Professional Documents
Culture Documents
INTRODUCTION
1.1 Motivation & Background
Of all loco motions nature created flying was always the most mystic one. From
Daedalus to Leonardo da Vinci, all tried to imitate the flapping of birds wings without
having the knowledge of aerodynamics which is present at that time. The lift force
generation of insects in hovering flight cant be explained by steady airfoil theory. Thus,
unsteady aerodynamics must be attendant. Most fascinating in flapping wing flight is
hovering of insects and hummingbirds. Flapping wing mechanisms are in one way able to
perform hovering flight and are also used as fixed wings when thermal lift is present for
long distant flights. The capability of combining these two skills with one apparatus
contains a huge potential. Also the extreme maneuverability that underlies in flapping
wings is still unreached by todays MAVs.
Over past few years interest in small unmanned aerial vehicles has increased
greatly. Many effects are going on to design aircrafts as small as possible for special
missions especially for military and surveillance. Most of them which are used today are
traditional methods such as propeller for thrust and fixed wing for gaining lift. Also
rotary system such as in helicopters, maneuverability is far behind the natural flying
creatures.
Now its a nice time to remember the quote given by Leonardo Da Vinci, Once
you have tasted flight, you will forever walk the earth with your eyes turned skyward, for
there you have been, and there you will always long to return. This is not only a quote
by the great inventor but also a hint to design aerial vehicles which mimics the motion of
small birds, bats or insects. By the improved technology we are fond of lightweight
materials, better batteries made this flapping wing Micro Aerial Vehicles (MAVs)
feasible.
Several Flapping Wing MAVs are already developed. Recently micro UAV
Dragon fly is the most successful MAV which have same maneuvering capabilities as
1
rotor powered aerial vehicle. Our aim is to design a MAV which is less than 15 cm in
length and less than 30gms weight. In addition to being compact transportable by a single
person with in his pocket, MAVs has other advantages, including its rapid deployment,
real time data acquisition capability, low radar cross section, and low noise. The potential
for low production cost is also an advantage.
The primary mission of a fixed wing MAV includes surveillance, detection,
communication and the placement of unattended sensors. Surveillance missions include
video recording (day and night), infrared images of battlefield (referred to over the hill
problem) which gives the number and position of opposition, also in hostage rescue.
MAVs are also used in traffic surveillance and for communication in urban areas.
The requirement of fixed wing MAVs includes a wide range of possible
operational environments such as urban, jungle, desert, maritime, mountain and arctic.
Furthermore, MAVs must be able to perform in all weather conditions (i.e.., precipitation,
wind shear and gusts). Because these vehicles fly at low altitudes where trees, buildings,
hills etc., present a collision avoidance system is also required.
Significant technical barriers must be overcome before MAV systems can be
realized. These include issues in small scale power generation and storage, navigation,
and communications as well as propulsion, aerodynamics and control. One of the most
interesting and least understood aspects of small-scale flight is the aerodynamics. The
combination of small length scale and low velocities results in flight regime with a low
Reynolds number. This places MAVs in a regime in a totally different from conventional
aircraft. The gross mass of MAVs and other flying objects is shown the figure below.
The long term goal is to develop aircraft systems with a mass less than 30g, a
wing span less than 8 cm and endurance of 20 to 30 min at speeds between 30 and 65
kmph. It appears that flapping or rotatory wing designs may have advantages over fixed
wing MAVs at a small size. One of them is generation of lift and thrust without excessive
size or weight. The interest in achieving insect like or birdlike flight performance has
focused attention on their wing dynamics and unsteady aerodynamics. However, our
understanding of aerodynamics of bird and insect flight is limited. One thing that is clear
from all investigations is that all creatures use two specific mechanisms to overcome
small-scale aerodynamic limitation of their: flexibility and flapping.
The essence of this two-part paper is the analytical, aerodynamic modelling of
insect-like flapping wings in the hover for micro-air-vehicle applications. Wing design is
an essential aspect of making a robust flapping device. The more popular approach to
obtaining an optimized wing is to completely mimic natural species. Rather, our study is
only inspired by these biological flyers, suggesting that the optimum design for an active
one degree of freedom flapping motion may differ. Error reduction is crucial to the
experimental optimization approach, building confidence that variations in thrust are
attributed to wing topology. To allow for consistency among replicate wings, the
fabrication process must be controlled and accurate along with data acquisition and
experimental setup. Digital image correlation and slow motion photography was used to
find subtle differences and gain more knowledge of the physics behind flapping.
Hovering MAVs serve as the primary application of the optimized wing.
A key feature of such flapping-wing flows is their unsteadiness and the formation
of a leading-edge vortex in addition to the conventional wake shed from the trailing edge.
What ensues is a complex interaction between the shed wakes, which, in part, determines
the forces and moments on the wing. Because of the high angles of attack encountered in
insect flight, separation is observed from both leading and trailing edges in the form of a
leading-edge vortex (LEV) and the usual trailing-edge wake. These wakes, together with
the unsteady motion of the wing, determine the forces and moments on the wing. By
enforcing the KuttaJoukowski condition at the wake-inception points and by observing
Kelvins law that the total circulation in a control volume enclosing the wing and its wake
is constant, the flow equations will be solved in accordance with Laplaces equation.
3
As a result, two non-linear wake integral equations were derived that, between
them, described the flow in its entirety. Although these equations are exact, they do not
have a closed analytical form and must be solved by numerical methods. Vortex methods
are used to implement this solution and study the validity of the model.
CHAPTER 2
LITERATURE REVIEW
2.1 Wing Kinematics
There are numerous reasons a large wing stroke is desired: for a given operating
frequency a larger stroke amplitude will result in larger instantaneous wing velocities.
Also, a larger stroke allows vortices to fully form and stabilize before the stroke reversal.
At a macro scale, this would be accomplished with a gear system. At the scale of an
insect, it is not feasible to produce gears with the necessary efficiency.
symmetry. The yb-axis is normal to the plane of symmetry pointing in the direction of the
right wing. The zb-axis then points downward in the MAV plane of symmetry,
completing the right-handed Cartesian system. In addition, the coordinates in the inertial
frame are denoted as (X, Y, Z) in this paper. The transformation between these two
frames can be accomplished by a rotational matrix R, satisfying
The EOM of the flapping-wing MAV can then be obtained by applying Newtons second
laws, given by
.....(2.1)
..(2.2)
where the external forces F include the weight of the vehicle, aerodynamical forces by
flapping wings, horizontal tail wings, and vertical tail. Those forces also generate
moments about the center of gravity (CG). The velocity can be decomposed as Vb = (u,
v, w), and the angular velocity can be decomposed as
= (p, q, r). The expansion of Eqs. (2.1) and (2.2) can be found in and also
included in Appendix.
2.3 Averaging Theory and Formulation of Forces
2.3.1 Applicability of Averaging Theory
Due to the periodic motion of the flapping wings, the averaging theory is usually
applied to analyzing the dynamics of a flapping-wing robot. The averaging theory is
6
applicable based on the assumption that the wing is much lighter than the body. As a
result, the flapping wing slightly affects the vertical motion of the vehicle. The model
verifies the validity of this assumption. Model flies forward in a velocity of around 1-3
m/s, but the fuselage still remains at an almost fixed height when the wings are flapping.
2.3.2 Averaged Force and Advance Ratio
In addition to the applicability of averaging theory, there was still one thing
unclear before. Although the averaging theory was assumed to be applicable to the
dynamical analysis of a flapping-wing robot, the researchers in control field usually
simulate the lift and thrust force with a simple function, such as a periodic triangular
wave. On the other hand, the researchers in aerodynamics field always formulate the lift
and thrust forces generated by a flapping wing as a function of the advance ratio, J,
defined as
(2.4)
Typically, unsteady-state flight has advance ratio J less than 1. Low advance ratio
is an indication that these flyers must flap their wings at high speed compared to the
speed of their flights in order to stay aloft. Therefore, the regime of J < 1 is dominated by
unsteady-state flight. On the other hand, for J > 1, the flight regime becomes quasi-steady
and approaches steady-state. For example, a fixed-wing airplane operates in the regime of
J near infinite because the wings flapping frequency is zero. The lift and thrust forces
can be expressed as functions of J.
.(2.5)
.(2.6)
where CL(J) and CT (J) are functions of J. Here we claim that the forces calculated from
the lift or thrust coefficient as a function of J can be treated as the averaged force, and can
be applied to the investigation of dynamics and control directly. A simple proof goes
below. Consider a very thing rectangular wing, as shown in the upper left of Figure 2.1,
with length b, width W, stroke angle , and flapping frequency f. Consider a small area
element on the wing, whose flapping motion is shown in the right of Figure 2.1, where vf
7
= 2fl is the flapping induced velocity. According to aerodynamics theory, the lift force
generated by this element is formulated as
..(2.7)
where U2 = u2 + v f 2, dA = W*dl, and
Figure: 2.2 Definition of wing parameters and velocity components and induced
velocity due to flapping.
..(2.8)
To simplify the notation we define l/b = J. Introducing the advance ratio we obtain
(2.9)
As a result, Eq. (2.9) can be reformulated as
.(2.10)
where S = W*b, and Consider the average force during the down stroke during time
interval Td, given by
(2.11)
where the subscript d denotes down stroke. Since the integrant is not an explicit function
of time, we can integrate with respect of time first and null out Td. Therefore,
(2.12)
Define
(2.13)
We obtain that
..(2.14)
Similarly, the average force during the upstroke is given by
..(2.15)
As a result, the average force generated during a complete flapping is given by
..(2.16)
We can see that the average force has the same formulation as Eqs. (2.15) and
(2.16).Although this is not a rigorous proof because many aerodynamics factors are not
considered, such as the stability of the air flow, the flexibility of the wing and so on, at
least this proof gives a qualitative link between the average force used in the dynamics
field and the most common way to formulate flapping lift and trust forces in the
aerodynamics field. In other words, if we have the lift and thrust coefficient curves at
hand, which are usually easy to obtain in aerodynamics journals, we can simply apply the
same methodology of analyzing a fixed-wing vehicle to the analysis of a flapping-wing
robot.
2.3.3 Formulation of Main-wing Forces
10
Having shown that the average forces over one flapping period by Eqs. (2.15) and
(2.16) are independent of time, we conclude that the methodology to analyze a fixed wing
vehicle can be applied to a flapping-wing vehicle. However, there are two differences.
First of all, the force coefficients CL and CT are no longer functions of AOA only, but also
functions of advance ratio. Second, considering the dynamics of the whole vehicle, we do
not use the AOA since it is not rigorously defined in flapping motion. Instead, the set
angle and stroke angle are introduced. A set angle is the angle between the horizon and
the chord of the main wing when no flapping occurs in the wing tunnel. Technically
speaking, this is similar to the definition of AOA for a fixed wing, since air always flows
horizontally in the following relation,
..(2.17)
the installation angle is 0 degree. The set angle can be simplified as
.(2.18)
Moreover, the induced angle equals 0 since the robotic bird does not have any vertical
motion in the wind tunnel. Consequently, in the wind tunnel test, the set angle simply
equals the pitch angle. Since lift and thrust forces can be formulated using Eqs. (2.15) and
(2.16), where the force coefficients are attainable from wind tunnel tests, FXb and FZb can
be obtained by considering the vector addition of the lift and thrust forces, given by
(2.19)
.(2.20)
In real flight, three things are worthy of mentioning. First, the MAV may suffer from nonzero induced angle resulted from vertical motion. The induced angle can be derived by
..(2.21)
present in the literature focus on the standard aircraft model and neglect the inertial
effects of the mass of the wings. The standard aircraft equations of motion, to include the
linearized model resulting from small perturbation theory, are extensively developed in.
Simulations are presented for a hover condition by utilizing a quasi-steady aerodynamic
model. The aerodynamic forces generated by the wings are transformed from the wing
frames to the body frame by using 2-3-1 Euler angles, but the inertial effects of the wings
are neglected. An aerodynamic model is developed, based on, that include rotational and
leading edge vortex effects. The coefficients for the aerodynamic model are determined
from a robotic flapper. The wing dimensions from the robotic flapper and the
mathematical model are used to present simulations of the FWMAV in a hover condition.
Many of the uses of the standard aircraft model for flapping wing flight dynamics are tied
to research areas conducting control research.
Lasek and Sibilski and Buleret al., respectively, derive the equations of motion
for a flapping wing micro air vehicle using the Gibbs-Appel Equations. The model
developed by Lasek and Sibilski is one of the first multi-body models of flapping wing
flight and is used to model ornithopters. Lasek and Sibilski develop a simulation
architecture and limit the wings to two degrees of freedom: flapping and lagging.
Alternatively, flapping is the deviation angle and lagging is the flapping (sweep) angle.
The feathering, or pitch angle, remains fixed in the study. Buler et al. model a flapping
wing micro air vehicle with two degrees of freedom for each wing: sweep in the stroke
plane and an angle of attack relative to the stroke plane in. The authors derived a flight
dynamics model with six degrees of freedom for the central body and two holonomically
constrained degrees of freedom for each wing, resulting in a system with 10 degrees of
freedom. Buler et al. use the model to numerically calculate a Jacobean linearization of
the system around a desired trajectory and present a linear quadratic regulator control
solution to track the trajectory.
Jackson et al. present a trajectory optimization problem for a flapping wing
micro air vehicle. The FWMAV is modeled as a system of three bodies, but the central
body is modeled as a point mass. The inertial effects of the wings are included, but
without the presence of a rigid body for the central body, the inertial coupling between
12
the wings and the central body is not accounted for in the simulations and trajectory
optimization problem.
Grauer and Hubbard derived the equations of motion of an ornithopter using the
Boltzman-Hamel equations and five rigid bodies: one for the central body, one for each
wing, and two determining the linkage for the tail. The work is motivated to provide
control of a relatively larger vehicle; namely an ornithopter with a four foot wingspan.
Each linked rigid body is allowed one degree of freedom. The angle of attack of the
wings is due to passive rotation and a quasi-steady aerodynamic model is used for
simulation efforts. Additionally, the equations of motion are transformed into a form
found often in spacecraft and robotics control in order to allow for the integration of
nonlinear control techniques. Grauer et al. have recently expanded the use of the model to
the testing and system identification of an ornithopte. Bolender derived the equations of
motion for a flapping wing micro air vehicle using Kane's Equations in. The novel
approach conducted by Bolender is to derive the equations of motion with four rigid
bodies: a central body, a tail, and two wings. The tail is used for pitch control of the
central body. Bolender includes the derivation of the inertial and active forces, necessary
for Kane's equations, but limits the derivation of the wings. The wings are allotted two
degrees of freedom, but the effects of the wings on the central body can be reconfigured
through tilting of the stroke plane. The simulations presented in the reference are for the
dynamic model with wing effects included and are simulated in an open loop fashion.
Furthermore, the presented simulations include the motion of the tail. The equations show
that without control, the magnitude of the pitch velocity of the central body steadily
increases.
Gebert et al. derived the equations of motion for a flapping wing micro air vehicle
using Newtonian methods, which requires the calculation of the constraint forces between
the wings and the body. The wings are not neglected, but simulations are not presented to
validate the efforts. Furthermore, Sun et al. claim in his work that the equations of motion
derived in reference contain errors and cannot be used. Dickson et al. present simulation
efforts for a model insect based on Drosophila that include the mass effects of the wings.
The method chosen uses physics engine software, similar to that used to make video
games and animated features, to model the wings and body. Simulations are presented
13
using the physics engine. If equations of motion were developed, which may be
important for the development of relevant control algorithms, the equations are not
presented.
F. Y. Hsiao, T. M. Yang, W. C. Lu : The aerodynamics performance in flappingwing animals consists of delayed stall, rotational circulation and wake capture. These
phenomenon and their functions can be explained by experiments and theories. However,
complete and exact analysis of the flapping flight is not available because of the
aerodynamic and mechanical complexity. As a result, Kim developed a smart flapping
wing with a macro-fiber composite actuator to mimic the flying mechanism to measure
the aerodynamic forces of flapping devices in wind tunnel test. As for the full dynamic
model of a flapping-wing MAV Zaeem built a longitudinal flight dynamics with timeaverage theory, but only in 2-dimension space. In our work, we intend to develop the
three-dimensional model, and the model will then be compared with real flight
trajectories.
inherently nonlinear system. Taylor et al. expand the work to nonlinear longitudinal
dynamics of the desert locust. The authors present a combination of stability and control
derivatives obtained from experiments. The derivatives are combined due to the inability
to distinguish between active and passive stability in insects. The derivatives are then
used for a nonlinear time-periodic (NLTP) model of the longitudinal dynamics. Both the
linear and NLTP model neglect the mass effects of the wings and are longitudinal models
only.
Sun and Xiong used the same rigid body approximation to analyze the hovering
flight stability of a bumblebee. The stability derivatives are obtained from computational
fluid dynamics. The aerodynamic forces and moments are cycle-averaged; the resultant
forces over one flapping cycle are used to determine the equilibrium flight condition in
the vicinity of a hover condition. The analysis results in the determination that, based on
the linearized system, the bumblebee has an unstable oscillatory mode for the
longitudinal axis. In an open loop setting, the aerodynamic pitching moment destabilizes
the longitudinal axis. Improper phasing between the pitching and flapping motion of the
wings will enhance the destabilizing effects of the pitching moment.
Sun et al. expand the analysis to four additional insect species. The same
methodology is used: coupling of the standard aircraft, rigid body equations of motion
with aerodynamic derivatives calculated from computational fluid dynamics techniques.
Sun et al. provided a justification for use of the standard aircraft equations of motion after
presenting a Newton-Euler multi-body derivation of the equations of motion. The
standard aircraft equations of motion are justified by assuming that either or the
gyroscopic effects of the wings are small, compared to the body effects, or that the effects
averaged over one flapping cycle are identically zero. The simulation model is used to
determine the equilibrium solution for hovering for the four insect species: drone fly,
crane fly, hoverfly, and hawk moth. The modal structure has two stable subsidence
modes, one fast and one slow, and one unstable, oscillatory mode. Sun et al. conclude that
the rigid body approximation may not be accurate for the larger insect species studied
(the crane fly and hawk moth). Furthermore, Sun et al. conclude that the assumptions at
15
the core of the analysis, namely that the effects of the wings are either small or timeaverage to zero, need to be validated with additional analysis.
Xiong and Sun continue the work presented and apply the methodology to
forward flight for a bumblebee. The linear stability analysis is conducted at various
forward flight speeds, ranging from 0 m=s to 4.5 m=s. The authors determined that at
slow forward flight speeds of less than 1.0 m=s, the modal structure of the system matrix
is qualitatively similar with the structure obtained in further studies. At forward flight
speeds of approximately 2.5 m=s, the modal structure for the longitudinal flight dynamics
is approximately neutrally stable. The eigenvalues exist in two pairs and both are stable,
although the magnitudes are close to the j-axis and noted as marginally stable. As
forward flight continues to increase, the modal structure switches back to the structure
near hover. The final modal structure contains four real eigenvalues: two stable and two
unstable. The analysis predicted a modal structure identical to the hovering modal
structure for even faster forward flight speeds. However, it is important to note that the
studies are conducted for two separate insect species and used different methods for the
determination of the aerodynamic derivatives.
Wu et al. present a method of obtaining the hover solution for two model insects,
the drone fly and hawk moth, by coupling the equations of motion with the Navier-Stokes
equations. The work is significantly different because the body is no longer assumed to
be fixed. Wu et al. present a method of solving the required parameters by using a
`shooting' method. The hover condition is solved for the longitudinal equations of
motion; the lateral motion of the insect model is neglected. The results for the hover
condition are qualitatively consistent with results obtained from biological studies of
drone flies and hawk moths. The hover solution is obtained while neglecting the coupling
effects of the wings. The validity of the assumptions made by Sun et al. in is evaluated
by Zhang and Sun . The validity of the assumptions is examined by comparing the
solutions obtained for hovering using the approximate theory to the solutions obtained.
Numerical simulations are presented that show, under the effects of the disturbances from
the hover condition, that the models are close for three flapping cycles. The results are
similar for the hawk moth and the drone fly. Zhang and Sun conclude that due to the
16
relatively low flapping frequency of the hawk moth (26 Hz), the approximate model
should be valid for all insects.
The results presented by Sun et al. are obtained independently, and using a
different method, by Faruque and Humbert . Faruque and Humbert model an insect using
the standard aircraft equations of motion and cycle-averaged forces and moments. The
forces and moments are calculated using a quasi-steady/blade-element aerodynamic
model, as opposed to the CFD aerodynamic model . The system matrix and control inputs
matrix, in vicinity of a hover condition, are obtained using frequency-based system
identification techniques. Faruque and Humbert obtain the same modal structure as
presented for an open loop system: two subsidence modes (one fast and one slow) and
one unstable, oscillatory pair. With halteres providing sensing and feedback applied, the
unstable oscillatory mode become stable.
The stable modal structure is independently obtained by Gao et al . Gao et al.
conduct a numerical analysis of hawk moth hovering using coupling of computational
fluid dynamics with the standard aircraft equations of motion. Gao et al.do not assume
simple sinusoidal motion for the wing stroke. Instead, actual wing kinematics are used.
The numerical analysis results in two subsidence modes, one fast and one slow, and a
stable oscillatory mode. The stable oscillatory mode is small in magnitude and close to
the j!-axis. The authors attribute the difference in the stability of the oscillatory mode to
the choice of wing kinematics. Faruque and Humbert expand their work to lateral
stability, in the vicinity of hover. Zhang and Sun present a similar approach to the
analysis of the stability derivatives for lateral motion. The lateral stability derivatives are
determined for the standard aircraft equations of motion, in the vicinity of a hover
condition, using computational fluid dynamics. Zhang and Sun's analysis reveals three
natural modes for the drone fly model: one unstable (fast) subsidence mode, one stable
(slow) subsidence mode, and a stable (slow) oscillatory mode. Faruque and Humbert use
the same techniques as outlined: a standard aircraft flight dynamics model, a quasi-steady
aerodynamic model, and frequency based techniques to identify the stability derivatives.
They explicitly include a passive damping term, as determined by Hedrick et al., named
flapping counter torque (FCT). Faruque and Humbert determine the closed-loop lateral
17
system to be stable with three natural modes: two subsidence modes, one fast and one
slow, and one stable oscillatory mode.
The difference in results between the two studies, a stable versus unstable system,
may be due to the inclusion of the flapping counter torque by Faruque and Humbert.
Bolender examines the open loop stability of a flapping wing micro air vehicle in hover.
The analysis is conducted by examining the orbital stability of the flapping wing micro
air vehicle, due to the periodicity of the flapping wing system. The vehicle is modeled as
a point mass with a quasi-steady, two-dimensional aerodynamic model transformed from
the wing frames. The body of the flapping wing micro air vehicle is first modeled as a
point mass and the stability analysis is conducted using Floquet Theory. The calculation
determines that the orbit of the point mass model is unstable. The analysis is also
conducted with a multi-body model, with the central body modeled as a rigid body. The
orbit for the multi-body model is also unstable.
The standard aircraft model is used to analyze the stability of the flight dynamics
of an ornithopter by Dietl and Garcia, the vehicle dynamics model is presented along
with the aerodynamic model. Dietl and Garcia present the longitudinal dynamics, trim
solutions and a limit cycle. The aerodynamic model used is developed in and used
throughout . The longitudinal dynamics are decoupled from the lateral dynamics in the
standard model. In , in addition to an analysis of the dynamics of the vehicle, control
solutions are presented, based on the discrete-time eigenvalues resulting from the
periodic solution. The periodic solution and stability analysis are conducted using Floquet
Theory, as in. The analysis of the dynamics neglects the inertial effects of the wings on
the central body and limits the wings to bird-like flapping: one degree of freedom with
passive rotation of the wings due to aerodynamic pressure on the wing.
The term micro air vehicle (MAV) or micro aerial vehicle refers to a class of
unmanned air vehicle (UAV) that is restricted by size and may have various levels of
autonomy. Today's MAVs are significantly smaller than those previously developed, with
target dimensions reaching a maximum of approximately 15 centimeters (six inches). The
origins of the micro air vehicle (or MAV) date back to about 1997 when DARPA
(Defense Advanced Research Projects Agency) launched a pilot study into the design of
hand-held (150mm) flying vehicles (Ansari, 2006). Flapping-wing micro air vehicles
18
(MAVs) are centimeter scale flying robots with potential applications in search andrescue, exploration, and reconnaissance. These robotic platforms frequently take design
cues from flying insects in an attempt to achieve a similar maneuverability, efficiency,
and hovering ability. Several prototypes of insect sized flapping-wing MAVs have shown
promise, including the Micromechanical Flying Insect [S. Avadhanula, 2006] and the
Harvard Micro robotic fly [Wood, 2007]. One of the main challenges in MAV design is
the selection of an actuation scheme that can provide sufficient power for autonomous
flight.
Figure: 2.3 Micro sized ornithopters. Top row left to right: Aerovironment/Caltec's
Microbat, University of Florida MAV, University of Toronto Mentor. Bottom row left
to right: Technical University of Delfts Delfly, Nathan Chronisters Hummingbird
and Petter Murens MAV.
20
CHAPTER 3
THEORITICAL BACKGROUND
3.1 Objectives
The aim of this draft is to design and built a flapping wing MAV capable of hovering.
First, an overview about existing platforms is given. Some preliminary studies are shown
and the unsteady aerodynamics are explained. In chapter 3 the conceptual design is
presented with the classification into actuation and mechanism. The equations of motion
are given. Also a general view over the amplitude analysis and first results are presented.
In the subsequent chapter 4 the hardware development such as wings and structure is
shown. The last chapter covers the whole testing part and the results and also the
controllability of the developed MAV.
21
ratio is reduced below a value of 2. While the maximum lift-to-drag ratio for most low
speed fixed wing aircraft including many UAVs (U < 50m/s ) is greater than 10, values
of insects and birds usually less than 10. Furthermore, to achieve these values for MAVs
at low Reynolds number, the wing must emulate bird and insect airfoils and should be
very thin (i.e., t/c < 0.06) with a modest amount of camber
Compared to Fixed wing Flight, Flapping the wings induce in general different
aerodynamic phenomena. Most of the airflow is turbulent and due to permanently
changed wing position and orientation, more the unsteady aerodynamics have to be
considered. Because these information could be of use for the development of a flapping
wing mechanism, in this section shortly the main aerodynamics phenomena of flapping
wings are described, using as main input.
23
24
Figure 3.3: Evolution of leading edge vortex in (A) two dimension and (B) three
dimensions during linear translation starting from rest
The evolution of the leading edge vortex for a translating wing starting from rest
is shown in figure 3.3. For the three-dimensional case as shown in figure 3.3, the leading
edge vortex is more stable and no trailing edge vortex forms. Several different studies try
to explain the stability of the formed leading edge vortex, which is only present for the
three-dimensional case. But still newer studies show, that the LEV has long been
underestimated and is far more complex than assumed so far.
geometry, which induces a bound vortex at the leading edge of each of the wings, and
each created vortex acts as a starting vortex for the other wing. As described by WeisFogh, this annihilation allows the circulation to be builted up more rapidly, because the
Wagner effect is suppressed. Another advantage of the clap is that the created vortices
during upstroke are vanishing during the clap, they cancel each other out as they are
oriented in opposite direction.
Figure: 3.4 Schematic representation of the clap (A-C) and fling (D-F)
Black lines show trajectory of the airflow, dark blue arrows represent the by the airflow
induced velocity, light blue arrow shows the net force on the airfoil.
Many insects make use of the fling to create a rotational airflow circulation, while the
clap is not performed by all insects. According to Ellington the clap is avoided by most of
the insects because the permanent clapping can damage the wings and more a 'almost'
clap is performed. Also for birds similar observations were made, for instance during the
takeoff of pigeons. Although no full clap and fling is performed, the wings almost touch
each other at the back and it is assumed, that in this way similar air circulations are
produced, which give additional lift.
26
27
Figure 3.5 Flow past a dead frozen birds body (upper), and an alive bird (lower) in
the wind tunnel.
Figure 3.5 shows that the alive bird knows how to adapt its body in flight to reduce
the skin friction drag. For example, a study on Teals showed that the body drag
coefficient of an alive bird is about 0.08 and a dead bird is about 0.4. The experiments on
dead birds imply that, if we intend to simulate bird flight, we may not be able to achieve
the same aerodynamic characteristics that nature gives to birds. Hence, we need to look
for ways to reduce the skin friction drag.
On the other hand, bodies of insects have no resemblance with streamlined bodies,
and they dont have feathers. Their bodies are much smaller than those of birds.
3.6.2. Wing structure
Bird wings have three joints elbow, wrist, and metacarpal joint that are covered
with feathers. The elbow and wrist joints allow motion only in one plane they work as
hinges. Wing span and wing area can be reduced to about a half in an instant by flexing
the elbow and wrist joints. The analogy to this motion is the closing/opening of a hand
fan. This instant flexing does not affect either mechanical strength or stiffness of the wing
and allows the mean cord to stay the same in the fully extended or flexed position.
However, hummingbirds almost do not flex their wings, but they use different wing
motions to enable different flights.
28
The wing has to be strong enough to resist bending and torsional moments. The wing
creates a lift force that results in a bending moment around the wing root. Also, when in
flight, a bird has to overcome the twisting moment that acts on its wings.
Birds wing feathers perform significant aerodynamic functions. First, feathers are
actually the material that has to withstand lift and drag forces as well as bending and
twisting moments on the wings. Feathers can help increase or decrease wing span and
wing area to adapt to particular flight conditions. In relation to aircraft, this is similar to
the variable swift of aircraft wings. Second, feathers can reduce flow separation, increase
lift, and act as a stall prevention device, similar to flaps used on airplanes during landing
at a low forward speed and a high angle of attack. When landing or taking off, a high
angle of attack is accompanied by flow separation that could reduce lift and eventually
cause stall. To avoid that, the pilot turns on flaps to increase lift and reduce flow
separation in order to have smooth landing or taking off. Figure 3.6 shows a bird that uses
feathers as flaps to make a safe landing at a low forward speed
therefore, speedy flight. Rounded or elliptical wings allow for better maneuverability.
Tapered wings allow for extremely high speed and maneuverability.
Hummingbirds possess pointed wings with slightly rounded tips. Figure 3.9 is the
photography of the wing of Ruby-throated hummingbird.
Figure 3.10 Elliptical (ideal) wing planform and the lift distribution.
31
-------------------(3.1)
Here, AR is the aspect ratio, b is the wing span, A is the wing area, and is the mean
cord. Figure 3.11 displays a schematic drawing of a bird that shows how to measure the
wing span and chord at the root and at the tip of the wing
32
CHAPTER 4
DESIGN
Design of efficient flapping mechanisms for flapping wing micro air vehicles is
mainly constrained by micro dimensions. However, the energetics of flapping wing
motion is not well understood. Insects and hummingbirds wing flapping at high
frequency is fascinating and still unreached by todays artificial wing mechanisms.
4.2 Aerodynamics
To capture an order of magnitude estimation of the aerodynamics forces steady
airfoil theory is applied. This approach is limited mainly by the velocity of the wings and
the unknown lift coefficients. The lift force generating unsteady aerodynamic effects as
described in 2 chapter cannot be explained with this theory. For estimation reasons it is
applied anyhow. Thus, the following equations are valid only for steady conditions. Lift
& drag forces in steady airflow. The main formulas are given in this section and for more
insight it is referred to the semester thesis of M. Naef.
F = c. /2.v2.A
(4.1)
This is the very general formula to start with. r is the air density, n is the airspeed,
A is the wing area and c is a coefficient that will be defined later. This formula must now
be applied to flapping wings (figure 4.1).
33
Figure 4.1: Flapping wing mechanism with defined angle for equation
The wing angle is marked as . With (t) being the angular velocity of the wing
the velocity in equation 4.1 can be derived with the following formula
v= .l
(4.2)
F(t) = c. /2.A.l2. 2
(4.3)
Both lift and drag forces are calculated with the same formula. The difference is
in the so called lift and drag coefficients, marked as c l and cd. These coefficients are
mostly dependent on the angle of attack and the wing parameters, such as chord and wing
length. For the simulations cl and cd are taken as constant in the first approximation and
their dependencies are considered later in the tests.
4.3 Mechanism
The basic question is how to realize a flapping mechanism with given actuators.
The list of possible actuation is indeed rather small. There are linear actuators and DC
motors. Other kind of actuation, such as piezo elements e.g., was not considered due to
small deflection. Linear motors have smaller force density than DC motors. Together
with the wide range of availability of DC motors considering the dimension they were
selected as actuators for the mechanism.
There were three different concepts that were considered and simulated.
34
35
36
CHAPTER 5
DEVELOPMENT
5.1 Wing Concept
For hovering flight the flapping of the wings are basically described by two
movements.
As seen in the next figure there is a horizontal motion indicated as up and down stroke.
The fly shaped design had to be extended by additional mechanics to keep the
passive rotation defined. These stoppers were placed on the wing axis and adjusted to the
desired angle. This angle of attack was constant over the whole wing span. Also two
supporting rods were placed on the wings in order to keep the structural integrity (figure
5.3). These supplementary structures resulted in heavy wings and slowed down the
passive rotation which is crucial for imitating the movements of flapping wings.
In the hummingbird design two wings are made of one piece (figure 5.5).
39
27:1 : 9 to 36/12 to 81
20:1 : 9 to 36/12 to 60
16:1 : 9 to 36/12 to 48
Together with the additional fixed distances between the gear axis the geometry of the
structure was set. The motors though were still to be placed and were chosen to be close
to each other and to the wing rotation axis allowing to keep the center of gravity
concentrated in the middle of the MAV (figure 5.7).
Figure 5.7: The main structure with the distances between the gear axis
The final structure of the MAV with two wings is seen in figure 5.8.
40
41
42
BEGIN DOWNSTROKE
END DOWNSTROKE
END UPSTROKE
BEGIN UPSTROKE
43
Down stroke is presented on the left column, starting at the top of the figure and
ending at the bottom. Upstroke begins at the bottom of the right column and continues to
the top of the right column. Based on the developed flapping-wing MAV, the dynamics is
studied and compared with flight test data. For a flapping wing, the average aerodynamic
forces over one flapping period are usually formulated
Flapping flight is a very efficient way to transport a unit of mass over a unit of
distance, even though it requires extremely high power output. For this reason, it is an
interesting field and a new generation technology for the flight configuration. There are
two kinds of flight configuration that is investigated in the literatures on natural flapping
flight: bird-like flight and insect-like flight. The focus of this paper is on bird-like flight.
The bird-like aerial robot we are investigating is developed by us.
44
5.3 FABRICATION:
Once the type of MAV or the associated driving mechanism is defined, there are
many parameters to be satisfied for its successful operations. Some of the general
requirements of MAV design and fabrication include:
1. Light weight construction; since wing loading (weight of MAV/wing area) and power
loading (weight of MAV/engine size) are some critical parameters.
2. Airframe structures which are covert and simple to operate and can be easily repaired.
3. Ability to accommodate electronic board, motor, servos, sensors and battery as
required for the MAV mission.
4. Ability to control and provide power or propulsion to maintain MAV in operation.
5. Ability to withstand and negotiate fast winds and maintain aerodynamic stability.
6. Reduced man hour in construction and assembly of the components.
7. Ability to take off vertically since sufficient takeoff area may not be available at the
concerned location Properties ABS-M30 Balsa Wood Ultimate tensile strength (MPa) 40
0 - 75 Compressive strength (MPa) 42 3.5 - 27 Specific gravity (Kg/m3) 1050 170 Coeff. of thermal expansion.(m/moC) 10 x 10-5 45 x 10-5 Thermal conductivity (W/moC)
0.2 0.03 - 0.07 Poissons ratio 0.35 0.48 Operating Temperature (oC) 60 -80 to 120
Density (kg/m3) 1024 130 Shrinkage (%) 0.1 0.6 to 2 s. All the components are made in
ABS (Acrylonitrile - Butadiene Styrene) polymeric material using STRATASYS
FORTUS 400mc machine.
45
46
This limitation of flight travel time mainly is because of physical constraint of rubber
band power.
surmounted by using polymeric materials like ABS, ULTEM etc. Although complex
shapes can be easily made using polymeric materials like ABS, it is observed that the
density is high compared to the conventionally used balsa wood material. However, 1020% weight reduction can be achieved with ABS material if the structures are made
either hollow or porous, which is possible using FDM rapid prototyping machines.
47
CHAPTER 6
STATIC ANALYSIS
Figure 6.1 Contours of static pressure shown on the wing design which has
been imported to FLUENT after designing in CATIA and Meshing.
Figure 6.3: Path lines colored by a particle shown on the wing design which
has been to FLUENT after designing in CATIA and Meshing from ICEM-CFD
Figure 6.4: A MATLAB post-processing graph showing how the Digital Image
Correlation measurements are used to see deflections at a certain point in time.`
49
CHAPTER 7
RESULT AND DISCUSSION
1) Stroke amplitude = 25 degrees
Arc length = r*theta = 130*25*3.14/180 = 56.6944 mm
2) Area= triangle area or other complicated designs by using plot digitizer = Area of
quarter ellipse = *a*b/4 = *130*70/4 mm2 = 7143.5 mm2
3) Volume= area*arc length = *130*70/4* 56.6944 mm3 = 404996.4464 mm3
4) Frequency= no of flaps per second = 3 flaps/second
5) Volume flow= 3 * v = 3*404996.4464 mm3 = 1.4883619 * 103 kg/sec
6) Thrust = Mass flow* velocity = 1.4883619 * 103 kg/sec * 4
= 5.95344 * 103
7) Force =
mass * Acceleration
3
Hence, by theoretical values , we get the Reynolds number as 19600 which found to be
reasonably agreeable with the practical values.
50
7.1 Validation
An experimental Analysis is done to determine the best values for numerical parameters.
The model is then validated against existing experimental data and very good agreement
is found both in terms of flow field representation and force prediction. The temporal
accuracy of the simulations is also noteworthy, implying that the underlying flow features
are well captured, especially the unsteadiness. The model also shows the similarity
between two-dimensional and three-dimensional flows for insect-like flapping wings at
low Reynolds numbers.
In this paper the averaging theory is applied to show the validity of this method.
Consequently, the dynamics of a flapping-wing MAV longer than a flapping period can
be analyzed using the same methodology as what we have done to a fixed-wing aircraft.
Physical parameters are also obtained using experiment data, or through computer aid
design programs. With the developed model and found parameters, numerical simulations
of potential motions are proposed. Although several papers have discussed similar topics,
comparing the numerical simulation with flight test data demonstrate the validity of our
assumptions and derivations. The developed model can be further utilized in flight
control and autonomous flight in the future.
51
CONCLUSION
This study will mainly focus on thrust production and velocity magnitude, since a
high offset for weight is necessary in hovering flight. The results from this work are
approximately taken up-to our level best by assuming in a correct and possible way
which we think will generate a large, well-documented data-base (i.e. 100s of wings) of
experimental results that can be used by numerical modelers for validation purposes, as
well as establish an approximate-efficient methodology to parameterize topologies to
streamline future modeling efforts . As for future work and advancements once the database is complete, a standalone hovering FWMAV can be manufactured.
52
FUTURE SCOPE
Present Progress of Research Work
A computational framework for the fluid-structure interaction problem was
developed by coupling a finite volume, pressure-based, Navier-Stokes solver, a finite
element structural solver, and a moving grid algorithm to handle shape changes and the
resulting grid deformation. The low aspect ratio wing, resulting from the size limitation
of MAVs, induces strong tip vortices increase the induced drag and substantially reduce
the lift-to-drag ratio. Efforts have been made to numerically probe the effect of endplates
in reducing the drag and increasing the lift characteristics of a 15 cm MAV wing. The
results demonstrate that the presence of an endplate parallel to the free-stream velocity
increases lift without a significant increase in drag. The increased lift is largely realized in
the tip region where the endplate reduces the downwash, and increases the effective angle
of attack. However, the effectiveness of the endplate diminishes as the angle of attack
increases, due to stronger wing tip vortices.
The impact of wing flexibility was studied on a 12.7 cm wingspan wing at 6o
angle of attack and Reynolds number of 7.1x104. The results showed that the membrane
wing configuration vibrates, even in a steady free-stream. Similar experimental (Waszak
et al., 2001) and computational (Lian et al., 2003c) investigations on MAVs found that
the membrane vibrates at O(100) Hz. Time dependent vertical displacements of the
membrane surface are assessed. Although the accuracy of the displacement field is
uncertain, as the rigidity of the battens is not precisely accounted for, results obtained
from the current approach are consistent with the experimental observations. In terms of
aerodynamic forces, for small to moderate angles of attack and before the stall limit, the
time-averaged lift and drag coefficients of the membrane wing are comparable to those of
a rigid wing. The membrane wing can extend the stall limit over the rigid wing by a
substantial margin.
The mechanisms of lift generation in flapping flight are investigated for two
hovering modes; one is based on the water treading mode and the other on the insect
53
normal hovering mode. The results for the water treading mode with a stroke
amplitude of 1 chord and at a Reynolds number of 1,700 are compared with similar
numerical simulations. The fluid solver coupled with the moving grid algorithm provides
good predictions in terms of force history and average lift coefficient.
The numerical results for the normal hovering mode are compared with
experimental and prior computational results of Wang et al. (2004). The force history and
peak values for a stoke amplitude of 1.4 chords at a Reynolds number of 75 are similar
with computational and experimental measurements of Wang et al. (2004). For the
normal hovering mode, with a stroke amplitude of 2.4 chords at a Reynolds number of
115, the flow patterns of the present computations share substantial similarities with the
experimental flow visualizations of Wang et al. (2004) and further confirm the
capabilities of the present numerical framework to resolve unsteady, low Reynolds
number flows. Different lift mechanisms are identified as contributing to the lift
generation for two hovering modes at a low Reynolds number of 100. The first mode,
based on the water-treading mode, indicates that the sinusoidal-type lift peaks are
primarily generated by the rapid pitching-up and delayed-stall mechanism. For the second
hovering mode, based on the normal hovering kinematics, the local lift peaks observed
at the beginning of the half-strokes indicates (besides rapid pitch-up and delayed-stall
mechanisms) that a wake capturing mechanism is present. The average lift coefficient
demonstrates that the water treading mode produced more lift than the normal
hovering mode for the studied Reynolds number.
A variation of the Reynolds number by one order of magnitude significantly
changes the force peak values for the hovering mode based on the water treading mode
with a large stroke amplitude (1.4 chords). Although the force patterns at both Reynolds
numbers share similarity, the stronger convective effect of the higher Reynolds number
case exhibit more pronounced asymmetry between forward and backward strokes.
Future Research Direction
Recent numerical and experimental investigations of the flapping flight provided
a better understanding of the unsteady lift generation mechanisms. Nevertheless,
important work lies ahead to clarify force generation mechanisms and implement the
54
gained knowledge to a working flapping wing MAV. The main research directions
proposed to further investigate the aerodynamics of flapping flight are:
Three-dimensional flapping wing simulations with comprehensive evaluation of the
kinematics (stroke amplitude, rotation modes, phase lag, etc.) and flight parameters
(Reynolds number, hovering, forward flight, etc) on the aerodynamic force
110 generation. It is important to probe the associated flow structures to offer clear
insight into the force generation mechanisms.
Behavior of the flexible structure of wings undergoing flapping motion and the effect of
flexibility on the lift, drag and thrust.
Three dimensional interactions of multiple wing configurations (dragonfly) and their
efficiency compared with a single wing pair configuration.
Parametric evaluation of lift, drag and thrust generation and flapping efficiency for
different configurations, leading to optimal selection of kinematics parameters for desired
flying characteristics.
Modeling of the laminar-to-turbulent transition observed in the Reynolds number range
of operation for micro air vehicles.
55
REFERENCES
[1] Wei Shyy, Hikaru Aono, Chang-Kwon Kang, Hao Liu 2013 An Introduction To
Flapping Wing Aerodynamics & Aerodynamics Of Low Reynolds Number Flyers,
Cambridge Aerospace Series.
[2] Christopher T. Orlowski, Flapping Wing Micro Air Vehicles: An Analysis Of
The Importance Of The Mass Of The Wings To Flight Dynamics, Stability, And Control,
University Of Michigan,2011.
[3] Dynamics Of Flapping-Wing Mavs: Application To The Tamkang Golden Snitch*
F. Y. Hsiao1*, T. M. Yang1 And W. C. Lu2 ,Department Of Aerospace Engineering,
Tamkang University, Tamsui, Taiwan 251, R.O.C.
[4] Dragos Viieru, Flapping And Fixed Wing Aerodynamics Of Low Reynolds Number
Flight Vehicles, University Of Florida, 2006
[5] Atilla Yilmaz, Design And Development Of A
Flapping Wing Micro Air Vehicle, University Of Zurich, 2010.
[6] S A Ansari, Non-Linear Unsteady Aerodynamic Model For Insect-Like
Flapping Wings In The Hover. Part 2: Implementation
And Validation, 2006.
[7]
57
[18] J.H. Wu, Y. L. Zhang, And M. Sun. Hovering Of Model Insects: Simulation
By Coupling Equations Of Motion With Navier-Stokes Equations. The Journal Of
Experimental Biology, 212:3313{3329, 2009.
[19] Http://Cdn.Instructables.Com , How To Build An Ornithopter.
[20] Rajkiran Madangopal, Biologically Inspired Design Of Small Flapping Wing Air
Vehicles Using Four-Bar Mechanisms And Quasi-Steady Aerodynamics, University Of
Delaware.
[21] William Paul Walker, Unsteady Aerodynamics Of Deformable Thin Airfoils,
Virginia Polytechnic Institute And State University, 2009.
58