Experimental Thermal and Fluid Science 61 (2015) 249258

Contents lists available at ScienceDirect

Experimental Thermal and Fluid Science

journal homepage: www.elsevier.com/locate/etfs

Reynolds number effects in the near-eld of a turbulent square jet

A. Ghasemi a, V. Roussinova b,, Ram Balachandar a,b, R.M. Barron a,c
a

Department of Mechanical, Automotive and Materials Engineering, University of Windsor, Windsor, ON N9B3P4, Canada
Department of Civil and Environmental Engineering, University of Windsor, Windsor, ON N9B3P4, Canada
c
Department of Mathematics and Statistics, University of Windsor, Windsor, ON N9B3P4, Canada
b

a r t i c l e

i n f o

Article history:
Received 23 May 2014
Received in revised form 22 September
2014
Accepted 28 October 2014
Available online 6 November 2014
Keywords:
PIV
Turbulence
Square jet
Reynolds number
Shear stress
Vorticity

a b s t r a c t
High-resolution particle image velocimetry (PIV) was used to study the ow characteristics in the neareld of a turbulent square water jet issuing from a smooth contraction nozzle. Mean velocity and turbulence statistics are investigated over a range of jet Reynolds number ReD (based on jet exit velocity and
equivalent nozzle diameter) varying from 10,000 to 41,400. The inuence of ReD on the shear layer
formation in terms of momentum thickness, Taylor length scale and the evolution of the turbulent/
non-turbulent interface are also studied. The velocity measurements reveal that the jet in the near-eld
is dependent on Reynolds number at the lower end of the range (ReD = 10,000) despite the fact that all
exit jet proles closely approximate a top-hat shape. It is shown that the jet shear layer grows faster
at lower Reynolds number which, combined with an increase of the spanwise turbulence component (vrms)
along the jet centerline, suggest rapid axis switching. Much weaker Reynolds number dependence of the
mean velocity, turbulence intensities, momentum thickness and jet centerline anisotropy (urms/vrms) was
found for square jets at ReD > 104. The Taylor length scale calculated along the jet centerline decreases
with increasing Reynolds number and asymptotes to a constant value for X/D > 4. The turbulent/nonturbulent (T/NT) regions of the jet have been identied using the velocity criteria proposed in the
literature. Evolution of the T/NT interface and the effect of ReD on the conditionally averaged streamwise
velocity and vorticity are also investigated in the near-eld of the jet. No jump of the conditionally
averaged streamwise velocity and vorticity proles was noted in the near-eld of the square jet.
2014 Elsevier Inc. All rights reserved.

1. Introduction
Turbulent jets are employed in a variety of engineering
processes such as mixing, combustion, ventilation and propulsion.
Jets are an important topic of turbulence research as they represent
a specic class of free shear ows that involve a rich variety of
large and small scale phenomena. It is well known that the initial
condition of the circular jet can substantially inuence the ownstream ow characteristics [1,2]. In particular, shear layer instability and vortex generation occurring in the near-exit region play an
important role in controlling the mixing process of the circular jet
[3]. Due to the potential for enhanced mixing and entrainment, jets
emanating from non-circular nozzles and orices have been studied by many authors. Jets produced from elliptical nozzles as well
as those with sharp corners, such as square, rectangular, triangular
[4,5] and daisy-shaped [6] nozzles, have been shown to yield
enhanced mixing and entrainment [7] due to the occurrence of
axis-switching [8]. While there are an extensive number of studies
that have explored the effect of Reynolds number on the mean ow
Corresponding author.
http://dx.doi.org/10.1016/j.expthermusci.2014.10.025
0894-1777/ 2014 Elsevier Inc. All rights reserved.

and turbulence statistics in round jets, the Reynolds number effect

on square jets has received signicantly less attention.
The effect of Reynolds number on the evolution of ow structures in round jets has been investigated by Dimotakis [9], who
put forward the idea of turbulent mixing transition. Dimotakis
[9] proposed a critical value of Reynolds number of 10,000, beyond
which ow properties such as mixing become a weak function of
Reynolds number. The jet Reynolds number Re (=DUj/m) was
dened based on the diameter of the nozzle D, jet velocity at the
nozzle Uj, and kinematic viscosity m. Focusing on free shear layers
and jets, Dimotakis [9] suggested that this critical Reynolds number is valid as long as the ow length scale is properly established.
For shear layers, the appropriate length scale was proposed to be
the width of the local shear layer. For round jets, the suggested
characteristic length scale was the local jet diameter. For Reynolds
numbers below the critical value (Re < Remin), mixing is enhanced
with increasing Re as a result of an increase in the interfacial area.
For Re > Remin, a fully developed turbulence state is achieved and a
weak Reynolds number dependence is observed for all ow
parameters. Evidence related to Re independence can be observed
from the experiments by Ricou and Spalding [10]. They were

250

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

Nomenclature
A
D
k
LCCD
LIA
LPIV
Lv
ReD
Dt
u
u0
urms
U
Uc

proportionality constant ()
equivalent diameter (m)
turbulent kinetic energy (m2/s2)
dimensions of CCD array (pixels)
dimensions of interrogation area (pixels)
smallest scale resolved by PIV (mm)
dimension of viewing area (mm)
Reynolds number ()
temporal interval (s)
instantaneous streamwise velocity (m/s)
uctuating streamwise velocity (m/s)
streamwise turbulence intensity (m/s)
mean streamwise velocity (m/s)
local centerline velocity (m/s)

among the rst to report that mixing transition occurs at a critical

Reynolds number based on nozzle exit diameter (ReD) of 25,000,
which is of the same order as reported later by Dimotakis [9].
Tandalam et al. [11] studied the near-eld of round jets emanating
from a smooth contraction nozzle at three different Reynolds numbers (ReD = 10,000, 30,000 and 55,000) using the PIV technique.
They identied large-scale structures by applying proper
orthogonal decomposition (POD) on the instantaneous velocity
elds. Based on the POD analysis, a signicant effect of Reynolds
number on the near-eld average vortex circulation was reported.
Beyond ve diameters, the average circulation proles become
independent of Reynolds number and collapsed onto each other.
At higher Reynolds numbers, the vortices were found to appear
closer to the nozzle exit.
Fellouah et al. [12] investigated the near and intermediate eld
of a round jet using stationary and ying hot-wire measurements.
They found that very near the nozzle the streamwise velocity
proles were independent of Reynolds number. Conversely,
streamwise turbulence intensity in the shear layer was found to
increase with Reynolds number, and decreased in the central
region of the jet. For ReD = 6000, the mean velocity prole preserves the initial jet condition for a longer distance. In other words,
it takes a shorter distance for the ow at ReD = 10,000 and 30,000 to
develop from initial top-hat prole to a Gaussian prole. From two
to ve diameters, the streamwise turbulence intensity was found
to be the lowest at the higher ReD = 30,000, while farther downstream, a Reynolds number independence was noticed. For the
same axial range, the streamwise velocity proles were shown to
be Re independent. A close match of the velocity proles at low
ReD = 6000 and 10,000 was noted, falling below the limit proposed
by Dimotakis [9]. Investigation of the velocity spectra showed the
appearance of the inertial sub-range at Reynolds numbers above
20,000, which was considered by Fellouah et al. [12] as onset of
mixing transition. Kwon and Seo [13] studied the Reynolds number
effect on the behavior of round jets using PIV. Exploring a range of
low Reynolds numbers (ReD = 1775142), they found that the
length of the zone of ow establishment decreased with increasing
ReD and the centerline velocity decayed more rapidly. They also
found that the Reynolds shear stresses increased with Reynolds
number.
Large eddy simulation (LES) of round jets has been carried out
by Bogey and Bailly [14] to study the effect of Reynolds number
(1.7  103 6 ReD 6 4  105). In their LES simulations, the inuence
of the Reynolds number on the jet development was investigated
by keeping identical initial conditions (shear-layer thickness,
inow forcing), except for the jet diameter. Based on the vorticity

hUi
Uj
u0 v 0

vrms
X, Y, Z
yi

e
g
h
kT

m
x
hxi

conditionally averaged velocity (m/s)

jet exit velocity (m/s)
Reynolds shear stress (m2/s2)
spanwise turbulence intensity (m/s)
Cartesian coordinates
interface location (m)
turbulent dissipation rate (m2/s3)
Kolmogorov length scale (m)
momentum thickness (m)
Taylor micro-scales (m)
kinematic viscosity (m2/s)
vorticity (s1)
conditionally averaged vorticity (s1)

contours, a transitional zone was located immediately after the

potential core. In the transitional zone, a large range of scales
was observed at a higher ReD = 4  105, whereas at ReD = 1.7  103,
the ne turbulent scales were rarely present. Further, as Reynolds
number increases, the generation of the vortical structures in the
shear layer was found to occur closer to the nozzle.
Flow characteristics of square jets have been studied by
Grinstein et al. [15], Grinstein and DeVore [16], Quinn and Militzer
[17], Sankar et al. [18], Tsuchiya et al. [19] and more recently by
Ghasemi et al. [21] using numerical simulations and experiments.
It is well established that the fundamental difference between
square and round jets lies in the initial vortex structures. In square
jets, the presence of sharp corners deforms the vortex structures
which break down into smaller scales. The process of vortex selfinduction initiates a complex interaction between azimuthal and
streamwise vorticity leading to the axis-switching phenomenon
[20]. Axis-switching is believed to be responsible for the higher
entrainment [8] and enhanced mixing in square jets compared to
round jets.
In the case of square jets, very few studies provide data for the
effect of the Reynolds number on the ow and entrainment
characteristics. Ai et al. [22] performed planar laser induced
uorescence measurements to study starting square jets at three
different Reynolds numbers ReD = 2360, 3560 and 4716. At these
ow conditions, the jets do not go through the mixing transition
as discussed by Dimotakis [9].
Recently, Xu et al. [23] studied the Reynolds number effect in a
jet issuing from a long square pipe. The main difference between
jets emanating from a long pipe and from a smooth contraction
nozzle is the presence of secondary ows. In the case of a jet from
a long pipe, the secondary ows change the ow features in the
near exit of the jet, which eliminates the potential core region. In
addition to the secondary ows, presence of the thick shear layer
at the exit of the long pipe eliminates the axis-switching mechanism. The mixing transition at ReD > 30,000 was reported by examining the centerline velocity decay rate, centerline streamwise
turbulence intensity and spectra. The critical Reynolds number
reported in [23] was ReD  30,000 which is higher than the critical
Reynolds numbers reported by Dimotakis [9], Ricou and Spalding
[10] and Fellouah et al. [12].
All of the studies discussed above indicate the need to further
research on the effect of the Reynolds number on ow characteristics of square jets. Of particular interest is the inuence of the neareld turbulence on the downstream development of the jet. In this
paper, ow and turbulence statistics in the near-eld of a square
jet that emanates from a smooth square contraction nozzle is

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

investigated for 10,000 < ReD < 41,400 using a high-resolution PIV.
In particular, effect of the Reynolds number on the mean velocity,
turbulence intensities and Taylor length scale are studied in detail.
The present study also investigates the evolution of the turbulent/
non-turbulent (T/NT) interface in the near-eld of the square jet,
where the ow is subjected to complex three-dimensional vortex
dynamics as a result of axis-switching. No prior knowledge exists
to predict the effect of the Reynolds number on the conditional
averaged velocity and vorticity proles at the T/NT interface in
the near-eld of square jet.

2. Experimental setup
Measurements were carried out in a water jet facility similar to
that used in earlier studies [11,21], consisting of a rectangular tank
2 m long, 1 m wide, and 0.7 m deep. The ow passes through a
nozzle with a smooth contraction, which terminates in a square
opening with 0.01 m sides as shown in Fig. 1. The center of the
jet nozzle was located 0.3 m above the bottom of the jet facility,
and 0.5 m away from both side walls of the tank. This facility has
been successfully used in other jet studies [11,21], demonstrating
that the ow development in the near eld of the jet is not affected
by the tank walls. For the coordinate system, X, Y and Z are taken as
the distance along the jet axis, along the spanwise direction and
along the vertical direction, respectively, with the origin located
on the jet axis at the nozzle exit.
The nozzle was mounted ush with the wall of the tank. The jet
discharge was provided by an overhead reservoir with a constant
head supply of 2.0 m. The ow from the overhead reservoir was
controlled by a valve to deliver a constant velocity at the nozzle
exit. The jet exit Reynolds number ReD = UjD/m based on the equivalent diameter D (= 4  area/wetted perimeter) and the jet exit
velocity Uj was varied in the range of 10,00041,400.
Particle image velocimetry measurements were acquired in the
horizontal (X, Y) plane through the jet centerline covering a length
up to ve jet diameters from the nozzle exit. A 1.0 mm thick laser
sheet, issuing from a 50 mJ/pulse Nd:YAG laser, was introduced
through the transparent side wall of the jet facility to illuminate
the silver coated hollow glass spheres (diameter = 12 lm; specic
gravity = 1.13) utilized as tracer particles. Care was taken to ensure
that the position of the laser sheet was oriented in the horizontal
plane through the jet axis. The TSI Powerview Plus 4MP CCD
camera with a resolution of 2048  2048 pixels was synchronized
with the laser pulse to capture the scattered light from the tracer
particles.
The size of the eld of view (FOV) used for all PIV measurement
was 56 mm  56 mm. Over 2000 pairs of sequential images were
captured with a frequency of 1.04 Hz. The temporal interval (Dt)

between two images was varied from 60 to 220 ls to

accommodate the local ow characteristics. An average particle
displacement of eight pixels was ensured on the jet centerline to
minimize velocity bias to approximately 1.4%. The images captured
were post processed using adaptive correlation and moving average validation options. The adaptive correlation uses a multi-pass
fast Fourier transform cross-correlation algorithm to determine
the average particle displacement within an interrogation area. A
three-point Gaussian curve t was used to determine particle
displacement with subpixel accuracy. The captured images were
interrogated using a 64  64 pixel rst interrogation window and
a 32  32 pixel second interrogation window, with 50%
overlapping. The smaller spatial resolution of the PIV is limited
and can be represented by

LPIV

LIA
Lv :
LCCD

Here LPIV is the smallest scale that can be resolved with the PIV, Lv is
the dimension of the viewing area in the ow, and LIA and LCCD are
the dimensions of the interrogation area (IA) and CCD array in pixels, respectively. Using Eq. (1), the spatial resolution for the present
PIV experiments is LPIV = 0.44 mm, which corresponds to 37 pixel/
mm. The ratio of LPIV to the equivalent jet diameter of the present
PIV measurements yields an estimated value of 0.044, which is
comparable to that of Khashehchi et al. [24], providing high spatial
resolution PIV. After processing all images, the velocity vectors were
validated using statistical tools and less than 1% of the vectors were
found to be erroneous and needed to be replaced by the
Gaussian-weighted mean of the neighboring vectors. Measurement
uncertainty in the data acquired was determined using the
methodologies outlined by Coleman and Steele [25] and Forliti
et al. [26]. At 95% condence level, the uncertainties in the mean
velocity, turbulence intensities, Reynolds shear stresses and
vorticity were estimated to be 2%, 3%, 5% and 8%, respectively.
3. Results and discussion
In the following section, mean ow parameters and turbulence
quantities are discussed. Unless otherwise noted, all distances are
normalized by the equivalent diameter of the nozzle (D = 0.01 m).
3.1. Inlet jet conditions
The near-exit conditions of the jet at different Reynolds numbers are assessed at X/D = 0.5. As shown in Fig. 2a, the mean velocity distributions have a top-hat shape and do not depend on the
Reynolds number. Proles of streamwise turbulence intensity urms/
Uj are presented in Fig. 2b. There is clearly a Reynolds number
dependence for the peak value of the turbulence intensity, which
decreases with increasing ReD, from 0.26 at ReD = 10,000 to 0.17
at ReD = 41,400. This trend coincides with the ndings of Deo
et al. [27] for a plane jet and those of Xu et al. [23] for a square pipe
jet. The urms/Uj proles in the central portion of the jet, where
urms/Uj is less than 2%, are not affected by the Reynolds number.
Shear layer development in the near-eld of the jet affects mean
ow parameters and turbulent structures downstream. Therefore,
it is worthwhile to investigate the Reynolds number effect on the
shear layer momentum thickness. The momentum thickness (h)
is calculated at different axial locations from

Z
0

Fig. 1. Schematic of the square jet nozzle and reference system.

251

1:8D



U
U
dY;
1
Uc
Uc

where Uc is the local centerline velocity. The momentum thickness

was obtained by performing independent numerical integration of
each half of the velocity prole and averaging the calculated values.

252

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

(a)

1.0
0.8
0.6

0.30
0.25
0.20

urms/Uj

U/Uj

(b)

ReD=10,000
ReD=17,000
ReD=25,000
ReD=32,400
ReD=41,400

0.4

0.15
0.10

0.2

0.05

0.0

0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Y/D

Y/D

Fig. 2. Jet near-exit conditions at X/D = 0.5 (a) mean velocity distributions and (b) turbulence intensity distributions.

different from that reported by Deo et al. [27] for a plane jet, where
it was reported that hX=0.5h decreases with increasing Reynolds
number. At X/D > 2, the momentum thickness for the lower
Reynolds number jet (ReD = 104) is consistently larger than for the
higher Reynolds number jets. This suggests a more rapid growth
of the turbulent boundary layer at lower ReD, and growth which is
independent of Reynolds number for ReD P 1.0  104.

ReD=10,000
ReD=17,000
ReD=25,000
ReD=32,400
ReD=41,400
ReD=55,000 Round jet

0.25

0.20

/D

0.15

3.2. Jet development in the near-eld

0.10

0.05

0.00
1

X/D
Fig. 3. Shear layer momentum thickness vs. axial distance.

As shown in Fig. 3, the present results at the axial location X/D = 0.5
rule out the dependence on the initial jet conditions and Reynolds
number with average value hX=0.5D = 0.025D. This trend is somewhat
X/D=1

1.0

(b)

0.8

0.6

0.6

0.4

X/D=2

1.0

0.8

U/Uj

U/Uj

(a)

0.8

Gaussian

0.4

0.2

0.0

0.0

1.0

0.0

0.5

1.5

(e)

0.8

U/Uj

0.6
0.4

0.4
0.2

0.0

0.0
0.0

0.5

1.0

Y/D

1.5

1.0

ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.6

0.2

0.5

X/D=5

1.0
0.8

Gaussian

0.0

Y/D

Y/D

X/D=4

1.0

1.0

Gaussian

0.4

0.2

Y/D

U/Uj

0.6

0.0
0.5

X/D=3

1.0

0.2

0.0

(d)

(c)

U/Uj

Fig. 4 shows the mean streamwise velocity proles of the

square jet at different Reynolds numbers. As expected, the central
portion of the velocity proles decreases with increasing X/D,
demarcating the potential core of the jet. The potential core of
the jet is usually dened as the region in the near eld of the jet
(0 < X/D < 7) where the streamwise velocity is constant, and it is
observed only for jets issuing from contraction nozzles. For a round
jet, this region is also known as a region of ow establishment,
characterized by merging of the shear layers which cause a
rapid development of turbulence. The streamwise velocity U/Uj
decreases when radially moving away from the jet centerline into
the region of the shear layers. Near the nozzle, the mean velocity
proles retain a top-hat prole. With increasing downstream

Gaussian

0.0

0.5

1.0

1.5

2.0

Y/D

Fig. 4. Spanwise distributions of the U/Uj at (a) X/D = 1; (b) X/D = 2; (c) X/D = 3; (d) X/D = 4; and (e) X/D = 5.

1.5

253

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

0.2

X/D=1

0.1

(b)

0.2

0.1

0.0
-1.5

-1.0

-0.5

0.0

0.5

1.0

0.0
-1.5

1.5

X/D=2

-1.0

-0.5

X/D=4

(e)

urms/Uj

urms/Uj

0.2

0.1

0.0
-1.5

-1.0

-0.5

0.0

0.5

1.0

(c)

0.2

0.1

0.0
-1.5

1.5

X/D=3

-1.0

-0.5

0.0

0.5

1.0

0.2

0.5

1.0

1.5

X/D=5
ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.1

0.0

1.5

0.0

Y/D

Y/D

Y/D

(d)

in the shear layers of the jet. With increasing downstream distance,

the jet entrains more ambient uid into the core region and the
streamwise turbulence intensity along the jet centerline gradually
increases. Unlike the mean velocity proles, the proles of
urms/Uj do not show strong transition with Reynolds number.
As illustrated in Fig. 6, spanwise turbulence intensity proles go
through a stronger transition with Reynolds number compared to
the streamwise turbulence intensity. The spanwise turbulence
intensity proles associated with ReD = 10,000 show deviations in
the vicinity of the shear layer as well as the jet centerline. Stronger
Reynolds number dependence for the velocity uctuations can be
connected to the additional spanwise deformations taking place

urms/Uj

urms/Uj

(a)

urms/Uj

distance the proles approach a Gaussian distribution. Up to X/

D = 3, the velocity prole at ReD = 104 deviates from the rest and
become wider. Velocity proles at ReD P 17,000 collapsed on to a
single distribution for all X/D, showing a Reynolds number
independence in agreement with the transition criterion suggested
by Dimotakis [9]. However, for X/D > 4 (beyond the potential core)
all the velocity proles, including for ReD = 104, collapse and
Reynolds number independence is observed.
Fig. 5 shows that the distribution of the streamwise turbulence
intensity has a weak ReD dependence. At X/D = 1, inside the
potential core of the jet the magnitude of urms/Uj for all Reynolds
numbers is below 3%. The maximum values of urms/Uj are observed

-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

Y/D

Y/D

Fig. 5. Spanwise distributions of urms/Uj for (a) X/D = 1; (b) X/D = 2; (c) X/D = 3; (d) X/D = 4 and (e) X/D = 5.

X/D=1

0.12

(b)

X/D=2

0.12

0.10

0.08

0.08

vrms /Uj

0.10

0.06

0.06
0.04

0.04

0.02

0.02

0.02

0.00

0.00
-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

-1.0

-0.5

0.0

X/D=4

0.12

(e)

0.12

0.10

0.10

0.08

0.08

0.06

0.04

0.02

0.02
-1.0

-0.5

0.0

1.0

1.5

0.00
-1.5

-1.0

-0.5

0.5

1.0

1.5

Y/D
Fig. 6. Spanwise distributions of

0.0

Y/D

X/D=5
ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.06

0.04

0.00
-1.5

0.5

Y/D

vrms /Uj

vrms /Uj

0.06

0.04

-1.5

X/D=3

0.12

0.08

Y/D

(d)

(c)

0.10

vrms /Uj

vrms /Uj

(a)

0.00
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

Y/D

vrms/Uj for (a) X/D = 1; (b) X/D = 2; (c) X/D = 3; (d) X/D = 4 and (e) X/D = 5.

0.5

1.0

1.5

254

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

X/D=1

0.010

(b)

0.005

-0.010
-1.5

0.000

-0.5

0.0

0.5

1.0

-0.010
-1.5

1.5

-1.0

-0.5

Y/D

(e)

0.5

1.0

-0.010
-1.5

1.5

-1.0

-0.5

0.010

1.0

1.5

ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.000

-0.005

-0.005

-0.010
-1.5

0.5

X/D=5

u'v' /Uj

0.000

0.0

Y/D

0.005

0.005

u'v' /Uj

0.0

Y/D

X/D=4

0.010

0.000

-0.005

-0.005

-1.0

X/D=3

0.010

0.005

u'v' /Uj

0.000

-0.005

(d)

(c)

0.005

u'v' /Uj

X/D=2

0.010

u'v' /Uj

(a)

-1.0

-0.5

0.0

0.5

1.0

1.5

-0.010
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

Y/D

Y/D

Fig. 7. Lateral distributions of the u0 v 0 =U 2j for (a) X/D = 1; (b) X/D = 2; (c) X/D = 3; (d) X/D = 4 and (e) X/D = 5.

(a)

0.10

ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.08

urms/Uj

0.06

0.04

0.02

0.00
0

X/D

(b) 0.10

ReD= 10,000
ReD= 17,000
ReD= 25,000
ReD= 32,400
ReD= 41,400

0.08

0.06

vrms/Uj

in the near-eld of square jets due to the axis-switching. Ghasemi

et al. [21] have shown that the instantaneous streamwise velocity
proles uctuate laterally in the near-eld of square jets.
Reynolds shear stress proles shown in Fig. 7 exhibit sharp
peaks at the edges of the jet which diminish to zero at the jet centerline. The lower Reynolds number case (ReD = 104) yields larger
values of normalised shear stress compared to the other values
of ReD. The higher values of the spanwise turbulence intensity
and larger Reynolds shear stress generation can be attributed to
the presence of larger vortical structures due to the thicker shear
layer. As can be seen, with increasing downstream distance, the
difference in the low and high Reynolds number shear stress
proles becomes smaller and beyond X/D = 4, all proles attain
similarity.
Axial distributions of streamwise and spanwise turbulence
intensity along the jet centerline are shown in Fig. 8a and b, respectively. Formation of the vortices at the shear layer and their continuous convection downstream is accompanied by strong
perturbations of the velocity uctuations. These uctuations penetrate into the jet core as the jet travels downstream, providing for
an increase in urms/Uj and vrms/Uj distributions. It should be noted
that this increasing trend is dominant in the near-eld, which is
the measurement range in the present study. It is expected that
beyond the potential core, where the vortex ring breaks down
and both shear layers are merged, the turbulence intensity will
approach an asymptotic value in the developing [21] and self-similar [12] jet regions. Similar to the lateral distributions discussed in
Figs. 5 and 6, Reynolds number dependence is more signicant for
the spanwise turbulence intensity distributions. Fig. 8b clearly
shows a deviation of vrms/Uj at ReD = 104.
The difference in the distributions of turbulence intensities
along the jet centerline is further highlighted by examining the
large scale anisotropy, dened as the ratio of urms/vrms. Ideally, a
value of unity satises the isotropic condition. However, previous
measurements in round jets [28] suggest values of urms/vrms
between 1 and 1.3. The distributions of urms/vrms at different ReD
are shown in Fig. 9, together with error bars representing a 5%

0.04

0.02

0.00
0

X/D
Fig. 8. Axial distributions of (a) urms and (b)

vrms along the jet centerline.

255

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

2.0

velocity eld in the plane, which allows for direct calculation of

the Taylor macro-scale. Here, the Taylor micro-scale was obtained
by [30]:

v
u 02
u u
kT t  :
@u0 2

1.5

It should be noted that the derivative in Eq. (3) was estimated

by using a central difference scheme, which reduces the errors
due to noise associated with the PIV measurements. To further
evaluate the limitation of the current PIV measurements, we also
require an estimate of the Kolmogorov length scale (for a local isotropic condition) given as

1.0

ReD = 10,000
ReD = 17,000
ReD = 25,000
ReD = 32,400
ReD = 41,400

0.5

0.0
0

@x

 3 1=4

m
e

X/D
Fig. 9. Axial distributions of urms/vrms along the jet centerline.

standard deviation. In the near-eld of the jet, the ow is highly

anisotropic and urms/vrms approaches unity only at X/D = 4, which
marks the end of the potential core. In the initial regions, higher
levels of anisotropy are noted for ReD = 104 compared to Reynolds
number higher than the transition level. In this region, increasing
the Reynolds numbers above the transition value can enhance
mixing. However, increasing the Reynolds number beyond the
transition level is not helpful in enhancing mixing beyond
X/D = 4. A similar conclusion may be drawn by considering the fact
that the larger structures formed in the ReD = 104 case reduce the
interfacial area and diminish mixing. It is also interesting to
observe that beyond the potential core, where the vortex ring
breaks down, urms/vrms is Reynolds number independent. This trend
beyond the potential core suggests that while near-eld mixing is
Reynolds number dependent, mixing is not inuenced by Reynolds
number in the far eld of the jet.
3.3. Taylor length scale
Evaluation of the turbulent length scales in a jet ow can enable
better prediction of the mixing processes. The turbulent jet contains a large spectrum of length scales covering the integral, Taylor
and Kolmogorov length scales. According to Chhabra et al. [29], in
the self-similar jet region, it is assumed that the integral length
scales are of the order of the local jet width. These large scales
are responsible for extracting energy from the mean ow without
being affected by viscosity. The intermediate length scales at the
top of the inertial range are known as Taylor micro-scales. By
applying the scale analysis, the Taylor micro-scales in the self-similar jet
underlocally
isotropic conditions can be obtained from
p

kT 10mk=e, where k is the turbulent kinetic energy and e is

the dissipation. Direct estimates of k and e are difcult to obtain
from velocity measurements and therefore Taylor length scales
are traditionally determined from stationary point hot-wires measurements invoking the Taylors hypothesis. Taylors hypothesis
converts a velocity time series into a space series and can lead to
large errors, especially in the ow regions of high turbulence uctuations. Both Taylors hypothesis and the assumption of locally
(small-scale) isotropic turbulence are not truly valid in turbulent
jet ows, as pointed out by Mi and Nathan [5]. Nevertheless,
calculations of kT using Taylors hypothesis have been reported
by Fellouah et al. [12] for the case of a round jet, as well as Mi
and Nathan [5] for jets issuing from different shaped nozzles. An
advantage of the PIV is its ability to provide information for the

where m is the kinematic viscosity and e is the dissipation. Using

equilibrium turbulence arguments, the dissipation can be reexpressed as [30]

eA

u03
;
l

where l is the integral length scale and A is a proportionality constant of approximately 1. Eq. (5) assumes that the turbulence is
homogeneous, isotropic and in local spectral equilibrium, which
might not be the case for the initial region of the square jet, where
the integral length scale is not a single value but rather varies from
region to region. Despite the shortcomings of Eq. (5), an approximate value for the dissipation can be obtained to assess the limitations of the PIV measurements. Substituting Eq. (5) into Eq. (4), and
noting that urms  (0.17  0.15)Uj (see Fig. 6) and l is equal to the
half-width of the jet [21], the Kolmogorov length scales are
g = (1232) lm for ReD = 10,00041,400. The spatial resolution of
the present PIV experiments (LPIV = 0.44 mm) is 1436 times larger
than the estimated Kolmogorov length scale and therefore it is more
appropriate to discuss the Taylor macro-scales. The resolved Taylor
length scales are approximately 1.53 times larger than LPIV.
Distributions of the Taylor scale for all square jets along the jet
centerline are shown in Fig. 10. To avoid cluttering in Fig. 10, 5%
error bars were included only for the lower and higher Reynolds
number cases, respectively.
The general trend of the Taylor scale for X/D < 3 shows an
increase downstream from the nozzle. This is not surprising since
in the near-eld of the jet the vortex activity is due to the shear

0.14
0.12
0.10

T /D

urms/vrms

Increasing ReD

0.08
0.06

ReD = 10,000
ReD = 17,000
ReD = 25,000
ReD = 32,400
ReD = 41,400

Increasing ReD

0.04
0.02
0.00
0

X/D
Fig. 10. Distributions of the Taylor length scale along jet centerline.

256

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

layer roll-up, vortex ring formation and axis-switching associated

with the large-scale structures. The Reynolds number effect is
clearly visible as the Taylor scale reduces with increasing Reynolds
number. As the jet progresses downstream 3 < X/D < 5, the largescale structures start to break-down and secondary structures
appear to dominate. It is anticipated that at the end of the jet
potential core, the Taylor scale slowly changes, while small scales
are generated. In the far eld of the jet at X/D > 10, kT/D grows
linearly with axial distance as shown by Mi and Nathan [5].
3.4. Evolution of turbulent/non-turbulent interface
According to the criteria discussed by Westerweel et al. [31],
Anand et al. [32] and Khashehchi et al. [24], the turbulent/non-turbulent (T/NT) interface of the jet can be detected from the PIV measurements. By identifying the location where the instantaneous
velocity satises the criterion that u/Uj > 0.03, the turbulent/nonturbulent interface can be captured. To locate the T/NT interface
a numerical procedure was developed that checks for the location
of the interface starting from the centerline of the jet and moving
radially outwards. The radial position at which the condition
u/Uj > 0.03 is rst violated is marked as interface (yi) for that
streamwise location. The code does not check for multiple
locations where this condition can be violated. The location of
the interface is calculated for the entire velocity data set.
Fig. 11a and b illustrates the instantaneous location of the T/NT
interface superimposed on the vorticity (x) contours for square jet
at ReD = 10,000 and 41,400, respectively. The region where
u/Uj < 0.03 corresponds to the non-turbulent zone, where the ow
is considered irrotational. Conversely, the jet region where

(a)

2
1.5
1

D/Uj
5
4
3
2
1
0
-1
-2
-3
-4
-5

Y/D

0.5
0
-0.5
-1
-1.5
-2

u/Uj > 0.03 is a turbulent region with intense vorticity. The instantaneous location of the T/NT interface encloses the shear layers
formed at the edges of the jet. The interface determined from the
velocity criterion is slowly expanding downstream as the jet
entrains more uid. The contractions and expansions of the interface can be related to the entrainment of the ambient uid behind
a passing vortex ring as discussed by Ghasemi et al. [21]. In the
core of the jet, the vorticity levels again become negligible. The
contours of the instantaneous vorticity also suggest a longer potential core for the case of ReD = 10,000, while at higher ReD = 41,400
the shear layers tend to merge at X/D  4.5 providing for additional
jet entrainment mechanism. The higher level of intermittency
observed at the end of the potential core can degrade the continuity of the jet interface. A recent study of the evolution of the T/NT
interface of a round jet by Khashehchi et al. [24] has shown no
jump in the conditional average velocity and vorticity proles for
X/D < 8. As can be seen beyond the interface, instantaneous
vorticity diminishes due to entering the irrotational ow region.
In Fig. 12, statistical probability density function distributions
of the location of the T/NT interface are shown at different X/D
locations for ReD = 41,400. At X/D = 2, a narrow distribution of yi
is obtained, with ymax
0:88D. With increasing X/D the
i
distributions of yi become spread out implying a high level of intermittency, possible due to increase of the vortex activity due to
breaking of the vortex ring and axis-switching. The ymax
location
i
of the interface progresses further outwards, which is consistent
with expansion of the jet due to the entrainment.
The evolution of the jet characteristics at the T/NT interface are
analyzed for a square jet at two Reynolds numbers. Conditionally
averaged velocity and vorticity proles are determined by averaging the data at xed radial distances relative to the interface location (yi) for the near-eld of the square jet. The mean conditionally
averaged velocity hUi normalized with the centerline velocity Uc is
shown in Fig. 13. All velocity proles are zero at the non-turbulent
side of the jet, while inside of the jet a rapid increase of the velocity
proles are observed. Both Reynolds numbers show similar behavior of the conditionally averaged velocity proles, where no jump
across the interface is noted.
Fig. 14 shows the evolution of the mean conditional averaged
vorticity for several different X/D locations. The conditional
averaged vorticity proles are normalized with Uc/(D/2). Vorticity
proles for X/D < 5 show no peak near the interface, which is
consistent with the observation of Khashehchi et al. [24] for the
case of a round jet. At X/D = 2, the maximum vorticity is observed
inside the jet for both Reynolds numbers. The vorticity gradually

X/D

(b)

X/D = 2
X/D = 3
X/D = 4
X/D = 5

200

1.5

D/Uj
5
4
3
2
1
0
-1
-2
-3
-4
-5

Y/D

0.5
0
-0.5
-1

150

Counts

100

50

-1.5
-2

X/D
Fig. 11. Instantaneous vorticity contours and envelop of the turbulent/nonturbulent (T/NT) interface for (a) ReD = 10,000 and (b) ReD = 41,400.

0
0.5

1.0

1.5

2.0

yi /D
Fig. 12. PDF distributions of the T/NT interface location for ReD = 41,400.

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

1.0

X/D=2
X/D=3
X/D=4
X/D=5

0.8

<U>/Uc

0.6

0.4

0.2

0.0

-0.5

0.0

0.5

1.0

1.5

2.0

2(y-yi )/D
Fig. 13. Proles of mean conditional streamwise velocity for several different
distances from the jet exit (solid symbols: ReD = 10,000; open symbols:
ReD = 41,400).

1.0

X/D=2
X/D=3
X/D=4
X/D=5

<>D/(2Uc )

0.8

0.6

0.4

0.2

0.0
-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2(y-yi )/D
Fig. 14. Proles of mean conditional averaged vorticity for several different
distances from the jet exit (solid symbols: ReD = 10,000; open symbols:
ReD = 41,400).

decays inside of the jet with increasing X/D locations. At X/D = 5, a

small increase in vorticity is noted near the T/NT interface. The
evolution of the conditional averaged vorticity proles appears
independent of the Reynolds number for the range studied. The
lack of a peak in the conditionally averaged vorticity proles in
the near-eld of the square jet implies that the jet is far from a
self-similar state where a formation of vortex sheet structures
has been reported at the interface for round jets by Westerweel
et al. [31]. The ow in the near-eld of a square jet appears to be
highly unstable with intense vortex activity which precludes
formation of a continuous interface.

4. Conclusions
Reynolds number (ReD = 10,00041,400) effect on the mean
ow and turbulence parameters, momentum thickness, centerline
jet anisotropy ratio urms/vrms and Taylor length scales in the neareld (X/D < 5) of a square jet is studied using high resolution PIV
experiments. The mean velocity proles become Reynolds number

257

independent at an axial location of X/D > 4 where they conform to

the Gaussian shape describing the velocity distributions in
axisymmetric jets. On the other hand, the turbulence intensity
proles show a disparity between the low Reynolds number case
(ReD = 10,000) and higher ReD, which implies that the turbulence
in the square jet requires larger X/D to attain a self-similar state.
Calculation of the momentum thickness reveals that the low
Reynolds number jet develops a thicker shear layer while all high
Reynolds number cases show a similar shear layer thickness. A
consistent deviation of the distributions of spanwise turbulence
intensities along the jet centerline is observed at lower
ReD = 10,000.
The jet centerline large scale anisotropy is investigated by
examining the ratio urms/vrms. In the near-eld of the jet, the ow
is highly anisotropic and urms/vrms approaches unity at X/D = 4,
which is considered the end of the jet potential core. In the region
X/D < 4, lower urms/vrms are observed for ReD = 10,000, compared to
Reynolds number higher than the transition level. This suggests
that increasing the Reynolds number above the transition value
could enhance mixing. A similar conclusion may be drawn by considering the fact that the larger structures formed at ReD = 10,000
reduce the interfacial area and thus diminish mixing. This effect
is negligible at X/D > 5 where the mixing is not inuenced by the
Reynolds number. Analysis of the Taylor length scales is available
from the present PIV experiments. As the jet progresses downstream in the range 3 < X/D < 5, the large-scale structures start to
break down, secondary structures appear to dominate and the Taylor scale reduces with increasing Reynolds number.
Evolution of the T/NT interface was analyzed using conditional
averaged streamwise velocity and vorticity elds. Using a streamwise velocity criterion, the instantaneous and time averaged turbulent/non-turbulent interface of the square jet was calculated
for different Reynolds numbers. The instantaneous interface clearly
separates irrotational and rotational ow regions as illustrated by
the vorticity eld. The statistics of the conditionally averaged
streamwise velocity does not show a clear jump across the interface for any of the examined jet cases. Similarly, vorticity proles
for X/D < 5 show no peak near the interface, which is consistent
with the observation of Khashehchi et al. [24] for the case of a
round jet. The ow in the near-eld of a square jet appears highly
unstable with intense vortex activity which precludes formation of
a continuous interface.
Acknowledgements
Discovery Grants from the Natural Sciences and Engineering
Research Council of Canada to RB and RMB are gratefully
acknowledged.
References
[1] A. Darisse, J. Lemay, A. Benassa, LDV measurements of well converged third
order moments in the far eld of a free turbulent round jet, Exp. Therm. Fluid
Sci. 44 (2013) 825833.
[2] A.P. Vouros, T. Panidis, Turbulent properties of a low Reynolds number
axisymmetric pipe jet, Exp. Therm. Fluid Sci. 44 (2013) 4250.
[3] D. Liepmann, M. Gharib, The role of streamwise vorticity in the near-eld
entrainment of round jets, J. Fluid Mech. 245 (1992) 643668.
[4] M. Azad, W.R. Quinn, D. Groulx, Mixing in turbulent free jets issuing from
isosceles triangular orices with different apex angles, Exp. Therm. Fluid Sci.
39 (2012) 237251.
[5] J. Mi, G.J. Nathan, Statistical properties of turbulent free jets issuing from nine
differently-shaped nozzles, Flow Turbul. Combust. 84 (2010) 583606.
[6] M. El Hassan, A. Meslem, Time-resolved stereoscopic particle image
velocimetry investigation of the entrainment in the near eld of circular and
daisy-shaped orice jets, Phys. Fluids 22 (2010) 126.
[7] A. Ghasemi, V. Roussinova, R.M. Barron, R. Balachandar, Analysis of
entrainment at the turbulent/non-turbulent interface of a square jet, in:
ASME Congress & Exposition IMECE2013-65355, 2013, pp. 17.

258

A. Ghasemi et al. / Experimental Thermal and Fluid Science 61 (2015) 249258

[8] K.B.M.Q. Zaman, Axis switching and spreading of an asymmetric jet: the role of
coherent structure dynamics, J. Fluid Mech. 316 (1996) 127.
[9] P.E. Dimotakis, The mixing transition in turbulent ows, J. Fluid Mech. 409
(2000) 6998.
[10] F. Ricou, D.B. Spalding, Measurements of entrainment by axisymmetric
turbulent jets, J. Fluid Mech. 11 (1961) 2132.
[11] A. Tandalam, R. Balachandar, R.M. Barron, Reynolds number effects on the
near-exit region of turbulent jets, J. Hydraul. Eng. ASCE 136 (2010) 633641.
[12] H. Fellouah, C.G. Ball, A. Pollard, Reynolds number effects within the
development region of a turbulent round free jet, Int. J. Heat Mass Transfer
52 (2009) 39433954.
[13] S.J. Kwon, I.W. Seo, Reynolds number effects on the behavior of a non-buoyant
round jet, Exp. Fluids 38 (2005) 801812.
[14] C. Bogey, C. Bailly, Large eddy simulations of transitional round jets: inuence
of the Reynolds number on ow development and energy dissipation, Phys.
Fluids 18 (2006) 114.
[15] F.F. Grinstein, E. Gutmark, T. Parr, Near eld dynamics of subsonic free square
jets: a computational and experimental study, Phys. Fluids 7 (1995) 14831497.
[16] F.F. Grinstein, C.R. DeVore, Dynamics of coherent structures and transition to
turbulence in free square jets, Phys. Fluids 8 (1996) 12371251.
[17] W.R. Quinn, J. Militzer, Experimental and numerical study of a turbulent free
square jet, Phys. Fluids 31 (1988) 10171025.
[18] G. Sankar, R. Balachandar, R. Carriveau, Characteristics of a three dimensional
square jet in the vicinity of a free surface, J. Hydraul. Eng. 135 (2009) 989994.
[19] Y. Tsuchiya, C. Horikoshi, T. Sato, On the spread of rectangular jets, Exp. Fluids
4 (1986) 197204.
[20] G.K. Batchelor, An Introduction to Fluid Dynamics, Cambridge University Press,
London, 1967.

[21] A. Ghasemi, V. Roussinova, R. Balachandar, A study in the developing region of

square jet, J. Turbul. 14 (2013) 124.
[22] J.J. Ai, S.C.M. Yu, Adrian W.K. Law, L.P. Chua, Vortex dynamics in starting
square water jets, Phys. Fluids 17 (2005) 112.
[23] M. Xu, A. Pollard, J. Mi, F. Secretain, H. Sadeghi, Effects of Reynolds number on
some properties of a turbulent jet from a long square pipe, Phys. Fluids 25
(2013) 119.
[24] M. Khashehchi, A. Ooi, J. Soria, I. Marusic, Evolution of the turbulent/non
turbulent interface of an axisymmetric turbulent jet, Exp. Fluids 54 (2013)
14491461.
[25] H.W. Coleman, W.G. Steele, Engineering application of experimental
uncertainty analysis, AIAA J. 33 (1995) 18881896.
[26] D.J. Forliti, P.J. Strykowski, K. Debatin, Bias and precision errors of digital
particle image velocimetry, Exp. Fluids 28 (2000) 436447.
[27] R.C. Deo, J. Mi, G.J. Nathan, The inuence of Reynolds number on a plane jet,
Phys. Fluids 20 (2008) 116.
[28] R.A. Antonia, P. Burattini, Small-scale turbulence: how universal is it? in: Proc.
15th Aust. Fluid Mech. Conference, Sydney, Australia, 2004.
[29] S. Chhabra, P. Huq, A. Prasad, Characteristics of small vortices in a turbulent
axisymmetric jet, J. Fluids Eng. 128 (2006) 439445.
[30] H. Tennekes, J.L. Lumley, A First Course in Turbulence, MIT Press, Cambridge,
MA, USA, 1972.
[31] J. Westerweel, T. Hofmann, C. Fukushima, J.C.R. Hunt, The turbulent/nonturbulent interface at the outer boundary of a self-similar turbulent jet, Exp.
Fluids 33 (2002) 873878.
[32] R.K. Anand, B.J. Boersma, Detection of turbulent/non-turbulent interface for an
axisymmetric turbulent jet: evaluation of known criteria and proposal of a
new criterion, Exp. Fluids 47 (2009) 9951007.