Citation
Development of a Parametric Software Tool for the Design and Manufacturing of Micro Air Vehicles

Material Information

Title:
Development of a Parametric Software Tool for the Design and Manufacturing of Micro Air Vehicles
Copyright Date:
2008

Subjects

Subjects / Keywords:
Aerodynamics ( jstor )
Aircraft ( jstor )
Aircraft wings ( jstor )
Airfoil camber ( jstor )
Airfoils ( jstor )
Carbon fibers ( jstor )
Computer aided design ( jstor )
Fuselages ( jstor )
Geometry ( jstor )
Wind tunnels ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
7/12/2007

Downloads

This item has the following downloads:


Full Text





DEVELOPMENT OF A PARAMETRIC SOFTWARE TOOL FOR THE DESIGN AND
MANUFACTURING OF MICRO AIR VEHICLES





















By

DANIEL J. CLAXTON


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2007



































2007 Daniel J. Claxton
































To my beautiful wife, Kristin.









ACKNOWLEDGMENTS

I thank everyone who helped and encouraged me on this project. First and foremost, I

thank my advisor, Dr. Ifju, for supporting my efforts and giving me the opportunity to prove

myself. I give special thanks to Mike Braddock and all of they guys in the machine shop, who

spent hours and hours helping me machine molds.

Thanks to all of the guys in the MAV Lab. They all made my job more interesting and

fun. Special thanks go to Bret Stanford and Frank Boria for providing me with advice,

inspiration, creative ideas and much-needed laughter. I thank Kristin for being patient with me

for all these years and encouraging me to push on, but do it quickly. Last, but not least, I thank

my parents who supported me financially and otherwise through all of my education. They

never stopped believing in me.









TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ..............................................................................................................4

L IS T O F T A B L E S ................................................................................. 7

LIST OF FIGU RE S ................................................................. 8

ABSTRACT ................... ............... ......... ... ...... ... .............

CHAPTER

1 INTRODUCTION ............... .......................................................... 13

M icro A ir V vehicles at a G lance .................................................. ................................ 13
M ircro Air V vehicles at the Research Level ........................................ ........................ 13
M otiv action and O v erview ......................................................................... ........................ 15

2 L IT E R A TU R E R E V IE W ........................................................................ .. ....................... 19

M icro Air Vehicle Design .................................... .. .......... .. ............19
C om putation F luid D ynam ics....................................................................... .................... 19
Manufacturing and Rapid Prototyping ............................................................................20

3 DESIGN PROCESS .................................. .. ... ..... .................. .. 22

M A V L A B .......................... ........................................................................... .................. 22
Scalar-B ased D esign Param eters......................................................... ............... 23
V ector-Based D esign Param eters ........................................................ ............... 25
A erodynam ic A nalysis........................................................ .. .. .. ....... .. ............28
Com puter Num erically Controlled M achining .............................. .................................... 28
C om posite C construction ................. ................................ ........... ...... .. .. ...30
Flexible W ing Fabrication ......................................... .................... ............... 31
R igid W ing Fabrication ........... .. ........................... ............. ..... 32
F u selage F abrication ........ .................................................................... ........ ........ 32
V acuum B aging and C during ........................................................................ ...................33
F u selag e A ssem b ly ......................................................... ................. 33
C om ponents In stallation ......... ..... ............ ................. .............................. ........................ 34

4 REPRESENTATIVE STUDY.............................................................. ...............51

C om puter M odel D description ......... .................................. ........................................51
W ing A erodynam ics Com parison................................................. .............................. 51
W ind Tunnel R results .................................................. .. ... ............ 52
A then V ortex L attice R esults.............................................................. .....................53
W ing M old Accuracy ................................... .. ........... .. ............54









C o n clu sio n ................... ...................5...................5..........

5 FUTURE WORK.......................... .................. ........... 62

O p tim iz atio n ..................................................................................................................... 6 2
A ero elasticity ............................................................................... 62
Full Aircraft M modeling .................. ................... .................. .......... .. ............ 63

A PPEN D IX U SER S' M A N U A L ........................................................................ ..................65

Basic D description of Features........................................................... .. ............... 65
G getting Started w ith M A V L A B .............................................................................. ............66
E x am ple U sag e ................... ...................6...................7..........
O pen and Saving W ing Files ............................................................................ ...... 67
M odify Som e P aram eters ........................................................................ .................. 67
Export W ing G eom etry ....................... .................................................................. 67
Run Athena Vortex Lattice Aerodynamic Analysis............................................68
Create Computer Numerically Controlled Toolpath.......................................................68
List of Functions ...................................................................... ........ 69
M ain Functions ...................................................................................................... 69
Sub-Functions ........................................................................................................ 70

L IST O F R E F E R E N C E S ...................................................................................... ...................76

B IO G R A PH IC A L SK E T C H .............................................................................. .....................79




























6










LIST OF TABLES


Table


3-1 List of basic MAVLAB wing parameters............................. ............... 35


page









LIST OF FIGURES


Figure pe

1-1 Typical composite airframe MAV developed at the University of Florida............................17

1-2 Dependency of aircraft aerodynamic performance on scale, where performance is
represented by maximum lift/drag and scale is represented by the non-dimensional
quantity R e. ................................................................................ 17

1-3 Example of a thin, under-cambered, flexible membrane Micro Air Vehicle wing ..............18

1-4 Relative scale and payload capacity of Micro Air Vehicles and Small Unmanned Air
V eh icle s ................... ......................................................... ................ 18

2-1 Black Widow Micro Air Vehicle developed by Aerovironment using multi-disciplinary
optim ization .............................................................................. 2 1

3-1 M AVLAB Graphical U ser Interface. ............................................ ............................. 36

3-2 Typical graphical display of a wing geometry in MAVLAB, with feedback for basic
parameters such as planform area, mean chord, aerodynamic center and aspect ratio......36

3-3 Wing span as defined by the linear distance from wing tip to wing tip. ..............................37

3-4 Wing root chord is defined as the distance from leading to trailing edge of the wing,
closest to the fuselage, or middle of the span. ................................. ........................... 37

3-5 The maximum camber is the ratio of the highest point of the camber line to the airfoil
chord length ..............................................................................38

3-6 The geometric dihedral angle is the angle the wing makes with the horizontal.................38

3-7 The sweep angle is the angle between the 14 chord line (dotted) and the line
perpendicular to the free stream velocity....................................... .......... ............... 38

3-8 Geometric wing twist, often called washout when it is negative and washin when it is
positive......................................................... 39

3-9 Typical example of CNC milling of a MAV wing mold out of resin-based tooling
b board .......................................................... ................................... 4 0

3-10 Scalloping or cusps created by consecutive passes of a ball (radial) end mill ...................40

3-11 Typical wing mold machined from high-density polyurethane tooling board ...................41

3-12 Typical ball end mill used for machining 3D freeform surfaces such as wing molds. ........41

3-13 A) Rigid carbon fiber wing. B) flexible membrane wing with batten reinforcement. ........41









3-14 Drawing layout pattern on the wing mold. ........................................ ........................ 42

3-15 Applying Teflon release film to the surface of the wing mold using spray glue .................42

3-16 Laying up of carbon fiber on the wing mold. ............................... .......................... 43

3-17 Application of K evlar to the leading edge ......... ........................................ ...................43

3-18 Cutting of unidirectional carbon fiber strips with custom cutting tool. ............................44

3-19 Trimming of excess carbon fiber during layup. ...................................... ............... 44

3-20 Trimming excess carbon fiber from wing after cure. ................................ .................45

3-21 Application of Teflon release film to fuselage mold. .................... .......................... 45

3-22 Fuselage placed on a form to maintain proper mate with wing.........................................46

3-23 Application of 1-2 layers of carbon fiber to the fuselage mold ................ ...............46

3-24 Vacuum bagging of wing tool. ..... ........................... .......................................... 47

3-25 Vacuum bagging of fuselage covered with peal-ply release film. .................................47

3-26 Assembly of wing and fuselage after cure....................................... ......................... 48

3-27 Masking prep and application of spray glue to top of wing for latex application................48

3-28 Application of latex membrane to top of wing............ ........................... .................49

3-29 After latex trimming, final gluing of latex to carbon fiber skeleton of the wing. ................49

3-30 Component installation, with servos lashed to inside of fuselage ......................................50

3-31 Tyvek hinge rudder and attachment of control rod to control surfaces .............................50

4-1 Thin, flexible wing model used to compare wind tunnel aerodynamics with Athena
Vortex Latice's computational aerodynamics approximation........................................56

4-2 Thin, flexible batten reinforced MAV wing with Icarex used in wind tunnel testing............56

4-3 Lift vs. angle of attack comparison between experimental wind tunnel data and
simulated results from AVL for a thin, under-cambered flexible wing.............................57

4-4 Pitching moment vs. angle of attack comparison between experimental wind tunnel data
and simulated results from AVL for a thin, under-cambered flexible wing.....................57

4-5 Drag vs. angle of attack comparison between experimental wind tunnel data and
simulated results from AVL for a thin, under-cambered flexible wing.............................58









4-6 Lift vs. drag comparison between experimental wind tunnel data and simulated results
from AVL for a thin, under-cambered flexible wing..................... ........................58

4-7 Visual image correlation 3D scanning overlay of out of plane displacement on painted
m old (speckle pattern)........... .................................................................... ........ .. ....... .. 59

4-8 Comparison of wing model geometry and actual wing mold, where color represents
variation in out of plane position in millimeters (note the yellow spot at the center of
the root of the wing, indicating a purposeful gouge in the mold as a reference point)......59

4-9 Comparison of MAVLAB wing model and machined wing mold cross section along the
w ing span .................................................................................60

4-10 Measured machining error along the span-wise cross section of the wing ........................60

4-11 Comparison of MAVLAB wing model and machined wing mold cross section along
the chord-w ise direction (airfoil). ........................................................... .....................61

4-12 Measured machining error along the chord-wise cross section of the wing. .....................61

5-1 Diagram of iterative aerodynamic analysis where the wing forces computed by AVL are
input into a FEA solver to determine the steady state deformation of a thin, flexible
w ing during flight. ....................................................... ................. 64

A-i M AVLAB Graphical User Interface. ............................................ ............................ 72

A -2 A V L G raphical U ser Interface...................................................................... ...................72

A -3 CN C Toolpath generator G U I. ..................................................................... ...................73

A -4 Z igzag toolpath exam ple............................................................................. .....................74

A -5 Spiral toolpath exam ple. ............................................................................. ......................74

A-6 Verification of milling toolpath via simulation (note that the minimum dimension of
the work piece was defined in the rendering). ......................................... ...............75









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

DEVELOPMENT OF A PARAMETRIC SOFTWARE TOOL FOR THE DESIGN AND
MANUFACTURING OF MICRO AIR VEHICLES

By

Daniel J. Claxton

May 2007

Chair: Peter Ifju
Major: Aerospace Engineering

Micro air vehicles (MAVs) are a special subset of unmanned air vehicles (UAVs) that

warrant a significant level of scientific interest. In general, MAVs are small, inexpensive and

often expendable platforms, flown by remote pilot, or autopilot. Because they maybe flown by

autonomous control or inexperienced pilots, they must have very reliable and benign flight

characteristics built into their design.

The University of Florida has developed a series of MAVs that adopt a flexible-wing

concept, most notably featuring a carbon fiber structure and a thin extensible membrane skin.

Because their design requirements mandate that they perform reliably in flight, careful thought

and consideration must go into a MAV design. Some of the design may come from intuition and

experience, but it must ultimately be verified through quantitative testing. In addition, the design

process must be performed in a way that is accurate and repeatable.

The purpose of my research was to develop an efficient and accurate methodology for

designing, producing and reproducing MAVs. My approach evolved into a rapid prototyping

process of designing and manufacturing MAVs while still maintaining geometric accuracy and

aerodynamic integrity. The solution was the development of a software-based design tool, called

MAVLAB, which incorporated specialized CAD design features, aerodynamic analysis tools and









rapid manufacturing through automated machining. My thesis includes an overview of the

University of Florida's design procedure, an example case study and a users' manual for

MAVLAB.












CHAPTER 1
INTRODUCTION

Micro Air Vehicles at a Glance

Micro air vehicles (MAVs), as defined by Defense Advanced Research Programs Agency

(DARPA), are a subset of aircraft with a maximum wing span of 15 cm (about 6 inches) [1]

(Figure 1-1) and a maximum flight speed no greater than 15 m/s [2]. This definition is important

because it corresponds to a regime of aircraft operating in a low Reynolds number (Re) region,

where aerodynamic behavior becomes more difficult to predict. This type of aircraft is of

particular interest to researchers in the scientific community as well as real world applications.

Small aircraft of this size can be useful for several reasons. They can be made relatively

inexpensive, expendable, portable, and undetectable, which lend well to military and surveillance

type applications. Much of the scientific interest in designing such vehicles comes from the

challenge of overcoming the loss of aerodynamic efficiency due to scale, as shown in Figure 1-2.

Mircro Air Vehicles at the Research Level

The challenges of designing an efficient, controllable MAV are what prompted the

University of Florida's involvement in MAV research. In order to overcome some of these

aerodynamic and technical challenges, the University of Florida developed a series of MAVs that

incorporated a unique, thin, under-cambered, flexible wing design (Figure 1-3) [1,3-6]. Studies

have shown that thin wings have a distinct advantage over thicker, volumetric wings at the low

Reynolds numbers where MAVs typically operate [3,7-8]. In general, the wings of these

vehicles are constructed of a carbon fiber skeleton and a thin flexible membrane, originally

inspired by the structure of bat wings and wind-surfing sails.









The University of Florida has been taking an active role in the investigation of MAVs,

researching and developing new MAV concepts for the past 9 years. In that time, the university

has come a long way in terms of development and testing procedure, while the main aircraft

design philosophy has remained relatively the same.

The basic philosophy of MAV research at the University of Florida initially evolved from

a design-build-fly approach with the main objective of creating the smallest possible flying fixed

wing aircraft. This was spawned by research grants with this goal in mind, and further inspired

by the creation of the International Micro Air Vehicle Competition. Over time, and after much

success, the University's focus shifted away from the concept of producing the smallest flying

aircraft in the MAV regime, but rather the smallest, mission-capable aircraft. This was

influenced mainly from the practical standpoint that many research funding institutions in this

field were defense contractors that wanted MAVs and small UAVs that were useful and mission

capable. This meant that they needed to be capable of flying missions that required longer flight

times, more payload capacity and maintained a greater degree of autonomy. In order to perform

these tasks, a greater understanding of flight dynamics was necessary. Figure 1-4 depicts the

relative trend in payload capacity for the MAV and small UAV scale.

This need to better characterize and understand small aircraft benefited MAV research at

the University of Florida. Eventually, it lead to the much-needed replacement of an aging open-

circuit low speed wind tunnel with a new closed circuit low speed wind tunnel. With this

addition, better analysis could be performed on MAVs through quantitative loads measurements.

In addition, a visual image correlation (VIC) system was purchased and used in conjunction with

the wind-tunnel to measure in-flight deformation of the thin, flexible wings. All together, these









advances in technology helped to better characterize and understand MAVs from quantitative

testing.

There was one part of the puzzle still missing. While there was some understanding as to

what a MAV design should look like, there was no official agreement on what guidelines the

most efficient design should uphold. This left researchers to build and test hundreds of designs

and try to form a relationship between design parameters and aerodynamic performance.

Motivation and Overview

Most of the University of Florida's MAV designs have revolved around the same central

design philosophy, with some variation on geometric design parameters, depending on the

desired performance characteristics and design requirements. However, even small changes to

any given design can drastically enhance or degrade aerodynamic performance. This is

especially true at these small scales. Further, any unintended deviation from the desired design

geometry would introduce unwanted effects. Thus a method of accurately designing, analyzing

and manufacturing MAVs was necessary.

In order to produce MAVs accurately, a process of forming composites over a machined

mold was initially developed. The construction of these molds can be described from the

following process. The airframe was first modeled in a 3D computer aided design (CAD)

program. The surface of the airframe was exported to computer aided machining (CAM)

software. The CAM software could then be used to generate a set of machine instructions for a

computer numerically controlled (CNC) to interpret and automate the milling of a mold. The

mold is then used in the production of the carbon fiber airframe structure. Depending on the size

and complexity of the airframe and the skill level of the designer, this process could take

anywhere from days to weeks to transition from concept to working model.









Wind-tunnel and flight testing are the preferred choice when it comes to analysis of new

MAV concepts and prototypes. Work has also been done to experimentally quantify wing

deformation using CFD [8], projection moire interferometry [9- 10] and photogrammetry [11].

None of these techniques were practical or possible before manufacturing. An intermediate and

iterative method was needed to analyze aerodynamics and aid in MAV design before production.

In order to streamline the MAV design process, a parametric software design tool was

developed. The idea was to have a simple interface that would allow a user to transform a "back

of the napkin" design to a CAD model in a matter of minutes. In addition the software would

allow the user the ability to adjust a few design parameters which controlled the entire geometry.

Further, the aforementioned "crude" comparative CFD approach could be implemented in

parallel to obtain relative performance of each design parameter. This approach was not created

to produce the exact optimum aircraft design with the press of a button, but rather a design tool

that would benefit an experienced MAV/aircraft designer. The end result of this tool provided

the user with a computer-simulated environment, called MAVLAB, in which the optimum MAV

configuration could evolve.




























Figure 1-1. Typical composite airframe MAV developed at the University of Florida.


103

Region of
Interest

102
CLq All Rigid, Smooth
CLD Max Airfoils
CDI Max

10

10 -


103 104 10" 106 107
Re, UC
Rec V
Figure 1-2. Dependency of aircraft aerodynamic performance on scale, where performance is
represented by maximum lift/drag and scale is represented by the non-dimensional
quantity Re.





























Figure 1-3. Example of a thin, under-cambered, flexible membrane Micro Air Vehicle wing.


ihwt
G*rl



/ -I'- E Umk


\ Ilunaa
I'H tr llr


SMALL
ITAVs


- I MAV~


o iI nl
W in g|Tan i^n"


Figure 1-4. Relative scale and payload capacity of Micro Air Vehicles and Small Unmanned Air
Vehicles.


Il
L


F'rllnrrr '~4~


r


Ct









CHAPTER 2
LITERATURE REVIEW

Micro Air Vehicle Design

The history of MAV design is a relatively short one, as the definition of MAVs has only

been around for just over a decade. In 1996 DARPA announced the Micro Air Vehicle Program

initiative, seeking to develop and test emerging technologies that could evolve into a mission

capable flight system for military surveillance and reconnaissance applications. The only

requirement was that the dimension of the vehicle should not exceed 15 cm. There were no other

restrictions on the design [1].

The DARPA initiative was what spurred the development of MAV designs in both

industry and academia. One of the first successful and practical MAVs was the Black Widow,

developed by Aerovironment (Figure 2-1). The design of the black widow was performed using

a multi-disciplinary optimization (MDO) approach [12]. In this way, the aircraft subsystems

such as propulsion, aerodynamics and airframe structure could be optimized to fit the design size

requirement. Aerovironment chose to use CAD modeling and a computer software environment

to design and evolve the vehicle. The results of Aerovironment's analysis and design of the

Black Widow MAV proved that a small, lightweight, mission capable, battery powered platform

could be remotely operated up to 1.8 km away and sustain flight times of up to 30 minutes. This

impressive performance would later prove to inspire and influence future MAV work, including

that at the University of Florida.

Computation Fluid Dynamics

There are several methods for performing computational analysis of aerodynamics, all of

which try to model single phase fluid flow. The most advanced method, commonly referred to

as "full blown" CFD uses the Navier-Stokes equations to determine the behavior of a fluid over a









body. Another approach is known as the panel method and attempts to solve linearized potential

flow equations. There is also another method that is similar to the panel method which is very

easy to use and is capable of providing remarkable insight into wing aerodynamics and

component interaction called the vortex lattice method (VLM). It was among the earliest

methods utilizing computers to actually assist aerodynamicists in estimating aircraft

aerodynamics. Vortex lattice methods are based on solutions to Laplace's Equation, and are

subject to the same basic theoretical restrictions that apply to panel methods [13]. Today, there

are dozens of free VLM software codes available for use in aerodynamic analysis, most notably,

Richard Eppler's Profil, and AVL, by Mark Drela and Harold Youngren. Due to better

availability and documentation, AVL was chosen as the core aerodynamics analysis tool for

MAVLAB.

Manufacturing and Rapid Prototyping

Rapid Prototyping (RP) can be defined as a group of techniques used to quickly fabricate a

scale model of a part or assembly using three-dimensional CAD data. Fused Deposition

Modeling (FDM), Stereolithography (SLA), Multi Jet Modeling (MJM), Electron Beam Melting

(EBM), and 3D Printing (3DP) were all common examples of rapid prototyping techniques [14].

While these methods of rapid prototyping were potentially viable processes for producing wing

molds, CNC machining was chosen in stead for the purposes of this research. CNC refers

specifically to a computer "controller" that reads ASCII text file instructions and drives the

machine tool, a powered mechanical device typically used to fabricate metal components by the

selective removal of metal or other material. CNC does numerically directed interpolation of a

cutting tool in the work envelope of a machine. The University of Florida used a basic 3-axis

end mill with a radial cutting tool to machine 3D freeform surfaces out of polyurethane tooling

board.









o-r"


Figure 2-1. Black Widow Micro Air Vehicle developed by Aerovironment using multi-
disciplinary optimization.









CHAPTER 3
DESIGN PROCESS

MAVLAB

The migration of an initial MAV design concept to a working model can be a complicated

and time consuming procedure. Before a MAV could be built and tested at the University of

Florida, it was first created in a computer simulated environment. Solid Works, Pro Engineer,

and Autodesk Inventor were the three major computer aided design (CAD) software packages

used by academia and industry at the time of this writing. These software packages were all well

documented and fully capable of creating parametric MAV designs. However, they had some

drawbacks. They were expensive, bulky, and required a steep learning curve. In addition, they

were not specialized to the University of Florida's approach of designing of aircraft. In an

attempt to increase productivity, aid design and reduce workload, a specialized Matlab-based

software design tool was developed, and named MAVLAB. This program was specifically

designed to translate basic, intuitive design parameters into a complex freeform airframe model.

Figure. 3-1 depicts a typical wing geometry in MAVLAB. In addition, MAVLAB could perform

aerodynamic analysis via a CFD plug-in. It also has the ability to export the geometry to other

CAD software. Ultimately, MAVLAB's most powerful feature was the ease in which it aided in

manufacturing. This process, called computer aided manufacturing (CAM) was what made the

University of Florida successful in verifying MAV designs. MAVLAB effortlessly converted

wing geometry into a computer numerically controlled (CNC) machine instructions, called a tool

path, used for automated machining and prototyping of molds. These molds would then go on to

be used in the fabrication of composite MAV wings.

The concept behind the CAD portion of MAVLAB was to produce an intuitive parametric

based model. The 7 basic parameters used to control the geometry are self explanatory in nature,









and relate to specific geometric descriptions of a generalized wing, rather than ambiguous

dimensions that have no direct relationship to aircraft dynamics. These parameters are listed in

Table 3-1 and are explained later in the chapter.

Scalar-Based Design Parameters

MAVLAB featured a simple GUI (Figure 3-1) designed to guide the user through the

construction of a wing via simple, scalar valued, geometric parameters. These parameters

included wing span, root chord, span scale, dihedral angle, sweep angle and geometric twist

angle, which were chosen because they had a direct correlation with wing aerodynamics.

Additionally, there was one parameter included for aesthetics and structural reasons, called the

leading edge reference. Each input could be changed by typing the value in a corresponding field

causing a 3D view of the wing to regenerate instantaneously in a neighboring window.

Additionally, parameters could be modified by clicking and dragging on an icon next to each

parameter. The user could then interact with the model by rotating and changing the viewpoint

with the mouse (Figure 3-2) to verify the geometry from different angles, in real time.

Wing span. The wing span is a simple feature often used to describe the length of a wing.

The wing span is measured from the left wing tip to the right wing tip in a straight line, and is

commonly denoted by the character "b". For clarification, Figure 3-3 illustrates the wing span

parameter on a generic wing however this definition also applies to MAV wings as well.

Root chord. In general, a chord refers to the length of an airfoil cross-section of a wing.

This is the linear distance from the leading edge of the wing to the trailing edge (Figure 3-4).

The root chord is the chord at the root of the wing; the point closest to the fuselage or middle of

the wing span. This parameter was used as a reference to scale the entire chord of the wing.

Airfoil camber. An airfoil is the cross section of the wing, whose shape is so important to

aerodynamics that it alone is often a good indicator of wing performance (Figure 3-5). In









general, airfoils are composed of a top and a bottom curve, creating a closed region. The mean

of those two curves is called the camber line. The distance between the top and bottom curves is

called the thickness distribution. The thin airfoils, characteristic of the University of Florida's

MAV wings were a simplification of this definition, with a thickness distribution of zero. It was

found that at the MAV scale, aerodynamics could benefit from thinner airfoils, and thus

MAVLAB was designed around this philosophy. The parameter that controls airfoil camber is

called "max camber", or simply "camber", and is the ratio between the maximum value of

camber and the length of the airfoil chord. This was beneficial to the user because it allowed

modification to the current airfoil coordinates selected for the wing.

Span scale. The span scale refers to a modifier parameter for the span shape, mentioned

later in the vectorized parameter section. The span scale allows the user to increase or decrease

the magnitude of the span shape by a scale factor, usually between 0 and 1. Negative numbers

inverted the span shape.

Dihedral angle. In geometry, the dihedral angle is defined as the angle between the

intersection of two planar surfaces. For aircraft, this is commonly seen as the angle between the

horizontal (ground) and the wing. The dihedral of a wing is a specific form of span shape, and is

defined separately in MAVLAB because it is so commonly used. Figure 3-6 illustrates the

dihedral angle parameter.

Sweep angle. Wing sweep is another common geometric feature found in aircraft wings,

shown in figure 3-7. It is defined as the angle between the 14 chord line and the line

perpendicular to the chord. The aerodynamic significance of wing sweep can provide both the

same effect as the dihedral angle, while moving the aerodynamic center (AC) further aft. This is









useful for flying wings and MAVs, where lack of volume renders it difficult to get the center of

gravity (CG) in front of the AC for stability.

Twist angle. The geometric twist angle (shown in figure 3-8) is the span-wise change in

airfoil angle of incidence, usually measured in degrees with respect to the root airfoil. This is the

parameter that can be controlled in MAVLAB and differs from aerodynamic twist, which can

provide the same results by a span-wise change in airfoil shape. Most aircraft have several

degrees of twist in the wing to prevent stall from occurring all at once. This is generally called

washout and refers to a negative twist angle. In MAVLAB, the direction of the twist angle is

reversed.

Leading edge reference. The leading edge reference or "edge ref' is a parameter that

differs from all the rest, in that it is not closely coupled with aerodynamic performance. When a

wing of arbitrary shape is generated in MAVLAB, an unintended artifact can occur. The span-

wise shape of the wing will be partially a function of the planform, due to scaling of the airfoil

along the span. This may or may not cause the wing to look funny, and may make the wing

difficult to mount to a fuselage or perform other structural duties. To alleviate this, the logical

parameter "edge ref' was introduced to allow the user to select between a leading edge reference

for the airfoil position, or a maximum camber reference. The latter of the two often produces a

much more aesthetically pleasing wing shape, and is consequently used most often.

Vector-Based Design Parameters

In addition to the basic MAVLAB parameters there were 3 wing parameters that were a

little more difficult to define, but equally as important. They included the planform shape,

airfoil, and, span-wise shape. Each of these parameters corresponded to the top, side and front

view of the wing, respectively, and were dependent on the aforementioned scalar parameters.









These parameters are referred to as vector-based (or vectorized) due to the fact that they needed

to be defined by a data set, rather than a single parameter.

Planform. The planform is the birds-eye-view, or top projection of the wing onto a flat

surface. This shape is closely tied to aerodynamic efficiency and over-all lift. Figure 3-3

illustrates the planform of a generalized airplane wing. While early releases of MAVLAB

defined the planform as a function of scalar based parameters, a more general method of

producing planform shapes was more desirable. The most common planform shapes are the

tapered wing and elliptical wing, the former of which is seen in most commercial aircraft. This

is because a tapered wing can be constructed in such a way that it is almost as aerodynamically

efficient as an elliptical wing, but lighter and more cost-effective. MAVLAB initially was

developed with these two shapes in mind, and consequently only let the user select between the

two, with variations of the taper ratio, and ellipse ratio. The ellipse ratio simply described a

more general elliptical shape composed of two ellipses commonly joined at their major axis, but

having different semi-minor axes. This ratio of semi-minor axes was called the ellipse ratio.

As MAVLAB developed it was clear that cost and weight restrictions were not a concern

in the composite construction used to make MAVs, and therefore more complicated planform

shapes could be explored. In MAVLAB the planform could be defined as any closed polygon

imaginable. This could be achieved in several ways. The planform could be input from an

ASCII text file, similar to common airfoil coordinate files. Alternatively, the planform could be

interactively drawn and manipulated using an interpolating spline.

Airfoil. As mentioned earlier, the airfoil is the single most critical part of a wing. The

airfoil shape largely governs the lift, drag and pitching moment of the aircraft [15]. Thus, the

selection and design of an airfoil is often critical. The original airfoil for the University of









Florida MAV concept was designed under the following constraints. It had to be a thin airfoil

with a neutral (i.e. zero) pitching moment. In order to produce this, a genetic optimization was

performed using a well known panel method viscous airfoil analysis software called XFOIL

(developed by the creator of AVL, Mark Drela). Since the airfoil was thin, it could be described

by a polynomial of order 6, and the coefficients of the polynomial were used as the variables for

optimization. The results of this analysis were what inspired the use of the thin, reflexed airfoil

largely used in MAV designs at the University of Florida. Figure 3-5 depicts a more general

thick airfoil, but the camber line in the illustration is more representative of what a thin airfoil

would look like.

MAVLAB offers the capability to fully define thin airfoils for a given wing design. This

could be performed in one of two ways. First, the user had the ability to read in normalized

airfoil coordinates from an ASCII text file, which described the airfoil geometry as a discritized

set ofx and y data points (tab or space delimited) from leading to trailing edge. Second, the user

could create a custom airfoil using an interactive interpolating spline curve. In addition, the

custom airfoil could be expressed as a function of the chord position in the range of 0-1. Only

one airfoil had to be defined for a given wing, but there could potentially be an unlimited number

of airfoils. Each airfoil was assigned a normalized span-wise position from 0 (the root) to 1 (the

wing tip). MAVLAB automatically and smoothly interpolated between each airfoil section to

create a smooth continuous wing surface.

Span shape. The span shape is a generalization of the dihedral of a wing. It basically

refers to the shape of the span-wise shape of the wing from the front view, as seen in Figure 3-6.

Like the planform, the span shape could potentially take any form. It was defined by a set of

data points describing the overall vertical displacement of the wing as a function of span. This









could be done by either importing a text file of coordinates or drawing one half of the symmetric

span shape with an interpolating cubic spline. Additionally, the shape of the span could be

defined as a explicit function on the range of 0 to 1, where 0 corresponded to the root and 1

corresponded to the wing tip.

Aerodynamic Analysis

In addition to providing the user with an efficient geometrical design interface, MAVLAB

also provides the capability to perform aerodynamic analysis. MAVLAB can export geometry

data to an extended vortex lattice CFD software called AVL. Created by Mark Drela, AVL

(Athena Vortex Lattice Model) was intended for rapid aircraft configuration analysis and could

compute aerodynamic metrics such as lift and drag coefficients as well as stability derivatives. It

also had the capability to model slender bodies, as well as performing trim flight calculations and

eigenmode analysis. While the vortex lattice method of solving for the fluid flow of a surface

has limitations, it gave reasonable lift and stability data as well as a relative trend in

performance.

Computer Numerically Controlled Machining

MAVLAB has the capability to export tool path files to a CNC milling machine. The

advantage of this feature is the ability to mill wing molds that were exact representations of the

CAD designs created in MAVLAB, reducing the likelihood of geometrical asymmetries in the

airframe. Figure 3-9 depicts a wing mold being machined on a CNC milling machine.

The molds were machined from high density tooling board. This material was chosen

because it was a tough, high-density polyurethane product intended for use in models,

prototypes, mold-tooling, and in composite mold applications where a uniform, grain-free,

dimensionally stable substrate is desired. In addition, it was designed to withstand the high

temperatures needed for composite curing and could be machined similar to aluminum.









However, the advantage over other materials such as aluminum was that the tooling board could

be machined at much higher feed rates, increasing material removal rate and reducing production

time. This advantage allowed the machined wing molds to be produced in as little as 25 minutes

on a standard 3-axis CNC end mill at a feed-rate of 45 inch/min. This time was dependent on the

size of the wing mold, but could potentially be dramatically improved if high-speed machining

was implemented. At speeds of 600 inches/min and up, the machine time could be reduced by

an order of magnitude.

After machining, the wing molds were not perfectly smooth due to the cusps created by the

scalloping of the ball end-mill (Figure 3-10). It was important that the wing mold be smooth, so

the remaining material from the scalloping had to be sanded away by hand. To assure that the

remaining material was removed evenly, a light trace coat of black spray paint was applied to the

mold surface before sanding. Then, a high-grit sand paper was applied to the surface until all

painted regions were gone. An example of a completed wing mold is shown in Figure 3-11.

The scallop height is a physical parameter which directly affects the surface finish of the

part. Mathematically, it is the maximum height of the ridges produced by the ball end-mill (see

Figure 3-12). The following equation relates the scallop height (hscallop) to the step-over distance

(xstepover) and diameter of the tool (Dtool).

1 2 22
scallop oo xepor 3-1

Step-over is the normal distance between adjacent paths of the tool.

In order to create a tool path, from the wing geometry, the surface had to be able to be

finely meshed according to the desired scallop height and direction of cut. The surface normals at

each of the grid points were calculated and scaled to the radius of the tool. Each normal is









placed at its corresponding grid point. The order in which these points were arranged determined

the toolpath.

Depending on the geometry, it could be necessary for the toolpath to follow various

patterns. MAVLAB's options for tool path patterns included a zigzag (chord-wise/span-wise)

and spiral path. In addition to creating the toolpath, MAVLAB also overlayed the toolpath

pattern over the wing surface and drew a box around the wing to signify the tool containment

boundary. The length, width, and height of this box are also returned for reference. Finally, in

order to verify that the toolpath was exactly what the user wanted, MAVLAB gave the user the

opportunity to visualize a simulated machining process, called "toolpath verification".

Composite Construction

The typical fabrication of a MAV revolved around the use of composite materials within

the wing and fuselage structure, including carbon fiber and Kevlar. The following documents

the custom fabrication process developed by the University of Florida [16-21]. The choice of

composite materials was influenced by their material properties. The primary composite utilized

was carbon fiber and thermoset epoxy in a pre-impregnated form which demonstrated an

inherently high strength to weight ratio, which was ideal for aircraft structures. This allowed the

possibility of creating a light weight wing with a thin airfoil while still being able to sustain

aerodynamic loading. The carbon fiber cloth used in construction was a 0/900 plain weave,

176.3 g/m2, 3000 fiber/yar pre-impregnated fabric. In addition, pre-impregnated unidirectional

carbon fiber is also used. Despite the high strength of the carbon fiber, 17 g/m2 Kevlar cloth

reinforcement was sparingly used to prevent crack propagation (brittle failure) in the 0/900

weave cloth on the wing leading edge. Kevlar, while not nearly as strong as carbon fiber, was

extremely resistant to shear stresses. Thus, Kevlar was concentrated around regions

susceptible to high impact loads, such as the leading edge of the wing.









The wing of the MAV was the most important airframe component with respect to flight

performance, requiring a great deal of care to be taken in its fabrication alone. No viable

automated process was available to lay up the wings, so development of a new procedure for

composite construction of the airframe was necessary. In addition, the University of Florida has

developed several different wing fabrication techniques. This section will outline the fabrication

process of a batten reinforced, flexible membrane wing and a rigid composite wing (both shown

in Figure 3-13). In addition, this section will also cover the procedure of fabricating a composite

fuselage.

Flexible Wing Fabrication

More complex than a solid composite layup, flexible wing designs first required a detailed

layup diagram to be printed on the tool. This was often drafted in a computer drawing program

and then printed to paper. The paper was then overlaid on the mold and traced out with a marker

(Figure 3-14).

Teflon release film was then adhered to the wing tool using Elmer's spray glue. The tool

was, first, lightly coated with spray glue, and then Teflon release film was delicately applied to

the surface (Figure 3-15) by tacking it down at one corner and then spreading the entire sheet

into the contours of the wing surface. Care was taken to prevent wrinkles from forming, as this

would adversely affect the final surface finish.

The leading 20% of the wing was constructed of a single layer of 0/900 cloth oriented at a

45 degree angle with respect to the chord line. A pattern was generally used to cut out a piece of

carbon fiber to the proper shape (Figure 3-16). The resulting patch of carbon fiber was aligned

with the pattern on the tool. The 0/900 cloth was then reinforced by a single additional layer

made of Kevlar cloth (Figure 3-15).









Two millimeter wide strips of unidirectional carbon fiber were cut from a sheet of pre-

impregnated material using a custom built carbon strip-cutting tool (Figure 3-17). The portion of

the wing that attached to the fuselage was made by carefully aligning layers of unidirectional

carbon fiber strips with the pattern drawn on the tool (Figures 3-18,19). The unidirectional

carbon fiber was layered for strength and rigidity and was interwoven at intersections to create

stronger mechanical joints. Elevator control surface hinges were created by interweaving a single

layer of Tyvek material into the gaps between the control surface and frame.

Rigid Wing Fabrication

Though the University of Florida was known for the development of the flexible

membrane wing, it also developed a bendable rigid wing. While the rigid wing did not benefit

from the desirable flight characteristics of the membrane wing, it was more suitable for the

design of the bendable wing. The rigid wing was simpler in construction than the flexible wing,

but required a separate fabrication process, documented below.

Rigid wing construction was largely similar to the flexible membrane wing, so the process

is only summarized in the following steps. First, Teflon release film was applied to the surface

of the mold with spray adhesive to prevent the resin from bonding to the mold. A single layer of

pre-impregnated carbon fiber weave was typically placed on the release film, biased at a +450

orientation. Kevlar weave was cut out and placed over the leading edge of the wing.

Additional layers of carbon fiber were added as needed to strengthen the wing where high

bending stresses were expected. Figure 3-20 shows an example of a rigid carbon fiber wing after

layup and cure.

Fuselage Fabrication

The fuselage lay-up was completed in a similar process. Carbon fiber weave was wrapped

around a Teflon lined male mold to produce the nose of the fuselage. Multiple layers of carbon









fiber and Kevlar were applied to the nose to provide adequate reinforcement for impacts. The

fuselage molds were not produced using MAVLAB, as the precision of their construction was

not as critical, and they were difficult to define by simple parameters (Figures 3-21-23).

Vacuum Bagging and Curing

The entire wing and fuselage were covered with porous Teflon peel-ply, to allow excess

resin to pull out of the carbon fiber without sticking to it. A layer of fabric breather material was

placed on top of the molds to ensure a vacuum would be applied evenly to the composite covered

mold surfaces. Finally, a small hole was cut in the vacuum bag to accept the hose attachment for

the vacuum pump. Vacuum bag tape was used to seal up both ends of the bag and the hose was

connected to the attachment and a vacuum was drawn to 0.01 atm absolute. Once the air had

been removed from the bag, it was inspected for leaks and air pockets. Leaks were sealed up

with additional tape and air pockets were alleviated from inside corners by repeatedly removing

the vacuum and forcing the bag into the corners. The vacuum bag was then cured in an oven at

1300 C for 2 hours (Figures 3-24,25).

Fuselage Assembly

Once cured, excess material was trimmed away as shown in Figure 3-20, and edges were

sanded to remove splinters. The fuselage and wing were mated together to ensure an acceptable

fit (Figure 3-26). The wing was often placed on the wing mold, and the fuselage is glued into

place, a method that helps to reduce asymmetries in the MAV.

For the flexible wing construction, the remaining open surface of the wing was then

skinned with 0.25 mm latex rubber. The latex was applied in a three step process. All but a 3

mm perimeter of contact surface around each of the open areas of the wing were masked with

masking tape (Figure 3-27). Next, Elmers spray glue was then lightly coated to the top surface

of the wing. The masking tape was then removed and the glue was left to set-up for thirty









seconds. Lastly, a latex membrane was adhered to the top surface of the wing. An aptly sized

rectangular section of latex was first stretched out over a block of foam and pinned down to

secure some pretension in the membrane and prevent any slack or wrinkles. This pretension was

not typically measured, however. It was important to note that the entire wing was skinned at

once in order to keep the relative membrane tension in both wings equal (Figure 3-28). The latex

was then pressed firmly against the wing, trimmed with a razor blade and secured with

cyanoacrylate glue (Figure 3-29).

Components Installation

Servos were attached to the fuselage by double sided tape (Figure 30), and lashed into

place with Kevlar thread. Control rods (1mm diameter steel) were then installed, connecting

the servos to the control surfaces (Figure 3-31). The rudder control surface was created by

cutting a section out of the vertical stabilizer, and then reattaching them with Tyvek hinges and

glue. Special care was taken to minimize any clearance or play in these linkages. Electronic

components such as the electronic speed controller, GPS receiver and autopilot were then placed

inside the fuselage and adhered with double-sided tape where necessary. An acceptable size hole

was bored into the nose of the fuselage for the motor using a Dremmel tool. Glue was used to

adhere the motor to the fuselage with the proper thrust angle.













Table 3-1. List of basic MAVLAB wing parameters
Parameter Dimension Type Description
Span Length Wing span
Chord Length Root chord length
Camber % Airfoil Maximum Camber
Z-scale % Scale factor for Z-direction
Twist Angle (degrees) Spanwise wing twist
Sweep Angle (degrees) Wing sweep angle
Dihedral Angle (degrees) Wing dihedral angle
Edge Ref Logical Span shape referenced at leading edge (True)












- .. .


- Parameters I- 3D View


Span F 24 'Q]
Chord 7
Camber 0.06 7
Z-Scale 0.25
Twist 0 A ']

Sweep 0 i]
Dihedral 0 &~


[ Mirror Wing
j Max Camber Z-Reference


Figure 3-1. MAVLAB Graphical User Interface.


-6 ""315
5


Figure 3-2. Typical graphical display of a wing geometry in MAVLAB, with feedback for basic
parameters such as planform area, mean chord, aerodynamic center and aspect ratio.


- (Click to edit),


- Span-wise Shape


0.67
-0.1219
12


PA =
MC =
AC =
AR =


134.3535
5.877422
0.2499742
4.083423










Span


Figure 3-3. Wing span as defined by the linear distance from wing tip to wing tip.


Figure 3-4. Wing root chord is defined as the distance from leading to trailing edge of the wing,
closest to the fuselage, or middle of the span.









Camber




----------------
-----------------
Figure 3-5. The maximum camber is the ratio of the highest point of the camber line to the
airfoil chord length.


Dihedral


Figure 3-6. The geometric dihedral angle is the angle the wing makes with the horizontal.


Figure 3-7. The sweep angle is the angle between the /4 chord line (dotted) and the line
perpendicular to the free stream velocity.

























Twist (washout)


Figure 3-8. Geometric wing twist, often called washout when it is negative and washin when it
is positive.































Figure 3-9. Typical example of CNC milling of a MAV wing mold out of resin-based tooling
board.


. radial step-over


*9


cusp
height


axial
depth
of cut


Figure 3-10. Scalloping or cusps created by consecutive passes of a ball (radial) end mill.


































Figure 3-11. Typical wing mold machined from high-density polyurethane tooling board.


________


Figure 3-12. Typical ball end mill used for machining 3D freeform surfaces such as wing molds.


A. B.
Figure 3-13. A) Rigid carbon fiber wing. B) flexible membrane wing with batten reinforcement.
































Figure 3-14. Drawing layout pattern on the wing mold.


Figure 3-15. Applying Teflon release film to the surface of the wing mold using spray glue


































Figure 3-16. Laying up of carbon


Figure 3-17. Application of Kevlar to the leading edge.


er on the wing mold.
































Figure 3-18. Cutting of unidirectional carbon fiber strips with custom cutting tool.
Figure 3-18. Cutting of unidirectional carbon fiber strips with custom cutting tool.


Figure 3-19. Trimming of excess carbon































Figure 3-20. Trimming excess carbon fiber from wing after cure.


Figure 3-21. Application of Teflon release film to fuselage mold.


































Figure 3-22. Fuselage placed on a form to maintain proper mate with wing.


Fer to the tuselage mold.


Figure 3-

































Figure 3-24. Vacuum bagging of wing tool.


Figure 3-25. Vacuum bagging of fuselage covered with peal-ply release film.
































Figure 3-26. Assembly of wing and fuselage after cure.


Figure 3-27. Masking prep and application of spray glue to top of wing for latex application.
































Figure 3-28. Application of latex membrane to top of wing.


Figure 3-29. After latex trimming, final gluing of latex to carbon fiber skeleton of the wing.































Figure 3-30. Component installation, with servos lashed to inside of fuselage.


Figure 3-31. Tyvek hinge rudder and attachment of control rod to control surfaces.









CHAPTER 4
REPRESENTATIVE STUDY

In theory, MAVLAB was a clearly useful solution for the design and manufacturing of

MAVs. In order to test this theory, a representative study was done on a MAV wing to verify the

accuracy in which MAVLAB could aid in predicting aerodynamics as well as the accuracy in

which it could generate wing molds.

Computer Model Description

Despite the name, MAVLAB was capable of designing and analyzing wings of all sizes,

not just 6 inch MAVs as originally defined by DARPA. In fact, in recent years, this dimensional

restriction for the definition of the MAV has been relaxed, as it was not seen necessary to

explore MAV designs only under 15 cm in maximum dimension. As such, the University of

Florida has done much research in the 6 24 inch realm of small UAVs. The following study

examined the design of a thin, flexible MAV shown in figure 4-1 and 4-2. The wing was

constructed using the custom fabrication process for flexible, batten reinforced wings as

described in chapter 3, with one exception. An inelastic light-weight, polycarbonate coated

nylon fabric called Icarex, was used instead of latex.

Wing Aerodynamics Comparison

The particular MAV wing design used in this study was originally designed to fly at a

cruise speed of 13 m/s with an angle of attack of 7-10 degrees. For this reason, testing was only

done at 13 m/s. This section compares the lift, drag and pitching moment of the computer

simulated model and actual wind tunnel results using the respective non-dimensional coefficients

CL, CD and Cm.









Wind Tunnel Results

All wind tunnel testing was performed in a newly installed Engineering Laboratory Design

(ELD) 407B wind tunnel at the University of Florida. The tunnel had two interchangeable test

sections. The smaller of the test sections had inner dimensions of 0.61 m x 0.61 m x 2.44 m and

has listed velocity capability ranging from 3 m/s to 91.4 m/s. The small test section was installed

downstream of the flow screens and plenum and was upstream of the straight walled diffuser.

The "large" test section has inner dimensions of 0.838 m x 0.838 m x 2.44 m and has listed

velocity capability ranging from 2 m/s to 45 m/s [22]. The large section was used for this study.

The University of Florida's low speed wind tunnel was equipped with a 6-component sting

balance used to measure aerodynamic loads. Under the given loads for this particular test case,

the sting balance could accurately predict lift and pitching moment, however drag was

questionable. This was due to the fact that the magnitude of the drag forces for this model were

on the same order of magnitude as the measured resolution of the drag axis on the sting

balance[23-24]. Thus drag measurements were not repeatable and likely questionable.

Loads measurements were take at a range of angles of attack (alpha), from -5 to 25

degrees. Figures 4-3 through 4-5 show the relationships of lift, pitching moment and drag vs.

angle of attack. Figure 4-6 represents a metric of aerodynamic efficiency, called the lift to drag

ratio. This data was based on the data from the first two figures. Figure 4-3 shows that the wing

demonstrated a zero lift angle of attack of roughly -2 degrees, a lift slope of 0.06 and stall at

about 17 degrees. The slope of the pitching moment vs. alpha (Cma) is an indicator of

longitudinal stability, where a negative slope is desirable. In this case, the wind tunnel results

showed that the wing had a Cma of approximately -0.013. Minimum drag appeared to occur at 5

degrees angle of attack; however this is questionable due to the reasons stated above.









Athena Vortex Lattice Results

The aerodynamic simulations performed in MAVLAB were done by automatically

exporting the wing data to a free VLM software called AVL [25]. AVL was capable of

predicting a variety of aerodynamic and stability metrics in a matter of seconds. But, since the

wind tunnel test results were limited to lift, drag and pitching moment vs. angle of attack, these

were the only values valid for comparison.

The vortex lattice method implemented by AVL does not fully model the behavior of a

body in fluid flow, and thus there were some assumptions that needed to be considered. This

method models fluid flow as inviscid, and as such, does not accurately predict wing stall or drag.

The drag that it does measure is induced drag, which should only be a small portion of the total

drag of a wing on this scale. In addition, AVL does not model the flexibility of the wing, and

cannot accurately model the steady state deformation of the wing during flight.

That being said, AVL was able to predict aircraft performance relatively well. The model

was partitioned into a grid of 24 span-wise points and 8 chord-wise points. It took roughly 15

seconds to compute the lift, drag and pitching moment at a range of angles of attacks between 0

degrees and 20 degrees. Results matched closely to wind tunnel tests and can be seen in Figures

4-3 through 4-6, with the exception of drag. This was to be expected for two reasons, since AVL

only predicts induced drag and the wind tunnel's measurements were questionable in the first

place. It should also be noted that though the pitching moment curves don't appear to be

consistent, Figure 4-4 is misleading. The slopes are nearly identical, but there is a slight vertical

shift. This is possibly a function of the rigid wing assumption in AVL, but it is not a serious

concern, as the slope of the curve is most important.









Wing Mold Accuracy

It was vitally important that the wing geometry be produced and reproduced as accurately

as possible. Any discrepancies in the wing mold geometry would lead to unwanted behavior in

aircraft performance. In addition, it would also weaken the correlation between simulated

aerodynamic analysis and experimental measurement. The following documents the process of

quantifying the accuracy of a wing mold produced from MAVLAB.

In order to experimentally quantify in flight elastic wing deformation of thin, flexible

wings, the University of Florida has employed the Visual Image Correlation (VIC) system,

developed at the University of South Carolina [6,20,23]. The VIC system was developed by

Helm et al. [27] in the mid 1990s and provided a global shape and deformation measurement for

3D geometries. The VIC system used two cameras to obtain highly accurate 3D measurements

of a surface prepared with a low luster, high contrast, random speckle pattern. Measurements,

both in plane and out of plane, were obtained by a comparison of the test subject in the deformed

state and a reference state. Error quantification was based on many variables and was different

for each set up. Errors of 0.05 mm have been reported [26].

Figure 4-6 illustrates the out of plane position of the wing mold as measured by the VIC

system. The speckled pattern can be seen behind the color overlay. It should be noted that the

VIC was not capable of measuring position up to the boundary of the region of interest, leaving

an approximately 3mm border around the perimeter of the wing that was not measured.

In order to quantify the difference between the machined wing mold and the computer

model, the experimental VIC data was read into Matlab and interpolated onto the same grid

used by the original computer model. The difference of the full-field out of plane position for

each was computed and is represented in Figure 4-7. The zero displacement region around the

edge of the wing was an artifact of the fact that the VIC could not measure all the way to the









border, as mentioned earlier. Note that there is also an apparent raised region (in yellow and

circled in Figure 4-8). This was due to an intentional circular gouge made in the mold by the

CNC at the time of machining in order to produce a reference point. Figures 4-9 through 4-12

depict slices of the wing for better comparison and visualization of error.

Conclusion

Overall, MAVLAB seemed to demonstrate remarkable results in this case study. Despite

rather large assumptions, AVL predicted aerodynamic performance relatively accurately, and

certainly well enough to give the user insight into the performance of the wing design before it

was even tested. Though it appeared, at first glance, that there were considerably large

differences between the actual wing mold and the modeled surface, the range of error between

the two was within the error of the VIC resolution of 0.05 mm. The user could be sure that the

computer modeled wing would be produced accurately via a precisely machined wing mold.

The final results of the VIC testing showed that the wing was machined within acceptable

tolerances.

















0.7695
0
8


Figure 4-1. Thin, flexible wing model used to compare wind tunnel aerodynamics with Athena
Vortex Latice's computational aerodynamics approximation.


Figure 4-2. Thin, flexible batten reinforced MAV wing with Icarex used in wind tunnel testing.


Area: 87.9
Mean Chord: 5.94
Aero Center: 25%
Aspect Ratio: 2.69






















a Experimental
--- AVL


0.80


5 0.60


0.40

0.20


S 0.0 5 10 15 20 25 30
-0.20
Alpha


Figure 4-3. Lift vs. angle of attack comparison between experimental wind tunnel data and
simulated results from AVL for a thin, under-cambered flexible wing.


15 20 25 30



S Experimental
AVL


- A


Alpha


Figure 4-4. Pitching moment vs. angle of attack comparison between experimental wind tunnel
data and simulated results from AVL for a thin, under-cambered flexible wing.


0.10


-0.10


-0.20


-0.30


-0.40


-0.50





















--A--- Experimental
-- AVL


A... 05


-5 0 5 10 15 20 25 30


Alpha


Figure 4-5. Drag vs. angle of attack comparison between experimental wind tunnel data and
simulated results from AVL for a thin, under-cambered flexible wing.







1.40


1.20

1.00

0.80


..---A. ...-.....-- ... -..


r ---A--- Experimental
5 0.60 A AVL

0.40

0.20

0.00 -
0. 0 0.10 0.20 0.30 0.40
-0.20
Cd


Figure 4-6. Lift vs. drag comparison between experimental wind tunnel data and simulated
results from AVL for a thin, under-cambered flexible wing.


0.40

0.35

0.30

0.25

0.20

0.15

0.10























Figure 4-7. Visual image correlation 3D scanning overlay of out of plane displacement on
painted mold (speckle pattern).


TT H


O-


7':~

-4,


hA~--


0.08

0.06

0.04

0.02

0

-0.02

-0.04

-0.06

-0.08


Figure 4-8. Comparison of wing model geometry and actual wing mold, where color represents
variation in out of plane position in millimeters (note the yellow spot at the center of
the root of the wing, indicating a purposeful gouge in the mold as a reference point).





































-60 -40 -20 0
Span (mm)


20 40 60 80


Figure 4-9. Comparison of MAVLAB wing model and machined wing mold cross section along
the wing span.


x 103


-4 I I I I
-80 -60 -40 -20 0
Span (mm)


20 40 60 80


Figure 4-10. Measured machining error along the span-wise cross section of the wing.


x 103


15 :


I........ MAVLAB model
Actual wing mold


I I I I I I I













0.05 i

......... MAVLAB model
-- Actual wing mold

-0.05


-0.1


S-0.15-


-0.2


-0.25 -


-0.3


-0.35
-100 -80 -60 -40 -20 0 20 40
Chord (mm)

Figure 4-11. Comparison of MAVLAB wing model and machined wing mold cross section
along the chord-wise direction (airfoil).


S103
-0.5










-2



-2.5 -
-3.


-100 -80 -60 -40 -20 0 20 40
Chord (mm)

Figure 4-12. Measured machining error along the chord-wise cross section of the wing.









CHAPTER 5
FUTURE WORK

MAVLAB has proven to be a useful aid in the design and analysis of MAVs. However, it

is a tool that is, by no means, complete. For instance, it cannot perform any automated

aerodynamic optimization. Also, with regard to the modeling of the aircraft aerodynamics, it

ignores the flexibility of the wing structure. In order to include the flexible model of the wing,

MAVLAB would require some type of finite element analysis (FEA). Some work has already

been done in this area and could be implemented fairly easily. Lastly, FEA and optimization

could be coupled together along with an entire aircraft model to obtain full aircraft dynamics.

Optimization

Matlab is a prime environment for numerical optimization routines because it can be

purchased with toolboxes designed with that in mind. This is an obvious advantage in that

optimization code does not necessarily have to be written from scratch. A good example of how

Matlab's optimization toolbox could be taken advantage of is the function FMINCON, which

tries to perform a constrained minimization of a nonlinear multivariate function. The aircraft

dynamics as a function of design parameters such as span, chord, camber, twist etc. are clearly

multivariate and likely highly non-linear in nature. If one wanted to find the maximum lift at a

given flight attitude by changing the airfoil camber, chord length and wing twist, FMINCON

could be used in conjunction with the output of AVL to converge on the solution fairly quickly.

Aeroelasticity

Aeroelasticity is the science which studies the interaction among inertial, elastic, and

aerodynamic forces. The construction of MAV wings developed and generally used by the

University of Florida were thin and flexible in nature. Under normal flight conditions and loads,

flexible wings tend to deform, thus introducing aeroelasticity. This fact is ignored when









performing aerodynamic analysis using AVL. One solution might be to perform an iterative

computational analysis by including FEA in the loop. Figure 5-1 depicts a diagram of the

iteration process.

Full Aircraft Modeling

As of the writing of this thesis, MAVLAB has the ability to send only wing geometry to

AVL for analysis. This is acceptable for analysis of just the wing, but often the case is more

complicated and requires the inclusion of the entire airframe. For example, one may need to

know how big and where to place a horizontal and vertical tail, if any is needed at all. AVL is an

excellent tool for doing this type of analysis, given a certain wing configuration, fuselage shape

and mass distribution. As of now, it is conceivable that one could modify the wing input file

MAVLAB generates for AVL, to include the rest of the airframe. This is a likely, but time

consuming step to perform more rigorous airframe analysis. A better and more long term

solution would be to incorporate code that allowed a user to select additional wing surfaces and

fuselage parts to be included in the automated generation of the AVL input file. Even further,

one could implement some level of optimization to increase wing and airframe efficiency.








AVL
Aerodynamic Analysis



*^ ^ ^^'- -S^


Matlab


FEA


Figure 5-1. Diagram of iterative aerodynamic analysis where the wing forces computed by AVL
are input into a FEA solver to determine the steady state deformation of a thin,
flexible wing during flight.


i


___~__ ~I
~~IL: I


J,









APPENDIX A
USERS' MANUAL

MAVLAB is a Matlab based toolbox designed to aid in the design and manufacturing of

MAVs. This section is not a manual for the Matlab programming environment, and assumes

the user has basic general knowledge of Matlab commands. For further information on

operating Matlab consult Mathworks website at http://www.mathworks.com/support/. This

users' manual is not intended to be comprehensive, but it does include a basic tutorial of

MAVLAB, an example usage of the program, as well as a list of important functions included in

the MAVLAB toolbox.

Basic Description of Features

Much of the basic description and important features of MAVLAB have already been

discussed in chapter 3. However, this section briefly outlines the major features of the toolbox.

MAVLAB offers a specific CAD suite designed for MAV wing creation, a plug-in for CFD

aerodynamic analysis using AVL (Athena Vortex Lattice, Drela) and a specialized CAM

software capable of exporting the wing geometry as a set of commands for CNC machining. The

CAD portion of the software allows full manipulation of a wing design through basic input

parameters (Table 3-1) and 3 vector parameters. The vector parameters describe the wing

planform, span-wise shape, and airfoil as a set of data points. The power of MAVLAB comes

into effect by allowing the user to either read the vector parameters in through saved ASCII text

files, or manipulate each through an interactive interpolating spline curve. Also, in order

maintain compatibility with other CAD software, the wing geometry can be exported to a

number of standard CAD formats. The aerodynamic analysis is performed by converting the

wing geometry to a format suitable for AVL and then calling a batch routine to run analysis. The

results are saved to a text file and finally read back in MAVLAB and printed to the screen. Once









the user has found the wing parameters that produce the desired results, MAVLAB can help

create a physical model of the wing. It does this by creating a set of instructions used by a CNC

machine to mill out a wing mold from stock material. The MAVLAB interface allows the user

to select from several different style toolpaths and other milling parameters. In addition, the user

can simulate the milling procedure to graphically verify the wing mold before it is created.

Getting Started with MAVLAB

MAVLAB is designed to be intuitive and easy to use graphical user interface (GUI).

Much of the underlying functionality is hidden to the user, who doesn't need to know how it

works. Most of the interaction with MAVLAB is prompted by icons and tool-tip descriptions of

what each function does. The main focus of this manual is the main MAVLAB interface. After

saving the main MAVLAB folder and subfolders to the search path, MATLAB can be called

from the command prompt by simply typing "mavlab" without quotes and in all lowercase

letters.



>> mavlab



This will start the main program (Figure A-i) and a window will pop up with a rather

uninteresting green square and some edit boxes and buttons. The green square is the default

shape for the wing, and assumes a rectangular planform with a wingspan of 1 and aspect ratio of

1. This, of course, is not ideal or reasonable for any practical aircraft design. Included with

MAVLAB are several test wing files. The example wing files are included in the main

MAVLAB directory under the "saves" folder. The user is encouraged to open one of these files

to see what MATLAB can do.









Sample Usage


Open and Saving Wing Files

Saved MAVLAB wing files can be found by clicking on the folder icon at the top of the

main window. This will open a dialog in which saved MAVLAB wing files with the extension

*.mat can be selected. Alternatively the user can load a wing file into MAVLAB at start up by

first typing "load wingfile.mat", where "wingfile.mat" is the name of the MAVLAB wing file.

This loads the wing file into the current workspace. Then the user can load the wing into

MAVLAB by typing "mavlab(wing)." In addition to loading existing files, MAVLAB can also

save them. To do this, simply click on the save button at the top of the MAVLAB main window.

A save dialog will appear to select a filename and destination.

Modify Some Parameters

MAVLAB offers a variety of parameters that can be manipulated and are explicitly

described in chapter 3. In order to modify the basic parameters simply type the value into the

corresponding field in the main MAVLAB window. The three vector parameters, planform,

airfoil and span shape can be modified by clicking on the respective images to the left of the

main MAVLAB window. For each, a new window will appear to modify or redefine each one.

This can be done by clicking on a the curve in the window and manipulating spline control

points, or simply selecting an ASCII text file of coordinates.

Export Wing Geometry

MAVLAB offers several useful and common industry formats in order to facilitate

exporting of the model. These include: IGES (International Graphic Exchange Specification),

STL, AutoCAD data exchange format (*.dxf), and Wavefront (*.obj) formats. This can be

performed by clicking on the export button at the top of the main MAVLAB window. A dialog

will then appear asking which format to save it in.









Run Athena Vortex Lattice Aerodynamic Analysis

Once the desired wing geometry has been achieved by manipulating the parameters in the

main MAVLAB window, aerodynamic analysis can be performed by running a plug-in called

AVL. To start AVL, simply click on the "AVL" button at the top of the main MAVLAB

window. A new window will appear (Figure A-2) with options and settings for AVL. The user

can select different parameters in the list box to the left. The parameters can then be modified by

typing in a numerical value or expression into the edit box to the right with the heading of the

current list box field. Multiple parameters can be set to the same value at once by making

multiple selections and entering the value. Default values are set for gravity and air density, and

are in metric units. In order to run AVL, at least velocity and angle of attack need to be set. In

addition to the list box, two other parameters need to be set. AVL divides the wing into discrete

chord-wise and span-wise sections. In the edit boxes near the bottom, these two corresponding

values can be modified. It appears that the default values are acceptable for most cases.

Create Computer Numerically Controlled Toolpath

The wing toolpath can be created relatively quickly in MAVLAB. To start the CNC

portion of MAVLAB, click on the button at the top of the window labeled "CNC." This will

open a new window with a graphical display of the wing and a dialog box asking for three

parameters (Figure A-3). These parameters refer specifically to machining parameters and are

the machine tool diameter, stepover and tolerance. The tool diameter can be any positive real

number, and will reflect the diameter of the tool going to be used for machining. The University

of Florida typically uses a V2 inch diameter tool. The stepover is a metric of the "roughness"

desired in the final wing mold. Experience has shown that a typical value of 0.050 inches was

acceptable for machining of wing molds in polyurethane tooling board. The final parameter,

called tolerance, reflects the relative resolution of the discretized toolpath. Since freeform curves









are approximated by piecewise linear segments in a toolpath, it is necessary to specify a value for

the tolerance depending on the complexity of the wing. If the tolerance is set too small, the

toolpath will become too large to send to the CNC computer.

Once the desired initial parameters are set, the user can select between two types of

toolpath shapes, zigzag and spiral (Figures A-5,6). Pressing the corresponding button at the top

of the window will generate the toolpath and display it with the wing. To verify that the toolpath

will turn out as desired, a feature has been included to graphically simulate the milling process.

By clicking on the "eye" button at the top of the menu, toolpath verification will initiate (Figure

A-6). Note that the dimensions of the minimum size work piece are displayed with the simulated

milled mold.

Finally, to machine the wing mold, the wing must be exported to the CNC machine.

MAVAB can do this by generating a text file of instructions for the CNC to follow called G-

code. This G-code is saved in a file called an NC file (*.nc) and can be created in MATLAB

after a toolpath has been generated. To generate a toolpath, click on the button at the top of the

CNC window with the "pencil and paper". A dialog will appear asking where to save the file.

List of Functions

Matlab code is generally composed of one or more files called m-files. Large programs,

such as MAVLAB, are typically broken down into tens and hundreds of m-files which each

perform a different function. Often, in Matlab, a collection of functions (or m-files) designed

to work together to perform similar tasks are termed a toolbox. These files will be grouped

together under one directory and given a toolbox name.

Main Functions

MAVLAB is more of a toolbox than a program. As such, it is composed of over 100

functions. The following are a summary of the main MAVLAB toolbox functions.









mavlab.m
dcAVL.fig
dcAVL.m
dcBuild.m
dcCNC.m
dcGetFoil.fig
dcGetFoil.m
dcGetPlan.fig
dcGetPlan.m
dcGetSpan.fig
dcGetSpan.m
dcLoadfoil.fig
dcLoadfoil.m
dcMain.fig
dcMain.m
dcSurf.m


Entrance function for MAVLAB called at command prompt
Supporting figure file for dcAVL.m
GUI for running AVL CFD
Main function for generating wing geometry from parameters
GUI for creating a CNC toolpath for airframe geometry
Supporting figure file for dcGetFoil.m
GUI for creating/modifying an airfoil
Supporting figure file for dcPlanFoil.m
GUI for creating/modifying the wing planform
Supporting figure file for dcGetSpan.m
GUI for creating/modifying the span-wise shape of a wing
Supporting figure file for dcLoadfoil.m
GUI for loading an airfoil into the database from a text file
Supporting figure file for dcMain.m
GUI main function for MAVLAB
Plot airframe geometry from parameter data


Sub-Functions

In addition to the main toolbox functions, which would generally be the most used, there

are dozens of helper functions. These helper sub-functions are not typically called on their own,

except for more advanced usage. The following is a list of the most useful advanced sub-

functions.


avl.m
mat2avl.m
nrbloft.m
nrbinterp.m
nrbWing.m
gnurbs.m
gspline2.m
scanfoil.m
mkWing.m
surf2spiral.m
surf2zigzag.m
surfature.m
verifytoolpath.m
offsetsurface.m
cncpost.m


Run AVL from wing geometry parameters
Convert surface data to *.avl file
Convert series of NURBS curves to a NURBS surface
Interpolate a NURBS curve to a set of data points
Create a NURBS surface from wing parameters
Interactively manipulate a NURBS curve/surface
Interactivley draw/edit a cubic spline curve
Low level reading of airfoil coordinate text files
Generate the wing parameter structure
Generate a spiral toolpath from a surface
Generate a zigzag toolpath from a surface
Compute Gaussian/mean surface curvature
Simulate milling process of a toolpath
Create offset/parallel surface for use with toolpath creation
Write a toolpath *.nc file from toolpath coordinates









Not all MAVLAB toolbox functions are listed above; just the ones that are of most interest

to the user. Included in MAVLAB are several sub-toolboxes, including a NURBS toolbox as

well as some needed functions from the SPLINE toolbox. Also included are two toolboxes not

currently being used by MAVLAB but helpful with regard to future work and additional

implementations of the MAVLAB toolbox. Those two toolboxes are a FEA toolbox specialized

for MAV wings and a MESH toolbox designed to help mesh wings for FEA. Neither of these

toolboxes are complete nor come with any guarantee. They are included only to help those

doing future work with MAVLAB.













- .. ..


- (Click to edit)


- Span-wise Shape


- Parameters

Span 24 1 'UQ]

Chord 7 j

Camber 006 i

Z-Scale 0.25

Twist 0 'A ]

Sweep 0-- 7

Dihedral 0 ~



[ Mirror Wing
j Max Camber Z-Reference


Figure A-1. MAVLAB Graphical User Interface.






Furn Pzrairi.ti-i- -


beta = 0
pb/2V = 0
qc/2V = O
rb/2V = 0
CL = 0
CDo = 0
bank = 0
elevation = 0
heading = 0
Hach = 0
velocity = 0
density = 1.225
grav.ace = 9.81
turn rad. = 0
loadfac = 0
X_cg = 0
Y_cg = 0
Z_cg = 0
mass = 0
Ixx = 0
Iyy = 0
Izz = 0
Ixy = 0
Iyz = 0
Izx = 0





Figure A-2. AVL Graphical User Interface.


- 3D View


- alpha





Name




Panel Layout

Nspan 1: Nchord 8



Run










pImqI 11


Bl E )


Figure A-3. CNC Toolpath generator GUI.


I I J. ( n -)1f-








....................................................... ....... .............................. .............................. .............................. ............................... .............................. .............................. ............................................................. ........................................ ...



















Figure A-4. Zigzag toolpath example.

rr
. ............ ............. .......................... .
FLT L.I" "i J UiiCEAj, t


Figure A-5. Spiral toolpatt example.











F E J- '. j .. r 1 1 I f. ...:. i
D Q3;^ 610 4^~TlD IHl 0a 0 -11U"


gure A-o. verricaton or milling toolpatn via simulation (note inat me minimum dimension or
the work piece was defined in the rendering).









LIST OF REFERENCES


1 Ifju, P.G., Ettinger, S., Jenkins, D.A., & Martinez, L., "Composite Materials for Micro Air
Vehicles," Proceedingfor the SAMPE Annual Conference, SAMPE Journal, Vol. 37,
2001, pp. 7-12.

2 Mueller, T., "Fixed and Flapping Wing Aerodynamics for Micro Air Vehicle
Applications," Proceedings of the Conference on Fixed, Flapping and Rotary Wing
Vehicles at Very Low Reynolds Numbers, Vol. 195, 2000, pp. 307-339.

3 Ifju, P.G., Jenkins, D.A., Ettinger, S., Lian, Y., Shyy, W., & Waszak, M.R., "Flexible-
Wing-Based Micro Air Vehicles," AIAA Annual Conference, AIAA Paper No. 2002-
0705, January 2002.

4 Jenkins, D.A., Ifju, P.G., Abdulrahim, M., & Olipra, S., "Assessment of the Controllability
of Micro Air Vehicles," Micro Air Vehicle Conference, Bristol, England, April 2001.

5 Ettinger, S.M., Nechyba, M.C., Ifju, P.G., & Waszak, M., "Vision-Guided Flight Stability
and Control for Micro Air Vehicles," Proceedings of IEEE International Conference on
Intelligent Robots and Systems, Vol. 3, 2002, pp. 2134-40.

6. Albertani, R., Stanford, B., Hubner, J., Lind, R., Ifju, P., "Experimental Analysis of
Deformation for Flexible-Wing Micro Air Vehicles,"AIAA Structures, Structural
Dynamics, andMaterials Conference, AIAA Paper No. 2005-2231, April 2005.

7. Albertani, R., Hubner, J.P., Ifju, P., Lind, R., and Jackowski, J., "Experimental
Aerodynamics of Micro Air Vehicles," SAE World Aviation Congress and Exhibition,
Paper No. 04AER-8, November 2004.

8 Lian, Y., Shyy, W., Ifju, P.G., "Membrane Wing Model for Micro Air Vehicles," AIAA
Journal, Vol. 41, No. 12, 2003, pp. 2492-94.

9 Fleming, G. A., & Burner, A. W., "Deformation Measurements of Smart Aerodynamic
Surfaces," 44th Annual SPIE International Symposium on Optical Science, Engineering,
and Instrumentation Optical Diagnostics for Fluids/Heat/Combustion and
Photomechanics for Solids, SPIE Paper No. 3783-25, July 1999.

10 Fleming, G. A., Bartram, S. M. Waszak, M. R., Jenkins, L. N., "Projection Moire'
Interferometry Measurements of Micro Air Vehicle Wings," SPIE Paper No. 4448-16,
November 2001.

11 Waszak, M. R., Jenkins, L. N., & Ifju, P. G., "Stability and Control Properties of an
Aeroelastic Fixed Wing Micro Aerial Vehicle," AIAA Paper No. 2001-4005, August 2001.

12. Grasmeyer, J. Keenon, M. "Development of the Black Widow Micro Air Vehicle", AIAA
Paper No. 2001-0127, January 2001.









13. Margason, R.J., and Lamar, J.E., "Vortex-Lattice FORTRAN Program for Estimating
Subsonic Aerodynamic Characteristics of Complex Planforms," NASA Paper No. TN D-
6142, 1971.

14. Chua C. K. and Leong K. F. Rapid Prototyping: Principles & Applications in
Manufacturing, John Wiley & Sons, Singapore, 1997.

15. Anderson, J. D., Aircraft Performance andDesign, WCB/McGraw-Hill, Boston,
MA,1998.

16. Mueller, T.R., Kellog, J.C., Ifju, P.G., Shkarayev, S.V., Introduction to the Design of
Fixed-WingMicro Air Vehicles, AIAA, Reston, VA, 2006, pp. 185-240.

17. University of Florida Micro Air Vehicle Team "University of Florida Competition Micro
Air Vehicles," 6th Annual International Micro Air Vehicle Competition, Brigham Young
University, Provo, Utah, May 2002.

18. M. Abdulrahim, R. Albertani, P. Bamswell, F. Boria, D. Claxton, J. Clifton, J. Cocquyt, K.
Lee, S Mitryk, and P. Ifju, "Design of the University of Florida Surveillance and
Endurance Micro Air Vehicles," 7th Annual Micro Air Vehicle Competition Entry,
University of Florida, Gainesville, FL, April 2003.

19. R. Albertani, P. Barnswell, F. Boria, D. Claxton, J. Clifton, J. Cocquyt, A. Crespo, C.
Francis, P. Ifju, B. Johnson, S. Jung, K. Lee, and M. Morton, "University of Florida
Biologically Inspired Micro Air Vehicle," 8th Annual Micro Air Vehicle Competition
Entry, University of Arizona, Tucson, AZ, April 2004.

20. Albertani, R., Boria, F., Bowman, S., Claxton, D., Crespo, A., Francis, C., Ifju, P.,
Johnson, B., Lee, K., Morton, M., Sytsma, M., "Development of Reliable and Mission
Capable Micro Air Vehicles," 9th Annual International Micro Air Vehicle Competition,
Konkuk University, Seoul, South Korea, May 2005.

21. Claxton, D., Johnson, B., Stanford, B., Sytsma, M., "Development of a Composite
Bendable-Wing Micro Air Vehicle," 10th Annual International Micro Air Vehicle
Competition, Brigham Young University, Provo, Utah, May 2006.

22. Sytsma, M., "Aerodynamic Flow Characterization of Micro Air Vehicles Utilizing Flow
Visualization Methods." Masters Thesis, University of Florida, Gainesville, FL, August
2006.

23. Albertani, R., "Experimental Aerodynamic and Static Elastic Deformation
Characterization of Low Aspect Ratio Flexible Fixed Wings Applied to Micro Aerial
Vehicles," Ph.D. Dissertation, Department of Mechanical and Aerospace Engineering,
University of Florida, Gainesville, Fl, December 2005.

24. Albertani, R., Hubner, P., Ifju, P., Lind, R., "Wind Tunnel Testing of Micro Air Vehicles
at Low Reynolds Numbers," SAE World Conference, SAE Paper No. 2004-01-3090,
November 2004.









25. Drela, M., Youngren, H., Software: AVL (Athena Vortex Lattice Model), Version 3.26,
http://web.mit.edu/drela/Public/web/avl, Mark Drela, Cambridge, MA, February 2007.

26. Helm, J. D., McNeill, S. R. and Sutton, M. A., "Improved 3-D Image Correlation for
Surface Displacement Measurement", Optical Engineering, Vol. 35(7), 1996, pp. 1911-
1920.









BIOGRAPHICAL SKETCH

Daniel J. Claxton was born to Linda and Erny Claxton in the small town of Orange

Park, Florida. He grew up in Orlando where he attended school through high school,

graduating from The First Academy. At an early age it was clear that he had an affinity

for engineering as he enjoyed taking his toys apart to see how they worked, and could

often be found constructing "inventions" and building gadgets and gizmos. As a 1st

grader, he aspired to be an inventor and by the time he was in high school it was clear

that his interest and proficiency in math and the sciences would lead him to pursue a

college education in some engineering discipline. After trying his hand at civil and

industrial engineering, he found his niche in the mechanical and aerospace engineering

program at the University of Florida.

The small college town of Gainesville became Daniel's home while he earned his

bachelor and master degrees. During that time he became involved in Campus Crusade

for Christ and made dozens of life-long friends. It was also about this time that he

became involved in the Micro Air Vehicle program at the University of Florida. He

became a part of an "elite" group of researchers looking to explore the realm and

capabilities of small aircraft. He was a part of the team for five years, and was proud to

take home five International Micro Air Vehicle Competition championships. Courtesy of

UF, he had the privilege to travel all over the world, including Eglin Air force Base (FL),

Utah, Arizona, South Korea and Germany. During the completion of his master's degree

he married the love of his life, and accepted an engineering position with United Space

Alliance in Huntsville Alabama




Full Text

PAGE 1

1 DEVELOPMENT OF A PARAMETRIC SOFTWARE TOOL FOR THE DESIGN AND MANUFACTURING OF MICRO AIR VEHICLES By DANIEL J. CLAXTON A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Daniel J. Claxton

PAGE 3

3 To my beautiful wife, Kristin.

PAGE 4

4 ACKNOWLEDGMENTS I thank everyone who helped and encouraged m e on this project. First and foremost, I thank my advisor, Dr. Ifju, for supporting my efforts and giving me the opportunity to prove myself. I give special thanks to Mike Braddoc k and all of they guys in the machine shop, who spent hours and hours helping me machine molds. Thanks to all of the guys in the MAV Lab. They all made my job more interesting and fun. Special thanks go to Bret Stanford and Frank Boria for providing me with advice, inspiration, creative ideas and much-needed laughter. I thank Kristin for being patient with me for all these years and encouragi ng me to push on, but do it quickly. Last, but not least, I thank my parents who supported me financially and ot herwise through all of my education. They never stopped believing in me.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................11 CHAP TER 1 INTRODUCTION..................................................................................................................13 Micro Air Vehicles at a Glance.............................................................................................. 13 Mircro Air Vehicles at the Research Level ............................................................................ 13 Motivation and Overview....................................................................................................... 15 2 LITERATURE REVIEW.......................................................................................................19 Micro Air Vehicle Design......................................................................................................19 Computation Fluid Dynamics................................................................................................. 19 Manufacturing and Ra pid Prototyping ................................................................................... 20 3 DESIGN PROCESS............................................................................................................... 22 MAVLAB...............................................................................................................................22 Scalar-Based Design Parameters..................................................................................... 23 Vector-Based Design Parameters.................................................................................... 25 Aerodynamic Analysis........................................................................................................... .28 Computer Numerically C ontrolled Machining ....................................................................... 28 Composite Construction.........................................................................................................30 Flexible Wing Fabrication...............................................................................................31 Rigid Wing Fabrication................................................................................................... 32 Fuselage Fabrication........................................................................................................ 32 Vacuum Bagging and Curing................................................................................................. 33 Fuselage Assembly.................................................................................................................33 Components Installation........................................................................................................ .34 4 REPRESENTATIVE STUDY................................................................................................ 51 Computer Model Description................................................................................................. 51 Wing Aerodynamics Comparison........................................................................................... 51 Wind Tunnel Results....................................................................................................... 52 Athena Vortex Lattice Results......................................................................................... 53 Wing Mold Accuracy............................................................................................................. 54

PAGE 6

6 Conclusion..............................................................................................................................55 5 FUTURE WORK.................................................................................................................... 62 Optimization................................................................................................................... ........62 Aeroelasticity..........................................................................................................................62 Full Aircraft Modeling......................................................................................................... ...63 APPENDIX USERS MANUAL................................................................................................ 65 Basic Description of Features.................................................................................................65 Getting Started with MAVLAB..............................................................................................66 Example Usage.......................................................................................................................67 Open and Saving Wing Files........................................................................................... 67 Modify Some Parameters................................................................................................67 Export Wing Geometry................................................................................................... 67 Run Athena Vortex Lattice Aerodynamic Analysis........................................................ 68 Create Computer Numerically Controlled Toolpath....................................................... 68 List of Functions.....................................................................................................................69 Main Functions................................................................................................................69 Sub-Functions..................................................................................................................70 LIST OF REFERENCES...............................................................................................................76 BIOGRAPHICAL SKETCH.........................................................................................................79

PAGE 7

7 LIST OF TABLES Table page 3-1 List of basic M AVLAB wing param eters............................................................................... 35

PAGE 8

8 LIST OF FIGURES Figure page 1-1 Typical composite airframe MAV deve loped at the University of Florida. ........................... 17 1-2 Dependency of aircraft aerodynamic perfor mance on scale, where performance is represented by maximum lift/drag and scale is represented by the non-dimensional quantity Re.........................................................................................................................17 1-3 Example of a thin, under-cambered, fl exible m embrane Micro Air Vehicle wing................18 1-4 Relative scale and payload capacity of Micro Air Vehicles and Sm all Unmanned Air Vehicles..............................................................................................................................18 2-1 Black Widow Micro Air Vehicle devel oped by Aerovironment using multi-disciplinary optimization.......................................................................................................................21 3-1 MAVLAB Graphical User Interface......................................................................................36 3-2 Typical graphical display of a wing geom etry in MAVLAB, with fe edback for basic parameters such as planform area, mean chord, aerodynamic center and aspect ratio...... 36 3-3 Wing span as defined by the linear distance from wing tip to wing tip.................................37 3-4 Wing root chord is define d as the distance from leading to trailing edge of the wing, closest to the fuselage, or middle of the span.................................................................... 37 3-5 The maximum camber is the ratio of the hig hest point of the camber line to the airfoil chord length.......................................................................................................................38 3-6 The geometric dihedral angle is the angle the wing makes w ith the horizontal..................... 38 3-7 The sweep angle is the angle between the chord line (dotted) and the line perpendicular to the free stream velocity.......................................................................... 38 3-8 Geometric wing twist, often called washout when it is negative and washin when it is positiv e....................................................................................................................... ........39 3-9 Typical example of CNC milling of a MAV wing m old out of resin-based tooling board..................................................................................................................................40 3-10 Scalloping or cusps created by consecutive passes of a b all (radial) end mill..................... 40 3-11 Typical wing mold machined from high-density polyurethane tooling board. .................... 41 3-12 Typical ball end mill used for machini ng 3D freeform surfaces such as wing molds......... 41 3-13 A) Rigid carbon fiber wing. B) flexible m embrane wing with batten reinforcement......... 41

PAGE 9

9 3-14 Drawing layout pattern on the wing mold............................................................................ 42 3-15 Applying Teflon release film to the surface of the wing mold using spray glue ................. 42 3-16 Laying up of carbon fiber on the wing mold........................................................................43 3-17 Application of Kevlar to the leading edge. ........................................................................... 43 3-18 Cutting of unidirectional carbon fiber strips with custom cutting tool................................ 44 3-19 Trimming of excess carbon fiber during layup. ................................................................... 44 3-20 Trimming excess carbon fi ber f rom wing after cure............................................................ 45 3-21 Application of Teflon release film to fuselage mold............................................................ 45 3-22 Fuselage placed on a form to maintain proper mate with wing............................................ 46 3-23 Application of 1-2 layers of carbon fiber to the fuselage mold. ........................................... 46 3-24 Vacuum bagging of wing tool..............................................................................................47 3-25 Vacuum bagging of fuselage covered with peal-ply release film ........................................ 47 3-26 Assembly of wing and fuselage after cure............................................................................ 48 3-27 Masking prep and application of spray glue to top of wing for latex application. ...............48 3-28 Application of latex m embrane to top of wing..................................................................... 49 3-29 After latex trimming, final gluing of latex to carbon fiber skeleton of the wing. ................ 49 3-30 Component installation, with servos lashed to inside of fuselage. .......................................50 3-31 Tyvek hinge rudder and attachment of cont rol rod to control surfaces................................50 4-1 Thin, flexible wing model used to com pare wind tunnel aerodynamics with Athena Vortex Latices computationa l aerodynamics approximation........................................... 56 4-2 Thin, flexible batten reinforced MAV wi ng with Icarex used in wind tunnel testing. ........... 56 4-3 Lift vs. angle of attack comparison between experim ental wind tunnel data and simulated results from AVL for a thin, under-cambered flexible wing............................. 57 4-4 Pitching moment vs. angle of attack co m parison between experimental wind tunnel data and simulated results from AVL for a thin, under-cambered flexible wing...................... 57 4-5 Drag vs. angle of attack comparison between experim ental wind tunnel data and simulated results from AVL for a thin, under-cambered flexible wing............................. 58

PAGE 10

10 4-6 Lift vs. drag comparison between experime ntal wind tunnel data and sim ulated results from AVL for a thin, unde r-cambered flexible wing......................................................... 58 4-7 Visual image correlation 3D scanning overl ay of out of plane displacem ent on painted mold (speckle pattern)........................................................................................................59 4-8 Comparison of wing model geometry and actual wing mold, where color represents varia tion in out of plane position in millimeter s (note the yellow spot at the center of the root of the wing, indicating a purposeful gouge in the mold as a reference point)...... 59 4-9 Comparison of MAVLAB wing model and m achined wing mold cross section along the wing span...........................................................................................................................60 4-10 Measured machining error along th e span-wise cross section of the wing. ......................... 60 4-11 Comparison of MAVLAB wing model a nd m achined wing mold cross section along the chord-wise direction (airfoil)....................................................................................... 61 4-12 Measured machining error along the chord-wise cross section of the wing. ....................... 61 5-1 Diagram of iterative aerodyna m ic analysis where the wing forces computed by AVL are input into a FEA solver to determine the st eady state deformation of a thin, flexible wing during flight............................................................................................................. .64 A-1 MAVLAB Graphical User Interface..................................................................................... 72 A-2 AVL Graphical User Interface.............................................................................................. .72 A-3 CNC Toolpath generator GUI...............................................................................................73 A-4 Zigzag toolpath example................................................................................................... .....74 A-5 Spiral toolpath example.........................................................................................................74 A-6 Verification of milling toolpath via simu lation (note that the m inimum dimension of the work piece was defined in the rendering).................................................................... 75

PAGE 11

11 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DEVELOPMENT OF A PARAMETRIC SOFTWARE TOOL FOR THE DESIGN AND MANUFACTURING OF MICRO AIR VEHICLES By Daniel J. Claxton May 2007 Chair: Peter Ifju Major: Aerospace Engineering Micro air vehicles (MAVs) are a special s ubset of unmanned air vehicles (UAVs) that warrant a significant level of scientific interest. In general, MAVs are small, inexpensive and often expendable platforms, flown by remote pilot, or autopilot. Because they maybe flown by autonomous control or inexperienced pilots, they must have very reliable and benign flight characteristics built in to their design. The University of Florida ha s developed a series of MAVs that adopt a flexible-wing concept, most notably featuring a carbon fiber structure and a thin extensible membrane skin. Because their design requirements mandate that they perform reliably in flight, careful thought and consideration must go into a MAV design. So me of the design may come from intuition and experience, but it must ultimately be verified th rough quantitative testing. In addition, the design process must be performed in a way that is accurate and repeatable. The purpose of my research was to devel op an efficient and accurate methodology for designing, producing and reproducing MAVs. My approach evolved into a rapid prototyping process of designing and manufacturing MAVs while still maintaining geometric accuracy and aerodynamic integrity. The solution was the devel opment of a software-based design tool, called MAVLAB, which incorporated specialized CAD design features, aerodynamic analysis tools and

PAGE 12

12 rapid manufacturing through automated machining. My thesis includes an overview of the University of Floridas design procedure, an example case study and a users manual for MAVLAB.

PAGE 13

13 CHAPTER 1 INTRODUCTION Micro Air Vehicles at a Glance Micro air vehicles (MA Vs), as defined by De fense Advanced Research Programs Agency (DARPA), are a subset of aircraft with a ma ximum wing span of 15 cm (about 6 inches) [1] (Figure 1-1) and a maximu m flight speed no greater than 15 m/s [2]. This definition is important because it corresponds to a regime of aircraft operating in a low Reynolds number (Re) region, where aerodynamic behavior becomes more difficult to predict. This type of aircraft is of particular interest to researchers in the scientif ic community as well as real world applications. Small aircraft of this size can be useful for several reasons. They can be made relatively inexpensive, expendable, portable, and undetectab le, which lend well to military and surveillance type applications. Much of the scientific interest in designing such vehicles comes from the challenge of overcoming the loss of aerodynamic effi ciency due to scale, as shown in Figure 1-2. Mircro Air Vehicles at the Research Level The challen ges of designing an efficient, controllable MAV are what prompted the University of Floridas involvement in MAV research. In order to overcome some of these aerodynamic and technical challenges, the University of Florida developed a series of MAVs that incorporated a unique, thin, undercambered, flexible wing design (F igure 1-3) [1,3-6]. Studies have shown that thin wings have a distinct a dvantage over thicker, volum etric wings at the low Reynolds numbers where MAVs typically operate [3,7-8]. In general, the wings of these vehicles are constructed of a carbon fiber skeleton and a thin flexible membrane, originally inspired by the structure of bat wings and wind-surfing sails.

PAGE 14

14 The University of Florida has been taking an active role in the i nvestigation of MAVs, researching and developing new MAV concepts for th e past 9 years. In that time, the university has come a long way in terms of development a nd testing procedure, while the main aircraft design philosophy has remained relatively the same. The basic philosophy of MAV research at the Un iversity of Florida initially evolved from a design-build-fly approach with the main objective of creating the smallest possible flying fixed wing aircraft. This was spawned by research gran ts with this goal in mind, and further inspired by the creation of the Internati onal Micro Air Vehicle Competition. Over time, and after much success, the Universitys focus shifted away from the concept of producing the smallest flying aircraft in the MAV regime, but rather the smallest, mission-capable aircraft. This was influenced mainly from the prac tical standpoint that many research funding institutions in this field were defense contractors that wanted MAVs and small UAVs that were useful and mission capable. This meant that they needed to be capab le of flying missions that required longer flight times, more payload capacity and ma intained a greater degree of au tonomy. In order to perform these tasks, a greater understa nding of flight dynamics was n ecessary. Figure 1-4 depicts the relative trend in payload capacity for the MAV and small UAV scale. This need to better characterize and understa nd small aircraft benefited MAV research at the University of Florida. Eventually, it lead to the much-needed replace ment of an aging opencircuit low speed wind tunnel with a new closed circuit low speed wind tunnel. With this addition, better analysis could be performed on MAVs through quantitative loads measurements. In addition, a visual image correl ation (VIC) system wa s purchased and used in conjunction with the wind-tunnel to measure in-flight deformation of the thin, flexible wings. All together, these

PAGE 15

15 advances in technology helped to better charact erize and understand M AVs from quantitative testing. There was one part of the puzzle still missing. While there was some understanding as to what a MAV design should look like, there was no official agreement on what guidelines the most efficient design should uphold. This left re searchers to build and test hundreds of designs and try to form a relationship between de sign parameters and aerodynamic performance. Motivation and Overview Most of the University of Floridas MAV designs have revolved around the same central design philosophy, with some variation on geometric design parameters, depending on the desired performance characteristics and design requirements. However, even small changes to any given design can drastically enhance or degrade aerodynamic performance. This is especially true at these small scales. Further, any unintended deviation from the desired design geometry would introduce unwanted effects. Thus a method of accurately designing, analyzing and manufacturing MAVs was necessary. In order to produce MAVs accurately, a proce ss of forming composites over a machined mold was initially developed. The constructi on of these molds can be described from the following process. The airframe was first modeled in a 3D computer aided design (CAD) program. The surface of the ai rframe was exported to computer aided machining (CAM) software. The CAM software could then be used to generate a set of machine instructions for a computer numerically controlled (CNC) to interpret and automate the milling of a mold. The mold is then used in the production of the carbon fiber airframe structure. Depending on the size and complexity of the airframe and the skill le vel of the designer, th is process could take anywhere from days to weeks to trans ition from concept to working model.

PAGE 16

16 Wind-tunnel and flight testing are the preferre d choice when it comes to analysis of new MAV concepts and prototypes. Work has also been done to experi mentally quantify wing deformation using CFD [8], projection moir in terferometry [910] and photogrammetry [11]. None of these techniques were practical or possi ble before manufacturing. An intermediate and iterative method was needed to analyze aerodynam ics and aid in MAV de sign before production. In order to streamline the MAV design proce ss, a parametric software design tool was developed. The idea was to have a simple interface that would allow a user to transform a "back of the napkin" design to a CAD model in a matte r of minutes. In addition the software would allow the user the ability to adjust a few design parameters which controlled the entire geometry. Further, the aforementioned crude comparativ e CFD approach could be implemented in parallel to obtain relative perf ormance of each design parameter. This approach was not created to produce the exact optimum aircra ft design with the press of a button, but rather a design tool that would benefit an experienced MAV/aircraft designer. The e nd result of this tool provided the user with a computer-simulated environm ent, called MAVLAB, in which the optimum MAV configuration could evolve.

PAGE 17

17 Figure 1-1. Typical composite airframe MAV developed at the University of Florida. Region of Interest10 3 10210010110310410510610 7 CLCLCLCD Max Rec U c v = All Rigid, Smooth Airfoils Figure 1-2. Dependency of airc raft aerodynamic performance on scale, where performance is represented by maximum lift/drag and scale is represented by the non-dimensional quantity Re.

PAGE 18

18 Figure 1-3. Example of a thin, under-cambered, flexible membrane Micro Air Vehicle wing. Figure 1-4. Relative scale and pa yload capacity of Micro Air Ve hicles and Small Unmanned Air Vehicles.

PAGE 19

19 CHAPTER 2 LITERATURE REVIEW Micro Air Vehicle Design The history of MAV des ign is a relatively short one, as th e definition of MAVs has only been around for just over a decade. In 1996 DAR PA announced the Micro Air Vehicle Program initiative, seeking to develop and test emerging technologies that coul d evolve into a mission capable flight system for military surveilla nce and reconnaissance applications. The only requirement was that the dimension of the vehi cle should not exceed 15 cm. There were no other restrictions on the design [1]. The DARPA initiative was what spurred th e development of MAV designs in both industry and academia. One of the first successf ul and practical MAVs was the Black Widow, developed by Aerovironment (Figure 2-1). The design of the black widow was performed using a multi-disciplinary optimization (MDO) approach [12]. In this way, the aircraft subsystems such as propulsion, aerodynamics and airframe stru cture could be optimized to fit the design size requirement. Aerovironment chose to use CAD modeling and a computer software environment to design and evolve the vehicle. The results of Aerovironments analys is and design of the Black Widow MAV proved that a small, lightwei ght, mission capable, battery powered platform could be remotely operated up to 1.8 km away and sustain flight times of up to 30 minutes. This impressive performance would later prove to in spire and influence future MAV work, including that at the University of Florida. Computation Fluid Dynamics There are several m ethods for performing computational analysis of aerodynamics, all of which try to model single phase fluid flow. The most advanc ed method, commonly referred to as full blown CFD uses the Navier-Stokes equati ons to determine the behavior of a fluid over a

PAGE 20

20 body. Another approach is known as the panel met hod and attempts to solve linearized potential flow equations. There is also another method that is similar to the panel method which is very easy to use and is capable of providing remarkable insight into wing aerodynamics and component interaction called th e vortex lattice method (VLM). It was among the earliest methods utilizing computers to actually a ssist aerodynamicists in estimating aircraft aerodynamics. Vortex lattice methods are based on solutions to Laplaces Equation, and are subject to the same basic theoretic al restrictions that apply to panel methods [13]. Today, there are dozens of free VLM software codes available for use in aerodynamic analysis, most notably, Richard Epplers Profil, and AVL, by Mark Drela and Harold Youngren. Due to better availability and documentation, AVL was chosen as the core aerodynamics analysis tool for MAVLAB. Manufacturing and Rapid Prototyping Rapid Prototyping (RP) can be defined as a gr oup of techniques used to quickly fabricate a scale m odel of a part or assembly using three-dimensional CAD data. Fused Deposition Modeling (FDM) Stereolithography (SLA) Multi Jet Modeling (MJM) Electron Beam Melting (EBM) and 3D Printing (3DP) were all comm on examples of ra pid prototyping techniques [14]. While these methods of rapid prototyping were potentially viable processes for producing wing molds, CNC machining was chosen in stead for the purposes of this research. CNC refers specifically to a computer "contr oller" that reads ASCII text f ile instructions and drives the machine tool, a powered mechani cal device typically used to fa bricate metal components by the selective removal of metal or other material. CNC does numerically directed interpolation of a cutting tool in the work envelope of a machine. The University of Florida used a basic 3-axis end mill with a radial cutting t ool to machine 3D freeform surf aces out of polyurethane tooling board.

PAGE 21

21 Figure 2-1. Black Widow Micr o Air Vehicle developed by Aerovironment using multidisciplinary optimization.

PAGE 22

22 CHAPTER 3 DESIGN PROCESS MAVLAB The m igration of an initial MAV design concep t to a working model can be a complicated and time consuming procedure. Before a MAV c ould be built and tested at the University of Florida, it was first created in a computer simu lated environment. Solid Works, Pro Engineer, and Autodesk Inventor were the three major computer aided design (CAD) software packages used by academia and industry at the time of this writing. These software packages were all well documented and fully capable of creating parame tric MAV designs. However, they had some drawbacks. They were expensive, bulky, and requ ired a steep learning curve. In addition, they were not specialized to the University of Florid as approach of designing of aircraft. In an attempt to increase productivity, aid design and reduce workload, a specialized Matlab-based software design tool was deve loped, and named MAVLAB. Th is program was specifically designed to translate basic, intuitive design parameters into a complex freeform airframe model. Figure. 3-1 depicts a typical wing geometry in MAVLAB. In addition, MAVLAB could perform aerodynamic analysis via a CFD plug-in. It also ha s the ability to export the geometry to other CAD software. Ultimately, MAVLABs most powerf ul feature was the ease in which it aided in manufacturing. This process, ca lled computer aided manufactur ing (CAM) was what made the University of Florida successful in verifyi ng MAV designs. MAVLAB effortlessly converted wing geometry into a computer num erically controlled (CNC) mach ine instructions, called a tool path, used for automated machining and prototypi ng of molds. These molds would then go on to be used in the fabricatio n of composite MAV wings. The concept behind the CAD portion of MAVLAB was to produce an intuitive parametric based model. The 7 basic parameters used to control the geometry are self explanatory in nature,

PAGE 23

23 and relate to specific geometri c descriptions of a generali zed wing, rather than ambiguous dimensions that have no direct re lationship to aircraft dynamics. These parameters are listed in Table 3-1 and are explained later in the chapter. Scalar-Based Design Parameters MAVLAB featured a simple GUI (Figure 31) designed to guide the user through the construction of a wing via sim ple, scalar valu ed, geometric parameters. These parameters included wing span, root chord, span scale, dihe dral angle, sweep angle and geometric twist angle, which were chosen because they ha d a direct correlation with wing aerodynamics. Additionally, there was one parameter included fo r aesthetics and structur al reasons, called the leading edge reference. Each input could be changed by typing the value in a corresponding field causing a 3D view of the wing to regenerate instantaneously in a neighboring window. Additionally, parameters could be modified by clicking and dragging on an icon next to each parameter. The user could then interact with the model by rotating and changing the viewpoint with the mouse (Figure 3-2) to verify the ge ometry from different angles, in real time. Wing span The wing span is a simple feature ofte n used to describe the length of a wing. The wing span is measured from the left wing tip to the right wing tip in a straight line, and is commonly denoted by the character b. For clarif ication, Figure 3-3 illustrates the wing span parameter on a generic wing however this defi nition also applies to MAV wings as well. Root chord. In general, a chord refers to the length of an airfoil cross-section of a wing. This is the linear distance from the leading edge of the wing to the trailing edge (Figure 3-4). The root chord is the chord at the root of the wi ng; the point closest to the fuselage or middle of the wing span. This parameter was used as a reference to scale the entire chord of the wing. Airfoil camber. An airfoil is the cross section of the wing, whose shape is so important to aerodynamics that it alone is often a good indica tor of wing performance (Figure 3-5). In

PAGE 24

24 general, airfoils are composed of a top and a bottom curve, creating a closed region. The mean of those two curves is called the camber line. The distance between the top and bottom curves is called the thickness distribution. Th e thin airfoils, characteristic of the University of Floridas MAV wings were a simplification of this definition, with a thickness distribution of zero. It was found that at the MAV scale, aerodynamics could benefit from thinne r airfoils, and thus MAVLAB was designed around this ph ilosophy. The parameter that controls airfoil camber is called max camber, or simply camber, and is the ratio between the maximum value of camber and the length of the airf oil chord. This was beneficial to the user because it allowed modification to the current airfoil coordinates selected for the wing. Span scale. The span scale refers to a modifier parameter for the span shape, mentioned later in the vectorized pa rameter section. The span scale allows the user to increase or decrease the magnitude of the span shape by a scale fact or, usually between 0 and 1. Negative numbers inverted the span shape. Dihedral angle. In geometry, the dihedral angle is defined as the angle between the intersection of two planar surfaces. For aircraft, this is co mmonly seen as the angle between the horizontal (ground) and the wing. Th e dihedral of a wing is a specifi c form of span shape, and is defined separately in MAVLAB because it is so commonly used. Figur e 3-6 illustrates the dihedral angle parameter. Sweep angle. Wing sweep is another common geomet ric feature found in aircraft wings, shown in figure 3-7. It is defined as the angle between the chord line and the line perpendicular to the chord. The aerodynamic significance of wing sweep can provide both the same effect as the dihedral angle, while moving the aerodynamic center (AC) further aft. This is

PAGE 25

25 useful for flying wings and MAVs, where lack of volume renders it difficult to get the center of gravity (CG) in front of the AC for stability. Twist angle. The geometric twist angle (shown in figure 3-8) is the span-wise change in airfoil angle of incidence, usually measured in degrees with respect to the root airfoil. This is the parameter that can be controlled in MAVLAB and differs from aerodynamic twist, which can provide the same results by a span-wise change in airfoil shape. Most aircraft have several degrees of twist in the wing to pr event stall from occurring all at on ce. This is generally called washout and refers to a negative twist angle. In MAVLAB, the direction of the twist angle is reversed. Leading edge reference. The leading edge reference or edge ref is a parameter that differs from all the rest, in that it is not clos ely coupled with aerodynamic performance. When a wing of arbitrary shape is genera ted in MAVLAB, an unintended ar tifact can occur. The spanwise shape of the wing will be partially a function of the planform, due to scaling of the airfoil along the span. This may or may not cause the wing to look funny, and may make the wing difficult to mount to a fuselage or perform other st ructural duties. To alleviate this, the logical parameter edge ref was introduced to allow the user to select between a leading edge reference for the airfoil position, or a maximum camber refe rence. The latter of the two often produces a much more aesthetically pleas ing wing shape, and is consequently used most often. Vector-Based Design Parameters In addition to the basic MAVLAB param eters there were 3 wing para meters that were a little more difficult to define, but equally as im portant. They include d the planform shape, airfoil, and, span-wise shape. Each of these parameters corres ponded to the top, side and front view of the wing, respectively, and were depend ent on the aforementioned scalar parameters.

PAGE 26

26 These parameters are referred to as vector-based (o r vectorized) due to the fact that they needed to be defined by a data set, ra ther than a single parameter. Planform. The planform is the birds-eye-view, or top projection of the wing onto a flat surface. This shape is closely tied to aer odynamic efficiency and over-all lift. Figure 3-3 illustrates the planform of a generalized airp lane wing. While early releases of MAVLAB defined the planform as a function of scalar based parameters, a more general method of producing planform shapes was more desirable. The most common planform shapes are the tapered wing and elliptical wing, the former of wh ich is seen in most commercial aircraft. This is because a tapered wing can be constructed in su ch a way that it is almost as aerodynamically efficient as an elliptical wi ng, but lighter and more cost-effective. MAVLAB initially was developed with these two shapes in mind, and cons equently only let the us er select between the two, with variations of the taper ratio, and ellipse ratio. Th e ellipse ratio simply described a more general elliptical shape composed of two ellipses commonly joined at their major axis, but having different semi-minor axes. This ratio of semi-minor axes was called the ellipse ratio. As MAVLAB developed it was clear that cost and weight rest rictions were not a concern in the composite construction used to make MA Vs, and therefore more complicated planform shapes could be explored. In MAVLAB the pl anform could be defined as any closed polygon imaginable. This could be achieved in several ways. The planform could be input from an ASCII text file, similar to common airfoil coordina te files. Alternativel y, the planform could be interactively drawn and manipulated using an interpolating spline. Airfoil. As mentioned earlier, the airfoil is the single most critical part of a wing. The airfoil shape largely governs the lift, drag and pitching moment of the aircraft [15]. Thus, the selection and design of an airfoil is often critical. The original airfoil for the University of

PAGE 27

27 Florida MAV concept was designed under the following constraints. It had to be a thin airfoil with a neutral (i.e. zero) pitching moment. In order to produce this, a genetic optimization was performed using a well known panel method visc ous airfoil analysis software called XFOIL (developed by the creator of AVL, Mark Drela). Since the airfoil was thin, it could be described by a polynomial of order 6, and the coefficients of the polynomial we re used as the variables for optimization. The results of this analysis were wh at inspired the use of th e thin, reflexed airfoil largely used in MAV designs at the University of Florida. Figure 3-5 depicts a more general thick airfoil, but the camber line in the illustrati on is more representative of what a thin airfoil would look like. MAVLAB offers the capability to fully define thin airfoils for a given wing design. This could be performed in one of two ways. First, the user had the ability to read in normalized airfoil coordinates from an ASCII text file, which described the ai rfoil geometry as a discritized set of x and y data points (tab or space delimited) from leading to trailing edge. Second, the user could create a custom airfoil using an interactive interpolating spline curv e. In addition, the custom airfoil could be expressed as a function of the chord position in the range of 0-1. Only one airfoil had to be defined for a given wing, but there could potentially be an unlimited number of airfoils. Each airfoil was assigned a normalized span-wise posit ion from 0 (the root) to 1 (the wing tip). MAVLAB automatically and smoothly interpolated between each airfoil section to create a smooth continuous wing surface. Span shape. The span shape is a generalization of the dihedral of a wing. It basically refers to the shape of the span-w ise shape of the wing from the front view, as seen in Figure 3-6. Like the planform, the span shape could potentia lly take any form. It was defined by a set of data points describing the overall ve rtical displacement of the wing as a function of span. This

PAGE 28

28 could be done by either importing a text file of coordinates or drawing one half of the symmetric span shape with an interpolating cubic spline. Additionally, the shape of the span could be defined as a explicit function on the range of 0 to 1, where 0 corresponded to the root and 1 corresponded to the wing tip. Aerodynamic Analysis In addition to providing the user with an e fficien t geometrical design interface, MAVLAB also provides the capability to perform aerodyna mic analysis. MAVLAB can export geometry data to an extended vortex lattice CFD softwa re called AVL. Created by Mark Drela, AVL (Athena Vortex Lattice Model) was intended for ra pid aircraft configurat ion analysis and could compute aerodynamic metrics such as lift and drag coe fficients as well as stability derivatives. It also had the capability to model slender bodies, as well as performing trim f light calculations and eigenmode analysis. While the vortex lattice met hod of solving for the fluid flow of a surface has limitations, it gave reasonable lift and stab ility data as well as a relative trend in performance. Computer Numerically Controlled Machining MAVLAB has the c apability to export tool path files to a CNC milling machine. The advantage of this feature is the ability to mill wing molds that were exact representations of the CAD designs created in MAVLAB reducing the likelihood of geom etrical asymmetries in the airframe. Figure 3-9 depicts a wing mold being machined on a CNC milling machine. The molds were machined from high density tooling board. This material was chosen because it was a tough, high-density polyuretha ne product intended for use in models, prototypes, mold-tooling, and in composite mold applications where a uniform, grain-free, dimensionally stable substrate is desired. In addition, it was designed to withstand the high temperatures needed for composite curing and could be machined similar to aluminum.

PAGE 29

29 However, the advantage over other materials such as aluminum wa s that the tooling board could be machined at much higher feed rates, increa sing material removal ra te and reducing production time. This advantage allowed the machined wing mo lds to be produced in as little as 25 minutes on a standard 3-axis CNC end mill at a feed-rate of 45 inch/min. This time was dependent on the size of the wing mold, but could potentially be dramatically improved if high-speed machining was implemented. At speeds of 600 inches/min and up, the machine time could be reduced by an order of magnitude. After machining, the wing molds were not perfectly smooth due to the cusps created by the scalloping of the ball end-mill (Figure 3-10). It was important that the wing mold be smooth, so the remaining material from the scalloping had to be sanded away by hand. To assure that the remaining material was removed evenly, a light tr ace coat of black spray paint was applied to the mold surface before sanding. Then, a high-grit sand paper was applied to the surface until all painted regions were gone. An example of a completed wing mold is shown in Figure 3-11. The scallop height is a physical parameter which directly affe cts the surface finish of the part. Mathematically, it is the maximum height of the ridges produced by the ball end-mill (see Figure 3-12). The following equati on relates the scallop height (hscallop) to the step-over distance (xstepover) and diameter of the tool (Dtool). 2 21 2scallop tooltoolstepoverhDDx 3-1 Step-over is the normal distance between adjacent paths of the tool. In order to create a tool path, from the wing geometry, the surface had to be able to be finely meshed according to the desired scallop hei ght and direction of cut. The surface normals at each of the grid points were calcu lated and scaled to the radius of the tool. Each normal is

PAGE 30

30 placed at its corresponding grid point. The order in which these points were arranged determined the toolpath. Depending on the geometry, it could be necessa ry for the toolpath to follow various patterns. MAVLABs options for tool path pa tterns included a zigzag (chord-wise/span-wise) and spiral path. In addition to creating the toolpath, MAVLAB also overlayed the toolpath pattern over the wing surface and drew a box around the wing to signify the tool containment boundary. The length, width, and hei ght of this box are also returned for reference. Finally, in order to verify that the toolpath was exactly what the user wanted, MAVLAB gave the user the opportunity to visualize a simu lated machining process, call ed toolpath verification. Composite Construction The typical fabrication of a MAV revolved ar ound the use of composite materials within the wing and fuselage structure, including carbon fiber and Kevlar. The following documents the custom fabrication process developed by the University of Florida [16-21]. The choice of composite materials was influenced by their materi al properties. The primary composite utilized was carbon fiber and thermoset epoxy in a preimpregnated form which demonstrated an inherently high strength to weight ratio, which was ideal for aircraft structures. This allowed the possibility of creating a light weight wing with a thin airfoil wh ile still being able to sustain aerodynamic loading. The carbon fiber cloth used in construction was a 0/90 plain weave, 176.3 g/m2, 3000 fiber/yarn pre-impregnated fabric. In addition, pre-impregnated unidirectional carbon fiber is also used. Despite the high strength of the carbon fiber, 17 g/m2 Kevlar cloth reinforcement was sparingly used to prevent cr ack propagation (brittle failure) in the 0/90 weave cloth on the wing leading edge Kevlar, while not nearly as strong as carbon fiber, was extremely resistant to shear stresses. Thus, Kevlar wa s concentrated around regions susceptible to high impact loads, such as the leading edge of the wing.

PAGE 31

31 The wing of the MAV was the most important airframe component with respect to flight performance, requiring a great deal of care to be taken in its fabrication alone. No viable automated process was available to lay up the wings, so development of a new procedure for composite construction of the airf rame was necessary. In addition, the University of Florida has developed several different wing fa brication techniques. This sect ion will outline the fabrication process of a batten reinforced, flexible membrane wing and a ri gid composite wing (both shown in Figure 3-13). In addition, this section will al so cover the procedure of fabricating a composite fuselage. Flexible Wing Fabrication More complex than a solid composite layup, fl exible wing designs first required a detailed layup diagram to be printed on the tool. This was often drafted in a computer drawing program and then printed to paper. The paper was then overlaid on the mold and traced out with a marker (Figure 3-14). Teflon release film was then adhered to the wing tool using Elmers spray glue. The tool was, first, lightly coated with spray glue, and th en Teflon release film was delicately applied to the surface (Figure 3-15) by tacki ng it down at one corner and th en spreading the entire sheet into the contours of the wing surface. Care was ta ken to prevent wrinkles from forming, as this would adversely affect the final surface finish. The leading 20% of the wing was constructed of a single layer of 0/90 cloth oriented at a 45 degree angle with respect to the chord line. A pattern was generally used to cut out a piece of carbon fiber to the proper shape (Figure 3-16). Th e resulting patch of carbon fiber was aligned with the pattern on the tool. Th e 0/90 cloth was then reinforc ed by a single additional layer made of Kevlar cloth (Figure 3-15).

PAGE 32

32 Two millimeter wide strips of unidirectional carbon fiber were cut from a sheet of preimpregnated material using a custom built carbon st rip-cutting tool (Figure 3-17). The portion of the wing that attached to the fuselage was made by carefully aligning la yers of unidirectional carbon fiber strips with the pattern drawn on th e tool (Figures 3-18, 19). The unidirectional carbon fiber was layered for strengt h and rigidity and was interwove n at intersections to create stronger mechanical joints. Elevator control su rface hinges were created by interweaving a single layer of Tyvek material into the gaps between the contro l surface and frame. Rigid Wing Fabrication Though the University of Florida was known for the development of the flexible membrane wing, it also developed a bendable rigi d wing. While the rigid wing did not benefit from the desirable flight characteristics of th e membrane wing, it was more suitable for the design of the bendable wing. The rigid wing was si mpler in construction than the flexible wing, but required a separate fabricati on process, documented below. Rigid wing construction was largel y similar to the flexible membrane wing, so the process is only summarized in the following steps. First, Teflon release film was applied to the surface of the mold with spray adhesive to prevent the resin from bonding to the mold. A single layer of pre-impregnated carbon fiber weav e was typically placed on the rel ease film, biased at a orientation. Kevlar weave was cut out and placed over the leading edge of the wing. Additional layers of carbon fiber were added as needed to strengt hen the wing where high bending stresses were expected. Figure 3-20 shows an example of a rigid carbon fiber wing after layup and cure. Fuselage Fabrication The fuselage lay-up was completed in a sim ilar process. Carbon fiber weave was wrapped around a Teflon lined male mold to produce the nose of the fuselage. Multiple layers of carbon

PAGE 33

33 fiber and Kevlar were applied to the nose to pr ovide adequate reinforcement for impacts. The fuselage molds were not produced using MAVLAB as the precision of their construction was not as critical, and they were difficult to define by simple parameters (Figures 3-21-23). Vacuum Bagging and Curing The entire wing and fuselage were covered w ith porous Teflon peel-ply, to allow excess resin to pull out of the carbon fibe r without sticking to it. A layer of fabric breather material was placed on top of the molds to ensure a vacuum w ould be applied evenly to the composite covered mold surfaces. Finally, a small hole was cut in the vacuum bag to accept the hose attachment for the vacuum pump. Vacuum bag tape was used to seal up both ends of the bag and the hose was connected to the attachment and a vacuum was drawn to 0.01 atm absolute. Once the air had been removed from the bag, it was inspected fo r leaks and air pockets. Leaks were sealed up with additional tape and air pockets were alleviat ed from inside corners by repeatedly removing the vacuum and forcing the bag into the corners. The vacuum bag was then cured in an oven at 130 C for 2 hours (Figures 3-24,25). Fuselage Assembly Once cured, excess material was trimmed away as shown in Figure 3-20, and edges were sanded to remove splinters. The fuselage and wi ng were mated together to ensure an acceptable fit (Figure 3-26). The wing was often placed on th e wing mold, and the fuselage is glued into place, a method that helps to reduce asymmetries in the MAV. For the flexible wing construction, the re maining open surface of the wing was then skinned with 0.25 mm latex rubber. The latex was applied in a three step process. All but a 3 mm perimeter of contact surface around each of the open areas of the wing were masked with masking tape (Figure 3-27). Next, Elmers spray glue was then lightly coated to the top surface of the wing. The masking tape was then remove d and the glue was left to set-up for thirty

PAGE 34

34 seconds. Lastly, a latex membrane was adhered to the top surface of the wing. An aptly sized rectangular section of latex wa s first stretched out over a bloc k of foam and pinned down to secure some pretension in the membrane and prev ent any slack or wrinkles. This pretension was not typically measured, however. It was importa nt to note that the enti re wing was skinned at once in order to keep the relative membrane tensi on in both wings equal (Figure 3-28). The latex was then pressed firmly against the wing, tr immed with a razor blade and secured with cyanoacrylate glue (Figure 3-29). Components Installation Servos were attached to the fuselage by doubl e sided tape (Figure 30), and lashed into place with Kevlar thread. Control rods (1mm diam eter steel) were then installed, connecting the servos to the control surf aces (Figure 3-31). The rudder control surface was created by cutting a section out of the vertical stabil izer, and then reattaching them with Tyvek hinges and glue. Special care was taken to minimize any clearance or play in these linkages. Electronic components such as the electronic speed controlle r, GPS receiver and autopilot were then placed inside the fuselage and adhered with double-side d tape where necessary. An acceptable size hole was bored into the nose of the fuselage for the motor using a Dremmel tool. Glue was used to adhere the motor to the fuselage with the proper thrust angle.

PAGE 35

35 Table 3-1. List of basi c MAVLAB wing parameters Parameter Dimension Type Description Span Length Wing span Chord Length Root chord length Camber % Airfoil Maximum Camber Z-scale % Scale factor for Z-direction Twist Angle (degrees) Spanwise wing twist Sweep Angle (degrees) Wing sweep angle Dihedral Angle (degrees) Wing dihedral angle Edge Ref Logical Span shape referenced at leading edge (True)

PAGE 36

36 Figure 3-1. MAVLAB Gra phical User Interface. Figure 3-2. Typical grap hical display of a wing geometry in MAVLAB, with feedback for basic parameters such as planform area, mean chord, aerodynamic center and aspect ratio.

PAGE 37

37 Figure 3-3. Wing span as defined by the linear distance from wing tip to wing tip. Figure 3-4. Wing root chord is de fined as the distance from leadi ng to trailing edge of the wing, closest to the fuselage, or middle of the span. Chord Span

PAGE 38

38 Figure 3-5. The maximum camber is the ratio of the highest point of the camber line to the airfoil chord length. Figure 3-6. The geometric dihe dral angle is the angle the wing makes with the horizontal. Figure 3-7. The sweep angle is the angle be tween the chord line ( dotted) and the line perpendicular to the free stream velocity. Dihedral Sweep Camber

PAGE 39

39 Figure 3-8. Geometric wing twist, often called washout when it is negative and washin when it is positive. Twist (washout)

PAGE 40

40 Figure 3-9. Typical example of CNC milling of a MAV wing mold out of resin-based tooling board. Figure 3-10. Scalloping or cusps created by consecutive passes of a ball (radial) end mill.

PAGE 41

41 Figure 3-11. Typical wing mold machined fr om high-density polyuret hane tooling board. Figure 3-12. Typical ball end mill used for mach ining 3D freeform surfaces such as wing molds. A. B. Figure 3-13. A) Rigid carbon fiber wing. B) flexible membrane wing with batten reinforcement.

PAGE 42

42 Figure 3-14. Drawing layout pattern on the wing mold. Figure 3-15. Applying Teflon release film to the surface of the wing mold using spray glue

PAGE 43

43 Figure 3-16. Laying up of carbon fiber on the wing mold. Figure 3-17. Application of Kevlar to the leading edge.

PAGE 44

44 Figure 3-18. Cutting of unidirectional carbon fiber strips with custom cutting tool. Figure 3-19. Trimming of ex cess carbon fiber during layup.

PAGE 45

45 Figure 3-20. Trimming excess carb on fiber from wing after cure. Figure 3-21. Application of Teflon release film to fuselage mold.

PAGE 46

46 Figure 3-22. Fuselage placed on a form to maintain proper mate with wing. Figure 3-23. Application of 1-2 layers of carbon fiber to the fuselage mold.

PAGE 47

47 Figure 3-24. Vacuum bagging of wing tool. Figure 3-25. Vacuum bagging of fuselage covered with peal-ply release film.

PAGE 48

48 Figure 3-26. Assembly of wi ng and fuselage after cure. Figure 3-27. Masking prep and application of sp ray glue to top of wing for latex application.

PAGE 49

49 Figure 3-28. Application of la tex membrane to top of wing. Figure 3-29. After latex trimming, final gluing of latex to carbon fiber skeleton of the wing.

PAGE 50

50 Figure 3-30. Component instal lation, with servos lashed to inside of fuselage. Figure 3-31. Tyvek hinge rudder and attach ment of control rod to control surfaces.

PAGE 51

51 CHAPTER 4 REPRESENTATIVE STUDY In theory, M AVLAB was a clearly useful so lution for the design and manufacturing of MAVs. In order to test this theory, a representative study was done on a MAV wing to verify the accuracy in which MAVLAB could aid in predic ting aerodynamics as well as the accuracy in which it could generate wing molds. Computer Model Description Despite the name, MAVLAB was capable of designing and analyzing wings of all sizes, not just 6 inch MAVs as originally defined by DAR PA. In fact, in recent years, this dimensional restriction for the definition of the MAV has b een relaxed, as it was not seen necessary to explore MAV designs only under 15 cm in maximu m dimension. As such, the University of Florida has done much research in the 6 24 inch realm of small UAVs. The following study examined the design of a thin, flexible MAV shown in figure 4-1 and 4-2. The wing was constructed using the custom fabrication pro cess for flexible, batten reinforced wings as described in chapter 3, with one exception. An inelastic light-weight, polycarbonate coated nylon fabric called Icarex, wa s used instead of latex. Wing Aerodynamics Comparison The particular MAV wing design used in this study was originally designed to fly at a cruise speed of 13 m/s with an angle of attack of 7-10 degrees. For this reason, testing was only done at 13 m/s. This section compares the lif t, drag and pitching moment of the computer simulated model and actual wind tunnel results using the respective non-dimensional coefficients CL, CD and Cm.

PAGE 52

52 Wind Tunnel Results All wind tunnel testing was performed in a newly installed Engin eering Laboratory Design (ELD) 407B wind tunnel at the University of Flor ida. The tunnel had tw o interchangeable test sections. The smaller of the test sections had inner dimensions of 0.61 m x 0.61 m x 2.44 m and has listed velocity capability ranging from 3 m/s to 91.4 m/s. The small test section was installed downstream of the flow screens and plenum and was upstream of the straight walled diffuser. The large test section has inner dimensions of 0.838 m x 0.838 m x 2.44 m and has listed velocity capability ranging from 2 m/s to 45 m/s [ 22]. The large section was used for this study. The University of Floridas low speed wind t unnel was equipped with a 6-component sting balance used to measure aerodynamic loads. Under the given loads for this particular test case, the sting balance could accura tely predict lift and pitching moment, however drag was questionable. This was due to the fact that the magnitude of the drag forces for this model were on the same order of magnitude as the measur ed resolution of the drag axis on the sting balance[23-24]. Thus drag measurements were not repeatable and likely questionable. Loads measurements were take at a range of angles of attack (alpha), from -5 to 25 degrees. Figures 4-3 through 4-5 show the relati onships of lift, pitchi ng moment and drag vs. angle of attack. Figure 4-6 represents a metric of aerodynamic efficiency, called the lift to drag ratio. This data was based on the data from th e first two figures. Figure 4-3 shows that the wing demonstrated a zero lift angle of attack of roughly -2 degrees, a lift slope of 0.06 and stall at about 17 degrees. The slope of the pitching moment vs. alpha (Cm ) is an indicator of longitudinal stability, where a ne gative slope is desirable. In this case, the wind tunnel results showed that the wing had a Cm of approximately -0.013. Minimu m drag appeared to occur at 5 degrees angle of attack; howev er this is questionable due to the reasons stated above.

PAGE 53

53 Athena Vortex Lattice Results The aerodynamic simulations performed in MAVLAB were done by automatically exporting the wing data to a free VLM softwa re called AVL [25]. AVL was capable of predicting a variety of aerodynamic and stability metrics in a matter of seconds. But, since the wind tunnel test results were limite d to lift, drag and pitching mo ment vs. angle of attack, these were the only values valid for comparison. The vortex lattice method implemented by AVL does not fully model the behavior of a body in fluid flow, and thus there were some assumptions that needed to be considered. This method models fluid flow as invisc id, and as such, does not accurately predict wing stall or drag. The drag that it does measure is induced drag, which should only be a small portion of the total drag of a wing on this scale. In addition, AVL does not model the flexibility of the wing, and cannot accurately model the steady state deformation of the wing during flight. That being said, AVL was able to predict aircraft performance relatively well. The model was partitioned into a grid of 24 span-wise points and 8 chord-wise poi nts. It took roughly 15 seconds to compute the lift, drag and pitching moment at a range of angles of attacks between 0 degrees and 20 degrees. Results ma tched closely to wind tunnel tests and can be seen in Figures 4-3 through 4-6, with the exception of drag. This was to be expected for two reasons, since AVL only predicts induced drag and the wind tunnels measurements were questionable in the first place. It should also be noted that though the pitching moment curves dont appear to be consistent, Figure 4-4 is misleading. The slopes are nearly identical, but there is a slight vertical shift. This is possibly a function of the rigi d wing assumption in AVL, but it is not a serious concern, as the slope of th e curve is most important.

PAGE 54

54 Wing Mold Accuracy It was vitally important that the wing geomet ry be produced and reproduced as accurately as possible. Any discrepancies in the wing mold geometry would lead to unwanted behavior in aircraft performance. In addition, it would also weaken the correla tion between simulated aerodynamic analysis and experimental measurement. The following documents the process of quantifying the accuracy of a wing mold produced from MAVLAB. In order to experimentally qua ntify in flight elastic wing deformation of thin, flexible wings, the University of Florida has employe d the Visual Image Correlation (VIC) system, developed at the University of South Carolin a [6,20,23]. The VIC system was developed by Helm et al. [27] in the mid 1990s and provided a global shape and deformation measurement for 3D geometries. The VIC system used two cameras to obtain highly accurate 3D measurements of a surface prepared with a low luster, high co ntrast, random speckle pattern. Measurements, both in plane and out of plane, were obtained by a comparison of th e test subject in the deformed state and a reference state. Error quantifica tion was based on many variables and was different for each set up. Errors of 0.05 mm have been reported [26]. Figure 4-6 illustrates the out of plane position of the wing mold as measured by the VIC system. The speckled pattern can be seen behind th e color overlay. It should be noted that the VIC was not capable of measuring position up to the boundary of the region of interest, leaving an approximately 3mm border around the perime ter of the wing that was not measured. In order to quantify the difference between the machined wing mold and the computer model, the experimental VIC data was read into Matlab and interpolated onto the same grid used by the original computer model. The diffe rence of the full-field out of plane position for each was computed and is represented in Figur e 4-7. The zero disp lacement region around the edge of the wing was an artifact of the fact th at the VIC could not measure all the way to the

PAGE 55

55 border, as mentioned earlier. Note that there is also an apparent raised region (in yellow and circled in Figure 4-8). This was due to an in tentional circular gouge made in the mold by the CNC at the time of machining in order to produce a reference point. Figures 4-9 through 4-12 depict slices of the wing for better comparison and visu alization of error. Conclusion Overall, MAVLAB seemed to demonstrate rema rkable results in this case study. Despite rather large assumptions, AVL predicted aerody namic performance relatively accurately, and certainly well enough to give the user insight in to the performance of the wing design before it was even tested. Though it appeared, at first glance, that there were considerably large differences between the actual wing mold and th e modeled surface, the range of error between the two was within the error of the VIC resolution of 0.05 mm. The user could be sure that the computer modeled wing would be produced accura tely via a precisely machined wing mold. The final results of the VIC testing showed that the wing was machined within acceptable tolerances.

PAGE 56

56 Figure 4-1. Thin, flexible wing model used to compare wind tunnel aerodynamics with Athena Vortex Latices computational aerodynamics approximation. Figure 4-2. Thin, flexible batten reinforced M AV wing with Icarex used in wind tunnel testing.

PAGE 57

57 -0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 -5051015202530 AlphaCl Experimental AVL Figure 4-3. Lift vs. angle of attack comparison between expe rimental wind tunnel data and simulated results from AVL for a thin, under-cambered flexible wing. -0.50 -0.40 -0.30 -0.20 -0.10 0.00 0.10 -5051015202530 AlphaCm Experimental AVL Figure 4-4. Pitching moment vs. angle of atta ck comparison between experimental wind tunnel data and simulated results from AVL fo r a thin, under-cambered flexible wing.

PAGE 58

58 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 -5051015202530 AlphaCd Experimental AVL Figure 4-5. Drag vs. angle of attack compar ison between experimental wind tunnel data and simulated results from AVL for a thin, under-cambered flexible wing. -0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 0.00 0.10 0.20 0.30 0.40 CdCl Experimental AVL Figure 4-6. Lift vs. drag comparison between experimental wind tunne l data and simulated results from AVL for a thin, under-cambered flexible wing.

PAGE 59

59 Figure 4-7. Visual image correlation 3D sca nning overlay of out of plane displacement on painted mold (speckle pattern). Figure 4-8. Comparison of wing model geometry and actual wing mold, where color represents variation in out of plane position in millimeter s (note the yellow spot at the center of the root of the wing, indicating a purposeful gouge in the mold as a reference point).

PAGE 60

60 Figure 4-9. Comparison of M AVLAB wing model and machined wing mold cross section along the wing span. Figure 4-10. Measured machining error along the span-wise cross se ction of the wing.

PAGE 61

61 Figure 4-11. Comparison of M AVLAB wing model and machined wing mold cross section along the chord-wise direction (airfoil). Figure 4-12. Measured machining error along the chord-wise cross section of the wing.

PAGE 62

62 CHAPTER 5 FUTURE WORK MAVLAB has proven to be a useful aid in the design and analysis of MAVs. However, it is a too l that is, by no means, complete. For instance, it cannot perform any automated aerodynamic optimization. Also, with regard to the modeling of the aircraft aerodynamics, it ignores the flexibility of the wing structure. In order to include the flexible model of the wing, MAVLAB would require some type of finite elem ent analysis (FEA). Some work has already been done in this area and could be implemente d fairly easily. Lastly, FEA and optimization could be coupled together along with an entire ai rcraft model to obtain full aircraft dynamics. Optimization Matlab is a prime environment for numerical optimization routines because it can be purchased with toolboxes designed with that in mind. This is an obvious advantage in that optimization code does not necessarily have to be written from scratch. A good example of how Matlabs optimization toolbox could be taken a dvantage of is the function FMINCON, which tries to perform a constrained minimization of a nonlinear multivariate function. The aircraft dynamics as a function of design parameters such as span, chord, camber, twist etc. are clearly multivariate and likely highly non-linear in nature. If one wanted to find the maximum lift at a given flight attitude by changi ng the airfoil camber, chord length and wing twist, FMINCON could be used in conjunction with the output of AVL to converge on the solution fairly quickly. Aeroelasticity Aeroelasticity is the science which studies the interaction among in ertial, elastic, and aerodynamic forces. The construction of MAV wings developed and generally used by the University of Florida were thin and flexible in nature. Under normal fli ght conditions and loads, flexible wings tend to deform, thus introducing aeroelasticity. This fact is ignored when

PAGE 63

63 performing aerodynamic analysis using AVL. One solution might be to perform an iterative computational analysis by including FEA in th e loop. Figure 5-1 depicts a diagram of the iteration process. Full Aircraft Modeling As of the writing of this thesis, MAVLAB has the ability to send only wing geometry to AVL for analysis. This is acceptable for analysis of just the wing, but often the case is more complicated and requires the inclusion of the enti re airframe. For example, one may need to know how big and where to place a hor izontal and vertical ta il, if any is needed at all. AVL is an excellent tool for doing this type of analysis, given a certain wi ng configuration, fuselage shape and mass distribution. As of now, it is conceiva ble that one could modify the wing input file MAVLAB generates for AVL, to include the rest of the airframe. This is a likely, but time consuming step to perform more rigorous airf rame analysis. A better and more long term solution would be to incorporate code that allo wed a user to select additional wing surfaces and fuselage parts to be included in the automated ge neration of the AVL input file. Even further, one could implement some level of optimizatio n to increase wing and airframe efficiency.

PAGE 64

64 Figure 5-1. Diagram of itera tive aerodynamic analysis where the wing forces computed by AVL are input into a FEA solver to determine the steady state deformation of a thin, flexible wing during flight.

PAGE 65

65 APPENDIX A USERS MANUAL MAVLAB is a Matlab based toolbox designed to aid in the design and m anufacturing of MAVs. This section is not a manual for the Matlab programming environment, and assumes the user has basic general knowledge of Ma tlab commands. For further information on operating Matlab consult Mathworks website at http://www.mathworks.com/support/ This users m anual is not intended to be comprehensiv e, but it does include a basic tutorial of MAVLAB, an example usage of the program, as we ll as a list of important functions included in the MAVLAB toolbox. Basic Description of Features Much of the basic description and important features of MAVLAB have already been discussed in chapter 3. However, this section briefly outlines the major features of the toolbox. MAVLAB offers a specific CAD suite designe d for MAV wing creation, a plug-in for CFD aerodynamic analysis using AVL (Athena Vortex Lattice, Drela) a nd a specialized CAM software capable of exporting the wing geometry as a set of commands fo r CNC machining. The CAD portion of the software allows full mani pulation of a wing design through basic input parameters (Table 3-1) and 3 vector paramete rs. The vector parame ters describe the wing planform, span-wise shape, and airfoil as a set of data points. The power of MAVLAB comes into effect by allowing the user to either read the vector parameters in through saved ASCII text files, or manipulate each through an interactive interpolating spline curve. Also, in order maintain compatibility with other CAD softwa re, the wing geometry can be exported to a number of standard CAD formats. The aerodyna mic analysis is performed by converting the wing geometry to a format suitable for AVL and th en calling a batch routine to run analysis. The results are saved to a text file and finally read back in MAVLAB and printed to the screen. Once

PAGE 66

66 the user has found the wing parameters that pr oduce the desired results, MAVLAB can help create a physical model of the wi ng. It does this by creating a se t of instructions used by a CNC machine to mill out a wing mold from stock mate rial. The MAVLAB interface allows the user to select from several different style toolpaths and other milling parameters. In addition, the user can simulate the milling procedure to graphically verify the wing mold before it is created. Getting Started with MAVLAB MAVLAB is designed to be intuitive and easy to use graphical user interface (GUI). Much of the underlying functionality is hidden to the user, who doesnt need to know how it works. Most of the interaction with MAVLAB is prompted by icons and tool-tip descriptions of what each function does. The main focus of this manual is the main MAVLAB interface. After saving the main MAVLAB folder and subfolders to the search path, MATLAB can be called from the command prompt by simply typing mav lab without quotes a nd in all lowercase letters. >> mavlab This will start the main program (Figure A-1) and a window will pop up with a rather uninteresting green square and some edit boxes a nd buttons. The green square is the default shape for the wing, and assumes a rectangular plan form with a wingspan of 1 and aspect ratio of 1. This, of course, is not ideal or reasonable for any practical aircraft design. Included with MAVLAB are several test wing files. The example wing files are included in the main MAVLAB directory under the saves folder. The us er is encouraged to open one of these files to see what MATLAB can do.

PAGE 67

67 Sample Usage Open and Saving Wing Files Saved MAVLAB wing files can be found by clic king on the folder icon at the top of the main window. This will open a dialog in whic h saved MAVLAB wing file s with the extension *.mat can be selected. Alternatively the user can load a wing file into MAVLAB at start up by first typing load wingfile.mat, where wingfile.m at is the name of the MAVLAB wing file. This loads the wing file into the current worksp ace. Then the user can load the wing into MAVLAB by typing mavlab(wing). In addition to loading existing files, MAVLAB can also save them. To do this, simply click on the save button at the top of the MAVLAB main window. A save dialog will appear to select a filename and destination. Modify Some Parameters MAVLAB offers a variety of parameters that can be manipulated and are explicitly described in chapter 3. In order to modify the basic parameters simply type the value into the corresponding field in the main MAVLAB window The three vector parameters, planform, airfoil and span shape can be modified by clicki ng on the respective images to the left of the main MAVLAB window. For each, a new window will appear to modify or redefine each one. This can be done by clicking on a the curve in the window and manipul ating spline control points, or simply selecting an AS CII text file of coordinates. Export Wing Geometry MAVLAB offers several useful and common industry format s in order to facilitate exporting of the model. These include: IGES (International Graphic Ex change Specification), STL, AutoCAD data exchange format (*.dxf), and Wavefront (*.obj) formats. This can be performed by clicking on the export button at the top of the main MAVLAB window. A dialog will then appear asking which format to save it in.

PAGE 68

68 Run Athena Vortex Lattice Aerodynamic Analysis Once the desired wing geometry has been achieved by manipulating the parameters in the main MAVLAB window, aerodynamic analysis can be performed by running a plug-in called AVL. To start AVL, simply click on the AVL button at the top of the main MAVLAB window. A new window will appear (Figure A-2) w ith options and settings for AVL. The user can select different parameters in the list box to the left. The para meters can then be modified by typing in a numerical value or e xpression into the edit box to the right with the heading of the current list box field. Multiple parameters can be set to the same value at once by making multiple selections and entering the value. Defau lt values are set for gravity and air density, and are in metric units. In order to run AVL, at least velocity and angle of attack need to be set. In addition to the list box, two other parameters need to be set. AVL divides the wing into discrete chord-wise and span-wise sections. In the ed it boxes near the bottom, these two corresponding values can be modified. It appears that th e default values are acceptable for most cases. Create Computer Numerically Controlled Toolpath The wing toolpath can be created relativel y quickly in MAVLAB. To start the CNC portion of MAVLAB, click on the button at the to p of the window labeled CNC. This will open a new window with a graphical display of the wing and a dialog box asking for three parameters (Figure A-3). These parameters refe r specifically to machining parameters and are the machine tool diameter, stepover and tolerance. The tool diameter can be any positive real number, and will reflect the diameter of the tool go ing to be used for machining. The University of Florida typically uses a inch diameter t ool. The stepover is a me tric of the roughness desired in the final wing mold. Experience has shown that a typical va lue of 0.050 inches was acceptable for machining of wing molds in polyur ethane tooling board. The final parameter, called tolerance, reflects the relative resolution of the discretized toolpath Since freeform curves

PAGE 69

69 are approximated by piecewise linear segments in a toolpath, it is necessary to specify a value for the tolerance depending on the comple xity of the wing. If the to lerance is set too small, the toolpath will become too large to send to the CNC computer. Once the desired initial parameters are set, the user can select between two types of toolpath shapes, zigzag and spir al (Figures A-5,6). Pressing the corresponding button at the top of the window will generate the toolpath and display it with the wing. To verify that the toolpath will turn out as desired, a feature has been includ ed to graphically simulate the milling process. By clicking on the eye button at the top of the me nu, toolpath verification will initiate (Figure A-6). Note that the dimensions of the minimum size work piece are displayed with the simulated milled mold. Finally, to machine the wing mold, the wing must be exported to the CNC machine. MAVAB can do this by generating a text file of instructions for the CNC to follow called Gcode. This G-code is saved in a file called an NC file (*.nc) and can be created in MATLAB after a toolpath has been generated. To generate a toolpath, click on the button at the top of the CNC window with the pencil and paper. A dial og will appear asking where to save the file. List of Functions Matlab code is generally composed of one or more files called m-files. Large programs, such as MAVLAB, are typically broken down in to tens and hundreds of m-files which each perform a different function. Often, in Matlab, a collection of functions (or m-files) designed to work together to perform similar tasks are termed a toolbox. These files will be grouped together under one directory and given a toolbox name. Main Functions MAVLAB is more of a toolbox than a program. As such, it is composed of over 100 functions. The following are a summary of the main MAVLAB toolbox functions.

PAGE 70

70 mavlab.m Entrance function for MAVLAB called at command prompt dcAVL.fig Supporting figure file for dcAVL.m dcAVL.m GUI for running AVL CFD dcBuild.m Main function for genera ting wing geometry from parameters dcCNC.m GUI for creating a CNC toolpath for airframe geometry dcGetFoil.fig Supporting figure file for dcGetFoil.m dcGetFoil.m GUI for creating/modifying an airfoil dcGetPlan.fig Supporting figure file for dcPlanFoil.m dcGetPlan.m GUI for creati ng/modifying the wing planform dcGetSpan.fig Supporting fi gure file for dcGetSpan.m dcGetSpan.m GUI for creating/modi fying the span-wise shape of a wing dcLoadfoil.fig Supporting figure file for dcLoadfoil.m dcLoadfoil.m GUI for loading an airfoil into the database from a text file dcMain.fig Supporting figure file for dcMain.m dcMain.m GUI main function for MAVLAB dcSurf.m Plot airframe geometry from parameter data Sub-Functions In addition to the main toolbox functions, whic h would generally be the most used, there are dozens of helper functions. These helper su b-functions are not typical ly called on their own, except for more advanced usage. The following is a list of the most useful advanced subfunctions. avl.m Run AVL from wing geometry parameters mat2avl.m Convert surface data to *.avl file nrbloft.m Convert series of NURBS curves to a NURBS surface nrbinterp.m Interpolate a NURBS curve to a set of data points nrbWing.m Create a NURBS surface from wing parameters gnurbs.m Interactively mani pulate a NURBS curve/surface gspline2.m Interactivley draw/edit a cubic spline curve scanfoil.m Low level reading of airfoil coordinate text files mkWing.m Generate the wing parameter structure surf2spiral.m Generate a spiral toolpath from a surface surf2zigzag.m Generate a zigzag toolpath from a surface surfature.m Compute Gaussian/mean surface curvature verifytoolpath.m Simulate milling process of a toolpath offsetsurface.m Create offset/parallel surface for use with toolpath creation cncpost.m Write a toolpath *.nc file from toolpath coordinates

PAGE 71

71 Not all MAVLAB toolbox functions are listed ab ove; just the ones that are of most interest to the user. Included in M AVLAB are several su b-toolboxes, includi ng a NURBS toolbox as well as some needed functions from the SPLINE toolbox. Also included are two toolboxes not currently being used by MAVLAB but helpful wi th regard to future work and additional implementations of the MAVLAB toolbox. T hose two toolboxes are a FEA toolbox specialized for MAV wings and a MESH toolbox designed to he lp mesh wings for FEA. Neither of these toolboxes are complete nor come with any guara ntee. They are included only to help those doing future work with MAVLAB.

PAGE 72

72 Figure A-1. MAVLAB Gra phical User Interface. Figure A-2. AVL Graphical User Interface.

PAGE 73

73 Figure A-3. CNC Toolpath generator GUI.

PAGE 74

74 Figure A-4. Zigzag toolpath example. Figure A-5. Spiral toolpath example.

PAGE 75

75 Figure A-6. Verification of milling toolpath via simulation (note that the minimum dimension of the work piece was defined in the rendering).

PAGE 76

76 LIST OF REFERENCES 1 Ifju, P.G., Ettinger, S., Jenkins, D.A., & Mar tinez, L., Composite Materials for Micro Air Vehicles, Proceeding for the SAMPE Annual Conference, SAMPE Journal, Vol. 37, 2001, pp. 7-12. 2 Mueller, T., Fixed and Flapping Wi ng Aerodynamics for Micro Air Vehicle Applications, Proceedings of the Conference on Fixed, Flapping and Rotary Wing Vehicles at Very Lo w Reynolds Numbers, Vol. 195, 2000, pp. 307-339. 3 Ifju, P.G., Jenkins, D.A., Ettinger, S., Lian, Y., Shyy, W., & Waszak, M.R., FlexibleWing-Based Micro Air Vehi cles, AIAA Annual Conf erence, AIAA Paper No. 20020705, January 2002. 4 Jenkins, D.A., Ifju, P.G., Abdulrahim, M., & O lipra, S., Assessment of the Controllability of Micro Air Vehicles , Micro Air Vehicle Conference, Bristol, England, April 2001. 5 Ettinger, S.M., Nechyba, M.C., Ifju, P.G., & Waszak, M., Vision-Guided Flight Stability and Control for Micro Air Vehicles, Proceedings of. IEEE International Conference on Intelligent Robots and Systems, Vol. 3, 2002, pp. 2134-40. 6. Albertani, R., Stanford, B., Hubner, J., Li nd, R., Ifju, P., Experimental Analysis of Deformation for Flexible-Wing Micro Air Vehicles,AIAA Structures, Structural Dynamics, and Materials Conference, AIAA Paper No. 2005-2231, April 2005. 7. Albertani, R., Hubner, J.P., Ifju, P., Lind, R., and Jackowski, J., Experimental Aerodynamics of Micro Air Vehicles, SAE World Aviation Congress and Exhibition, Paper No. 04AER-8, November 2004. 8 Lian, Y., Shyy, W., Ifju, P.G., Membr ane Wing Model for Micro Air Vehicles, AIAA Journal, Vol. 41, No. 12, 2003, pp. 2492-94. 9 Fleming, G. A., & Burner, A. W., Def ormation Measurements of Smart Aerodynamic Surfaces, 44th Annual SPIE International Sy mposium on Optical Scie nce, Engineering, and Instrumentation Optical Diagno stics for Fluids/Heat/Combustion and Photomechanics for Solids, SPIE Paper No. 3783-25, July 1999. 10 Fleming, G. A., Bartram, S. M. Waszak, M. R., Jenkins, L. N., Projection Moire Interferometry Measurements of Micro Air Vehicle Wings, SPIE Paper No. 4448-16, November 2001. 11 Waszak, M. R., Jenkins, L. N., & Ifju, P. G ., Stability and Control Properties of an Aeroelastic Fixed Wing Micr o Aerial Vehicle, AIAA Paper No. 2001-4005, August 2001. 12. Grasmeyer, J. Keenon, M. Development of the Black Widow Micro Air Vehicle, AIAA Paper No. 2001-0127, January 2001.

PAGE 77

77 13. Margason, R.J., and Lamar, J.E., Vor tex-Lattice FORTRAN Program for Estimating Subsonic Aerodynamic Characteristics of Co mplex Planforms, NASA Paper No. TN D6142, 1971. 14. Chua C. K. and Leong K. F. Rapid Prototyping: Principles & Applications in Manufacturing, John Wiley & Sons, Singapore, 1997. 15. Anderson, J. D., Aircraft Performance and Design, WCB/McGraw-Hill, Boston, MA,1998. 16. Mueller, T.R., Kellog, J.C., Ifju, P.G., Shkarayev, S.V., Introduction to the Design of Fixed-Wing Micro Air Vehicles, AIAA, Reston, VA, 2006, pp. 185-240. 17. University of Florida Micro Air Vehicle T eam University of Flor ida Competition Micro Air Vehicles, 6th Annual International Micro Air Vehicle Competition, Brigham Young University, Provo, Utah, May 2002. 18. M. Abdulrahim, R. Albertani, P. Barnswell, F. Boria, D. Claxton, J. Clifton, J. Cocquyt, K. Lee, S Mitryk, and P. Ifju, Design of th e University of Florida Surveillance and Endurance Micro Air Vehicles, 7th Annual Micro Air Vehicle Competition Entry, University of Florida, Gainesville, FL, April 2003. 19. R. Albertani, P. Barnswell, F. Boria, D. Claxton, J. Clifton, J. Cocquyt, A. Crespo, C. Francis, P. Ifju, B. Johnson, S. Jung, K. Lee, and M. Morton, University of Florida Biologically Inspired Micro Air Vehicle, 8th Annual Micro Air Vehicle Competition Entry, University of Arizona, Tucson, AZ, April 2004. 20. Albertani, R., Boria, F., Bowman, S., Clax ton, D., Crespo, A., Francis, C., Ifju, P., Johnson, B., Lee, K., Morton, M., Sytsma, M., Development of Reliable and Mission Capable Micro Air Vehicles, 9th Annual International Micro Air Vehicle Competition, Konkuk University, Seoul, South Korea, May 2005. 21. Claxton, D., Johnson, B., Stanford, B., Sytsma, M., Development of a Composite Bendable-Wing Micro Air Vehicle, 10th Annual Internationa l Micro Air Vehicle Competition, Brigham Young Univer sity, Provo, Utah, May 2006. 22. Sytsma, M., Aerodynamic Flow Characteriza tion of Micro Air Vehicles Utilizing Flow Visualization Methods. Master s Thesis, University of Flor ida, Gainesville, FL, August 2006. 23. Albertani, R., Experimental Aerody namic and Static Elastic Deformation Characterization of Low Aspect Ratio Flexib le Fixed Wings Applied to Micro Aerial Vehicles, Ph.D. Dissertation, Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville, Fl, December 2005. 24. Albertani, R., Hubner, P., Ifju, P., Lind, R., Wind Tunnel Testing of Micro Air Vehicles at Low Reynolds Numbers, SAE World Conference, SAE Paper No. 2004-01-3090, November 2004.

PAGE 78

78 25. Drela, M., Youngren, H., Software: AVL (A thena Vortex Lattice Model), Version 3.26, http://web.mit.edu/drela/Public/web/avl Mark D rela, Cambridge, MA, February 2007. 26. Helm, J. D., McNeill, S. R. and Sutton, M. A., Improved 3-D Image Correlation for Surface Displacement Measurement, Optical Engineering, Vol. 35(7), 1996, pp. 19111920.

PAGE 79

79 BIOGRAPHICAL SKETCH Daniel J. Claxton was born to Linda and Erny Claxton in the sm all town of Orange Park, Florida. He grew up in Orlando where he attended school through high school, graduating from The First Academy. At an ear ly age it was clear that he had an affinity for engineering as he enjoyed taking his toys apart to see how they worked, and could often be found constructing inventions a nd building gadgets and gizmos. As a 1st grader, he aspired to be an inventor and by the time he was in high school it was clear that his interest and proficiency in math and the sciences would lead him to pursue a college education in some engineering discip line. After trying his hand at civil and industrial engineering, he found his niche in the mechanical and aerospace engineering program at the University of Florida. The small college town of Gainesville b ecame Daniels home while he earned his bachelor and master degrees. During that time he became involved in Campus Crusade for Christ and made dozens of life-long friends. It was also about this time that he became involved in the Micro Air Vehicle progr am at the University of Florida. He became a part of an elite group of rese archers looking to explore the realm and capabilities of small aircraft. He was a part of the team fo r five years, and was proud to take home five International Micro Air Vehicle Competition championships. Courtesy of UF, he had the privilege to travel all over th e world, including Eglin Air force Base (FL), Utah, Arizona, South Korea and Germany. During the completion of his masters degree he married the love of his life, and accepte d an engineering pos ition with United Space Alliance in Huntsville Alabama