You are on page 1of 14

Exp Fluids DOI 10.

1007/s00348-011-1110-6

RESEARCH ARTICLE

The formation of vortex rings in a strongly forced round jet


E. Aydemir N. A. Worth J. R. Dawson

Received: 12 October 2010 / Revised: 8 March 2011 / Accepted: 25 April 2011 Springer-Verlag 2011

Abstract The periodic formation of vortex rings in the developing region of a round jet subjected to high-amplitude acoustic forcing is investigated with High-Speed Particle Image Velocimetry. Harmonic velocity oscillations ranging from 20 to 120% of the mean exit velocity of the jet was achieved at several forcing frequencies determined by the acoustic response of the system. The time-resolved history of the formation process and circulation of the vortex rings are evaluated as a function of the forcing conditions. Overall, high-amplitude forcing causes the shear layers of the jet to breakup into a train of large-scale vortex rings, which share many of the features of starting jets. Features of the jet breakup such as the roll-up location and vortex size were found to be both amplitude and frequency dependent. A limiting time-scale of t/T & 0.33 based on the normalized forcing period was found to restrict the growth of a vortex ring in terms of its circulation for any given arrangement of jet forcing conditions. In sinusoidally forced jets, this time-scale corresponds to a kinematic constraint where the translational velocity of the vortex ring exceeds the shear layer velocity that imposes pinch-off. This kinematic constraint results from the change in sign in the jet acceleration between t = 0 and t = 0.33T. However, some vortex rings were observed to pinch-off before t = 0.33T suggesting that they had acquired their maximum circulation. By invoking the slug model approximations and dening the slug parameters based on the experimentally obtained time- and lengthscales, an analytical model based on the slug and ring energies revealed that the formation number for a
E. Aydemir N. A. Worth J. R. Dawson (&) Department of Engineering, University of Cambridge, Trumpington Street, Cambridge CB2 1PZ, UK e-mail: jrd37@cam.ac.uk

sinusoidally forced jet is L/D & 4 in agreement with the results of Gharib et al. (J Fluid Mech 360:121140, 1998).

1 Introduction Although free round jets are self-similar in the far eld, it is well known that the initial development region exhibits large-scale orderly structures. The classical experiment of Crow and Champagne (1971) observed the formation of large-scale vortical puffs in the near eld of an excited round jet. The phenomena of vortex pairing in an excited jet was investigated by Zaman and Hussain (1980). These and numerous other experiments over the years have shown that small levels of forcing applied to the base ow affects how the ow eld develops by amplifying instability growth rates, enhancing the transition to turbulence, affecting sound generation, and mixing characteristics (Becker and Massaro 1968; Gutmark and Ho 1983; Lee and Reynolds 1985; Raman et al. 1989). This paper investigates the effect high-amplitude forcing has on the structure of the near eld of a jet, particularly the onset of large-scale unsteadiness characterized by the periodic formation of vortex rings. The term high amplitude in this paper refers to uctuations in the exit velocity being orders of magnitude greater than the turbulent uctuations u0 x; t: Vortex rings formed by starting and unsteady jet ows play a key role in a variety of emerging technologies. For example, unsteady propulsion devices such as pulse detonation engines and pulse combustors generate periodic vortex rings whose formation time-scales correlate with improved thrust augmentation (Opalski et al. 2005; Krueger and Gharib 2005; Mason and Miller 2006). They are also an integral feature of unsteady aerodynamics found in bio-inspired propulsion such as underwater autonomous

123

Exp Fluids

vehicles (UAVs) (Krieg and Mohesni 2008) or where the controlled apping of wings causes the ow to separate and roll-up into vortices (Rival et al. 2009). In combustion instability, vortices are a source of ame sheet kinematics responsible for nonlinearity that leads to limit-cycle oscillations and is a signicant development issue in lean burn gas turbine combustion (Kulsheimer and Buchner 2002; Balachandran et al. 2005). Fundamental investigations into vortex rings formed by starting jets discharging into a quiescent uid have elucidated many aspects of the formation process together with models to predict a variety of ring parameters of interest such as trajectory, circulation, and vorticity distribution in the core (Pullin 1979; Didden 1979). Using a piston-cylinder apparatus to generate vortex rings from starting jets, Gharib et al. (1998) showed that for increasing aspect ratio of the uid slug (L/D) the circulation of the vortex ring increased until L/D & 4. Further increases in L/D did not increase the ring size and circulation with the excess vorticity ux forming a trailing jet in the wake of the vortex. This transitional state between a single ring without a trailing jet and a ring with a trailing jet is commonly referred to as the formation number, which is L/D & 4. Linden and Turner (2001) analytically demonstrated that the limiting process described by the formation number corresponds to an optimal vortex ring as it possesses the maximum impulse, circulation, and volume for a given energy input. The presence of counter and coow on the formation number were also investigated by Dabiri and Gharib (2004) and Krueger et al. (2006), respectively. Johari (2006) investigated the vortex formation in a fully pulsed jet in crossow. However, no studies have investigated whether or not such a limiting process exists in forced round jets with a mean ow. The focus of this paper is to study the formation of vortex rings that occur when a round jet is subjected to high-amplitude acoustic forcing. Emphasis is placed upon the formation time-scales and the circulation of the vortex rings as a function of forcing conditions. A particular aim is to identify whether or not a formation number exists for forced jets and if so what are the dening nondimensional parameters. For this purpose, experiments have been performed using 2D High-speed particle image velocimetry (HSPIV) for the quantitative examination of the near eld of an acoustically forced round jet.

2 Experimental methods A diagram of the apparatus and experimental setup is shown in Fig. 1. The apparatus was based on the setup of

Balachandran et al. (2005) to study forced ames and was designed to provide large-amplitude, sinusoidal velocity oscillations at the nozzle exit by exciting various acoustic modes of the system. This experimental arrangement could generate velocity amplitudes C100% of the mean exit ow for a discrete set of frequencies. The apparatus consists of a 200-mm long-cylindrical plenum chamber with an inner diameter of ID = 100 mm. Flow comes in through the bottom of the plenum chamber and passes into a 300 mm long tube with an ID = 35 mm. The converging exit nozzle of the round jet was knife-edged and designed with a matched cubic prole having an exit diameter of D = 10 mm to give a top-hat exit velocity prole. A larger diameter exit nozzle D = 23 mm was also used for some experiments but most of the results presented herein were obtained with the 10 mm nozzle. To generate harmonic velocity oscillations at the nozzle exit, a TTi 40 MHz waveform generator was used to provide monochromatic sinusoidal input signals, which were amplied to drive two diametrically opposed loudspeakers tted to the plenum chamber. The air mass ow rates were set by an Alicat MC series mass ow controller and passed through an oil seeder consisting of double Laskin nozzles to produce an oil mist with mean diameters ranging from 0.3 to 0.5 lm for the PIV measurements. The rst step was to carry out a frequency sweep and measure the amplitude response A of the round jet by placing a hot-wire anemometer (HWA) at the center of the exit nozzle, where the A u0 f =U: The value of u0 is the magnitude of the Fourier transform centered at the forcing frequency, and U is the time-averaged velocity from the hot-wire data. As will be discussed shortly, it is important to note that physically meaningful denitions of A and other ow variables need to be carefully dened in order to relate the vortex formation process with the forcing conditions. Figure 1 shows that the largest amplitude response occurs at forcing frequencies of 40, 150, and 260 Hz corresponding to the acoustic modes of the system. The mean exit velocity proles under forced and unforced conditions were obtained from both hot-wire anemometry (HWA) and the high-speed PIV measurements (HSPIV) and are shown in Fig. 2. The velocity proles are approximately top hat and show good agreement apart from the shear layer region where the HSPIV results are not as spatially resolved as the HWA measurements due the relatively large eld of view required to capture the vortex formation process. Figure 2 shows the phase-averaged exit velocity at the jet centerline obtained from HSPIV can be well approximated as sinusoidal and scales linearly with forcing amplitude, A. This permits the adoption of a modied slug model to characterize vortex ring formation

123

Exp Fluids

(a)

(b)

0.8 0.7 0.6 0.5

0.4 0.3 0.2 0.1 0 0 50 100 150 200 250 300

f (Hz )
Fig. 1 A schematic of the experimental setup a and b the acoustic response as a function of forcing frequency, f Fig. 2 The exit velocity prole measured by HWA and PIV in a for both forced and unforced conditions, b shows the variation in centerline velocity as a function of forcing amplitude, A

(a)
1

(b)

2
Increasing A

0.8

1.5

U/ U

0.6

U(t)/ U
HWunforced HWforced PIVunforced PIVforced

0.4 0.5

0.2

0 0.6 0.4 0.2 0 0.2 0.4 0.6

0 0 0.2 0.4 0.6 0.8 1

r/D

t/ T

from a sinusoidal velocity programme superimposed upon a mean background ow. Following the usual notation, we can decompose the velocity into a mean and uctuating component Ut u0 t U: Here, the prime denotes the large-amplitude sinusoidal disturbances generated by the acoustic forcing noting that the turbulent uctuations are comparatively small. The time-averaged mean velocity of the jet is 1 U T ZT Ut dt:
0

Neglecting the effects of phase, the velocity at the exit of a jet subjected to harmonic forcing is Ut U 1 Asin2pft 2

where the amplitude A u0max =U and u0max is the peak amplitude of the sinusoidal velocity oscillation calculated from the PIV measurements. As hot-wire measurements were only acquired to calibrate the boundary conditions,

this denition of A serves as a suitable proxy to the frequency centered Fourier transform method. In order to make meaningful comparisons with the literature on vortex formation, the forcing parameters need to be expressed in terms compatible with the slug model (Lim and Nickels 1995; Shariff and Leonard 1992). This requires suitable denitions for the slug length (L), velocity (Up), and discharge time (t) in terms of the forcing conditions. The slug length is a derived Rt parameter where L 0 Ut dt and in these experiments, both L and t are functions wavelength k and the inverse of forcing frequency f, respectively. These parameters are used to describe a complete vortex formation process each forcing cycle and are treated as quasi-steady independent events. From physical arguments, we dene the slug discharge time over half a forcing period from t = 0 to t = T/2, i.e., when the jet velocity is greater than U: Setting the limits of integration to the slug discharge time, the mean slug which we term the effective piston velocity can be dened as

123

Exp Fluids
tT=2 Z

1 Up T

Ut dt:
t0

Table 1 List of experimental conditions including dimensional and nondimensional forcing parameters f (Hz) 40 U ms1 10.40 10.56 6.86 6.89 7.05 6.72 150 9.68 9.41 9.78 6.36 6.51 6.60 6.37 260 10.28 10.02 10.01 6.37 6.34 6.45 6.30 Up ms1 15.85 16.71 10.33 11.10 11.86 8.46 15.47 15.18 16.52 9.50 10.61 11.49 7.52 14.10 14.97 15.99 10.03 11.02 11.93 7.74 A 0.82 0.91 0.80 0.96 1.07 0.41 0.94 0.96 1.08 0.78 0.99 1.17 0.28 0.58 0.78 0.94 0.90 1.16 1.33 0.36 Re 6630 6735 4373 4396 4493 4285 6176 6003 6237 4055 4150 4206 4061 6555 6393 6383 4062 4044 4116 4017

Where this integral is evaluated from the ensemble averaged exit velocity of the jet obtained from the PIV measurements and forms the relevant velocity scale in terms of vortex formation. Evaluating Up over a discharge time of half the forcing period also relates the slug length L to the appropriate forcing length-scale k/2 via the frequency as t = T/2 = 1/2f giving a stroke length of L k Up  : 2 2f 4

It is important to note that the denominator 2f is the slug discharge time expressed in terms of the forcing frequency. As will be discussed later in this paper, identication of the formation number of a given ow hinges on the correct determination of L. The above expressions provide a direct analogy with the vortex ring experiments of Gharib et al. (1998) and Krueger et al. (2006). Combining Eqs. (2) and (3) and integrating, we obtain   2A Up U 1 : 5 p Equation (5) shows how the mean slug velocity Up is a function of both mean ow U and the forcing amplitude A. The forcing conditions can be nondimensionalized using Eq. (4) by incorporating both A and f L Up : D 2fD 6

U and Up calculated from the ensemble averaged PIV measurements. Re is based on D and U

This denition of L/D describes both the aspect ratio of the uid slug and the nondimensional formation time for vortex rings in sinusoidally forced jets. In other words, it expresses the magnitude of the forcing in terms of the length- and time-scales relevant to vortex formation. Using this equation, we can determine the formation number for high-amplitude forced jets. Up relates the forcing period T/2 and wavelength k/2 to the slug length where L is the effective slug volume displaced by the swept length of a piston stroke over the characteristic discharge time t = 1/2f. However, later on in this paper, we demonstrate that for sinusoidal forcing, the integration limits that determine the circulation of the vortex ring are actually from t = 0 to t = T/3, which can only be determined empirically and changes L : k/3 and equation (6) accordingly. Table 1 lists the forcing conditions investigated with 2D HSPIV. The HSPIV system was comprised of a PegasusPIV dual pulse 527 nm Nd:YLF laser with a maximum repetition rate of 10 kHz, a set of sheet forming optics and a Photron Fastcam SA-1.1 monochrome camera with a

maximum frame rate of 5.1 kHz at 1MP (max) resolution. The camera timing and the synchronization of the laser pulses were achieved using a LaVision high-speed controller. A 1 mm thick laser sheet was passed through the jet cross-section as shown schematically in Fig. 2 with the camera positioned normal to the imaging plane. The elds of view (FOV) were adjusted to cover the developing region of the jet up to x/D * 6, giving the images spatial resolutions between 0.06 and 0.09 mm/pixel. The time delays between image pairs were varied between 10 and 60 lm to ensure in-plane particle translations of 45 pixels depending on inlet conditions. In most of the cases, 33 image pairs per forcing period T were obtained requiring sampling frequencies of 1,320 and 4,950 Hz for f = 40 and 150 Hz, respectively. In order to maintain maximum camera resolution, 19 image pairs were obtained with a sampling rate of 4,940 Hz when the jet forcing frequency was f = 260 Hz. The cross-correlations were performed using Davis software. Firstly, image preprocessing which includes sliding background subtraction and particle intensity normalization was applied to the raw images. Second, a four pass cross-correlation scheme was applied, with a circular weighting function and window sizes of 32 9 32 pixels for the rst two passes and 16 9 16 pixels for the nal two

123

Exp Fluids

passes. A 50% overlap of adjacent windows was applied to each image pair providing a ow eld resolution of 0.50.7 mm. A vector postprocessing step including; median lter, smoothing, and lling up of empty spaces was applied. In Fig. 3, a sample raw PIV image with the corresponding velocity vector elds show the forming vortex ring. In order to determine the circulation, C of the vortex rings the vorticity elds were obtained using a 2nd order least-squares method following the recommendation of Raffel et al. (1998). The vortex ring circulation, Cvr was calculated using Stokes theorem combined with a suitable threshold I ZZ Cvr udl x dS: 7
C S

3 The evolution of vortex rings as a function of forcing conditions The formation of a vortex rings over a normalized forcing period t/T are shown by iso-vorticity contour plots in Figs. 4, 5, and 6, which correspond to forcing frequencies of f = 40, 150 and 260 Hz, respectively. For comparison, each gure shows the effect of two values of A on the formation process. The vorticity is normalized by the mean velocity x xD=U for contour ranges of 20 B x* C -20 with increments 1 with normalized time-stamps shown at the top of the gures. For reasons that will become apparent later, the vortex ring formation will be described with reference to the forcing frequency f and amplitude A rather than in terms of their slug aspect ratio L/D. In each of the gures, the initial roll-up, growth, pinchoff, and advection of the vortex rings is clearly seen. In general, an increase in f promotes earlier breakup of the jet into smaller vortex rings. For a xed f, an increase in A causes the formation process to move upstream closer to the jet exit. This effect is nicely illustrated in Fig. 4 by comparing the vortex formation process when A = 0.41 and 0.96. For A = 0.96, the vortex ring rolls up earlier in the forcing cycle and closer to the jet exit as shown when t = 0.15T. Consequently, the vortex ring formed for A = 0.96 is much bigger by t/T = 0.45. This A-dependence amounts to a phase shift between the formation timescales of the vortex ring and the forcing period. This effect is also evident when the jet is forced at f = 150 and 260 Hz. Figures 5 and 6 show that the number of rings visible within the FOV decreases from two rings to one and from three rings to two, respectively, as the spacing between consecutive rings increases with A. This increase in spacing between consecutive rings occurs because during formation the ring convects with Up which scales with both U and A according to Eq. (5). These effects are discussed further in Sect. 5.

The circulation of a vortex ring can be quantied by measuring the circulation around the largest iso-vorticity contour which denes its boundary. However, dening the boundary of a vortex ring is problematic given that the formation process is a transient event. This has an impact on dening other aspects of vortex ring formation such as pinch-off, separation, and the core size. To this extent, we adopt a similar approach to others (Gharib et al. 1998; Krueger et al. 2006) by dening a threshold around an appropriate iso-vorticity contour level. A sensitivity analysis showed that this threshold value can signicantly inuence the value of vortex ring circulation since it affects the area and the sum of bounded vorticity. Krueger et al. (2006) used a threshold with a normalized vorticity of x* = 0.91, which corresponded to approximately 20% of the peak vorticity in the pinched off ring. In the current study, we found that the smallest value which gave sensible results was x* = 1, for a normalization based on the mean jet velocity U; making the vortex ring boundaries and the separation events clear and identiable. In the worst cases, this value corresponds to approximately 5% of the peak vorticity of the ring, which is suitably resolved.

Fig. 3 Mie-scattering image showing the roll-up of a vortex ring in a and b the corresponding vector eld

(a)

(b)

4 10 m/s

0 2

1.5 1

0.5

0.5

1.5

0 2

123

Exp Fluids
*

t = 0.00 T

t = 0.15 T

t = 0.30 T

t = 0.45 T

t = 0.61 T

t = 0.76 T

t = 0.91 T

6 5 4

10

x/D

3 2

5 1 0
t = 0.00 T t = 0.15 T t = 0.30 T t = 0.45 T t = 0.61 T t = 0.76 T t = 0.91 T

10 * 10

6 5 4

x/D

3 2

5 1 0 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 10

r/D

r/D

r/D

r/D

r/D

r/D

r/D

Fig. 4 Normalized iso-vorticity contours x* illustrating the effect of amplitude A on the formation of a vortex ring over a forcing period, f = 40 Hz, A = 0.41 (top) and 0.96 (bottom)
*

t = 0.00 T

t = 0.15 T

t = 0.30 T

t = 0.45 T

t = 0.61 T

t = 0.76 T

t = 0.91 T

10 5

4 3

x/D

2 1 0
t = 0.00 T t = 0.15 T t = 0.30 T t = 0.45 T t = 0.61 T t = 0.76 T t = 0.91 T

0 5 10 * 15 10 5 0 5

4 3

x/D

2 1 0

10 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 15

r/D

r/D

r/D

r/D

r/D

r/D

r/D

Fig. 5 Normalized iso-vorticity contours x* illustrating the effect of amplitude A on the formation of a vortex ring over a forcing period, f = 150 Hz, A = 0.28 (top) and 0.9 (bottom)

For f = 40 and 150 Hz, the existence of a trailing jet is evident, with the former resembling the type of ow eld encountered in the experiments of Gharib et al. (1998) for starting jets with long nondimensional stroke ratios (L/D). In Fig. 4, the separation of the leading vortex ring from the

trailing jet occurs when part of the ring has convected out of the FOV by t & 0.5T. This only occurs when the jet is forced at f = 40 Hz due to the longer wavelength, whereas the formation process is fully captured in Figs. 5 and 6 and lasts up to t & 0.5T. Forcing at f = 150 Hz shows some

123

Exp Fluids
*

t = 0.00 T

t = 0.16 T

t = 0.32 T

t = 0.47 T

t = 0.63 T

t = 0.79 T

t = 0.95 T

15 10 5 0 5

4 3

x/D

2 1 0
t = 0.00 T t = 0.16 T t = 0.32 T t = 0.47 T t = 0.63 T t = 0.79 T t = 0.95 T

10 15

20 10

4 3

x/D

2 1 0 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1

0 10 20

r/D

r/D

r/D

r/D

r/D

r/D

r/D

Fig. 6 Normalized iso-vorticity contours x* illustrating the effect of amplitude A on the formation of a vortex ring over a forcing period, f = 260 Hz, A = 0.36 (top) and 1.16 (bottom)

interesting phenomena. First of all, the lead ring separates from the trailing jet much later in the forcing period and secondly, the vorticity left in the wake of the leading ring starts to roll-up before it is entrained into the oncoming ring. For A = 0.9, the entrainment process is completed by the end of the forcing period as shown at the time-step t = 0.91T in Fig. 5. It is also interesting to note that the magnitude of the wake vorticity reduces when A is increased due to the formation of a larger ring. This entrainment phenomena is somewhat similar to the leapfrogging effect reported by Gharib et al. (1998). Another distinguishing feature between the forcing frequencies is the extent of the shear layer disruption. When the jet is forced at f = 40 Hz with A = 0.41, the vorticity eld clearly uctuates in response to the velocity oscillations, however, for the most part x*-contours remain attached to the jet exit extending throughout the near eld. As A is increased, the shear layer is weakened as more vorticity is rolled-up into the vortex ring, yet there is evidence that the shear layer remains attached to the jet exit. Forcing at wavelengths longer than the near eld of the jet, i.e., x/D & 5, did not promote complete breakdown of the jet shear layer. The other extreme is found when the jet is forced at f = 260 Hz where the entire near eld is dominated by the formation of a train of vortices each forming within 1D of the jet exit with no discernible structure associated with the mean ow remaining. The relative phase between the time- and length-scales between vortex formation and the forcing conditions in the near eld of jet ows have important consequences on a variety of technical applications, such nonlinear ame dynamics in lean

premixed ames that result from thermo-acoustic instability, and thrust augmentation techniques discussed in Balachandran et al. (2005), Mason and Miller (2006), respectively. To further elucidate the effects of forcing amplitude, the vorticity elds from A = 0 to the maximum value are plotted at the same normalized time-step of t/T = 0.2 for f = 150 Hz (top) and f = 40 Hz (bottom) in Fig. 7. The unforced jet shows the natural development of the Kelvin Helmholtz instability along the jet shear layer, which is locked into the forcing frequency as A is increased. This gure provides a nice illustration of how increasing A causes the vortex roll-up progressively closer to the jet exit. The value of A at which this occurs is clearly f-dependent, with higher values needed as f is decreased. Figure 7 shows that for f = 150 Hz, the critical value of A lies between 0.4 and 0.5. whereas for f = 40 Hz, the value of A [ 1. The increase in spacing between consecutive rings is also captured for the f = 150 Hz case. These results suggest that in the case of forced jets other L/D-dependent regimes might be considered in addition to the formation number. For example, the minimum L/D required to form a coherent vortex. This was investigated by Kulsheimer and Buchner (2002) in a swirling jet with and without combustion excited by a siren. These authors showed that the minimum amplitude required to form a coherent vortex decreased hyperbolically with increasing nondimensional frequency (Strouhal number). A hint of this trend can be seen in Fig. 7, however, lower values of A were not investigated in this study. Moreover, their denitions of A and St were based on r.m.s. and mean

123

Exp Fluids
*
20 4 10 3

A = 0.00

A = 0.28

A = 0.53

A = 0.78

A = 0.99

A = 1.17

x/D

2 1 0

0 10 20

A = 0.00
4

A = 0.41

A = 0.59

A = 0.80

A = 0.96

A = 1.07

10 5

x/D

0 2 5 1 0 10 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1

r/D

r/D

r/D

r/D

r/D

r/D

Fig. 7 The amplitude dependence on the roll-up location for f = 160 Hz (top) and f = 40 Hz (bottom) at the same normalized time-step

quantities and are distinctly different from the derived slug parameters in Eqs. (5) and (6) making exact comparisons difcult. The effect of forcing conditions on the length- and timescales associated with the formation process can also be examined in terms of the spatial location where lead ring separates from the jet and moves off at its own induced velocity. Separation is dened as the moment when the lowest iso-vorticity contour of the lead ring separates from the trailing jet. As will be discussed in Sect. 4, the difference between separation and pinch-off of a vortex ring is signicant and care should be taken not to confuse the two. The vorticity elds for all the experiments were examined to identify when and where separation occurred during the forcing cycle. The separation time ts and location xs/D obtained from all the experiments are plotted in Fig. 8. The x-axis is the nondimensional time dened as t ts Up =D: The location of separation as a function of nondimensional time follows an approximately linear trend whose slope corresponds to a velocity of &Up/2. These results show good agreement with those of Didden (1979) and Mohseni and Gharib (1998).

5 4 3 2 1 0 0 2 4 6 8 10
Re = 4,250 Re = 6,750 Re = 10,500 Re = 7,500 Re = 7,000 Re = 15,000 Re = 22,500

decreasing frequency

xs / D

t
Fig. 8 Normalized downstream position xs/D of all vortex rings at the rst observed separation from the trailing jet in formation time t* = Ut/D

4 The circulation of vortex rings as a function of forcing conditions This section of the paper discusses how the forcing conditions relate to the size of the vortex rings in terms of their

circulation, Cvr . Figure 9 shows the total and vortex ring circulations as a function of normalized period for a range of forcing conditions. As before, each graph compares the effect of different A for forcing frequencies of f = 150 in Fig. 9a, b and f = 260 Hz in Fig. 9c, d. The circulation values are normalized by C C=UD: The total circulation C is represented by the solid lines, whereas the circulation of the vortex rings Cvr are denoted by the vertical dashed lines. Normalizing the circulation with U opposed to Up t=2; which would collapse the data, was done to illustrate the effect of different forcing conditions. For consistency, when calculating Cvr in Fig. 9a, b, the trailing vorticity left over from the previous cycle as shown in
2

123

Exp Fluids Fig. 9 Vortex ring circulation as a function of normalized period for different forcing amplitdue A according to Table 1. Top a, b f = 150 Hz, Bottom c, d f = 260 Hz

(a)

(b)

8 7

5 6 4 5

vr at A=1.17 at A=1.17 vr at A=0.99

4 3

2 1 0

vr at A=1.08 at A=1.08 vr at A=0.94 at A=0.94

at A=0.99 vr at A=0.78 at A=0.78

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

t/

t/

(c)

(d)

vr at A=1.33

vr at A=0.94

at A=1.33 vr at A=1.16 at A=1.16 vr at A=0.90 at A=0.90

at A=0.94 vr at A=0.78 at A=0.78 vr at A=0.58 at A=0.58

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

t/

t/

Fig. 5 was excluded as the entrainment process continues after the ring has left the FOV for some values of A. During the rst half of the forcing cycle, the total circulation grows steadily before levelling off around midcycle after which the circulation of the vortex ring can be determined. The results show that both C and Cvr increase with A suggesting that the vortex rings in these cases are not optimal in terms of their circulation, in other words, they are below the formation number following the terminology of Gharib et al. (1998). However, a closer look at Fig. 9 reveals a relationship between C and Cvr : In each of the gures, the vortex ring circulation is approximately equal to the total circulation discharged within a narrow range of time-scales from t/T = 0.28 to 0.35 with the average being t/T & 0.33 or a third of the cycle. Physically, this means that the vorticity ux issuing from the jet exit up to this time determines the circulation budget of the forming vortex ring. For these cases, it also corresponds to the time when pinch-off occurs. As mentioned by Dabiri (2009), the difference between pinch-off and separation

requires succinct denition. Pinch-off refers to the moment when vorticity from a discharging slug stops owing into the developing ring. By denition, the dynamics of this process cannot be resolved experimentally as there is no viscous length-scale associated with a vortex sheet; however, the experiments of Gharib et al. (1998) showed that the moment of pinch-off can be inferred from time-scales obtained in the C plots, which for a wide range of forcing conditions was found to be t/T & 0.33. Separation on the other hand refers to when the lowest iso-vorticity contour around the vortex ring disconnects from the trailing jet ow as shown at t/T = 0.76 in Fig. 5(bottom) and is sensitive to the value selected for thresholding. This time-scale was found in all the experiments listed in Table 1 with the exception of some cases where pinch-off happens earlier due to the formation of optimal vortex rings, which is discussed shortly. Further insight can be gained by comparing the ring velocity Wr with the shear layer velocity. Figure 10 plots the time-dependent core location in the x-direction, which

123

Exp Fluids

(a) 3.5
3 2.5 2 1.5 1

(b)
VR Core Velocity Shear Layer Velocity

1.2 1 0.8 0.6 0.4 0.2

X VR /D

0.5 0 0 0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8

t/ T

U(t)/ U

t/ T

Fig. 10 The position of the vortex ring core for the case in Fig. 9 as a function of normalized period a, b the corresponding shear layer and ring core velocity

is used to calculate Wr and compare with the shear layer velocity in Fig. 10. The core location was found by tracking the peak vorticity, which in these data was found to be more reliable than tracking the centroid, which would include vorticity left in the wake of the previous ring. The ring velocity was obtained from the core position using a central differencing scheme at each time-step to compare with the shear layer velocity Ut=2: As shown in Fig. 10, Wr exceeds the shear layer velocity at t/T & 0.35, which is consistent with the range of time-scales centered around t/T & 0.33. It is interesting to note that this time-scale is the same as the pinch-off time-scale of the leading edge vortex (LEV) formed by a harmonically plunging aerofoil in the experiments of Rival et al. (2009). There are several interesting aspects to this time-scale. Recall from Sect. 2 that integration limits of t = 0 to t = T/2 were selected in order to dene the discharge time of the uid slug hence the mean slug velocity Up : In all the cases, shown in Fig. 9, the vortex ring pinches off at the earlier time-scale of t & 0.33T. However, we could have obtained the same value for Up by setting the integration limits from t = 0 to t = T/4 due to symmetry in the forcing conditions. This would require pinch-off to occur at t = T/4, which is an inexion point i.e., is when dU/dt goes from positive values at t = 0 to zero at t = T/4. The wide range of experimental results suggest that it is the change in sign of dU/dt that initiates pinch-off at t/T & 0.33, which means that the value of Up is slightly underestimated when calculated over half of the forcing period. The next section in this paper discuss how t/T = 0.33 represents an empirically derived kinematic constraint for jets subjected to harmonic forcing and cannot be analytically determined from energy arguments. Moreover, it enables the

effective slug length L to be determined and hence the formation number. So, if a formation number exists in forced jets, the forcing conditions need to be arranged to deliver the maximum impulse before or at t = 0.33T. Figure 11 plots the change in C and Cvr when the jet is forced at f = 40 Hz for a range of A. The vortex ring circulation intersects the total circulation curve at t/T & 0.17 in Fig. 11 and t/T & 0.12 in Fig. 11. The fact that the time when C % Cvr occurs before t = 0.33T suggests the formation of optimal rings. In these cases, the mechanism of pinch-off is due to the energy constraint described by Gharib et al. (1998) and Shusser and Gharib (2000) rather than the kinematic limit described previously supporting the supposition of optimality. Comparing the pinch-off times in Fig. 11 with the time when Wr equals or exceeds the shear layer velocity in Fig. 12 further illustrates the change in pinch-off mechanism and optimal ring circulation. According to Fig. 11,Wr intersects the shear layer velocity at t/T = 0.31, however by this time pinch-off has already happened. By contrast, separation of the lead ring from the trailing jet occurs several time-steps later as shown in the bottom of Fig. 4. These results suggest that if the formation number can be applied to harmonically forced jets, it must account for the kinematic constraint at t/T & 0.33. In other words, the formation number L/D & 4 occurs when the forcing conditions are arranged to provide maximum circulation at t = 0.33T. 5 The formation number of sinusoidally forced jets By investigating the process of vortex formation in highamplitude forced jets, we are asking what effect does the

123

Exp Fluids

(a)

16
vr at A=1.07

(b)
at A=1.07 vr at A=0.96 at A=0.96 vr at A=0.80 at A=0.80

16
vr at A=0.91

14 12 10

14 12 10

at A=0.91 vr at A=0.82 at A=0.82

8 6 4 2 0

8 6 4 2 0

0.1

0.2

0.3

0.4

0.1

0.2

0.3

0.4

t/ T

t/ T

Fig. 11 Vortex ring circulation as a function of normalized period for different f = 40 Hz for a range of velocity amplitudes A

(a)

(b)

1.2

0.8

X VR / D

U(t)/ U

0.6

0.4

0.2
VR Core Velocity Shear Layer Velocity

0.1

0.2

0.3

0.4

0.1

0.2

0.3

0.4

t/ T

t/ T

Fig. 12 The position of the vortex ring core for a case in Fig. 11 as a function of normalized period a, b the corresponding shear layer and ring core velocity

presence of a mean ow U have on the formation number? In this section of the paper, we will show that the limiting time-scale t = 0.33T serves as both a kinematic constraint and the maximum slug discharge time-scale associated with the formation number of forced jets. Figures 9, 10, 11, and 12 demonstrated that two different pinch-off mechanisms were at play depending on the forcing conditions. For a wide range of experiments, pinch-off occurred at t = 0.33T, whereas for small range of experiments pinch-off happened earlier. In the latter case, the mechanism of pinch-off is not kinematic but related to the relative energy of the forced jet and the vortex ring. For vortex rings formed with long stroke ratios, i.e., L/D C 4, Shusser and Gharib (2000) proposed

that pinch-off occurs when the jet is no longer able to deliver energy at a rate compatible with the energy of a steadily translating vortex ring. In other words, the ring will continue to accumulate vorticity as long as the energy of the jet exceeds the energy of the ring, Ej [ Er. However, in the forced jet case, there is a constraint on the maximum time available for the ring to acquire its maximum circulation, which is t = 0.33T and means that the condition Er C Ej must be met at or within this time constraint or else the ring will pinch-off. By taking this into account, we can analytically determine the value of L/D when Ej = Er at t = 0.33T. Employing the slug model approximations for the circulation C; impulse I, and the energy E, we obtain

123

Exp Fluids

1 C LUp ; 2 1 2 I pD qLUp ; 4 1 2 E pD2 qLUp : 8

8 9 10

Evr;nd

^2 t

 2 L 9a p : D 2p

17

where Up is dened according to Eq. 5 and L is the slug length. Generalizing these expressions in terms of the forcing parameters where L Up t where the discharge time t is a function of f and the normalized time ^ t=T ft, we obtain t Cj Ij E ^ 2 t Up ; 2f 11 12 13

^ t 2 pD2 qUp ; 4f ^ t 3 pD2 qUp : 8f

The energy of a steadily translating vortex ring Evr can be calculated according to Shusser and Gharib (2000) by q Evr a qIC3 14 where the limiting value for the dimensionless coefcient a was experimentally determined by Gharib et al. (1998) as a & 0.33. It has to be mentioned that this denition of Evr makes some assumptions about various properties of the vortex ring. For example, it is assumed that both the vorticity distribution in the core and the radius of the ring is xed. However, the most important assumption is that the formation of the vortex ring is a quasi-steady process. The quasi-steady assumption is implicit in the denitions of Up and the other slug parameters which treats the formation of each vortex as an independent quasi-steady event occurring once per cycle. During the time where Ej [ Evr, the jet transfers energy to the ring, which is manifested by an increase in circulation. At the moment when Ej = Evr, pinch-off occurs restricting any further growth of the ring by cutting-off the vorticity ow. In order to plot the growth in the jet and ring energies as a function of the normalized forcing period (t/T), we use the dimensionless kinetic energy End 4E pD3 qUp
2

When Ej,nd = Evr,nd Eqs. 16 and 17 can be combined to solve for the value of L/D at that instant to obtain the formation number for harmonically forced jets. Recall that Up was evaluated from Eq. 5 over T/2 and not T/3. Consequently, the values of Up in Table 1 are all slightly underestimated. Figure 13 shows that the % change of Up as a function of A between the two different integration limits is only a few % up to A = 1, which is within experimental accuracy. However, the change to the slug discharge time, which is expressed in terms of frequency is signicant, i.e., t = 1/2f to 1/3f and needs to be accounted for if the effective slug length L and hence formation number is to be correctly determined. The growth in the nondimensional jet and ring energies as a function of normalized period is shown in Fig. 14 for a selection of forcing conditions. For a given arrangement of Up ; U; f ; and A, a set of curves is produced illustrating how the relative growth rates of the jet and ring vary over a normalized forcing cycle. Figure 14 shows the growth of the jet and ring energies when the jet is forced at f = 150 Hz with A = 1.17. For this case, the jet energy exceeds the vortex ring energy up to t/T = 0.5. This means that purely from an energy argument the energy of the jet can feed the ring until this time, however Fig. 9 shows that the growth of the ring is halted due to the kinematic con-

% errorin Up

15

0.5

1.5

to cast the normalized jet and ring energies in Eqs. 13 and 14 into their nondimensional forms where Ej;nd 3^ L t 2D 16

A
Fig. 13 Percentage change in the mean effective piston velocity Up when calculated from Eq. 3 with empirically determined limits from t = 0 to t = 0.33T

123

Exp Fluids Fig. 14 The nondimension vortex ring and jet energy as a function of normalized forcing period for a variety of experimental conditions and the analytically determined value of L/D at the kinematic time-scale t = 0.33T

(a)

3
Ej L/D=2.55 Evr L/D=2.55

(b)
Ej L/D=3.80 Evr L/D=3.80

3
Ej L/D=8.61 Evr L/D=8.61

2.5

2.5

Ej L/D=13.21 Evr L/D=13.21

E nd

1.5

E nd
Kinematic Restriction

1.5

0.5

0.5

0.2

0.4

0.6

0.8

0.05

0.1

0.15

0.2

t/ T

t/ T

straint occurring at t/T = 0.33. The nondimensional energy of the ring is not equal to the jet at this time, and therefore, the ring can be considered suboptimal. Based on the values from Table 1, the forcing conditions correspond to L=D Up =3fD 2:55, which is below the formation number. Figure 14 also plots the moment when Ej = Evr, which is coincident with the time-scale t = 0.33T. This condition is met when L/D = 3.80, which shows remarkable agreement with the formation number in starting jets originally proposed by Gharib et al. (1998). Figure 14 on the other hand shows two forcing cases where the nondimensional energy of the ring exceeds the jet prior to t/T = 0.33. In these cases, the jet was forced at f = 40 Hz for values of A = 0.8 and 0.82. By comparing with the experimental results in Fig. 11, we see that the analytical model correctly predicts the pinch-off times. The fact that pinch-off occurs so early in the forcing cycle is striking and is due to the fact that they correspond to nondimensional stroke ratios of L/D = 8.6 and 13.2. This highlights the importance of recovering the correct slug parameters needed to correctly dene L, particularly the discharge time as a function of the forcing conditions, which needs to be found empirically. Incorporating the kinematic time-scale into Eq. 6, the formation number for harmonically forced round jets can be expressed as L Up % % 4: D 3fD 18

et al. (2009) reports such a shift. This kinematic restriction was also observed but not commented upon in the experiments of Gharib et al. (1998) for vortex rings formed with a slow-ramp velocity programme shown in their Fig. 2 (which is a good approximation of the positive half of a sine wave). By cross-referencing their Fig. 2 with the corresponding circulation plot in their Fig. 12 shows that the actual pinchoff time occurs just after the peak velocity which agrees with our experiments (note: the value of L/D in the captions of their Figs. 12 and 13 should be swapped).

6 Conclusion The periodic formation of vortex rings in the development region of a round jet subjected to high levels of acoustic forcing has been performed for a wide range of experimental conditions. When subjected to harmonic velocity oscillations, the near eld of the jet breaks up into vortex rings of various size and circulation depending on the forcing conditions. The spacing between consecutive rings and the initial roll-up location are both amplitude and frequency dependent. By considering the total and vortex ring circulation over the forcing period revealed that a limiting time-scale of t/T = 0.33 kinematically restricts the growth of a vortex ring for any given arrangement of jet forcing conditions. This kinematic constraint is imposed by the translational velocity of the vortex ring exceeding the shear layer velocity which emanates from the change in sign in the jet acceleration. This time-scale was also reported by Rival et al. (2009) for the separation of a vortex ring from the leading edge of an harmonically plunging aerofoil, which lends credence to the notion that this time-scale may have a more universal character for vortex rings formed in other congurations which experience a sinusoidal velocity history.

where t = 1/3f corresponds to the discharge time of the uid slug which occurs over one third of the forcing period. It is worth pointing out that there is some evidence that this kinematic restriction is more universal in character and may apply to any time-dependent velocity programme, i.e., velocities programmes where the acceleration changes sign. The kinematic time-scale will shift according to when the acceleration changes sign in the velocity programme, Rival

123

Exp Fluids

By considering the energy of the jet and a steady translating vortex ring in conjunction with the kinematic time-scale of t/T = 0.33, the formation number for sinusoidally forced jets was analytically determined to be L/D = 3.80, which agrees well with the results of Gharib et al. (1998). This illustrates that care needs to be taken when determining the slug discharge time in order to recover the effective stroke length L in a forced jet and will most likely need to be experimentally determined for a given velocity programme.
Acknowledgments Dr James Dawson is indebted to Dr Tim Nickels for his many suggestions and helpful discussions on vortex rings related problems, he will be sorely missed. Dr James Dawson is supported by EPSRC under the Advanced Research Fellowship scheme.

References
Balachandran R, Ayoola B, Kaminski C, Dowling A, Mastorakos E (2005) Experimental investigation of the nonlinear response of turbulent premixed ames to imposed inlet velocity oscillations. Combust Flame 143(12):3755 Becker HA, Massaro TA (1968) Vortex evolution in a round jet. J Fluid Mech 31:435448 Crow SC, Champagne FH (1971) Orderly structure in jet turbulence. J Fluid Mech 48:547591 Dabiri JO (2009) Optimal vortex formation as a unifying principle in biological propulsion. Annu Rev Fluid Mech 41:1733 Dabiri JO, Gharib M (2004) Delay of vortex ring pinchoff by an imposed bulk counterow. Phys Fluids 16:L28L30 Didden N (1979) On the formation of vortex ringsRolling-up and production of circulation. Z Angewandte Math Physik 30:101116 Gharib M, Rambod E, Shariff K (1998) A universal time scale for vortex ring formation. J Fluid Mech 360:121140 Gutmark E, Ho C (1983) Preferred modes and the spreading rates of jets. Phys Fluids 26:29322938 Johari H (2006) Scaling of fully pulsed jets in crossow. AIAA J 44:792801

Krieg M, Mohesni K (2008) Thrust characterization of a bio-inspired vortex ring generator for locomotion of underwater robots. IEEE J Ocean Eng 33:123132 Krueger PS, Gharib M (2005) Thrust augmentation and vortex ring evolution in a fully pulsed jet. AIAA J 43:792801 Krueger PS, Dabiri JO, Gharib M (2006) The formation number of vortex rings formed in uniform background coow. J Fluid Mech 556:147166 Kulsheimer C, Buchner H (2002) Combustion dynamics of turbulent swirling ames. Combust Flame 131(1-2):7084 Lee MJD, Reynolds WC (1985) Bifurcating and blooming jets. In: 5th Turbulent shear ow symposium, Cornell University, Ithaca, NY Lim T, Nickels T (1995) Vortex rings. In: Green S (ed) Fluid vorticies. Kluwer, Dordrecht Linden PF, Turner JS (2001) The formation of optimal vortex rings, and the efciency of propulsion devices. J Fluid Mech 427:6172 Mason SA, Miller RJ (2006) The performance of ejectors driven by sinusoidally unsteady jets. In: 44th AIAA aerospace sciences meeting and exhibit AIAA, Reno, NV Mohseni K, Gharib M (1998) A model for universal time scale of vortex ring formation. Phys Fluids 10:24362438 Opalski AB, Paxson DE, Wernet MP (2005) Detonation driven ejector exhaust ow characterization using planar PIV. In: 41st AIAA aerospace sciences meeting and exhibit AIAA, Reno, NV Pullin DI (1979) Vortex ring formation at tube and orice openings. Phys Fluids 22:401403 Raffel M, Willert C, Kompenhans J (1998) Particle image velocimetry: a practical guide. Springer, New York Raman G, Zaman KBMQ, Rice EJ (1989) Initial turbulence effect on jet evolution with and without tonal excitation. Phys Fluids 1:12401248 Rival D, Prangemeier T, Tropea C (2009) The inuence of airfoil kinematics on the formation of leading-edge vortices in bioinspired ight. Exp Fluids 46:823833 Shariff K, Leonard A (1992) Vortex rings. Annu Rev Fluid Mech 24:235279 Shusser M, Gharib M (2000) Energy and velocity of a forming vortex ring. Phys Fluids 12:618621 Zaman KBMQ, Hussain AKMF (1980) Vortex pairing in a circular jet under controlled excitation. I. General jet response. J Fluid Mech 101:449491

123

You might also like