of 56 inches (1422.
4 mm); the other two
A COMPARATIVE STUDY
72 inch diameter two blade and 56 inch diameter
three blade propellers designed for 200 mph
airplanes with 100 horsepower engines.
by E. Eugene Larrabee
Emeritus Professor, M.I.T.
525 Victoria St.
Costa Mesa, CA 92627
Airplanes such as the VariEze and
the Prescott Pusher, designed to
maximize laminar boundary layer de-
velopment through pusher propeller in-
stallations, tend to be penalized by un-
dersize propellers necessary for ground
clearance during nose-up rotation at
take off. Limitations on pylon height
places similar restrictions on propeller
90
80
in
O
t"
diameter for flying boats and amphi-
bians such as the Sea Hawker. Tractor
engine airplanes, on the other hand,
can easily swing much larger propellers
whose size is limited by transonic blade
section drag rise and excessive fly-over
noise associated with tip speeds near
sound velocity. The following study
quantifies the relative aerodynamic per-
formance of two fixed pitch propellers
matched to a common design point:
they absorb 100 hp (74 570 W) at 2700
rpm at sea level and 200 mph (89 m/s).
One has three blades and a diameter
absorb a+
effect re
V/n = 77.
\
2O 'in, 3O
for-
blades and a diameter of 72 inches
(1828.8 mm) The difference of perfor-
mance of these two propellers is sur-
prisingly small.
The study is made possible by a 30
kbyte propeller analysis code. ELICA
(the Spanish word for propeller, cog-
nate with "helix") written in Pascal by
James T. Grimes to run on an IBM per-
sonal computer. He translated it from
an earlier code HELICE (the French
word for propeller), written in Fortran by
Susan Elso French. Both codes depend
on algorithms developed by the author
and his former student, Mark Drela, now
an Assistant Professor at M.I.T. The al-
gorithms take advantage of approxima-
tions associated with Betz-Prandtl ra-
dial loading, which is to propellers as
Munk's elliptic span loading is to wings.
An account of the theory will be found
in references 1, 2 and 3.
Figure 1 compares blade planforms
of the two propellers and their radial
pitch distributions. The pitch distribu-
tions given assume that the body carry-
ing the engine has negligible effect on
the flow through the propeller disc. Al-
ternate PRACTICAL pitch distributions
are sketched in for a typical tractor in-
stallation for the two blade propeller and
a typical pusher installation for the three
sea
SPORT AVIATION 31
 0.10 r 1

blade propeller. The idea in adapting a flow velocities near the center. The celled by the increased buoyancy drag
propeller to the body flow field is to re- inner portions of the propeller will then of the body in the axially varying pres-
duce the inner radius pitch enough so produce more shaft thrust with the sure field of the propeller; a tractor fuse-
that the design radial lift distribution will same lift, because the blade path helix lage feels increased aft positive pres-
be preserved, even though the body angles have been reduced. The in- sure, and a pusher fuselage feels in-
disturbance field reduces the axial in- creased shaft thrust will be nearly can- creased aft suction pressure.
The calculated pitch distributions are
based on blade angles measured be-
S> —
T
' X» _
-£)
"
tween the chord lines joining the leading
(.7. = —J — (- D =» ————— and trailing edges of the blade sections
Oes/gn />o/'*r/ and the plane of rotation; for Clark Y
profiles this angle is about two degrees
42 y ^^
more than for a tangent to the flat under
V • ayo«/' *^
___ ^*\
,P = 74.57 IfV ffoafie) surface. All airfoils are assumed to have
r enough camber to develop a lift coeffi-
t ^^-C \ 2?OOrpm cient of 0.4 at zero angle of attack, for
N >v C example, the NACA 44XX airfoil series.
F --'"" "/ \ ^ y P \ V- &m/f(f99./w,)
It will be noted that the three blade pro-
/ C
r\ \ 1 peller, having the higher disc loading,
(ses /'MS/J requires a higher geometric pitch for a
0.1 O.S / given effective pitch; its "slip" is inhe-
\ \ rently greater. Propellers designed by
\
ELICA would have radially constant
\\. 5fco.;A-f geometric pitch if all blade sections had
sufficient camber to develop their de-
\\\ ' "** sign lift coefficients at zero angle of at-
tack in two-dimensional flow.
a.t> o.e ( <z Figures 2 and 3 present the esti-
a ^ i . \\ . mated variations of thrust coefficient,
y
<f. f0.2 A #3 n% 0.4- 0.f A6 t[/s-lr-W power coefficient and efficiency of the
two and three blade propellers, respec-
Fin. 3 Three b/dc/e prope//€r ch&scferisficf tively, as given by the versatile ELICA
code, which even predicts blade stall
32 JULY 1986
 to

0.2 r/g 4.4- 46

.4 c/iord

realistically as the advance ratio is de- blade area and the profile losses at the 0.65 presents no problem with the three
creased. It will be noted that propeller design point, leading to a lower design blade propeller, which can have blade
efficiencies greater than 90% are pre- point efficiency. sections as much as 12% thick right out
dicted for both propellers at the design Figure 5 presents the radial variation to the tip. The blade tip Mach number
point, which is probably optimistic be- of blade section Mach number at the of 0.8 for the two blade propeller is more
cause ELICA tends to underestimate design point. The tip Mach number of challenging, requiring thinner blade tip
the induced velocities (or the "slip")
slightly, and also because the blade
section profile drags may be slightly op-
timistic. Nevertheless, it should be
noted that the two blade propeller has
consistently higher efficiency at all ad- 0. 8-,
vance ratios, mainly because it is larger
and the energy left in its slipstream is
less, even though the slipstream vel-
ocities are more periodic, there being
two blade vortex sheets per revolution
instead of three; more would make the Mo.
slipstream more uniform.
In either case blade stall, shown by 0.4.
saturation of the thrust co-efficient,
tends to occur at about 64% of the de-
sign advance ratio. The stall advance
ratio could be decreased by increasing
the estimated blade maximum lift coef-
ficient of 1.2, which may or may not be
appropriate to the blade Reynolds num-
bers shown for the design condition in
Figure 4. Alternatively the blade design
lift coefficient of roughly 0.5 could be 4-2 0-8
reduced, which would increase the stall . -5 number
margin at low advance ratios. A lower
design lift coefficient would increase the
SPORT AVIATION 33
 80^
P

. so

to

20 -
- 20

4000 rpm 20oo 3006

fit. £> fa// engine
sections to avoid transonic drag rise dimensional near the tips, and because blade propellers are usually made of
caused by shock-induced boundary the Mach number decreases rapidly as duralumin in spite of the weight penalty
layer separation. Nevertheless the flow one moves inward. The thin tip sections and fatigue sensitivity of the material.
around propeller tips is much less liable — say less than 10% thick — are much Because of the unpredictable effects
to cause drag rise than that around easier to obtain with duralumin blades, of blade section radial boundary layer
wings because it is so strongly three which explains why larger diameter two growth on airfoil characteristics (not
given in THE THEORY OF WING SEC-
TIONS!), and because of the equally
hard to estimate fuselage interference
effects, it would seem prudent to make
go . separate blade-and-hub ground adjust-
able pitch propellers for prototype in-
p /ere/ stallations on new machines. Commer-
cially available one piece two blade
duralumin propellers can be repitched
by brute force for a price.
Although little propeller efficiency is
lost through running duralumin propel-
lers at tip Mach numbers of 0.9 (or even
4o • 1.07 for the Curtiss-Reed propeller on
the Curtiss 1925 racer!), the fly-over
noise level is very objectionable be-
cause the sound pulses propagated
along the line of hearing by high tip
&rfr*me speed propellers stay organized into
s/toir highly structured waves, spaced at
too m/vt blade passage period times sound
speed, straight from the blade moving
20 V 40 H>/f 60 SO 100 toward the auditor to his ears. The in-
tensity of the sound one hears is a
.7 strong function of the component of
blade Mach number along the line of
hearing. It is a maximum when the line
34 JULY 1986
 8 Disturbance ef a 2 6/at/e pusher
of hearing coincides with the plane of
rotation — when one is broadside on to
the airplane flying past — and what one
hears is the blades coming towards
one, the sound of which quite over-
whelms the engine exhaust tone. Tip
Mach numbers of 0.6 are almost inaudi-
ble, but tip Mach numbers of 0.8 create
the typical "airplane noise" which airport
neighbors dislike so much.
Although Figures 2 and 3, which
show propeller performance in standard
dimensionless coefficient form, give the
impression of very different characteris-
tics for the two and three blade propel-
lers, when these characteristics are
united with the full throttle engine
characteristic, given in Figure 6, to pro-
duce engine-propeller combination
properties, the overall "power avail-
able", as given by Figure 7, shows a
very marked similarity for the two en-
gine-propeller combinations. To derive
the "power available" curves on Figure
7, it is necessary to calculate several
values of the power coefficient:
Cp = Power/pn3D5 ; p = air density;
n = rpm/60; D = diameter
for several engine speeds at full throttle.
These values then determine corres-
ponding propeller advance ratios:
X = Speed/ilR; il = 2-irn (shaft
speed in radians/s); R = radius neces-
sary to absorb the power from Figs. 2
and 3. The corresponding airspeeds are
given by
V = X (ftR).
The propeller efficiency at these ad-
vance ratios, when multiplied by the en-
gine power absorbed, then gives the
"power available" as a function of the
airspeed. Consistent units must be
used in calculations.
Figure 7 also includes a typical
"power required" to maintain straight
flight for a small airframe such as the
VariEze or Lancair configurations. It will
be seen that the high tip speed two
blade propeller-engine combination is
slightly better at all airspeeds, giving
both a higher top speed (power required
= power available) and a higher rate of
climb (more difference between power
available and power required at all
speeds below top speed). It does so at
the expense of more fly-over noise.
Figure 8 shows why the two blade
propeller might not be suitable for Vari-
Eze. Every time the propeller becomes
horizontal the two blades simultane-
ously encounter sharply defined wing
wakes. The wing wakes momentarily
deprive the blades of a large fraction of
their inflow velocities, leading to large,
pulse like increases of blade angle of
attack and forward bending of both
blades. There are, of course, simultane-
ous anti-symmetric excitations of blade
in-plane bending due to the associated
torque disturbances. These distur-
bances have a good chance of produc-
ing resonant response in solid duralu-
min blades, which ring like a bell, and
are notoriously susceptible to fatique.
Two blade wooden or plastic propellers
are more fatique resistant. A three
blade propeller should be better yet,
since only one blade encounters the
wing wake at a time.
References
1. Larrabee, E.; "Propeller Design
and Analysis for Modelers"; 1979 NFFS
International Symposium Report,
Bakersfield, CA, pps 9-25. Parallel En-
glish and French text; detailed al-
gorithms given.
2. Larrabee, E.; "Five Years Experi-
ence with Minimum Induced Loss Pro-
pellers — Part I, Theory' Part II, Appli-
cations"; SAE preprints 840026 and
840027.
3. Larrabee, E. and Drela, M.; "De-
sign and Analysis of Efficient Propel-
lers"; manuscript submitted to AIAA
Journal of Aircraft, December 1985.
SPORT AVIATION 35