Вы находитесь на странице: 1из 15

AIAA 2017-0012

AIAA SciTech Forum


9 - 13 January 2017, Grapevine, Texas
55th AIAA Aerospace Sciences Meeting

Design, Construction, and Flight Testing of the World’s


Fastest Micro-Scale Quadcopter

Richard B. Bramlette,1 Taylor A. Johnston,2 and Ronald M. Barrett3


University of Kansas, Lawrence, KS, 66045

In recent years, the quadrotor configuration has resurfaced, enabled by access to low-cost,
high performance sensors and electric motors. At the same time, the “Micro Aerial Vehicle”
(MAV) world has yet to fully take advantage of the platform in transition flight. This paper
involves a new class of transitioning aircraft proposed in 2013 (denoted the XQ-139) that
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

circumvented the complexity and weight penalties of tilt-rotor/wing aircraft and was scaled
down to explore the aircraft family’s full design envelope. Specifically, it details the scalability
of the general XQ-139 design involving the development of a micro-scaled version of the
originally-proposed XQ-139A. It further describes the design evolution of the aircraft’s
aerodynamic shell from the perspective of drag reduction and exploiting the effects of
structural scaling to build a monocoque airframe which provided both smooth curvature and
a structural capability better than carbon composites in terms of both effective strength-to-
weight and survivability. Data from controlled bench tests of the small-scale rotors were
investigated. The pertinent flight test parameters that could be experimentally verified are
discussed. This includes a thorough investigation of boosting the aircraft’s top speed with a
commercially-available solid rocket motor. This achieved a top speed of 133 mph (11,700
ft/min) verified photogrammetrically and is believed to be the world’s fastest quadcopter,
particularly at this scale having a motor-to-motor span of just 4.75 in. (12 cm). The paper
concludes with a discussion of the direct potential applications of the design as a consumer
technology and future capabilities of this unique small-scale architecture.

Nomenclature
A = rotor disk area
ABS = acrylonitrile-butadiene-styrene (thermoplastic)
COTS = commercial, off-the-shelf
DC = direct current
FM = rotor figure of merit
KU AAL = University of Kansas Adaptive Aerostructures Lab
L = rocket lift capability
m = mass
MAV = micro-aerial vehicle
T = thrust force
P = power
PLA = polylactic acid
UAV = unmanned/uninhabited aerial vehicle
V = velocity
VTOL = vertical take-off and landing
W = weight
Ω = propeller angular speed

1
Post-Doctoral Researcher, Aerospace Engineering, 2120 Learned Hall, AIAA Student Member.
2
Undergraduate Student, Aerospace Engineering, 2120 Learned Hall, AIAA Student Member.
3
Professor, Aerospace Engineering, 2120 Learned Hall, AIAA Associate Fellow.
1
American Institute of Aeronautics and Astronautics

Copyright © 2017 by Richard Bramlette, Taylor Johnston, Ron Barrett. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
I. Introduction and Background

T HE history of the fixed-wing aircraft runs largely parallel to the development of the rotary-wing aircraft. The only
reasons for the delay in development of the helicopter and vertical flight in general is owed largely to the greater
power requirements and additional control complexity compared to fixed-wing aircraft.1 The modern helicopter
provides a critical capability that conventional fixed-wing aircraft cannot – vertical take-off and landing (VTOL).
Fixed wing aircraft must move from one prepared airfield to another whereas VTOL aircraft can operate from nearly
any improvised flat ground. They also have the unique ability to hover on-station providing critical mission capabilities
like precision cargo delivery, reconnaissance, search and rescue, and close air support to name a few. Fixed-wing
aircraft are limited in this ability by their stall speeds below which the aircraft cannot provide enough lift to remain
airborne. Instead, to loiter over a mission area, fixed-wing aircraft must orbit the site or make periodic strafing passes
both of which limit the amount of time the aircraft is actually mission-capable.
In terms of raw performance, however, fixed wing aircraft have longer ranges and more endurance than
helicopters because they only need to use power to overcome a small drag force whereas helicopters must use power
to generate all the aircraft’s lift force. Conventional helicopters are also at a strong flight speed disadvantage with the
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

rotor system ultimately limiting top speeds where fixed wing aircraft are fundamentally only limited by the amount
of onboard power and eventually high surface temperatures at supersonic Mach numbers. It is clear, then that an
aircraft that could exploit the higher top speeds of a fixed-wing aircraft but retain the VTOL capability of helicopters
should represent a formidable multipurpose vehicle and potentially allow new mission capabilities. Such an aircraft
would have to take-off vertically, transition to high speed forward flight to cruise to a destination, and transition back
to a vertical configuration. This sort of mission profile can significantly extend the range beyond that of a conventional
helicopter and shorten response times of both fixed-wing aircraft and conventional helicopters.
This is all based on the development histories of more than a century of manned flight. Over the last few decades,
however, development of small UAVs and micro-aerial vehicles (MAVs) has explored similar design goals at much
smaller scales. These aircraft not only probed the effects of scaling on aerodynamics, onboard power requirements
and capabilities, structural strength, stability and control and autonomous flight, some resulted in aircraft designs with
potential mission capabilities.2-9 The majority of MAV designs, however, conform to conventional fixed-wing2-4,
rotary-wing5, 6, or flapping-wing7 aircraft designs at small scales including investigation into autonomous swarms of
micro-scale multirotors. Unlike manned aircraft, however, there has been little investigation into transitioning MAVs.
This research effort explores one such concept of a unique transitioning MAV based on a recently-proven transitioning
UAV (referred to as the XQ-139) designed for high speed flight.

A. Transitioning Aircraft
The design of the XQ-139 transitioning aircraft draw on the long and varied development history of manned
transitioning aircraft which is discussed in detail by many authors1, 10-12 and from the focused perspective of top speed
in this work’s companion paper.13 That discussion is summarized visually in Figure 1. These two charts show the disk
loading (the amount of weight that can be lifted by the available rotor area) of a wide range of VTOL designs compared
to their power loading (the amount of weight a VTOL aircraft can lift normalized by the amount of power required to
lift it) and top speed. The power loading can be thought of as an efficiency parameter with a high power loading
corresponding to greater range and endurance. The trend with disk loading shows that in general, aircraft with larger
rotors tend to have greater hover efficiency. These larger rotors, however, more strongly impose drag penalties (in
addition to the typical tip sonic and blade retreating stall) limiting their top speeds.
Compound helicopters like the AH-56 Cheyenne are able to unload their rotors with small fixed wings sized for
cruising flight which alleviates some of the blade retreating stall limitations and produce extra forward thrust with a
separate propeller or turbine.1 The extra propulsion weight required for high speed flight necessarily cuts into the
design’s payload capability, the top speed improvement is minimal, but the hover efficiency of this design can clearly
be competitive with the conventional helicopters. On the other end of the spectrum are the jet-powered VTOLs,
particularly the jet tailsitters and the fighter jets with lift-fan and/or thrust-vectoring capability. These aircraft have the
highest top speeds of any VTOL aircraft owing to their low drag profile in cruise but necessarily have the highest disc
loadings and thus have the lowest hovering efficiency. The latter was so detrimental that aircraft like the Harrier was
intended only to take-off and land vertically as a means of operating without runways. It was only capable of short
hovering flights limited by the amount of onboard fuel. The tiltwings and tiltrotors are an attractive alternative from
the perspective of both hovering efficiency and top speed, but again suffer from significant weight growth and
complexity associated with maintaining a “level deck.” UAVs and by extension MAVs rarely have tilt-sensitive
onboard cargo or personnel that require a level cargo volume thus the necessity of a level deck is erroneous and the
associated weight growth and complexity cannot be justified.

2
American Institute of Aeronautics and Astronautics
12 700
Helicopters Harrier GR.3
Compounds
Multirotors
600
10 Jet Tailsitters XFY-1
Vectored Jets
Power Loading, WMAX VTOL/P (lbf/hp)

Tilt Rotors / Fans XFV-1


XH-51 Tilt Wings 500

Max Airspeed, VMAX (knots)


8 C.450
Prop Tailsitters X-19
Ideal Inflow
XC-142
400
XV-15
6
XV-15 V-22 X-13
300 X3
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

AH-56 XH-51
4 AH-56
V-22 200

X3 X-13
2
X-19 100
XFY-1
XC-142 C.450
XFV-1
0 0
1 10 100 1000 1 10 100 1000
Disk Loading, T/A (lbf/ft2) Disk Loading, T/A (lbf/ft2)
Figure 1: Power Loading (Gross Takeoff Weight Per Unit Shaft Power) and Maximum Airspeed Data Trends with Disk
Loading (Hover Thrust per Rotor Disk Area) for a Variety of VTOL-Capable Aircraft

The tailsitters, named for the way they sat on their tails with their noses pointed upward for vertical take-off and
landing, are comparatively lighter and simpler. Operating off of ships at sea, they were intended to take-off vertically,
shortly thereafter pitch over into forward flight, and perform a high speed cruise to intercept attacking aircraft. 14, 15
Their position in Figure 1 with regard to hovering efficiency and top speed is somewhat misleading. The configuration
was never explored beyond the limited development of the manned XFV-1 and XFY-1 aircraft built by Lockheed and
Convair respectively. These two competing designs are shown in Figure 2. They were designed with top speed as the
ultimate goal intending only to take-off and land vertically with no on-mission vertical flight. As a result, like the jet
VTOLs that came afterward, they suffered from high disk loadings and low hovering efficiency. The tailsitter concept,
however, remains partially untapped. A design that mixed this planform architecture with that of the more efficient
multirotors should be able to boost the efficiency of the tailsitter design’s low blockage area and cruise drag profile
necessary for high top speed.

Figure 2: Tailsitters On the Ground and in Flight: The Lockheed XFV-1 “Salmon” with Temporary Landing Gear for
Conventional Takeoff and Landing (Left)16 and the Convair XFY-1 “Pogo” (Right)16

In just the last few years, there has been a resurgence in VTOL UAV designs that consider transition capabilities.
The most famous of these are the “hybrid quadrotors”17, 18 that are functionally just small-scale compound helicopters.
3
American Institute of Aeronautics and Astronautics
They use a set of rotors for vertical take-off and landing, and transition into forward flight (while keeping the deck
level) at which point the hover motors are stopped and conventional fixed-wing flight is supported by wings and
horizontally-oriented propellers. Although these designs present low development risk with decoupled flight
propulsion, they suffer from the same increment in empty weight that full-scale compound helicopters do. They also
appear to rely on comparatively large wings which incrementally increases cruise drag and weight. A few alternatives
to this design do exist,19-23 but all strangely appear to still rely on large wing areas for aircraft that are supposed to
transition to a hover on vertical thrust, not wing lift.
A new multirotor design was conceived of that properly sized its wings for a low drag, high speed cruise, sized
its powerplant and installation for low blockage, efficient hover, and was still capable of pitching its whole body like
the tailsitters to convert between vertical and horizontal flight modes. That new design is the XQ-139 (shown in Figure
3) and is discussed in more detail in this work’s companion paper. 13 The design draws from years of experience
developing the XQ-138 coleopter-based convertible UAV which successfully demonstrated hovering and forward
flight mission capabilities in December of 2001 powered by an internal combustion engine. The newer all-electric
XQ-139 was instead built on a multirotor backbone yet was designed to provide the same kind of mission capabilities.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

This makes it similarly ideal for a wide variety of UAV roles including infrastructure inspection, search and rescue,
battle damage assessment, and collocated close air support.

Figure 3: A Comparison of the XQ-138 Coleopter-Based Convertible UAV (Left) and the Original-Scale XQ-139
Multirotor-Based Convertible UAV (Right)

B. Scaling Down the XQ-139 Design


The baseline XQ-139 design is fully-scalable as shown in Figure 4. The original airframe has a wingspan of 8.0
in. (20.3 cm) and a gross take-off weight of 315 g (11.1 oz) lifted by just 175 W (0.235 hp). This scaled up to a fully
mission-capable design with a wingspan of 19.0 in. (48.3 cm), a gross take-off weight of 6.8 kg (15 lb) lifted by 1750
W (2.35 hp) as well as down to the MAV scale with a wingspan of 12.1 cm (4.75 in.), a gross take-off weight of 33 g
(1.16 oz) and 3.75 W (0.005 hp) of onboard power. All are capable of efficient hover due to their streamlined airframes
and low blockage areas particularly compared to conventional multirotors. They are also all capable of controlled
transition to sustained high speed horizontal flight operating on commercial, off-the-shelf (COTS) electronics.
The smallest design in this family of aircraft, designated the XQ-139µ for its micro-scale and nicknamed the
“QuadSparrow,” is the primary focus of this work with a more thorough discussion of the original scale (the XQ-
139A) left to this work’s companion paper.13 Scaling the design down brought not only unique design challenges but
also unique material requirements. At the size of the XQ-139A, the semi-monocoque carbon fiber skins were already
using the thinnest single-ply dry cloth available vacuum-bagged with minimal resin to maximize stiffness and
strength-to-weight and were still structurally overdesigned in regard to flight load cases. The wings were stiffened
with a combination of expandable foam injected into the airframe to simultaneously bond parts together and prevent
crushing and buckling failures. This was further augmented with embedded carbon fiber tube spars as primary load
paths. None of this was feasible for a scaled-down aircraft.
4
American Institute of Aeronautics and Astronautics
Scaled Up

Original
Design
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

Scaled Down

XQ-139B XQ-139A XQ-139µ


“QuadEagle” “QuadHawk” “QuadSparrow”
Figure 4: From Left to Right, The XQ-139 Family of Aircraft Showing Scaling Up to a Nominal Flight Mass of 2500 g
WTO (XQ-139B), Original Size at 250 g (XQ-139A) and the Smallest at 20 g (XQ-139µ)

The first prototypes were constructed from a carbon fiber/foam extrusion similar to the larger scale XQ-139
aircraft with engine pods at the tips of comparatively thinner wings but without the integrated carbon fiber spars. A
cutout was made between the main wing and empennage surfaces providing an open cavity for the propellers. This
configuration (shown with others of the QuadSparrow design evolution in Figure 5) performed adequately but was
clearly structurally over-designed. There was, however, nowhere to remove weight from the smooth and streamlined
micro-scale carbon shell. The thinnest carbon skins available were simply too thick.
The next iteration relied more heavily on experimenting with different foams eventually settling on the stiffness,
low density, and ready availability of depron (a closed-cell polystyrene foam). Small carbon spars were used to stiffen
1 mm thin sheets of depron foam bonded with cyanoacrylate glue. The combination of the thin, brittle, and stiff carbon
spars, mated to stiff foam, and bonded with similarly brittle cyanoacrylate glue resulted in an airframe was light with
enough surface area to transition to horizontal flight at low speeds but suffered in crashworthiness. These prototypes
resulted in the first micro-scale transition flights but during flight testing, spars and motors would crack free of the
foam, and airframes eventually deteriorated.
While true at larger scales approaching sizes that can carry human passengers, the effects of the common square-
cube law and the non-availability of thinner carbon cloth and the brittleness of small carbon tubes instead favored a
search for other structural materials. That search led to polycarbonate sheet. Not only is polycarbonate more tough
and damage-resistant than carbon fiber (it is commonly used to form bullet-resistant windows) it is available in thinner
sheets, is optically clear, and is formable without the use of additional resin systems. Moreover, at MAV scales it is
competitive in terms of both strength and stiffness in large part because thinner sheets can be used. Carbon’s superior
strength-to-weight is fundamentally crippled because all the resulting structures were overweight and despite their
superior raw strength, were eventually destroyed by handling and edge load cases rather than normal operational loads.
Polycarbonate structures did not suffer from these problems because they could handle the brief edge cases with
temporary elastic deformation rather than permanent yielding or fractures.
The resulting operational design of the micro-scale QuadSparrow, thus, was made from thermally vacuum-formed
polycarbonate shells that resulted in a streamlined monocoque structure that could incorporate smooth, compound
5
American Institute of Aeronautics and Astronautics
curves around the necessary electronics as opposed to joined flat plates. This is the final design shown in Figure 5).
To manufacture the main wings in a single top-down vacuum-formed part, the base design was reduced to the
empennage surfaces and a streamlined centerbody with a removable, ogival nose cone held in place with a friction fit.
The wing trailing edges were formed by a final closeout vacuum-formed polycarbonate shell thereby minimizing the
base drag of the airframe and providing a resilient structure to absorb landing loads.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

Figure 5: Design Evolution of the XQ-139µ QuadSparrow Convertible MAV via Carbon/Foam Extrusion (Left),
Reinforced Foam Sheets (Middle) and Polycarbonate Vacuum-Forming (Right)

C. The Unique Design Challenge of a Rocket-Boosted Multirotor


Not only does the QuadSparrow represent what may be the smallest transition-capable aircraft, but with a slight
modification to accommodate a small solid rocket motor, its low-drag airframe allows it to reach speeds that also make
it the world’s fastest quadrotor at this size and possibly any size to date. This modified airframe (shown compared to
the unmodified QuadSparrow in Figure 6) was given the nickname QuadRocket to fit its new capability. The tailcone
was cut to fit a standard 13 mm (0.51 in.) phenolic tube with an endcap to protect the internal polycarbonate structure
from the heat of both launch and the rocket motor’s ejection charge. In a conventional model rocket, this ejection
charge typically blows forward removing the nose cone and deploying a small parachute. In That modification and
addition of a solid rocket motor affected the longitudinal center of gravity location pulling it aftward and reducing the
static margin. These modifications raised the nominal rocket-boosted take-off mass to 30 g which dropped to 28.3 g
as the motor burns out and 24.4 g as it is subsequently ejected from the centerline tube, leaving the aircraft to easily
and controllably fly down.

Figure 6: A Comparison of the QuadSparrow (Left) and QuadRocket (Right) Polycarbonate Airframes
6
American Institute of Aeronautics and Astronautics
The design of a small quadrotor with rocket boosting necessitated some unconventional comparisons. Because
the design has two competing modes of VTOL thrust generation (rocket lift and electric lift), they must be compared
simultaneously. The quadrotor must not have a mass high enough such that the rocket thrust to weight is too low for
a conventional lift-off. Selected data published on the smallest, most commonly-available rocket motors is included
in Table 1. Based on the motor data and the specifications of a large selection of model rocket kits designed to use
them, a nominal thrust to weight is on the order of 20 – 50. Based on the range of motor maximum rated lift, LMAX,
however, would decrease the minimum rated thrust to weight to a range of 6 – 20. The most efficient of the rocket
motors commercially available from Estes (in terms of the ratio of maximum thrust, T MAX, to motor burnout weight,
WBO) are the 1/2A3-2T and the A10-3T with a TMAX/WBO of 220 and 320 respectively. While the A10-3T may seem
more desirable for its greater impulse, the 1/2A3-2T has significantly lower initial and burnout masses, mM,0 and
mM,BO, respectively making it a better candidate for lifting via electric propulsion. It is still possible to launch the
QuadRocket via the rocket-boost with a takeoff weight greater than the electric motors can lift (more than 32 g of
mass for instance) but for quadrotor recovery, the flight mass after burnout must be below the maximum electric
VTOL mass of 32 g.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

Table 1: Manufacturer’s Performance Data on 13 mm (0.51 in.) Diameter Consumer Rocket Motors 24
Impulse m0 mBO TMAX LMAX TMAX/ WBO
Motor
N-sec g g N g -
1/4A3-3T 0.625 5.6 4.75 4.9 28 105
1/2A3-2T 1.25 5.6 3.85 8.3 57 220
1/2A3-4T 1.25 6.0 4.25 8.3 28 199
A3-4T 2.50 7.6 4.10 6.8 57 169
A10-3T 2.50 7.9 4.12 13.0 85 322

A nominal rocket-boosted flight for the QuadRocket involves an unpiloted conventional model rocket take-off
stabilized aerodynamically via a good static margin. Upon burnout, the aircraft is still coasting to apogee and may
either be allowed to do so or the flight may be interrupted by the pilot remotely increasing throttle to artificially
stabilize it in a hover. At this point, conventional quadrotor flight is possible and the pilot may perform any desired
maneuvers or fly the model back to the launch pad where a new motor can be installed and the launch repeated.

II. Airframe Rapid Manufacturing Processes


The QuadSparrow and QuadRocket airframes both share the same manufacturing processes with the exception of
the installation of the phenolic tube to accommodate the rocket motor. This general process lent itself to the philosophy
of rapid prototyping. Compared to conventional methods of composite construction involving machining of metal
tooling and fiber layups, this method allowed rapid construction and modification of arbitrary tooling as well as rapid
prototype construction to facilitate early testing and design iteration. What would normally take a week or more was
reduced to as little as a day limited largely by two-part system cure times.

A. Additive Manufacturing
The process begins with a three-dimensional (3-D) computer-aided design (CAD) model of the desired part that is
converted to a stereolith file (.stl filetype) and printed via an available 3-D printer. Typically, 3-D printers use a two-
axis worm gear system to articulate a heated print head that is fed material. The material melts in the print head and
is extruded onto a build plate along a path defined by the stereolith file tracing out the perimeter of the model at a
single layer of a carefully-controlled thickness on the order of 0.1 mm. Once the current layer’s perimeter is completed,
the build plate or print head increment one layer’s thickness away from the other and prints the next layer’s perimeter
on top of the prior one. Following many repetitions, this results in a hollow print of the CAD model. The largest 3-D
printer available in the University of Kansas Adaptive Aerostructures Lab is the Makerbot Replicator Z18 (shown in
Figure 7) and uses polylactic acid (PLA) as the print material. The resulting 3-D print represents a solid copy of the
intended part to be eventually thermally vacuum-formed out of thin polycarbonate sheet.
The resulting prints are a close representation of the CAD models except for the model’s surface finish. The
printing process with PLA (as with most thermally-extruded materials) leaves ridges between the print layers. For
some projects this may not be an issue, but for the creation of tooling, these ridges must be smoothed. The most time-
and labor-efficient method of smoothing these ridges appears to be the following: (1) Remove any excess print material
such as a build raft or supports. (2) Sand any internal faces and assemble parts as necessary using a thin layer of fast
curing epoxy between parts. (3) Make a first pass at sanding down the surface ridges with a medium-grit (~150)
7
American Institute of Aeronautics and Astronautics
sandpaper and then wipe away any dust. (4) Cover Surface in a thin but liberal coating of glazing putty. (5) Once dry,
sand with a high-grit sandpaper (~220-320) until the surface is smooth. Reapply glazing putty and repeat if large ridges
are still felt. (6) Coat the print with a filler primer and sand smooth with high-grit sandpaper (~320+). At this point
the surface should be smooth and ready for molding and casting.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

3-D Printed
Nose Cone Plug

Figure 7: The MakerBot Replicator Z18 3-D Printer (Left), and an Example Sanded Nose Cone Print (Right)

B. Plug-and-Mold Casting
The smoothed print is unfortunately not capable of surviving the temperatures necessary for thermal vacuum-
forming of polycarbonate. Polycarbonate has a softening temperature of approximately 150 °C (302 °F)25 while the
PLA the plug is printed out of has a much lower transition temperature of just 57 – 63 °C (134 – 145 °F)26. This means
that if a PLA plug is used directly as a thermal vacuum-forming mold, the combined effects of temperature and force
of the vacuum suction will permanently deform the mold. It also means that care should be taken not to allow a curing
mold material to reach high exothermic temperatures. This prevents use of many hard resin systems as a mold material
without other means of absorbing the released heat via water-cooling for instance. Use of a hard resin is not desirable
for the mold anyway as it will not be used under pressure as in an autoclave or vacuum-bagged and cured in an oven
and removing the plug from a hard mold will be more difficult than removal from a flexible mold.
For those reasons, a low durometer, two-part silicone was chosen to make a mold of the PLA plug. The cure
temperature in representative batches is low enough to not pose a thermal risk to the PLA, the low durometer means
the cured silicone will be flexible but strong while resisting tearing, and the use of silicone in general means most
casting materials will naturally not stick to the mold. This eliminates the need for a separate mold release material. A
mold shell was built from scrap aluminum sheet and assembled using polyester tape to prevent the liquid silicone
mixture from leaking until it cures. The plug was fixed to the floor of the shell with double-sided tape to simultaneously
prevent the plug from either shifting or floating up off the bottom of the mold. An appropriate volume of silicone was
then mixed and poured into the mold shell with at least ¼ in. of silicone depth above the plug’s highest point to
maintain dimensional stability of the cured mold.
After 6 – 8 hours, the silicone was cured and the new mold was removed from the mold shell. The part was then
removed from the silicone mold and any silicone flashing cut away with a razor blade. A fast-curing, thin, two-part
casting resin system with a high temperature tolerance was chosen as it could be poured into small parts that more
viscous resins may not completely fill, quickly cure, and then survive the elevated temperatures and forces of vacuum-
forming. An appropriate volume of the resin was mixed and poured into the mold and allowed to reach a full-hard
cure. The resin casting was then removed and required a one hour thermal soak in the oven as a post-cure to fully set
the cast part. At this point, the casting is examined for any surface imperfections which can be filled with glazing
putty, and/or sanded smooth prior to use as a thermal vacuum-forming mold. At this point, minor changes to the part
can also be made – sanding areas down or building them up to meet unforeseen needs – and immediately tested on the
vacuum-former. The most time-consuming parts of the build process are related to the building of the initial silicone
mold and casting of the resulting vacuum-forming molds. Mold and casting materials were chosen primarily for their
thermal properties and their fast curing time but these steps still require most the total time necessary to bring a new
concept from a CAD model to a finished prototype.

8
American Institute of Aeronautics and Astronautics
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

3-D Printed Resin Casted


Nose Cone Plug Silicone Mold Nose Cone Plug
Figure 8: The 3-D Printed Mold Plug (Left), the Resulting Silicone Mold (Middle), and a Resin Casting (Right)

C. Thermal Vacuum-Forming
Following formation in the silicone mold, the cast plug was then used as a mold in the vacuum-forming process
(Shown below in Figure 9). This typically involves the use of a heating element to warm a sheet of a thermoplastic
material fixed at the edges on a rigid frame. The material used in this case was polycarbonate but other materials like
styrene and acrylonitrile-butadiene-styrene (ABS) are common as well. Polycarbonate was selected for its high
strength to weight, damage tolerance, and optical clarity. For these reasons, it is commonly used to make remote
control car bodies as well as model airplane canopies. At the time of publishing, however, the XQ-139µ appears to be
the first aircraft made completely out of polycarbonate.

Heating Element

Vacuum Table Vacuum Table

Figure 9: Heated Polycarbonate Sheet Sagging Over the Molds Ready on the Vacuum Table (Left) and After Lowering
the Sheet Allowing it to Form Over the Molds Under Vacuum Pulled from the Perforated Table (Right)

The thermoplastic material is fixed to the rigid frame and placed under the heating element to absorb heat to the
point that it begins to sag. Meanwhile, the vacuum-forming molds are placed on a vacuum table attached to a solenoid
valve and a vacuum chamber. Typical hobbyist units use air compressor tanks evacuated with a vacuum pump. Once
the polycarbonate has passed the glass transition temperature and is sagging an appropriate amount, the frame and
polycarbonate are removed from the heat source. The solenoid valve is opened exposing the table to the vacuum tank
9
American Institute of Aeronautics and Astronautics
and the polycarbonate on the frame is pressed over the part forming a pressure seal. The vacuum tank removes all the
air between the polycarbonate sheet and the mold drawing it into small crevices and taking the shape of the mold. In
the absence of the heat source, the polycarbonate rapidly cools in a few seconds to equilibrium with the surroundings
and sets into the shape provided by the mold. The solenoid valve is closed, the tank pressure evacuated again via the
vacuum pump, and the polycarbonate part removed from the vacuum table. In most cases, the plug removes easily but
troublesome shapes can also be coated in Teflon to reduce mold/material interfacial friction.
The most profound benefit of this method is its fast turnaround time. From the time the polycarbonate is placed
under heat to when it is fully formed and the mold is removed can be as fast as just a few minutes. Subsequent pulls
are faster than the first because the amount of vacuum pressure that must be renewed in the vacuum tank is small after
it is initially pumped down and the residual heat in the heating element speeds up the time necessary to reach the
polycarbonate’s forming conditions. As a result, rapid manufacturing consistent with prototype iteration and high rate
production is possible. Thermal vacuum-forming, thus, represents a new means of building strong, damage tolerant
MAV structures.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

D. Assembly and Post-Processing


The XQ-139µ (both the QuadSparrow and QuadRocket airframes) used a total of five vacuum-formed parts. These
include a two-part nose cone (formed as two halves), an internal floor to support the multirotor control board, the main
wing part forming the wing leading edges and motor mounts, and lastly the combination tail cone and wing trailing
edge part that closes the aft edges of the airframe. The fully-enclosed structure is strong and stiff despite being made
of polycarbonate thicknesses as thin as 0.17 mm (0.0067 in.) owing to effects of both the square-cube scaling law and
the nature of beam bending stiffness as it relates to monocoque structures. All the material stiffness of the closed
structure is as far from the cross-sectional neutral axis as possible and thus the material optimally contributes its
stiffness to the finished structure.
The vacuum-formed parts are cut from the sheet to a uniform edge using a purpose-built hot wire cutter (shown in
Figure 10) relying on driving power through a nichrome wire which gets hot enough to cut through the polycarbonate.
The hot wire cutter was built into a fixed stand such that the depth of the cut could be adjusted and locked in place via
the commonly-available aluminum T-slot rail system. Not only does this allow rapid removal of the part from the
vacuum-formed sheet, it prevents the formation of stress-concentrations that tend to form when mechanically cutting
the polycarbonate sheet (for example with scissors). The hot-wire cutting melts the edge of the polycarbonate adding
a small amount of edge thickness and therefore strength and stiffness.

Fixed Rails

Adjustable Rails
Cut
Airframe

Vacuum-
Formed Nichrome Excess
Airframe Wire (Discarded)

Figure 10: A Vacuum-Formed Airframe Still on the Sheet (Left), the Hot Wire Cutting Tool Built with Depth Adjustment
(Middle) and a Part Cut From the Vacuum-Formed Sheet (Right)

The parts are then trimmed as necessary to fit together on a purpose-built precision assembly tool to maintain
dimensional stability of the polycarbonate until the airframe is fully assembled. The COTS electronics (a multirotor
control board with an integrated power plug and brushed DC electric motors) are installed into the main wing part.
The power plug is fed through the top of the airframe and fixed in place with Permatex 84115 Epoxy. This was selected
following the early tests using cyanoacrylate showing the importance of the bonding material on damage tolerance
and crashworthiness of the MAV structure and integrated parts. It provides a strong but slightly pliable plastic-plastic

10
American Institute of Aeronautics and Astronautics
or plastic-metal bond necessary to help absorb high transient loads rather than pass it directly into the polycarbonate
structure causing stress concentrations and ultimately failure of the structure. It also has a fast set time of 5 minutes
and a strength of up to 24 MPa (3500 psi) following a full 24 hour cure cycle.
The polycarbonate parts were bonded using Weld-On 3 styrene-butyrate-polycarbonate glue. Unlike the thicker
Permatex epoxy, the Weld-On glue is a thin, clear solvent that functionally melts the two surfaces together instead of
bonding to the plastic-plastic interface. The solvent method was desirable for both its optical clarity and its rapid
effective cure time. It requires just a few seconds from application of a thin film to the bonding area and fitting parts
together before they are functionally “welded” together. This method was used to assemble the rest of the airframe.
The resulting clear aircraft could then be flown or painted with common spray paints to aid in visibility and in-flight
vehicle orientation.

III. Experimental Testing


The XQ-139 MAVs were tested in a variety of conditions corresponding to separation of the individual modes of
operation the design would encounter. These are split into the separate discussions below of the static bench testing
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

of the small electric powerplants, flight testing in a hover, flight testing in transitioned horizontal flight, and rocket-
boosted flight. These include the limited performance data that could be determined from these flights.

A. Static Bench Testing


To quantify the available performance of the COTS brushed (DC) electric motors and the recommended 2 in. (5.08
cm) diameter propellers, a bench test of a single motor was performed. The test setup and resulting static thrust force
resulting from applied input power plotted against propeller angular speed are included in Figure 11. The motor was
attached on a long moment arm to magnify the thrust measurement and scaled based on the ratio of pivot moment
arms. DC power was supplied to the motor via a variable power supply on which voltage was controlled. At a selected
voltage, the current and thrust were recorded from the power supply and scale, respectively while the propeller angular
speed was determined using a strobe tachometer. The thrust and power data showed a strong growth with increasing
propeller speed with a hover speed of just below 10,000 rpm and 0.5 W (2W total power). These conditions are
identified on the combined thrust and power plot. This also shows that while it hovers on approximately half the total
thrust capability, this is approximately equivalent to about 40% the total power capability.
100 1.4
Thrust
Digital 90
Scale Power 1.2
Pivot 80

Electrical Power Input, PIN (W)


Static Thrust Force, T (mN)

Assembly 1.0
70

60
0.8
Hover Condition
50
0.6
40

30 0.4

20
0.2
10

Motor and 0 0.0


Propeller 0 2 4 6 8 10 12 14
Angular Speed, Ω (1000 rpm)
Figure 11: Static Test Setup (Left) and Combined Thrust and Power Data (Right) for a Single Motor and Propeller

To judge the efficiency of the motor/propeller combination, the propeller figure of merit (FM, defined as the ratio
of the ideal power to the actual measured power) was calculated. At large scales like those found on manned
helicopters, the figure of merit may be as high as 0.8 – 0.91 with carefully-engineered rotor blades operating near
design conditions. As rotors and propellers scale down, the effects of lowering Reynolds number progressively rob
11
American Institute of Aeronautics and Astronautics
the blades of this power efficiency. Scaling down to the 2 in. (5.08 cm) propeller diameter used on the available COTS
motors appears to diminish the figure of merit to a maximum of 0.35 at full power.
0.40

0.35 Hover Condition


Figure of Merit, FM (-)

0.30

0.25

0.20

0.15

0.10
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

0.05

0.00
0 10 20 30 40 50 60 70 80 90 100
Thrust Force, T (mN)
Figure 12: Static Bench Testing Propeller Figure of Merit Data of a Single Motor and Propeller

B. Hover Flight Testing


The XQ-139µ QuadSparrow MAV exhibited high maneuverability and control authority when flown at the
nominal flight mass of 20 g (0.70 oz). Flights were performed in the KU AAL’s closed laboratory space with minimal
influence of air currents and in controlled temperature conditions as shown in Figure 13. Including the 4.3 g (0.15 oz)
150 mAh capacity one-cell lithium polymer battery, hover testing showed a useful load of up to 17.4 g (0.61 oz)
corresponding to a maximum controllable take-off mass of 33 g (1.16 oz). Based on an average hovering endurance
of 8 minutes and the battery’s rated capacity, the nominal hover power was estimated at 3.75 W.
Hover flight testing was also performed outdoors on calm days as well as windy days. On calm days the handling
and performance matched closely with that experience indoors. Flights were attempted in winds gusting up to 40
km/hr (25 mph) and although the airframe presented a nontrivial sideward drag profile, there was always available
thrust and control authority to maintain position in a hover by exploiting the XQ-139’s natural weathercock stability.
The static stability of the airframe allows it to naturally point into the ambient wind direction at which point the pilot
need only maintain thrust equivalent to the airframe drag to remain in position and minor control adjustment to hold
orientation. In this way, the transitioning quadrotor architecture may be able to penetrate wind conditions that
comparable helicopters, multirotors, and other low wing-loading MAVs may not.

Figure 13: XQ-139µ “QuadSparrow” MAV Hover Flight Testing Indoors (Left) and Outdoors (Right)
12
American Institute of Aeronautics and Astronautics
C. Horizontal “Missile Mode” Flight Testing
Following the experience with outdoor hover testing in high winds, it became clear that transition flights could be
easily tested in these conditions. The weathercock stability mode of the design allows for testing of the intermediate
transition flight conditions between hovering flight and fully-transitioned horizontal flight. The latter was denoted
“missile mode” flight as a combined reference to the horizontal flight path and the airframe’s convertible missile
lineage. In the presence of increasing wind speeds, the aircraft’s ground speed could be controlled directly by the
airframe’s pitch angle and throttle control. The transition to missile mode flight was thus flight tested by progressively
expanding the flight envelope by flying at top speed limited by the controlled airframe pitch and then returning to a
stabilized hover. After several of these incremental transition flight tests, the XQ-139µ “QuadSparrow” MAV fully
transitioned into missile mode flight. This is shown in Figure 14 as a series of pictures demonstrating transition from
a hover taken during outdoor flight testing.
The small scale of the aircraft made in-flight data gathering particularly difficult. At the time, there were no small
COTS, flight-logging, brushed control boards compatible with the design. In the absence of onboard or telemetry data,
flight speed measurements gathered via radar gun also failed to reliably resolve the speed of the small airframe even
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

with aluminized mylar to augment radar reflectivity. The aircraft was able to successfully penetrate wind speeds of 40
km/hr (25 mph) with an approximate ground speed of 8 – 16 km/hr (5 – 10 mph) thus suggesting a top speed of 48 –
56 km/hr (30-35 mph) based on passing landmarks. This speed, however, is not likely to be the airframe’s maximum
potential as the COTS electronics and most importantly, the propellers, are intended for efficient hovering as opposed
to high speed forward flight. Keeping the available electric power constant, but increasing the propeller angular speed,
Ω, either by changing the supply voltage or the mechanics of the DC motors should result in a higher electrical top
speed. In short, the QuadSparrow design is limited in its top speed only by the multiple propellers’ ability to penetrate
the wind as opposed to the airframe’s base drag for instance.

Figure 14: XQ-139µ “QuadSparrow” MAV Outdoor Flight Testing Demonstrating Transition November 2015

D. Rocket-Boosted Flight Testing


Lastly, the rocket-equipped version of the XQ-139µ “QuadRocket” MAV was also flight tested. It’s primary mode
of flight is the rocket-boosted launch but hovering and transition to missile mode flight were also tested. The minor
increase in empty mass (without the mass of the 1/2A3-2T solid rocket motor) had a negligible effect on the general
performance compared to the baseline QuadSparrow aircraft it was originally modified from. This includes the
qualitative handling and maneuverability as well as the endurance, gross take-off mass, and the top speed under electric
power.
The nose cones of the QuadRockets use an integrated launch lug similar to those found on conventional model
rockets to maintain alignment during initial acceleration. Rocket ignition and subsequent launch is also initiated the
same way it is on a conventional model rocket. An igniter is inserted into the rocket motor, held in place with a plastic
launch plug, and the igniter is activated from a safe distance with an electronic launch controller.
13
American Institute of Aeronautics and Astronautics
Over a dozen rocket-boosted flights were performed to test the response of the airframe, the COTS flight controller,
and the stability margins of the design. The original QuadSparrow’s short wing chord was enough for static stability
in horizontal flight with a positive static margin. But the addition of the solid rocket motor at the aft-most location
pulled the center of gravity far enough back as to result in a statically unstable configuration. To alleviate this
condition, the aerodynamic center was pulled aftward by extending the wing area aftward as well. Following this
modification, the QuadRocket accelerated up the launch rail reaching a burnout speed of 214 km/hr (133 mph) and
continuing on a stable climbing flight path to an approximate apogee of 200 ft. The top speed was estimated based on
frame-by-frame photogrammetrical analysis of high speed footage. Still images taken from footage of one of these
rocket-boosted flight tests are included in Figure 15. Subsequent tests reduced the size of these wing extensions until
the QuadRocket was no longer stable after leaving the launch rail.
This wing sizing data was used to modify the existing thermal vacuum-forming molds to have greater depth forcing
the polycarbonate to draw further down under vacuum and extending the wing area aftward. This appeared to
noticeably affect the minimum throttle setting required to hover which resulted in a greater margin of available control
power and top speed. This may be a result of the increased chord without proportionally increasing the wing thickness
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

or body diameter. This would functionally decrease the wing airfoil’s thickness-to-chord ratio and better streamline
the wings with a shallower angle post maximum thickness. In other words, this may be the result of improved flow
separation control and therefore a further reduction in the effective base drag of the airframe.

Figure 15: XQ-139µ “QuadRocket” MAV Rocket-Boosted Launch Stills from High Speed Footage (1000fps)

IV. Conclusions and Future Work


The design, construction and flight testing of a new kind of transitioning micro-aerial vehicle (MAV) was
presented including the rapid prototyping and manufacturing methods used to develop it. This aircraft design was
scaled down from an original UAV design that is also capable of efficient hovering flight and transition to high speed
horizontal or “missile mode” flight. These formed the XQ-139 family of aircraft designs. Rapid prototyping of the
MAV designs was also demonstrated through a series of steps that allow CAD models to be affordably created and
quickly turned into useful tooling. These methods should also be applicable to other MAV designs that may reap the
benefits of the optically clear, damage tolerant, and effectively high strength-to-weight of the polycarbonate sheets
used in thermal vacuum-forming.
Testing at small scale showed high maneuverability and good stabilization provided by the low-drag airframe and
commercial, off-the-shelf (COTS) electronics installed inside the aerodynamic shell. Efficient hovering flight was
demonstrated in indoor and outdoor flight tests on a single-cell 150 mAh lithium polymer battery owing in part to the
low blockage area and streamlining of the polycarbonate shell. Stability in high winds was also demonstrated via the
airframe’s static stability and thus natural tendency to point into the prevailing wind. This effectively reduced the
problem of crosswind station keeping to one of throttle to maintain lift and counter the airframe drag. This was also
effective in providing a way to incrementally test transition flight between hover and missile mode by reducing the
ground speed at which transition occurs. Further flight tests of the QuadSparrow MAV demonstrated reliable control
and performance when pitched over into full missile mode with an approximate top speed of 56 km/hr (35 mph) likely
limited by the propeller pitch speed. Testing of the rocket-equipped QuadRocket MAV (modified from the baseline
QuadSparrow MAV) showed very similar results while launch under rocket power was able to boost the top speed to
14
American Institute of Aeronautics and Astronautics
214 km/hr (133 mph) equivalent to a climb rate burst of 11,700 ft/min. This opens up the potential for a new class of
MAVs and potentially UAVs that could launch vertically to an observational altitude and then enter either an efficient
hover or transition to forward flight to perform a high speed dash to a mission location.
Future tests of the airframe will include replacing the current motors with ones that can spin the same or higher
pitch propellers faster which should increase the advance ratio and thus the top speed and gust penetration capability
of the QuadSparrow. Further testing with newer electronics that can log onboard flight data should also help to quantify
many of the effects discussed in this study. These future tests also include scaling up the rocket-boosting capability to
the XQ-139A as well as studying if there exists a vehicle scale that could maximize the resulting top speed or altitude
using COTS rocket motors. All of these changes will require minor modifications of the aerodynamic shells which
are made easier using the rapid prototyping techniques described in this study. Meanwhile, the basic principles of the
XQ-139 family of designs and their scaling still apply.

References
1Leishman, J. G. Principles of Helicopter Aerodynamics. Cambridge University Press, New York, 2006.
Downloaded by UNIVERSITY OF COLORADO on January 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/6.2017-0012

2Zyluk, A., and Sibilski, K. "The Gust Resistant MAV - Aerodynamic Measurements, Performance Analysis, and Flight

Tests," AIAA Atmospheric Flight Mechanics Conference. American Institute of Aeronautics and Astronautics, 2015.
3Thipyopas, C., and Moschetta, J. M. "From Development of Micro Air Vehicle Testing Research to the Prototype of TYTO:

Low Speed Biplane MAV," 26th AIAA Applied Aerodynamics Conference. American Institute of Aeronautics and Astronautics,
2008.
4Hwang, H., Chung, D., Yoon, K., Park, H., Lee, Y., and Kang, T. "Design and Flight Test of a Fixed Wing MAV," 1st UAV

Conference. American Institute of Aeronautics and Astronautics, 2002.


5Yeo, D., Shrestha, E., Paley, D. A., and Atkins, E. M. "An Empirical Model of Rotorcrafy UAV Downwash for Disturbance

Localization and Avoidance," AIAA Atmospheric Flight Mechanics Conference. American Institute of Aeronautics and
Astronautics, 2015.
6Stiltner, B., Olien, C., and Faber, J. "Preliminary and Conceptual Design of a Remotely Piloted Ducted Fan MAV," 48th

AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition. American Institute of
Aeronautics and Astronautics, 2010.
7Hsu, C.-K., Evans, J., Vytla, S., and Huang, P. G. "Development of Flapping Wing Micro Air Vehicles -Design, CFD,

Experiment and Actual Flight," 48th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace
Exposition. American Institute of Aeronautics and Astronautics, 2010.
8Henry, J., Schwartz, D., Soukup, M., and Altman, A. "Design, Construction, and Testing of a Folding-Wing, Tube-Launched

Micro Air Vehicle," 43rd AIAA Aerospace Sciences Meeting and Exhibit. American Institute of Aeronautics and Astronautics,
2005.
9Grasmeyer, J., and Keennon, M. "Development of the Black Widow Micro Air Vehicle," 39th Aerospace Sciences Meeting

and Exhibit. American Institute of Aeronautics and Astronautics, 2001.


10Markman, S. Straight Up: A History of Vertical Flight. Schiffer Publishing, 2000.
11Anderson, S. B. "Historical Overview of V/STOL Aircraft Technology," Ames Research Center, 1981, NASA TM-81280
12Harris, F. D., Center, A. R., and Monterey Technologies, I. Introduction to Autogyros, Helicopters, and Other V/STOL

Aircraft. National Aeronautics and Space Administration, Ames Research Center, 2011.
13Bramlette, R. B., and Barrett, R. M. "Design and Flight Testing of a Convertible Quadcopter for Maximum Flight Speed,"

AIAA Sci-Tech Forum and Exposition, Grapevine, TX, American Institute of Aeronautics and Astronautics, Reston, VA, 2017
14Yenne, B., and Jenkins, D. Convair Deltas: From SeaDart to Hustler. Specialty Press, 2009.
15Coleman, S., Ginter, S., Long, B. J., and Chana, B. Convair XFY-1 Pogo. Ginter Books, 1994.
16Ginter, S., Lockheed, and Koenen, R. Lockheed XFV-1 VTOL Fighter. Naval Fighters, 1996.
17Textron. "Latitude Engineering: Hybrid Quadrotor Technology," 2016, https://latitudeengineering.com/products/hq/
18Amazon. "Amazon Prime Air," 2016, https://www.amazon.com/b?node=8037720011
19AeroVironment. "AeroVironment: UAS Advanced Development: SkyTote (UAV)," 2012,
https://web.archive.org/web/20121113174211/http://www.avinc.com/uas/adc/skytote/
20Stone, R. H., Anderson, P., Hutchison, C., Tsai, A., Gibbens, P., and Wong, K. C. "Flight Testing of the T-Wing Tail-Sitter

Unmanned Air Vehicle," Journal of Aircraft Vol. 45, No. 2, 2008, pp. 673-685.
21Stone, R. H. "The T-Wing Tail-Sitter Research UAV," 2002 Biennial International Powered Lift Conference and Exhibit.

American Institute of Aeronautics and Astronautics, 2002.


22xCraft. "xCraft: X-PlusOne: A Quadcopter Plus a Flying Wing," 2016, http://xcraft.io/x-plusone-drone/
23Transition-Robotics. "Transition-Robotics: Quadshots!," 2016, https://transition-robotics.com/collections/quadshots
24Estes Rockets. "Estes Engine Chart," 2016, http://www2.estesrockets.com/pdf/Estes_Engine_Chart.pdf
25Covestro LLC. " Safety Data Sheet: HYGARD MS1250 CLEAR-A00 Polycarbonate Thermoplastic Polymer Sheet," 2016,

http://www.eplastics.com/pdf/SDS-makrolon-gp-clear.PDF
26UL Prospector. "Prospector Materials and Ingredients Search Engine: Polylactic Acid (PLA) Typical Properties," 2016,

https://plastics.ulprospector.com/generics/34/c/t/polylactic-acid-pla-properties-processing

15
American Institute of Aeronautics and Astronautics

Вам также может понравиться