American Institute of Aeronautics and Astronautics
1
Small UAV Research and Evolution in Long Endurance
Electric Powered Vehicles
Michael J. Logan
1
, Julio Chu
2
, Mark A. Motter
3
NASA Langley Research Center, Hampton, VA 23681
Dennis L. Carter
4
, Michael Ol
5
, Cale Zeune
6
USAF Air Force Research Laboratory, WPAFB, OH, 11111
This paper describes recent research into the advancement of small, electric powered
unmanned aerial vehicle (UAV) capabilities. Specifically, topics include the improvements
made in battery technology, design methodologies, avionics architectures and algorithms,
materials and structural concepts, propulsion system performance prediction, and others.
The results of prototype vehicle designs and flight tests are discussed in the context of their
usefulness in defining and validating progress in the various technology areas. Further areas
of research need are also identified. These include the need for more robust operating
regimes (wind, gust, etc.), and continued improvement in payload fraction vs. endurance.
Nomenclature
UAV = Unmanned Aerial Vehicle
fps = feet per second
RPM = Revolutions per minute
J = advance ratio (V/nD)
V = velocity (fps)
n = revolutions per second
D = propeller diameter
Ct = propeller thrust coefficient
42
Dn
T
Ct
∗∗
=
ρ
Cq = propeller torque coefficient
52
Dn
Torque
Cq
∗∗
=
ρ
η = propeller efficiency
ρ = air density
T = propeller thrust
BART = Basic Aerodynamics Research Tunnel
I. Introduction
oday’s small UAVs are the result of an evolution in the various enabling technologies that compose the
vehicle and the processes used in its design. These technologies include power storage improvements,
innovative motor design, avionics miniaturization, design and optimization techniques, and others. Evolution in each
of these areas as they apply to the small UAV are discussed below.
1
Head, Small Unmanned Aerial Vehicle Laboratory, Langley Research Center, Hampton, VA, Member AIAA.
2
Aerospace Engineer, Langley Research Center, Hampton, VA, Member AIAA.
3
Senior Controls Researcher, Langley Research Center, Hampton, VA, Member AIAA.
4
Senior Engineer, AFRL/VA, WPAFB, OH, Member AIAA.
5
Senior Engineer, AFRL/VA, WPAFB, OH, Associate Fellow AIAA.
6
Aerospace Engineer, AFRL/VA, WPAFB, OH, Member AIAA.
T
American Institute of Aeronautics and Astronautics
2
II. Power Storage Improvements
Many small UAVs are electrically powered. Furthermore, for these electrically powered vehicles, the power
storage system, in most cases a battery, represents the largest component by weight in the vehicle. Improvements in
power storage represent the largest “target of opportunity” to decrease the weight of the vehicle and/or improve the
performance. Table 1 shows an evolution of rechargeable batteries used as the primary power source for several
small UAVs. While in general it can be said that chemistry improvements provide weight savings, other factors may
influence whether there is a net gain. For example, in the case of recent developments in rechargeable Lithium based
batteries, the radio-controlled model aircraft demands for higher current draw batteries has caused a decrease in the
total storage capacity for a given weight. Such tradeoffs are useful for applications such as 3-D aerobatic aircraft
where run time is limited and thrust-to-weight is a primary motivating factor. However, for longer endurance UAVs,
the surge current requirement is likely to be far less than 5C so there would be a net penalty involved in using the
higher current draw rated battery. As battery chemistry and manufacturing technologies improve, they will need to
be focused on the small UAV application needs in order to provide an overall improvement to the vehicle system.
Testing will also be required to ensure that these new battery types are capable of withstanding the duty cycles
intended for UAV applications
5
Battery
Date
Manufactured
Nominal
Voltage
Capacity
(Ah) Weight (g)
Energy
density
(W-h/kg)
Nickel Cadmium 1990 10 1.5 410 36.59
Nickel Metal Hydride 2003 12 3 590 61.02
Li-Ion AA cells 2001 12 3.4 288 141.67
Li-Ion Cylindrical cells 2006 7.4 2.9 174 123.33
Li-Poly 3C rated cells 2003 10.5 3.3 201 172.39
Li-Poly 10C rated cells 2005 11.1 8 486 182.72
Li-Poly 25C rated cells 2007 11.1 6.2 492 139.88
Table 1. Battery Pack history.
Energy Density
0.00
50.00
100.00
150.00
200.00
1985 1990 1995 2000 2005 2010
Year of Manufacture
Energy Density (Wh/kg)
Energy Density
In looking forward, there may be a practical limit to the use of secondary batteries as the primary power source
for the UAV. For example, there is a practical limit on the total vehicle weight for hand launching simply due to
human factors issues. As such, the vehicle weight cannot grow regardless of desired endurance. When this happens,
alternate energy storage systems, such as primary batteries or fuel cells, must be used.
III. Electric Motor Evolution
In addition to power storage evolutions, the primary propulsion means have also undergone an evolution. Figure
1. shows a picture of three electric motors designed for a similar application but using different technologies. The
motor on the left is a “traditional” brushed motor and gearbox circa 2001 that has a mass of 269g. The middle motor
is a brushless motor replacement which has a mass of 209g, a savings of 22.3%. The motor on the right is an
“outrunner” motor or external can motor which drives the propeller directly. This high torque motor eliminates the
necessity for a gearbox thus providing a further weight savings of 13.3% over the inrunner brushless motor. Also
seen is as a reduction in the number of moving parts, enhancing its reliability. Costs for these improved motors also
seem to be decreasing as their utility becomes more widespread amongst mass-market users.
One potential drawback to the outrunner motor appears in the form of integration. Since the majority of the
external surface is rotating, there is no direct way to attach a heat sink to aid in cooling, as would be the case for the
American Institute of Aeronautics and Astronautics
3
other motor types. This requires more consideration for cooling airflow and the associated impacts of that flow on
the overall system design.
Testing of these motors indicate that all three have similar initial performance characteristics at certain specific
design points. However, the brushed motor typically degrades more rapidly with use than the other two. During a
recent series of wind tunnel tests, the geared inrunner motor was tested with the same propeller as an outrunner
motor of the type shown. At 60fps, the motors were consuming similar power levels (168 watts vs. 173 watts),
produced similar net thrust values (1.15lb vs. 1.2lb.) and had similar propulsive system (i.e. combined motor and
propeller) efficiencies (55.7% vs. 56%).
IV. Avionics Miniaturization
One of the most dramatic size, weight, and power reductions for small UAV components has come from
avionics miniaturization. Not only have these systems become dramatically smaller, but they have also become more
capable. Table 2. shows a collection of inertial measurement systems from various time periods. Note both the
weight improvement as well as functionality have risen concurrently.
Table 2. Avionics Miniaturization
Unit
Year
produced Weight (g) Functions Picture
Exdrone 2-Axis
Wing leveler
1985 770
1 gyro, wings
leveling only
COTS
Autonavigation
5-axis
2002 159
3-axis gyro,
2-axis
accelerometer
LaRC Gen1 5-
axis IMU
2003 105
3-axis gyro,
2-axis
accelerometer,
airspeed,
altimeter, GPS
LaRC Gen2 6-
axis IMU
2004 54.5
3-axis gyro, 3
-
axis
acclerometer,
airspeed,
altimeter, GPS,
microphone,
temperature
Figure 1. Similar motors using different configuration technologies.
American Institute of Aeronautics and Astronautics
4
Continued improvement from technologies such as micro-mechanical machines (MEMS) appear to be
accelerating more functionality into smaller form factors. For example, the MEMS devices being used in Figure 2.
are all single-element devices, i.e. one axis gyro or accelerometer. MEMS devices are currently available which
package 3-axis accelerometer or dual-axis gyros in a single chip. This added functionality can either be used directly
to lower the part count and board surface area or it can be used to provide redundancy to improve reliability. Other
types of microelectronics, such as counters (for measuring RPM), analog-to-digital converters, sensors (pressure,
temperature), and others are also benefiting from advances in packaging and microcontroller improvements.
V. Design Improvements
Several improvements of a general nature have occurred in the last few years which have application to small
UAVs. Widespread commercial availability of inexpensive yet robust materials such as Expanded Poly-Propylene
(EPP), have served to offer the convenience of simple homogeneous structures with the durability of core-sheeting
multi-layered structure. In some cases, other foam types, beyond the traditional expanded polystyrene (EPS) foams
have begun to be used as primary structure. These foams include Depron, Zepron, Arcel, extruded polystyrene, and
polyurethane foam types.
Design methods are currently being developed to help improve the design optimizations of small UAVs. Figure
3. shows a comparison of one currently available electric motor propulsion prediction code with wind tunnel data.
Analysis indicates that the predicted vs. actual thrust values can vary by +/-25%. Variances in the predicted power
required to generate a specific thrust value vary by a similar amount. Optimization using the predicted values could
easily lead to poor optimization in the actual system.
Development of a more robust propeller analysis and/or design capability for this class of vehicle is clearly needed.
One such development is being undertaken by AFRL. Motivations for a new development include, first the need: the
lack of thoroughly vetted non-proprietary propeller design codes, and the relevance of those codes to the low-
Reynolds number flowfields encountered by the propeller blades of small UAVs; and the means: wind tunnel and
especially static thrust-stand tests of off-the-shelf propellers driven by electric motors are in principle
straightforward and amenable to university-type experiments. Merchant and Miller
1
and Brandt
2
tested large
collection of off-the-shelf propellers designed for radio-controlled hobby aircraft, with electric or internal-
Fi
g
ure 3. Com
p
arison of
p
redicted vs. wind tunnel measured thrust.
Prediction vs. Wind Tunnel Data
0.00
0.50
1.00
1.50
2.00
2.50
0 20406080100
Airspeed (fps)
Net Thrust (lb.)
Graupner 10x8
Measured
E-Calc
Aeronaut 10x8
E-Calc
Graupner 10x8
Aeronaut 10x8
Measured
American Institute of Aeronautics and Astronautics
5
combustion engines. The present work includes a related study at the Langley Aeronautical Research Center’s Basic
Aerodynamics Research Tunnel (BART). A standard 6-component internal balance was used to measure thrust
(axial force) and torque (rolling moment) from a series of propeller-motor-speed controller combinations. These
were then compared with a spreadsheet-driven analytical prediction based on momentum and blade-element theory.
Figure 4 compares the BART experimental data, the analytical prediction and experimental data from BrandtError!
Bookmark not defined., for the Graupner 10” diameter 8” pitch “cam slim” propeller
3
designed for electric motors.
The analytical prediction used blade chord distribution as reported by Brandt, and two alternative approaches to
twist distribution: that reported by Brandt, and a standard twist distribution implied by the manufacturer’s value of
blade pitch. The prediction assumes blade sectional airfoil properties based on XFOIL
4
computations for a NACA
2412 section at Re = 100,000. Such an approach is patently flawed on numerous grounds; Reynolds number will
vary significantly depending on flight speed, propeller rotation speed and the blade station; the local flow
disturbance intensity – which affects airfoil lift and drag considerably – is unknown; and the actual propeller blade
section is unknown and itself varies with blade station. However, if the deep-stall lift and drag curves are modeled
with curve fits, some experience suggests that blade section and Reynolds number effects are relatively subordinate
to chord and twist distribution. That is, the accuracy of the prediction depends far more on capturing the correct
chord and twist distribution, than on sectional profile and Reynolds number.
BART wind tunnel data in Figure 4 are a composite of four separate runs. Repeatability is good at every setting
except zero free-stream (J = 0), but inferior to Brandt’s dataError! Bookmark not defined.. Brandt’s data clusters
around higher torque coefficient values than for the other data sets, thus giving a low propeller efficiency.
Analytical prediction using the nominal pitch value shows excellent agreement with BART thrust coefficient data.
Both analytical approaches grossly underpredict low-J torque, but prediction based on Brandt’s twist data shows a
good fit at higher J. Efficiency is overpredicted by analysis, due to the underprediction of torque – though, again,
the fit is reasonable near the maximum values of J.
0 0.25 0.5 0.75 1
J
0
0.02
0.04
0.06
0.08
0.1
0.12
LaRC BART data
Brandt
Analysis, fitted twist
Analysis, nominal twist
C
T
0 0.25 0.5 0.75 1
J
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
LaRC BART data
Brandt
Analysis, fitted twist
Analysis, nominal twist
C
Q
