Professional Documents
Culture Documents
for the degree of Masters of Applied Science Graduate Department of Aerospace Science and Engineering University of Toronto
Acquisitions et
seMeas bibliographiques
The author has grauteci a nonexclusive Licence aiiowing the National Liof Canada to reproduce, loan, distribute or seii copies of this thesis m microfom, papa or ekcronic fomats.
L'auteur a accord une licence non exclusive permettant la Bibliothque nationale du Canada de reproduire, prter, distriiuer ou vendre des copies de cette thse sous Ia forme de microfichelf;ilm, de reproduction sur papier ou sur format lectronique.
The author retains ownershfp of the copyright in t h i s thesis. Neither the thesis nor substantid extracts tiom it may be printed or othewise reproduced without the anthor's permission.
L ' a u t e u r conserve la proprit du droit d'aueur qai protge cette thse. Ni la thse ni des extraits substantiels de celle-ci ne doivent tre imprims ou autrement reproduits sans son autorisation.
Abstract
An experimental investigation into the stabiiii of a flappiug wing micro air vehicle was perfomied at the University of Toronto Institute for Aerospace Studies. A thtee-degree of W o m force balance was designed and constnicted to measure the forces and moments exhibiteci by a set of flapping wings through 180' of rotation at varied fiee-stream velocities. The same apparatus was a h used to test two tail configurations.
A two-dimensional simulation program was wtitten using MA'IZAB software to
ident* stable whicle configurations at or near the hoveiliig condition. A total of four case studies were performed, and each revealed the vehicIe had inherent stabiiity. The presence of a tail on the vehicle produced only margmal effects. Of crucial importance was the placement of the vehicle center of gravity with respect to the wings. A preferred
distance of 3.5 cm h m the c.g. to the leading edge ofthe whgs allowed for stable flight
under ali cases studied.
Acknowledgements
Many individuals need to be thanked for their time and guidance during the course of this research. W i u t them I do not beiieve that 1 could ever have completed
this research on my own, First, I would like te *hdc Mr. Dave Loewea W i i u t his input
and experience, the work completed herein could w t even have been started His patience in keeping me fiom h b l i n g about the workshop was much appreciated.
Dr. James DeLaurier also deserves much credit for adding his wealth of
knowledge and experience in supervishg my work. His office door always seemed to be
Finally, 1 wouid like to acknowledge my fbkh m God 1do not think the stniggks
that arose both inside and outside my studies over the past 18 months couid ever have been overcome without a steadfst devotion to Him.
Table of Contents
Absrnet
.. ................... ..........."....-... A~knowledgememts .................................................. ..........................................UI Table of Cmtens ... ....... ...................................... . .................................................N ... List of Figures............................................................................................................. List of T a b h ........................................................................................................ .......xi Chapter 1: Introduction ................................................................................................ 1 MAV Project at UTIAS ....-..... ................................ ......................i 1.1
H
**-
VIII
1.1.2 ResearchObjectives................................................................................... 3
1.1.3 Year 3 hject Metamorphosis ................................................................... 4 1.1.4 About the Vehicle Components Used ......................................................... 5
Chapter 2: Force BaIance Design
9 .................................................................................. Rationale for Sekcted Design ........ .......... ............................... ..................9 Design Sperifiertioas , . ., .. . ................... ....... . . 10
IO
2.1
2 2
12
13
14
..
23
BaianceCalibrati6n
....................
....................17
17
3.1
3.2
3.2.1 Initial Resuhs ........................................................................................... 24 322 Revised Design ........................................................................................ 24 3.2.3 Calibration Procedure .............................................................................. 26
Chapter 4: Ekperimenls
4.1
4.1.1 Methodology............................................................................................ 29
4.1.2 Taring ...................................................................................................... 34
4.2
...........................................................................39
...............................................................................43
5.1
Wings
.......................................................................................................... 43
..
5.1.1 Repeatabriity ............................................................................................ 43 5.1 2 Longitudinal (2-axis) Forces ....................................................................44 5.1.3 Lateral (X-mis) Forces ............................................................................ 48
5.1.4 Moments (about Y-axis) ........................................................................... 49
52
Taib
............................................................................................................ 51
53
Ampliticition of ah
52
5.3.1 Z Forces................................................................................................... 52
. . . . . . . . . . m . . . . .... . . . .................................. m 58
6.1
. . . . . . . . . . . . . . . . . . 6 . 1 . 2 Lookup Tables .
. . ........................................................... 6
Do
6 . 1 . 3 Numerical Procedure.............................................................................. 6 4
6.2
Initial Results
............................. . . . . . . . . . . . . . . . . . . .................................71
......................................................................72
75 77
6 . 4 . 1 T e s t Cases..........................................................................................
6.42 Case I - Hovering Condiiion with Tiltmg Disturbance .............................79 6 . 4 . 3 Case ii - Slight Ascent with T W g Disturbance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .82 6 . 4 . 4 Case i - SLight Descent with T i Disturbance ................................... 85
88 6 . 4 . 5 Case IV - Lateral Gust.............................................................................
Chapter 7: Conclmkns
7.1
8.1
8.2
Appendices:
List of Figures
Chapfer I Figures
Chapter 2 Figures
Figure 2.1. Side View of Force Balance Design.................................................... 1 1
Figure 2-82: Combined X and Z Forces (Top View) ............................................... 20 Figure 2.8.3. Combined X, 2 Forces with Moment (Top View) ............................... 2 1
Chripier 3 Figures
Figure 3 . 1 .1. Open End Wmd Tunnel at UTIAS ..................................................... 23 Figure 3 . 1 2:Open End Wmd Tunnel at UTlAS .....................................................23
Figure 3 . 2 . Sample VeIocity Field (with Cone)..................................................... 2 5
26 Figure 33: Pitot Tube and Manometer Setup...........................................................
Chapter 4 Figures
Figure 4.1 :Wing Testing Procedure ........................................................................ 30 Figure 4.2. Original Mounting Bracket .................................................................... 35 Figure 4.3. Original Mounting Bracket (Top View) ..............................................
36
Figure 4.4. "Gooseneckn Mounting Bracket ............................................................ 37 Figure 4.5. Foam Shroud and Mounting Bracket .................................................... 38 Figure 4.6. Exaggerated Mounting Misalignment ................................................... 39 Figure 4.7. Tail Designs .......................................................................................... 40 Figure 4.8. Tail Dimensions .................................................................................... 40 Figure 4.9. Tai1 Testing Mount ................................................................................ 41
Chapter 5 Figures
43 Figure 5.1 :Lateral (X-axis) Force vs.Angle, J = 0.735 ..........................-*.***............ Figure 5.2. Longitudinal (Z-axis) Discontinuity at 90"................ . . .......................44 Figure 5.3: Longitudinal (Z-axis) Force vs. Angle with Liaear
52
Figure 5.9. Thrust Ratio vs.Free-Stream - Frequency Ratio (Origina[ Data) ........... 54
Figure 5.10: T h -
Figure 5.12. Effectof Flapping Frequency on X Force ........................................... 57 Figure 5.13. Effect of FIapping Frequency on Y Moment ........................................ 58 Figure 5.14. X Force Cornparison to initiaiiy Assumed Values................................ 59
60
Chapter 6 Figures
Figure 6.1 : Mode1 Representation.........................................................................62 Figure 6.2. Disturbed Condition ............................................................................ 63
..
. . .70 Figure 6.6. Force and Moment Summation Example (Tail Only) ..........................
Figure 6.7. Initiai T e s t Case Wahout Tai1 ................................................................ 71 Figure 6.8.1 :Disc Dampuig Experimental Setup ..................................................... 73 Figure 6.8.2. Dise Damping Experimental Setup, Perturbeci Condition .................... 73 Figure 6.9. Disc Dampmg Apparanis....................................................................... 75 Figure 6.10. Example Plot of Osdiatory Decay ................................................ 76 Figure6.11.1.CaseI-NoTa
Il =75 cm......................................................... 79
Figure 6.1 1.2. Case 1-No Ta. EfEct in the Reduction of 11................................... 80 Figure 6.1 1.3. Case II -No Tail, EEct m the Reduction of 11 .................................83
Figure 6.1 1.6: Case iII -No Tail, Effect in the Reduction of 1, ..............................
86
Figure 6.1 1.7: Case III - With Tail, h = 12.5 cm and -12.5 cm, II = 7.5 cm .........,... 87
Figure 6.1 1.8: Case HI - With and Without Tail, i2 = 12.5 cm, II = 2 cm ................. 88
Figure 6.1 1-9: Case IV -No Ta& Effect in the Reduction of 1, ...................... ..... 89
List of Tables
Chupter 2
Table 2.1: K-Value Summary ....................................................................
*Note: Figures and fables in the Appendices are nos listed here. however the Appendix title and introductory poragraph should a h w the reader to determine w h t rypes of
Chapter 1: INTRODUCTION
1.1
MAVPmjectatUTlAS
current and irnpeodiag developments in rnicroelectronics t a h g place at the the. The intent of the project was to create a small airborne platform capable of perforrning
should not exceed 15 cm, and have a total vehicle mass between 30 and 50 gram. Such a vehicle would be expected to carry a variety of sensors, yet remai. portable and durable
a c k Hence, a priority was enough so that it could be easily transponed inside a sofdier's p
placed upon the devetopers to mate a tightweight, robust and efficient design that wodd satisfL the demands of the agency.
DAWA awarded several research contracts to various hitutions and k m
across the United States. iacluded amongst these was a contractai partnership between SRI International of Menlo Park, California and the University of Toronto Institute for Aerospace Studies (TiAS). Together, this team sought to evoive a vehicle design that would combine the technologies of fiapping-wing propulsion and artifid muscle actuation. This particular f o d a would stand apart h mother proposais in that it would
be directly aimed at producing a MAV capable of bvering ftight. F
i 1.1 depicts an
This marriage of expertise between SRI and LmAS began in M a y of 1998, with
the total contract duration encompassing 3 years, With its strong background, knowledge and expex-ience in hpping-wing flight, TiAS wouid focus on developiug a successtl design for wing propulsion as well as the vehicle aerodyaamics. Alternatively, SRI would direct its work toward perfectiog its technologies in Electrostrictive Polymer A r t i f i d Muscles (EPAMs) as the wing actuating mecbanism, in addition to incorporahg the vehicle's necessary electronics.
wing design tbat would provide sufficient thrust to cary a mass budget of approximately
50 grams. The second year continued with the wing research, addressing such areas as
flow visualization and developing numerical tools for analysis of this flight regime. At the outset of the contract's finai year, there was a reorganization of the UTIASfSRI position in the DAWA administration. What soon foiiowed was a subsequent reclassification of the TiASlSRI effort to E d l under the direction of the Micro-Adaptive Flow Control (MAFC) branch of DARPA, rather than the original MAV group.
Also i n the early stages of the
"flight simulator" of t h e MAV for the purpose of allowing rapid evaluation of stability and control under different vehicle configurations. This would become an invaluable tool for facilitating prototype design. The analytical model used by the simulator required experimental data (i.e., forces and moments) for the MAV wings and t d under diffierent flight conditions. The initial data that was King used were simply "best estimates" of
what performance could be expected
e r e aIso under different angles to the k-stream veiocity. Tai1 lift and drag data w
determined, Together, these were t o be used wih the aforementioned simuiation code to
coduct case studies of possiile tail-wing coufiguratr*onsthat would lead to saishtory
s realkd Udbrtunately due to Iogistical controI and stability when a fiynig modei i
problems, the author muid not personay conduct such case studies with the SRI simulation program. As an alternative, a 2-Dprogram was produced s o that these studies couid still be performed, albeit at a somewhat less sophisticated level of programming.
research. With this change came tbe aiieviation of some of the restrictions piaced upon the project in terms of size limitations. No longer did the vehicle need to conform to a 15 cm maximum dimension; however the pmject would sti rernain hcused on producing a platfom useful to the military. One of the main issues impeding progress of the initial
MAV prototypes was the lack of energy density available with even the latest generation
of batteries and capacitors. Free flyers powered in this mamer were very limited in their tiight duration Thus, much of the finai year of contract work hcused on developing a 30 cm span flyer that would achieve successfl flight. It was thought that by going to a larger span, more thrust wouid be produced anci therefore the ability to use heavier, gaspowered forms of propulsion wouid be made possible (which in tuni wouid extend flight duration times). Indeed, at the time of this writhg, a gas-powered R/C flyer designeci at
That being saki, the reader shouid be reminded that d i the tests, experitnental
results and data, as weU as the 2-D simulation code, a revert back to the original 15 cm
MAV flight modeL The idea being that if the contract were to continue beyond 3 years,
the initiai 15 cm platform may be revisited In order to meet DARPA1soriginal criteria. Even if a contract renewai were not to materiaiize, or if the 15 cm flyer is completely abandoned, the research into the stability of a haif scde mode1 of the existmg gaspowered prototype would most certainly be beneficial as a f k t approximation in evaluating the i t s stability.
obtain true qualitative data on the latest MAV wing design. Due to the unsteady nature of the lifi mechanisrns involved, an analyticd method of i d e u t m g the forces and moments of such a wing configuration was yet to be fiilly devebped Thus, an entireiy empirical approach was taken in detennining this information
i n g design showing the most promise had the ability to produce 50 The latest w
gram of thnist when flapping at appximately 40 Hz [l]. Caiied the BAT-12, this
design is depicted in figure 1.2. It was this type of wing was used to evaiuate the test data
m this author's research The wings are constnicted using a unidirectional carbon
fibre
Since EPAMs remained unavailable for use in the tests, the wings were actuated
attached. Driving these shafts were two wnnecting rods attached to a d DC eIectric
d fabricated by SRI during the early stages of motor, Such a rnechanism was designed a
the project. Figure t .3 shows this device (named ProtoSouth).
e b.ansfomd into tube rotation. The rods extend 10.2 cm to a crank extending rods to b
h m the DC electric motor. When actuated, the motion of the tubes i s nearly sinusoidal.
T h i s motion i s transfemd to the wings momted on hubs attached to the tubes. With two
wings per hub (m an opposing orieatation), they are able to tlap and rotate against one another. Such motion produces the cIapfling effect
mechanisms sought to produce the required iift. For f k h r msight into this and other
through in one cycle of craak rotation, It is governeci by varying the Iengths of the
vertical links in the four bar mechanism. Unfortunately, the abiiity to vary the amplitude
was not a feature made available in the construction of ProtoSolnh, The flappmg
amplitude of this mechanisrn was fixed at a value of 60 degrees. Previous research fiom
[II revealed that 72 degrees of amplitude was a more desirable value. However, since the
existing prototype was both readily available for testing in addition to beiag more durable than other existing mechanisrns, it was thought that it would be sufficient to evaiuate the desired data. No previous research had been done to mvestigate an optimum t a i i design, nor in
the placement of a tail with respect to the fklage of the MAV, except for some
conceptual drawings and sketches. The simulation code requkd only the coefficients of
tifi and drag of the tail through 360 degrees of rotation m a flow field, It was believed at
the outset of rhis research ihat the orientation of the tail (Le., above or below the wings) would be the most important factor in govwning vehicle stability. Therefore more emphasis was directed to investigating tail positionhg rather than on exhaustive testing of various taii designs.
tail,
sensitive enough to measure the inherentiy small forces to be encotmtered. Such a force
balance was buiit at the UTIAS Lab specifidy for these tasks, and its design i s
descriibed in the foiiowing chapter.
2.2
Design Specifications
First and hremost, the fundamental design had to b e scaled down m order F o r it to
adapt to the anticipated forces encountered with the MAV wings. Essentiaiiy, the revised
concept consisted of an aluminum tray suspended h m a fixed upper pIate via thin wires. This m y could translate i n two directions as weli as t w i s t .A munting piece attached to
the Iower tray extended up through a hole i n the centre o f the 6 x 4 upper plate. It was to
this mounting piece that the ProtoSouth iiapping mecbanism attached and was able to
transfer loads. Siran gauges were rnounted to the h e d upper plate and reacted to aay
translations of the suspendeci lower tray. Figures 2.1 and 2.2 are simple depictions b
more cleariy illustrate how loads were transferred to the strain gauges, as weU as their
layout. Three gauges were useci, each labeiieci #1, #2 and #3 as m figure 2.2.
measurwient of
forces and a moment, which was preciseiy what was desired h r the planned testhg
to follow.
Fixed Plate
Btacket
-.
I
F -
Sbain Gauge
Longitudinal Lateni
Gauge #1
th
Gauge #2
loads were resisted by gauge #3. Longitudinal loads were determineci through a summation of the readings h m gauges #I and #2. Any appiied moment manihted itself
as a diierence m these two gauge readings and was determined knowing the distance "8'
between them, using the simple formula:
axis) extends
through the centre of the MAV dong the tlapping axis. The x-axis (labeled the laterai
axis) was aiways oriented so that its cosine component was pomted downstream. Hence,
when the simulation depicted the vehicle rotating past 180 degrees in a crossfiow, the xaxk instantaneousiy changed to maintain its direction inro the wind. The y-axis
completed the orthogonal triad m the right-handeci seme. Figure 2.3 superimposes ttiis
system over a simple sketch of the MAV.
Adapting this system to the gauge Iayout in figure 2 2 , tfme forces dong the z-axis
as lateral Ioads.
minute forces yet still retain some durability so as not to be easiiy damaged. Thus, during
constmction. an effort was made to allow 6 r adjustment m order to make the device
more rigid or relaxed. With a fieroile desigu, it was believed that if the completed
bahce perfomied u~wtisfactory, i t wuId be easily modined without scrapping the
entire device and sbrting over. One level of adjustnsent was t h e abiiity to alter the distance separating the two plates. Taken in the extreme sense, a very short distance
wouid d
e the balance very "SM" with respect to applied moments an forces, whereas
rigidly a n c h o ~ g them into position. Leaving the wires long p d e d the bwer plate to descend h h e r shouid the mecl arise. Figure 2.4 ilbistrates how this was done.
io F
d L rper Plate
Another pararneter that codd be changed, albeit somewhat less conveniently, was the distance "d" separating gauges #1 mi #2 9i figure 2.2. A larger distance wodd d o w
The gauges used were corrmietcially purchased AC Sensor Mode1 6000 PlanarBeam Force Sensors [4]. Each sensor contameci a fu bridge stnm gauge mtegrated ont0 a thin-film s t a d e s steel element of 0.004 i n thiclrness, T h i s particular mode1 sensor was
the Iowest capacity (114 pound) avaiiable h m AC Sensor. It was decided that such a commerciay manufactured product wouM tte more reliat,le and accurate than design& and sizing appropriate flexures m-house. indeed, h m the detaiIs that foiiow m this chapter and those ahead, this assirmptioa proved to be ue. ,I
mounted cantilevered as shown m fgwe 2.5.1. It was discovered however that this type
pardel beam fashion as a means of cornpeasating any applied moment and reducing
errors in off-centre loading. In other words, the gauges were consained to react to pure
To transmit the appiied loads to the strain gauges, angle brackets were mounted to
the Fiee lower plate. Each bracket was aligned with one gauge (refr to figures 2.1 and
2.6). Extendhg fkom the bracket to the gauge was a piece of heavy piano wk,which
acted as a rigid rod
folIows: an applied force h m the apping rnechanism i s passed through its mount d o m
to the k e Iower plate, which m tum is transmitted through the angIe brackets, through
the piaao wire, and finallv is resisted by the gauge. A photo of the fmished balance i s
shown m figure 2.6. Figure 2.7 shows the balance together with ProtoSouth, attacheci to a
lripod as it was during actual testing.
2 . 3 Balance Calibradion
As rnentioned previousiy, a cantilevered strain gauge design was scrapped
in
hvor of the paralle1 bearn configuration. What foiiows focuses on the calibration of the
gauges m their latter form.
d&
for this effort. First, s k e the *es attaching the angIe brackets to the gauges were glued mto p k , these was m, way of "umloingy this uai step. Secondly, a cumpke gstm caiiition (Le., with angle brackets attachai) wodd require the assumption that the
apped longitudinai loads were equally shared 50150 between gauges #I and #2. By
The resuits fiom these idependent triais revealed that there was no appreciable change m the gauge slopes before and d e r the fird attachment of the angle brackets.
loads. By using a pdey system, a series of known masses providing the forces were
appiied in both lateral and longitudinal directions. Output voltage readings were recorded, dowing the determination of each gauge's slope, dehed as:
k = AV/m
(2-2)
where AV represents the change m the gauge output volage between the loaded and
unloaded condition (measuted in millivolts), and m is d e W as the applied m a s providing the force (in gram)-
a series of loads in the x direction (both in the positive and negative sense), ensuring no force component emerged dong any 0
t h axk
and #2 in the z direction. As expected, each gauge exhiiited near behavior in response
to loading, as weii as the remarkable virtue of zero crosstaIk among the gauges. With
zero crosstalk, a gauge's output was wt comrpted h m loadings outside of its intended
axis of measurement. T h i s particular test's resutts are detailed m Appendix B under case
banage of tests would serve two purposes. F i , by using the h M l y derived k-values,
an estimate of the percent error incurred under ciifferent force conditions could be
evaluated. Second, new k-values codd b e derived h m these extra test cases, allowing the abity to assess any gross change th& magnitudes. This rather elaborate procedure of the balance, and aid m wouid gamer a deeper insight into the o v d perform~tnce determiniag the final k-values to use during actual experbentation. Complete data for
each test case are included in Appendix B. What wl fbllow will be a brief description of
each subsequent case and summarize its d t s .
Pute Applied Moment
The 6rst case entailed the application of a pure positive moment about the y-axs.
This was achieved by bolting a daluminum a m to the ercisting mouut, as ilhistrated
in figure 2.8.1. The distance " P ' between the applied force T and the center of the plate
couid be varied aiong the arm, which allowed the magnitude of the applied moment to be
adjusted. A mass of 40.3 gram was applied at 1 cm increments outward aiong the arxn
Results for this scenario were exemplary. Using the mitialiy derived k-values, al
erros for mass and moment were on or about 5%. No crosstak was observeci in gauge
#3.
Cmbined X and Z Forces
A mass of 17.6 gram was applied dong a diagonal, such that it allowed a
component of its force to appear m both the x and z directions. Figure 2.8.2 depicts this
scenh.
Choosing a diagonal travelling exact& through the corner of the retangulac plate fcilitated proper alignment. Simple geometry determined the angle 0 to be 53.04". Again
the balance performed admirabiy, save for mstances of small loads (below 7 grams). Combined X and Z Forces wih Moment
This final test case w a s compteted by using the previous a m attachment aligned
dong a similar diagonal as shown m figure 2-8.3. Agaiu, the appiied test m a s w a s 40.3
granis-
values, as one could view each scenario m its& as a calibration method includmg the
+ Z Force
0.0541
0.0871
+ X Fort4
nia
0.0563
nia
0.0820
da
da
0.0528
- Z Force
- X Force
+ Y Momsnt
CornMnad X, Z
nla 0.0546
0.0524 0.0554
nla
0.0897
nia
0.0537
J
Forcer
Combinsd X, Z and
0.0815
0.0829
Y Moment
0.0537
margin of dEerence no greater that 10%. E t was aiso kk that the finai k's chosen stiould
greater reflect test cases which involveci multiple forces. After much thought, it was
decided that the t e d s deriwd kom the combined x, z forces and y moment condition
would be the best representationof the overa system's calibration cuefiients.
tests. This tunnel was built by Mr. Patrick Zdunich (a Master's student at TTAS involved with the MAV project) to mvestigate flow visualization aspects of his research.
Mr. Zdunich later decided to pursue other flow visualization methods, thus leaving his
wind tunnel available for use. T h i s was a fortunate circumstance, as the larger wind
tunnet in the subsonic lab would likely produce velocities much higher than those desired
for testing a small MAV. The tunnel measured some 49.5 in long, and was consmted fiom medium density fibre (MDF).Air was accelerated by a small 18 m diameter indusiriai fn initially
wide. Figures 3.1.1 and 3.1.2 are photos of this tunnel at UTIAS.
3.2
3.2.1 M i a l Resulfs
Idedy, a wind tunnel should create a b w field tbat is entirely d o m in its cross-section, Such an ideai is never t d y realized due to boundary Iayer effects dong the
w a k of the tunnel as weU as turbulence in the flow. Early cali'bration tests were
performed by Mr. Darcy AUison, an undergraduate student who worked during the
revealed a somewhat
disappointhg "weli" or "dip in the center of the velocity protile. Some steps were
required in order to rectify this problem.
in an attempt to dirninish this chanrcteristic, a cone was built by Mr. Allison that
f i e d to the fn of the tunnel. This was expected to accelerate the air more uniformiy, as
the rather gewrically designed fiin was by no means coostnicted with be purpose of
wind tunnel testing in nhd. The cone addition yielded somewhat better results, however
the undesirable velocity dip was sti apparent, as depicted in figue 3.2 (in three
dimensions).
I
I
I
I
I
,
I 1
1
Mer much consideration of these eariy results, it was decided that this trait of the flow field might not be as great a hindrance as initially expected. Although the tlow field
as a whole was decidediy non-dom, the velocities in the central "pocket" of the flow
were in ict firiy consistent. The shallow dip measured roughiy 15 cm in height by about 20 cm in width. Recalling the span of the MAV wings were 15 cm, i t was decided that provided d i of the tests were con6ne to this "sweet spot", respectable resuhs would be
attainable. As is d e m i m later chapters, this assumption proved to be accurate. For
3 . 2 . 3 Calibrafion Procedure
A pitot tube together with a nianometer was used to m e m e the flow field
velocities. The pitot tube was anchored to a rod supported by a U-shaped fiame situateci
in front of the tunnel exit, as shown in figure 3.3. The probe was positioned to take
sample readiigs at 1 in hcrements verticaiiy and 1.5 in horizontay. There was also the
ability to position the probe at various distances away fiom the tunnel exit. This aiiowed the degree of velocity decay away from the tunnel exit to be observed.
The standard method for detennining velocity using a manometer was used, whereby a change m the nianometer reading was translated mto a dynarnic pressure, which in turn was used to calculate the air velocity at that pomt. The pressure P exerted
by a manometer fluid with a density puu& at a depth h is given by:
P=p&&h
(3-1)
where g is the acceleration due to gravity. The change in the manometer reading h m the
zero vetocity condition constituted the value k Thetefore! P would quai the dynamic
pressure exerted by the air. The dynamic pressure q of the air is d e W as
q~ = sPurv2
(3-2)
The density of the air during testing was evaiuated by knowing the ambient temperature
w@)
(3-3)
a s the change m manometer reading (mches), with the manometer tluid being where L w
kerosene. T h i s formula was easily modified for use with decane, as the onIy property that changed was the fluid density. Thus, the equation becarne
q = 0.2199 L
(3-4)
Of course for consistency, the results were convened and reported in metric units (Pa).
For the 6nai caiiition test, the mawrneter normally used with the large wind tume1
was empbyed, as it was fomd a more convenient apparatus. It read m mches of water, so
no speciai forrnuia for q was required. Equations (3-1) and (3-2) were set equal and
inimediateiy solved for the air velocity V.
voltages up to 110 volts, which therefore dictaid the maximum attamable wind velocityin each case, a specific voltage setting was correlated to a certain calibrateci wind speed. Complete velocity protiles for the tunnel settings used in the experhmtal tests
are included in Appendix C . It shouid be noted that h m the prelEnioary tests performed
by Mr. Allison, it was discovered t h there was mniimal decay in the velocity field as
one moved away Eom the tunnel exit (i.e., m the order of 15 cm or less). Smce it would
be quite simple to constrain aii testing to within this distance, it was decided to take
cali'bration readings for a 15 cm square region centered only at the tunnel exit. No additional profles were sampled at dhances away fiom the exit.
Chapter 4 : EXPERIMENTS
4. i WSng Testing Procedure
4.1.1 Mefhodology
It would be wise for the reader to c e themselves with figure 2.3 in
Chapter 2, whih iilustrated the wind-hub axis system used by the simulation code. It was
decided that the best testing procedure wouid measme these forces directly, Le., have the
force bahce continually aligned with this body-tixed axes system This wouid elaninate
the need to convert the results with tngonornetry into the desired axiai components. Such
an added step may bave produced undue error. The SRI simulation program required data for the MAV wings' lateral and
longinidinai forces and moments for 180" rotation in various ke-strem velocities.
These measurements wouid be perftorrned m the static sense, meanhg the wings wouid be
positioned at a ked angle of incidence to the crosdow, and then the forces would be recorded whiie flapping at a steady state. The tests wouid not address the dynamic scenario, whereby the mechanism wouid be rotated through the crosslow at a constant
anguiar velocity while simuitaneouslytaking readngs.
Due to the nature of wbg actuation in ProtoSouth (see figure 1.3), it was
irmnediately apparent there wouId be problem when the whgs were oriented past 90 m a
crodow. In the extreme sense, with the wings positioned at the 180 mark, the flapping
mechanhm (as wel as the mount attachai to it) wouid be upstream of the wiags. T h i s
sort of flow blockage would be totaiiy unacceptabie. R
obtain data for the wings alone (Le., mau.s any driwig mecbanism). Since it was
mipossible to completeiy isolate the whgs h m the main body of ProtoSouth, some alternative method of testing was necessary m order to record data at angles beyond 90".
A simple solution emerged whereby the wings were mounted backwards (Le., inverted)
t was possible to accumulate iformation for on the mast of ProtoSouth. By dohg this, i
the extreme angies of crossfiow. Figure 4.1 illustrates this wing testing procedure.
45O
90'
This figure iustrates the two basic steps in the testmg ptocess. Step 1 depicts the wing mounting used during the first 90" of rotatioa At the 90" point, the wings were detached a d remounted as shown in step 2, suh that the i d h g edge of the wing was now upstream of the l d m g edge. This allowed the remahhg angIes to be tested.
One remaining drawback ofthe procedure was observeci during the mitialOO - 90
rotation phase. During these angles the wiugs were orienteci such that they were tbnistmg d o m upon the flapping mechanism and mount, which acted to b k k the thnist. It would have been more desirable if the mast of ProtoSouth w e r e much longer than its current 5.5
cm Length, Such au elongated uwt would have acted as a shg, thereby dowing fess
downwash on the main body of RotoSouth. Effort was taken however i n design& a momt that would not add coderabiy to this b w impedance. Short of rebuilding
ProtoSouth, this was a that could k done. Uniess the ensuing resuits appeared
completeiy out of sorts, no such recoastnrction would be atternpted.
Communication with SRI'S Tom Low, the progmmmr who developed the
simulation revealed that the code worked using a series of lookup tables. The computer
wodd evaiuate the MAV vehicle's flight condition based on the advance ratio J and the
vehicle's orientation in the fiee-tre;un.M r .Low defined J of the vehicle as:
where V i s the fiee-stream velocity, b i s the span of the wings (15 cm), o is the hpping kequency (in Hertz), and 8 i s the flappuig amplitude (in radians). Once the computer determined the vahe of J, i t would reference the tables and interpohte where necessary
to acquire the forces and moments acting upon the MAV.
Wbat beame mimediateiy apparent was that J had dimensions of revolutiom?,
which meant that it was a kqueracy dependent variabIe. S k e the wuigspan and Eiapping
amplitude were fixed, the ody parameters that coukl be varied were the fiee-stream
velocity and the flapping kquency. M r . Low had mitialiy programmeci his code with
advance ratios of 0.5, 1.0, 1.5 and 2.0. Matheniatidy speakin& there was an infinae
number of V and o combiions that couid produce these desired P s . However, one
must t e -
this k
t with
condition, For example, knowing the top speed of the ninnieI to be about 7 mis, it can be
I 1 Hz However, would this be a realistic fIapping fkquency? In the context of the MAV vehicie, the m e r was absoluteiy not. Refrring to the research performed both by M r .
Bilyk [Il and Ms. EEKhatib [2], such a low hquency would not produce suflicient
thrust, nor wouid the wiugs twist in order to perform m t h "clap-hg" region so mveted in this scde of ight, Therefore, the set of experiments wouid have to be performed in
such a way as to be meaaingfiil and approximae the tme &ght conditions.
As an initial approach, i t was decided to perform tests at 40 Hz (a value
hovering of the anticipated MAV, Unforturaately, it was dikcovered thai this was a padcularly demanding hquency in ternis of wing and motor durabiiity. in hct, once the
crossfiow cornponent was appiied to such fiapping, the wings were found to disintepte
(approlcmiateiy 22 gram). The conclusion was therefore to @nn ail testing a t 30 Hz,
with uniy the variations in the fk-stream velocity king the method of aIering the
advance ratio.
Refrring to the wnmd tunnel caii'bration resdts [Appendix C), the maximum attaiuabIe vebcity was 7.0 mls. With the other parameters d o n e d above, this limited the rmxbmm advance ratio to 0.743. Given this d u e , and wah fkther disckon
with
SM,it was decided to acquire data for three othet ratios of about 020, 0.50 and 0.65,
Each of these would require a specinc velocity. Fiow k I c vetocities wete based on an
average value of several manometer readiogs. Hem, it was extremeiy difiicult to make these average values match to hose specied by the advance ratios above. An
effort was made to approach these as best as possible, and as a result, the &g values were 0.19,0.55,0.66 and 0.74, which were deemed acceptable by M r .Low.
h i s greatiy The force balance was mounted to a tripod for ail trials perfonned. T
fcilitated Ieveliing of the system, as the tripod had numerous adjustments for this
purpose. In addition, the tripod aiiowed the tialance to be raised or lowered m the flow,
such that the wings wouid aiways be p k e d m the optimum "sweet spot" m the tunnel's flow field.
The balance was wired to a Fluke NetDAQ data acquisition system attached to a
laptop cornputer. The NetDAQ monitored four channeis, aamey the three gauge outputs
as weU as the applied excitation voItage. The NetDAQ proved to be a very convenient t s accompanying Windows softwareprovided many options with regards to apparatus, as i
sample times and output formats. With some advice fiom Mr. Dave Loewen, a 5 second sample t h would be recoded to a data tile at 0.006 second intervais. Thus, a typical test
n i n wouid
the wings in motion. The gauge outputs w h k the wings were flappbg were quite
oscillatory, as can be expected by the nature of the motion. In order to determine the
mean change i n voitage h m the zero cornlition, these oscinating outputs were smply
averaged over the 5 second tirne i n t d This was proven to be a valid assumption as a graphical plot showed these aUctuathg outputs akhg place about a lxed mean due.
A h , simple thrust tests (with no crusdow) produced tbrusts m close approximation to
Each test run consisted of positioning the wings and tripod together at he desired
angles in the crodow. The kremental change in aogle was chosen to be IO0, whkh
proved to be of acceptable resolution. As illustrated in step 1 of figure 4.1, the wings
were swept MaIy to 90, wiih an added test doue at 100' prior to inverting the e
s .
This was done to provide some overiap i n the results. As descriid earlier, the wings
were inverted and cepositioned (or swept back) to compIete the fidi 1%O0 rotation. Each of
these weeps was performed three times for each advance ratio to detennine spread of
data and he! degree of repeatability.
4.f.2 Taring
One of the greater
involveci the tare values of the force baiance mount. Tare dues are the force and
moment conttriutions mide by components other than the wiags during testing. It was of
great importance to keep the tare to a minimum percentage of the total reading, as iarger
values tend to contaminate the r e d s . i n this case, the muunhg bracket and hpping
mechanism w m susceptible to the crossflow and thus transmitted drag forces to the
balance. Fortunately, the fieestream did not affect the force bahnce k i f as it was
[a otder to
The method for calculating the tare of the mount and flapping mecbanism was to
simply record their longitudiuai and laterd forces and moments (without the Wmgs
attacheci) under the same crossfiow conditions as those to be tested wiih the wings. As
can be seen lom figure 4.1, the tare values fiom O" to 90" wodd be completely analogous to those fiom 90" to 180'. The 6rst mounting bracket used w a s a disappomtment. Sketched m figure 4.2 below, it consisted of a simple post with gussets extendhg outward to support ProtoSouth.
measured moments (about the centre of the baiance) to a position on the MAV wings.
The problem was with the laterai force's contriiution, which had a large lever arm, which
m turu
increased the mgnitude of the readmgs. This is depicted more clearly in @re
43.
Lever
Darire Moments
c
Figure 4.3: Original Mouniing Bmcket (Top View)
force readiigs met no such problems m taring, and th& data couid be obsewed.
A lesson was leamed h m this rather dispieashg start, and much greater thought
went into the design of the second mount. Two issimes were addressed. Fi,the size of
the bracket's fiontal area was minimised at d e s near 90' to reduce lateral tares.
Secondly, there was the need to have a h e d reference point by whih moments would be calculated about, rather than attempting to transfr the moment to a selected point on the
Discussion with M r . Low supported this decision, and revealed that his code couid be adapted to d o w the moment to be r e h e d anywhere on the MAV body. No removaI of the laterai force's moment contriion wodd be perfonned, With these issues i n
mirad. a "gooseneck" type m u a t was constructeci (shown in ligure 4.4), which enableci
the leading edge of the MAV wiags (in either a f o d or inverteci attachment) to be
The irnprovement was still far h m perfct. At least in this instance the values
were les scattered and a trend was beginnnip to emerge. Apparently, the moments of the
MAV wings were either exceptiody small, the tare was d l too Large, or both
Determinhg their values with confidence continued ta be a challenge. Two nnai options emerged, The h t was a redesign of the force bahce to d e
it les s t 8 in an effort to attenuate its sensitivitynsitivity Alternatively, an attempt to s h u d the
muutmg bracket wh some type of shieki wouid reduce the tares even finrther. Tbe decision feu to the latter, as it would be he quickest and e!zisiestto impIement.
A two-piece b a r n "cocoon" (seen m figure4.5) was cut and mounted about the
braket. The reduction m tare values was astonishing, reducing their dues by 64%. The moment data (reportecl in more detail m the next chapter) became immediately more clear, and an identifiable and repeatable trend was observed. The tare reduction effort had
actual tests, a series of trials were performed without a crossflow to determine the magnitude of this misalignment. Until this tare was minimised (through numerous
ajustrnents), the test would not proceed. In addition, at certain pomts in the test procedure these trials were repeated to ensure that the thtare had not changed. If it
had, the test was either repeated or modified to reflect the new tare.
4.2
previously, more emphasis was to be d e on their placement with respect to the MAV
body in the simulation program. An m-depth and exhaustive study to create an optimum
tail consgUration was not the intention oLthis research,
The hst design was a basic rectanguIar sbape, wbile the secoiad was a simple
half-mon. Both were constructed of 1/16" batsa and glued to a metal rod wbich attacbed
to
4.2.2 Methodoiogy
The taiIs were tested i n a similar fshon to the methods above; save in this
instance the gooserteck mount was mit use& In its place was a verticai post with a r d
attachent as iu figure 4.9.
Mr. Low's simulation required ody CLand CD c w e s for the tails d e r a 180'
rotation. Thus, rnoment meaSurements were w t requned duriug the procedure. This k t ,
i n combination with the smaller motrnting bracket and absence of a fhppng mebaniJm
meant there was no need for a shroud to duce the tare. T a dues were f o u d to pose no difiicuity whatsoevet.
It was decided to test the taiIs at a wirmd velocity correspondmg to the typicai downwash velocities detecmined from M s . El-Khatb's [2] research with hot-wire anemometry. However, given the probable s k an MAV size tail (iess than 7 cm),the issue of taring problems ernerged once more. Thus it was decided to double the scak of the tail dimensions, but test at haLf the velocity, This was to ensure that the Reynolds
nurnber remahed m a smiilar regime. Ms. El-Khati'b's research showed air velocities of
about 4 mis at positions 15 to 18 cm below the wings. Unfortunateiy, the enlarged wings tested at 2 d s yielded very minute forces i n the order of 3 grams or less. T h i s greatly pushed the sen~itivity iimits of the baIance, causing unrealistic drag and lift curves. Mer much consideration, it was coneded that the only option was to test at a higher velocity of 5.24 d s . Although effectively more than doubling the Reynolds nutnber, it was sti of low value (below 25,000) such that there would be minimal error m the n~n-~onal
Lift and drag curves. Auy discrepancy w u l d likeiy manifest itseif m the f& curve's
5.1. 1 Repeafabiiity
As descn'bed i n Chapter 4, a totai of four advame ratios were investigated in the
experimental anaiyses. Also mentioned was that for each advance ratio, a total of three
180 weeps were performed in order to establish the repeatability of the measwements
i r mor bands. In al cases, the degree of scatter in the recordeci data and the size of k
was low, especialIy with respect to the lateral (x-axis) forces. Recaiiing the tahg
challenge that was encountered with the moment measurement, it was a pleasant
experience to h d y iden@ clear and repeatable trends for this data. Figure 5.1 shows a sample plot of the lateral (x-axis)force vs. crossflow angle for 3 triai rum.
X Force (9) vs. Angle 3 Trials
l
40.0 -[
'
Upon inspection of the longitudinal (2-axis) data, i t was readily observai that a sharp discommuity o c 4 at the 90" mark. The uiitiai conclusion was that the
inversion of the w b g s (recd section 4.1.1) was the source of this abrupt "jump"in the
rneasurements. Why the force decreased m magnitude however, was somewhat mysterious. One would intuitively expect that with the wings mverted, they would b ek e
l+om the b w blockage caused by the Dapping mechanism a d shroud and subsequentiy
produce mure thnist. Yet it appeared the oppsite was me. Of particular interest mthe
k t that this dikcontinuity was absent w k n the a d m e ratio e q d e d 0.19. Figure 5.2
illustrates the characteristic, and its non-appearance when J = 0.19.
Due
to
about the fiont of the shroud (Le., the region immediately afi of the e s ) , it may have interacted differentiy wih the complex vortex shedding of the wings, serving to ampli@ their thrust. Another reason may have been that the shroud and flapping mechanism served to block the mcomhg air to the inverted wings, demashg their thrust. Or perhaps
the higher velocities disnipted the mtake of the whgs m their inverted condition in aii
iikelihood, it may be an elaborate combination of al1 of ttiese tactors that contriuted to the problem. No clear solution was apparent, and no remedy seemed to remove or lessen the trait unless ProtoSouth codd be refitted with an elongated
mast.
Overlapping
readings were recordeci at 100" d e r the conventional aa mverted attacbments, and reveaied the discontinuity continumg past the 90" mark. CompIete raw data for these tests are mcluded m Appendix D.
A few mteresting points were made atler a ktrend h e was passed through
the data for each of the advance ratios. in ati cases (except for J = 0.19), it was
immediatety apparent that divergence h m the trend line began at 70" and ceased at 90"
(see figure 5.3). AU data points afier 90" were very near the trend Zinc, with those below
70" conforming as weU. With these outlymg points temoved, and the trend line reapplied,
i t was discovered that the equation of the trend 1Eie changed oaly slightiy (figure 5.4).
Hence, the culprit causing this jump would most Likely be found by focusing an mvestigation m the region between 70" 90".Evadently, inversion of the wings was not the contriuting fctor, but rather some intefaction ernerging near the 70" pomt.
Figure 5.3: Longitudinal (Zab) Force vs. Angle with Linear Trend Line, J = 0.735
Z Forca vs. Angle, J = 0.735 (Anomalies Removed)
Figure 5.4: Longitudinal (Zd) fiire vs Angle wak Liuear Trend Line, J = 0.735, (Odijdng Anornulits Runo@
The data was reported to M r . Low at SRI m its origiual form, Le., without any adjustments or alteration. Of course, it wodd t e keiy that such changes would (and certainly should) occur, however the author felt it best to report the accumuiated data m its purest sense, without any manipulatioa hother observation made kom the addition of the trend lines was the remarkable linearity in their slopes, as indicated by their ' R vahies (again with the exception of the case where J = 0.19). This lead to the conchision hat for advance ratios above 0.5, the relationship between longitudinal Grce and the angle of incidence to the crossflow was of
a nearly linear nature. T h i s conchision was Limited to these higher advance ratios (which
of course corresponded to higher velocities), as indicated by the marked difference in bebaviour when J = 0.19. Dirring the accumulation of data, discussion with the M r . Low and his coiieague Bruce Knoth showed they had a preference for data at the higher advance ratios (i-e., above OS), aad thus no f i r tests were performed to determine if a
similar dope relationship couid be made for the Iower region of advance ratios. It i s of
importance to remind the reader that the simulation code worked on the principle of
lookup tables, rather than c o m t e mathematicd fornulas, to determine the forces on the MAV. Thus, the above dope observation was an experimd conclusion reiated to this thesis, but not reported w r desired by the SRI software developers.
Figure 5.2 shows a clear correlation between the force magnitude ami advance
ratio (Le., a logicai progression of highex advance ratios correspoedmg to higher forces). However, again there was a contrase wben J = 0.19 which, as already mentioned, iacked
What was certainly conmion to aii four ratios was their magnitude at 90, which was approximately 22 grams. This corresponded to the static th& value the BAT-12
wings produce at 30 Hz. This made sense because, when at right angles to the oncoming tlow, the wings (at least m the longitudinai sense) were not "seeing"any component fiom
the crosstlow, and thus produced t h e h conventional thnist values t h e regardless of the
magnitude of the Eee-stream.
apparent. The close symmetry of ali plots about the 90" pomt also foiiows as one might expect.
Avg. X force vs. Angle
--
longitudinal data) data did not show similar rehtionships for any advance ratios. A 6nai
observation can be made on the disruption in the curve for J = 0.19, while the curves for the other ratios remained srnooh Again, one mut assume that the source of the error i s kely due to the change i n whg aitachment, as the dip m measurement appeared near
90". A surnmary of the raw data used to generate these figures i s mcluded m Appendix D.
5.1.4 Moments (about Y-axis) The moment data obtained Eom the experirnents was by Far the most interesting,
and revealed a striking contrast ktween the higher advance ratios and the lower J = 0.19 condition. For the high advance ratios (0.55, O656 and 0.735), the similanty in trends were obvious, with the lowest magninade of moment fonning an apex about the 6 0 '
- 70"
-70.0 J
Angle m g )
Figure 5.6: Moment (about Y-k)
vs. Angle, Al1 Advance Ratios
With regards to when J = 0.735,it was observeci that the moment did not return to zero at 180 as one might expect. Since this was the highest advance r a t i o (and thus the highest ee-strearn velocity), it was probable that any minor misalignments of the
apparatus were amplied at 180,conmbuting to the zero o&t. Even through repetition,
this data point rernained an outlier fiom the zero pomt. Of course, one couid d
Paruarucuiariy interesting was the pronouuced positive moment at aogles above 110,
whereas the oher ratios for the most part were entirely negative. When J = 0.55, there
was a simiIar positive moment region, aithough here it was tightIy cordird between 160
and 180". One can speculate b t were the advance ratio lowered fiom 0.55 to 0.19, this
positive zone would expand to encompass a wider berth of angles. At the h i @ ratios of 0.656 and 0.735, there were no positive regions present.
adme
5.2
lails
5.2.1 Results
The CLand CDcurves for the two tails descriid in Chapter 4 are mcluded below.
In cornparison to the whg tests, ihese experimens were relativeiy simple, and therefore
only two 180" sweeps were perfonned, Very W e scatter was encountered between readings. Unlike the wings, the simukition code required curves for 360 rotation. Thus,
the data was simply mirrored about the 180" mark to satisfj. these requirements. Figures
FigmS .7 : CLC
especiaiiy wih regards to the CL cuves, which were nearly on top of one another. Any
performance difference would most likeiy d e s t itself i n the slightty higher Co values
producd by the taii #2 design, As was discussed m Chapter 4, it was expected that
placement of the taii (and n o t some extraordinary tail stiape) would be the more miportant fktor in determining a stable coafiguration.
Z Forces
For reasons aimdy meatioaed, al1 data was taken at a fiqping f k q u e q of 30
HL However, the projected mass of the actuaI MAV was expected to be close to 50 g, a
thrust value tbat could ody be reached by flapprag at a 40 H z fkqwncy- Thus came the
issue of scaIing the data mteigently so as to represent the hrces kurred at 40 H z A
forces was anything but iiuear- Consuitation with M r . Bilyk and Dr. DeLaurier raised the
bypothesis that the thnist produced by the MAV was directiy rekted to the axial
component of the fiee-stream velocity impmging upon it. This was taken fiom the fact that when the MAV was oriented 90" to the flow (i.e., with no axial k-stream comportent), it produced a thrust values neariy identical to those when the ke-stream
was absent. A nondiinsionai relationship was devised to properly test this theory, and
it included the above variables together with the f.lapping fkquency and span of the MAV
wings. The Girst nondimensional group compriseci the ratio of measure hmst (Le., that
h n i s t pmduced when be crossflow recorded during testing) to the static thrust (ie., the t
was absent), and i s herein referred to as the tisrust ratio.
between the axiaI component of the k-stream velocity to the product of the flapping tiequency and wnig span, and was labeiied the k-stream kquency ratio.
Aii four advaace ratios were reduced to this f o r m a t , and plotted as shown m
figure 5.9. The scatter plot showed a remarkable lineanty in this telationship. However,
this did not entireiy validate the hypothesis. AU the advance ratios were produced h m
30 Hz data. Some fkther investigation was required at other fkquewies to better
Th t u i Ratio v a W a W
2.50 1
I
Werent lapping fkquencies. Tbnist data for 25 Hz and 35 H z was obtaiued in a 524
d s crosdow, and is shown with the previous data m figure 5.10.
4.30
-020
4.10
0,W
0.10
0.20
0.30
/
1 l
VaxhW b
Figure 5.1 O: Tkrrrst Ratio us, Free-Streum Frequency Ratio (Orig. and Ertrp Daia)
From t h e d t s , it was apparent that this relationshrp extendeci to other
keqwncies, with a tolerable degree of scatter m the plots. The next task was thus to
extrapolate h m this to anah the &ta for 40 H z A simple addition of a trend iine through tbe data ( h w n m figure 5.10) allowed for this, remit& in the eqation
where b is the span of the wings and o is the apping ikqueclcy. The q u e of mterest
was now Lat 40 Hz for each tested advance ratio. Thus, by hwing the Vd
components for each advance ratio, ami knowing the satk thnrst at 40 Hz to be 50
grams, it was a relatively minor task to obtain the required curves. These are sbwn m
figure 5.1 1.
Extrapolated ZForee vs. Angle (40 Hz)
Of course, due to the chaage m Dapphg fiequency to 40 H z , the advance ratios were altered accordiigiy. It is this step that emphasizes how J is a iequency dependent
variable, as the 40 Hz extrapoIated d
From testing
experience, it was mtuitiveiy beLieved that the f h p m q effct wodd b e niargmal as the
x forces were feh to be largely due to a sectional area drag, and y moments about ttie
Ieading edge of the wing were essentiaiiy a by-product of these drag forces. Intuition of
strain gauges on the balance was damaged (kely due to fatigue M u r e ) . Enough evidence w a s present however to conclude that the effect of equency on the lateral forces and Y moments w a s not significant.
1
I
5.4
essentiaiiy educated guesses as to what type of aerodynamic perfbrmance could be expected fiom the MAV. Early test nms with this estimateci data showed the aircrail to be
unstable, and therefore it was important to determine (br better of for worse) what
degree of instabiiity truiy existed. This section briefiy compares the differences between the estimated and experimeatal data.
The most convenieat comparisons can be made between the eqerhm&d resuis
recordeci at an advance ratio of 0.55, and the a s s d vahses for J = 0.50. The ciifferences
encouniered bmamn ihem were astonishing. Both lateral forces and monients dfkxd by neariy two orders of mgnitude. One couid infr h m these substantiai increases,
parti'cuiarty i n the
shaiiow @es
forces t o the MAV i f it were disturbed fiom a steady hoverhg position. Figures 5.14 and
5.15 illustrate the moment and lateral force cornparison t o the initiay.a s s u d values.
X Force Comparison of Measured (J = 0.55) v a h u m e d (J = 0.50)
1
40.0
1-ksuedXForceJ=0.55
35.0
1
I
-.
1
I
I
I
1
3.0
la0
150
I 290
1
I
Angle (de91
1 -Assumed
Y Moment J = 0.50 j 1
With respect to the Z forces, the assumed values feii more m h e with the actuai
(albeit amplined) data Yet a ciifference between them was stdi readiy apparent, Both
were nearly linear in shape, but thei. dopes dBired simcantiy. This simply meant a
also important to note that the assumed values showed a considerable thnist surplus (to a
value of roughiy 80 grams), which also gave a discrepancy.
--
Chapter 6: 2 0 SIMULATDN
6.1
Numerical Model
in order to simulate the MAV dynamically, Newton's iaws were appiied directly,
using a mathematid mode1 coded m the MATLAB version 5.0 programming language.
&. The distance 11 represented the Iength betwea the vehicle centre of gravity and the
leading edge of the wings. This parameter was d d k d by the fact thai al1 forces recocded
during the experimentai testing were resolved about the wings' leadimg edge. I n the same vein,
h represented the dktance fbm the W s quarter chord to the vehicle centre of
Pvity-
z
Flight Path from Origin
. , .,
\,
longitudinal axis to coincide w i t h how the data h m Chapter 5 was recorded. The same can be said of the drag force h m the vehicle's wings b and Dm of course reptesented
Values for thnist and drag of the MAV were necessary to properly calculate the vehicle acceIeration at each tirne step. This was performed by using Iookup tables generated h m the experimental results of Chapter 5. Smce these results depended upon
both the angle to the k-stream and the fke-stream velocity, a double interpoiation
a matter of resoiving them appropriately into the global coordinate system duriag the
summation of forces and moments upon the vehicle.
tyjically only displaceci h m the vertical with ail other conditions zero). The second step
was to use the lookup tables to evaiuate
forces iiom the tail were determined fiom a mathematical equation taken h m a trend Sie placed tfrough its CLVS. a and CDvs. a graphs (see Appendix D). These coefficients were then simply muitipiied by the tail area and dynamic pressure to give totd Iift and drag h m the t a i l Step three mvoived evaiuatiog the angular acceleration of the MAV body through a summation of moments, which was then mtegrated twice over the time
step to get the new anguiar displacement, The dynamic equations goveming this step are
Iya = D* II
(6-1)
anCui = a*dt +
(6-2)
en,
= uacw*dt + 00ld
(6-3)
where a represents the ringular acceImtion about the vehicIe over the tirne step dt, o is the angular velocity, and 8 i s the angular isphcement 6om the vertical. 1 , represents the moment of inertia of the a i r c d about its y-axis (see figure 6.2), D denotes drag force
and 1 ,i s dehed as in figure 6.1. The "old" subscript refers to a variablets integrated value
at the end of the previous tirne step, whereas the "new" subscript indicates the updated value of the variable at the end of the current time step.
Similarly with steps four and tive, the z and x acceleraions were calculateci and
integrated to yield the new velocities and positions at the end of the time interval. Referring to the global coordinate system iliustrated in figure 6 2 , the dynamic equations goveming this step were
(6-4)
= a,*dt = v*,*dt
+ Vzold
+ &Id
(6-5)
(6-6)
Z n ,
(6-7)
(6-8) (6-9)
+ v,
X m = vxncw*dt + &id
where a,
and a, are the h e a r acceierations of the vehicle, v, and v, are the hear
velocities, and z and x are the total displaced positions of the vehicle fiom its initial position. Again, subscripts "OH" and "newnrefr to the vaiue of the variable at either the end of the previous time step or the newly updated vdue at the end of the iatest time step during the integration process.
At this point d i the variables had k e n updated and the process was repeated for the next increment in the. A complete iisting of the code i s included (with conments) in Appendix E. One procedure m the program that required carehl h u g h t and planning deserves some elaboration here. This was conceniing the method for evafuating the magnitude of the fie-stream velocity and the angle of attack, which had to be assesseci with respect to
both the leading edge of the wiags and the quarter chord of the tail. T h e fk-stream
velocity was simply the magnitude of the resultant vector generated by the x and z velocities at the current time step. Note that due to rotation, the h-siream velocity at the tail would not be qua1 to that at the leading edge of the wings. A more cornplicated scheme was required however in determinhg the angle these vectors made to either the wings or tail. Taking the inverse tan of the ratio of the velocity components was not suficient to defme the angle corredy under al1 vehicle conditions. For example, since the experiments m Chapter 5 were perfomred with al1 crosdow angks measured with respect to the longitudinal axis of the vehicle, it was necessary to define the ee-stream angle for the MAV wings m the same way. Hence, the "me" angle to the b s t r e a m incorporateci mt only the inverse tan of the velocity components, but aIso the current tirne step's tilt angle h m the vertical (in the global coordiite system). Figure 6.3 better iiiustrates the situation.
VeIoci@Angie
The figure shows the vehicIe in a typicai displaced codiion with horizontal and
vertical velocities together w i t h a tilt angle of 0. The angle of interest i s Ob, which is tlme
angle the resuitant velucity vector makes with the IongitudinaI axis of the MAV. The
o d y hoid true for the above situation It was neessary to wosider the various
combinations of angies and velocities (in both the positive and negative sense) such that the correct Oa wouid be calculatedevery time. A mire complicated example is illustrateci
i n figure 6.4.
values of x and z velocities. The angle the total velocity vector (Vd)makes with the longitudinal axis O f the vehicle is Qf, = 90 9,
+ 8.
Sirnilar cases can be made with x and z velocities king negative together with
positive and negative tilt angles. Seven general cases were necessary to encompass al1 these possibiiities, a d are included in the program in Appendix E. In the case of the tail,
ody six such cases were necessary.
More effort was still required in order to resolve the forces into the global coordinate system used by the program. To correctiy d e t h the appropriate signs for the
drag, thnist and Lift forces, one must know the tilt angle and the direction of the kstream velocity. For example, with regards to the wings the direction of the drag force
wouid aiways be oriented in the same direction as the ke-stream. Also, the tilt of the vehicIe to the left or nght of the vertical wouid determine which direction the wings'
thnist component wouid be orienteci m the x direction. Figure 6.5 depicts an example of
such a tlght condition.
CM = D * l i
Again, the reader is reminded of the use of lookup tables in the simulation
procedure. AU drag and thrust values were recordeci as positive, and there was no mathematicai formula to reIy upon to take care of the sign convention. Thus the onus fell
upon the author to ensure any possible combination of velocity and tih angles
encountered would always sum the forces wih the correct signs.
The tail forces relied on an angle of attack caldation, and thus W Korientation
relied on a cornbition of x and z velucities aad It angle. This necessitatecl eight separate cases, each involving a specinc mmbiaation of x and z velocities and tilt angles that in turn yielded a unique force or moment summation. See figure 6.6 for an example.
Tail t in
summation of forces and moments for this particular scenario wouki become
XF, = L*sin' + D*sint CFz = L*cose' D*sin' XM = (L*cos 6 + Dssin6)*12
(6-13)
(6-14)
(6-1s)
where l2 is the distance h m the quater chord to the vehicie c.g. as depicted in figure6.1 and 0' in this nstance is dehed as 90 - 0 - Os degrees. Again, caution had to be taken to ensure the correct sign convention was obtained d e r ail possible combinations of 0 and linear velocities. The reader i s referenced to the code in Appendix E F o r firrther details to gain insight on this procedure.
On a f i n a l note, it shouid also be mentioned that the code used Euler inegration,
and the program's results were checked for convergence by coritinually haiving the tirne step util no perceptiile changes in the output couid be obsaved.
6.2
Initial Resulfs
6.2. i Simple Hovenng Condition The code was rst used to analyse the MAV under a disturbed condition with and without a tail. In both cases, the vehicle was placed into a hovering state (where thnist equaiied the vehicle mass), but with a 2 O initial disturbance fiom the vertical, The ensuing motion was found to be a steady oscillation between
decayed sigdlcantly (ie., neutral stabiiii). This occurred regardless of the presence of
the tan. Apparently, the force of the MAV wings were much larger than those of the tail,
mainly because the k-stream velocities at the tail w e r e low which, in turn, reduced the amount of dynamic pressure. Oniy if the tail was made ridiculously large did one begin to
see its influence on the system. An example of the oscillation without the tail is show in
figure 6.7.
7
-301
i
1
1 0
Tirne (sec)
R g v r e 6 7: lilitial T e s t Glse W d h t T d
nieans
reaiism, as one wouid expect some sort of convergence or divergence if the vehicle was to be truly placed under such conditions. Hence, it became a matter of h d h g a way to incorporate such an element of reality.
6.2.2 Rotational Dise Damping Mer discussion with Dr, DeLaurier, it was felt that the code lacked a disc
damping tem. T h i s damping would be due to the physicai act of ''tibg" the hppingwing disc plane about the y-axis, which in turn would become a sink for retnoving energy
from the system It made sense that such an effect would be absent m the static tests
morrned in Chapter 5, as it was a dynamic property of the vehicle. The term wouid appear during the summation of the moments on the body as sorne yet-unknown coefficient multiplied by the angular veiocity of t&e vehicle. The task then was to determine experimentalIy the value of this unknowncoefficient.
countemeight
A
1 2
pendulum mass
6 . 3 . 2 Dynrmic Equations
The oscillaocy equations of motion for a peoduium are WU documenied in any
dynamics or viiions text. As seen h m figure 6.82, three components serve to
dampem the motion, Bearing fliction and aerodynamic drag were combined h o ooe
prameter (labelled F in the d i ) which, a s a first approximation, was multiplieci by the anguiar velocity of the apparatus to caiculate tbe resistance. More important, however, are the two remaining damping parameters. Daqing due to the bbsurging" of the wing disc area in the z direction was labelIed quai to Ci2 The damping due to %itingW of the wing disc a m about the y-axis was, m turn, labeiied Cz. By sumrning the moments about the pivot point in the perturbed condition, it is hund that
r,b@
- mg*h*sin 0 - C& - ~ 2 - 8F 6
(6-16)
with 1, being the mass moment of inertia about the y-ais, m king the mas of the
pendulurn weight, and 11 and h as defineci in figure 6.8.1. Simpli@ing through the use of the small angle approximation and letting 'z = It0,
e
where
In the most general case of such an oscillatory system, fiom reference solution to the dflrential equation written in (6- 17) is given as
[q the
The bracketed terms are responsble for the osdations m tbe system, while the exponentiai term outside gives the damping. E t is eady m u that in the absence of the damping terms Ci, CZand F, then D would @zero and t6ere would be no exponentiai decay in the soiution. Thus it becornes a matter of empmcaily determinhg the exponent in the patameter e4"' and solving for the unknown coefficients of interest.
6.3.3 Experimenf
The experiment was set up sunilady to figure 6.8.2, and a photo of the appamtus i s shown in figure 6.9.
h was caicuiated by detefmiLUng the centre of gravity of both the aiuminum arm and
penduium mas together with respect to the pivot point. Attached at 90" to this ann was a
slender steel rod PmtoSordh was attachai to t kend of the rod in such a way as to allow
it to siide up or down the length, thus allowing variance m 1,. A compass and nede were
mounted above the pivot point so that angular displacements could be accurately measured.
In order to monitor the osciilations precisely, each test was videotaped using a
Canon digital video camera. Knowing tha there were 30 fiames per second and replaying
the video in a h
e by h
produced. An example of such a plot i s s b w n in figure 6.10 for the tare damping of the system As mentioned, the envelope of decay in the oscillations i s govemed by the term e4DR''.Plotting the upper peaks of oscillation alone and adding au exponential trend line
in Microsofi Excel determineci this value of interest. Hence, it becarne a simple matter of
taking the exponent, equating it to D/2, and solving for the unknown parameters.
30.0
Prior to testing, it was wcessary to calculate the system moment of mertia (I,,), which can be found in Appendix F. T h e f b t test then hvolved the caiculation of the loss parameter F of the system, T h i s parameter wuid be viewed essentially as the ~
'
'
damping in the system, and was simply measured by obseMng the pendulu.motions in
the absence ofany wing flapping. W i a solution for F, the next s t q s were to observe
the decaying oscillations while the wings were flapping. Since there were two remaining
unknowns (Ci and C2), it was necessary to generate two separate equations. Thus the expriment was performed twice but with a different It during each test.
6.3.4 Resulfs
Complete raw data for the disc damping tests are included in Appendix F. It was
i to discovered that F had a value of 0.00 1688 N*m-s/rad,and the ensuing tests revealed C
equal0.0002673 N.s/rad/m and C2 t o be 0.001734 N-mdrad. In terms of the simulation
code, the effects of surging (represented by CI) should have aiready manifested themselves in the lookup tables, as this was simply the motion of the wings translating through a flow fieId. Therefore, the newly implemented parameter was C2, which would make its appearance in the summation of moments about the vehicle cg. at each time step. Its effects were substantiai, and discussed in the sections to foUow.
6.4
Case Studies
n hopes of establishing a stable mterest, and then varying the geometry of the aircrafl i
configuration, as weU as gainhg insight into the behaviour of the vehicle in tlight. Smce the number of possiile variations was nearly idhite, it was decided to focus on situations where the MAV was at or near the hoverhg state. This was the flight reghe
for which the MAV was m o s t intended, and so it seemed a nahuai flight condition m which to investigate.
A total of four scenarios were devised, and each was evaluated both with and
without the presence of a tail. The geometry of the vehicle was initially taken directly fiom the early 15 cm span k e flyer whose body moment of inertia and centre of gravity location were determineci fiom a Solidworks mode1 of the MAV. The first scenario consisted of a typical hovering condition with a 2" initial disturbance h m the vertical,
with ail other initial conditions zero. The second case would involve the MAV i n a slight
ascent (by simply by lowering the mass) and disturbhg the MAV by 2' fiom the vertical. Similarly, a third case would put the vehicle into a slight descent together with a 2" tilt
disturbance. F i y , he fourth case wodd simulate a lateral gust of 2 m/s with the MAV
initially unperturbeci Eom the vertical. In al1 cases the motion of the vehicle would be observed to determue if d converged to a 0" vertical displacement. Divergence was certainly a possibiiity, and indeed it was hoped that if such situations were encountered
that they could be remedied with appropriate modification to
configuration. Parameters that could be modified included the vehicle's mas, the tail geometry, the taii's position above or below the wings, and the cg. position with respect to both the
wings and tail. Each case wouki have the MAV begin m what the author defines as the
"standard configuration". This meant that the Ieading edge of the wings would be 7.5 cm above the c.g. of the vehicle (labeiled 1, m figure 6.1), which geometrically msitched the prototype drawn m Solidworks. From this initial configuration, each case study wodd be
run and the above parameters wouM be mdited to observe their impact on stability. in
al1 cases, the parameters were never increased beyond the 15 cm maximum dimension
estabshed in the MAV project requirements.
disturbed by an angle of 2" and observing its ability to right itseif to a steady hovering state. Ail other initial conditions were kept at zero. The resuits for the vehicle without a
taii in the standard configuration are shown in figure 6.1 1.l.
Theta w.Time
Time, sec
vertid disphcement. It then becarne a question of determining the effect of changing the
79
c.g.
location with respect to the leading edge of the wings (11). Reduing 11 to 3.5 cm
revealed improved convergence and indeed with a value of 2 cm there were even better results. Figure 6.1 1 2 shows this trend caused by the reduction o f 11.
Theta W. Time
10
1s
Time. sec
of the projeted wing area was below the c.g. Anaiysis of the moment and faeral force
data obtained h m the wind tunnel experiments maleci on average that the wings' center of pressure was siightiy less than 2 cm below the kadhg edge. Values of Il
s d e r
tban this would yield misleading results. For example, the simulation would
resolve the drag force acting through the leading edge of the wings 0.5 cm above the c.g., even though the center of pressure was obviously below. Hence caution would be needed to take in the interpretation of such results. Instances in which 1, extended beyond 7.5 cm only proved to be less satisfactory than those in figure 6.1 1.2 Continuhg one step M e r and truiy placing the wings below the c.g. added nothhg to aid stability. In k t , the vehicle immediately began to diverge catastrophically. The next step was to evaluate the effects of the tail. Under such conditions it intuitively made sense to place the tail above the wings in hopes of enhancing convergence. Since the properties of the two tail designs tested m Chapter 5 were closely matched, either wodd suEce in conjunction with the code. For al1 the case studies, tail #2 was chosen to cornpiete the analyses. The tail's area was made 0.007 mZand placed at an initial distance of 12.5 cm (h) above the c.g., but its effects on the performance of the
MAV were truiy negiigible. The results were almost an exact dupiicate of the
performance without a tail (note that Il was kept at 7.5 cm in thk case). Similar tests at distances of 15 cm, 1 cm, and even below the c.g. remained ineffective. T h i s was because
the velocities encountered by the tail were very small and hence theh abity to produce
aerodynamic forces were limited by a lack of dynamic pressure. Further investigation showed the tal drag forces to be four orders of magnitude greater than those of the
wings.
Doubling the tail area (to 0.014 m2)and combmmg this with the above d u e s of
12
h i s wouId mvolve a tail d c e area more than double the size of the investigated, as t
wings, a tnily ridiculous notion (To aid tbe reader conceptually, the tail area of 0.007 m2
would be an area siightly d e r than that of the wings). Thus it was conciuded that m
6.4.3 Case II Slight Ascent wifh Tilting Disturbance in this study the
niass
Under the standard configuration without a ta& the vehicle began to diverge noticeably
d e r 6 seconds to a peak displacement of roughly 9.5'. after which the oscilIation neither
grew nor decayed. However, lowering the distauce 1, produced profound stabilising
effects, as evinced in figure 6.1 1.3. At lrther distaitces of 3.5 cm and 2 cm, the abiiity to retum to a 0 tilt angle was even more effective.
The addition of the tail to the standard configuration (11 = 7.5 cm) only made the
response worse. In this instance, the tail area was again set to 0.007 m2and h made 12.5
5
Tirne, sec
10
Figure 6.11.4: Case Ll- Wuh Tail, III= -12.5 cm, 1, = 7.5 cm
This noticeable improvement comes with a caveat however, as the tail would be
situated i n the region of downwash of the wings. This phenonnon was not modeUd in
CeMvi f there were some form of active contml mrfkes on the tail, then this type of
configursttion could be beneficial. But due to the its passive nature, however, conciusions
and combining hem reveald an overd bene& i n performance. That is, with I r set to 2
cm and l2 set to -12.5 cm, the vehicle dispiayed the best convergence trajectory of a i i as
me, sec
Along the sirnilar nes a s in Case II, the MAV mas was inreased to a value of
52 grams in order to impose a ttinist deficit and thus mate a siight descent. In addition, a
3" vertical disturbance was added, with d i otber initial conditions remaining at zero. in the absence of a tail under the standard configuration, the vehicle showed a converging oscillation. As was seen m Cases I aml II, redmtion m the value of 11 produced better
results. Likewise, placement of the wings bdow the c.g. caused diergence. Figure 6.1 1.6
displays he effects of demashg II.
Theta u s . Time
2
1.5
1
d
CD
4 0.s
a i -
0.5
k!"'"L
-1
, 10
Time, sec
15
Wiih the addition o f the taii (hawig the same geometry as that in Cases 1 and iI),
the motion was found to be only siightly better for values o f h king eitber 12.5 cm or
-12.5 cm. The results for both cases are shown in figure 6.1 1.7.
Tme, sec
The same caution must be taken here as i n Case II with respect to downwasti
effects. Upon wmbining the best fiom bot&scenafios (with and without a tail), one fhds
10 Tirne, sec
-- - - - -
Figure 6.11.8: Case L With and Without Tai, lz= 12.5 cm, 1, = 2 cm
To rationalise the minor effects of the tail, it was again determineci that the taii forces were many orders of magnitude smaller than those of the wings. It was concluded that under this condition the best performance would be achieved without a tail; however
a tail's presence would do nothhg to M e r an e v d convergence.
In this 6nai case shdy, the MAV was distrrrbed fiom a steady hovering condition
by a laterd gust of 2 mis. Ttiere was
M>
conditions were set to zero. in the absence of the tail, the standard con.fguration showed
a remarkable ability to right itseif der the disturbance. It is noticed h m figure 6.11.9
t'urther t o 2 cm
Time, sec
T h e addition ofa tail (ofsame geometry as the previous cases) above or below the
wings had negligible effcts, again due t o minute taii forces. The vehicle retained its
abirity to converge. i n this case,the best performance occurred i n he absence ofa tail.
Chapter 7: CONCLUSIONS
7.1
judicious placement of the vehicle's c.g. position a stable design was entirely possible without the presence of a taiL Hence, future MAV prototypes should stress component layout such that the cg. falls much cbser to the wuigs' leading edges, preferably at
distances near 3.5 cm. The addition of a tail thetefore may be viewed as optiod, and
indeed its presence for the most paa did ttle to augment the inherent stability of the
system. In the interest of saving weight, this appears to be a t d y bemficial
i t h any simulation, and real worId characteristic. However caution must be taken w
experimental analyses would certainly be required to estabtish the vdidity of this
statement.
The final MAV prototype will obviously need a method of flight controi, and thus
some sort of active control surfaces will have to be incorporated. It is cornforthg to kmw
that the presence of a tail in the conditions descn'bed in Chapter 6 does not hinder the
vehicle's abiiity to stabilise ilself d e r a disturbance. The case studies reveaI that the best taiI placement would be below tfme c.g. at a distane of 12.5 cm to the quarter chord of the
k. This would be coupIed with the wings' leadmg edges placeci 2 cm above the c.g.
This Iayout, under ali case studies, was a weU balanceci configuration with respect to
overall performance. Also, with the taiI piaced in the downwash of the wing thrusr, it could be suggested that better pedonnance of the control surfices wouid b e encotmtered.
Improvements that could be made in this study e w g e primarily i n the area of the wind tunnel velocity profiIe. It is believed t h t the industrial type fn used in the wind tunnel i s the Likely culprit in the non-uniform flow field. Another recommendation could
be made towards increasing the sensitivity of the s W gauges, as some of the data was
obtained in their lower range. At the tirne of the experiments these were the most sensitive gauges available commerciaily, but newer versions may have ernerged since that time.
This provides the necessary closure for this body of work, and lends optimism to
the realisation of a stable and controable fiapping-wing MAV. Future research should focus on the development of an actual flight vehicle testbed fiom which observations of stability can be made. Appropriate modifications should then be applied based upon both the observed flight characteristics of the vehicle and the conclusions d r a m Eom this thesis. Work should also continue with the 3 dimensional simulation code developed at
SRI using a sirnilar case snidy analysis perforrned in this document. Its conclusions
should be compared to those of the 2 dimensional code to see what descrepancies may exist between them Outputs of both programs under similar initial conditions should in the very least be cotnplimentary. In closing, it is sincerely hoped that the hdings herein wiii benefit the ture
References
Bilyk, Derek. The Developnient of Flapping Wingsfor a Hovering Micro Air Vehicle. University of Toronto Institute for Aerospace Studies; Master of Applied Science Degree, 2000.
111
El-Khati'b, Jasmine. Flow Msu(11isationfor a Micro Air Vehicle. University of 121 Toronto Institute for Aerospace Studies; Master of Applied Science Degree, 2000.
Loewen, Dave C. An Experimenfd Investigation of Closely Spaced Membrane Aifloils. University of Toronto; Bachelor of Applied Science Degree, 1991.
Pl
'Mode1 6000: Planar-Beam Force Seasor." Advanceci Custom Sensors Inc. 19 141 July 2000. < h t t p ~ I ~ ~ ~ . a c s e n s o r . w m l P a g ~ o d e l - 6 0 0 . Thompson, William T. and Dahleh, Marie Diiion Throry of Vibrationivith Applications. 5 ' ed. Upper Saddle River, NJ:Prentice Hali, 1998, p. 27 - 3 1.
FI
8.2
Bibliography
SRI International, UTIAS. FIapping Wing Proplsion Udng EIectrosiriclive Polymer Artifcial Muscle Actuators: Semi-And Report. 14 December 2000.
Fiuke NeetDAQ Data Logger User's Marnral, Fidce Corporation, 1995.
Anderson, John D., Ir. Introduction ro Fighr. 3d ed, New Yotk: McGraw-Hill, 1989.
selected rather arbitrarily, with the provision for adjustment should the need arise. The essential design was a scaled dom version derived fiom an existing cantilevered balance residing in the UnAS subsonic aerodynamics lab. The CAD drawings that depict tbe overail dimensions and layout of the design are included on the foiiowing page. Al1 measurements show are in mm and the drawings are not to scale.
series of test cases were perfonned. These wodd determine if there were any adverse effects on the gauges, or their behavior would change under d i i e n t load combinations. The foilowing appendix is divided into sections correspondhg to each different test case. Each section wiii include a description and summary of the caiiiration results.
beam balance. As mentioned m the main body of the text, this particular orientation
compensated for load misaiignments, allowhg for the measutement of pure forces ody. For simplicity, the author will refer to a parailei beam configuration as a single unit and cal1 it a gauge. Hence, three of these "gauges" were mounted on the balance. Prier to their final attachment to the lower plate, these gauge units were individually dihatecl through the appiication of k w w n masses and the recordmg their voltage outputs. This
was performed by using a Keithley 177 Microvott DMM (digital multimeter) m
the appiied input voltage was 10 VOL. This value was monitored both before and afler
the c a i i i i o n tests to ensure that it did w t drift appreciably during the readings. Proper
strain relief of the leads extending fiom the gauges was crucial in order to obtain
y to a rigid surface
so that they could not move during caliibration. Known masses were applied to each gauge by using an attached looped thread.
The thread passed over a pulley and ended in a hook onto which these masses were hung
in the positive axis direction ody. Aii gauges exhibiteci Iinear behavior, with the dope
curves shown below. The k-values correspond to the siope of the trend lime equations show on the graphs.
fist load condition to determine these new k-values was an application of a pure load
dong a single axis only (in both positive and negative directions). Loads applied dong the z-axis were considered to be shared equally between gauges #1 and #2. Under these conditions another set of k-values were determined. Linearity with respect to Ioading was tetained, with gauge #1 yielding a k-value of 0.0541 mVlg in the positive z direction and
0.0563 mV/g in the aegative z direction. Gauge #2 produceci a value of 0.0871 mVlg and 0.0820 in the positive and negative z directions respectively. Finally, Gauge #3 gave k-
values of 0.516 mV/g (positive x direction) and 0.0528 mV/g (negative x direction).
Upon cornparison between these and the previous independent tests, one can immediately
deduce that there was ttle effect on the gauge slopes due to their final attachment. Thus,
the initial assumption of equaiiy shared Ioading between gauges #1 and #2 was
considered vaiid. It i s also worth noting that tbere was no crosstalk observed. The hiIo wing graph iIlustrates the results of this test.
Wi the initial set of k-values determineci h m case 2 above, it was then desired
to apply a variety of load conditions to see if any appreciable change occurred in the gauge slopes. Ideally, these values should not change- These extra tests however, would
h i s was important, as the loading shed light on the overd behavior of the balance. T
conditions expected durhg actual testing would be quite variable. By means of a small arm attached to a vertical post extending fiom the lower tray,
a pure moment was applied to the balance. Loads were attached at various positions
dong the length of the arm to aIlow variation in the moment's magnitude. The results of the test are shown in table B-1 below. An error anaIysis was performed using the k-dues iiom case 2, and revealed most of the mors did not stray fat. kom the 5% value. In addition, k-values could be derived fiom the test condition using a system of equations as follows: let AVl = change in voltage of gauge #1 between loaded and unloaded conditions, and similarly let AV2 = change in voltage of gauge #2. ktine xl and xt as AVIIm and AV2/m respectively, with m king the apptied mas. The distance separatiag gauges #1 and #2 is labeled d. Together, this data reduces to a system of two equations
(2)
These are easily solved hr simuhaneously for the imknowns XI and x2, which in
turu are the inverse of the gauge k-vahies. These equations w e r e used for each moment
apptied in the test. Hence, since seven different moments were used, seven different dopes could be deduced. An average of these values reveaied that the k-value for gauge
#1 was 0.0546 mV/g and gauge #2 was 0.0897 mV/g. On account of the load orientation,
no component of the force occurred in the x direction, and therefore no k-due could be deduced for gauge #3. One final comment should be made on the caiculated slope of
h i s number m e d gauge #2 for the instance where the applied moment was 322.4 g c m T
out to be 0.2238 mV/g, which was decidedly out of sync with the rest of the calculated
slopes for gauge #2. It was therefore not included in calculating the overail average and considered an anomaly due to the hi& moment Ioading on the gauge. It was t d y unlikely that such a large moment would be encomtemi in practice.
an angle to the center of the lower plate. The appIied force was simpIy reduced mto its
component vectors m order to detemine the forces aiong these orthogonal axes. As before, use of the initiai k-dues h m case 2 gave percent mors m the m g e of 5%, with
poorer performance occurring only in the extremely tight loading condition. The test
resuits are depicted below.
62.2
46.8
-21.7
-23.3
64.3
3.3
-45.0
4.0
Simply plotting the gauge outputs vs. mass showed the k-values to be 0.0524
mV/g for gauge #1,O.O8 15 mVlg for gauge #2 and 0.0537 mV1g for gauge #3. These are shown in the fo Uowing graphs.
OZ-
92-
This test was a dupliation of case 4 except the load was off center of the plate by
use of an
components. The results are summarized below, and reveal errors (using k-values fiom
case 2) no greater than 5%. Again, average k-values were bund to be 0.0554 mV/g for
gauge #1,0.0829 mV/g for gauge #2 and 0.0539 mVIg for gauge #3.
33.8
33.6
4.9
4.3
0.0 1.8
2.3
33.4
33.6
3.7
0.7 1.0
0.7
33.8
34.0
-24.5
-24.4
' l n al1 cases the applied mass was -40.4 g. Placement along the bar uttachment uilowed variation in the appked moment.
0.0562
-0.2144 0.0681
0.0700
0.0688 0.08#1**
0.534 0.537
0.0655
0.0549
0.0540
0.0543
0.0639
0.0538
Average:
0.0554
*In cuch i m n c e rhe applied Z Mas WPT -242g und the applied X M m was 32.2g. as indicared in Table B-3. T h e vaiues 0.0688 and -L?l.(1were not includedin culculution of the average.
prelirninary calidnation tests, with the remainder performed by the author. A drawiug of
/ crosssection
*Each station helght was separated by a distance of 2,M om (7 in.) and each station wldth by 3.81 cm (1.5 in.).
a i n body of this tests performed on both the BAT- 12 w i q s and tails, as descri'bed in the m
document. Each table gives the data accumulateci for one advance ratio. Each table is Further divided into sections for IateraVlongitudinal forces and y moments. As each advance ratio was repeated h e e times, a final column correspondhg to the test average
was used to determine the overd trend of each force or moment vs. angle to the
crossfiow. Following these tables i s another table giving the amplified data deduced for the z forces. Finaily, the 1st table contains the CJCD data obtained for the two tail designs.
CO J's cf O , ~ I . L C , 9 . 5 5 , 0 . i 5 and 0 . 7 3 5 and a s h row f o r a n g l e s O t h r u t o 180 d e g r e e s - 7 - lookug t a n l e as above, c x c e p t f o r t h e t h r u s t vaiues of t h e wings :heta - a n g u i a r d i s p i a c e m e n t from Che vertical, d e g r e e s - - - a l i i t u d e (metzes!, x - i a t e r a l d i s p l a c e m e n t i m e t r e s i , a l p h a angular accel.irad/sa2i cmeqa anguhar v e l o c i t y Ired/si , x a o t l a t e r a l v e l o c i t y !rn/s!, xdd - l a ~ e r a : accel ( n / s " Z ' i zd - v e r t i c a l v e l o c i t y ! m / s ) , x!d - v e c t i c a l accel. [rn/sa2) t h e t a v - a n g l e t h e f r e e s t r e a r n makss w i t h t h e h o r z . d u e O : boch z r x v e i c c t l e s ac l .e. of r i n g s L - d i s r a n c e fzom che c.g. t o t h e 1.e. of Che wings im: in - n a s s of Che v e h i c l s : k g ] , y - inass rnomen: o f i n c e r i a anout t h e y a x i s , k g m"2i rho - air i s n s i q : k g / m " 3 ) , 5 = t a i l s u r f a c e a r e a ( m A 2 ) , cl - d i s c damping c c e f f i c i e ~ t :!cg n A 2 / s ) x v t - t c t a l l a t e r a l v e l c c i t y a t 1 . e . o f t h e wings d u e t o : r a n s l a t i o n and r o t a t i o n :m/si z v t - t o t a l v e r t i c a i v e i c c i t a t 1 . e . o f t h e wings d u e CO ::anslation and r o t a t i o n Im/s I c h e t a f s - maqnitude o f iree Stream a n g l e t o t h e l e a c i n g edge of t h e wlngs x r t t h e v e r t i c a l ! d e q r e e s i Thr - t h r u s t from wings :gi , Drag - d r a g frcm wings ( g i .- . . . -.- * ... ~- - * - - - - - . * * - - - - -.- - ..- . . . . . . . . .-*.--*..**".* ' . - - * - - - - - - . . - ~ . - - * - - -.---.
. -
- p o p u i a t e lookup =abLes
.89 0;7.85 6.98 4.93 3.40 0;15.88 12.94 10.10 6.02 0;23.81 18.4 14.46 7.7 0;29.58 21.79 17.36 9.79 0;33.23 24.6 19.33 10.67 0;35.6 26.4 19.33 10.67 0;35.6 26.4 20.96 12.01 0;37.03 27.01 21.41 12.56 O; ... 35.59 25.81 21.1 11.13 0;34.74 25.28 20.64 9.21 0;34.02 24.66 20.41 10.08 0;30.92 22.52 17.92 10.27 0;27.35 18.07 16.08 8.61 0;24.24 15.25 12.53 5.41 0;20.63 12.23 9.43 3.12 0;16.06 8.63 5.14 0.74 0;10.22 5.07 2.39 -.O6 0;4.99 3.29 1.06 -.6 0;-1.56 -95 1.43 -.66 O];
-specify initial conditions theta=(2/180*pi);z=0;x=O;xdot=O.O;alpha=O; xdd=0.O;thetaplot(1)=theta/pif18O;xdotplot(1)=xdot;alphaplot(1)=alpha; dragplot (l)=O;thrplot (lI=SO;tplot(l)=O;xpIot (l)=O;zplot (l)=O; omegaplot(l)=0;velplot(l)=0;thetav(l)=0; cl = -0.0017345: -orner variables l=O.OiS; dt=0 -001; m=O -050; Iy=0.000031; -se loo~ptobles to evaluat t h e curreat D, T, using xdot as -refrence veiocit? and reference rheta and assuming al1 flapping done
-at 4gRr
-designate Che velocities "heaaings" for eacn marrix column in lookup - raDles vl=7.04;v2=6,28;v3=5.24;v4=l.87;v5=0; thetav(1)=O; -beu:n los-ing chrough tirne steps tplot (1) =O;i=l;q=O; -Icc beins 3t dt, and al1 variables ger updated based on avg.
-accelerations -thr=ugh r b a t Fncrenentai rime step. The values a t t h e end of Srepresnt Lhe vaiues sf zhat v r i a b l e at ~ n d ttirne at.
3
stec
fsr t=dt:dt: 15, i=i+l; tplot (il=t; - t c z a i x velocity at leaaing edge of wings x v t (il=xdoc+omegaf 1-0s Cabs (thetal1 ; -cotai -~el=city at leading edge of wings zvt (il=zd+omega*l*sin (abs( t h e t a ) ) ;
- d e r e m n e thetav, cne angie zhe free Stream makes with tne h o : z . -due to 50th vert & korz translations ( x v t and zvt; F f zvt(i)==O & xvt(i)==O, -prevent undefined 0 / 0 when using arctan function! thetav(i)=O; eise thetav(i) =atan (zvt(il/xvt (i)1 ; end;
<=O if z v t (i) if .wt(i)<O, if theta<=O, thetafs=pi/Z+theta+thetav(i}; eise thetafs=pi/2+thetav(i) +theta; enci; end; end; if zvt(i)C=O if xvt(i)>O, if theta>=O, thetafs=pi/2-theta-thetav(i);
end;
end;
thetafs=pi/2-thetav(i1-theta; end;
end;
end;
-case
etc;
end; end; ~f zvt(i)< 0.00001, ~f zvtti)>-0.00001, if xvt (i)<0.00001, if xut(i)>-O ,00001, thetafs=O; q=q+ 1; ena; end;
nc ;
ena; tfs (i) =thetafs*lEO/pi; -thetafs is the H?GtITUDE of the angle between the free stream -and the vertical wrt 1.e. of wings, *determine uhat range the velocity falls under -the higher veLocity calumn Vel=(xvt(i) "2+~vt(i)^2)~.S; velplot (il=Vel; if Vel<vl h Vel>=v2, columnl=l;column2=2;vlow=v2;vhigh=vl; end;
columnl is always
i f Vel<v2
&
&
col~l=3;column2=4;vlow=v4;vhigh=v3;
angle = abs(thetas/pi+l801; ang ( i) =angle; : f anglecl0 & anqle>=O rowl=2;row2=l; end; if angle<20 & angle>-10 rowl-3; row2=2; end; rf anglec30 6 angle>=20 row2=3; rowl=4; eria; if anglec40 & anqle>=30 rowl=5;row2=4; nd; :f angle<50 & angle>=40 ; rowl=6;r0~2=5 md; 15 anglec60 & angle>=50 rowl=7;row2-6;
ena;
rowl=8; row2=7; end; if angleC80 & angle>=70 rowl=9;row2=8; end; i f angle<90 & angle>=80 rowl=lO;row2=9; end; if angle<100 & angle>=90 rowl=ll;row2=l0; end; if angle<llO h anqle>=lOO rowl=12;row2=ll; end; if angle<120 & angle>=llO rowl=l3;row2=12; end; if anglecl30 & angle>=l20 rowl=L4;row2=13; end; i f anglecl40 & angle>=130 rowl=15;rou2=14; end; if anglecl50 & angle>=L40
rowl=16; row2=lS ;
end;
- v e l s c i c y pair a c l o w e r angle VLowl=T (row2,columnl) ;VHighl=T (row2, column2); -velocity- pair for higher a n q l e vLow2=T(rowl,columnl);VHighS=T(roi~l,~0I~mn2); - i n t a r p o i a c i wich t h e s e p o i n t s a t t h e currenc u e l c c i t y ThrLow~ngle=~~owl-~Highl)/(vhigh-vlaw)~(Ve~-vlow~+VHigh~; ThrHiqhAnqle=(VLow2-VHiqh2}/(vhigh-vlow)*(VeL-v10w)+VHigh2;
- i n t e : c o i a t e =o ?et a: the desired angle Thr = (ThrHiqhAngle-ThrLUwAnqle]/IO' (angle-anglelow) +ThrLowAngle; thrplot [ i l =Thr; -F.epeat rhe same cCinq f o r getting Che a r a 9 - v e i o c i t : ~? a i r at i a u e r angit VLowT=D (row2,coLumn1) ;VHighl=D (row2,c o l u 2 ) ; - v e l o c i t y pair for higner a n g l e VLow2-D ( rowL, columnl);uHigh2=D ( rowl, c o l u 2 1 ; -icrerpoLace -&th these poincs a t t h e current v d o c i t y ~rag~ow~ngl (VLowl-VHighl) e/ (vhigh-vlow) (Vel-vlow)+VHighl; DragHighAngle= (VLow2-VHighZ]/ (vhigh-vlow) el-vlow] tVHigh2; - i n t e r p o l a t e t a get Drag a t :ne
desired angle
Drag =(DragKigh~nqLe-Drag~ow~ngle]/10*(angle-anglelow)+DragLowRngle;
-evaiate a n g u l a r a c r e l e r a t i o n , ensuring Drag f o r c e always points - 0 p p o s i t e to Che a i r e c t i o n o f zhe velocity thetaold=theta; ornegaold=amega; draqpLot (i1=Drag;
- t h e o r i g i f i a l alpha at the b e g i n n i n g of the current rime i n t e r v a l alphaold=alpha; discdampl = cI*omegaold;
if x u t ( i ) > = O , a l p h a = -Drag/L000t9.81+L/Iy+discdampl/Iy;
tlse
alpha
en;
~rag/1000*9.81*1/Iy+discd~l/Iy;
alphaplot ( i l =alpha;
- i n t e g r a t e alpha t w i c e t a get angular p o s i t i o n -once:
-total omega at the END of current time interval omega = alphafdt+omegaold; omegaplot ( i =amega; twice: -total theta at the END of current time interval theta = omegafdt+thetaold; thetaplot(i)=theta/piW; if thetaplot (il>80, fprintf ( ' 1 AM OUT OF CONTROL ! ! ! ' ) end; -evaluate vertrcal accieration zcid, and incegrace twice for -"aLticde" postition zddold=zdd; if xvt (i) >=O, if thetaold>=O, zdd=Thr/1000*9.81/m*cos (abs(thetaold) 9.81+Drag/1000*9.8i/mfsin(abs(thetaold) l g a
else
-integrate twice zdold=zd; rd = zdd*dt+zdold; zold=z; z = zd*dt+zold; zplot (il=z; zdplot (i) =zd; -evaluate horizontal accieration xdd, integrate once for x velocity, -then twice for horizontal position xddold=xdd; -use appropriate equation aepending upon vhich side of the theta -equals O nark. - m u t use theta at s t a r t of this interval to update -xdot if thetaold>=O, if xvt(i)>O, ) xdd=Thr*9.81/1000/m*sin tabs (thetaold) Drag/l000*9.81/m*cos(abs(thetaold~~-omeg*zd; else xdd~hr*9.81/1000/mfsin[abs (thetaold) 1 +D~ag/1000*9.8l/m*cos(ab s (thetaold) 1 -omega*zd; end; else if xvt(i)>O, xdd=-Drag/1000*9.81/m~os(abs(thetaold))Thr/1000*9.81/rn*sin(abs~thetao~d))-O-;
else
xdd=Drag/1000*9.Bl/m*cos(abs(thetaold))Thr/1000+9,8l/m*sin(abs (thetaold))-omega*zd;
end; end;
xdotold=xdot; xdot=xddfdt+xdotold;~thexdot at the END cf this interval xdotplot (il=xdot; xold=x; x = xdotcdt+xold; xplot (i) =x; -ail v a r i a b l s have now Seen updated. Laop. end;
figure ( 1 ) ; plot(tplot,thetaploti;title('Theta
vs. Time');
clear all; . . . . . .
.....*.*..**....*-.
.,
'.
3F K R I - U L E S
-;iL;--L;;;;;;;;-;;i-iL-i-L-----.., .. . .. . .
cf f i l , Q . l 9 , ~ ~ . 5 5 , O . E 5 6 and 0 . 7 3 5 a n d e a c h row f o r a n g l e s 0 t h = = t o l e 0
- - ,ockup -
.--
: t a b l e a s &ove, e x c e p t f o r ~ h t eh r u s t v a l u e s o f c h e wings r n e t a - a n g u l a r d i s p l a c e m e n t Ercm t h e v e r t i c a l , d e g r e e s - a l ~ i t u d c : m e t r e s ) , x - l a t e r a l d i s p i a c e r n e n t irneczesl, alpha a n q u l a r a c c e i . [ rad/s a2 1 zmega - a n q u i a r , r e l a c r t y ; r a a / s ! , xdot - L a t e r a l v e l a c i t y ! d s i , xdd - l a c e r a l a c c e i !m/sA2: =c - . f e r z i c a l v e l o c i r y ; r n / s i , zCd - v e r t i c a l a c c e l . im/SmSi xddwing - L a t e r s i a c c e l . c o n t r i b u t e c i by~ wings ; m / s A 2 : zadwing - v e r ~ l c a la c c e l . c o n t r i b u t e d by wings (m/s'2i :<ad:ail - L a t e r a l a c c e i . z o n t r i b u t e a By c a i l im/sA2i z d d t a i l - verticai accei. c o n t r i b u t e d by t a i i Irn/sAS> awing - u i q z o n t r i b u c t a n t o t o t a i anqu'ar a c c e l e r a t i o n a b o u t r h e
co t o c a l a n q u i a r a c c e l e r a t i o n a b o u t che
--
--~
zk.ecaq; - a q i t :kit free s t r e m ;riakts witb =ha n o r t . due Co bcth h ;: : i v e l o c t i e s a c 1.e. f w i n q s I - a i s t a n c e fzom the c.g. to zhe l . e . o f che wings i m ) in - zass cf =he v e h l c l e [ k g ! , Iy - mass marnent o f i n t e r i a a b o u t t h e y azis i k g m ' 2 L? - d i s t a n c e from the r . g . t o : h e q u a r t e r c h o r d o f Che t a i l ;rit) r h c - al: c e n s i t y tkg/mA3), 5 = tail s r f a c e area ( n " Z l , c l - csc d a r g i n g =aef f i c i e n t kq m A Z / s xvc - t o t a i l a t e r a l v e l o c i t y a t 1.e. of che wings due t o :ransla:ion and r o t a c i o n i d s i z t c ~- t o t a l v e r t i c a l v e l o c i t y a t i , e . of t h e wings d u e to t r a n s l a t i o n a n d rocaticn ' d s i xvtail - z u t a l l a t e r a l velocity a t c a i l q u a r t e r c h o r d due to z r a n s l a c i o n and rotaticn ! m / s ) s v t a i L - c o r a i v e r t i c a l v e l o c i t y a t t a i l q u a r t e r chord due t o translation and r o t a t i o n I m / s ) r h e t a f s - rnagniude o f f r e e S t r e a m angle t o the i e a d i n g e d g e of t h e w i n g s wrc the v e r t i c a l ( d e g r e e s ; t h e t a v t a l l - angle t h e f r e e stream makes w i t h the horz. d u e O : both z S x veloc=ies a t c/4 af t a i i Idegrees) c h e t a f s t a i i - .magnitude o f free stream a n g l e t o zhe p a r t e r cnoca o f c a i l w r t t h e v e r t i c a l idegreesi L t a i l - i i f t d u e t a tail (NI, D t a i i - drag d u e ta tail ! N i , Thr t h r u s t f rom winqs ! gj 3raq - d r a q from winqs (g) ~ ~ ; ; ; ; .; ; . . ~ . ; ~ i . . ; . ; . ~ , ;. L. .~ .~. ~ i L ~ i ~ -~ i > . 2 > ~ ~ i ~ ~ G G + . .
- p c p u l a t a Lookup tables f o r L i f t and Drag a a c a -.Il -89 0;7.85 6.98 4 - 9 3 3 - 4 0 0;15.88 12.94 10.18 6.02 0;23.81 18.4 14.46 7 . 7 0;29.58 21.79 17.36 9.79 0~33.23 24.6 19-33
D = [1,98 0 . 3
10.67 0;35.6 26.4 19.33 10.67 0;35.6 26.4 20.96 12-01 0;37.03 27.01 21.41 12.56 O; ... 35.59 25.81 21.1 11.13 0;34.74 25.28 20.64 9-21 0;34.02 24.66 20.41 10.08 0;30.92 22.52 17.92 10.27 0;27,35 18.07 16.08 8.61 0;24.24 1 5 - 2 5 12.53 5.41 0;20.63 12.23 9.43 3.12 0;16.06 8.63 5.14 0.74 0;10,22 5.07 2.39 -.O6 0;4,99 3.29 1.06 . 6 0;-1-56 -95 1.43 -.66 O];
-spec: f y ~nit;ai tondicions theta=(2/180+pi);z=0;x=O;xdot=O.O;alpha=O.OO;omega=O;zdd=O.OOOOO;zd=0.00;xdd=0.0;thetaplot(l)=theta/pi*l8O;xdotplot(l~=xdot; alphaplot (1) =alpha;dragplot (1) =O;thrplot (1) =SO;tplot (1) =O; xplot(l)=0;zplot(l)=O;omegaplot(l~=O;velplot~l)=O;thetav~1~=O; xddwing(1) =O; zddwing(1) =O;xddtail(l)=O;zddtail(l)=O;awing(l) =O; atail(l)=O; -othsr v a r i a 8 i . ~ l=O.O75; dt=0.001; m=0.050; Iy=0.000031; 12=0.125; -distance t o tail . c i 4 rho=l,225;S=0,007; cl=-0.0017345; -use Lockup tables ta -valuate the surrent E, T, using xdot as .refsrence velocity and reference theta -and assurmng al1 flapping aone at 40Hz .designate che velocities "headings" for each matrix ealumn in lookup -tables
v1=7.04;~2=6.28;~3=5.24;~4=1.87;~5=0;
thetav ( 1 1 =O; -begin Looping through ~ i m e ateps tplot (1) =O;i=l; -lccp ~eginsat dt, ana aii variables get updated based on avg. -acceierations -thzough chat inczemental time step. The values at the end of a step -repzesent the -values of that variable at that time dt. for t=dt:dt:l5, i=i+l;tplot (il=t; ;total x velocity at leading edge of wings xvt(i)=xdot+omega*l*cos(abs(theta)); rtotal z velocity at leading edqe of wings z v t (il=zd+omega*l*sin (abs (theta) ); 'determine thetav, the angle the free strevn makes with the horz. -due to both vert h horz translations (xvt and z v t ) if zvt(i)=O, if xvt(i)=O, thetaw(i)=O;iprevent undefined 0/0 when using arctan function!
end;
eise
anc; snc; end; 15 zvt (i) >O if xvt ( i l >O, ;f theta>=O, thetafs=pi/Z-theta-thetav0 ; eise thetafs=pi/2-thetav(1-theta; ana; end; i f zvt(i)>O if xvt(i)<O, F E thetat=O, thetafs=pi/S+theta+thetav(i); else thetafs=pi/2+theta+thetav(i) ; end;
E-II
rowl=12;row2=ll; en; if angle<l20 & angle>=llO rowl=l3;row2=l2; end; if angle<l30 & angLe>=120 rowl=14;row2=13; end; if anglecl40 & angle>=130 rowl=1S;row2=14; end; ~f anqle<150 & angie>=140 rowl=16;row2=15;
en;
if angle<l60
end ; if angle<170 & anglo>=160 rowl=18;row2=i7; end; . iZ angle<l80 & angie>=l70 rowl=19;row2=18; rn; anglelow= (row2-LI +IO; -star: i o c b l e interpolation for thrus: ... ,a -A ?Lcity: -veiccity p a r at Lower angle VLowl=T ( raw2,columnll ;VHighl=T (row2, column2); - v e l c c i t ; ~p u r for higher angle Vtow2=T (rowl,columnl);VHigh2=T(rowl, colrimn2); -interpolate with chese points a t the current velocicy ThrLowAnqle= (VLowl-VHighl) / (vhigh-vlow) (Vel-vlow) +VHighl; ThrHighAngle= (VLow2-VHigh2) / (vhigh-vlow) *(Vel-vlow) +Vkiigh2;
+
-interpoiate =a get at the desired angle Thr = (ThrKighAngle-ThrLowAngle)/l0*(angle-anglelow)+ThrLowAngle; thrplot (i) =Thr; -Repeat the same thing for getting the drag -veioci= gai1 at lower angle VLowl=D(row2,columl);VEIighl=D(row2,column2); -veiocity ?ai= fcr higher angle colu.1~2 1; VLow2=D ( rowl,c o l m l ) ;VHigh2=D (rowl, -interpoiate with these points at the eurent velocity DragLowAngle=(VLuwl-VHighl) / (vhigh-vlow) Y Vel-vlow)+VHighl; DragHighAngle=(VLow2-VHigh2) /(vhigh-vlowl *(Vel-vlow)+Wigh2; -inteqaiate to get Drag at the desired angle Drag = (DragHighAngle-DragLowAngle)/lO*(angleanglelow)+DraglowAngle;
---..-.*.- * * . - ..* *. -. . -. . . . . .
~
.~
-1'-'----.-. - . -. " -. - -. - - .. -7 . f f. -.. . . TAIL FORCES ;total < : velocity at c/4 of tail ); xvtail (i)=xdot+omega*12*cos (abs(theta] -total z velocity at c / 4 of tail =zd+omega*l2*sin(abs(theta)) ; zvtail(i)
. . . . . . . . . . . . . .
............
....-............
. . .
determine cnetamaii, cne angie cne free scream maices witn the -horz. due to both vert h horz translations ixvttail and zvtail) if zvtail(il==O & xvtarl (i) =O, -prevent undefinea 0 / 0 when using a r c t a n function! thetavtail (il=O; else ) ; thetavtail (i) =atan (zvtail( i l /xvtail (i) end:
if zvtail(il<=O r f xvtail(i)<=O, if theta<=O, thetafstail=pi/2+theta+thetavtail(i); alse thetafstail=pi/2+thetavtail(i)+theta;
nna;
end;
and; if zvtail(i) <=O ~f xvtail (i) >=O, rf theta>=O, thetafstail=pi/2-theta-thetavtailti); els thetafstail=pi/2-thetavtaiL(i1-theta; end; ad; cna; F f zvrail(i)>=O ~f xvtail(i)>=O, r f theta>=O, thetafstail=pi/2-theta-thetavtail(i);
ISS
thetafstail=pi/2-thetavtailcil-theta; ma;
ena;
thetafstail=pi/2+thetafth~ftavtail(i);
end;
2r.d;
end; rf xv-tail(i) == 0, if zvtail (i) >=O, -case where wings are rising purely vertically thetafstail=theta; end; if zvtail(i)<=O, thetafstail=pi-abs(theta1;-case of pure descent of wings end; end;
if xvtail (i) =O, thetafstail=O; end; end; tfstail(i)=thetafstail; -thetahtail is the PlAGNITUDE of the tail's angle -ci arrack between the free stream velocity vector -and the ocdy-fixed t-axis (longitudinal axis -der,nine magnitude of : a i l free stream velocity Veltail=(xvtail (i) *2+zvtail(i)^2) ^.5; -calculate : h e CL, cd of : n e tail at the free-stream angle anqletail=abs(thetafstail/pi+l80); cl = 3*10A(-6) +angletail^3,0008*angletai1''2+0.0487+angletai1+0.006l; cd = 2*lO"(-8)*angletailA4-7*lOn(-6)fangletailA3+O.OOO6*anqletailA20.001*angletai1+.0984; a i l -caicula:e :oral drag and lift frcm : Ltail=1/2*rho+Veltailn2*S*cl; Dtail=1/2*rho*VeLtailA2+S*cd; Ltl(i)=Ltail;Dtl(i)=Dtail: :in body-fixeci frame:
-evalratt angular acreleration, ensuring 9 r a q farce alwzys pornts - o p p o s i L :O t n direction o f the velocit thetaold-theta; megaold=omega; dragplot ( i ) =Drag;
-the original alpha a: the beginning of the current time interval alphaold=alpha; discdampl = cl*omegaold; -qet the zoncrinution of the wings to alpna if xvt (i)>=O, awing(i1 = -Drag/1000+9.81+1/Iy+discdampl/Iy; eLse awing(i1 = Drag/1000+9.81+1/Iy+discdampl/Iy;
end;
-gec the contribution of the taii to alpha -CASES Wt?'ERE Thetaold < 0 : - M E ia: ThetaCo, xvtail<O, zvtailC0, thetafstail<90deg. if thetaold<=O h xvtail(i)<=O h zvtail(i)<=O d thetafstail<=pi/2, atail(i1 = (Dtailfsin (abs(thetafstail)) +Ltail'cos (abs(thetafstail1 ) 1 *12/1y; end; -CASE Lb : ThetaCo, :rvtail<O, zvtail<O, thetafstail>=90deg. rf thetaold<=O h xvtail (il<=O h zvtail(i) <=O h thetafstail>pi/2, atail (i) = (Dtail*cos(abs(thetafstail) -pi/2) Ltail*sin ( a h(thetahtail)-pi/2) ) *12/Iy; end; ;CASE 3: Theta<O, :wtail<O, zvtail>O, thetafstail<oOdeg. if thetaold<=O h xvtail (il<=O & zvtail (i) >O h thetafstail<=pi/Z, atail(i) = (Dtail*sin(abs(thetafstail)1 +Ltailtcos (abs(thetafstail)) 1 *12/1y;
enci;
- Q U E Ja: Theta<, .uvcaii>, zmaii>, cnetafscaii<50aeg. if thetaold<=O & m a i l (i) >O h zvtail(il>=O & thetafstail<=pi/2, )atail (i) =(-Dtail*sin (abs(thetafstail) Ltail*cos(abs(thetafstail)l )*l2/Iy; end; -CASE 5B: Theta<O, xvcail>O, zvtaib0, thetafstail>=90deg. if thetaold<=O & xvtail (il>O & zvtail (il>=O h thetafstail>pi/2, ) +Ltailfcos(piatail (i) =(-Dtail*sin (pi-abs(thetafstail) abs (thetafstail) ) ) *l2/Iy; enci; -CASE 7 : Theta<O, xvtail>a, zvtail<O, thetaLstail>=90deg. if thetaold<=O & xvtail(i)>O & zvtail(i)<=O & thetafstail>pi/2, atail(i1 =(-Dtailhsin(pi-abs(thetafstai1))+Ltail*cos(piabs (thetafstail) ) ) *U/Iy;
end;
-CASES WHERE TheCa > 0: -CASE 2: Theta>O, xv:ail<O, zlrtail<O, thetafstail>=90deg. L E thetaold>O & xvtail ( i l <=O & zvtail (i) <=O & thetafstail>pi/2, atail(i) =(Dtailcsin(pi-absithetafstai1))-Ltail*cos(piabs(thetafstai1) l )*12/Iy; end; -CASE 4a: The:a>O, :wtail<O, =vtaFf>O, thetafstail<90deq. if thetaold>O 6 xvtail (i) <=O & zvtail (i) >=O h thetafstail<pi/2, atail(i) = (Dtail*sin(abs(thetafstail) l+Ltailtcos(abs(thetafstail)))*12/Iy;
-CAS= I b : The:a>O, xvtarl<O, z~rtaib0,:hetafstail>=90deg. L f thetaold>O & xvtail (i)<=O h rvtail(i)>O & thetafstail>pi/2,
atail (i) = (Dtail*cos(abs(thetafstail) -pi/2)Ltail*sin(abs(thetafstail)-pi/2))*12/Iy; end; - W E 5: Theta>O, :tmail>O, zviail>O, thetafstail<=?Odeg. ~f thetaold>O & xvtail(i)>O h zvtail(i)>O h thetafstailcpU2, atail(i) = (Dtail*sin(abs(thetafstail) )+Ltail+cos(abs~thetafstail)))*l2/Iy; ezd; -CASE Sa: Theta>O, xvtail>O, z v t a i l < O , =hetafstail<90deg. if thetaold>O & xvtail (i) >O 6 zvtail (il<=O & thetafstail<pi/2, atail (il = (-Dtail*sin(abs (thetafstaill 1 Ltail*cos(abs(thetafstail) 1 *12/Iy; enc; - C E E ; ab: Theta>r),xvLail>O, zvtail<O, thetafstail>90deg. if thetaold>O h xvtail (i) >O & zvtail (i) <=O & thetafstail>pi/2, atail(i1 = ( - D t a i l * s i n ( p i - a b s ( t h e t a f s t a i l + c o s ( p i i t b s (thetafstail) ) ) *l2/Iy; end;
-integrate alpha twice to get angular position -once: -total omega at the D I D of current time internai omega = (alpha *dt+omegaold); ; twice : -totai rheta at the END of current time inte,yal theta = omegafdt+thetaold;
thetaplot(i)=theta/pi*180; if ~hetaplot(i)MO, fprintf('STOP!! - 1 AM OUT OF CONTROL!!!!'); end; -evaluate vertical accleration zdd, and integrate twice for ;"altitudew postitin zddold=zdd;xddold=xdd;
-get the wing =ontribution to the z acceleration if x v t (i) >=O, rf thetaold>=O, zddwing (i) =Thr/l000+9.8l/m*cos (abs(thetaold) )9.81+~raq/1000*9.81/m+sin(abs(thetaold))+omeqa*xdot; ise zddwinq(i)=Thr/1000+9.8l/m~os(abs(thetaold))-9.81~raq/1000+9.81/m*sin(abs(thetaold)~+omeqa*xdot; end;
?Ise
: f thetaold>=O, ) -9.81zddwinq (i) =Thr/Z000+9,8l/m+cos (abs(thetaold) ~raq/1000+9.81/m+sin(abs(thetaold) l+omega*xdot; else =~hr/2000*9.8l/m*cos (abs(thetaold) 1zddwing (i)
9.81+Drag/1000*9.81/mfsin(abs(thetaold))+omeqa*xdot; end; end;
-aec cte winq toncributron t u the x acceleraticn -use appropriate equation depending upon which side of rhe theta 1 ;nar~. -equals 1
-must use theta at starc of r h i s m t e ~ v a lco upaace xact thetaold>=O, :f x v t (il>O, xddwing(i)= ~ h r9.81/1000/m*sin(abs * (thetaold) 1Drag/l000*9.81/m*cos(abs(rhetao~d~ 1-omega*zd;
1 :
eise
xddwinq(i)=~hr+9.81/1000/mfsin(abs(thetaold))+Draq/1000+9.8l/m +cos(abs(thetaold) 1 -omega*zd; end; else if xvt(i)>O, ) xddwinq (i) =-Drag/1000+9.81/mtcos (abs(thetaoid) Thr/1000*9.81/m*sin(abs(thetaoldH-omega*zd; else xddwing(i)=Drag/1000*9.8l/m*cos (abs[thetaold) Thr/1000+9,8l/m*sin(abs(thetaold)) -omeqa+zd; end; end;
-qet the tail's contribution to the z and x accelerations -CASES WHERE Theta < O: ;ASE ia: Theta<O, xvtail<O, z M a i l 4 , thetafstail<gOdeg. if thetaold<=O & xvtail(i)<=O d zvtail(i)<=O h thetafstail<pi/2, zddtail(i) = (Dtailtsin (abs(thetaddl +abs (thetafstail) pi/2)+Ltail+cos(abs(thetaold) +abs (thetafstaill-pi121 1 lm;
xddtail (i) = (Dtail*cos (abs(thetaoldl+ab9 (thetafstail) -pi/2)Ltail'sin (abs(thetaold) +abs (thetahtail) -pi/2)) /nt; 2nd; -O.SE Lb : ThtaCo, x-taii<O, rvtaii<O, thetafstail>=?0deg. if thetaold<=O & xvtail (i) <=O & zvtail(i)<=O & thetafstail>=pi/2, zddtail(i)=(Ltail*sin(pi-abs(theEafstai1)abs (thetaold) ) +Dtailfcas(pi-abs( t h e t a f s t s(thetaold) 1 1 lm; xddtail(i)=(Dtail*sin(pi-abs(thetafstail)-abs(thetaold))Ltail'cos (pi-absi thetafstail)-abs (thetaold) ) 1 lm; end; -CASE 3: Theta<O, xvtail<O, zvtaib0, thetafstail<9Odeg. L thetaold<=O & xvtail (i) <=O & zvtail(il>=O & thetafstail<pi/2, zddtail(i) = (Ltailfcos (pi/2-abs(thetaold) -ab9 [thetafstail)1 Dtail'sin (pi/2-abs (thetaold) -abs (thetafstail) 1 1 /m;
xddtail(i)=(Dtail*cos(pi/Z-abs(thetaold)-
abs (thetafstail) ) +LtailCsin(pi/2-abs (thetaold) -abs (thetafstail) 1 1 lm; end; - &-SE 5a: ThetacCl, xvtaib0, z7n;aii>0,chetafstail<9Odeg. if thetaold<=O & xvtail (i) >=O 6 zvtail(il>=O 6 thetafstail<=pi/2, zddtail (i)=(ltail*sin(abs (thetafstail) -abs (thetaold) 1Dtai14cos(abs(thetafsrail) -abs (thetaold) 1 lm; xddtail (i)=(-Dtailtsin(abs(thetafstail) -abs (thetaold) 1 Lral*cns !abs (rhetafstail) -abs lthetaold)) ) /in; ena; -CASE 5 ~ ThetaCo, : :wtail>O, zvcail>O, :hetaf staii>=SOdeq. if thetaold<=O & xvtail (il>=O & zvtail (i) >=O & thetafstail>=pi/2, zddtail (i) = (Ltailf cos (pi/2+abs(thetaoldl-abs (thetafstail) 1Dtail4sin(pi/2+abs(thetaold]-abs(thetastail]))/m; xddtail (i) =(-Dtail+cos(pi/2iabs(thetaoldl-abs(thetastail)1 ~tail*cos(pi/2+abs(thetaold)-abs(thetafstai1)) )/m; end; -CASE 7: Theta<[), xnaii>O, zvtail<O, thetafstaii>=?Odeg. if thetaold<=O & xvtail(i)>=O & zvtail(i)<=O & thetafstail>=pi/2, = (Ltail*cos(ab8(thetafstailbpil2zddtail (i) abs(thetaold))+Dtail*sin(abs(thetafstail)-pi/2-abs(thetao~d)))/m; xddtail (i)=(-Dtailfcos(abs (thetafstail) -pi/2abs (thetaold)) +Ltail'sin (abs{thetafstail)-pi/Z-abs (thetaold)) ) /m; end; -CASES -HERE Theta > 0: -CASE 7: ThetaiO, xvcail<O, zvtail<O, thetafstail>=gOdeg. if thetaoldHI & xvtail(i)=O & zvtail(i)C=O & thetafstail>pi/2, zddtail(i) = (Ltail'cos (abs(thetafstail) -abs (thetaold) pi/2)+~tail+sin (abs(thetafstail) -abs (thetaold}-pi/2)) /m; xddtail (i) = (Dtail*cos (abs (thetafstail) -abs (thetaold) -pi/2) Ltail*sin (abs(thetafstail) -abs (thetaold) -pi/2) 1 /m; t n c i ; -CASE 4a: TbetaiO, a v t a i i C 0 , Z~tail>0,thetafstail<gOaeg. if thetaold>O & xvtail(i)<=O & zvtail(iI>=O 6 thetafstail<pi/2, = (Ltaif *cos(abs(thetaold) +pi/2-ab5 (thetafstail) )zddtail(i) Dtail*sin(abs (thetaold] +pi/2-abs (thetafstail} 1 ] /m; xddtail(il=(DtaiL*cos Iabs (thetaddl+pi/2abs(thetafstail))+Ltailfsin(abs(thetaold~+pi/2-abs(thetastail) ) )lm; ena; -CASE 4b: ?heta>O, xvtaild?, zvtaib0, thetafstail>=?Odeg. <=O 6 &ail (i) >=O & thetafstaiDpU2, if thetaold>O & xvtail (i) zddtail(i) =(Ltail*cos (abs (thetaold) -abs (thetafstail) +pi/2)Dtailesin(abs(thetaold) -abs (thetafstail)+pi121 ) /m;
PW
xddtail (i) =( Dtail*cos (abs(thetaold) abs (thetafstail) +pi/2)+Ltail%in i a b s (thetaoldlabs (thetafstail) +pi/2)1 /m;
end;
.CASE 6: Tbeta>O, xvtail>O, z v t a i l > , t h e t a f s t a i l < = o O d e q . if thetaold>O 5 &ail (i) >=O & zvtail(i)>=O & thetafstail<pi/2, zddtail(i)= (Ltail'cos (pi/2-abs (thetaoldl-abs (thetafstail) 1Dtailtsin(pi/2-abs(thetaold)-abs(thetafstail)))/m; xddtail( i)= (-Dtail'cos ( P X - a b s (thetaold) -abs (thetahtail) Leail*sin(pi/2-abs(thetao1d)-abs(thetafstai1))) / m i
T h e t a > Q , xvtail>O, zvtailC0, thetafstail<cOdeq. thetaolcb0 & m a i l (il>=O h zvtail (il<=O & thetafstail<pi/S, zddtail (il=(Ltail*cos (absI t h e t a f s t a i l ) +abs (thetaold) pi/2)+Dtail4sin(abs(thetafstail)+abs(thetaold)-pi/2))/m; xddtail (i1 = (-Dtail+cos (abs(thetdfstaill+abs(thetaoldlpi/2)+Ltail*sin(abs(thetafstail)+abs(thetaold)-pi/2))/m;
:f
Theta>O, xvtailz0, zvtail<0, t h e t a f s t a i l > ? O d e g . ~f thetaold>O & xvtail(i)>=O b zVtail(i)<=O & thetafstail>pi/2, zddtail(i)= (Ltail*sin(pi-abs(thetafstail) abs(thetao1d))+Dtail'cos(pi-abs(thetafstai1)-abs(theta01d)) )lm; xddtail (i) = (-0tailfsin (pi-absIthetafstail) abs (thetaold)) +Ltail'cos (pi-abs( t h e t a f s t s (thetaold)1 1 /m;
ena;
xdotold=xdot;
- t h e adot at tke END of this i n t e r v a l xdot=xdd*dt+xdotold; xold=x; x = xdottdt+xold; -al1 v a r i a b l e s have now been updated. Locp.
na;
are plots of the oscillatory decay observed for the dBerent configurations descriid in Chapter 6.
0.0814
0.187
0.0152
0.00987
0.122
0.00 120