Design and Testing of an Autorotative Payload Delivery System

Anne Brindejonc Jayant Sirohi Inderjit Chopra

Graduate Research Assistant Assistant Research Scientist Alfred Gessow Professor and Director
Alfred Gessow Rotorcraft Center, Department of Aerospace Engineering
University of Maryland, College Park, MD

The design, testing, and analysis of an autonomous autorotative payload delivery system called the Autobody is presented.
The Autobody must be capable of passively deploying a payload from a conventional aircraft, by means of an autorotative
rotor. Operational requirements specify the Autobody to have a four bladed rotor with a diameter of four feet, a total mass of
2.27 kg (5 lb) and a maximum steady state descent velocity of 4.57 m/s (15 ft/s). A novel rotor hub design incorporating negative
pitchflap coupling in conjunction with negative blade pitch and a negative precone is implemented to passively achieve the
transition to steady autorotation. An analysis is developed to predict the steady state behavior of the Autobody. Only vertical
autorotation is considered as it will result in a conservative design and is the simplest state to analyze. Wind tunnel tests were
performed on a scaled model rotor to validate the analysis and to investigate the effect of different rotor parameters. The
analysis was then used to perform a parametric study of the effect of several rotor variables on the system performance, from
which an optimum full scale configuration is identified. An instrumented full scale prototype was flight tested by dropping
it from a hot air balloon. For an Autobody of mass 2.27 kg, with a 41 pitch-flap coupling angle, a 10 fixed collective
pitch, and a 4 precone, a steady state descent velocity of 4.11 m/s (13.5 ft/s) was observed. Based on the predictions and
the flight tests, it was concluded that the proposed Autobody design satisfactorily meets all operational requirements.

Nomenclature T rotor thrust

Tana predicted thrust
aoa coefficient of thrust predicted by analysis Texp measured thrust
aoe coefficient of measured thrust uT velocity in-plane of rotor disk
ba coefficient of rotational speed predicted by analysis up velocity out-of-plane of rotor disk
be coefficient of measured rotational speed Vd descent velocity
Cd drag coefficient vh induced velocity in hover
Cl lift coefficient angle of attack
Cl lift curve slope flapping angle
CQ torque coefficient p precone angle
CT thrust coefficient 3 angle of flapping hinge (pitch-flap coupling angle)
dD elemental drag force  change in flapping angle
d Fz vertical force at blade element  change in blade pitch angle
dL elemental lift force blade pitch angle
dQ torque on annular element o blade root pitch angle
dr width of annular element TW blade twist rate
dT thrust on annular element induced velocity ratio
ET error between measured and predicted thrust d descent velocity ratio
E RPM error between measured and predicted rotational speed o nonrotating flapping frequency
F resultant force e rotating flapping frequency
Ib blade flapping inertia blade solidity
k blade flapping stiffness rotor rotational speed
M AF total nondimensional aerodynamic flapping moment ana predicted rotational speed
mb blade mass exp measured rotational speed
Q rotor torque
r radial location
Introduction
Corresponding author; email: [email protected].
Presented, in part, at the 61st Annual Forum of the American Helicopter Society Autonomous unmanned aerial vehicles (UAVs) find numerous ap-
International, Grapevine, TX, June 13, 2005. Manuscript received August 2006; plications in hazardous civilian and military environments, which in-
accepted July 2007. clude aerial reconnaissance and surveillance of enemy territory, traffic

360
OCTOBER 2007 DESIGN AND TESTING OF AN AUTOROTATIVE PAYLOAD DELIVERY SYSTEM 361

monitoring, and search and rescue operations. Recently, there has been Sirohi et al. (Ref. 12). This study proposed a rotor with a negative pre-
considerable interest in the design and development of versatile, cost- cone and a negative pitch-flap coupling to achieve the required change
effective, autonomous UAVs, which are capable of deploying electronic in collective pitch in flight. The rotor started from rest and underwent a
instruments on ground from conventional aircraft. Such a vehicle is re- transient state before reaching steady autorotation. While the basic con-
quired to decelerate the payload to a low descent velocity in order to mini- cepts were discussed and a parametric study performed, the theoretical
mize the impact load on the sensitive onboard electronic equipment. A de- analysis needed to be refined and validated with wind tunnel data and
livery system called the Autobody is envisaged to safely deploy a payload flight test data. The present study complements this work.
dropped from an aircraft. The main feature of the Autobody is an autorota- The present paper describes the design of an autorotative rotor with
tive rotor to produce lift, thus eliminating the need for an on-board engine. completely passive operation. The development of the theory to predict
An autorotative rotor was preferred over other decelerating devices, such the steady state behavior of the system is presented. The analysis was
as a parachute, for several reasons (Ref. 1). For example, an autorotative validated with wind tunnel tests performed on a quarter scale model. The
system eliminates the sudden deceleration during deployment, smoothing validated analysis was used to carry out a parametric study to minimize
out the descent. In addition, once deployed, power can be provided to the the rate of descent of the Autobody. A full scale prototype was fabricated
on-board instrumentation with solar cells embedded in the rotor blades. and flight tested by dropping it from a hot air balloon. This represents
The objective of the present study is to design a passively controlled, vertical autorotation, and ignores any effects of forward wind velocity
autorotative rotor with minimum mechanical complexity and minimum that the Autobody might encounter while being deployed from an aircraft.
weight penalty. The vehicle must satisfy certain operational constraints Data measured during the flight test is used to further validate the analysis.
imposed by the specific deployment scenario. These include a maximum
gross weight of 2.27 kg (5 lb), a four bladed rotor of 1.22 m (4 ft) diameter, Physical Principles
and a maximum steady state rate of descent of 4.57 m/s (15 ft/s). It is
known that the minimum rate of descent occurs for autorotation with some The rotor of the Autobody starts from rest, and the rotor rotation needs
forward velocity, hence designing the rotor for the specified maximum to be initiated in the correct direction. As the system descends, the rotor
rate of descent in vertical autorotation results in a conservative design. spins up and has to produce an increasing thrust. In the steady state,
In addition, vertical autorotation presents the simplest case for analysis. the thrust must be equal to the total weight of the system. The descent
Therefore, the present study only considers purely vertical autorotation. velocity should be as low as possible to permit a safe landing. This can
Several authors have discussed the phenomenon of steady autorota- be achieved by maximizing the rotor thrust for a given descent velocity.
tion (Refs. 25). Because momentum theory is not valid in this rotor The resultant force F acting on a rotor blade as it starts from rest is
operating condition, semi-empirical models are typically used for anal- shown in Fig. 1. Just after the vehicle is released, the rotor rotational speed
ysis. Although steady autorotation is well documented and understood, (RPM) and the in-plane velocity of the blades, u T are both equal to zero.
there is little data on rotors starting from rest and experiencing a transient An out-of-plane velocity, u p is present, and is equal and opposite to Vd , the
state before reaching autorotation. descending velocity of the system. If the blade is set at a positive pitch,
The concept of an autorotative vehicle dropped from an aircraft has the resultant force F will have an in-plane component acting towards
been previously explored by some researchers. During the second world the trailing edge. This causes the rotor to start rotating trailing edge first.
war, Hafner developed a manned autogyro glider, the Rotachute, capable If the blade is set at a negative pitch, the in-plane force causes the rotor to
of delivering a human payload from a large transport aircraft (Ref. 6), and spin up in the correct direction: leading edge first. However, a low value
a larger system named the Rotabuggy. In 1942, Flugzeugwerke developed of blade pitch results in a low steady state thrust. Therefore, to obtain
the Fa-33 (Ref. 6), which was an autogyro towed behind a submarine for a high steady state thrust, a passive increase in pitch is needed as the
surveillance purposes. In 2003, Bartz and Miklosovic (Ref. 7) investi- RPM increases. To achieve this, a negative 3 angle is incorporated at
gated the effect of airfoil camber on the autorotation and deceleration the blade flap hinge, yielding a negative pitch-flap coupling. As a result,
performance of an Autorotor. However, the Autorotor was equipped with an increase in flap angle will result in an increase in blade pitch. The
a controllable outboard flap. relationship between blade pitch and flap angle is expressed as
The idea of an autorotative rotor mounted on a spacecraft to lower
its rate of descent during re-entry has also been widely investigated.  =  tan(3 )
Wernicke (Ref. 8) presented a preliminary experimental study of a model = ( p ) tan(3 ) (1)
Spacecraft Rotor Landing System. Kretz (Ref. 9) developed a rotor-based
re-entry system, called the Space Rotor, that incorporated actively con-
trolled blade collective pitch. Levin and Smith (Refs. 10, 11) performed F
wind tunnel tests on a rotor/launcher assembly to determine the achiev- Motion
able performance gains. The Roton, developed by the Rotary Rocket Rotor plane
Company (Ref. 1), was capable of taking off vertically like a rocket un- up
der the control of a crew, delivering a 3,175-kg (7,000 lb) payload to low
earth orbit, and slowing down during re-entry by means of an autorotative Positive collective pitch
rotor.
These concepts, while being innovative, were actively controlled.
Conventional piloted helicopters enter into autorotation in case of en- F
gine failure. A key function of the pilot during autorotation is to ensure Motion
that the rotor speed does not decrease below a certain acceptable value.
Rotor plane
Flying in this state requires constant pilot inputs to control the blade pitch
and therefore the rate of descent. In the case of the Autobody, the rotor up
starts from rest, and must transition to steady autorotation.
Negative collective pitch
A theoretical study of the feasibility of a fully autonomous, autorota-
tive system with no on-board active control systems was performed by Fig. 1. Effect of blade collective pitch on the direction of rotation.
362 A. BRINDEJONC JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

where  is the change in pitch angle,  the change in flap angle, and The RPM sweep analysis assumes purely axial flow and sea level
p is the blade precone angle. Note that if p < 0, then  is greater conditions. The blades are assumed rigid and undergo flap and pitch
than it would be if p > 0. In other words, introducing a negative precone degrees of freedom. The analysis includes blade dynamics and kinematic
angle further increases the blade pitch. pitch-flap coupling in conjunction with an asymptotic solution procedure
Based on these observations, the final hub design incorporates the based on a refined blade element theory. The lift coefficient at each blade
following features: 1) a negative blade pitch angle o to initiate rotation element is determined by Cl = Cl , where Cl = 5.73 /rad is the lift
in the correct direction, 2) a negative precone angle p to allow a large curve slope, and is the angle of attack. The drag coefficient at each
change in flap deflection, and 3) a negative 3 angle to convert the change blade element is taken to be constant, Cd = 0.04. This value is slightly
in flap deflection to an increase of blade pitch. While descending, the higher than the typical value as the rotor blades are in a low Reynolds
rotor speeds up leading to an increase in lift and centrifugal force on each number environment. Given a descent velocity and a rotational velocity,
blade. As a result, the blade flaps up, which because of the 3 coupling, the inflow, aerodynamic angles and elemental forces are computed at
results in an increase in blade pitch. This causes a further increase in the each blade element until the blade flap angle converges. The integrated
lift and torque on the rotor. The rotor RPM increases continuously until values of rotor thrust and torque are then determined. The basic algorithm
the equilibrium condition is arrived: a constant RPM and a zero net rotor is as follows:
torque. In this manner, a passive increase in the rotor collective pitch, and 1) Step 1: Inputs and initial conditions. Input data include the assumed
a transition to a state of steady autorotation is achieved. RPM, rotor parameters (including blade pitch, precone and pitch-flap
coupling), and descent velocity.
Analytical Model: The RPM Sweep Method 2) Step 2: Calculation of inflow. The inflow at a rotor annulus of
width dr at a radial station r , is calculated by the blade element momen-
The aim of the analysis, called the RPM sweep method, is to predict the tum theory as
rotor thrust and RPM when it has reached the steady state of autorotation.
 2

The analysis is validated with experiments performed on a scaled model d Cl   d Cl Cl r
rotor in an open-jet wind tunnel, and is then used to design a full scale = + + (2)
2 16 2 16 8
prototype. Note that typical estimates of autorotative rates of descent
for helicopters rely on empirical data (Refs. 2, 4), and some researchers
where is the rotor solidity, is the pitch angle and d is the descent
have experimented with powered rotor models in wind tunnels in order
velocity ratio.
to improve the understanding of these empirical relations (Ref. 13). The
3) Step 3: Calculation of aerodynamic angles and elemental forces.
goal of the present analysis is to design an optimum rotor geometry to
Based on the inflow calculated in Step 2, and the pitch angle at each blade
minimize the descent velocity of the Autobody.
element, the aerodynamic angles are calculated. The spanwise pitch angle
For a given descent velocity Vd , a value of rotor RPM is chosen for
is given by
which the rotor thrust T and torque Q are calculated from blade element
momentum theory (BEMT). The chosen RPM starts from a value that (r ) = o + r TW +  (3)
represents a large negative ratio of descent velocity to hover induced
velocity (ratio Vd /vh ) and gradually increases until it approaches the where o is the blade root pitch, TW is the twist rate and  is the change
ratio Vd /vh = 2. In this way, the chosen RPMs sweep through the of pitch, written as
windmill brake state of the rotor. Calculation is stopped if Vd /vh 2,
beyond which momentum theory is not valid. The calculated rotor torque  = tan(3 )
coefficients are plotted versus RPM and the curve is extrapolated to find = tan(3 )( p ) (4)
the point of zero rotor torque (Fig. 2). This point corresponds to steady
state autorotation. The rotor thrust and torque are calculated based on the The change in flap angle, , is initially equal to zero. The elemental lift
RPM at this point. d L and drag d D as well as the components of the elemental forces along
the horizontal and vertical axes are calculated at each blade segment. The
elemental thrust dT and torque d Q are then computed on each annulus.
0.05 The total nondimensional aerodynamic flapping moment on the blade,
M AF , is
0
Cq = 0 R
d Fz
Torque coefficient, CT

RPM = 2477 0
-0.05 AF =
M (5)
2 I b
-0.1 where d Fz is the vertical force at the blade element and Ib is the moment
of inertia of the blade about the flapping hinge. The steady flap angle
-0.15 is calculated from the steady state flapping equation as

AF
M 2o
-0.2 = + p (6)
2e 2e
-0.25 where the nonrotating flapping frequency o and rotating flapping fre-
0 1000 2000 3000 4000
quency e are defined as
RPM

Fig. 2. Sample result of rotor torque coefficient versus RPM given by k
o = (7)
the RPM sweep method. Ib 2
OCTOBER 2007 DESIGN AND TESTING OF AN AUTOROTATIVE PAYLOAD DELIVERY SYSTEM 363


e = 2o + 1 (8)

The change in pitch angle is then obtained using Eq. (4), and several
iterations are performed until converges. The total rotor thrust T , torque
Q, thrust coefficient C T and torque coefficient C Q are then calculated.
4) Step 4: Check rotor operating regime. If 2 < Vd /vh < 0, momen-
tum theory is no longer valid, and the calculation is stopped. The curve
of C Q versus RPM is extrapolated to find the RPM at which C Q = 0. If Fig. 3. Close-up view of the model rotor hub with the negative root
Vd /vh < 2, then the computation proceeds to the next RPM. pitch angle, o , and the negative precone angle, p .

Wind Tunnel Tests

The performance of a scaled model of the rotor was evaluated by

mounting it in an open jet wind tunnel such that the wind flowed through
the rotor in a condition of axial descent. The main goal of the wind tunnel
tests was to validate the RPM sweep method. Steady RPM and thrust were
measured for different sets of rotor parameters. Because the wind tunnel
offered a controlled environment, the influence of blade pitch and precone
and pitch-flap coupling on the steady state could be investigated.

Model rotor construction

A scaled model rotor of diameter 0.33 m (13 inches) was sized to

fit the dimensions of the test section (0.533 m (21 inches) 0.533 m
(21 inches)) of an open-jet wind tunnel. The characteristics of the model
rotor are shown in Table 1 and the main features of the rotor are shown Fig. 4. Close-up view of the model rotor blade grip with the 3 hinge.
in Fig. 3. Note that while the figures show a four-bladed rotor, the final
wind tunnel tests were performed with a two-bladed rotor in order to
minimize errors inherent in manually setting the blade pitch. The model
blade grips feature a flexure based flap hinge with a 3 angle machined
at 30 or 50 (Fig. 4). As a result of the finite width of the flexure, it
was found that the effective flap hinge line was not along the machined
hinge axis. The actual 3 angles were measured to be 29.74 for the
50 case and 17.33 for the 30 case. Grips without 3 were also
manufactured to test the case of no pitch-flap coupling. The hub features
a negative precone of p = 0 , p = 4 or p = 6 . A negative blade
pitch can be set (Fig. 3) by rotating the grip around the hub fixture. The
blades are made of 31-Rohacell foam, of density 32 kg/m3 , and covered
with one ply of carbon fiber prepeg.

Wind tunnel test setup

Autorotation was simulated by a free-spinning hub (the torque on hub

is zero at steady state) and the wind tunnel velocity reproduced the upward
flow of velocity Vd on the rotor (Fig. 5). The rotor RPM was measured
by a Hall switch installed on the rotor stand in conjunction with a magnet
attached to the free spinning hub. A load cell between the rotor shaft and
the rotor stand measured the rotor thrust. The rotor stand was designed to

Table 1. Parameters of the scaled rotor model Fig. 5. Wind tunnel test setup.
Rotor diameter, m (inches) 0.33 (13)
Blade number, Nb 2
Blade span, m (inches) 0.1524 (6) provide a stable support to the rotor/load cell assembly while at the same
Blade chord, m (inches) 0.0287 (1.13) time minimizing any vibrations that could adversely affect the quality
Blade twist, deg/m (deg/inch) 0 (0) of the load cell measurement. The stand was also designed to minimize
Blade mass (g) 5.2 any flow disturbances. The free stream velocity Vd was measured using a
Airfoil NACA 0010
Pitot probe connected to a pressure transducer. All data were acquired by a
Pitch angle (deg) 0, 6, 8, 12
data acquisition computer running an in-house developed LabView based
Pitch-flap coupling angle, 3 (deg) None; 17.33; 29.74
Precone angle, p (deg) 0 ; 4; 6 software. The test data were collected over a range of rotor parameters:
blade pitch angle o , pitch-flap coupling angle 3 , blade precone angle p ,
364 A. BRINDEJONC JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

5 5
Experiment Experiment
Analysis Analysis
4 4

o = 6o

Thrust (N)
Thrust (N)

3 3
p = 6
2 o = 8o 2

1 o = 12o 1
p = 0

0 0
0 2 4 6 8 10 0 2 4 6 8 10

Descent velocity (m/s) Descent velocity (m/s)

(a) Steady state thrust
(a) Steady state thrust
Experiment
Experiment 4000
4000 Analysis
Analysis
o = 6
o

3000 p = 6
3000
o = 8
o

RPM
p = 0
RPM

2000
o = 12
o
2000

1000
1000

0
0 0 2 4 6 8 10
0 2 4 6 8 10
Descent velocity (m/s)
Descent velocity (m/s)
(b) Steady state RPM
(b) Steady state RPM Fig. 7. Analytical and experimental results from a 2-bladed rotor
Fig. 6. Analytical and experimental results from a 2-bladed rotor with a pitch-flap coupling 3 = 29.74 and pitch o = 12 .
with stiff flap hinges, no pitch-flap coupling, precone p = 0 .

and wind tunnel velocities Vd ranging from 1 to 9 m/s. The values of the blade pitch. This is because a lower blade pitch effectively translates
tested pitch angles, 3 angles and precone angles are shown in Table 1. to a lower steady state angle of attack. The same trend is observed in
other groups of tests that feature different values of precone angle and
Experimental Results pitch-flap coupling, and is independent of the value of Vd .

The measured steady state rotor thrust and RPM are plotted as a
function of descent velocity Vd (wind tunnel velocity) in Figs. 68. The Effect of precone p . Figure 7 shows the steady state rotor thrust and
values of thrust and RPM calculated using the RPM sweep method are RPM as the precone angle decreases from p = 0 to p = 6 . For these
also plotted. From these figures, the effect of rotor parameters on the tests, the pitch-flap coupling angle was set at 3 = 29.74 and the blade
performance of the rotor, as well as the correlation of the measurements pitch was o = 12 . For Vd = 4.5 m/s, the thrust increases from 0.35 to
with analysis, can be evaluated. 0.54 N and the RPM increases from 1,440 to 1,500. It can be concluded
that the thrust and RPM increase with more negative precone. Note that
for the cases with the rigid flap hinge, the measured thrust and RPM are
Influence of rotor parameters
independent of the precone angle. This is because the main influence of
decreasing the precone angle is to increase the effective change in flap
The influence of the design parameters on the performance of the
angle at steady state. Therefore, the precone angle is not effective in the
rotor can be evaluated by comparing the steady thrust and RPM of the
case of a rigid flap hinge.
rotor at a specific descent velocity, while changing the rotor parameters
between each test run, as indicated in Table 1.
Effect of pitch-flap coupling angle 3 . Figure 8 shows the steady state
Effect of blade pitch o . The variation of steady state thrust and RPM as a rotor thrust and RPM as the pitch-flap coupling angle is changed from
function of descent velocity for blade pitch angles of o = 6 , 8 , and 3 = 29.74 to 3 = 17.33 . The blade pitch was kept at o = 12
12 is shown in Fig. 6. For these tests, the rotor was configured with a and the precone was kept at p = 6 for these tests. For Vd = 6.5 m/s,
rigid flap hinge as well as zero precone. At Vd = 8 m/s, the thrust decreases the thrust decreases from 0.99 to 0.85 N as the 3 angle is increased, and
from 2.5 N to approximately 1 N and the RPM decreases from 3,400 to the RPM decreases from 2,160 to 2,100. Therefore the thrust and RPM
2,200. It can be concluded that the thrust and RPM decrease with lower increase with more negative pitch-flap coupling.
OCTOBER 2007 DESIGN AND TESTING OF AN AUTOROTATIVE PAYLOAD DELIVERY SYSTEM 365

5 Correlation with analysis

Experiment
Analysis In order to compare the trends of the measurements, and to correlate
4
the analysis with measured results, it is useful to represent the data in
Thrust (N)

terms of a polynomial curve fit. From the experimental results, it was

3 observed that the variation of thrust with wind velocity was quadratic
3 = 29.74 and the variation of RPM with wind velocity was linear. For each plot of
2 thrust versus Vd and RPM versus Vd , a curve can be fitted to the experi-
mental data points. Similarly, the values of thrust and RPM calculated at
1 3 = 17.33 each descent velocity using the RPM sweep method can be expressed in
terms of polynomials. The measured thrust, Texp , measured RPM, exp ,
0 analytical thrust, Tana and analytical RPM, ana , are given by
0 2 4 6 8 10
Texp = aoe Vd2 (9)
Descent velocity (m/s)
exp = b Vde
(10)
(a) Steady state thrust
Tana = aoa Vd2 (11)
Experiment
4000 ana = b Vda
(12)
Analysis
3 = 29.74 where the superscript e refers to coefficients from measured data and
3000 the superscript a refers to coefficients from the analysis. Comparing the
coefficients aoe with aoa and be with ba provides a convenient means to
RPM

2000 3 = 17.33 evaluate the accuracy of the analytical predictions with respect to the
experimental data. Table 2 shows the aoe , aoa , be and ba coefficients for
each test case. The error between the analysis and measured values is
1000 given in terms of the error in thrust, E T , and the error in RPM, E RPM .
In most cases the analysis underestimates the thrust and RPM. In
0 general, the analysis shows the correct trends and around 10 to 15%
0 2 4 6 8 10 magnitude error in both thrust and RPM. However, for some cases the
Descent velocity (m/s) correlation is less accurate, with errors of up to 25%. The errors do not
appear to exhibit any definite trends with the test variables; however, in
(b) Steady state RPM general the analysis is less accurate for the cases with 3 = 17.33 .
Fig. 8. Analytical and experimental results from a 2-bladed rotor The discrepancy between the analysis and measurements can be at-
with a precone p = 6 and pitch o = 12 . tributed to two main causes. Firstly, the underlying physics impose

Table 2. General wind tunnel test matrix

Case No. o (deg) 3 (deg) p (deg) aoe aoa ET (%) be ba ERPM (%)
1 6 None 0 0.0427 0.0396 7.2 403.71 362.93 10.1
2 8 None 0 0.0254 0.0279 9.8 343.93 326.54 5.0
3 12 None 0 0.0138 0.0141 2.1 273.77 262.1 4.2
4 6 29.74 0 0.0452 0.0442 2.2 404.05 373.8 7.4
5 8 29.74 0 0.0372 0.0307 17.4 385.99 336.11 12.9
6 12 29.74 0 0.016 0.0149 6.8 290.45 263.55 9.2
7 6 29.74 4 0.092 0.0708 23.04 459.8 354.1 22.9
8 8 29.74 4 0.0688 0.0637 7.41 459.92 355.18 22.77
9 12 29.74 4 0.0297 0.0225 24.2 352.84 302.48 14.2
10 6 29.74 6 0.0833 0.0699 16.0 421.28 399.52 5.1
11 8 29.74 6 0.0669 0.0561 16.1 415.04 393.86 5.1
12 12 29.74 6 0.0333 0.0279 16.2 346.04 323.99 6.3
13 6 17.33 0 0.076 0.0581 23.5 446.8 344.02 23.0
14 8 17.33 0 0.0486 0.0505 3.9 423.5 327.19 22.74
15 12 17.33 0 0.0181 0.0146 19.3 309.96 261.35 15.6
16 6 17.33 4 0.0808 0.0628 22.27 467.22 354.71 24.0
17 8 17.33 4 0.0548 0.0557 1.6 420.08 336.66 19.8
18 12 17.33 4 0.0227 0.0181 20.2 330.22 281.13 14.8
19 6 17.33 6 0.089 0.0679 23.7 499.5 376.14 24.7
20 8 17.33 6 0.0538 0.0413 23.2 437.05 366.17 16.2
21 12 17.33 6 0.0242 0.0202 16.5 334.1 291.74 12.6
366 A. BRINDEJONC JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

fundamental limits on the analysis. Autorotation occurs in the vortex 3

ring state of the rotor where there is no well-defined slipstream. The flow
behind the rotor is highly unsteady, and no analytical model appears to 2.5
exist that can capture the behavior of the rotor in this state. The RPM

Thrust (N)
2
sweep method represents an approximation to this physical condition, ob-
tained by extrapolating the rotor torque versus RPM curve. As the curve 1.5
reaches zero rotor torque (Fig. 2), the slope of the curve decreases and be-
1
comes close to zero. Consequently, it is difficult to obtain the exact point
of intersection of the curve with the zero torque axis. This presents an 0.5
additional approximation in the analysis. Second, the steady rotor thrust
0
and RPM are extremely sensitive to the blade pitch angle. An error of 4 6 12 45
even 1 in setting the blade pitch angle on the rotor model will result in Pitch angles
a significant error in the predicted value of rotor thrust and RPM. It can
(a) Thrust
be concluded that the error in the analysis is acceptable within the limits 3000

Rotational velocity (RPM)

of the simplistic approach of the RPM sweep method, and the present
analysis is satisfactory as a preliminary design tool. 2500
It is interesting to investigate the sensitivity of the analysis to the 2000
choice of the airfoil profile drag coefficient. For instance, if Cd = 0.02
is used instead of Cd = 0.04 the calculated RPMs were closer to the 1500
measured data. For example, in case No. 19, for Cd = 0.02 the RPM error 1000
E RPM = 11%, whereas for Cd = 0.04 the RPM error is E RPM = 24.7%.
However, using a lower Cd value increases the error in thrust predictions. 500
For case No. 19, the thrust error E T = 23.7% for Cd = 0.04, and in the 0
case of Cd = 0.02, the thrust error is E T = 52.3%. Hence, the use of 4 6 12 45
Cd = 0.04 in the analysis is justified. Pitch angles
(b) RPM
Parametric study
Fig. 9. Thrust and rotational velocity for different pitch angles.
The wind tunnel tests gave an understanding of the behavior of the
system and of the influence of the initial collective pitch, pitch-flap cou-
pling, and precone angles. Using the RPM sweep method, a parametric
study is performed on the rotor design variables. The descent velocity is
2.5
chosen as 4.57 m/s, and the variables of the parametric study are the blade
pitch angle o , pitch-flap coupling angle, 3 , precone angle, p , and the
2
blade mass m b . The baseline case is chosen as 1) blade pitch, o = 12 ;
Thrust (N)

2) pitch-flap coupling angle, 3 = 17.33 ; 3) precone angle, p = 6 ;

1.5
and 4) blade mass m b = 0.0052 kg.
The effect of the design variables is summarized in Figs. 912 and in 1
Table 3. It can be seen that thrust and RPM vary monotonically with each
design parameter, and no optimum value exists. The values for the full 0.5
scale Autobody must take into account other mechanical considerations
and are chosen as follows: 0
1) The value of blade pitch should be chosen as high as possible to 6 17.33 34.66 45 50
maximize the thrust; however, the pitch at 3/4 radius should be negative Pitch-flap coupling angles
to initiate rotation in the correct direction. (a) Thrust
2) For 3 angles lower than 50 , the analysis showed that the blade 3000
Rotational velocity (RPM)

gets stalled. Therefore, the optimum 3 value for the Autobody was cho-
2500
sen to be 45 to avoid stall and achieve a high steady state thrust.
3) An ideal value of precone would be as negative as possible. How- 2000
ever, a large precone would result in a decrease in disk area and high
1500
stresses at the blade root. A value of 4 is chosen for the full-scale
Autobody. 1000
4) The lowest possible blade mass will result in the best performance.
However, care must be taken to maintain an acceptable blade stiffness. 500
In addition, it should be noted that the quantitative influence of each 0
parameter is different. Table 3 presents the variation in steady state thrust
6 17.33 34.66 45 50
and RPM for a change of 100% in each parameter from the baseline Pitch-flap coupling angles
values. It can be seen that the parameters with the greatest influence
(b) RPM
on rotational velocity and thrust are the 3 angle and the blade pitch.
The precone and the blade mass have less influence on the thrust and Fig. 10. Thrust and rotational velocity for different pitch-flap cou-
RPM. pling angles.
OCTOBER 2007 DESIGN AND TESTING OF AN AUTOROTATIVE PAYLOAD DELIVERY SYSTEM 367

3 Table 3. Sensitivity of steady state thrust and RPM

on design parameters
2.5
Parameter Parameter Variation (%) Variation (%) T Variation (%)
Thrust (N)
2
o +100 25 63
1.5 3 +100 +30.2 +58.3
p +100 +11.7 +39
1 mb +100 0.51 1.5

0.5

0
0 4 6 12 30
Precone angles
(a) Thrust
3000
Rotational velocity (RPM)

2500

2000

1500

1000

500

0
0 4 6 12 30
Precone angles
(b) RPM
Fig. 13. Full-scale Autobody configuration.
Fig. 11. Thrust and rotational velocity for different precone angles.

Flight Tests

A full-scale prototype Autobody was designed based on the results

of the parametric study, and was flight tested with on-board instrumen-
tation to measure the descent velocity and the rotor RPM. The data were
0.72
correlated with a modified version of the RPM sweep analysis.

0.71 Prototype construction

Thrust (N)

0.7 The full-scale prototype is shown in Fig. 13. The fuselage was a
long cylindrical body of diameter 4 inches, terminated by a heavy round
nose. The prototype incorporated a 48-inch diameter, four-bladed rotor
0.69
attached to the body by means of a free spinning hub. The full-scale
rotor was designed with an adjustable blade pitch, a negative precone
0.68 p = 4 , and a negative pitch-flap coupling angle 3 = 45 .
0.0052 0.0104 0.1 1
Blade mass (kg)
The blade grips of the full-scale rotor consist of two parts, linked
by a clamped spring steel shim, which acts as a flexure and constitutes
(a) Thrust
1755
the 3 hinge (Fig. 14). The actual value of the 3 angle was measured
to be 41 due to the finite width of the flexure. The flapping stiffness
Rotational velocity (RPM)

1750
of the hinge could be changed by varying the thickness of the spring
steel shim. A lag hinge was incorporated in the blade grip so that the
1740
rotor blades could rotate about this hinge to absorb the energy during
landing impact. The blades were manufactured with a positive twist and a
rectangular planform, and consisted of a foam core covered with two plies
1730 of IM7/8552 carbon fiber weave. Two sets of blades were manufactured,
each with a different mass. The parameters of the full-scale rotor are
summarized in the flight test matrix shown in Table 4.
1720
0.0052 0.0104 0.1 1
The blades had a tip twist of +7.75 to achieve a more uniform distri-
Blade mass (kg)
bution of angle of attack along the span of the blade. In order to initiate
rotor rotation in the correct direction, the blade pitch at 3/4 span must
(b) RPM
be negative, i.e., 0.75 < 0. Therefore, the blade root pitch must meet the
Fig. 12. Thrust and rotational velocity for different blade masses. condition o 5.8 .
368 A. BRINDEJONC JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

15
13

Vertical acceleration (m/s 2)

11 Test #1
9
7 Test #3
5
Test #2
3
1
1
3
5
Fig. 14. Features of the full-scale rotor blade grip with a 3 of 41 . 0 5 10 15
Time (sec)
Fig. 15. Comparison of vertical acceleration data time history
Table 4. Flight tests matrix
recorded during different flight tests.
Test Number 1 2 3
Rotor diameter, m (inches) 1.22 (48) 1.22 (48) 1.22 (48)
365
Number of blades, Nb 4 4 4
Blade span, m (inches) 0.508 (20) 0.508 (20) 0.508 (20)
Blade chord, m (inches) 0.0762 (3) 0.0762 (3) 0.0762 (3) 305
Blade airfoil SC 1095 SC 1095 SC 1095
Test #1
Blade tip twist, deg +7.75 +7.75 +7.75 245

Height (m)
Root pitch angle, deg 10 9 10
3 angle, deg 41 41 41 185
Precone angle, p , deg 4 4 4
Test #2
Blade mass, g 84.6 61.3 84.6 125
Flapping stiffness, k , Nm/rad 313 313 94 Test #3
65

0
Instrumentation 0 10 20 30 40
Time (sec)
On-board instrumentation was installed to acquire rotor RPM, ac-
celeration along the rotor shaft axis and height data. The sensors are Fig. 16. Comparison of autobody height data time history recorded
connected to an on-board micro-recorder, the Tattletale TT8V2, and during different flight tests.
the data were recorded as a function of time. The instrumentation con-
sisted of 1) a Hall switch to measure RPM in conjunction with a mag-
net mounted on the rotor shaft; 2) an ADXL202 two-axis accelerome- Figures 1517 show the vertical acceleration, height, and rotor RPM
ter mounted in the Autobody such that one of the sensing directions is as a function of time for the three flight tests. The transient as well as
aligned with the longitudinal axis of the Autobody. The vertical accel- steady state response of the Autobody can be clearly seen from the vertical
eration is used to identify the steady state condition; and 3) a pressure acceleration shown in Fig. 15. The Autobody was held in the balloon for
transducer to measure the static pressure inside the body. The height 0.8 s and then released. During this time, the Autobody was stationary,
can be obtained from these readings and is used to obtain the descent and its vertical acceleration was zero. At this point, its height was constant
velocity. and equal to the reference altitude of 305 m (Fig. 16). From Fig. 15, it
can be seen that at the instant of release, the Autobody experiences an
Test matrix acceleration of 1 g. Thereafter, the rotor starts producing an increasing
amount of thrust and decelerates the system. From 2.5 to 6.5 s from
Three different configurations were tested (Table 4) to determine the release, the vertical acceleration becomes negative because the thrust
which one achieved a rate of descent lower than 4.57 m/s. The full-scale produced by the rotor is greater than the weight of the vehicle. After
configurations were tested by dropping them from a hot air balloon, re- approximately 6.7 s from the release, the vehicle stabilizes and enters the
leased from altitudes of 305 (1,000 ft), 244 (800 ft), and 250 m (820 ft), steady state of autorotation, characterized by zero vertical acceleration.
respectively. It was observed that after approximately 13 s, the Autobody started
precessing. This can be seen as an oscillation in the vertical acceleration
Flight test results and RPM (Fig. 15 and 17) values. Because of the precession, the thrust
decreased and the rate of descent increased, as seen in Fig. 16. The pre-
In each flight test, the vertical acceleration and rotor RPM data were cession was caused due to small dissimilarities between the blades, such
recorded as a function of time and the steady state values were validated as differences in mass due to the manufacturing process, and differences
with the analysis. Based on the results of each test, the rotor parameters in setting the blade pitch angle. Elimination of the precession will involve
were changed appropriately in order to achieve the goal of steady state tracking the rotor before launch, and will be an important practical design
descent velocity less than 4.57 m/s. consideration in terms of the final application of the Autobody.
OCTOBER 2007 DESIGN AND TESTING OF AN AUTOROTATIVE PAYLOAD DELIVERY SYSTEM 369

Figure 17 shows the rotor RPM as a function of time. Because of a Correlation with analysis
data capture error, the RPM data could not be acquired during the third
drop test. However, the value of the steady state RPM for the first two Analytical results are compared with data measured during the three
drop tests can be extracted from the recorded data. flights in Fig. 18. The analysis shows good agreement with the measured
It can be seen that the height obtained from the pressure transducer has descent velocity. The maximum error between analysis and experiment
a significant noise level (6 m (20 ft)). Therefore, the descent velocity is 22.2% and occurs in the third flight test. Note that the third test was
was obtained by first calculating a moving average of the height data, and performed with the lowest flapping stiffness k and it is expected that the
then performing a numerical differentiation. Note that as the sampling dissimilarities between the k of the blades had a larger effect. The mea-
rate of the on-board datalogger was limited, it was not possible to obtain sured descent velocity for the third test is 4.11 m/s (13.5 ft/s). Therefore,
the descent velocity by integration of the acceleration data. this Autobody design meets the requirement of a descent velocity lower
than 4.57 m/s. The measured RPM also shows good agreement with the
analysis, with a maximum error of less than 15%.

Test #1 Summary and Conclusions

800
Analytical and experimental studies have been carried out to investi-
gate the behavior of an autonomous vehicle with passive controls, called
600 the Autobody. It is designed to passively deploy a payload from an aircraft
by means of an autorotative rotor, with a steady state descent velocity
of less than 4.57 m/s. Only purely vertical autorotation is considered
RPM

as it results in a conservative design, and presents the simplest case for

400 Test #2 analysis.
A negative blade pitch was incorporated to start the rotor rotation in the
correct direction (i.e., leading edge first). A novel mechanism consisting
200 of a negative pitch-flap coupling in conjunction with a negative precone
was incorporated to passively increase the pitch of the rotor blades as the
rotational speed increased. These concepts ensured that the rotor rotation
was initiated in the correct direction and that enough steady state thrust
0 was generated to minimize the descent velocity of the system.
0 5 10 15 20 25 30 An analytical study, called the RPM sweep method, based on blade
Time (sec) element momentum theory, was developed to predict the behavior of
the Autobody in steady state of autorotation. The method consists of
Fig. 17. Comparison of rotor RPM data time history recorded during sweeping through a range of RPMs such that the rotor remains in the
different flight tests. windmill brake state, and calculating the thrust and torque of the rotor
at each RPM. The torque versus RPM curve is extrapolated to the zero
torque point, which indicates the RPM at which autorotation occurs.
9 Wind tunnel tests were performed on a scaled model rotor to establish
Experiment
Descent velocity (m/s)

Analysis the proof-of-concept of the rotor design, to investigate the influence of

2.5 % error 4.7 % error 22.2 % error the rotor design parameters on the steady state performance of the rotor
6 and to validate the analysis. The thrust and RPM were measured for
several values of wind velocities Vd , blade pitch, pitch-flap coupling,
and precone angles. It was observed that variation of thrust with Vd was
3 quadratic and the variation of RPM with Vd was linear. Furthermore, it
was observed that the experimental data and analytical predictions show
good agreement with a maximum error of around 25%.
0 A parametric study was carried out to investigate the influence of
Test 1 Test 2 Test 3
blade pitch, pitch-flap coupling angle, precone angle, and blade mass on
(a) Descent velocity
the steady state thrust and RPM. Based on this quantitative parametric
1000 study and on mechanical considerations, an optimum full-scale configu-
Experiment
Rotational velocity (RPM)

14.8% Analysis
ration was determined to achieve the objective of steady state descending
800 11.7% velocity less than 4.57 m/s. Instrumented flight tests on this configuration
600
were conducted to obtain rotor RPM, vertical acceleration, descending
velocity and altitude of the Autobody during the transient and steady
400 states of autorotation. The Autobody was dropped from a hot air bal-
loon and the data were recorded on an on-board microrecorder. It was
200 observed that a steady state rate of descent of 4.11 m/s was attained
for a rotor with a blade pitch angle o = 10 , a pitch-flap coupling
0
Test 1 Test 2 Test 3
angle 3 = 41 , and a precone angle p = 4 and having blades of
mass m b = 85.4 g each. This configuration met all the requirements and
(b) RPM
is therefore a successful design. Furthermore, the full-scale flight test
Fig. 18. Correlation of steady state descent velocity and rotor RPM experimental results showed good agreement with the analytical predic-
data with analytical predictions for the three flight tests. tions. The maximum error between analysis and experiment was 22.2%
370 A. BRINDEJONC JOURNAL OF THE AMERICAN HELICOPTER SOCIETY

for the prediction of the descending velocity during one of the flight 4
Prouty, R. W., Helicopter Performance, Stability, and Control,
tests. Krieger Publishing Company, Malabar, FL, 2002.
An analysis of the transient state of the flight should be developed to 5
Prouty, R. W., Torque Distribution in Autorotation, Vertiflite, 48,
describe more accurately the behavior of the Autobody during its flight. (2), 2002, p. 28.
Improvements must also be carried out in the design and manufacturing of 6
Lambermont, P., and Pirie, A., Helicopters and Autogyros of the
the 3 hinge. Repeatable and precise manufacturing process of the flexure World, Cassell, London, 1958.
hinge would ensure that all the hinges have equal flapping stiffness and 7
Bartz, J., and Miklosovic, D. S., An Experimental Analysis of Cam-
therefore allow each blade to flap and pitch in a similar fashion. This ber Effects of a 6-Bladed Flapped Autorotational Aerodynamic Decel-
would assure a tracked rotor and stable flight without precession. Finally, erator, AIAA-2003-2143, 17th AIAA Aerodynamic Decelerator Sys-
more reliable instrumentation should be used to perform further flight tems Technology Conference and Seminar, Monterey, CA, May 1922,
tests. Specifically, a higher sampling rate microrecorder should be used 2003.
and means of measuring height and descending velocity more accurately 8
Wernicke, R. F., Preliminary Tests of Model Spacecraft Rotor Land-
should be developed. ing System, Technical Report, Bell Helicopter Corporation, Fort Worth,
TX, 1959.
Acknowledgments 9
Kretz, M., Space RotorA European Project for Recovery of
Heavy Launch Vehicles (Reentry and Recovery of Space Vehicles in Re-
The authors acknowledge Mr. Seibert Murphy from Progeny Systems liable Shock Free Landings, Using Modified Helicopter Rotor for Hyper-
who provided the initial impetus for this research. We thank Dr. Marat sonic Flow and Thermal Problems), Spaceflight, Vol. 8, 1966, pp. 369
Tishchenko for his help and advice, Dr. V. T. Nagaraj for his advice, and 373.
Mr. Nitin Gupta for his help in the experimental testing. 10
Levin, A. D., and Smith, R. C., An Analytical Investigation of the
Aerodynamic and Performance Characteristics of an Unpowered Rotor
References Entry Vehicle, Technical Report, NASA TN D-4537, April 1968.
11
Levin, A. D., and Smith, R.C., Experimental Aerodynamics of a
1
Hudson, G. C., Roton Development and Flight Test Program, Rotor Entry Vehicle, Journal of Aircraft, Vol. 6, (4), 1969, pp. 330335.
12
AIAA-1998-5258, AIAA Defense and Civil Space Programs Conference Sirohi, J., Nagaraj, V. T., and Chopra, I., Design and Testing of a
and Exhibit Proceedings, Huntsville, AL, October 2830, 1998. Rotor for Autonomous Autorotation, American Helicopter Society 60th
2
Leishman, J. G., Principles of Helicopter Aerodynamics, Cambridge Annual Forum Proceedings, Baltimore, MD, June 710, 2004.
13
University Press, New York, 2000. Castles, W., Jr., and Gray, R. B., Empirical Relation between In-
3
Leishman, J. G., The Autogyro: The First Rotating-Wing Aircraft. duced Velocity, Thrust, and Rate of Descent of a Helicopter Rotor as
Part 1 : That Curious Phenomenon of Autorotation, Vertiflite, Vol. 49, Determined by Wind-Tunnel Tests on Four Model Rotors, NACA TN-
(2), 2003, p. 48. 2474, 1951.