| 1. Preamble.
There are a number of criteria for propeller design for the
big USRA Unlimited pylon racers. Perhaps the most important
relates to avoiding the SHITDS. This is the acronym for "Smoking
Hole In The Desert Syndrome". Such holes appear regularly
in the desert around LA, and their appearance is sometimes
hard to explain. However, propeller failure can produce these
unfortunate geographical features, and in less time than it
takes to say "SHITDS
Next on the list is performance. At 9000 RPM and 240 MPH,
a prop needs a considerable blade angle: Not as much as the
real "Rare Bear" or the Tu 95, but still plenty
enough to produce a reasonable degree of profile efficiency.
However, with a tip speed above Mach .9, the aerodynamics
starts to get interesting, and it is here that the challenge
lies for the prop designer.
Now to examine these points in detail.
2. Propeller failure modes.
Regrettably there are a number of good ways of destroying
a perfectly good propeller. One is the manner in which it
is mounted to the engine. A second relates to the high instantaneous
torque produced when the engine fires. And finally, at low
speeds and high RPM, as at takeoff, stall flutter may occur.
2.1. Prop retention.
There are 2 common ways of retaining the propeller. Some
engines have a single axial shaft with a single nut retaining
the prop. Others have multiple bolt fixings, which vary
from 4 to 6 bolts with an axial stud for prop centering.
Both are effective, though I prefer the single stud method.
But why? The potential for disaster if the prop comes off
is very high, so the following should be considered with
care.
Engine torque has to be transmitted by the prop driver to
the propeller, which absorbs the torque. ie engine produces
torque, prop absorbs torque. Now it is nice if these are
equal: but with incorrect mounting they may not be, and
that lost torque has to go somewhere.
The question thus becomes: what is the safest way to transfer
the engine torque into the prop? Well, we have a prop driver,
and an assortment of bolts. I guess if we tighten them up
enough on to the the prop hub, we have this licked. Wrong,
guess again.
If there is low friction or indexing of the prop driver
with the prop hub, then there is every probability of slippage
between the adjacent faces of the driver and the hub. Due
to the high instantaneous torque of firing, this can and
has happened. The resultant oscillations produce heat which
can melt the epoxy within moments, and then, SHITDS. The
prop comes loose, the high tensile bolts break in bending,
and its all over red-rover.
The message here is that the bolts are not intended to transfer
engine torque to the prop. On full scale aircraft, the bolts
do not even touch the prop, but have clearance away from
the prop. In the case of wooden props, it is friction drive.
For forged aluminium props, there are interference fit studs
mounted on the prop driver. We need to pay attention to
these guys, they learn this stuff by killing pilots.
The reason I like a single axial bolt fixing is that in
this eventuality we are not snapping bolts. Those snapping
bolts don't come off evenly; the resultant side forces being
free to rip out the engine as well. Some have survived this
engine-loss catastrophe, but its not something you would
want to test for yourself. Better to be old than bold. With
the single-bolt fixing, engine loss is much less likely.
But hey! we have enough problems already without this, so
what is the solution?
Well, we gotta stop the torque-induced slippage. For 60
years at least, model engines have had knurled prop drivers.
The knurls are there to provide friction, and if they are
sharp as well, then they also provide mechanical indexing,
or maybe registration is a better word. The knurls must
bite into the prop hub: they must and are required to damage
the surface of the hub.
There are engine outfits out there that sell engines with
smooth, polished prop drivers, and they get away with it,
most of the time. But your big Herbie or whatever won't
take long to throw the prop off, believe me. So the rule
is: transfer torque with friction/indexing, and use the
bolts to crush the prop onto the knurls.
End of story. Here endeth sermon 2.1.
2.2 Torque failures
There are three directions that interest us with propellers.
These are axial, radial and tangential. The tangential direction
corresponds to forces acting in the torque plane, which
lies in the propeller disc.
The problem we have is that the instantaneous torque can
be very high, due to engine firing for only a short period
in the rotation of 360 degrees. If the engine detonates
or has advanced ignition timing, then the torques can be
very high indeed, to the point where the propeller experiences
a failure mode seldom discussed in the literature, but fairly
common on models.
Most people are aware of the high tension forces which exist
due to prop rotation. However, tension failures are quite
rare, and almost non-existent in composite propellers, whether
of wood or carbon/glass/kevlar construction. It can even
be argued that the tension force strengthens these materials,
and certainly stiffens the prop against torque and thrust
loads.
Most failures occur in nylon/plastic props, simply because
these materials are just not strong enough in tension for
the high rotation speeds we like to use. Aluminium props,
if properly forged are fine, but have a lifetime determined
by the fatigue properties of the metal. This is also true
of wood props. I have an F3D prop which has 200 runs on
it. But it is retired, because the epoxy in the hub is cracking
up: the carbon fibre in the blade is just fine.
So where is this headed? Well, torque plane failures do
occur. What is interesting is where and how they occur.
They appear as compression failures on the front blade surface
near the hub. The appearance is that of wrinkles running
fore and aft of the airfoil section, maybe 20% of the radius
out from the hub.
This is a bit weird. Thrust failures would appear on the
other (rear or face) surface, but they don't. Tension failures
might be expected to from on both surfaces, but they don't.
So what is it about torque failures?
Well, near the hub we usually have a high blade angle. This
effectively makes the prop appear "thin" in the
torque plane, thereby reducing resistance to the prop trying
to fold back on itself in the torque plane. This is consistent
with both the location of the failure and the comression
nature of the failure.
So why worry? Just make that area stronger. Well, that is
not so easy. We are not talking about carbon fibres in tension
here. Rather, we have resin/fibres in compression: not nearly
so nice.
I wish I had a solution to this problem which would work
all the time, but I don't. So my approach is to add thickness
as a projection in the torque plane; avoid high torques
due to detonation; and in the case of big fat radials, keep
the blade angle as low as reasonable in front of the cowl.
This latter turns out to be an OK thing to do in the aerodynamic
sense as well, so that at least is win-win.
So what to do about it? Well, inspect the prop after every
run and look for compression failure marks on the front
near the hub. You'll recognise them by the way a chill runs
down your spine when your subconscious brain recognises
them first!
2.3 Stall flutter. Ok, now we get a chance to throw the
tips off! Much more fun, and a bit less likely to give us
the SHITDS. Of course, you do have a safety wire from the
engine to the airframe, don't you? They don't glide too
well with the engine out (and I do mean OUT).
There is a condition referred to as "propeller stall".
It is a complex subject, and I won't go into it here, as
it seems to provoke disputes with the wing expert guys.
But tis a nasty you need to avoid.
Consider this condition. Full revs and the airplane stationary
on the ground. As a first approximation, you can imagine
that this puts the tip airfoil at a high angle of attack.
High enough in fact to stall. The problem is, that this
stall can be transient and be followed by another stall.
ie, lift, lose lift, lift again. My friends, the devil invented
this oscillation, and he has many victims who found their
way into hell due to flutter.
Now to keep this simple, stall flutter does occur, however
you choose to describe the process. The trick is to recognise
and know you are on the wide and slippery slope.
Many years ago, I made an F3D prop (28000 RPM, 160 MPH)
which worked fine at home here in Oz. But in Florida, at
the World Champs, it decided to misbehave on the flight
line. Folks, it made so much noise that the other contestants
could not hear to tune their own engines. Wish I could claim
credit for such a brilliant tactic, but it was accidental.
My guy came 7th, so that was OK.
Well, it was stall flutter making that noise. Model engine
noise is typically composed of a harmonic at the frequency
of the engine revs/second, with a set of equally spaced
harmonics above that. The strength of each harmonic seems
to be a bit unpredictable, but they do weaken a lot above
the 7th or so. Mostly.
If you have stall flutter, this all changes. A couple of
the harmonics may grow massively in intensity, so that the
beautiful unsymphonic engine roar becomes a howl, with beats
forming due to the strong harmonics. The little F3d props
(7.5" diameter) are tough enough to take this, but
on your Unlimiteds you can kiss your tips goodbye. This
may also diminish performance, but I'll leave you to imagine
where and how!
The solution? If stall flutter is present, it immediately
diminishes and ceases as the aircraft gains speed. Best
on take-off to ease the throttle open slowly as the airplane
runs down the strip: then the condition that leads to stall
flutter (high RPM, low speed) does not occur. Be kind to
your props!
Now we have finished with prop failure modes. Now doubt
there are inventive types out there who can create others,
but that is the best we can do for now.
3. Aerodynamics. There is something nice about those big
fat radial cowls on Bearcat, Corsair and Sea Fury racers (not
to mention the fighter version of the T6). Very nice, and
warming to the heart of the propeller designer. It is this.
The inflow field of the air entering the propeller disc is
axially symmetric. Even better, there is a simple result from
fluid mechanics that permits one to perform an approximation
of the perturbed flow field. This is not the case for poor
old GR7, P51 and the hopeless pusher types.
Flow field aside, there is the interesting situation where
prop tip speeds approach Mach .92, a region where low Reynolds
number and high Mach number effects converge to throw the
design problem into confusion.
So we have 3 regions of interest. Prop disc in front of cowl,
prop disc outside cowl but relative wind at less than M =
.7, and outside M=.7 where shock waves may reasonably be expected
to form.
3.1. Cowling/spinner flow field. Air has to flow out and
around the spinner /cowl assembly. This means that, compared
to the free field flow, the air gains a radial component
of flow and loses an axial component of flow velocity. The
propeller has no response to the radial component, and hence
sees only a reduced axial velocity at the prop disc. As
a consequence, the blade angle has to be reduced to provide
the best angle of attack for maximum L/D.
As the air may be treated as incompressible, this reduction,
to lessening degrees, extends all the way to the tips: indeed,
the axial component of flow may still only be 0.9 times
the airspeed of the aircraft. Now I call that interesting!
Due to the fortunate existence of a theorem in fluid dynamics
related to the sum of sources of flow, this radial gradient
of axial flow may be calculated and the correct (or nearly
so) blade angle determined.
Propellers that do not allow for this change in flow field
may lose a considerable degree of performance. My AT6 prop
designed on these lines is 10% faster than a conventional
prop, a huge difference in terms of propeller efficiency
where designers would kill their grandmother to get just
a 1% gain.
3.2 Outside cowl, axial flow less than M = .7. This region
of the propeller disc has flow speeds for which there is
known data: unfortunately, not for these Reynolds numbers
and model-size conditions. From tests with F2A, F2C and
F3D propellers, it would appear that Suzuki and Connover
type sections out perform any other. These simple shapes,
consisting of a flat bottom, with a circular upper-surface
leading-edge and sharp-leading edge outperform all others.
No idea why, but suspect our friend Reynolds and engine
vibration effects on the flow field are involved. Pitch
is made to rise as we move away from the cowl.
3.3. Shock wave region, M>.7. Now the fun really starts.
If a designer has his way, the tip speed is held below M
= .8 because he has no idea what is happening above this
Mach number. At least, unless he has been here before. Shock
waves form when the local speed of the airflow on the section
falls below Mach 1. Behind the shock, turbulence and flow
separation may occur. In front of the shock, the conditions
which at M<.7 produced lift, now reverse the lift, which
is bad.
This looks like no win. However, due to the strongly 3-dimensional
flow at the tip, there is a reduction in shock wave formation
equivalent to a 10% reduction in Mach number. So now we
can get M = .8 before we get too confused. Studies of shockwave
formation using Schlieren photography in wind tunnel test
in the range .7 < M <1 offer some other insights.
For example, reducing the convexity of the airfoil-section
upper- surface delays and weakens the upper shock. Furthermore,
drooping the trailing edge moves the lower shock forward
and simultaneously weakens it. Nice. Now we have an airfoil
like the so-called 'supercritical sections' which revolutionised
airliner wing design in the early 1970's. Tests on F3D props
showed no loss in performance when an airfoil was adopted
with these features, plus a sharp leading edge. This variation
from the leading-edge design of the supercritical sections
can be excused on the basis of Reynolds number effects and
the empirical results. All this despite having only a guess
at the zero-lift of the crudely constructed profile. Not
bad!
4. Features of the 25X30. This prop has all the features
dictated by the above observations and guesses. That is, low
pitch in the hub region, increasing progressively outside
the cowl all the way to the tip. The hub airfoils, basically
covered by the spinner, are intended to provide sufficient
strength to resist impulsive torque loads. The airfoils outside
the cowl are radially graded to progress from Suzuki/Connover
thru to empirical super-critical section at the extreme tip.
The diameter may be in excess of requirement. It is hoped
that the engine willnot be able to turn the prop as supplied
above 9000 RPM. The tip may then be trimmed off to give whatever
RPM the operator finds gives best performance in the course,
noting that best straight line speed does not give best time
around the course.
The hub has been rubbed to remove the shiny epoxy surface
produced by the mould, thereby increasing friction.
The red coloured points on the hub rear are where holes have
been drilled and lead added to provide radial balance. No
attempt has been made to give lateral balance.
The carbon rovings are radially graded in quantity to provide
maximum strength in tension. The hub is wound with fibreglass
rovings to fill the remaining hub volume. It is CNC drilled
to a PCD of 1.5", with 4 1/4" orthogonal holes drilled
for the retaining bolts.
The blade is tapered in a manner intended to produce an optimal
radial distribution of circulation, thereby maximising induced
efficiency.
5. Testing. This is a new prop design on an engine of unknown
characteristics. testing must proceed with extreme caution,
both to protect the operator and avoid un-necessary damage
to the propeller. The following recommendations should be
considered.
5.1. The engine should be mounted on a test stand for initial
test. The prop driver must be deeply and sharply knurled.
The retaining bolts should be sufficiently tightened so
that the knurls bite into the hub and register there.
5.2 The engine should be started on low throttle and brought
up to 6000 RPM for 1 minute. A this point the retaining
bolts should be checked for torque and tightened again.
5.3 Retain this procedure in steps of 500 RPM up to full
throttle. If stall flutter occurs, immediately close the
throttle. Do not exceed this RPM again, at any time when
the airplane is stationary on the ground.
5.4. At each RPM above, check for compression failures
in the region of the spinner diameter. If compression effects
are present, cease running the engine and return the prop
to the supplier.
5.5. If the prop survives the ground tests, it should be
mounted and flown in an old airframe to reduce the pain
of SHITDS should things go awry. On every launch, the engine
must not be run at full throttle. If the argument is made
that the engine cannot be tuned unless it is run at full
throttle on the ground, sack the lazy operator and find
someone competent. Advance the throttle only after the model
has been released, as before to avoid stall flutter. Even
if you made it on the ground-test to full throttle without
getting flutter, remember that the conditions for stall
flutter may vary with airframe, engine firing and weather.
It will jump up and bite you when you least desire it.
5.6. After every flight, check for prop cracks everywhere,
with emphasis on the front face of the hub. Stop flying
when any anomaly at all appears. Check with Supercool, who
is actually the Greek God of propellers reincarnated. He
will only fire bolts of lightning at you if you withhold
all or any observations, no matter how minor.
5.7. After a dozen flights or so in the test airframe,
and with confidence established in the strength and characteristics
of the prop, place the prop on your race plane and start
again at
5.5. A different airframe will induce difference vibration
modes into the prop. You can never relax on fault inspections
after every flight. 5.8. If the test prop does not perform
up to race standard, contact Supercool who stands by his
product, but despite his divinity can have no knowledge
which you do not supply. Good flying.
|
