AERODYNAMICS OF FLIGHT AT SPEEDS UNDER 5m/s

by
P.B.S. Lissaman, Vice President AeroVironment Inc.
H.R. Jex, Senior Aerodynamicist, Systems Technology Inc.
P.B. MacCready, President, AeroVironment Inc.
1. Introduction
The Gossamer Condor, with its flight speed of less than 5 m/s, operated in a relatively unexplored flight realm. Here we explore various aspects of its aerodynamic
characteristics, particularly those aspects which during the vehicle development
program were investigated with advanced analysis techniques: stability and control, and airfoil design. The focus is always on the Gossamer Condor (in its final
form or in earlier versions), but the discussion will provide some insight about
certain phenomena, which will be encountered with other extremely slow flying
aircraft. A brief history of the Gossamer Condor development has already been
published (MacCready 1978) which includes a treatment of power required this
paper omits that aerodynamic subject.
Appendix A gives details on computer simulation of the flight dynamics of the
configuration which made the Kremer Prize flight. It was prepared by Henry Jex,
utilizing program and computer courtesy of Systems Technology Inc.
2. A Perspective
An underlying philosophy of the design of the Gossamer Condor was the vehicle
should be made so large and light that there would be a substantial margin of
Dower to handle any problems which might arise. Power required for an airplane
of a fixed shape varies as (weight)^3/2 and (span)^-1; thus if you keep weight low
while increasing span you can produce a vehicle which can fly with any arbitrarily
low power value. We had no preconceived idea of how stability and control would
be handled just the faith that they could be handled effectively if there was a, reasonable power margin. It turned out this faith was justified. The solutions yielding
a satisfactory design were fairly simple, although finding them was not. Coping
with the stability and control phenomena required assimilating clues from flight
tests on many different vehicle configurations, from model tests in air and water
and from computer analyses. The configuration still empirical, based on the clues
and some intuition. Subsequent numerical analyses presented here in Appendix A
substantiated the conclusion from flight tests that the configuration handled stabil-

ity and control satisfactorily. The challenge of achieving stability and control came
especially from the scale of things (so large, so light).
A scale factor of special importance in assessing vehicle dynamics is mass which
ratios the mass of the vehicle to the mass of a cube of air could enclose the vehicle.
Table I gives this relative mass for representative devices.
Table 1 relative Mass of Flight Systems
Vehicle

Relative Mass

Commercial Jet (Boeing 707)

Bird (Sea Eagle, Haliasetus Albuilla)
Sailplane (Standard Cirrus)
Dragonfly Aeschna Cyanea)
Human powered aircraft (Gossamer Condor)

0.86
0.24
0.077
0.0033
0.0031

The Table shows the Gossamer Condor is in the lightweight flying insect range,
several orders of magnitude removed from the range of normal aircraft for which a
background of aeronautical tradition and insight has been built up over the years.
The same fundamental equations still determine vehicle dynamics but designers
experience is of less value than normal. (in some ways it can even be of negative
value) and flight tests with model or full scale vehicles of proper relative mass are
of greet value. Since model tests on the dragonfly scale are difficult, it was logical
to do model tests with a normal size (~ 1m span) model in water. So that the low
relative mass ratio could be approximated.
While we should not give the impression that the Gossamer Condor was designed
according to some random cut and try technique it is certainly true that much of
design was roughed out and then proven in the ultimate medium flight. A deliberate effort was made to avoid overkill in design or many standard aerodynamic
calculations which could in principal have been made with some precision were
executed in greatly simplified form, on the that provided the performance was
approximately in the desired range, flight performance would be satisfactory. It is
fair to say that in this area the collective experience of the design team and of
many experts who gave advice proved an invaluable aid. This approach then permitted us to focus or what appeared to be the more difficult aerodynamic problems,
which discussed in this paper and here we used an amalgam of methods, experience, observation and test. We do not suggest that this is necessarily suitable for
design of other types of vehicles, although we believe that, in all cases, detailed
design analysis is necessary only if it meets the criteria that the results of analysis
are of sufficient importance to merit the effort.

3. Philosophy of Controllability VS stability

the starting concept was to ignore stability and figure that controllability was all
that was needed. The flight speed was very slow (under 5 m/s and initially as low
as 3 m/s), and certainly damping would be high because of the vehicle proportions,
and so if controllability were satisfactory the pilot could easily react quickly
enough to put, the vehicle into any desired attitude as it turned out the final vehicle
seemed to be near neutral stability around all axes and hence the pilots control
task was not difficult. Any mental or physical effort for control decreases significantly the power the Pilot has left to propel the aircraft. Thus we found that it was
essential to have control forces light. This dictated the use of ailerons on the canard
(stabilizer) to cause the surface to roll, rather than using the pilots power directly to
roll the surface directly via wires. These ailerons took very little force to manipulate - Just an easy wrist twist but they added to the sluggishness of control; the
canard first had to roll, powered by the slow moving oncoming wind, and then it
could give yawing moment.
4. Illuminating the challenge
The first vehicle had a 29m (95 loot) span (96' was intended, being pieces of 12'
tubing, but the hanger at Mohave airport could not accept 96') and a 3.7m (12 foot)
chord, see Figure l. Structural problems left the airfoil so misshapen that the drag,
was high. Nevertheless, by a sequence of many changes, the airplane was able to
make a 2 1/2 minute flight. From the very first, pitch control proved to be adequate. This had already been indicated by tests on two 2.5m span models and a
27m span full scale model which the crew could run with and control in pitch
without a pilot on board. As the flights moved toward exceeding a minute, for the
first time we began looking hard at the problems of lateral stability and control. It
was obvious the vehicle was not stable laterally. This was not surprising since there
were no vertical surfaces of any sort, just the horizontal wing and horizontal canard. In flight the Gossamer Condor would slide off to one side or the other and
contact the ground. We introduced the tilting (rolling) stabilizer in configurations
which permitted it to be controlled by the pilot or automatically by tabs which
sensed the amount of yaw of the vehicle. We tried ailerons of the wing tips, fixed
and movable vertical surfaces the of the wing 3/4 of the way out, and used several
kinds of spoilers on the top surface near the tips. We also tried adding vertical area
above the by filling the space between the king post and the rearmost wire mylar
sheet.
The tests with all these configurations were not encouraging. In fact it seemed that
addition of and manipulation of the various surfaces had very little effect on the
subsequent motion of the aircraft. One significant clue did start emerging, although

somewhat unclearly. The rapid raising of a spoiler (0.6m high, 2m long) near one
tip would momentarily cause a small, sharp yaw reaction, but keeping it extended
would not produce much further effect. Also, once the spoiler was up the pilot
could then not keep the vehicle flying level; more power was needed. These phenomena were somewhat obscured by the fact that more propulsion power was also
required when the turbulence was stronger. Finally, on one day when we were
doing crude tow tests, dragging the aircraft behind a car with an observer reading
force with a spring scale, we began correlating drag with spoiler control motions.
The main interpretation of all these clues was that the drag added, impulsively, bar
the spoilers was obviously high, affecting a large circulation, and yet big events
happening out near a tip had little effect on roll.
When one pushed the vehicle into the air, the amount of force required was surprisingly large, and the pilot-engine could not keep the plane aloft for more then a few
seconds. There seemed to be an effect of getting on the step, when the thrust
required would be much less. The 40 second flight on December 26th, 1976 (Figure 1) was the first really sustained flight. Evidently it got to the right speed and
stayed there, without acceleration and putting energy into the starting vortex which
arises during acceleration. The 2-1/2 minute flight a month later was powered by a
strong, championship-level cyclist. It was made in very quiet air, involved only one
operation of a spoiler for direction control.
The likely importance of apparent mass (the resistance to vehicle acceleration
caused by the coupling to the air) became evident when we did the calculation
showing that the roll inertia of the wing was increased by an order of magnitude by
the apparent mass. Eventually rolling the wing would require large roll forces and
be expensive as regards energy.
5. Redesign
A first cut at calculating, controllability factors by means of a computer model
showed one striking; effect that a large lift change on one wing would have very
little direct effect on roll. The limited modelling did not suggest how to improve
the situation, but one approach was evident. Decrease the chord at the tips. This
cuts down apparent mass effects there as a function of the
square of the chord, and also cuts down roll damping in direct proportion to the
chord. This design change decreased wing area and hence increased the flight
speed. Me speed increase required more attention be given to cutting the parasite
drag coefficient, which meant putting a streamline housing around the pilot area.
The only other major redesign factor was adopting a 2 surface airfoil. This served
two vital functions: a) it permitted the use of an airfoil which was much less criti-

cal to angle of attack changes from turbulence than the original rather sharp-edged
single surface airfoil (the drag bucket was broadened), and b) it permitted a single
spar to be located near the center of pressure, facilitating and lightening the structure, while letting this spar be completely out of the airstream. The low pitching
moment coefficient of the double surface airfoil was an added bonus.
The resulting second-generation vehicle, built at Shafter Airport, went through
many tests and design stages. However, there was really little difference, beyond
airfoil thickness and structure, between the final config configuration which made
the Kremer Prize flight August 1977 (Figures 2 & 3) and the very first configuration flown at Shafter Airport on March 15, 1977. There were some important control changes. We tried varying center of gravity positions (i.e., varying loads on the
canard), and varying methods of effecting the roll of the canard. We experimented
briefly with a vertical fin on the canard and then with a rudder positioned on a
frame far behind the propeller, and finally we provided a wing twisting capability.
During the redesign stage in February 1977, clues for the design came from the
above computer analyses, from tests on various configurations of the Mohave
vehicle, from short glide tests with a 2.5 m balsa and tissue paper model and a 1m
span balsa sheet model, under water tests with a 1 m span balsa sheet model, and
the computer analyses. The water tests were especially helpful. They were performed so as to get a feel for phenomena when apparent mass effects would
dominate (when the relative mass ratio eras very small). As one hushed the model
through the water and felt the force it was evident that coordinated turns were
feasible with low drag, but that accelerations involving yaw without roll or roll
without yaw created large drag The tests provided encouragement that a vehicle
could be gentled but not horsed around a turn. The gentling involved motion
around an axis perpendicular to the wing surface, i.e. tilted slightly back from the
vertical. Certainly at zero speed such motion involved the least apparent mass for a
rotation. In ordinary flight the situation was the same, providing: in effect a decisive yaw-roll coupling coincidence with turning with little additional power.
In summary, there was an expectation but no computational assurance that the new
Shafter Airport configuration would turn without large turning force being required, and that a rolling canard would have more effectiveness on the new vehicle
than on the old. There was no real information about how to effect coordinated
turns. The tests with the new vehicle at Shafter Airport showed the rolling canard
to be very effective in producing well coordinated (draw-roll) direction changes of
some 200, but it was not satisfactory for handling greater direction changes. By the
time the Gossamer Condor made its last flight late in 1977, the situation with respect to controllability was that the vehicle worked well, there was some appreciation for the reasons underlying the good control and stability characteristics, but
there was no complete understanding. The development was based on clues de-

rived from many areas, with a very strong

emphasis on flight tests with various vehicle shapes and controls. The computer
modeling of vehicle dynamics had served its purpose but was hastily done and not
evaluated completely. The more complete modeling presented in Appendix A. is
more informative - and in general agreement with the flight observations.
The vehicle was still a quiet air performer. It was occasionally operated in winds as
high as 3.5 m/s flying only straight into the wind; turns were not attempted at wind
speeds exceeding 1.5 or 2 m/s. As convective conditions began in mid morning the
control was just able to permit safe flight, but ground handling became unsafe
since a 1.5 m/s cross or tail wind could readily damage the plane on the ground.
6. Some Analytical Design features
Only three aspects of the design were derived with the aid of detailed analysis; the
airfoil section, the propeller design and the lateral dynamic stability.
Initially it was hoped that a single surface airfoil, with a finite nose radius formed
by the main spar tube, would be satisfactory. This appeared to perform tolerably at
design attitude but had a very limited low drag range with angle of attack. A
number of different rigid leading edge fairings were examined, analysed on a computer and tested in flight. An example of some of the many airfoils investigated is
shown in figure 4. It will be observed that small leading edge changes, imperceptible in the figure, have substantially reduced the leading edge pressure peak. Flight
tests verified the higher lift coefficient obtainable with the leading edge more carefully shaped. In this design process a simple quasi-linear computer airfoil design
program, providing a rapid turnaround with direct plots of airfoil shape and pressure distribution proved very useful. It was possible to iteratively change the airfoil
at the computer console; and, with experience, one could develop half a dozen
airfoils in an hour, each aimed at correcting some undesirable feature of the previous one.
It was finally decided that a normal finite thickness airfoil would be required for
the Shafter version of the Gossamer Condor. The requirements of the airfoil were
that its design CL should be about 1.0, and that it should have a low drag at design
attitude, with a reasonably wide low drag range; that the pitching moment should
be smaller than -0.04, and that the CLmax should be about 1.5. Very important
other requirements were that it should be easy to build, that there should not be
extreme curvatures (particularly concave which would not be achievable with the
flexible covering, and that the performance should tolerate imprecise construction

and billowing of the covering. A further feature which was felt to be desirable in
the layout and construction of the airfoil was that most of the lower surface should
be flat. It was found essential to have some nose camber to suppress pressure peaks
in that vicinity, so the final design was contrived to be flat from 100/6 chord aft.
The airfoil used on the Gossamer Condor was the Lissaman 7669 (Figure 5). The
two-dimensional characteristics of the airfoil have never been determined, but
flight observations with tufts indicated that flow was fully attached at cruise lift
coefficient. When the same airfoil was employed on a later airplane, the Gossamer
Albatross, with much more accurate and rigid surfaces, the cruise lift coefficient
was found to be 20-30% higher, and the section drag coefficient considerably
lower than on Gossamer Condor. A slightly different airfoil, differing primarily in
being 13.5% thick instead of 11% thick, was tried in flight on the Gossamer Condor and found to have much higher profile drag.
The same 11% airfoil was used on the canard, and on the propeller. The propeller
design philosophy was to use the simplest geometry compatible with good performance, which predicated a constant chord blade with the appropriate twist. A
number of twist distributions were examined on the AeroVironment propeller design computer programs, and the performance for a range of advance ratios was
examined, to select a propeller design which appeared to have the most desirable
performance over the operational range. Here, as with the airfoil considerable
effort was spent in iterating through design variable and computer performance
plots with the final design being selected as one which showed an acceptable wide
range of good performance. Testing of the propeller was done on the aircraft, in the
air. No measures of the propeller performance are available, but estimates of the
aircraft drag and pilot power output suggest it is in excess of 75%. Two different
propellers were tested on the Gossamer Condor both of 3.8m diameter with one of
40cm chord and one of 30cm chord. Pitch was adjusted by twisting the blades on a
shaft, to give the pilot a pedaling rate near 90 rpm at cruising speed. Both propellers appeared to have comparable performance. It is believed that the location of
the propeller directly aft of the wing and fuselage was very helpful in that these
flying surfaces produced a rectifying or guidvane effect, straightening the inflow to
the propeller.
7. Final Comments and Conclusions
(1) For an aircraft which flew extremely slowly (<5 m/s) and hence has low relative mass (vehicle mass v. mass of the air to which it couples), there is so little
background experience that standard intuition is a poor guide to vehicle dynamics.
Thus it is beneficial to treat the problem using computer simulation based on fundamental concepts - particularly for lateral stability and controllability.

(2) Computer analysis was also a valuable tool for selecting and design of specialized aerodynamic features involved in the airfoil and propeller.
(3) Flight tests showed that the less the CL of the canard, the less the pitch stability.
There was always adequate pitch controllability. With the large ditch damping of
the vehicle, the pilot response time was comfort ably faster. Than any pitching
motions and so pitch stability was not critical. With the canard lightly loaded, say
operating at a CL 0.3, in turbulence the pilot would have to be correcting for pitch
rather often, which was a slight distraction Paid probably harmful to efficient flit
both aerodynamically and as regards pilot power potential. Of course the higher the
loading on the canard, she stronger the yawing torque as the canard was rolled. A
satisfactory compromise operation of the canard was Cl ~ 0.6, which gave the
potential of a wide Cl range for control, adequate loading for yaw control, and
satisfactory total stability.
(4) Model tests at low relative mass in water suggest that a wing with large apparent mass influences rotates most July in its own plane. This couples yaw and roll in
a manner which opens up practical mechanisms for effecting turns. The complete
computer analysis of the Gossamer Condor dynamics is consistent with this particular coupled motion, as are the full scale flight tests. A manifestation of this
apparent mass phenomenon is that large up or down forces put out on the wings by
the ground crew would have very little effect on roll. Small backward or forward
forces near the tips, however, would have a large and immediate effect on roll.
(5) Coordinated turns, at radii down to about 200', would be initiated and continued by wing warping, with the canard roll used for trim. The wing with wash-in
would be on the inside of the turn, just opposite to the case of initiating a turn in a
normal airplane with strong yaw stability.
(6) The final configuration had controllability completely satisfactory for the purpose for which the vehicle was developed. Evidently the amount and location of
the vertical fuselage area was just what was needed to provide the sort of yaw-roll
coupling which made the rolling stabilizer twisting wing system yield coordinated turns. It is worth noting here that the rolling stabilizer twisting wing concept is not new; soaring birds have been using it for one hundred million years.
Incidentally, the power requirements during turns seemed to increase more than
expected, say about 20% for a 75m radius.
(7) The computer simulation of lateral and longitudinal stability/controlability
given in Appendix A gives insight into why the vehicle dynamics of the Gossamer
Condor were satisfactory.
(8) The designers might like to take credit for creating every successful solution to

flight dynamics problems, but they appreciate the large role played by luck in the
design process. For example, the size and positioning of the vertical area turned out
to be a good choice. However, the choice was forced on the design by 1) the need
for the pilot to be close to the center of gravity since he dominates the vehicle
mass, 2) the need for him to be enclosed in order to decrease parasite drag (which
might not have been required if the tip chord had remained large and the flight
speed low, but lateral control dictated a narrow tip chord) and 3) the need for the
fuselage to be deep to provide a low support point for the flying wires.
(9) The vehicle dynamics analysis of Appendix A is ix A is illuminating but also
shows that a more careful evaluation would be justified, one which explores many
configurations and control variations.
Reference
MacCready, P.B., 1978, Flight on 0.33 Horsepower: Gossamer Condor. AIAA
Preprint

Reproduced from: "Man-Powered Flight the Channel Crossing and the Future."
Proceedings of the Third Man-powered Aircraft Group Symposium at the Royal
Aeronautical Society, London, 6th February 1979.
J McIntyre 2004.

Figure 3. Three view drawing of final Shafter configuration.

Shown in flying position. No wires shown.
Span 96'
Height 19'6"
Length 30'
Scale 1:125

-5.0
Cp
-4.0

Airfoil 7604

-3.0

-2.0

-1.0

chord
1.0

Figure 4. As shown airfoil and pressure distribution.

All Values in Percent Chord

x
0
1.25
2.5
5.0
7.5
10
15
20
30
40
50
60
70
80
90
95
100

upper
surface
0
2.25
3.34
4.96
6.15
7.06
8.40
9.26
9.92
8.97
6.96
4.86
3.16
1.81
0.84
0.41
0

lower
surface
0
-1.64
-2.01
-2.30
-2.30
-2.16
-1.70
-1.38
-1.06
-0.91
-0.75
-0.60
-0.45
-0.30
-0.16
-0.08
0

Nose radius 1.84, center of nose circle (1.84, 0.14)

Trailing edge angle from chord line
Upper surface 4.50, Lower surface -0.90
Figure 5. as shown 7669 airfoil.