Sie sind auf Seite 1von 13

2005-01-3198

Wake Structure Diagnostics of a Flapping Wing MAV


Manikandan Ramasamy J. Gordon Leishman Beerinder Singh
Department of Aerospace Engineering
Glenn L. Martin Institute of Technology
University of Maryland, College Park, MD 20742

Copyright 
c 2005 Society of Automotive Engineers, Inc.

SUMMARY flapping wing system of high aerodynamic and mechani-


cal efficiency with low weight, there are claimed aerody-
Experiments were performed to better understand the namic advantages in using flapping wings at low opera-
aerodynamic flow field of a flapping-wing micro air vehicle. tional Reynolds numbers relative to other hovering MAV
High-resolution laser sheet flow visualization and particle concepts such as rotors. Certainly, nature has found a
image velocimetry (PIV) analyses have shown the pres- way for insects and some classes of birds to use compli-
ence of folded vortex filaments that are trailed from the tip cated three-dimensional vortical flows and unsteady aero-
and root of the wing, which are combined with a shed dy- dynamic effects to hover efficiently and perform very de-
namic stall vortex with a strong spanwise flow toward the sirable flight maneuvers without much noise generation,
wing tip. This leading-edge vortex gains strength as the all of which are the ideal requirements for MAVs. The suc-
translational motion of the wing accelerates through mid- cessful development of a flapping wing MAV with good ef-
stroke. There is a subsequent shedding of this vortex, but ficiency stems, in part, from the better understanding and
with the simultaneous formation of another leading-edge efficient exploitation of these complex aerodynamic phe-
vortex. The generation of the second vortex occurs before nomena. Even then, however, the viability of a flapping
the first vortex reaches mid-chord, enhancing overall lift. wing concept for an MAV remains to be seen.
This second vortex moves along the chord during supina-
tion, before finally being shed from the trailing-edge of the Experiments have been made using living insects to mea-
wing. A starting vortex forms near the trailing-edge as the sure and understand their flow field [1, 2]. However, this
wing starts to accelerate during the downstroke/upstroke work has met with limited success from a fundamental
of the flapping cycle. This starting vortex grows larger fluid mechanics perspective. This is because of the issues
in size, gaining energy from further shed vortices, until in making detailed quantitative measurements on flapping
the wing reaches the mid-point of the cycle. The folded wings and also because of the difficulties in separating
root and tip vortices that trail from the flapping wing have the aerodynamic forces from the inertial forces on the in-
been found to be relatively strong, and move inward and sect [3, 4]. Larger-scale mechanical models that mimic
axially downward as the wing moves through its flapping the kinematics of a flapping wing have been built to allow
cycle. The close proximity of the starting vortex, as well measurements that help more fully understand the under-
as the trailed root and tip vortices, has a large influence lying physics of flapping wing-borne flight [5–8]. Compu-
on the downwash over the wing. This suggests that any tational fluid dynamic models have also been developed
modeling techniques used to predict the lift on flapping to simulate the aerodynamic flow field surrounding a flap-
wings must fundamentally take into account the three- ping wing [9,10], but have given limited additional physical
dimensional, unsteady effects associated with its complex insight into the flow field. This is, in part, because of the
vortex wake structure. difficulties in predicting three-dimensional wake vorticity
to relatively old ages in terms of wing flapping cycles. In
INTRODUCTION both cases, a complete understanding of the underlying
wing motion kinematics and a faithful reproduction of the
Flapping wing based systems are being considered for aeroelastic deformation of the wing is essential for suc-
application to hovering micro air vehicles (MAVs). De- cessfully simulating the complex three-dimensional flow
spite mechanical complications in producing an effective field in the laboratory. Developing the mechanics of a flap-

1
ping wing that properly emulates insect wing kinematics The failure of steady or quasi-steady aerodynamics to ex-
and aeroelastics is, in itself, an extremely difficult task. plain the enhanced lift produced by the flapping wings
[11] has caused various investigators to analyze this
Comparing various aerodynamic aspects of the flapping problem based on unsteady separated flow mechanisms.
wing with its counterpart, the rotating wing, can help un- Ellington [8] suggested that the insects generate en-
derstand the inherent difficulties involved in making reli- hanced lift using a spilled leading-edge vortex (LEV) that
able measurements in the flow field of a flapping wing. arises from the combination of high angle of attack and
Unlike rotating wings, which have constant angular veloc- low Reynolds number. The presence of a LEV was identi-
ity through out the rotational cycle, the reciprocating mo- fied by performing flow visualization on a hovering hawk-
tion in the stroke plane of a flapping wing has varying an- moth Manduca Sexta [1] and also on the mechanical “flap-
gular velocity (accelerating in the early part of translation per” that mimicked the wing movements of the hawkmoth
followed by deceleration). As a result, defining the three- [18, 19]. Because the LEV is formed during the transla-
dimensional wake structure for a flapping wing becomes tional motion of the flapping wing and not from wing ro-
much more difficult because of the powerful dependence tation (i.e., not because of pronation or supination), Van
on the wing stroke location. For example, the strength of den Berg [18] argued that this vortex is a form of dynamic
the leading-edge vortices (LEVs), which are hypothesized stall. A LEV is expected to shed as it gains energy from
to be mostly responsible for the lift produced by flapping the continuous translation of the wing in the stroke plane.
wing insects, will be a function of the stroke position. Also, However, Ellington made an hypothesis that the LEVs stay
the location and strength of the vortices that trail from the on top of the wing during most part of the flapping wing cy-
tip and root of the wing play a significant role in defining cle because of spanwise flow [8]. It was claimed that this
the overall induced flow field (i.e., the non-uniform angle of spanwise flow prevents the LEV from gaining more en-
attack of the wing), and the convection velocities of these ergy and, as a result, prevents its shedding from the wing.
vortices will be a function of the stroke location. Further- The LEV that is smaller in size and weaker near the root
more, the vortical or apparent aerodynamic mass effects of the wing increases its size and strength until it reaches
at the start of each stroke (and deceleration at the latter 75% of the wing span. It was claimed that this results in
part of each stroke), as well as the large angular rota- a spanwise pressure gradient along the axis of the LEV,
tions at the end of each stroke (pronation and supination), and is the source of the spanwise flow. Clearly, the con-
make the predictive analysis of a flapping wing concept tinuous presence of LEV on top of the wing reduces the
an extremely challenging goal. pressure on the top surface of the wing and enhances the
overall lift produced. The spanwise velocity was reported
Applying the existing knowledge base of conventional, to be of the order of the tip velocity of the flapping wing.
linear, small disturbance quasi-steady aerodynamic the- Van den Berg & Ellington [18] also proposed a vortex ring
ory to understand the physics of flapping-wing flight [11] structure in the wake of a flapping wing, which is the result
is fundamentally inadequate, despite some recent reat- of combined shed and trailed wake vorticity.
tempts and claimed successes [12]. There are also
other problems as well, including the differences in many Birch & Dickinson [20], in an alternative hypothesis, sug-
aerodynamic aspects of the problem such as the very gested that the spanwise flow that prevents the LEV from
low operating Reynolds number of the wings and also shedding is dependent on the Reynolds number (based
in effectively characterizing the degree of unsteadiness on wing chord) of the flow. From the results of an ex-
of the flow. It has been shown that the aerodynamic periment where the spanwise flow of the flapping wing
performance of stationary airfoils decreases at very low was restricted using fences and baffles, Birch & Dick-
Reynolds numbers [13, 14]. Other than the domination inson [20] suggested that the spanwise flow may not
of viscous forces (a characteristic of low Reynolds num- stabilize the LEV. However, they hypothesized that at a
ber flow), the measure of unsteadiness in flapping wings Reynolds number matching the flows relevant for most in-
(which is usually defined by a form of reduced frequency) sects (Re =150), the downward flow induced by the tip
is considered to be large. Defining reduced frequency vortices would be the primary source that limits the growth
using the peak flapping velocity in the stroke (the usual of the LEV. It should, however, be noted that the experi-
convention), unfortunately, suggests that the reduced fre- ments performed by Ellington & Usherwood [21] on a ro-
quency is an inverse function of the aspect ratio of the tating wing (to eliminate the unsteady effects) at different
wing [15]. The often used argument that the larger aspect Reynolds numbers (10,000 to 50,000) suggested that the
ratio wings (such as helicopter rotor blades) have smaller lift coefficients at higher Reynolds numbers dropped sig-
unsteady effects and smaller aspect ratio wings (such as nificantly when compared to low Reynolds numbers. This
flapping wing) have higher levels unsteadiness is not very suggests the less pronouced formation of LEVs at higher
useful or insightful. Studies have been conducted to un- Reynolds numbers. The effect of Reynolds number on the
derstand the effects of low aspect ratio wings on the per- stability of LEVs, or on its mere existence, is still not clear.
formance of micro-scale fixed wings that operate at low It is, therefore, essential to have a better understanding
Reynolds numbers [16, 17]. However, no equivalent sys- of the effects of Reynolds number because flapping wing
tematic studies have been made for flapping wing sys- MAVs will operate in this Reynolds number range (103 to
tems. Consequently, current results for the effects of re- 105 ).
duced frequency remain speculative.

2
Multiple hypotheses have been put forth by various re-
searchers for the enhanced lift produced by flapping
wings. While Willmott & Ellington [1] suggested the pres-
ence of LEV as the principal source of enhanced lift, Weis-
Fogh [22] argued that a mechanism called “clap and fling”
is responsible for higher lift production. Other variations
of this mechanism include “clap and peel” and “near-clap
and fling.” However, it is found that most insects rarely
clap in free flight and, as a result this may not explain the
physics behind the overall increased lift shown by flapping
wings [23]. Later, Birch & Dickinson [24] (from a PIV anal-
ysis) suggested that the enhanced lift has its source not
only from the LEV, but also from wake capture. The total
wing force, based on that proposed by Birch & Dickinson,
comes from four sources: (1) the acceleration-reaction
force when the wing starts its motion, (2) force from LEV
during translation, (3) force that results from increased cir-
Figure 1: Insect based flapping wing model developed at
culation as a result of rotation at the end of the strokes,
the University of Maryland. (Tarascio et al. [25])
and (4) an unsteady wake-capture when the wing flips and
starts accelerating on the upstroke of its motion. Net Force
Supination
Pronation
The aforementioned observations and perhaps contra- Wing Path
dicting hypotheses suggest the need for much more fun-
Stroke
damental research on flapping wings, including high- Plane
resolution quantitative measurements, that can help bet- Wing
ter understand their complex flow structures. Clearly, the Downstroke Section
successful design of a mature flapping wing MAV concept
largely depends upon the complete understanding of un- Upstroke
steady aerodynamics that is inherent to its low Reynolds
number flight regime. The objectives of the present study
were to develop effective techniques to allow high-fidelity
measurements in the flow field of a flapping wing and to
further a fundamental understanding of its overall aerody- Figure 2: Basic wing kinematics of a flapping insect that
namics. The work suggests features of the flow similar to constitute one full cycle.
that found by other investigators, but also several impor- attached to the shaft, make contact with Delrin ball ends
tant new observations were made, including the role of at the end of each half-stroke. This causes the shaft to
the trailed vortex wake system. pitch and, hence, generate the wing flip at the end of each
flapping stroke.
EXPERIMENTAL SETUP
The rotation of the shaft or “flip” at the end of each half-
A biomimetic insect based flapping wing MAV has been stroke is generated by the pitch assembly, which also
built at the University of Maryland [25] that is capable of serves to fix the pitch angle of the shaft during the transla-
emulating insect wing motion kinematics – Fig. 1. The tional phases of the wing motion [25]. The pitch assembly
model, which is similar to the typical motion of an insect consists of main shaft that is rigidly attached to the cam
wing as shown in Fig. 2, consists of four parts: (1) down- and is, in turn, held by a Delrin slider and compression
stroke, in which the wing translates with a constant pitch spring. In combination with the pitch stop, the entire as-
angle; (2) supination, where the wing flips through a large sembly is bi-stable in that it allows the shaft to rest in only
angle of attack range to produce a positive pitch angle on two positions. The model has 80◦ flapping stroke in the
the upstroke; (3) upstroke, again a translation motion with horizontal plane.
constant pitch angle; (4) pronation at the end of upstroke,
where the wing flips back again to have a positive pitch Aluminum/mylar wings were attached to the flapping wing
angle for the downstroke. mechanism, the planform of which is based on a scaled-
up fruit fly wing (similar to the Robofly wings in Ref. 7),
The required flapping and pitching motion was produced as shown in Fig. 3. The wing was made using a 0.02 inch
by a brushless motor, which is controlled by a sensorless thick aluminum frame. The pitch angle of the flapping wing
speed controller. This was operated in combination with during the translational stroke was set to 45◦ . The operat-
a microprocessor-based precision pulse generator. The ing Reynolds number based on maximum stroke velocity
motor shaft is rigidly attached to a rotating disk, which in and mean chord was approximately 19,500.
turn is attached to a pin that drives a scotch yoke. As the
shaft is actively flapped, pitch actuators, which are rigidly The flapping wing model was mounted on a test stand (as

3
Pressure Chamber

Smoke inflow

Smoke inflow

y Honey comb
structure
θ

r
z Laser sheet
x
x Camera Wing

Laser optics

Image plane
Figure 3: Flapping wing used in the experiment.

Figure 5: Schematic of the experimental apparatus for the


flapping wing.

to capture the images of the LEVs. A phase delay was in-


troduced so that the laser could be fired at any flapping
phase angle. A simple schematic explaining the experi-
mental apparatus used for the flow visualization is shown
in Fig. 5.

Seed was produced by vaporizing a mineral oil into a


dense fog. Oil was broken down into a fine mist by adding
nitrogen under pressure, forced into a heater block and
heated to its boiling point, where it became vaporized. As
the vapor escaped from the heat exchanger nozzle, it was
mixed with ambient air, rapidly cooled, and condensed
into a fog. The fog/air mixture was passed through a se-
ries of ducts and introduced into the flapping wing flow
field using a plenum and honeycomb structure, as shown
in Fig. 5. The plenum reduces the velocity of the smoke
Figure 4: Experimental apparatus for flapping wing flow that enters the measurement flow field to such an extent
visualization. that the smoke never reaches the region of focus unless
the mechanism is operating. The honeycomb screens
shown in Fig. 4) at about ten effective wing spans from the help ensure that the flow is eddy free.
ground. The experimental apparatus was placed inside
a flow conditioned test cell of volume 362 m3 . Because
the flapping wing model was tested with a single semi- PIV SYSTEM The important parts of the stereoscopic
span wing in the present work, a large image plane was PIV system included a pulsed laser light sheet for illumi-
constructed to provide nominal symmetry to the flow. nating the region of focus in the flow field, a CCD cam-
era to acquire the images, and a calibration grid to obtain
quantitative values from the acquired images. The cam-
FLOW VISUALIZATION Flow visualization images were era was operated synchronously with the laser in double
acquired by seeding the flow using a mineral oil fog exposure mode to acquire two non-interlaced full frame
strobed with a laser sheet. This light sheet was produced images during a single frame interval. Images were ac-
by a dual Nd:YAG laser and was located to the desired quired using a 8-bit CCD camera having a sensor array of
orientation in the flow using an optical arm. Images were 1K X 1K (1 MP) pixels.
acquired using a Nikon D-70 6.1 mega-pixel digital cam-
era that was placed perpendicular to the laser light sheet. To measure the flow field in a plane perpendicular to the
The laser was triggered using a once-per-revolution sig- stroke plane of the flapping wing, the camera was focused
nal, which was obtained from an encoder attached to the on the laser light sheet located along the wing span. The
mechanism. The laser light sheet was placed parallel to camera covered the region of interest, which was 135 X
the span of the wing to capture the root and tip vortices, 100 mm. This means that each pixel is a distance of about
and was placed along the chord at different span locations 0.1318 mm and 0.0976 mm in the x-axis and y-axis, re-

4
spectively. A grid of 81 horizontal nodes by 60 vertical towards trailing-edge before being shed – see Figs 9(a)
nodes was constructed in the aforementioned region of through 9(e). The shed vortices that result from supina-
interest. This results in 1.6 mm between adjacent nodes tion can be clearly seen in Fig. 9(e).
in either direction. A laser pulse separation delay of 30 µs
for a fluid flow of approximately 8 m/s (tip velocity of the It is apparent from the foregoing discussion that the LEV
flapping wing) will result in 0.24 mm movement of individ- is not at all stable on the wing and it continuously sheds
ual seed particles. This is well within the 1.6 mm distance from the leading-edge as the translational motion of the
between the adjacent nodes. The cross correlation and all wing continues, a phenomenon similar to the known clas-
other calculations were performed using commercial PIV sic features of dynamic stall. This observation directly
software. contradicts the stable LEV concept proposed by several
previous researchers, who claim a balance between the
RESULTS generation of vorticity at the leading-edge and the trans-
port of the vorticity into the wake. This difference may
The qualitative and quantitative measurements that were be attributed, in part, to the higher Reynolds number at
performed to understand the flow field surrounding a flap- which the flapping wing mechanism is operated in the cur-
ping wing MAV are explained in the following categories: rent study (19,500) when compared with the flapping wing
(1) flow visualization studies and (2) PIV measurements. mechanisms used by both Ellington, Dickinson and others
(150-1,400). Wisdom from helicopter blade studies shows
that an aft movement of the LEV will result in pitching mo-
LEADING-EDGE VORTEX Chordwise flow visualiza- ment stall that will be followed by a significant reduction in
tion images, which were acquired at fixed span location lift when the LEV sheds from the trailing-edge. However,
(50% of the half-span) through half-stroke of the flapping the physics is clearly different in the case of flapping wing
wing, are shown in Figs. 6 through 9. Because the laser because as the LEV moves aft over the chord of the wing
sheet was projected from the right-hand side of the wing in a new spilled vortex simultaneously forms at the leading-
this case, a shadow behind the wing appears in all the im- edge. This continuously sustains lift, a novel characteris-
ages. The time period, ζ, in all the images is represented tic that has been shown by flapping wings at low Reynolds
in such a way that the entire flapping stroke (pronation, number.
supination, downstroke and upstroke) is of 360◦ , with 0◦
representing the mid-point of pronation and 180◦ repre- The presence of multiple vortices on top of the wing at a
senting the mid-point of supination. given time depends not only on the Reynolds number but
also on the position of the wing stroke and the spanwise
It can be observed from Fig. 6 that a LEV does not form location on the wing. In the present study, because the
until the completion of pronation, as also suggested by region of initial interest was at the mid-span of the wing,
Van den Berg [18]. The shed vortices that result from the presence of multiple vortices was identified beyond
the wing rotation are clearly visible at the trailing-edge of the mid-point of translation during the stroke cycle. A sim-
the wing, as shown in Fig. 6(c). As the wing starts to ilar observation has been reported earlier by Tarascio et
accelerate after pronation, a small starting vortex forms al. [25] at the 25% span location for a thin rectangular wing
near the trailing-edge of the wing, as shown in Fig. 6(e). operated at the same Reynolds number as in the present
As the wing continues its translational motion, the size of experiment. The presence of multiple axial velocity peaks,
the starting vortex increases and appears to stay closer to as observed by Birch et al. [27] for the Robofly operat-
the trailing-edge – see Figs. 6(e) through 7(d). This phe- ing at a Reynolds number of 1,400 at 65% span might
nomenon has significant effect on the downwash velocity be because of the presence of multiple vortices. Ongoing
and, hence, on the lift produced by the flapping wing. PIV analysis of the flow at various stroke locations and
spanwise wing sections is now giving a better overall pic-
Figures 6(f) through 7(e) show that a small LEV forms at ture about the aerodynamic stability of LEVs on flapping
around 60◦ and it continues to grow with the translational wings.
motion of the wing. The LEV is then shed and moves aft
over the chord, as shown in Fig. 8. This behavior is a Chordwise flow visualization images that were acquired at
well-known property of a classic dynamic stall vortex [26]. different span locations at mid-point of translational stroke
However, in this case this shedding is immediately fol- (upstroke) are shown in Fig. 10. It can be observed that
lowed by the formation of another new vortex, thereby the LEV is totally absent at the root of the wing. The
producing multiple vortices at a given time, as shown in separated flow region is clearly identified by the eddies
Figs. 8(b) and 8(c). It can be observed from images that in Fig. 10(a). However, at the 25% span location, the sep-
correspond to 117◦ through 137◦ that for some times there arated flow reattaches at approximately 30% chord result-
are clearly two vortices over the top surface of the wing. ing in a LEV. This point of reattachment moves aft when
As the wing continues its cycle, the first LEV sheds off moving towards the wing tip – see Figs. 10(b) through (e).
the trailing-edge creating a very unsteady flow field be- Very close to the tip, it can be observed from Fig. 10(f)
hind the wing. Another clean single LEV is produced over that the flow separates again. This is because of the in-
the top of the wing, which is shown in Fig. 8(d). As the creased pitch angle that results from elastic wing bend-
wing starts to supinate, the second LEV starts to move aft ing, which has its source in the inertial forces produced

5
(a) (b) (c)

(d) (e) (f)

Figure 6: Flow visualization images obtained during pronation: (a) Schematic of the flow structure during pronation; (b) Mid
point of pronation, ζ=0◦ ; (c) During pronation, ζ=13◦ ; (d) End of pronation, ζ=25◦ ; (e) Accelerating wing, ζ = 40◦ ; (f) During
translational motion, ζ=60◦ .

Figure 7: Flow visualization images obtained at during downstroke: (a) ζ=65◦ ; (b) ζ=75◦ ; (c) ζ=80◦ ; (d) Mid-point of down-
stroke, ζ=90◦ ; (e) ζ = 97◦ ; (f) Schematic of the flow structure around the flapping wing at ζ = 90◦ .
on the wing. A similar form of flow separation near the Fig. 11. The images are explained in terms of wake age,
tip for a rectangular wing has already been reported by which is defined as the time elapsed since the wing refer-
Tarascio et al. [25]. ence axis was aligned with the laser light sheet. For ex-
ample, a wake age of 0◦ would mean that the wing shaft
axis is in line with the laser light sheet and a wake age of
WING ROOT AND TIP VORTICES Spanwise visualiza- 90◦ means that the wing shaft axis has moved one quar-
tion behind the trailing-edge of the wing has shown the ter of the flapping wing cycle from the laser light sheet. It
existence of relatively strong root and tip vortices – see can be observed from Fig. 12 that there are two sets of

6
(a) (b)

(c) (d)

Figure 8: Flow visualization images obtained during downstroke: (a) Aftward movement of LEV, ζ = 107◦ ; (b) Formation of
second LEV, ζ=117◦ ; (c) Presence of two vortices, ζ=137◦ ; (d) Second LEV after first one sheds, ζ=153◦ .

Figure 9: Flow visualization images obtained during supination: (a) Start of supination, ζ=157◦ , (b) During supination,
ζ=173◦ , (c) Mid-point of supination, ζ=180◦ ; (d) During supination, ζ = 188◦ ; (e) End of supination, ζ=196◦ ; (f) Schematic of
the flow structure around the flapping wing during supination.
root and tip vortices. The first set, which is closer to the Centrifugal forces, which have their source from the
wing, corresponds to those that are at 90◦ of wake age higher swirl velocities in the flow produced by the root and
and the second set corresponds to the vortices that were tip vortices, push the seed particles away from the vortex
formed during the wing motion on the previous half-stroke. centers. This results in a very low density of seed parti-
Because the laser sheet was placed precisely at the mid- cles near the vortex core axis to reflect the laser light, and
cycle of the wing translation, the second set of vortices so appears as a seed void at the approximate center of
are 270◦ older. vorticity in all the images. For tracking and targeting, the
geometric center of the seed void can be assumed to be

7
Figure 10: Chordwise flow visualization images along the span at mid-point of translation: (a) At the root of the wing; (b)
20% span location; (c) 40% span location; (d) 50% span location; (e) 75% span location; (f) 85% span location.

LEV

Root vortex

Laser light sheet

Wing

Root vortex
Tip vortex

Figure 11: Root and tip vortices behind the flapping wing
at 0◦ wake age.
Figure 13: Spatial locations of the root and tip vortices
trailing the flapping wing in the mid-stroke plane at various
wake ages.

the center of the vortex. The location of the root and tip
Tip vortex vortices relative to the flapping wing at various wake ages
ζ = 90 deg were determined by using the location of the center of the
seed void relative to a calibration grid. Because there is
always an inherent amount of aperiodicity associated with
this type of unsteady flow, both the average and the stan-
Root vortex dard deviation of these trailed vortex locations is shown in
ζ = 90 deg Fig. 13. These vortices are of sufficient proximity to each
other that they produce mutually induced effects. Notice
Root vortex
Tip vortex ζ = 270 deg that the root and tip vortices move radially towards each
ζ = 270 deg
other and axially downward with increasing time, suggest-
ing a contracting wake structure. The location of the root
Figure 12: Root and tip vortices behind the flapping wing
vortices could be observed until 300◦ , and for the tip vor-
at 90◦ wake age.
tices only until 150◦ of age. This is because of the spin-

8
(b)
(a)

(c) (d)

Figure 14: PIV analysis for 0◦ wake age (both laser light sheet and the wing are at the mid-point of downstroke: (a)
Representative image used for PIV analysis; (b) Ensemble averaged velocity plot; (c) Vector plot before ensemble averaging
showing all unsteady wake interactions; (d) Close-in view of root and tip vortices.
down of the vortex cores through viscous and turbulent field. Eighty such pairs were used to obtain an ensem-
effects, which results in less distinct seed voids at older ble average of the velocity field. The results for the first
wake ages. case are shown in Fig. 14(b). It can be seen that the av-
erage velocity field, as shown in Fig. 14(b), eliminates all
the aperiodic and turbulent wake fluctuations. The veloc-
QUANTITATIVE ANALYSIS The results from the PIV ity field obtained before performing ensemble averaging is
measurements that were made in a plane perpendicular shown in Fig. 14(c). A closer view of root and tip vortices
to the stroke plane of the flapping wing are explained in for the ensemble phase-averaged vector plot is shown in
this section. The measurements shown here were made Fig. 14(d).
at two wake ages: (1) when both the wing shaft axis and
laser light sheet are at 90◦ (mid-point of the translation The significant spanwise velocity can be seen in Fig. 14,
stroke), and (2) when the wing is at 160◦ and the laser which Ellington has previously hypothesized to be the
light sheet at 90◦ , i.e., essentially at 70◦ wake age. The source of the stability of the LEV. The root to tip flow mea-
laser light sheet in both the cases was centered along an sured here unambigously proves the presence of a span-
axis perpendicular to the stroke plane. wise flow. The spanwise flow in the LEV was found to be
significant, and of approximately the same magnitude as
A representative image from which the flow velocity vec- the maximum wing tip velocity during the flapping cycle.
tors across the flow field were obtained is shown in The vortex-like velocity vectors at approximately mid-span
Fig. 14(a). Two such images were taken with 30 µs phase of the wing suggests that the LEV is considerably three-
delay and were cross-correlated to obtain the velocity dimensional and is slightly inclined to the stroke plane.

9
(a) (b)

(c) (d)

Figure 15: PIV analysis for 70◦ wake age (laser light sheet at the mid-point of downstroke and the wing is at 70◦ away from
the mid-point of downstroke: (a) Representative image used for PIV analysis; (b) Ensemble averaged velocity plot; (c) Vector
plot before ensemble averaging showing all unsteady wake interactions; (d) Closer view of root and tip vortices.
The signatures of the root and tip vortices that were The ensemble average obtained for 70◦ wake age is
formed during the previous half stroke can also be seen shown in Fig. 15(b). It can now be observed that the root
in Fig. 14. The approximate peak swirl velocity of the root and tip vortices are 36% and 42% of the maximum tip
and tip vortices at 180◦ wake age at mid-point of down- velocity, respectively, which is large enough to have sig-
stroke was found to be 28% and 36% of the maximum tip nificant effect on the angle of attack of the flapping wing,
velocity of the flapping wing at its mid-stroke, respectively. a mechanism Dickinson hypothesized for the stability of
This is a relatively substantial velocity and will have sig- LEVs. Despite having such a large induced velocity that
nificant effect on the induced velocity distribution over the is comparable to the magnitude of the tip velocity of the
wing. The term “approximate” peak swirl velocity is used flapping wing, the LEVs were found to shed. As explained
here because of various issues that are not addressed earlier, one notable difference between the current exper-
here in accurately determining the core size and peak iment and other experiments is the higher Reynolds num-
swirl velocity of the root and/or tip vortices, such as the ber at which the present experiment was conducted.
inherent aperiodicity of the flow. Correcting the spatial lo-
cations of the vortices for effects of aperiodicity in the flow The velocity vectors measured along the chord to esti-
would generally result in a reduced vortex core size and mate the magnitude of leading-edge vortex is shown in
an increase in peak swirl velocity [28]. Fig. 16. The image on the left shows the original im-
age used to find the velocity vectors and the velocity vec-

10
(a) Raw PIV image (b) PIV vectors

Figure 16: PIV diagnostics of the dynamic stall vortex on top of the flapping wing at its mid-point during the translational
stroke. (a) Raw PIV image, (b) PIV velocity vectors.
Spanwise
flow
Separated wing. This would mean either that the spanwise flow is
flow
not large enough to prevent the shedding at this Reynolds
number or that the spanwise flow is not responsible for the
continuous attachment of the LEVs.

CONCLUSIONS
Root
vortex
x High-resolution flow visualization and PIV analysis have
Spilled lead-
ing edge been performed on a flapping wing MAV concept. The
vortex
following are the conclusions that have been derived from
the analysis of the measured results:
Separated
flow

1. A strong starting vortex forms during the early part


z
of translation that stays close to the wing for most
of the flapping stroke. The strength of the starting
Tip vortex
vortex increases during the flapping stroke and has a
substantial influence on the induced velocity field at
y the wing.

Figure 17: Schematic of the three-dimensional flow over 2. A LEV forms and gains strength during the transla-
the top of the flapping wing at its mid-point during the tional motion of the wing. This is followed by the
translational stroke. shedding of the LEV. A new LEV is formed before
the first LEV sheds from the trailing-edge. The pres-
tors are shown on the right. The image was obtained at ence of at least one LEV over the wing throughout
the mid point of upstroke. This suggests that the LEV is the translation stroke helps explain the sustained lift
present on either side of the translation (both downstroke characteristics shown by flapping wings.
and upstroke) and is of a magnitude that is comparable to 3. The presence of a spanwise flow that has simi-
the maximum tip velocity of the flapping wing. lar magnitude as the wing flapping velocity is not
large enough to prevent the shedding of LEVs at this
A schematic of the overall flow structure on top of the flap- Reynolds number. This LEV is three-dimensional in
ping wing, which is based on interpretations of both the nature and its axis of vorticity lies out of the stroke
flow visualization and PIV analysis, is shown in Fig. 17. plane.
Despite the presence of such a significant spanwise flow,
the LEVs were still in the process of shedding from the 4. A strong root and tip vortex pair has been observed

11
to trail behind the flapping wing. The peak swirl ve- [11] Ellington, C. P., “The Aerodynamics of Hovering In-
locities of these trailed vortices were measured to be sect Flight. IV. Aerodynamic Mechanisms,” Philosoph-
comparable in magnitude to the the wing tip velocity. ical Transactions of the Royal Society of London. Series
The root and tip vortices move radially towards each B, Biological Sciences, Vol. 305, No. 1122, Feb. 1984,
other and axially downward for increasing time, with a pp. 79–113.
contracting wake structure suggesting substantial lift
generation. [12] Zbikowski, R., Knowles, K., Pedersen, C. B., and
Galinski, C., “Some Aeromechanical Aspects of
5. The entire flapping wing wake structure, and the in- Insect-Like Flapping Wings in Hover,” Proceedings of
terconnectivity between the LEVs and the tip and root the I Mech E Part G Journal of Aerospace Engineering,
vortices, is still not fully understood and still remains Vol. 218, No. 6, pp. 389–298, 2004.
the subject of ongoing research.
[13] Laitone, E. V., “Aerodynamic Lift at Reynolds Num-
ber Below 7×104 ,” AIAA Journal, Vol. 34, No. 9, Sept.
1996, pp. 1941–1942.
REFERENCES
[14] Laitone, E. V., “Wind Tunnel Tests of Wings at
[1] Willmott, A. P., Ellington, C. P., and Thomas, A. L. R., Reynolds Number Below 70,000,” Experiments in Flu-
“Flow Visualization and Unsteady Aerodynamics in ids, Vol. 23, 1997, pp. 405–409.
the Flight of the Hawkmoth, Manduca Sexta,” Philo-
sophical Transactions of the Royal Society of London. Se- [15] Steven Ho., Hany Nassef., Nick Pornsinsirirak., Yu-
ries B, Biological Sciences, Vol. 352, No. 1351, March Chong Tai., and Chih-Ming Ho., “Unsteady Aerody-
1997, pp. 303–316. namics and Flow Control For Flapping Wing Flyers,”
Progress in Aerospace Sciences, Vol. 39, pp. 635–681,
[2] Srygley, R. B., and Thomas, A. L. R., “Uncon- 2003.
ventional Lift-Generating Mechanisms in Free-Flying
Butterflies,” Nature, Vol. 420, Dec. 2002, pp. 660– [16] Zimmerman, C. H., “Aerodynamic Characteristics of
662. Several Airfoils of Low Aspect Ratio,” NACA TN. 539,
Aug. 1935.
[3] Buckholz, R. H., “Measurements of Unsteady Peri-
odic Forces Generated by the Blowfly Flying in Wind [17] Pelletier, A., and Mueller, T. J., “Low Reynolds Num-
Tunnel,” The Journal of Experimental Biology, Vol. 90, ber Aerodynamics of Low Aspect Ratio Wings,” AIAA
1981, pp. 163–173. Paper 99-3182, 1999.
[4] Zanker, J. M., and Gotz, K. G., “The Wing Beat of
[18] Van den Berg, C., and Ellington, C. P., “The Vortex
Drosophila Melanogaster II. Dynamics,” Philosophi-
Wake of a ”Hovering” Model Hawkmoth,” Philosoph-
cal Transactions of Royal Society of London. B, Vol. 327,
ical Transactions of the Royal Society of London. Series
1990, pp. 19–44.
B, Biological Sciences, Vol. 352, No. 1351, Mar. 1997,
[5] Maxworthy, T., “Experiments on the Weis-Fogh pp. 317–328.
Mechanism of Lift Generation by Insects in Hover-
ing Flight. Part I. Dynamics of the “Fling”,” Journal of [19] Van den Berg, C., and Ellington, C. P., “The Three-
Fluid Mechanics, Vol. 93, 1979, pp. 47–63. Dimensional Leading Edge Vortex of a ”Hovering”
Model Hawkmoth,” Philosophical Transactions of the
[6] Bennett, L., “Clap and Fling Aerodynamics – An Ex- Royal Society of London. Series B, Biological Sciences,
perimental Evaluation,” The Journal of Experimental Bi- Vol. 352, No. 1351, Mar. 1997, pp. 329–340.
ology, Vol. 69, 1977, pp. 261–272.
[20] Birch, J. M., and Dickinson, M. H., “Spanwise Flow
[7] Dickinson, M. H., Lehmann, F. O., and Sane, S. P., and the Attachment of Leading-Edge Vortex on In-
“Wing Rotation and the Aerodynamic Basis of Insect sect Wings,” Nature, Vol. 412, No. 6848, 2001,
Flight,” Science, Vol. 284, No. 5422, 1999, pp. 1954– pp. 729–733.
1960.
[21] Ellington, C. P., and Usherwood, J. R., “Lift and
[8] Ellington, C. P., Van den Berg, C., Willmott, A. P., and
Drag Characteristics of Rotary and Flapping Wings,”
Thomas, A. L. R., “Leading Edge Vortices in Insect
Fixed and Flapping Wing Aerodynamics for Micro-Air Ve-
Flight,” Nature, Vol. 384, 1996, pp. 626–630.
hicle Applications, Edited by Thomas, A. Mueller,
[9] Wang, Z. J., “Two Dimensional Mechanism for In- Vol. 195 of AIAA Progress in Aeronautics and Astronau-
sect Hovering,” Physics Review Letters, Vol. 85, 2000, tics, Chap. 12, AIAA, Reston, VA, 2001, pp. 231–248.
pp. 2216–2219.
[22] Weish-Fogh, T., “Quick Estimates of Flight Fitness
[10] Liu, H., and Kawachi, K., “A Numerical Study of In- in Hovering Animals Including Novel Mechanisms for
sect Flight,” Journal of Computational Physics, Vol. 146, Lift Production,” The Journal of Experimental Biology,
1998, pp. 124–156. Vol. 59, 1973, pp. 169–230.

12
[23] Sane, S. P., “The Aerodynamics of Insect Flight,”
The Journal of Experimental Biology, Vol. 206, 2003,
pp. 4191–4208.

[24] Birch, J. M., and Dickinson, M. H., “The Influence of


Wing-Wake Interactions on the Production of Aero-
dynamic forces in Flapping Flight,” The Journal of Ex-
perimental Biology, Vol. 206, pp. 2257–2272, 2003.

[25] Tarascio, M., Ramasamy, M., Chopra, I., and Leish-


man, J. G., “Flow Visualization of MAV Scaled Insect
Based Flapping Wings in Hover,” Journal of Aircraft,
Vol. 42, No. 2, March 2005, pp. 355–360.

[26] Leishman, J. G., Principles of Helicopter Aerodynam-


ics, Cambridge University Press, New York, 2000,
Chapter–9.

[27] Birch, J. M., Dickson, W. B., and Dickinson, M. H.,


“Force Production and Flow Structure of the Lead-
ing Edge Vortex on Flapping Wings at High and Low
Reynolds Numbers,” The Journal of Experimental Biol-
ogy, Vol. 207, pp. 1063–1072, 2004.

[28] Leishman, J. G., “Measurements of the Aperiodic


Wake of a Hovering Rotor,” Experiments in Fluids,
Vol. 25, 1998, pp. 352-361.

13

Das könnte Ihnen auch gefallen