You are on page 1of 11

ARTICLE IN PRESS

Experimental Thermal and Fluid Science xxx (2007) xxxxxx www.elsevier.com/locate/etfs

Comparison of turbulent jets issuing from rectangular nozzles with and without sidewalls
Ravinesh C. Deo
*,1,

Graham J. Nathan, Jianchun Mi

School of Mechanical Engineering, The University of Adelaide, SA 5005, Australia Received 3 December 2006; accepted 26 June 2007

Abstract This paper reports a systematic study of a turbulent jet issuing from a rectangular slot nozzle of high-aspect ratio, AR ( w/h, where w and h are the long and short sides of the slot, respectively) tested with and without sidewalls. The solid sidewalls were ush with each of the slots short sides vertically and extend axially along the streamwise direction of the jet. Hot-wire measurements were conducted at a Reynolds number based on slot-width (h) and exit centerline velocity of Reh % 7000 (for AR = 30 and 60) and at 10,000 (for AR = 30) up to 160 h downstream. All jets have a potential core in which the local centerline velocity is approximately constant, followed by a transition region and then a statistically two-dimensional (2-D) region where the centerline mean velocity, Uc $ x1/2. The potential core of the jet without sidewalls is shorter than that with sidewalls. Near eld power spectral analysis reveals that the primary vortex shedding rate is higher for the jet without sidewalls than the jet with sidewalls. The 2-D region of the jet with sidewalls is found to extend over a longer axial distance than that of the jet without sidewalls. It is also demonstrated that both the decay and spread rates of the jet within the 2-D region are lower for the case with sidewalls. Beyond the 2-D region, the jet without sidewalls enters into a far eld transitional phase and then tends to behave statistically like an axisymmetric jet with Uc $ x1. The centerline turbulence intensity of the jet with sidewalls becomes asymptotic closer to the nozzle exit than the jet without sidewalls. The skewness and atness factors conrm further statistical dierences between the jet with and without sidewalls. Crown Copyright 2007 Published by Elsevier Inc. All rights reserved.
Keywords: Plane jet; Planar jet; Rectangular jet; Eect of sidewalls

1. Introduction The standard nozzle conguration used to produce a planar jet comprises by a rectangular slot of dimensions w h and high aspect ratio, AR ( w/h, where w and h are nozzles long and short sides) with two parallel plates attached as sidewalls to the slots short sides. The sidewalls extend axially in the direction of jet propagation. This arrangement ensures statistical two-dimensionality because
Corresponding author. Tel.: +61 8 8303 5460; fax: +61 8 8303 4367. E-mail address: r.deo1@uq.edu.au (R.C. Deo). 1 Present address: Center for Remote Sensing and Spatial Information Sciences, School of Geographical Sciences and Planning, The University of Queensland, Brisbane 4072, Australia. 2 Present address: Department of Energy and Resource Engineering, College of Engineering, Peking University, Beijing 100871, China.
*

the jet is forced to entrain the ambient uid only in the direction normal to the nozzles long sides. For instance, the early work of Heskestad [1], Bradbury [2] and Gutmark and Wygnanski [3] warrant the use of sidewalls to create a statistically two-dimensional (planar) jet. In contrast, a jet issuing from a similar rectangular slot but congured without sidewalls are termed rectangular even though the jet is only truly rectangular at the exit plane. For example, Sforza et al. [4], Trentacoste and Sforza [5], Sfeir [6] and Quinn [7] have used the terminology rectangular jet to assess the development characteristics of a jet from a rectangular nozzles without sidewalls. We also adopt this terminology for consistency, where applicable in this paper. If the aspect ratio of a rectangular nozzle is suciently large, this jet, too, behaves like a planar jet up to a certain downstream distance [8]. This downstream distance,

0894-1777/$ - see front matter Crown Copyright 2007 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.expthermusci.2007.06.009

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
2 R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx

Nomenclature AR e Fu h Ku Ky Re Su u u0 Uc Uo,c nozzle aspect ratio (AR = w/h) dissipation rate of this energy, assuming isotropy with e 15tU 2 ou=ot2 c centerline atness factor, Fu = hu4i/(hu2i)2 slot-width of a rectangular nozzle decay rate of mean centerline velocity jet spreading (widening) rate Reynolds number Reh  Uo,c h/t centerline skewness factor, Su = hu3i/(hu2i)3/2 uctuation component of mean velocity in streamwise (x) direction root-mean-square (rms) of the velocity uctuation, u 0 = hu2i1/2 local mean velocity on the centerline mean exit centerline velocity w x01 x02 xp y0.5 m x,y,z z0.5 slot-span of a rectangular nozzle virtual origin of the normalized mean centerline velocity virtual origin of the normalized velocity halfwidth length of the jets potential core velocity half-width, calculated at the y-location at which U x 1 U c x 2 kinematic viscosity of the air (% 1.5 105 m2 s1 at 20 C) streamwise or axial (x), lateral (y) and spanwise or transverse (z) coordinate velocity half-width, calculated at the z-location at which U x 1 U c x 2

however, depends on the nozzle aspect ratio [9,10]. It is perhaps for these reasons, or others, that some previous investigations have treated the two congurations as if they could be used interchangeably. For example, Krothapalli et al. [11], Hussain and Clark [12] and Namar and Otugen [13] have used the term planar to describe a jet that issued from rectangular nozzles of high-aspect ratio but congured without sidewalls. This approach is reasonable given that the mean velocity decay rates in the self-similar region of both jets obey the relation Uc $ x1/2, and thus exhibit statistically two-dimensional behaviour. Namar and Otugen [13] have checked the jets spanwise velocity distribution to further conrm statistical twodimensionality of jets from rectangular nozzles without sidewalls. They found that the spanwise distribution of mean velocity and turbulence intensity were fairly uniform in the downstream region. Furthermore, Mi et al. [8] veried experimentally that the relation Uc $ x1/2 also holds true in the 2-D region of rectangular jets, but only within a limited axial distance. Nevertheless, given the sensitivity of all shear ows to boundary conditions, some dierences can be expected to result from the presence or absence of sidewalls. This issue remains to be explored at present. A study undertaken by Hitchman et al. [14] to identify dierences between jets with and without sidewalls is inconclusive. It showed that the mean centerline velocity of the jet without sidewalls decays slower than for the jet with sidewalls, while its spreading rate is higher than the jet with sidewalls. This clearly contradicts the law of mass conservation, since a jet which decays faster must also spread faster, by virtue of a higher rate entrainment of ambient uid. No other direct comparisons of a jet with and without sidewalls currently exist, thus no detailed statistical information is available. Hence the present investigation seeks to address this need. An assessment of the impacts of the presence or absence of sidewalls is also of relevance to the broader understand-

ing of the eects of boundary conditions on turbulent shear ows. It has been well established by George [15] and George and Davidson [16], both analytically and experimentally, that dierences in boundary conditions propagate into the greater ow eld. However, the extent of such inuences is not so well known. The present investigation aims to provide more detailed statistical comparison of otherwise identical jets of large aspect ratio with and without sidewalls over a ow region of up to 160 h using hot-wire measurements. 2. Experimental details The present nozzle facility, shown schematically in Fig. 1, consists of an open circuit wind tunnel of hexagonal cross section built to generate uniform ows. The wind tunnel, made of wooden modules with polished inner surfaces, was driven by a 14.5 kW aerofoil-type centrifugal fan mounted rmly to the oor so that the entire rig is vibration free. The tunnel employed a wide angle (25) diuser spanning 2100 mm in length into a settling chamber. Within the diuser are two screens of 60% porosity at intervals of 600 mm that aid in reduction of velocity defect in the turbulent boundary layer and streamlines the incoming airow. The settling chamber (of length 1400 mm) contained a honeycomb of length 150 mm assembled via drinking straws aligned horizontally with the main stream ow to reduce swirl. Following the honeycomb are three screens apart followed by a 150 mm and a further 600 mm before the start of the contraction which assists in eective ow conditioning to reduce turbulence levels. The wind tunnel contraction had an area ratio of 6:1 and employs a smooth contraction prole based on third order polynomial curve to enhance ow uniformity. The test section consisted of a rectangular nozzle with a 12 mm radial contraction prole on its long sides (Fig. 1a). This was mounted to the end of the tunnel contraction

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx 3

Fig. 1. Schematic view of the experimental setup: (a) the wind tunnel details and nozzle attachment; (b) the side view showing nozzle parameters and the case with sidewalls; (c) jet development characteristics; and (d) the experimental apparatus. Note that diagrams drawn are not to scale.

using tight rubber seals at the edges of contact to avoid leakage. Where present, the sidewalls were mounted ushed to the slots short sides and aligned in the xy plane (Fig. 1b). The sidewalls extended 2000 mm downstream and 1800 mm vertically, and were secured tightly by bolts to the laboratory ceiling to avoid vibrations during the tunnel operation. The slot-width (h), aligned with the lateral direction (y-coordinate) of the nozzle, was xed at 5.60 mm. The slot-span (w), aligned along the z-direction, was 340 mm for the large aspect ratio, AR  w/h = 60,

which was tested with and without sidewalls. The smaller nozzle had a 36-mm radial contraction prole and 5 mm slot-height but half the span, so that w = 170 mm and the sidewalls were moved inwards so that AR = 30. This was only tested without sidewalls so that the eect of slot aspect ratio on statistical two-dimensionality of the jet could be assessed. The normalized radius of curvature of r/ h % 2.14, and 7.20 conforms to our previous nding that a nozzles mean exit ow closely approximates top-hat prole for r/h > 2.0 [8,17].

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
4 R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx

The entire jet facility was located in a laboratory of dimensions 18 m (long) 7 m (wide) 2.5 m (high), and was acoustically isolated from externally driven noise. The distance from the jet exit to the front wall of the room was approximately 1400h and that between the jet and the ceiling/oor was approximately 125h, allowing these isothermal jets to entrain stagnant air freely. Based on a similar the approach to that of Hussein et al. [18], the present momentum loss is <0.5% at 160h downstream. This conrms that present jets resemble those in an unconned innite space. The facility was operated at a jet discharge velocity of Uo,c = 18.4 m s1 for AR = 60 (with and without sidewalls) and 26.3 m s1 for AR = 30 (without sidewalls), resulting in Reynolds numbers, Reh = Uo,ch/ t % 7000 and 10,000, respectively. Here t is the kinematic viscosity of air at the test conditions. Since it is known that nozzle aspect ratio inuences the transition regions of jet ows [10,17], the nozzle of smaller aspect ratio (AR = 30) is primarily used to characterize the decay of the jet without sidewalls. This cannot be assessed with the present large aspect ratio nozzle (see Fig. 3) since these regions occur too far downstream for reliable measurement. However, detailed systematic comparisons of the other ow parameters from both the jet with and without sidewalls are undertaken using the AR = 60 nozzle. Velocity measurements were performed over the axial (x) range from 0 6 x/h 6 160 using a hot-wire anemometer under isothermal conditions of ambient temperature of approximately 20 C. A specially designed three-dimensional traversing mechanism, driven by stepper motors, was employed to achieve precise probe traversing for measurements along the axial (x) and lateral (y) directions (Fig. 1d). The custom-built hot-wire (tungsten) sensor used was %5 lm in diameter and %0.8 mm long, aligned parallel to the long side of the nozzle. To minimize aerodynamic probe interference on the sensor, the probe was carefully mounted perpendicular to the ow with prongs parallel to the jet. Ambient conditions in the laboratory were isothermal. The overheat ratio of the hot wire was kept xed at 1.5. To optimize the bandwidth of the sensor and/or anemometer, and to check its operation, a square wave test was performed, revealing a sensor frequency response of 15 kHz. Calibrations of the hot wire were conducted using a standard Pitot static tube connected to a digital manometer, located side by side with the hot-wire probe at the exit plane (where hu2i1/2/Uc % 0.5%) prior to and after each measurement. Both calibration functions were tested for discrepancies, and the experiment was repeated if velocity drift exceeded 0.5%. Data points were converted from voltages to velocities using a fourth order polynomial curve, similar to the one proposed by George et al. [19]. The average accuracy of each calibration function was found to be 0.2%. Velocity signals obtained were low-pass ltered with a high and identical cut-o frequency of fc = 9.2 kHz at all the measured locations to remove excessive high-frequency noise. The voltage signals were oset to within 0 3 V and amplied and then digitized on a personal com-

puter at fs = 18.4 kHz via a 16 channel, 12-bit PC-30F A/ D converter (Fig. 1d) of signal input range 05 V as a precautionary measure to avoid signal clipping (Tan-Atichat et al. [20]). The sampling period was approximately 22 s, during which 400,000 instantaneous data points were gathered. Using inaccuracies in calibration and observed scatter in present measurements, the uncertainties correspond to a mean error of 4% at the outer edge of the jet and 0.8% on the centerline. The errors in the centerline root-mean-square (rms) velocities were found to be 1.8% and in the skewness (Su) and atness (Fu) factors of 3% and 8%, respectively. The errors in the momentum integral quantities are 8%. 3. Results and discussion The exit ows of the nozzle of AR = 60 were characterized by measuring the velocity proles in each isothermal jet with and without sidewalls at x/h % 0.25 along the lateral (y) direction over the range 0.70 6y/h 6 0.70. These are presented in Fig. 2a and b. The measurements conducted along the lateral (y)-direction at the center-plane of both nozzles, i.e. at (x/h, z/h)% (0.14, 0) conrm that these nozzles, like conventional smoothly-contracting ones, produce an approximately top-hat mean velocity prole, which are uniform (U/Uo,c % 1) for both cases within 5% (Fig. 2a). However, there are subtle dierences between the two exit velocity proles, in particular, the prole measured with sidewalls being slightly atter than that measured without sidewalls. Likewise, although the exit centerline turbulence intensity of both jets is about 0.5%, turbulence intensity within jet shear layers vary from about 3.6% for the jet without sidewalls to about 2.8% for the jet with sidewalls (Fig. 2b). This variation is signicant, and indicates that even in the very near eld; sidewalls produce a signicant eect on the exit ow. However, despite an 8%
0.06 1.0 0.8
U/Uo,c

without sidewalls with sidewalls

0.05
/ U o,c (<u >)
2 1/2

0.04 0.6 0.4 0.2 0 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 y/h 0.03 0.02 0.01 0

Fig. 2. Lateral proles of the mean velocity, U/Uo,c (denoted by symbol O, h), and turbulence intensity, u 0 /Uo,c (denoted by symbol j, ) measured at x/h = 0.25 for AR = 60, Reh = 7000.

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx 5

uncertainty in the momentum integral, the dierences are signicant enough to demonstrate that the sidewalls, and probably also the shape of the contraction, do inuence the initial shear layer intensities. We characterize the jet exit conditions by estimating the boundary-layer displacement (dd) and momentum thickR h=2 ness (hm) via the R integral equations, dd 0 1 h=2 U =U o;c dy and hm 0 1 U =U o;c U c =U o;c dy using the mean exit velocity proles. These estimates are somewhat crude due to the coarse measurement grid. In addition, the momentum integral does not take into account the slight overshoot of the mean velocity for the case without sidewalls. These contribute to an estimated uncertainty in integral quantities of up to 8%. For both cases, the present values of dd and hm are about 0.15h and 0.059h. Likewise, the corresponding shape factors, H = (dd)/(hm), which represents the atness (uniformity) of the mean velocity proles [21], are %2.55 compared with a value of %2.60 for a Blasius exit velocity prole. Given the closeness of the shape factors to that of an ideally uniform mean exit velocity prole, both jets are characterized as having laminar boundary layers at their nozzle exit [22]. Fig. 3 presents, with a logarithmic abscissa, the normalized mean centerline velocity, Uo,c/Uc for jets without sidewalls measured at (Reh, AR) = (7000, 60) and (Reh, AR) = (10,000, 30) and the jet with sidewalls measured at (Reh, AR) = (7000, 60) (see Table 1). The jets are qualitatively similar, but quantitatively dierent in the rst 40
10 with sidewalls; AR = 60 without sidewalls; AR = 60 without sidewalls; AR = 30 5 Quinn (1992); without sidewalls Quinn [7]; AR = 20
rectangular; without sidewalls

Uc~ x

-1

1 1

Uc Uo,c

Uc~ x

-1/2

10

x/h

100

300

Fig. 3. The normalized centerline mean velocity Uo,c/Uc for AR = 30, Reh = 10,000 and AR = 60, Reh = 7000. Also shown are the rectangular jet data of Quinn [7] for Reh = 36,000 and AR = 20.

slot-heights downstream and then diverge signicantly from each other into the far eld. Both jets exhibit the well-known non-decaying potential core region, i.e. Uc % Uo,c, followed by a 1/2-power law inverse decaying region, where Uc $ x1/2. Further downstream, the jet without sidewalls exhibits a transition region, where Uc does not appear to obey any systematic dependence on x, to an inversely-linear decaying region, where Uc $ x1. In contrast, the inverse square decaying region extends throughout the measured range for the jet with sidewalls. (However, it should be noted that, further downstream (i.e. x/h ) 160), the jet with sidewalls, too, will enter a transition phase; see Deo et al.[10]). The axial extent of the statistically two-dimensional region depends on nozzle aspect ratio. Fig. 3 also plots the decay of the AR = 30-jet without sidewalls, which exhibits statistically two-dimensional behaviour only up to x/h % 30. This dependence of the axial extent of statistical two-dimensionality on nozzle aspect ratio for both rectangular (without sidewalls) and planar (with sidewall) jets has already been demonstrated [8,10]. However, the present work shows that, without sidewalls, a jet has a longer region of statistically two-dimensional ow only if the nozzle aspect ratio is suciently large. Importantly, the mean eld implies that without sidewalls, the jet appears to transform into an axisymmetric form, characterized by Uc $ x1, in the far eld. Our ndings are consistent with those of Trentacoste and Sforza [5] and Ser [6] whose mean velocity and scalar decay of rectangular jets (without sidewalls) behaved statistically like axisymmetric jets in the far eld. Likewise, the rectangular jet data of Quinn [7], reproduced in Fig. 3, shows these four regions in the decay of Uc. It is important to note that the logarithmic scale in the abscissa is essential to identify the points of transition from statistically two-dimensional to three-dimensional mean ow. Previous investigators, e.g. Quinn [7], Tsuchiya et al. [23] did not use this approach and therefore, did not specically identify this transition region. The length of the jets potential core, xp (the maximum axial (x) distance up to which Uc % Uo,c) is dierent for the jets with and without sidewalls. For the case without sidewalls, xp % 4h, while for that with sidewalls, xp % 6h. This dierence, which cannot be attributable to an experimental error, implies a signicantly higher rate of near eld entrainment by the jet without sidewalls. The inuence of the sidewalls on the spread in the spanwise direction cannot explain such a dramatic dierence in xp on the nozzle axis for this high-aspect ratio jet. Rather it is important to note that the initial shear layer of the jet without sidewalls has

Table 1 The measured jet congurations and initial conditions for the cases with and without sidewalls. Conguration With sidewalls Without sidewalls Without sidewalls Reh 7000 7000 10,000 AR 60 60 30 dd 0.15h 0.15h hm 0.059h 0.059h H = dd/hm 2.55 2.55 u0c =U c 0.5 0.5 0.5 u0p =U c 2.6 3.6 3.6

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

U o,c/ U c

ARTICLE IN PRESS
6 R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx

higher turbulence intensity than that with sidewalls (Fig. 2b). Since the length of the potential core is dominated by the growth of the shear layer in the xy plane [24], the higher initial turbulence intensity of the jet without sidewalls is consistent with its shorter potential core. Next, we assess whether the higher shear exit layer turbulence intensity (Fig. 2) and a consistently shorter potential core (Fig. 3) might be caused by the jet without sidewalls undergoing larger amplitude near eld apping than the jet with sidewalls. Were such a apping to be present, it would be evident in a low-frequency oscillation in the power spectrum. To determine whether this was present, we have calculated the power spectra of centerline velocity uctuations at the beginning and the end of the potential core (i.e. x/h = 3, 4 and 5) of the jet with and without sidewalls by ltered instantaneous velocity uctuation data at fc = 250 Hz. The spectra (Appendix 1) show that there exist no genuine signicant dierence in the low-frequency spectra, and no signicant peaks for either case (A small peak at 50 Hz was found in each case, which is attributed to electronic noise). Hence it is deduced that the higher shear layer intensity is not attributable to such apping motions, but rather to increased three-dimensionality of the jet without sidewalls, due to dierences in their underlying vortex motions (as discussed in detail below). Although the measurements of other turbulence parameters (e.g. shear stresses) could conrm this, it was beyond the scope of present investigation since data here are limited to single component velocity measurements. To investigate the near eld ow structure of the two AR = 60 jets more closely, we have plotted the power spectra of the centerline velocity uctuations within the potential core region (i.e. at x/h = 3) shown in Fig. 4. Here, the dominant peak is normalized so that the vortex shedding
-2

10

f* 0.22 f* 0.36 10
-3

10

-4

10

-5

with sidewalls without sidewalls


0 0.5 1.0 1.5

f* = f h / Uo,c
Fig. 4. Power spectra, Uu(f*), of the centerline velocity uctuations measured at x/h = 3 for the jets of AR = 60, Reh = 7000.

R frequency, f*  fh/Uo,c, and Uu f df 1. The appearance of broad peaks in power spectra conrms that both nozzles generate regularly-occurring primary vortices, a feature which is well known to occur in other smoothly contoured nozzles as well. The smoke visualization experiment of Tsuchiya et al. [23] on rectangular jets revealed streamwise vortex structures between 04 nozzle-widths downstream, as did the planar jet ow visualization of Shlien and Hussain [27]. However, it is particularly interesting to note that the present jet without sidewalls sheds vortices at f* = 0.36 whereas, for the jet with sidewalls, vortex shedding occurs at f* = 0.22. This signicant dierence in f* conrms distinct dierences in their underlying near eld ow structure. It is hypothesized that the dierences in spectral peaks can be attributed to structural dierences in vortices (e.g. a higher degree of three-dimensionality) in the near eld ow of the jet without sidewalls [26]. It is known that a rectangular jet, congured without sidewalls, is known to produce ring-like three-dimensional primary vortices (Grinstein [28], Grinstein et al. [29]) while the 2-D rollerlike vortices are found in a planar jet, congured with sidewalls (Browne et al. [30]). The dierences in the vortex dynamics of both jets are described below. The primary ring-like vortices produced by a jet without sidewalls (e.g. a rectangular jet) are probably initially interconnected in a loop that stretches around the circumference of the jet [25]. These vortex loops undergo substantial twisting and turning at the corners, and cause more rapid spread along the shorter side than the longer side as the jet spreads in the spanwise (z) direction. The non-uniform growth rates of the long and short sides causes increased distortion of the ring, leading to axis switching and the rings eventual destruction. Evidence of this behaviour can be found from the ow visualization experiments of Trentacoste and Sforza [5]. They studied decay rates of primary vortex rings in rectangular jets, noting that the vortex rings rstly issue as an ellipse, whose major is axis aligned with the major axis of the nozzle. This then proceeds downstream and become circular and then, still further downstream become elliptical again, but with its major axis rotated 90 to that of the nozzle slot. Typically, multiple axis switching is observed in rectangular jets which are not constrained by sidewalls in the xy plane (Piaut [31]). Another useful description of the ring-like vortices in rectangular jets can be found in the ow visualization obtained by DNS of Grinstein et al. [28]. They found, in addition to the primary ring-like vortices, secondary (hairpin or braid) vortices extending longitudinally. Importantly, their ow topology revealed these large-scale spanwise vortex ring and streamwise braid vortices to have strong inter-scale interactions, which increases threedimensionality and the degree of disorganization. In contrast, the jet with sidewalls produces counterrotating streamwise vortices within the near nozzle region that are more two-dimensional than the vortices produced by a jet without sidewalls. The early ow visualization of

u( f*)

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx 7

Brown [32] and of Beavers and Wilson [33] revealed symmetrical (cylindrical) vortices on alternate sides of the potential core. These initial vortices grew into larger vortices by coalescence with their adjacent counterparts as the jet propagates downstream. The process of coalescence typically depends on the exit conditions. For instance, Sato [34] showed that acoustic excitation at a frequency close to the natural vortex shedding frequency, causes these vortices to coalesce closer to the nozzle exit. Importantly, unlike jets without sidewalls, the counter-rotating vortices in jets with sidewalls are mostly two-dimensional (double roller) in nature. Thus increased three-dimensionality of the rectangular jet over the planar jet is likely to be signicant in its higher near eld entrainment implied by the shorter potential core (Fig. 3). The structural dierences in the vortices of jets with and without sidewalls are also evident in the power spectrum (Fig. 4). Not only do their dominant vortex shedding frequencies dier but it is also apparent that the spectrum of the jet without sidewalls exhibits a multimodal behaviour with four distinct sub-peaks (harmonics) in f*. These are evidence of multiple length scales in the ow and hence three-dimensional motions. Fig. 5 investigates the decaying behaviour of the mean velocity eld by presenting U o;c =U c 2 versus x/h within the statistically two-dimensional region of the jets with and without sidewalls. The self-similar relationship of the form (Uo,c/Uc)2 = Ku(x/h + x01/h) holds true over the axial range 10 6 x/h 6 40. Here, Ku and x01 are the jet decay rate and virtual origin, respectively. Both jets have distinct differences in velocity decay rates, presented for clarity in the insert of Fig. 5. In particular, the jet without sidewalls (Ku % 0.20) decays at a rate of some 15% higher than does the jet with sidewalls (Ku = 0.17). This implies that the far eld entrainment rate of the jet with sidewalls is lower than the jet without sidewalls. To assess the spreading behaviour of both jets, Fig. 6 presents the axial evolution of velocity half-widths, y0.5 in
30
8

20 open symbol (circle): y0.5 closed symbol (circle): z0.5 16

yo.5 /h; z0.5 /h

12

8 Quinn (1992); AR = 20 Quinn [7]; AR = 20 rectangular, without sidewalls rectangular; without sidewalls 4 with sidewalls; AR = 60 without sidewalls; AR = 60 0 0 20 40 60 80 100 120 140 160 180 200 x/h

Fig. 6. The mean velocity half-width for the jets of AR = 60, Reh = 7000.

25 20
(Uo,c / Uc )
2

4 0 0

15 10 5 0 0 20

10 20 30 40 50 x/h

with sidewalls; AR = 60 without sidewalls; AR = 60


40 60 80
x/h

100

120

140

160

180

Fig. 5. The proles of (Uo,c/Uc)2 versus x/h for jets of AR = 60, Reh = 7000.

the lateral (y) direction. The half-width is the location at U x; y 0:5 1 U c x obtained from the mean velocity pro2 les (Fig. 1c). Also presented the velocity half-widths of Quinn [7] measured along both lateral (y) and spanwise (z) directions. The values of y0.5 vary linearly with x, and follow the relationship y0.5/h = Ky(x/h + x02/h), where Ky is a measure of the spreading rate and x02 is the virtual origin of that jet spread. In the self-similar region, y0.5 $ x, further conrming statistical two-dimensionality. The jet without sidewalls spreads at a rate of about 24% higher (Ky % 0.14) than does the jet with sidewalls (Ky % 0.11). Importantly, Figs. 5 and 6 are internally consistent, since both nd the decay and spreading rates of the jet without sidewalls to be higher than that with sidewalls. They are also consistent with a shorter potential core of the jet without sidewalls, which again suggests it has a higher entrainment rate. A jet which is not constrained by sidewalls in the xy plane spreads along the spanwise (z) direction of the nozzle as well. From Quinns [7] data, we can see that, generally, for jet without sidewalls, y0.5 grows initially at a substantially greater rate than does z0.5. This is probably due to stretching of the vortex rings as discussed earlier. The growth of y0.5 is approximately linear until it crosses over z0.5 where the so called axis switch occurs. It is apparent that z0.5 decreases slightly, and then continues to increase at a similar rate to that of y0.5. Further downstream, y0.5 enters a transition phase, where it decreases in magnitude, while z0.5 continues to increase linearly. In the region x/ h > 130, both spreading rates (i.e. z0.5 and y0.5) converge into each other. This provides further evidence of axisswitching phenomenon of the unconned jet. Fig. 7 presents the local centerline turbulence intensity, 1=2 u0c =U c of both jets. The term u0c hu2 i where ucis the cenc terline velocity uctuation, and Uc is the mean centerline velocity. Also included in the gure are the u0c =U c values of Quinn [7] measured at (Reh, AR) = (36,700, 20) for a

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

(U o,c/U c)

ARTICLE IN PRESS
8 R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx

0.25

Mi etet al. [35] Mi al. (2000) round jet round jet

0.20

0.15

u'c / Uc

Antonia et al. (1982);AR = 44 Antonia et al. [34]; AR = 44 with sidewalls planar; with sidewalls
Quinn [7]; AR = 20 Quinn (1992); AR = 20 rectangular; without sidewalls without sidewalls

0.10

0.05
without sidewalls; AR = 60 with sidewalls; AR = 60

0 0 20 40 60 80 100 120 140 160 180

vortex structures in the jet with sidewalls. This analogy is drawn from published information. For example, a smoothly contoured round nozzle produces analogous coherent vortex structures and also exhibits a near eld hump, while a pipe jet with much less coherent vortices does not (Mi et al. [40]). As such, the magnitude of the humps is expected to reect the strength of the overall interaction between coherent vortical structures and the induced ambient uid (Mi et al. [41]). The turbulence intensity of the jet with sidewalls reaches an asymptotic value of about 0.21, which is very close to that of a round smooth contraction jet. In contrast, for the jet without sidewalls, u0c =U c decreases signicantly to values of about 75% of those with sidewalls over the range 30 6 x/h 6 100. Note that the small variations of u0c =U c for the case with sidewalls are within the experimental uncer-

x/h
Fig. 7. The normalized centerline turbulence intensity u0c =U c of AR = 60, Reh = 7000.

5 4

rectangular jet (without sidewalls), Antonia et al. [35] for a planar jet (with sidewalls) measured at (Reh, AR) = (7000, 44) and Mi et al. [36] for a free circular jet. The overall shape of graph of the data of Quinn [7] is similar to that of our jet without sidewalls. The qualitative features in the data of Quinn [7] are similar to that of our jet without sidewalls. This is despite the fact that the magnitudes and axial locations of major features (e.g. humps) are dierent. The dierences reect the dierent initial conditions between our jet without sidewalls and that of Quinn [7] (i.e. nozzle aspect ratio, AR = 60 versus 20, Reynolds number, Reh = 7000 versus 36,000, nozzle contraction prole, radially contoured versus sharp-edged orice plate). All jets exhibit a rapid increase of u0c =U c in the near eld, reecting the streamwise growth of the shear-layer instabilities associated with the large-scale structures [35,37]. Such large-scale structures produce large-scale engulfment of ambient uid, resulting in higher velocity uctuations and greater decay of jets mean velocity. For the present data, there is negligible dierence in the turbulence intensity of both jets between the exit plane and x/h = 10. Although, both jets produce a hump in u0c =U c between 10 6 x/ h 6 12, the hump for the jet without sidewalls is discernibly higher than the jet with sidewalls. It also appears that for both jets, this hump is about 10% higher than their respective asymptotic values. The occurrences of humps in turbulence intensity have been attributed to the laminar states of the boundary layer at the nozzle exit (e.g. [38] and Fig. 2a). The appearance of the humps in both jets is attributable to the collision of double roller structures, emanating from alternate sides of the shear layers beyond the potential core region, e.g. Mumford [39], Browne et al. [30]. The dierence in the magnitude of the humps suggests that the presence of sidewalls increases the coherence of the shear layer

Fu= <u >/(<u >)


3 2 1

S = <u >/(<u >)


u

3/2

0 -1 0 2 4 6 8 x/h 10 12 14 16

5 4 Fu= <u >/(<u >) 3 2 without sidewalls with sidewalls 1 0 -1 0 20 40 60 80 x/h


Fig. 8. Streamwise evolutions of the skewness, Su, and atness, Fu, for the jets of AR = 60, Reh = 7000, measured in: (a) the near eld and (b) the entire eld.
4 2 2

Su= <u >/(<u >)

3/2

100

120

140

160

180

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx 9

tainty. The convergence of the turbulence intensity of present jet with sidewalls to an asymptotic value is consistent with the planar jet of Antonia et al. [35] of (Reh, AR) = (7000, 44) issuing from a smoothly contoured nozzle although their asymptotic value is somewhat 10% lower. In contrast, the far eld values of u0c =U c of the present jet without sidewalls are only about three-quarters of the magnitude of the jet with sidewalls. To further show the dierence between the jets with and without sidewalls, Fig. 8 displays the centerline evolutions of the skewness, Su = hu3i/(hu2i)3/2 and atness, Fu = hu4i/ (hu2i)2 factors of AR = 60-jets. Here, Su and Fu were determined from a reasonably large sample, having approximately 400,000 data points of the instantaneous velocity. This ensures that the convergence of the calculations is good. In the near eld (0 6 x/h 6 20), the location of the peaks in Fu and Su are further downstream for the jet without sidewalls, consistent with it having a shorter potential core. For each jet, Su displays a local maximum at x/ h % 4 (without sidewalls) and x/h % 4 (with sidewalls) and a local minimum at x/h = 5 (without sidewalls) and x/ h = 7 (with sidewalls). The minimum value of Su occurs at almost the same location as the maximum in Fu for each jet. Such evolutions of Su and Fu are typical of two-dimensional jets, and have been reported previously (e.g. Deo et al. [10], Deo [17]). Within the region where the two jets both exhibit statistically two-dimensional behaviour in the mean eld (10 6 x/h 6 40), both Fu and Su are signicantly dierent

for the jets with and without sidewalls. For example, at x/h = 30, their respective values are 2.7 and 2.9 for Fu and 0.02 and 0.34 for Su for the jets with and without sidewalls, respectively. Hence, even in this region, the two jets behave signicantly dierently. Further downstream, the Su and Fu factors of the jet with sidewalls remains approximately constant, at values close to the Gaussian, as has been reported previously. In contrast, those for the jet without sidewalls are never truly Gaussian anywhere in the measured range, although they are closest to it in the range 30 6 x/h 6 110. For x/ h > 120, Su and Fu depart signicantly from the Gaussian values, consistent with jet having not yet completed to its transition to the asymmetric ow (Fig. 3). The characterization of the ow dynamics of jet with dierent boundary conditions falls short without attributing the role and dependence of large-scale coherent structures on jet boundary conditions. Since this study has used a single wire probe and thus does not enable us to quantify the dominance/relevance of inter-scale interactions of coherent structures in the two jet ows directly, present statistics do indicate the inherent role played by these structures in producing widely dierent near and far eld states. The statistical dierences in present results support the deduction that sidewalls impose a modication of the jet turbulence structure, and therefore, warrant further in-depth investigation (e.g. a ow visualization or twopoint cross correlation analysis) on the role of large-scale eddies on dierent exit boundaries. Although this could

Fig. 9. Side and top views of the typical ow eld of: (a) the jet without sidewalls; and (b) the jet with sidewalls.

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
10 R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx

be taken as a very useful exercise, it was beyond the scope of the present study, and awaits another independent investigation. 4. Conclusions In summary, the ow elds of jets with and without sidewalls from rectangular nozzles are consistently statistically dierent. Indeed, perhaps surprisingly, even the exit velocity proles very close to the exit (x = 0.25h) are also changed slightly by the presence or absence of sidewalls. The near eld ow also reveals clear distinctions in the length of the potential core being greater for the case with sidewalls which implies a lower rate of near eld entrainment. The near eld vortex shedding rate of the jet with sidewalls has both a lower dominant shedding frequency and higher frequency peaks. This, together with low-frequency spectra, implies that a jet with sidewalls has a less three-dimensional near eld structure. These dierences in the near eld ow structure are possibly to be caused by dierent vortex types (ring-like in jet without sidewalls and roller-like in jet with sidewalls). In common, both jets have characteristic regions of two-dimensional (2-D) mean velocity decay, which follow the scaling Uc $ x1/2 in their respective self-similar elds. However, the decaying and spreading rates are lower for the case with sidewalls than the jet without sidewalls. This is also consistent with a more three-dimensional underlying structure for the case without sidewalls. Further downstream following the 2-D region, the mean velocity decay of the jet without sidewalls follows the scale Uc $ x1 after an apparent transition zone. This provides further evidence that a jet unbounded by sidewalls follow-

ing axis switching, undergoes a transition to axisymmetric ow. Other jet properties, e.g. turbulence intensity, skewness and atness factors of the velocity uctuations are also dierent, conrming that ows from the two nozzle congurations are statistically dierent. It must, therefore, be recognized that, although rectangular nozzles without sidewalls of suciently large aspect ratio generate statistically two-dimensional mean velocity elds, their other ow development characteristics do not resemble those of planar nozzles which are congured with sidewalls. This is primarily because the presence of sidewalls has a profound eect on both the near and far eld ow. This, together with low-frequency spectra, implies that a jet with sidewalls has a less three-dimensional near eld structure. The far eld of ows from rectangular nozzles without sidewalls exhibit more three-dimensional underlying large-scale coherent motions than those from rectangular nozzles with sidewalls, which produce more two-dimensional coherent structures (see Fig. 9). Acknowledgement This paper reports results from the rst authors PhD research completed through the support of Endeavor International Postgraduate Scholarship, Adelaide Achievers Scholarship and an ARC Discovery Grant in collaboration with FCT. We also thank the two anonymous reviewers whose comments have improved the overall clarity of our paper. Appendix 1. See Fig. A1.

10-1

with sidewalls
10

-1

without sidewalls

x/h=5 x/h=4 x/h=3

10-3
u (f)

10

-3

10-5

10

-5

10-7

10

100 f (Hz)

1000

10000

10

-7

10

100

1000

10000

Fig. A1. Power spectra of velocity uctuations at x/h = 3, 4 and 5 for jets with and without sidewalls.

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

ARTICLE IN PRESS
R.C. Deo et al. / Experimental Thermal and Fluid Science xxx (2007) xxxxxx 11

References
[1] G. Heskestad, Hot-wire measurements in a plane turbulent jet, Trans. ASME, J. Appl. Mech. 32 (1965) 721734. [2] L.J.S. Bradbury, The structure of a self-preserving turbulent planar jet, J. Fluid Mech. 23 (1965) 3164. [3] E. Gutmark, I. Wygnanski, The planar turbulent jet, J. Fluid Mech. 73 (3) (1976) 465495. [4] P.M. Sforza, M.H. Steiger, N. Trentacoste, Studies on three-dimensional viscous jet, AIAA J. 4 (5) (1996) 800806. [5] N. Trentacoste, P. Sforza, Further experimental results for threedimensional free jets, AIAA J. 5 (5) (1967) 885890. [6] A. Sfeir, Investigation of three-dimensional turbulent rectangular jets, AIAA J. 17 (10) (1979) 10551060. [7] W.R. Quinn, Turbulent free jet ows issuing from sharp-edged rectangular slots: the inuence of slot aspect ratio, Exp. Therm. Fluid Sci. 5 (1992) 203215. [8] J. Mi, R.C. Deo, G.J. Nathan, Characterization of turbulent jets from high-aspect ratio rectangular nozzles, Phys. Fluids 17 (6) (2005) 68102. [9] J. Bashir, S.M. Uberoi, Experiments on turbulent structure and heat transfer in a 2-D jet, Phys. Fluids 18 (4) (1975) 405410. [10] R.C. Deo, G.J. Nathan, J. Mi, The inuence of nozzle aspect ratio on plane jets, Exp. Therm. Fluid Sci. (2006). <http://dx.doi.org/10.1016/ j.expthermusci.2006.08.009>. [11] A. Krothapalli, D. Bagano, K. Karamcheti, On the mixing of rectangular jet, J. Fluid Mech. 107 (1981) 201220. [12] A.K.M.F. Hussain, A.R. Clark, Upstream inuence on the near eld of a planar turbulent jet, Phys. Fluids 20 (9) (1977). [13] I. Namar, M.V. Otugen, Velocity measurements in a planar turbulent air jet at moderate Reynolds numbers, Exp. Fluids 6 (1988) 387399. [14] G.J. Hitchman, A.B. Strong, P.R. Slawson, G. Ray, Turbulent planar jet with and without conning walls, AIAA J. 28 (10) (1990) 1699 1700. [15] W.K. George, The self-preservation of turbulent ows and its relation to initial conditions, in: Recent Advances in Turbulence, Hemisphere, New York, 1989, pp. 3973. [16] W.K. George, L. Davidson, Role of initial conditions in establishing asymptotic behavior, AIAA J. 42 (3) (2004) 438446. [17] R.C. Deo, Experimental investigations of the inuence of Reynolds number and boundary conditions on a plane air jet, Ph.D. thesis, School of Mechanical Engineering, The University of Adelaide, Australia (2005). ADT Program :<http://thesis.library.adelaide.edu.au/public/adt-SUA20051025. 054550/index.html>. [18] H.J. Hussein, S.P. Capps, W.K. George, Velocity measurements in a high Reynolds number, momentum-conserving axisymmetric turbulent jet, J. Fluid Mech. 258 (1994) 3160. [19] W.K. George, P.D. Beuther, A. Shabbir, Polynomial calibrations for hot wires in thermally varying ows, Exp. Therm. Fluid Sci. 2 (1989) 230235. [20] (Sec. 15.15) J. Tan-Atichat, W.K. George, S. Woodward, Use of computer for data acquisition and processing, in: A. Fuhs (Ed.), Handbook of Fluids and Fluids Engineering, Wiley, NY, 1996, pp. 10981116. [21] S.B. Pope, Turbulent Flows, Cambridge University Press, UK, 2002.

[22] H. Schlichting, Laminare strahlausbreitung, Z. Angew. Math. Mech. 13 (1933) 260263. [23] Y. Tsuchiya, C. Horikoshi, T. Sato, M. Takahashi, A study of the spread of rectangular jets: (The mixing layer near the jet exit and visualization by the dye methods), JSME Int. J. 32 (Series II (1)) (1989) 1117. [24] R.A. Antonia, L.W.B. Browne, S. Rajagopalan, A.J. Chambers, On organized motion of a turbulent planar jet, J. Fluid Mech. 134 (1983) 4966. [25] J.C. de Gortari, V.W. Goldschmidt, The apparent apping motion of a turbulent plane jet further experimental results, in: ASME Winter Annual Meeting, November 1978, Chicago, Ill, pp. 1621. [26] A. Sau, Generation of streamwise vortices in square sudden-expansion ows, Phys. Rev. E 69 (2004) 056307. [27] D.J. Shlien, A.K.M.F. Hussain, Visualization of large-scale motions of a plane Jet, ow visualization III, in: Proceedings of the Third International Symposium, September 69, 1985, Ann Arbor, MI, Washington, DC, pp. 498502. [28] F.F. Grinstein, Dynamics of turbulent free jets, Internet Resource, Last accessed 3rd May 2007. <http://www.uottawa.ca/publications/ interscientia/inter.2/jet.html>. [29] Chapter III F.F. Grinstein, M.N. Glauser, W.K. George, Vorticy in jets, in: Fluid Vorticity, Kluwer Academic Publisher, Netherlands, 1995, pp. 6594. [30] L.W.B. Browne, R.A. Antonia, S. Rajagopalan, A.J. Chambers, Interaction region of a two-dimensional turbulent plane jet in still air, in: Structure of Complex Turbulent Shear Flows, IUTAM Symposium, 1982, Marseille, pp. 411419. [31] V. Piaut, Axis-switching in square coaxial jets, Masters thesis, Mechanical Engineering, Louisiana State University, USA, 2003. [32] G.B. Brown, On vortex motion in gaseous jets and origin of their sensitivity in sound, P. Phys. Soc. Lond. 47 (1935) 703732. [33] G.S. Beavers, T.A. Wilson, Vortex growth in jets, J. Fluid Mech. 44 (1) (1970) 97112. [34] H. Sato, The stability and transition of a two-dimensional jet, J. Fluid Mech. 7 (1960) 5380. [35] R.A. Antonia, B.R. Satyaprakash, A.K.M.F. Hussain, Measurement of dissipation rate and some other characteristics of turbulent plane and circular jets, Phys. Fluids 23 (4) (1980) 695700. [36] J. Mi, G.J. Nathan, R.E. Luxton, Centerline mixing characteristics of jets from nine dierently shaped nozzles, Exp. Fluids 28 (2000) 93 94. [37] S.V. Gordeyev and F.O. Thomas, Visualization of the topology of the large-scale structure in the planar turbulent jet, in: Proceedings of the 9th International Symposium on Flow Visualization, 2000, pp. 13. [38] P. Bradshaw, The eect of initial conditions on the development of a free shear layer, J. Fluid Mech. 26 (1966) 225236. [39] J.C. Mumford, The structures of large eddies in fully developed shear ows. Part 1. The plane jet, J. Fluid Mech. 118 (1982) 241268. [40] J. Mi, D. Nobes, G.J. Nathan, Inuence of exit conditions of round nozzles on the passive scalar eld of a free jet, J. Fluid Mech. 432 (2001) 91125. [41] J. Mi, G.J. Nathan, D.S. Nobes, Mixing characteristics of axisymmetric free jets from a contoured nozzle, an orice plate and a pipe, J. Fluids Eng. 123 (2001) 878883.

Please cite this article in press as: R.C. Deo et al., Comparison of turbulent jets issuing from rectangular nozzles ..., Exp. Therm. Fluid Sci. (2007), doi:10.1016/j.expthermusci.2007.06.009

You might also like