48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference<br>15th AIAA 2007-2234

23 - 26 April 2007, Honolulu, Hawaii

Development of a Long-Range Small UAV for Atmospheric

Monitoring

J. B. Kosmatka*
University of California, San Diego, La Jolla, CA 92093-0085

A small low-cost all-composite Unmanned Aerial Vehicle (UAV) was developed to carry
atmospheric monitoring instrumentation on long endurance (30+ hour) missions. The
design process was initiated by determining the payload size and weight requirements, and
then selecting an engine that would provide the needed power and fuel economy for the
flight duration. The fuselage was designed to carry the payload, flight hardware (auto-pilot,
Iridium satellite communication) equipment, engine, landing gear, and required fuel tank.
A computational fluid dynamics (CFD) software package (FLUENT/GAMBIT) was used to
shape the fuselage and tail configuration followed by quarter-scale wind-tunnel testing. The
resulting aircraft was fabricated using the RTM and VRTM process using carbon/epoxy
composite materials. The aircraft is flying and the performance properties were measured.

I. INTRODUCTION

T he role of the Unmanned Aerial Vehicle (UAV) has increased in the past several years from military
surveillance to scientific observations. In the science community, UAVs are used to gather data for geographic
surveys, agriculture inspections, animal migrations, and long-range atmospheric sensing. The appeal of UAVs arose
from lower operational costs and longer flights, since there is no need for a pilot for these long missions. Specialty
UAV’s are being developed for most scientific applications because the payloads and missions are unique. For
example to monitor global warming and changes to the Earth’s albedo it is desired to have a low-cost aircraft
capable of carrying a range of scientific instruments (5 kg, 700 in3) at speeds of (60-80 mph) for a long endurance
mission (30+ hours) while at a low altitudes (6000-10,000 ft SL). At this altitude, the aircraft can fly below and
through the low-level clouds to monitor the solar radiation and the presence of aerosols which contribute the most to
the planetary albedo changes [1]. The Pacific Basin provides science researchers with various conditions for
monitoring aerosol changes due to pollution from Asia and North America and pristine air from the equatorial
marine regions.
The needed on-board scientific equipment includes: Condensation Particle Counter (CPC), a Cloud
Condensation Nuclei Counter (CCN), an Optical Particle Counter (OPC), a Cloud Droplet Probe (CDP),
Pyranometer, a Spectral Radiometer, a Data Acquisition (DAQ) and an Aerosol Inlet. Each of these sensors and
DAQ require a degree of miniaturization to reduce weight and size. In addition, an onboard computer is required to
record and store the data on flash memory and send samples via satellite (Iridium communications network) to the
ground station. A pusher aircraft design is required so that an aerosol inlet attached to the fuselage nose can sample
clean undisturbed air. Moreover, the pyranometers sensors must be attached horizontally above and below the
aircraft and thus a landing gear system is required to protect the equipment during take-off and landing.
There are several small commercially available UAVs that could potentially be used for atmospheric monitoring.
They include the Aerosonde, the RnR APV-3, and Advanced Ceramics Research’s Manta B. The Aerosonde UAV
was initially developed with the sponsorship of US Office of Naval Research (ONR) and Australian R&D to
conduct meteorological observations in remote and inaccessible regions. This aircraft is capable of flying for nearly
30+ hrs when carrying no payload for nearly 20 hrs when carrying small (2 Kg) payloads. The aircraft is powered
by 1.6HP fuel injected engine which allows for a much higher flight envelope (up to 23,000 ft). The RnR APV-3
UAV was developed by RnR Products for surveillance flight research under sponsorship by NASA-Ames and
Lockheed-Martin. The aircraft is powered by large, 86cc 2-stroke motor in a conventional tractor configuration with
endurance of 8 hours. The Manta-B UAV is an inexpensive UAV developed by Advanced Ceramics Research,
which is capable of carrying a 5 kg payload at 60-80 mph on a six hour endurance mission. These three small
UAV’s clearly offer the opportunity to do atmospheric monitoring flight research, but none of them are capable of
*
Professor, Department of Structural Engineering, MC-0085, Associate Fellow

1
American Institute of Aeronautics and Astronautics

Copyright © 2007 by J. B. Kosmatka. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
flying the aforementioned atmospheric mission profile. Larger military-developed UAVs, such as the General
Atomics Predator-B could be adapted for this application but it is considered a very expensive option that is really
an over-design with its 400 kg payload capacity, 10,000 lb take-off weight, 250 mph cruise speed, and 50,000 ft
flight envelope. All four commercial candidates are presented in Figure 1.

Figure 1. I Aerosonde, RNR APV-3, ACR Manta, and General Atomics Predator-B

The University of California, San Diego with support from the National Science Foundation, has a developed a
small UAV that specifically meets the needs of the atmospheric science research community. This aircraft was
initially designed around the payload requirements (5 kg, 700 in3) and mission requirements (30+ hours, 60-80 mph,
10,000 ft altitude). Numerous design iterations were performed based upon engine performance testing (horsepower
versus specific fuel consumption), since there are few reliable 2- and 4-stroke small (1 to 6 Hp range) engines that
are capable of efficient long duration reliable flights. The engine selection was critical to sizing the fuel tank, wing
area, aircraft structure, and flight performance properties. The aircraft shape was designed using a computational
fluid dynamics (CFD) software package (FLUENT/GAMBIT) and then verified with quarter-scale wind-tunnel
testing. The resulting aircraft has two different wing span (11.7 ft and 13.7 ft) configurations for different mission
requirements. The final aircraft was fabricated with laminated carbon/epoxy composite materials. Finally, the
needed electronics (servo’s, on-board data acquisition, auto-pilot, Iridium satellite communications and data
acquisition) were developed and installed. The aircraft design is presented in Figure 2. A comparison of the UCSD
aircraft specifications with commercially available aircraft is presented in Table 1. The resulting aircraft was
initially fight tested over a six-month period to measure its performance and compare them to the predicted values.
The resulting aircraft has since flown multiple 10+ hour missions and plans are underway to fly a long range (35+
hours) mission.
Table 1: Comparison of Aircraft Specifications [2].

General Atomics
Aerosonde RNR APV-3 ACR Manta Predator-B UCSD Sterna
Range (miles) 1875 400 360 7500 2300
Endurance (hrs) 30 8 6 30+ 35
Cruise (mph) 50-94 60-90 45-103 250 50-80
Ceiling 23,000 10,000 10,000 40,000 10,000
Take-Off Weight (lb) 33 54 52 10,000 70
Payload Weight (kg) 3 10 7 400 5
Wing Span (ft) 9.5 12.3 8.75 66 11.7 & 13.7

2
American Institute of Aeronautics and Astronautics
 Figure 2: UCSD Sterna UAV

II. WIND TUNNEL TESTING

Once the design sizing and aerodynamic modeling exercises were completed, a (1/4) scale model was tested in the
wind tunnel to validate the CFD calculations (lift, drag, pressure distributions) and to assess the performance of the
aircraft with three different tail designs. All testing was performed at the San Diego State University low speed
wind tunnel which has a test section of 45 inches wide by 32 inches tall by 68 inches long. The inlet contraction
ratio is 6.3:1 which allows for testing speeds up to 180 mph and it utilizes turbulence damping screens to reduce the
inlet turbulence intensity factor in the test section to 1.27. A six-axis external balance is used to measure three
forces and three moments at one point on the body. A ¼ scale model of the aircraft (33.5 inch wing span) was
fabricated to allow five inches between the aircraft and any wall surface to eliminate wall turbulent boundary layer
interactions. Based upon aerodynamic (Reynolds number) scaling and wind tunnel speed limitations, experimental
results up to true vehicle speeds of 45 mph could be studied. The aircraft wind tunnel model was fabricated in three
sections: (1) combined wings and fuselage, (2) tail booms, and (3) tail, where the wings, fuselage and tail are
fabricated using three-dimensional SolidWorks CAD drawings and stereolithography (rapid prototyping by
selectively curing a liquid photopolymer by exposing it to an ultraviolet laser). Three different tail configurations
were fabricated to study the performance of an H-tail, inverted V-tail, and the proposed Lambda tail. The three
different tail designs are presented in Figures 2 and 3. The wind-tunnel model tail booms are fabricated of
Aluminum tubing with threaded ends to ease in switching the different tail configurations. The fuselage model also
has seven pressure ports molded into the fuselage and wings in order to measure pressure at identified low pressure
(high velocity) locations. One pressure port is at the nose of the aircraft to sample stagnation pressure and the
remaining six are distributed along the fuselage centerline and wing upper quarter chord. Each port is simply a
1/16” vein molded in to the model with an exit at the fuselage tunnel mount that are connected to the wind tunnel
pressure transducers.

Figure 3: H-Tail and Inverted V-tail designs

3
American Institute of Aeronautics and Astronautics
 The wind tunnel model with all three tail configurations is shown in Figure 4. The rear of the aircraft was
mounted to a threaded rod that could be turned from outside the tunnel to move up and down, changing the angle of
attack of the model without opening the tunnel. A digital level placed on the boom allowed correlation between
these numbers and the different angles of attack of the model. Once the model was secure, the tunnel was closed and
testing could begin. For each tail configuration (H tail, inverted V tail, lambda tail), the model was tested at two
Reynolds numbers. Testing speeds are shown in Table 2. Data recording was done on a computer connected to the
balance using a custom LabVIEW program written specifically for the tunnel. After data was recorded, the model
was removed from the tunnel and the mounting apparatus remained. Data was then measured at the two different
speeds and all angles of attack (or extension of the tail mounting point in this case) to get baseline values for the
forces on the balance. The data collected during this tare run is subtracted from the data collected while the model is
in the tunnel to get the actual forces on the balance caused by the airflow over the model. After the forces were
measured, tubing was connected to the pressure tap ports and run down the mounts and out of the tunnel to the
pressure transducer.

Figure 4: Wind-tunnel model with H-, inverted V, and Lambda-tail configurations.

Table2: Wind Tunnel Speeds

The aircraft lift (CL) and drag (CD) coefficients versus angle of attack and the three different tail configurations at
a Reynolds number of 415,000 is presented in Figure 5. The lift coefficient (CL) of both the H-tail and Lambda tail
are nearly identical over the entire range where a maximum value of 1.4 (no flaps) is observed at 10 degrees. At
greater angles a slight wing tip stall was observed. The lift coefficient of the inverted V-tail had the same slope (as
expected) as the other tails, but the curve was shifted down and to the right. This shifting can be visualized as a
fixed loss of aircraft lift throughout the entire range that is associated with the lower lift produced by the tail
surfaces. The drag coefficient (CD) versus angle of attack is nearly identical for all three tails, where the H-tail and
the inverted v-tail recorded the lowest minimum drag coefficient of 0.028, whereas the Lambda-tail recorded a
minimum value of 0.03. These experimentally measured values were nearly half that the drag value calculated by
the CFD simulations (CD = 0.058). It was noted that the CFD predicted drag was greater for all angles of attack,
getting much worse as the angle of attack increased.
In Figure 6, the drag coefficient (CD) is plotted versus the lift coefficient () for the three different tail designs,
where the H-tail had the lowest drag over the widest lift range, whereas the inverted V-tail had low drag for low lift
but greatly increased as the lift coefficient increased. The lift-to-drag (L/D) ratio is also presented, where all three
tail designs are optimized at 6 degrees angle of attack with the H-tail producing a value of 30 followed by the
Lambda tail at 26 and the inverted H-tail at 25. All three of these measurements are excellent. The lambda tail
appears to offer a good compromise between the H- and V-tail but having the advantage that with its four surfaces
(two horizontal and two inclined) it offers plenty of elevator and rudder redundancy. Thus if a servo should fail in
the tail, there is a very good chance the aircraft will still be controllable.

4
American Institute of Aeronautics and Astronautics
 Figure 5: Aircraft lift (CL) and drag (CD) coefficients versus tail type (Re=415,000).

Figure 6: Aircraft drag (CD) versus lift (CL) and (L/D) versus tail type (Re=415,000).

III. AIRCRAFT FABRICATION

The aircraft is fabricated from intermediate modulus carbon fiber and epoxy using various manufacturing
techniques including VRTM, matched molding, and wet lay-up. Material stiffness and strength allowables were
measured by performing ASTM tests on representative (minimum of six) coupons. All fabrication steps were
directed and supervised by experienced aircraft fabricators (Aaron Valdes and Don Vangundy). The wings were
fabricated in several steps using matched polished
Aluminum tools for the upper and lower sandwich
wing skins. Composite fuselage tools were
fabricated from machined fuselage plug halves.
The upper and lower fuselage sandwich skins
were fabricated with flange attachments for the
payload tray, the landing gear and housings, the
forward bulkhead and aft firewall. Separate
hatches (access points) were fabricated for
instrumentation access, payload access, engine
access, and landing gear bay doors. In Figure 7,
the payload cover is removed to show the payload
tray with ample room for instrumentation.

Figure 7 Payload bay

5
American Institute of Aeronautics and Astronautics
 IV. PERFORMANCE FLIGHT TESTING
The aircraft was completed during the summer of 2005 and the first test flight was conducted on August 28,
2005 at Camp Pendleton Marine Base, California. Aaron Valdes is the test pilot. During the next six months (75
hours of flight testing), the aircraft was flown under a wide variety of payload and atmospheric conditions so as to
measure its performance and compare with predicted values. During its initial flights a larger engine (Zenoah-80cc)
was installed to insure it had enough power for short runway take-offs, to recover from planned testing stalls, and for
flight stability studies. Later smaller engines (3-W 28 cc and 3-W 56 cc) and various carbon propellers were
evaluated for fuel economy and performance. The three different tail configurations were fabricated, flight tested,
and evaluated. In Figure 8, the aircraft is shown during its first test flight (August 28, 2005) with an H-tail
configuration and the landing gear down and in a later flight with an inverted V-tail configuration and the landing
gear retracted. During the test flights the aircraft performance properties (vehicle speed, altitude, GPS location, air
temperature, pressure, engine temperature, engine RPM) were measured, recorded, and transmitted wirelessly (2.4
GHz) back to the ground station for immediate review.
During the initial flights, most of the efforts concentrated on assessing the ground characteristics and stability of
the aircraft during take-off and landing (i.e. take-off and landing distances, rate of climb, sink), followed by in-flight
measuring of the aircraft stall speed with the flaps up (cruise condition) and flaps down (take-off and landing
conditions), followed by measuring the rate of climb properties, and finally testing the yaw and roll stability and
damping by performing doublets. All of these initial flights and tests were performed with the landing gear in the
down position to reduce the risk of accidents. The aircraft gross take-off weight was measured as 46.67 lbs with a
CG location at 22.8% MAC. The low aircraft weight was a result of carrying no instruments and very little fuel.
Some lead mass was added to the fuselage nose to property place the CG. A take-off speed of 38 mph was
measured with a traffic radar gun and verified by the on-board data acquisition system. A take-off distance of 234 ft
was measured. A stall speed of 34 mph with the flaps and landing gear down, which resulted in a maximum lift
coefficient at stall of 1.62. A cruise speed of 59.6 mph and a maximum rate of climb of 15.2ft/sec (37.7 mph) were
measured were the landing gear in the down position.

Figure 8: Flight Testing. Initial flight with an H-tail configuration and landing gear down (left). Later flights
with an inverted V-tail configuration and landing gear retracted (right). The Lambda-tail installed
on the aircraft.

To test aircraft stability doublets in pitch, roll, and yaw were performed by the pilot. The results of this test are
qualitative because they are based on the pilot observations. Doublets are characterized by moving a control stick
up, down, up and down, while holding the stick for a second at each of the four stations. The amplitude of

6
American Institute of Aeronautics and Astronautics
disturbance is increased for each test case: quarter-stick, half-stick and three-quarter-stick. Then the stick is
neutralized and controls are used to stabilize the aircraft. During the first test flight, the doublets were performed
with flaps at cruise condition. The pitch doublet at quarter-stick experienced a wing tip stall and caused the aircraft
to roll and required immediate correction to return to level flight. Roll doublets at cruise were performed and the
aircraft passed the test. Lastly, yaw doublets were performed and the aircraft was deemed well damped. Lastly, the
stall was checked at various speeds.
Once all this was completed and flight confidence was achieved, then all the tests were repeated with the landing
gear retracted. A smaller engine (3-W 56cc) was installed with a lower weight configuration. In this configuration,
a stall speed of 29 mph was recorded with the gear down and 34 mph with the landing gear retracted. A top speed of
131 mph was recorded. An example of the GPS snail trail of a flight test at Camp Elliot Navy Base near UCSD is
presented in Figure 9.

Figure 9: An example of a GPS snail trail during flight testing (Camp Elliott, San Diego, CA).

V. Conclusion
A small low-cost all-composite Unmanned Aerial Vehicle (UAV) was developed to carry atmospheric
monitoring instrumentation on long endurance (30+ hour) missions. The design process was initiated by
determining the payload size and weight requirements, and then selecting an engine that would provide the needed
power and fuel economy for the flight duration. The fuselage was designed to carry the payload, flight hardware
(auto-pilot, Iridium satellite communication) equipment, engine, landing gear, and required fuel tank. A
computational fluid dynamics (CFD) software package (FLUENT/GAMBIT) was used to shape the fuselage and tail
configuration followed by quarter-scale wind-tunnel testing. It was shown that there is a significant advantage of
using either a standard H-tail or a new proposed Lambda-tail as opposed to the more common invert-V tail. The
resulting aircraft was fabricated using the VRTM process using carbon/epoxy composite materials.

Acknowledgements
The author wishes to acknowledge the support of National Science Foundation (Jay Fein, monitor). The author
wishes to thank the team of UCSD graduate (Mark Shtayerman, Ed Reed, Amilio Aviles) and undergraduate
students (David Klein, Chris Franich, Jon Carter, Wen Hsu), our world-class pilot and flight test engineer (Aaron
Valdes), fabrication and flight test expert (Don Vangundy) and industry consultant (Jose Panza).

References
1
Ramanathan, V., Crutzen, P. J., Kiehla, J. T., & Rosenfeld, D. 2001. Global Albedo Project – How do Aerosoles and Clouds
Regulate the Global Albedo. Science, December, 2119–21244.
2
Jackson, Paul, Munson, Kenneth, & Peacock, Lindsay (eds). 2003. Jane’s All The World’s Aircraft, Jane’s Information
Group Limited.
3
Raymer, Daniel P. 1999. Aircraft Design: A Conceptual Approach. Third edn., American Institute of Aeronautics and
Astronautics, Inc.

7
American Institute of Aeronautics and Astronautics