag
ratio is of interest, the true measure of the benefit of
winglets is reflected in how they influence the overall
cross-country performance. To demonstrate this, the
percentage change in average cross-country speed
relative to that of the baseline aircraft, without ballast
and without winglets, is presented in Fig. 9. The
winglets improve the cross-country performance for all
the thermals considered, that is, for thermals having a
500 ft. radius and strengths, averaged across the
diameter, of up to 10 kts. As expected, the performance
gains are significant for weak thermals, as the winglets
allow for some climb rate, whereas without winglets, it
is minimal or zero. As the thermal strengths increase,
the benefit due to winglets decrease; however, for this
glider, winglets are always a benefit to cross-country
speed, even for average thermal strengths of 10.0 kts.
and above. 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, which corresponds to a
predicted climb rate of about 5.2 kts. with full ballast.
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. In addition, the glider with winglets
can carry ballast to slightly weaker conditions without
penalty than can the glider without winglets.

Other Considerations
In designing winglets for a variety of sailplanes, as
well as for a few non-sailplane applications, it seems to
be true 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).
The case presented here is, in fact, one of the most
difficult designs undertaken thus far. As an example of
how critical these designs can be, the effect of the
winglet toe angles on the Discus 2 winglet is presented
in Fig. 11. As shown, even a small deviation from the
optimum can cause the winglet to hurt performance.
Furthermore, as many of these parameters are unique to
each type of glider, each must have winglets tailored
specifically for it. Generalities regarding winglet
geometries can be disastrous. In the course of this work
on winglets, one thing has become clear; it is much
easier to make a glider worse with winglets than it is to
make it better!
In some winglet design cases, it has been found that
the winglets fix some problem of the original wing. For
example, in the case of a flapped glider, it is important

An Unrestricted Span Design Example

Based on results from some of the early work on
minimizing induced drag, it has long been accepted that
when wingspan is unrestricted, a pure span extension
will generally result in a greater performance gain than
can be achieved with winglets. When profile drag
considerations are included in the analysis, however,
this conclusion is not so clear. For example, the
minimum induced drag depends on maximizing both
the span and the span efficiency. Unfortunately,
because it is harder to maintain an elliptical spanwise
load distribution as the span increases, the span
efficiency usually decreases with increasing span. For
wings of lower aspect ratios, the benefit of increasing
the span generally far outweighs the penalty due to
decreased span efficiency; however, for wings having
very high aspect ratios, the benefit of increasing span is
less assured. Because the lift distribution of a very high

Reichmann, H., Cross-Country Soaring, Seventh

Edition, Soaring Society of America, Inc., 1993.
3
Horstmann, K.H., Ein Mehrfach-Traglinienverfahren und seine Verwendung fr Entwurf und
Nachrechnung
nichtplanarer
Flgelanordnungen,
DFVLR,
Institut
fr
Entwurfsaerodynamik,
Braunschweig, DFVLR-FB 87- 51, 1987.
4
Drebler, U., !Aerodynamic Design of Winglets
for a Standard-Class Glider," Technical Soaring, Vol.
Vlll, No. 4, July 1984, pp. 118-126.
5
Marsden, D.J., !Winglets for Sailplanes,"
Technical Soaring, Vol. XV, No. 4, Oct. 1991, pp. 119124.
6
Hoffstadt, B.A., Analysis and Design of Winglets
for Standard-Class Sailplanes, M.S. Thesis, Dept. of
Aerospace Engineering, Penn State Univ., University
Park, PA, 1997.
7
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.
8
Heyson, H.H., Riebe, G.D., and Fulton, C.L.,
Theoretical Parametric Study of the Relative
Advantages of Winglets and Wing-Tip Extensions,
NASA TP-1020, 1977.
9
van Dam, C.P.,!Analysis of Non-Planar WingTip-Mounted Lifting Surfaces on Low-Speed
Airplanes,"NASA Contractor Report 3684, 1983
10
Masak, P.C., !Design of Winglets for
Sailplanes," Soaring, June, 1993, pp. 21-27.
11
Mortara, K.W. and Maughmer, M.D., A Method
for the Prediction of Induced Drag for Planar and NonPlanar Wings, AIAA Paper 93-3420, Aug. 1993.
12
Anderson, R.P., Sailplane Winglet Design Using
Non-Linear, Non-Planar Techniques, M.S. Thesis,
Dept. of Aerospace Engineering, Penn State Univ.,
University Park, PA, 1993.
13
Kunz, P.J., Development of a Software Package
for the Assessment of High-Performance Sailplanes,
M.S. Thesis, Dept. of Aerospace Engineering, Penn
State Univ., University Park, PA, 1997.
14
Maughmer, M.D. and Kunz, P.J., !Sailplane
Winglet Design," Technical Soaring, Vol. XXll, No.
4, Oct. 1998, pp. 116-123.
15
Blackwell, J.A. Jr., Numerical Method to
Calculate the Induced Drag or Optimum Loading for
Arbitrary Non-Planar Aircraft, Vortex-Lattice Utilization, NASA SP-405, May 1976.
16
Yates, J.E. and Donaldson, C.du P., A
Fundamental Study of Drag and Assessment of
Conventional Drag-Due-To-Lift Reduction Devices,
NASA CR 4004, Sept. 1986.

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
for the optimum spanwise loading. Although only a
small portion of the wing is seemingly influenced, it
results in very significant induced drag increase. In
these cases, cutting the tip back to the aileron in order
to mount the winglet can result in gains, especially at
high speeds, that would not be expected by the addition
of winglets.
It has been found from experience and flight test
that winglets often result in unanticipated performance
gains and improved handling qualities. In particular, it
has been found that winglets improve the flow in the tip
region and thereby improve the effectiveness of the
ailerons. One of the benefits of greater control
effectiveness is that smaller aileron deflections are
required for a given rolling moment. This not only
results in less drag for a given roll rate, but it also
allows for higher roll rates. Woolen tufts attached to
glider wings have shown that much of the flow over the
inboard tip during turning flight is separated, and this is
nearly eliminated by the presence of a properly
designed winglet. In addition to the resulting reduction
in drag, this benefits safety in that the aileron
effectiveness is retained much deeper into the stalled
region than before.
Closing Comments
Although the performance gains achieved with
winglets are only a few percent at moderate thermal
strengths, such small differences can be an important
factor in determining the outcome of many crosscountry flights or contests. For example, for the U.S.
Open Class Nationals in 1999, just 68 points separated
the first six places. As the winner had 4882 points, the
difference between first and sixth places amounted to
less than 1.5%. This is far less than the performance
gains that have been achieved using winglets.
Since their shaky introduction a number of years
ago, the notion that winglets can produce performance
advantages is now widely accepted. At the World
Championships in 1991, out of 105 competing gliders,
only 19 used winglets. In the Championships of 1998,
essentially every glider had winglets or some type of tip
treatment. It is clear that the benefits are far reaching.
If properly designed such that the profile drag penalty is
of no consequence over the range of airspeeds at which
the glider operates, then there is no reason whatsoever
not to take advantage of the benefits that winglets offer
to both performance and handling qualities.
References
Thomas F., Fundamentals of Sailplane Design,
Judah Milgram, translator and contributor, College Park
Press, College Park, MD, 1999.
1

17

Bose, S., !Investigation on a Winglet of the

Glider ASW-24," Studienarbeit, Technical University
of Berlin, 1993.
18
Thorsen, O.R., !Theoretical and Experimental
Analysis of the Winglet Designed for the High
Performance Sailplane ASW-27," Low-Speed
Aerodynamics Laboratory, Delft University of
Technology, March 1999.
19
Munk, M.M., Minimum Induced Drag of
Aerofoils, NACA Technical Report No. 121, 1921.
20
Cone, C.C. Jr., !Theory of Induced Lift and
Minimum Induced Drag of Nonplanar Lifting
Systems," NASA Technical Report R-139, 1962.
21
Selinger, P.F., Segelflugzeuge Vom Wolf zum
Discus, Motorbuch Verlag, Stuttgart, 1989.
22
Eppler, R., Die Entwicklung der Tragflgetheorie, Z. Flugwiss, Nov. 1997, pp. 133-144.

Fig. 2 Schempp-Hirth Ventus 2 sailplane with

winglets.

Fig. 3 Detail of winglet on a Schleicher AS-W 27

sailplane.
V (kts)

VS (kts)

40

60

80

100

Discus, 705 lbs

Predicted
Flight Test
6

Fig. 4 Comparison of predicted and flight-test

results for the straight-flight speed polar of the
Schempp-Hirth Discus.

Fig. 1 Geometric quantities used to define winglet.

V (kts)
0

40

60

80

100

ASW-22B, 1192 lbs

VS (kts)

10

Theory: Flap pos. 4

Flap pos. 3
Flap pos. 2
Flap pos. 1
Flight Test: Flap pos. 4
Flap pos. 3
Flap pos. 2
Flap pos. 1

120

Fig. 5 Comparison of the predicted and flight-test

results for the straight-flight speed polar of the
Schleicher ASW-22B.

Fig. 8 Percentage gain in predicted lift-to-drag

ratios due to winglets for unballasted and ballasted
Discus 2.

V (kts)
0

40

60

80

100

120

VCC %

VS (kts)

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

Baseline: Discus 2, 685 lbs

Discus 2 WL, 685 lbs

Discus 2, 1160 lbs
Discus 2 WL, 1160 lbs

8
0
2

Fig. 6 Predicted straight flight polars of unballasted

and ballasted Discus 2, with and without winglets.

-1

50

10

Thermal Strength (kts)

Fig. 9 Percentage gain in predicted average crosscountry speed due to winglets and ballast relative to
unballasted Discus 2 without winglets.

40

L/D
15

BASELINE:
ASW-22B- 1150
1150 lbs
LBS.
Baseline: ASW-22B

30

20

10

VCC%

Discus 2, 685 lbs

Discus 2 WL, 685 lbs
Discus 2, 1060 lbs
Discus 2 WL, 1060 lbs

10

Tip Extensions - 1150 lbs

Winglets - 1150 lbs
ASW-22B - 1620 lbs
Tip Extensions - 1620 lbs
Winglets - 1620 lbs

0
40

60

80

V (kts)

100

120
0
2

10

Thermal Strength (kts)

Fig. 7 Comparison of predicted lift-to-drag ratios

for unballasted and ballasted Discus 2, with and
without winglets.

-5

Fig. 10 Percentage gain in predicted average crosscountry-speed due to tip extensions and winglets
relative to an unballasted ASW-22 without winglets.

Discus 2 WL, 685 lbs

Discus 2 WL, 1160 lbs

VCC %

L//D %

Baseline: Discus 2, 685 lbs

Winglet Toe Angles
(From zero-lift angle)
-3 deg
1 deg
5 deg

0
40

60

80

100

120

V (kts)

-1
0
2
-1

11

Thermal Strength (kts)

10

Fig. 11 Percentage change in predicted average

cross-country speed as it depends on winglet toe
angle for an unballasted Discus 2.

12