MARS UNMANNED AIRCRAFT 2003-2004

ODYSSEUS TEAM
CORNELL UNIVERSITY

ADVISOR
PROFESSOR MICHEL Y. LOUGE

TEAM LEADER
ALEXANDER CHEFF HALTERMAN

DATE
APRIL 1, 2004
THE TEAM:

Team Members:
Alicia Billington arb351 BEE (2006)
Emmanuel Franjul ef35 MAE (2005)
Jian Gong jg253 ECE (2004)
Alexander Halterman ach22 MAE (MEng 2004)
Yen-Khai Lee yl245 ECE (2004)
Jeremy Nersasian jbn5 MAE (MEng 2004)2
Cem Ozkaynak co37 ECE (2005)
Jing Pei jp292 MAE (2004)
Mikiko Ujihara mu23 MAE (2004)

Advisor:
Professor Michel Louge myl3

1 Student ID numbers are the students e-mail address (ID#@cornell.edu i.e. [email protected])
2
Ceased doing work after December 2003 due to January 2004 graduation
 Abstract

As space exploration progresses, Mars gains more focus as the next frontier in human exploration.
Manned missions to Mars have been discussed and planned to a certain degree. However, before humans can set
foot on Mars, a wealth of information about Martian conditions will need to be provided by satellites, unmanned
vehicles and a myriad of other data collection instruments. The goal of the Odysseus team is to design an unmanned
aerial vehicle (UAV) for flight in the lower regions of the Martian atmosphere. Such a vehicle would collect specific,
high resolution topographic data for speculated landing sites. The data returned by this aircraft will be of the utmost
importance to the success of any Mars landing mission.

This paper focuses on the aerodynamics, propulsion, structures and electrical systems of an unmanned aerial
vehicle for flight on Mars. The Martian environment, as well as the mass sensitive nature of current space
exploration, present a set of conditions by which an aircraft must be designed. Such a design must optimize mass,
volume, flight time, power, and instrumentation in order to create an aircraft that can be sent to Mars with existing
spacecraft while satisfying its exploratory goals once it reaches Mars. A UAV design for Mars must incorporate
the aerodynamic properties for sustained flight in a low density atmosphere, propulsion in an atmosphere lacking
sufficient oxygen for combustion, structural integrity with minimal mass and electrical controls for unmanned flight.
Our UAV design overcomes these daunting constraints and provides a robust platform for reconnaissance of Mars.

The final UAV design consists of a 10.8 kg aircraft with fuselage length of 2.1 meters, maximum
fuselage diameter of 0.25 meters, and a wingspan of 2.078 meters. For propulsion we have chosen a single 2.27-
meter diameter three-bladed propeller mounted aft of an inverted V-tail. The propeller motor as well as the
topographical and control instrumentation aboard the UAV is powered by multiple lithium-ion SPE batteries.
The data produced during flight will be continuously transmitted to satellites orbiting Mars that relay the signals to
Earth. Flight control and navigation is accomplished through feedback from on-board sensors that detect
acceleration, pitch and roll. This design provides a flight time of 2.3 hours at a cruising speed and altitude of 130
m/s and 500 meters respectively. Further details of our design choices and potential alternatives are discussed
within the following pages.
Introduction fuselage designs. Electrical evaluated the capabilities of
Mars is indisputably the centerpiece of current various electrical components and communication
space exploration with both the scientific community’s network designs. Through a series of presentations to
space exploration efforts and the general public’s Professor Michel Louge that focused on our
interest focused on recent investigative missions to the conceptual, preliminary, and final design, we narrowed
Red Planet. In the spirit of human exploration the down the initial trade-studies to a single UAV design
Odysseus Team is designing an Unmanned Aerial optimized for the Martian environment.
Vehicle to fly on Mars as part of the Revolutionary Our design process focuses only on the
Vehicles: Concepts and Systems University Student Competition technical aspects of flight on Mars, from the time the
for 2004 sponsored by NASA. A UAV serves as a UAV is deployed through its expected life-cycle. Other
reconnaissance platform for future manned missions to aspects of the mission, such as launch costs and
Mars. The aircraft will carry topographical and imaging procedures, Earth-to-Mars transit routes, aero braking
instrumentation to produce valuable data regarding the techniques upon reaching Mars, and atmospheric
Martian environment. deployment feasibilities were not considered in our
The design of a UAV must optimize mass, design. Additionally, cost, environmental impact,
volume, flight time, power, and instrumentation in political motivation and human safety were not
order to create an aircraft that can be sent to Mars with prevalent issues for us; however, technological and
existing spacecraft while satisfying its exploratory goals mission feasibility were carefully evaluated throughout
once it reaches Mars. Such a design for Mars must the design process. Our final design only employs
incorporate the aerodynamic properties for sustained technologies that are currently available or on the
flight in a low density atmosphere, propulsion in an horizon while acknowledging the many aerodynamic,
atmosphere lacking sufficient oxygen for combustion, structural, propulsion, and electrical challenges of
structural integrity with minimal mass and electrical deploying a fully-autonomous UAV on Mars for long
controls for unmanned flight. Our UAV design periods of time.
overcomes these daunting constraints and provides a
robust platform for reconnaissance of Mars. Atm ospheric Constants
T0 (K) P0(Pa) (m -1 ) R 
General Design Methodology: 223 700 9.00E-05 192.1 1.289
The design process began with the high-level
abstract evaluation of the various objectives and Table 1: Martian Atmospheric Constants
constraints. Members of the team met twice a week to
develop the mission profile, such as the scientific Martian Atmosphere and Environment:
motivations behind a high-resolution topographical Designing an aerial vehicle requires
map of the Martian surface and atmospheric profile, knowledge of the environmental conditions the craft
both of which are not currently possible using existing will be flying in. One must know temperature,
satellites or landers. pressure, density and viscosity as functions of surface
Following the mission profile, we conditions and altitude. We have equations published
consolidated the various engineering aspects – by NASA that give acceptable and reasonable fits for
aerodynamics, structures, propulsion, electrical systems atmospheric data. The equations are valid below 7000
– into a systems-level hierarchy of the conceptual UAV Meters, which is above our cruise altitude, in an effort
design. Each level of the hierarchy takes into to capitalize on the largest density possible, these
consideration safety and technological feasibility in equations work well.
order to identify critical paths and achieve our stated
mission objectives. For instance, much of the fuselage T  T0  .000998 * h (1)
design relied on the availability of a light-weight
electrical engine to provide thrust needed to stay aloft
in the low-pressure environment. As such, several P  P0 e   h (2)
interdependencies were stated early on and constantly
revised as we progressed. Equations ( 1 ) and ( 2 ) give the temperature
Once the systems-level picture was developed, and pressure profiles respectively based on surface
each sub-team developed its own set of trade-studies values. Where  is a constant (9e-5m-1) and h is altitude
and constraints. Aerodynamics compared many airfoil in meters. From temperature and pressure we
and wing designs using Matlab and Excel. Propulsion extrapolate data for density and speed of sound using
compared the feasibility of propellers, chemical rockets, known values and laws. The ideal gas law ( 3 ) gives us
and jet engines given the atmospheric constraints on density as a function of temperature, pressure and the
Mars. Structures required light-weight but durable gas constant R, which is 192.1 on Mars. Equation ( 4 )

1
 solves for viscosity as a function of temperature. Speed therefore flight at the correct altitude becomes
of sound, a, is found using equation ( 5 ) and known important. Thrust specific fuel consumption (TSFC), a
values such as the ratio of specific heats,  and relationship between distance and fuel required, is
temperature. ultimately a function of air density and thus altitude is
the determining factor for fuel efficiency. TSFC sets an
  T pR (3) altitude for efficient cruise flight, from which a wing
loading can be chosen to attain cruise conditions at the
desired altitude. By choosing a specific wing loading
 408.17  the designer can fix the wing area for a craft with a
  36.592T 1.5    10
10
(4)
target weight.
 T  120  Our design did not have the luxury of using
thrust specific fuel consumption to fix wing area. Our
a sonic  RT (5) aircraft will be propeller driven and run off electrical
energy. Since our energy source has no dependence on
 is usually seen to take the value 1.4 because that is its pressure or density, we can not set an optimum cruise
number for the air on Earth.  is 1.289 on Mars due to altitude. So we begin the design process without a
the abundance of CO2 (over 95%) that makes up the specified wing area. With this area we would have been
Martian atmosphere. This lower value, along with the able to find appropriate airfoils, find maximum CL/CD
lower ambient temperature, causes a lower speed of values and proceed to optimize the aspect ratio
sound, approximately 75% of that on Earth. This dependant on wing weight and induced drag from wing
means that speeds need to be further limited to avoid end conditions. Without it, wing area becomes another
sonic conditions. parameter we need to optimize.
Other considerations on Mars include the Next we begin our search for suitable airfoils.
rampant dust storms that can spring up unexpectedly Since the Martian atmosphere is approximately one
and make flight very difficult. These dust storms are hundredth the density of Earth’s and the craft is small
seasonal, allowing a wise launch and flight time to in comparison to commercial aircraft, Reynolds
reliably eliminate this potentially devastating threat. numbers will be very low. Traditionally, planes fly well
into the turbulent boundary layer regime, with a
Cruise Conditions Reynolds number on an order of 106. With the
h (m) T (K) P (Pa)  (kg/m 3)  A (m/s) conditions we are given, Reynolds number values will
500 222.5 669.2 1.57E-02 1.45E-05 234.72
be between 40,000 and 80,000, with 80,000 being an
extreme value that is unlikely.
Table 2: Cruise Velocity Atmospheric Conditions Dealing with such low Reynolds numbers
poses a problem; boundary layers are largely laminar,
which are notorious for flow separation due to low
Aerodynamics Design inertial forces. A craft flying in a laminar regime must
utilize an airfoil designed specifically for low Reynolds
Introduction flows. Traditional airfoil shapes are designed for
As stated previously, the design of an aerial turbulent conditions and will not suffice in the Martian
vehicle for Mars is a tricky process due to the planet’s atmosphere. On the advice of Professor David
harsh environmental conditions. Normally one begins Caughey of Cornell University, we considered research
an aircraft design process by defining take-off and done by Professor Michael Selig of the University of
landing scenarios, as well as approach to cruise altitude, Illinois at Urbana-Champaign. Professor Selig has
but the fact that this vehicle will be launched from orbit done a remarkable amount of work with low Reynolds
makes this unnecessary. airfoils and has a wealth of data available, including lift
Our first step is to determine the wing loading and drag coefficients at various angles of attack, as well
necessary for the most efficient flight in cruise. Wing as coordinates that can be used to numerically generate
loading is the force per unit area on the wing during airfoils. With his data, we proceed with an airfoil
steady state flight conditions, which is important to design.
determine for two reasons. First, it fixes the area of the We use Matlab and Excel to search through
wing. Knowing the wing loading simplifies the roughly 1500 of Professor Selig’s airfoils to find those
minimum drag analysis by tying the wingspan and mean most suitable for our applications. Rough calculations
aerodynamic chord (M.A.C.) together. Optimization show that for velocities limited to Mach 0.6 and
then becomes a question of choosing the correct aspect Reynolds numbers between 40,000 and 65,000, lift
ratio and taper ratio. The second reason involves the coefficients are best chosen to be 0.3-0.6. This is a
fuel efficiency; aircraft designed with combustion relatively low value, but reasonable for this particular
powered engines are sensitive to atmospheric changes; application. With low Reynolds numbers, due to the

2
need for a large chord and a small wing area, a lift equations, described in the previous Martian
coefficient that is too high would tend to limit aspect Atmosphere and Environment section of the report,
ratios. Induced drag becomes overwhelming in this that provide a rough model of the Martian atmosphere.
situation, causing inefficient flight. We also assume the Selig airfoil data to be
With a rough value for the lift coefficient, we correct for all of the airfoils he tested. His tests were
use Matlab to inspect different airfoils for high lift-to- done in a 3 foot wind tunnel using a rectangular wing
drag ratios in the correct regime. Fifteen airfoils are with a 33.375 inch span and a 12 inch chord. These
selected that have good characteristics around the dimensions suggest Selig’s decision to minimize 3D
aforementioned CL values. Finding a high ratio of aerodynamic effects and that the data collected was
CL/CD for 2D data does not guarantee a good airfoil. analogous to that of a 2D airfoil. The fact that the
Rather, Reynolds number, maximum thickness-to- span of the airfoil was so large in comparison to the
chord ratio and actual lift coefficient are also important. tunnel strongly supports this theory. If this was not the
Since the relationships are complicated and hard to case, blockage effects would be a large factor, and the
judge by inspection we input the potential airfoils into tests would be inaccurate. Efforts for backing up this
Excel solver to find the optimum geometry and assumption are outlined later.
minimum drag for each airfoil. Finally we assume that the wing weight
This completes the initial wing design. All estimation we use is accurate. Equation ( 8 ), later in
that remains is to select the proper sweep angle that the paper, uses a series of constants, as well as
approximates an elliptical lift distribution. This can be geometric and dynamic conditions, to make an estimate
done after the geometry is largely set, then optimized a for wing weight given conditions. We use this data in
second time to come up with the most efficient wing optimizations to limit span. The values from this
possible. equation agree with the values the structures team
obtain using the software Pro-Engineer and a suitable
Design Assumptions: material. This appears to be a valid assumption.
Before we begin the design process, we need
to make certain assumptions to determine the optimal
wing structure. These assumptions are made to Lift Coefficient (CL) selection:
account for the fact that we are not in possession of We begin the wing design by finding a lift
accurate data for every scope of our design process. If coefficient. Since the airfoils we are dealing with are
this design is to be finalized for physical flight on Mars, for laminar flow, the drag data is very erratic, making it
accurate weather data and atmospheric gradients need nearly impossible to find a valid curve that fits the data.
to be obtained to verify or refute our current Although the lift curve slopes from the data are very
calculations. This kind of data collection is out of the close to linear, the fact that drag is so far off makes
scope of this project and therefore we assume that our curve fitting to find continuous points virtually
atmospheric representation is correct. We have a list of impossible. Generally speaking, drag data from airfoils

Figure 1: Drag vs. a for a Standard Airfoil Figure 2: Drag vs. a For a Low Reynolds Airfoil

3
will follow a parabolic curve (see Figure 1) in the region C L vs. Reynolds Number
0.6
of the drag bucket. This allows you to fit a second
order polynomial to the data, and find values of drag at
continuous points on the curve. Figure 2 shows an 0.55

example of drag data from one of the Selig airfoils we

are considering. It can be seen that in low Reynolds 0.5
airfoils laminar bubbles and possibly hysteresis in the
switching from laminar to turbulent boundary layer C L 0.45
conditions result in erratic data. A parabolic curve of
the form C D  C D  kC L2 , where k is an arbitrary 0.4
0

constant, can give an accurate fit to the drag data in

0.35
Figure 1. Using this parabolic fit and the easily
obtained lift curve slope, we can find CL and CD values
0.3
for any angle of attack, given a wing planform and an
35000 45000 55000 65000 75000
aspect ratio. Reynolds Number
Since the data does not yield a valid curve fit,
we are forced to use the discrete values that are
provided with the Selig data. The best option for the Chart 1: Optimal CL Values
laminar regime is to find a wing that is suitable for our
purpose and test it for numerous angles of attack and Re 40000 45000 50000 55000 60000 65000
use the data acquired. This is a time consuming C L 0.555 0.521 0.468 0.453 0.403 0.359
process and is unfeasible given the time and resources
available, so our choice is to use discrete data in place Table 3: Optimal CL for a Given Re
of more expansive experimental data.
With discrete data we cannot determine an 0.35 is optimal. At the other extreme, for the low
exact lift coefficient, as the chance of finding a suitable Reynolds number case, a value of approximately 0.55 is
airfoil with that specific data point is unlikely. Instead, optimal. These optimized CL values are determined
we consider a range of lift coefficient values dependant using an Excel spreadsheet to generate feasible
on Reynolds number. As Reynolds number increases, planforms for a given Reynolds number. With
with all other variables being held constant, velocity constraints set by the user and a specific value for
increases and results in a lower lift coefficient needed Reynolds number, the solver generates optimal
for the same net lift. The opposite is also true; a low planforms for that condition. Angle of attack is one of
Reynolds number has a lower velocity, requiring a the constraints, so an optimal value for CL is found for
higher lift coefficient. For the high end values of each airfoil and entered Reynolds number.
Reynolds number, around 65,000, a CL value of around Since drag is related to velocity and we desire
a low value for drag, a low Reynolds number will yield
a lower drag. A CL of 0.55 is set as the design lift
coefficient with a corresponding Reynolds number of
40,000. However, this estimation does not take into
account base drag of different airfoils, which play a
small part in the overall drag. Further work shows that
this CL value is optimal.

Computational analysis:
Computational analysis makes up a large
percentage of the work involved in obtaining a sound
model for the airfoil characteristics. Sorting through
650 airfoils is a manageable task with Matlab analyzing
each airfoil and linking it to an Excel spreadsheet. The
initial stages involve writing code to take the data and
put it in a user friendly form. The design of a graphical
user interface (GUI), shown in Figure 3 further eases
the process of airfoil selection. The user can browse
various airfoils based on target Reynolds number and
vary plots of data until desirable curves are found.
Saving the layout allows us to return to configurations
at a later point for further review. This allows for
Figure 3: Matlab User Interface for Airfoil
Analysis
4
 Design Constraints standpoint, a large AR, between 7 and 10, is a well
Re mach number wing span Aspect Ratio designed wing. However, a large span means larger
specified Ma > .55 1.5 < b <4 Ar >3 bending moments in the wing structure, due to lift
taper ratio altitude 1/4 chord sweep loads acting farther from the root of the wing, resulting
 >.2 h >400 0 < c¼ <10 in additional weight needed to withstand the increase in
bending moment; something discussed in the
Table 4: Design Constraints Structures section.
Taper ratio is the ratio of tip chord to root
simultaneous generation of several acceptable chord, in our case 0.2. A wing with a low taper ratio,
planforms for various airfoils. referred to as “a highly tapered wing,” tends to have
The ‘Target Re’ field allows the user to search lower lift coefficients on the outer portion of the wing,
for the airfoils closest to the desired Reynolds number. as the downwash pattern changes, toward an elliptical
The fields that follow are outputted data relevant to the lift distribution. Low taper ratio also results in larger
current airfoil. This is useful for visualizing the current chords and wing thickness inboard where the bending
wing design to verify that the geometry is acceptable on moments are the largest, moving the lift in towards the
aesthetic and packaging standpoints. Buttons allow the craft, reducing the aerodynamic bending moments.
user to browse various angles of attack and cycle Both of these effects are favorable for wing structural
through different Reynolds number airfoils with ease. weight. However, low taper ratio wings have a
The user is also able to select between graphs of CD vs. tendency to stall at the tip, which is prevented with
CL, CL vs. , CD vs.  and CD/CL vs.  depending on wing twist.
the desired lift and drag characteristics of an airfoil. With the final structural and electrical weights
Excel parameterizes the planform layouts and we find that a slight increase in wing area, to produce
reduces the design problem to four variables with given more lift, is required for cruise flight. We decided that
geometric constraints: wing span, root chord, tip chord we will keep the span constant and obtain the
and sweep angle. (See Table 4) Since the initial additional required platform area by increasing the
optimization objective is to find the lowest possible taper ratio. The final taper ratio is 0.35, which leads to
drag for reasonable geometric constraints, overall wing a platform area of 0.6 m2, a new AR of 7.2 and an
weight is a secondary factor in comparison to drag increased safety factor of greater than 1.2
optimization, and thus is not included as one of the Geometric twist is the equivalent of taking a
constraints. straight wing, and applying a moment about its axis,
causing the tip of the wing to be at a different angle
than the root of the wing, in our case, negative three
Airfoil Selection: degrees. A positive value of twist refers to an increase
Using the Matlab GUI with the Selig data we in angle of attack along the span of the wing. A
can narrow down the 650 airfoils to 9. This is done by negative twist value, known as “washout,” greatly
selecting airfoils that have high CL/CD values at the increases the stability of the craft by causing stall to
design lift coefficient of 0.55 and a Reynolds number occur at the root of the wing before occurring at the
between 40,000 and 65,000. As Reynolds number wing tip. When this happens, control can still be
increases, the CL/CD values required for a feasible maintained due to the moment that can be generated
airfoil increase due to the need for a lower aspect ratio from the tips of the wings to control the rolling motion
and the resulting increase in induced drag. This results of the craft.
in fewer airfoils that meet our goal as the Reynolds
numbers increase.
With secondary optimization of the nine
airfoils selected, six have favorable characteristics. Only
one of these surpasses the others in both low drag and Wing Drag(N) Mass(kg) b(m) Sweep c root (m) C tip (m)
low weight (see Table 5). Airfoil gm15 makes possible
gm 15 2.4442 0.82 2.078 4.9 0.422 0.084
a planform of low weight, short wing span and low s6063 2.9398 1.86 3.1 7.1 0.966 0.193
drag. Note that sweep value refers to the sweep angle s7012 2.5381 1.41 2.922 4.9 0.627 0.125
added in addition to the sweep induced from the taper rg14 2.8761 2.22 3.645 7.1 1.131 0.226
ratio. The actual leading edge sweep angle will be rg14 2.8331 1.33 2.848 5 0.628 0.126
higher than this value. gm 15sm 2.6638 2.55 4 6.2 1.087 0.217
Based on the geometric properties for our gm 15sm 2.4501 1.59 3.162 4.6 0.631 0.126
selected airfoil, our calculations yield an aspect ratio of sd7003 2.5914 2.62 4 6.6 1.164 0.233
8.22 and a wing platform area of 0.525 m2. The aspect sd7003 2.5313 2.34 4 4.8 0.84 0.168
ratio is defined as b2/S; for a given wing area, S, a large
aspect ratio means a large span. From a drag Table 5: Multiple Optimization Results for 6 wings

5
Mean Aerodynamic Chord & Center lift at the tip. A sharp edge makes it more difficult for
Mean aerodynamic chord ( c , M.A.C.) is a the air to flow around the tip, because the flow often
parameter, directly associated with the Reynolds separates at these edges. A winglet or endplate blocks
number ( Vc /  ). Equation ( 6 ) shows the M.A.C. the flow from the bottom to the top of the wing. This
offers the greatest benefit to low aspect ratio wings
as a function of the taper ratio () and the exposed root whose wing tip vortex is strong, by increasing the
chord (CR). effective wingspan, which further decreases induced
drag. Our concern with using a winglet is the
2 
M . A.C.  C R (1    )  0.307 (6) additional wetted area, which will create a larger
3 1  parasite drag nullifying any reduction in induced drag.
An unswept wing tip curves upward to increase the
Y, the distance of the M.A.C. from the centerline of
effective wing. We are using this design since it is
the aircraft is 0.436 meters. This distance is dependent
similar to adding a winglet, without an increase in total
upon the taper ratio and the wing span. Aerodynamic
wetted area.
center is the point on the aircraft where the airfoil
pitching moment is constant with a change in angle of
Wing Weight and Structural Considerations:
attack. It determines where to position the wing, and is
With any extraterrestrial mission weight is a
important in stability calculations. In subsonic flow,
top priority. Our mission is no different. Generally,
the aerodynamic center is typically located at the
wing weight selection is an iterative process between a
quarter-chord point on the mean aerodynamic chord
structures team and an aerodynamics team. Due to
line, which is found to be 0.0768 meters from the
time constraints, we approximate the wing weight with
leading edge of the wing.
equation ( 8 ), a formula based on historical data used
From equations historically used in aircraft
by many aircraft manufacturers as an initial wing
design, such as Equation ( 7 ), we conclude that the
weight. The formula is a relationship between,
aerodynamic center will be located 0.703 meters from
dynamic pressure, q, aspect ratio, AR, total aircraft
the nose of the UAV.
weight, Wdg, thickness-to-chord ratio, t/c, load factor, n,
A.C.  1.5C R  0.25c (7) taper ratio, , wing area, Sw, sweep angle, , and a
multitude of constants, C1 through C14, that have been
obtained using years of data for three classes of aircraft:
Wing Vertical Position fighter, transport and general aviation. Using the
The wing’s vertical placement with respect to general aviation constants and the planned weight of
the fuselage can be at three locations: atop, below or the craft, 10 Kg, a circular reference of wing area, lift
through the middle of the fuselage. A high wing design force, total craft weight and wing weight is created in
is used primarily for cargo planes, allowing the fuselage Excel. Turning on the iteration command in Excel
to be placed closer to the ground. However, the causes the values to converge to a steady state solution
passing of the wing box over the fuselage will increase for weight analysis, solving with ease a process
the parasite and pressure drag due to the increase in otherwise overly complicated by hand.

 c
frontal area. This increase in frontal area is also present C8
in a low wing design, which is used by virtually all Wwing  C1C2C3WdgC4 n C5 S wC6 AC7 t
commercial transport aircraft due to the advantages in (8)
landing stowage. Since we are not concerned about ...C9    cos  C
C10 11
S Cf12 q C13W fwC14
landing, we have no need for a low wing design. The
advantage to a mid-wing design is that it gives the UAV
more maneuverability while having a lower frontal area
than the high or the low wing design. Due to the
advantages associated with a mid-wing design, we are
placing the wing in the middle of the fuselage.

Wing Tip Selection

A wing tip can prevent high pressure air
beneath the wing to “escape” around the tip of the
wing to the low pressure region above, resulting in a
loss of lift at the tip of the wing, which is highly
undesirable. Four different wing tip designs are (a) (b)
considered: rounded, sharp, winglets and unswept. A
smooth-rounded tip is precisely what we want to avoid. Figure 4: (a) Upswept Wingtip (b) Inverted V-Tail
It easily permits air to flow around the tip and reduces

6
Tail Arrangement Selection them. Plotting the CL/CD vs. angle of attack () data
Tails act as small wings; their purpose is to for all four, we find the tail desired angle of attack to be
provide trim, stability and control to the craft. Trim is between 3 and 5 degrees. By comparing the values of
the generation of the proper lift force to balance CL/CD at 3 degrees for each candidate, it is apparent
pitching moment about the center of gravity. Stability that SD7003 is the best airfoil at the value we need for
and control are the tail’s ability to restore the aircraft angle of attack.
from a perturbation in pitch, yaw, and roll, which is
discussed in detail in the section of stability and Stability and Control
control. Stability and control is an integral part of
There are a variety of possible aft-tail designing an aircraft. It is vital that the aircraft is stable
arrangements. Our design focus is on reducing parasite and able to handle moments, from various
drag. We can thus narrow our search to four possible disturbances, while maintaining control. An aircraft
tail configurations based primarily on the wetted area possesses three degrees of freedom, pitch, roll, and
of each tail configuration: conventional, V-tail, inverted yaw, and has two types of stability, static and dynamic.
V and Y-tail. A system is statically stable if forces and moments
Conventional tails, used on over 70% of all acting on a body, as a result of a disturbance, initially
aircraft, have the typical vertical and horizontal tail seen act to return the body towards its equilibrium position.
on most commercial airliners. It provides adequate A system is dynamically stable if it eventually returns to
stability and control at a reasonably light weight. With and maintains its equilibrium position over a period of
the V-tail, as the name suggests, the vertical and time. For our case, our top concern is longitudinal
horizontal tail components are combined in an attempt static stability, involving the pitching moments about
to reduce the wetted area. The horizontal and vertical the center of gravity. Though, as with any aircraft,
forces, on the V-tail, are the resultant of their lateral-directional static stability and control involving
respective projections from the two angled surfaces. yaw/roll moments are also important.
In order to provide the proper movement, the The steps to design a stable aircraft are as
rudder and elevator on a V-tail are combined to create follows:
“ruddervators.” The problem with a V-tail is the 1) Make an assumption for the location of the
production of a rolling moment in opposition to the center of gravity with respect to the nose of
desired direction of turn, known as “adverse roll-yaw the aircraft
coupling.” This produces a spiraling tendency when the 2) Make an educated guess regarding the
UAV is making a turn. The inverted V-tail avoids this placement/sizing of the tail
problem; it instead produces a desirable “proverse roll- 3) Determine the moment about the center of
yaw coupling”. gravity due to the wing, fuselage, tail, and
The “Y-tail” is similar to the V-tail, with a payload
reduced dihedral angle and a third surface mounted 4) If the moment coefficient, calculated in
vertically beneath the V, giving the UAV more yaw equation ( 9 ), at zero lift (Cm L=0) is positive
control. A drag penalty involved with adding another and the slope of the moment coefficient
control surface causes the Y-tail design, like the versus angle of attack (dCM/d) is negative,
conventional design, to not fit our design goals of the aircraft is longitudinally and statically
minimizing drag. Using an inverted V-tail gives us the stable
low drag required with greater stability than the 5) Reiterate the process if necessary
standard V-tail. 6) Determine the static margin
Tail Airfoil Selection M cg
The goal in selecting an airfoil for the tail is CM  (9)
similar to the wing, in that we want an airfoil with high
1
2 V 2 Sc
lift-to-drag ratios. Since tails are small wings
themselves, we look at the final low Reynolds number Equation ( 9 ) solves for the moment coefficient about
airfoils from the wing selection process. The main the center of gravity, where M is the moment
purpose of tails is not to generate lift, but to provide contribution about the center of gravity,  is the
stability and control. For this reason, airfoils used for density of the atmosphere, V is the cruise speed, S is
tails typically have little to no chamber to them. Since the wing platform area and c is the mean aerodynamic
our wing airfoil has high camber, the chosen wing chord, dependent on taper ratio.
airfoil, gm15, is not applicable for the tail. Based on lift We use equation ( 10 ) to determine the
and drag data, as well as amount of camber, we limit moment coefficient about the center of gravity at zero
our choices to four airfoils: SD 7003, S6063, S7012, lift; recall that this value must be positive for the
and RG14, all of which have only slight camber to aircraft to be stable. The first term on the right hand

7
 side of the equation is the moment contribution from St, to be 0.15 square meters and the tail moment arm,
the wing and the fuselage, about the center of gravity. Lt, to be 1 meter. Solving for tail moment coefficient
The second term is the moment contribution from the yields a value of 0.25, which more than compensates
tail. CM(payload) is the moment induced by the payload. for the moment coefficient from wing, fuselage and the
The moment coefficient about the center of gravity payload.
from the wing and fuselage, CM,CGwb, is approximated as
  d  
  h  hacwb   V h  t /  1 
the sum of the moment contribution of the wing body dCm

d 
about the aerodynamic center and the moment ( 12 )
d  
generated by the lift force from the wing,
M cgw  M acw  LW cos  w (hc  hacw ) . Drag terms are With the first stability criteria satisfied, we can
not included in the calculation of wing and fuselage solve equation ( 12 ) for the second criteria, negative
moment coefficient, as they are negligible compared to dCM/d Where  and tare the lift curve slopes of the
the other terms. Solving equation ( 10 ) gives us a wing and the tail respectively, the quantity (h - hacwb) is
value of 0.041 for CM,CGwb. the distance between the center of gravity and the
aerodynamic center, and d/d is the rate of change of
C 
M ,cg L 0  C M ,cgwb  VHat (it , eo )  C M ( payload ) ( 10 ) the downwash angle with respect to the angle of attack
for the wing, approximately 0.45. dCM/d satisfies the
To determine the moment coefficient from second stability criteria with a value -0.03
the payload, we establish a series of point loads to With both of the criteria satisfied the UAV is
approximate the weight distribution. This layout can longitudinally and statically stable. The neutral point is
be seen in Figure 5. According to our calculations the a fixed point on the UAV behind the center of gravity
center of gravity will be 0.85 meters aft of the nose of where dCM/d is equal to zero, and must be aft of the
the UAV. Moment calculations yield a moment center of gravity in order to achieve longitudinal
coefficient value of -0.153 (negative moment being in stability. Setting dCM/d to zero and solving for h gives
the clockwise direction.) the location of the neutral point at 1.06 meters from
In order to satisfy the first stability criteria, a the nose of the craft. The static margin, the difference
CM,cg greater than zero, the tail must be large enough to between the neutral point and the center of gravity, is
balance the clockwise moment produced by the 0.217m. This parameter is directly related to the
payload and lift forces. According to our calculations stability of the UAV, the larger the static margin, the
the tail moment coefficient must be larger than 0.194. greater the pitching moment must be to cause a change
The moment coefficient is defined as Vhat(it+eo), where in the pitching angle. However, too large a static
Vh is the tail volume coefficient, at is the lift curve slope margin may cause the flight controller to go unstable
of the tail airfoil (~ 0.1), it is the tail setting angle (~3 due to unacceptably high reaction latency. On the
by standard convention), and eo is the downwash angle, other hand, too low of a static margin will yield an
which can be neglected. aircraft that is inherently unstable in regards to pitching
lt St motion, and will require very fast control response to
Vh  ( 11 ) maintain steady state cruise conditions. Our static
cw S w margin is between these two extremes, allowing
stability without an over-active controller.
Vh, the tail volume coefficient, is proportional to lt , the In many ways, the lateral-directional analysis
distance of the tail from the center of gravity, St ,the resembles the longitudinal analysis. Lateral-directional
platform area of the tail, cw , the M.A.C. of the wing and stability is the tendency of the UAV to return to a
Sw , the wing platform area. Equation ( 11 )shows that wing-level attitude after being displaced from a level
the further away the tail is from the center of gravity, attitude by roll or yaw moments, from such things as
the smaller the area of the tail needs to be. We want to turbulent air. There are two primary factors for lateral-
make the tail large enough to give us adequate stability, directional stability: wing dihedral angle and wing
but not as to further increase the parasite drag. To sweepback angle. Dihedral angle is the angle at which
optimize both parameters, we set the tail platform area, the wings are slanted upwards from the root to the tip;

0.25 0.65 0.75 0.85 1.60 1.75

0m 2.1
cg m
1.17kg Lidar 4.2kg 0.61kg 1.8kg 0.43kg 0.6kg
Controls Battery Camera Motor drive shaft Propeller

Figure 5: Payload Point Mass Layout for Aerial Vehicle

8
its main purpose is to correct roll moments. The Furthermore, being a 6 degree of freedom system, a
stabilizing effect of dihedral occurs when an aircraft soundly designed multi-input/multi-output, or MIMO
sideslips slightly as one wing is forced down in controller to be used is required for an aircraft. The
turbulent air or during a turn. This sideslip results in a designing of such controllers are currently beyond our
difference in the angle of attack between the higher and expertise.
the lower wing. The increased angle of attack on the
lower wing produces an increase in lift which helps the Cruise Performance
wing return to its level position. Research leads us to a The stall speed is determined directly by wing
dihedral angle of 3.5 degrees. loading and the maximum lift coefficient. Stall speed is
Sweepback is the angle between the line a major contributor to flying safety, indicating the
formed from the front of the wing and the line minimum speed that will keep the UAV aloft. When
perpendicular to the centerline, in the plain of the an aircraft flies below the stall speed, flow around the
aircraft. The effect of sweepback in producing lateral airfoil begins to separate; as a result, a rapid loss in lift
stability is similar to that dihedral angle. A yaw will be experienced. At that point, if the velocity is not
moment increases the sweepback angle in one wing increased beyond the stall speed, the aircraft will lose
panel and decreases it for the other side of the aircraft. altitude and thus lose control. Equation ( 13 )
The change in sweep alters the effective dynamic determines the stall speed where W is the aircraft
pressure normal to the quarter-chord line of the wing weight, is the fluid density, S is the wing platform
panel, increasing the lift on one side of the wing, area and CLmax is the maximum lift coefficient of our
lowering it on the other side, and producing a restoring airfoil, obtained from the graph of CL vs. angle of
moment. Historical trends in wing sweep back give us attack. Our aircraft has a stall speed of 83.4 m/s; we
a sweep angle of 5 degrees. must fly above this speed in order to maintain adequate
lift.
Tail Geometry Since our main goal is to map as much terrain
Having found the platform area of the tail, St as possible, it is necessary to maximize the range. To
based on stability constraints, it is important that we do so, we must fly at the speed where lift to drag ratio
obtain the tail geometry. Determining the aspect ratio is greatest, given in equation ( 14 ) where CDP is the
is crucial; having a large aspect ratio corresponds to a parasite drag coefficient and AR is the aspect ratio of
small chord, which further leads to an unusually low the wing.
Reynolds number. Likewise, a small aspect ratio will
lead to a substantial increase in induced drag. After Vstall 
2W
( 13 )
much consideration we decided on an aspect ratio of 4.  SCl MAX
Using a taper ratio similar to that of the wing, 0.35, we
determined the length of the tail root and tip chord.
Similar to that of the wing, the tail mean aerodynamic 2W
V L / D max  ( 14 )
chord ( ct ) is a function of the root chord and the taper  S C DP   AR  e
ratio. Based the ct value, we determined the tail Range is maximized when VL/D is 130 m/s
Reynolds number, which is approximately 30,000. This (Mach number of 0.56). If the cruising speed is set to
value is in fact only half of the optimum Reynolds be greater than this value, then there will be a
number suggested for our selected airfoil (SD7003). substantial increase in drag. Values deviating from this
However, because the tail angle of attack will be small, velocity will result in a loss of range.
boundary layer separation would almost be nonexistent.
Therefore it is ok for the operating Reynolds to be Turning Performance
smaller than the optimal value. Maneuverability plays an important role in the
design of an aircraft. Unlike combat planes that
Control perform sharp turns, our UAV only needs to be able to
The primary aerodynamic controls available slowly turn to avoid physical obstacles that it may
are ailerons, elevators, and rudders. Because we are encounter. A crucial parameter in trying to figure out
incorporating a V-tail design, the functions of elevators the turning performance of the aircraft is the load
and rudders will be combined into one. Ailerons are factor, n, defined as the ratio of lift-to-weight. In our
the primary roll-control device, which operate by case, n is approximately equal to 1.095. The wing bank
increasing lift on one wing and reducing it on another. angle,  defined as cos-1(1/n), is approximately 24
They range from 50 to 90 percent of the wingspan and degrees. Ailerons control the wing bank angle by equal
20 percent of the wing chord length. Since the aircraft and opposite deflection of the two wings trailing edges,
is unmanned, a suitable, redundant controller must be one up and one down; thus increasing lift on one side
designed to stabilize and maneuver the aircraft. and decreases lift on the other side of the aircraft,

9
 CLvs. Angle of Attack & CL/CD vs. Angle of Attack
inducing a roll moment about the centerline. A rolling 1.4 13.8
moment banks the airplane and tilts the lift vector to
13.6
one side. The horizontal component of the lift vector 1.2
accelerates the aircraft laterally, thereby curving the 13.4
flight path. Equation ( 15 ) solves for the minimum 1.0 13.2
radius of turn, R.
13.0
0.8

C L /C D
2
V CL 12.8
R  ( 15 ) 0.6
g n 12
12.6

0.4 12.4
Because of our relatively small load factor, n, we obtain
a value of 10.5 km for our turn radius. The UAV is not 12.2
0.2
capable of performing a sharp turns; this is not a 12.0
problem as we will have adequate warning for any turns 0.0
Cl SELIG Data Cl/Cd
11.8
that need to be made and the controller can take into 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
account minimum turn radius. Equation ( 16 ) Angle of Attack
describes the turning rate (degrees/time) for the UAV, Figure 7: CL vs. for CFD (Green) and Selig (Red)data
how large your turn is, in degrees, per unit time.
CL/CD vs.  for CFD (Blue)
d g n^2  1
 ( 16 )
dt V
outlining the 2D airfoil obtained from Professor Selig’s
Fluent Analysis (Selig Data Verification)
database and resizing it for our chord length.
It is stated in the aerodynamic assumptions
Once FLUENT reads the mesh we specify
section that Professor Selig’s data for his numerous
the atmospheric and flying conditions. In this case the
airfoils are correct. Using Fluent, a computational fluid
closest to Martian atmosphere that can be used is an
dynamics (CFD) package, we justify this assumption
environment of carbon dioxide. In FLUENT we set
with appropriate calculations. While the Selig airfoils
are normalized by Reynolds number, Martian the fluid properties to the values from Table 1 in the
conditions vary greatly from those in Professor Selig’s section on atmospheric data and a gravitational
wind tunnel. constant of 3.72 m/s2, roughly 4/10ths the magnitude
The first step to CFD is creating a mesh of gravity on earth. Boundary conditions are set that
containing the airfoil in a large space with boundary specify the pressure far from the airfoil and the velocity
conditions to simulate the Martian environment for our in terms of x-y components and Mach number. The x-
airfoil. This is done, using GAMBIT, from grid points y components allow us to vary the angle of attack
without the need of creating a new mesh for each angle
of attack we want to test
With fluid and environmental properties set
we determine the appropriate method to use to
perform calculations. Since we are flying in with a low
Reynolds number we use a laminar boundary regime.
To verify our assumption about the Selig data
it is necessary to run simulations at several angles of
attack. With a chart of this data we can compare the
CFD data with that from Selig and find that, while they
do not lie directly on top of each other, they both yield
an acceptable coefficient of lift for an angle of attack
between two and three. The CFD data gives slightly
higher CL values than the Selig data. Since we are
backing up an assumption based on experimental data,
not determining values, this inconsistency is acceptable.
If this error were on the side of lower lift we would
have to do more calculations to make sure there is no
problem. As this is not the case, it stands that our
assumption regarding the validity of the Selig data is
Figure 7: FLUENT Pressure Gradient Around Airfoil acceptable.
(Red = High Pressure Blue = Low Pressure)

10
Propulsion Design .
Propeller Design Overview
Introduction From past Mars aircraft concepts and high
Low atmospheric density and the lack of altitude, low speed Earth aircraft, propellers have been
appreciable amounts of atmospheric oxygen complicate the preferred choice. Our choices for powering an
the propulsion for a Mars airplane. These constraints electric motor are: batteries, fuel cells and solar cells
lead to the consideration of propulsion options that are (see discussion on solar cells in the Electrical section.)
more restrictive than those of Earth. As a result of the lower speed of sound on Mars, due to
The analysis carried out in the propulsion low temperatures, and density about one-hundredth
section of this report is based on an airplane that is not that of Earth’s, our effort focuses on generating the
landing intact on the surface of Mars once flight is necessary amount of thrust, as well as keeping the tip
completed; if an airplane is intended to land or take-off speed of the propeller below supersonic conditions. If
from Martian soil, a new set of design specifications the tip Mach number reaches 0.85, there will be a large
need to be considered. drop in the propeller efficiency due to the flow
separation and formation of shockwaves.
Propulsion Selection
Since the use of a combustion engine is not Propeller Placement
feasible due to the lack of oxygen, our choices for the Examining the advantages and disadvantages
propulsion subsystem are limited. There are two of propeller placement along the fuselage places the
methods to propulsion we consider for Martian aerial propeller at the rear of the fuselage. The main
flight: chemical propulsion and propeller driven advantage in using a pusher is in the aircraft’s capability
propulsion by an energy source. to fly in undisturbed air. With a tractor propeller, the
The use of monopropellant rocket thrusters aircraft flies in the turbulence from the propeller wake,
enables combustion without the need for atmospheric which could lead to additional drag.
oxygen, by carrying chemical compounds that burn
spontaneously when ignited. It provides the UAV with Drag Calculations
uniform thrust; however, once ignited, the process The UAV will be operating at steady, level
cannot be stopped until the fuel runs out. flight, where all the forces will be in equilibrium,
Bipropellant thrusters, on the other hand, meaning that thrust must balance the drag in order to
carry fuel and oxidizer separately. They are more keep the UAV at a constant cruising speed.
practical in this case since the thruster can be turned on C D  C DP  C DI  C DC ( 17 )
or off in order to maintain cruise speed at V(L/dmax).
Bipropellant thrusters, however, tend to be more Equation ( 17 ) calculates the total drag on the
complicated to design. The thrusters found for our aircraft, where CDp is the parasite drag coefficient (also
design constrains are capable of generating anywhere known as skin friction drag coefficient,) CDi is the
from 5 to 20 N of thrust and have a specific impulse,
induced drag coefficient and CDc is the drag due to
Isp, in the range of 300 to 350 seconds. Isp, a key
performance parameter for rockets, is defined as the compressibility. CDc becomes significant when the
thrust that can be obtained with a propellant weight craft approaches sonic condition. However, because
flow of 1 unit per second. Modern large scale rockets, the UAV will operates at subsonic speed, CDc is be
like the one found on the Shuttle, can achieve a neglected.

C f k swet 
maximum Isp of around 450 s.
The second approach to propulsion we C DP ,wing  ( 18 )
consider is the use of energy from an on-board battery, 4
nuclear device or an off-board energy source, such as
solar energy, to power a propeller. Solar powered The parasite drag, coefficient from the wing is
airplanes must have a large projected area to collect solved in equation ( 18 ), where Cf and k are roughness
sufficient solar power and are inefficient when the solar constants based on the wing Reynolds number and Swet
intensity is low, as it is on Mars. is the total wetted area for the wing. Similarly, we are
Solid rocket propulsion is inherently simple; able to determine the parasite drag coefficient for the
yet, as mentioned before, there is no way to control the tail and the fuselage. Assuming that the wing, fuselage
thrust once ignited. Bipropellant thrusters run the risk and tail contribute to 95% of the total skin friction
of explosion due to low atmospheric temperature on drag, the overall parasite drag coefficient is determined
Mars. Our calculations also indicate that for the same to be 0.02225.
weight, a battery driven propeller would yield a much
greater range than for a bipropellant thruster, so we
select propeller as our form of propulsion

11
 addition, the use of propfans, which feature 8 to 10
2
C wide, short blades of sweptback planform are
C DI  L  0.016 ( 19 )
considered for blade configurations. If a propfan blade
Ae configuration can be utilized, being powered by an
Using equation ( 19 ) we solve for the induced electric engine, opposed to the standard turboprop
drag coefficient, where CL is the wing lift coefficient, A engine, it would be an option worth considering.
the wing aspect ratio, and e the Oswalt efficiency factor. Although counter-rotating blades and propfans were
Combining the parasite and induced drag coefficients investigated as ways to improve propulsion, the
gives an overall drag coefficient of 0.0394. aforementioned roll moment is beyond our current
Thus we can determine total drag force level of expertise.
experienced by the aircraft, from equation ( 20 ), to be
3.14 N. D prop  0.54 Pengine  2.27 m ( 22 )
1 
D  CD  V 2 S   3.14 N ( 20 )
2  Propeller Efficiency
As noted earlier, in the section it is essential
that we keep the tip speed of the propeller under M =
Motor Power Specification
0.85 or 195 m/s. The helical speed, the tip velocity on
With the thrust required for level flight
a moving aircraft, is the sum of the rotating speed at
known, we specify the amount of power the motor
the tip of the propeller and the freestream velocity,
needs to produce in order for the propeller to generate
calculated in Equation ( 23 )
that much thrust.

TVo   prop Pengine  ( 21 ) V 

tip helical  Vtip  V freestream
2 2
( 23 )

We determine the engine brake horsepower, Pengine,

where T is the thrust required to maintain level flight, Vtip  nD ( 24 )
V is the flight velocity and η is the efficiency of the
propeller. Since the UAV will be cruising at V(L/Dmax) The stationary tip velocity is calculated in
the flight velocity is 130 m/s (see aerodynamic section Equation ( 24 ),where n is the rotation speed in
for more details.) For a propeller efficiency of 95%, revolutions per second and D is the diameter of the
The engine will produce a power of 430 W or 0.58 HP. propeller. With a helical tip speed of 195 m/s, we
Although calm flight conditions are assumed, obtain propeller rotation speed of 20 revolutions per
it is highly likely that there will be significant wind gust second, or 1200RPM.
that will increase the drag value. Taking this factor into The overall propeller efficiency is expressed in
account, we impose a safety factor of 1.2. The terms of the advance ratio v/nD, thrust coefficient CT,
propeller must therefore be capable of generating 3.77 and power coefficient CP in equation ( 25 )), where CT
N of thrust if necessary to maintain leveled flight; this and CP are defined in equations ( 26 ) and ( 27 ). The
corresponds to a maximum engine power of 490 W or resulting propeller efficiency of approximately 93% is
0.675 HP. very close to the 95% value we assumed initially in
calculating power.
Propeller Diameter
v CT
Using Equation ( 22 ), we determine the  ( 25 )
diameter of the propeller from the brake horsepower n * D CP
of the engine. Note, this diameter is equivalent to the
length of our wing span; as the propeller rotates an
T
induced roll moment is generated. We further explore CT   0.0188 ( 26 )
the possibility of using counter-rotating blades. n 2 D 4
Counter-rotating blades have mainly two advantages:
they are more efficient at high Mach numbers than a ( 27 )
P
single propeller configuration and they allow a smaller CP   0.056
diameter blade, allowing them to spin at higher RPM n 3 D 5
without a loss in aerodynamic efficiency. By having
counter-rotating blades, our propeller diameter will be
decreased considerably while maintaining the same
efficiency. However, the extra blades will increase the
weight compared to the use of a single propeller. In

12
Propeller Pitch blade will be traveling at different speeds. A small twist
The pitch is the theoretical distance the in the propeller blade must be incorporated to ensure
propeller will advance along the axis of rotation in one that each section advances forward at the same rate
complete revolution. There are two types of propellers: which stops the propeller from bending.
fixed pitch and variable (controllable) pitch. Designing propeller blades takes a great
In a fixed-pitch propeller, the pitch is set by amount of expertise and years of experience. Since it is
the manufacturer and cannot be changed by the pilot. beyond our level of expertise, we will not determine the
There are two types of fixed pitch propellers: the climb exact pitch, shape, twist, and airfoils for the propeller.
propeller and the cruise propeller. The climb propeller
has a lower pitch, which therefore leads to less drag. Propulsion Summary
This results in the capability of higher RPM and more As previously mentioned, the power
horsepower being developed by the engine; such will consumed by the propulsion system will be 430 W/hr.
increase performance during takeoffs and climbs but The 4.2 kg of battery that is carried onboard will
decrease performance during cruising flight. On the generate a total of 1050 W, which allows the UAV to
other hand, the cruise propeller has a higher pitch and stay aloft for approximately 2.3 hours. At a cruising
therefore more drag which results in lower RPM and speed of 130 m/s, this corresponds to a range of 1076
less horsepower capability. Performance during takeoff km.
and climb is therefore decreased; yet, efficiency during
cruising flight is increased.
Contrary to the fixed pitch, a variable pitch Structural Design
propeller permits the pilot to select a pitch that will
result in the most efficient performance for a particular Introduction
flight condition. Since we are solely dealing with cruise In structural aspects, the objective is to design
flight conditions, we select the fixed pitch cruise and verify the safety, stability, and reliability of the
propeller for its simplicity and performance. unmanned aerial vehicle. Both the wing and fuselage
will be hollow in order to minimize weight and house
Propeller Blade Design instrumentation needed for the mission. The structure
In propeller design, deciding the number of must also be able to withstand outside forces as well as
blades to incorporate is essential. An optimization its own weight.
among efficiency, thrust and weight shows that a three-
bladed propeller is preferred. Not only is it capable of Structural Design/Fabrication
producing more thrust than would a two-bladed Two procedures are considered in designing
propeller, it is also lighter and more efficient than using the structural body of the unmanned aerial vehicle.
a four-blade propeller. The first process consists of making a skeleton with
The tip section of the propeller revolves faster trusses and placing coatings of outer layers on top of
than the root section; therefore, the Reynolds number this system to form the external shell. The second
along the propeller changes as the radius increases. As option is to make a mold out of Styrofoam or a similar
a result, one would have to select a different airfoil for solid foam material in the exact shape of the aircraft
each section of the propeller blade. and then coating the outer layers of skin on top of the
mold. After the layers are set, the inner mold is
removed and only the thin shell remains, but shaped in
the form of the aircraft.
The option implemented in this project is the
latter process of coating a mold. This is primarily due
Figure 8: Propeller Blade Illustration to weight considerations. Although trusses increase the
weight of the aircraft, a few trusses will be used for
The optimum airfoil thickness will be around support. The trusses will serve solely for structural
15 to 18 percent near the root, progressively thinning purposes, and not to form the shape as do the trusses
to 10 percent at the tip. in the first process.
Propeller blades are in fact wings themselves, The coating of the aircraft will consist of three
producing a resultant aerodynamic force that may be layers: a base layer, a middle body layer, and a surface
resolved into a force pointing along the axis of the finish. The materials chosen for the first two layers
airplane. Thus, similar to the airfoil for our wings, the must have a low density to optimize weight, but must
blades should have a high aspect ratio in order to also be structurally sound. In addition, the materials
minimize drag. An elliptical-based shape blade with a must be capable of withstanding the extreme
rounded tip would yield optimal performance. temperatures in the Martian environment, which reach
As the propeller spins, each section of the on average -63°C on the surface. The material picked

13
for the bottom layer is 0.1mm aluminum. The body
layer will be 0.5mm carbon fiber, and the outer layer
will be polished aluminum.

Wing
Several approximations are made in designing
the wing of the aircraft. First, the wing is treated as a
cantilever beam and second, as being hollow. The Figure 9: Wing Model
following sections will explain further analyses on the The thickness at the center of gravity, “y”, is
wing. also highly approximated—the front face of the wing is
modeled as a 2-dimensional trapezoidal figure. From
Wing: Material Considerations our trapezoidal face, we can find a linear relationship of
In choosing the wing material, the following thickness, y, to position along the length of wing, x.
parameters are of importance:
 Low density, ranging from approximately  y@root  y@tip 
y   x  y@root  0.0116x  0.0185 ( 28 )
1300 kg/m3 to 1700 kg/m3  l 
 High fracture toughness (FT) approximately  total wing 
1.0 x 105 Pa-m1/2 Upon determining the horizontal component of the
 High tensile strength that is capable of: center of gravity, we can find the thickness at this
a) Supporting instrumentation weight point, y. Values can be found in table 7.
b) Resisting forces of lift, 44.0 N
c) Resisting forces of drag, 3.20 N Carbon
Material Aluminum
d) Resisting forces of gravity Fiber
Density(kg/m3) 1500 2700
 Usage temperature between -80oC and 40oC
Center of Gravity (m) 0.4342 0.4342
 Easily molded, shaped, and machined Thickness (m) 0.0005 0.0001
Base (root/tip) (m) 0.422 0.152 0.422 0.152
Given these material parameters, we identify carbon Height (root/tip) (m) 0.037 0.0129 0.037 0.0129
fiber to be optimal; its material characteristics are Length (m) 1.039 1.039
shown in Table 6: Y (m) 0.0135 0.0135
Property Value I (m4) 1.781e -6 1.781e-6
Density (kg/m3) 1500 Total Volume (m3) 2.469e-4 2.469e-4
Fracture Toughness (Pa-m1/2) 5.7
Tensile Strength (MPa) 13.9 Mass (kg) 0.3704 0.6667
Force, weight (N) 1.370 2.467
Young’s Modulus (GPa) 71 Force, lift 22.0 22.0
Hardness-Vickers (HV) 42
Moment (Nm) 8.957 8.481
Temperature, min (C) -273 Stress (Pa) 8.347e5 7.903e5
Temperature, max (C) 2002
Table 7: COG thickness and Aluminum Stress
Table 6: Carbon Fiber Properties Analysis
Wing: Stress Analysis
It is crucial that the structure of the UAV be Varying wing thickness corresponds to
strong enough to withstand aerodynamic forces while varying values of mass—and therefore, varying values
in flight. We therefore perform stress analysis on the of gravitational force. Because the aircraft is
wing to verify its capability of flying in Mars. The lightweight, the force of gravity is nearly negligible.
overall wing dimensions are known based on This force of gravity and 44 N force due to lift provide
calculations from aerodynamics; the thickness of the the moment used to determine bending stress. The
hollow wing is dependent upon the yielding of carbon moment of inertia for the approximated rectangular
fiber. The wing is approximated as a cantilever beam cross-section is determined at the center of gravity.
with its airfoil modeled as a rectangular cross-section,
as seen in Figure 9. I  I outer  I inner 
1
12

bh 3  b  2t h  2t 
3
 ( 29 )

For a wing of 0.5mm thickness, we can

calculate the bending stress.
My
   0.8347 MPa ( 30 )
I

14
 Thus, since the yield stress of carbon fiber is the fuselage; the batteries would therefore be on the
known to be 300MPa, we can conclude that a wing of top and mid floor.
0.5mm thickness will not yield.
Performing similar calculations for the inner Fuselage: Thermal Design
aluminum core layer, we calculate a bending stress of (Note: all the equipments will be placed inside
0.7903MPa. In comparison to its yield strength of the fuselage; therefore we are solely concerned about
approximately 20MPa, we confirm that the aluminum designing a thermal control system for the fuselage
will not yield. Results are depicted on table 7. only)
The main purpose of having a thermal control
Wing: Buckling system is to maintain batteries and instruments at their
After determining that the wing will not yield, optimal operating temperature. There are two basic
it is important to confirm that the wing will not buckle. approaches to the design of a spacecraft’s thermal
Pcr, critical load, is found at the tip of the control system – passive and active. Passive control
wing, where buckling is most likely. The load that the operates by using appropriate materials and surface
UAV will experience due to lift and gravitational force finishes so that the fuselage temperature remains within
of each material is approximately 26 N, well under the acceptable range of temperatures. The latter uses
critical load of 1613 kN. Thus, the wing will not mechanical or thermoelectric devices. For instance, the
buckle. UAV would consist essentially of a central thermal
transfer bus, a fluid loop transporting the heat from the
 2 EI radiator to the individual components. We choose a
Pcr   1.613e6 N ( 31 ) passive thermal control system because it is more
L2
reliable and easier to design than an active one.
The ideal operating temperature range for our
Fuselage instruments is approximately 270-290 K. Due to the
Once we have established the dimensions of the wing, extremely low temperature on Mars and the fairly high
we must now consider the design of the fuselage. The velocity that we will be flying at, heat losses due to
shape of the fuselage will be one similar to sailplane conduction and convection would be significant.
design, for weight minimization and aerodynamic Therefore, we want to incorporate a high-absorptance
purposes.  and low-emitting metallic surface. This would
allow the UAV to trap as much heat from radiation as
Fuselage Interior possible while emitting relatively a little amount. The
In order to accommodate the various optimal fuselage equilibrium temperature would be
instruments required in this mission, the interior of the obtained by varying values of  and through the
fuselage will consist of three floors. These floors will following heat balance equation:
be made of honeycomb sandwich structure (Figure 10).
Among the advantages of using honeycomb are A s / c  T eq  Q conduction  Q convection
4

lightweight, high crush strength and stiffness, structural ( 32)

integrity and high fatigue resistance.  A s J s  A P  J a  A P  F12 J P  Q int ernal
The left hand side of the equation consists of
the heat that is escaping the system. The term
As/cTeq4 is the heat loss due to radiation. Qconduction
and Qconvection are heat losses due to conduction and
convection, respectively. As/c is the total surface area
of the UAV.  is the Stefan-Boltzman constant
Figure 10: Honeycomb Structure (5.67x10-8 W/m2K4). The right hand side of the
equation is the sum of the heat being absorbed by the
The honeycomb structure material will be system. The term AsJs is the heat addition from solar
made of aluminum, due to its smooth and thin cell flux of the sun. ApJa is the heat addition due to
walls, in addition to the high strength-to-weight and planetary albedo. ApF12Jp is the heat addition from
stiffness-to-weight ratio. planetary radiation. Finally Qinternal is the internal heat
Particular instruments must be laid out in generated by the various instruments. {As and Ap are
specific locations within the fuselage; for example, the the projected areas of the UAV in the directions of the
LIDAR will be placed on the base floor. For the sun and the planet respectively. Js is the solar radiation
LIDAR to serve its purpose, this part of the fuselage intensity on Mars (590 W/m2). Ja is the albedo
base will have glass material as a “window.” The shaft contribution to the total radiation input to the UAV,
and motor will be further back, towards the tail end of which is defined as the product of Js, a (average Mars

15
albedo, 0.15), and visibility factor F, which is dependent of the materials that we considered along with their
on the altitude of flight and the bearing angle.} properties:
We can model heat loss due to conduction as
a 1 -dimensional conduction problem. Equilibrium
MATERIAL a e a/e Temperature
(K)
Polished Aluminum Surface 0.35 0.004 8.75 265
Polished Stainless Steel 0.50 0.130 3.85 n/a
Polished Copper 0.28 0.130 2.20 n/a
Figure 11: Model of heat loss as conduction Grafoil 0.66 0.340 1.90 n/a
Vapor-blasted Stainless Steel 0.60 0.330 1.80 n/a
The total heat transfer via conduction can be Gold/Kapton/Aluminum 0.53 0.420 1.260 n/a
expressed in equation ( 33).
Gold-plate on Aluminum 0.30 0.040 7.50 261
Tatm  Teq
qx  Table 8: Metallic Surface Materials - Property Table
1 L1 L2 ( 33 )
[( )  ( )( )]
hA KalA KcfA The temperatures of Steel, Copper, Grafoil,
where Tatm is the average ambient temperature (230 K) and Gold/Kapton/Aluminum of Table 8 are found to
at our cruise altitude, h is the convective heat transfer be well under 265K and therefore not applicable.
coefficient, A is the surface area of entire fuselage, L1 is Thus, based on the values and given equilibrium
the thickness of the aluminum (0.1mm), L2 is the temperatures, in Table 8, our chosen external thermal
thickness of the layer of carbon fiber (0.5mm), and K is layer will be one of polished Aluminum surface.
the thermal conductivity of aluminum and carbon fiber
respectively. After plugging in values for all the Fuselage: Material Selection
variables, the equation is reduced down to 1.4(230-Teq). The fuselage itself is modeled as a pressure
Next, we model heat losses due to convection vessel. The entire structure is assumed to be sealed off
by the fuselage as fluid flowing pass a flat plate. The when it is built; therefore, it would contain Earth’s
total heat transfer via convection is defined in equation atmospheric pressure (100 kPa). However, the ambient
( 34). pressure on Mars is so low (700 Pa) that there would be
significant pressure drop between the pressure inside
Q  hATatm  Teq  ( 34 ) the fuselage respect to the atmospheric pressure.
Because air in a high pressure region tends to move to
Equation ( 34) expresses h as the convective heat that of a lower region, there would be a tremendous
transfer coefficient, A as the surface area of the force expanding outwards. The function of a pressure
fuselage, Tatm, Teq as the atmospheric and equilibrium vessel is to essentially contain pressure P. The main
fuselage temperature, respectively. In order to obtain objective is to do so while minimizing weight. A
the convective heat transfer coefficient h, we use the constraint involving this would be that the pressure
Nusselt correlation of equation ( 35) for flow over a flat vessel must leak before it breaks. This ensures that if a
plate. crack exists, the leak would release pressure gradually
and thus safely.
Nu L  0.664 Re 1 / 2 Pr 1 / 3 ( 35 ) Based on these objectives and constraints, it is
essential that we determine the appropriate material
The Reynolds number is approximated to be indices, which would help us assess the optimal
280,000. The Prandtl number is somewhere around materials. We can idealize the pressure vessel as a thin-
0.76. As expressed in equation ( 36), h is simply a walled cylinder with an average radius R and thickness
function of the Nusselt number, conduction coefficient t. The wall thickness is chosen so that at a certain
k of the ambient, and L the length of the flat plate. pressure difference, the stress is less than the yield
Nu L K strength of y of the material. The stress should also
h ( 36 ) be less than the fracture stress, at which point a crack
L
would propagate in the vessel. It is however, notable
Now that all the terms in equation ( 32) are in that there would be no worry of crack propagation if
terms of  and Teq and Teq, we can select metallic the stress is kept under the yield stress—this ensures
surface finishes with reasonably  and low values that stable deformation. Such is expressed in equations
would give us comfortable temperature for the ( 28 ) - ( 37 ).
instruments to be operating at. The following is the list

16
 2
M4 is expressed in Figure 13 —strong, light
K  materials lie near the top of the figure.
 a c  C 2  1C  ( 37 )
  y 
Maximizing the material index, M1, will
maximize tolerable crack size.
K 1C
M1  ( 38 )
y
Leaks in the pressure vessel caused by a crack
can be detected if the crack is just the size to penetrate
the inner and outer surface while maintaining vessel
stability. We must note that the wall thickness is such
that it will not yield. Thus, a maximum value of the
material index M2 indicates safe containment of
maximum pressure.
Figure 13: Specific Strength Chart of Materials
y
M4  ( 39 ) Based on these material indices, carbon fiber

is chosen for the fuselage structure.
While maximizing M1 and M2, we keep in
mind that minimizing wall thickness is an important Propeller Shaft
objective. Since small values of wall thickness indicate A cylindrical shaft in the tail end of the
high numbers of yield stress, we try to maximize yield fuselage interior will connect the propeller and the
stress for our material. Thus, another material selection motor. Conventionally, the shaft is 1/20 of the
criterion would be to maximize index M3. fuselage diameter. We will perform calculations of
torsion to determine the minimum size that is required
M3  y ( 40 ) of a shaft of carbon fiber.
The total torque created by the propeller is
Because our vessel is in fact an aircraft, weight approximately 5330 Nm; this is calculated by the force
is crucial. For minimizing weight, we must further find generated by the propeller, taking all three blades into
a material in which the material index M4 is maximized. account. Using the Tresca yield criterion, where
K 1C
max=y/2, we use equation ( 42) for maximum shear
M1  ( 41 ) stress to find an optimum shaft radius.
y
T
Once the four material indices have been  max 
found, a material selection chart is referred to. Figure
 c3 ( 42 )
2
12 is one of fracture toughness (K1c) versus elastic
limit (y). The diagonal line corresponds to M2, With a safety factor of 2, we result in a shaft
expressing the constraint that the vessel must leak diameter of approximately 0.0712m. Performing
before it breaks. weight calculations based on density, we find that using
the given diameter, the shaft will weigh 6kg—this
exceeds our initial weight constraint. We further
continue our shaft analysis by making the shaft hollow.

C2
C1

Figure 14: Shaft Diagram

Figure 12: Fracture Toughness vs. Elastic Limit

17
The equation for maximum shear stress is now finish, the total mass is considered negligible. Weight is
illustrated in equations ( 43)-( 44). therefore calculated as the product of the surface area,
the thickness, and the density of the material. A small
Tc 2 area of glass fiber, to accommodate the LIDAR
 max  ( 43 )
instrument at the base of the fuselage, is noted.
J Including the shaft and propeller as part of the
structure, the UAV will have a structure of 2.97kg.
J 

2
c 2
4
 c1
4
 ( 44 ) Note Table 9 for specific weights.

Performing calculations, we achieve a shaft Table 9: Specific Weights of Chosen Materials

radius of approximately 0.03m. For a safety factor of 2, Structures Conclusion
this will increase to approximately 0.0815m. The mass For our given structure, bending moment and
is still too large, at around 1.73 kg. It is therefore buckling analysis is done on the wing, in addition to
necessary to decrease the length of the shaft to around pressure vessel analysis on the fuselage. We can
0.25m, in which case the mass will decrease therefore verify that our structure is both stable and
considerably, to around 0.43kg. Thus, the motor must sound—the wing will not yield nor buckle and the
be designed to fit into a tube with a diameter of fuselage will not break given our dimensions. Our total
approximately 0.09m. structure weight achieved is approximately 3 kg.
Surface Total
Material/
Area
Thickness Density
Mass Electrical and Systems Design
Section (m) (kg/m3)
(m2) (kg)
Carbon Fiber Introduction:
Fuselage 0.827 0.0005 1500 0.6203 The electrical and systems sub-team
Wing 1.200 variable 1500 0.6000 concentrates on mapping the topography of Mars given
weight parameters set by the structural sub-team. An
Tail 0.125 0.0005 1500 0.0938
electrical system needs to power and control the plane,
Shaft n/a n/a 1500 0.4300 as well as gather topographical data and flight
Propeller n/a n/a 1500 0.6000 information. Listing flight-critical sensors and data-
gathering equipment is the first step in the design
Aluminum process. We compile a chart of mass and power
Fuselage 0.827 0.0001 2700 0.2233 consumption based on initial trade studies, then
Wing 1.200 variable 2700 0.3600
consider instrument networks to provide data storage
and distribution capabilities. Once a general idea of
Tail 0.125 0.0001 2700 0.0338
equipment and power supply forms, we reduce mass in
Polished Aluminum light of flight time optimization by using lighter and
Fuselage 0.827 0.0001 n/a Negligible more energy efficient sensors along with more
Wing 1.200 0.0001 n/a Negligible advanced energy sources.
The overall mass for the instruments is 7.80
Tail 0.125 0.0001 n/a Negligible
kg with an energy consumption of 459 watt-hours.
Glass Fiber (18%)
Assumptions:
Fuselage 0.009 0.0007 2500 0.0151 The following assumptions are made in the
design of the UAV electrical systems:
Honeycomb floor 0.1000
TOTAL 2.9761 1. The operating temperature inside the UAV is
greater than 0C.
Overall Structure Weight 2. There are no directional magnetic fields
Keeping the structure stable and lightweight is present to adversely affect navigation.
crucial for the UAV. Overall structural weight is based 3. There will be a network of Mars-orbiting
on the surface area and material thicknesses. The UAV satellites as outlined by NASA. This will
can be divided into three sections, being the tail, wing, enable direct line-of-sight communication
and fuselage. Each of these three sections will have with the UAV at all times.
three layers of material—the external thermal layer and 4. All instruments are customizable for the UAV
inner aluminum core layer each being 0.1mm thick, and and Martian environment. Electronics will be
the structural carbon fiber layer being 0.5mm thick. radiation-hardened to ensure adequate
Because the polished aluminum is merely a surface

18
 performance during transit and on Mars.
5. All wiring will be shielded to prevent Manufacturer & Product Function Power
electromagnetic interference and to reduce Name
transmission losses. Aeroflex RadHard MCU 0.48 W
6. The mean-time-to-failure (MTF) of all UT80CRH196KDS
instruments is much greater than the expected Aeroflex ACT5108 RadHard Motor 0.2 W
duration of the mission, which includes transit Motor Driver
time and time spent on Mars. Cognex MVS-8100D Digital Video 3W
Frame Grabber
Constraints: DSP Arch. DSP24 24-bit HP Comm 2.48 W
The following constraints are imposed on the Digital Signal Processor
design of the UAV based on current technologies and Aeroflex UT28F256 LV Memory 1.5 W
design methodologies: PROM
SEAKR NV-CPCI Non- Memory 3W
1. The total mass of all instruments must be less Volatile, Solid-State FLASH
than 8 kg to accommodate the aerodynamic
and structural constraints listed earlier. Table 11: Micro-controller Instrumentation
2. The data rate used for communications is
limited to 480 MB/hr based on a two-antenna Microcontroller Unit and Memory:
design and UHF frequencies. Table 11 lists the components that comprise
3. The oxygen-deficient atmosphere prevents the main micro-controller unit. All circuit components
the efficient use of fuel-cells as a potential are radiation-hardened to at least 300 Krads during the
power source fabrication, design, and layout processes to ensure they
perform as expected after a three to six month transit
Instrument Selection: period and during the two hour mission on Mars.
Table 10 lists the UAV instruments and their Electronic circuits are exposed to approximately 1.75
respective mass, power consumption, operating Krads[Si]/year in space when shielded with 50 mils of
temperatures, and physical dimensions. aluminum, with a slightly higher number (10 krads[Si])
The following sections discuss the trade-offs during solar storms. Therefore, the radiation-hardened
and specifications for each instrument. Important components are capable of withstanding several years
sections such as power management, communication, of exposure without side-effects.
and navigation are considered in more detail followed However, radiation-hardening requires larger
by a high-level block diagram of the system. layout footprints and more conservative transistor
designs, resulting in slower switching speeds and larger
Name Mass (kg) Power Op. Temp Dimensions
(W) (C)
Analog Devices ADT7317 (x3) 0.003 0.045 -40 to 125 7 mm x 7 mm x 3 mm
Endevco 32394 Si MEMs Pressure Sensors (x2) 0.004 0.125 -54 to 120 1.65 m m x 1.2 m m x 0.4 m m
Motorola MPXV5004G6U Low-Pressure Sensor 0.002 0.053 0 to 85 18.9 m m x 17.5 m m x 7.6 m m
Safety, navigation, control transducers (m echanical 0.5 0.1 > -5 variable
flaps, rudders, elevators)
Analog Devices ADXL150 Accelerom eter (x3) 0.015 0.054 -40 to 85 7 mm x 7 mm x 3 mm
Analog Devices ADXL105 Accelerom eter (x3) 0.006 0.041 -40 to 85 7 mm x 7 mm x 3 mm
Analog Devices ADXRS150/300 Gyroscope 0.062 0.126 -40 to 85 7 mm x 7 mm x 3 mm
Cognex CDC-100 CMOS CCD 0.11 2.5 0 to 45 34 m m x 31 m m x 47 m m
General Atom ics LIDAR Receiver 0.12 1 > -10 53 m m x 64 m m x 33 m m
Triple-Junction Photovoltaic Com pass 0.015 N/A > - 10 90 m m (diam ) x 240 m m
Li-Ion SPE Batteries 4.2 -1050/hr > -20 140 m m x 140 m m x 140 m m
Microcontroller Instrum entation 0.21 10.2 0 to 50 136 m m x 93 m m
DC Electric Engine 1.8 430 N/A -
Patch Antenna (x3) 0.75 15 > -10 50 m m x 50 m m
Total 7.80 459/hr 0 C (min)

Table 10: Instrument Selection

19
 areas compared to commercial designs. The trade-offs Software programs, such as navigation algorithms and
are improved durability in harsh-environments and communications protocol will be stored on the internal
lower power consumption due to lower transistor SRAM to allow for faster data-transfer rates to the
densities - both of which are critical factors in the UAV MCU registers.
design.
The UT80 MCU is a 16-bit microcontroller LIDAR and Camera:
designed to run on 20 MHz clock and industry- The LIDAR unit, provided by General
standard MCS-96 RTR architecture, which ensures Atomics, includes a frequency-doubled Nd:YAG pulse
compatibility of out-sourced software design. The 1 laser, dual microchannel plate CCD detector, scanning
KB of internal SRAM is insufficient for all sensor and mirror, and light-weight collection lens required for
image data; therefore, the UT28F256 PROM and numerous high-resolution images. The unit is located
SEAKR FLASH external memory are added to in the nose of the UAV. In order to provide maximum
augment data-storage capabilities. The UT28F256 coverage of the landscape, the scanning mirror,
external LV PROM adds an additional 256 KB of non- powered by a low-power DC motor, provides a fast
volatile memory, which is used to primarily store the horizontal scanning motion at several thousand points
temperature, pressure, acceleration, and gyroscope per scan. Based on the method developed by Lathrop
sensor data using a sample rate of once per second. et al., the CCD detector gathers all of the reflected laser
The SEAKR NV-CPCI FLASH chip allows for 1 GB intensity per scan. A horizontal resolution of much
of non-volatile memory with a maximum data rate of better than two meters is possible with this unit, which
27 Mbps and 3 W of power dissipation. The FLASH is a significant improvement over the LIDAR
chip will be primarily used to store high-resolution instrumentation onboard current Mars-orbiting
images and LIDAR data. Table 12 lists the properties satellites. The receiver specified by General Atomics
and memory bits required for each sensor and camera provides a resolution of 320x320 pixels at 4 bits/pixel
type. Extra bits are required for the Endevo pressure and a total of 50 kB/scan.
sensors to provide the required 8.62 mV/psi sensitivity. The Cognex CDC-100 HiRes CCD provides
Similarly, due to the aerodynamic precision required 1280x1024 resolution at 8 bits/pixel of color with no
once in flight, at least 10-bits of resolution is needed compression, which requires a total of 1.25 MB/photo.
Sensor Type Quantity Total Bits The CCD interfaces with the Cognex MVS-8100D
listed in Table 11.
ADXL150/105 3/3 60 bits @ 10-bits each
ADXRS150/300 3/3 60 bits @ 10-bits each
Power Management:
Endevo 32394 2 24-bits @ 12-bits each Initial energy source selections for the UAV
Motorola 1 8-bits @ 8-bits each consist of (1) solar cells, (2) alternative electromagnetic
MPXV5004G6U sources, (3) fuel cells, and/or (4) batteries. The primary
ADT7317 2 20 bits @ 10-bits each requirement for these energy sources is to provide
adequate energy to on-board instruments and motor,
ThermalTab RTD 1 10-bits @ 10-bits each specified by a minimum of 460 W for one hour.
Total 18 182 bits (22 bytes) Moreover, they must function within Martian
atmosphere; that is, they must function despite various
gas compositions, temperatures, distances from the
Table 12: Memory allocation and data storage sun, and other factors. They must also be optimized in
for distributed sensor package light of the ratio kilowatt-hour (energy) per kilogram.
Trade-off analysis for the various selections
for the gyroscope and accelerometer data in order to shows that, first, solar cells do not provide enough
resolve the analog output. required energy for flight alone or regeneration. This
Given the memory requirements listed in fact is expressed by a specific area of 263 W for 1 m2 of
table 12 below, the Aeroflex UT28F256 LV PROM can a solar panel with a Martian efficiency of 28%, yielding
store over twelve thousand data sets given a sample only 73 W/m2. Note that given a wingspan of around
rate of once per second and 22 bytes per set. This 2 m x 0.5 m, only 73 W can be achieved from a full
allows for over two hours of storage time, which is solar cell array. Additionally, since the energy per mass
adequate given the current mission specifications. The value is only 32.2 W/kg after Martian considerations,
SEAKR FLASH chip is also able to store hundreds of the extra mass might as well be spent on batteries or
images at any one time. In order to ensure a timely and fuel cells without regeneration.
accurate transfer of data to orbiting satellites, a first-in- The second alternative suggests implementing
first-out (FIFO) queue structure will be implemented in a land microwave or laser electromagnetic targeting
memory where old data will be transmitted first. source to beam energy towards the UAV while it is in

20
flight. The reason for the rejection of this alternative is liquefaction. Moreover, pressurization requires an
simple – there is no guarantee that any form of a land entire system of compressors with the addition of extra
station will be available for this purpose, while satellites volume. The same volume concern holds if a
are too far away to be able to transmit and pinpoint at refrigeration system is used. As the volume tradeoff
these wavelengths accurately. The 0.3 cm-30 cm shows, fuel cells without the best pressurization are not
wavelengths for microwaves and hundreds of optimal for an aircraft of this size, even if the mass
nanometer wavelengths for lasers implies that the required for the energy supply is small.
energies cannot be precisely targeted from the orbit. Until the Mars In-situ Propellant Production
We attempt using fuel cells in the third can provide immediate oxygen production, or until the
alternative. The fuel cell is an aspiring future volume can be drastically reduced by using absorption
technology that generates a very high energy per mass material or other Martian gases like nitrogen, fuel cells
value. However, all existing fuel cell technologies require the for the UAV will have to hold.
use of hydrogen and oxygen for the generation of
electric current. On the other hand, Mars is dominated Batteries:
by CO2, nitrogen, and argon. One solution to the Batteries are by far the more convenient and
paucity of hydrogen and oxygen is to bring along readily available technology compared to fuel cells, and
pressurized/refrigerated gas tanks; another solution is require no peripheral equipment at a much more
to produce oxygen via compressing CO2 in the manner compact volume. A variety of battery types are
of the Mars In-situ Propellant Production (MIP). commercially available and under development, as
As for now, the weight and size of a custom listed in Table 13 below:
MIP device is not readily available, nor has it made for
rapid collection of CO2 (only at night and low Name kWh/kg
temperatures of around 200 K) although an 8.5 kg, Lead Acid 0.035
40cm x 24cm x 25cm device has already been
NiMH 0.07
demonstrated.
A fuel cell system will require the following Li+ 0.15
components to carry the necessary hydrogen and NaS 0.11
oxygen onboard the UAV: Li+ SPE 0.25
Table 13: Comparison of battery energy/mass
 Tanks (hydrogen, oxygen) with insulation ratios
 Fuel cell pressurization maintenance (or even
refrigeration) system
 Fuel cell array and delivery Mass and Containment Considerations:
Lithium ion Solid Polymer Electrolyte (SPE)
 Control system
is the best contender among these existing
technologies. It provides the highest energy density,
Sources show that 1g of hydrogen fills 11
around 1.5 to 2 times more than the currently existing
liters at 0°C and 1 atm. In order to reduce the volume
Lithium-ion battery technology. The sample that we
of the tank to reasonable levels, we can try to
have chosen is under development by Ultralife since
pressurize the volume to around 2.2 liters at 5 atm on a
2001, and is a feasible power source due to its non-
Martian surface pressure of 0.01 atm, which is a very
atmospheric requirements.
generous pressurization value. If the temperature in
A simple calculation at an allowance of 4.2 kg
the fuel tanks can be maintained at around -100°C, via
gives 1050 Whr, while the volume of the battery is
simple insulation and without extra refrigeration cycles,
around 400 Wh/liter or 400 kWh/m3; 1050 Whr yields
the volume can fall to around 1.4 liters for 1 gram, or
0.0026 m3 of theoretical battery space, or around 14cm
0.7143 g/liter or 0.7143 kg/m3.
x 14cm x 14cm of volume. Compared to fuel cells, the
Given that we have at most 0.008 m3, or a
given mass-volume tradeoff is extremely reasonable for
20cm x 20cm x 20cm, volume available for a H2 tank,
our considerations.
5.7 mg of H2 is required. In this case, the tank will fill
up 10% of the wingspan and nearly 82% of the
Navigation:
maximum fuselage area. Now consider that 1 kg of H2
The absence of magnetic poles makes
gives 86 MJ of energy, which means that multiplying
navigation on Mars particularly difficult without a GPS-
5.7 mg by 86 MJ/kg we will have 491 kJ, or around 137
like system. Since the flight of the UAV will be
Whr, which is still less than the best batteries (>200
decided upon in advance and, in general, will be
Whr) at the cost of a much larger volume.
relatively direct across the surface of Mars, the UAV
This result only takes into consideration the
can take advantage of its on-board sensors such as
hydrogen tank. The oxygen tank adds to another part
accelerometers and gyroscopes to detect any deviations
of the fuselage, and is usually larger without

21
from its path. Such sensors should be sufficient to when the sun is directly overhead. This problem can
ensure that the UAV stays on course once it has begun be avoided ahead of time by taking this problem into
its flight path. However, for the UAV to begin on the consideration when planning the flight location and
correct flight path our design requires additional time. Another complication is that our measurements
sensors. are accurate only when the UAV is flying parallel to the
The photovoltaic compass (PV or sun Martian surface. This means that measurements made
compass), based on InGaP/GaAs/Ge technologies, during ascent/descent or turbulent flight should be
can be used in the initial state of the plane’s launch to discarded. Fortunately we will ensure that the PV
detect the proper orientation of the plane relative to the compass is used only when the plane is in the proper
sun. Precise knowledge of the launch area on Mars will orientation by checking the accelerometers and
make it possible to know what the proper angle to the gyroscopes.
sun should be and this can be checked by the PV
compass throughout the flight. The design of the PV Communication:
compass is based on 26 small rectangular Triple- In order to transmit the information gathered
Junction solar cells that are arranged in an octagonal- by the array of sensors aboard the UAV, we have
cylinder fashion where the inward facing cells form the included three UHF patch transceiver antennas in our
walls and bottom of the cylinder. The top surface is design. Only two of these antennas will be operating at
covered by anti-reflective fused Silica industrial-grade any one time, with the third antenna serving as a back-
glass that has a low refraction coefficient relative to the up. We estimate the power requirement of two
Martian atmosphere, and low coefficient of thermal antennas to be 15 W with a total mass of 0.75 kg for
expansion. The glass part of the PV compass will be the three antennas. One antenna will be one on each
exposed to the Martian atmosphere at the top of the wing with the third on top of the fuselage. The patch
UAV’s fuselage without altering the laminar flow over antennas have a flat profile which can further be
the UAV. Incident sunlight will enter the PV compass reduced by placing them into indentations in the
and strike certain solar cells. Based on the current surface structure.
produced by each cell we can determine the angle of These particular antennas are compatible with
the incident sunlight and thus the angle to the sun. existing satellite/rover communications equipment.
Rather than attempt to transmit directly to Earth, we
assumed that the UAV would be operating in an
environment where there would be multiple
90 mm opportunities to transmit data to Mars-orbit satellites.
By avoiding direct transmission to Earth, we save
power and thus weight aboard the UAV. Also, we
reduce the probability of corrupting the transmitted
signal. If the UAV could continuously transmit to a
satellite, our communications uplink capability would
be 8 MB/min per antenna. Implementing this design
with only the current satellites orbiting Mars, the
Odyssey and the Global Surveyor, would permit one 8
minute window during the flight to transmit all of our
215mm
data, which would have to be limited to 64 MB.
A future satellite communications
infrastructure around Mars would increase the value of
the UAV’s mission by allowing for the transmission of
higher resolution images. Furthermore, this network of
communications satellites could serve as the backbone
of a navigational system for this and other missions to
37 mm Mars.
37 mm
37 mm
Software:
Figure 15: Photovoltaic Compass Software will be written in a combination of C
and Assembly based on industry-standard MCS-96
RTR instruction set architecture. External events will
With this design, we can detect 26° of be monitored using multi-level state diagrams clocked
incidence across each solar cell and can determine the at 20 MHz and eighteen separate interrupt service
angle to the sun with high precision based off of routines (ISRs) such as the timer COMPARE interrupt
partially lit cells. One shortcoming of the PV compass, and UART RXC interrupt. A real-time microcontroller
however, is the ambiguity of direction that remains

22
operating system can be used to provide near- except the ACT5108, which is instead located on a
instantaneous responses to external events. All separate board near the DC engine. The sensor
software packages will conform to current IEEE Code package is distributed throughout the UAV’s fuselage
of Ethics and ACM standards. to provide maximum coverage. The batteries are
regulated with a separate controller chip and a series of
Block Diagram: DC-DC converters to maintain the voltage and provide
Figure 16 shows the high-level block diagram adequate Vcc for the various components.
of the electrical systems. The central microcontroller
unit contains all the components listed in Table 11

Hi-Res
LIDAR
Camera

Software ADC

Video
Antenna Txmitter DAC DSP MCU Card
MEM DSP DAC Txmitter
Antenna

Rxver ADC ADC Rxver

Mechanical DAC DAC

Mechanical
Transducers Transducers
Left Wing ADC Right Wing

Sensor ADC
Solar
Compass
Package
DAC

Tail
DC Motor Controller
Controller

DC Electric Engine Batteries

Mechanical
Propeller Transducers
Copyright 2004, CU Odysseus Team

Figure 16: Functional Block Diagram of Electrical System

23
Conclusion that interprets data from accelerometers,
gyroscopes, and a photovoltaic compass.
Throughout the design process, the Microcontrollers and the associated software will
Odysseus Team has striven to maximize the process sensor data to check for deviations from
versatility of its UAV proposal. The team flight path and will implement the necessary
optimizes factors such as mass, power, speed, and corrective flight adjustments. For communication
flight duration to devise an aircraft that will meet between the UAV and Mars-orbit satellites we
the demands of a Martian scouting mission. The have, for redundantly, chosen three UHF patch
product of such systematic and collaborative transceiver antennas that can continuously
design process is a 10.8 kg aircraft that can fly transmit data produced by the UAV
unaided by a human controller for 2.3 hours at 130 instrumentation. Finally, a power supply
m/s and a cruising altitude of 500 meters with a consisting of multiple lithium-ion SPE batteries
maximum flight range exceeding 1000 km. will be responsible for powering the
Equipped with topographical and imaging instrumentation and propeller motor. Lightweight
instrumentation, such a UAV will be able to batteries are chosen upon detailed consideration of
produce the detailed information necessary for watt-hours per kg of all potential power sources
future manned missions to Mars. including solar cells.
The UAV blueprint outlined in this paper Due to the constraints of time and
includes a 2.1-meter long fuselage, a 0.25-meter expertise, our UAV design has excluded discussion
maximum diameter for the fuselage, and a 2.078- of the induced roll due to a propeller.
meter wingspan. The design of such components Furthermore, we have not specified an exact
takes into consideration low atmospheric density design for the contours of propeller itself. These
which causes reduced lift and drag. The elements of aircraft design are extremely complex
dimension of wingspan is the result of detailed lift and exceed the expertise of the team. Additionally,
and airfoil analysis. To propel the UAV in an power supply imposes limitations on flight
atmosphere that lacks the necessary amount of duration. In the future it is possible that potential
oxygen for combustion we have chosen a single power supplies, which we have also explored, will
2.27-meter diameter propeller mounted aft of an improve to a point where they are feasible for use
inverted V-tail. The decision to use an inverted V- on an unmanned Martian aircraft, extending its
tail is the result of a compromise between the flight duration. Despite our omissions, we our
higher control stability but higher drag of a confident that the team’s UAV design is an
conventional tail and the lower drag but lower optimal solution with existing technologies and
control stability of a V-tail. To control the UAV will provide a reliable platform for Martian
flight path we have designed a feedback system exploration.

24
 References

Aeroflex UTMC. (2002, December). UT28F256LV Radiation-Hardened 32Kx8 PROM.

Aeroflex UTMC. (2003, May). UT80CRH196KDS Microcontroller Datasheet.

Aeroflex UTMC. (2003, May). ACT5108 & ACT5109 3-Phase Brushless DC Motor Drivers.

Aircraft Design & Aeroflightdyanmics Group (ADAG). Aircraft Designed at DPA. Retrieved February 20, 2004
from http://www.dpa.unina.it/adag/eng/aircraft-design.htm

Anderson, J.D. (2000). Introduction to Flight (4th ed.). Boston, MA: McGraw-Hill Companies, Inc.

Archer, R.D. (1996). Introdution to Aerospace Propulsion. Upper Saddle River, NJ: Prentice-Hall, Inc.

Artemis Society International. (2001). Artemis Project: Hydrogen-Oxygen Fuel Cells for Lunar Habitat Energy Storage.
Retrieved from http://www.asi.org/adb/04/03/03/h2power.html

Ashby, M. & Cebon, D. (1996). Case Studies in Materials Selection. Retrieved October 2003 from
http://www.analog.com.

Beer F. & Johnston, R. (2001). Mechanics of Materials (3rd ed). New York, NY: McGraw-Hill.

Borst, H.V. (1973). Summary of Propeller Design Procedures and Data (Vol. 1). Springfield, VA: National Technical
Information Service.

Cesarone, R., Hastrup, R., Bell, D., Lyons, D., & Nelson, K.. Architectural Design for a Mars Communication &
Navigation Orbital Infrastructure, AAS 99-300.

Cognex. (2001). MVS-8100 Datasheet.

Cognex. (2002). CDC Series Digital Machine Vision Cameras.. http://www.cognex.com.

Darooka, D.K. & Jensen D.W. Advanced Space Structure Concepts and Their Development, AIAA-2001-1257
http://www.ilcdover.com/WebDocs/AdvStct.pdf

DSP Architectures. (2004, January). RHDSP24 Radiation Hardened Scalable DSP Chip.

EADS Space Transportation. (2003). EADS Space Propulsion. Retrieved October 27, 2003 from
http://cs.space.eads.net/sp/

Edie, D. & Buckley, J. (1993). Carbon-Carbon Materials and Composites. USA: William Andrew Publishing.

EG&G Services Parsons, Inc. (2000, October). Fuel Cell Handbook (5th ed). Retrieved from
http://www.fuelcells.org/fchandbook.pdf.

Endevco, Inc.. (2003). Silicon MEMS Pressure Sensor, Model 32394. http://www.endevco.com/

Ezell, E.C., & Ezell, L.N. (1984). On Mars: Exploration of the Red Planet, 1958-1978. Washington, DC: NASA

Hall, D., Tsai, K.C., Galbraith, D., & Parks, R.W. (1997). Conceptual Design of the Full-Scale Vehicle
Propulsion System Concept. Airplane for Mars Exploration. NASA/Ames Research Center, CA: David Hall
Consulting.

Hepperle, M. (2004). Aerodynamics for Model Aircraft. Retrieved February 18, 2004 from http://www.mh-
aerotools.de/airfoils/index.htm

1
Hexweb. (2004). Hexweb: Honeycomb Attributes and Properties. Retrieved February 20, 2004 from
http://www.hexcelcomposites.com/Markets/Products/Honeycomb/Hexweb_attrib/hw_p11.htm

James, D. (2003). NASA Quest: Atmospheric Flight. Retrieved November 2003 from
http://quest.arc.nasa.gov/aero/planetary/atmospheric/structure.html

Kaplan, D.I. et al. (2000). In-Situ Propellant Production on Mars: The First Flight Demonstration. Retrieved from
http://powerweb.grc.nasa.gov/pvsee/publications/mars/MIP_LPSC.html

Kroo, I. & Shevell, R.S. (2003). Aircraft Design: Synthesis and Analysis. Stanford, CA: Desktop Aeronautics, Inc.
Retrieved February 21, 2004 from http://adg.stanford.edu/aa241/AircraftDesign.html

Lathrop, J., Snider, R., Becraft, W., Doll, D., & Cianciotto, F. UAV and LIDAR Mine Detection in the Surf Zone:
Operational Considerations. General Atomics Aeronautical Systems, Inc.

Mason, L. (2000). Prototype Solar Panel Development and Testing for a Mercury Orbiter Spacecraft. AIAA-2000-2881.
Cleveland, OH: NASA Glenn Research Center.

Morton, O. (1999, December). MarsAir: How to build the first extraterrestrial airplane. Air & Space Magazine,
Vol. 14. Retrieved October 28, 2003 from http://quest.nasa.gov/aero/planetary/MarsAir.html

Motorola, Inc.. (2004). http://www.motorola.com.

Mueller, T.J. (1985). Low Reynolds Number Vehicles. AGARDograph, No. 288.

Murray, J.E. & Tartabini, P.V. (2001). Development of a Mars Airplane Entry, Descent, and Flight Trajectory. Retrieved
October 2003 from AIAA.

Nice, K. (2003). How Fuel Cells Work. How Stuff Works. Retrieved October 20, 2003 from
http://science.howstuffworks.com/fuel-cell.htm

Nicolai, L.M. (1975). Fundamentals of Aircraft Design. Fairborn, OH: E.P. Domicone Printing Services.

Ralph, E.L. & Linder, E.B. (1996, May 13). Advanced Solar Panel Designs. IEEE Xplore Journal. 25th PVSC.
Washington D.C.

Raymer, D.P. (1999). Aircraft Design: A Conceptual Approach (3rd ed.). Reston, VA: AIAA Education Series.

SEAKR Engineering, Inc. (2002, October). NV-CPCI Non-Volatile Compact PCI Memory Card.

Shevell, R.S. (1989). Fundamentals of Flight (2nd ed.). Upper Saddle River, NJ: Prentice-Hall, Inc.

Stackhouse, D. (1999, January). Conventional vs V-Tails. Charles River Radio Controllers. Retrieved March 5, 2004
from http://www.charlesriverrc.org/articles/design/donstackhouse_conventionalvsvtail.htm

Stanford Audio Phonic Photographic Infra Red Experiment. (1995, February 22). Space Environmental Study.
Retrieved from http://students.cec.wustl.edu/~sapphire/design/log/spacenv.html

Stevens, B.L., & Lewis F.L. (2003). Aircraft Control and Simulation (2nd ed.). Hoboken, NJ: John Wiley & Sons,
Inc.

Teofilo, V. L., Brown, R., & Cuellar, E.A. (2001). Advanced Lithium Ion Solid Polymer Electrolyte Battery
Development.

United States Department of Energy. (2004, January 15). Energy Savers: Electric Vehicle Batteries. Retrieved from
http://www.eere.energy.gov/consumerinfo/factsheets/fa1.html.