A study of mechanical

flapping-wing flight
J. D. D E L A U R I E R
Institute for Aerospace Studies,
University of Toronto,
Downsview, Ontario, Canada

and J. M. HARRIS
Battelle Memorial Institute
Columbus, Ohio, USA

(like a propeller), and fixed wings provide the lift. All of the
ABSTRACT above designs have been called "ornithopters" by their builders,
but only those which meet the "bird like wing" definition will be
The feasibility of mechanical flapping-wing flight has been stud- referenced and discussed.
ied by analyses and experiments. The key results from this work Although nature's flapping-wing examples had been humanity's
include the development of an efficient wing with unique features original inspiration for achieving flight, the successful mechanical
for twisting and lift balance, as well as a lightweight and reliable realisation of this has been limited largely to small rubber-
drive mechanism. These were incorporated into a radio-con- powered ornithopters built by aeromodellers*1*, which all derive
trolled, engine-powered, flapping-wing aeroplane (ornithopter), from the 1874 model flown by Alphonse Penaud*2'. A current
whose flight tests have been the proof-of-concept focus of this commercial example of this is the "Tim Bird" (Figure 1) which,
research. In September 1991, this aircraft achieved successful sus- like Penaud's, uses a twisted rubber band operating a crank
tained flights, demonstrating the practicability of this particular system to flap the wing spars.
solution for mechanical flapping-wing flight. A more sophisticated series of rubber-powered ornithopters
were built by von Holst<3' for his study of bird flight. These have a
very intricate stick and tissue construction, with a drive mecha-
nism incorporating an eccentric drum which varies the supplied
moment to the flapping wings. Therefore, maximum flapping
NOMENCLATURE forces are applied to the wings when needed during the cycle.
As for motorised flapping models, successful examples are few
b wingspan and sparsely documented. Those that warrant particular mention
CLe local lift-curve slope wrt flapping axis angle are the engine-powered ornithopters designed by Percival Spencer
c local chord (reported by Dwiggins<4>) and David Atkins (reported by Brooks,
y spanwise coordinate et a/<5>)- However, no design details or performance figures were
% chordwise distance from a fixed reference point to the given.
local aerodynamic centre A notable exception is the 18 ft-span robot pterosaur designed
xac chordwise distance from the reference point to the and constructed by AeroVironment of Monrovia, California<5'.
wing's aerodynamic centre This remarkable aircraft, with its computerised stability

INTRODUCTION

The purpose of this work has been to assess the feasibility of

mechanical, powered, flapping-wing aircraft. These are called
"ornithopters", which means "bird like wing". The authors have
interpreted this definition to mean that, although such an aircraft
need not look like a bird per se, its means of flight are clearly bird
like, even to a casual observer. Therefore, tail first (canard) or
multiwing insect like configurations were excepted. Further, it
was judged important that the flapping wings should provide all of
the thrust and nearly all of the lift. Thus, exception was also taken
to configurations where the flapping surfaces provide thrust only

Paper received 19 August 1992, revised versions received 24 March 1993

and 13 May 1993, accepted 1 July 1993.
Paper No. 1917. Figure 1. "Tim Bird" rubber powered model ornithopter.

Aeronautical Journal October 1993 DeLaurier and Harris 277

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 augmentation and flight control systems, provided a valuable
Support struts |K4 stlown
reinterpretation of the appearance and flight mechanics of these
prehistoric creatures. However, as discussed in Ref. 6, the require-
ment for biological realism and the on-board battery weight did
not allow the aircraft to achieve sustained flapping flight.
For full scale piloted flight, the authors are not aware of any
successful engine powered ornithopters. However, flights have
been achieved with human powered ornithopters, the most
successful of which was tested by LippischC) in 1927. This was
capable of extended glides, but not sustained flight. Similar per-
formance was also achieved by human powered aircraft tested by
Vogel<8> and Hartmann<9>. Clearly, this ancient goal of flight has
posed formidable technological obstacles.
The present approach to mechanical flapping-wing flight owes
little to the previous examples. As described in Ref. 10, the mem-
brane wing design of the Penaud-type flappers was found to be
0 1ft too inefficient to be attractive for a larger engine powered aircraft.
Also, although it appears to be bird like in its actions, its low ad-
vance-ratio, vectored thrust, flight mechanics are quite at variance
with those for large birds in level cruising flight. Therefore, it was
established that the ornithopter's flapping wings should provide
horizontal thrust, with the resulting flight speed acting on the
wing's chordwise-rigid camber to generate a mean lift. However,
it should be stated that this wing design also differed from birds'
wings in several important ways. Early in the project, slow-
motion movies of birds in level equilibrium flight were studied in
order to gain some insight on typical wing motions. Very little
was learned from this because there is a complexity of motions
("tip flip", feather spreading, fore-and-aft swinging, semispan
variation, etc) which would be mechanically intimidating if each
motion were considered to be of equal importance. Instead, it was
hypothesised that the important fundamental kinematics for
Figure 2. 24 September 1986 version.
efficient flight consist of a simple harmonic up-and-down flapping
motion accompanied by phased pitching (wing twisting). Further,
it was assumed that efficient flapping could be achieved with a
constant semispan wing incorporating a smooth-skinned (non
feathered) aerofoil.
Finally, it should be noted that the drive mechanism and wing
flapping kinematics were also original developments. As
discussed further, the lift balancing multipanel wing design was
harmonically driven by a lightweight and reliable transmission
which reduced the engine's high rotational speed to the low fre-
quency required for the wing's flapping. This approach is different
from previous ornithopter practice, and is also unlike the muscle
and tendon actions of birds.

WING DESIGN

Figures 2-4 show the various evolutions of the ornithopter. The

wing details vary, but in all cases they are characterised by
consisting of three hinged panels with chordwise rigid ribs and
constant spanwise lengths. The three panel feature, which was in-
vented by Harris, serves to balance the time-varying lift seen by
the fuselage. That is, the centre portion of the wing is flapping in
one direction while the outer portions are flapping in the opposite
direction. Whereas it is not possible to completely balance the
inertial and aerodynamic loads for all flight conditions, this ten-
dency to achieve a dynamic balance would be a valuable feature
for a human carrying aircraft.
Additionally, for ornithopters of any scale, the three panel wing
provides a means for evening out the power required from the
engine during the flapping cycle. That is, the engine sizing for
ornithopters is determined by the peak power required during the
cycle. For traditional two panel designs, the power required for
Figure 3. 24 October 1989 version. the downstroke may be several times greater than that for the

278 DeLaurier and Harris Aeronautical Journal October 1993

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 Support struts not shown

Figure 5. 1985-1987 aerofoil.

system, but it had the aerodynamic disadvantage of producing

large, variable, spanwise gaps between the centre and outer pan-
els. Pieces of stretchable fabric were used to cover the gaps; but
this, aerodynamically, was not a particularly efficient solution.
In 1989, the support struts were changed to a system incorpo-
rating hinged vertical links. These allowed the sliding hinges at
the centre panel to be replaced with simple non-sliding units. The
gap then remained small and could be aerodynamically sealed
with narrow abutting extension strips attached to the end faces of
the panels. This system operated reliably during the course of the
flight tests in 1989 and 1991.
The outer panels between 1985 and 1987 incorporated a single-
surface aerofoil and two spar structure as shown in Fig. 5.
Attached to the ribs were short spanwise tubular pieces through
which the spars passed. Therefore, the spars were free to rotate
within the tubes. The original idea was that since the single sur-
face aerofoil has torsional compliance, the outer panel's pitching
distribution could be driven by spanwise rocking of the two spars.
This would be accomplished by pitching the centre panel, since
the hinges acted as universal joints of the Hooke/Cardan type. The
spars would then rock slightly out of phase with each other, direct-
ly producing both plunging and pitching distributions along the
^==UB^ span. As it turned out, sufficient outer panel pitching was pro-
duced by structural deformation alone, and pitching input from the
Figure 4. 4 September 1991 version. centre panel was not required. It was also determined that the centre
panel's relative angles of attack would remain below stall values
upstroke. Therefore, the required engine may be much larger than for plunging only. Therefore, its pitching mechanism was dis-
average power considerations would dictate. This would result in pensed with from 1986 onwards.
an aircraft heavier than a fixed wing propeller driven aeroplane of In detail, the outer panel front spar bending resistance was
comparable performance. Hence, omithopter designers have increased by means of bracing wires (as seen in Fig. 2), and the
sought ways to store energy on the upstroke. For birds and bats, panel's torsional stiffness was determined primarily by the bend-
some storage may be provided by their tendons. It is more likely, ing stiffness of the rear spar. This was varied considerably during
though, that the upstroke is the period of repose required by the the course of the flight tests by changing the rear spar's length and
downstroke muscles. Such repose is not needed by mechanical material. Initially, in 1985 and early 1986, the rear spar was sim-
systems, and energy storage has typically used elastic means, such ply a 74 in diameter birch dowel extending about halfway along
as the bungee cords incorporated into the AeroVironment the panel's semispan. There was no analytical guidance for this. It
pterosaur replica. However, an omithopter isn't required to be a was just a first attempt at what seemed to be a reasonable design.
scale model of an animal, and the three panel wing thus provides Later, as it became evident from flight tests and the developing
an attractive means for power evening as well as dynamic analysis that the wing had excessive torsional compliance, the rear
balance. spar was extended and a portion of it changed to a ]U in diameter
As seen in Figs 2-4, the wing's flapping motion is actuated fibreglass rod. In fact, there was a flight test session where the rear
through the centre panel's vertical oscillation, driven in simple spar was a 74 in diameter solid carbonfibre rod.
harmonic motion by a mechanism which will be described in the Guided by the analysis, the ornifhopter's performance improved
next section. The centre panel is constructed to be rigid in bending steadily during this course of development. However, it was never
and twisting, and incorporates a simple flat-bottom aerofoil (nom- near to achieving sustained flight. That is, the flight mode was a
inally, a Clark-Y). It was originally envisaged that the centre panel "powered glide", where the flapping would clearly extend the dis-
would require a direct pitch articulation in conjunction with its tance flown compared with the non-flapping cases, but it would
vertical (plunging) motion; but it was found, as described later, not climb or maintain altitude.
that it operated most effectively in plunging only. The reason, revealed by windtunnel tests*11), was that the actual
The outer panels operate in a spanwise-rocking ("seesaw") aerodynamic characteristics fell short of the estimated values used
fashion, driven at one end by hinge connections to the centre panel. as inputs in the analysis. In particular, the aerofoil had high chord-
There were two systems of support struts and hinges for the outer wise drag (Cdf = 0-063), poor leading-edge suction (r)s = 0-29),
panels in the course of their evolution, as seen in Figs 2-4. From and separated flow on one side or the other at all angles of attack.
1985 to 1987 the strut assembly was rigid, with a single hinge In comparison with modem double surface sections, this aerofoil
connection to the outer panel. The kinematic requirements were was a very poor performer (typically, for a Reynolds number of 2
then met by means of spanwise-sliding hinges to the centre panel. x 105, a modem aerofoil should give Cdf = 0-012 and r|s = 0-90). It
That is, the driving hinges on the end of an outer panel had lubri- was desired, however, to retain the single-surface feature because
cated cylindrical extensions which slid, piston-like, into the centre of its torsional compliance, and to find ways to improve such an
panel's hollow spar tubes. This proved to be a very reliable aerofoil's aerodynamic performance.

Aeronautical Journal October 1993 DeLaurier and Harris 279

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 Figure 6. 1989 aerofoil.

As discussed in Ref. 11 and shown in Fig. 6, a candidate sec-

tion was found. Note the absence of the rear spar. It was decided
to rigidly attach the ribs to the front spar, and let the torsional
compliance be determined by that spar's characteristics. This has
Figure 8. Top view of the drive mechanism.
the merit of improved aerodynamic cleanness (measured
Cdf• = 0-019, t| s = 042) but one also loses the ease with which the Finally, it was found from the ComboWing calculations that the
compliance could be changed. Therefore, one has to have a great Mark-8 panels are "design robust" in the sense that the optimum
deal of confidence in the ability of the analysis to accurately characteristics lie on a fairly broad plateau where normal con-
predict the performance of the "as built" panel. struction variations would have small effects on the average
Simultaneous with the flight tests and aerofoil experiments, the thrusts and lifts. This is a valuable feature because even slight
analysis had been steadily developed, extended, and evaluated aerodynamic asymmetries from the long outer panels would
with windtunnel tests. As described in Ref. 10, its culmination require excessive control trim for straight flight.
was a comprehensive computer code named ComboWing which
calculates a flapping wing's aeroelastic response and resulting
aerodynamic performance. This was used to design a new outer
panel, the "Mark-6 wing" shown in Fig. 3. Notice that bending
stiffness still required short external guy wires on the front spar. DRIVE MECHANISM
However, the flapping system now used the vertical link arrange-
ment, with the resulting aerodynamically improved smaller gaps The first design goal of the drive mechanism was to convert, in a
between the panels. simple and lightweight fashion, the rotational output of a small in-
The expectation of better performance was borne out by glide ternal combustion engine to a 3 in amplitude (6 in peak to peak)
tests. However, a series of problems involving control, premature vertical sinusoidal oscillation in the range of 3-5 Hz. The
engine cut out, and launching compromised a fair evaluation of second goal was to efficiently couple this oscillatory drive to the
the Mark-6 wing in powered flight. It should be mentioned that centre panel of the three panel wing assembly. If this could be ac-
tests on 24 October 1989 showed climbing flight for the short du- complished, the centre panel and strut-supported pivots would
ration of the engine run (14 s). It was difficult to state, though, that take care of passing the driving effort to the outer panels. The
this constituted true sustained flight because the aircraft did not fly mechanical means selected were two stages of fibre-reinforced
far enough to be unequivocally away from the influence of the toothed belting for the rotational reduction, and a scotch-yoke for
launching hill's ridge lift. the conversion from rotation to reciprocation.
Flight tests with the Mark-6 wing would have continued into The drive underwent several overhauls and modifications dur-
1990 except that, during this time, a new outer panel design was ing the ornithopter's flight test evolution, but always conformed to
being developed which promised a very significant performance the basic configuration illustrated in Figs 8 and 9. The engine is an
improvement. Reference 10 describes this new panel, called the O.S. Max 45 FSH model helicopter version (enlarged cylinder
"Mark-8 wing", in detail. Briefly, this has a unique, torsionally- head for extra cooling capacity) of 0-45 in3 displacement. Its
compliant, structure with an efficient double surface aerofoil. That output shaft carries a cooling fan, the first stage input pulley, and a
is, a split trailing edge feature gives what is, essentially, two sin- cylindrical starting collar. Extended socket head set-screws fasten
gle-surface wings lying one over the other and rigidly attached to the collar to the shaft and mate with a cross-slotted steel
a common leading edge (described as "Shearflexing"). Therefore, dog mounted on the shaft of a handheld starting motor. This
one has the advantage of a single-surface wing's torsional compli- arrangement was found reliable, convenient, and more positive
ance along with the ability to incorporate a modern aerofoil with than the conventional rubber friction inserts supplied with hand-
cross-sectional thickness. The aerofoil used, the S1020 (Fig. 7), held electric starters. The engine's measured torque and power
was designed especially for this project by Michael Selig (then at characteristics (with the custom built muffler) are shown in Fig.
Pennsylvania State University) to have a wide angle of attack 10, where it is seen that the maximum power output is about 1 hp
range for attached flow and high leading edge suction efficiency. at ~ 12 000 rpm and full throttle. Estimates of the maximum in-
Windtunnel tests bore out his predictions (r|s = 0-92) and the stantaneous power required by the ornithopter are shown, in the
Mark-8 wing'benefited accordingly. When incorporated on the or-
nithopter (Fig. 4) the aircraft performed unequivocally sustained
and climbing flights, as described later.

Figure 7. 1991 Selig S1020 aerofoil. Figure 9. Side view of the drive mechanism.

280 DeLaurier and Harris Aeronautical Journal October 1993

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 foundations were strengthened. The final value of the overall drive
T • arilh muilei
ratio was 54-5 to 1, composed of 6-67 to 1 in the first stage and
» o wtttwut muffler 8-18 to 1 in the second stage. Engine rpm at 3 Hz flapping
frequency was 9810.
~- Estimated Hp
From the outset of the programme, modifications of all kinds
were tested with static bench runs. In particular, repaired wings or
Calculated max ftslamansojs
newly-designed wing panels were flight qualified by enduring a
required erijinaHp for
, 4Sep!.'910mltioc&(
specified amount of static running, usually 20-30 s, at full flight
O
frequency. The frequency was measured by attaching a
microswitch where the centre panel would actuate it once per
Estimated Torque cycle. The switch drove an electronic counter which displayed the
- — Calculated max instantaneous
period (reciprocal frequency) in seconds.
requfred engine torque lor
4Sept.$10m»iopter
A critical factor indirectly associated with the drive mechanism
was the often vexing problem of simply keeping the engine run-
14 15
ning after a launch. This difficulty was suppressed for a long time
RPM (x1000)
because aerodynamic limitations restricted all flights to the pow-
Rapping freq. • 23 Hz S.SHz
ered glide mode with durations of 27 s or less. However, the two
3.0 Hz
particularly promising flights of 24 October 1989 were cut short
by early engine stoppage, and it was recognised that such events
Figure 10. Power characteristics of O.S. Max 45 FSH engine. were increasingly less tolerable as the capability to sustain flight
approached realisation. In June 1991 an extended bench test was
lower portion of the graph, to be less than 0-5 hp; thus the engine
run in which the ornithopter was cycled through nose-up and
is adequately sized to provide a comfortable power margin. The
nose-down attitudes and the fuel system components were closely
fuel used for these power tests, as well as all flight tests, was
observed. Fuel foaming in the tank and bouncing of the tank's
methyl alcohol mixed with castor oil and 7% nitromethane
flexible internal intake line were identified as the most likely cul-
(marketed commercially as "Apollo 7").
prits. Stiffening of the intake line and placement of slosh-damping
It should be noted that an internal combustion engine was
plastic scouring pad material in the tank cured the engine failure
selected because it has a higher power to weight ratio than any al-
problem in time for the successful flights of the following September.
ternatives, such as an electric motor with batteries. Also, it was
felt that many of the lessons learned in incorporating such a pow- Although nearly every component in the drive train, and for
erplant could be applicable to the drive train design for a full scale that matter virtually every part of the ornithopter, was modified or
ornithopter. replaced at least once during the flight test series, notable excep-
The external second stage belt is wider and coarser pitched than tions were the Thomson recirculating ball bushings used in
that of the first stage and has an adjustable idler on the slack side the scotch-yoke slider. These antifriction bushings, the linear
to control tension. The scotch-yoke crank and its ball bearings are equivalent of ball bearings, were originally selected not for wear
supported by a hollow plywood box structure affixed to the resistance, but because ultra-low friction was essential to
"thorax" floor and right side (see Fig. 14 for an illustration of minimise the required depth of the thorax. The higher the slider
main-component designations). At the tip of the crank is a xli in friction, the larger would have to be the vertical separation of the
diameter ball bearing roller which runs in the steel tracked slot of bushings to avoid binding on the support posts under the eccentric
a slider unit. This, in turn, reciprocates through a 6 in total stroke loading of the crank roller. Consequently, the thorax would have
on a pair of 3/s in hollow steel rods affixed at the top and bottom to grow deeper to contain the slider over its whole stroke. The ball
of the thorax. The slider is bolted to the underside of a glass bushings proved trouble free throughout the vehicle's history.
fibre/balsa/carbon "highrise" structure which passes the driving An important basis for the selection of simple harmonic drive
forces to the centre panel's aluminium leading edge spar tube. motion was a study by Fairgrieve and DeLaurier<l2> which showed
Modifications to the drive mechanism were generally in the that non-sinusoidal flapping would be unlikely to offer aerody-
areas of strength enhancement and ratio increase. During the early namic advantages. In addition, it was felt that practical difficulties
flight trials of 1985-86, the 40 DP (diametral pitch) first stage would be likely to arise from the tendency of any finite mass
drive belt stripped twice and was replaced with a coarser-pitched oscillating mechanism to react with sharply increased peak loads
belt. A little later the shaft supporting the scotch-yoke crank was to motions containing significant harmonic activity above the
remachined from semi-hardened steel after it twisted during a fundamental.
crash load transient. When the wing support arrangement was It should be noted that even the use of a classical sinusoidal
changed in 1988-89 to incorporate vertical links, there was also a motion generator, such as a scotch yoke, does not assure realisa-
planned increase in flapping amplitude from ±17° to ±30° accom- tion of the desired output. The "sinusoidality" of the scotch yoke
panied by a decrease in frequency from 5 Hz to approximately is conditional on a uniform rotation rate at the crank. This, in turn,
3 Hz. The net effect of this conversion was to mandate an increase depends on the response of the drive to the unavoidable variable
in drive ratio to keep the engine speed reasonably near its power back loading from the wings. If the loading variation is sufficient
peak. Additionally, a tradeoff of factors produced higher loads to significantly affect the instantaneous rpm of the prime mover,
throughout the drive and caused an intolerable amount of tooth- then the output will deviate from simple harmonic motion.. The
jumping (cogging) at the second stage input pulley. As a result, a best defence against this problem is a large drive ratio, the benefi-
series of drive modifications was undertaken. The output pulleys cial effects of which can be visualised in two complementary
of both stages were enlarged as much as physical space would ways. Taking the engine's point of view, it has a large (55 to 1)
permit, the belts were widened, the tooth form was changed to a mechanical advantage on the final-drive crank, and also sees the
cog-resistant semicircular profile, and the second stage tensioning rotational equivalent of the wing inertia diminished by a factor of
idler was rebuilt and its foundation reinforced. To combat the squared reciprocal of the drive ratio, or 1/2970 for the subject
slot-track scoring and occasional structural damage due to sudden ornithopter. Looking the other way, the crank sees the small
wing stoppages during crashes, the slider was rebuilt using 'A in inertia of the engine multiplied by 2970, hence the engine's net
aircraft plywood incorporating steel rather than aluminium roller flywheel effect is far from negligible. An excellent illustration of
tracks. For similar reasons the aluminium main crank arm was re- the opposite situation occurs with most rubber powered or-
inforced with steel near the hub, and the crank-support box and its nithopters. The rubber motor is an almost perfect, inertialess,

Aeronautical Journal October 1993 DeLaurier and Harris 281

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 torque producer, and is usually connected with no reduction gear- The downwash was calculated by using the method described
ing at all. Consequently, the rotation rate at the crank varies both in Ref. 13, with the wing positioned at its mean flapping angle.
widely and wildly with the rapidly changing instantaneous back Standard aeroplane methods were also used for estimating the sta-
loading from the wings. The net result is a sharp, violent snapping biliser's and fuselage's influence on the neutral point calculations.
at the ends of the wing strokes. Advanced rubber model builders The result for the 1991 ornithopter was a neutral point located
have recognised this problem and devised ingenious remedies 5-83 in aft of the rotational axis of the scotch yoke crank (the cen-
such as variable diameter pulleys and multiple wing sets flapping tre of the last pulley). Since the mass centre is 0-39 in aft of this
out of phase with each other. point, one obtains that this aircraft has a generous static margin of
55%.
The trim condition required a decalage (based on chord lines)
between the centre panel and stabiliser of 6°. However, note that
the effective decalage is less because of the washout produced by
STABILITY AND CONTROL the wing's mean spanwise twist distribution.
As for lateral stability, it was particularly important for this to
Since the ornithopter was not required to look identical to a flying be inherent because the ornithopter has no direct roll control from
animal, it was decided to provide stability and control with a tradi- ailerons. This was accomplished by giving a mean dihedral to the
tional aeroplane-like empennage assembly located aft of the wing. outer panels; that is, the upstroke flapping angle is larger than the
As seen in Figs 2-A, this consists of a stabiliser and fin mounted downstroke angle. For the 1991 aircraft, this mean value is 7-09°.
on the "rear fuselage" which is attached to the thorax. Also, the As with the longitudinal stability setup, this dihedral angle was
tail and control surface areas were sized in accordance with chosen based on the assumption that the proportions and angles
typical proportions for fixed wing radio controlled models. for stable fixed wing aeroplanes would also be appropriate values
The mass centre was located forward of the estimated neutral for the mean geometry of the ornithopter.
point, as is the usual practice for fixed wing aeroplanes. However, Dihedral is also important for turning, which is achieved by the
note that this is not usually so for Penaud-type ornithopters, which roll/yaw coupling produced through rudder deflection in conjunc-
have an aft mass-centre location to give an upward vector to the tion with wing dihedral angle. Additionally, the angular displace-
wing's thrust and to obtain significant lift from the tail. As men- ment of the flapping outer panels gives a large effective vertical
tioned in the Introduction, this is not the manner in which large area against which the rudder moment may act to produce a turn-
cruising birds fly, nor was this the flight mode chosen for this or- ing force. As discussed in the next section, when the ornithopter
nithopter. It was hypothesised that the route to efficient flapping had flights of sufficient duration so that its turning behaviour
flight was with the thrust vector being essentially horizontal, pro- could be evaluated, its performance proved to be very satisfactory
ducing the forward velocity necessary to generate mean lift from and comparable to stable and controllable fixed wing radio
the wing. Since this is the same thrust and lift arrangement as for controlled models.
fixed wing aeroplanes, it was assumed that similar trim and static
margin requirements apply. It should be mentioned that at the
time this design decision was made, there was no stability analysis
performed on the ornithopter. Flight tests were counted upon
to provide guidance on the workability of this notion. In the event, FLIGHT TESTS
this basic assumption proved to be correct, although certain
adjustments in static margin, decalage, and control surface sizes The chronology of the flight tests is given in Appendix A. These,
had to be made, as will be described in the flight tests section. in fact, were the focus of this work. The analyses and laboratory
The flight tests section also describes the Launch computer pro- experiments were in support of the flight tests, and would be per-
gram, which is a nonlinear three degree-of-freedom simulation of formed only to the amount required to give confidence for a flight
the ornithopter during the launch transient. Such a program gives attempt. As it turned out, it required six years of effort before
equilibrium flight information as a special case, and it allowed a sustained flight was achieved. However, these tests provided the
confirmation of efficient stable flight for the chosen trim and static
motivation for, and the most demanding assessment of, this
margin setups of the 1989 and 1991 configurations.
research into the feasibility of mechanical flapping-wing flight.
It should be mentioned that the neutral point calculations are
Although ornithopter design details and certain support personnel
based on the assumption that the wing is fixed at its mean posi-
changed during the course of the flight tests, the basic methodology
tion. That is, although the wing's geometry (and corresponding
wake) is continuously varying throughout the flapping cycle, it remained the same: the aircraft was started while supported on a
was assumed that its mean configuration may be used for calculat- portable work stand, and then hand launched off a hill (or ridge)
ing the aerodynamic centres and downwash angles. In particular, into a wind. Note that the ornithopter had no means for stopping
for the wing's aerodynamic centre calculation, Combo Wing is the engine with the outer panels at a guaranteed positive (stable)
used to obtain the mean spanwise twist distributions. Lifting line dihedral angle, so it was necessary to land while flapping. Addi-
theory is then used to obtain the corresponding distributions of tionally, the landing gear had been removed in order to
local lift coefficients. This is done for several values of flapping reduce weight and drag. Since the wingtips extend below the fuse-
axis angles with respect to the flight direction so that the local lage and could contact the ground at full downstroke, it is easy to
lift-curve slopes, CLe, may be calculated. This information, imagine that such landings would be damaging. As it turned out,
along with the wing's geometry, allows the calculation of the this posed no practical problem. If the throttle was cut back just
aerodynamic centre from: before touchdown, the aircraft usually landed with no injury to its
structure or mechanism. This is not necessarily a recommended
procedure for an operational ornithopter, but it did turn out to be
x =^ acceptable for this research aircraft.
•^ac b
The first flight tests were on 3 October 1985, and they affirmed
certain of the crucial basic assumptions discussed earlier. First of
For the 1991 wing incorporating the Mark-8 panels, it was calcu- all, during these short, straight flights the aircraft was stable with
lated that xac./c~ = 0-315, where c~ is the mean aerodynamic chord no perceptible fuselage heaving or pitching. This assuaged
(9-86 in) and xac is measured aft from the most forward leading concerns regarding the workability of the wing's three-panel
edge position. balance as well as effects of its unsteady shed wake on the tail.

DeLaurier and Harris Aeronautical Journal October 1993

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 Second, it was clear that the wing's flapping was extending the Glide tests showed slopes less than 4:1, which is much poorer
glide. This affirmed the notion that its horizontal thrust could performance than one would calculate from standard aeroplane
produce the velocity required for lift, without the flapping, itself, drag polar methods. A windtunnel test of the ornithopter minus its
significantly reducing the mean lift. Because a weakness of the outer panels gave fairly high drag values (0-832 lb at 46 ft/s), but
front spars' bracing wires required the flapping frequency to be it was found that the major shortcoming was in the assumed aero-
kept below 4 Hz, it was assumed, then, that successful flight only dynamic efficiency of the outer panels' aerofoil. The windtunnel
required stronger wires capable of withstanding 5 Hz. experiments described in Ref. 11 showed that the attached flow
Several changes were made prior to the next flight tests. First, angle of attack range and leading edge efficiency had been greatly
thicker, stranded, wing bracing wires were installed. Second, be- overestimated. When the corrected values were used in the analy-
cause the outer panels appeared to have adequate aeroelastic sis, the predicted performance confirmed the flight test results.
twisting without any required articulation from the centre panel Therefore, a major redesign of the outer panels and support struts
(as discussed earlier), the pitching mechanism was removed and was undertaken, resulting in the Mark-6 wing described previously.
the centre panel was mounted rigidly to the highrise. Third, be- The design, construction, and windtunnel testing of the Mark-6
cause full up-elevator had been required to achieve longitudinal panels took all of 1988. Bench runs also revealed the need to
trim during the 1985 flights, the stabiliser was changed. That is, strengthen the drive mechanism and the centre panel. Therefore,
the original stabiliser had been built with a cambered, flat bottom flight tests with the new outer panels did not occur until 1989. The
aerofoil in accordance with usual free flight aeromodelling first series, in June, was basically a disaster. The new wing didn't
practice. This was replaced with a stabiliser of equal area incorpo- really have a chance to be evaluated because of poor launches (13
rating a symmetrical section (RAF 27), which thus increased the June) and inadequate elevator power (19 June). On the two occa-
decalage. sions when it did pull out, the engine immediately quit. In truth,
the expression "ground-breaking research" acquired additional
The next flights took place in June 1986, and a new problem meaning for this project. However, there was one glide test which
became evident. Soon after launch the ornithopter's flight path confirmed the aerodynamic improvement provided by the Mark-6
diverged and it spiralled into the ground. Although pilot inexperi- panels, in that the measured slope was in excess of 6:1.
ence might have been a factor, it was judged that the aircraft's
During repairs it was seen that the elevator control horn had
lateral stability was inadequate despite its high wing configura-
loosened. It was not known if this occurred because of the crash,
tion. At this time, the wing had no mean dihedral. That is, the
or was the cause of the ornithopter's failures to pull up from the
midstroke flapping angle was zero. Therefore, along with repair-
launch. However, it did serve to focus attention on the strength
ing the extensive crash damage, the support strut geometry was
and operation of the controls. Besides making the horn mounting
changed to give a midstroke flapping angle of 2°. and its control runs more robust, the torque outputs of the servo-
The repaired and modified ornithopter was next flown on 17 mechanisms were measured and found wanting in comparison
September. Several launches were made, but the aircraft had a with calculated hinge moments. Therefore, the original radio con-
perverse left turn tendency. This had nothing to do with spiral in- trol system (Kraft Series Seventy-Seven) was replaced with a
stability, for throttling back would allow recovery. It was clearly a Futaba FP. The ornithopter, minus the outer panels, was placed in
thrust asymmetry, and field corrections were attempted by in- a windtunnel and the controls were found to be capable of full ac-
creasing the rudder area and diverting the lateral cooling fan flow tuation at a speed of 52 ft/s, which is higher than the estimated
at the nose. Finally, straight, full throttle flight was achieved; but cruise speed of 45 ft/s. At this same time, the drag was remea-
it was evident that the ornithopter was not capable of sustaining sured and found to be 0-63 lb, which scales to 0-47 lb at 45 ft/s. It
such flight. was clear that the new, more streamlined, support struts gave a
At this time, the flapping wing analyses were being developed. lower overall drag despite the larger second stage external pulley.
The aerodynamic model had been derived, as described in Ref. 6, The launching technique was also given attention. A nonlinear
and coded in a program called Flapping. A separate program, three degree-of-freedom computer simulation was developed in
named Dynflex, calculated the aeroelastic response of the two spar order to study the launch transient, identifying optimum release
outer panel. The information from Dynflex was incorporated into techniques and post launch control strategies. Figure 11 shows a
Flapping to predict the outer panel's average lift, thrust, and representation of the mathematical model, and Fig. 12 shows a
propulsive efficiency. It was found that a significant performance typical trajectory from the program (named "Launch"). It was
improvement could be achieved by stiffening the outer panels' found that a successful launch could be obtained with a level
rear spars, and these changes were made before the next flight release at speeds of about 25 ft/s, low flapping frequencies, and
tests on 24 September. full up-elevator. Immediately after, the frequency should be
The ornithopter flew well, achieving the longest duration yet
(26 s). In fact, a significant portion of the flight was at nearly the
same altitude as the launch point. It only seemed to descend be-
cause the first stage drive belt kept stripping its teeth (evidence
that the drive mechanism would have to work harder as the outer
panels' efficiency increased). It was tempting, at this point, to
claim successful sustained flight. However, the authors had
established a criterion that success could only be claimed if the
ornithopter flew higher than the launch point and performed a
discretionary landing.
The major modification before the next flight tests was to
strengthen the first stage drive, as described in the drive mecha-
nism section. This gave no problems during the 1987 series of
flights, but it also became evident that the ornithopter was still not
capable of sustained flight. That is, although numerous tests were
made during 1987, only one flight had longer duration (by one
second) than that attained on 24 September 1986. The rear spar
stiffness was tinkered with in various ways, but it was clear that
the aircraft's aerodynamic efficiency had to be reassessed. Figure 11. Physical model for nonlinear flight simulation.

Aeronautical Journal October 1993 DeLaurier and Harris 283

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 * rile Edit
I LBUNCH8HP Bpl 1
Unoh Sp«id(ft/Mc)- 25 F*u1 Thrust(lt)" .»S tontriKM- .88
FlnftlSpMdttttocy 34.4 F™1 X«roCto)=-.25 rintimt/K' <2asi
hltM Pitch AneMteg)- 0 VtlflhtOb)- 6.S93999
F«»t
FlMlPltehAngllUtj)- 4.42 XSUfnCl))* .01 CO L«4tion(iiO= 14
+55
Hill «ngk(<l>9)> 30 ZsumOb)* .16

Pr»« Spact Bjr for Anothtr l«moh

-10 Fr«4tt7-K«is)
MV
-« !< [7s~\ ,
125 I I 230

Figure 12. Typical launch trajectory from flight simulation.

increased while the elevator angle is brought back to zero. Note Figure 13. Omithopter in flight, 4 September 1991.
that the simulation's predictions are conservative in that ridge lift The durations were long enough to adequately evaluate the or-
and head winds are not modelled. nithopter's performance, stability and control. The flight videos
Additionally, a plywood mockup of the omithopter had been showed speeds of approximately 50 ft/sec at flapping frequencies
constructed and launch tests were performed in order to gain ex- from 3-0 to 3-8 Hz (averaging 3-3 Hz). According to the pilot,
perience with attaining the strong level throws dictated by the these evidently exceeded the minimum conditions for level flight.
simulation. This proved to be a valuable training tool as well as a The aircraft was flown "hot", staying at altitudes below 300 ft for
means for obtaining launch speed data. documentation by the ground based cameras. Therefore, the previ-
These efforts paid off on 24 October with two excellent launch- ously predicted minimum flight conditions of 45 ft/s at 3 Hz are
es. For the first time, the omithopter rose higher than the launch probably valid.
altitude and appeared to be sustaining. However, the engine Turns were attempted, for the first time, and these were execut-
prematurely failed in both cases, so the landings did not meet the ed smoothly and easily with no unstable tendencies. The rudder
criterion of being discretionary. Also, it was unclear if the or- worked with the mean dihedral, as designed, to give the required
nithopter flew free of ridge lift, so unequivocal success could not bank angles. A slight amount of asymmetrical thrust was readily
compensated for with rudder trim. The longitudinal stability was
be claimed. It appeared then that a solution to the engine failure
likewise very good, confirming the adequacy of the chosen static
problem (which had plagued this project off-and-on from 1985)
margin. The elevator trim for level flight was found to be essen-
was now the top priority. tially zero, as calculated. In all, the omithopter appeared to be
This was dealt with by increasing the first stage drive ratio and inherently stable.
improving the fuel feed. Meanwhile, a new windtunnel rig had When in flight, there was no perceptible fuselage heaving or
been developed*10' which allowed a flapping Mark-6 outer panel pitching in response to the flapping. Undoubtedly, there is some
to be tested for average lift and thrust. The results showed that the unbalanced force imposed on the fuselage; but the three-panel
omithopter would only have marginally sustained flight. At the wing apparently worked at reducing this. In fact it was striking,
same time the development of the Mark-8 wing showed such looking at the videos, how smoothly the omithopter flew.
promise that it was decided to make no further flights with the Finally, both landings were very satisfactory, with no damage
Mark-6 wing. Instead, the omithopter was preserved for tests with for the first one, and only slight damage to the subrudder for the
the new wing. second. If required, that could have been a field repair.
The Mark-8 wing was ready by June 1991, having undergone
windtunnel tests which confirmed its performance potential*10).
Also, both bench runs and run ups at flight speeds on a truck
mounted rig were conducted to proof test the wing structure, drive CONCLUDING REMARKS
mechanism and fuel feed system. The omithopter was then ready
for flight testing, and the opportunity for this occurred on The flights of 4 September graphically demonstrated the feasibili-
4 September. ty of this particular approach to engine powered flapping wing
The location was near Newton-Robinson in rural Ontario (about flight. It is readily acknowledged that there may be other solutions
30 miles north of Toronto). The winds were from the north at ap- for attaining this; but the omithopter design resulting from this
proximate speeds of 5-15 mph, and the aircraft was launched from work has certain technologically attractive features. First of all,
the top of a northward facing ridge. Both launches were excellent, the wing design operates very efficiently because the Shearflex
smooth and level; and in both cases the omithopter climbed and feature allows it to incorporate a thick, modern, aerofoil with a
performed sustained flight (shown in Fig. 13) until it was decided wide attached-flow angle of attack range and high leading-edge
to land. The pilot kept the elevator angle initially at zero, planning suction efficiency. Next, the three-panel feature substantially re-
only to increase it as required. As it turned out, the initial up- duces two of the main design problems for omithopters, namely
elevator suggested by the Launch simulation was not needed, the unbalanced lift imposed on the fuselage and the difference
between downstroke and upstroke power. Further, a drive mecha-
probably because of ridge-lift effects.
nism was found that was lightweight, reliable, and capable of
The flight duration was limited only by the amount of on board
delivering simple harmonic motion to the wing. Note the
fuel, which is a maximum of 4 oz. Since a fair amount of run up important fact that the drive reduction provided such an apparent
was performed before the first launch, it was decided to land back inertia at the engine that no flywheel had to be added. That
before the tank emptied. If that occurs, the wing might lock at an is, the engine was able to run as comfortably through its throttle
unstable flapping angle. Therefore, the first flight was 1 min 46 s. range as if it were rotating a propeller.
The second flight was longer, at 2 min 46 s.

284 DeLaurier and Harris Aeronautical Journal October 1993

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 The notion that the ornithopter's stability and control is deter- project, the pilots have been Donald Uffen, Sunjoo Advani, and
mined by its mean configuration was confirmed. That is, the usual Eric Edwards; and the launchers have been Darius Mavalwalla,
geometric and mass balance criteria which stabilise a fixed wing Christopher Lewis, and David Loewen. Continuous help, both in
aeroplane were shown to be also appropriate when applied to the the lab and for the tests, has been provided by research engineers
ornithopter's time averaged geometry. It is worth noting at this William McKinney and Eric Edwards, along with Sing Wong,
point that flapping wing aerodynamics could have had several un- Henry Kwok, and David Loewen.
pleasant surprises to nullify that notion, such as strong unsteady Also, several individuals have stepped forward at crucial times
wake effects on the stabiliser and significant mean lift loss with to provide their expertise. These include Gus Rinnella with his
flapping frequency. In the event, these problems did not occur. materials knowledge, Marcus Basien with his wing construction
The control sizes were proportioned somewhat large as dictated ideas, James Winfield with his skill at both still and video photog-
by flight test experience. The 1989 flights had shown an apparent raphy, Brian Alsop with his key idea for fuel-slosh damping, and
requirement for a strong low speed nose-up pitching moment William Unger who made the structures lab available for load
during launch; and this was provided for with a generously sized testing the vertical struts.
elevator. Even though the 1991 flights did not utilise this full ca- Special thanks go to Michael Selig who, because of his interest
pability, it could be needed for launches into weaker winds. The in the project, designed the excellent SI020 aerofoil which
rudder was enlarged during the 17 September 1986 flights to cor- performed so well on 4 September 1991.
rect a strong turning tendency. Even though that amount of rudder Finally, acknowledgement should be given to those who always
power was never again required, the enlargement was retained be- seemed available to help in any way needed, such as carrying
cause of the possibility of asymmetrical thrust from the two outer equipment to flight test sites, handling photography and video-
panels. The 1991 flights, however, appeared to confirm the pre- graphy, obtaining food and materials, etc. These include Chris
dicted "design robust" feature of the Mark-8 panels in that even Hayball, James Lowe, Matthew Malone, Darin Graham, Susan
when hand constructed by two different workers at two different Haigh, Justin Amann, David Ahier, Karl Stoll, and Roland
times, they performed very nearly in concert. Lorenz. This accomplishment is shared by all listed.
It must be noted that although the three-panel wing works to-
wards providing a lift balance, it will generally never achieve this
perfectly. Therefore, the fuselage will always experience some os-
REFERENCES
cillation at the flapping frequency. Furthermore, the three-panel
design requires support struts, with their corresponding weight
and drag. Additional weight, as well as mechanical efficiency loss, 1. SCHOENKY, P. Omithopters, 1951-52 Model Aeronautic Year Book,
ZAIC, F. (ed) Model Aeronautic Publications, New York, 1952, pp 188-90.
is likewise a consequence of an ornithopter's drive reduction
2. GIBBS-SMITH, C. H. The Eighteen Seventies and Eighties, A History
system compared with a simple shaft mounted propeller. of Flying, B.T. Batsford, London, 1953, p 172.
Serving to offset these losses, however, is the flapping wing's 3. HERZOG, K. Demonstration des Tierfluges an Hand von Vogelmod-
large actuator area. The high efficiency of a large diameter, slowly ellen und Schwingenflugapparaten, Anatomie und Flugbiologie der
rotating propeller is well known. A flapping wing offers the same Vogel, Gustav Fischer Verlag, Stuttgart, 1968, pp 136-169.
aerodynamic advantage, imparting small velocity increments to a 4. DWIGGINS, D. Flapping wings... Spence's secret life, Sport
large amount of air. Since this is also the requirement for a low Av, August 1977, pp 19-22.
aeroacoustic signature, omithopters therefore hold the promise of 5. BROOKS, A.N., MACCREADY, P.B., LISSAMAN, P.B.S. and MORGAN,
W.R. Development of a wing-flapping flying replica of the largest
being the quietest powered aircraft.
Pterosaur, AIAA Paper 85-1446, 1985.
The analytical foundation and structural design resulting from 6. DELAURIER, J.D. An aerodynamic model for flapping-wing flight,
this work should be applicable to the design and construction of a Aeronaut J, April 1993, 97, (964), pp 125-130.
full scale aircraft. That is, this particular solution to mechanical 7. LIPPISCH, A.M. Man Powered Flight in 1929, Aeronaut J, July 1960,
flapping wing flight appears to be capable of being scaled up to a 64, pp 395-398.
human carrying engine powered ornithopter. There would be con- 8. BERANEK, L. Der Schwingenflug, eine Technische Realitat, Deutsche
siderable development involving takeoff techniques and drive Flugtechnik, 1959, Bd 1.
mechanism design, but a major result from this research is that 9. Man-powered aircraft a-coming, Sci Mech, April I960, pp 106-108.
such an achievement is now feasible. 10 DELAURIER, J.D. The development of an efficient ornithopter wing,
Aeronaut J, May 1993, 97, (965), pp 153-162.
Finally, although this work has concentrated on an ornithopter DELAURIER, J.D. An experimental study of low-speed single-surface
11
designed for cruising flight, it is interesting to speculate that a airfoils with faired leading edges: In Low Reynolds Number Aerody-
future advanced flapping wing aircraft might be capable of both namics; Proceedings of the Conference, Notre Dame, Indiana,
efficient Vstol and high speed subsonic flight. This would require Springer-Verlag, Berlin, 1989, pp 161-173.
a major research and development effort involving mechanisms 12 FAIRGRIEVE, J.D. and DELAURIER, J.D. Propulsive performance of
and wing articulation drives; but note that the basic requirement two-dimensional thin airfoils undergoing large-amplitude pitch and
for a large actuator area is inherent in the ornithopter's design. plunge oscillations, UTIAS Technical Note No 226, July 1982.
13 Royal Aeronautical Society, Data Sheet Aircraft 08. 01. 04.

APPENDIX A: CHRONOLOGY OF FLIGHT TESTS

ACKNOWLEDGEMENTS
3 October 1985
This project was primarily funded by the authors, with support for Location: Morningside Hill, Scarborough, Ontario
certain incidental expenses provided by the Natural Sciences and Pilot: Donald Uffen
Launcher: James DeLaurier
Engineering Research Council. Funding was also obtained from
Flight 1: powered, 12 s. Flight 2: powered, 13 s. Flight 3: glide, 8 s
Exhibitions Canada, during 1991, which allowed the completion
of the ornithopter in time for the 4 September flights. These were
9 June 1986
filmed by the Canadian National Film Board for incorporation
Location: Morningside Hill, Scarborough, Ontario
into an IMAX movie, Momentum, shown at Expo 92. Pilot: Donald Uffen
Furthermore, this project has benefited greatly from the enthusi- Launcher: James DeLaurier
astic help of several volunteers. During the course of this long Flight l: powered, 5 s (crashed*)

Aeronautical Journal October 1993 DeLaurier and Harris 285

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press

 14 June 1986 APPENDIX B: COMPONENT BREAKDOWN OF THE
Location: Morningside Hill, Scarborough, Ontario
4 SEPTEMBER 1991 ORNITHOPTER
Pilot: Donald Uffen
Launcher: James DeLaurier Figure 14 shows a cutaway perspective drawing of the 4 September 1991
Flight l: glide, 12 s. Flight 2: glide, 8 s. Flight 3: powered, 13 s (crashed). ornithopter. Compared with the 3 October 1985 aircraft which weighed
6-5 lb, this version weighs 8-7 lb. As discussed in this article, these
17 September 1986
increases were due to modifications for improving performance and
Location: Morningside Hill, Scarborough, Ontario
reliability. The specific component breakdown is given below:
Pilot: Sunjoo Advani
Launcher: James DeLaurier
Flight 1: powered, 9 s. Flight 2: powered, 8 s. Flight 3: powered, 9 s.
Flight 4: powered, 12 s. Flight 5: powered, 12 s. Flight 6: powered, 9 s Component Primary Materials Mass
(crashed). (9)
24 September 1986 Outer panels Kevlar/carbon/foam/balsa/ 590
Location: Morningside Hill, Scarborough, Ontario (including plywood/plastic/aluminum/
Pilot: Sunjoo Advani hinge fittings) steel
Launcher: James DeLaurier
Flight l: powered, 21 s. Flight 2: powered, 26 s. Centre panel Kevlar/carbon/aluminum/ 184
balsa/Mylar
4 June 1987
Location: Mono, Ontario
Support struts plywood/foam/fibreglass/ 480
Pilot: Sunjoo Advani
(with pins) steel/brass
Launcher: James DeLaurier
Flight l: powered, 16 s (crashed).
Vertical links carbon/plywood/steel/brass 116
13 June 1987 (with upper pins)
Location: Newton-Robinson, Ontario
Pilot: Sunjoo Advani Thorax (including steel/aluminum/plywood/balsa/ 2319
Launcher: James DeLaurier drive mechanism fibreglass
Flight 1: powered, 12 s. and radio control,
but no fuel)
15 June 1987
Location: Newton-Robinson, Ontario Rear fuselage balsa/plywood/Kevlar/Mylar 190
Pilot: Sunjoo Advani
Launcher: James DeLaurier Stabiliser & fittings balsa/plywood/Mylar/plastic 83
Flight 1: powered, 27 s. Flight 2: powered, 6 s. Flight 3: powered, 9 s.
Flight 4: powered, 13 s. Flight 5: powered, 15 s. Flight 6: powered, 12 s.
21 September 1987 Total: 3962 g
Location: Mono, Ontario (8-74 lb)
Pilot: Eric Edwards
Launcher: James DeLaurier The wing area is 7-69 ft2, so the wing loading is 1-14 lb/ft2.
Flight 1: powered, 3 s. Flight 2: powered, 13 s. Flight 3: powered, ~ 8 s.
22 September 1987
Location: Newton-Robinson, Ontario Editors note: Two previous papers by Professor DeLaurier concerning his
Pilot: Eric Edwards work on ornithopters were published in the April (96 (964)) and May (96
Launcher: James DeLaurier (965)) issues of The Aeronautical Journal.
Flight l: glide, 6 s. Flight 2: glide, 6 s. Flight 3: glide, 8 s
Flight 4: powered, 8 s. Flight 5: powered, 10 s. Flight 6: powered, 1 s.
Flight 7: powered, 8 s. Flight 8: powered, 21 s. Flight 9: powered, 18 s.
13 June 1989
Location: Newton-Robinson, Ontario
Pilot: Eric Edwards
Launcher: James DeLaurier
Flight l: glide, 3 s. Flight 2: powered, 2 s. Flight 3: powered, 9 s.
Flight 4: powered, 2 s (crashed).
19 June 1989
Location: Mono, Ontario
Pilot: Eric Edwards
Launcher: Darius Mavalwalla
Flight l: glide, 13 s. Flight 2: powered, 8 s. Flight 3: powered, 3 s (crashed).
24 October 1989
Location: Mono, Ontario
Pilot: Eric Edwards
Launcher: Christopher Lewis
Flight l: powered, 26 s. Flight 2: powered, 23 s (crashed).
4 September 1991
Location: Newton-Robinson, Ontario
Pilot: Eric Edwards
Launcher: David Loewen
Flight 1: powered, 1 min: 46 s. Flight 2: powered, 2 min: 46 s.

Total powered flights = 38

Total glides = 8
* "crashed" means that the damage was too extensive for a field repair. Figure 14. Sectional drawing of 4 September 1991 ornithopter.

DeLaurier and Harris Aeronautical Journal October 1993

https://doi.org/10.1017/S0001924000026452 Published online by Cambridge University Press