<%BANNER%>

Optimization of a Morphing Wing Geometry Using a Genetic Algorithm with Wind Tunnel Hardware-in-the-Loop

Permanent Link: http://ufdc.ufl.edu/UFE0021478/00001

Material Information

Title: Optimization of a Morphing Wing Geometry Using a Genetic Algorithm with Wind Tunnel Hardware-in-the-Loop
Physical Description: 1 online resource (58 p.)
Language: english
Creator: Boria, Frank J
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: aerodynamic, air, aircraft, algorithm, biorobots, boria, change, composite, dic, ga, genetic, hardware, loop, mav, morph, morphing, optimization, shape, uav, vehicle, vic, wing
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Aerospace Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advancements of compliant actuator technologies and flexible composite structures have enabled true wing morphing capabilities. The resultant effect of these new capabilities is optimal wing shapes for all flight regimes. With the ever-increasing complexity of these morphing wing structures comes the challenge of proper shape management for achieving desirable, real-time flight performance. Using a wind tunnel with sting balance as a hardware-in-the-loop objective function, a genetic algorithm is used for optimizing multiple design variable morphing wing structures. This evolutionary approach to optimization provides a more efficient means of convergence over that of other conventional optimization techniques. There are four primary subsystems utilized in the experimental setup. These include the optimization software, the servo controller, the wind tunnel, and the data acquisition software. Through the use of a simple text-format software, data is passed between each of these subsystems, thereby automating the procedure. A visual image correlation system is used independently during testing to determine the three-dimensional wing shapes of noteworthy servo positions. To demonstrate the procedure, a two design-variable morphing wing is fabricated with the capability of altering maximum camber and trailing edge reflex. Two experiments are conducted for maximizing the lift coefficient and maximizing efficiency. These experiments resulted in optimal servo position versus angle of attack shape functions. Having set the wing shape as a flat plate and again as a medium camber wing with reflex, data is again collected for comparison. These experiments resulted in a comparatively benign stall characteristic, a 63.4% increase in lift coefficient, and up to 62.1% increase in maximum efficiency over that of the fixed wing with reflex. Plotting the performance of the morphing wing with that of the two arbitrary fixed wing shapes illustrates the dramatic improvement achievable by morphing wing structures. Utilizing these procedures allows for real-time adaptive wing morphing to adjust to the instantaneous changes of flight attitude and wind conditions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Frank J Boria.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Ifju, Peter.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021478:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021478/00001

Material Information

Title: Optimization of a Morphing Wing Geometry Using a Genetic Algorithm with Wind Tunnel Hardware-in-the-Loop
Physical Description: 1 online resource (58 p.)
Language: english
Creator: Boria, Frank J
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: aerodynamic, air, aircraft, algorithm, biorobots, boria, change, composite, dic, ga, genetic, hardware, loop, mav, morph, morphing, optimization, shape, uav, vehicle, vic, wing
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Aerospace Engineering thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advancements of compliant actuator technologies and flexible composite structures have enabled true wing morphing capabilities. The resultant effect of these new capabilities is optimal wing shapes for all flight regimes. With the ever-increasing complexity of these morphing wing structures comes the challenge of proper shape management for achieving desirable, real-time flight performance. Using a wind tunnel with sting balance as a hardware-in-the-loop objective function, a genetic algorithm is used for optimizing multiple design variable morphing wing structures. This evolutionary approach to optimization provides a more efficient means of convergence over that of other conventional optimization techniques. There are four primary subsystems utilized in the experimental setup. These include the optimization software, the servo controller, the wind tunnel, and the data acquisition software. Through the use of a simple text-format software, data is passed between each of these subsystems, thereby automating the procedure. A visual image correlation system is used independently during testing to determine the three-dimensional wing shapes of noteworthy servo positions. To demonstrate the procedure, a two design-variable morphing wing is fabricated with the capability of altering maximum camber and trailing edge reflex. Two experiments are conducted for maximizing the lift coefficient and maximizing efficiency. These experiments resulted in optimal servo position versus angle of attack shape functions. Having set the wing shape as a flat plate and again as a medium camber wing with reflex, data is again collected for comparison. These experiments resulted in a comparatively benign stall characteristic, a 63.4% increase in lift coefficient, and up to 62.1% increase in maximum efficiency over that of the fixed wing with reflex. Plotting the performance of the morphing wing with that of the two arbitrary fixed wing shapes illustrates the dramatic improvement achievable by morphing wing structures. Utilizing these procedures allows for real-time adaptive wing morphing to adjust to the instantaneous changes of flight attitude and wind conditions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Frank J Boria.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Ifju, Peter.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021478:00001


This item has the following downloads:


Full Text





OPTIMIZATION OF A MORPHING WINTG GEOMETRY USING A GENETIC
ALGORITHM WITH WINTD TUNNEL HARDWARE-IN-THE-LOOP


















By

FRANK JO SEPH BORIA


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
































O 2007 Frank Joseph Boria




























To my bride, Rikki









ACKNOWLEDGMENTS

This work has been made possible through funding and facilities provided by the

University of Florida and BioRobots, LLC. Many thanks in particular go to Dr. Peter Ifju for his

guidance, leadership, and ability to meld science with low-quality humor. Another thank you

goes to Dr. Roberto Albertani for his guidance on wind tunnel operation and his incessant

support of the metric system, despite its obvious illogical basis (long live the slug). And yet

another thank you goes to Dr. Bruce Carroll for so strongly supporting my work during this final

task.

I thank Bret Stanford for his thoughts and input throughout my research, whose insight

and experience helped to create the framework of these experiments. I am very appreciative of

Scott Bowman for his design, fabrication, and programming of a sophisticated morphing wing

controller that was used to conduct this research.

I am forever grateful and indebted to Frank and Joanne Boria for raising me with love and

compassion while teaching me faith, respect and commitment. They are both remarkable parents

and phenomenal role models. I will always look to them for guidance.

To Julie Badger, Carl Boria, and Elizabeth Nettles, I am thankful for their strength of

support, their being there to listen, and their reminding me of birthdays and anniversaries. To

you, my brother and sisters, I gratefully dedicate the last figure of the fifth chapter.

I would like to thank my children, Addie Reith for contributing her skills in photography,

Olivia Reith for her remarkable contribution of digital photo editing, and Christian Boria for

thinking that morphing airplanes are really, really cool. I love them all.

Finally, I thank my bride, Rikki, for her never-ending support, love, and inspiration. She

has brought me great j oy, true happiness, and has taught me to live life with abandon. I am

forever thankful to her for being genuine, refreshingly enthusiastic, and mine.












TABLE OF CONTENTS


Page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ........._._ ...... .__ ...............7....


LIST OF FIGURES .............. ...............8.....


AB S TRAC T ............._. .......... ..............._ 10...


CHAPTER


1 INTRODUCTION .............. ...............12....


Emulating Biological Structures ................. ...............12........... ....
Optimization of Morphing Wing Structures ................. ...............13........... ...
Motivation............... ...............1
Overvi ew ................. ...............14.................


2 LITERATURE REVIEW .............. ...............16....


Emulating Desirable Wing Shapes .............. ...............16....
Actuation of a Morphing Wing Structure ................. ............ ............... 17. ...
Optimization Techniques for Acquiring Desired Wing Shapes .............. ....................18
Hardware for Data Collection ................. ...............19................
Wind Tunnel Uncertainty Analysis .............. ...............19....
Visual Image Correlation System ................. ...............19................

3 EXPERIMENTAL SETUP AND DATA ANALYSIS ................... ...............2


W ind Tunnel .............. ...............20....
Servo Controller............... ...............2
Software Description .............. ...............21....
Visual Image Correlation System .............. ...............22....

4 ONE DESIGN VARIABLE MODEL .............. ...............27....


Model Description and Purpose ................. ...............27................
A Study of Aerodynamic Hysteresis .......__ ..........._ .......___ ....._ .........28
Initial Alpha Sweep for Predicting Convergence .............. ...............29....
Validation of Optimization Procedure ....._____ ................... ............3

5 OPTIMIZATION OF A HIGH-FIDELITY MORPHING WINTG .............. ..................40


Model Description .............. ..... ...............40.
Comparative Lift and Efficiency Curves ................ ...............41........... ...












Maximizing Coeffieient of Lift Curve ................. ...............41........... ...
Maximizing EfH ciency Curve............... ...............43.


6 CONCLUSION AND RECOMMENDATIONS FOR FUTURE WORK ................... .....52


Conclusion ............... ........ ........... .........5
Recommendations for Future Work. ........._._.. ....__. ...............53..


LIST OF REFERENCES ........._.___..... .___ ...............55....


BIOGRAPHICAL SKETCH .............. ...............58....










LIST OF TABLES


Table Page

4-1 Small coefficient of variation between the ten base data sets and the random data sets
ensures minimal effect of aerodynamic hysteresis on the coefficient of lift. ................... .33

4-2 Drag data show acceptably low variation between the ten base data sets and the
random data sets............... ...............33..

4-3 Efficiency is found to have a variance of the same magnitude as drag, which is
sufficient for the procedure to correctly converge upon the optimum servo position.......33










LIST OF FIGURES


Figure Page

3-1 Closed-loop wind tunnel at the University of Florida is pictured with a 24-inch test
secti on ........... ............... ...............24.......

3-2 Digital protractor is used to measure the angle of the sting balance that mounts to the
U-shape model arm. ............. ...............24.....

3-3 The servo controller (background) is used to set and hold servo positions, while the
serial port adapter (foreground) converts signals from RS232 to TTL. ............................24

3-4 Data exchange between software applications. ............. ...............25.....

3-5 The VIC system hardware setup about a wind tunnel test section.. ............ ..................26

4-1 The one design variable, rectangular planform model used to validate the
optimization process has a single point of actuation that alters camber. ...........................34

4-2 The full field, three-dimensional image, developed by the VIC system, is overlaid on
an image of the morphed wing illustrating deformation in the Z-direction. ...................34

4-3 Wing shape attained by servo position 1 with velocity at 0 and 16 m/s at 100 AOA.
The top left image shows the wind off Z-coordinates of the wing in millimeters. The
top right image shows the wind on wing deformation ................. .......... ...............35

4-4 Wing shape attained by servo position 250 with velocity at 0 and 16 m/s at 100 AOA....35

4-5 Plot of a series of airfoil shapes attained as a function of servo position. .........................36

4-6 Results of an aerodynamic hysteresis study depict the servo scatter and data
coefficient of variance for multiple servo positions. ................ ................. ..........37

4-7 Lift coefficient, drag coefficient, and efficiency versus angle of attack are plotted for
several servo positions of the one design variable model ................. ................ ...._.3 8

4-8 Resulting data illustrate the effectiveness of the optimization procedure. A) Initial
population. B) Final population. C) Coefficient of lift servo scatter. D) Morphing
wing performance comparison for lift coefficient. E) Efficiency servo scatter.
F) Morphing wing performance comparison for efficiency. ............. .....................3

5-1 The two design-variable model morphs the wing geometry through change of camber
and reflex. ............. ...............45.....

5-2 The full field, three-dimensional image, developed by the VIC system, is overlaid on
an image of the morphed wing illustrating deformation in the Z-direction. ...................45










5-3 Morphed wing surface (Servo Positions: 250,230) with airfoil slices from the 20%
and 80% span yields a negative camber and negative reflex ................. .....................46

5-4 Morphed wing surface (Servo Positions: 1,1) with airfoil slices from the 20% and
80% span yields a positive camber and positive reflex. ............. .....................4

5-5 An illustration of the intermediate wing shapes from the 20% span shows the
morphing capability of the two design-variable model. ................ ............... ...._...47

5-6 The lift coefficient and efficiency is plotted using set flat plate and reflex wing
geom etries. .............. ...............47....

5-7 These sample plots illustrate the convergence of the genetic algorithm upon the
optimal servo position for maximum lift at 100 angle of attack. ............. ....................48

5-8 Morphing wing data for optimized lift coefficient. A) The morphing wing
performance as compared to the fixed reflex wing and flat plate. B) The optimal
servo positions as a function of angle of attack with the corresponding wing shape........49

5-9 These sample plots illustrate the convergence of the genetic algorithm upon the
optimal servo position for maximum efficiency at 100 angle of attack. ................... .........50

5-10 Morphing wing data for optimized efficiency. A) The morphing wing performance
as compared to the fixed reflex wing and flat plate. B) The optimal servo positions
as a function of angle of attack with the corresponding wing shape. ............. .................5 1









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

OPTIMIZATION OF A MORPHING WINTG GEOMETRY USINTG A GENETIC
ALGORITHM WITH WINTD TUNNEL HARDWARE-IN-THE-LOOP

By

Frank Joseph Boria

August 2007

Chair: Peter Ifju
Major: Aerospace Engineering

Advancements in compliant actuator technologies and flexible composite structures have

enabled true wing morphing capabilities. The resultant effect of these new capabilities is optimal

wing shapes for all flight regimes. With the ever-increasing complexity of these morphing wing

structures comes the challenge of proper shape management for achieving desirable, real-time

flight performance.

Using a wind tunnel with sting balance as a hardware-in-the-loop obj ective function, a

genetic algorithm is used for optimizing multiple design variable morphing wing structures. This

evolutionary approach to optimization provides a more efficient means of convergence over that

of other conventional optimization techniques.

There are four primary subsystems utilized in the experimental setup. These include the

optimization software, the servo controller, the wind tunnel, and the data acquisition software.

Through the use of a simple text-format software package, data is passed between each of these

subsystems, thereby automating the procedure. A visual image correlation system is used

independently during testing to determine the three-dimensional wing shapes of noteworthy

servo positions.









To demonstrate the procedure, a two design-variable morphing wing is fabricated with the

capability of altering camber and trailing edge reflex. Two experiments are conducted for

maximizing the lift coefficient and maximizing efficiency. These experiments result in optimal

servo position versus angle of attack shape functions. Setting the wing shape as a flat plate and

again as a medium camber wing with reflex provides comparative data. These experiments

resulted in a comparatively benign stall characteristic, a 63.4% increase in lift coefficient, and up

to 62. 1% increase in maximum efficiency over that of the Eixed wing with reflex. Plotting the

performance of the morphing wing, with that of the two arbitrary Eixed wing shapes, illustrates

the dramatic improvement achievable by morphing wing structures. Utilizing these procedures

allows for real-time adaptive wing morphing to adjust to the instantaneous changes of flight

attitude and wind conditions.









CHAPTER 1
INTRODUCTION

No longer are aircraft confined to fixed wing geometries suiting a single flight regime.

The advancements of actuation and material technologies have enabled true morphing

capabilities. The resultant effect of these new capabilities is optimal wing shapes for all flight

regimes. With the ever-increasing complexity of these morphing wing structures comes the

challenge of proper shape management for achieving desirable, real-time flight characteristics.

The work presented in this thesis is one approach to such a challenge.

Emulating Biological Structures

The study of ornithology shows that each species of bird has evolved to become uniquely

specialized within their habitat. From the perspective of the aerodynamicist, designing an

aircraft to emulate the performance of the falcon for high speed dives, the eagle for highly

efficient soaring, the hovering capability of the humming bird, or the agility of the swift would

be of great value. Ultimately, however, melding each of these desirable flight characteristics on

a single aircraft is the true obj ective. Such a versatile vehicle could achieve remarkably high lift,

for maximum payload capacity, high speed, to quickly move from waypoint to waypoint, and

high efficiency. Utilizing a morphing wing structure for asymmetric deformations would yield

the ability to maneuver with high agility through complicated urban environments. Such a

vehicle could also maintain stability during the most volatile of flight conditions, such as erratic

and gusting winds among buildings and structures.

Utilizing thin plate composite materials with multiple actuation points, it is possible to

create a morphing wing capable of complex deformation. A noteworthy study on similar

materials with regard to creep is presented by Richard Thomas Haenel [1], which details the









evaluation of creep with respect to material deformation. Careful design of such a wing will

yield configurations that significantly outperform conventional Eixed wing geometries.

Optimization of Morphing Wing Structures

When a given structure is created to emulate the complexities of what exists in nature, a

method must also be developed to command the structure to take a desired shape. A high Eidelity

morphing wing with several actuation points will require a complex control system. The once

rigorous determination of aerodynamic performance for conventional Eixed wing geometries has

become an even greater undertaking in multi-variable shape optimization.

While the Hield of computational fluid dynamics continues to improve, modeling complex

structures of composite makeup remains expensive. It is also difficult to accurately model the

aeroelastic effects on such a structure when it is under load. This becomes increasingly the case

when utilizing thin plate wings. While the use of a wind tunnel may also be expensive, it

provides a source for aerodynamic force data, which take full account of wing deformations due

to aeroelastic effects.

Regardless of the method used for determining aerodynamic forces, a procedure for

optimization must be properly matched to the capability of the wing structure. While a response

surface approach may be well suited for morphing wings with minimal design variables, high-

fidelity morphing wing structures may be more efficiently optimized through the use of a genetic

algorithm. The genetic algorithm is better capable of finding a global optimum more efficiently

than many strategies.

Motivation

The use of a genetic algorithm for optimizing high-fidelity, multiple design variable

morphing wing structures provides a more efficient means of convergence over that of other

conventional optimization techniques.









For example, collecting data for a one design variable wing of 10 position increments over

an alpha sweep of 10 angles of attack requires 100 data points. To ensure data validity, these

same data points are typically collected multiple times to ensure precision, which increases the

number of required data points to a minimum of 200. Increasing the models capability through

the use of two design variables increases the number of required data points to 2,000.

Ultimately, the required number of data points increases by an order of magnitude with each

additional design variable. Of course this analysis assumes only 10 position increments are

possible, while the actuators used for this study have 250 position increments.

The use of response surface techniques is a reasonable approach for wings with up to three

and even four design variables. But as the number of design variables grows, it becomes

inefficient. Finding the optimal position of the determined response surface becomes more

difficult to solve. The complexity of the response surface also raises concerns for convergence

upon false, local optimums. The use of this and similar techniques for finding maximum

performance shapes for high-design variable morphing wing structures is ineffective.

Overview

To establish an effective procedure for determining optimal wing shape of a multi-variable

morphing wing, the use of a genetic algorithm with wind tunnel hardware-in-the-loop is

implemented. This procedure is used to determine a specific desired flight characteristic, such as

maximum lift or maximum efficiency. By conducting these optimization tests at several angles

of attack, it becomes possible to obtain optimal wing shapes throughout the range of angles of

attack. Utilizing these data in a control system make it possible to automate real-time wing

morphing for optimal performance during flight. It is the intent of this thesis to present and

implement such a procedure with its proficiencies and detriments.










This thesis consists of six chapters. The first chapter outlines the intent and scope of the

investigation. The second chapter details a literature review of prior work related to morphing

wing structures, optimization procedures, and wind tunnel operation and data collection. The

third chapter describes the experimental setup with detail on the wind tunnel and the visual

image correlation system hardware. Also described in this chapter are the optimization software,

the interaction among the programs, and the experimental procedure for error minimization. The

fourth chapter describes the preliminary one design variable model used to validate the

optimization procedure. The fifth chapter describes the multi design variable model used to

illustrate the improved performance the morphing wing has over that of conventional fixed

wings. The sixth chapter closes with a conclusion of the investigation and suggests

recommendations for future work.









CHAPTER 2
LITERATURE REVIEW

Through the use of unmanned aerial vehicles it becomes possible to conduct dangerous

missions without endangering the life of a pilot. These vehicles are becoming smaller with

missions that require significant agility for navigating complex urban environments. To see a

bird navigate through such an environment will inspire the concept of its emulation. Through the

ever-maturing field of material science, the morphing capabilities of an aircraft wing are more

capable of simulating a bird in flight.

Emulating Desirable Wing Shapes

The complex system of structural members and joint articulation of a bird' s wing [2] is

difficult to duplicate mechanically with regard to function as well as low weight and high

durability. Studies associating wing shape to function are conducted to differentiate

requirements for static aerodynamics, physiology, and flapping control [3].

Understanding this differentiation leads to morphing wing structures of less complexity

than that of the birds wing. For example, while birds have many layers of feathers that are

moved around to adjust to the specific maneuver they need to perform [4], morphing wing

structures can be fabricated of thin sheet materials, which are manipulated to yield the desired

wing shape.

Currently, many unmanned aerial vehicle designs are based on a flexible wing design

developed at the University of Florida [5],[6]. These thin, flexible wings are beneficial for

dampening the effects of wind gusts and increasing stability during adverse flight conditions.

With regard to wing morphing, they are also easily manipulated for desired control and overall

flight performance using minimal actuation points. An example of this is morphing in the form

of asymmetric wing twist making it is possible to surpass the roll rates of conventional aileron









configurations [7]. Another example is the exercise in design optimization conducted by Gano et

al. [8],[9] for a single, high endurance wing that separates into a second wing for greater agility.

Much work has been done to assess the flight characteristics of these flexible wings.

NASA has performed a control assessment and simulation of a micro air vehicle with aeroelastic

wings which adapt to the disturbances during flight [10]. Morphing wing weight equations and

an approach to size morphing aircraft [1 1] are developed to replace current methods of general

heuristics based on fixed wing aircraft. Also, extensive wind tunnel tests on a variety of flexible

wing configurations showed a decrease in wing incidence, an increase in dihedral and a shift in

maximum camber position that results in a favorable change in pitching moment and lift.

Ultimately, smoother flight characteristics are achieved over that of conventional fixed wings

[12]. A variety of experimental flow tests are conducted on these wings using surface oil flow

visualization, laser based flow visualization, and particle image velocimetry with results that

validate previously determined CFD results [13].

Actuation of a Morphing Wing Structure

Advancements in actuator technologies for these compliant wing structures, such as Macro

Fiber Composites (MFC), Electroactive Polymers (EAP), and Smart Memory Alloys (SMA), are

yielding lightweight, capable devices. The thin, compliant MFC is a type of piezoceramic

composite actuator that is constructed of orthogonal layers of unidirectional piezoceramic fibers

and copper electrodes encased in layers of acrylic and Kapton [14]. These actuators have been

reported to produce high strain levels on the order of 2000 CLE [15]. Alternatively, the EAP

actuator is similar to muscle tissue with regard to stress and force capability. This actuator is

capable of large-scale linear motion [16]. Another source of actuation is through the use of an

SMA mechanism. These mechanisms consist of a metallic alloy that act as linear actuators by

contracting when heated and returning to their original shape when cooled [17]. This actuator is










lightweight and very effective although precaution must be taken to prevent variations in

temperature, which will affect its performance.

These, and many other, compliant forms of actuation have been developed and utilized for

morphing wing structures. Extensive work on adaptive shape change problems, which determine

methods for morphing a given curve or shape into a target curve or shape, is providing solutions

with a minimal number of actuators [18]. This work leads to the challenge of morphing a wing

not to a single target shape, but rather to adaptively morph the wing real-time during flight to

adjust to the instantaneous changes of flight attitude and wind conditions.

Optimization Techniques for Acquiring Desired Wing Shapes

The use of evolution strategies for optimization of complex design problems began back in

the early sixties [19] and has been gaining momentum over the past fifteen years. Such

strategies emphasize selection, recombination and mutation of a given population to determine

the makeup of the next population set [20]. Each of the members of this new generation is then

evaluated for fitness. Poor performing members are rej ected from the next generation while high

performing members are retained. This results in convergence upon an optimal solution after

multiple generations.

The use of a genetic algorithm (GA) has advantages over other techniques when searching

within a complex domain. Due to its evolutionary tactics, the GA is resistant to convergence on

local optimums making it better suited for finding true global optimums. It is also capable of

working with many design variables, which can be discrete, continuous, or mixed [21], [22],

[23].

As morphing wing structures become more complex, the GA solver becomes better

preferred. Studies have already shown the effective use of the GA to optimize fixed airfoil and

wing shape [24] and to better develop composite wings for desired structural characteristics [23].









Another study used a sequential surrogate optimization approach with hardware-in-the-loop as

its objective function [25]. While the study was specific to determining the maximum walking

speed of a humanoid robot [26], this use of hardware-in-the-loop is well suited for determining

optimal wing shape for a morphing wing.

Hardware for Data Collection

Wind Tunnel Uncertainty Analysis

By testing the morphing wing in a wind tunnel with a six degree-of-freedom sting balance

to measure forces and moments, an array of performance data is collected. Using the sting

balance as a hardware-in-the-loop objective function, the GA morphs the wing shape for each

member of a given generation. A fitness value for each member is calculated from data received

from the sting balance. Critical to the use of any hardware-in-the-loop optimization procedure is

a determination of the measurement uncertainty for evaluation of data quality. Such an

evaluation of wind tunnel and sting balance performance is documented in [27] and [28].

Visual Image Correlation System

Using the digital image correlation (a.k.a. visual image correlation or VIC) system, a three-

dimensional non-contact full-Hield measurement of strain and displacements can be attained for a

given structure of composite material [29]. It has successfully been used to determine strains in

specimens of fiber reinforced polymer composites [30]. This system is well suited to determine

wing shapes for differing morphed positions while under flight loads. The supporting principles,

specifications, calibration procedures, and applications are well documented in [31].









CHAPTER 3
EXPERIMENTAL SETUP AND DATA ANALYSIS

There are four primary subsystems utilized in the experimental setup. These include the

optimization software, the servo controller, the wind tunnel, and the data acquisition system.

Through the use of simple text-format software, data is passed between each of these

subsystems. This cooperative scheme of utilizing previously independent software programs

provides a means for automating the optimization procedure. The visual image correlation

system is used independently during testing to determine the three-dimensional wing shapes of

noteworthy servo positions.

Wind Tunnel

The primary piece of hardware used for this optimization procedure is the closed-loop

wind tunnel model 407B (Figure 3-1), which is manufactured by Engineering Laboratory

Design. This low-speed, low-turbulence tunnel has been reported to have less than 0.2%

turbulence [13] along the centerline of its 33-inch test section. The tunnel is capable of

producing airspeeds of 2 m/s to 45 m/s with its 250HP motor and 2-stage axial fan.

Extending from the sidewall of the test section is a U-shape model arm (Figure 3-2) that

supports the 6 degree-of-freedom sting balance and rotates the balance/model assembly to the

commanded angle of attack. It is the sting balance that provides all force and moment data to the

software through a National Instruments SCXI 1000 frame. The sting balance, produced by

Aerolab model 01-15, is capable of measuring loads on the order of 0.01 N [13].

Several LabVIEW virtual instruments (VI) are used to control the wind tunnel speed,

temperature, model angle of attack, and sting balance voltage data acquisition. As detailed later

in this chapter, automation of the optimization procedure is achieved by calling these subroutine

VIs from within a primary VI which exchanges loads data with the genetic algorithm software.









Servo Controller

Each morphing wing developed for these experiments utilizes Futaba S3102 servo

actuators to attain the surface deformation. These servos supply 4.6 kg-cm of torque and a rate

of 600 in 0.2 seconds. Using Matlab, servo number and position are commanded through the

serial port in a 1200-baud rate, 8-bit, 1-stop, and no-parity form. A sub-module converts the

RS232 signal to transi stor-to-transistor logic. Receiving the signal on the morphing wing

structure is a custom servo controller (Figure 3-3) designed and fabricated by Scott Bowman.

This controller features 8 kilobytes of flash memory, a 16-bit high-resolution servo timer, and an

8 MHz ATMEL microprocessor. This servo controller is designed to set and hold commanded

servo positions while the wing undergoes aerodynamic loading.

Software Description

The two primary software packages used for this study are LabVIEW for wind tunnel

control and data acquisition and Matlab for wing morphing control and optimization. A flow

chart of how the information is passed between the programs is shown in Figure 3-4. To initiate

the optimization procedure, an initial population is prescribed to the genetic algorithm. When

the algorithm is started, the first member of the initial population is sent to the subroutine that

commands the servo controller. The servo controller then sets and holds each servo position

thereby morphing the wing. Simultaneously, the servo positions are sent to a text file. At this

point, the genetic algorithm waits for the force and moment data that is associated with the servo

positions to be added to the text file. It determines when this happens by monitoring the text file

size.

While the optimization software is on hold for data, the wind tunnel is brought up to the

required velocity. After the angle of attack is set, the force and moment data are determined

from the sting balance. The text file previously created by the optimization software is accessed









and the data is appended to the same line. The resulting line will consist of servo positions

followed by the lift coefficient, drag coefficient, coefficient of side force, coefficient of pitching

moment, and coefficient of yawing moment. The rolling moment is omitted because the system

yields invalid data. This is of no detriment to these sets of experiments since rolling moment has

no contribution in force and is not utilized as a constraint. At this point, the data acquisition

software waits for the next set of servo positions to be added to the text file. As with the

optimization software, it does this by monitoring the file size. When the new servo positions are

added to the text file, the file size will increase thereby triggering the data acquisition software

for the next set of data.

After the force and moment data have been added to the text file, the optimization software

identifies the increase in file size and retrieves the data. The next servo positions are

commanded, and thereby begin the next cycle.

As with any sting balance, there does exist the concern for signal drift. This sting balance

contains a network of strain gauges that is used to measure the elastic strain of the balance.

These strain gauges are susceptible to signal drift due to temperature variations caused by both

environment and self-heating. To manage the signal drift, the wind tunnel software counts the

number of data sets collected. After the fifteenth data set, the tunnel velocity is set to zero and

the model arm is rotated to the original angle of attack. At this point, the data acquisition

software performs an offset nullification of the strain gauge voltages. The tunnel is then brought

back up to the velocity and angle of attack required by the experiment. This simple procedure

ensures data quality and is set to coincide with the genetic algorithm population size of fifteen.

Visual Image Correlation System

Determining the shape of the morphed wing is essential to quantitatively validate the

obtained solutions. Through the use of the visual image correlation (VIC) system, a non-contact









full-Hield shape and deformation tool, the wing shape can be captured and evaluated for any

given morphed configuration. A typical experimental setup with the VIC system is illustrated in

Figure 3-5. This process was developed in the mid 1990s by Helm with a reported resolution on

the order of + 0.05 mm [32].

The specimen is first uniformly painted with a low-luster color, and then painted with a

highly contrasting, random speckle pattern. Two digital cameras carefully focused on the

specimen. The cameras are triggered such that an image from each is acquired instantaneously.

A plate with high contrast dots of known diameter and spacing is used to calibrate the camera

system. Photos of the plate in different locations of the image frame and at varying angles and

rotations are taken and processed for calibration. After the calibration procedure is complete, the

precise orientation of one camera to the other is known. At this point, the system is now ready to

photograph the specimen. From these images, the software is capable of developing the full-

Hield three-dimensional shape of the specimen using stereo triangulation.

The system is also used to determine deformation of a specimen under load. By first

taking a no-load reference image of the specimen, then taking another image of the loaded

specimen, the software is capable of calculating the displacement Hield. This is useful for

determining how much deformation a wing undergoes when in flight.























Figure 3-1.


The closed-loop wind tunnel at the University of Florida is pictured with a 24-
inch test section (image provided by Michael Sytsma).


Figure 3-2.


A digital protractor is used to measure the angle of the sting balance that mounts
to the U-shape model arm.


Figure 3-3.


The servo controller (background) is used to set and hold servo positions, while
the serial port adapter (foreground) converts signals from RS232 to TTL.












Loop 1


Sing Balance
Data
Acquisition


Loop 2


Genetic
Algorithm


Figure 3-4.


These flowcharts illustrate how data is appended to the shared text file.


Servo
Controller











































Figure 3-5.


This illustration depicts the VIC system hardware setup about a wind tunnel test
section.









CHAPTER 4
ONE DESIGN VARIABLE MODEL

Model Description and Purpose

To validate the procedures for data collection, a simple one-design-variable model is

utilized. This model is a thin plate wing of rectangular planform with a single point of actuation

that changes the wings camber. The wing consists of a single layer, plain weave carbon fiber

skin and horizontally aligned unidirectional spars that run from wingtip to wingtip every one

inch from the leading edge (illustrated in Figure 4-1). These spars transfer loads such that when

the wing shape is changed at the root, the same shape will be attained at the tips effectively

resulting in an extruded airfoil along the span of wing.

An attachment point at the leading edge permits rotation only. An attachment point at the

70% chord permits both rotation and longitudinal motion. The unconstrained trailing edge allows

for significant deformation when under load. This deformation is an aeroelastic effect caused by

the interaction among inertial, elastic, and aerodynamic forces on the wing. Typical CFD is

expensive when attempting to model this aeroelastic effect, let alone optimize a morphing wing,

and is why this tunnel-in-the-loop procedure is of particular value. Using the VIC system, the

wing shape is determined. It is shown as a topographical overlay on an image of the wing itself

in Figure 4-2. The wind-on deformations, and the airfoils associated with each, are plotted in

Figure 4-3 for servo position 1 (a negatively cambered wing) and Figure 4-4 for servo position

250 (a highly cambered wing).

The model is capable of altering the maximum camber between -7% through 19%.

Utilizing the VIC system, 9 servo positions are selected and photographed to determine the range

of attainable wind-off wing shapes. The normalized airfoils are plotted in Figure 4-5, which help

to illustrate the magnitude of camber change.










By collecting data for set servo positions over an array of angle of attacks, the performance

trends of the model become clear. The now predictable performance of this wing is used to

determine whether proper convergence of the genetic algorithm occurs. When it is confirmed

that convergence is achieved repeatedly, the process can then be used to optimize performance

for complex, high fidelity wings capable of sophisticated wing morphing.

A Study of Aerodynamic Hysteresis

Aerodynamic hysteresis occurs when the detached, low Reynolds number flow

inconsistently reattaches thereby resulting in varied performance. Repeatability of wing

performance is essential when conducting optimization tests since the genetic algorithm will

semi-randomly alter servo position while conducting its search. In an effort to determine if

aerodynamic hysteresis exists, a series of tests are conducted.

Five servo positions are used to represent the extremes and intermediate positions within

the range of servo motion. While holding the angle of attack and airspeed constant, ten

consecutive data sets are collected for each of the five servo positions. The standard deviation is

calculated for each of these data sets. This standard deviation is a measure of instrument

precision and is independent of aerodynamic hysteresis.

Data is then collected as the servo positions are varied between the extremes and

intermediate servo positions to measure any effect caused by aerodynamic hysteresis. The data

is sorted into common servo positions and the standard deviation is calculated. If the standard

deviation of the original data sets and the varied data sets coincides within a common range,

aerodynamic hysteresis will have been proven not to exist.

Tests are conducted at 00, 100, and 22.50 angles of attack, which represent low drag, high

lift, and near-stall flight regimes. Within Figure 4-6, eight plots are provided to illustrate the

results of the study. In Fig. 4-6 (A), the servo position variation used to induce aerodynamic










hysteresis is shown. Several sweeps from maximum to minimum and intermediate servo

positions are performed to measure the effect of the hysteresis on the lift and drag coefficients.

The drag coefficient versus servo position, shown in Fig. 4-6 (B), is provided to show the

precision of the data. It is selected since it is the least precise of all data provided by the

instrumentation; this is typically an order of magnitude less than that of the lift coefficient. The

subplots shown in Fig.4-6 (C-H) are magnified data sets for lift and drag coefficients at servo

positions 1, 125, and 250. While the plots clearly show aerodynamic hysteresis does exist, the

standard deviation within a given data set is acceptable with a coefficient of variance of no more

than 2.3% for coefficient of lift (Table 4-1), 2.9% for coefficient of drag (Table 4-2), and 3.4%

for efficiency (Table 4-3).

This study shows the effect of the aerodynamic hysteresis is minimal and has little

influence on the data collected. It is in the author's opinion, as supported by this study, that the

effects due to aerodynamic hysteresis can be assumed negligible.

Initial Alpha Sweep for Predicting Convergence

To validate the procedure for determining the optimum wing shape, a preliminary set of

data was collected. The servo used in the model is capable of 1200 rotation, which is

commanded by the servo controller using position points 1 through 250. This range of servo

position is divided into 9 positions that are used to populate lift, drag, and efficiency plots. For

each servo position, data is collected for angle of attacks ranging from -50 through 300 in 2.50

increments (Figure 4-7).

In doing this it can be seen that the lift curves systematically increase as camber increases.

The drag plots show an interesting inversion in trend shape as the camber increases, which is

likely due to the aeroelastic effects of the unconstrained trailing edge. Finally, the efficiency plot









shows a complex trend, which would have been difficult to predict. By plotting the convergence

data against these plots it becomes clear if the genetic algorithm is finding the true optimum.

Validation of Optimization Procedure

Using the genetic algorithm with the wind tunnel hardware in the loop, tests for

determining servo positions associated with optimal lift and efficiency are conducted. The

optimization procedure is first tested multiple times at set angles of attack to ensure repeatable

convergence on the optimum position. This demonstrates the data collected to be both

sufficiently accurate and precise for this procedure. The optimization procedure is then tested at

several other angles of attack and is compared to the previously acquired alpha sweeps presented

in Figure 4-7. Showing that the converged upon optimal values coincide or exceed predicted

values validates the optimization procedure.

Since the lift forces yield a lower standard deviation than that of the drag forces, it is used

for the initial convergence tests. When calculating efficiency, the lift is divided by the drag

resulting in a much greater coefficient of variance than that of the two individual components.

These tests are conducted after convergence for maximum lift is achieved.

The initial population is user defined and is equally distributed throughout the servo

position range. Setting the initial population in this manner, shown in Figure 4- 8 (A),

significantly reduces the required number of generations to convergence. This occurs because

the lift and efficiency curves as a function of servo position are simple polynomial curves

without complex local minimums. A sample of the final generation is plotted in Figure 4- 8 (B),

which shows convergence on an optimum with subtle deviations from that optimum.

The genetic algorithm selected servo positions, or servo scatter, are plotted in Figure 4- 8

(C) and (E). Note the cluster of data points near the optimum servo positions where the

algorithm narrowed its search. This plot is useful to visually ensure the algorithm searched









sufficiently throughout the range of servo positions. A large gap in the scatter will raise

concerns that the true global optimum may have been missed.

By setting the angle of attack to 100, high lift and drag forces are ensured. Three

optimization experiments are conducted to determine which servo position maximizes lift. As

shown in Figure 4- 8 (D), each experiment converged on a value that exceeded or matched the

predicted value. All three experiments converged to servo position 250, which is associated with

the greatest camber. Due to the variance in lift data, the lift coefficient associated with the

optimal servo position will never converge closer than 2.3% of one another. This does not,

however, prevent the genetic algorithm from properly converging on that optimal servo position.

Due to the low coefficient of variance and the rapid convergence on the optimum servo

position, it is in the author' s opinion that the procedure will accurately converge to the optimum

value regardless of angle of attack. For this reason, the second sets of convergence experiments

for optimal effciency are conducted.

Continuing to hold the angle of attack at 100, three convergence experiments for

determination of maximum efficiency are conducted. Since the coefficient of variance for

effciency is much higher than that of lift, other angles of attack are tested. As shown in Figure

4- 8 (F), each of the three experiments converges on servo position 134 with a maximum

effciency of 5.56. While this value is below that of the predicted value, it does exist within the

previously calculated variance and did converge on the predicted servo position.

Because the coeffcient of drag is very low at 00 and lift will become negative at -50 angle

of attack, the efficiency may vary considerably. For this reason they are tested to ensure proper

convergence. These results are plotted in Figure 4- 8 (F). As illustrated, the optimization










procedure did accurately converge upon the correct servo positions. This is because although the

drag forces may be very low, they are consistently low; i.e. precise though not accurate.

These results support high confidence in the optimization procedure. The genetic

algorithm is capable of proper convergence despite the variance in collected wind tunnel data. It

is in the author' s opinion that the procedure will effectively determine optimal wing shapes for

wings with multiple design variables and of higher morphing capability.











Table 4-1. Small coeffcient of variation between the ten base data sets and the random data sets
ensures minimal effect of aerodynamic hysteresis on the coeffcient of lift.


0 Deg


0.14
1.36
1.83
0.34
1.17
0.72
0.19
0.83
0.66


10 Deg


0.32
2.20
1.48
0.21
0.98
0.63
0.38
0.25
0.40


22.5 Deg


1.79
2.28
2.27
0.25
1.45
1.05
0.30
1.36
0.96


Lift coefficient of variations


Base
Random
Overall
Base
Random
Overall
Base
Random
Overall


SP 1



SP 125



SP 250


Table 4-2. Drag data show acceptably low variation between the ten base data sets and the
random data sets.


O Deg


3.07
2.63
2.80
0.56
0.87
0.77
1.29
1.47
1.42


10 Deg


3.40
2.29
2.88
1.85
1.22
1.60
1.05
0.69


22.5 Deg


1.90
1.67
2.04
0.81
1.26
1.47
0.39
1.24


Drag coefficient of variations


Base
Random
Overall
Base
Random
Overall
Base
Random
Overall


SP 1



SP 125



SP 250


Table 4-3. Effciency is found to have a variance of the same magnitude as drag, which is
sumfcient for the procedure to correctly converge upon the optimum servo position.

Efficiency coefficient of variations / Deg 10Dg 25 e


Base
Random
Overall
Base
Random
Overall
Base
Random
Overall


3.05
3.13
3.39
0.54
0.46
0.94
1.22
1.00
1.39


2.23
1.88
2.98
0.92
0.97
1.04
0.47
1.11
0.94


SP 1



SP 125



SP 250







































Figure 4-1.


The one-design-variable, rectangular planform model used to validate the

optimization process has a single point of actuation that alters camber.


Z [mm]
11 307
10 264
9 22102
817805
71350B


400618
2 96321
-1 92024


Figure 4-2.


The full field, three-dimensional image, developed by the VIC system, is overlaid
on an image of the morphed wing illustrating deformation in the Z-direction (left

image has 19% camber, right image has -7% camber).





.r




w/e


lip ir


_ _~ _~


IEtsr ~I


-0.0723


-0.0287


0.0149 -0.00953
Serve Position I


0.0004C62


0.0105


0.05 I-


-0.05


- Wind On


I


-0 1


0,1 0.2 0.3 0,4 0.5 0.6 0.7 0.8 0,9


Figure 4-3.


Wing shape attained by servo position 1 with velocity at 0 and 16 m/s at 100
AOA. The top left image shows the wind off Z-coordinates of the wing in
millimeters. The top right image shows the wind on wing deformation.


I 1


Y

r


-0.0104

0.2

0.15



0.05


0,0913


0.193 -0.0110
Serve Porsition 250)


0.0303


0.0715


0 0. I 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 I
x/c

Wing shape attained by servo position 250 with velocity at 0 and 16 m/s at 100
AOA. The top left image shows the wind off form of the wing in millimeters.
The top right image shows the wind on wing deformation, also in millimeters.


Figure 4-4.


I i


--~IIIIEFC-i
ICL-Z1F I













0.2

0.15

0.1

(i 0.05


-0.05

-0.1






Figure 4-5.


0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 I
x/c


Plot of a series of airfoil shapes attained as a function of servo position.






































Servo Position 1


Servo Position I


A 250


B


0.45

0.4

0 35

0.3

0 2(


200

150

100

50


0 10 20
Data Set Number


II x Consecu~tive
Scattered



50 100 150 200 250
Servo Position


C -0.206


0.057

0.056

0D 055

0.054

0.053

0.052


-0.208

S-0.21

u -0.212

-0.214


-0.216


0.167


S0.166

e 0.165

0.164


0.061


0 0605


0 Og


Servo Position 125


Servo Position 125


0.594

0.592

0.59
0.588

0.586
0.584


9 .2

S0.198




0,'194


Servo Position 250


Servo Position 250


Figure 4-6.


Results of an aerodynamic hysteresis study depict the servo scatter and data
coefficient of variance for multiple servo positions.














- SPI
---- SP30
----SP60

~--------- SPl20

SPl50
----- SPI80
~-----SP215
-- SP250










-- SPI
---- SP30
----SP60
----SP90
-- --SPl20
SPl50
- SPl80~
----SP215
~-----SP250










---- SPI
- SP30,
~-----SP60
--------SP90
----SPl20

----SPI80
-----SP215
-- SP250


I

-l

0p 0.5

0












0.5
do


' 0.4



8 0.2


0.1











"4





za


0 5 10 15 20 25
Angle of Attack


0 5 10 15 20) 25
Angle of Attack


0 5 10 15 2025
Angle of Attack


Figure 4-7.


Lift coefficient, drag coefficient, and efficiency versus angle of attack are plotted
for several servo positions of the one design variable model.











A 250

200

S150




0 ~o






C 1.5





g .
"e


200

S150
o,
E 10

50to

0o


1.


1 23 4 567 89 1011
Member


1 2 345 67 89 1011
Member


50 100 150 200 250
Servo Position


-5 0 5 10 15 20 25 30
Andec of Attack


E 7

6

>.5


50 100 150 200 250
Servo Position


-5 0 5 10 15 20 25 30
Angle of Attack


Figure 4- 8. Resulting data illustrate the effectiveness of the optimization procedure. A) Initial
population. B) Final population. C) Coeffieient of lift servo scatter.
D) Morphing wing performance comparison for lift coefficient. E) Efficiency
servo scatter. F) Morphing wing performance comparison for efficiency.









CHAPTER 5
OPTIMIZATION OF A HIGH-FIDELITY MORPHING WING

A high-fidelity morphing wing model is fabricated to demonstrate the advantage of the

morphing wing geometry over that of conventional Eixed wing geometries. Initial coefficient of

lift and efficiency curves are developed for flat plate and reflex wings. This data is used as a

comparative tool to measure the performance of the morphing wing shapes. The optimization

procedure is conducted for maximum lift and maximum efficiency through a range of angle of

attacks. All experiments are conducted with an unconstrained genetic algorithm.

Model Description

This model is an elliptical planform, thin plate wing (Figure 5-1) consisting of a single

layer, plain weave carbon fiber skin and arced unidirectional carbon fiber spars, which follow the

outer curvature of the planform. As with the one-design-variable model, these spars transfer

loads such that when the wing shape is altered at the root, a similar shape will transfer

throughout the span of the wing. While the model is not capable of transferring the exact airfoil

from the root to the tip it does significantly change the wing shape, which is desired for this

experiment.

The model has two points of actuation, which are positioned at the quarter chord and the

trailing edge. There are two attachment points fixing the wing to the fuselage structure, which

are located at the leading edge and trailing edge. The attachment point at the leading edge

permits rotation only. The attachment point at the trailing edge permits both rotation as well as

longitudinal, linear motion. The quarter chord actuation point significantly modifies the wing

camber on the order of a 1 1%, while the trailing edge point of actuation allows for & 4% wing

reflex.









Using the VIC system, a full-field, three-dimensional image of the morphed wing shape, as

shown in Figure 5-2, is developed at 16 servo positions. To illustrate how the airfoil at the wing

root relates to the airfoil at the tip, a topographic plot with airfoil "slices" from the 20% and 80%

span (Figure 5-3) is shown. These servo positions (250,230) yield a +7.1% camber and -2.6%

reflex. For comparison, these plots are also provided for servo positions 1,1 (Figure 5-4) which

is effectively an inverted wing shape of the previous. As plotted in Figure 5-5, an array of airfoil

shapes from the 20% span is illustrated to show the morphing capabilities of the model.

Comparative Lift and Efficiency Curves

In order to determine how well the morphing wing performs, the coefficient of lift and

efficiency versus angle of attack are determined for two Eixed wing shapes. The first shape is an

approximate flat-plate wing obtained by setting servo positions to (130,100) and the second is a

reflex wing obtained using servo positions (250,110). Note the abrupt stall of the flat plate

(Figure 5-6) at approximately 110 and its poor efficiency throughout the alpha sweep. The wing

with reflex has less aggressive stall characteristics at approximately 150 angle of attack with a

much higher lift coefficient than that of the flat plate wing. It also has a much higher efficiency,

which carries over a large range of angles of attack.

This data is used as a comparative tool to measure the performance of the morphing wing

shapes. The optimization procedure for maximum lift, then maximum efficiency for several

angles of attack is performed. The results from these experiments are plotted with the two fixed

wing shapes and evaluated for improved performance.

Maximizing Coefficient of Lift Curve

The optimization procedure is conducted for six angles of attack from -50 to 200 in 50

increments, which is sufficient for evaluation of its performance. The result of the experiments

yield wing shapes that produce maximum lift for a given angle of attack. By fitting a curve









through these datum points a simple polynomial equation is calculated. It is this equation that

correlates maximum lift to angle of attack.

The genetic algorithm is capable of converging within just six generations when using a

well distributed initial population as illustrated in Figure 5-7 (bottom left). The topmost plot in

the same figure shows the rapid convergence of the mean performance with the best

performance. The rate of convergence can also be seen from the lack of diversity of the servo

scatter. The pattern of the initial population is still easily defined because the algorithm

immediately hones in on the region of the optimal position. Note that in this case, a second

optimization procedure is conducted, this time with an initial population narrowed around the

optimal region. Once again the algorithm hones almost immediately on the optimum servo

positions, which is well illustrated in the bottom right plot in Figure 5-7. This is a special case

due to the simplicity of the objective function and the small coefficient of variance in the data.

In the following maximum efficiency tests it is shown that convergence will not occur so rapidly.

The performance of the morphing wing as compared to the two fixed wing geometries is

plotted in Figure 5-8 (A). By morphing the wing throughout the alpha sweep, a much greater lift

coefficient is obtained. Note the benign stall characteristics of the morphing wing, which has

little decrease in lift from 15-200. No tests are conducted beyond the 200 angle of attack due to

the vibrations encountered within this stall region.

Having completed the tests for maximization of lift coefficient, the plot of optimal servo

positions versus angle of attack is acquired and plotted in Figure 5-8 (B). As is expected, the

maximum lift at any given angle of attack will have no reflex and is why the second servo

always converges to position 1. These servo position curves can now be used in a control

system, with angle of attack as input, to yield maximum lift performance.









Maximizing Efficiency Curve

Efficiency is calculated by dividing the lift coefficient by the drag coefficient. As

determined in Chapter 4, the coefficient of variance with such a calculation is significantly

greater than that associated with the lift coefficient alone. For this reason, it may take as many as

three times the number of generations for the genetic algorithm to converge on the optimal servo

positions as compared to the prior set of experiments. The optimization procedure is conducted

at seven angles of attack, which are sufficient to populate an efficiency curve. This efficiency

curve is then compared to the efficiencies of the flat plate and reflex wings.

It is seen in Figure 5-9 (top) the algorithm requires approximately 20 generations before

convergence. Note how the mean performance varies throughout the experiment while the best

performance will improve incrementally. This is due to the increased complexity of the

obj ective function and the increase in the variance in data. As a result, the genetic algorithm

selects much broader and more diverse servo positions to find the optimal region. The scatter

plot in Figure 5-9 (center) best illustrates the diversity in selected servo positions. The final

graphs of the same figure show the varied, well distributed initial population (bottom left) and

the final, converged upon optimum servo positions (bottom right).

The performance of the morphing wing with the two fixed wing geometries are plotted in

Figure 5-10 (A). Morphing the wing to the determined optimal shapes through the alpha sweep

provides a significant improvement in efficiency up to the region of stall. Notably, at 2.50 angle

of attack, the morphing wing has twice the efficiency value as that of the other wings. Having

completed the tests for maximization of efficiency, the plot of optimal servo positions versus

angle of attack is acquired and plotted in

Figure 5-10 (B).









Above the servo positions are the optimized wing shapes rotated to the corresponding

angle of attack. Using the equations of these curves in a control system, maximum efficiency as

a function of angle of attack can be achieved. These unpredictable curves illustrate the need for

optimization procedures for highly complex morphing wing structures.




























Figure 5-1.


The two-design-variable model morphs the wing geometry through change of
camber and reflex.


Z [mm]
1100603


Figure 5-2.


The full field, three-dimensional image, developed by the VIC system, is overlaid
on an image of the morphed wing illustrating deformation in the Z-direction.











80% Span
0."1
-0.1 ---
m 0 05


xr,


II :


80)% Span



D 0.5
x/c


x
lli ~I


-026 0.011 .270.060 0 07Il


Figure 5-3.


Morphed wing surface (Servo Positions: 250,230) with airfoil slices from the 20%
and 80% span yields a negative camber and negative reflex.


O
~r


20% Span



010 0 5 1
x/c


-0).078 -0.05 -0.022 0.0066 .035


Figure 5-4.


Morphed wing surface (Servo Positions: 1,1) with airfoil slices from the 20% and
80% span yields a positive camber and positive reflex.


20% Span
0.1 -

-0.1 * * *'
0 0.5 1









































~I ~ ~I


*******~~ Filat Plate


167


8~ t


Serve 2


Figure 5-5.


An illustration of the intermediate wing shapes from the 20% span shows the
morphing capability of the two-design-variable model.


...........,- Flat Plate

0 10 20 30
A ngle of Attack


0.2 t


-0.2

-0.4L
-10


0 10
Anrgle of Attack


Figure 5-6.


The lift coefficient and efficiency is plotted using set flat plate and reflex wing
geometries.





0 50 100 150 200 25
Servo 1 Position


Initial Population


S200 -


o 100.

n *


Final Population









5 10 15 20 25
Member


-0.5~

-1


...8r*********


Mean *Best


1.5 E


Generation





x






x






X
X
X


200E


160 E


100 E


XXX
XXx


.1 200


o '100


0 5 10 15
IMember


20 25


Figure 5-7.


These sample plots illustrate the convergence of the genetic algorithm upon the
optimal servo position for maximum lift at 100 angle of attack.






























































5 I I I I .I


0.6

0.4 I-


** .......,**** Flat Plate
-0.2 -- -- Morphing Wing

-OA4
-5 0 5 10 15 20
Angle of Attack


250 -

200

150

100


a ..s







Servo 1
-- S~ervio 2


Angle of Attack.


Figure 5-8.


Morphing wing data for optimized lift coefficient. A) The morphing wing
performance as compared to the fixed reflex wing and flat plate. B) The optimal
servo positions as a function of angle of attack with the corresponding wing
shape.














. Mean Best


. *
r***. e*


0


h-2


Generation


x x





x
x
x


200 E


xxx


150 E


100 E


X X
x


0 50 100 150
Servo 1 Position

Initial Pop~ulation


Final Pop~ulation


Member


Member


Figure 5-9.


Sample plots showing the convergence of the genetic algorithm on the optimal
servo position for maximum efficiency at 100 angle of attack.


.~ 200 ~ 200

I 11 11111111111111 1 II I ~

~'"I IIIII1IIIIIII~II~I~L LI~1 "100
rn 1 1111111111111111111 1111 v,
0 I~IIILLLLL.IlILIIL.IL11-111.1 0
0 5 10 15 20 25 0 5 10 15 20 25

























ii ******* Flat Plate

sj -- orphing Wing
-5 0 5 10 15 20 25
Angle of Atlacki










oj 250 a
o
ta 200 -' -
S1o e.I


100

50 C r Servo I
-*- Servo 2

-5 0 5 10 15
Angle of Attack



Figure 5-10. Morphing wing data for optimized efficiency. A) The morphing wing performance
as compared to the fixed reflex wing and flat plate. B) The optimal servo
positions as a function of angle of attack with the corresponding wing shape.









CHAPTER 6
CONCLUSION AND RECOMMENDATIONS FOR FUTURE WORK

Conclusion

A procedure has been established to achieve proper shape management of morphing wing

structures. It has been demonstrated that utilizing a genetic algorithm with wind tunnel

hardware-in-the-loop is an effective means of determining optimal wing shape functions. The

use of these shape functions enables a morphing aircraft to achieve optimal flight performance in

all flight regimes.

To demonstrate the procedure, a two design-variable morphing wing is fabricated with the

capability of altering maximum camber and trailing edge reflex. Two experiments are conducted

for maximizing the lift coefficient and maximizing efficiency, which resulted in optimal servo

position versus angle of attack shape functions. For a performance comparison, the wing shape

is set as a flat plate and again as a medium camber wing with reflex. These fixed wing shapes

are tested for lift and efficiency performance from -50 to 250 angle of attacks.

Plotting the performance of the morphing wing with that of the two fixed wings illustrates

the dramatic improvement achieved by morphing wing structures. In particular, while the slope

of the lift coefficient curve remains the same for each data set, the morphing wing shows a

63.4% improvement over the fixed reflex wing. Another remarkable attribute of the morphing

wing is its benign stall characteristics. While more data must be collected to determine the

wings performance at angles of attack greater than 200, it is clear that such a wing can be used to

control and mitigate wing stall.

The experiments for determining maximum efficiency show an equally impressive increase

in performance. Considering the range of -50 to 50 angle of attacks, typical for straight and level

flight and subtle altitude changes, efficiency increases as much as 62.1%.









These results support high confidence in the optimization procedure. The genetic

algorithm is capable of proper convergence despite the variance in collected wind tunnel data. It

is in the author' s opinion that the procedure will effectively determine optimal wing shapes for

morphing wings of many design variables and higher morphing capability.

Recommendations for Future Work

There are many valuable wing shape functions that can be attained using this procedure

which have yet to be explored. By simply applying constraints to the genetic algorithm many

new ideas for flight optimization arise.

One such experiment could be to vary the wing shape as a function of velocity to attain a

desired lift. For a given angle of attack, a high velocity would yield minimal camber while a low

velocity would yield a comparatively high camber. Such a shape function would minimize drag

thereby increasing efficiency.

Another interesting study would be to apply the constraint of a desired pitching moment

for each angle of attack to ensure longitudinal stability. Using this constraint while maximizing

lift at very high angles of attack could result in the mitigation of wing stall. Such a wing shape

function would yield a controlled, rate of descent flight mode useful in an auto-land feature.

Of course these shape functions are of no value unless they are utilized in an on-board

control system for real-time wing morphing. Such a control system will receive data such as

angle of attack or velocity and in response, command the position of the actuators. The servo

controller produced by Scott Bowman, which is utilized exclusively for this research, is capable

of such a task.

In a given mission profie for an aerial vehicle, several flight modes may be commanded.

For example, at takeoff the vehicle may require a wing shape that maximizes rate of climb.

When required altitude is met, another wing shape may be commanded to maximize speed to a









specified waypoint. At this point the vehicle may be required to conduct surveillance, which

would require a wing shape that would yield maximum efficiency for extended flight time. After

completion of surveillance the vehicle may be commanded to return to another waypoint where it

is required to perform a steep descent angle for landing in a small, confined field. The ability to

switch between these distinct optimal wing shape functions is necessary to maximize the overall

performance of a morphing winged aircraft as it transitions between flight modes.









LIST OF REFERENCES


[1] Haenel, R. T., "Creep Deformation of Pin-Jointed Structures," MS Thesis, The
Pennsylvania State University, Pennsylvania, 1955.

[2] Gooders, J., and Pledger, M., "Birds of North America," Edison, New Jersey: Chartwell
Books Inc., 1987.

[3] Davidson, J., Chwalowski, P., and Lazos, P., "Flight Dynamic Simulation Assessment of
a Morphable Hyper-Eliptic Cambered Span Winged Configuration," 2003 AIAA Atmospheric
Flight Mechanics Conference and Exhibit, Austin, TX, AIAA Paper 2003-5301, 2003.

[4] Shyy, W., Berg, M., and Ljungqvist, D., "Flapping and Flexible Wings For Biological
and Micro Air Vehicles," Progress in Aerospace Sciences, Vol. 35, No. 5, 1999, pp. 455-506.

[5] Ifju, P., Jenkins, D., Ettinger, S., Lian, Y., and Shyy, W., "Flexible-Wing Based Micro
Air Vehicles," AIAA Publication, AIAA Paper 2002-0705, 2002.

[6] Albertani, R., Hubner, J., Ifju, P., Lind, R., and Jackowski, J., "Experimental
Aerodynamics of Micro Air Vehicles," SAE World Aviation Congress and Exhibition, Reno,
NV, 2004.

[7] Stanford, B., Abdulrahim, M., Lind, R., Ifju, P, "Investigation of Membrane Actuation
for Roll Control of a Micro Air Vehicle," Journal of Aircraft, Vol. 44, No. 3, 2007, pp. 741-749.

[8] Gano, S., Renaud, J., Batill, S., and Tovar, A., "Shape Optimization for Conforming
Airfoils," 44th AIAA/ASME/ASCE/AHS Structures, Structural Dynamics, and Materials
Conference, Norfolk, Virginia, AIAA Paper 2003-1579, 2003.

[9] Gano, S., and Renaud, J., "Optimized Unmanned Aerial Vehicle with Wing Morphing for
Extended Range and Endurance," 9th AIAA/ISSMO Symposium and Exhibit on
Multidisciplinary Analysis and Optimization, Atlanta, GA, AIAA Paper 2002-5668, 2002.

[10] Waszak, M., Davidson, J., and Ifju, P., "Simulation and Flight Control of an Aeroelastic
Fixed Wing Micro Aerial Vehicle," 2002 AIAA Atmospheric Flight Mechanics Conference and
Exhibit, Monterey, CA, AIAA Paper 2002-4875, 2002.

[1l] Skillen, M., and Crossley, W., "Modeling and Optimization for Morphing Wing Concept
Generation," NASA Publication CR-2007-214860, March 2007.

[12] Albertani, R., "Experimental Aerodynamic and Static Elastic Deformation
Characterization of Low Aspect Ratio Flexible Fixed Wings Applied to Micro Aerial Vehicles,"
Ph.D. Dissertation, University of Florida, Gainesville, FL, 2005.

[13] Sytsma, M., "Aerodynamic Flow Characterization of Micro Air Vehicles Utilizing Flow
Visualization Methods," MS Thesis, University of Florida, Gainesville, FL, 2006.










[14] Bilgen, O., Kochersberger, K., Diggs, E., Kurdila, A., and Inman, D., "Morphing Wing
Aerodynamic Control via Macro-Fiber-Composite Actuators in an Unmanned Aircraft," 2007
AIAA Conference and Exhibit, Rohnert Park, CA, AIAA Paper 2007-2741, 2007.

[15] Williams, R., Inman, D., and Keats-Wilkie, W., "Nonliniear Mechanical Behavior of
Macro Fiber Composite Actuators," Center for Intelligent Material Systems and Structures,
Department of Mechanical Engineering, Virginia Polytechnic Institute and State University,
Undated material.

[16] Claverie-Bohn, C., "Dynamic Antifouling Structures and Actuators Using EAP
Composites," Ph.D. Dissertation, University of Florida, Gainesville, FL, 2004.

[17] Strelec, J., Lagoudas, D., Khan, M., and Yen, J., "Design and Implementation of a Shape
Memory Alloy Actuated Reconfigurable Airfoil," Journal ofhitelligent Material Systems and'
Structures, Vol. 14, No. 4-5, 2003, pp. 257-273.

[18] Lu, K., and Kota, S., "Design of Compliant Mechanisms for Morphing Structural
Shapes," Journal ofhitelligent Ma'~terial Systems and' Structures, Vol. 14, No. 6, 2003, pp. 379-
391.

[ 19] Rechenberg, I., "Evohitionsstrategie: Optintierung technischer Systeme nach Prinzipien
der biologischen Evohition," Frommann-Holzboog, Stuttgart, 1973.

[20] Whitley, D., "An Overview of Evolutionary Algorithms," Journal ofhifornzation and'
Software Technology, Vol. 43, 2001, pp. 817-831.

[21] Rasheed, K., and Hirsh, H., "Learning to be Selective in Genetic-Algorithm-B ased
Design Optimization," Artificial hItelligence for Engineering Design, Analysis and'
Manufacturing, Vol. 13, 1999, pp. 157-169.

[22] Herrera, F., Lozano, M., and Molina, D., "Continuous Scatter Search: An Analysis of the
Integration of Some Combination Methods and Improvement Strategies," European Journal of
Operational Research, Vol. 169, No. 2, 2006, pp. 450-476.

[23] Herrera, F., Lozano, M., and Sanchez, A., "Hybrid Crossover Operators for Real-Coded
Genetic Algorithms: An Experimental Study," Soft Comput, Vol. 9, 2005, pp. 280-298.

[24] Nauj oks, B., Willmes, L., Haase, W., Back, T., and Schtitz, M., "Multi-point Airfoil
Optimization Using Evolution Strategies," European Congress on Computational Methods in
Applied Sciences and Engineering, Barcelona, 2000.

[25] Liu, B., "Two-Level Optimization of Composite Wing Structures Based on Panel Genetic
Optimization," Ph.D. Dissertation, University of Florida, Gainesville, FL, 2001.

[26] Hemker, T., "Hardware-in-the-Loop Optimization of the Walking Speed of a Humanoid
Robot," CLAWAR 2006, Brussels, Belgium, 2006.











[ 27] Kammeyer, M., Wind Tunnel Facility Calibrations and Experimental Uncertainty, The
Boeing Company, American Institute of Aeronautics and Astronautics, 1998.

[28] Springer, A., "Uncertainty Analysis of the NASA MSFC 14-Inch Trisonic Wind Tunnel,"
1999 37" AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, AIAA-99-0684, 1999.

[29] Lopez-Anido, R., El-Chiti, F., Muszyfiski, L., Dagher, H., Thompson, L., and Hess, P.,
"Composite Material Testing Using a 3-D Digital Image Correlation System," American
Composites Manufacturers Association, Tampa, Florida, 2004.

[30] Muszyfiski, L., Lopez-Anido, R., and Shaler, S., "Image Correlation Analysis Applied to
Measurement of Shear Strains in Laminated Composites," SEM IX International Congress on
Experimental Mechanics, Orlando, Florida, 2000.

[31] Schmidt, T., Tyson, J., and Galanulis, K., "Full-Field Dynamic Displacement and Strain
Measurement Using Advanced 3-D Image Correlation Photogrammetry," Part I. Experimental
Techniques, Vol. 27, No. 3, 2002, pp. 47-50.

[32] Helm, J., McNeill, S., and Sutton, M., "Improved 3-D Image Correlation for Surface
Displacement Measurement", Optical Engineering, Vol. 35, No. 7, 1996, pp. 1911-1920.









BIOGRAPHICAL SKETCH

Frank Boria was born in Portsmouth, New Hampshire and spent most of his childhood in

York, Maine. When he was 13 years old, he and his family moved to Fernandina Beach, Florida.

During his junior year of high school, Frank won a scholarship to train for his private pilot

license. Having trained in a Citabria 7ECA, a tandem two-seat tail-dragger, Frank received his

pilot license and developed a passion for aviation.

After high school he moved to Tallahassee, Florida where he attended Lively Technical

Center and received his Airframe and Powerplant certifieations. With these certifieations he

moved to Lake City, Florida where he spent 7 years working at TIMCO, a third-party aircraft

repair station. His experience was primarily focused on aircraft structures for DC9, Boeing 727,

and C130 aircraft. With the understanding of how to fly and maintain an aircraft, his interests

leaned toward understanding the physics of flight and the process of design.

During the final 2 years of work at TIMCO, Frank attended evening classes at Santa Fe

Community College where he received his Associate of Arts degree. He transferred to the

University of Florida where, in 2004, he received his Bachelor of Science degree in aerospace

engineering. He continued his education at the University of Florida, and in 2007, he received

his Master of Science degree in aerospace engineering with a focus on solid mechanics, design,

and manufacturing. Frank now resides in Gainesville, Florida with his beautiful bride, Rikki,

and three wonderful children, Addie, Olivia, and Christian.





PAGE 1

OPTIMIZATION OF A MORPHING WING GEOMETRY USING A GENETIC ALGORITHM WITH WIND TUNNEL HARDWARE-IN-THE-LOOP By FRANK JOSEPH BORIA 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 1

PAGE 2

2007 Frank Joseph Boria 2

PAGE 3

To my bride, Rikki 3

PAGE 4

ACKNOWLEDGMENTS This work has been made possible through funding and facilities provided by the University of Florida and BioRobots, LLC. Many thanks in particular go to Dr. Peter Ifju for his guidance, leadership, and ability to meld science with low-quality humor. Another thank you goes to Dr. Roberto Albertani for his guidance on wind tunnel operation and his incessant support of the metric system, despite its obvious illogical basis (long live the slug). And yet another thank you goes to Dr. Bruce Carroll for so strongly supporting my work during this final task. I thank Bret Stanford for his thoughts and input throughout my research, whose insight and experience helped to create the framework of these experiments. I am very appreciative of Scott Bowman for his design, fabrication, and programming of a sophisticated morphing wing controller that was used to conduct this research. I am forever grateful and indebted to Frank and Joanne Boria for raising me with love and compassion while teaching me faith, respect and commitment. They are both remarkable parents and phenomenal role models. I will always look to them for guidance. To Julie Badger, Carl Boria, and Elizabeth Nettles, I am thankful for their strength of support, their being there to listen, and their reminding me of birthdays and anniversaries. To you, my brother and sisters, I gratefully dedicate the last figure of the fifth chapter. I would like to thank my children, Addie Reith for contributing her skills in photography, Olivia Reith for her remarkable contribution of digital photo editing, and Christian Boria for thinking that morphing airplanes are really, really cool. I love them all. Finally, I thank my bride, Rikki, for her never-ending support, love, and inspiration. She has brought me great joy, true happiness, and has taught me to live life with abandon. I am forever thankful to her for being genuine, refreshingly enthusiastic, and mine. 4

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES...........................................................................................................................7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................10 CHAPTER 1 INTRODUCTION.............................................................................................................12 Emulating Biological Structures........................................................................................12 Optimization of Morphing Wing Structures......................................................................13 Motivation..........................................................................................................................13 Overview............................................................................................................................14 2 LITERATURE REVIEW..................................................................................................16 Emulating Desirable Wing Shapes....................................................................................16 Actuation of a Morphing Wing Structure..........................................................................17 Optimization Techniques for Acquiring Desired Wing Shapes........................................18 Hardware for Data Collection............................................................................................19 Wind Tunnel Uncertainty Analysis............................................................................19 Visual Image Correlation System...............................................................................19 3 EXPERIMENTAL SETUP AND DATA ANALYSIS.....................................................20 Wind Tunnel......................................................................................................................20 Servo Controller.................................................................................................................21 Software Description.........................................................................................................21 Visual Image Correlation System......................................................................................22 4 ONE DESIGN VARIABLE MODEL...............................................................................27 Model Description and Purpose.........................................................................................27 A Study of Aerodynamic Hysteresis..................................................................................28 Initial Alpha Sweep for Predicting Convergence..............................................................29 Validation of Optimization Procedure...............................................................................30 5 OPTIMIZATION OF A HIGH-FIDELITY MORPHING WING....................................40 Model Description.............................................................................................................40 Comparative Lift and Efficiency Curves...........................................................................41 5

PAGE 6

Maximizing Coefficient of Lift Curve...............................................................................41 Maximizing Efficiency Curve............................................................................................43 6 CONCLUSION AND RECOMMENDATIONS FOR FUTURE WORK........................52 Conclusion.........................................................................................................................52 Recommendations for Future Work...................................................................................53 LIST OF REFERENCES...............................................................................................................55 BIOGRAPHICAL SKETCH.........................................................................................................58 6

PAGE 7

LIST OF TABLES Table Page 4-1 Small coefficient of variation between the ten base data sets and the random data sets ensures minimal effect of aerodynamic hysteresis on the coefficient of lift.....................33 4-2 Drag data show acceptably low variation between the ten base data sets and the random data sets.................................................................................................................33 4-3 Efficiency is found to have a variance of the same magnitude as drag, which is sufficient for the procedure to correctly converge upon the optimum servo position.......33 7

PAGE 8

LIST OF FIGURES Figure Page 3-1 Closed-loop wind tunnel at the University of Florida is pictured with a 24-inch test section............................................................................................................................24 3-2 Digital protractor is used to measure the angle of the sting balance that mounts to the U-shape model arm............................................................................................................24 3-3 The servo controller (background) is used to set and hold servo positions, while the serial port adapter (foreground) converts signals from RS232 to TTL.............................24 3-4 Data exchange between software applications..................................................................25 3-5 The VIC system hardware setup about a wind tunnel test section....................................26 4-1 The one design variable, rectangular planform model used to validate the optimization process has a single point of actuation that alters camber............................34 4-2 The full field, three-dimensional image, developed by the VIC system, is overlaid on an image of the morphed wing illustrating deformation in the Z-direction.......................34 4-3 Wing shape attained by servo position 1 with velocity at 0 and 16 m/s at 10 AOA. The top left image shows the wind off Z-coordinates of the wing in millimeters. The top right image shows the wind on wing deformation.......................................................35 4-4 Wing shape attained by servo position 250 with velocity at 0 and 16 m/s at 10 AOA....35 4-5 Plot of a series of airfoil shapes attained as a function of servo position..........................36 4-6 Results of an aerodynamic hysteresis study depict the servo scatter and data coefficient of variance for multiple servo positions..........................................................37 4-7 Lift coefficient, drag coefficient, and efficiency versus angle of attack are plotted for several servo positions of the one design variable model..................................................38 4-8 Resulting data illustrate the effectiveness of the optimization procedure. A) Initial population. B) Final population. C) Coefficient of lift servo scatter. D) Morphing wing performance comparison for lift coefficient. E) Efficiency servo scatter. F) Morphing wing performance comparison for efficiency..............................................39 5-1 The two design-variable model morphs the wing geometry through change of camber and reflex...........................................................................................................................45 5-2 The full field, three-dimensional image, developed by the VIC system, is overlaid on an image of the morphed wing illustrating deformation in the Z-direction.......................45 8

PAGE 9

5-3 Morphed wing surface (Servo Positions: 250,230) with airfoil slices from the 20% and 80% span yields a negative camber and negative reflex.............................................46 5-4 Morphed wing surface (Servo Positions: 1,1) with airfoil slices from the 20% and 80% span yields a positive camber and positive reflex.....................................................46 5-5 An illustration of the intermediate wing shapes from the 20% span shows the morphing capability of the two design-variable model.....................................................47 5-6 The lift coefficient and efficiency is plotted using set flat plate and reflex wing geometries..........................................................................................................................47 5-7 These sample plots illustrate the convergence of the genetic algorithm upon the optimal servo position for maximum lift at 10 angle of attack........................................48 5-8 Morphing wing data for optimized lift coefficient. A) The morphing wing performance as compared to the fixed reflex wing and flat plate. B) The optimal servo positions as a function of angle of attack with the corresponding wing shape........49 5-9 These sample plots illustrate the convergence of the genetic algorithm upon the optimal servo position for maximum efficiency at 10 angle of attack.............................50 5-10 Morphing wing data for optimized efficiency. A) The morphing wing performance as compared to the fixed reflex wing and flat plate. B) The optimal servo positions as a function of angle of attack with the corresponding wing shape.................................51 9

PAGE 10

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 OPTIMIZATION OF A MORPHING WING GEOMETRY USING A GENETIC ALGORITHM WITH WIND TUNNEL HARDWARE-IN-THE-LOOP By Frank Joseph Boria August 2007 Chair: Peter Ifju Major: Aerospace Engineering Advancements in compliant actuator technologies and flexible composite structures have enabled true wing morphing capabilities. The resultant effect of these new capabilities is optimal wing shapes for all flight regimes. With the ever-increasing complexity of these morphing wing structures comes the challenge of proper shape management for achieving desirable, real-time flight performance. Using a wind tunnel with sting balance as a hardware-in-the-loop objective function, a genetic algorithm is used for optimizing multiple design variable morphing wing structures. This evolutionary approach to optimization provides a more efficient means of convergence over that of other conventional optimization techniques. There are four primary subsystems utilized in the experimental setup. These include the optimization software, the servo controller, the wind tunnel, and the data acquisition software. Through the use of a simple text-format software package, data is passed between each of these subsystems, thereby automating the procedure. A visual image correlation system is used independently during testing to determine the three-dimensional wing shapes of noteworthy servo positions. 10

PAGE 11

To demonstrate the procedure, a two design-variable morphing wing is fabricated with the capability of altering camber and trailing edge reflex. Two experiments are conducted for maximizing the lift coefficient and maximizing efficiency. These experiments result in optimal servo position versus angle of attack shape functions. Setting the wing shape as a flat plate and again as a medium camber wing with reflex provides comparative data. These experiments resulted in a comparatively benign stall characteristic, a 63.4% increase in lift coefficient, and up to 62.1% increase in maximum efficiency over that of the fixed wing with reflex. Plotting the performance of the morphing wing, with that of the two arbitrary fixed wing shapes, illustrates the dramatic improvement achievable by morphing wing structures. Utilizing these procedures allows for real-time adaptive wing morphing to adjust to the instantaneous changes of flight attitude and wind conditions. 11

PAGE 12

CHAPTER 1 INTRODUCTION No longer are aircraft confined to fixed wing geometries suiting a single flight regime. The advancements of actuation and material technologies have enabled true morphing capabilities. The resultant effect of these new capabilities is optimal wing shapes for all flight regimes. With the ever-increasing complexity of these morphing wing structures comes the challenge of proper shape management for achieving desirable, real-time flight characteristics. The work presented in this thesis is one approach to such a challenge. Emulating Biological Structures The study of ornithology shows that each species of bird has evolved to become uniquely specialized within their habitat. From the perspective of the aerodynamicist, designing an aircraft to emulate the performance of the falcon for high speed dives, the eagle for highly efficient soaring, the hovering capability of the humming bird, or the agility of the swift would be of great value. Ultimately, however, melding each of these desirable flight characteristics on a single aircraft is the true objective. Such a versatile vehicle could achieve remarkably high lift, for maximum payload capacity, high speed, to quickly move from waypoint to waypoint, and high efficiency. Utilizing a morphing wing structure for asymmetric deformations would yield the ability to maneuver with high agility through complicated urban environments. Such a vehicle could also maintain stability during the most volatile of flight conditions, such as erratic and gusting winds among buildings and structures. Utilizing thin plate composite materials with multiple actuation points, it is possible to create a morphing wing capable of complex deformation. A noteworthy study on similar materials with regard to creep is presented by Richard Thomas Haenel [1], which details the 12

PAGE 13

evaluation of creep with respect to material deformation. Careful design of such a wing will yield configurations that significantly outperform conventional fixed wing geometries. Optimization of Morphing Wing Structures When a given structure is created to emulate the complexities of what exists in nature, a method must also be developed to command the structure to take a desired shape. A high fidelity morphing wing with several actuation points will require a complex control system. The once rigorous determination of aerodynamic performance for conventional fixed wing geometries has become an even greater undertaking in multi-variable shape optimization. While the field of computational fluid dynamics continues to improve, modeling complex structures of composite makeup remains expensive. It is also difficult to accurately model the aeroelastic effects on such a structure when it is under load. This becomes increasingly the case when utilizing thin plate wings. While the use of a wind tunnel may also be expensive, it provides a source for aerodynamic force data, which take full account of wing deformations due to aeroelastic effects. Regardless of the method used for determining aerodynamic forces, a procedure for optimization must be properly matched to the capability of the wing structure. While a response surface approach may be well suited for morphing wings with minimal design variables, high-fidelity morphing wing structures may be more efficiently optimized through the use of a genetic algorithm. The genetic algorithm is better capable of finding a global optimum more efficiently than many strategies. Motivation The use of a genetic algorithm for optimizing high-fidelity, multiple design variable morphing wing structures provides a more efficient means of convergence over that of other conventional optimization techniques. 13

PAGE 14

For example, collecting data for a one design variable wing of 10 position increments over an alpha sweep of 10 angles of attack requires 100 data points. To ensure data validity, these same data points are typically collected multiple times to ensure precision, which increases the number of required data points to a minimum of 200. Increasing the models capability through the use of two design variables increases the number of required data points to 2,000. Ultimately, the required number of data points increases by an order of magnitude with each additional design variable. Of course this analysis assumes only 10 position increments are possible, while the actuators used for this study have 250 position increments. The use of response surface techniques is a reasonable approach for wings with up to three and even four design variables. But as the number of design variables grows, it becomes inefficient. Finding the optimal position of the determined response surface becomes more difficult to solve. The complexity of the response surface also raises concerns for convergence upon false, local optimums. The use of this and similar techniques for finding maximum performance shapes for high-design variable morphing wing structures is ineffective. Overview To establish an effective procedure for determining optimal wing shape of a multi-variable morphing wing, the use of a genetic algorithm with wind tunnel hardware-in-the-loop is implemented. This procedure is used to determine a specific desired flight characteristic, such as maximum lift or maximum efficiency. By conducting these optimization tests at several angles of attack, it becomes possible to obtain optimal wing shapes throughout the range of angles of attack. Utilizing these data in a control system make it possible to automate real-time wing morphing for optimal performance during flight. It is the intent of this thesis to present and implement such a procedure with its proficiencies and detriments. 14

PAGE 15

This thesis consists of six chapters. The first chapter outlines the intent and scope of the investigation. The second chapter details a literature review of prior work related to morphing wing structures, optimization procedures, and wind tunnel operation and data collection. The third chapter describes the experimental setup with detail on the wind tunnel and the visual image correlation system hardware. Also described in this chapter are the optimization software, the interaction among the programs, and the experimental procedure for error minimization. The fourth chapter describes the preliminary one design variable model used to validate the optimization procedure. The fifth chapter describes the multi design variable model used to illustrate the improved performance the morphing wing has over that of conventional fixed wings. The sixth chapter closes with a conclusion of the investigation and suggests recommendations for future work. 15

PAGE 16

CHAPTER 2 LITERATURE REVIEW Through the use of unmanned aerial vehicles it becomes possible to conduct dangerous missions without endangering the life of a pilot. These vehicles are becoming smaller with missions that require significant agility for navigating complex urban environments. To see a bird navigate through such an environment will inspire the concept of its emulation. Through the ever-maturing field of material science, the morphing capabilities of an aircraft wing are more capable of simulating a bird in flight. Emulating Desirable Wing Shapes The complex system of structural members and joint articulation of a birds wing [2] is difficult to duplicate mechanically with regard to function as well as low weight and high durability. Studies associating wing shape to function are conducted to differentiate requirements for static aerodynamics, physiology, and flapping control [3]. Understanding this differentiation leads to morphing wing structures of less complexity than that of the birds wing. For example, while birds have many layers of feathers that are moved around to adjust to the specific maneuver they need to perform [4], morphing wing structures can be fabricated of thin sheet materials, which are manipulated to yield the desired wing shape. Currently, many unmanned aerial vehicle designs are based on a flexible wing design developed at the University of Florida [5],[6]. These thin, flexible wings are beneficial for dampening the effects of wind gusts and increasing stability during adverse flight conditions. With regard to wing morphing, they are also easily manipulated for desired control and overall flight performance using minimal actuation points. An example of this is morphing in the form of asymmetric wing twist making it is possible to surpass the roll rates of conventional aileron 16

PAGE 17

configurations [7]. Another example is the exercise in design optimization conducted by Gano et al. [8],[9] for a single, high endurance wing that separates into a second wing for greater agility. Much work has been done to assess the flight characteristics of these flexible wings. NASA has performed a control assessment and simulation of a micro air vehicle with aeroelastic wings which adapt to the disturbances during flight [10]. Morphing wing weight equations and an approach to size morphing aircraft [11] are developed to replace current methods of general heuristics based on fixed wing aircraft. Also, extensive wind tunnel tests on a variety of flexible wing configurations showed a decrease in wing incidence, an increase in dihedral and a shift in maximum camber position that results in a favorable change in pitching moment and lift. Ultimately, smoother flight characteristics are achieved over that of conventional fixed wings [12]. A variety of experimental flow tests are conducted on these wings using surface oil flow visualization, laser based flow visualization, and particle image velocimetry with results that validate previously determined CFD results [13]. Actuation of a Morphing Wing Structure Advancements in actuator technologies for these compliant wing structures, such as Macro Fiber Composites (MFC), Electroactive Polymers (EAP), and Smart Memory Alloys (SMA), are yielding lightweight, capable devices. The thin, compliant MFC is a type of piezoceramic composite actuator that is constructed of orthogonal layers of unidirectional piezoceramic fibers and copper electrodes encased in layers of acrylic and Kapton [14]. These actuators have been reported to produce high strain levels on the order of 2000 [15]. Alternatively, the EAP actuator is similar to muscle tissue with regard to stress and force capability. This actuator is capable of large-scale linear motion [16]. Another source of actuation is through the use of an SMA mechanism. These mechanisms consist of a metallic alloy that act as linear actuators by contracting when heated and returning to their original shape when cooled [17]. This actuator is 17

PAGE 18

lightweight and very effective although precaution must be taken to prevent variations in temperature, which will affect its performance. These, and many other, compliant forms of actuation have been developed and utilized for morphing wing structures. Extensive work on adaptive shape change problems, which determine methods for morphing a given curve or shape into a target curve or shape, is providing solutions with a minimal number of actuators [18]. This work leads to the challenge of morphing a wing not to a single target shape, but rather to adaptively morph the wing real-time during flight to adjust to the instantaneous changes of flight attitude and wind conditions. Optimization Techniques for Acquiring Desired Wing Shapes The use of evolution strategies for optimization of complex design problems began back in the early sixties [19] and has been gaining momentum over the past fifteen years. Such strategies emphasize selection, recombination and mutation of a given population to determine the makeup of the next population set [20]. Each of the members of this new generation is then evaluated for fitness. Poor performing members are rejected from the next generation while high performing members are retained. This results in convergence upon an optimal solution after multiple generations. The use of a genetic algorithm (GA) has advantages over other techniques when searching within a complex domain. Due to its evolutionary tactics, the GA is resistant to convergence on local optimums making it better suited for finding true global optimums. It is also capable of working with many design variables, which can be discrete, continuous, or mixed [21], [22], [23]. As morphing wing structures become more complex, the GA solver becomes better preferred. Studies have already shown the effective use of the GA to optimize fixed airfoil and wing shape [24] and to better develop composite wings for desired structural characteristics [23]. 18

PAGE 19

Another study used a sequential surrogate optimization approach with hardware-in-the-loop as its objective function [25]. While the study was specific to determining the maximum walking speed of a humanoid robot [26], this use of hardware-in-the-loop is well suited for determining optimal wing shape for a morphing wing. Hardware for Data Collection Wind Tunnel Uncertainty Analysis By testing the morphing wing in a wind tunnel with a six degree-of-freedom sting balance to measure forces and moments, an array of performance data is collected. Using the sting balance as a hardware-in-the-loop objective function, the GA morphs the wing shape for each member of a given generation. A fitness value for each member is calculated from data received from the sting balance. Critical to the use of any hardware-in-the-loop optimization procedure is a determination of the measurement uncertainty for evaluation of data quality. Such an evaluation of wind tunnel and sting balance performance is documented in [27] and [28]. Visual Image Correlation System Using the digital image correlation (a.k.a. visual image correlation or VIC) system, a three-dimensional non-contact full-field measurement of strain and displacements can be attained for a given structure of composite material [29]. It has successfully been used to determine strains in specimens of fiber reinforced polymer composites [30]. This system is well suited to determine wing shapes for differing morphed positions while under flight loads. The supporting principles, specifications, calibration procedures, and applications are well documented in [31]. 19

PAGE 20

CHAPTER 3 EXPERIMENTAL SETUP AND DATA ANALYSIS There are four primary subsystems utilized in the experimental setup. These include the optimization software, the servo controller, the wind tunnel, and the data acquisition system. Through the use of simple text-format software, data is passed between each of these subsystems. This cooperative scheme of utilizing previously independent software programs provides a means for automating the optimization procedure. The visual image correlation system is used independently during testing to determine the three-dimensional wing shapes of noteworthy servo positions. Wind Tunnel The primary piece of hardware used for this optimization procedure is the closed-loop wind tunnel model 407B (Figure 3-1), which is manufactured by Engineering Laboratory Design. This low-speed, low-turbulence tunnel has been reported to have less than 0.2% turbulence [13] along the centerline of its 33-inch test section. The tunnel is capable of producing airspeeds of 2 m/s to 45 m/s with its 250HP motor and 2-stage axial fan. Extending from the sidewall of the test section is a U-shape model arm (Figure 3-2) that supports the 6 degree-of-freedom sting balance and rotates the balance/model assembly to the commanded angle of attack. It is the sting balance that provides all force and moment data to the software through a National Instruments SCXI 1000 frame. The sting balance, produced by Aerolab model 01-15, is capable of measuring loads on the order of 0.01 N [13]. Several LabVIEW virtual instruments (VI) are used to control the wind tunnel speed, temperature, model angle of attack, and sting balance voltage data acquisition. As detailed later in this chapter, automation of the optimization procedure is achieved by calling these subroutine VIs from within a primary VI which exchanges loads data with the genetic algorithm software. 20

PAGE 21

Servo Controller Each morphing wing developed for these experiments utilizes Futaba S3102 servo actuators to attain the surface deformation. These servos supply 4.6 kgcm of torque and a rate of 60 in 0.2 seconds. Using Matlab, servo number and position are commanded through the serial port in a 1200-baud rate, 8-bit, 1-stop, and no-parity form. A sub-module converts the RS232 signal to transistor-to-transistor logic. Receiving the signal on the morphing wing structure is a custom servo controller (Figure 3-3) designed and fabricated by Scott Bowman. This controller features 8 kilobytes of flash memory, a 16-bit high-resolution servo timer, and an 8 MHz ATMEL microprocessor. This servo controller is designed to set and hold commanded servo positions while the wing undergoes aerodynamic loading. Software Description The two primary software packages used for this study are LabVIEW for wind tunnel control and data acquisition and Matlab for wing morphing control and optimization. A flow chart of how the information is passed between the programs is shown in Figure 3-4. To initiate the optimization procedure, an initial population is prescribed to the genetic algorithm. When the algorithm is started, the first member of the initial population is sent to the subroutine that commands the servo controller. The servo controller then sets and holds each servo position thereby morphing the wing. Simultaneously, the servo positions are sent to a text file. At this point, the genetic algorithm waits for the force and moment data that is associated with the servo positions to be added to the text file. It determines when this happens by monitoring the text file size. While the optimization software is on hold for data, the wind tunnel is brought up to the required velocity. After the angle of attack is set, the force and moment data are determined from the sting balance. The text file previously created by the optimization software is accessed 21

PAGE 22

and the data is appended to the same line. The resulting line will consist of servo positions followed by the lift coefficient, drag coefficient, coefficient of side force, coefficient of pitching moment, and coefficient of yawing moment. The rolling moment is omitted because the system yields invalid data. This is of no detriment to these sets of experiments since rolling moment has no contribution in force and is not utilized as a constraint. At this point, the data acquisition software waits for the next set of servo positions to be added to the text file. As with the optimization software, it does this by monitoring the file size. When the new servo positions are added to the text file, the file size will increase thereby triggering the data acquisition software for the next set of data. After the force and moment data have been added to the text file, the optimization software identifies the increase in file size and retrieves the data. The next servo positions are commanded, and thereby begin the next cycle. As with any sting balance, there does exist the concern for signal drift. This sting balance contains a network of strain gauges that is used to measure the elastic strain of the balance. These strain gauges are susceptible to signal drift due to temperature variations caused by both environment and self-heating. To manage the signal drift, the wind tunnel software counts the number of data sets collected. After the fifteenth data set, the tunnel velocity is set to zero and the model arm is rotated to the original angle of attack. At this point, the data acquisition software performs an offset nullification of the strain gauge voltages. The tunnel is then brought back up to the velocity and angle of attack required by the experiment. This simple procedure ensures data quality and is set to coincide with the genetic algorithm population size of fifteen. Visual Image Correlation System Determining the shape of the morphed wing is essential to quantitatively validate the obtained solutions. Through the use of the visual image correlation (VIC) system, a non-contact 22

PAGE 23

full-field shape and deformation tool, the wing shape can be captured and evaluated for any given morphed configuration. A typical experimental setup with the VIC system is illustrated in Figure 3-5. This process was developed in the mid 1990s by Helm with a reported resolution on the order of 0.05 mm [32]. The specimen is first uniformly painted with a low-luster color, and then painted with a highly contrasting, random speckle pattern. Two digital cameras carefully focused on the specimen. The cameras are triggered such that an image from each is acquired instantaneously. A plate with high contrast dots of known diameter and spacing is used to calibrate the camera system. Photos of the plate in different locations of the image frame and at varying angles and rotations are taken and processed for calibration. After the calibration procedure is complete, the precise orientation of one camera to the other is known. At this point, the system is now ready to photograph the specimen. From these images, the software is capable of developing the full-field three-dimensional shape of the specimen using stereo triangulation. The system is also used to determine deformation of a specimen under load. By first taking a no-load reference image of the specimen, then taking another image of the loaded specimen, the software is capable of calculating the displacement field. This is useful for determining how much deformation a wing undergoes when in flight. 23

PAGE 24

Figure 3-1. The closed-loop wind tunnel at the University of Florida is pictured with a 24-inch test section (image provided by Michael Sytsma). Figure 3-2. A digital protractor is used to measure the angle of the sting balance that mounts to the U-shape model arm. Figure 3-3. The servo controller (background) is used to set and hold servo positions, while the serial port adapter (foreground) converts signals from RS232 to TTL. 24

PAGE 25

Loop 1 Loop 2 Figure 3-4. These flowcharts illustrate how data is appended to the shared text file. 25

PAGE 26

Figure 3-5. This illustration depicts the VIC system hardware setup about a wind tunnel test section. 26

PAGE 27

CHAPTER 4 ONE DESIGN VARIABLE MODEL Model Description and Purpose To validate the procedures for data collection, a simple one-design-variable model is utilized. This model is a thin plate wing of rectangular planform with a single point of actuation that changes the wings camber. The wing consists of a single layer, plain weave carbon fiber skin and horizontally aligned unidirectional spars that run from wingtip to wingtip every one inch from the leading edge (illustrated in Figure 4-1). These spars transfer loads such that when the wing shape is changed at the root, the same shape will be attained at the tips effectively resulting in an extruded airfoil along the span of wing. An attachment point at the leading edge permits rotation only. An attachment point at the 70% chord permits both rotation and longitudinal motion. The unconstrained trailing edge allows for significant deformation when under load. This deformation is an aeroelastic effect caused by the interaction among inertial, elastic, and aerodynamic forces on the wing. Typical CFD is expensive when attempting to model this aeroelastic effect, let alone optimize a morphing wing, and is why this tunnel-in-the-loop procedure is of particular value. Using the VIC system, the wing shape is determined. It is shown as a topographical overlay on an image of the wing itself in Figure 4-2. The wind-on deformations, and the airfoils associated with each, are plotted in Figure 4-3 for servo position 1 (a negatively cambered wing) and Figure 4-4 for servo position 250 (a highly cambered wing). The model is capable of altering the maximum camber between % through 19%. Utilizing the VIC system, 9 servo positions are selected and photographed to determine the range of attainable wind-off wing shapes. The normalized airfoils are plotted in Figure 4-5, which help to illustrate the magnitude of camber change. 27

PAGE 28

By collecting data for set servo positions over an array of angle of attacks, the performance trends of the model become clear. The now predictable performance of this wing is used to determine whether proper convergence of the genetic algorithm occurs. When it is confirmed that convergence is achieved repeatedly, the process can then be used to optimize performance for complex, high fidelity wings capable of sophisticated wing morphing. A Study of Aerodynamic Hysteresis Aerodynamic hysteresis occurs when the detached, low Reynolds number flow inconsistently reattaches thereby resulting in varied performance. Repeatability of wing performance is essential when conducting optimization tests since the genetic algorithm will semi-randomly alter servo position while conducting its search. In an effort to determine if aerodynamic hysteresis exists, a series of tests are conducted. Five servo positions are used to represent the extremes and intermediate positions within the range of servo motion. While holding the angle of attack and airspeed constant, ten consecutive data sets are collected for each of the five servo positions. The standard deviation is calculated for each of these data sets. This standard deviation is a measure of instrument precision and is independent of aerodynamic hysteresis. Data is then collected as the servo positions are varied between the extremes and intermediate servo positions to measure any effect caused by aerodynamic hysteresis. The data is sorted into common servo positions and the standard deviation is calculated. If the standard deviation of the original data sets and the varied data sets coincides within a common range, aerodynamic hysteresis will have been proven not to exist. Tests are conducted at 0, 10, and 22.5 angles of attack, which represent low drag, high lift, and near-stall flight regimes. Within Figure 4-6, eight plots are provided to illustrate the results of the study. In Fig. 4-6 (A), the servo position variation used to induce aerodynamic 28

PAGE 29

hysteresis is shown. Several sweeps from maximum to minimum and intermediate servo positions are performed to measure the effect of the hysteresis on the lift and drag coefficients. The drag coefficient versus servo position, shown in Fig. 4-6 (B), is provided to show the precision of the data. It is selected since it is the least precise of all data provided by the instrumentation; this is typically an order of magnitude less than that of the lift coefficient. The subplots shown in Fig.4-6 (C-H) are magnified data sets for lift and drag coefficients at servo positions 1, 125, and 250. While the plots clearly show aerodynamic hysteresis does exist, the standard deviation within a given data set is acceptable with a coefficient of variance of no more than 2.3% for coefficient of lift (Table 4-1), 2.9% for coefficient of drag (Table 4-2), and 3.4% for efficiency (Table 4-3). This study shows the effect of the aerodynamic hysteresis is minimal and has little influence on the data collected. It is in the authors opinion, as supported by this study, that the effects due to aerodynamic hysteresis can be assumed negligible. Initial Alpha Sweep for Predicting Convergence To validate the procedure for determining the optimum wing shape, a preliminary set of data was collected. The servo used in the model is capable of 120 rotation, which is commanded by the servo controller using position points 1 through 250. This range of servo position is divided into 9 positions that are used to populate lift, drag, and efficiency plots. For each servo position, data is collected for angle of attacks ranging from -5 through 30 in 2.5 increments (Figure 4-7). In doing this it can be seen that the lift curves systematically increase as camber increases. The drag plots show an interesting inversion in trend shape as the camber increases, which is likely due to the aeroelastic effects of the unconstrained trailing edge. Finally, the efficiency plot 29

PAGE 30

shows a complex trend, which would have been difficult to predict. By plotting the convergence data against these plots it becomes clear if the genetic algorithm is finding the true optimum. Validation of Optimization Procedure Using the genetic algorithm with the wind tunnel hardware in the loop, tests for determining servo positions associated with optimal lift and efficiency are conducted. The optimization procedure is first tested multiple times at set angles of attack to ensure repeatable convergence on the optimum position. This demonstrates the data collected to be both sufficiently accurate and precise for this procedure. The optimization procedure is then tested at several other angles of attack and is compared to the previously acquired alpha sweeps presented in Figure 4-7. Showing that the converged upon optimal values coincide or exceed predicted values validates the optimization procedure. Since the lift forces yield a lower standard deviation than that of the drag forces, it is used for the initial convergence tests. When calculating efficiency, the lift is divided by the drag resulting in a much greater coefficient of variance than that of the two individual components. These tests are conducted after convergence for maximum lift is achieved. The initial population is user defined and is equally distributed throughout the servo position range. Setting the initial population in this manner, shown in Figure 48 (A), significantly reduces the required number of generations to convergence. This occurs because the lift and efficiency curves as a function of servo position are simple polynomial curves without complex local minimums. A sample of the final generation is plotted in Figure 48 (B), which shows convergence on an optimum with subtle deviations from that optimum. The genetic algorithm selected servo positions, or servo scatter, are plotted in Figure 48 (C) and (E). Note the cluster of data points near the optimum servo positions where the algorithm narrowed its search. This plot is useful to visually ensure the algorithm searched 30

PAGE 31

sufficiently throughout the range of servo positions. A large gap in the scatter will raise concerns that the true global optimum may have been missed. By setting the angle of attack to 10, high lift and drag forces are ensured. Three optimization experiments are conducted to determine which servo position maximizes lift. As shown in Figure 48 (D), each experiment converged on a value that exceeded or matched the predicted value. All three experiments converged to servo position 250, which is associated with the greatest camber. Due to the variance in lift data, the lift coefficient associated with the optimal servo position will never converge closer than 2.3% of one another. This does not, however, prevent the genetic algorithm from properly converging on that optimal servo position. Due to the low coefficient of variance and the rapid convergence on the optimum servo position, it is in the authors opinion that the procedure will accurately converge to the optimum value regardless of angle of attack. For this reason, the second sets of convergence experiments for optimal efficiency are conducted. Continuing to hold the angle of attack at 10, three convergence experiments for determination of maximum efficiency are conducted. Since the coefficient of variance for efficiency is much higher than that of lift, other angles of attack are tested. As shown in Figure 48 (F), each of the three experiments converges on servo position 134 with a maximum efficiency of 5.56. While this value is below that of the predicted value, it does exist within the previously calculated variance and did converge on the predicted servo position. Because the coefficient of drag is very low at 0 and lift will become negative at -5 angle of attack, the efficiency may vary considerably. For this reason they are tested to ensure proper convergence. These results are plotted in Figure 48 (F). As illustrated, the optimization 31

PAGE 32

procedure did accurately converge upon the correct servo positions. This is because although the drag forces may be very low, they are consistently low; i.e. precise though not accurate. These results support high confidence in the optimization procedure. The genetic algorithm is capable of proper convergence despite the variance in collected wind tunnel data. It is in the authors opinion that the procedure will effectively determine optimal wing shapes for wings with multiple design variables and of higher morphing capability. 32

PAGE 33

Table 4-1. Small coefficient of variation between the ten base data sets and the random data sets ensures minimal effect of aerodynamic hysteresis on the coefficient of lift. 0 Deg 10 Deg 22.5 Deg Lift coefficient of variations % % % Base 0.14 0.32 1.79 Random 1.36 2.20 2.28 SP 1 Overall 1.83 1.48 2.27 Base 0.34 0.21 0.25 Random 1.17 0.98 1.45 SP 125 Overall 0.72 0.63 1.05 Base 0.19 0.38 0.30 Random 0.83 0.25 1.36 SP 250 Overall 0.66 0.40 0.96 Table 4-2. Drag data show acceptably low variation between the ten base data sets and the random data sets. 0 Deg 10 Deg 22.5 Deg Drag coefficient of variations % % % Base 3.07 3.40 1.90 Random 2.63 2.29 1.67 SP 1 Overall 2.80 2.88 2.04 Base 0.56 1.85 0.81 Random 0.87 1.22 1.26 SP 125 Overall 0.77 1.60 1.47 Base 1.29 1.05 0.39 Random 1.47 0.69 1.24 SP 250 Overall 1.42 1.01 1.11 Table 4-3. Efficiency is found to have a variance of the same magnitude as drag, which is sufficient for the procedure to correctly converge upon the optimum servo position. 0 Deg 10 Deg 22.5 Deg Efficiency coefficient of variations % % % Base 3.05 3.49 2.23 Random 3.13 3.03 1.88 SP 1 Overall 3.39 3.22 2.98 Base 0.54 1.64 0.92 Random 0.46 1.40 0.97 SP 125 Overall 0.94 1.51 1.04 Base 1.22 1.31 0.47 Random 1.00 0.53 1.11 SP 250 Overall 1.39 1.00 0.94 33

PAGE 34

Figure 4-1. The one-design-variable, rectangular planform model used to validate the optimization process has a single point of actuation that alters camber. Figure 4-2. The full field, three-dimensional image, developed by the VIC system, is overlaid on an image of the morphed wing illustrating deformation in the Z-direction (left image has 19% camber, right image has % camber). 34

PAGE 35

Figure 4-3. Wing shape attained by servo position 1 with velocity at 0 and 16 m/s at 10 AOA. The top left image shows the wind off Z-coordinates of the wing in millimeters. The top right image shows the wind on wing deformation. Figure 4-4. Wing shape attained by servo position 250 with velocity at 0 and 16 m/s at 10 AOA. The top left image shows the wind off form of the wing in millimeters. The top right image shows the wind on wing deformation, also in millimeters. 35

PAGE 36

Figure 4-5. Plot of a series of airfoil shapes attained as a function of servo position. 36

PAGE 37

Figure 4-6. Results of an aerodynamic hysteresis study depict the servo scatter and data coefficient of variance for multiple servo positions. 37

PAGE 38

Figure 4-7. Lift coefficient, drag coefficient, and efficiency versus angle of attack are plotted for several servo positions of the one design variable model. 38

PAGE 39

Figure 48. Resulting data illustrate the effectiveness of the optimization procedure. A) Initial population. B) Final population. C) Coefficient of lift servo scatter. D) Morphing wing performance comparison for lift coefficient. E) Efficiency servo scatter. F) Morphing wing performance comparison for efficiency. 39

PAGE 40

CHAPTER 5 OPTIMIZATION OF A HIGH-FIDELITY MORPHING WING A high-fidelity morphing wing model is fabricated to demonstrate the advantage of the morphing wing geometry over that of conventional fixed wing geometries. Initial coefficient of lift and efficiency curves are developed for flat plate and reflex wings. This data is used as a comparative tool to measure the performance of the morphing wing shapes. The optimization procedure is conducted for maximum lift and maximum efficiency through a range of angle of attacks. All experiments are conducted with an unconstrained genetic algorithm. Model Description This model is an elliptical planform, thin plate wing (Figure 5-1) consisting of a single layer, plain weave carbon fiber skin and arced unidirectional carbon fiber spars, which follow the outer curvature of the planform. As with the one-design-variable model, these spars transfer loads such that when the wing shape is altered at the root, a similar shape will transfer throughout the span of the wing. While the model is not capable of transferring the exact airfoil from the root to the tip it does significantly change the wing shape, which is desired for this experiment. The model has two points of actuation, which are positioned at the quarter chord and the trailing edge. There are two attachment points fixing the wing to the fuselage structure, which are located at the leading edge and trailing edge. The attachment point at the leading edge permits rotation only. The attachment point at the trailing edge permits both rotation as well as longitudinal, linear motion. The quarter chord actuation point significantly modifies the wing camber on the order of 11%, while the trailing edge point of actuation allows for 4% wing reflex. 40

PAGE 41

Using the VIC system, a full-field, three-dimensional image of the morphed wing shape, as shown in Figure 5-2, is developed at 16 servo positions. To illustrate how the airfoil at the wing root relates to the airfoil at the tip, a topographic plot with airfoil slices from the 20% and 80% span (Figure 5-3) is shown. These servo positions (250,230) yield a +7.1% camber and .6% reflex. For comparison, these plots are also provided for servo positions 1,1 (Figure 5-4) which is effectively an inverted wing shape of the previous. As plotted in Figure 5-5, an array of airfoil shapes from the 20% span is illustrated to show the morphing capabilities of the model. Comparative Lift and Efficiency Curves In order to determine how well the morphing wing performs, the coefficient of lift and efficiency versus angle of attack are determined for two fixed wing shapes. The first shape is an approximate flat-plate wing obtained by setting servo positions to (130,100) and the second is a reflex wing obtained using servo positions (250,110). Note the abrupt stall of the flat plate (Figure 5-6) at approximately 11 and its poor efficiency throughout the alpha sweep. The wing with reflex has less aggressive stall characteristics at approximately 15 angle of attack with a much higher lift coefficient than that of the flat plate wing. It also has a much higher efficiency, which carries over a large range of angles of attack. This data is used as a comparative tool to measure the performance of the morphing wing shapes. The optimization procedure for maximum lift, then maximum efficiency for several angles of attack is performed. The results from these experiments are plotted with the two fixed wing shapes and evaluated for improved performance. Maximizing Coefficient of Lift Curve The optimization procedure is conducted for six angles of attack from -5 to 20 in 5 increments, which is sufficient for evaluation of its performance. The result of the experiments yield wing shapes that produce maximum lift for a given angle of attack. By fitting a curve 41

PAGE 42

through these datum points a simple polynomial equation is calculated. It is this equation that correlates maximum lift to angle of attack. The genetic algorithm is capable of converging within just six generations when using a well distributed initial population as illustrated in Figure 5-7 (bottom left). The topmost plot in the same figure shows the rapid convergence of the mean performance with the best performance. The rate of convergence can also be seen from the lack of diversity of the servo scatter. The pattern of the initial population is still easily defined because the algorithm immediately hones in on the region of the optimal position. Note that in this case, a second optimization procedure is conducted, this time with an initial population narrowed around the optimal region. Once again the algorithm hones almost immediately on the optimum servo positions, which is well illustrated in the bottom right plot in Figure 5-7. This is a special case due to the simplicity of the objective function and the small coefficient of variance in the data. In the following maximum efficiency tests it is shown that convergence will not occur so rapidly. The performance of the morphing wing as compared to the two fixed wing geometries is plotted in Figure 5-8 (A). By morphing the wing throughout the alpha sweep, a much greater lift coefficient is obtained. Note the benign stall characteristics of the morphing wing, which has little decrease in lift from 15-20. No tests are conducted beyond the 20 angle of attack due to the vibrations encountered within this stall region. Having completed the tests for maximization of lift coefficient, the plot of optimal servo positions versus angle of attack is acquired and plotted in Figure 5-8 (B). As is expected, the maximum lift at any given angle of attack will have no reflex and is why the second servo always converges to position 1. These servo position curves can now be used in a control system, with angle of attack as input, to yield maximum lift performance. 42

PAGE 43

Maximizing Efficiency Curve Efficiency is calculated by dividing the lift coefficient by the drag coefficient. As determined in Chapter 4, the coefficient of variance with such a calculation is significantly greater than that associated with the lift coefficient alone. For this reason, it may take as many as three times the number of generations for the genetic algorithm to converge on the optimal servo positions as compared to the prior set of experiments. The optimization procedure is conducted at seven angles of attack, which are sufficient to populate an efficiency curve. This efficiency curve is then compared to the efficiencies of the flat plate and reflex wings. It is seen in Figure 5-9 (top) the algorithm requires approximately 20 generations before convergence. Note how the mean performance varies throughout the experiment while the best performance will improve incrementally. This is due to the increased complexity of the objective function and the increase in the variance in data. As a result, the genetic algorithm selects much broader and more diverse servo positions to find the optimal region. The scatter plot in Figure 5-9 (center) best illustrates the diversity in selected servo positions. The final graphs of the same figure show the varied, well distributed initial population (bottom left) and the final, converged upon optimum servo positions (bottom right). The performance of the morphing wing with the two fixed wing geometries are plotted in Figure 5-10 (A). Morphing the wing to the determined optimal shapes through the alpha sweep provides a significant improvement in efficiency up to the region of stall. Notably, at 2.5 angle of attack, the morphing wing has twice the efficiency value as that of the other wings. Having completed the tests for maximization of efficiency, the plot of optimal servo positions versus angle of attack is acquired and plotted in Figure 5-10 (B). 43

PAGE 44

Above the servo positions are the optimized wing shapes rotated to the corresponding angle of attack. Using the equations of these curves in a control system, maximum efficiency as a function of angle of attack can be achieved. These unpredictable curves illustrate the need for optimization procedures for highly complex morphing wing structures. 44

PAGE 45

Figure 5-1. The two-design-variable model morphs the wing geometry through change of camber and reflex. Figure 5-2. The full field, three-dimensional image, developed by the VIC system, is overlaid on an image of the morphed wing illustrating deformation in the Z-direction. 45

PAGE 46

Figure 5-3. Morphed wing surface (Servo Positions: 250,230) with airfoil slices from the 20% and 80% span yields a negative camber and negative reflex. Figure 5-4. Morphed wing surface (Servo Positions: 1,1) with airfoil slices from the 20% and 80% span yields a positive camber and positive reflex. 46

PAGE 47

Figure 5-5. An illustration of the intermediate wing shapes from the 20% span shows the morphing capability of the two-design-variable model. Figure 5-6. The lift coefficient and efficiency is plotted using set flat plate and reflex wing geometries. 47

PAGE 48

Figure 5-7. These sample plots illustrate the convergence of the genetic algorithm upon the optimal servo position for maximum lift at 10 angle of attack. 48

PAGE 49

Figure 5-8. Morphing wing data for optimized lift coefficient. A) The morphing wing performance as compared to the fixed reflex wing and flat plate. B) The optimal servo positions as a function of angle of attack with the corresponding wing shape. 49

PAGE 50

Figure 5-9. Sample plots showing the convergence of the genetic algorithm on the optimal servo position for maximum efficiency at 10 angle of attack. 50

PAGE 51

Figure 5-10. Morphing wing data for optimized efficiency. A) The morphing wing performance as compared to the fixed reflex wing and flat plate. B) The optimal servo positions as a function of angle of attack with the corresponding wing shape. 51

PAGE 52

CHAPTER 6 CONCLUSION AND RECOMMENDATIONS FOR FUTURE WORK Conclusion A procedure has been established to achieve proper shape management of morphing wing structures. It has been demonstrated that utilizing a genetic algorithm with wind tunnel hardware-in-the-loop is an effective means of determining optimal wing shape functions. The use of these shape functions enables a morphing aircraft to achieve optimal flight performance in all flight regimes. To demonstrate the procedure, a two design-variable morphing wing is fabricated with the capability of altering maximum camber and trailing edge reflex. Two experiments are conducted for maximizing the lift coefficient and maximizing efficiency, which resulted in optimal servo position versus angle of attack shape functions. For a performance comparison, the wing shape is set as a flat plate and again as a medium camber wing with reflex. These fixed wing shapes are tested for lift and efficiency performance from -5 to 25 angle of attacks. Plotting the performance of the morphing wing with that of the two fixed wings illustrates the dramatic improvement achieved by morphing wing structures. In particular, while the slope of the lift coefficient curve remains the same for each data set, the morphing wing shows a 63.4% improvement over the fixed reflex wing. Another remarkable attribute of the morphing wing is its benign stall characteristics. While more data must be collected to determine the wings performance at angles of attack greater than 20, it is clear that such a wing can be used to control and mitigate wing stall. The experiments for determining maximum efficiency show an equally impressive increase in performance. Considering the range of -5 to 5 angle of attacks, typical for straight and level flight and subtle altitude changes, efficiency increases as much as 62.1%. 52

PAGE 53

These results support high confidence in the optimization procedure. The genetic algorithm is capable of proper convergence despite the variance in collected wind tunnel data. It is in the authors opinion that the procedure will effectively determine optimal wing shapes for morphing wings of many design variables and higher morphing capability. Recommendations for Future Work There are many valuable wing shape functions that can be attained using this procedure which have yet to be explored. By simply applying constraints to the genetic algorithm many new ideas for flight optimization arise. One such experiment could be to vary the wing shape as a function of velocity to attain a desired lift. For a given angle of attack, a high velocity would yield minimal camber while a low velocity would yield a comparatively high camber. Such a shape function would minimize drag thereby increasing efficiency. Another interesting study would be to apply the constraint of a desired pitching moment for each angle of attack to ensure longitudinal stability. Using this constraint while maximizing lift at very high angles of attack could result in the mitigation of wing stall. Such a wing shape function would yield a controlled, rate of descent flight mode useful in an auto-land feature. Of course these shape functions are of no value unless they are utilized in an on-board control system for real-time wing morphing. Such a control system will receive data such as angle of attack or velocity and in response, command the position of the actuators. The servo controller produced by Scott Bowman, which is utilized exclusively for this research, is capable of such a task. In a given mission profile for an aerial vehicle, several flight modes may be commanded. For example, at takeoff the vehicle may require a wing shape that maximizes rate of climb. When required altitude is met, another wing shape may be commanded to maximize speed to a 53

PAGE 54

specified waypoint. At this point the vehicle may be required to conduct surveillance, which would require a wing shape that would yield maximum efficiency for extended flight time. After completion of surveillance the vehicle may be commanded to return to another waypoint where it is required to perform a steep descent angle for landing in a small, confined field. The ability to switch between these distinct optimal wing shape functions is necessary to maximize the overall performance of a morphing winged aircraft as it transitions between flight modes. 54

PAGE 55

LIST OF REFERENCES [1] Haenel, R. T., Creep Deformation of Pin-Jointed Structures, MS Thesis, The Pennsylvania State University, Pennsylvania, 1955. [2] Gooders, J., and Pledger, M., Birds of North America, Edison, New Jersey: Chartwell Books Inc., 1987. [3] Davidson, J., Chwalowski, P., and Lazos, P., Flight Dynamic Simulation Assessment of a Morphable Hyper-Eliptic Cambered Span Winged Configuration, 2003 AIAA Atmospheric Flight Mechanics Conference and Exhibit, Austin, TX, AIAA Paper 2003-5301, 2003. [4] Shyy, W., Berg, M., and Ljungqvist, D., Flapping and Flexible Wings For Biological and Micro Air Vehicles, Progress in Aerospace Sciences, Vol. 35, No. 5, 1999, pp. 455-506. [5] Ifju, P., Jenkins, D., Ettinger, S., Lian, Y., and Shyy, W., Flexible-Wing Based Micro Air Vehicles, AIAA Publication, AIAA Paper 2002-0705, 2002. [6] Albertani, R., Hubner, J., Ifju, P., Lind, R., and Jackowski, J., Experimental Aerodynamics of Micro Air Vehicles, SAE World Aviation Congress and Exhibition, Reno, NV, 2004. [7] Stanford, B., Abdulrahim, M., Lind, R., Ifju, P, Investigation of Membrane Actuation for Roll Control of a Micro Air Vehicle, Journal of Aircraft, Vol. 44, No. 3, 2007, pp. 741-749. [8] Gano, S., Renaud, J., Batill, S., and Tovar, A., Shape Optimization for Conforming Airfoils, 44th AIAA/ASME/ASCE/AHS Structures, Structural Dynamics, and Materials Conference, Norfolk, Virginia, AIAA Paper 2003-1579, 2003. [9] Gano, S., and Renaud, J., Optimized Unmanned Aerial Vehicle with Wing Morphing for Extended Range and Endurance, 9th AIAA/ISSMO Symposium and Exhibit on Multidisciplinary Analysis and Optimization, Atlanta, GA, AIAA Paper 2002-5668, 2002. [10] Waszak, M., Davidson, J., and Ifju, P., Simulation and Flight Control of an Aeroelastic Fixed Wing Micro Aerial Vehicle, 2002 AIAA Atmospheric Flight Mechanics Conference and Exhibit, Monterey, CA, AIAA Paper 2002-4875, 2002. [11] Skillen, M., and Crossley, W., Modeling and Optimization for Morphing Wing Concept Generation, NASA Publication CR-2007-214860, March 2007. [12] Albertani, R., Experimental Aerodynamic and Static Elastic Deformation Characterization of Low Aspect Ratio Flexible Fixed Wings Applied to Micro Aerial Vehicles, Ph.D. Dissertation, University of Florida, Gainesville, FL, 2005. [13] Sytsma, M., Aerodynamic Flow Characterization of Micro Air Vehicles Utilizing Flow Visualization Methods, MS Thesis, University of Florida, Gainesville, FL, 2006. 55

PAGE 56

[14] Bilgen, O., Kochersberger, K., Diggs, E., Kurdila, A., and Inman, D., Morphing Wing Aerodynamic Control via Macro-Fiber-Composite Actuators in an Unmanned Aircraft, 2007 AIAA Conference and Exhibit, Rohnert Park, CA, AIAA Paper 2007-2741, 2007. [15] Williams, R., Inman, D., and Keats-Wilkie, W., Nonliniear Mechanical Behavior of Macro Fiber Composite Actuators, Center for Intelligent Material Systems and Structures, Department of Mechanical Engineering, Virginia Polytechnic Institute and State University, Undated material. [16] Claverie-Bohn, C., Dynamic Antifouling Structures and Actuators Using EAP Composites, Ph.D. Dissertation, University of Florida, Gainesville, FL, 2004. [17] Strelec, J., Lagoudas, D., Khan, M., and Yen, J., Design and Implementation of a Shape Memory Alloy Actuated Reconfigurable Airfoil, Journal of Intelligent Material Systems and Structures, Vol. 14, No. 4-5, 2003, pp. 257-273. [18] Lu, K., and Kota, S., Design of Compliant Mechanisms for Morphing Structural Shapes, Journal of Intelligent Material Systems and Structures, Vol. 14, No. 6, 2003, pp. 379-391. [19] Rechenberg, I., Evolutionsstrategie: Optimierung technischer Systeme nach Prinzipien der biologischen Evolution, FrommannHolzboog, Stuttgart, 1973. [20] Whitley, D., An Overview of Evolutionary Algorithms, Journal of Information and Software Technology, Vol. 43, 2001, pp. 817-831. [21] Rasheed, K., and Hirsh, H., Learning to be Selective in Genetic-Algorithm-Based Design Optimization, Artificial Intelligence for Engineering Design, Analysis and Manufacturing, Vol. 13, 1999, pp. 157-169. [22] Herrera, F., Lozano, M., and Molina, D., Continuous Scatter Search: An Analysis of the Integration of Some Combination Methods and Improvement Strategies, European Journal of Operational Research, Vol. 169, No. 2, 2006, pp. 450-476. [23] Herrera, F., Lozano, M., and Snchez, A., Hybrid Crossover Operators for Real-Coded Genetic Algorithms: An Experimental Study, Soft Comput, Vol. 9, 2005, pp. 280. [24] Naujoks, B., Willmes, L., Haase, W., Bck, T., and Schtz, M., Multi-point Airfoil Optimization Using Evolution Strategies, European Congress on Computational Methods in Applied Sciences and Engineering, Barcelona, 2000. [25] Liu, B., Two-Level Optimization of Composite Wing Structures Based on Panel Genetic Optimization, Ph.D. Dissertation, University of Florida, Gainesville, FL, 2001. [26] Hemker, T., Hardware-in-the-Loop Optimization of the Walking Speed of a Humanoid Robot, CLAWAR 2006, Brussels, Belgium, 2006. 56

PAGE 57

[27] Kammeyer, M., Wind Tunnel Facility Calibrations and Experimental Uncertainty, The Boeing Company, American Institute of Aeronautics and Astronautics, 1998. [28] Springer, A., Uncertainty Analysis of the NASA MSFC 14-Inch Trisonic Wind Tunnel, 1999 37th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, AIAA-99-0684, 1999. [29] Lopez-Anido, R., El-Chiti, F., Muszyski, L., Dagher, H., Thompson, L., and Hess, P., Composite Material Testing Using a 3-D Digital Image Correlation System, American Composites Manufacturers Association, Tampa, Florida, 2004. [30] Muszyski, L., Lopez-Anido, R., and Shaler, S., Image Correlation Analysis Applied to Measurement of Shear Strains in Laminated Composites, SEM IX International Congress on Experimental Mechanics, Orlando, Florida, 2000. [31] Schmidt, T., Tyson, J., and Galanulis, K., Full-Field Dynamic Displacement and Strain Measurement Using Advanced 3-D Image Correlation Photogrammetry, Part I. Experimental Techniques, Vol. 27, No. 3, 2002, pp. 47-50. [32] Helm, J., McNeill, S., and Sutton, M., Improved 3-D Image Correlation for Surface Displacement Measurement, Optical Engineering, Vol. 35, No. 7, 1996, pp. 1911-1920. 57

PAGE 58

BIOGRAPHICAL SKETCH Frank Boria was born in Portsmouth, New Hampshire and spent most of his childhood in York, Maine. When he was 13 years old, he and his family moved to Fernandina Beach, Florida. During his junior year of high school, Frank won a scholarship to train for his private pilot license. Having trained in a Citabria 7ECA, a tandem two-seat tail-dragger, Frank received his pilot license and developed a passion for aviation. After high school he moved to Tallahassee, Florida where he attended Lively Technical Center and received his Airframe and Powerplant certifications. With these certifications he moved to Lake City, Florida where he spent 7 years working at TIMCO, a third-party aircraft repair station. His experience was primarily focused on aircraft structures for DC9, Boeing 727, and C130 aircraft. With the understanding of how to fly and maintain an aircraft, his interests leaned toward understanding the physics of flight and the process of design. During the final 2 years of work at TIMCO, Frank attended evening classes at Santa Fe Community College where he received his Associate of Arts degree. He transferred to the University of Florida where, in 2004, he received his Bachelor of Science degree in aerospace engineering. He continued his education at the University of Florida, and in 2007, he received his Master of Science degree in aerospace engineering with a focus on solid mechanics, design, and manufacturing. Frank now resides in Gainesville, Florida with his beautiful bride, Rikki, and three wonderful children, Addie, Olivia, and Christian. 58


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101113_AAAARG INGEST_TIME 2010-11-14T01:13:20Z PACKAGE UFE0021478_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 1053954 DFID F20101113_AAEAPP ORIGIN DEPOSITOR PATH boria_f_Page_01.tif GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
1a19fdbec396a328465b280dd90f5138
SHA-1
54fcac8e23091ad9290abbcc67ea3a31a0a4e2d5
25271604 F20101113_AAEAKS boria_f_Page_45.tif
f51b7f987b2824254a3db5afd80b6767
04b0c1478413c4359218b23df9e46d160a4ecec3
34703 F20101113_AAEAUM boria_f_Page_52.QC.jpg
dde2ad0e52d1f012f34dd2baf782e53b
4d5891893e15be15d8a8274d6442cad4c5c14168
F20101113_AAEAPQ boria_f_Page_02.tif
d541ca31977cab0896594de98a4ddaa7
9d59c67192f30e72ad7c4a903f1381452ec42ea8
1573 F20101113_AAEAKT boria_f_Page_33.txt
94663e5fd8e7d8305d8d0b6a28716ff5
aca377d2f28bdc4ed17b3ac222e8dc7e15beceb7
7808 F20101113_AAEAUN boria_f_Page_35thm.jpg
859c39dd7f1ae4a4df2de76aaeec42a0
12d16e5fad374066d60b3ccd291f8843aeb2e100
F20101113_AAEAPR boria_f_Page_04.tif
c5e44ffe5ac76a696ae7a6e170946555
e460a737071203a5a40769a7204210778e7dac58
8697 F20101113_AAEAUO boria_f_Page_14thm.jpg
0ed101217d94d0c6130cd1c8e240e844
b4909d343b326ccf4973e81a4578f198d40b459c
F20101113_AAEAPS boria_f_Page_05.tif
f7811052b1652ac132a21186fb9d5b86
dc0803e2449229229cea47c32747b3f40dfa024c
116651 F20101113_AAEAKU boria_f_Page_28.jp2
08c5e7656dd358dc7b704550847f9e86
ed795a637db37d5f00552fb97ed1198bad3a3533
6923 F20101113_AAEAUP boria_f_Page_10thm.jpg
b9fbf6a0d474c038ea0a78955908bfc4
8d8a77ff3349eab6ee0257119a31c27202d80db1
F20101113_AAEAPT boria_f_Page_12.tif
412dd76f84cca889988c0b2af32599a2
4e98054c927d29e0b9f062dedafe4d01969cf762
17058 F20101113_AAEAKV boria_f_Page_47.QC.jpg
b5fb47172ea624466304746d5aaa8093
5f1db34db565cda98e6bae20ba820f8298421450
3189 F20101113_AAEAUQ boria_f_Page_36thm.jpg
3fc6cf2811a1fda1fce716396989d692
86de70c2c263c17294ba31df23d76e326bdb2087
F20101113_AAEAPU boria_f_Page_13.tif
c91f7dee21c060b46857e0dd45cb791a
0cd82ea81ef743e8b36d8d984bf325a6dada8d35
73372 F20101113_AAEAKW boria_f_Page_33.jp2
157585be7c68cc08d7a0e3155ab1fdb9
c0d22a5feaf8d573e385fb10b5b5378cfb406a2d
11259 F20101113_AAEAUR boria_f_Page_54.QC.jpg
4bcd91604d3feacbc0996325554a73f4
fd13d06a9291c1116cd042b54ad3a7151f48220c
F20101113_AAEAPV boria_f_Page_14.tif
cfd9a8171fc90271354cf58c96bbbaea
9198d0ad8e6953c731a538b9aecdaa88c3bc8cd4
26917 F20101113_AAEAKX boria_f_Page_10.QC.jpg
a643a5990318ac17c17f9730009dc424
ba751ec30ce410998854edad099a78887d3310ce
26029 F20101113_AAEAUS boria_f_Page_05.QC.jpg
8955c8728379c1a56e68ed60232bc3bf
49615c8a05576c84b3e73005eeeb02b89258b852
F20101113_AAEAPW boria_f_Page_15.tif
31b770125ae360a852409f83686d772e
4f22ba1433330518dc029cabe9ef54bcd2acf270
546558 F20101113_AAEAIA boria_f_Page_26.jp2
c18bfa8c94c4da54678eb828920df30f
a11234bfd49338beafc078e9a819c2dfa5f6db7a
118258 F20101113_AAEAKY boria_f_Page_41.jp2
855e04577315e21c5a4801850faa99f5
220bf9fad5ae1ee84003f9c5e151f6d6f9fd3f1f
29051 F20101113_AAEAUT boria_f_Page_19.QC.jpg
f70b71325e909e9bd93fc87705aeb266
e4af0dfac73a658f33219bc3547dc8298b0b3901
F20101113_AAEAPX boria_f_Page_16.tif
82c8b10341d2aba4f885b18a4968b4c8
675dccb56af67a961a37f3f833b2e501e1ca264b
16246 F20101113_AAEAIB boria_f_Page_48.QC.jpg
74181fe296ce9658d0b6d3a7ea6e36f1
e58b3238f02b1b1cebf657f2fcfac931e44a1008
F20101113_AAEAKZ boria_f_Page_58.tif
146f304807db156acc4bb3da92d1717b
83fd0c9eb5ffddcb884ef8562d0f6e970279af25
18589 F20101113_AAEAUU boria_f_Page_26.QC.jpg
e12221a03466513d809d19d305bcd4fe
58b7a54de73e06f76fcc1efec0bb0a1b9931533b
F20101113_AAEAPY boria_f_Page_18.tif
036e1c14af6c90c8caadc68cf3cec551
5cb885831941475fcaee2283a6e98789ebe3bded
F20101113_AAEAIC boria_f_Page_57.tif
1553fc61b37e73f76ce0b90d703ec607
e84e229f2e8750c255b2eaf40987c84553a030ad
530 F20101113_AAEAUV boria_f_Page_02thm.jpg
3296af98a970450f90ec1ff9c8f50f33
10d196a5c29f6827080da43ef8f447de468a6390
66061 F20101113_AAEAID boria_f_Page_55.pro
d3ae106d5b934c23d1104168f9941d30
89d93b279fd7108f605a8d18c7c1eb6201f1ef44
5371 F20101113_AAEAUW boria_f_Page_50thm.jpg
4f9e3950b538ad6b4d8c322ed4320c9c
d0f25db822428873255edb488e7f2782c45c7b50
146270 F20101113_AAEANA boria_f_Page_08.jpg
e81d33b004bdbb8ee789673a13208d10
091966467a8f4cc7f69b79c2bf0b6dd958b01c3d
F20101113_AAEAPZ boria_f_Page_19.tif
9f039b99f304a56511b82a6b37e123c7
72f98f11c7799704286662202aa2fd6c149582f1
35676 F20101113_AAEAIE boria_f_Page_28.QC.jpg
6fd6e19f0f897dce8c0d86843091b245
c3644faef658fc37e5cf6a163ba91c1db657d115
32751 F20101113_AAEAUX boria_f_Page_12.QC.jpg
0173acf8bc81329d77273f62eaa17c1f
112b3d6ae0e36b26b9d4286e5cd61dc3aaeced44
86671 F20101113_AAEANB boria_f_Page_09.jpg
a98a168973049c9f2ce0a83637f09aee
d9f4078ee560aef5621ee5694a845fd66c3bc826
117041 F20101113_AAEAIF boria_f_Page_17.jpg
131420403811f124c31ba10867d4cc54
033d95f2ca427b0f4d559ee6ca1bc0790b0d8dc3
8032 F20101113_AAEAUY boria_f_Page_12thm.jpg
bbc29af92a2f2e536e2d2df10b166489
3041cd24de45632ee448fc51e909059f023f94dd
101083 F20101113_AAEANC boria_f_Page_12.jpg
9c51be445a4b2c2b13d3805950e21ab2
7a1a5105c09dea6d76dbe0184666ce8837bec706
36880 F20101113_AAEAIG boria_f_Page_30.QC.jpg
d46a46ebae0d6dee1a4e98e9144f42b8
a4df72e8476f779be114ff12417ddbec2bb1bc72
54149 F20101113_AAEASA boria_f_Page_42.pro
fb26103862d1b3c2a6d284a5b865a858
249a9e77544bdc428106e446686e34f3913940f6
33712 F20101113_AAEAUZ boria_f_Page_20.QC.jpg
fc6047587a99aad638bcf84f00b8d251
65553b85d0d536725bf814e2b3882c180d44ebc7
106118 F20101113_AAEAND boria_f_Page_14.jpg
3f8e95aaefd1c70aa9045490ae18582d
756cbbbab569e467158dd72d13f1125d18951ee4
5243 F20101113_AAEAIH boria_f_Page_51thm.jpg
a7550eba49fb06d57fb2a4a363bc1cbd
17b5c2587199153da3094014c9ab700c43876cb1
8130 F20101113_AAEASB boria_f_Page_45.pro
b113fe8799879aede0d19ce084e77076
71665f7c960144cd1301e56d8c88d5fd3825b22f
53716 F20101113_AAEANE boria_f_Page_15.jpg
bbd729a0287593a77bef2400429557c8
369776f654dfb4c52159ae16170783879294ebc3
51076 F20101113_AAEAII boria_f_Page_43.pro
64e5c5a8f6663735b81cb0dc97df1215
6e8a588bcbe36ea2bc40029fe9391b66089edb9a
15464 F20101113_AAEASC boria_f_Page_46.pro
3def4c3e58823507bfdd0fb80d091347
c022eb34f772bee8f6a51db54fc710fa27386990
101130 F20101113_AAEANF boria_f_Page_16.jpg
4cf084c2156cc0fda061e020eedca0c2
d69f5b19630be9a2a53aceb772ca57b66123e879
7074 F20101113_AAEAIJ boria_f_Page_34thm.jpg
a3efbae8ab8161442ced03f26b987902
53b423b289a9b7f017958fa95dab997c2f233c7f
12262 F20101113_AAEASD boria_f_Page_48.pro
2adce497bc11ded0b35b7e22fc6e3f5f
3cc3a65894006e772a9312f417359bde6aa8bdbb
107365 F20101113_AAEANG boria_f_Page_18.jpg
8bf6728cfd8965c87800a079a83bbfaa
397a38f9bc09679657177c67088e2b2864547fb5
47479 F20101113_AAEAIK boria_f_Page_51.jpg
240d684d848aa0ad13ef3f208c2d39c9
04e57ec73516209291fd2fd284d3acc7f6827097
12926 F20101113_AAEASE boria_f_Page_49.pro
e4e18cd3d62e25e70f44c60af14fea52
5e64cf6c960aaa99f36cb81e837ce5f54cf5882e
89812 F20101113_AAEANH boria_f_Page_19.jpg
ff6c529f3a65cc43f78b0728ef754c89
6da2feca40c2673d1239cbb7baa5ca03530e2240
44963 F20101113_AAEAIL boria_f_Page_47.jpg
89c73b781d5803c0acce11955fb84c02
f18311c627ca1ed0d22c21a29a839bf0e0eafdde
13391 F20101113_AAEASF boria_f_Page_51.pro
13527233e98d159515e110cd1a94c82a
50c2500f9b4302e6a652885af15db59cf7cda7e0
105334 F20101113_AAEANI boria_f_Page_21.jpg
b545cad3c1de631ab740d2e9eb8a113b
5f4cded57f9f8e3c4901a7a9148c0b350beaa688
F20101113_AAEAIM boria_f_Page_20.tif
88fe6350ecb64606be4130d603d87dee
f19406ca48b7de65618cdbcdc572189aff25f930
49406 F20101113_AAEASG boria_f_Page_52.pro
63b4be610c0cffe1d02a9463a85efd20
14487235e3df1f40a7f415bfe89d36ead9178159
110630 F20101113_AAEANJ boria_f_Page_22.jpg
837c4946583f536a30639d49c4d93911
5f4b23d77952db61f8e6598dc0f6139d16f84e2e
F20101113_AAEAIN boria_f_Page_09.tif
ca2ece3c3174c9b366e9adbf2a2005ef
0de3d9deaf5ad348cddc0ee29631d65b0f663d78
14914 F20101113_AAEASH boria_f_Page_54.pro
8a35f80240f8fede81742064bf4e03c7
b56db8b442f6c697c130c0dd3f91baab11829fe9
79903 F20101113_AAEANK boria_f_Page_23.jpg
d090bb44cca0f08b584b34cb250af538
c3401da80426481768a5b2a763e131dee494c1dc
F20101113_AAEAIO boria_f_Page_47.tif
51bf0991067ca8a2325942ccb17ca963
8b0f9968aef33586aead67e95a56eda5bb5e7448
88 F20101113_AAEASI boria_f_Page_02.txt
9bb108fd105b0e050d6db3f05914bd28
1192bf5c9030b33602d30ac5c232c8a036b1e211
76355 F20101113_AAEANL boria_f_Page_24.jpg
2638e211d42c3381d0584b52d247cbc5
0dea6b02e7e8848ccc59acd26819263ede1fd866
27896 F20101113_AAEAIP boria_f_Page_01.jpg
60a4ca810076dfaa35d67f337c26c1ca
2d1169d9439db5c94407836ef95c4a9eaed11327
73 F20101113_AAEASJ boria_f_Page_03.txt
762abb58cce5d1e708dfebff4f13a150
b700f123134f7fd39d8ae50cc28e6e4d89e1b4c0
61326 F20101113_AAEANM boria_f_Page_25.jpg
5dac8d243a8b01c1bdcf0ddede3f2952
32d2cd79de1391e8b44d0351c429a50eceec8800
9419 F20101113_AAEAIQ boria_f_Page_56thm.jpg
5342e84a0dadd68d7cbc251cabe206d8
51ac61a56e374d3b53590fba5aa921359411ad29
610 F20101113_AAEASK boria_f_Page_06.txt
ee0571c9eb9e40c68d070f4bf51871cf
dc20bfb04b7780201ed8829d2ea1702433cfa4b1
56161 F20101113_AAEANN boria_f_Page_26.jpg
8f7cdf42db84c697d3fec4e3d328aa34
24b59ac4e130bcc867ea7978c2e9fd89c7580772
2651 F20101113_AAEAIR boria_f_Page_56.txt
35c89d6a345068a35aa32daf9bb6945e
a8f53c0c679a4c96a90496cd62525900af96d136
1734 F20101113_AAEASL boria_f_Page_09.txt
896f7caddabdc399c1cdefec3e92a487
f28ec0213de6278fb9676c5cd57fec895d1e2497
104216 F20101113_AAEANO boria_f_Page_27.jpg
98ce242437a1f797150b6d3a702247f3
1199241af15f2a342d1bc5d77bccf7c1341494da
2046 F20101113_AAEASM boria_f_Page_12.txt
5a7e2c03e7c83ed60774dc819a195642
a533ce4709e2172cbc0af32e43fffea932a7e5c1
107839 F20101113_AAEANP boria_f_Page_28.jpg
826ed846c6f5b31b010c16d2c7d192cf
998b2f98e879ca153b213c3ca1af504df81af554
2036 F20101113_AAEASN boria_f_Page_13.txt
6c5a326924d90dccccb4a73633a2205b
3f6698b53cd5734cf838aa933f33c6cf6f2f1eb7
111518 F20101113_AAEANQ boria_f_Page_30.jpg
9a9ba25be096e737056417a955e7c477
7d8f5c8ea390f0c4c76c95502d4db1c0da17d536
2169 F20101113_AAEAIS boria_f_Page_28.txt
bc1d001d5dbd19515e5ed51fbba568c1
9877c4aa8e5071cdf6f327ad1544d0f1cf1322d2
2132 F20101113_AAEASO boria_f_Page_14.txt
1ea52e16347d5c64b74e3046ee505217
711eb50bb372247248b9305087d281affb93aff7
103719 F20101113_AAEANR boria_f_Page_31.jpg
b03d756df9d312a113aeb8e407210b50
2d8925fd758d73f7fa90ae9f7b37ace28c00d750
F20101113_AAEAIT boria_f_Page_36.tif
c3149cdfd589de3769d25c5f1eab3f60
120423ef728f7a7dc06f05c846948f8c5f16106f
1008 F20101113_AAEASP boria_f_Page_15.txt
0516ad9af981909f8f4c27f668af6078
946029106bedf1751dc5200a1e28fc1afb152ea6
72918 F20101113_AAEANS boria_f_Page_33.jpg
5b3fb58001ddae17632c44b906bff252
1e0df110a2fb0dc993051ec89332430eed17d329
17277 F20101113_AAEAIU boria_f_Page_50.QC.jpg
71e6387275f9dd5dc4bf48c3374691c5
8a5fc03f6708c75b2956c1fd082176a50ab0b354
2048 F20101113_AAEASQ boria_f_Page_16.txt
c56e6a2725f9c90621f62cac9f0b749f
6e92023be4575b303a53fc0f98ce0293234b86bd
73367 F20101113_AAEANT boria_f_Page_39.jpg
6d1102b03d9265ada4c5fb48ecc8e507
8fd1bb3813bc1a1b0024c1aac69129933f2d2d20
51922 F20101113_AAEAIV boria_f_Page_53.pro
ba15d2d21ef306c09bf676b7967ee8d9
b3ae8692ee5aaffce557f9bcb8caeb7c4ca9c5b9
2093 F20101113_AAEASR boria_f_Page_18.txt
ce4f5842df655e8111b3780635600c3d
ef5dd89a682fb9e28a5214d79e7d4f2176167264
110159 F20101113_AAEANU boria_f_Page_41.jpg
3b17543c04459bd86e5dfe78a0ef1588
1a67447413ab36a4e395e4612c6e4ff68f18ed30
1178 F20101113_AAEAIW boria_f_Page_03.QC.jpg
5e1cc0a69a23e6c18020159e1f58db80
0333dcb307d1ad9c3a404b392334d994ef2064db
2084 F20101113_AAEASS boria_f_Page_20.txt
6609bcfee6408cbfca9df0dc91347baa
9bc3ecc6068c87f854ca7a6186b83e4b6f20f5a7
108091 F20101113_AAEANV boria_f_Page_42.jpg
27f2f78b1b923bd77d21706b5664947a
c068432e3c441935c0cfcb7594ea156cfa241f42
38578 F20101113_AAEAIX boria_f_Page_55.QC.jpg
6c76b002073bb3e8172492e059faf090
a9b1f674c67849f58de31c9b4fd89d7aff6acc22
2143 F20101113_AAEAST boria_f_Page_21.txt
d0ec504a0bc7c9e3d32f68b245e7e27f
b46a66a9be85a529485a400c6a56b9e90ba16939
103775 F20101113_AAEANW boria_f_Page_43.jpg
a542b87039fe4c535fe4511abae6bf88
7a52452006d21f0f3031e362cada59c68a34dd34
2840 F20101113_AAEAIY boria_f_Page_05.txt
750793ef6218c45b2d7b4d5a6f49b44e
5e70c9eddd5bf46a2aaa21ef56fb095a31b4e91a
1546 F20101113_AAEASU boria_f_Page_23.txt
b6fad17902109ccb202e3c522043030a
cfd953507bdc313f8d5e144886c18fc95cd463c6
F20101113_AAEAIZ boria_f_Page_06.tif
b693ee1b826b509e1ab3b4b090c42441
f361cc5ea5951a2b16ffcb5446b26c1a9f03c136
505 F20101113_AAEASV boria_f_Page_24.txt
7bc3303f09c283c7d88275488e0b3198
4f7746302aac10a23fef3fed045862000ed0bdb2
71038 F20101113_AAEANX boria_f_Page_45.jpg
06ee82d08999231db161dc325973d450
1381f1f95b4919e35fe560c179a9c38ec2ccccc1
192 F20101113_AAEASW boria_f_Page_26.txt
91916d5d4f9b24eebb88d97c85d3f225
c6182da63e3176c252f8f19ea7e73598cd74c59e
2232 F20101113_AAEALA boria_f_Page_41.txt
2808242f5dd5281dd8fda58cab165079
05d940018a69ca82c982f02831154710d0decd0a
45883 F20101113_AAEANY boria_f_Page_48.jpg
33cc6c1b3afcb27058a96e0484143e23
61599fbf1956879e9acaeb0dab466ff9c4afa45c
F20101113_AAEASX boria_f_Page_27.txt
5547d0cfefed82f46467153c8d30f893
5b7fb8591c321de4ad938579918c72aa27b07f9a
139195 F20101113_AAEALB boria_f_Page_55.jp2
61a6d637c9ea6ffdb3172290897e80a9
531cdee47ae28dbca23577c283526a17d414202a
45693 F20101113_AAEANZ boria_f_Page_49.jpg
db09e9ebe361d843438aebf62e6d1b18
487e321ed4cb7802a0b01603e3ca555329305c01
2086 F20101113_AAEASY boria_f_Page_29.txt
ea182bfcb219ea630a57c1d7fd35ecd7
aa2527b254cc77adb45ea8a48d52d89c758e9702
5569 F20101113_AAEALC boria_f_Page_48thm.jpg
49897ad7dc04f034c55e29d911144e7c
04559ac64b7445fe99d574b9aac0f0ba216e84a8
2225 F20101113_AAEASZ boria_f_Page_30.txt
03f17f7df28915d28f5bbe60df03ca96
ae51b2f522befa8d36f01580f1bf807e0f0b0aad
763530 F20101113_AAEALD boria_f_Page_46.jp2
fe255c43f44669e0c84f98ff16502461
daaba8c6d70a9ea64ad5c00d81aea8f0d1dc8586
F20101113_AAEAQA boria_f_Page_21.tif
32f4d5aa2e1c7c50f16d59bd685dc6bc
5526851f23b1b7f7b6c0899ec2fc4049147e17d9
F20101113_AAEALE boria_f_Page_22.tif
0d5d57598c5e946a0b217a5f30d34557
c77f66ce8869848598d8b4a7694ad79f1b7174cd
F20101113_AAEAQB boria_f_Page_23.tif
b645f9c70f76241485fb2ec22a1ff19f
e280879fd249bd4c44b4ee7affd46805d0a81837
783 F20101113_AAEALF boria_f_Page_46.txt
49e90a221c0e3ca5939c4714ab7cd1a8
cc0184790b47d1a2e16ad4566180858ff34814c7
F20101113_AAEAQC boria_f_Page_25.tif
7d13ffcb4beccf2e6562ce3180052362
7dc4dbb5b7b2e1692628740de6db583feaab2eb9
38226 F20101113_AAEALG boria_f_Page_41.QC.jpg
05e458b68b17b025c35a8d02835ff1bb
224f1e439811f9b7757b7543273626f89980ad7a
37249 F20101113_AAEAVA boria_f_Page_22.QC.jpg
ba7068f19abf9b0478e5dac35492efb4
128064347797d55521f39869d661cee77bf05008
F20101113_AAEAQD boria_f_Page_30.tif
65c87cd8f31666e8bdb0392a0155ab2f
21ef15eb89fbf2805ff1cdd3b80f0bde28cb0aa9
104578 F20101113_AAEALH boria_f_Page_29.jpg
e8ef433fda7758a6f2b33b89192c0fd5
69b2a5ad95f47f58bed1b410030fa12415e535e1
7773 F20101113_AAEAVB boria_f_Page_40thm.jpg
fb4f0639f18f063f74a9df934159fa94
a6bb01731b745d9b663d113d11233233c6fcb719
F20101113_AAEAQE boria_f_Page_31.tif
da4b03ff89695c13d62847ec4459c933
90b5503126d688b9a26034933b19ac52d7f5d2f3
6483 F20101113_AAEALI boria_f_Page_38thm.jpg
5358c0e5557c795e3cc04c7cb2bdd114
f054a9d6d6ee05e6d94a82fb3dfbbd684281dd00
9452 F20101113_AAEAVC boria_f_Page_55thm.jpg
9dfed183eb08a33b96c741c783bf6e79
f5b5716ca0832f5d161a878f1b3af28c156d1922
F20101113_AAEAQF boria_f_Page_32.tif
fce99357d628c3eafd34f6a93bc39024
d041fd5f8803dd9ea3228d5bc570299a261cf740
27258 F20101113_AAEALJ boria_f_Page_23.QC.jpg
d5b2a803d990906eda16112734113f24
95c41e9d66f15a2d746c09703dfd3570c5a2a3dd
32842 F20101113_AAEAVD boria_f_Page_16.QC.jpg
1fc7e71b133ade75e64a60ae40269b9d
96710a6f473c57006d0eb6828dff902b9db35a96
F20101113_AAEAQG boria_f_Page_33.tif
feb37e4fbf5e9431c4f028a6f8fe1819
5db6db7fedb2f8c470dd708d45caecc164a7785a
28644 F20101113_AAEALK boria_f_Page_58.QC.jpg
be9055d2b477254f5bb2c41e1390ee99
48950479b25780c8b84b1f243cca7ba4462985cb
39784 F20101113_AAEAVE boria_f_Page_08.QC.jpg
457336175b05f464b721087fda7b1ed9
ead3a5a4b60be7cd30a48d9283feadf3ec4157bb
F20101113_AAEAQH boria_f_Page_34.tif
6c49acd59b1ea00ba230f9a7edc32350
bd733aefb75e358f47940db065e7b0ddb1eca826
73918 F20101113_AAEALL boria_f_Page_34.jpg
e66ab3997e4b4852f22d02e7bbd12830
71264064f3e0566af916f2f76e624edd0918da21
2924 F20101113_AAEAVF boria_f_Page_07thm.jpg
b2170095780734359c5f6d425b5d291f
1062360cf27ac1bc8eb794dd9ec0173bcb5531de
F20101113_AAEAQI boria_f_Page_35.tif
14d7e74dd216541a54b99be36a646871
705bcade1bd739fc9ec342f045f14df74d8e2e06
1196 F20101113_AAEALM boria_f_Page_02.QC.jpg
2561ced76fdd62d3e141815a48ea32d2
8502fea611836daa0b631db454fdbddeb097de4d
7232 F20101113_AAEAVG boria_f_Page_44.QC.jpg
1d9bdc6ed61d0dc8120463673b6c4ba8
3591cf8098253799c094d7df605b99dfd5093409
F20101113_AAEAQJ boria_f_Page_38.tif
e9355278548e7b193f47f10aee581b47
cc5fa7b0a9d20386a3b6b9cefa523c4e98c59c82
6061 F20101113_AAEALN boria_f_Page_37thm.jpg
48b996b5213e97e8c57e1ceb7454e1e1
f46c80abaccffd38c1f9ee2418f6751218d4383a
38669 F20101113_AAEAVH boria_f_Page_17.QC.jpg
ce4b83dc7d45297f1f203cf979ffedaf
688cbed3788e5bf4f0ca8b6aa1f450acd1e03b50
F20101113_AAEAQK boria_f_Page_39.tif
b7e3a49e64f16bdaa5d0140466c5e2b4
1490798af8ef133c9685d519387ced020fe345f5
56209 F20101113_AAEALO boria_f_Page_22.pro
f3d3a5d466c43fd219de5ae6f961cfb1
c4c362ee500fbb7f64ebb7bbcc4b9e27f9a60186
4850 F20101113_AAEAVI boria_f_Page_57thm.jpg
8ca4e63dadf3d094f29841105f729b71
3f9ac06ddfd62f830fdb63982e51f307ad338ecf
F20101113_AAEAQL boria_f_Page_41.tif
f09077ec9d0305e7dfe23cf02e2a55bc
d22388a3704ffb004866d417b631f2b54a9454b7
8322 F20101113_AAEAVJ boria_f_Page_43thm.jpg
293045225daa3ae7372c5537c018251c
d44a51d68057150ff55769ef4a04b12802aa79a7
F20101113_AAEAQM boria_f_Page_46.tif
e8bc52459bbc5a01a032616e8247ade9
0a40777b470b49368baf3ed889d109176391e2e0
7224 F20101113_AAEALP boria_f_Page_39thm.jpg
2c8cfd8827897eb470b252b516373e8d
9116c2d50c4112b9fdee3d2aff4e80c7cbf596c4
2056 F20101113_AAEAVK boria_f_Page_44thm.jpg
c74960bcf9166449cf7f53bf2c7706c7
3f7d4fad279315953d0f6d3bacee02ffc0de110f
F20101113_AAEAQN boria_f_Page_48.tif
861fc88edb6454ac395a58b931b63c04
de1d146492faf5eb3267dddbeed3e0d5010a761d
33618 F20101113_AAEALQ boria_f_Page_43.QC.jpg
fde42c60777414abf5362ad59be0b274
21e526b5cdbfea724686778aa198918d282bef08
9148 F20101113_AAEAVL boria_f_Page_22thm.jpg
2c702b2b55c7fafd6742d8f7e3cf3692
af8fbfa7dcf59f5eeca6b162c170e5c945458440
F20101113_AAEAQO boria_f_Page_49.tif
167f3203a1b2aa4346e87f7a35d6eb4c
f1db58fe5a2bf0546be75ac827903a9445f8ee46
113812 F20101113_AAEALR boria_f_Page_18.jp2
95864f66fcb0a8ffc394d52b83daf58a
dfa8605d6aa88189404b0913e290d0c273834389
7912 F20101113_AAEAVM boria_f_Page_24thm.jpg
d3564d8344989a49dfb3d7d9126ad72f
db1028aa4eb3e9d6efe7b5f9b87634893344600e
F20101113_AAEAQP boria_f_Page_50.tif
b8a23bd6b3f692905cbb6f5023e7924e
8d9453197b50fffe2b687b5e182ee3683992b9b3
8790 F20101113_AAEALS boria_f_Page_01.pro
174c816ebcddbdc5354f3f4d32dbe713
c95b76ef763be9b7948fbf1e1822f40b92b63996
6558 F20101113_AAEAVN boria_f_Page_05thm.jpg
44b6e25a38c3a0dd0d2bb7ca91e9d8b8
74ed66c314450a2d9c8235e67f156afe2cc43a42
F20101113_AAEAQQ boria_f_Page_51.tif
b4fd8a8e3c0fb6a358d6d3a9aedd401c
93a74486430146c0aa938f6e365b8828c3914839
F20101113_AAEALT boria_f_Page_10.tif
c2a08d4285389863d6c03f4324bf63b8
f1202bf7bbbbaba79c51d9817d14c5fc95aaca72
33658 F20101113_AAEAVO boria_f_Page_31.QC.jpg
15d1248c338013bcb92d63ce8f0c051f
3f0ec5e525889af31c1c56a5f59ddfa4f58cd471
54507 F20101113_AAEAGW boria_f_Page_11.jpg
93dd14abb681fc71e1d961128e5adf8d
bb7daa6b577afc2259c05f7cc78fe18faecc1c91
F20101113_AAEAQR boria_f_Page_52.tif
b5d6a8de3285e940b580e7c20340beea
a44335a0c8a22b02f14df194acf06cb4ea1aa392
123230 F20101113_AAEALU boria_f_Page_17.jp2
a77b6a48746dc1f8d2e11e802767c929
9751448d1cb4e1bd30948d7b4f72e41f3d0f0317
8164 F20101113_AAEAVP boria_f_Page_13thm.jpg
47161060d2e97e8ab98c80c1bfeb23da
627500024b64afed4605d66594aac1f041ad8ae9
2282 F20101113_AAEAGX boria_f_Page_17.txt
708648b951ecf1bee6e168a6a8401abc
69dc873f669ef93021fe018b899556b275396930
F20101113_AAEAQS boria_f_Page_54.tif
1123d6b016caff8765d8aedcc0986fc9
c7af276fad81c8333238af210f7e158967e5cc2a
22864 F20101113_AAEAVQ boria_f_Page_33.QC.jpg
ed3ceb8d437a05ccbf5cfe7dd0ba66cf
bf777c69759d835ada9a5657b27096ff5a50ce2b
F20101113_AAEAGY boria_f_Page_53.tif
ae1c1616ab162ad0fba01b4c98c417af
60a5ba7c12fa5b775d1238f7da71bcc036e72cb1
F20101113_AAEAQT boria_f_Page_55.tif
1c66bbba15a4646a464ad0ba52c19c03
43a11412d7d39ea17f0ae81fa84c636ec50a1e61
22084 F20101113_AAEALV boria_f_Page_44.jpg
85a6f3f62d84568d68cb60fb7dc2580c
ecfc1a6ed326a53469a26c48f8fc6e219932b252
21135 F20101113_AAEAVR boria_f_Page_45.QC.jpg
49d615e0e12a4b0eedc050529a71f089
7545d18199db1b1181771c832bbcf6cbed3f0f37
46841 F20101113_AAEAGZ boria_f_Page_40.pro
ae755046fe38575dfb0f270ed9c2ae0f
b6f8bcb42a9aed19a998e3e37ef1031bb13e95b5
F20101113_AAEAQU boria_f_Page_56.tif
865473c81bb31c41cefce51c539239f3
f7e23df385417f4a1118a9d1e14e56404a6db3c0
519132 F20101113_AAEALW boria_f_Page_49.jp2
e4398dbba4a6b5eda8828e61d6d46432
3c0ef959270988064a946d541f7b2d2b03b9d7f3
8268 F20101113_AAEAVS boria_f_Page_20thm.jpg
17f0ed306c02d1a7d9aedc16764dcdbd
ca9dc5a925de9b87f8e34e8286b3f5b34a814d68
932 F20101113_AAEAQV boria_f_Page_02.pro
c057d426b6f921124c72429463d3e141
00b07e7b00352fd5a4edbe2a13a461366c95efb1
485 F20101113_AAEALX boria_f_Page_01.txt
0985833bd57f4f2e42b401360bbd8f72
6b8e8560c4809929108fb76e96f55d95a94b1479
33963 F20101113_AAEAVT boria_f_Page_27.QC.jpg
17ad9bd3488d91b6e4d72fcfb1a02001
e0571c8530b63d41c119fac7423d816e72a7c1ae
750 F20101113_AAEAQW boria_f_Page_03.pro
f06d954f734b473c5f21c504f1d7e37e
bada1c15d8c7e217f34b4beb65e2d96cf87987b9
57627 F20101113_AAEAJA boria_f_Page_17.pro
98d0c29fbf3dc001dfd970f7b8e2e90d
5a75b80f4bd911049de4b67f792c19bbf152fd66
F20101113_AAEALY boria_f_Page_42.tif
33bbe854d7c7724d9fa246793f0bff5d
68ae9cdd9af48696a6ecc82a200bab01ee1caa24
8436 F20101113_AAEAVU boria_f_Page_27thm.jpg
c91508871218778ef42445d00ddb4824
2089c9f029588b4a001ce097f4fd16a24984109d
52007 F20101113_AAEAQX boria_f_Page_04.pro
c5bee2657d9c03827a76506e919d87d5
01319c6fdd0645b0d4c3712ac17458607c522b91
1834 F20101113_AAEAJB boria_f_Page_10.txt
d38836aaca380c3b2a2416aef90627ae
67e5ba806f9340d317f64be5de035376c361c947
101419 F20101113_AAEALZ boria_f_Page_13.jpg
09aa40fd15d624b3e39a2d5965478d43
e3302252a9d78a468e86de71367a8b9c325b23ac
6734 F20101113_AAEAVV boria_f_Page_23thm.jpg
845d61b4e8fe661f0e223938ba810565
208b6e4f801fbd182201110228f33483a84d7ea2
63568 F20101113_AAEAQY boria_f_Page_05.pro
26940603b1c3ca725b9f12dfe84fc35b
5ac178ab442b45911ca01f80e2fd2b752b548df2
F20101113_AAEAJC boria_f_Page_44.tif
76d00c554a916220510b97325a7e1cdd
aca79cb374292139f7b70682210d8297f94a2972
10978 F20101113_AAEAVW boria_f_Page_36.QC.jpg
3d8a5b724ebfbde0b313fdb943cc62dc
cb870295a662109664a35b21a695e635d50dd8e9
48231 F20101113_AAEAOA boria_f_Page_50.jpg
f6d3c1f6aafee3f24bdf9c7197a7603b
1c4a844f0b86e2b7ad21fa5caf55450cf98207df
14149 F20101113_AAEAQZ boria_f_Page_07.pro
558e4522beccdb628619d242ea0c3fdf
380132c6d639cea6c62558371641a932f7e27920
14655 F20101113_AAEAJD boria_f_Page_50.pro
29d569555f345df104957a09972f7502
7031c0e341f70f63747c68d9918b795e70a5df72
17841 F20101113_AAEAVX boria_f_Page_15.QC.jpg
bdcd33badcaa2846c8f804d87a213f81
98c520208d54cd5ef3967423549e084ec0e9247a
102233 F20101113_AAEAOB boria_f_Page_52.jpg
62744ef4cd8e8fecf373f2eba7aba4ec
218c5c1c1a5ed9b34bc8f7a11e894da035f05039
2553 F20101113_AAEAJE boria_f_Page_55.txt
e22681c2c7890fa5b327bf064ac127f0
40cafedc453e850495df7b772c40a419e681c4a5
104727 F20101113_AAEAOC boria_f_Page_53.jpg
334f181b30dee2103da74fd300ac24a2
4b68ca85929d814c39061bbeda79deb981e6e39e
9346 F20101113_AAEAJF boria_f_Page_44.pro
5a5c6c7c1e6c47263414055a7610edac
e10aa945a80014138090e610fb6c1ece2c24bf9d
34329 F20101113_AAEAVY boria_f_Page_14.QC.jpg
bbd7e13ae1d6d1bdf1a9062e5a1b31bf
3ba031c80981fb068857bf18716c97bde691e6fe
32712 F20101113_AAEAOD boria_f_Page_54.jpg
5e94f958ca100d4c176eadbf62a130df
a4d5f489c79f5c4efa384672c606b230de63fe7d
41179 F20101113_AAEAJG boria_f_Page_10.pro
c3d3bdabe4b14e7bc566e4c5c473ec10
a8bc385532d7da532969b03d528104690db2a7ec
2040 F20101113_AAEATA boria_f_Page_31.txt
2a98cbd7da63a70039d79a623d3e3a39
bf255fd158cacae493ed169613ad7a268e5c650f
88957 F20101113_AAEAVZ UFE0021478_00001.xml FULL
231df1152424b82a067350ae79b5e506
77670864d00f078af6f93cd6e415199029f52648
135302 F20101113_AAEAOE boria_f_Page_55.jpg
a2b11d936776ce1fde06b70f0ee5e33e
827ca636855fc4a03e3dc60af41d7a5d28ea97d8
6050 F20101113_AAEAJH boria_f_Page_09thm.jpg
39a58a966ba1af8d7cd4d518f1fc332f
df637a33cf1b2a5e030eb4c9bc876e940e6e2f24
575 F20101113_AAEATB boria_f_Page_32.txt
f45d87c188ef638ab1a884ef0ff225b6
029dd6f9364c640544e7074463e8bd8cd0c7eca9
136505 F20101113_AAEAOF boria_f_Page_56.jpg
315db308ca6ad7188f40a0101f10c8c1
fcb7442d595b9ef025405a5181326e2545d7d2d4
96303 F20101113_AAEAJI boria_f_Page_40.jpg
f9c9aee6784c14b117ceb07487acd72e
623b046c6f4748954ad8f3c1e706603d619ecf9b
567 F20101113_AAEATC boria_f_Page_34.txt
09a1d357286709892c8712a5b95923ec
c9cf22f8f14aa54a3f95eca253e205d103a587a5
88691 F20101113_AAEAOG boria_f_Page_58.jpg
e434b22c065668714e15c3667ff0a2c0
96bfaca01e2dd81f8c8288c5c40dd914c5b6c840
554801 F20101113_AAEAJJ boria_f_Page_48.jp2
088077555649588e9fa3cc96b5dca761
9bd8b417ad82f44d7b68377d90ae5c0f3778cbd8
1066 F20101113_AAEATD boria_f_Page_35.txt
adf7eb0da0be9983263a4c64e773daf1
9b98e314ac853ae9003db8e68226d4b10ad0a898
26257 F20101113_AAEAOH boria_f_Page_01.jp2
43f0f81c6011e4e17c4d1101ccceb27b
4b86fd54c8783344dcadbf56e06230ebc7cb73eb
56718 F20101113_AAEAJK boria_f_Page_37.jpg
8ea792259e2d50822d6b99fe0823f14d
9e250d3ef011f655a1b6c094967bb63f94b1e719
900 F20101113_AAEATE boria_f_Page_37.txt
2650fca948f3012c7aa96123c5819e7d
24ee9a655e07f9f5583ee75eedc62290d823e19d
5575 F20101113_AAEAOI boria_f_Page_02.jp2
30ef076b1ac5ca1873a407491c172278
93f99cc5feda94e7a990fc19c08dee3a9ee80942
F20101113_AAEAJL boria_f_Page_37.tif
bc33e1cd1670799e2c4fab65c3cb1e9d
0182baeb5c7cd839c0dea5e61c1306d40bce48d0
830 F20101113_AAEATF boria_f_Page_38.txt
e8a2d024397871ae8ebcb4440d3da451
d53e77b51be663fb41da14282c075adc50befa6f
4618 F20101113_AAEAOJ boria_f_Page_03.jp2
e06c9e75952a932ef1d041c01acf9522
159facbd9ea250ee972221d0eab01dd0537b06a1
2103 F20101113_AAEAJM boria_f_Page_04.txt
6d6ebcae02333135f89880823e2e9931
1280b749ebd01c22c901d25291f472a06fdd47e7
874 F20101113_AAEATG boria_f_Page_39.txt
b7a2327da95462dc5597aed91e20b21e
b9ddceb617a9506e32e6d0b145701e0fd7e2a7f5
110562 F20101113_AAEAOK boria_f_Page_04.jp2
56833c8c1def7a319e6f417f1b69d9ce
ca4444e7693872a2548ebe4e476e4eeffb1c558a
F20101113_AAEAJN boria_f_Page_07.tif
1e361489f8a913d1d2db97c27c35f398
8d5a2b002e73d1332d284d719223c8378f4a0ee9
1952 F20101113_AAEATH boria_f_Page_40.txt
5b152f954355eba51c87590bfcacc167
ba2cee0f6383d264375c7b6b52e98d2423086ba0
109646 F20101113_AAEAJO boria_f_Page_13.jp2
9d9a4b61494f3c70b0e3375096394264
60a9d1fc81cd69986140356332445610626b0573
2135 F20101113_AAEATI boria_f_Page_42.txt
7d941b651be6a31433d4209b92488586
e13da6a9a77dc2ba0527fd801cd0deffa3048a39
556549 F20101113_AAEAOL boria_f_Page_06.jp2
426a64badbff7188ebff9e66921a0f44
8bf4ff896d679abd4c8f33046777fcd5d81aab39
F20101113_AAEAJP boria_f_Page_29.tif
88bb6d484d473a725bd0d4096fff7000
cf2cebd71194d41bca1eaef7138a6a929acfbb3b
2057 F20101113_AAEATJ boria_f_Page_43.txt
3ff4ca22a38a5cb1cd7c7d11cba7d9d2
cf8db5a88fd4b39ff5149cb6decd63ccb0168dc6
654959 F20101113_AAEAOM boria_f_Page_07.jp2
1f8892684e74904f46bfb5c6a7dbdbef
cea687754a5d0fb7292d69749d91e5db1529c25b
1241 F20101113_AAEAJQ boria_f_Page_57.txt
b3cb22d6d657d525e91c76c7e1587d91
77a762e65479b916e866c530b6684f6d09b13408
380 F20101113_AAEATK boria_f_Page_44.txt
07c57a45190c1e59172ab5fcc1df389e
c734fa80a4aea2582b47eef4f8600a91eff642c0
1051974 F20101113_AAEAON boria_f_Page_08.jp2
60a22780a2e932d4119e6575766e7976
61f7c2545c321542aadaabc34beb104c6e8b04f7
F20101113_AAEAJR boria_f_Page_11.tif
81489f7c56181662f4d39a7c2c432b51
80e04f51b0852c1f3708c76eabce72648d8b71dd
367 F20101113_AAEATL boria_f_Page_45.txt
57c6d50558842bcca6bf55ca2b516456
b1f280fb7c7d3c50685150433c5cec0a94fd9eaf
58326 F20101113_AAEAOO boria_f_Page_11.jp2
3d833279a0c21e8a4c91576a7338a7d7
20a8fcf1f090c2e1ca1b722c20ecc5e4e47579f7
72568 F20101113_AAEAJS boria_f_Page_38.jpg
e8322cb17ba7d49a62911c2687271ffc
9580ce02bfa046e0b2bd498deafa8697f8d216fe
568 F20101113_AAEATM boria_f_Page_47.txt
f5b540bef17c8fe91d3eac03bb2ef5eb
b2118335043cda76f3c41e664014c9928deae713
106550 F20101113_AAEAOP boria_f_Page_12.jp2
530ed013d6daf70484c6606ba302cf55
54d49bc9a58119819313fc48873f0bc43e2d1b98
708 F20101113_AAEATN boria_f_Page_48.txt
160990a8f2a56e51cd2dff270d36b31b
e99c88dc696ae98828c27a481c2bf96c79187634
56356 F20101113_AAEAOQ boria_f_Page_15.jp2
fe893b07d027b3da97bdaee3d6308d51
bc3e74cfb275fb3c30d1f4320000a48be5bb8065
15588 F20101113_AAEAJT boria_f_Page_49.QC.jpg
b939308dd50ed8b94d58c3c4790512d6
8482824144f574d7372fd9123b646bec8c710d1b
726 F20101113_AAEATO boria_f_Page_50.txt
42e16aed120f6b497374c7526bf92ae2
ac09fca0a2f94522bb27c3377b0e39e7bda32554
106038 F20101113_AAEAOR boria_f_Page_16.jp2
ecf8e23d955f14e1b8f6f056b168ba0a
530c3af42f053c2f745398dde520268af9b78615
F20101113_AAEAJU boria_f_Page_43.tif
01522404b656479e89e9e2f02ea40aaa
4cf10c292fb5e8913cecc0d979cad50336d9f507
2050 F20101113_AAEATP boria_f_Page_52.txt
b7fba6520aef5de979465aa5322a93ea
94404db5d751a31fcf6958eda7474570bb3127a0
94052 F20101113_AAEAOS boria_f_Page_19.jp2
51c36d9aea7defa520ba144caa0886d4
917fefa8c09ac6880ed9386ff05377b019da9ef0
584 F20101113_AAEAJV boria_f_Page_07.txt
4ec809044f0c1c4d4a2014c49c74e4de
a6159e13c1d769ede16917b66977a6ca6bfb791c
2096 F20101113_AAEATQ boria_f_Page_53.txt
279eed68fffef23e6cdf7ea59223aba8
f7f057c8f3f7d7a2b822a8e160cb4235a1964ea5
109541 F20101113_AAEAOT boria_f_Page_20.jp2
aebe4405aca9139a8e809a1da82ba6bb
4f566bf22ea9c1b7f0d37bd4661958c0d21a7da5
4515 F20101113_AAEAJW boria_f_Page_11thm.jpg
b529fc07a52cd1ca8faa58575dd13e70
14593943a4974fb32a8b993a7da48bf253ec3a96
592 F20101113_AAEATR boria_f_Page_54.txt
cb795c117d2245d68fcf9bf4a94e1b85
bcfe61122422c11b1d0381242af18146e5956a24
112189 F20101113_AAEAOU boria_f_Page_21.jp2
b60be393b5ebd0caf6aa3fd17374ed6a
090c63039007e31279969fd3d63f38d374a2256a
8355 F20101113_AAEAJX boria_f_Page_31thm.jpg
913b4a30da1797cb307e0dc29225ef51
ac8f5ab9174851ca476d230950edf3dcd73f3a9b
1710 F20101113_AAEATS boria_f_Page_58.txt
910008866f9278bebca52ea4ee22a643
a6636e26eb58223f5da1b2ac6e4c80b2619a7970
85815 F20101113_AAEAOV boria_f_Page_23.jp2
801996fd37792b7ea084d1505a5d9441
3af0eb022f5963f106bc723dff464f0747ce883c
10923 F20101113_AAEAJY boria_f_Page_32.QC.jpg
fc3cfc92d47a1578328e7593fd482008
0aeaf788c6574bd75bf973979aace458952a4df1
2108 F20101113_AAEATT boria_f_Page_01thm.jpg
62d0fba07d0f9c756cc46ae05ed98d19
2302a4ed3f503e7fd7daa7e749f0c2453a13a916
563766 F20101113_AAEAOW boria_f_Page_25.jp2
66ed0fcad04ffec83b1952ad4e2b8882
6043f0a2c311e58c7bfcc98b2420649f267538f9
F20101113_AAEAHA boria_f_Page_24.tif
b73bac1abdcfa8bbe3db00e558b9f17f
ddd2cc0884fc5430e9d8244676e7c780d3d4aa8b
67076 F20101113_AAEAJZ boria_f_Page_57.jpg
dbcbb86006a284507f78f3583641c6f6
5570020837b5cc1ff757362942f968bb98b92231
1711051 F20101113_AAEATU boria_f.pdf
9981502feda91530df6d0509370662b9
4e9f476659d90738592990aca6b5039156b4bb42
108983 F20101113_AAEAOX boria_f_Page_27.jp2
1eb4463363497adba443f67a64b16283
b917dce167709b43acbbc1bcf8e5bfdb4d925d60
52206 F20101113_AAEAHB boria_f_Page_29.pro
d8caf2bc4cbc831a390a2365797eebe7
8fd99fdc740fbf84a87b4ef894c72540178cc69b
8532 F20101113_AAEATV boria_f_Page_29thm.jpg
4991310bb0910d5ecc5b69a762368037
3393118c42823e1723de8cf0fdad1ac2ecb4253b
11734 F20101113_AAEAHC boria_f_Page_47.pro
aec7778a6d725f44b29354a50b77c137
297dff27a780820307aaf3e90e8a938b3da92b4f
9669 F20101113_AAEATW boria_f_Page_08thm.jpg
269da10a606fa0491287b82c0fdef79b
960afd29c7362a821c4f9d0448d6462bc03c2dbd
55729 F20101113_AAEAMA boria_f_Page_30.pro
6b57fb8b66eec4890116176c00693547
b2afdac9b957e1e40f1a34b72c5e89af2771bddd
110989 F20101113_AAEAOY boria_f_Page_29.jp2
0058635fa04aa0c19cbc6802194ba70c
94ea13edd975f6fb5c05b365bd57b8c4db97cec6
23929 F20101113_AAEAHD boria_f_Page_44.jp2
e8d66301d946b84f34d6e440d4efe18c
c07ae6d9ceeedc6edd557bbff6fd50a790ce8418
8720 F20101113_AAEATX boria_f_Page_01.QC.jpg
f594ccaef30e5bae629d9bbc18a832a5
469cddeb16860ef03af4929822330534ccfb9166
15929 F20101113_AAEAMB boria_f_Page_38.pro
468eef0392ae278dc961e329b7524224
45092f298ed53714b95e45aa34f6696995a28147
117831 F20101113_AAEAOZ boria_f_Page_30.jp2
47f8114452817f2fb8b11c059305a2d8
8c4367d23cc89fb04b5d46978a896892016a52a3
35722 F20101113_AAEAHE boria_f_Page_53.QC.jpg
7f1fcd9af054050fa7a08e402310eba8
33282519cd4104b64e85001885869f47b1ea776a
33814 F20101113_AAEATY boria_f_Page_29.QC.jpg
7c066f7b3a3175c87a03cff2a42229ca
df1840cb3b9c4112553f3b2413cc144a9a11dd4e
31736 F20101113_AAEAMC boria_f_Page_32.jpg
8bb9b44cf363d199434933ef896d381e
6e93243b54cfe55f867354cdf1343afce347ee59
1051986 F20101113_AAEAHF boria_f_Page_24.jp2
19647f8cbd641151c21ce9a658f6d8fb
b76f8e56407e45459950e12cf25315e4385743bc
70433 F20101113_AAEARA boria_f_Page_08.pro
0e3aef2e1c0f02f595c6a90e6fbad8ea
e6f2b560f3dc92ff92dcf353940cca144490839a
6767 F20101113_AAEATZ boria_f_Page_25thm.jpg
ddab426e001a2196d1c7de2bdf318547
e660c954ad7573b23683a6f85d98b1ddbfb40fd0
35059 F20101113_AAEAMD boria_f_Page_04.QC.jpg
15164560c23382f76376540ec3c040d6
f9f7b1817fefb4fc5e8e31bb4683a6be0d2031f5
765 F20101113_AAEAHG boria_f_Page_49.txt
7e75ec3ad2a7323f3cc5657be653d20b
91d2a2bd5995cbef89c29f5da8576951f86e9d27
41884 F20101113_AAEARB boria_f_Page_09.pro
b880d5076ddc4c6531230e647a4e7dfb
26e1244d0bd98c7f10f7ffe7ed48eba403e3875e
21222 F20101113_AAEAME boria_f_Page_46.QC.jpg
8830b2d216c87164021c141e6f5dc2be
5c2eae8283149690572575eda6229e7804b88a1b
1735 F20101113_AAEAHH boria_f_Page_19.txt
0577ff204994f751d579d58403406877
e0c428a3d14ef5ae25f0d1c887fe707f159274ce
25538 F20101113_AAEARC boria_f_Page_11.pro
2de7555e4ae7758b3825d86a4054fb0f
2d15fbc6b31c36b6f346da17d8051ba4ba6062e3
112041 F20101113_AAEAMF boria_f_Page_53.jp2
b7ac59b0dcfd40ee718466c0c8b0b870
061f561b5db1af19fb33dcc9e5c1afdaa96cc0ae
F20101113_AAEAHI boria_f_Page_08.tif
3e3b65d0109303b5038a1f06673a46e1
dbb177a638272e5e5b6aed7c2c50ddcceacd22c5
49039 F20101113_AAEARD boria_f_Page_12.pro
ee52c2f18a2b9cd601533a86e98027b7
5c3948348236b28a51579ce1edb1eb67d3ea104c
1051960 F20101113_AAEAMG boria_f_Page_34.jp2
9cccf86529fbe01a1b557b85a1307b2a
6b060e01b79845eaa197ee5aaa7e1d81afe54829
8351 F20101113_AAEAWA boria_f_Page_04thm.jpg
4d253eb01dd6e2919a1d9d2e52d56289
2d2d8ddf5124b6895c9e62e322251bc007561530
32949 F20101113_AAEAHJ boria_f_Page_36.jpg
455c73cdffa6c839f0e79c6c6621a6df
d4f0fc898557c5b953271090f9a7fa21c9ebb099
49563 F20101113_AAEARE boria_f_Page_13.pro
77173ae9b8678490552732e5ce45c4e8
a43acfcd1bead634b333367b9f0579fe642b796c
104080 F20101113_AAEAMH boria_f_Page_20.jpg
827d75fa0aecac95d89848db160817c1
09ff82fd3e218cd62b1cd1f5dd29a761b6d6a7c4
2318 F20101113_AAEAWB boria_f_Page_06thm.jpg
a43ff1da917c7f672a4ace27e5c7e94c
cec096cbed686319a789a5160f3e6c4ceb5b2a29
26207 F20101113_AAEAHK boria_f_Page_24.QC.jpg
a0eb2708f354731aee6b208d953a66d4
8f29e9126854680d14dd8b73fbef72fb6860f6f6
52990 F20101113_AAEARF boria_f_Page_14.pro
46d0503344ca678f727deaaef825b61d
9bf5d5611679da77ef76ecde4c3fb92b827866f9
F20101113_AAEAMI boria_f_Page_28.tif
45e2bd2aecacc63b40fd3174346e104b
dcbc1836b016eb3a8f5b7a7054bfb78d5fa2cc3e
17986 F20101113_AAEAWC boria_f_Page_11.QC.jpg
90ec5e5c088207cc1f7c423c9a335a90
cf068b135410b898a0a96dbb2065c9df9c31ab54
31586 F20101113_AAEAHL boria_f_Page_57.pro
4097a0b6591e380c450f762250e49236
928931e9206c682bf58c599a4cea132ff65a71a2
25206 F20101113_AAEARG boria_f_Page_15.pro
9c26a5b039fc1675e99b53fd4a5664a3
7a966c102346ea248f6e1964da368fe5d1ff3441
75160 F20101113_AAEAMJ boria_f_Page_35.jpg
136c95e8c96de7db17aeaadc75f9441e
7bd6c415041797d8424c1e8949f1a42e9ce5549f
34426 F20101113_AAEAWD boria_f_Page_13.QC.jpg
710644c9e7db057d5fd94c9080a95a63
626fb352f8c5f134645cf9f8e42938904d3c4617
39056 F20101113_AAEAHM boria_f_Page_56.QC.jpg
5c3d7890e9ac67db6aca29b1b6a2c6cf
36e1167709deca9932ef5c19c6cafce4b3308468
1051985 F20101113_AAEAMK boria_f_Page_05.jp2
52a18c77828765dc3eedf79faafdbba1
951110eb82f89b07db83b88545b2ae026ef40ba2
4357 F20101113_AAEAWE boria_f_Page_15thm.jpg
dcc0b154bec1af1818bc468fb16f3b92
3ad9d2a19ecf0dfc0ba25c6e30749b269862fd65
35074 F20101113_AAEAHN boria_f_Page_42.QC.jpg
6e9206e0dc5d0bde55cbe5ab31942952
108d5790dcad2a6bce499c8266fb3b7c91961340
49224 F20101113_AAEARH boria_f_Page_16.pro
96f79b0d36e4ea8aa49c8331be234b37
c917d64c30dd37862fa051025ed9fc9e45a14eaa
8213 F20101113_AAEAML boria_f_Page_16thm.jpg
9f133edda0fb4ef056080e1298686f3e
81ea3069b9402206f42a324fde0138d49a23fcf0
8989 F20101113_AAEAWF boria_f_Page_17thm.jpg
d8316e4331523a170cff5f393377b2f8
c8a85286ef626c2f2bc6a079a6989e48f0f1f2f6
41918 F20101113_AAEAHO boria_f_Page_58.pro
04453f6057f69278b829e6830db010f0
c0a01d9e89dd3a615c6df9ae56a10e7df7f02564
52401 F20101113_AAEARI boria_f_Page_18.pro
dccebc0d33cca5bef2a94ac20f93115d
b4106b816bd97676ae1230d7fe347a81f694d45c
14517 F20101113_AAEAMM boria_f_Page_06.pro
a2ad44e60dc99e61eb62ea333efe2f02
41979f70b3990f4f50f7404198a29ea8eb134c3f
8794 F20101113_AAEAWG boria_f_Page_18thm.jpg
29565a80a554794a066d206b66792a31
6b472f00d7867c60e8659ed27d5de02576839a6f
55084 F20101113_AAEAHP boria_f_Page_41.pro
0813f8a2219dfbd4b9392b366f2df1de
1dd4daa8843e4d9fe1554a20000dc10dd2fe34c5
42802 F20101113_AAEARJ boria_f_Page_19.pro
0853f2a5845a0893da56ff686e094571
b3d81c423daef64ea42c393b7b31e571574c920e
35206 F20101113_AAEAMN boria_f_Page_21.QC.jpg
5a72f0d158748f5c402b276099489ad6
15d623c1a257762e2ef340c8134552f30f3d40a4
5216 F20101113_AAEAWH boria_f_Page_26thm.jpg
e138b9cc4003db10df47f58cc35f1a13
f69377d734488cd3b41f341e0c1cb7c7a9998736
F20101113_AAEAHQ boria_f_Page_26.tif
ed21cfc836129a8b9dd358bbe9d96aef
86dc23c45fb97b590b50dd10b970d6f857865caa
50250 F20101113_AAEARK boria_f_Page_20.pro
608e1f5d162d407b4888918d4d3ab5f2
5e0615520b7a3e3389f5d83197afbd05060ceca6
6593 F20101113_AAEAWI boria_f_Page_33thm.jpg
b96a924699fcabbe00ca96afbbd1907b
4b40e84d3fedddb12bd7f78c8a9bcf01399f3152
F20101113_AAEAMO boria_f_Page_17.tif
529fd14f9654b1fa1d45cf03a4060b92
d3bbda0143eee79c225b28bafd64158a08f694b3
52443 F20101113_AAEARL boria_f_Page_21.pro
7611a5930565bdbfb697e52bfe98609c
f2690bf19637e42513457aaaa6a9262d0b1ddb14
21145 F20101113_AAEAWJ boria_f_Page_34.QC.jpg
d188bc19992dd7dc733045bace102b78
6546f994da5c27dbded44fd674428bb3f6b4f627
117289 F20101113_AAEAMP boria_f_Page_22.jp2
a34906bd5638b52a03e6197e3cc69584
4f1df333b5484cace6fa6abe1f07d6be7ed57f04
92423 F20101113_AAEAHR boria_f_Page_10.jp2
0feab88fb2ea75676f94568ccd02dff6
32b5ec25160e82e82cd71b89c4734a23463b436d
38716 F20101113_AAEARM boria_f_Page_23.pro
df7024df0d6c3a8c4ac5fb7d905752c1
c0a077a157d76bb0afebedf8d8661e0928cb9feb
20239 F20101113_AAEAWK boria_f_Page_37.QC.jpg
1e7392581f7a4b28f20bfc5cb7830740
a68925b81b5ed8635c15bbf890650abaeb541655
1019 F20101113_AAEAMQ boria_f_Page_11.txt
529d67460ce659bc31e198620f528c4a
32c4b7b7e823ae602bfc2b216949539d66ea4b0c
32560 F20101113_AAEAHS boria_f_Page_06.jpg
a23e57febb889572f7b40970ac139cb9
925b7e6015620120e4e9f927b154f09d98d29fce
12275 F20101113_AAEARN boria_f_Page_24.pro
a80e7c84f4ea0274db92dce702e19733
6578096fc8a55f4983f02b3925274429267084f9
9015 F20101113_AAEAWL boria_f_Page_41thm.jpg
ad3a3d94eeb1639db8859a4a198146ce
103f725e0d4bfbaea8479a12a9c9f453338c8e6c
8987 F20101113_AAEAMR boria_f_Page_30thm.jpg
66facd4b2cc24e81bf4d53fdcb3d04cb
1eae650db481238439e4ed5b32620dea4bc91b23
111294 F20101113_AAEAHT boria_f_Page_05.jpg
c850d1c482f3078a3d173b5d978fc03e
6f6deba9228747fa8cfb140f08c74aad2b02ffc2
5176 F20101113_AAEARO boria_f_Page_25.pro
baed1fdd4949b512350bc8af96c2af0a
34de9137c50a6c2c062b223806d1c900c207c660
8881 F20101113_AAEAWM boria_f_Page_42thm.jpg
4cba02e10d7729ea6b14466cb1aaf53c
e9836714d6da70f87763a078a6ea6877a0ff7598
61376 F20101113_AAEAMS boria_f_Page_46.jpg
42d8a5ae4425818d7503fd2822aafabf
46d42c340db019b4f5a334b20adfb8e1be709b6c
1051971 F20101113_AAEAHU boria_f_Page_09.jp2
3c99650ff1ec6d0766152b550c6405de
ce2235e2c54f4aeacb6dcf8a7d5e63cc87afae4f
2998 F20101113_AAEARP boria_f_Page_26.pro
2621dddae6714d26f2467ab6b0e1eb25
93c7104be1a405939518f4523e1ba304fb915ae4
5109 F20101113_AAEAWN boria_f_Page_49thm.jpg
607dcbf4cad69184a76643095a0d3bbb
860117b77c420ff19132cf96b84c7652a405997b
69116 F20101113_AAEAMT UFE0021478_00001.mets
e11efdaa501d35be5588d7b962721949
897cac8cfb5db05f33b706f42edc2ec7e17dd89b
114076 F20101113_AAEAHV boria_f_Page_14.jp2
ebffa03d4940955461bb97dd68325e1b
42b02de99cd62678d2998bb5603a2c8e7ae5ca53
50324 F20101113_AAEARQ boria_f_Page_27.pro
bccc25cf0504c0870b82cc6afb744bbe
93408646e65589c2a2b8b716ba872e226f0df0dc
16563 F20101113_AAEAWO boria_f_Page_51.QC.jpg
130875f454316700e1dcce260809a59a
e691be5734d5416d0c7509e3ee383176926c02af
228 F20101113_AAEAHW boria_f_Page_25.txt
f24ff48a885527c99f6b87da8c86123e
d9451685be39c27e069dfe49f2552511ca91430c
54014 F20101113_AAEARR boria_f_Page_28.pro
293e87fb8dca9eb3df7294c17e5b61a2
103ee87797c4782eee9feb99a874f8e143b164f0
8641 F20101113_AAEAWP boria_f_Page_52thm.jpg
b578b251c372ea316b301f36e6a1d38f
f7a33358dde3cbf76152a5a6093be0c77083dd5b
473 F20101113_AAEAHX boria_f_Page_03thm.jpg
eed42481d9e9f9a8639fcd56d3f731af
af9f2ab309c3fbdb6949bf776f7660f38cc70b45
51450 F20101113_AAEARS boria_f_Page_31.pro
0972ee82f31c761945e95ae4b01113d9
230ef2bc9a7238c585154228d7faf98c10f6e7d6
8417 F20101113_AAEAWQ boria_f_Page_53thm.jpg
9b557f55a0008a1ce8aeb4fb70ca8c9a
70b480be7a6bef55268dee50c58f935b6e25aec2
315 F20101113_AAEAHY boria_f_Page_36.txt
3f43171925858a16d71bcb154563c4ab
7dc49ae6574a13aadf938c2ea368f3a4a9bfcd77
14267 F20101113_AAEART boria_f_Page_32.pro
209e800b41d380808b0718b3adf43cbe
f3f535b4b92ea117f2bf4b710771525f270afada
4123 F20101113_AAEAMW boria_f_Page_02.jpg
db3c161429dda2cd280fceebc30f33f0
067aa6da492c7d54a404dfd6574b82ee1248024e
24277 F20101113_AAEAHZ boria_f_Page_39.QC.jpg
9c888ae8b2e14257e69dd06a1e760ce0
4d8a6a6137a1f84a37a749c2e6367ad3713a5037
34730 F20101113_AAEARU boria_f_Page_33.pro
3d182034f5d9470e3667f5dd78f2e451
e643f87d376e6cc0f420d52a8042ca46e3e14f9a
3663 F20101113_AAEAMX boria_f_Page_03.jpg
b9c6e1f6fdb5d0ea70fe1cb810bfbd3b
a42d563eb0b7ab130da8ace0f215fa76452ed330
12471 F20101113_AAEARV boria_f_Page_34.pro
09c6477506c49b8a61e19b344045d73b
b7ad1ac49a50b97de41f1371e225f6283b63e03a
106096 F20101113_AAEAMY boria_f_Page_04.jpg
76be05526f70995e80f383ecdb9fb5ea
638ae8fe8601ee9356fc6fb67033788f7f6b427e
24135 F20101113_AAEARW boria_f_Page_35.pro
b86aab31076046435dfd2e694d0c9e75
86bc1be35cedabe795928cf3f9e4910b97c9a4aa
902 F20101113_AAEAKA boria_f_Page_51.txt
348da614822f945010ca065807169eed
a3255a6234907a22635f961c8eb00ec7476aaf63
37654 F20101113_AAEAMZ boria_f_Page_07.jpg
1693c8dfe716779d1435c88d8e6f9f6a
bbb8f07d04306e4b0db17ca90db5b011c446209c
4741 F20101113_AAEARX boria_f_Page_36.pro
cb9a14a96b552f47569661699f52465f
33bf8fbb981739cbfcea40edae265dd9775a515b
68530 F20101113_AAEAKB boria_f_Page_56.pro
b2d9c45890d1a9eccb809226a40400a1
2e1e34758dec98701135221bf38e332754540a78
17613 F20101113_AAEARY boria_f_Page_37.pro
7a910a9416a8d311bb233c75dfd013e5
d40b7565086f3d8d03c3ef1b91911cfe7da3b0a0
25287 F20101113_AAEAKC boria_f_Page_35.QC.jpg
35b852fcff7c5272db02279944b21cf4
0a7923eeba99ec2860bfd89a43fdf679e60ee8c4
110179 F20101113_AAEAPA boria_f_Page_31.jp2
7ade46f4dca7c0e263273f91a9c5fda9
0c87b893df88d1fc99edf04fab52935d454100b4
17076 F20101113_AAEARZ boria_f_Page_39.pro
0fa874345de38de70cd2965be821d0a4
5c734af15c50f47857da3d4a29043d7b6423a39b
1025415 F20101113_AAEAKD boria_f_Page_35.jp2
32949a3aa9371170cc5feaf7df6c0122
ca9255b5ad2edb58507666e4262fcda06dab0ae0
33715 F20101113_AAEAPB boria_f_Page_32.jp2
1d77ee636b1b6d3391f4f55d49a7020b
4a04ceece10f0ad2affac4a1ad2541b3d122d7a7
1051984 F20101113_AAEAKE boria_f_Page_45.jp2
b4cd1451bd0ab66719767533096cdb1b
2f947a4dbb2f3e05a44eb8b9ed6223b66a6ca019
464887 F20101113_AAEAPC boria_f_Page_36.jp2
814c60100a3c54857236bef1bca87fff
7a3c65def49b15e799a85216feb4573fa94a7691
2885 F20101113_AAEAKF boria_f_Page_08.txt
2e0533e55c8aa369ff9d3d2a64a8bf22
70b3e112066e78598db43d4a5ce098f4c54e8b2c
6960 F20101113_AAEAUA boria_f_Page_46thm.jpg
d02af2355f5d117048b13c4da2a7cf8d
8e1dffe576f7977921ad841884d17fe4ea08cef8
662656 F20101113_AAEAPD boria_f_Page_37.jp2
2774a243f0d73088bc1a91b489d95ce4
dc321bb8ab5236f5a5c4275d40593033533dcd10
2241 F20101113_AAEAKG boria_f_Page_22.txt
ce7316fa24955d7571c080a1eb12c83e
4bee9447aeaa58a8427eeea3a4338f5dbc26c0ee
2745 F20101113_AAEAUB boria_f_Page_54thm.jpg
8cdd7f600de17251dc0945fc7d69b3b2
59e9e4f4c9ee242fb5778237dfe905997af6faaa
F20101113_AAEAPE boria_f_Page_38.jp2
ad720870399a15a25610612606c75214
c48a90aae9808b54c1716444688be4257cd5194c
8655 F20101113_AAEAKH boria_f_Page_28thm.jpg
1124309e6f9b5be596c89107baab4c86
28398012c5caa573d8b1538745f92ec91be43bf4
6756 F20101113_AAEAUC boria_f_Page_58thm.jpg
52328b59ff8c1ff7e7adb3508e449e4a
759aff268111d4e55faf8b6bf115af97600f8e91
975741 F20101113_AAEAPF boria_f_Page_39.jp2
9cfcba13b6bdab09df402611f912d1ec
bbbc6793e80751b59c16d5c01644be23ad8a9daa
F20101113_AAEAKI boria_f_Page_27.tif
8ee00152f5d4ceef8786c0b8b6e2bca0
cb28d054bbdce15b67be3c8decd1cc981d5fd927
101704 F20101113_AAEAPG boria_f_Page_40.jp2
52b7b58b47963354a5c007fd3ab93d4d
75e80bc627c0f3a59dbab7e10db63e20da93d592
537852 F20101113_AAEAKJ boria_f_Page_51.jp2
f466161713188715dd7b922298eb798d
5617cdde94fd519b4775a5fee7b6f7f62a1264f2
10358 F20101113_AAEAUD boria_f_Page_07.QC.jpg
4e68acebfacfa16973b994ecdc0c0303
f23423c9e637b450b6c2328c7a58c11668be44e7
114068 F20101113_AAEAPH boria_f_Page_42.jp2
d8578cdbdec10490ea93b1a3ecafe8de
d781cae5d82c1b644a821443f844c3b98ea70853
F20101113_AAEAKK boria_f_Page_03.tif
7b30f37fe9d20b141297492e57d679e5
acf22d1b04344cb922d3bfc53c01e760b3e43b1c
5384 F20101113_AAEAUE boria_f_Page_47thm.jpg
89315bf5adf58854b86ec151875a3fb6
654d922be697174f6903e0edab1a2238003aa928
109735 F20101113_AAEAPI boria_f_Page_43.jp2
936a280666484e3d0cc4fd44b26cd3f5
5edd44d3d5dae687797d3cc6b4e66842c9febc85
108880 F20101113_AAEAKL boria_f_Page_52.jp2
e5ee3380377e8f100969544be841b46d
bed56fc6d86ec0951725f11340da5a24a0eb7627
31589 F20101113_AAEAUF boria_f_Page_40.QC.jpg
b5827b5bc44acc115af751b205b57c78
567e7fcc9ff6b118cd391f8661e3dae869a70071
435449 F20101113_AAEAPJ boria_f_Page_47.jp2
2b08ac9406df128514a7590677d8aa8b
b9a204566a1e8081cb1e523792636c14d3a8507e
8505 F20101113_AAEAKM boria_f_Page_21thm.jpg
a145773998d4edf66ca73c7b413723b0
5c07bcadcabe617c5a3f5e14d6ca29bd5667ea8e
35968 F20101113_AAEAUG boria_f_Page_18.QC.jpg
e39765784e00d4afaffcadacb1e3ec30
ebc4834201b4bee40028378aa169dab0f2d77054
606782 F20101113_AAEAPK boria_f_Page_50.jp2
757a50423a72d52e5da59a50fa8cec8c
900ad56c4b63dbb8f60aa4946e025a98937b159a
23022 F20101113_AAEAKN boria_f_Page_38.QC.jpg
4451cde757ffcc14e2030a8390db5b99
09a194ac5c525face9570514102b46d6b3c036cc
19248 F20101113_AAEAUH boria_f_Page_57.QC.jpg
bce4462a96c100c2ae6df9898778af5e
27a51d94b3a806f5f26a05b844ad8097996845bf
35051 F20101113_AAEAPL boria_f_Page_54.jp2
a3498b0cd7de9d4907133d67ad45e0c2
6c410240ca786584e47fe3be1fd30e4fc4f77326
F20101113_AAEAKO boria_f_Page_40.tif
4def7f73f3bb0b26b244566294d15002
8e34b3339c40bea1bc3dc645a84feb6e79992345
8225 F20101113_AAEAUI boria_f_Page_06.QC.jpg
556cee195d5bf2bf4f1adf21de000d05
fc7667511527f7a9a7571e52a649ee674be1ef9c
145593 F20101113_AAEAPM boria_f_Page_56.jp2
4a279ce0481477164c7cfa45177adc6b
6fe400f493baecb03cb4d6494a276ff337e7621e
2674 F20101113_AAEAKP boria_f_Page_32thm.jpg
fcd0d62f1652b0dd5636d0a3b6eb017b
aa5cfd3feda7730587beb9fc704de8c7132f6438
7339 F20101113_AAEAUJ boria_f_Page_19thm.jpg
d9af88e201773113eac4f74d8660c9e1
98c5aacf6024eff5fdb8bd1fed6dc413abe6ca5b
70528 F20101113_AAEAPN boria_f_Page_57.jp2
f0af4e52f4304c9672d4994da4cfb623
70bdb272878c652c0f216521ff6b59577bb516ba
24308 F20101113_AAEAKQ boria_f_Page_09.QC.jpg
3ff0aee65968566bb8fdfbe4c4894207
99323fd082f25fcfed586fa5910c279430a20277
20735 F20101113_AAEAUK boria_f_Page_25.QC.jpg
210bbb0c05af1f73b2c2d826ce488bbc
2a2cf0232d32ba47ad165f344bbfdc51286585f8
93406 F20101113_AAEAPO boria_f_Page_58.jp2
4329e019ddeb7af667b1f9fdea3eec8d
dd5374178376152c85026cddf013ab710eaae32a
87860 F20101113_AAEAKR boria_f_Page_10.jpg
1d9a2f844d1b9f43ead3f1a1dad8a9b6
25b99e621a10e5047283d9fc39a13b27f35a0c64
6761 F20101113_AAEAUL boria_f_Page_45thm.jpg
fefa515b7f9df270ccc83f833565f262
6a41e27053dd77f67962a6b015f0c3e1df1320ca