The idea started in January 1999 while flying from the Gariep Dam situated on
the Orange River in South Africa at 30° 37' 22.55" S - 25° 30' 22.86"
Soaring conditions here are mostly excellent, but the out-landing
possibilities because of the rugged terrain, are extremely hazardous if not at
Under these conditions a self-launcher would seem to be highly desirable,
however, the prohibitive cost of a self-launcher led me to consider
building one of my own.
The pop-up concept, because of the adverse effect it has on the glide angle
at engine start-up, and various other uncomfortable factors, encouraged me to
reject this principle.
My thoughts were to build a pusher-type craft with folding propeller at the
rear, joined to the engine situated behind the wing by means of a drive shaft
similar to that employed by a BD5, Mini-Moke or Graal.
My decision to go ahead with the project was taken some time in mid
I decided to name the project “Selene” after the Polynesian goddess of
the moon, having been influenced by the fact that the general consensus of
opinion at the time was that my idea was Utter Lunacy.
The wings would be from ASW 20.
The power plant would be a
single cylinder, two-stroke, water-cooled engine of 30 kW.
would extend from just behind the ASW 20 cockpit to the leading edge of the fin,
and would mount wings under-cart and house the drive train.
would have composite removable covers.
To give the propeller ground
clearance, the fin would need to be both dorsal and ventral.
to whether this system could be adapted to an existing 15m flapped glider
appeared to be positive.
Order the engine.
Build a hanger to house the project.
Design and build
a prototype drive-train test rig.
Acquire an ASW 20.
These steps took place over the following few years, but not necessarily
in the order stated above.
By the beginning of 2003 the hanger was ready and
the design was on its way.
An ASW 20 had not yet been found but the design
and building of the propulsion system had begun.
The drive-train test-rig was built using rectangular tubing and fitted to a
small trailer for ease of maneuverability, which also made it possible to
perform dynamic tests on the system.
The first drive-train assembly produced excessive vibrations, which were
severe enough to adversely affect the carburetor function.
The reason for the
poor performance of the original engine mounting system was simple but not
easily identified, and turned out to be that the propeller was not directly
connected to the engine as it would be on most other applications.
that the flywheel effect of the propeller no longer stabilized the
Various mounting systems were tried and eventually a 3 mount system
and suitable flywheel made the engine run satisfactorily.
The drive-train at
this stage consisted of a 3 piece shaft with a maximum propeller speed of 2350
rpm. And no torsional vibration issues were evident throughout the operating rev
At this point it seemed that the development would be less difficult
The engine was temporally covered as it would be in the aircraft and the
cooling system evaluated.
This led to the fitment of a lighter aluminum heat
exchanger having greater cooling capacity and a more powerful fan.
dynamic testing was performed by towing the rig up and down the runway at
varying power outputs and simulating air starts.
These tests revealed nothing
undesirable. However, the theory that forward motion would soften the violent
propeller deployment when the engine is started with the propeller folded was
Starting the engine when hot caused the propeller to deploy with more
violence than would be acceptable.
This was because the engine would fire on
the first compression stroke and the propeller would still be partially folded
at this time.
Keeping the magneto’s switched off until the starter motor had
fully deployed the propeller usually resulted in a flooded engine.
keeping the throttle open until the starter motor had deployed the propeller did
not work consistently.
A fully closed throttle was needed to start hot or
cold and the engine would not idle smoothly at this throttle
Propeller speed was too slow and needed to be increased.
During mid 2004 an ASW 20, ZS-GMU had a bad out-landing which
resulted in some damage to the fuselage and I subsequently had the opportunity
of purchasing the aircraft.
Now having an ASW 20 to work with, allowed the mass and balance
considerations to be further investigated as the drive-train masses were
beginning to be known.
By the end of 2004 the mass and balance issues were
fully understood and the entire design was well on its way.
The cockpit section which will attach to the space-frame was repaired and the
attachment ring temporally fitted to it.
The minor damage to the wings were
repaired, wing-tip wheels fitted and provision was made for fitting the fuel
tanks into the wings.
At this time it was decided to buy the machinery needed to perform all the
machining as the cost and time consumption of outsourcing the work was
It should have been understood from the outset that a project of
this nature would be protracted and would be best accomplished in-house.
In 2005 the material for the space-frame was imported and the building of it
The wings and mixing box were fitted and control rods made.
tank and prototype wiring were installed.
The engine with new reduction ratio
and revised drive shafts, bearings and folding propeller hub were fitted.
Early in the year the revised drive-train was run in the space-frame for the
first time with some surprising and frustrating results.
torsional vibration problems which had come about due to the ratio change and
space-frame mounting. This took many months to resolve.
dampers were experimented with, flywheels altered and coupling types
By mid 2006 the drive-train had 4 shafts and was performing well
again with a minor transient vibration at 3500 – 3600 engine rpm.
The remaining problems were still:
Engine start-up reliability.
Inability to idle at start throttle
The violent propeller fling-out when the engine is started hot and
with the propeller folded as will be the case when the glider is flying.
All these problems were eventually solved by replacing the carburetor with
A locally developed engine management system and a throttle
body from a motorcycle were used to confirm that the fuel injection system would
solve all the remaining problems, but the motorcycle throttle body would not fit
into the final aircraft constraints.
Three throttle bodies were fabricated
before optimization of the system was achieved.
At the end of 2006 cooling trials were again successfully conducted.
The space-frame and drive-train combination had been successfully run in
constraint conditions, which closely approximate flying, taxiing and
The system had accumulated a total of 36 hours of test running, 11 in the
original test rig and the balance in the space-frame.
Some test results as at December 2006:
Static thrust at 5900 engine rpm is 78kg and it is reasonable to expect 6100
rpm and 80kg+ at take-off speed.
An estimated 7 minute launch simulation:
30 second warm-up.
1 min 30
sec full power.
4 min climb power.
1 min idle.
This used 0,75 litre of
Possible cruise power settings, 3200 engine rpm at 3 litre per hour and 3800
rpm at 5 litre per hour.
In January work on the composite tail section started.
Michael Charl, an aircraft constructor who had learned his trade working for
Shremp Hirth in Germany, ably consulted on the design aspects and performed all
the many tasks required to produce aircraft quality components.
It was decided to build a new horizontal stabilizer and elevator.
the ASW 20 unit weighed 9,2kg and a saving of 3kg could be realistically
Reduced weight at the rear of the aircraft is a good thing in this
The elevator will remain in the same geometric position as it is an
ASW 20 and will retain the same profile, but will be extended in span by
The original unit was used in the construction of the mould.
Fin-Rudder and Rear Fuselage Cone:
The fin-rudder area and profile was used and the aspect ratio
Polystyrene foam was hot-wire cut utilizing CNC cutting
The plugs were made from these and subsequently the moulds from
Calculations for the composite lay-up strength were performed.
The “mapping”, programming of the engine control unit was finalized in the
winter of 2007, as cold weather was needed for this exercise.
under-cart which had been designed in 2004 was fabricated and fitted to the
The cockpit was connected to the space-frame at the estimated position and
the control system fitted.
A dummy fin and rear fuselage cone was made using the mould for the first
time, which was then fitted to the space-frame.
The elevator control system was fitted.
The engine cowl system was designed and a prototype made and fitted complete
with its actuator system
The need for engine instruments and the lack of space to fit them led to the
application of a glass cockpit instrument system.
The original fuel supply system which consisted of two wing mounted aluminum
tanks of 10 liter each, feeding a newly constructed fuselage mounted header tank
of 1 litre was pressure tested and fitted.
However, there were just too many
pipes which necessitated the design, manufacture and fitment of a new fuselage
mounted fuel tank.
At this stage all components were evaluated for potential
mass reduction and where feasible lightening was implemented.
The next goal was to perform ground run tests with all the systems assembled
as an aircraft, using the dummy fin and elevator, all controls, ancillaries,
fuel lines and wiring installed as when the aircraft is ready to fly.
The drive-train ran well and there were no damaging vibrations in any of the
The 3500 – 3600 rpm range still had a vibration which made this
rev range unsuitable for continuous operation.
It was decided to have the cooling fan on when the engine was running to
avoid the complexity of a thermostatic switch or the unreliability of a manual
This resulted in a net battery discharge whenever the engine was
The fuel pump originally selected was of a higher flow rating than
required and a more suitable one was acquired which gave the battery a small
positive charge when engine revs were above 4500 rpm.
The aircraft was weighed with estimated masses added where components yet to
be made would be situated.
The results indicated that the mass and balance
were within the parameters envisaged and would not pose any problems.
The engine intake noise was significantly reduced by utilizing long, large
diameter ducting joined to a Donaldson air filter
At this stage it was decided to change the method of connecting the
space-frame to the cockpit.
Originally a 4130 steel tube was to be glassed to
the cockpit structure.
This method was discarded in favor of making a
composite flange to bolt the two assemblies together.
The cockpit was
disconnected and the composite flange made after which the front section of the
space-frame was revised to suit the new connecting method.
Next a flying tail-cone, fin and rudder were manufactured using aircraft
The rear fuselage cone, fin and rudder assembly complete
with tail wheel, control actuators and rudder mass balancing was fitted to the
Pilot-static and total energy probe was fitted to the nose of the
The horizontal stabilizer and elevator were made, and when fitted
enabled the aircraft to undergo taxi testing.
Initial taxi testing indicated the main wheel not being far enough forward of
the center of gravity, causing the aircraft to pitch nose down too easily.
a result the main wheel was moved 160mm forward which optimized ground
Ground vibration tests were performed on the aircraft in January 2009, at the
CSIR (Council for Scientific and Industrial Research) in Pretoria.
Next, the covers were made and fitted.
Originally there would have been
blisters in the covers to accommodate the carburetor on the right side and the
exhaust outlet on the left.
As it subsequently turned out, the injector
throttle body was small enough to enable the engine to be moved over by 50mm
eliminating the need for blisters.
This repositioning of the engine and
realignment of the drive-shaft had a very positive overall effect, inasmuch as
the entire rev range became continuously usable with no perceivable torsional
This would be due to the added damping effect caused by the
The aircraft was disassembled, components inspected and reassembled using new
fasteners with a view to airworthiness.
The wiring harness was remade,
reconnected and tested.
The aircraft was weighed and removable ballast fitted
to bring the flying center of gravity to the forward third of its allowable
limits for the pilot concerned.
Control surface deflections were measured and
Stabilizer to wing incidence was measured and adjusted.
tests were performed with all covers fitted. During these tests the propeller
leading edges were eroded by runway debris, which had not occurred during the
previous taxiing tests which had been done without covers fitted.
had most likely lodged in the fan and radiator during the previous tests.
wheel guard was developed and fitted to prevent this and propeller leading
edge protection was applied.
Finally, an annual inspection procedure was performed by two competent
persons and the appropriate forms completed.
First Test Flight:
The first test flight took place on 23rd December 2009 at 6am
CofG: 30% behind forward
Ambient Temp.: 20° C
Maximum height: 6500ft
speed: ASI malfunction due to poor static port
Ground roll to lift-off: ±500m
Maximum decent rate: 600ft/min
duration: 26 min
Take-off was achieved when the flap lever was moved to the second positive
No elevator input needed.
Climb-out was kept quite flat because of
the lack of airspeed indication, but felt stable and safe.
At 6500ft the
roll, pitch and yaw were gently checked and found to be good.
settings with decent power (engine idle) and air-brakes were checked and the
landing configuration decided on.
An approach to stall in the landing configuration was tried, to get a feel of
the aircraft on final approach.
Next, a low power level flight in neutral flap was tried and felt stable and
safe, but the engine lost power at this low throttle setting and only resumed
when the throttle was advanced.
The landing happened with the tail wheel
touching down slightly before the main.
There was no bounce and the
roll-to-stop was ±300m.
In general the handling of the aircraft was very
Unfortunately the flight data recording facility of the Enigma
instrument was not correctly set up resulting in the flight data not being
Second Test Flight:
The second flight took place on the 31st December 2009 also at 6am.
For this flight the static port was disconnected and cockpit static was
The propeller pitch was increased by half a degree which brought static
rpm down from 5900rpm to 5600rpm.
Take-off was good and climb-out was done at
At 6500ft the engine was shut down, cowl closed and
This process was reversed without any
Again at a low power setting the engine lost power and recovered on
application of power.
This seemed to be more pronounced than on the first
Landing was made as before with indicated airspeed of 90kph over the
On this flight, again the flight data recorder did not work due to
The aircraft was then de-rigged and the covers removed.
On inspection all
was ok except that the fuel line to the injector appeared to be pressing against
Redesign and remake of the injector fitting solved this
A prolonged ground run with covers fitted, which required special cooling of
the cover in the exhaust outlet area, indicated that the fuel in the feed pipe
to the injector was becoming overheated.
This was suspected as the cause of
the engine misbehavior.
Changing the fuel-line plumbing so that fuel is
returned from the injector back to the tank, instead of being dead-ended at the
injector, stopped the fuel heating. However, it did not solve the erratic engine
This was eventually traced to one of the wire terminals connecting
the pick-up to the engine control unit.
The terminal connection became
intermittent due to engine vibrations and the method of connection.
method of connecting wiring to the engine was revised at all terminals to
prevent a reoccurrence of this problem.