Eur. Phys. J. Special Topics 182, 97–112 (2010)
c EDP Sciences, Springer-Verlag 2010
DOI: 10.1140/epjst/e2010-01227-x
THE EUROPEAN
PHYSICAL JOURNAL
SPECIAL TOPICS
Regular Article
The art and science of flow control – case studies
using flow visualization methods
F.S. Alvi1,a and L.N. Cattafesta III2,b
1
2
Department of Mechanical Engineering, College of Engineering, Florida A & M University and Florida
State University, 2525 Pottsdamer Street, Tallahassee, FL 32310, USA
Interdisciplinary Microsystems Group, MAE Department, University of Florida, 231 MAE-A,
PO Box 116250, Gainesville, FL 32611-6250, USA
Received 17 March 2010 / Received in final form 20 May 2010
Published online 6 July 2010
Abstract. Active flow control (AFC) has been the focus of significant research in
the last decade. This is mainly due to the potentially substantial benefits it affords.
AFC applications range from the subsonic to the supersonic (and beyond) regime
for both internal and external flows. These applications are wide and varied, such
as controlling flow transition and separation over various external components
of the aircraft to active management of separation and flow distortion in engine
components and over turbine and compressor blades. High-speed AFC applications include control of flow oscillations in cavity flows, supersonic jet screech,
impinging jets, and jet-noise control. In this paper we review some of our recent
applications of AFC through a number of case studies that illustrate the typical
benefits as well as limitations of present AFC methods. The case studies include
subsonic and supersonic canonical flowfields such as separation control over airfoils, control of supersonic cavity flows and impinging jets. In addition, properties
of zero-net mass-flux (ZNMF) actuators are also discussed as they represent one
of the most widely studied actuators used for AFC. In keeping with the theme
of this special issue, the flowfield properties and their response to actuation are
examined through the use of various qualitative and quantitative flow visualization methods, such as smoke, shadowgraph, schlieren, planar-laser scattering, and
Particle image velocimetry (PIV). The results presented here clearly illustrate the
merits of using flow visualization to gain significant insight into the flow and its
response to AFC.
1 Introduction
Active flow control has witnessed explosive growth in recent years due to its potential to
revolutionize thermal-fluid systems. This multidisciplinary field has obviously benefited from
the development of numerous sophisticated mathematical, computational, and experimental
tools. But the development and implementation of such tools often precludes their routine use
in the early stages of a research project. However, flow visualization methods are quite valuable
a
b
e-mail: alvi@eng.fsu.edu
e-mail: cattafes@ufl.edu
98
The European Physical Journal Special Topics
Fig. 1. Time-averaged (top) and stroboscopic (bottom) smoke-flow visualization of a synthetic jet
operating at a frequency of 234 Hz [12].
in this regard. The objective of this paper is to demonstrate through examples how flow visualization is a very valuable tool to: (1) identify phenomenon that lead to problems and need to
be controlled; (2) provide guidance on control approaches; (3) provide information about the
effectiveness of control, once implemented; and (4) examine flows over a large range of length
scales from O(m) down to O(sub-mm) without considerable expense.
We will provide representative results from a number of flow control applications where
our research was guided by and greatly benefited from visual results. These include subsonic
flows, such as zero-net mass-flux (ZNMF) actuators and their use in separation control and
also supersonic flows, such as impinging jets and cavity flows and their control using microjet
actuators. Qualitative flow visualization results are supported by quantitative measurements
that confirm visual evidence and reaffirm the value of the increasingly-neglected, at a minimum
significantly under-utilized, tool of flow visualization.
2 Subsonic flow examples
2.1 Case 1: Flow from a ZNMF in a quiescent medium
Although most studies of ‘synthetic’ or ‘zero-net mass-flux’ jets in the literature date back
only to the past decade, the successful demonstration of the synthetic jet may be traced back
at least as far as [12], in which they identified and characterized acoustic streaming around
circular orifices [17]. Their setup used a plane-wave tube in which sound waves were generated
at one end using a speaker. The other end contained an adjustable plunger which allowed the
orifice plate to be situated at the quarter-wavelength of the sound wave and thus eliminate the
impedance of the cavity [5]. A total of 25 orifice plates of varying height and diameter were tested
with varying sound pressure levels at a frequency range of 100 Hz–1 kHz. The flow around the
orifice was visualized with smoke particles. Both constant and stroboscopic illumination were
used, allowing for time-averaged and instantaneous photographs of the flowfield to be acquired.
These photographs showed that four distinct “regions” of flow established themselves around
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
99
Fig. 2. Qualitative effect of Reynolds number and orifice height-to-diameter aspect ratio on synthetic
jet formation; white vertical line represents the approximate location of the orifice and the white
horizontal bars represent the approximate diameter of the orifice. The Reynolds number increases from
bottom to top, and orifice height-to-diameter ratio increases from left to right. Adapted from [6].
the orifice as a function of the frequency of the sound wave, the sound pressure level, and
the geometry of the orifice. An example of one such region is shown in Fig. 1, illustrating the
synthesis of a time-averaged jet that actually consists of a train of vortex rings.
Many years later, [6] used flow visualization to investigate the qualitative effect of the cavity
and orifice geometry on synthetic jet formation. A shaker-driven modular setup was constructed
which allowed for interchangeable cavity depths and circular orifice plates with varying heightto-diameter aspect ratio. The fluid in the cavity was seeded with smoke particles and visualized
through the use of a 50 Hz pulsating light sheet and video recorder, which allowed for aliased
“movies” of the vortex ring formation to be generated when the oscillator was run at a frequency
of 50.1 Hz. Fig. 2 shows a two-dimensional matrix of images as a function of Reynolds number
and orifice height-to-diameter aspect ratio. The videos suggested that after a maximum value
of circulation for a vortex ring was reached, the excess vorticity emerged as a tail behind the
ring, which rolled up to form a secondary ring if the Reynolds number was high enough. They
also discovered that increasing the orifice height-to-diameter ratio while holding the Reynolds
number constant tended to increase the amount of circulation in the ring, presumably because
a more developed velocity profile arose in the orifice.
Both the power and relative simplicity of flow visualization is clear from the above
examples, and flow visualization lays the foundation for subsequent quantitative measurements
(see the discussion in [7]). But the above results pertain to the case of circular orifices and
thus lend themselves to simpler setups that exploit (where possible) axisymmetry. However,
many practical ZNMF applications employ finite aspect-ratio slots, in which case the threedimensional nature of the resulting flowfield is of primary interest. For example, [24] performed
detailed measurements in the x-y plane, as shown in Fig. 3a, of quiescent synthetic jets issuing
from a 0.5 mm slot extending 75 mm in the spanwise (z) direction using single and cross-wire
100
The European Physical Journal Special Topics
(a) Coordinate system of a ZNMF
slot, where h, d, and w are the
dimensions of the slot in the x, y,
and z directions, respectively.
(b) Schlieren image of a rectangular synthetic jet. Provided by Barton
Smith. The artistic similarity of the image to “The Scream” by
Norwegian artist Edvard Munch is noted.
Fig. 3. Coordinate system of the rectangular ZNMF jet and corresponding schlieren image.
(a) View of x-y plane (23 mm by 35 mm).
(b) View of x-z plane (43 mm by 65 mm).
Fig. 4. Flow visualizations of a ZNMF jet with ReU0 = 84, StU0 = 0.86, h/d = 1.46, w/d = 14.88 [11].
hot-wire anemometry. But they first performed both phase-locked schlieren (in the x-y plane)
and smoke-flow visualization (in the x-z plane) to assess the two-dimensional nature of the
flowfield. For the schlieren (see Fig. 3b), the air inside the cavity is slightly heated using a
thin-film surface heater. Their smoke-flow visualizations (not shown – see Fig. 4 of [24]) clearly
show the formation of secondary vortices in the spanwise direction, features which are easily
missed by assuming two-dimensional flow. One may argue that flow visualization is essential
prior to performing more extensive and time-consuming flow measurements.
As an example, [11] sought to unify numerous prior investigations by recasting previous
results in terms of a single set of dimensionless parameters (namely Reynolds number, Strouhal
number, h/d, w/d). By doing so, he was able to identify a relevant range of dimensionless parameters for potential investigation. Nonetheless, the parameter space was still intractable, so he
used planar smoke-flow visualization to screen potential cases. The goal was to find cases where
the time-averaged flowfield near the slot was sufficiently two-dimensional to warrant subsequent
detailed study using PIV.
Flow visualization images were acquired using a digital camera with various lenses to provide
highly resolved images of the flow near the slot. The camera shutter speed was adjusted over
a wide range to obtain clear images of the particle pathlines in the thin (<0.5 mm thick) sheet
of the continuous light sheet. An example is shown in Fig. 4, where the motion of the smoke
particles is clearly revealed. In this case, the growth of the jet in the x-y plane is accompanied
by a narrowing of the jet in the x-z plane (see Fig. 3a). These and other flow visualization
results of [11] reveal the strong three-dimensional character and potential major axis switching
of finite-aspect ratio (w/d) ZNMF quiescent jets. Indeed, care is required when interpreting
PIV measurements in the x-y plane of the slot beyond a few slot dimensions, since the flow is
far from two-dimensional as illustrated in [11] and [4].
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
101
text
splitter plate
(2.4 mm thick)
Dynamic Pressure
Transducer Locations
Piezoelectric
Diaphragms
Cavity
y
NACA 0025
Airfoil
Synthetic Jet
x Actuator Pairs
text
z
text
Preamp
oard
Circuit B
text
Actuators
One Two
piezoceramic
disks
slots
(0.5 mm wide)
6"
text
(a)
side view
top view
(b)
Fig. 5. Platform for separation control experiments. (a) cross section of the NACA 0025 airfoil with
the synthetic jet actuator arrays shown in (b) [10].
2.2 Case 2: Separation control using ZNMF actuators
Beyond experiments designed to characterize ZNMF actuators, these devices have been used
extensively for various AFC applications [7]. Here, we describe a leading edge separation control
experiment (Fig. 5). A two-dimensional, six inch (152.4 mm) chord, NACA 0025 airfoil serves
as the test model in a low speed wind tunnel. A cross sectional drawing of the airfoil is shown
in Fig. 5a. The wind tunnel test section measures 30.48 cm by 30.48 cm, and the span of the
model is 29.21 cm. The nominal Reynolds number based on chord length is Rec = 105 , and the
angle-of-attack was set at 12◦ , which causes massive leading-edge separation.
The airfoil is fitted with two pairs of synthetic jet arrays in the central spanwise region of
the airfoil as shown in Fig. 5b. The first array is located near the leading edge of the airfoil, at
approximately 3% chord, while the second array is inactive and covered. All five piezoceramic
disks in each cavity are controlled via an arbitrary function generator. Two synchronized function generators are used to independently control each synthetic jet on the leading edge actuator
pair. The signals are phase-locked to the first synthetic jet and the relative phase between the
two excitation signals is continuously adjustable from 0◦ to 360◦ . The sinusoidal signals from
each function generator are then amplified. The piezoceramic diaphragms are driven with a
voltage of 50 Vpp . The frequency of excitation for this study is approximately 1500 Hz.
The objective here is to examine high frequency excitation effects. In particular, at high
excitation frequencies, the time scale of the synthetic jet formation process is quite small compared to the convective time scale of the incoming boundary layer. In other words, the actuation
frequency is more than an order-of-magnitude greater than the characteristic frequency, U∞ /c,
associated with the flow past the airfoil. For a six-inch chord airfoil in presence of a 10 m/s
freestream, the characteristic frequency is about 66 Hz. Of particular interest here is the effect
of actuator phasing between the first two actuator slots for a given amplitude, etc. As shown in
Fig. 6, the relative phase angle between the actuators is defined as ∆ϕ, while any given phase
angle on a phased-locked cycle is defined as θ.
Before expanding significant time on quantitative phase-locked PIV measurements, a simpler flow visualization method was used to explore the influence of ∆ϕ. Prior to conducting
the experiments, we expected that ∆ϕ could be adjusted to enhance the effect of an actuator
by timing the phase of maximum expulsion of the second actuator with the arrival of a vortex structure from the first actuator. Using conventional fog fluid seeding in the freestream,
flow visualization images are acquired near the actuator in the presence of the boundary layer.
102
The European Physical Journal Special Topics
expulsion
starts
max
expulsion
ingestion
starts
max
ingestion
0
o
0
o
+180
o
+90
o
-90
0
90
180
θ (degree)
270
360
Fig. 6. Relative phases between two successive actuator arrays. The clock face and the solid line
indicate the upstream actuator (actuator 1). The legend refers to the downstream actuator [10].
a)
b)
c)
d)
e)
f)
g)
h)
i)
j)
k)
l)
m)
n)
o)
p)
Fig. 7. Selected frames from aliased movies showing the effect of a cross flow in the vicinity of the
synthetic jets, region of 3%–10% chord shown, vertical lines show evolution of the vortex structure from
the second actuator, (a)–(d) ∆ϕ = 0◦ , (e)–(h) ∆ϕ = 90◦ , (i)–(l) ∆ϕ = 180◦ , (m)–(p) ∆ϕ = −90◦ [10].
In these experiments, a high-speed imaging system was not available, so a light sheet from a
Nd:YAG laser is used to illuminate the flow at precisely 15 Hz, and the actuator frequencies are
adjusted slightly to just above 1500 Hz to create an intentional aliasing effect. This results in a
qualitative phase-locked “movie” which reveals several important features of the flow, as shown
in a false color scale in Fig. 7.
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
103
Acoustic Waves
Large-scale
structures
Fig. 8. Instantaneous shadowgraph of a Mach 1.5, ideally expanded impinging jet. Strong acoustic
waves can be clearly seen in the ambient environment along with large, turbulent structures [16].
The interaction with the boundary layer produces vortex structures that remain attached
to the surface and are swept downstream within a thin region over the surface of the airfoil.
This suggests that high momentum fluid is being drawn in from the edge of the boundary
layer toward the surface. Horizontal rows of images in Fig. 7 show the evolution of the vortex
structure for a fixed relative phase between the first and second actuators at the four different
phases indicated in Fig. 6. For example, the first row corresponds to the sequence of images
during the four phases of the cycle when the two actuators are operated in phase. On the
other hand, vertical rows of images show the flow phase locked to the upstream actuator for
a range of relative phases of the downstream actuator. The locations of the vortex structures
are clearly dependent on ∆ϕ, but the overall flow structure seems relatively unaffected by ∆ϕ.
For the range of actuation conditions considered, the strongest vortex structures consistently
appear to originate from the second actuator. This is illustrated by comparing the diagonal
images in Fig. 7, which essentially give the flow structure phase locked to the beginning of the
expulsion stroke of the downstream actuator. Selective PIV and LDV measurements indeed
confirm that (1) ∆ϕ has no appreciable effect on the controlled velocity field and (2) the
second actuator is indeed more effective than the first, emphasizing the dominant importance
of actuator placement on control effectiveness.
3 Supersonic flow examples
3.1 Case 1: Supersonic impinging jet
Impinging jets in general and supersonic impinging jets in particular are often dominated by
large coherent structures that make their visualization appealing, aesthetically speaking, as
well as insightful from a scientific perspective. The supersonic impinging jet flowfield is also of
interest from both fundamental and applications perspectives. This flow contains complex shock
wave structures, viscous/inviscid interactions (especially in the impingement region) such as
shock-shock and shock wave-shear layer interactions; flow-acoustic coupling and coherent, large
scale structures among others. This makes it a challenging but very interesting flow to study
[1,14]. It is also well-known that the high-speed, in particular supersonic, impinging jet flowfield
is very unsteady. A number of studies have clearly established that the self-sustained, highly
unsteady behavior of the jet and the resulting aeroacoustic properties are due to a feedback
mechanism [1,14,19,20,25]. Very briefly, instability waves in the jet shear layer at the nozzle
exit evolve into large-scale structures as they propagate downstream towards the impingement
surface. The impingement of these structures on the ground plane generates strong acoustic
waves, which travel upstream and excite the highly receptive shear layer near the nozzle exit.
Figure 8 shows an instantaneous shadowgraph image of a Mach 1.5, ideally-expanded
impinging jet flow where the impingement plane is four diameters from the nozzle exit. The
104
The European Physical Journal Special Topics
Fountain Flow
Fig. 9. Instantaneous shadowgraph of dual Mach 1.5, impinging jets. Visible is the coupling between
the two jets, the fountain flow and the turbulent structures in the jet shear layer [16].
shadowgraph shown here was obtained using a white-light spark source in a conventional
“z-type” Toepler arrangement [23] where the knife-edge has been removed. The instantaneous
image clearly shows the presence of multiple, strong, acoustic waves in the ambient environment. Waves travelling in both directions are observed in this image; the upstream propagating
waves, seen as concave downwards, originate near the ground plane, they are reflected from a
plate flush mounted with the nozzle exit (seen as a black line on the top of the image) and
are visible as waves that are concave upwards. These waves are responsible for the feedback
mechanism discussed above. The fact that these ‘waves’ are so clearly visible suggests their considerable strength; they are also responsible for the overall flow unsteadiness and high-amplitude
impingement tones which are ubiquitous in such flows. Another notable visual feature in the
instantaneous shadowgraph is the presence of large-scale structures in the jet shear layer, the
most prominent of these has been marked in this figure. In Fig. 9, we show an instantaneous
shadowgraph for dual Mach 1.5 impinging jets. This flow is more complex and richer in flow
features than a single impinging jet. This is partly due to the presence of the fountain flow
between the two jets and also because of the additional coupling between the two jets. This
shadowgraph reveals a number of the salient features typical of these flows, as follows: similar
to the single jet, this flow is also highly unsteady. The main jet columns are highly ‘sinuous’ in
appearance due to the presence of large-scale structures in the jet shear layer, as in the single-jet
case, and the highly oscillatory fountain flow between the jets [16].
The shear layer turbulent structures, which constitute one leg of the feedback loop, can be
more clearly seen in representative instantaneous PIV images for the single and dual impinging
jets shown in Fig. 10. Such spatially coherent, large-scale structures are unusual in compressible shear layers (the shear layer at the jet periphery) and visually indicate an unusually rapid
amplification of instabilities in the shear layer, a characteristic of this flow. The images shown
in Fig. 10 were obtained by seeding the main jet with sub-micron droplets and seeding the ambient air with smoke particles produced by commercial fog generators. The flow is illuminated
using a thin light sheet generated from a double-pulsed Nd:YAG laser (see [2] and [14] for more
details). In addition to providing considerable insight into the global flowfield properties, such
double-pulsed images can be used to obtain the planar, two-component velocity field using PIV
processing algorithms [14]. Two examples of the instantaneous velocity fields obtained from
such PIV images are shown in Fig. 11.
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
105
Fig. 10. Instantaneous PIV images of Mach 1.5 single and dual impinging jets. (Single Jet: [14], Dual
Jet: [16]).
0
0
-10
-10
-20
-20
-30
-30
X (mm)
x (mm)
-40
-50
-60
-40
-50
-60
-70
-80
-70
-90
-80
-100
-50
-25
0
r (mm)
25
-90
-60
-45
-30
-15
0
15
30
45
60
r (mm)
Fig. 11. Instantaneous PIV images of Mach 1.5 single and dual impinging jets (Single Jet: [14], Dual
Jet: [16]).
Fig. 12. Instantaneous phase-conditioned schlieren images of a rectangular supersonic screeching air
jet, phase increases in images from left to right [15].
Before discussing quantitative results for the above flows, we show another example of a flow
that is dominated by flow-acoustic resonance and where the effect of this coupling is visually
striking. In Fig. 12 we see instantaneous, phase-conditioned, schlieren images obtained by [15] of
a rectangular supersonic screeching air jet, operating at NPR = 3.5. In the three different phases
of the screech cycle shown here, one clearly sees the cylindrical acoustic wave, originating near
the third shock cell, as it propagates further out in each image. These images further illustrate
the value of using phase-conditioned imaging in extracting periodic features from a flowfield,
features that provide valuable insight into the fundamental flow physics.
106
The European Physical Journal Special Topics
BW-nc-s1LPmic-h3-may31.lay
190
NPR= 3.7 h= 3.0
No Control
180
Ground Plane Kulite
Lift Plate Kulite
Microphone
170
Prms (dB)
160
150
140
130
120
110
100
5000
10000
15000
20000
25000
30000 35000
Frequency(Hz)
Fig. 13. Unsteady surface pressure and microphone spectra for Mach 1.5 impinging Jet [3].
The visual evidence seen in the shadowgraph and the instantaneous PLS/PIV images clearly
suggest that the flowfield is highly unsteady. Furthermore, the presence of coherent structures
and acoustic waves point to the presence of a resonance loop. The visual evidence is confirmed
by quantitative measurements, shown in Fig. 13, for a representative case of the single impinging jet. The narrow-band spectra in this figure (from [3]) shows microphone measurements
of the near-field noise as well as the unsteady pressures measured on the impingement plane
(referred to as the ground plane in this figure) and on the plate flush mounted with the nozzle
exit (referred to as the lift plate). The noise and the unsteady pressures are shown in terms
of dB (re 20 µPa). Among the most notable features in this spectra are the discrete, highamplitude, multiple peaks which are indicative of impingement tones due to the flow-acoustic
coupling. Such high amplitudes tones, coupled with overall high levels of the noise and unsteady
pressures, confirm the highly unsteady behavior noted in the visualizations. Furthermore, a comparison of the microphone and unsteady pressure measurements data reveals that the resonant
tones occur at nominally identical frequencies at all three transducer locations. This supports
the earlier observation based on the visual evidence that the flow is globally unsteady as expected in a feedback-governed flow.
Finally, if, as the visual evidence suggests and the quantitative measurements confirm, the
flow is indeed dominated by a resonance loop, then the most efficient way to attenuate this is to
disrupt it at or near the nozzle exit, a region where the shear layer is highly receptive. A control
approach whereby steady and pulsed microjets were injected near the nozzle exit was adopted
by [3]. This approach was rather successful in that it dramatically reduced the flow unsteadiness
and the accompanying impingement tones and noise. The effect of microjet control of the single
impinging jet can be visibly seen in the shadowgraph shown on the left in Fig. 14. The effect
is visually dramatic when compared to the baseline, un-controlled case, shown in Fig. 8. Note
that the strong acoustic waves in the near field seen in Fig. 8 have essentially been eliminated.
This is accompanied by a dramatic reduction in the size and frequency of large-scale structures
in the jet shear layer. Also visible in this shadowgraph are the streaks generated by the supersonic microjets. Such streaks have been used as a qualitative indicator of streamwise vorticity
where tabs and other passive devices have been used to control jet aeroacoustics [22] and [26].
Subsequent velocity-field measurements in the single impinging jet flow by [2] have confirmed
that microjets in fact generate significant streamwise vorticity which plays a significant role in
attenuating the flow-acoustic coupling and resonance in this flow. The visual effect of control on
the dual impinging jet is equally impressive as seen in the shadowgraph on the right in Fig. 14.
As before, a comparison to the corresponding uncontrolled image (Fig. 9) leaves no doubt that
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
107
Microjet
Streaks
Microjet
Streaks
Fig. 14. Shadowgraphs illustrating the visual effect of microjet-based control on the impinging jet
flowfield. Left: Single impinging jet [3]; Right: Dual impinging jet [16].
150
NPR=3.7 h/d= 4 Microphone
(b)
No control
100 psia
140
SPL (dB)
130
120
110
100
500 0
10 000
150 00
2 000 0
250 00
30 000 3 500 0
Frequency
Fig. 15. Effect of microjet control on the near-field noise for a single impinging jet. Solid lines: baseline
flow; dashed line: flow with microjet control [18].
the unsteady oscillations of the dual jets have been significantly reduced. As in the single impinging jet, also gone are the large shear layer structures which dominated baseline flow. The
reduction in oscillations of the primary jets and the attenuation of the large-scale structures
results in a more stable fountain jet. Finally, microjet-induced streaks are also clearly present
in the controlled dual impinging jet.
The visual impact of control is accompanied by a correspondingly significant reduction in the
flowfield unsteadiness. As an example, Fig. 15 shows the narrowband spectra for the near-field
noise for a representative case. Upon comparing the uncontrolled spectra (solid lines) to the controlled case (dashed lines), it is clear that the primary impingement tone has been substantially
attenuated – by more than 20 dB, and perhaps more significantly, that this tonal attenuation
is accompanied by a notable reduction in the broadband noise levels. Similar reductions were
also measured in the unsteady pressures on the impingement and lift plates. Microjet control
also resulted in comparable reductions for the dual impinging jet case [16] again providing evidence that appropriate use of even simple flow visualization provides information that can be
used to better understand the flow and provide queues to where and what type of quantitative
measurements should be used.
108
The European Physical Journal Special Topics
Fig. 16. Cavity flowfield visualizations, from low subsonic to transonic [13].
3.2 Case 2: Supersonic cavity flows
High-speed cavity flows are also governed by a feedback mechanism similar to that in impinging
jets [21]. The disturbances in the separating shear layer at the cavity leading edge evolve into
large-scale structures. Upon impingement on the cavity trailing edge, the structures generate
acoustic waves or disturbances which propagate upstream to the cavity leading edge – for
subsonic cavity flows this occurs both inside and outside the cavity; the waves only propagate
within the cavity when the freestream flow is supersonic. These waves further excite the shear
layer at the leading edge and, when there is a match in the frequency and the phase of the
acoustic waves and the shear layer instabilities, resonance is achieved. This produces very high
unsteady pressure loads on the surfaces within and in the vicinity of the cavity. One of the
earliest discussions of the feedback loop for cavity flows was provided by Krishnamurty (1953);
he was also perhaps the first to provide dramatic visual evidence of upstream propagating
acoustic waves in high-speed cavity flows. In Fig. 16 we see an example of his visualizations of
a cavity flow – for the three subsonic cases acoustic waves are seen to propagate upstream in
the outer freestream, whereas for supersonic flow, these waves are swept downstream by the
freestream flow.
More recently, [27] and [28] have examined supersonic cavity flows and their control using
visualizations and quantitative measurements. Figure 17 shows two instantaneous shadowgraph
images of a Mach 2, three-dimensional, cavity flow, where the cavity sidewalls are made of
Schlieren-quality glass to allow optical access to the internal flowfield. The flow direction is
from left to right and large turbulent structures are visible in the cavity shear layer in both
images; the edge of some of these structures has been outlined by a dashed line in the top
image. In addition, a number of other prominent features, essentially wave fronts, have also
been labeled from I to V as follows: Type I waves are generated near the cavity leading edge
by the flapping motion of the separating shear layer. A compression wave is generated when
the shear layer moves up and an expansion wave when the shear layer flaps down. In literature
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
109
Fig. 17. Instantaneous shadowgraphs of a Mach 2 cavity flow showing the presence of various structures
and wavefronts [27, 28].
Fig. 18. Instantaneous PIV image and velocity filed of a Mach 2 cavity flow [27, 28].
similar waves have been referred to as quasi-steady compression shocks [8]. Type II waves are
formed by the convecting, large-scale structures in the cavity shear layer. The impingement of
these structures on the trailing edge creates bow shocks marked as type III waves. The acoustic
waves – generated by the impact of these structures, propagate upstream within the cavity at
the local speed of sound. In doing so, they perturb the cavity shear layer and generate type
IV waves in the free stream. Since the convective speed of the cavity-induced perturbations is
higher relative to the shear layer turbulent structures, they are more swept when compared to
type II wavefronts. These four types of waves have been observed in supersonic cavity flows
by other investigators, e.g. [8] and [9]. However, here we also clearly see an additional feature,
marked as a type V wave inside the shadowgraphs. It is suggested that this is an example of
the upstream propagating acoustic waves/perturbations generated by the impact of the shear
layer on the cavity trailing edge.
The excursions of the large shear layer structures into the supersonic freestream flow can
be more clearly seen in the instantaneous PIV snapshot shown in Fig. 18 on the left. Also, as
discussed in the context of impinging jets, such double-pulsed images can be used to extract
velocity-field data using PIV algorithms. A representative instantaneous velocity field extracted
from such PIV images is shown on the right in Fig. 18. The presence of spatially coherent
structures is seen in the shear layer as well as inside the cavity.
110
The European Physical Journal Special Topics
160
SPL (dB)
150
Mode1
Mode2
Mode3
Mode4
A1
A3
A5
A6
140
130
120
Freq (khZ)
5
10
15 20
Fig. 19. Unsteady Pressure Spectra at various locations in a Mach 2 cavity [28].
Fig. 20. Instantaneous showing the effect of microjet control [27, 28].
Further quantitative evidence of the flow unsteadiness is seen in Fig. 19, which shows the
spectra measured inside this cavity at various locations using high-frequency response, flushmounted Kulite transducers. Similar to the impinging jet spectra (Fig. 13), the flow is dominated
by high-amplitude, discrete frequency ‘cavity tones’ which suggest the presence of a global instability in the form of the feedback or resonance loop. The various tones have been indicated
as “Modes” in the figure since such tones are also commonly referred to as ‘Rossiter modes’ in
the context of cavity flows. Also note that the overall unsteadiness increases as one approaches
the trailing edge of the cavity, from transducer A1 to A6, since the amplitude of the shear layer
flapping increases in this region. Given that the unsteady cavity and the impinging jet flowfields
are both governed by a similar feedback loop, an array of microjets was integrated at the cavity
leading edge with the aim of controlling the flow unsteadiness by efficiently disrupting feedback
loop.
The effect of this control can be seen in the instantaneous shadowgraph in Fig. 20. The
activation of the steady microjet array generates an oblique shock just upstream of the array. A study of a series of images similar to those in Fig. 20 reveals that microjet control
results in a significant reduction in the flow unsteadiness, in a manner similar to the control of impinging jets. Nearly all the waves seen in Fig. 17, the uncontrolled case, have either
been eliminated or dramatically weakened. Only type III waves are still visible in the shadowgraphs although at a reduced strength. Since the frequency and strength of these waves,
especially type V inside the cavity, are an indication of the flow unsteadiness, their dramatic weakening suggests that the feedback loop has been significantly disrupted. The unsteady pressure spectra shown in Fig. 21 provide convincing evidence that this is indeed the
case. The pressure spectra (black – no control; gray – with control), measured with transducer A1 located at the cavity leading edge, show the following. The dominant cavity tone
is attenuated by more than 23 dB while the secondary tones are reduced by 5 dB or more.
Revealing the Invisible – Classical Methods of Flow Visualization Revisited
111
Control Off
Control On
180
SPL dB
170
160
150
140
130
5
10
15
20
Frequency KHz
Fig. 21. Effect of microjet control on unsteady pressures inside the cavity [27, 28].
Moreover, the attenuation in the tones is accompanied by a significant reduction in the broadband noise levels as well, making this control technique rather effective. (See [27] and [28] for
more details.)
4 Concluding remarks
Using flow visualization methods as the primary diagnostic tool, a number of examples or
case studies of Active Flow Control have been presented. These case studies represent typical
subsonic and supersonic flowfields, where AFC has provided a notable effect. Subsonic flow
control is demonstrated through the control of separation over airfoils using Zero-Net MassFlux (ZNMF) actuators. Supersonic AFC is discussed in the context of supersonic cavity flows
and supersonic impinging jets using microjet arrays. A discussion of ZNMF actuators, among
the most ubiquitous of actuators, is also provided, starting with one of its earliest versions to
the more recent evolutions.
In both subsonic and supersonic regimes, the use of appropriate AFC has a significant,
favorable impact on the flowfield, an impact that is clearly visible in smoke visualizations,
shadowgraphs/schlieren images and planar laser scattering images. The visual evidence of this
favorable response to AFC, be it separation reduction, lift enhancement or noise attenuation, is
supported by more quantitative measurements as seen by the representative examples discussed
herein. The correspondence between the visual and more quantitative measurements clearly
illustrates the benefits of visual diagnostics. By using relatively simple flow visualization methods one can better understand the flow behavior, especially in terms of its global properties.
This in turn provides queues to where and what type of quantitative measurements should
be used and, in the context of AFC, where and how control should be applied. Using simple
methods, it is clear that “Seeing is Believing,” or perhaps “Seeing is Understanding,” still rings
true even in this era of complex diagnostics.
The authors gratefully acknowledge support from AFOSR and NASA for a significant portion of the
work presented here. Much of the subsonic flow and supersonic flow results presented here are based
on research by R. Holman (ZNMF actuators), H. Lou (impinging jets) and N. Zhuang (cavity flow).
References
1. F.S. Alvi, K. Iyer, AIAA Paper 99–1829 (1999)
2. F.S. Alvi, H. Lou, C. Shih, R. Kumar, J. Fluid Mech. 613, 55 (2008)
112
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
The European Physical Journal Special Topics
F.S. Alvi, C. Shih, R. Elavarasan, G. Garg, A. Krothapalli, AIAA J. 41, 1347 (2003)
M. Amitay, F. Cannelle, Phys. Fluids 18, 054101 (2006)
D.T. Blackstock, Fundamentals of Physical Acoustics (John Wiley & Sons, Inc., New York, 2000)
A. Crook, N.J. Wood, 9 th International Symposium on Flow Visualization (Edinburgh, Scotland,
2000)
A. Glezer, M. Amitay, Ann. Rev. Fluid Mech. 34, 503 (2002)
H.E. Heller, D.B. Bliss, Air Force Flight Dynamics Laboratory Technical Report, No. AFFDL-TR74-133 (1975)
H. Heller, J. Delfs, J. Sound Vib. 196, 248 (1996)
R. Holman, Q. Gallas, B. Carroll, L. Cattafesta, AIAA Paper 2003-3709 (2003)
R. Holman, An experimental investigation of flows from zero-net mass-flux actuators, Ph.D. thesis,
Mechanical and Aerospace Engineering (University of Florida, 2006)
U. Ingard, S. Labate, J. Acoust. Soc. Am. 22, 211 (1950)
K. Krishnamurty, Sound radiation from surface cutouts in high speed flow, Ph.D. thesis (California
Institute of Technology, 1956)
A. Krothapalli, E. Rajakuperan, F.S. Alvi, L. Lourenco, J. Fluid Mech. 392, 155 (1999)
A. Krothapalli, D. Baganoff, Y. Hsia, AIAA Paper No. 83-0727 (1983)
A. Krothapalli, F.S. Alvi, AIAA Paper 2002-5977 (2002)
S.J. Lighthill, J. Sound Vib. 61, 91 (1978)
H. Lou, F.S. Alvi, C. Shih, AIAA J. 44, 58 (2006)
G. Neuwerth, NASA TT F-15719, 1974
A. Powell, J. Acoust. Soc. Am. 83, 515 (1988)
J. Rossiter, Aeronautical Research Council, Rept. 3438 (1964)
M. Samimy, K.B.M.Q. Zaman, M.F. Reeder, AIAA J. 31, 609 (1993)
G.S. Settles, Schlieren and Shadowgraph Techniques: Visualizing Phenomena in Transparent Media
(Springer-Verlag, Berlin, 2001)
B.L. Smith, A. Glezer, Phys. Fluids 10, 2281 (1998)
C.K.W. Tam, K.K. Ahuja, J. Fluid Mech. 214, 67 (1990)
K.B.M.Q. Zaman, M.F. Reeder, M. Samimy, Phys. Fluids 6, 778 (1994)
N. Zhuang, F.S. Alvi, C. Shih, AIAA Paper 2005 (2005)
N. Zhuang, F.S. Alvi, M Alkilsar, C. Shih, AIAA J. 44, 2118 (2006)