Post on 01-Oct-2021
transcript
FLOW VlSUALlZATlON FOR A MICRO AIR VEHICLE
Jasmine El-Khatib
A thesis submitted in conformity with the requirements for the degree of Yaster of Applied Science
Graduate Deparbnent of Aerospace Science and Engineering University of Toronto
O Copyright by Jasmine EiXhatib (2000)
National Library I*l of Canada Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliogaphic Services services bibliographiques
395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 Ottawa ON K1A ON4 Canada Canada
The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant a la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distribute or sel1 reproduire, prêter, distribuer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de rnicrofiche/film, de
reproduction sur papier ou sur format électronique.
The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts barn it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimes reproduced without the author's ou autrement reproduits sans son permission. autorisation.
Abstract
In the development of a tlapping-wing micro air vehicle ( M N ) , the need for
research in the field of flapping-wing flight is evident. One tool which can give insight
into this field is flow visualization. The purpose of this thesis is to obtain a qualitative
and quantitative presentation of the flow-field created by a MAV. This was
accomplished using smoke flow visualization and hot-wire anemometry. The tesuits
have indicated that strong Ieading-edge vortices and the clap-fling e f k t are the high-lift
mechanism of a MAV.
Acknowledgements
It is a pleasure to thank Dr. DeLaurier for his supervision and for giving me the
opportunity to explore the làscinating field of small-scale flapping flight.
Thanks also to Dr. Johnston for lending the hot-wire anemornetry equipment so that
this research may be perfonned.
As well, appreciation is given to al1 those who have offered help and advice during
the course of this research, in particular Mr. David Loewen and Mr. Derek Bilyk.
I am indebted to Mr. Samir Fahs for information about hot-wire anemornetry, Mr.
Bruce Woodcock for advice on hot-wire welding and Mr. Jurgen Schumacher for
technical help with Tecplot . A word of thanks is also mentioned for the numerous CO-op and summer students who
participated in this project. They include Mr. M&ar Ahmed, Ms. Theresa Robinson,
Ms. Yaeko Yamamoto, and Mr. Joseph L m
1 wodd also like to thank my farnily, fiends and colleages at the University of
Toronto Institute for Aerospace Studies for their support and encouragement.
The fuiancial assistance of the Defense Advanced Research Projects Agency is
gratefully acknowledged.
Table of Contents
Chapter 1: THE MICRO AIR VEHICLE
1.1 The Motivation Behind this Research
1.1.1 Research Objectives
1.12 RationalefortbisResearch
1 . Summary
13 Background Literature
1.3 Flapping Wings & Muscle Actuato~
Chapter 2: FZO W WSUALLZ4 TION TECHNIQUES
2.1 Common Flow Visualization Techniques
2.11 Flow Visualization by Addition of Foreign MateriPb
2.1.2 Optical Methods for Fiow Vkualuatioi
2.13 Quantitative Methods fur Flow Visualilntion
2.2 Fluid Flow Created by the Micro Air Vehicle
2.3 Flow Visuaikation Techniques for a Micro Air Vehicle
Chapter 3: FLOW YISUALIZQTION USI1VG SM-
3.1 Smoke Fiow V i s u a t i o n
3 3 Experiment #I
3 3 Experiment #2
3.4 Experiment #3
3 5 Fnrther Experimentation
Chapter 4: FZO W WSUALIZ4TION USING HOT-MRE ANEMOmTRY:
THE EXPE-NTAL SET-UP
4.1 Principle of Hot-Wire Anemometry
4.1.1 The Hot-Win Probe
4.1.2 ModesofOperation
4.1.3 H a t Transfer
4.2 Selecting The Type of Hot-Wire Anernometer
4.2.1 Sensor Specifications
4.23 h b e Specifïcations
4 3 Weîding Hot-Wire h b e s
4.4 Caiibration of Hot-Wire Anemometem
4.5 Construction of a Three-Dimensional Traverse
4.6 Data Acquisition System
4.7 Measnring Velocity & Turbulence
4.8 Summary of the Experimental Set-Up
Chapter 5: FLOW VISUALIZA TION USING HOT- FWRE AlVEMOMETR Y:
THE RESULTS
5.1 Velocity Flow Field Under Ow BAT42 Wing
5.2 Velocity Flow Field Under Two BAT-12 W i q
5.3 Velocity Flow Field Under Four BAT-12 W I i
5.4 Velocity Flow Field Under One Eiiipticai Wing
5.5 F u i t k r Discussion on the &sults
Chapter 6: CONCLUSION & FUTURE WORg
6.1 Hot-Wire Anemometry Teating
6.1.1 Testing of MAV Wings
6.13 Improvements to Apparatos
6.13 Other Applications
6.2 Smoke Testing
6.3 Free Flight Vehicle
6.4 A d y t i d M d e l
6.5 Conclusion
Chupter 7: REFERENCES & BIBLIOGRAPHY
7.1 References
7.2 Bibliograp hy
List of Tables
Table 4.1. Properties of Single-Sensor Miniature Wire Probes
vii
List of Figures
Figure 1.1.
Figure 1.2.
Figure 1.3.
Figure 1.4.
Figure 1.5.
Figure 1.6.
Figure 3.1.
Figure 3.2.
Figure 3.3a
Figure 3.3 b.
Figure 3.4.
Figure 3.5.
Figure 3.6.
Figure 4.1.
Figure 4.2.
Figure 4.3.
Figure 4.4.
Figure 4.5.
Figure 4.6.
Figure 4.7.
Figure 4.8.
Figure 4.9.
Figure 4.10.
Figure 4.1 1.
Figure 4.12.
Figure 4.13.
Figure 4.14.
Figure 4.15.
Figure 5.la
A Four-Wing Flapping MAV
The Clap-Fling Effect
A Vortex Ring Gait versus a Continuous Vortex Gait
The MAV Test Rig
The BAT42 Whg
The Elliptical Wing
Fog-Fluid Generator
Smoke Flow Visualization of the MAV
Flow Patterns for a BAT42 Whg
Flow Patterns for a BAT-12 Wing
Flow Patterns for an Eiliptical Wing
Modified Apparatus for Smoke Flow Visualization
Laser Light Sheet
Hot-Wire Anemometer Probe
Constant Temperature Anemometer
Typical Wire Sensor
The Heat Balance for a DEerential Element of a Hot-Wue Sensor
Hot-Wire Probes With Either 1,2, or 3 Sensors
Spot- Welding Equipment
Calt'bration of a Hot-Wire Anemometer
CTA Output as a Function of U for Robe # 1
Construction of a Three-Dimensional Traverse
Scale in Degrees
Hot W Ï Ï Anemometer & BAT42 Wings
Unis lide Traverse
Velmex Controller/Driver for Stepper Motors
Velocity Components at the Sensor
Schernatic Diagram of Experimental Set-Up
Top View SLice of the Velocity Flow Field Under One BAT42 Wing
Figure 5.lb.
Figure 5.2a
Figure 5.2b.
Figure 5.2~.
Figure 5.3a
Figure 5.3b.
F i y r e 5.4a
Figure 5.4b.
Figure 5.5a
Figure 5.5b.
Figure 5.5~.
Figure 5.6a
Figure 5.6b.
Figure 5.7a
Figure 5%
Figure 5.8a
Figure 5.8b.
Figure 6.1.
Figure 6.2.
Figure 6.3.
Top View SLices of the Velocity Flow Field Under One BAT-12 Wing
Side View Slice of the Velocity Flow Field Under One BAT4 2 Wing
Side View Slices of the Velocity Flow Field Under One BAT-1 2 Wing
Side View Slices of the Velocity Flow Field Under One BAT42 Wing
Top View Slices of Velocity Flow Field Under Two BAT-12 Wings
Top View Slices of the Velocity Flow Field Under Two BAT-12 Wings
Side View Slices of the Velociry Flow Field Under Two BAT42 Wings
Side View Slices of the Velocity Flow Field Under Two BAT-12 Wings
Top View Slices of the Velocity Flow Field Under 4 BAT42 Wmgs
Top View Slices of the Velocity Flow Field Under 4 BAT-12 Wings
Top View Slices of the Velocity Flow Field Under 4 BAT-12 Wings
Side View SLices of the Velocity Flow Field Under 4 BAT-12 Wings
Side View Slices of the Velocity Flow Field Under 4 BAT-12 Wings
Top View Slices of the Velocity Flow Field Under One Elliptical Wig
Top View Slices of the Velocity Flow Field Under One Elliptical Wmg
Side View Slices of the Velocity Flow Field Under One Elliptical Wing
Side View Slices of the Velocity Flow Fieki Under One EUipticai Wing
New Test Rig Design
Free-Flight MAV Mode 1
Wing Design Used in Free-Flight Mode1
Chapter 1
THE MICRO AIR VEHICLE
1 . The Motivation Behind this Research
1.1.1 Research Objectives
A micro air vehicle is defhed by the Defense Advanced Research Projects Agency
(DAWA) as "an airbome phtform with no dimension exceeding 15 cm (6 in)"'. These
vehicles have a wide variety of applications both in military and civilian use. They are,
however, primarily designed as reconnaissance vehicles that c m 'Till a nÿveillance blind
spot left by today's military satellites and spy planes"2. Equipped with a surveillance
camera, a MAV can provide a soldier with an unobstructed view of the banlefield. In
order to meet this condition, the vehicles must be large enough to carry optical cameras
and innared sensors on bard, yet smail and light enough to fit into a soldier's backpack
and fly in cluttered environments.
The construction and design of a micro air vehicle has shown considerable challenge.
The aerodynamics and flight performance, the propulsion system, flight control, the
communication components and the imaging sensors all need to be considered in the
development of a micro air vehicle. In 1997, DAWA contracted several groups across
the United States to take into account all these Factors and develop micro air vehicles
capable of performing military tasks. One of these researc h groups was SRI International
of Menlo Park, California With the help of the University of Toronto Institute for
Aerospace Studies (UTIAS) subsonic a e r o d y d c s group, SRI accepted a three year
contract to develop a micro air vehicle which combines two unique technologies,
fiapping-wing propulsion and Electrostrict ive Po lymer Artficial Muscle (EP AM)
actuation.
Since May of 1998, the UTIAS subsonic aerodynamics group, under the direction of
Dr. DeLaurier, has k e n involved with studying micro air vehicles operated by flapping
wing propulsion. It is hoped that MAVs using this type of propulsion wiil have several
advantages. For example, they are expected to have
and slow flight capabilities; greater stability and
greater payload for a given ~ i n ~ s ~ a n " ~ . An early
propulsion is shown in Figure 1.1.
"great er energy enic ienc y; hovering
controllability; greater stealth; and
sketch of a MAV with this type of
(source: Drawlng by ~i ~av id ~&wen)
Indeed, flapping wing propulsion seems like the ideal choice for MAVs with
reconnaissance applications.
The overall goal of the UMAS aerodynamics group is to design MAV wings and a
body to mimic not only a bird's outer appearance (for camouflage purposes) but also,
more importantly, the flight of a tlappiug bird. In particular, the UTIAS group has k e n
researching hummingbird flight due to the bird's excellent hovering capabilities, which
would be beneficial for a MAV. However, in attempting to imitate the flight of a
hummingbird, several diniculties have arisea The field of flapphg flight at the s d l
scale of a MAV is relatively new and unexplored. To develop such vehicles, research
m u t be performed to gain insight into this type of flight. One experimental tool that can
give tremendous amounts of information about flapping flight is the technique of flow
visualizat io n.
Thus, the purpose of this research will be to visualize the flow patterns and motion of
the flow around a flapping MAV. A qualitative and quantitative representation of the
velocity flow field, created by the MAV, is specifically desired. Indeed, as the fo llowing
section will demonstrate, a great deai of information fan be inferred Eom visualking the
flow of such a vehicle. This information can lead to a p a t e r understanding of the
aerodynamics of man-made flappers as well as elucidate the performance of the current
design and shape of the MAV.
9.1.2 Rationale for th& Research
The interaction of an object with a rnoving fluid medium cari be detected in the
structure of the wake. However, flow patterns in the wake are often invisible and short-
lived and, therefore, flow visuaüzation techniques need to be applied to the patterns.
Flow visualization techniques are capable of yielding a qualitative, so metimes
quantitative, macroscopic picture of the overall f low-field. The images, resuhing fiom
flow visualization, are a valuable aid in interpreting the complex flow pattans in the
wake around the O bject; in our case, this would be the MAV wuig. The images are able
to show in detail how the wake is generated by the fiapping wings and how it develops
after it has been shed.
There are many incentives for performing flow visualization techniques on a flapping
MAV. The flow-pattern pictures could potentially provide informat ion about the
mysterious high-lift mechanism of s m a l l flapping vehicles (Ellington CP et ai., 1996).
The images couid also be an aid in calculating the forces acting on the wings,
detemiinhg the optimum wing design of the MAV and providing the basis for theoretical
models of flapping-wing flight. Each of these reasons provides motivation for
visualking the flow of a MAV. They will al1 be descnbed in tum.
Hinh-Lift Mechanisms for Small Flapoers:
The aerodynamics of srnall-scale flapping wing flight is presently not yet weii
understood due to the large-scaie unsteady motions involved and the complexity of the
voxticity structure detected in the wake (Liu and Kawahi, 1998). T b applies to d l
small-scale flapping devices, be it a man-made flapper or a bird or insect. The lift forces
of their flapping wings disagree with the values computed fiom conventional
aerodynamic theories. However, by visualizing the flow of the flapping wing, the
location and strength of vortices may imply where this source of extra lift cornes fiom.
Thus, flow visualization experiments have the means to show how Lift is being generated
to support the O bject's weight.
In constructing a MAV, one's aim is to design a vehicle capable of having a high üft
coefficient since the vehicle wiil be carrying its own power source, payload and other
components. Thus, with the aid of flow visualization images, one cm adjust and modify
the design of a MAV to achieve a higher Mt coefficient. In other words, the images may
provide insight into why MAVs have high-li. aerodynamic performance.
Forces and E n e r a
Besides studying the iift mechanism of a fiapping MAV with the use of flow images,
one is also able to calculate the forces acting on the wings as well as its mechanical
energy output. Observing the vortex s û u c ~ s in the wakes of the MAV flapping wings
(Le. their strength and location) can provide valuable insight into the flight performance
of the vehicle. Because the wake vortex strength and, therefore, the strength or
circulation of the vortices bound on the wing are constant in t h e (Rayner J, 1997), it is
possible to compte the induced airflow around the wing and thus to determine lift,
thnist, drag, and mechanical energy output. Indeeà, it is fiom the flow visualization
images of the full tirne-varying velocity and vorticity fields of the wings that these
aerodynamic components may be calculated.
Optimum Wing Desian:
The design and shape of the entire wing is very important for the aemdynamics
involved. In nature one f i d s that "sustained and more economic fliers tend to have
elongated wings of high aspect ratio, while animals which need to fly in cluttered habitats
tend to have shorter and more rounded wings9? The wings of a MAV follow the
example of wings found in nature. They are designed to minimize the totai dmg,
minimize the wing weight for a given lift and maxmiize the thrust. In gened, successful
MAV wings shodd rnaximize force and minimize power required for flight.
The shape of the wing greatly affects the vortices m the wake. A slight deviation in
the wingtip shape, for example, could severely alter the strength and location of these.
Thus, flow images are useful because they help to determine the effects of varying the
wing shape of a MAV and, in turn, help to discover the optimum wing design.
At the tirne of this research work, two favorable MAV wing designs were available.
They are known as the BAT-12 wing and the Elliptical wing. More detaiis regarding
these two different whg designs are given in section 1.3. The point to be made, however,
is that these two types of wings will generate different flow patterns and, fiom those
patterns, the researcher may then formulate a hypothesis of which one outperfomis the
other.
Theoretical Modelinn:
As previously mentioned, the techniques of flow visualization provide information on
the geometry and strength of the flow fields. This information can also be predicted by
theoretical modeling, which uses the knowledge gained experimentaiiy of the vortex
wake structure to formulate the models. The models can then be used for any wing shape
and geometry. Theoretical models would be able to generate the aerodynamic properties
of wings more quickly than experirnental work would. This is another incentive for
visualking the flow of flapping-wing MAVs.
1.1.3 Summaty
Of the four reasons descriid above for performing fiow visualization techniques on a
flapping MAV, only two are of relevance to this research The flow visualbation images
coiiected in this work will be examined to gain insight into the Lift mechanism of the
M N . As well, the optimum wing design for this vehicle will be considered by
comparing the flow patterns of two successful MAV wing designs. These two motives
are intenelated since the optimum wing will be the one which produces the most
lift/power under the Limited size constraints.
The caicuiations for the forces acting on the MAV wings have k e n performed
previously by Mr. Derek Büyk as part of his own Master's research work (Bilyk D,
2000). He used a strain gauge balance to measure these forces and thus the calculations
are not repeated in this work. The interested reader is referred to his thesis.
As well, the development of an analytical model capable of representing the unsteady
and highly complicated flow of the MAV is not attempted here. The flow visuaiization
data couected throughout this research, however, will form the bais of Mr. Patnck
Zdunich's Master's thesis. Mt. Zdunich will focus his attempts on producing a
theoretical model.
In summary, the main objectives of this research are:
1) to tiimiliarize oneself with the techniques of fluid measurement, in particular flow
visualizaion,
2) to 0btai.n a qualitative and quantitative representation of the velockj flow field
created by flapping MAV wings,
3) to c o q m e the overall performance of two diffêrent MAV wing designs, and
4) to use the information gained to provide greater insight into the relatively new field
of d - s a l e flapping-wing flight (Le. the high-lifi mechanisms present in insects
and birds).
1.2 Background Literature
The micro air vehicle studied in this research is essentially a mechanical
hurnmingbird. It is designed to mirnic the flight and hovering capabilities of the bùd and,
thus, the flow patterns generated by this vehicle may be assumed to be similar to the flow
experienced by a hummingbird or similar flying animal.
Unfominately, there is Little existing literature on flapping-wing flight at the srnall
scale of a MAV due to the field king fairly new. However, there does exist some
information rrgarding the high-lift mechanism of severai flapping insects and birds. As
well, information can be found on the optimum wing design for a bird with wing sizes
close to that of the MAV. Although not much was discovered about flow visualization
performed on man-rnade flappers, numetous flow visualization studies in the past have
ken accomplished with real life birds and insects. Because al1 of this literatue is
relevant to the research performed here, a brief description will be given
The High-Lie Mechanism of Insects and Birds:
It was discovered that imects, both in forward flight and hovering, attahed lift
coefficients too high compared to the ones calculatecl fimm steady state principles
(Ellington et al., 1996). These higher lift coefficients were also found in flapping bird
flight. The question was then posed to scientists and researchers: "Where does this extra
mysterious source of lift corne hm?'
One solution was proposai by Weis-Fogh in 1973. He predicted that the generation
of tremendous amounts of lifi is due to a c lap-hg phenomenon. The basics of this
phenomenon are shown in Figure 1.2.
Cla~-Flina LiR Auamentation
Air is pushed downward t t
Figure 1.2. The Cbp-Fling Effect (Soume: Oawing by Mr. David Loewen)
As the wings separate, a "super-circulation" effect occurs which generates the increased
lift detected in birds and insects.
Oatimum Wing Design for Birds:
In most textbooks, it is stated that the optimum wing design for a flying animal is a
planar elliptical wing because of its better aerudynamic performance. The advantages of
this wing are that it minirnizes induced drag for a given lift and given wingspan, and it
represents a well-established theoretid solution to the lifting line problem for a flat wing
(Rayner et al., 1997). This finding should be considered when daigning the MAV wing.
The wingtip shape is dso meaningful in the design of a flapping wing because it
determines how the vortices are shed from the wing and if they roll up into a concentrated
vortex core. As described previously, the observation of the created vortices dows one
to determine the energy required to generate the wake (i.e. the induced clrag) and thus in
turn the mechanical output (Rayner et al., 1997). It has been s h o w that a rounded wing
tip produces more t h and permits greater accelerations in taking off nom the ground
and is thus a more advantageous shape.
Flow Visualization Emeriments:
Flow visualization experiments have ken perfonned on various types of birds and
insects in the past. For example, there are published midies on the wakes of dragonflies,
hawk mo ths, pigeons, kestrels, and butterflies.
The first flow visualization experiments with birds identified a vortex ring gait. The
experiments were largely concemed with slow fly ing animals (Magnan et al. 1 93 8).
Flight s peed Figure 1.3. A Vortex King Gait versus a Codnuous Vortex üait
(Source: Rayner et al., 1997)
By flow visualization experiments, Rayner and his colletigues at the University of
Bristol later discovered two gaits used by flying birds and bats: the vortex ring and the
continuous vortex gait. The dBerences are shown in Figure 1.3. The vortex elements
are eKptical or near circuiar in the vortex ring gait. On the other han& in the continuous
vortex gait, the elements follow close to the path of the wing tip. The vortex ring gait can
be detected in the wakes of birds and bats in slow flight. This type of gait is also detected
in species undergoing M e r flight but having shorter wings and a lower aspect ratio.
Longer-winged anhais at higher or cruising speeds, however, use the continuous vortex
gait . From Figure 1.3, one can predict that the M N , having a low aspect ratio and low
flight speed, may have more features of the vortex ring gait than the continuous vortex
gait in its wake. However, Rayner's study did not examine the flight of a bird in
hovering mode. Recall that this mode is what one is mainly interested in since the MAV
behaves as a mechanid hummingbird.
Several studies exist conceming the merences between forward flapping flight and
hovering flight. During sustained forward flapping flight, the trailing edge vortex is shed
and is the dominant flow feature (Freymuth, 1988). This can be modeled by steady state
principles. However, during hovering flight, where the animal is suspended in still air by
the actions of the wings, leading edge separation is dominant during the clap fling
mechanism (Maxworthy, 198 1). Luttges in 1989 also showed the occurrence of leading
edge separation for hovering dragon aies. The strong leading edge vortex, found by flow
visualization expehents, bas haen used to explain the mysterious source of high lift
present in small insects and birds (Ellington et al., 1996). From these studies, one may as
a remit expect to find leading edge vortices in the wake of a high-lift flapping-wing
MAV.
Indeed, the flow visualization experiments of the pst have increased our
understanding of the clapfihg mechanism as well as the leadhg edge vortex shedding
phenornena which is beyond classical steady-state aerodynamics. These two issues will
be investigated for the flapping-wing MAV in this research
1.3 Flapping Wings 8 Muscle Actuafors
As mentioned previously, the construction and design of a micro air vehicle using
flapping wing propulsion and EPAM actuators is a joint effort of SRI International and
the UTIAS subsonic aerodynamics group. The researchers at SRI mainly concentrate
the+ attention on designing, hbricating and testing the EPAM actuators. Once the
actuators are in operation, the researchers will then be accountable for integrating these
on a flapping MAV. On the other han& the UTTAS group is focused on the design,
fabrication and testing of the MAV wings as well as the overall ve hicle aerodynamics.
This section will describe the current development statw of the flapping-wing M N .
The EPAM actuation system is currently still not a part of the UTIAS micro air vehicle.
The purpose of this thesis has Iittle to do with this system and therefore it wiil not be
described. The interested reader, however, is refened to Mr. Bilyk's Masters thesis for a
description. Instead, uK main ernphasis in this section will be on the current MAV test
rig, which was built and designed to test the performance of various MAV wings, and the
present MAV wing designs. This information is needed in order to M a r i z e oneself
with the equipment with which flow visualbation testing is perfomed.
The MAV Test Rie:
Mr.David Loewen, a research engineer at UTIAS, built a bench-top test rig in 1998.
The test rig is shown in Figure 1.4. It allows up to four MAV wings to be attached to it.
The basic mechanism of the rig is that it converts the ci~cular motion of an electric mtor
to a sinusoidally varying flapping motion
(Source: ~hotograph Taken by theuthor)
Note that four MAV wings are attached to the rig. As Figure 1.1 indicates, the
mearchers at UTIAS proposed that the completed MAV would operate by a flapping,
four-wing propulsion system. The nason for this king that the lateral forces are
balanced and that the ciapfling effect, which greatly enhances Mt, occurs twice as ofken
as for a two-wing flying MAV.
In the pst, the test rig was used for measuring the iift produced by the wings with a
highly sensitive strain gauge bridge circuit. nie resuits of these tests may be found in
Mr. Bilyk's Masters thesis.
The bench top test rig, illustrated in the figure above, is a crucial piece of equipment
used for this thesis research. Because a fke-flying MAV has not yet been developed, the
test-rig is needed to hold the wings in place and allow for flow visualkation data to be
coilected around the wings. A fiequency meter, used to measure the frequency of the
flapping wings, is also shown in the background of Figure 1.4.
MAV Wing Construction:
In the first year of the contract, many whg designs were constmcted and tested. The
goal was to build four MAV wings, capable of producing a lift of 50 grams. The basic
approach was to build the wings rapidly and inexpensively. From this research one wing
design showed the most promise. It is called the BAT-12 wing and is depicted in Figure
1 S. This wing shape generates 50 gram of thnist using only about 5.7 W of mechanid
input power (Bilyk D, 2000). These values were measured at a flapping wing fiequency
of 39.3 Hz.
(Source: ~ k n n e d Image Taken by the Author)
In the following year of research, a new wing design was constnicted fhm the ideas
of Dr. Dehinier. This is known as the Elliptical wing because of its shape. It is shown
in Figure 1.6.
(Source: sknned Image Taken by the Author)
As descriid in the literature review, an eliiptical wing design has k e n known to
attain better aemdynamic performance. Therefore, as part of this thesis, the two wing
designs, the BAT42 and EUipticai wing, can be compared by examinhg the flow
patterns they create while flapping at their operating frequencies.
Both wings are constnicted using PEEK, which is a uni-directional carbon fibre. The
BAT42 wing consists of a s t 8 leading edge spar of PEEK and several lighter structural
members emanating fiom the spar. These extra members ~ p p a the shape of the light
mylar covering. The EUiptical wing also has three members to support its mylar
covering.
Chapter 2
FLOW VlSUALlZATlON TECHNIQUES
Fluid flows are studied in numerous fields including engineering, physics,
oceanopphy, chemistry and geology. A researcher often requires information abour a
particular fluid flow. For example, the parameters çuch as the velocity, pressure,
temperature, volume, mass and fiow patterns of the fluid may be required. Obtaining
flow patterns of the fluid is the basis of flow visualization. It provides insight into a
physical process if the flow pattern is produced or related to the process. Over the years,
advances have been made to develop techniques capable of producing flow patterns.
Chapter 2 examines several widely used flow visualization techniques. Only very
brief descriptions of these techniques are given Their advantages and disadvantages are
also mentioned. It is hoped that this information will help to determine which of these
techniques are best suited to test and provide insightfbl data conceming the airflow
created by the MAV. However, before a technique (or techniques) may be chosen, two
main issues need to be addressed. First, d visualization techniques depend on the nature
of the flow. Therefore, assumptions about the MAV airfiow need to be made ahead of
time and these are provided in Section 2.2. Secondly, before undertaking any flow
visuali7iitio~ a clear decision should be made with regards to the purpose of the
measmement. This was not only explained in Section 1.1 and 1.2 but d l also be
describecl in Section 2.3. Finaily, provided with the information about the types of flow
visualhtion techniques, the purpose of this research and the nature of the MAV flow,
appropriate visualization techniques for a MAV may be chosen.
2. .1 Common Flow Visualization Techniques
Currently, there exist numerous techniques for obtaining visual pictures and
presentations of fiuid flows. This section briefly examines some of the more common
techniques. Not ody do these techniques provide a qualitative global image of the flow
pattern, but several of them also allow the researcher to derive quantitative data h m the
flow image. As well, it will be shown in Section 2.1.3 that several flow-measuring
instruments, once surveyed through an entire field, can produce a picture of the
distribution of the flow quantity measured. Again, this displays quantitative data of the
flow-field in a visual presentation Indeed, because of the high quality of Uiformation
provided by flow-visualization techniques, flow visualization has become a very usefbl
and unique tool in fluid dynarnics.
2. f . f Flow Visuelixation by Addiüon of Fomign Mafen'als
Because rnany £lui& are transparent, contarninants need to be introduced into the
fluid so that the motion of the fluid can be tracked. This technique of adding foreign
materials to the Buid works only if the foreign material is visible. The basis of this
technique is that if the particles, forming the material are small enough, the assumption
is that the motion of these particles is equivalent to that of the fluid. Therefore, this type
of visuaiization is known as an indirect method since it is not the fluid motion that one is
capturing in the flow image, but rather the motion of the foreign particles.
Typical foreign materials added to ~isualize gaseous fluid flows are smoke, helium
bubbles, dust particles and glowing h n particles. On the other han& to visualize the
flow of liquids, materials like dyes, particles, neutrally buoyant spheres and hydrogen
bubbles are used. For reliable resdts, the foreign particles are given a densiîy sllnilar to
that of the fluid particle, the reason king that it would minimize the differences between
the movement of the foreign and fluid particles. Thus, in compressible flows, where the
density of the fluid varies, flow visualization results will not be as precise.
In hct, any factors that can cause a difTerence in the foreign and fiuid particle motion
are undesirable. Not ody does a dEerence in density cause this effect but also a
difference in the thermodynamic properties. If the therrnodynamic properties of the fluid
are different h m those of the foreign material, this c m pose a problem. As a result, this
visualization technique is excellent for stationary flows but not for the case of unsteady
flows due to the h i t e size of the foreign particles.
Selecting an appropriate foreign material and a method to capture the flow pattern
images are not easy tasks to accomplish. It has already been described how the choice of
a foreign material depends greatly on the properties of the fluid one is testing. Once the
type of foreign material is chosen for a particular fluid fiow, the photographic equipment
is the next important consideration The illumination of the flow, the visibility of the
foreign material and the recording device al1 play an important role in acquiring good
flow images.
A final point to be made is that quantitative data, such as fluid velocity, cm be
derived fiom a flow pattern M e obtained using this particular method. A single
foreign particle, acting as a tracer, is seeded into the fluid and mves dong with the flow.
Appropriate recordhg devices allow one to measure the velocity of the tracer particle.
The main problem with this type of velocity measurement lies with the assurnption that
the velocity of the tracer partic le is equal to the velocity of the fluid. This, however, rnay
not be the case.
2.1.2 Optical Meîhods for Flow Visualilation
ûptical techniques, uniike the first method, are better suited for testing flows of
compressible fluids since optical methods rely cjn the fluid's varying density. Note that
fluid density is a function of the refiactive index of the fluid, so any variations in density
leads to a change in the index. Because changes in the rehctive index are invisible to
the naked eye, optical methods attempt to make these changes visible and, in doing so,
some property or properties of the flow are determined.
The basic approach to this technique is as follows: a beam of light passes tbrough the
fiuid flow-field and is disturbed because of the inhomogeneous distribution of the
refkctive index in the flow. As a result, the light bearn deflects fiom its original
direction and the phase of the disturbed light wave (i.e. the light wave in the flow) is
shifted with respect to the undisturbed light (Le. the hght not in the flow). Using this
phase change, one can visualize the flow. Aiso, quantitative density data may be
obtained with this method aithough the relatiooship between the fluid density and
rehctive index must be known.
Methods that use the optical approach to gain insight into the density variation of a
flow include schlieren, shadowgraph, and interferometnc techniques. As descriid
above, these rnethods depend on a change in refbctive index However, each of these
techniques measures dflerent quantities. The schlieren system measures the iïrst
derivative of the index of refhction (normal to the light beam) whereas shadowgraph
systems measure the second derivative. Iatderometers, on the other band, focus on the
differences in optical path length to show the index of r e h t i o n field within the flow.
Interested readers c m l e m more about these methods in (Goldstein R, 1983) and
(Merzkirch W, 1987).
To sum up, optical methods are valuable for visualiziag flow in which there exists a
deosity variation in the flow. These flows include compressible flows, plasma flows and
stratified flows. The methods can also be used in the case of studying fluid mixing,
where the fluids have dEerent densitieq and combustion. Not only do optical techniques
give a picture of the flow field but they also give quantitative data such as density,
pressure and temperature variations in the flow. This rnakes these techniques extremely
vaiuabie in the study of fluid flows.
2.1.3 Quantitative Meaiods for Fio w Wsualization
nie two techniques, described above, explain how a visual image of the flow-field
can be captured. From the images, quantitative &ta may be derived. However, the 80w
visualkation technique discussed in this section works in the opposite way. First
quantitative data is coilected with a measurement probe and then fiom the recorded data.
the flow patterns are extracted. This is unique in that quantitative data is not denved
fiom the flow images as before, but rather the quantitative data is experimentaily
collected to fom the flow image. Therefore, instead of stniggling to capture the flow
pattern images, one's focus is primarily on measuring a fluid's property throughout the
range of the flow-field.
Probes are tools that rneasure a specific fluid property at a single point in space within
the flow. Their output is in the form of an electrical signal. This alone cannot give one
an image of the flow. Yet, if the probe is traversed throughout the entire flow-field, one
can obtain a map of the distribution of the measured quantity in the flow-field. This map,
displayed in the fom of cornputer generated graphies, then represents the flow pattern of
the fluid.
Fluid properties, such as temperature, pressure and velocity, are d quantities that can
be measured by probes. A travershg total pressure probe is able to generate isopressure
maps. Hot-wire anemometry probes, on the other han& are instruments for measuring
velocity of the fluid and can give a visual presentation of the fluid's velocity flow-field.
Indeed. these measurement probes can offer great d e t d about a particular flow.
Unlike the other two flow visualization techniques, measurement probes may be used
in dl types of flows. This includes compressible and incompressible fluids. Another
advantage that this technique has is that it allows for more preck rneasutement of fluid
quantities in cornparison with the fnst t w ~ rnethods. Recd that the other rnethods are
essentially for O btaining qualitative flo w visualization pictures. From these p ictures, the
fluid quantities are inferred. This can be deceiving and can htroduce erm, as was
briefly discussed. Measurement probes, however, eliminate this error since the fluid
quantities are measured directly from the fluid flow and not Eorn an image.
There are also several disadvantages that exkt when using rneasurement probes. The
fkst is that the probes, although minute in size, still offer some disturbance to the flow
one is measuring. The other flow visualization techniques do not offer this disturbance.
Another main disadvantage is that the flow pattern images pmduced by meamernent
probes are formed by data collected during a period of time rather thsn during an instant
of tirne. Thus, for consistent results, the experimentalist must ensure that durhg the data
collection (i.e. as the probe traverses the flow-field), the conditions of the flow and
testing apparatus do not change.
2.2 Fluid Flow Created by the Micro Air Vehicle
A few assumptions need to be formed about the fluid flow created by the Micro Air
Vehicle. These assumptions are made in order to determine which flow visualization
techniques are best suited for obtaining information about the MAV's flow-field. Recall
that d flow visualization techniques depend on the nature of the flow king tested, and
thus the predictions about the flow surroundhg the MAV are of utmost importance.
The predictions of the flow- field created by the MAV are based primarily on previous
studies and research in the field of flapping-wing flight. (See section 1.2). In particular,
insect and bûd flight are examined since the MAV wings are designed to mimic the flight
and hovering capabilities of a hummingbird. Therefore, the most sensible prediction
about the fiow-field is to assume that the MAV wings produce a similar flow-field as the
ones created by birds and insects.
The low Reynolds number regime and unsteady aerodynamics makes this particular
application of nuid flow diffcult to study. Insect wings are known to operate at
Reynolds numbers less than ld. Their Moi i characteristics have not been sufficiently
investigated (Sunada et aL, 1997) since most studies concentrate on the Reynolds nurnber
regime in which conventional a i r c d operate (Re 1 103. The flows at such low
Reynolds numbers behave in an rmsteady maMer and are accompauïed by compiicated
vortex break-down, separation and reattachment quite unlike the fiows at high Reynolds
numbers (Liu H and Kawachi K, 1998). The airtlow is unsteady due to the motion of the
flapping wings, king dominated by harmonic reciprocating motions (Vest MS and Katz
J, 1996).
Viscous effects are confined near the wings and the wake shed behind the wings.
Because most vorticity production is caused by viscous stresses, vorticity is confined to
the boundary layer and wake.
If the wings are flapping at a slow speed (much slower than the speed of sound)
through air, then the airflow can be assumed to be incompressible (Vest and Katz, 1 996).
Other predictions about the airflow surrounding the MAV can be made simply by
one's physical touch. By placing a hand below the wings, it is noticeable that the flow
speed is very low. The MAV is most likely not generating airfiow velocities greater than
lOm/s. One can also feel the rough dimensions of the airfiow around the MAV. The
dimensions of the ffow-field depend greatly on the flapping fiequency of the wings.
2.3 Flow Visualization Techniques for a Micro Air Vehicle
From section 1.1 of this thesis, the purpose of this work has been explained. The
main interest lies in king able to vimalize and measure the velocity flow-field around
the M N . By achievllig a qualitative image of the flow-field, one is hoping to see the
vortices created by a set of flapping wings. Before pedorming this study, the location
and strength of the vortices c m only be guessed. However, this information is extremely
important d e n designing the MAV wings. Ideaily, wings should have vortices that
remain attached as long as possible to the wing before sepamtting. As welI, the
. -
quantitative vebcity flow-field is desired because it would not only give the position of
the vortices once again, but it would also be an indication of the strength of the vortices.
Regions in space havhg faster air movement aiso have higher energies. Ideally, MAV
wings that produce fàster moving air beneath them show promise of king able to lift
with a heavier load. As a remit, the velocity flow-field can give many details about a
certain MAV wing design.
Now that numerous flow visualization techniques have been described, the most
promising ones can be selected for this type of application.
Immediately from the discussion of section 2.1.2, Qow visualization by optical
methods can be mled out as a method in testing the flow of a MAV. These optical
methods are for visualhg compressible fluid flows, which are usually hi&-speed
gaseous flows. Obviously, the flow created by a MAV does not corne close to these
velocities and thus this type of technique is not applicable.
Section 2.1.1 describes the flow visualization technique of adding a foreign material
to the fluid and recording the £low patterns on camera. The end result of this method is a
qualitative image of the flow's motion. Since this image would be insightfbl this
technique is attempted. The foreign material chosen to test the air surrounding the MAV
is smoke. This type of material has been used in numerous other studies and has given
favorable results.
Besides fiow visualization using smoke, the probe technique described in section
2.1.3 aiso looks promising. This technique is suited to give a quantitative presentation of
the flow-field. Since the a h of this thesis is primarily concemed with the velocity flow-
field, hot-wire anemometer probes are chosen to give the desired results. As explained in
the section, these probes rnust be able to traverse the entire flow-field dected by the
flapping MAV wings in order to give a map of the flow pattan.
In conclusion, flo w visualization using smo ke and ho t-wire anemo metry show the
greatest promise for O btaining both a qualitative and quant &ive representation of the
flow-field. The technique of using smoke was attempted first and the experiments and
results are given in Chapter 3. Flow visualization with the use of hot-wire mernometers
was performed next and is discussed in Chapters 4 and 5. Chapter 4 de& with the
preparatory work completed before testing while Chapter 5 gives the results and
explanat ions of the test.
Chapter 3
FLOW VlSUALlZATlON USlNG SMOKE
Flow visualization, with the use of smoke, is one of the simplest and easiest
techniques to achieve a visual presentation of the flow direction This technique cm give
an informative map of the flow patterns around flapping MAV wings. These patterns
may then be captured and displayed by camera or video. Because of the technique's
simplicity and quick resdts (i.e. no data analysis is required), this was the first flow
visualization technique attempted.
This C hapter will focus on giving details conceming the various experiments
perfomed using smoke testing. As part of this thesis work, three experiments were
performed. The main differences among them are the way in which the smoke was
generated and the method that was used to capture the flow visualization images. Both
the experimental set-ups and the renilts for these experiments are given. The chapter is
then concluded with a section about the M e r experimentation in this field continued by
undergraduate students in the summer of 1999. A brief outline of their progress is
described and a few words are mentiowd regarding the cment status of smoke testing in
the UTAS subsonic Iab.
3. i Smoke Flow Visualiza tion
Before descn'bing the experiments perfomed in this thesis project, it is useful to
familiatize onese!f with the technique of smoke flow visualization. The technique
consists of several main steps. They are: generating the smoke, introducing the smoke
into the flow, illuminating the smoke, and photographing the resultant flow patterns.
Each of these steps will be described in turn.
The term "smoke" is misleading because it can also include smoke-like materials.
For example, smoke flow visualization can be accomplished with vapor, a e a q aerosols,
mists and fumes. Notice how these materials are visible without the help of optical
methods. As described in section 2.1.1, smoke particles need to be s d l so that they
closely follow the flow pattern studied. Besides king well visible and cornposed of
small particles, the matenais should also be nontoxic because of the exposure the
experimenten have with it. Yet another requirement that must be fuüilled by the smo ke-
like materials is that they should not adversely affect the mode1 king studied. In this
case, it would be the MAV wings and test ng.
There are several ways to generate smoke. "The basic types of producing smoke are
buniing or moldering tobacco, wood or straw; vaporizing mineral oils; producing mist as
the result of various chernical substances; and condensing steam to form a visible fog"'.
The smoke generators that exist, and are cornmercially available, are for vaporizing
hydrocarbon O ils suc h as kerosene.
Of the smoke-like materials, the most practical ones for smoke flow visualization
include tobacco, min, carbon black and oil smoke. Each of these have particle &es in
the range of 0.01 pn to 1 pn which rnakes their particles srna11 enough to follow the
flow yet large enough (> 0.1 5 pm) to scatter light m order to make them visible. For
cornparison, the sue of an average water vapor particle (fog) is much larger than 1 p,
ranging up to 50 p Aithough these combustion particles are superior Li ternis of size,
they do pose a hazard in te= of their toxicity. In that way, vapors and mists are more
acceptable.
Once the smoke material has been chosen and produced by one of the methods
described above, the next concem is introducing the smoke into the Buid flow. This can
be accomplished by releasing the smoke fiom a pipe or a series of pipes positioned
parailel to the main air stream. Another alternative method is the smoke wire. Here,
evaporating oil from an electrically heated wire can create smoke. If the wire is coated
with both a paste of dye and oil, one can produce colored smoke. The smoke wire
method is used for delivering fine sheets of controllable smoke lines. Besides mechanical
means of bringing smoke into a fluid, there are also other methods which will not be
described here.
Illuminating smoke is also an important task in attaining smoke flow visualization.
To obtain the best visible images, the direction of illumination mua be chosen. This is
based upon the direction that shows the maximum scattering characteristics of the smoke
particles. Typical illumination devices are mercury lamps, halogen lamps, spot lights and
strobe lights. For viewing the flow of a d e , one is interested in illuminating only a
plane sheet so that particular flow structures are visible. This is accomplished with a
light sheet created by a laser.
To capture the smoke flow patterns, a wide variety of photographic equipment may
be used. Still, stereo, cinematic and stereo-cinematic cameras are able to take flow
visuaikation photographs. High-speed rnovies cm also be made with cameras capable of
speeds anywhere fiom 1000 to 8000 bnedsecond. A very high intensity light (1000
and 2000 W) must be used for m a h g these movies. Video cameras, which are cornputer
compatible, can offer additional information about the Biid flow king studied. Indeed,
capturing successful flow pattern images requires much effort and experimentation.
3.2 Experiment #7
The first attempt at achieving a clear presedation of the aKnow around the MAV was
performed with the use of fog. The design of the apparatus used for this study is similar
to the one found in the smoke tunnel of the University of Toronto undergraduate
aerodynarnic laboratory and is shown in Figure 3.1.
Figure 3.1. Fogm Generator (~ourcei ~hotognphs Taken by the Author)
To generate the smoke, a small quantity of fog fluid is placed at the bottom of a flask.
A wire is fitted through a glas tube and then wrapped around the outside of the tube,
which is covered in fibre glas cloth. This glass tube is then inserted into the flask such
that the cloth is immersed in the fog fluid and the two wire ends stick out of the fiask. A
rubber stopper with two openings then covers the flask. Each opening has a tube attached
to h By heating the wire with the use of a power supply, fog is generated and builds up
in the flask. By blowing into the flask through one tube, the fog is drawn up the flask by
the fibre cloth, which acts as a candlewick. The fog then exits through the other tube.
To test this apparatus, the steady stream of fog exiting the tube is placed over the
flapping wings and test rig. The room is darkened and the fog is illuminated with a short
interval (= 20 0s) strobe light. The illumination is provided at the front since this gives
the most visibility. A video camera was used to detect the flow patterns and capture them
on still photographs.
However, it was observed that although the flow patterns, created by the flappbg
wings, are visible with the naked eye, they become too difficult to detect with the video
equipment. This is most kely due to the fact that the fog is not dense enough and that
one tube did not offer enough fog to seed the entire flow-field with smoke.
There were also problems in regards to achieving a steady stream of fog. The fog
exiting the tube is dependent on the rate that the fog fluid condenses and also the arnount
of pressure in the flask. There needs to be a perfect baiance between the rate of fog
generation and the rate of fog exiting the flask.
Also, recail fkom the discussion in section 3.1 that fog particles are larger than the
particles of other smoke-like materials Therefore, the choice of using fog to detect flow
images may not have k e n the best decision since there is a greater chance that the fog
particles are not foiiowing the same motion as the fluid particles.
As a result of the dficulty in achieving flow patterns using the fog generator, this
method was abandoned. Experiment #1 has shown that another moke materiai may
produce better resuts and, as weil, more suitable photographie equipment would be
beneficial.
On May 14", 1999 the UTIAS aerodynamic iab had an oppominity to acquire a
high-speed digital video camera for a &y. With this improved video equipment, the flow
images in Figure 3.2 were collected.
Figure 3.2. Smoke F~OW Visualkation of the MAV (Source: Video Images Capturecf by the Aerodynamics Group)
Figure 3.2 shows a series of flow patterns collected at h e s captured a time
increment apart. The pattern were photographed while smoke was blown over a BAT-
12 MAV wing. Throughout the experiment, the wing had a flapping fiequency of 40 H i .
Birming paper generated the smoke for this experiment. Mr. David Loewen, a research
engineer at UTIAS, inhaled the smoke and then siowly blew directly over the wings so
that a steady Stream of smoke exited his mouth and came into contact with the upper edge
of the BAT-12 wing. Diiring this test, ail the iights in the room were tlnned off with the
exception of a strobe light. The strobe Light was positioned to illuminate the h n t of the
images.
Figure 3.2 shows some improvement over the poor resuhs of Experiment #1. This
time, the smoke is actually visible on the pictures. The reason for this is most Likely due
to a thicker smoke being produced. The smoke also covered a greater area, unlike the
area covered by the smoke exiting a narrow tube as in experiment #1. It appears as if the
smo ke formed by cornbust ion methods may be more appropriate for flo w visualizat ion.
This supports what other researchers have found.
The improved video equipment gives more information about the flow because as the
nurnber of W e s per second increases. the easier it is for one to O btain more detail about
the motion of the fluid fiow. That is, each of the h e s shows an image of the flow at
that particular point in time and so one will have less difficulty inferring the fluid motion
over tirne fkom a greater number of images.
However, regardless of d l these improvements, the flow images in Figure 3.2 are
still not acceptable. The shed vortices are very difficult to detect. The main reason for
this is that the through-flow velocity of the smoke is not large enough. Therefore, the
smoke becomes blmed and smudged by the flapping wing before king able to reach the
bottom of the wing, where the vortices are shed.
3.4 Experiment #3
The third major attempt at achieving flow visualktion images occurred on January
17", 2000. On this day, another high-speed camera was used dong with cigarette smoke
to produce flow visualization images.
The UTIAS subsonic lab had the opportunity to use the RedLake imaging
MotionScope PCI 500 L system. This image acquisition system consists of a high-speed
camera, comecting cable, PCI board and Windows-based software. The system
specializes in recording fast motion applications, with recording rates fiom 50 to 500
b e s per second.
At this point in time, the design for the eiiiptical wing existed. nius, the smoke
visualization tests were p e r f o d for both a BAT42 whg, operating at 40Hz, and an
EUiptical wing, operating at 30Hz
The cigarette mwke in experiment #3 was generated in a similar manner as in
experiment #2. Mr. Loewen inhaied the smoke fiom a cigarette and blew it over the
flapping MAV wings. Once again, a strobe light was used to illuminate the fiont of the
wings and flow patterns. The camera was set at a recording rate of 500 b e s per
second and a shutter speed of 1f5000.
(Source: Wdeo Images Taken by the Aetodynarnics Group)
(Source: Video Images Taken by the Aerodynamics Group)
The images captured for both the BAT-12 and Elliptical wing look promising and are
show in Figures 3 .39 3.3b and 3.4. These figures coosist of a series of fhmes (2 ms
apart) taken f?om a movie file made by the image acquisition system The smoke images
for the BAT42 wing show the flow patterns for an entire cycle of the BAT whg's
flapping amplitude. The smoke images for the EHiptical wing are also shown for a
complete cycle of its flapping amplitude. nie images are ordered fiom Ieft to right.
Indeed, the smoke produced by combustion of tobacco offers greater potential for
visualiPng the flow created by a MAV. Figures 3.3% 3.3 b and 3.4 are very informative
because the smoke is thick and dense enough to be captured in the image. AU figures
show vortices for the two different types of wings, although the vortex pattern are more
pronounced in the case of the Ellipticd wing.
There is some indication that severai of the vortices, shown in Figures 3 . 3 4 3.3b and
3.4, may be leading-edge vortices. These vortices are known for their swirling motion
around the top leading-edge of the wiog. As already mentioned, one assumes that
leading-edge vortices are present in the MAV wings although this was never verified.
The Ieading-edge vortices are known for giving wings highlift and thus they would be a
favorable effect if they did occurr on the MAV wing. Thus, the results fiom this
experiment are agreeable because they hint that these types of vortices exist on flapping
MAV wings.
Over the summer of 1999, two undergraduate students, Ms. Yaeko Yamamoto and
Mr. Joseph Lan, continued research in smoke flow visuaiization. They experimented
with numerous smoke-like materials and various smoke generation systems.
-
(Source: Photogisph Taken by aie Author)
One attempt they had was a modification of the equipment in experiment #1. The
major change was that they used other chemicals and oils rather than fog fluid. Instead of
using the flask, they designed a tube to hold the smoke-generating chemicals. The
students also constnicted a series of tubes exiting the tube so that the smoke exited more
than just one tube as before. The purpose of this was to obtain thicker and denser smoke
throughout the entire flow-field of the MAV. Theû apparatus is depicted in Figure 3.5.
Notice how the series of tubes and pipes span over the MAV wings.
Unfortunately, the efforts of the undergraduate students were h i l e u and no suitable
tlow visualization images were coHected. Indeed, one quickly learned that smoke flow
visualization is a very dficult process, which requires patience and much
experimentation.
Currently, there is no M e r flow visuaikation testing k ing conducted in the UTIAS
aerodynamics lab. There are, however, plans to continue testing in the near future.
These h t w e tests will involve using a thin light-sheet for illumination. It has been found
that one can visualize vortical structures in a flow, seeded with smoke, using this type of
illumination.
r laser ¶M!TI
gure 3.6. Laser (Source: Metzkirch 1987, p.33)
The laser light sheet is formed as shown in Figure 3.6. The sheet needs to be placed
normal to the main flow direction. Srnoke, creaîed by cigarettes, will most likely be used
for future tests since it showed the most encouraging results.
Chapter 4
FLOW VlSUALlZATlON USlNG HOT-WIRE
ANEMOMETRY: THE EXPERIMENTAL SET-UP
The hot-wire anemometer has been used for more than 50 years as a valuable
instrument for research in fluid dynamics. With this tool one is able to determine mean
and fluctuating variables in fluid flows. These variables include the direction and speed
of the fluid, fluid temperature, turbulent properties of the fluid, and gas mixture
concentrations. The hot-wire anemometer is also unique in that it may be used in a wide
variety of fluids such as: air, water, oil, glycerine, blood, mercwy, polymer solutions and
luminous gases. Because of its excellent fkquency response, high sensitivity at low fluid
velocities, good spatial remlution and an output signal that is convenient for data
analysis, hot-wire anemometers have grown in popularity among researchers.
As previously mentioned, this thesis work is concemed prirnaril y wit h the
measurement of airflow velocity amund the MAV wings. In order to use hot-wire
anemometers for this type of application, it is important to understand and fàmiliarize
oneself with their operation. Chapter 4 of this thesis is devoted to descriiing the
expetimentd laboratory set-up needed to O btain thne-averaged velocity data using ho t-
wire anemometers. The chapter first delves into a discussion on the p ~ c i p l e of hot-wire
anemometry and then into the type of mernometer used in this experiment. Finally, a
few words are mentioned about additional work pediormed in order to carry out the
experiment. This work includes the construction of a the-dimensional computerized
traverse and confipiiring a data acquisition system to record and store the data
4.1 The Pnnciple of Hot-Wire Anemometry
4.1.1 TheHot-WtePmbe
A hot-wire anemorneter consists of a probe, attached to a cable, and an electronics
package. A hot-wire probe is shown in Figure 4.1 below. This type of single-sensor
probe is the most cornmon, the least expensive and is also the type of probe used in this
experimental researc h project .
Wire Sensor Probc Body
2
Figure 4.1. Hot-Wire Anemometer Probe (Source: Lomas 1986, p.2)
As Figure 4.1 illustrates, the probe contains a sensor, which is typicdy a thin wire
with a diameter as little as a few micrometers. The wire is usuaily made of hingsten or
platinum and may ako be plated with a different metal. The wue sensor, as shown in the
illustration, is suspended between the tips of two prongs and is electrically heated. The
prongs, or sensor supports, are made of stainless steel and tapered to give an end d a c e
of about O.lmm. Epoxy or cetamic materid n o d y forms the body of the probe. At
the other end of the probe body, there is an electrical connecter plated with gold to reduce
resistance. One should also note the small suR of the hot-wire probe. This makes the
anemometer an excellent device for shidying flow details since it does not disturb the
flow to a great extent
4.7.2 Modes of Operation
Hot-wire anemometers measure fluid velocity by the coohg effect the fiuid velocity
has on the heated sensor. A hot wire anemometer may operate in two different modes.
They are a constant temperature mode and the less popular constant c m n t mode. A
constant temperature anemometer (CTA) supplies a sensor with a current that varies with
the fluid velocity in order to maintah constant sensor resistance and thus constant s e m r
temperature. On the other hanci, a constant current anemometer (CCA) supplies a sensor
with a varying temperature so that a constant current is maintained. The only type of
anemometer used in this thesis project was the constant temperature anemorneter and so
Eom this point on, only this type of anemometer is discussed and made reference to.
Figure 4.2 shows a block diagram of a constant temperature anemometer.
Pmbe
gure 4.2. Constant i emperahire Anemometer (Source: Lomas 1986, p.3)
Notice how the wire sensor acts as one resistor of a Wheatstone bridge circuit and the
remaining resistors, two h e d and one adjustable, are supplied by the electronics package
that d e s up the complete hot-wire anemometer. The seosor temperature is kept
constant under all fiow conditions with the use of the feedback amplifier. Before the
system is in operation, the adjustable resistor is set to a higher value than that needed to
Mance the bridge. When the system is powered, the feedhck amplifier increases the
sensor heatmg current, which causes the sensor tempenitine and resistance to increase
untii the bridge is balanced. Whea the case occurs such that the fluid temperature,
composition and pressure are aii constant, then only the fluid velocity affects the heat
transfer from the sensor. An increase in velocity cools the wHe sensor and unbalances
the bridge. The feedhack amplifier then increases the sensor cunent so that the bridge
reaches equiiibrium again. Because the feedback amplifier responds quickly, the sensor
temperature rernains constant as the velocity changes. The voltage dBerence across the
bridge is then proportional to the fluid velocity.
4.1.3 Heat Transfer
As discussed previously, a hot-wire anemometer measutes the fluid velocity by
sensing changes in the heat transfer from a sensor exposed to the fluid motion. This
section will briefly touch upon the equations goveming the operation of constant
temperature anemometer and in particular focus on the heat transfer fiom a wire sensor.
It will be shown, with a simplified analysis, how the voltage output of an anemometer
relates to the velocity of the fluid that the sensor is placed in.
Sensor dimensions: lengai - 7 mm
wre supports (StSt. meciles)
Sensor (thin wire)
(Source: DANTEC Measur&Ïwnt Technology website, 1999)
Figure 4.3 shows a wire sensor attached to the tips of two support needles. A current,
1, is passed through the wire and heat, Qw, is genefated as a result. For equilibrium to be
reached, the heat must be balanced by heat loss to the surroundings, QH. This may be
expressed in the form of an equation:
/dt = Qw - QH (4- 1)
where QE is the thermal energy stored in the wire, Qw is the power generated by electrical
heating and QH is the heat transferred to the mundings . The thermal energy, QE, cm
be caiculated fiorn:
QE = cw T, (4-2)
where Cw is the heat capacity of the wire and T, is the temperature of the sensor. The
power generated, Qw, depends on both the current, 1, and the resistance of the wire, Rw.
This relationship is given in Equation (4.3).
Qw =12& (4.3
The heat loss to the surroundings, QH, is the sum of three contributions. They are the
heat loss by convection to the fluid Q , heat loss by conduction to the support needles
Qd and heat loss by radiation to the cooler surroundings QR. Thus, the value of QH may
be expressed as:
Q, = C @ n + ~ d + ~ R ) (4.4)
Because the losses due to radiation are minimal, QR is ofien neglected in calculations. To
derive equations for each type of contriiution, one must tirst consider a di£Ferential
element made up of a s d length of the sensor. Figure 4.4 illustrates such an element
44
having a length of rk and a cross-sectional ma, A. Note how the origin of the coordinate
system is located at the center of the wire sensor.
Hot Wirc Sensor- A
I Heat out by convection
Figure 4.4. The Heat Balance for a Differential Eiement of a Hot-Wire Sensor (Source: Lomas 1986, p.56)
The conduction heaî-tramfer rate in at the left end of the differential element is:
where k, is the coefficient of themial conductivity for the sensor material and x is the
length meanired dong the sensor. The hansfer rate out at the right end is:
Therefore the total conduction heat transfer rate out of the differential element is given
dLi, = - k , ~ ( i P ~ , / & ~ ) d x (4-7)
The convection heat-transfer rate, on the other han& can be expresseci by equation (4.8).
d ~ ) , = m ; l h ( ~ , -~,)dx (4.8)
The sensor diameter is given by d, h is the coefficient of convective heat transfer and Tf is
the temperature of the fluid. Lastly, the radiation heat ûansfer rate can be determined
fiom equation 4.9.
d~~ = - T:, (4.9)
where o is the Stefan-Boltzmann constant, s is the ernissivity of the sensor and T,, is the
temperature of the surroundings.
In order to express the heat balance equation for the dBerential element of the wire
sensor, the values for Qw and QE need to be modified to account for the element.
Equation (4.2) bec0 mes:
d e , = PCA(~T, Pt& (4. I O)
where, p is the density of the sensor material, c is its specific heat, and t is tirne.
Equation (4.3) also changes to:
4% = (12~, l~)dx
The resistivity of the sensor material is given by p,.
Findy, equations (4.7) through (4.1 1) combine in the form of equation (4.1 ) to g k
the differential equation for the heat balance in a hot wire sensor. This important
equation is given below:
pi(aT$t) = ( r Z p , / ~ ) - & k S ~ ( d 2 ~ / a 2 ) + &(T, -TI) +mim(q4 - ~2Jc6c (4.12)
For equilibrium conditions, the heat storage is zero because n/ût = O. In other
words, Qw = QH. This only works if one can assume that the radiation and conduction
Iosses are small and uniformity exists throughout the wire length. Sethg QE equal to
zero and neglecthg the contnbut ion fiom QR, equation (4.1 2) for a hot-wire anemometer
simplifies to:
( I ~ ~ , / A ) + k , ~ ( i 3 ~ ~ , / i b c ~ ) - d(~ , - T,)= O (4.13)
Several investigators such as King, Corrsin, Davies, Fisher, and Champagne have all
attempted to solve equation 4.13. King's law is the most well known of the heat transfer
laws in hot-wire anemometry. His law will not be denved here but will be given In the
forced convection regime, he found that:
I ~ R ~ = V' = (A+ BU") (4.14)
where A and B are constants and the exponent, n, is approximately 0.5. Equation (4.14)
c learly shows the relationship between the anemometer output voltage, V, taken across
the Wheatstone bridge and the velocity of the fluid, U. It is this relationship that ailows
one to measure fluid velocity with the hot-wire anemometer.
4.2 Selecfing The Type of Hot-Wire Anemometer
There is a wide variety of hot-wire anemometers available for research. Part of
achieving excellent experimental data is h d h g the hot-wire anemometer appropriate for
the application one is testing. This usually depends on the iab fiicilities one is using and
the amount of funding one has to spend. This section will briefly describe the types of
anemometers that exist and specificdy focus on the type used in this experiment. A
short explanation as to why this type of anemometer was chosen will also be given.
4.Zm 1 Sensor Specitications
As d e m i in Section 4.1.1, a hot wire anemometer consists of a wire sensor. The
choice of sensor material depends on the properties the experimeatalist wishes to acquire.
These properties include maximum flow sensitivity and highest possible mechanical
strength. Sensors may be made from nickel, platinum, and tungsten, among other
materials. It is found that sensors made fiom tungsten are superior for most testing
applications due to its figure of ment (i.e. 0.041@2ecm4/~) and also its excellent
mechanical strength. As a result, the sellsors used in th& thesis project were platinum
plated hingsten wires. The plating is necessary in order to weld the wire to the support
needles. This will be d e s c r i i in Section 4.3.
482m2 Probe Spe~ific~tions
Every probe contains a wire sensor, sensor supports, a probe body and an electrical
connecter. The probe may have anywhere fkom one to three sensors for use in one-, two-
, or three-dimensional fluid flows. There exist two different types of probes: wHe probes
and film probes. The sensor, in a wire probe, is a thin wire suspended between two
prongs whereas the sensor, in a film probe, is a thin metal film deposited on an
electrically insulating substrate. The main clifference between the two types of probes is
that wire probes are typicaily used in gases and in non-conducting Liquids. Füm probes,
on the other hancl., are mainly used in water and other conducting liquids. Because the
aim of this thesis is to measure airflow, wire probes are the naturd choice between the
two. These types of probes allow measurements of velocities in gases fiom a few cmls
up to supersonic velocities. Their seasors have high flow sensitivity aiad the highest
fiequency response. However, they are limited in theu mechanical strength and particle
contamination is a constant worry.
Figure 4.5 illustrates three hot-wire probes: a single-sensor probe, a dual-selisor probe
(X-probe) and a triple-sensor probe (Tri-axial probe). The first two sketches are
miniature wke probes while the last sketch is gold-plated wire probe.
Figure 4.5. Hot-Wire Probes Wth Eiier 1,2, or 3 Sensors (Source: DANTEC 1996, p5-7)
Miniature probes have 5 pm diameter, 1 .Zmm long plathum-plated tungsten wire
sensors. The entire length of the wire acts as the sensor because the wires are welded
directly unto the prongs. Gold-plated wire probes, on the other hanci, have 5pm diameter,
3mm long platinun-plated tungsten wire sensors. The wire ends are copper- and gold-
plated to a thickness of about 15pn so that ody 1.25mm of wire, in the middle of the
sensor, is active.
Single-sensor probes are designed primariiy for memernent in one-dimensional
flows. However, dual-sensor probes are used mainly for two-dimeasional flows. In this
case, the sensors are amuiged in an X-array where they form an angle of 90' with one
another. Note that a triple seasor probe has three sensors that are used to measure
parameters in three-dimemional flows. The triple-sensor probe, as show in Figure 4.5,
has three mutually perpendicular sensors consisting of gold plated wires. The senson
form an orthogonal system with an acceptance cone of 7 0 . 4 O . One can use this type of
probe for measuring the three velocity components in an unstationary three-dimensional
flow field.
For each wire probe there are a number of sensor configurations available. By having
a different prong ben& the correct probe for almost any measurement situation can be
found. Indeed, the variety of probe types and sensor configurations is endless.
Recall that the motivation behind this research is to obtain a three-dimensional map
of the velocity flow field around flapping MAV Wmgs. Keeping in mind the discussion
in Chapter 2 conceming the predictions of the airflow around the M N , a suitable hot-
wire probe for this application may w w be selected. Triple-sensor probes seem to be the
appropriate choice. These probes d o w mean velocity and instantaneous flow direction
measurements to be made for a three-dimensionai flow. However, despite the suitable
choice of a triple sensor probe, this type of probe was not used because the UTIAS
nibsonic aerodynâmics lab did not have access to this type of probe design. Instead, the
probes which were available were the single-sensor miniature wire probes, one of which
is depicted in both Figure 4.1 and the first sketch in Figure 4.5. The single-sensor probes
are the cheapest type of probe and are relatively easy to repair. Table 4.1 gives some
properties of the probe used in this research project.
The major ciifference between the single-sensor probe and the triple sensor probe is
that the triple sensor probe dows for more quantities to be measured. W i . a triple
sensor probe, one can make ail the same meanirements that one d e s with a single
probe, but may also masure instantaneous flow direction, high turbulence intensities,
turbulent shear stress and spatial turbulence components. This, indeed, would have been
the ideal tool for measuring the three-dimensional, turbulent flow around the MAV.
However, the single-sensor probes were the only ones availabie. Still, the single-sensor
probes can stiU give the important information required for this research. As the hot-wire
probe coflects mean speed data at every point in space below the MAV wings, a general
trend as to where the air is flowing can be obtained.
- (Source: DANTEC 1996, p.10)
Thus, the single-sensor probe is the one used for aii hot-wire experimentation in this
thesis. Several of the following sections in this chapter describe the procedures to
prepare the hot-wire probe for testing. This includes the repair of the probe (by spot-
welding) and the calibration of the probe.
4.3 Welding Hot-Wire Probes
The miniature wire probes are advantageous with respect to other probes in terms of
the ease in which they are repaired. WE breakage can occur due to contamination fiom
the fluid or smundings, electrical short-circuiting, vibrations or overall clumsiness of
the researc her .
To repair the wire probe, the damageci tungsten wire is removed fiom the sensor
supports (prongs). The prongs are then polished with fine-grade wet-grinding paper and
cleaned with acetone in order to ensure that they are fke fkom traces of grease. Finaily,
the new wire can be fastened between the prongs by spot-welding.
Spot-welding is a tedious process due to the s d l wire size one must work with. The
equipment for spot-welding the wires is shown in Figure 4.6.
(Source: OANTEC Measurement Technology w e b b , 1999)
The equipment consists of a spot-welding generator and a micromaaipulator. A
stereomicroscope is also required but is not shown in the figure. The micromanipuiator
holds the spool of wire and the probe body. It allows one to accurafely place a thread of
wire directly over the two prong tips of the probe body. The wire must not be stretched
over the two prongs since the tightness may later cause wire breakage due to v i i t i o a
Once the thread is aligned and in the proper position, the spot welding generator is used
to spot-weld the wire at the two tips. The welder sends out charges of approximately 150
pA when comected to the prong tip. A few welds are made at each prong tip so that the
wire is well-fastened. The wire at either ends of the prong tips are then broken off by
s d e t electncal charges.
Spot-welding hot-wire probes is a time-consuming task that eventually beco mes
easier with practice. Once the probe is welded, a visual examination can detennine
whether it will be fhctional. The wire sensor between the prongs should not look
damaged in any way. Sometimes the electrical curent passes through one prong tip and
traveh dong the wire to the other prong tip. This results in a very weakened wire sensor,
which cm be detected by looking at the wire through the stereomicroscope. As well, if
the wire has been attached correctly, it d l not appear taut across the prongs but rather
have some flexibility, formhg a slight S-shape between the two prongs.
For the experiment conducted in this thesis project, several hot-wire probes had to be
welded. The lab acquired three probe bodies and purchased a spool of sensor wire.
Throughout the experimentation, wire breakage occurred several times due to
inexperience with using hot wire probes and the very dirty and dusty environment in
which the testing occurred Thus, one can see that the technique of spot-welding hot-wire
probes needed to be learned and applied.
4.4 Calibrafion of Hot-Wire Anemometers
Before any hot-wke anemometry testhg cm be perfbrmed, the hot-wire probes must
be calibrateci. For this research, three hot-wire, single-seosor anemometers were loaned
to the UTIAS aerodynamics lab. Each of these mernometers had their own Wheatstone
bridge. Thus, before calibrating, the bridge for each probe needs to be balanced.
Baiancing the bridge is a process whereby the value of the adjustable resistor in the
Wheatstone circuit is set to the appropriate value depending on the sensor's resistance.
Once the bridge has k e n balanced for each of the three hot-wire probes, the probes
may then be caiibrated. Recd that the flow below the MAV wings is expected to be
approximately in the range of O to 1 Ods. Single-sensor probes are capable of rneasuring
velocities d o m to 0 . 2 M s . Thus, with special care and attention, the hot-wire probes
can be calibrated for the low velocities encountered beneath the MAV wings.
As discussed previously, the output signa. of an anetnometer is in the form of a
voltage readhg. This voltage value, V, is related to the fluid velocity, U. Section 4.1.3
described this dependence and gave one of the most popular relationships, which was
proposed by King (equation (4.14)). However, there is no universal equation that relates
both the voltage output signal fiom the anemometer and the fluid velocity. Thus, the goal
of caiibration is to determine the equation that holds for the particular fluid velocity range
one is interested in.
Figure 43. Calibration of a Hot-Wire An (Source: DANTEC Meawrement Technoloqy mkit8)
In order to calibrate the probes, the experimental set-up, show in Figure 4.7, is required.
The figure illustrates a flow unit that is attached to an air compressor. A valve controls
the air compressor so that the airflow, entering the flow unit, c m be varied. The flow
unit creates a low-turbulent, f ke jet at the exit where the probe is placed. A h , at the
exit, a pitot tube is positioned to measure the Merence between the static and total
pressure. A miromanometer gives the pressure readings. The pressure readings are then
converted to velocity values by using Bernoulli's equatioa
To perform the calilration, the air compressor valve is placed at a setting such that
the akflow out of the flow unit's exit is roughly 1 Om/s. At this flow setting, the pressure
readings are taken with the pitot tube and micromanometer. Then at the same airflow
setting, the voltage signal Born the anemometer is read off of a voltmeter comected to
the output of the CTA bridge and anemometer. This is repeated two or three times to
ensure accuracy in both the voltage and pressure readings. The air compressor valve is
then changed to dcaease the airflow at the jet exit and the measurements are repeated.
This process continues until at least 10 data points are collected in the velocity range of O
to 1 W s . Finally, the pressure readings can all be converted to velocities and calibration
curves can be obtained showing fluid velocity as a function of the output voltage.
The calibration curve for Robe # 1 is shown in Figure 4.8. Notice how the response is
nonlinear and that the awmometer voltage output increases with fluid velocity, U. A
polynornial curve fit was determincd to be an appropriate method to represent the
relationship between voltage and velocity. The c w e nt goes through most points with
the exception of one that o c c d at about 1.3m/s. The reason why the data at extremely
low velocities is hqder to c w e fit is most likely due to the fkct that sensitivity incqxses
with decreasing fluid velocity. Thus, at the lower velocities, it is more diacult to obtain
accurate calibration curves. Attention needed to be given to positioning the pitot tube
and memernometer probe in exactly the same position in the jet exit stream. As well, the
micromanometer, akhough a very precise tool, also intmduced a source of error. It was
difficult to read off the pressure readiugs (in hches o f water) because there was very little
change in the height of water when one was measuring extremely low velocities.
I Figure 4.8. CTA O . utput as a Function of U for Probe #l
Appendix A contains the caiibration cuves for probes 2 and 3. Polynomiai curves
were used to fit the data It can be seen that equations having a quadratic degree in
velocity gave sufncient curve fis.
Al1 three of the CTA probes available were calirated so that in case of breakage, a
probe could be quickly replaced without too great an interruption to the experiment.
4.5 Construction of a Three Dimensional Traverse
Once calibrated, the three hot-wire anemometers are ready to be used in research.
However, an anemometer ody gives a velocity value for that point in space where it is
located. Thus, to determine a map in space of the velocity flow-field beneath the MAV
wings (which is the goal of this research), a method of accurately positionhg a probe in
space is needed. Therefore, the next step in preparing for hot-wire testing is to determine
how to accurately change the position of the hot-wire probe such that it avers all regions
of the flow-field created by flapping MAV wings.
The solution to this problem was to constnict a traverse that would allow the probe to
be easily attached to and offer it three dimensions of W o m . This traverse is shown in
Figure 4.9.
(~oÜrce: Photogtaph Taken by Mt. Loewen)
A sturdy, steel base forms the bottom part of the traverse. Attached to this base are two
UnisLide traverses, one long and one short. The long one provides horizontal movement
of the probe while the short one ailows the probe to move vertically. The third dimension
is provided by a keaded rod attacheci to a stepper motor at the bottom of the steel base.
The rod swivels the entire base and hence the probe in a circular movement around the
wings. There is a scale in terms of degrees at the bottom of the base to allow for accurate
measwement. The scale, although diE~uit to see, is displayed in Figure 4.1 0.
(source: Picture Taken by the Author)
The traverses were assembled together on top of the steel base while taking into
account the testing area under the MAV wings. There is an attachment plate and a probe
holder attached to the short traverse. Once the center of the base is aiigned perfèctly
below the MAV test ng, the hot wire probe attached to the traverse is in perfect position
to take data, as shown in Figure 4.9 and 4.1 1.
(Source: Photograph Taken by Mr. Loewen)
This discussion would not be complete without more details concerning the traverses
themselves and how they are powered. Figure 4.12 illustrates a typical UniSlide short
traverse. In the figure, it is evident that the traverse is stepper motor driven. A stepper
motor is unique in that the motor is incremented a predetermined number of steps to
achieve the desired position. The stepper moton dong with accurate lead screws provide
a means to accurately move to a target position.
I
Figure 4.12. UniSI (Source: UnSIide Catalog M-99, p.3)
Notice nom Figure 4.9 that the the-dimensional traverse is composed of three
stepper-motors, one for each direction of the probe. The stepper-motors used in the
traverse are 3.0 Volt, 4 Amp motors. These specinc stepper-motors take 200 steps per
revo lution of the leadscrew.
The stepper-motors, and consequently the iraverse, can be powered either mnnually
or by a computer with the use of the Velmex ControllerDriver. This piece of apparatus,
illustrated in the bottom right corner of Figure 4.13, has connections for three stepper
motoa. Once comected to the motors, a "run" button on the front panel of the controller
can be pressed and the motor in question wiU move the traverse. On the other han& the
"jog" button dows one to vary the speed of the traverse. This was not found to be very
appropriate for testing purposes because the exact location of the probe had to be
measured by a d e r each time aller moving the traverse. Instead, it was observed that a
more accurate way to move the probe is to have the stepper m t o n conîrolled and driven
by a computer. The controlleddriver also has on its fiont panel an RS-232 connection,
which can be directly conwcted to a serial port on a computer. A 486 computer was
purchased for this experirnent. It was comected to the driver and a few short commands
were written in BASIC Ianguage. These commands instnicted a particular rnotor to move
the traverse a certain number of aeps forwards or backwards at a particular velocity and
ramp çpeed. It was found that 1000 steps moved the traverse exactly 1 inch. Finally, a
precise and accurate way of determining the position of the hot-wire probe and moving it
in space had b e n found.
(Source: Photograph Taken by the AU~IIO~)
Ms. Theresa Robinson, an undergraduate summer student, attempted to d e a more
detailed code, consisting of several paths the probe could take. By imputing variables
such as the number of data points to collect and the geometry of the path to take, one
could let the program nin and it could control the traverse to move in the specified
direction, collecting data points quickly and efficiently, without much work fiom the
experimentalist. However, this program was never implemented because the code was
not written for the particuiar stepper-motors used and thus the code could not run.
4.6 Data Acquisition System
ûne of the final steps in preparing for the hot-wire anemornetry experiment is to
acquire a data acquisition system capable of recording and storing the data. This consists
of selecting the proper AID board and the proper sampling rate and number of samples.
The A / ' board is used to convert the analogue signal fiorn the anemometer into digital
information (Le. voltage readings). Typically, when performing hot-wire meanwments,
the sampling rate of the board is set to twice the highest hquency in the flow. The
number of samples depends on what is sufficient to provide stable statistics. Because
there is not rnuch knowledge available about the flow created by the MAV, the pmper
choice of sampiing rate and samples was difficult to assess. Measurements were taken by
changing these two variables and noting the outcome of the test. It was observed that a
sampling rate of 3.125 kHz and a sample number of 10,000 data points at each location in
space gave accurate measurements. Mr. David Loewen configured the data board to
these settings.
4.7 Measuring Velocity & Turbulence
The hot-wire anemometer is d y an instrument used for measuring the speed and
direction of fluid flows. This section will discuss how mean velocity measurements and
turbulence measurements are obtained fkom the voltage signals of single-sensor hot-wire
anemo met ers.
I V = Velocity Vectw
V, = 1 Sensor, 1 1 Supports I V, = 1 Sensa , l Supports V, = 1 I Sensor
I Figure 4.14. Velocity Components at the ç
(Source: Goldstein 1983, p.118)
Figure 4.14 below illustrates the velocity components at
velocity vector, U, is decomposed into three orthogonal
a single wire sensor. The
components. They are the
normal component UN, the tangential component, UT, and the binormd component, UBN.
The effective velocity of the sensor is then:
I/ V' = (v: + k : ~ : + k;viN)/* (4.15)
where kT and kN are empirically determined factors. If the mean flow is in the UN
direction, then the other velocity components equal zero and thus the mean effective
velocity is equal to the mean normal velocity.
tg = V v (4.16)
It is because of equation (4.16) that single sensor hot-wires are better suited for making
measurements in one-dimensional fluid flows. The sensor should be oriented such that it
is perpendicular to the flow in order to get the maximum response. If the probe sensor
had been pardel to the flow, a minimum reading would have been obtained since the
value of the kT is small, ranging fiom O to 0.2. Therefore single-sensor probes are usuaily
placed perpendicular to the flow.
In the case of rneasuring the airflow velocties beneath the M N , the air beneath the
MAV is expected to mainiy flow vertically downwards fiom the vehicle. Thus, the CTA
probe is positioned such that its sensor is horizontal in space (Le. perpendicular to the
flow).
The experimental procedure may now be dem'bed. While the MAV wings are
flapping, the region in space that is affécted by the wings needs to be determined. This
will be the region tested by hot-wire anemometry since it is the area of most mterest. The
next step is to choose the number of data points one shouM coilect in order to obtain
sunicient results. Once this has d been decided upon, the computer records and stores
an array of 10,000 voltages, Vi, for that point in space where the probe is positioned.
Recall that the data acquis t ion card is set to take 10,000 data points. Then the probe is
moved to the next position, using the computerized three-dimensionai traverse, and
another array of voltages is collected. This process is repeated until data has been
collected for the entire area in Wace aEected by the flapping MAV.
The data analy sis and reduction for this expriment is quite simple. Each array of Vi
values is converted to instantaneous velocities, ui, with the use of the calibration equation
for the particula. probe used. The mean velocity, U, at a certain point in the flow field, is
then given by:
where N is the sample size. In this case, N is equal to 10,000. The fluctuating velocity,
h, can be found by:
Once the rk, is fouci, a measure of the turbulence is simply:
Using equations (4.17) and (4.19), a measure of the mean velocity and turbulence at any
position in the fluid can be obtained.
4.8 Summary of Experimental Set-Up
The overall experirnentai set-up is shown in the xhematic diagram of Figure 4.1 5.
Figure 4.1 5. Schernatic Diagnm of Experlrnental Set-üp (Source: Sketch Drawn by the Author)
The MAV wings are attached to the test rig and the anemometer probe is placed near
the wings. The probe is attached to a cable connecting it with the CTA bridge. The
output voltage si@s of both the probe and bridge are then sent to the data acquisition
board. The board collects the data and sen& it to the computer, which stores the data.
The probe is also comected to the three-dimensional traverse. The traverse receives
input from the controlier/driver, which in turn receives its commands fiom a PC
computer. The schematic diagram, shown in Figure 4.15, is a simple presentation of al1
the factors that need to be addressed before any experimentation can take place.
Chapter 5
FLOW VlSUALlZATlON USlNG HOT-WIRE
ANEMOMETRY: THE RESULTS
Between the months of February 2000 and April 2000, the single-sensor hot-wire
anemometer was used for colIecting velocity data for sets of BAT-12 wings and EUiptical
wuigs. Because one is most interested in the wake shed by the wings, the hot-wire
anemometer probe traversed the ent ire flow-field area beneath the flapping wing S. This
area was determined by roughly estimating the dimensions with the use of one's hand
irnmersed in the wake. However, usually an area greater than the flow-field was tested in
order to ensure that the important details and structures in the flow were captured. AU of
the hot-wire testing occurred while the wings flapped at their operating frequencies. For
the BAT42 wing, its flapping fiequency is 40 Hz while for the Elliptical wing, it is 25
Hz. Both a strobe light and fiequency meter were used to ensure that the flapping
fkquency remaineci constant throughout the duration of the testing (i.e. the traverse of the
probe).
This Chapter gives the results of the hot-wire testing. The velocity data was collected
for a single BAT-12 wing, a set of two BAT-1 2 wings, a set of four BAT42 wings and a
single Elliptical wing. The velocity data was then plotted by a graphics program, Tecplot
Version 7.0, which illustrates the velocity flo w- field as two-dimensio na1 images. These
resuhs with some discussion are given in separate sections of this Chapter dependhg on
the set of wings king tested The accuracy of the results is discussed in the ha1 section
of this Chapter. The section Sicludes a discussion about the ciiflïculties encountered in
hot-wire testing as weil as an investigation into the various sources of enor in this
experiment .
5. i Velocity Flow Field Under One BAT4 2 Wing
The first hot-wire anemometry test was performed on one BAT-12 wing, flapping at
40 Hz. Because of the concem of wire breakage, the hot-wire sensor was never placed
closer than 1.27 cm (0.5 in) beneath the wing and 1.27 cm ftom the test rig center. The
velocity of the flow was measured for a cylindrical area hahg a radius of 27.94 cm
(fiom the test ng center) and a height of 20.32 cm (beneath the wing).
Due to symmetry, one expects that the flow field on one side of the wing's mid-stroke
would equal the other side. This was found to be tme and so velocity data was collected
oniy for one side of the mid-stroke. A few angles on the other side of the mid-stroke
were measured mainly for cornparison reasons. The mid-stroke angle is denoted as O
degrees and so angles on one side are given a positive notation ami, on the other side,
they are given a negative one. It was detemiined that the flow field could be fùlly
represented in a span of 108 degrees on one side of the mid-stroke. Although the traverse
is only capable of spanning 70 degrees, it can be moved physically and repositioned to
span another 70 degrees elsewhere in the flow-field.
Tecplot Version 7.0, a powemil plotting program, was used to illustrate the velocity
data in a series of two-dimensional images of the flow-field. Once the data points are ail
entered into the program, a contour flood is used to fili the entire space of the flow field.
The fiow-field can be presented in two rnpaningful ways: top view slices and side view
slices. The BAT-12 wing is drawn to sale on these plots.
Figure 5.1 illustrates a top view slice of the hw-field. The two black lines, drawn on
the figure, represent the 72 degree flapping amplitude of the BAT42 wing and their
length represents the span of the wing. The velocity color scale shown on the side of the
figure is the same scheme used in ail figures in this Chapter. Note tbat the velocity is
measured in units of m/s.
Top View of Test Rig Velocity Flow FieM 1.27cm Under I BAT Wing
-20 -1 O O 10 20 30 Disfance (cm)
tgure 5.1 a. Top View Slice of Velocity Flow Field Under One BAT42 Win!
From Figure 5. la, one can see that at the end of the wing stroke, the air is pushed
outward by the wing. This was also seen in the smoke fiow visualization images of
Chapter 3. The following Figure 5.1 b shows a series of topview slices moving vertically
downwards Eom the wing. The 5pping amplitude of the wing is drawn on all of the
figures as two black hes, emanating firom the center of the test rig (the white circle).
D m m (an)
Dciinm (em)
Figure 5.1 b. Top Viiw Slices of the '
oiirino (an)
The topview slices of Figure 5 . lb show how the airflow eventually separates Uito
two distinct vortices by about 5 cm below the BAT- 12 wing. The separation grows as the
two concentrations of air move slowly apart. These images show, as did the smoke flow
images, that the vortices king shed off the wing at the two end-strokes move at a
diagonal down- and outwards (Le. away fiom the center of the test rig).
Figure 5.1 not only gives an indication where the vortices are located in the flow-field
but also a sign of their strength The air closest to the wing (i.e. 1.27 cm away fiom the
h g ) has the greatest velocity with speeds reaching 8.5ds beneath the mid-stroke of the
wing. However, as common sense tells us, the fbrther the anemometer probe is fiom the
wing, the less velocity it will measure. It is a fact that the air velocity eventually
dissipates, yet it is interesting to note how long the air in the vortices keep their velocity.
From about 3cm to 10 cm below the wing, the air velocities are still in the range of 3.5 to
5.5 m/s (falling into the green color range).
Figure 5.2a.
Vekcÿ F kw Fieid Under 1 BAT Whg O Degm Sam .-
;ide View Slice of the Velocity Flow Field Under One ;AT4 2 Wing
Much information can dso be gathered fiom exaniining the side-view slices of the
flow-field. These are shown in Figures 5.2% 5.2b and 5 . 2 ~ . Again, the color legend is
the same as previously. Figure 5.2a is the side-view slice of the velocity field at O
degrees (i.e. at the mid-stroke angle) while Figures 5.2b and 5 . 2 ~ contain the side slices
of other angles in the flow-field. The BAT- 12 wing is drawn to scale only in Figure 5.2a
Figure 5.2b. Side View Slices of the Vt
O 10 20 ~ f f a n T u t l ? b C n l w ( a n )
Figure 5.2~. Side View Slices of the Vek y Flow Field Under One BAT4 2 Wing
Figure 5.2b shows the slices for the 1 8 and 36 degree angle case on either side of the
wing's mid-stroke. As the figure shows, symmetry does exist between the + 1 8 O and -1 8"
case as well as the +36* and -36' case. Thus, one is justifïed in assuming that aU other
angles have a similar symmetry and so experimentai data was only recorded for the
negative angles (with the exception of the O", +18", and +36" cases).
Figure 5 . 2 ~ illustrates the remaining side-view slices of the flow-field, created by a
flapping BAT-12 wing. It is apparent fiom these image slices that the airflow, or
vortices, spread out at the larger angles and move away fiom both the test-rig and the
BAT-12 wing.
5.2 Velocity Flow Field Under Two BAT-1 2 Wings
There is rnuch interest in the clap-fling phenornenon that was described in Chapter 1.
As a resuit, the next hot-wire anemometry test was to obtain the velocity flow-field
beneath two BAT42 wings, experiencing the clap-fling effect. Each BAT-12 wing had a
flapphg-amplitude of 72 degrees and this is shown in al1 the topview slices. The two
wings were positioned such that there was a 45 degree Merence between their mid-
m k e angles. k u g h o u t the test, the wings were flapped at a fkquency of 40 Hz.
The results of this experiment are once again illustnited in top and side-view slices of
the velocity flow field under the BAT-12 wings. This tirne, however, data was collected
for a c y h ~ c a i area of radius 20.32 cm and height 25.4 cm. Velocity data was rneamred
for the range of angles +90 to -54. The resuits were piotted utilizing the Tecplot program
with the same color Iegend that was used earlier. The next few pages contain the results
of this experiment. A brief discussion will follow shortiy thereaf'ter.
Top View of Velocity Fbw Field 1.27cm Bebw 2 BAT Wings
Distance (cm) I
' Figure 5.3a.TopViewSlicesofVeloci Yow Field Under Two BAT4 2 Wings
Figure 5.3b. Top View Slices of the Velc
-20 -15 -10 -5 O S t O 15 ZO 25 10 Dliiinei (an)
ty Flow Fkld Under Twa BAT4 2 Wings
1 Figure S.&. Side View Slices of the Veloc Flow Field Under Two BAT4 2 Wings
Figure 5Ab. Side View Slices of the Vek
O 5 10 15 D#noicm,TrtRlgCrilw(an(
Flow Field Under Two BAT42 Wngs
Again, for this experiment, one expects some symmetry to exist. The O degree slice
was chosen to be the slice exactly in between the mid-strokes of the two wings (see
Figure 5.3a). Thus +9 degrees and -9 degrees would be the slices dkectly below the end-
strokes of the two wings. Therefore, if the wings are symmetricai, then ideally there
should be symmetry on either side of the O degree slice. Looking at the side-view slices,
this symmetry is apparent although there are slight Merences. These differences are due
to the experimental testing conditions (i.e. the wings are not exactly identical) and
sources of error, which will be described in Section 5.5.
In comparing the flow-field under one BAT4 2 wing with that of two BAT- 12 wings,
it is clearly evident that the airtlow is stronger for the two-wing case. Closest to the
wings, the flow-field reaches velocities as high as 8.5 m/s. This high velocity value was
also detected in the one-wing case. However, in the one-whg case, this hi& velocity
was only detected in the flow visualization image recorded at 1.27cm beneath the wing
(Figure 5.1a). For the two-wing case, however, the high velocity can still be detected
well below the wings at 5.08cm. The high velocity region, shown in red and orange, also
occupies a greater area in the top view slices for the two-wing case in cornparison with
the one-wing case.
It is also interesthg to note where the higher velocity regions occur. Recail that in
the one-wing case, the highest measured velocities were initially Iocated below the rnid-
stroke of the wing. However, for the two-wing case, the higher velocities occur directly
below the location where the two wings meet and 'clapfling'. The slices beneath the
mid-strokes of the two wings, on the other hanci, experience very low velocities. Thus
the flow-field under the two wings is showing a 'clapfling' effect because, as they meet,
they push the air that was once between them straight down beneath the wing, and
thereby give that region in space a higher velocity.
From O bserWlg the side view images of Figure 5.4, one c m also notice how the clap-
h g effect is directing the air to move vertically downwards as opposed to the motion of
the air in the one-wing case. For the one-wing case, the wing pushed the air outwards
away Eom the rig and wing. As a result, there was not much air movement beneath the
wing to cause a lifting effect. In cornparison, the air under the 2 wings is more confined
and this will contribute to lifting the wings dong with the entire body of the W.
Indeed, air moving vertically downwards fiom the MAV and not spreading out far past
the wing's edges wül be beneficial in the flight of a hovering MAV because such a
concentrated wake should produce more lift.
5.3 Velocify Flow Field Under Four BAT42 Wings
After seeing the flow visualization results of Section 5.2, one may ask, "What
happens when the clapfling effect is doubled?" This leads to the next experiment, which
is to obtain the velocity flow-field underneath a set of four BAT-12 wings. Like the
previous tests, the same experimental procedure is used. However, it was found that the
velocity flow-field covered a greater area in space below the 4 wings. Thus to capture
the unique features of the flow-field, data was coilected for a cylinder with a radius of
20.32 cm and a height of 45.72 cm. Meamrements were taken for slices spanning h m O
degrees up to + 90 degrees, thus forming a quart- of a cylinder. Due to symmetry,
however, the data can be mirrored to make up a complete full cylinder. The results are
79
now presented in the form of top and side-view slices with the BAT wing end-strokes
drawn in as black lines.
Top VIew of Ve Flow Field 1.27cm Below 4 ""t BA Win-
Figure 5.S. Top View Slices of the VeL :ity Flow Field Under 4 BAT4 2 Wings
ocity Flow Field Under 4 BAT42 Wngs
I Top Ykw ollomt R b V . l o b y F b w F k l d j b . 6 ~ ~ 4 B A T ~ 1
I I Figure 5 .5~. Top View Slices of the Velc ty Flow Field Under 4 BAT42 Wngs
The top view slices show a similar pattern to thse formed in the two-wing case. The
highest velocity regions in the flow-field appear to be directly below where the four
wings meet and clap-fling. This time, however, the high velocity of the air (show by red
and orange regions) remains much longer, only disappearing at about 12 cm below the
wings. This indicates the greater strength of the vortices siace, in the two-wing case, the
high velocity region disappeared at about 5 cm. Indeed, doubling the clapfling effect has
made a significant change in the flow-field.
Vmüxüy Fbw FWd Undrr4 BAT Whqr 18 OegmmSlœ
a-
Figure 5.6a. Side View Slices of the Vt ity Flow Fkld Under 4 BAT42 Wings
Figure 5.6b. Side View Slices of the ~ & i t y Flow Field Under 4 BAT42 Wings
The side-view Unages clearly show once again that noticeable velocities can be
detected fàr below the wing. Unfortunately, the test rig is ody capable of measuring a
height of 45.72 cm below the wing. However, by using one's hand, significant air
movement could still be felt at 70 cm below the wings. The extra clap-flhg eEect has,
indeed, helped to increase the air movement beneath the wings. Because the air seems to
be moving verticaily downwards fiom the wings and not spreading out too far past the
wing's edges, this indicates a high lift mechanisa
From Figure 5 . 5 , one can see that if dl four wings are symmetrical then one rnight
expect to see symmetry in the O and 9 degree slices, the 18 and 72 degree slices and the
36 and 54 degree slices. This can be checked by examining the side-view slices, where it
is seen that a similar flow pattern does indeed exist. Any minor dEerences are due to the
sources of error, which will be discussed in Section 5.5.
No further testing ushg the BAT-1 2 wings was performed. The results of the tests all
show that m order to achieve high Iüt, one would want to increase the clapfling effect.
For example, the tests indicate that a MAV with eight flapping-whgs may be more
efficient in generating lift due to the effect of having four simultaneous clap-fling events
with each wing kat. This structure (Le. an eight wLig MAV) will be considerd as a
&tue design because it uses the advantageous clapfling effect to its full potential.
Velocity Flow Field Under One Elliptical Wing
The BAT42 wing was first designed over a year ago. It had been under development
for quite some t h e , so it was robust enough to withstand the duration of the hot-wire
testing. The Elliptical wing, on the other hand, was a new design, not yet fully developed
before this hot-wire testing began. Thus it was found that throughout the testing, the
wing was incapable of remaining intact. The design did not allow for long periods of
flapping. Therefore, its operathg flapping kquency for hot-wire testing was chosen to
be 25 Hz, considerably lower than that for the BAT42 wing. This lower flapping
kquency reduced the stress on the wing and dlowed the experiment to be completed.
The same experimentd procedure was followed for the Elliptical wing test. A
cylinder of radius 27.94 cm and height 20.32 cm was chosen as the testing area. The hot-
wire probe traversed a span of -99 degrees to +36 degrees.
The following Tecplot figures show the results in the fonn of t o p and side-view
slices. On all topview slices, the two end-strokes of the Elliptical whg are h w n to
scale. The Elliptical wing had a 7 2 O flapping amplitude for the tests. The wing is also
drawn to scale on the side-view image for the O0 slice. This slice is the mid-stmke plane
of the wing.
Top View of Vekc
72 d-
127m Beiow 1 Hdrqin.
Distance (cm)
ml ' \ Figure 5.7a. Top View Slices of the Ve
Wing
-- (ml
ocity Flow Field Under One Elliptical
D i a m (an)
Dlirino (an)
tioc?
25 - O L-L- 10 20
owi#rcornTwtRlgCrCr(an)
Figure 5.8a. Side View Slices of the iocity Flow Field Under 0 n i
Figure 5.8b. Sidc One
The one-wing Elliptical tests can be compared with the one BAT wing tests. The first
major clifference between the two flow-fields is that the BAT42 wing separated the air
into two separate, distinct vortices. The flow under the EUiptical wing, however, remains
as one mass until about 8 cm below the flow-field.
Another observation regarding the flow-fields of the two different wing designs is
that the velocities rneasu~ed in the Elliptical wing's flow-field are l e s than those
measured under the BAT uing. This is most iikely due to the fact that the Ellipt ical wing
was flapping at 25 Hz whereas the BAT wing was openited at 40 Hz
Unfortunately, a detailed cornparison between the two wing types cannot be made.
Because the Elliptical wing's construction could not withstand the long duration of the
testing, the flow-field under a set of two and four Eliiptical wings could not be captured
The stress of the clapfling effect would break the Elliptical wings. Yet, h m existing
literatw, it was found that an elliptical wing shape is the most optimum design.
Therefore, it is of interest to study this type of wing design m e r . Before doing so,
however, the wing must be modified in terms of the type of materials used to constnict
them and the way in which they are built. It is hoped that stiffer, more robust elliptical
wings will be able to 1st throughout the testing and outperform the BAT42 wing.
5.5 Fumer Discussion on the Results
The resdts have k e n presented as cornputer-generated velocity flo w-field images for
both the BAT wing and the EUptical wing. The results have shown the importance of
the clsp-fling mechanism in generating lift. They have also shown how the variation in
wing design changes the m d flow-fields. However, before accepting these r e w
attention should be paid to theu accuracy. Nurnerous sources of enor were present
throughout the experiment. This section wül examine these sources.
The location of the testing apparatus posed a pro blem The equipment was placed
near a door, which when opened and closed would create a d d t near the equipment.
This could increase the velocity of the air king measured. A h , any people passing by
the equipment could also create a draft of air, which would aEkt the redts. Ideally, the
equipment should have been placed in an enclosed space where the surrounding air could
not affect the flow-field being measured. For these tests, care was taken to record
velocity data when the door remained closed and no people passed by.
The surrounding air could also af5ect the testing if the air camed contaminants such
as dirt, oil and dust. There have been studies in the past showing the effect of
contaminants on a seosor, which were found to change the caiibration cuve of the probe.
RecaIl that anything that dters the heat tmnsfer of the wire sensor greaîly influences the
output of the anemometer and hence the results. Therefore, the laboratory space near the
testing equiprnent was kept as clean as possible so as not to introduce cuntarninants in the
air.
The location of the three-dimensional traverse is another concern. The traverse was
perfectly aligned at the beginning of each experiment below the wings. However, the
traverse is ody capable of spanning 70 degrees. As a result, the traverse had to be
physicdy moved and repositioned once or twice durhg the testing. This introduces
some error. The ody way to remove this error is to des ign a traverse capable of
rotating a greater circderence aroimd the wings.
A signifïcant source of error in this testing was the deterioration of the wings during
the tests. Numerous times, the stresses of the wing flapping at a high fkquency for hours
caused the wing covering to peel off or the spar to break. The wing then had to be
replaced with an identical wing. However, data taken with a new whg may be slightly
dBerent tha . data taken with an older wing. The newer wing would be stifFer whereas
the older whg might have more flexiility and be able to bend and flex a greater amount.
It was aiready s h o w in these tests that a variation in wing shape could alter the resuhs.
Therefore, each t h e a wing had to be repiaced, the flow-field panans may have
c hanged.
A final source of error that wül be described regards the vibration of the hot-wire
anemometer. Although the hot-wire anemometer is attached to a sturdy steel base, its
attachent to this traverse is more flirnsy. Each time the traverse rnoved, it introduced a
slight vibration in the sensor. This vibration changed the values of the voltage king
recorded Thus, every tirne the hot-wire probe was moved to a new position, the
experimentaiist had to mit a few seconds before recording data to aliow the vibration to
dissipate. The downside to this is that the experiments then take a longer time to
complete, which means the wings experience more stress and perhaps more breakage.
Chapter 6
CONCLUSION 8 FUTURE WORK
This Chapter will focus on giving details conceming the remaining research to be
completed at UnAS within the next few months. A short discussion about
improvements to both the equipment and testhg procedures wili also be inciuded.
6.1 Hot- Wire Anemometry Testing
6.1.1 Testing of MAV Whgs
As seen in Chapter 5, the velocity flow field, using hot-wire anemometers, was
captured beneath sets of 1, 2 and 4 BAT-12 wings as well as for one Elliptical wing. As
mentioned previously, it is hoped that the Elliptical wing design will eventually
outperform the BAT- 12 wing design. Therefore, once the Eiiiptical wing is redesigned to
d o w for extra strength and stifniess, the velocity flow field can then be detemiined for
sets of 2 and 4 Elliptical wings. At that point, a betîer cornparison between the two wing
designs can then be made.
There are also plans to collect velocity data adjacent and above the MAV wings. The
work in this thesis was primanly concemed with collecting data below the wings where
the vortices are shed. However, in modeling the unsteady flow created by flapping
wings, experimental data surrounding the entire wing set is needed.
Any fùture wing designs can also be easily tested with the hot-wire apparatus now
that the equipment has been assembled and is fùnctioning. The hot-* testing wili
enable one to quickly determine the velocity flow field around different wings and assess
whether a particular wing design will be successful or not.
6.1.2 lmpmvements to the Apparatus
Before more hot-wire testing is performed, severai important improvements to the
apparatus and the method of data collection need to be addressed.
First, the 3-dimensional traverse should be modified to give extra support to the hot-
wire probe holder. It was noted that as the traverse moved a step, it Uitroduced small
vibrations that translated to the hot-wire probe holder and thus the wire sensor. As seen
in section 5.5 of this thesis, any vibration of the wire results in a change in the velocity
value rneasured at that point. This, in tum, affects the results of the study. To remove
this source of error, one mut be aware that the base of the traverse is so lid enough and so
the problem anses with the flimsy way in which the probe holder is attacheci to the
traverse plate. A solution cm be found by removing the present attachent and
constructing a solid metd goose neck holder that permanently attaches to the hot-wire
probe holder and traverse plate.
Another improvement will be made to the test rig. The test rig used in this thesis
work was over a year old and had several faiures. The motor was slowly losing its
ability to f'unction and this affected the MAV flapping fkquency. As the motor changed
its speed, while having a constant power input, the flapping frequency also changed
proportionately. For this study, Ï t was crucial that the flapping hquency remaineci
94
constant for ai l tests. Any changes in fiequency would give poor velocity data and an
inaccurate representation of the flow-field. Other problems with the test ng include joints
that needed to be welI oiled continuously and electrical connections that were corning
apart.
The design for the new test rig is shown below in Figure 6.1. Mr. Dave Loewen
designed this test rig and wiU be the primary person involved in its construction It is
hoped that by June 1 ', 2000 the new rig will be operational.
kgum 6.1. New Test Hig Design (Source: Sketch Drawn by MI. Loewen)
The new test rig will allow for more accurate flow visualization data by ensuring that
the data is coilected at constant fiapping fiequemies. The rig also is designed to measure
the net thnist produceci by a set of four or eight wings. By using a 6 - g a g e balance on
the wing root and by senhg the wing beat fiequency as weli as the flapping amphde,
one will be able to measure the instantanenus mot bending moments on the wing spars.
Besides mechanical improvements to the equipment, other revisions are in store for
increasing the ease with which hot-wire data is to be collected. The most notable one is
that a computer code d l control the traverses to trace out paths in space while collecting
hot-wire data simultaneously. In the past, the traverse, although controlled by the
computer, had to be commanded to move each step. It would cut down on the data
collection t h e if the traverse was able to move and stop on its own. Ms. Theresa
Robinson, a 1999 m e r undergraduate student, wrote a code with this in rnind. The
code was written in C* and aliowed one to input the coorduiates in space for the
traverse to follow with several commands allowing the traverse to stop in order for the
hot wire to take data before moving on. However, her code was written for a different
type of stepper motor than the ones that permanently control the traverse, and thus her
code was never implemented.
Earlier in the thesis it was explained that a triple-sensor wire probe would be the most
ideal anemometer for measuring the MAV flow. These probes could give greater
information and thus will be used in future hot-wire anemometry tests.
6. f.3 Other Applications
The wide versatility of the hot-wire anemometer has already become apparent in the
UTIAS subsonic aerodynamics lab. Not only has the hot-wire equipment been used for
this study but also for the thesis work of an undergraduate student, Ms. Rachelle
Lemieux. Part of her study deait with studying the flow created by a mode1 helicopter
rotor. What she discovered was that the flow was quite d o m beneath the rotor and did
not spread out d a i l y pst the edges of the rotor.
6.2 Smoke Testing
Smoke testing in the past was W e d due to complications in producing the smoke
and clearly displayhg the flow-visuaiization images. With the addition of improved
high-speed digital equipment and smo ke generation, the possibility of revisit h g this
technique exists. There are also plans in the fbture to use a laser sheet for vimalizing the
vortices. Existing literanire has shown the potential o f using this type of illumination.
6.3 Free Flight MA V Model
On March 16~, 2000, a MAV model fiee-flew for the kt the . The flight was
achieved with no control surfàces. The modei, designed by Mr. Paîrick Zdunich, Mr.
Dave Loewen and Mr. Derek Bilyk, is show in Figure 6.2.
Figure 6.2. Free-Flight MAV Model (Source: Picture Taken by Mr. Loewen)
The model is powered by four 3.3F, 2.5V capcitors in series, charged to 14 Volts. It
has three 7 inch carbon-fibre rods, stemming downward h m the fwlage, for launch
purPo=-
The wing, which is depicted in Figure 6.3, is based on the BAT- 12 design but has a
3/8 inch extension at the mot, thus making the entire span of the vehicle approxhately
6.75 inches with a total weight of 40 gnims. Currently, more testing is king done with
this model. Control surfaces are king added and the wing shape is king modined to
achieve a wing span of less than the nominal 6 inches.
(source: Scanned Image Taken by Mr. Loewen)
6.4 Analyücal Mode1
Mr. Zdunich is currently developing an analytical tool for the design, development
and optirnization of flapping in the MAV's Reynolds number regime as part of his own
Master's thesis work. He will be applying potential-flow panel methods to model the
unsteady aerodynamics associated with the flapping wing of the MAV. Because the fiow
is unsteady, viscosity effects must be taken into account with his model. As well a large
leading edge vortex is predicted on the leeward side of the wing; so this, too, must be
present in the potential-flow model. A strip theory approach with corrections for
spanwise location will be implemented to change the model to three dimensions. Once
the mode1 is complete, it can be compared to the experimental data colIected &om the
hot-wire testing. This model will be a very useful tool for enabling o w to get valuable
information and i;Isight into various wing designs without laborious experknental work.
6.5 Conclusion
As staîed in the Introduction, the work presented here is a thorough study on flow-
visualization techniques applied to an aerospace application. The two particular
techniques emphasized were flow visualization using smoke and flow visukation using
hot-wire anemometry. While one technique gave a qualitative image of the velocity
flow-field around the MAV wings, the other quantifieci the flow-field. Both techniques
gave similar visual resuits although the tests done with smoke were show to be more
dficult to achieve.
The micro air vehicle project at UTIAS is now entering the third year of its contract.
The previous year's work deait mainly with a trial-andsrror approach when it came to
designing wings. However, with the capability of performing quick flow visualization of
the velocity field, one cm easily determine the pefiormance of each wing design. By
examining how the vortices are k ing generated and shed off the wing, one can modify
the wing's properties, i.e. shape and stifiess, in order to achieve an optimum wing
design. Much of this thesis work was concerned with sethg up the equipment for the
flow-visualization tests. This took a gxat Iength of tirne to prepare since rnost of the
equipment, such as the 3-dimensional traverse, had to be specifically tailored for these
tests. However, now that the testing equiprnent is assembled, flow-vidization tests in
the future can be more efficiently and rapidly completed. W h a matter of a few days,
one can obtain an entire map of the velocity flow-fie Id under a particular wing.
The wfuhess of such velocity flow-field plots have been demonstrated in this thesis.
The flow-field created by the BAT-12 wings and EUiptical wings could be compared.
Also, much insight came fiom studying the performance of BAT wings experiencing the
clapfling effect. These plots showed a higher velocity region where the two wings meet
to clap-hg. Flow visualization has thus been shown to be a ver - insightfùl tool in
studying Bapping-wing flight.
This final chapter has show that more testing using hot-wire anemometers and using
smoke is planned for the near hture. As weli, this Chapter bas also show that the data
coiiected for this thesis may be applied to other research work, such as development of
the analytical model. To sum up, flow visualization has shown its usefulness and
applicability in the development and construction of high-performance MAV wings.
Chapter 7
REFERENCES & BIBLIOGRAPHY
7.1 References
[l] Johnson DC (editor). Micro Air Vehicle Missions and Technology Assessment,
Project Report MA V- 1. Massachusetts: Massac husens Institute of Techno logy, Linco ln
Laboratory, 1997, p. 1.
[2] Palmtop Planes. New Scientist 1997; 2076: p.36.
[3] SRI International, UTIAS. Flopping Wing Propulsion Using Electrostricfive
Polymer Artifcial Muscle Actuators: Fîrst Semi-Annual Report. October 1998, p.2.
[4] Rayner J, Gordon R. Yi-I izat ion and Modelling of the Wakes of Flying Birds.
Biona Report (12) 1997.
[SI Merzkirch W. Flow Visuulizution, T<' edîtion. London: Academic Press, Inc.,
1987, p.25.
Archer RD, Sapuppo J , Betteridge DS . Propulsion chmcteristics olflopping wings.
Aeronautid journal 1979; 355-371.
Bilyk D. The Development of Flapping Wings for a Hovering Micro Air Vehicle.
University o f Toronto Institute for Aerospace S tudies; Master o f Applied Science Degree,
2000.
Brodsky A K Vortex formation in the tetheredflighr of the peacock butterjZy. J o d
of Experimental Biology 1991; 161: 77-95.
Bruun, H. Hot-Wîre Anemomehy: Principles and SigMI Ancrlysis. Oxford: Oxford
University Press, 1995.
Campbell JF, Chambers JR. Patterns in the Sky: Natwal v'l~tl~lization of Aircr@
Flow Fields. Virginia: N A S A Langley Researc h Center, NASA SP-5 14, 1994.
DANTEC Documenation Department, ScientSc Research Equipment Divisioa
Probes for Hot- Wire Anemomehy. Denmark: Publication #196-105-01, 1996.
DANTEC Documenation Department, Scientific Research Equipment Division.
Imîruction Manual: 56C17 CTA Bridge. Denmark.
Dantec Measurernent Technology Intemet Site: riuwdantccmt.com. accessed on July
8", 1999.
Doebeh E. Meanvernent Systems: Application and Design. New York: McGraw-
Hill Book Company, 1966.
Dracos, TH, edito r. Three-Dimensional Velocity and Vorticity Meanving and Image
Andysis Techniques. London: Kluwer Academic Publishers, 1 996.
Ellington CP. The aerodynamics of hovering insect jlight. M. L f t and Power
Requirements. Philos. Tram Roy. Soc. London Ser. B 1984; 305: 145.
Ellington CP. Umteady aerodynumics of insecijlighr. Symp. Soc. Exp. Biology 1995:
49: 109.
EUington CP, Van Den Berg C, Willmott AP. Leading-edge vortices in insectflighl.
Nature 1997; 384: 626-630.
Freymuth P. Propulsive VorticaZ Signature of Plunging and Pitching Ai$oiI. AI AA
J o d 1988; 26: 881-883.
Goldstein R. Fluid Mechanics Meawemenis. Washington: Hemisphere Pubiishing
Corporation, 1 983.
Johnson DC (editor). Micro Air Vehicle Missions and Technology Assessment,
Project Report UA V-2. Massachuseîts: Massachusetts Institute of Technology, Lincoln
Laboratory, 1997.
Kokshaysky NV. Tracing the wake of a f l ' g bird. Nature 1979; 279: 146- 148.
Kornbluh R, P e b e T, DeLaurier J. Flnpping wingpropuision ushg electroslrictive
polymer muscle actuators. EMU 97-067, SRI International, 1997.
Liu H, Kawachi K. A numerical snrdy of insect flight. Journal of Computatiooal
Physics 1998; 146: 124-156.
Liu H, Ellington CP, Kawachi K, Van Den Berg C, WiUmott AP. A computatiomZ
dynamic stu* of hawkmoth hovering. The Journal o f Experimental Biology 1998; 201
(4): 46 1-477.
Lomas C. Fundamentals of Hot Wire Anemometry. New York: Cambridge University
Press, 1986.
Maxworthy T. Experiments on the Weis-Fogh mechanism of lift generafion by insects
»i hoveringflght. J o d of Fluid Mechanics 1979; 93: 47-63.
Merzkirch W. FIow C/isualization, yd edition. London: Academic Press, Inc., 1987.
Rayner J. A vortex theory of animulflight. J o d of Fluid Mechanics 1979; 91 : 697-
763.
Rayner J. Wake vortex dyamics in swimming andjlying vertebrates. Symp. Soc. Exp.
Biol. 1995; 49: 131-155.
Rayner J, Gordon R Visuaiization and Modeiling of the Waks of Flying Birdr. Biona
Report 1997; 12.
Smol'yakov AV, Tkachenko V M . The Meanvernent of Tvbulenf Fluctuations: An
Introduction to Hot- Wire Anemomehy and Reluted Tmducers. Berlin: Springer-Verlag,
1983.
Spedding GR. The wak of a kstrel (Falco tinnunculus) inflappingflighi. Journal o f
Experimental Biology 1987; 127: 59-78.
Spedding GR, Maxworthy T. The genercation of circulation and liji in a rigid two-
dimemionalfling. Journal o f Fluid Mecbanics 1986; 165: 247-272.
Sunada S, Sakaguchi 4 Kawachi K Ai$oil section characteristics at u low Reynolds
mmber. Journal of Fluids Engineering 1 997; 1 1 9: 1 29.
Van Den Berg C , Ellington CP. The three-dimensional leadnig-edge vortex of a
"hovering" model hawkmoth. Philos. T m . Roy. Soc. London Ser. B 1997; 352: 329.
Vest M S , Katz J. Unrteady aerodynumic model of flopping wings. AIAA Joumal
1996; 34: 1435-1440.
Weis-Fogh T. Energetic of hovering flight in humrningbirdr and in Drosophila.
J o u d of Experimental Biology 1972; 56: 79-104.
Wells DJ. Muscle Peformunce in hovering hummingbirh. Journal o f Experimental
Biology 1993; 178: 39-57.
Appendix A
Calibration Cuwes
Calibration Curve for Probe #Z
O 1 2 3 4 5 6 7 8 9 ?O
Velocity (rnls)
Calibration of Hot Wire Probe #3