<%BANNER%>

Investigating Manufacturing Techniques, Testing, and Design to Enhance Confidence in Thrust Production for Synthetic Fle...

MISSING IMAGE

Material Information

Title:
Investigating Manufacturing Techniques, Testing, and Design to Enhance Confidence in Thrust Production for Synthetic Flexible Small Flapping Wings
Physical Description:
1 online resource (89 p.)
Language:
english
Creator:
Rue, Jason T
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Aerospace Engineering, Mechanical and Aerospace Engineering
Committee Chair:
Ifju, Peter G
Committee Members:
Haftka, Raphael Tuvia
Mohseni, Kamran

Subjects

Subjects / Keywords:
flapping -- manufacturing -- thrust -- wings
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:
Flapping wing technology exists and continues to expand on many fronts, although without a solid understanding behind the physics of flapping flight, a purely experimental approach can be taken. This study focuses on small synthetic wings which were biologically inspired by hummingbirds as they are comparable in size, shape, and flapping frequency. The focus was aimed at the average thrust production from a one degree of freedom flapping mechanism and the weight of each wing. The motivation arises with the final application of a standalone hovering device but brings to attention a question of whether or not to trust the current manufacturing process that consists of a carbon fiber hand lay-up method. The main objective of this work was to create a high-fidelity manufacturing process having repeatable and robust wings resulting in improved results where minute variations in average thrust production can be detected while in hover mode. To advance the possibilities and expand the testing envelope four distinct methods were proposed, the carbon fiber hand lay-up, a Teflon® CNC milled mold, a milled plastic frame combined with a carbon fiber rod for support, and a unique attachment method with a 3M transfer tape. A multitude of methods were performed along with a validation comparison of two different geometric structures and several wing duplicates. Conclusions heeded a significant drop of approximately 74% in the coefficient of variation describing the weight distribution between duplicate wings and an over 84% decrease for the average thrust. Now that this is accomplished, there are hopes that further research can be done to find how the characteristics such as structural stiffness and geometric topology have an effect on flight.
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 Jason T Rue.
Thesis:
Thesis (M.S.)--University of Florida, 2013.
Local:
Adviser: Ifju, Peter G.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045987:00001

MISSING IMAGE

Material Information

Title:
Investigating Manufacturing Techniques, Testing, and Design to Enhance Confidence in Thrust Production for Synthetic Flexible Small Flapping Wings
Physical Description:
1 online resource (89 p.)
Language:
english
Creator:
Rue, Jason T
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Aerospace Engineering, Mechanical and Aerospace Engineering
Committee Chair:
Ifju, Peter G
Committee Members:
Haftka, Raphael Tuvia
Mohseni, Kamran

Subjects

Subjects / Keywords:
flapping -- manufacturing -- thrust -- wings
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:
Flapping wing technology exists and continues to expand on many fronts, although without a solid understanding behind the physics of flapping flight, a purely experimental approach can be taken. This study focuses on small synthetic wings which were biologically inspired by hummingbirds as they are comparable in size, shape, and flapping frequency. The focus was aimed at the average thrust production from a one degree of freedom flapping mechanism and the weight of each wing. The motivation arises with the final application of a standalone hovering device but brings to attention a question of whether or not to trust the current manufacturing process that consists of a carbon fiber hand lay-up method. The main objective of this work was to create a high-fidelity manufacturing process having repeatable and robust wings resulting in improved results where minute variations in average thrust production can be detected while in hover mode. To advance the possibilities and expand the testing envelope four distinct methods were proposed, the carbon fiber hand lay-up, a Teflon® CNC milled mold, a milled plastic frame combined with a carbon fiber rod for support, and a unique attachment method with a 3M transfer tape. A multitude of methods were performed along with a validation comparison of two different geometric structures and several wing duplicates. Conclusions heeded a significant drop of approximately 74% in the coefficient of variation describing the weight distribution between duplicate wings and an over 84% decrease for the average thrust. Now that this is accomplished, there are hopes that further research can be done to find how the characteristics such as structural stiffness and geometric topology have an effect on flight.
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 Jason T Rue.
Thesis:
Thesis (M.S.)--University of Florida, 2013.
Local:
Adviser: Ifju, Peter G.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045987:00001


This item has the following downloads:


Full Text

PAGE 1

1 INVESTIGATING MANUFACTURING TECHNIQUES, TESTING, AND DESIGN TO ENHANCE CONFIDENCE IN THRUST PRODUCTION FOR SYNTHETIC FLEXIBLE SMALL FLAPPING WINGS By JASON THOMAS RUE 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 2013

PAGE 2

2 2013 Jason Thomas Rue

PAGE 3

3 To my family, friends and all that supported me throughout the years

PAGE 4

4 ACKNOWLEDGMENTS I would like to start with my sincere gratitude for my advisor Dr. Peter Ifju. His knowledge and guidance led to a passionate work environment where I was able to flourish. I am thankful for his open door policy, h is willingness to help, and for bringing me in to his lab on short notice, for staying at the University of Florida in the Flapping Wing Lab was the best choice I could have made. Next, I have great appreciation for my parents and all their support. They h ave made my success possible and are continuously there to help in any way they can. I thank my girlfriend for believing in me and helping with everything along the way, like bringing me dinners on long nights Special thanks to all of my graduate friends and my lab mate Kelvin Chang along with the Air Force Office of Scientific Research ( AFOSR ) for providing research funding.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 Literature Review ................................ ................................ ................................ .... 14 Background Work ................................ ................................ ................................ ... 16 Motivation ................................ ................................ ................................ ............... 20 Goals and Ob jectives ................................ ................................ .............................. 22 2 EXPERIMENTAL CHARACTERIZATION OF MANUFACTURING PROCESSES 24 Wing Structure ................................ ................................ ................................ ........ 24 Flapping Mechanism ................................ ................................ ............................... 26 Mechanism Version 1 ................................ ................................ ....................... 27 Mechanism Version 2 ................................ ................................ ....................... 28 Mechanism Version 3 ................................ ................................ ....................... 29 Mechanism Version 4 ................................ ................................ ....................... 30 Data Col lection ................................ ................................ ................................ ....... 31 Visual Inspection ................................ ................................ ................................ ..... 35 3 MANUFACTURING ................................ ................................ ................................ 37 Hand Lay Up Process ................................ ................................ ............................. 39 Process One ................................ ................................ ................................ ..... 39 Issues ................................ ................................ ................................ ............... 40 Teflon Mold Method ................................ ................................ .............................. 42 Process Two ................................ ................................ ................................ ..... 42 Issues ................................ ................................ ................................ ............... 45 Plastic Frame ................................ ................................ ................................ .......... 49 Process Three ................................ ................................ ................................ .. 50 Issues ................................ ................................ ................................ ............... 52 Transfer Tape Technique ................................ ................................ ........................ 52 Proc ess Four ................................ ................................ ................................ .... 53 Issues ................................ ................................ ................................ ............... 53

PAGE 6

6 Other Attempts ................................ ................................ ................................ ........ 54 4 EVALUATION OF TECHNIQUES ................................ ................................ ........... 56 Two Design Assessment ................................ ................................ ........................ 56 Hand Lay up Process ................................ ................................ ............................. 58 Teflon Mold Method ................................ ................................ .............................. 60 Plastic Frame ................................ ................................ ................................ .......... 62 Transfer Tape Technique ................................ ................................ ........................ 62 Results ................................ ................................ ................................ .................... 63 Weight Comparison ................................ ................................ .......................... 64 Average Thrust Comparison ................................ ................................ ............. 67 Conclusions ................................ ................................ ................................ ............ 75 5 REVIEW AND FUTURE WORK ................................ ................................ ............. 77 LIST OF REFERENCES ................................ ................................ ............................... 83 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 88

PAGE 7

7 LIST OF TABLES Table page 4 1 Average weight of the 160 individual wings. ................................ ....................... 67 4 2 Average thrust values at 30 Hz, given in grams, for each of the eight wing groups. ................................ ................................ ................................ ............... 74 4 3 Average trial thrust CV measurement for all eighty pairs. ................................ ... 74 4 4 Standard deviation data for the thrust measurements at 30 Hz. ......................... 75

PAGE 8

8 LIST OF FIGURES Figure page 1 1 ................................ ......... 16 1 2 Full field deformation measurements taken with DIC software and two sets of cameras. ................................ ................................ ................................ ............. 17 1 3 Three main wing designs focused on by McIntire including one based on a hummingbird, a leading edge only design, and one reinforced on its perimeter. ................................ ................................ ................................ ........... 19 1 4 horizontal error bars represent the range of values between flap angles 5 and 5 with the mean denoted as a marker. ................................ ....................... 20 1 5 WowWee t oy Dragonfly [35]. ................................ ................................ ............ 21 1 6 AeroVironment Hummingbird [20] ................................ ................................ ..... 21 1 7 The DelFly autonomous ornithopters [29] ................................ .......................... 22 2 1 Examples of wing combinations with different frame topologies. ........................ 25 2 2 Wing area along with notation used in this paper. ................................ .............. 26 2 3 Flapping Mechanism 1 capable of 45 Hz and 60 flap angle. Comprised of many hand scupted parts and numerous screws.. ................................ ............. 27 2 4 Flapping Mechanism 2 equipped with a nylon block and machined parts. ........ 29 2 5 Flapping Mechanism 3 furnished with aluminum frame and fewer parts. ........... 30 2 6 Flapping Mechanism 4 with screw attachment and bearings. ............................ 31 2 7 LabVIEW virtual in struments tha t control the testing procedure ........................ 32 2 8 Filtering raw data to help interpretation.. ................................ ............................ 34 2 9 MATLAB GUI interface. ................................ ................................ .................... 34 2 10 Image of a wing manufactured by a hand lay up method having considerable visual inconsistencies such as the battens differing in width, the LE is a function of span, and the triangular section has smashed out beyond the desired dimensions. ................................ ................................ ............................ 36

PAGE 9

9 3 1 fiber taking roughly two hours. These steps were programmed into the oven and followed exactly. ................................ ................................ .......................... 38 3 2 Problem of carbon fiber on a flat plate loosing structural stiffness.. .................... 41 3 3 Typical deviations in carbon fiber from the hand lay up process.. ...................... 41 3 4 A CNC mill cutting the Teflon mold shown in the upper right.. .......................... 43 3 5 Custom strip cutter created to not only cut carbon fiber strips consistently but cut multiple at a time, becoming more efficient and saving time. ........................ 43 3 6 The Teflon Mold in two views.. ................................ ................................ .......... 44 3 7 A quick visual comparison regard ing the wings from the hand lay up and mold processes.. ................................ ................................ ................................ 45 3 8 One problem with the Teflon mold process was the difficultly of syncing the mold depth and amount of carbon fiber correctly so that a rectangular cross section was formed.. ................................ ................................ ........................... 46 3 9 Buckling possibly due to pressure distribution or thermal expansion. ................. 47 3 10 Example of cured carbon fiber frame where breaks occurred along weak joints. ................................ ................................ ................................ .................. 48 3 11 Completed plastic frame wings with the carbon fiber rod, frame, and CAPRAN already combined. ................................ ................................ ............. 50 3 12 Zoomed view of the plastic frame.. ................................ ................................ ..... 51 3 13 A figure with the steps of the process from milling to the plastic white frame to the finalized wing.. ................................ ................................ .......................... 51 3 14 54 4 1 Shown are the two designs chosen for the validation process. Although created in the past, these were randomly chosen from a group of wings that had decen t test results.. ................................ ................................ ..................... 57 4 2 Hierarchy breakdown of the experiment explaining a total of 80 duplicates were constructed with 800 runs. ................................ ................................ ......... 57 4 3 Dimensions, layout, and various examples of the processes for the two design confirmation. For brevity, the transfer tape technique was excluded due to looking exactly like the plastic frame ................................ ....................... 58 4 4 A look at the flat plate and hand lay up procedure.. ................................ ........... 59

PAGE 10

10 4 5 A close up of Design 1 post cure. ................................ ................................ ....... 60 4 6 A glance at the Teflon mold process. ................................ ................................ 61 4 7 Plastic frame with the LE carbon fiber rod attached. ................................ .......... 62 4 8 Steps involved with applying the transfer tape.. ................................ .................. 63 4 9 For the ten pairs (20 total) of wings in e ach category, this illustrates the CV of the weight in grams. Each method improves this measure, making the duplicate wings much more consistent. ................................ .............................. 65 4 10 Histogram showing a drop in weight variation due to the enhanced processes for the radial batten wings. ................................ ................................ ................. 66 4 11 Histogram showing a drop in weight variation due to the enhanced processes for the parallel batten wings. ................................ ................................ ............... 66 4 12 For the ten wings in each category, this illustrates the CV of thrust in grams. For each batten structure, the upgraded manufacturing technique lowers this statistic, allowing for more reliability. ................................ ................................ .. 68 4 13 Histogram showing the drop in thrust variation due to the enhanced processes. The increase in thrust for the plastic frames is quite visible here as the plastic frame/transfer tape bars are dramatically shifted to the right. ....... 70 4 14 For the parallel batten wings, the thrust production for the hand lay up wings is rather scattered and pulls together with the Teflon mold. The plastic wings decrease in the number of bins and shift to the right. ................................ ......... 71 4 15 Graph of the 40 Radial Batten wings presenting how the variation decreases with each new method ................................ ................................ ....................... 72 4 16 Graph of the 40 Parallel Batten wings presenting how the variation shrunk with each new method.. ................................ ................................ ...................... 73 5 1 Static deflection DIC.. ................................ ................................ ......................... 79 5 2 Three dimensional space optimization. ................................ .............................. 80 5 3 A bubble plot presenting static deflection results with a 0.5 gram load obtained for different carbon rod percentages that occupy the plastic LE. The size of the bubbles specifies the amount of thrust obtai ned for the wings. ......... 80

PAGE 11

11 LIST OF ABBREVIATIONS AR Aspect Ratio, b 2 /S b Total wingspan (wing tip to wing tip) CA Cyanoacrylate glue CAD C omputer aided design program, which in this case was SolidWorks CNC Computer numerically controlled automated machine CV Coefficient of Variation DAQ Data acquisition device DIC Digital Image Correlation FWMAV Flapping Wing Micro Air Vehicle LE Leading edg e MAV Micro Air Vehicle PTFE Polytetrafluoroethylene a synthetic polymer commonly known as Teflon S Entire wing area UAV Unmanned Aerial Vehicle

PAGE 12

12 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Master of Science INVESTIGATING MANUFACTURING TECHNIQUES TESTING, AND DESIGN TO ENHANCE CONFIDENCE IN THRUST PRODUCTION FOR SYNTHETIC FLEXIBLE SMALL FLAPPING WINGS By Jason Thomas Rue August 2013 Chair: Peter Ifju Major: Aerospace Engineering Flapping wing technology exists and continues to expand on many fronts, although without a solid understanding behind the physics of flapping flight, a purely experimental approach can be taken. This study focuses on smal l s ynthetic wings which were biologically inspired by hummingbirds as they are comparab le in size, shape, and flapping frequency. The focus was aimed at the average thrust production from a one degree of freedom flapping mechanism and the weight of each wi ng The motivation arises with the final application of a standalone hovering device but brings to attention a question of whether or not to trust the current manufacturing process that consist s of a carbon fiber hand lay up method. The main objective of this work was to create a high fidelity manufacturing process having repeatable and robust wings resulting in improved results where minute variations in average thrust production can be detected while in hover mode To advance the po ssibilities and expand the testing envelope four distinct methods were proposed, the carbon fiber hand lay up a Teflon CNC milled mold a milled plastic frame combined with a carbon fiber rod for support and a unique attachment method with a 3M transfer tape

PAGE 13

13 A multitude of methods were performed along with a validation comparison of two different geometric structures and several wing duplicates. Conclusions heeded a sign i ficant drop of approximately 74 % in the coefficient of variation describing the wei ght distribution between duplicate wings and an over 84 % decrease for the average thrust Now that this is accomplished there are hopes that further research can be done to find how the characteristics such as structural stiffness and geometric topology h ave an effect on flight.

PAGE 14

14 CHAPTER 1 INTRODUCTION Literature Review Ever since human powered flight was imagined fixed and rotary wing aircraft have been prominently used in designs. As technology has developed these designs have taken on a new realm, shrinking in size and gaining the ability to fly with controls given on the ground. Small unmanned aerial vehicles (UAVs) and micro air vehicles (M AVs), usually not exceeding a range of 150 mm [1] continue to grow in popularity. After smaller UAVs and MAVs were introduced, traditional propulsion continued to be on the forefront of thought, although a new track started t o develop: flapping wings. Fixed and rotary wings, especially at a lesser scale level, introduce numerous problems that flapping could potentially solve. Stability issues at the reduced scale along with the inability to hover or sustain slow flight plague fixed wing aircraft while rotary wings have relatively low efficiency and a high noise signature. Flapping wing micro air vehicles (FWMAVs) theoretically could have the ability to hover in addition to forward flight, require only a short or vertical takeoff, and have high maneuverability, all advantages over the others. These properties are ideal for various indoor applications or in situations where a disguised vehicle is critical. W ith optimization, flapping technology has the potential shown from nature, to have a superior efficiency. For these reasons, flapping is worth acknowledging as a propulsion source. Work in the flapping wing field entails both experimental and modeling efforts. Computational analysis, such as that done by Lee et al. [2] shows the complexities of the fluid structure interactions for flapping objects and the realities of low Reynolds number situations Even studies like Ansari et al. [3] that characterize d the effect of wing

PAGE 15

15 geometry for example, finding a nearly linear relationship between aspect ratio and mean lift make major assumptions. Here, the wing sections were assumed to be rigid flat plates and the simulations were obtained for just one wing, l eaving out the possibility for wing interactions. The harsh nature of calculating accurate unsteady aerodynamic loads generated throughout the cycle and lack of a complete multidimensional, multi degree of freedom high fidelity dynamic model has driven th is study to focus on developments by means of experimentation. Familiarity with such testing also came from previous graduate students within the department, which is explained in the next section. Numerous studies have focused on flapping mechanics or aerodynamics [4 9] These touched on the effects of flexibility like Zhao et al. or Mazaheri or even creating an insect mimicking flapping device where a roughly 10 g flapper was able to sustain 25 Hz and produce 3 g of average thrust [6] Another group emphasized on n atural species [10 17] like insects or birds to learn what nature has provided. Several also touched on specific animals such as the H awkmoth ( Manduca sexta ) [18 19] or hummingbirds [20 23] As with many FWMAVs, so much effort goes into areas like keeping weight restraints or building the flapping mechanism, that repeatability in wing manufacturing may not be considered. As additional topics are researched and more is understood, the uncertainty of manufacturing n eeds to be quantified The question of whether studies such as [5 6, 10, 24 29] can be duplicated arises since they did not show an in depth look as to whether manufacturing could be consistently reiterated This even spreads to past experiments do ne at the University of Florida [1, 30 32] and makes the point that attention should be brought to this area

PAGE 16

16 Background Work The pursued investigation was the successor of flapping wing research done by Pin Wu [33] and Justin McInti re [34] Both scholars worked extensively with small two dimensional flapping wings that consisted of a carbon fiber frame and a thin nylon based membrane. The wingspan of 150 mm (25 mm root) was chosen by Wu so that the wing w ould remain within the maximum definition of a MAV and the aspect ratio of a hummingbird Each wing took on a Zimmerman shape, formed by two ellipses which intersect at the quarter chord point and had a triangular section to which it was attached to the f lapping mechanism This is depicted in Figure1 1 A. His work was developed by employing a single degree of freedom flapping mechanism where he gathered data, such as the averaged thrust o ver a period of time, with a load sensor. With two sets of cameras an d D igital I mage Correlation (DIC) methods, he was able to produce deformations along the entire flapping cycle. Figure 1 1. frame lay out used in the test. B) Displays the variables in this particular experiment.

PAGE 17

17 [12] of measuring the material properties of a real beetle wing as exceptional creativity. A comprehensive look at the deformations is given in Figure 1 2. Figure 1 2. Full field deformation measurements taken with DIC software and two sets o f cameras. A vacuum chamber enabled him to find differences in flight and collect information about wing loading. Wu found after several evaluations that passive wing deformation can be used to enhance aerodynamic performance. The deformation reflects the aeroelastic effects produced by both the aerodynamic and inertial loads. One significant finding was that wing stiffness and mass distribution are both highly influential to thrust production. This was a similar finding to Zhao et al. who found that as fle xibility of the wing increases, the ability to generate aerodynamic forces decrease [4] Mazaheri established that while in hovering mode, more flexible wings produce

PAGE 18

18 approximately 20% lower thrust albeit for frequencies much lo wer than used in this study [5] Six pairs of wings were part of the stiffness test and were labeled as L(#)B(#), for L gives the number of carbon fiber layers cured on the leading edge and B for the number of layers on each ba tten. Wu also looked at changing the wing membrane to better suit the characteristics required for the experiments. After deliberating over several materials including Mylar Tyvek Kapton or latex rubber, it was concluded that CAPRAN fit the lightwei ght, strength, and surface opacity needs for the flapping wings. The idea of using different manufacturing processes was also noted by Wu who attempted to inertia. By using a CNC milled aluminum mold with a release coat, he filled a mold with carbon fiber to above the top surface, cured the material, and proceeded to sand down any excess As a precursor to this work, this thought brought about the inclination to use a mold as a tool to b Wing fabrication has also been a topic for several experimentalists including Robert Wood of Harvard University. He explained in [35] and [36] that advances in Smart Composite Microstructures allowed him to produce a 60 70 mg insect like vehicle that used two 16 mm carbon fiber flapping wings ( ~ 100 Hz) to generate lift. The recent micromanufacturing technology gave the group confidence that they could mass produc e identical parts. He then worked to constrain the variability in the component assembly to produce repeatable and reliable wings. It was concluded that the prototype vehicle had not been symmetric and that it was necessary for this fabrication to

PAGE 19

19 construc t a reliable, symmetric vehicle. Even on a different scale, this proves a link between wing fabrication and repeatability. McIntire worked to understa nd if increasing twist leading edge or altering other associated deformations would produce larger average thrust values. He continued aspects set forth by Wu while also establishing a finite element model confirmed by thrust testing and DIC in hopes to reduce testing time. He began with a Zimmerman planform and identical wing manufacturing proc edures as Wu while focusing on three main frames as illustrated in Figure 1 3. Figure 1 3. Three main wing designs focused on by McIntire including one based on a hummingbird, a leading edge only design, and one reinforced on its perimeter The first, Hu m Wing is described as having radial battens that were biologically inspired by a hummingbird. The LEO Wing (leading edge only) had a strip of carbon fiber along the leading edge and one along the root length. Thirdly, the PR Wing was reinforced along the entire perimeter. Multiple series of these wings were tested using a load sensor along with full field

PAGE 20

20 highly correlated with the average thrust output. Figure 1 4 exemplifies a trend found linking the Z Centroid with thrust and adding merit to his hypothesis. Figure 1 thrust The horizontal error bars represent the range of values between flap angle s 5 and 5 with the mean denoted as a marker. The finite element model was completed in Abaqus and fine tuned with results from the aforementioned wings. Various wing designs were modeled, tabulated, and verified experimentally. After a few slight modifi cations, he was able to find wings with increased torsional aptitude as well as higher average thrust values. Motivation Crucial aspects of flapping technology have intrigued hobbyists and toy manufacturers, such as the WowWee Flytech TM Dragonfly. Release d in 2007, this 28.35 gram [37] forward flying dragonfly shows that even a simple design can endure repeated flights and require little wing optimization to be effective. As advancements ensued, universities and companies began additional in depth research. In 2011, a company called AeroVironment Inc. unveiled a 19 gram hovering UAV which was biologically motivated by a hummingbird. Despite taking about six years to complet e and having a significant expense which was funded by the Defense

PAGE 21

21 Advanced Research Projects Agency (DARPA), the project constructed a vehicle worthy of praise. Figure 1 5 WowWee t oy Dragonfly [37] Having a wingspan of 165 mm, the Nano Hummingbird demonstrated the ability to hover with a flap rate of 30 Hz fly forward, and transmit a live color video to a remote ground station [20] As impressive as this technology may be, one drawback of the l imited tools showed in the inadequate endurance of a measly four minutes. Figure 1 6 AeroVironment Hummingbird [20] Another study, extensively described in [29] showed the progression of a 50 cm wingspan (21 g) and 28 cm w ingspan (16.07 g) ornithopter and how that developed into the DelFly Micro, the smallest flying ornithopter with a camera and transmitter onboard at 10 cm ( 3.07 g ) Two of the flapping vehicles are pictured in Figure 1 7. The ability to

PAGE 22

22 observe places that are inherently dangerous are continuous reminders of the possibilities for FWMAV s and the need for more research A B Figure 1 7 The DelFly autonomous ornithopters [29] A) Top view of DelFly II. B) DelFly Micro Even though interest exists within the world of flapping technology, accurate flapping models remain difficult to produce while little is known about the physics behind such movements. Until unsteady aerodynamics and wing structure optimization is further understood, an experi mentalist approach will be the most direct and convincing strategy for future progress. Goals and Objectives To further the knowledge of this field, this project took on the objective to expand/explore the design space for synthetic flapping wings by adva ncing fabrication techniques and implementing experiment based optimization [38] Along with Dr. Peter Ifju at the University of Florida (UF) with the knowledge of micro air vehicles two additional professors were brought onto the assignment due to their expertise Dr. Raphael Haftka at the University of Florida for optimization and Dr. Tony Schmitz at the University of North Carolina at Charlotte (UNCC) for advanced fabrication

PAGE 23

23 The main objective of this paper was to create a high fidelity manufacturing process having repeatable and robust wings resulting in improved results where minute variations in average thrust production can be detected while in hover mode The goal was to cre ate a systematic approach based on previous studies that limits errors and breeds full confidence in the accuracy of data collection. If done correctly, it would lend others a superior method to produce wings and allow for a more precise optimization. The ultimate goal after a thorough understanding of the flapping process is completed will be to create a free standing hovering mechanism that utilizes the gained knowledge of flapping wing technology. Although this was not researched directly in this paper, the idea rests that certain aspects, such as wingspan or flapping frequency, should not exceed power constraints or size limitations available on that type of MAV and is carried out in the entirety of this paper

PAGE 24

24 CHAPTER 2 EXPERIMENTAL CHARACTERIZ ATION OF MANUFACTURING PROCESSES Wing Structure The chosen overall wing dimensions were essential in this study to remain within the realm of restrictions for a standalone hovering MAV. In terms of size, the wings need to be small enough to limit inertial effects yet large enough to ultimately hover and carry a payload. Continuing from the work of Wu and McIntire, a 150 mm span (b) wing (75 mm half span) with an a spect ratio [39] of 7.64 was chosen. The wing also remained flat without a camber. This falls within the range of a honeybee ( Apis mellifica ) at an AR of 6.65 an d a Rufous Hummingbird ( Selasphorus rufus ) at 9 [7] Keeping this general size was also significant, as nature follows a similar trend in requiring small wings to hover Warrick et ata from analyses of aerod ynamic models and from empirical studies of the mechanical power and metabolic cost of flight at different speeds in birds all agree that hovering flight is much more expensive than intermediate speed forward flight [21] If similar to nature, it could be expected to use wings smaller than many fixed wing, forward moving, MAV wings. He continues to explain that due to the high power loads necessary, hovering has become only an attainable goal for small birds. Since these wi ngs are on the same scale as a hummingbird, the dimensions were deemed acceptable. A popular approach among experimentalists to obtain a wing structure has been to mimic natural species like Nguyen mimicking a beetle [10] Rat her a separate methodology takes on the understanding that nature may not develop a geometrically

PAGE 25

25 optimized wing structure in terms of thrust output for it takes into account nutrients, a complicated multi degree of freedom flapping motion, other flight d ynamics, and an A certain bird or insect may need to grow to a specific size to fend off predators in the wild while possibly surpassing ideal dimension s for hovering flight Although inspired by these natural flyers, the wing frames estab lished in this study had the opportunity to evolve past natural configurations to include complicated topologies. It should also be noted that flyers, like hummingbirds, commonly have intricate flapping strokes including many active degrees of freedom and multiple hinge joints, comprising of asymmetrical stroke paths [22] Therefore the simplistic wings actuated with a single degree of freedom created here may behave differently and hold a varied optimized structure for maximum thrust output due to their altered kinematics, inability to change during flight, and complicated nonlinear geometries. The Zimmerman planform used by Wu and McIntire was abbreviated to a quarter elliptical area to ease manufacturing and gain repeatability Figure 2 1 expresses a fraction of the frame topologies possible to investigate Figure 2 1. Examples of wing combinations with different frame topologies

PAGE 26

26 Together with using the verbiage of quarter elliptical area in this paper, a few other terms in Figure 2 2 are defined Half span will describe the distance the wing extends along the leading edge (LE) while root is the distance in the chord direction at its inner most edge. Figure 2 2. Wing area along with notation used in this paper Flapping Mechanism A challenge arises when numerous goals are being kept in mind. The flapping mechanism must be robust and able to bear months of continuous use although it must actuate the wings symmetrically, have an easily fixable structure, and have t he wings detach effortlessly. As this project began, the first version of the flapping mechanism was in use and able to flap wings up to roughly 45 Hz limited by the frame, motor, and controller components This was deemed acceptable for the subsequent ve rsions for it compares to examples in na ture with similar wingspans. A H awkmoth ( Manduca sexta ) with a span of 96.6 mm flaps at 26.1 Hz while a Rufous Hummingbird ( Selasphorus rufus ) with a span of 109 mm flap s 41 Hz [7] Also, the wings would need to produce enough thrust so that in hover mode, a standalone flapping mechanism could offset its own weight. Nguyen et al. molded a 10.26 g device that flapped wings with a span of

PAGE 27

27 125 mm up to 37 Hz [11] This coincides well with the work done by previous students and adds a proof of concept. Mechanism Version 1 The first flapping mechanism was fabricated by a former PhD student, Pin Wu and was still in use under Justin McIntire. This mechanism, and each described thereafter, is powered by a brushless 15W EC16 DC Maxon motor (http://www.maxonmotorusa.com ) with a 57/13 reduction ratio planetary gear head, a 256 counts per revolution encoder, and an EPOS 24 controller. The Maxon motor is able to turn an output crank module up to 45 Hz or a torsional component of 21 Nmm. This motion was transferred to linear single degree of freedom by a push rod to a mount and reciprocator to create a flapping motion as seen in Figure 2 3 This was largely done for the sake of reducing weight and requiring fewer parts, along with adequately resembling a hovering MAV. Wings were then glued onto the carbon fiber wing mounts with cyanoacrylate ( CA ) adhesive Figure 2 3. Flapping M echanism 1 capable of 45 Hz and 60 flap angle. Comprised of many hand scupted parts and numerous screws. Courtesy of Pin Wu.

PAGE 28

28 Unfortunately, this version ha d significant issues with unnecessary friction and screw misalignment along with being constantly under repair. The frame and parts wer e put together by many M1 screws that were loosened by the vibrations of the push rod and also needed lubrication amidst trials Due to the complexity of the device and the way it was designed, fixing certain parts, like the reciprocator, would require a l arge majority of the flapper to be disassembled. Another concern was how the wings were attached as glue residue built up on the mount s and carbon fiber frame. This led to difficulties if either a wing was to be tested again or removed for a variety of rea sons. The motor was controlled by a LabVIEW program that was later exhaustively overhauled by fellow lab mate Kelvin Chang to improve efficiency and add viable options Mechanism Version 2 The next iteration held slight upgrades to Version 1 to condense the number of parts and decrease certain errors. The first adjustment was to remove the former reciprocator and replace it with a CNC (computer numerical control) milled block of nylon. The ad vantages to this change included the reduction of parts for the reciprocator from around 12 to 1 and not needing to rely on frequent applications of lubricant. Next, the wing mount s and rockers were milled out of aluminum to provide easily replaceable parts, considerably enhanced connections, and a more symmet ric flapping motion as opposed to the hand made originals. The mounts also had a recessed area to better repeatedly position the wings. The actual machining for this and future versions, was done by Dr. Tony Schmitz and the graduate students in his depart ment. Lastly, the mechanism was painted white, largely to help with DIC measurements where black dots

PAGE 29

29 were tracked on the wings. In Version 1, there were times where the edge of the wing was indistinguishable with the device in the background. Figure 2 4 Flapping M echanism 2 equipped with a nylon block and machined parts Courtesy of Justin McIntire. Even though great strides were taken with this step, u ltimately, the complications that propagated throughout this version lead to the creation of Version 3 Mechanism Version 3 A substantial renovation was done for V ersion 3 to reach a flapper that could be refurbished quicker with fewer parts while having less friction and wear on the motor. To start, a brand new motor, encoder, and gear head a direct copy of those used prior were coupled to an all aluminum frame, as can be viewed in Figure 2 5. A set of nylon gears transfer red the spinning motion of the motor to two cranks and pushrods while the wing mounts pivot ed around a screw. This removed the problematic reciprocator along with the friction that accompanied it. Four holes in the cranks allowed the flapping angle, the angle above and below mid plane the wings travel, to be altered to four settings: 21, 32, 48, 52. One large improvement was the move to larger and fewer screws, permitting less slop and maintenance.

PAGE 30

30 Figure 2 5. Flapping M echanism 3 furnished with aluminum frame and fewer parts Courtesy of Justin McIntire Again, small issues crept up as e xperiments continued, especially that of how the wings were connected, although, this version stood as a significant upgrade to any past mechanism. Mechanism Version 4 Many of the s ame options exist on the most current flapping mechanism as in Version 3. T he main difference for this flapper is the insertion of two bearing s where the wing mounts pivot to further reduce friction. This creates an effortless movement that improves controllability and lengthens the life of the motor components. Another huge adva ncement was in the wing attachment points. Up until this flapper, the tester would need to glue the wings into place taking up time and leaving behind residue. Instead, the wing would be inserted into a corner slot and held in place by a tightened screw ( Figure 2 6). This reduces the time needed to change wings, elongates the life of the wings, and still allows for proper placement.

PAGE 31

31 Figure 2 6. Flap ping M echanism 4 with screw attachment and bearings Photo courtesy of Jason Rue. Other than wear and tear with typical expected replacement parts, this version suits every need for the current experiments and can be changed quickly with new aluminum parts if new ideas arise, such as a different flapping angle or wing attachment Data Collection A 6 axis (3 force, 3 moment) force/torque sensor Nano17 [40] was used to measure the thrust and lift with a resolution of 0.318 g and read through to the computer 30k times per second by a National Instruments NI USB 625 1 16 bit data acq uisition (DAQ) device ( http://www.ni.com ) while being controlled by LabVIEW Figure 2 7 shows the elaborate control panel for the experiment. The leftmost screen has the options to regulate the motor and adjust features such as the device settings, acceler ation, the testing ramp period, frequencies, and a custom file naming system. The dial also gives the commanded target frequency along with what frequency the motor is currently

PAGE 32

32 running. The right panel controls the DAQ device and details like the record r ate while dumping the data into text files. Every file was labeled with a custom naming scheme to not only stay organized but to later read them through MATLAB Text files began with a seven digit number the takes into account a specified wing number, the frequency, and the tr ia l number ( WWWFFTT ) and followed by several terms relating back to either lift or thrust. A B Figure 2 7. LabVIEW virtual instrument s that control the testing procedure A) Motor control. B) Sensor Readings. Changes were implemented to adjust the rate at which the data was read because a correlation amongst rate and overall thrust scatter was discovered. 30k readings per second were chosen to still benefit but not bog the computer down. The load cell generates a rather noi sy signal making it challenging to interpret any results before post processing occurs, therefore, to ease deducing results a filter was inserted into the LabVIEW code. A point by point low pass Butterworth filter was used real time in the LabVIEW softwar e code to process the noisy signal to present the low frequency thrust offset data that is desired for each frequency. This filter provides the operator of

PAGE 33

33 the experiment real time performance that is easy to interpret, serving as a check to assure consist ency and create a warning sign for possible delamination or damage to the wing between trials [41] This gave the experimenter a look into the data graphically on LabVIEW as previous codes failed. It should be noted that avera ge thrust and lift values remained the same before and after applying the filter to the system. Each continuous run was done using the 48 flap angle at three frequencies, 20 Hz, 25 Hz, and 30 Hz, recording for six seconds each and allowing a ramping per iod of 3.2 seconds in between each frequency. These ramps let the motor reach and settle on a frequency before data was collected. To prevent outliers from degrading the numbers, each wing was ideally tested ten times and averaged. The standard deviation was also noted to realize the spread. Occasionally, wings failed by either delamination of the CAPRAN (membrane) or breakage of the frame In these cases, the situation is noted and repaired if possible unl ess a substantial disaster occurs, in which case the wing is remade The wings were situated in such a way (Figure 2 6) so that gravity could be neglected from any measurements. At the beginning of each run, the program read off a four second tare measurem ent without the mechanism moving to take into account The length of time chosen for each section of the ramp was picked to be sure a sufficient amount of data was taken while the wings were in the correct frequency range without overburdening the flapping mechanism or damaging the wings. Figure 2 8 gives a brief example of how the data was observed before and after a filter was implemented. The cyclical nature of the 25 Hz signal and relatively flat 20/30 H z data occurred habitually.

PAGE 34

34 A B Figure 2 8 Filtering raw data to help interpretation. A) An example of complete raw data ranging from 20 to 30 Hz. B) A separate instance after filtering for a similar flapping run. The MATLAB code previously used was extensively revamped and rewritten by the author to not only be exceedingly efficient as opposed to what the lab had used previously and include a GUI interface to be more user friendly but export data to Microsoft Excel after the data was averaged Detai ls of the interface are shown in Figure 2 9. Figure 2 9. MATLAB GUI interface.

PAGE 35

35 After a few user inputs like the wing number, trial, weight, and frequency range, the code is run simply by pressing three buttons. T he written code reads in the text files th at LabVIEW created and averages the raw data over the specific run time to get a single averaged lift and thrust value for each frequency. A ddition ally other numbers such as the average frequency over the run (due to the controller, the frequency is not h eld completely constant), current draw, and minimums/maximums are either saved automatically into a data structure or graphically displayed and saved as JPEGs at the Any important calculated values that are warranted for manipulation are then exported to Microsoft Excel along with a comment that typically would describe the current study The data was then manipulated in Excel to extract more graphs and analysis. Visual Inspection For as many comparison tests that exist to define differen ces between geometries one that demonstrated to be most level of acceptable accuracy expected within this study, especially for nominally identical wings, far exceeds what one might get if two visually incomparable wings are run. Meaning that if wings have a tendency to display inconsistent batten widths from wing to wing carbon fiber that has strayed or bled away from the structure or uncontrollable dimensions such as varying strip widths created within the manufa cturing process the scatter associated with the gathered data will be too great. Figure 2 10 indicates how a wing might appear if a hand lay up method is taken. As mentioned in Chapter 1, if minor disparities are to be measured, then every detail througho ut the procedure of measuring average thrust would require scrutiny.

PAGE 36

36 Figure 2 10 Image of a wing manufactured by a hand lay up method having considerable visual inconsistencies such as t he battens differ ing in width, the LE is a function of span, and th e triangular section has smashed out beyond the desired dimensions. Photo courtesy of Jason Rue. In the figure, the battens have different widths the triangular attachment point is not distinct, and the leadi ng edge increases in width toward the tip. In t rying to reduce scatter within the data and adjusting manufacturing techniques to keep the necessary meticulousness, the eye test was the first hurdle to overcome.

PAGE 37

37 CHAPTER 3 MANUFACTURING The wings take advantage of passive wing deformation, much like hummingbirds or insects, as described by Wu [33] The wing compliance and LE stiffness have positive aerodynamic effects and, while using passive deformation, can sim plify kinematic actuation while reducing parts and weight. Having wings that are repeatable and robust are essential to any further developments in this area. Wings created in this paper can be divided into constituents, frame and the membrane. This chapte r outlines several ways to produce the wing frame s each evolving to enrich the latter. The material used for the membrane remain ed a constant throughout the entire experiment al program and originated with Wu An extremely thin (14 microns) nylon based fil m by Honeywell, called CAPRAN 1200 Matte, is used for the membrane, although according to their specification sheet, general uses consist of thermal lamination for book cover applications [42] This material has a relatively h igh tensile strength at rupture (234 276 MPa) and puncture strength (925 g) while being very lightweight. In comparison, another material previously used at the University of Florida with a similar elongation Mylar has a n ultimate tensile strength of 200 MPa (M achine D irection )/234 MPa (T ransverse D irection ) [43] Initially, a spray adhesive by 3M named Super 77 was used but the glue remained tacky long after it was applied. McIntire performed peel tests on an Instron machine with several different adhesives [34] He found that a thin cyanoacrylate (CA) adhesive held much tighter as it had a more brittle bond. Even after numerous additional trials during that inquiry, no better option was found. The first three processes in this paper use CA

PAGE 38

38 while the fourth manufacturing process introduces a solution to frequent problems such as membrane delamina tion, an adhesive transfer tape from 3M In addition to the 3M 9471LE Transfer Tape, a few future and ongoing projects are also described in Chapter 5 to improve the bond between these surfaces The wings used by McIntire weigh roughly 0.1 2 g to 0. 22 g so new methodologies were designed to keep the wings in a similar range not only for consistency but to not overload the motor. For each progression involving prepreg carbon/epoxy detailed in this chapter, regardless of how pressure is applied during t he curing cycle recommended two hour controlled temperature profile is used and was detailed in Figure 3 1 As for the carbon fiber itself, a ~35% toughened epoxy resin unidirectional carbon prepreg was purchased from The Composite Store ( http://www.cstsales.com ) The fiber and the AR250 Series Resin originated from Aldila Composite Materials The woven bidirectional carbon epoxy prepreg (AX 51 6 0) cut for the triangular sections w as from Axiom Materials ( http://axiommaterials.com ) Figur e 3 1. Manufacturer s recommended o ven configuration to cure ~35% resin carbon fiber taking roughly two hours These steps were programmed into the oven and followed exactly.

PAGE 39

39 Hand Lay Up Process The first step to creating a high fidelity experimental process was to improve a commonly used method for wing assembly, as Nguyen [6] Wu [33] and McIntire [34] performed which entails a hand lay up technique where carbo n fiber strips are delicately laid on a flat plate in a certain pattern and cured in a vacuum bag with in an oven Bidirectional carbon fiber ma de up the triangular attachment area whereas the unidirectional composes the rest of the frame The main advantag e to this process was that it was done entirely in house and was easily customizable. Process One Stages involved with this route are critical as the hand lay up serves almost as an art form. Once a certain design had been set, a thin outline of the featur es was printed and glued to a flat plate with Super 77, being careful to keep the surface as pristine as possible. The surface was covered by a very thin ( 0.0254 mm, 0.001 in) thermally stable release film called D7000 ( http://www.decomp.com/ ) to prevent t he carbon fiber from sticking Cautiously 6 mm by 12 mm triangles were cut by hand from the bidirectional prepreg and narrow, roughly 0.8 1.0 mm strips were trimmed from a larger sheet of unidirectional prepreg. The triangular area was then laid in three layers with a leading edge sandwiched in between, which was composed of one to three layers depending on the desired stiffness. A single carbon fiber layer made up each other support batten After the entire fram e was sufficiently oriented and aligned with the printed pattern, another piece of release film was taped on top to help secure the carbon fiber and keep it from sticking to the vacuum bag The whole plate was then placed into a

PAGE 40

40 sealed vacuum bag, compress ed by 30 in Hg to insure proper consolidation and positioned in the oven. Once the cycle was complete, the wings were delicately removed from the flat plate and prepared for the gluing process. Each wing was held with tweezers while CA was meticulously sp read over the carbon fiber. If too much glue was placed on the surface, the excess would spread out onto the CAPRAN and change the membrane properties. On the other hand, if an insufficient amount is dabbed onto the carbon fiber, delamination would almost certainly occur during testing. Again, the advantages are listed below. Done completely in house Method easily altered for new wings Issues This process is routine and typically used but holds many uncertainties and concerns. Obstacles include the loss of the cross sectional shape as it generally flattens out under pressure losing stiffness, along with the point that it is also fairly difficult to keep the fibers aligned or cut the strips with consistent accuracy. Figure 3 2 illustrates the matter of losing the defined rectangular shape of the cross sectional area thus reducing the moment of inertia Since Wu found that the leading edge stiffness and minimizing weight play a large role in thrust production, one can immediately make a generalized statement that the wings laid up by hand are not optimal for thrust production and are less efficient. For a certain level of thrust production, these wings also weigh more than they should due to that fact, as the carbon fiber could be utilized more efficiently.

PAGE 41

41 A B Figure 3 2 Problem of c arbon fiber on a flat plate loosing structural stiffness A) Diagram description c ourtesy of AFOSR Flapping Proposal. B) Actual picture of the cross section of a three layer carbon fiber LE. Photo courtesy of Jason Rue. Therein lays another predicament with the struggle of commanding the battens to a definite position as very little restrains the carbon fiber while it cures In Figure 3 3, two pictures explain this point. The battens are off kilter the LE developed a bow, and the triangular area is by no means distinct. These wings fail the visual eye test and could not possibly be reproduced. For optimization to produce reliable results and to improve upon average thrust production, the wings must be repeatable. The thrust values obtained through experimentation must be trusted with little variation. A B Figure 3 3 Typical deviations in carbon fiber from the hand lay up process A) Overflowing triangular area and u neven widths B) Curved LE and imprecise angles Photo s courtesy of Jason Rue.

PAGE 42

42 Each wing when it is glued to the CAPRAN was then cut by hand to the desired curve. This is difficult to mimic and can be a major drawback since lift is generally proportional to the wing area. A summ ary of the deficiencies include: Hand cut prepreg carbon/epoxy Loss of cross sectional area Difficult to keep fibers aligned Many human components to process CA bond and shelf life Teflon Mold Method To correct several of these issues, a CNC milled Teflo n (PTFE) mold was created The mold was resistant to sticking, held the carbon fiber in place during the curing cycle and kept the cr oss sectional shape intact. As a n alternative to the vacuum bag, a silicon e mat with a temperature range of 50F to 500F, well within the oven profile was layered to apply pressure and then was clamped down bounded by two 6.35 mm ( 0.25 in ) aluminum plates. The specific plate thickness was to deter warping when the screws were tightened. Process Two The p hases involve d with this process started with a 3 mm thick sheet of PTFE. Via Chris Tyler, Figure 3 4 depicts a 1 mm diameter, four fluted square endmill bit on a three axis milling machine (Haas TM 1) cutting the sheet. This design was custom and could be redone simply by altering the CAD model. These Teflon molds could be milled in a relatively quick period of time and allowed for several consecutive carbon fiber curing cycles as they were reusable.

PAGE 43

43 Figure 3 4. A CNC mill cutting the Teflon mold shown in the upper right. P icture courtesy of Chris Tyler, graduate student of Dr. Tony Schmitz. Once the mold was shipped to the University of Florida, carbon fiber strips were cut similar to the previous method. One large difference was the construction of a custom multi bladed strip cutter (Figure 3 5 ). Instead of continuing past efforts to struggle with consistency, this device allows up to eight strips to be cut at one time while each having a very dependable width of 0.8 mm. The quan tity of strips could be controlled by placing a certain number of razor blades within the apparatus. Figure 3 5. Custom strip cutter created to not only cut carbon fiber strips consistently but cut multiple at a time becoming more efficient and saving t ime Photo courtesy of Jason Rue. Strips would then be skillfully positioned into the mold, like Figure 3 6B indicates, and then sandwiched with release film and a 40A durometer silicone mat ( 1.5875 mm, 0.0625 in) between two aluminum plates. Figure 3 6A e ncompasses a cross sectional

PAGE 44

44 drawing of how the layers are assembled. It is noteworthy to mention that initial Teflon sheets were permanently coupled to aluminum plates post milling. Subsequently, large shipping costs and warping led to a detached sheet that extended out to the border where the screws held the sheet in place during the curing process This halted many of the warping issues and allowed only the PTFE to be shipped from the UNCC campus to UF, drastically reducing costs. A B Figure 3 6 T he Teflon Mold in two views A) An exploded view of the aluminum sandwich structure B) Photo of post cure mold Photo courtesy of Jason Rue. After the curing cycle in the oven, the wings were carefully removed from the mold attached to CAPRAN with CA and cut out by hand As can be seen below, the frames are considerably improved and compare well with the hand lay up frames. Figure 3 7A illustrates the advances made as the carbon fiber strips hold an even width are much straighter, and the triangular section is much more defined. The second picture lays duplicate wings side by side matching b etter than previously possible with the methods demonstrated by those at the University of Florida A rehash of the advantages follow s the figure

PAGE 45

45 A B Figure 3 7 A quick visual comparison regarding the wings from the hand lay up and mold processes. A) C omparison between two carbon fiber frames created with the hand lay up and mold methods Visually, one can see the improvement brought by the mold process. B) A series of carbon fiber frames that meet the visual test and show promising repeatability Photos courtesy of Jason Rue. Much more consistent strips Was able to keep the cross sectional shape better No vacuum bag saved material and time Reusable molds Iss ues Some issues still remained, especially those with uneven pressure distribution warping, thermal expansion, and mold depth M any times the resulting specimen contained batten cross sections that lacked repeatability. The result of a shallow mold channe l or too much carbon fiber was that accumulated above the mold and is shown in Figure 3 8A H ere a leading edge was cured, cut, surrounded by epoxy, and polished so that an approximately 150x photo could be taken. If the mold channel was too deep or a deficient amount of carbon fiber was placed into the groove, a recess consistent with the pressure distribution formed caus ing air bubbles and less contact area when the top surface was glued to CAPRAN A side experiment wa s performed to to particular mold depths, althoug h some variation still existed.

PAGE 46

46 One downfall of the strip cutter was that the blades needed to be positioned precise ly or when the tool was pressed down to the material, a few blades pierced further than the others Occasionally this would even cause a blade to not penetrate the carbon fiber and an entire strip would be skipped. Also, the carbon fiber would, at times, b unch up or move slightly as the cutter moved through it. As mentioned above, this led to having extra or too little fiber in the mold. One solution could have been to just fix the carbon fiber sheets down to the cutting table or establish a pretension A B Figure 3 8 One problem with the Teflon mold process was the difficultly of syncing the mold depth and amount of carbon fiber correctly so that a rectangular cross section was formed. A) Zoomed in view (approx. 150x ) of the cross sectional shape post cure B) Looking at cross section al shape variations Photo courtesy of Jason Rue. An irregular pressure distribution would also occur if screws on the periphery were tightened incorrectly. If a poor compression transpired by screws that were under t orqued, a lack of carbon fiber consolidation would ruin the frame. The screws would also present a dilemma if one side of the mold was squeezed stronger than the other. Plate warping became a nuisance and led to the change from thinner aluminum to the 6.35 mm ( 0.25 in ) plates. Originally, the Teflon mold spread only up to the screw holes, leaving a sizable gap between the outer portions of the plates when sandwiched. A bowing ensued and managed to form an added reason for unbalanced pressure.

PAGE 47

47 Thermal expansion was deemed an explanation to the occasional buckling as it continued to appear While still mainly in the mold under the post cure phase, the carbon fiber would lift up. Figure 3 9, provides a look at this phenomenon and hints to the ques tion of whether the carbon fiber cured properly When this happened the structural integrity seemed to stay intact and the LE flattened once out of the mold. Figure 3 9 Buckling possibly due to pressure distribution or thermal expansion Photo courtesy of Jason Rue. As the molds run through multiple curing cycle s excess epoxy from the prepreg b uild s up and leaves residue In Figure 3 9 a discoloration reveals evidence of this point. After extended applications, the carbon fiber gains a tendency to adhere to the mold making it impossible to remove, thus only work ing for so long Weak joints caused by the sudden merge of a flexible one layer batten and either the LE or the triangular attachment area, in many cases, nullifies the wing before data can be taken. Figure 3 10 gives a glimpse of an incident where the root batten completely detached Extra epoxy also frequently extends from the frame and must be cleaned off.

PAGE 48

48 Figure 3 10 Example of cured carbon fiber frame where b reak s occurred along weak joints Photo courtesy of Jason Rue. A few alternatives of the mold design occurred in conjunction with the aforementioned, including a Teflon sheet that did not stretch to the border of the aluminum plate and one with thinner aluminum plates with fewer screws, both of which established warping concerns. For any mold where the Teflon stopped short of the outside edge, the PTFE was adhered to the aluminum plate. Assorted solutions to help stop the sheet from moving were attempted including: a d ouble sided carpet tape that stuck well although partially melted, the CA or Super 77 that was not strong enough to last through the oven cycle, and a 30 min Epoxy that ended with mixed results. Any thick adhesive commonly dried with lumps, worsening the p ressure distribution. These were ultimately discarded and replaced without a fastener and a larger sheet fixed in place by the through screws for reusability and costs. Because any plates slimmer than 6.35 mm ( 0.25 in ) bent when the screws were tightened a nd molds with fewer than four screws per side held lower pressure on the carbon fiber, all styles incorporating these features were rejected. One design where a thicker Teflon mold was placed into a vacuum bag to apply pressure had decent results but had trouble pressing into the crevices of the mold and curing correctly Ultimately, a list of the issues is listed below in bullets

PAGE 49

49 Uneven pressure distribution Warping Thermal expansion Buckling of cured fibers Mold Depth Cutting prepreg carbon/epoxy consis tently Epoxy build up on molds Weak joints on frame Travel time between universities Cross section repeatability A new design took an entirely new mold CA bond and shelf life Plastic Frame U ltimately cost and precision needs drove the study to the current method. Instead of molds, the author switched to a plastic [44] which was CNC milled into the shape of the desired frame and fitted with a commercially available 0.5 mm ( 0. 0 2 in ) diameter Graphlite [45] carbon fiber circular rod The rod slid into a 0.2 mm trench on the LE for stiffness created by a 0.4 mm endmill One huge advantage to this procedure was that the y could be machined in less than two mi nutes per wing as opposed to the hours necessary with carbon fiber. Strong stiffness and adequate fatigue properties were the reason behind acetal resin. This was a much more cost effective manufacturing process, given that the core study calls to test hun dreds of wings. It also incorporated less human error, benefitting from machine precision batten placement and thickness A dding CNC precision throughout the wing provides repeatability and keeps craftsmanship as a priority The wing area has an effect on the thrust output, so to maximize repeatability and to further development; faint marks were printed onto the CAPRAN as cutting guides before it was adhered to the frame An ink jet printer was utilized so that heat did not affect the material properties of the membrane.

PAGE 50

50 Process Three Following the design being drawn in a CAD program, t he Delrin frame was milled at the University of North Carolina at Charlotte and shipped to the University of Florida There, the author, t o keep a solid bond between the frame and rod, applied rubber toughened cyanoacrylate glue to the LE and attached the rod. This CA was sought after due to its added strength from the thin CA used to bond the membrane. T he frame was then adhered with CA to CAPRAN like previous wings. The weight of the wings compares well with the previous carbon fiber wings, ranging from approximately 0.1 to 0.25 grams per wing. Several plastic frame wings, laid out in Figure 3 11, justify the CNC mill as each wing manifes ts fine detail Stiffness wise, this process is far more controllable than any other especially with the pre cured carbon fiber rods. For a closer look, Figure 3 12 displays three photos that demonstrate the precise nature of the CNC milled wings along wit h a cross sectional view at the tip of a wing. A B Figure 3 11. Completed plastic frame wings with the carbon fiber rod, frame, and CAPRAN already combined. A) 0.5 mm circular carbon fiber rod glued to LE. B) Size and shape example of completed wings shown with a metric [cm] scale. Photos courtesy of Kelvin Chang.

PAGE 51

51 A B C Figure 3 1 2 Zoomed view of the plastic frame A) Close up of the triangular section. B ) LE trough that shows the detail that can be manufactured. C ) Cross section with the rod attached to the frame. Photos courtesy of Jason Rue. White Delrin was used in some of the wings to help with DIC measurements Figure 3 1 3 arrang es three photos, each covering the process. The same mill used previously sculpts the wing while the other photos give an example of a frame before and after the rod is adhered to the LE Figure 3 1 3 A figure with the steps of the process from milling to the plastic white frame to the finalized wing. A) The mill carves out the acetal resin. B) W hite Delrin frame with carbon fiber rod attached Even though the rod is black, the black LE is mainly due to the rubber toughened CA. C) White Delrin frame. P ic ture A courtesy of Chris Tyler, graduate student of Dr. Tony Schmitz. Photo B C courtesy of Jason Rue.

PAGE 52

52 Cost effective Dramatically less time to create wing LE has commercial repeatability Color allowed for better DIC measurements Less human error More cons istent wing area Issues Almost all parts of this wing are machine controlled except for attaching the carbon fiber rod to the LE and the frame to the CAPRAN The amount of glue build up is difficult to control by hand and can lead to slight weight differences between wings. The manufacturing process also created a few residual stresses within the frame, leading to slight warping and a curvature to a fraction of the wings. Multiple problems arose when the CA would reach the end of its shelf life as i t thickened and lost effectiveness. The existing problems are listed as follows. Fastening the rod to the LE Attaching the plastic frame to CAPRAN as it comes off frequently Travel time between universities Residual stresses from manufacturing which cause slight warping CA bond and shelf life Transfer Tape Technique A dhering CAPRAN to the plastic frame became problematic quite frequently during testing periods and work was done to find new techniques to attach the membrane with new adhesives and vacuum pressure or compression By far, this seemed to be the most pressing issue with wing development. This led to more human interaction and repeated attempts to re glue the bond. Additional testing and curing time was also a small concern. A few options of acrylic or rubber based adhesives with high shear and initial tack that typically occupy the back of permanent stickers stood as the primary materials. The potenti al here was to virtually peel and stick the frame onto the

PAGE 53

53 CAPRAN removing weight from excess glue and human error along with speeding the procedure. A solution was a 58 micron 3M High strength double sided Acrylic Adhesive [46] typically used for plastic nameplates, graphic overlays, or attaching identification material to lightly oily surfaces. This removed the thin CA from production and while substituting a durable, long lasting, and time saving material. Process Four All i nitial steps were barrowed from the plastic frame process including the CNC milled frame and circular carbon fiber rod. A change occurred next as the frame needed to be adhered to the membrane. The frame was positioned onto the one inch wide transfer tape carefully to make sure the whole area made contact. Once settled in place, a piece of white computer paper (cut to size) sandwiched the frame with the tape backing. Pressure was applied along the edges of the frame with a finger until the tape could be pee led away completely from the backing. At this point, the frame could be loosened as well and pushed onto the CAPRAN The wing was pressed down to remove any gaps and was then cut out similarly to prior procedures. While a custom press applied the needed c ompression required for this step, vacuum pressure could also strengthen the bond and reduce voids although was found to be lengthy and less effective A few of the advantages are bulleted below. Two year shelf life Extremely smooth and consistent High initial adhesion No need for glue to dry Issues Even with being a sound material and procedure, a few issues still existed. Obviously the carbon fiber rod was still glued on with rubber toughened CA and the

PAGE 54

54 plastic frame had to be created at UNCC as well as the stresses in the frame. Opposed to the widely available thin CA, the 3M transfer tape cannot be purchased in small quantities. Lastly, the tape sticks immediately and has a strength that is not conducive to fixing an ill place frame. Attention was ne eded to appropriately position the frame to the CAPRAN Fastening the rod to the LE Travel time between universities Residual stresses from manufacturing which cause slight warping Sticks immediately and allows no play Not available in small quantities Other Attempts As previously described, a few manufacturing processes have been developed to help diminish scatter and spawn superior results, although, not every attempt worked. Before the Teflon CNC milled molds were created, Dragon Skin High Performan ce Silicone Rubber ( http://www.smooth on.com ) was used as a surrounding to help keep as pictured in Figure 3 1 4 This effort was unsuccessful since the silicone lacked the strength to hold the carbon fiber in place. A B Figure 3 1 4 Effort to use silicon e shape. A) After laid on a flat plate, the silicon e rubber was poured on top of the ca rbon fiber. B) The silicone rubber was treated as a female mold. Apart from changing the depth of the mold, different durometer silicone mats specif ically a super soft 10A and 20A with varying thicknesses were used to apply

PAGE 55

55 pressure with the hope of varying the penetration and pressure distribution within the Teflon mold channel This would also exhibit failure and added more variables to validate. The softer material generally spread outward and into the groves as planned although was not solid enough to apply a high pressure. If the mats were clamped differently and not allowed to expand outward, it might have been possible to administer the correct level of pressure. The LE thickness was shown to hold importance due to the increased moment of inertia As a result, thicker plastic L Es were constructed but were identi fied as being too heavy. Most overloaded the controller and refused to run. In the attempt to choose a n ew material for the frame a few other routes were tested. The thought behind switching was mainly due to accuracy but also to decrease dramatically the manufacturing time. Isotropic materials such as a luminum, titanium, and ABS plastic, as well as a milled 0 90 bidirectional cured carbon fiber plate were machined. The wings were either too soft or not stiff enough for this application, although it led t o the composite plastic/carbon fiber method currently adopted. Each method had advantages, although many had a slew of shortcomings. The hand lay up process became archaic and quickly fell behind in terms of reliability, although wings by this method could be erected in house with ease. The Teflon mold worked well but struggled to deliver the needed precision Delrin plastic frames with an inserted carbon fiber rod evolved into the best method to date.

PAGE 56

56 CHAPTER 4 EVALUATION OF TECHNIQUES Again, to touch on the goal of this paper, a past manufacturing approach was to be made repeatable and reliable so that small variations in thrust production between wings could be concluded as a definitive gain or loss. If the thrust measurements of a sin gle design recorded in recurring attempts are too scattered, then optimization and further experiments would prove to be very difficult. This is based on the idea that a certain frame design should result in the same thrust production every time it is prod uced. Even though the flapping process is highly dynamic and controller driven, it is an assumption that constant thrust production is plausible. To determine how the manufacturing procedures related with respect to average thrust production, duplicate fra me s were created with each method described in Chapter 3 to ascertain whether the scatter in the data was reduced The w ing replicates (separate copies of the same nominally equal design) would then show if a tested result could be trusted. Ideally, every duplicate would yield the same result. As duplicates were produced, attention was also focused to identifying how weight differences propagated throughout the batch. Two Design Assessment In this paper, two frame designs were fashioned with four manufactur ing styles : a hand lay up, a Teflon mold sandwiched by aluminum plates, a CNC milled plastic frame and a new membrane attachment method using transfer tape Plain frame drawings, seen in Figure 4 1, describ e both design s : one characterized as having parallel battens and one with a radial pattern. As mentioned in Chapter 2, a 75 mm x 25 mm wing was tested throughout this study. A group of patterns were chosen since they

PAGE 57

57 had been used in the past with success without major deformation, delamination issu es, or inadequate thrust and these two were randomly taken from that selection A B Figure 4 1. Shown are the t wo designs chosen for the validation process Although created in the past, these were randomly chosen from a group of wings that had decent test results A) Design 1, described as having parallel battens. B) Design 2, defined as radial battens. To further understand the scope incorporated within this study and to visually show explanations regarding data collection Figure 4 2 illustrates the hierarchy breakdown for one design and the total quantity associated to each level. For every manufacturing type, ten duplicates of a single design were assembled leading to eighty pairs in total, with each comprisin g of ten trials. Figure 4 2. Hierarchy breakdown of the experiment explainin g a total of 8 0 du plicates were constructed with 8 00 runs The specific dimensions for both designs are arranged in Fi gure 4 3 followed by samples f r om each manufacturing technique. Three layers of carbon fiber ma d e up the

PAGE 58

58 LE in both the hand lay up and Teflon mold versions while the same 0.5 mm ( 0. 0 2 in ) diameter carbon fiber rod was inserted onto the plastic frame. All other battens were a single laye r of carbon fiber for wing support. Any thicker battens inside the wing would have generally made the wing heavier without adding structure stiffness. A B Figure 4 3 Dimensions layout, and various examples of the processes for the two design confirmation For brevity, the transfer tape technique was excluded due to looking exactly like the plastic frame. A) Dimensions for Design 1 along with a look at examples of the three manufacturing systems B) Measurements for Design 2 accompanied by a similar view o f the three methods. Photos courtesy of Jason Rue. Hand Lay up Process The hand lay up process taken from past graduate studies at the University of Florida is pictured in Figure 4 4, where the radial design was printed glued to a flat plate, and coated with release film. This marks the beginning step within the process and continues to include cutting the carbon fiber, placing it on a flat plate, and vacuum bagging the end product

PAGE 59

59 A B C D E F Figure 4 4 A look at the flat plate and hand lay up procedure. A) A flat plate with Design 2 printed on paper and glued to the surface. B) The pattern with release film on top. C) Hand cut prepreg carbon/epoxy strips. D) Parallel batten wings laid up before curing cycle. E) Flat plate placed into a vacuum bag. F) Radial batten wings post cure before they are removed from the flat plate. Photos courtesy of Jason Rue.

PAGE 60

60 Next, the carbon fiber was cut by hand with a single blade laid out with great detail, and cured under pressure in a vacuum bag during a heating cycle. Once the oven cycle completed, the frames were carefully removed from the aluminum flat plate (Figure 4 5) and glued to CAPRAN The problem of unreliable strip width remains and serves as a reinforcement of the need for superior methods. Figure 4 5 A close up of Design 1 post cure. Photo courtesy of Jason Rue. To stay constant with this execution the outline of the wing was cu t with scissors however no guide kept the area absolute. Teflon Mold Method Two Teflon mold s w ere carefully crafted for this study and include d dimensioning from Figure 4 3. The carbon fiber strips were now carved out with the custom multi bladed strip cutter. Figure 4 6 displays a picture of the Teflon mold procedure that includes the mold, the carbon fiber once they are meticulously placed in position the clamped sandwich structure the post cured wings, and one example where that was remove d and prepared for the CAPRAN

PAGE 61

61 A B C D E F Figure 4 6 A glance at the Teflon mold process. A) Design 1 mold. B) The pre cured Design 1 Teflon mold. C) Mold with the silicone mat and release film placed on top. D) Two molds clamped and ready for the oven. E) View of the Teflon mold wings after the curing process (Design 2). F) Design 1 post cure out of the mold. Photos courtesy of Jason Rue. Even wit h the author having practice in making hundreds of wings with the hand lay up method and a keen eye, the mold delivers a more repeatable product every time.

PAGE 62

62 Plastic Frame Delrin plastic frames were milled with precision at a constant thickness. After the f rames were received, the commercial rod was added to the LE with rubber toughened CA. That combination was then put on CAPRAN that had been printed on previously with outlines. Figure 4 7 depicts the milled frame and final product against a red background so that details of the white plastic can be seen easily. A B Figure 4 7 Plastic frame with the LE carbon fiber rod attached. A) White CNC milled Delrin plastic frame. B) Final wing with the carbon fiber rod glued to the LE and the frame to CAPRAN Photos courtesy of Jason Rue. Transfer Tape Technique The time saving technique solved a problem that had plagued some wings throughout testing. During the flapping experiments with wings where the transfer tape was utilized, not a single wing developed de lamination. This point all but neglected human interaction during the trial testing. The tape thickness also was far more consistent than the layers of thin CA applied to the frame. Figure 4 8 lays out the steps taken to adhere the 3M transfer tape to the plastic frame. A few supplies were examined as potential materials for the step in Figure 4 8E including vacuum bag plastic and paper. Paper was chosen for its cost, reliability, and ease to use. Figure 4 8F portrays the wing once the backing was removed. Once the frame is carefully peeled off, the plastic remains with a consistent layer of adhesive and can be pressed down onto the CAPRAN

PAGE 63

63 A B C D E F G Figure 4 8. Steps involved with applying the transfer tape. A) The same plastic CNC milled frame used in the previous step. B) The plastic frame with a carbon fiber rod attached with rubber toughened CA. C) An example of the radial design. D) The frame is placed on to the one inch wide transfer tape. E) A piece of white computer paper sandwiches the frame with the tape backing. F) The backing is removed to display the frame with a complete layer of transfer tape attached. G) Custom jig designed to apply high pressure to the frame/CAPRAN and to remove air gaps between the two. Photos courtesy of Jason Rue. Results After all the manufacturing processes, eighty pairs of wings were ready for testing. Individual wings were weighted on a Gemini 20 Portable Milligram Scale and then mounted on the flapping mechanism, going through ten consecutive trials. Each

PAGE 64

64 trial consisted of the flapping frequencies of 20, 25, and 30 Hz. The average thrust production in each of the ten trials was then combined to give a single measured val ue linked to that particular wing pair. Below explains the comparisons between the eight ( two designs, four manufacturing methods) categories in areas of interest ( w eight and thrust p roduction ) The trends show an ideal situation for both variables as they decrease in variation and validate the work accomplished to update manufacturing. A few parameters were graphed in the evaluation including histograms of the actual values and the coefficient of variation, which is defined as followed. H istogram bins were arbitrary chosen to split the minimum an d maximum in to ten equal groups while the C V was sou ght after to normalize the data and compensate for a changing mean. As seen in the comparisons the designs finished with similar trends although very different paths. This leads to the conclusion that many of the measurements could be design specific. Wei ght Comparison Eighty wing pairs completed for this test were broken into the respective groups, either by design or process. For the weight assessment the 16 0 individual wings were weighed and tabulated. For each of the eight groups, the coefficient of v ariation was calculated (Figure 4 9 ). As the production methods developed, the values dropped and added confidence that the more refined wings were more consistent. First off, the radial batten design C V dropped by ~20% or more for each process, for a tota l of 60 .7%. The parallel batten wings had a similar reduction, but totaled 88 %

PAGE 65

65 Figure 4 9 For the ten pairs (20 total) of wings in each category, this illustrates the CV of the weight in grams. Each method improves this measure, making the duplicate wings much more consistent. As a look at the actual weight distribution eight combined histograms are lo cated below with the same horizontal axis. The weight bins range from 0.126 g to 0.199 g. Figure 4 10 encompassing the four radial batten types, shows how the y became more reliable and less scattered while Figure 4 11 clarifies the four parallel batten types These reflect how the weight gathered around an average and how t he scatter was reduced with each additional approach The hand lay up method had a much more spread out data set, especially with the parallel batten wings. This endorses the work to better the initial manufacturing procedure and states that this method is less desirable for optimization. The Teflon mold reduced the number of bins associated to the distribution as expected with consistent wings. Although somewhat repeatable, characterized by these variables, there still exists a necessity for the plastic m illed frames. For both designs, the plastic frames, attached by CA or transfer tape, were in reduced clusters and a closer averaged value. 0% 2% 4% 6% 8% 10% 12% 14% 16% Hand Lay-up Teflon Mold Plastic Frame Transfer Tape CV, Weight [g] Manufacturing Process Radial Batten Parallel Batten

PAGE 66

66 Figure 4 10 Histogram showing a drop in weight variation due to the enhanced processes for the radial batten wings Figure 4 1 1 Histogram showing a drop in weight variation due to the enhanced processes for the parallel batten wings. 0 2 4 6 8 10 12 Number of Wings Weight [g] Radial Hand Lay-up Teflon Mold Plastic Frame Transfer Tape 0 2 4 6 8 10 12 14 Number of Wings Weight [g] Parallel Hand Lay-up Teflon Mold Plastic Frame Transfer Tape

PAGE 67

67 The average weight, in grams, for the eight types of wings is in Table 4 1. The plastic frame wings were found to be roughly 4 % lighter for the radial batten style than the carbon fiber frame alternatives although increased again when the transfer tape was used Here, the reduction in C V was enough to offset the slight gain in weight for the fourth procedure. For the parallel bat ten wings, a n improvement of a pproximately 7% and 14% was found from the Teflon mold and hand lay up methods respectfully. In a similar trend as the radial batten wings, the weight increased marginally with the transfer tape. Table 4 1. Average weight of the 16 0 individual wings. Manufacturing Process Radial Batten Parallel Batten Hand Lay up 0.15 5 0.153 Teflon Mold 0.158 0.142 Plastic Frame 0.150 0.132 Transfer Tape 0.157 0.135 With continuous reductions in weight, the wings created by the multiple manufacturing methods become more efficient, taken that similar thrust values are measured The large changes in standard deviation demonstrate d the need for the new approaches. Averag e Thrust Comparison This section expands upon the main reason behind this paper. The experiments of p rior UF graduate students using carbon fiber and a hand lay up have been shown to be problematic, especially with dependability. Average thrust production was measured for each pair of wings, and, much like the weight, the C V was examine d as well. The thrust data in this section was measured at 30 Hz. Throughout many of the studies, the 30 Hz mark was focused on to fit within the c onstraints of the flapping mechanism and to produce a level of thrust that could counteract the weight of a standalone hovering

PAGE 68

68 device The duplicate wing C V values are obtained in Figure 4 1 2 where a 43% 56% and 24% reduction occurred for the radial ba tten manufacturing progress es The parallel batten wings saw a 46% and 76% dip for each of the initial two step s The transfer tape process added an unfavorable 9% at that step from 2.52% to 2.74%. Since the testing was done with a finite population of ten wings, the loss was taken as negligible and the values treated as equivalent. Figure 4 1 2 For the ten wings in each category, this illustrates the C V of thrust in grams. For each batten structure, the upgraded manufacturing technique lowers this statistic, allowing for more reliability. Altogether, the radial batten wings saw a decline in C V of 81 .1% from the hand lay up to the plastic frames adhered with transfer tape In being even more pronounced, the parallel batten ones noticed a staggering 8 6 % drop. Small variations from wing to wing could not be regularly differentiated with the hand lay up process. If the plastic frames are considered in the future, significantly more assurance would be held that, based on thrust production, one wing design was superior or inferior over another. 0% 5% 10% 15% 20% 25% Hand Lay-up Teflon Mold Plastic Frame Transfer Tape CV, Thrust [g] Manufacturing Process Radial Batten Parallel Batten

PAGE 69

69 An effort was made to try and quantify the uncertainty throughout the process based on thrust production from flapping wings while being in hover mode A concurrent study of 6 7 designs with the plastic wing frames wa s conducted to determine how the uncertainty would split between the manufacturing variability and testing errors. Uncertainty brought by manufacturing was mostly due to machine tolerance and human error while the testing uncertainty was present from senso r and testing conditions. Multiple wings with geometrically nominal designs were compared to extract manufacturing uncertainty while the testing uncertainty was quantified by looking at repeated trials. By using surrogates and an optimization algorithm cal led Efficient Global Optimization, Anirban Chaudhuri, graduate student of Dr. Raphael Haftka, was able to use 33 designs for a final analysis. His work was described in [47] and concluded a conservative testing uncertainty of 3.55% along with a conservat ive manufacturing uncertainty of 3.15%. Since ten trials of a wing are customary, the testing uncertainty is reduced to 1.12%. The total uncertainty was calculated to be 3.34%, a number acceptable to start any further optimizati on as it was lower than an initial 5% goal This value is slightly higher than the 2 4% or 2.7 % given in the last part of Figure 4 1 2 due to having a limited number of samples for the calculation and being conservative in nature Even though there is a slight discrepancy, there is a strong belief that the testing uncertainty is much lower than the manufacturing error. Through the optimization, Chaudhuri was able to increase the maximum thrust output within the study by 7% to 11.9 g at 30 Hz. This value exceeds the error and can be trusted as a true improvement.

PAGE 70

70 Aside from the differences in C V the scatter of actual thrust values is described within the eight gathered histograms below ( four radial and four parallel) The thrust bins extend from 3.85 g to 8.01 g and was gathered at 30 Hz Figure 4 1 3 Histogram showing the drop in thrust variation due to the enhanced processes. The increase in thrust for the plastic frames is quite visible here as the plastic frame/transfer tape bars are dramatically shifted to the right. Even though two designs were investigated, the aforementioned range is considerably larger than a tolerable breadth. Much like the case of weight, the thrust values see a substantial range for the hand lay up pr ocedure. This does not lend itself well to optimization where an average thrust value must be trusted. The Teflon mold helped, again, in the two design cases by contracting and centralizing the values. The red (plastic frame) and purple (transfer tape) color outlines an even further subtraction, being in only 2 3 bins. Ultimately, these amounts should be as close as possible to instill faith in the thrust production. For all four categories, the radial batten wings seem to be confined to fewer bins. It may be the case that the topology is suited 0 1 2 3 4 5 6 7 Number of Wings Average Thrust [g] Radial Hand Lay-up Teflon Mold Plastic Frame Transfer Tape

PAGE 71

71 to flap in this range and is more stable than the parallel batten wings. Since the parallel batten wings embodied a parallel trend in the weight description, a possibility persists that the frame geometry could b e correlated to the performance. Figure 4 1 4 For the parallel batten wings, the thrust production for the hand lay up wings is rather scattered and pulls together with the Teflon mold. The plastic wings decrease in the number of bins and shift to the r ight. For a view of the larger picture, Figure 4 1 5 and Figure 4 1 6 break up the data into the two design sets. The graphs give all eighty wings and the results of the thrust measurements for frequencies of 20, 25, and 30 Hz Based on the four colors and the stages shown in A through D the thrust advantage and diminished variation s are clear Each step distinctly illustrates a reduction in scatter of the ten wings given on each graph. Visually, the comparison of the hand lay up versus the transfer ta pe is undeniably profound. These figures create the purpose of this experiment by themselves and speak the importance of manufacturing procedures. Stage E in both figures combine s the averages from each into an overall assessment where it can be 0 1 2 3 4 5 6 7 8 9 Number of Wings Average Thrust [g] Parallel Hand Lay-up Teflon Mold Plastic Frame Transfer Tape

PAGE 72

72 seen clear ly that the plastic frame and transfer tape wings produce higher thrust than the previous two. A B C D E Figure 4 15. Graph of the 40 Radial Batten wings presenting how the variation decreases with each new method. A) Hand Lay up. B) Teflon Mold C) Plastic Frame. D) Transfer Tape. E) A single averaged line for each method plotted together.

PAGE 73

73 A B C D E Figure 4 16. Graph of the 40 Parallel Batten wings presenting how the variation shrunk with each new method. A) Hand Lay up. B) Teflon Mo ld. C) Plastic Frame. D) Transfer Tape. E) A single averaged line for each method plotted together. In terms of thrust force, the plastic frames produced more than the carbon fiber counterparts without question. On average, t he radial batten wings gaining roughly 3 5 % while the parallel batten achieved 3 3 %. Exact values are found in Table 4 2. A r eason

PAGE 74

74 for the increase could be simply th at the LE beca m e stiffer with the use of the rod This is understandable although is a significant achievement since the ov erall weight lessened as well for the parallel batten Although this was not an original goal, the successful outcome bred a stronger and more efficient wing. Table 4 2. Average thrust values at 30 Hz, given in grams, for each of the eight wing groups. Man ufacturing Process Radial Batten Parallel Batten Hand Lay up 5.77 5.73 Teflon Mold 5.49 5.62 Plastic Frame 7.52 7. 37 Transfer Tape 7.68 7.69 A further look into the individual trial CV calculations (different from above) for each of the eighty pairs of wings for 20 Hz, 25 Hz, and 30 Hz proved interesting. These eighty values were formed from the ten tria ls that each pair was comprised and grouped by manufacturing. For simplicity, the data was averaged in each group and ca n be seen in Table 4 3. Generally, the CV would decrease as the frequency increase d and for each additional manufacturing advance. For the radial design at 30 Hz, the averages held fairly constant between 1.9% and 2.4%. Similarly, the parallel design was a round 2.3% for the first two and then dropped to roughly 1.5%. This could be due to the geometry topology differences Table 4 3. Average trial thrust CV measurement for all eighty pairs. Manufacturing Process 20 Hz Ave 25 Hz Ave 30 Hz Ave Radial Hand Lay up 6.020% 3.047% 1.945% Teflon Mold 4.762% 3.347% 2.490% Plastic Frame 2.793% 2.146% 2.420% Transfer Tape 1.929% 1.677% 1.960% Parallel Hand Lay up 7.446% 5.862% 2.332% Teflon Mold 6.013% 4.338% 2.351% Plastic Frame 3.886% 2.484% 1.675% Transfer Tape 2.959% 1.867% 1.381%

PAGE 75

75 To take a look at the standard deviation data and make sure that the CV values were not skewed due to higher means in the later processes, Table 4 4 shows a decrease for each manufacturing technique. The slight increase for the parallel batten wings put together with transfer tape is rather miniscul e and, overall, the procedures acted the way they were intended. Table 4 4. Standard deviation data for the thrust measurements at 30 Hz. Manufacturing Process Radial Batten Parallel Batten Hand Lay up 0.730 1.113 Teflon Mold 0.395 0.591 Plastic Frame 0.237 0.186 Transfer Tape 0.184 0.211 Conclusions Over the life of this entire study, the central objective was to acknowledge existing problems with manufacturing or experiment techniques and to provide viable solutions. For additional work and optimization to proceed, stronger confidence was needed in the average thrust production produced by a certain wing design. A check for repeatability was developed, includin g testing duplicate wings and focusing on how the quantities differed. A n element discovered was the fact that several of the quantities seemed to be design specific. This could speak to the precision of the mill, the accumulated errors from data collectio n, human inaccuracies when materials are glued, or possibly the idea that the geometric configuration does not just play a role in overall stiffness but in performance. The topology of the frame could control how the wing billows under aerodynamic forces, affect how quickly the wing returns to a neutral plane a fter deformation, and the chord wise twist an outcome which McIntire found to adjust the thrust production.

PAGE 76

76 After completion, the milled p lastic frames proved to deliver less scatter than the other op tions in both categories of weight and average thrust output. These also contributed a possibility for wings with lower weight, higher thrust, and considerably more reliability. The scatter in the weight and average thrust were convincingly bolstered, adding to the validity and necessity for such a study. The new techniques have brought about an appropriate conversation as to whether the construction of flapping wings, and albeit all small to medium sized carbon fiber hand lay ups, are highly repeatable without extreme attention to detail. The information delivered by this paper offers a discovery to help scholars conceive of innovative manufacturing and hopefully will lead to a better route to the understanding of how different features of a wing link t o flight.

PAGE 77

77 CHAPTER 5 REVIEW AND FUTURE WORK As f lapping wing technology expands and the advantages are further defined from other propulsion options, more work will be encouraged by the scientific community. In hopes that this paper wil l generate enthusiasm and build upon the push to understand the physics behind flapping flight, a few major conclusions were established. The main goal of this paper was to produce a high fidelity procedure so that synthetic wings could be tested and minuscule variations in thrust production could be decidedly measured as a gain or a loss over another wing. Great strides were taken toward this goal incorporating a flapping mechanism and a manufacturing techn ique Data collection has been enhanced locally within the lab by coding LabVIEW and MATLAB correctly to heighten efficiency and productivity. This became pivotal when mass testing was required. Monitoring the w eight of 1 6 0 artificial wings built by separate means demonstrated the erraticism in hand constructed frames. The plastic manufactu ring stage s accrued nearly 75 % less scatter and an average weight savings of 5 %. Extensive t hrust comparisons promoted the importance of this paper when attempting to compare wings. The C V in thrust (at 30 Hz) for the two designs was reduced by over 80 % by adjusting how the wings were prepared. This method can be adopted by other researchers to help contribute to more reliability and credit to final data. Variability in typical hand lay up methods have been proven larger and more uncontrollable than originally thought This paper should a t least bring attention to the fact that inconsistencies exist in manufacturing techniques use d for flapping wing experiments and provide a viable solution. Future work entails a continuation to learn the physics behind flapping flight and how certain characteristics can be related to thrust production. Since small flapping wing analysis can be cumbersome and inefficient if lengthy trials of flapping are inevitable before any data is scrutinized, charts or a series of curves could be

PAGE 78

78 formulated from experiments to associate relationships and predict thrust production. This particular point is well conceived in Figure 4 2, as for just two designs and four manufacturing techniques 800 trials were needed. These relationships should be comparing thrust production with certain non flapping tests. As an example, i f a relationship of predetermined curves that correlate wing properties, such as structural stiffness or specific mode shapes, to the average thrust output is organized, the process becomes much simpler for evaluation and leads to an effective intuition. With this method, implementing a simple test could then yield an average thrus t value without having to run a more intensive experiment. Also, if a certain thrust should be necessary, the curves would allow for one to realize wing properties before manufacturing occurs. Given the full range of deformation and frequencies that the wi ngs are exposed to, the natural frequencies and mode shapes could be examined to recognize how the with a Laser Doppler Vibrometer (LDV) to take non contact measurements The Polytec scanning vibrometer includes a PSV I 400 scanning head, a PSV 400 junction box, and a OFV 500 controller module The wings attached via screw s to a Low Dynamic Stiffness (LDS) V201 permanent magnet shaker that sweeps from 1 to 500 Hz. Software an alyzes the wing and displays a f ast Fourier transform (FFT) with peaks at natural frequencies and an animation of the shapes. Knowing the link of these to thrust production may contribute to how wings are engineered in the future. Considering there exists a relation ship associating stiffness to thrust production s tatic DIC measurements have started to try and relate deflection to average thrust production. Here, the wings are speckled with black enamel paint and Kevlar string at

PAGE 79

79 the tips (Figure 5 1 ) After a load is applied, two Point Grey Research Flea2 cameras VIC 3D to calculate the deflection. Figure 5 1A shows a picture of a wing with the glued on Kevlar strings Currently, six different measurements are taken including three from the LE string and three from the root string. The three data collections differed by adding various weights (0.3 g, 0.5 g, and 1 g) to the string and photographing the deflection like w hat can be seen in Figure 5 1B. One ongoing study called to optimize wings in a three dimensional space, having three variables containing: aspect ratio (AR) an angle (taken from the LE) for the single batten (root batten always there) and percentage tha t the carbon fiber rod extends down the LE. These variables were chosen exercising the intuition from prior experiments because they correlated strongly to thrust production. Figure 5 2 places the six ARs that were tested (only six for simplicity) plus exa mples of the other two variables. A B Figure 5 1. Static deflection DIC. A) Mounted wing with weights hanging from Kevlar string. B) Camera set up for DIC. Photos courtesy of Jason Rue.

PAGE 80

80 Figure 5 2. Three dimensional space optimization Seeding a three dimensional volume with near equally spaced designs, wings were flapped and recorded. Figure 5 3 graphs the elongation percentage of the rod along the LE versus the d eflection of the wing tip. Figure 5 3 A bubble plot present ing static deflecti on results with a 0.5 gram load obtained for different carbon rod percentages that occupy the plastic LE The size of the bubbles specifies the amount of thrust obtained for the wings. The general exponential curve follows expectation, indicating little de flection with wings that have a rod close to 100% of the LE. An interesting behavior of these bubbles

PAGE 81

81 is how a few have drifted to the left. This possibly suggests that the single batten has reinforced the LE or the smaller ARs behave stiffer. A real benef it may come from this research trying to correlate certain variables to thrust. Then the space can be expanded to larger ARs and more batten arrangements. A Nano17 Titanium sensor was newly purchased to help increase resolution of the readings although d ue to low overload on the sensor, a new flapping mechanism would need to be modeled and designed in such a way that the sensor is shielded from strong torques or forces. This instrument could allow thrust data to be refined even tighter and for smaller variations to be read. With the idea of replacing current parts, fabricating a new flapping mechanism could easily enlarge the testing envelope to scrutinize other factors. Besides this change, a new controller and motor combination might permit an increase in flapping symmetry or tolerate heavier wings. Even adjusting the acceleration rates and the PID constants on the current system could conce de better results. Other future experiments could take place to redo inconclusive tests done in the past with insufficient manufacturing approaches Sweep Angle Variations Further Recording Time Tests Loose CAPRAN Snap Through Sensor Interaction and how it plays a role Wing Mount Stiffness Motor Controller and Acceleration Long Term Wing Decay Varying diameter of carbon fiber rod Effectively taper the LE carbon fiber rod These tests may not lead to any significant knowledge advanceme nts or they could be a breakthrough.

PAGE 82

82 One question remains from the research done thus far about if there is a driving feature in flapping wing thrust product ion or if a multitude of variables have an effect. Presently, combined overall structural stiffness and geometric topology are two main paths that could be explored. kinematics at elevated frequencies which brought insight to perplexities. This technology, should it be found useful could help drastically in learning about unanticipated occurrences. Work on new fabrication methods would be beneficial, for the improvement reached in this paper allowed more tests and new possibilities to prosper. Now that wings can be relied upon, sci ence can proceed to fulfill the basic understanding behind flapping flight.

PAGE 83

83 LIST OF REFERENCES [1] P. Wu, P. Ifju and B. Stanford, "Flapping Wing Structural Deformation and Thrust Correlation Study with Flexible Membrane Wings," AIAA Journal, vol. 48, no. 9, pp. 2111 2122, September 2010. [2] J. Lee, J. Shin and S. Lee, "Fluid structure interaction o f a flapping flexible plate in quiescent fluid," Computers & Fluids, vol. 57, pp. 124 137, 2012. [3] S. Ansari, K. Knowles and R. Zbikowski, "Insectlike Flapping Wings in the Hover Part 2: Effect of Wing Geometry," JOURNAL OF AIRCRAFT, vol. 45, no. 6, p p. 1976 1990, 2008. [4] L. Zhao, Q. Huang, X. Deng and S. Sane, "Aerodynamic effects of flexibility in flapping wings," Journal of The Royal Society Interface, vol. 7, pp. 485 497, 2010. [5] K. Mazaheri and A. Ebrahimi, "Experimental investigation of the effect of chordwise flexibility on the aerodynamics of flapping wings in hovering flight," Journal of Fluids and Structures, vol. 26, pp. 544 558, 2010. [6] Q. Nguyen, Q. Truong, H. Park, N. Goo and D. Byun, "Measurement of Force Produced by an Ins ect Mimicking Flapping Wing System," Journal of Bionic Engineering [via SciencDirect.com], vol. 7, pp. S94 S102, 2010. [7] W. Shyy, H. Aono, S. Chimakurthi, P. Trizila, C. Kang, C. Cesnik and H. Liu, "Recent progress in flapping wing aerodynamics and ae roelasticity," Progress in Aerospace Sciences, vol. 46, pp. 284 327, 2010. [8] D. Lentink and M. Dickinson, "Rotational accelerations stabilize leading edge vortices on revolving fly wings," The Journal of Experimental Biology, vol. 212, pp. 2705 2719, 2009. [9] D. Lentink, F. Muijres, F. Donker Duyvis and J. van Leeuwen, "Vortex wake interactions of a flapping foil that models animal swimming and flight," The Journal of Experimental Biology, vol. 211, pp. 267 273, 2008. [10] P. Bhayu, Q. Nguyen, H. Park, N. Goo and D. Byun, "Artificial Cambered Wing for a Beetle Mimicking Flapper," Journal of Bionic Engineering [via ScienceDirect.com], vol. 7, pp. S130 S136, 2010. [11] Q. Nguyen, H. Park, N. Goo and D. Byun, "Characteristic and a Flapping Wing System that Mimics Beetle Flight," Journal of Bionic Engineering [via ScienceDirect.com], vol. 7, no. 1, pp. 77 86, 2010.

PAGE 84

84 [12] T. Jin, N. Goo, S. C. Woo and H. Park, "Use of a Digital Image Correlation Technique for Measuring the Material Properties of Beetle Wing," Journal of Bionic Engineering, vol. 6, pp. 224 231, 2009. [13] C. Ellington, "THE NOVEL AERODYNAMICS OF INSECT FL IGHT: APPLICATIONS TO MICRO AIR VEHICLES," Journal of Experimental Biology, vol. 202, pp. 3439 3448, 1999. [14] X. Mou, Y. Liu and M. Sun, "Wing motion measurement and aerodynamics of hovering true hoverflies," The Journal of Experimental Biology, vol. 214, pp. 2832 2844, 2011. [15] G. Spedding, A. Hedenstrom, J. McArthur and M. Rosen, "The implications of low speed fixed wing aerofoil measurements on the analysis and performance of flapping bird wings," The Journal of Experimental Biology, vol. 211, pp. 215 223, 2008. [16] S. Combes and T. Daniel, "Flexural stiffness in insect wings," The Journal of Experimental Biology, vol. 206, pp. 2979 2987, 2003. [17] W. Shyy, M. Berg and D. Ljungqvist, "Flapping and flexible wings for biological and micro air vehicles," Progress in Aerospace Sciences, vol. 35, pp. 455 505, 1999. [18] A. Agrawal and S. Agrawal, "Design of Bio inspired Flexible Wings for Flapping Wing Micro sized Air Vehicle Applications," Advanced Robotics, vol. 23, pp. 979 1002, 2009. [19] R. O'Hara and A. Palazotto, "The morphological characterization of the forewing of the Manduca sexta species for the application of biomimetic flapping wing mic ro air vehicles," Bioinspiration & Biomimetics, vol. 7, no. 4, 2012. [20] M. Keennon, K. Klingebiel, H. Won and A. Andriukov, "Development of the Nano Hummingbird: A Tailless Flapping Wing Micro Air Vehicle," in 50th AIAA Aerospace Sciences Meeting incl uding the New Horizons Forum and Aerospace Exposition Nashville, TN, 2012. [21] D. Warrick, T. Hedrick, M. Fernandez, B. Tobalske and A. Biewener, "Hummingbird flight," Current Biology, vol. 22, no. 12, pp. R472 R477, 2012. [22] T. Hedrick, B. Tobal ske, I. Ros, D. Warrick and A. Biewener, "Morphological and kinematic basis of the hummingbird flight stroke: scaling of flight muscle transmission ratio," Proceedings of The Royal Society, vol. 279, no. 1735, pp. 1986 1992, 22 May 2012.

PAGE 85

85 [23] D. Altshuler, K. Welch, B. Cho, D. Welch, A. Lin, W. Dickson and M. Dickinson, anna)," The Journal of Experimental Biology, vol. 213, pp. 2507 2514, 2010. [24] J. Mason, A. Je nnings, J. Black, A. Sharp, J. Blandino and J. Lysher, "Validation of a Finite Element Analysis of a Flapping Wing Against Inertial and Aeroelastic Responses," in 54th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference Boston Massachusetts USA, April 2013. [25] H. Hu, A. Kumar, G. Abate and R. Albertani, "An experimental investigation on the aerodynamic performances of flexible membrane wings in flapping flight," Aerospace Science and Technology, vol. 14, pp. 575 586, 2010. [26] E. Swanton, B. Vanier and K. Mohseni, "Flow visualization and wall shear stress of a flapping model hummingbird wing," Experiments in Fluids, vol. 49, no. 3, pp. 657 671, 2010. [27] C. S. Lin, C. Hwu and W. B. Young, membrane wings with simple flapping motion," Aerospace Science and Technology, vol. 10, pp. 111 119, 2006. [28] A. Gogulapati, P. Friedmann, E. Kheng and W. Shyy, "Approximate Aeroelastic Modeling of Flapping Win gs in Hover," AIAA Journal, vol. 51, no. 3, pp. 567 583, March 2013. [29] G. de Croon, K. de Clercq, R. Ruijsink, B. Remes and C. de Wagter, "Design, aerodynamics, and vision based control of the DelFly," International Journal of Micro Air Vehicles, vol 1, no. 2, pp. 71 97, 2009. [30] B. Stanford, P. Ifju, R. Albertani and W. Shyy, "Fixed membrane wings for micro air vehicles: Experimental characterization, numerical modeling, and tailoring," Progress in Aerospace Sciences, vol. 44, pp. 258 294, 2008 [31] L. Xie, P. Wu and P. Ifju, "Advanced Biologically Inspired Flapping Wing Structure Development," in Proceedings of the SEM Annual Conference Indianapolis, Indiana USA, June 2010. [32] P. Wu, B. Stanford and P. Ifju, "Insect Inspired Flapping Wing Kinematics Measurements with Digital Image Correlation," in Proceedings of the SEM Annual Conference Albuquerque New Mexico USA, June 2009. [33] P. Wu, "Experimental Characterization, Desig n, Analysis and Optimization of Flexible Flapping Wings for Micro Air Vehicles," University of Florida Disseration, Gainesville, 2010.

PAGE 86

86 [34] J. McIntire, "Investigating Torsional Compliance of Flapping Wings to Maximize Thrust Capability," University of F lorida Thesis, Gainesville, 2011. [35] R. Wood, "Design, fabrication, and analysis of a 3DOF, 3cm flapping wing MAV," in IEEE/RSJ IROS San Diego, CA, Oct 2007. [36] K. Ma, S. Felton and R. Wood, "Design, Fabrication, and Modeling of the Split Actuato r Microrobotic Bee," in 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems Vilamoura, Algarve, Portugal, 2012. [37] WowWee, "DRAGONFLY," 2013. [Online]. Available: http://www.wowwee.com/en/support/flytech dragonfly. [Accessed 24 M arch 2013]. [38] P. Ifju, T. Schmitz and R. Haftka, "Expanding the Design Space of Synthetic Flexible Flapping Wings by Advancing Fabrication and Optimization Methodologies," University of Florida AFOSR/RSA Proposal. [39] J. Anderson, Jr., Fundamentals of Aerodynamics, Fourth Edition ed., New York: McGraw Hill, 2007. [40] ATI Industrial Automation, "F/T Sensor: Nano17," 2013. [Online]. Available: http://www.ati ia.com/products/ft/ft_models.aspx?id=Nano17. [Accessed 23 Ma rch 2013]. [41] K. Chang and J. Rue, "Analysis of Thrust Production in Small Synthetic Flapping Wings," in 2013 Annual Conference on Experimental and Applied Mechanics Lombard, IL, 2013. [42] Honeywell International Inc, "CAPRAN 1200 Matte," 16 2 201 0. [Online]. Available: http://www51.honeywell.com/sm/capran/common/documents/PP_Capran_1200_m atte_Specification_sheet.pdf. [Accessed 26 2 2013]. [43] DuPont Teijin Films, "Mylar Polyester Film Product Information," June 2003. [Online]. Available: http:/ /usa.dupontteijinfilms.com/informationcenter/downloads/Physical_And_Therm al_Properties.pdf. [Accessed 6 May 2013]. [44] McMaster Carr, "More About Plastics," 2012. [Online]. Available: http://www.mcmaster.com/#8574kac/=lnff8t. [Accessed 26 2 2013]. [45] Aviasport, "Avia Composites," 2011. [Online]. Available: http://aviasport.net/composites/. [46] 3M, "Adhesive Transfer Tapes with Adhesive 300LSE," September 2002.

PAGE 87

87 [47] A. Chaudhuri, R. Haftka, P. Ifju, D. Villanueva, K. Chang, J. Rue, C. Tyler and T. Schmitz, "Experimental Optimization and Uncertainty Quantification of Flapping Wing of a Micro Air Vehicle," in 10th World Congress on Structural and Multidisciplinary Optim ization Orlando, FL, 2013.

PAGE 88

88 BIOGRAPHICAL SKETCH Jason was born t o a wonderful family in south Florida. From early on, Jason has strived to continuously apply his abilities and tries to build upon an ever expanding skillset. From running a small, hobby driven, business as a teen to attending a world array of knowledge that will allow for contributions to highly motivated ethical organizations in the future Jason started as a teen and an owner of a business, creating wooden pens and pencils from exotic woods. The opportunity taught crucial skills about marketing, accounting, and learning from an early age to keep business records. The pens were sold from California to New York and he was even highlighted by the Penn State He continued his diversification as a teen by partaking in multiple varsity sports, playing both the alto and tenor saxo phone, having involvement with numerous community service opportunities, working as a chef in a local caf, and keeping a goal of higher education. Continuing past high school, Jason finished his next step at the University of Florida in 2011 with a Bachel or of Science in a erospace e ngineering. As a student, Jason was not only graduated Cum Laude, but kept busy with extracurricular activities He worked as an undergraduate in the fluid d ynamics and w ind t unnel l abs as an ass istant and held a lead position in the annual UAV competitions put on by Cessna, Raytheon, and the AIAA student group. In the competitions, Jason help lead UF to its best finish by maintaining responsibili ti es such

PAGE 89

89 as keeping group d eadlines, finding solut ions to challenging p roblems, and strong leadership. Jason remained persistent through re ceiving a Master of Science in aerospace e ngineering in 2013. While accomplishing this feat, he was a teaching assistant for several classes and a research assistant working to further flapping wing technology. Along with using his knowledge and critical thinking skills to further the research objectives set forth within the department, Jason also presented at the 2012 SEM XII International Co nference the 2012 SEM Southeast Graduate Student Symposium, 2013 SEM Symposium, and passed the Fundamentals of Engineering Exam. In terms of other concentrated areas of interest, Jason would like to continue his education throughout his career experimenti ng with flight characteristics and analysis of supersonic aircraft, missiles, and other reconnaissance vehicles. His goal would be to elongate the duration of operation, look at stress strain relations, work on vehicle and product design or development, en hance payload capacities, increase maximum velocities during flight, and to create better maneuverability. Other learning includes specifics about shock analysis, steady and unsteady aerodynamics, finite element analysis, plastic stress strain nonlinear re lations, rotorcraft, and daily launch operations among other subjects of aerodyn amics and structural composure. After graduation, Jason plans to continue his career at Lockheed Martin as a Systems Engineer. He will work to capitalize on technology to incre ase efficiencies and take market share while addressing crucial issues and providing great products to customers.