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)