On Steady and Pulsed Low-Blowing-Ratio Transverse Jets

J. Fluid Mech. (2013), vol. 714, pp. 393433.
doi:10.1017/jfm.2012.482
c Cambridge University Press 2013

393
On steady and pulsed low-blowing-ratio

transverse jets
G. Bidan and D. E. Nikitopoulos
Department of Mechanical Engineering, Louisiana State University, Baton Rouge, LA 70803, USA
(Received 19 March 2012; revised 20 August 2012; accepted 28 September 2012)
The present experimental and numerical study focuses on the vortical structures
encountered in steady and pulsed low-blowing-ratio transverse jets (0.150 6 BR 6 4.2),
a configuration hardly discussed in the literature. Under unforced conditions at low
blowing ratio, a stable leading-edge shear-layer rollup is identified inside the jet pipe.
As the blowing ratio is increased, the destabilization and evolution of this structure
sheds light on the formation mechanisms of the well-known transverse jet vortical
system. A discussion on the nature of the counter-rotating vortex pair in low-blowingratio transverse jets is also provided. Under forced conditions, the experimental
observations support and extend numerical results of previous fully modulated jet
studies. Large-eddy simulation results provide scaling parameters for the classification
of starting vortices for partly modulated jets, as well as information on their threedimensional dynamics. The counter-rotating vortex pair initiation is observed and
detailed in both Mie scattering visualizations and simulations. The observations
support a mechanism based on stretching of the starting vortical structures because
of inviscid induction and partial leapfrogging. Two modes of cross-flow ingestion
inside the jet pipe are described as the pulsed jet cycles from high to low values of
blowing ratio.
Key words: jets, vortex dynamics, vortex interactions
1. Introduction
Smoke stacks, vertical and/or short take-off and landing (VSTOL) aircraft, and
fuel injectors are few of the many engineering systems involving transverse jets, also
called jets in cross-flow, and have triggered numerous flow studies over the past
70 years. Among them, Fric & Roshko (1994) documented the four fundamental
vortical structures involved in this type of flow configuration: the shear-layer vortices,
counter-rotating vortex pair (CRVP), horseshoe vortex and wake vortices.
The shear-layer vortices are near-field structures, typically taking the form of
consecutive vortex rings convected in the free stream and generally considered to
be the result of KelvinHelmholtz type instability of the jet cylindrical shear layer
(see Fric & Roshko 1994; Kelso, Lim & Perry 1996; Yuan, Street & Ferziger 1999),
although other theories have been formulated by Blanchard, Brunet & Merlen (1999),
for instance.
Email addresses for correspondence: guillaume.bidan@gmail.com, medimi@lsu.edu
394
The CRVP is commonly considered to be the principal mixing structure in

transverse jets, and has been the topic of many studies concerned with mixing
enhancement or prevention. Formed downstream of the jet exit and dominating the
far field, the CRVP consists of a pair of quasi-streamwise vortices. Several initiation
mechanisms have been proposed, among which that of Yuan et al. (1999), who
identified a pair of hanging vortices at the base of the jet column as the origin of the
CRVP vorticity. Other research groups suggested that the CRVP vorticity was directly
issued from the jet shear layer being reoriented under the influence of the cross-flow
(Andreopoulos & Rodi 1984; Haven & Kurosaka 1997). In particular, studies of the
shear-layer vortical structures by Sykes, Lewellen & Parker (1986) and Kelso et al.
(1996) showed that vertical vorticity was provided to the CRVP by the folding of
the shear-layer vortices through their side arms, while the upstream and downstream
rollups cancelled each other. This mechanism was later supported by the transient
numerical simulations of Cortelezzi & Karagozian (2001) and later of Marzouk &
Ghoniem (2007) using vortex simulation methods, which showed the initiation of the
CRVP through the folding of the starting structures.
The horseshoe vortex constitutes the third characteristic vortical structure of
transverse jets and is the result of cross-flow boundary-layer separation upstream
of the jet exit, comparable to the flow separation upstream of a wall-mounted cylinder.
Krothapalli & Lourenco (1990) and Kelso & Smits (1995) conducted in-depth studies
of the horseshoe vortex system, identifying a dynamic coupling with the jet shear-layer
structures leading to streamwise oscillation of the leading horseshoe vortex ahead of
the jet exit.
Finally, the wake vortices were described by Fric & Roshko (1994) as tornado-like
quasi-vertical vortices, located downstream of the jet exit and below the jet core,
originating from cross-flow boundary-layer separation due to the adverse pressure
distribution past the jet column. According to this mechanism, initially spanwiseoriented rollers are reoriented and stretched vertically by interaction with the jet core
(Rivero, Ferre & Giralt 2001).
Numerous studies have dealt with the characterizations of general quantities such
as jet penetration, spread and mixing rate, and have identified the jet to cross-flow
mass-flux ratio (blowing ratio, BR) and the jet to cross-flow momentum ratio as two
of the principal scaled parameters for transverse jets. Although a considerable number
of studies have been documenting transverse jets, unanimous agreement has yet to be
reached on the formation mechanisms of the different vortical structures, a result of
the high complexity associated with these highly three-dimensional flows.
Moreover, since mixing enhancement and VSTOL propulsion have been the main
concern in the past studies, most of them have focused on jets with rather high
blowing ratio, while only a few studies have considered blowing ratios less than 1.0.
Among them, Gopalan, Abraham & Katz (2004) showed that the vortical structures
at low blowing ratios (BR < 0.5) could be fundamentally different from previously
described jets. These systems exhibit reattaching jets associated with the formation
of a recirculation region downstream of the jet exit, enclosed by a semi-cylindrical
vortical layer originating from the jet shear layer. Other studies, such as the ones
from Acarlar & Smith (1987a,b) or Hagen & Kurosaka (1993), focusing on boundarylayer transition, have provided descriptions of the dynamics of individual structures
also encountered in low-blowing-ratio jets in cross-flow. Using a hemispherical
protuberance or long streamwise injection slits to generate hairpin vortices, they
established their overall dynamics in a near-wall region. Although some of these
investigations were carried out using transverse jet configurations, the significant
On steady and pulsed low-blowing-ratio transverse jets
395
impact of jet hole geometry, as evidenced in work from Haven & Kurosaka (1997),
New, Lim & Luo (2003) or New, Lim & Luo (2006), led to sets of vortical structures
and interactions quite different from the one encountered in round transverse jets.
Forced transverse jets have been primarily investigated for their potential in mixing
and penetration enhancement. Gogineni, Goss & Roquemore (1998) obtained up to
30 % increased penetration by simply exciting the jet shear layer using piezoelectric
actuators, while Johari, Pacheco-Tougas & Hermanson (1999), reached mixing rate
enhancements of the order of 50 % using fully modulated jets. In the latter study, two
forced jet regimes were observed. The first one for long injection times consisted of
successive puffs resembling the steady-state jet during injection, and the second for
shorter injection times was associated with the formation of a starting vortex ring at
the jet impulse, yielding significant penetration and mixing enhancements. Both studies
agreed on the fact that duty cycle and forcing frequency could greatly affect the jet
behaviour.
Eroglu & Breidenthal (2001) conducted detailed water tunnel experiments,
examining both steady-state and fully modulated transverse jets. Their observations
confirmed the formation of series of starting vortex rings in the forced jet
configuration, penetrating deeper in the cross-flow compared to the unforced jet. At
the same time, MCloskey et al. (2002) attempted penetration amplification using
acoustically excited jets in cross-flow coupled with a digital compensator and obtained
significant improvement for a specific optimal pulse width, found to be in agreement
with previous work from Gharib, Rambod & Shariff (1998). In their study, Gharib
et al. (1998) used impulsively started jets in a quiescent environment, and identified a
maximum stroke ratio beyond which the leading vortex ring could not gain additional
circulation and would depart from the wall, followed by a series of smaller trailing
ring vortices. In Johari (2006), a mapping of the different types of starting vortices
was provided for strong forced jets in cross-flow (BR > 3) as a function of the duty
cycle and the stroke ratio, highlighting the existence of four different types of starting
vortices: distinct vortex rings, vortex rings/puff plus tail, turbulent puffs and
elongated steady jet-like structures.
More recently, Sau & Mahesh (2008) investigated the formation and behaviour of
vortex rings in cross-flow emanating from a fully modulated forced jet using direct
numerical simulations (DNSs), including rather low average blowing ratios below
2.0. They found that, for blowing ratios below 2.0, large hairpin vortices would be
generated at the impulse instead of vortex rings. They also provided a mapping of
the starting vortex regimes as a function of the blowing ratio and the stroke ratio
with three distinct forms: discrete vortex ring, vortex ring with trailing column and
hairpin-like vortex structures. These results constitute a generalization of the findings
of Gharib et al. (1998) on the existence of a formation number (maximum stroke ratio)
beyond which a trailing column is formed for forced jets in a quiescent environment,
although in the case of forced transverse jets in, the transition stroke ratio varies with
the blowing ratio.
Recently, pulsed jets in cross-flow at lower average blowing ratios have been
considered for improved film cooling performance. Ekkad, Ou & Rivir (2006)
were the first to implement actively controlled film cooling jets and reported some
improvement in adiabatic effectiveness and heat transfer over the continuously flowing
jet when using full modulation. On the other hand, Coulthard, Volino & Flack (2007)
and later Womack, Volino & Schultz (2008) stated that fully modulated jets in
cross-flow were overall detrimental to film cooling applications even when used in
periodically perturbed cross-flow conditions. Interestingly, the only cases presenting
396

(a)
Contraction 16:1
Three-axis
traverse
To fan
Laser sheet
(b)
Periodic
Velocity profile /
Outflow
Computational
domain
TiCl4 ports
Flow meter
Conditioning
Needle
screens
valves
z
y
1.4 bar
x
Solenoid
valve
Y
No slip /
adiabatic
Modulated
plug velocity /
F IGURE 1. (Colour online) (a) Experimental apparatus; and (b) numerical domain.
improved metrics in the last study were high-forcing-frequency cases where 100 %
modulation was not achieved.
In view of these past results, the current study aims at investigating the vortical
structures generated in low-blowing-ratio partly modulated transverse jets. In the first
part of this paper, a detailed steady-state survey is presented to provide a description
of the vortical structures found in low-blowing-ratio transverse jets as well as their
evolution as the blowing ratio is increased to reach a more commonly studied detached
configuration. In the second part, partially modulated transverse jets are described,
with a particular focus on the transient regimes introduced by jet forcing. This
study is based on experimental results using hot-wire anemometry and Mie scattering
visualizations while monitoring the jet flow rate in a time-resolved manner. A largeeddy simulation (LES) as implemented in Ansys FluentTM is also used to provide
additional insights into the behaviour and the dynamics of the vortical structures.
2. Experimental apparatus
2.1. Wind tunnel experiments
Experiments were conducted in an open-loop aerodynamic wind tunnel with a 9 m
long test section and a 0.9 m 0.6 m cross-section, schematically presented in
figure 1(a). A set of four conditioning screens followed by a contraction with an
area ratio of 16 : 1 was located directly upstream of the test section to provide
stable inlet conditions during the experiments. Optical access was available through a
set of transparent acrylic walls constituting the top and one of the sidewalls of the
test section, respectively allowing visualizations in planes parallel to the bottom wall
(XY) and parallel to the jet symmetry plane (XZ). The coordinates x, y and z were
respectively associated with the streamwise, spanwise and vertical directions of the
flow, as shown in figure 1(a), and the quantities Xj , Yj and Zj corresponded to the
normalized coordinates with respect to the jet diameter Dj (respectively x/Dj , y/Dj and
z/Dj ). The origin was taken at the centre of the jet exit.
The free stream flow in the test section had turbulence intensity less than 0.5 %,
with a boundary-layer Reynolds number at the jet level of Re = U / = 1700
(with being the 99 % boundary-layer thickness and the air kinematic viscosity). In
this study aimed at the observation of the vortical structures generated under pulsed
397
conditions, the choice of a laminar boundary layer was made in order to limit the
number of factors affecting the jet behaviour. The jet exited from a 25.4 mm (Dj )
round exit located 760 mm downstream of the test section inlet and was mounted
flush to the bottom wall. The air supply for the jet was provided by an industrial
compressor, dried using an inline desiccant dryer regulated at 1.4 bar. At the jet inlet,
the set-up was composed of a main and a bypass branch, each equipped with a
metering. The bypass also had a computer-controlled solenoid valve and was used to
pulse the flow during forced flow experiments (see figure 1a).
This system provided the ability to set independently and accurately low (BRl )
and high (BRh ) jet blowing ratios in forced experiments, recorded in a time-resolved
manner by an inline flow meter. The jet and cross-flow were at ambient temperature,
and therefore the density ratio was unity. The overall jet pipe was over 12Dj long
to ensure fully developed flow inside the pipe. The experimental set-up was studied
using laser sheet Mie scattering visualizations, hot-wire anemometry and time-resolved
flow-rate records. An in-house seeding system injected in two locations water vapour
and titanium tetrachloride (TiCl4 ) vapour inside the jet to generate sub-micrometre
titanium dioxide (TiO2 ) particles used as a tracer in Mie scattering visualizations.
Turning off the jet water seed allowed for reactive Mie scattering visualizations since
the reaction would only occur in the test section as TiCl4 vapour reacted with moist
cross-flow in the mixing layers. To control the solenoid valve and acquire data from
the different sensors, a set of analogue-to-digital acquisition modules from National
InstrumentsTM coupled with a fully automated three-axis traverse system placed on top
of the test section were used to conduct anemometric measurements and visualizations
in a synchronized manner. Flow-rate records were acquired with at 1 kHz sampling
frequency through long enough periods of time to extract statistically significant
averaged and phase-averaged quantities, with more than 30 cycles per record during
pulsed flow experiments. Hot-wire measurements, used for spectral characterization,
were performed at a sampling frequency of 10 kHz using appropriate Nyquist filtering
at 5 kHz by means of a single wire probe (TSI model 1212-T1.5) with a sensor
length of 1.27 mm (Dj /20). Records used for statistical purposes were acquired with a
sampling interval longer than the local integral time scale, including enough points to
ensure less than 5 % uncertainty on the computed averaged quantities. Wavelet analysis
was performed on high sampling records to provide better insight into deterministic
signatures observed in unsteady flow conditions, and often masked in Fourier analysis.
The wavelet computations were made using a modified version of the algorithm
presented in Torrence & Compo (1998) with a Morlet wavelet of characteristic
parameter 6 giving satisfactory temporal and spectral resolutions between 0 and 50 Hz.
Contrast limited adaptive histogram enhancement (CLAHE) from MATLABTM was
used in Mie scattering visualizations to locally improve the contrast.
2.2. Numerical simulations set-up
Numerical simulations were carried out parallel to the experiments in order to provide
more detailed information on the vortical structures and their formation mechanisms
under forced and unforced conditions. A commercial solver (Ansys Fluent) was
used to simulate the unsteady, turbulent flow through incompressible LES using a
dynamic Smagorinsky subgrid-scale model with second-order accuracy for both spatial
and temporal discretizations. Figure 1(b) shows the simulated domain and applied
boundary conditions consisting of a rectangular box representing a part of the wind
tunnel test section and of the jet pipe. The main computational domain was 18Dj
long (streamwise x direction), 8Dj wide (spanwise y direction) and 6Dj tall (vertical z
398

U
(m s1 )
Ti
(%)
/Dj
/Dj
/Dj
1.6
<0.5
0.630
0.172
0.077
2.23
TABLE 1. Inlet boundary-layer characteristics at Xj = 6, Yj = 0.
direction) and counted 174 92 76 cells along the respective dimensions. The jet
exit centre was located 6Dj downstream from the domain inlet and the jet pipe domain
was 8Dj long to allow development of the artificial inlet boundary conditions. The jet
tube mesh consisted of an O-grid type mesh with 3060 cells in the cross-section and
150 cells along the pipe axis. The full mesh was structured and included overall two
million cells. The first cells in contact with solid walls were on average 0.03Dj tall in
the normal direction, resulting in y+ values below or close to unity.
The simulation process started with an initial Reynolds averaged NavierStokes
(RANS) simulation to initialize the flow domain and obtain a maximum viscous
dissipation rate () estimate in the domain. While the pressure, velocity and
temperature fields were used to initialize the LES runs, the viscous dissipation rate
was used to compute
the time step used in LES according to the Kolmogorov
time scale, K = (/). Thus in the current simulations, the time step varied from
5 104 s at BR = 0.150 to 5 105 s at BR = 4.3. To obtain statistically significant
data, several thousand iterations were performed for each case on high-performance
computing platforms at Louisiana State University (IBM Power 5+ at 1.9 GHz), using
an average of 24 processors per run. Velocity characteristics and boundary-layer
profiles for the inlet of the computational domain were obtained from experimental
hot-wire measurements performed in the wind tunnel as summarized in table 1, where
, and H correspond, respectively, to the boundary-layer displacement thickness,
momentum thickness and shape factor.
At the inlet of the jet pipe, uniform velocity was set so as to equal the volumetric
flow rate of the experiment. In the pulsed jet simulations, the velocity was modulated
by using the signal of the unsteady volumetric flow-rate measurement from the
flow meters during experiments. Spatial profile perturbations were provided using
the spectral synthesizer method implemented in Ansys Fluent with a characteristic
length scale equal to Dj and a cross-flow characteristic boundary-layer length scale
equal to 0.4. Based on the experimental measurements presented in table 1, a 0.5 %
perturbation level was imposed at the cross-flow inlet. Since the experimental set-up
did not allow measurements 8Dj inside the jet pipe, the perturbation levels at the
jet inlet were adjusted so that simulated and experimental perturbation levels at the
jet exit would match. The jet and test section walls were set as adiabatic. Periodic
boundary conditions were applied to the side boundaries of the test section box
and a standard convective outflow boundary condition was used on the outlet plane.
Finally, conversely to the experiment, the jet and cross-flow fluids were maintained
at constant temperatures of, respectively, 300 and 330 K similarly to what was done
in the experiments of Ekkad et al. (2006). The density difference associated with
temperature gradients was sufficiently low as not to affect the velocity field.
Figure 2 shows a comparison between experimental and predicted velocity
magnitude profiles at two streamwise locations, Xj = 1.0 and Xj = 3.5, in the
symmetry plane Yj = 0 for three values of the blowing ratio. Since experimental
399

(a)
(b)
(c)
2.0
2.0
2.0
1.5
1.5
1.5
1.0
1.0
1.0
0.5
0.5
0.5
F IGURE 2. Experimental (symbols) and simulated (solid line) velocity magnitude profiles
in the symmetry plane at Xj = 1 and Xj = 3.5 for: (a) BR = 0.150; (b) BR = 0.250; and
(c) BR = 0.465.
profiles were obtained using single-component hot-wire anemometry, no distinction

was made between streamwise and vertical velocity components, and thus velocity
magnitudes were chosen as a comparison basis. In addition, recirculatory flow present
near the wall at Xj = 1.0 was not measured because of hot-wire signal rectification,
and thus no experimental data are reported in this region. The velocity profiles in
figure 2 compare reasonably well at BR = 0.15 and BR = 0.465 for both streamwise
locations, while at BR = 0.25 the simulations appear to under-predict the velocity
magnitude in the jet wake. This result is consistent with a mismatch in the transition
behaviour between experimental and numerical data, which will be described later.
Overall, the agreement of the rudimentary mean flow comparisons between simulation
and experimental results is quite good and reinforces the qualitative and mechanistic
comparisons between the simulation results and the visualizations presented in the
following sections. Thus, both physical and simulated flows were comparable enough
to be used concurrently in interpreting and understanding fundamental processes.
3. Unforced jet
A rather extensive study of the unforced jet over a wide range of blowing ratios
from BR = 0.150 to BR = 4.3 was conducted as a baseline for the subsequent pulsed
jet study. Based on the characteristic vortical structures of the jet at various blowing
ratios, the two known attached and detached jet regimes were identified at blowing
ratios below 0.275 and above 0.6, respectively. The latter exhibited classic transverse
jet vortical structures (horseshoe vortex, ring shear-layer vortices, wake vortices). At
intermediate values of the blowing ratio (0.275 6 BR < 0.600) the transition cases
exhibited features from both types of flow.
3.1. Attached jet
Similarly to the detached configuration, the attached jet exhibits a set of characteristic
vortical structures resulting from the interaction between jet and cross-flow. In the
reactive Mie scattering visualizations of figure 3(a) at BR = 0.188, a recirculation
400

(a)
(b)
F IGURE 3. (a) Reactive Mie scattering visualization in the plane Yj = 0 at BR = 0.188.

(b) Fully reacted Mie scattering visualization in the plane Yj = 0 at BR = 0.188. H, horseshoe
vortex; Iv, inner vortex; Hp, hairpin vortices.
0
1
F IGURE 4. (Colour online) Time-averaged pressure field and streamlines in the vicinity of
the jet exit and at the bottom wall (Zj = 0) from LES at BR = 0.188. P+ , pressure iso-surface,
high-pressure region; P , pressure iso-surface, low-pressure region; H, horseshoe vortex; Iv,
inner vortex.
region located directly downstream of the jet exit is revealed by reacted seed and is
also evidenced in the LES results of figure 4 by a low-pressure region (P ). In the
time-averaged streamlines of figure 4, the recirculation region is mainly supplied by jet
fluid issuing from the sides of the jet tube, although a certain amount of cross-flow
fluid (moist air in the experiment) is entrained and reacts to generate the seed particles
in figure 3(a) in that region. In the experimental visualizations, the mixing layers
appearing as bright lines of reacted seed at both the leading and trailing edges depict
the presence of the respective shear layers near the jet exit. This assumption is only
valid in the vicinity of the jet exit, as the mass diffusivity of titanium dioxide particles
is typically two to three orders of magnitude lower than the momentum diffusivity of
air. At such low blowing ratios the jet trajectory is strongly diverted by the cross-flow
with significantly higher momentum, and both upper and lower shear layers merge on
average within the first one to two jet diameters downstream from the jet exit.
Figures 3(a) and 5 show the typical shear-layer structures found in the attached
jet configuration consisting of interlocked hairpin vortices (Hp). These are similar
to the hairpin structures described by Perry & Lim (1978) in co-flowing jets, or
the ones generated in the wake of a semicircular protuberance in Acarlar & Smith
(1987a) or through fluid injection from a high-aspect-ratio streamwise slit into a
laminar boundary layer in Acarlar & Smith (1987b) and Hagen & Kurosaka (1993).
Comparable vortical structures have also been previously observed in fully detached
transverse jets by Camussi, Guj & Stella (2002) at blowing ratios above 2.0 and by
Lim, New & Luo (2001) at BR = 1.0, though at much lower jet Reynolds numbers
(Rej = (U Dj )/) of, respectively, 100 and 350 relative to the current value of 2700.
These vortices are formed periodically in the lower shear layer with increasing
formation frequency as the blowing ratio is increased. The multiple self-induction
401

Hp
IHp
H
S
F IGURE 5. Reactive Mie scattering visualization in the plane Zj = 0.5 at BR = 0.188. Image
enhanced using CLAHE. H, horseshoe vortex (solid line); Hp, hairpin vortices; S, side
vortices (dash-dotted line); IHp, inverted hairpin vortices (dashed line).
(a)
(b)
(c)
Hp
S2
S1
z
y
x
Mirrored image
(d)
S1
S2
y
Mirrored image
Mirrored image
Hp
F IGURE 6. Induction mechanisms for hairpin vortices showing mirrored induction (hollow
arrows), mutual induction (solid arrows) and vortex line induction (dashed arrow): (a) side
view; (b) front view. Induction effects between hairpin (Hp) and side vortices (S1 and S2) and
their respective mirrored images: (c) front view; (d) top view.
and mirrored induction effects dictating the behaviour of the hairpin vortices as they
are convected downstream by the cross-flow are described in great detail in the work
of Acarlar & Smith (1987b) and are only mentioned here for the sake of clarity.
The schematics of figure 6(a,b) show how the mirrored induction with respect
to the bottom wall (figure 6a,b hollow arrows), causes the legs of the hairpin
vortices to converge towards the symmetry plane (Yj = 0) and tends to reduce the
convection velocity of the head, while the convective effect from the cross-flow
carries the structures downstream. Simultaneously, the mutual induction from the legs
(figure 6b solid arrows) generates a vertical velocity component, which is reinforced
by the converging effect of the mirrored induction as the legs approach each other.
Finally, the hairpin vortices are also subject to line vortex induction due to curvature,
which in the present case corresponds to vertical and a counter-streamwise component
(figure 6a dashed arrow).
A time series of instantaneous simulated flow fields at BR = 0.188 is presented
in figure 7. Iso-surfaces of the Laplacian of the pressure, with grey scale to reflect
streamwise vorticity levels, are generated to identify vortical structures based on the
402
(a) CH
(b) Hp3
(c)
(d)
Hp3
IHp
Hp2
Hp1
Hp2
IHp
Hp2
Hp3
2
IHp
Hp1
Hp1
S1
S2
IHp
Hp2
S1
Hp1
S1
S2
6
S2
8
10
2
F IGURE 7. An XY view of the Laplacian of the pressure iso-surfaces at 1P = 1.2 kPa m2

extracted from LES at BR = 0.188, using streamwise vorticity with grey scale from
black (negative) to white (positive). Successive time instants are such that: (a) tc = t0 ;
(b) tc = t0 + 2.2 (35 ms); (c) tc = t0 + 5.4 (85 ms); (d) tc = t0 + 8.5 (135 ms).
principle that vortex cores are associated with local pressure minima. This choice
was made based on the proportionality between the second invariant of the velocity
gradient tensor (Q ) and the Laplacian of the pressure in incompressible flows, and on
the equivalence between Q and the second eigenvalue (2 ) of the tensor S 2 + 2 as
described in Jeong & Hussain (1995), where S corresponds to the stress tensor and
to the rotation-rate tensor. The reported time scale is the convective time scale such
that tc = tU /Dj , with t the time in seconds. The hairpin vortices observed in the
experiments are also present in the simulations and denoted as Hp1, Hp2 and Hp3
in figure 7. Upstream from the hairpin vortices Hp1 and Hp2, the spatial separation
between the upper and lower jet shear layers is extremely low, therefore yielding
sharp velocity gradients and resulting in the formation of an inverted hairpin vortex
denoted as IHp. Compared to regular hairpin vortices, inverted ones have identical
y vorticity in the head but opposite x vorticity in the legs. Consequently, instead of
converging towards the symmetry plane, the legs of inverted hairpins tend to spread
away from it due to opposing induction effects from their respective mirrored images.
A comparable structure can be observed in figure 10 of the work of Zhou et al.
(1999) directly downstream of the secondary hairpin vortex (SHV), although this
structure is not discussed nor described by the authors. Because inverted hairpins
are formed near the wall as well as in between and below regular hairpins, their
presence in the Mie scattering visualizations can only be hinted at in figure 3(a)
by a rollup located directly beneath the most downstream hairpin vortex, and in
figure 5 where V-shaped structures precede the regular hairpin vortices. Finally,
secondary hairpin vortices are found to develop upstream of the regular one as seen in
figure 7(bd) directly upstream and on top of Hp1 in agreement with the observations

(a)
CH
403
(b)
S
y
x
F IGURE 8. (Colour online) (a) Laplacian of the pressure iso-surface at 1P = 1 kPa m2 and
local velocity profiles with spanwise and vertical components only, in the vicinity of the jet
exit from LES at BR = 0.188. H, horseshoe vortex; Iv, inner vortex; CH, counter-horseshoe
vortex; Hp, hairpin vortex; S, side vortex. (b) Authors representation of the interlocked
shear-layer structures in attached jet configuration, connected to the inner vortex.
of Acarlar & Smith (1987a). Overall the dynamics of the hairpin vortices observed in
the simulations was found to be consistent with the one described in the experiments.
In figures 3 and 5, a stable horseshoe vortex (H ) is visible above and on the leading
edge of the jet exit. This vortical system is the one described by Fric & Roshko
(1994) and Kelso & Smits (1995), resulting from the cross-flow boundary-layer
separation due to the jet blockage effect. Relatively small at the lowest blowing ratios
(approximately 0.19Dj in radius at BR = 0.15), the primary vortex becomes larger in
diameter with increasing values of the blowing ratio (up to 0.26Dj at BR = 0.275)
while being pushed further upstream from the leading edge of the jet. Although only
a single rollup is clearly visible in figures 3 and 5, at least one other small-scale
rollup was intermittently seeded and observed upstream of the first one. The LES in
figure 7 confirms the presence and location of the horseshoe vortex at BR = 0.188.
In figure 3, the horseshoe vortex is seeded with jet tracer particles entrained from
the upstream lip of the jet into the horseshoe vortex system, suggesting transport of
jet fluid by the structure. This is in agreement with the time-averaged LES results of
figure 4 showing streamlines exiting from the jet into the horseshoe vortex system.
Additional streamwise vortices were also found in both Mie scattering visualizations
and numerical simulations. In figure 5 these side vortices (S) are located on the
outside of the horseshoe vortex and appear to interact downstream with the shear-layer
vortices. In figure 7, the side vortices have opposite streamwise vorticity compared to
the horseshoe vortex and are denoted by CH near the jet exit, as counter-horseshoe
vortices.
The origins of these secondary vortices are explained in figure 8(a) by a strong
positive spanwise (or locally radial outwards) velocity gradient in the vertical (wallnormal) direction due to jet fluid injected around the jet hole, as well as cross-flow
deflection imposed by the jet blockage. Conversely to previous studies at higher
blowing ratio (e.g. Kelso & Smits 1995), describing complex horseshoe vortex systems
with multiple positive and negative horseshoe vortices ahead of the jet exit, no
evidence of negative spanwise vorticity was found upstream of the jet exit. This
suggests that the current counter-horseshoe vortex is not the result of cross-flow
boundary-layer separation due to adverse pressure gradient at the jet leading edge, but
rather is the result of cross-flow deflection and induced vertical velocity gradients as
presented in figure 8(a). Further downstream, as shown in figure 7, both horseshoe and
counter-horseshoe vortices interact with the shear-layer structures to form, respectively,
404
positive and negative side vortices S1 and S2 (named after their streamwise vorticity
sign in the first space quadrant). The relatively stronger opposite-sign vorticity from
the legs of the hairpin vortices tends to weaken the negative side vortices, while
the positive ones are strengthened by the cross-flow inrush effect generated by the
hairpin legs as described by Acarlar & Smith (1987b). Owing to their closeness to the
wall and their opposite vorticity, the two types of side vortex behave differently as they
are convected.
Figure 6(c) illustrates first-order mutual induction effects as well as mirrored
induction effects. On the one hand, the negative side vortex (S1) has a mirrored
induction motion towards the symmetry plane, whereas its interaction with the positive
side vortex (S2) tends to entrain it away from the wall. On the other hand, the positive
side-vortex induced velocity entrains it away from the symmetry plane, while its
interactions with both the negative side vortex and the hairpin vortex have an opposite
yet weaker influence, such that the positive side vortex tends to have a downward
motion. Thus, as they are convected downstream, the negative side vortices will have
a tendency to ride on top of the positive ones while converging towards the symmetry
plane, whereas the latter will diverge from the symmetry plane. The different relative
motions of the side vortices result in X-patterned structures lying on each side
of the jet core, which can be observed in both the Mie scattering visualizations of
figure 5 and the LES visualizations in figure 7, and are illustrated in the schematic of
figure 6(d).
Among the few low-blowing-ratio studies of the past, Gopalan et al. (2004)
suggested that, at low blowing ratios, the spanwise vorticity from the boundary layer
resulting in the formation of the horseshoe vortex interacted with the jet upper shearlayer vorticity and ultimately cancelled it so that, conversely to the higher-blowingratio cases, no vortical structure was formed in the jet upper shear layer. However, in
the current study, the side visualizations of figure 3(b) realized in fully reacted Mie
scattering experiments show the presence of a stable rollup of reacted seed, or inner
vortex (Iv), resulting from jet flow separation inside the pipe and located at the leading
edge. The inner vortex is prevented from shedding by the blockage exerted by the
cross-flow on the jet as suggested by the relatively advanced position of the horseshoe
vortex. Accordingly, in figure 4, a high-pressure region due to cross-flow deflection or
pressure cap is located upstream and above the jet exit and forces part of the jet fluid
exiting from the pipe to roll up at the jet leading edge. The time-averaged streamlines
show that jet fluid from the upstream region of the jet exits on the sides of the jet pipe
as well as upstream, underneath the primary horseshoe vortex. The closed vortex line
in figure 8(a) (grey line) also shows that the inner vortex and the consecutive hairpin
vortices form a series of interlocked rings illustrated schematically in figure 8(b). This
vortical structure is to be related to the leading-edge rollup observed in the LES
of Guo, Schroder & Meinke (2006) or the hovering vortex of Kelso et al. (1996),
although none of these has been documented to remain stable inside the jet pipe. To
the authors knowledge, this constitutes the first observation of such stable vortical
structure located inside the jet pipe for round transverse jets.
In the time-averaged streamwise vorticity field of figure 9, a pair of streamwise
counter-rotating vortices is located on each side of the symmetry plane (filled arrows).
This result is a common feature in transverse jets and is usually identified as the
CRVP. According to the instantaneous observations, the vorticity of this vortex pair
is generated by the legs of the hairpin vortices. Initially observed in the average flow
field of the detached jet, the CRVP has also been observed in the instantaneous flow
field in Rivero et al. (2001). It could be tempting to directly interpret the vortex pair
405
5%
5%
5%
5%
5%
5%
5%
10
8
6
4
2
F IGURE 9. Time-averaged streamwise vorticity (filled) and normalized temperature contours

by increments of 15 % (line) from LES at BR = 0.188.
found in figure 9 as the usual CRVP, although a few considerations lead us to do

so with caution. First, the perceptible continuity of the vortex pair in the average
vorticity field of figure 9 is in fact induced by the downstream convection of the
hairpin shear-layer vortices. Instantaneously though, this vortex pair is discontinuous,
as seen in figure 7(d) near Xj = 9, where no particular vortex pair is found. This
constitutes a first difference from the usual CRVP structure, which, although it can be
asymmetric in strength with respect to the symmetry plane (Smith & Mungal 1998;
Rivero et al. 2001), has never been documented as discontinuous. Furthermore in
figure 9 at Xj = 12, the vorticity in the vortex pair appears twice as low as the one
in the average side vortices found near the wall (hollow arrows), which is somewhat
in contradiction with the fundamental notion that the CRVP is the dominant structure
in the far field of transverse jets. This last point is supported by the distribution
of the passive scalar concentration (in our case the normalized temperature field
( = (T T )/(Tj T )), which is highly distorted near the wall in an inverted
T-shape. This is fundamentally different from the typical kidney-shape distribution in
the jet exit vicinity or the circular distribution in the far field encountered in highblowing-ratio configurations. It also reveals that the side vortices are predominantly
participating in the scalar quantity transport. Therefore, although it appears similar to
the familiar structure found in most high-blowing-ratio studies, the streamwise vortex
pair present in the average vorticity field of figure 9 exhibits significant differences
when compared to the traditional detached jet CRVP.
At blowing ratios in the range 0.225 6 BR 6 0.275, the vortical structures are
overall identical to the one found at lower blowing ratios, with the exception of
random horseshoe vortex transports on top of the jet upper shear layer as evidenced
in figure 10. Originally located upstream of the jet exit in figure 10(a), the horseshoe
vortex is shown to ride on top of the jet/cross-flow interface in figure 10(b). This
destabilization is probably provoked by an interaction between the horseshoe vortex
and the inner vortex, which is partially outside the jet pipe in figure 10(b). Once
detached, the horseshoe vortex is convected downstream, where it eventually merges
with hairpin vortices of identical spanwise vorticity sign. Although not immediately
406

(a)
(b)
(c)
(d )
F IGURE 10. Mie scattering visualizations in the plane Yj = 0 for BR = 0.25 at: (a) tc = t0 ;
(b) tc = t0 + 6 (95 ms); (c) tc = t0 + 7.9 (125 ms); (d) tc = t0 + 9.5 (150 ms). Hn, horseshoe
vortex number n; Iv, inner vortex.
(a) 1
(b)
(c) 1
(d)
400
0
1
F IGURE 11. Instantaneous spanwise vorticity XZ slices and streamlines at BR = 0.25. Time
sequence (a)(d) corresponding to figure 10. Hn, hairpin vortex number n; Iv, inner vortex.
seeded and entirely visible, a second horseshoe vortex very quickly replaces the
convected one in the cross-flow boundary layer upstream of the jet, as seen in
figure 10(d). This transport is also occurring in the simulation results of figure 11
at identical blowing ratio, where a second horseshoe vortex H2 is instantaneously
formed while the first one, H1, is transported downstream. The shedding, as it affects
the pressure distribution at the jet exit, also affects the inner vortex, which can be seen
oscillating inside the jet pipe in both experimental and numerical visualizations. In the
numerical results of the sequence shown in figure 11, the transported horseshoe vortex
(figure 11a) loses its coherence as it is convected and diffused in the jet upper shear
layer. In the experimental Mie scattering visualizations, however, the shed vortices
were found to stay more coherent and visibly transported over the jet.
3.2. Transitional jet
Although the horseshoe vortex transport constitutes an early sign of the jet
transitioning from the low-blowing-ratio regime to the high-blowing-ratio one, the
majority of the vortical structures are still analogous to those of the attached jet
configuration. For blowing ratios between 0.275 and 0.6, however, the jet behaviour
is significantly modified as the previously stable inner vortex starts to shed. A typical
407

(a)
(b)
(c)
(d)
F IGURE 12. Fully reacted Mie scattering visualizations at Yj = 0 and BR = 0.365: (a) tc = t0 ;
(b) tc = t0 + 6 (95 ms); (c) tc = t0 + 7.9 (125 ms); (d) tc = t0 + 9.8 (155 ms). Hn, horseshoe
vortex number n; Iv, inner vortex; Hp, hairpin vortices.
(a)
(b)
(e)
0
1
0
1
(c)
(d)
1
0
1
F IGURE 13. Laplacian of the pressure iso-surfaces at (ad) 1P = 10 kPa m2 and (e) 1P =
2.5 kPa m2 from LES at BR = 0.465 using streamwise vorticity with grey scale from
black (negative) to white (positive). Time stamps: (a) tc = t0 ; (b) tc = t0 + 1.6 (25 ms);
(c,e) tc = t0 + 2.2 (35 ms); (d) tc = t0 + 3.5 (55 ms). Hn, hairpin vortex number n; Rvn,
ring vortex number n; Ivn, inner vortex number n.
convection sequence of this vortical structure is presented in figure 12. The inner
vortex (Iv), still partly inside the jet tube in figure 12(a), is convected in figure 12(b),
which is followed by an ingestion of cross-flow fluid inside the jet pipe at the leading
edge. The perturbation of the jet/cross-flow interface also triggers the transport of the
horseshoe vortex (H1) in figure 12(c), followed by the formation of a large hairpin
vortex (Hp) in figure 12(d). A second horseshoe vortex (H2) replaces the convected
one ahead of the jet exit almost instantaneously. At values of the blowing ratio
approaching 0.6, the size of the transported inner vortex becomes comparable to the
downstream shear-layer vortices, and the ingestion of cross-flow fluid that follows the
inner vortex transport is gradually decreased until it completely disappears beyond
BR = 0.6.
The LES results at BR = 0.465 presented in figure 13(ad) show inner vortex
transport sequences consistent with the experimental observations of figure 12. In
addition, the simulations show a clear connection between the inner vortex and the
downstream hairpin vortices, thus forming a ring vortex (Rv1), a characteristic vortical
structure encountered in higher-blowing-ratio regimes. Once shed, the upstream rollup
408

5%
5%
5%
5%
5%
5%
10
5%
8
6
4
2
F IGURE 14. Time-averaged streamwise vorticity (filled) and normalized temperature

contours by increments of 15 % (line) from LES at BR = 0.465.
is folded and quickly dissipated in the vicinity of the jet exit by the overall positive
spanwise vorticity of the shear layers. After the inner vortex (or ring vortex Rv1) is
transported, the horseshoe vortex (H1) is subsequently destabilized and convected (e.g.
figure 13c,e) and a second pair of inner (Iv2) and horseshoe (H2) vortices is formed,
respectively, inside the jet pipe and ahead of the jet exit. Hairpin vortices, typical of
lower-blowing-ratio jets, are also found in figure 13(e) (Hp) downstream of the ring
vortices.
The average streamwise vorticity field in figure 14 shows a decrease in the strength
of the side vortices (hollow arrows) near the wall compared to the principal vortex
pair (solid arrows). Consequently, the normalized temperature (scalar) concentration
distribution is clearly influenced by the vortex pair and becomes more comparable
with the typical distribution of transverse jets at higher blowing ratios (see Smith &
Mungal 1998). It should be noted that the inner vortex (Iv) in the experiments was
destabilized and convected at blowing ratios above 0.275, while similar sequences
were clearly observed in the LES at BR > 0.415. This discrepancy was attributed to
inconsistencies in the jet and cross-flow inlet fluctuation spectra between experiment
and simulation, to which the interaction between horseshoe vortex and inner vortex is
expected to be highly sensitive.
3.3. Fully detached jet
In the current study, cases with blowing ratios above 0.6 exhibited completely
detached jets, as shown in figure 15. This threshold value was found to be consistent
with previous studies, such as Harrington et al. (2001). In figure 15, shear-layer
vortices (Rv) appearing as well-defined rollups on both the upper and lower jet/crossflow interfaces shed periodically while wake vortices (Wv) located under the jet core
are evidenced by seed particles, suggesting that these structures transport jet fluid.
In figure 16, the LES at BR = 1.00 show that, on the downstream side of the jet,
the shear layer rolls up outside the jet pipe, whereas the upstream part rolls inside
it. The different pressure conditions at the jet exit on the windward (adverse pressure
gradient) and leeward (favourable pressure gradient) sides are in part responsible for
409
F IGURE 15. Fully reacted Mie scattering visualization in the plane Yj = 0 at BR = 1.5. Rv,
ring vortices; Wv, wake vortices.
(a)
(b)
1
0
0
1
2 1
(c)
1
2 1
2 1
(d)
1
0
0
1
2 1
F IGURE 16. (Colour online) Laplacian of the pressure iso-surfaces at 1P = 20 kPa m2 and
streamlines extracted from LES at BR = 1.0 using spanwise vorticity with grey scale from
black (negative) to white (positive). Contour slices show streamwise vorticity from pale grey
(blue online; negative) to dark grey (red online; positive). Successive time instants such that:
(a) tc = t0 ; (b) tc = t0 + 0.63 (10 ms); (c) tc = t0 + 1.25 (20 ms); (d) tc = t0 + 2.9 (30 ms). H,
hairpin vortex; Rvn, ring vortex number n; Wv, wake vortex.
this difference. The dynamics of the ring vortices observed in figure 16 is comparable
to that described by Kelso et al. (1996). After shedding, the vortex rings are folded
in such a way that the lower and upper rollups approach each other, thus increasing
their induced upward velocity, while the connecting side arms are stretched in the
downward direction. The horseshoe vortex, invisible because unseeded in the Mie
scattering visualizations, is located upstream of the jet exit and oscillates back and
forth due to the fluctuating pressure field associated with the shedding of shear-layer
vortices, in agreement with the results from Kelso & Smits (1995). The streamlines
in figure 16 show the formation of wall and wake vortices from the cross-flow
410

5%
5%
5%
5%
5%
5%
10
5%
8
6
4
2
F IGURE 17. Time-averaged streamwise vorticity (filled) and contours by increments of

15 % (line) from LES at BR = 1.0.
boundary-layer separation around the jet exit and entrainment by the side arms of
the shear-layer structures. A pair of vortex tubes joining the individual vortex rings
and consistent with the CRVP is also identified in the instantaneous Laplacian of the
pressure iso-surfaces. An initialization mechanism for this structure is described in the
following sections. In figure 17, the average streamwise vorticity field exhibits the
typical kidney-shaped distribution in the near field of the jet exit. In the far field, the
counter-rotating vortex pair appears to be the predominant feature in terms of both
size and strength, and accordingly the normalized temperature () distribution contours
are in good agreement with the previous results in detached jet experiments mentioned
earlier.
3.4. Characteristic frequency modes
Single-component hot-wire measurements were performed at chosen locations selected
based on the experimental visualizations in order to identify frequency modes
associated with the previously described structures. Measurements were carried out
at the jet exit, in the upper shear layer above the jet exit, and at a downstream location
(Xj = 3.5). The velocity signals were subjected to both Fourier and wavelet spectral
analyses, the latter between 0 and 50 Hz (corresponding to Strouhal numbers St
between 0 and 0.794), guided by the results from the former. The results are presented
in figure 18, summarizing the fundamental frequencies found at each location, scaled
by jet diameter and mean jet velocity (St j = fDj /Uj ) in figure 18(a) and scaled by jet
diameter and cross-flow velocity (St = fDj /U ) in figure 18(b).
In figure 19 wavelet analysis mappings show that measurements performed at the jet
exit (and above) revealed no particular frequencies for BR 6 0.2 due to the weakness
of the shear-layer instability at this location for low blowing ratios. On the other hand,
measurements performed at Xj = 3.5 exhibited no clear signature for BR > 0.3. Results
from the attached jet regime show an increase in Strouhal number with increasing
blowing ratio, corresponding to the rate of formation of hairpin vortices. Similarly,
411

(a)
(b)
1.5
0.6
1.0
0.4
0.5
0.2
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
1.0
F IGURE 18. Fundamental frequencies scaled using: (a) time-averaged jet velocity; (b) crossflow velocity. In the experiment these were identified from hot-wire records using wavelet
analysis. Symbols: , jet exit (Xj = 0, Zj = 0); O, location (Xj = 0, Zj = 0.25), and H, location
(Xj = 0, Zj = 0.5), both inside the upper shear layer; , fundamental, and , its subharmonic
from downstream location (Xj = 3.5, Zj = 0.75); , location (Xj = 3.5, Zj = 1.25); ,
characteristic frequencies from the LES at the centre of the jet exit (Xj = 0, Zj = 0).
results from the detached jet at BR = 0.6 show a nearly linear rate of increase in
Strouhal number with blowing ratio at both locations, corresponding to the formation
rate of ring vortices. The simulation results show an identical trend but with overall
lower frequencies and a slightly lower rate of increase. This is in agreement with
previous results reported by Megerian et al. (2007) for BR < 3.5, even though their
study showed Strouhal numbers increasing at a rate of 1/5 while in the current study
the rate of increase is approximately 2/5. However, it should be noted that, in the
previously cited study, stronger jets with top-hat exit profiles were involved, whereas
in the current investigation the jets were weaker, with larger shear-layer momentum
thicknesses. Fundamental frequencies obtained for the attached jet appear to be
significantly higher than equivalent signatures obtained at higher blowing ratios. This
is justified by the fact that data for the first regime were obtained at Xj = 3.5 near the
wall, where measurements were influenced by the passage of inverted hairpin vortices
formed in between primary ones, thus doubling the apparent passage frequency of the
primary structures.
To provide a more relevant comparison base, the corresponding sub-harmonics for
those measurements (also found in the wavelet records but with lower energy content)
are also provided in figure 18 (black diamonds) and show better agreement with
characteristic frequencies at higher blowing ratios. The usefulness of the wavelet
analysis in steady state was found in the transition regime, where events (or series of
events) have a tendency to occur rather intermittently, as seen in figure 19(c), making
the identification of characteristic frequencies difficult with Fourier spectrum analysis.
Overall the value of St appears relatively constant around approximately 0.2 over
the range 0.225 < BR < 0.425 while St j decreases continuously from 0.78 to 0.32 at
the jet exit. This indicates that the dominant features, in this case the transports of
the horseshoe and the inner vortices, are better scaled by the cross-flow velocity and
thus suggests that the transport mechanism is initiated by the horseshoe vortex. On
the other hand, for 0.425 6 BR < 0.6, St j values tend to reach saturation while St
412
(a)

4.0
4.5
5.0
5.5
6.0
0.6
0.6
0.4
0.4
0.2
0.2
5.75
4.5
6.00
5.0
6.25
6.50
5.5
6.75
7.00
6.0
7.25
0.6
0.4
7.75
8.00
8.25
8.50
8.75
(d) 16.25
16.50
16.75
17.00
17.25
0.6
6.00
6.25
11.00
6.50
6.75
11.25
7.00
11.50
7.25
0.2
16.25
16.50
16.75
17.00
17.25
11.75
( f ) 6.00
6.25
6.50
6.75
7.00
6.50
6.75
7.00
0.6
0.6
0.4
0.4
0.2
0.2
1.0
1.0
0.8
0.5
0.6
10.75
8.75
0.2
0.4
10.75
8.50
0.4
0.2
(e)
8.25
0.6
0.4
0.2
5.75
8.00
0.6
0.1
4.0
7.75
0.8
0.2
(c)
(b)
11.00
11.25
Time (s)
11.50
11.75
0
6.00
6.25
Time (s)
F IGURE 19. Wavelet analysis mapping of hot-wire time records in the plane Yj = 0 for:
(a) BR = 0.188, Xj = 0, Zj = 0; (b) BR = 0.188, Xj = 3.5, Zj = 0.75; (c) BR = 0.3, Xj = 0,
Zj = 0; (d) BR = 0.3, Xj = 3.5, Zj = 1.0; (e) BR = 0.6, Xj = 0, Zj = 0; (f ) BR = 0.6, Xj = 3.5,
Zj = 1.5.
is clearly increasing, indicating a shift in the transport initiating structure from the
horseshoe to the inner vortex.
4. Forced jet
4.1. System characterization and preliminary observations
Forced jet experiments were performed using a nominal square-wave actuation signal
for the solenoid valve shown in figure 1. A wide range of forcing parameter values for

Case no.
c
1
2
3c
4c
5
6
7a
8a
9a,c
10b,c
11b,c
BRm
BRl
BRh
BRpp
DC ( %)
0.25
0.25
0.35
0.35
0.35
0.35
0.75
1.25
1.25
0.365
1.275
0.188
0.075
0.188
0.188
0.225
0.175
0.200
0.200
0.200
0.200
0.700
0.438
0.325
0.836
0.513
0.475
0.425
2.20
4.20
2.20
2.20
3.00
0.250
0.250
0.648
0.325
0.250
0.250
2.00
4.00
2.00
2.00
2.30
25
70
25
50
50
70
25
25
50
8
25
413
TABLE 2. Forced case parameters: a no flow-meter record available; b no experiments at

these conditions; and c simulations carried at these conditions.
BRm , BRl , BRh , BRpp and DC (respectively, mean, low, high and peak-to-peak blowing
ratios over a cycle, and duty cycle) was covered, a part of which is presented in more
detail herein and summarized in table 2. Each individual case was observed at four
distinct forcing frequencies (ff ) of 0.5, 1.0, 5.0 and 10.0 Hz, corresponding to Strouhal
numbers of St = 0.008, 0.016, 0.079 and 0.159, respectively. These frequencies
were selected to stay below the shear-layer natural frequencies of the unforced jet
(see figure 19b) so as to minimize the potential of direct resonant amplification
(see Nikitopoulos et al. 2006). However, resonance with sub-harmonics and other
lower harmonics giving rise to stronger nonlinear effects (see Nikitopoulos & Liu
1987, 2001) cannot be ruled out.
Phase-locked Mie scattering visualizations supplemented by hot-wire anemometry
were used to document the forced cases. Both visualization planes for every case
were documented using 50 phase-locked positions at St = 0.008 and 0.016, and
10 such positions at St = 0.079 and 0.159, each position being acquired over
10 distinct cycles. The use of time-resolved flow-meter measurements allowed for
precise verification of the forcing conditions. Typical flow-rate profiles from case 4
and corresponding jet exit velocity magnitudes are presented in figure 20. These
records attest to the good conservation of the square-wave features such as the duty
cycle and the peak-to-peak values, if one excludes the initial transients from the low
to the high part of the cycle and vice versa. Calibrated duty cycle values were used
to account for valve response delays at higher frequencies, so that effective duty
cycles of 50 % were obtained with actuation DC of 40 and 30 % at, respectively,
St = 0.079 and 0.159. An oscillatory signature consistent with volumetric resonance
was observed in the flow rate at the valve opening and closing with a constant
frequency of St = 0.75 (47.5 Hz) and an exponential decay rate regardless of the
forcing parameters. This frequency was well above the jet shear-layer characteristic
frequencies for BR 6 1 so that no significant direct interaction was expected with the
natural jet modes. The phase-averaged value of the first blowing-ratio overshoot (BRos )
scaled relatively well using the bilinear dependence on BRl , BRh and BRpp , such as
BRos = 1.5BRh 0.5BRl = BRh + 0.5BRpp , as evidenced in figure 20(c).
The oscillating time in the high-blowing-ratio part was almost constant over the
considered range of blowing ratios, with a small sensitivity to the BRh value. The
average e-folding time was 24 3 ms and the average 99 % attenuation time was
414
(a)
(c)
1
0.6
0.4
0.2
(b)
0
8.4
8.6
8.8
0.8
0.6
0.8
1
0.6
0.4
0.2
0
1.0
9.0
0
8.2
8.3
Valve signal
Valve signal
0.8
0.4
0.4
0.6
0.8
1.0
Time (s)
F IGURE 20. (a,b) Typical flow rate (grey line), vertical velocity at the jet exit (black line) and
valve signal (dashed line) records in forced experiments corresponding to case 4 at forcing
frequencies of (a) St = 0.016 and (b) St = 0.079. (c) Overshoot scaling.
110 15 ms, shorter times being achieved at higher values of BRh . Oscillations in
the low part of the cycle after the valve closing had an average e-folding time of
41.3 7 ms and a 99 % attenuation time of 192 32 ms. Because the forcing cycle
periods in this study ranged from 2 s down to 100 ms, cases at St = 0.008 and
0.016 had cycle periods much longer than the transient time scales, whereas cases
at St = 0.079 and 0.159 had comparable or shorter cycle periods. From these basic
preliminary considerations, the jet was expected to behave differently at low forcing
frequencies compared to higher ones, where the transient dynamics introduced by the
forcing would dominate the jet behaviour. These observations are supported by the jet
exit velocity records of figure 20(a) at St = 0.016, showing distinct behaviour after
the transient period of the high part of the cycle that are not found in figure 20(b)
at St = 0.079. Finally, in view of the large number of parameters attached to partly
modulated transverse jets, the use of the cross-flow Strouhal number St defined in 3
was preferred over the parameter St j , necessitating the identification of characteristic
velocity scales for each forced case. In the rest of this paper, the dimensionless
time t = tff will be used to report instants inside a cycle. Using this parameter, the
transition from low to high blowing ratio occurs at t = 0 while the transition from
high to low blowing ratio occurs at t = DC.
The study of the Mie scattering visualizations leads to a qualitative identification
of the different vortical structures generated during a forced cycle. As expected, the
jet behaviour was found to be different at forcing Strouhal numbers of St = 0.008
and 0.016 compared to higher forcing Strouhal numbers (St = 0.079 and 0.159).
Figure 21 presents visualizations from case 3 at forcing frequencies of St = 0.016
(top) and St = 0.159 (bottom) at identical phase-locked positions. In figure 21(ad)
the forced case at St = 0.016 exhibits four distinct phases: a transition from low
to high blowing ratio with starting vortices (figure 21a); a quasi-steady plateau in
the high-blowing-ratio part of the cycle with structures typically encountered in the
unforced jet at BR = BRh (figure 21b); a transition from high to low blowing ratio with
cross-flow ingestion inside the jet pipe (figure 21c); and finally a quasi-steady plateau

(a)
(b)
(c)
(d)
(e)
(f)
(g)
(h)
415
F IGURE 21. Mie scattering visualizations in the plane Yj = 0 for case 3: at St = 0.016 for
(a) t = 0.1, (b) t = 0.2, (c) t = DC + 0.025 and (d) t = DC + 0.63; and at St = 0.159 for
(e) t = 0.1, (f ) t = 0.2, (g) t = DC + 0.025 and (h) t = DC + 0.63.
in the low part of the cycle with typical structures encountered in the unforced jet at
BR = BRl (figure 21d). In figure 21(eh), however, only two overlapping phases are
found: a transition from low to high blowing ratio with the formation of a starting
structure (figure 21e,f ); and a transition from high to low blowing ratio with crossflow ingestion inside the jet pipe (figure 21g,h). These observations are in agreement
with the findings of Ligrani et al. (1996), who observed quasi-steady behaviour in
free stream bulk flow pulsation experiments for St < 2.4. In their study, however,
jet forcing was only passive and imposed through sinusoidal cross-flow pressure
fluctuations, which could explain the discrepancy in the limiting Strouhal number
values. Similarly, Johari et al. (1999) mention that in their experiment the jet behaves
as a portion of the steady jet in the far field for injection times greater than 300 ms.
Among the four possible phases encountered in a typical pulsed case, the quasiunforced behaviours in both high and low blowing ratios were found to be very similar
to the ones described in the previous section for unforced jets and therefore will not be
further discussed. The next subsection will then mainly focus on the transient phases
from low to high blowing ratio and vice versa.
4.2. Low to high blowing-ratio transition: starting vortices
At the transition from low to high blowing ratio, a starting vortex is formed at
the edge of the jet exit due to the higher shear generated across the jet/cross-flow
interface by the bulk flow-rate increase. These starting structures have been observed
and described in many pulsed jets in cross-flow experiments and simulations for their
potential in mixing and penetration improvement, though generally at higher average
blowing ratios and forcing frequencies than in the current study.
Over the range of forcing parameters covered in the present study, two principal
types of starting vortices were observed: vortex rings formed at rather high values
of BRh (figure 22a,b, and hairpin vortices very similar to the one observed under
unforced conditions for lower values of BRh (figure 22c,d). As in Johari (2006) and
Sau & Mahesh (2008), the types of starting vortex regimes could be divided into
two sub-categories depending on the pulse width ( = DC/ff ) corresponding to
the duration of the high part of the cycle. At low values only a single starting
416

(a)
(b)
(c)
(d)
F IGURE 22. Mie scattering visualizations in the plane Yj = 0 for: (a) case 3 at St = 0.159,
single starting vortex ring; (b) case 3 at St = 0.016, leading starting vortex ring and
trailing vortices; (c) case 1 at St = 0.159, single starting hairpin vortex; and (d) case 2 at
St = 0.079, leading starting hairpin vortex and trailing hairpin vortices. Sv, starting vortex;
Tv, trailing vortices.
(a)
(b)
(c)
(d)
F IGURE 23. Mie scattering visualizations in the planes (a,b) Yj = 0 and (c,d) Zj = 0.125 for
case 1 at St = 0.016: (a,c) t = 0 ; (b,d) t = 0.1.
vortex (Sv) was generated (figure 22a,c), while for longer pulse widths, a trailing
column composed of smaller vortices (Tv) was formed in the wake of the principal
starting vortex (Sv) (figure 22b,d). These observations are in good agreement with the
results presented in previously mentioned work, although trailing vortices in the case
of starting hairpin vortices have not been clearly documented before. The dynamics of
the different starting vortices is described and compared to the DNS results from Sau
& Mahesh (2008) in the following subsections.
4.2.1. Starting hairpin vortex
As mentioned previously, the starting hairpin vortices are in essence identical
to those extensively described in steady state. In several cases, such as case 1
presented in figure 23, starting hairpin vortices were generated even though the BRh

(a)
417
(c)
(b)
(d)
F IGURE 24. (a,b) Laplacian of the pressure iso-surfaces at 1P = 15 kPa m2 and streamlines
for case 4 at St = 0.016: (a) t = 0.08, starting hairpin vortex; (b) t = DC + 0.02, crossflow ingestion. (c,d) Laplacian of the pressure iso-surfaces and streamlines for case 1 at
St = 0.159 for: (c) 1P = 25 kPa m2 at t = 0.15; (d) 1P = 20 kPa m2 at t = 0.4.
value (BRh = 0.438 in case 1) was significantly higher than the unforced jet
transitional blowing-ratio threshold (BRtr = 0.275). In figure 22(c), the inner vortex
can be seen partially out of the jet pipe without yet being convected, and eventually
pushed back inside for the rest of the cycle. The mechanisms responsible for
this behaviour will be described later with the aid of LES results. Figure 23(a,b),
respectively before and after the transition from BRl to BRh , shows that starting hairpin
vortices are considerably larger than the ones observed in steady or quasi-steady
state and penetrate deeper into the cross-flow. The horseshoe vortex diameter is also
significantly increased after the transition between figure 23(a,c) and figure 23(b,d).
The side vortices are entrained in the wake of the leading starting hairpin vortex and
reoriented in the vertical direction in a similar manner to the formation mechanism of
the tornado-like wake vortices described by Fric & Roshko (1994).
A case with higher BRh value corresponding to case 4 was investigated both
experimentally and numerically. Qualitative comparison between LES results and
experimental visualizations can be found in Bidan & Nikitopoulos (2011). The
instantaneous streamlines in figure 24(a) issuing from both the jet and the cross-flow
show significant flow entrainment near the starting hairpin vortex legs, in a pattern
comparable to the one observed in figure 23(d). A quasi-streamwise vortex (Wv) is
formed in the wake of the starting vortex and is reoriented in the vertical direction as
the starting hairpin vortex is convected, which is also consistent with the experimental
observations.
As pointed out earlier, starting hairpin vortices are formed for a certain period
of time after the high-blowing-ratio onset, although the value of BRh is higher than
the transitional blowing ratio found in unforced conditions. The absence of transport
of the inner vortex leading to the formation of a starting hairpin vortex rather than
a complete ring vortex is explained in the two-dimensional streamline visualizations
from the simulations of case 4 at St = 0.016 in figure 25. Upon increase in blowing
ratio, the inner vortex is pushed outside and upstream from the jet pipe, as seen in
figure 25(b), repelling the horseshoe vortex further upstream compared to its previous
location (also supported by experimental observations in figure 23d). In figure 25(c),
as the initial overshoot decays, the inner vortex weakens due to its interaction with the
418

(a)
(b)
(c)
(d )
(e)
(f)
0.5
0.5
0.5
0.5
1.5
1.0
0.5
1.5
1.0
0.5
0 1.5
1.0
0.5
F IGURE 25. Instantaneous pressure difference contours P Porigin and two-dimensional

streamlines for case 4 in the plane Yj = 0 at St = 0.016 at time instants: (a) t = 0 ;
(b) t = 0.017; (c) t = 0.065; (d) t = 0.265; (e) t = 0.33; (f ) t = 0.370.
horseshoe vortex of opposite spanwise vorticity and is pushed back inside the pipe by
the cross-flow momentum and induction effects from both the horseshoe vortex and its
image vortex. The increase in jet flow rate results in an increased blockage onto the
cross-flow, which leads to an augmentation of the primary horseshoe vortex diameter
(also observed in figure 23b) as well as an extension of the boundary-layer separation
and the formation of a secondary horseshoe vortex upstream of the previous one.
With the overshoot decay, the horseshoe vortex system relocates downstream, closer
to the jet leading edge although further above the wall, thus causing the entrainment
of jet fluid in the upstream region of the jet exit. In figure 25(d,e) as the inner
vortex strengthens again, the induction between the two structures increases and the
horseshoe vortex is pulled closer to the wall, interrupting the flow of jet fluid to
the upstream region of the jet exit. Finally, as quasi-unforced regime is reached in
figure 25(f ), the inner vortex is convected in agreement with corresponding unforced
cases at BR = BRh , entraining the horseshoe vortex in its wake while the upstream part
of the shear layer rolls up inside the pipe to form the next inner vortex.
4.2.2. Starting single vortex ring
In the remainder of this paper, the tilting directions of vortex rings will be described
with respect to rotation around the y axis of the vertical unit normal vector to the
vortex ring plane. Using this convention, negative rotation around the y axis will be
described as upstream rotation, and positive rotation described as downstream rotation.
In figure 21(eh) and figure 22(a), both corresponding to case 3, a single starting
vortex ring is formed. The starting vortex tilts towards the upstream direction as it
is convected downstream and away from the wall. This is in good agreement with
observations from both Chang & Vakili (1995) and Sau & Mahesh (2008) in which
a tilting angle of 34 (compared to approximately 20 for case 3) was observed
and attributed to a KuttaJoukowski lift effect exerted by the cross-flow on the ring.
419

(a) 3
(b)
(c)
(d)
(e)
(f)
2
0
1
0
1
1
0
F IGURE 26. An XZ view of the Laplacian of the pressure iso-surfaces at 1P = 80 kPa m2

for case 10 at St = 0.159 for time instants: (a) t = 0.085; (b) t = 0.19; (c) t = 0.29;
(d) t = 0.39; (e) t = 0.49. The spanwise vorticity has grey scale from black (negative) to
white (positive). (f ) Jet exit velocity profiles at t = 0 (), t = 0.005 (), t = 0.075 ( ),
t = DC + 0.005 (4), and spanwise vorticity profile (grey) at t = 0.005.
This explanation was, however, refuted by experimental and numerical studies from,
respectively, Lim, Lua & Thet (2008) and Cheng, Lou & Lim (2009), showing that the
tilt was more likely to be attributed to an interaction between the cross-flow boundary
layer and the upstream part of the vortex ring, and that the extent of the tilt was
related to their vorticity magnitude ratio. Sau & Mahesh (2008) also reported a size
imbalance between the upstream and downstream rollups attributed to the shearing
effect of the cross-flow on the structure. In the current study, size imbalance appeared
as well in the early stages of the vortex ring formation as observed in figure 21(f ),
while the vortex ring was still near the wall, but did not significantly increase as the
starting vortex departed from it.
Among the experiments monitored using a time-resolved flow meter, only case 3
at St = 0.159 exhibited a single vortex ring. This case was also simulated using
the experimental flow-rate records. The starting vortex formed in the simulations
is shown in figure 24(c,d). Although in figure 24(c) a vortex ring is formed, the
upstream rollup is not convected as fast as the downstream rollup and results in
the elongated starting vortex in figure 24(d). This was attributed to the same causes
responsible for the mismatch in the value of the transitional blowing ratio under
unforced conditions. To observe the dynamics of single starting vortex rings, a forcing
signal using an idealized square wave at St = 0.159 was used such that BRl = 0.200,
BRh = 2.20 and DC = 8 % (identified as case 10 in table 2). Iso-surfaces of the
pressure Laplacian corresponding to case 10 are presented in figure 26 at the transition
from low to high blowing ratio. The vortex ring dynamics agrees fairly well with
that found in the experimental visualizations of figure 21 and provides additional
information on the three-dimensional evolution of the structure. As the starting vortex
departs from the bottom wall (figure 26b), the structure acquires an initial tilt angle
towards the upstream direction, which appears to increase progressively as shown in
the frame sequence of figure 26. Under the action of the cross-flow, the vortex ring is
progressively folded such that the upstream and downstream parts approach each other
and the side arms are relatively dropping lower. The dynamics is comparable to the
folding mechanism of the shear-layer ring vortices presented in the steady detached
jet of figure 16. According to the previously mentioned studies, the rollup size
imbalance and the upstream tilting of the starting vortex are due to a weakening of the
windward part of the structure from an interaction with the incoming boundary layer
of opposite vorticity. However, the velocity profiles in figure 26(f ) show a small, yet
consistent momentum and vorticity deficit at the leading edge compared to the trailing
edge throughout the starting vortex formation process. This observation suggests that
420
(a)

(b)
(c)
(d)
(e)
F IGURE 27. Mie scattering visualizations in the plane Yj = 0 for case 8 at St = 0.079,
showing leading vortex ring and trailing column formed at the jet transition from BRl to BRh .
The first ring of the trailing column (Tv1) is ingested by the leading vortex ring (Sv).
the circulation and initial convection velocity imbalances also originate from the
characteristic asymmetry of transverse jet velocity profiles (maximum velocity shifted
downstream). These two effects combined cause the unevenness in circulation and
speed between upstream and downstream rollups affecting the early dynamics of the
single starting vortex ring.
4.2.3. Starting vortex ring and trailing column
For longer injection times, a significantly different behaviour was observed and
the leading starting ring vortex was followed by a series of smaller vortex rings
constituting a trailing column. A typical formation sequence is shown in figure 27
from the experiments of case 8. In figure 27(a), the leading vortex ring initially tilts
in the upstream direction due to interaction with the cross-flow boundary layer and
jet exit velocity profile asymmetry. As it departs from the wall and the first trailing
vortex ring is formed in figure 27(b), the leading vortex ring tilting is reversed towards
the downstream direction swallowing the downstream part of the first trailing vortex
as shown in figure 27(c). This partial leapfrogging of the first trailing vortex as
well as the downstream tilting of the leading vortex ring can be explained using
two-dimensional inviscid considerations.
Figure 28 presents qualitative inviscid vortex interactions based on the positions of
the vortices in figure 27, albeit with arbitrary vortex strengths. Induction velocities
were estimated based on the inverse of the vortex core distances. The convective effect
of the cross-flow on the leading starting vortex shifts its axis downstream such that
the downstream rollup of the first trailing vortex is placed in its central velocity field
(figure 28a). Consequently, the vertical velocity component of the starting vortex (Sv)
downstream rollup is decreased and its streamwise velocity increased. At the same
time, the downstream part of the trailing vortex (Tv1) gains in vertical and negative
streamwise velocity, resulting in a velocity vector pointing towards the upstream rollup
of Sv, in a manner comparable to leapfrogging of concentric vortex rings (see Lim &
Nickels 1995). The interaction between the upstream part of Sv and the downstream
part of Tv1 generates higher vertical velocities for both of them. Conversely, the
upstream part of Tv1 interacts with the upstream part of Sv providing it with additional
upward velocity as it is convected downstream. The relative difference in velocities
experienced by the upstream and downstream rollups of Sv results in an overall
rotation of the structure towards the downstream direction, whereas Tv1 tilts towards
421

(a)
(b)
(c)
Cross-flow
z
x
Jet
F IGURE 28. (a) Interactions between leading starting vortex and first trailing vortex based on
inviscid considerations; (b) corresponding to figure 27(c); (c) corresponding to figure 27(e).
Mutual induction effects are represented (matching grey arrows), as well as overall resulting
vortex velocity (black arrows).
(a)
(b)
(c)
(d )
(e)
F IGURE 29. Reactive Mie scattering visualizations in the plane Yj = 0 for case 7 at
St = 0.008. Images enhanced using the CLAHE algorithm show the passage of the jet from
transient structures (trailing column) to transverse jet structures (upper and lower shear-layer
rollups and CRVP roots).
the upstream direction. In figure 28(b,c) the downstream part of Tv1 approaches
the upstream part of Sv so that their vertical velocity increases while the upstream
rollups of Tv1 and downstream rollups of Sv have relatively lower velocities, thus
accentuating the tilting mechanisms. The following trailing vortices (Tv2, Tv3 and Tv4)
experience a velocity field comparable to the one encountered by Tv1 in figure 28(a),
and successively tilt in the upstream direction. In addition to the induction mechanism,
the upstream parts of the trailing vortex rings are exposed to the high-pressure
region generated by the cross-flow deflection in this region while their downstream
counterparts are facing a lower pressure in the jet nascent wake, which promotes
tilting in the upstream direction as they depart from the wall.
The reorientation of the trailing vortices is an important feature since it generates
vertical vorticity from the side arms of the trailing column ring vortices and therefore
provides a source for the initialization of the CRVP. The reactive Mie scattering
visualizations in figure 29(ad) show the passage from transient to quasi-steady
422
(a) 3
(b)
(c)
(d)
(e)
2
1
0
1
(f) 3
(g)
(h)
(i)
( j)
2
1
0
1
F IGURE 30. (ae) The XZ views and (f j) the YZ views of the Laplacian of the
pressure iso-surfaces at 1P = 100 kPa m2 for case 9 at St = 0.159 for time instants:
(a,f ) t = 0.085; (b,g) t = 0.25; (c,h) t = 0.33; (d,i) t = 0.39; (e,j) t = 0.49. The grey
scale for (ae) spanwise and (f j) vertical vorticity ranges from black (negative) to white
(positive).
behaviour of the jet in case 7. Overall the starting vortical structure assembly is very
much the same as the one described previously. However, the transparency offered
by this type of visualization technique, where only mixing regions are visualized,
permits the identification in the background of a quasi-vertical structure joining the
leading vortex ring to the base of the jet. These hanging vortices as described
in the LES results from Yuan et al. (1999) have been observed in many jets in
cross-flow experiments, such as Kelso et al. (1996), and have been associated with
the initialization of the CRVP. In figure 29(b,c) quasi-vertical vortices are visible in
the jet wake and appear connected to the downstream part of the trailing ring vortices.
Structures consistent with the well-known wake vortices formed in the jet wake are
shown in figure 29(d). These visualizations confirm that wake vortices originate from
an interaction between the downstream shear-layer vortices and the boundary layer, as
mentioned by Fric & Roshko (1994). A similar interaction was observed previously in
the dynamics of the hairpin starting vortices of figure 24(a).
The simulated dynamics of the leading vortex ring and the trailing vortices for
case 9 is presented in figure 30 through the rendering of iso-surfaces of the pressure
Laplacian. These are in good qualitative agreement with the experimental observations
despite the use of an idealized square wave for the excitation signal in the numerical
simulations owing to a lack of time-resolved flow-rate records from the corresponding
experiment. The sequence of snapshots in figure 30 confirms the partial leapfrogging
of the trailing vortices as well as the tilting dynamics in this higher-blowing-ratio
case as well. Additional insight is gained into the behaviour of the side arms of
the trailing vortices as they are convected in the free stream and away from the
wall. The first trailing vortex is highly distorted into an L-shape due to the rapid
ingestion of its downstream rollup by the starting vortex ring. The next trailing
vortices (Tv2, Tv3, Tv4, . . . ) follow similar trajectories so that the overall reorientation
of the trailing column generates coherent vertical vorticity on the lee side of the
423
jet as seen in figure 30(f j) and results in the formation of a quasi-vertical vortex
pair consistent with the CRVP of the unforced jet. Both experimental and numerical
observations indicate a CRVP initiation mechanism based on partial leapfrogging of
trailing vortices, combining tilting, stretching and deforming globally, in agreement
with the mechanisms described by Kelso et al. (1996), Cortelezzi & Karagozian (2001)
or Marzouk & Ghoniem (2007).
4.2.4. Classification of starting vortices
In view of the good qualitative agreement with the numerical results from Sau
& Mahesh (2008), a comparable classification of the starting vortices observed in
the current experiments was conducted. At this point, a set of characteristic forcing
parameters had to be identified in order to provide a proper classification. In the
results from Gharib et al. (1998), the piston stroke ratio SR = L/D was equated to the
ratio of length to diameter of the fluid column ejected such that L/D = hUp i/Dj ,
where hUp i was the average piston velocity and the duration of the velocity
program (using Gharibs terminology). Similarly, in the fully modulated jet numerical
simulations of Sau & Mahesh (2008), the characteristic stroke ratio was extrapolated
to L/D = hUjexit i/Dj , where hUjexit i was the average jet nozzle exit velocity and
the pulse width. While both of these studies dealt with flow/no-flow type of jet
forcing, the current study involving non-zero values of BRl leads us to consider
at least three characteristic jet velocities to define the stroke ratio: Uh , Upp and
Um , corresponding respectively to BRh , BRpp and BRm . While the values of Uh and
potentially Ul (thus Upp = Uh Ul ) are expected to affect the starting vortex formation
mechanism, the duration of the low part of the cycle ( (1 DC)/DC) is unlikely
to do so. Consequently, the average value Um , which changes as the low part of
the cycle duration is extended when all other parameters are held constant, was set
aside. The pulse width was logically taken as the duration of the high-blowingratio regime so that = DC/ff . Therefore, two sets of parameters were used in the
current paper to classify the vortical structures: (SRh , BRh ) and (SRpp , BRpp ), such that
SRx = Ux /Dj = BRx U DC/(ff Dj ) = BRx DC/St , with x = pp or x = h.
The parameters used in the classification of the starting vortices were computed
from the values measured by the flow meter and are presented in figure 31(a,b).
Limiting cases were defined when multiple types of starting structures were
observed in consecutive cycles. Overall, both sets of parameters exhibit a distribution
comparable to the one provided by Sau & Mahesh (2008) with a differentiation of
the four different starting vortices. A better separation between the regimes is achieved
when using the set (SRh , BRh ) with less overlapping of the zones compared to the one
obtained with (SRpp , BRpp ). The distinction between the two principal starting structure
regimes observed in figure 31(a) shows that hairpin vortices are formed in cases with
BRh < 0.584 3 %, while this separation occurs for cases with BRpp < 0.294 12 % in
figure 31(b). Based on these observations, the set (SRh , BRh ) appears to provide a more
precise classification map of the starting vortices.
To confirm the better scaling using Uh over Upp , a set of three distinct cases
using idealized square-wave excitations of different amplitudes and pulse widths
were simulated. Using a two-dimensional assumption, the total circulation was
computed at the jet exit by integrating the spanwise vorticity on the domains
{1 : 2 6 Xj < 0, Yj = 0, 0 < Zj 6 2.5} and {2 : 0 < Xj 6 2, Yj = 0, 0 < Zj 6 2.5},
respectively, at the leading and trailing edges of the jet exit. Figure 32 presents
the scaled results using the two scaling parameters used in figure 31: Uc = Uh and
Uc = Upp . The initial time chosen in figure 32 corresponds to the transition from low
424
(a)
(b)
1.2
1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
F IGURE 31. Classification of the starting vortices observed in forced experiments with
respect to characteristic stroke ratios and blowing ratios: (a) SRh , BRh ; (b) SRpp , BRpp .
Symbols: , leading vortex ring and trailing column; , single vortex ring; B, limiting
case vortex ring with/without trailing column; , leading hairpin vortex and trailing
hairpins; , single hairpin vortex; , limiting case hairpin/ring vortex; O, limiting case
hairpin vortex with/without trailing vortices. The vertical line F0 = 3.6 is the asymptotic
formation number from Sau & Mahesh (2008). Transition stroke number is fitted to
(SRx )tr = F0 A1 exp(A2 BRx ): (a) (SRh )tr with A1 = 3.6, A2 = 1.51; (b) (SRpp )tr with
A1 = 3.6, A2 = 1.26.
(a) 3
(b) 3
40
F IGURE 32. Total two-dimensional circulation from LES scaled using (a) Uh and (b) Upp for
case 9 (), case 10 (, high flow on; , low flow on), and case 11 (4) in domains 1 and 2 .
to high blowing ratio, while the final time corresponds to the moment when either one
of the downstream or upstream parts of the vortex ring leaves 1 or 2 . According to
the results from Rosenfeld, Rambod & Gharib (1998) and Krueger, Dabiri & Gharib
(2006), using the proper velocity scale (Uc ) should result in a fixed proportional
relationship between the scaled total circulation at the jet exit ( = /(Uc Dj )) and
the formation time (t = tUc /Dj ) at any forcing conditions. Based on this criterion, the
high-blowing-ratio velocity Uh provides once again an overall better scaling of the rate
of creation of circulation during the formation of the starting vortex (figure 32a)
compared to the peak-to-peak velocity Upp (figure 32b), as the total circulation

(a)
425
(b)
F IGURE 33. Mie scattering visualizations in the plane Yj = 0 showing evidence of:
(a) leading-edge ingestion and horseshoe vortex transport in case 1 at St = 0.079; and
(b) peripheral ingestion in case 9 at St = 0.016.
curves collapse better using Uh . This result supports the experimental observations
of figure 31 and confirms that the jet average high velocity Uh should be used as the
scaling parameter instead of the peak-to-peak velocity.
The limiting stroke number at fixed high blowing ratio between single starting
vortex and starting vortex with trailing vortices is presented in figure 31 as the
transition stroke number (SRx )tr using the nomenclature introduced by Sau & Mahesh
(2008). An exponential fit was used to obtain a continuous limiting value, again
according to the previously mentioned study. The decrease in the value of the
transition stroke number (SRx )tr for decreasing values of BRh (or BRpp ) is consistent
with the numerical results from Sau & Mahesh (2008) and can be explained by
previous research from Rosenfeld et al. (1998) and Krueger et al. (2006). Indeed, in
the first study the value of the formation number was directly affected by the jet
velocity profile and decreased formation number was observed for parabolic profiles
(weak jet) compared to more uniform profiles (strong jet). Such variation in the jet
velocity profile is expected in the current experiments, as the jet velocity profile
becomes weaker when the blowing ratio decreases. In the study by Krueger et al.
(2006) on vortex ring formation in co-flowing flows, a decrease of the formation
number was observed as the co-flowing velocity ratio (equivalent to 1/BR here) was
increased. In the current work, the co-flowing component of the free stream with
the jet flow increases as the blowing ratio decreases and the jet deflection in the
streamwise direction is more important.
Although the current results agree qualitatively well with the observations from Sau
& Mahesh (2008) in terms of the type of starting vortices generated, their dynamics
and global classification, a discrepancy was found in the threshold value of BRh
between the formation of starting hairpin vortices and starting ring vortices. In Sau
& Mahesh (2008) a threshold value of BRh = 2.0 was established, while a value
of BRh = 0.584 was found in the current experimental results. It should be noted,
however, that in the current numerical simulations the threshold was also found to be
significantly higher (around BRh = 2.20) than the experimental threshold based on the
available simulations. As for the discrepancies in transitional thresholds in unforced
conditions, this disparity was attributed to mismatching perturbation levels and initial
length scales between simulations and experiments as well as potential inaccuracies
introduced by the LES model.
4.3. High to low blowing-ratio transition: cross-flow ingestion
At the jet shutdown, most of the cases with BRpp > 0.150 exhibited clear ingestion
of unseeded cross-flow fluid inside the jet feeding tube. As shown in figure 33, two
426
(a)
(b)
(c)
(d)
(e)
0.5
0
0.5
1.0
1.5
1.0 0.5
1.0 0.5
1.0 0.5
1.0 0.5
1.0 0.5
0.5
F IGURE 34. Instantaneous temperature contours and two-dimensional streamlines for case 3
in the plane Yj = 0 at St = 0.016 at time instants: (a) t = DC + 0.018; (b) t = DC + 0.022;
(c) t = DC + 0.13; (d) t = DC + 0.19; (e) t = DC + 0.37.
types of ingestion were observed. In figure 33(a), at low BRh values, the ingestion
principally occurred at the leading edge of the jet exit and triggered the downstream
convection of the horseshoe vortex. However, in figure 33(b), for higher values of BRh ,
the cross-flow ingestion occurred over the whole circumference of the jet exit. The
lack of seed in the cross-flow and the absence of visibility in the jet pipe prevented
a more detailed characterization of this transition using Mie scattering visualizations.
Numerical simulations were then used to gain additional physical insight on these
transient regimes.
4.3.1. Leading-edge ingestion
Figure 34 corresponding to case 3 at St = 0.016 shows that, at the jet shutdown,
the horseshoe vortex is partially ingested inside the jet pipe (figure 34b) while the
inner vortex is pulled further inside the jet pipe. The stream traces show flow
separation at the jet leading and trailing edges around Zj = 1.5. The higher shear
generated by the backward flow entrains the rollup of the jet shear layer to form
a complex vortical system evidenced in figures 34(c) and 35(a,b) and composed of
three vortices of negative spanwise vorticity and two of opposite vorticity near the jet
upstream wall. In the three-dimensional representation of figure 35(b) corresponding
to figure 34(c), the streamlines exit on the side of the jet and merge to form the
legs of the hairpin vortex seen in figure 35(b). Over time, the multiple inner vortices
eventually pair in figure 34(d,e) so that only one stable rollup persists in the jet pipe,
according to the corresponding unforced regime configuration. The time stamps of
figure 34 show that the formation time scale of the inner vortex system inside the jet
pipe at the conditions of case 3 are of the order of 100 ms (t = 0.1), whereas the jet
recovers from the transient regime into a quasi-steady state after approximately 400 ms
(t = 0.4). These time-scale considerations explain the lesser extent of ingestion
observed in simulations at higher forcing frequencies of St = 0.159, as the vortical
system does not have time to develop fully until the next pulse. Ingestion of cross-flow
fluid can also be observed in the snapshot of figure 24(b) corresponding to case 4 with
a BRpp almost half the one in case 3. Although significant ingestion is also occurring,
the extent of it is found to be somewhat lessened, with no horseshoe vortex ingestion.
These observations suggest that the BRpp parameter, directly related to the mass deficit
between high and low parts of the cycle, plays a significant role in the scaling of the
427

(a)
(b)
0.5
0
0.5
z
y
1.0
(c)
(d )
1.0
F
CRVP
0.5
0
0.5
Separation
y
0.5
0.5
1.0
1.5
2.0
F IGURE 35. Ingestion. (a,b) Instantaneous three-dimensional streamlines for case 3 at

St = 0.016 at t = DC + 0.13: (a) XZ side view and spanwise vorticity contours;
(b) YZ front view; streamlines indicated by grey scale of spanwise vorticity from negative
(black) to positive (grey). (c,d) Streamlines for case 9 at St = 0.016 at t = DC + 0.02:
(c) instantaneous two-dimensional streamlines and spanwise vorticity contours from negative
(black) to positive (white); (d) three-dimensional streamlines and vertical velocity profiles at
Zj = 0 and Zj = 1 from negative (black) to positive (white) values.
ingestion extent. This is also supported by the fact that no ingestion was observed for
BRpp < 0.15.
4.3.2. Peripheral ingestion
Peripheral ingestions such as the one evidenced in figure 33(b) were also witnessed
in the numerical simulations of case 10, for instance, as evidenced by the jet exit
vertical velocity profile of figure 26(f ) showing negative vertical velocity values at
both the leading and trailing edges at t = DC + 0.005. In figure 35(c), corresponding
to case 9 at t = DC + 0.02 after the transition from high to low flow rate, streamlines
from both the upstream and downstream regions converge towards the jet exit and
enter the jet pipe. The flow inside the jet pipe separates over the total circumference
of the pipe, therefore reducing the effective flow area and increasing the effective
blowing ratio. The jet shear layer is clearly disrupted with the formation of regions
of vorticity opposite to the natural shear-layer vorticity at the jet exit. The negative
values of the vertical velocity all around the jet exit and the streamline trajectories in
figure 35(d) show the peripheral character of the ingestion. Only two regions at the
jet exit, evidenced in figure 35(d) by two tailless hollow arrowheads, maintain positive
velocity values and correspond to the location of the roots of the CRVP.
The extent of the separation regions in both cases and the evacuation of transient
structures inside the feeding pipe explain the rather chaotic character of the flow
during the early moments of the low part of the cycle.
428

7
(a)
10
(b)
0.6
0.6
0.4
0.4
0.2
0.2
0.5
1
0
7
8
6
(c)
10
15.2
(d)
0.6
0.6
0.4
0.4
0.2
0.2
3
15.4
15.6
1
0
0
(e)
4.0
BR
5
4.1
4.2
7
4.3
4.4
(f)
4.5
7.3
0.6
0.4
0.4
0.2
0.2
1
1
BR
4.1
4.2
4.3
Time (s)
4.4
4.5
BR
BR
15.2
0.6
0
4.0
BR
BR
0.5
BR
0.5
0
0.5
7.3
15.4
7.4
7.5
15.6
7.6
7.7
1
BR
7.4
7.5
7.6
7.7
BR
Time (s)
F IGURE 36. Wavelet analysis mapping of hot-wire time records (upper panels) and
corresponding normalized velocity records and instantaneous flow-rate time records (lower
panels) in the plane Yj = 0 for: (a) case 1 at Xj = 0, Zj = 0, St = 0.016; (b) case 1 at
Xj = 3.5, Zj = 1.25, St = 0.008; (c) case 2 at Xj = 0, Zj = 0.5, St = 0.016; (d) case 4 at
Xj = 0, Zj = 0, St = 0.079; (e) case 6 at Xj = 0, Zj = 0.5, St = 0.159; (f ) case 5 at Xj = 3.5,
Zj = 1.25, St = 0.159.
4.4. Characteristic frequencies

The time-dependent capabilities of the wavelet analysis were particularly useful
for forced experiment records where several regimes and characteristic frequencies
may be encountered throughout a single cycle. Figure 36 shows hot-wire records
taken at several locations under different forcing conditions and analysed using
wavelet decomposition. As for the unforced records, the hot-wire measurements were
obtained at the centre of the jet exit (Xj = 0, Yj = 0, Zj = 0), in the upper shear
layer directly above the jet exit (Xj = 0, Yj = 0, Zj = 0.5) and at a downstream
429
location (Xj = 3.5, Yj = 0, Zj = 1.25). Records taken at the jet exit (figure 36a,d) show
that the shape of the excitation signal is fairly well preserved throughout the jet tube
in most cases. The characteristic acoustic frequency of the system is clearly identified
at the jet exit (St = 0.75) but does not appear in any other records at the other
locations, supporting the fact that this volumetric resonance has little effect on the
global jet behaviour. In agreement with previous observations, low-forcing-frequency
cases (St = 0.008 and 0.016) exhibit distinct characteristic signatures associated
with quasi-steady regimes that are absent from higher-forcing-frequency records
(St = 0.079 and 0.159). In figure 36(ac), corresponding to cases 1 and 2, signatures
at St = 0.155, 0.28 and 0.23, respectively, are found during either the high or low
quasi-steady regimes of the cycle, and are comparable to the ones found in unforced
cases at corresponding blowing ratios: St = 0.145 for BR = 0.438, St = 0.302 for
BR = 0.188 and St = 0.216 for BR = 0.325. The first harmonic and sub-harmonic
are also present in figure 36(a). At higher forcing frequencies of St = 0.079 and
0.159 only the structures associated with the transient regimes were observed in the
visualizations, which is consistent with the wavelet decomposition. Figures 36(d) and
36(f ) are overwhelmed by their respective forcing frequency signatures at St = 0.079
and 0.159 as well as their first harmonics corresponding to the passage of the starting
vortices at the probe location and generated only once per cycle.
5. Summary and conclusions
This study has been oriented towards the description and understanding of vortical
structure dynamics involved in steady and pulsed transverse jets at low blowing ratios.
The investigations were carried out using experimental visualizations and hot-wire
anemometry along with numerical LES methods.
The unforced transverse jet study, which serves as a baseline for the pulsed jet
study, provides insights into the dynamics of vortical structures and its evolution as jet
blowing is increased from relatively low values. The attached jet regime is observed
up to BR = 0.225, where the jet exhibits four dominant vortical structures: hairpin
shear-layer vortices, quasi-streamwise side vortices, horseshoe vortex and inner vortex.
Secondary inverted hairpin vortices also form in the lower jet shear layer due to
wallvortex interactions. Such structures have been observed in many studies dealing
with hairpin vortices but not of the inverted type, the more so in the jet in cross-flow
context. Side vortices are born from shear caused by the injection of jet fluid along
the jet exit perimeter, cross-flow deflection around the jet exit and flow inrush near
the wall caused by the hairpin legs. The authors argue that mutual and wall-mirrored
induction mechanisms acting on those side vortices are responsible for the formation
of X-patterned structures located on each side of the jet core and constitute dominant
features in the far field. Although a counter-rotating vortex pair is identifiable in
the mean vorticity field, the relative strength and the spatial discontinuity of the
associated instantaneous hairpin vortex structures make the resemblance with the wellknown CRVP of the detached jet only tenuous. The horseshoe vortex, resulting from
boundary-layer separation, is found to occupy a stable position partly on top of the jet
exit. The resulting pressure cap above the jet leads to the separation and rollup of the
jet upstream shear layer to form a stable inner vortex inside the jet pipe. While rollups
of the leading-edge shear layer have been documented in the past, they have always
been described as unstable and shedding in a periodic or quasi-periodic manner. This
is the first time this structure has been found to occupy a stable position inside the jet
pipe. The interaction between the horseshoe vortex and the inner vortex leads to the
430
convection of the former for 0.225 6 BR < 0.275 and of both for 0.275 6 BR < 0.6.
The convection of the inner vortex marks the transition from the attached to the
detached jet regime in which they evolve into the well-known upper shear-layer rollups
observed in many detached jet studies. The side vortices eventually become the wake
vortices at higher blowing ratios. Beyond BR = 0.6, the jet column and its associated
pressure gradient become strong enough to prevent the convection of the horseshoe
vortex, which remains near the jet exit leading edge as the detached jet regime is
established.
The current work adds to the knowledge of the interactions of transverse jet vortical
structures at low blowing ratios and shows that the well-known detached jet structures
can be traced back to ones observed in the attached jet regime. It also shows that the
nature of their interactions (i.e. horseshoe vortex/inner vortex and side vortices/shearlayer vortices) and their relative strength change significantly as the jet lifts off,
therefore leading to considerably different dynamics. As such, the evolution of the jet
from one regime to another is seen as a competition between the different vortical
structures. On the one hand, the horseshoe vortex/inner vortex interaction is such that
at low BR values the inner vortex is rather weak and maintained inside the jet pipe
by a relatively stronger horseshoe vortex. However, as the blowing ratio is increased,
the inner vortex gains in strength, pushes the horseshoe vortex away from the jet exit
and becomes free to be convected to form the well-known upper shear-layer vortices
in the detached jet configuration. On the other hand, the shear-layer/side vortices
interaction is such that at low BR values the side vortices are relatively stronger than
the hairpin vortices and play a significant role in the far field, effectively challenging
the notion of CRVP due to their physical proximity. However, at higher blowing ratios,
the shear-layer structures gain in strength and lift off the wall, thus changing the
nature of their interaction with the side vortices, which evolve into the well-known
wake vortices. This transition is also manifest in the characteristic formation frequency
signatures isolated using wavelet analysis with a tipping point around BR = 0.425.
The forced jet exhibits two distinct transient phases during a forcing cycle at the
transition from low to high blowing ratio and the reverse. Quasi-steady phases are
present between the two transients for cases with sufficiently low forcing frequencies
(St 6 0.016). At the low to high blowing-ratio transition, starting hairpin or ring
vortices are generated, due to sudden increase in shear at the jet exit. Trailing
structures are formed in the wake of the leading starting vortices, be they hairpin
or ring, if the stroke number (SRh = BRh U DC/(ff Dj )) exceeds the transition stroke
number value at the corresponding BRh . Starting hairpin vortices are formed at low
BRh values and their dynamics conforms to that observed in the smaller regular
hairpin vortices under unforced conditions. A strong interaction between starting
vortex and horseshoe vortex is observed, which leads to the reorientation of the
latter in a similar way as wake vortices are reoriented in the unforced detached jet to
generate wake vortices. The current work shows that forcing the jet can significantly
affect the interactions of vortical structures. In particular, a discussion is provided
on the modified dynamics between horseshoe vortex and inner vortex, leading to the
generation of large hairpin (starting) vortices at blowing ratios significantly above the
transitional threshold found in steady state. The experimentally observed dynamics of
single starting vortex rings is in agreement with previous numerical studies, indicating
upstream tilting of the structure due to interaction of its upstream part with the
cross-flow boundary layer.
The authors also identify the inherent asymmetry of the jet exit velocity profile
as a potential source of the velocity imbalance between upstream and downstream
431
rollups responsible for the tilt. A two-dimensional inviscid induction mechanism

accounting for the partial leapfrogging of the first trailing ring vortices is proposed
as an explanation for the dynamics of the starting ring vortex with a trailing column.
The reorientation of the structures of the trailing column leads to the formation of
a coherent counter-rotating vortex pair on the lee side of the jet responsible for
the initialization of the CRVP. Estimations of the rate of generation of circulation
under multiple forcing conditions have led to the identification of SRh and BRh as
a suitable pair of scaling parameters for the classification of the starting vortices
in partly modulated forced transverse jets. Using those parameters a limiting value
of BRh = 0.584 is found below which hairpin starting vortices are generated and
above which starting ring vortices are formed. This experimental value evidences a
discrepancy between experimental and numerical results, which was found to be closer
to BRh = 2.0 in the current simulations as well as previous work. The discrepancy is
attributed to differences in jet inlet conditions between experiments and simulations
but also to potential numerical model inadequacies. In addition to illuminating the
dynamics of transitions and the associated interactions of vortical structures, this study
also constitutes an experimental verification of past numerical studies in terms of both
the dynamics of the starting vortical structures as well as the classification proposed
by Sau & Mahesh (2008). Beyond that, it further extends those results to partly
modulated transverse jets with non-zero low blowing ratios.
At the transition from high to low blowing ratio, cross-flow ingestion inside the jet
pipe is found to occur for BRpp values above 0.15 from the mass deficit generated
as the blowing ratio is suddenly decreased. Two ingestion mechanisms are observed
both in experiments and in numerical simulations. Leading-edge ingestion occurs at
rather low values of BRh when the jet is attached during the high part of the cycle.
Conversely, ingestion along the jet exit perimeter occurs at rather high values of BRh
when the jet is detached during the high part of the cycle. In both cases, the mass
deficit leads to a jet flow separation and backflow inside the feeding pipe. In the case
of leading-edge ingestion, the separation triggers the formation of a complex vortical
system, which is slowly dissipated as the jet settles in the low part of the cycle. The
parameters BRpp and St are found to have a direct impact on the extent of the
cross-flow ingestion.
Acknowledgements
This work was supported by the U.S. Air Force Office of Scientific Research under
grants FA9550-06-1-0335 and FA9550-08-1-0440, as well as the Louisiana Boards
of Regents through grant LEQSF(2006-07)-ENH-TR-14. Computing resources were
available through the Louisiana Optical Network Initiative (LONI) at Louisiana State
University.
REFERENCES
ACARLAR, M. S. & S MITH, C. R. 1987a A study of hairpin vortices in a laminar boundary layer.
Part 1. Hairpin vortices generated by a hemisphere protuberance. J. Fluid Mech. 175, 141.
ACARLAR, M. S. & S MITH, C. R. 1987b A study of hairpin vortices in a laminar boundary layer.
Part 2. Hairpin vortices generated by fluid injection. J. Fluid Mech. 175, 4385.
A NDREOPOULOS, J. & RODI, W. 1984 Experimental investigation of jets in a crossflow. J. Fluid
Mech. 138, 93127.
B IDAN, G. & N IKITOPOULOS, D. E. 2011 Fundamental study of modulated transverse jets from a
film-cooling perspective. AIAA J. 49 (7), 14981510.
432
B LANCHARD, J. N., B RUNET, Y. & M ERLEN, A. 1999 Influence of a counter rotating vortex pair
on the stability of a jet in a cross flow: an experimental study by flow visualizations. Exp.
Fluids 26 (1), 6374.
C AMUSSI, R., G UJ, G. & S TELLA, A. 2002 Experimental study of a jet in a crossflow at very low
Reynolds number. J. Fluid Mech. 454, 113144.
C HANG, Y. K. & VAKILI, A. D. 1995 Dynamics of vortex rings in crossflow. Phys. Fluids 7 (7),
158397.
C HENG, M., L OU, J. & L IM, T. T. 2009 Motion of a vortex ring in a simple shear flow. Phys.
Fluids 21 (8), 081701.
C ORTELEZZI, L. & K ARAGOZIAN, A. R. 2001 On the formation of the counter-rotating vortex pair
in transverse jets. J. Fluid Mech. 446, 347373.
C OULTHARD, S. M., VOLINO, R. J. & F LACK, K. A. 2007 Effect of jet pulsing on film cooling
Part I: Effectiveness and flow field temperature results. J. Turbomach. 129 (2), 232246.
E KKAD, S., O U, S. & R IVIR, R. B. 2006 Effect of jet pulsation and duty cycle on film cooling
from a single jet on a leading edge model. J. Turbomach. 128 (3), 564571.
E ROGLU, A. & B REIDENTHAL, R. E. 2001 Structure, penetration, and mixing of pulsed jets in
crossflow. AIAA J. 39 (3), 417423.
F RIC, T. F. & ROSHKO, A. 1994 Vortical structure in the wake of a transverse jet. J. Fluid Mech.
279, 147.
G HARIB, M., R AMBOD, E. & S HARIFF, K. 1998 A universal time scale for vortex ring formation.
J. Fluid Mech. 360, 121140.
G OGINENI, S., G OSS, L. & ROQUEMORE, M. 1998 Manipulation of a jet in a cross flow. Exp.
Therm. Fluid Sci. 16 (3), 209219.
G OPALAN, S., A BRAHAM, B. M. & K ATZ, J. 2004 The structure of a jet in cross flow at low
velocity ratios. Phys. Fluids 16 (6), 20672087.
G UO, X., S CHRODER, W. & M EINKE, M. 2006 Large-eddy simulations of film cooling flows.
Comput. Fluids 35 (6), 587606.
H AGEN, J. P. & K UROSAKA, M. 1993 Corewise cross-flow transport in hairpin vortices the
tornado effect. Phys. Fluids A: Fluid Dyn. 5, 3167.
H ARRINGTON, M. K., M C WATERS, M. A., B OGARD, D. G., L EMMON, C. A. & T HOLE, K. A.
2001 Full-coverage film cooling with short normal injection holes. J. Turbomach. 123 (4),
798805.
H AVEN, B. A. & K UROSAKA, M. 1997 Kidney and anti-kidney vortices in crossflow jets. J. Fluid
Mech. 352, 2764.
J EONG, J. & H USSAIN, F. 1995 On the identification of a vortex. J. Fluid Mech. 285, 6994.
J OHARI, H. 2006 Scaling of fully pulsed jest in crossflow. AIAA J. 44 (11), 27192725.
J OHARI, H., PACHECO -T OUGAS, M. & H ERMANSON, J. C. 1999 Penetration and mixing of fully
modulated turbulent jets in crossflow. AIAA J. 37 (7), 842850.
K ELSO, R. M., L IM, T. T. & P ERRY, A. E. 1996 An experimental study of round jets in cross-flow.
J. Fluid Mech. 306, 111144.
K ELSO, R. M. & S MITS, A. J. 1995 Horseshoe vortex systems resulting from the interaction
between a laminar boundary layer and a transverse jet. Phys. Fluids 7 (1), 1538.
K ROTHAPALLI, A. & L OURENCO, L. 1990 Separated flow upstream of a jet in crossflow. AIAA J.
28 (3), 414420.
K RUEGER, P. S., DABIRI, J. O. & G HARIB, M. 2006 The formation number of vortex rings formed
in uniform background co-flow. J. Fluid Mech. 556, 147166.
L IGRANI, P. M., G ONG, R., C UTHRELL, J. M. & L EE, J. S. 1996 Bulk flow pulsations and film
cooling I. Injectant behaviour. Intl J. Heat Mass Transfer 39 (11), 22712282.
L IM, T. T., L UA, K. B. & T HET, K. 2008 Does Kutta lift exist on a vortex ring in a uniform cross
flow? Phys. Fluids 20 (5), 051701.
L IM, T. T., N EW, T. H. & L UO, S. C. 2001 On the development of large-scale structures of a jet
normal to a cross flow. Phys. Fluids 13, 770.
L IM, T. T. & N ICKELS, T. B. 1995 Vortex rings. In Fluid Vortices (ed. S. I. Green). Kluwer.
M ARZOUK, Y. M. & G HONIEM, A. F. 2007 Vorticity structure and evolution in a transverse jet.
J. Fluid Mech. 575, 267305.
433
MC LOSKEY, R. T., K ING, J. M., C ORTELEZZI, L. & K ARAGOZIAN, A. R. 2002 The actively
controlled jet in crossflow. J. Fluid Mech. 452, 325335.
M EGERIAN, S., DAVITIAN, J., A LVES, L. S. DE B. & K ARAGOZIAN, A. R. 2007 Transverse-jet
shear-layer instabilities. Part 1. Experimental studies. J. Fluid Mech. 593, 93129.
N EW, T. H., L IM, T. T. & L UO, S. C. 2003 Elliptic jets in cross-flow. J. Fluid Mech. 494,
119140.
N EW, T. H., L IM, T. T. & L UO, S. C. 2006 Effects of jet velocity profiles on a round jet in
cross-flow. Exp. Fluids 40, 859875.
N IKITOPOULOS, D. E., ACHARYA, S., O ERTLING, J. & M ULDOON, F. H. 2006 On active control
of film-cooling flows. In ASME Conference Proceedings, vol. 2006, pp. 6169. ASME
(4238X), http://link.aip.org/link/abstract/ASMECP/v2006/i4238X/p61/s1.
N IKITOPOULOS, D. E. & L IU, J. T. C. 1987 Nonlinear binary-mode interactions in a developing
mixing layer. J. Fluid Mech. 179 (1), 345370.
N IKITOPOULOS, D. E. & L IU, J. T. C. 2001 Nonlinear three-mode interactions in a developing
mixing layer. Phys. Fluids 13, 966.
P ERRY, A. E. & L IM, T. T. 1978 Coherent structures in coflowing jets and wakes. J. Fluid Mech.
88 (3), 451463.
R IVERO, A., F ERRE, J. A. & G IRALT, F. 2001 Organized motions in a jet in crossflow. J. Fluid
Mech. 444, 117149.
ROSENFELD, M., R AMBOD, E. & G HARIB, M. 1998 Circulation and formation number of laminar
vortex rings. J. Fluid Mech. 376, 297318.
S AU, R. & M AHESH, K. 2008 Dynamics and mixing of vortex rings in crossflow. J. Fluid Mech.
604, 389409.
S MITH, S. H. & M UNGAL, M. G. 1998 Mixing, structure and scaling of the jet in crossflow.
J. Fluid Mech. 357, 83122.
S YKES, R. I., L EWELLEN, W. S. & PARKER, S. F. 1986 On the vorticity dynamics of a turbulent
jet in a crossflow. J. Fluid Mech. 168, 393413.
T ORRENCE, C. & C OMPO, G. P. 1998 A practical guide to wavelet analysis. Bull. Am. Meteorol.
Soc. 79 (1), 6178.
W OMACK, K. M., VOLINO, R. J. & S CHULTZ, M. P. 2008 Combined effects of wakes and jet
pulsing on film cooling. J. Turbomach. 130 (4), 041010.
Y UAN, L. L., S TREET, R. L. & F ERZIGER, J. H. 1999 Large-eddy simulations of a round jet in
crossflow. J. Fluid Mech. 379, 71104.
Z HOU, J., A DRIAN, R. J., BALACHANDAR, S. & K ENDALL, T. M. 1999 Mechanisms for
generating coherent packets of hairpin vortices in channel flow. J. Fluid Mech. 387, 353396.

On Steady and Pulsed Low-Blowing-Ratio Transverse Jets

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

On Steady and Pulsed Low-Blowing-Ratio Transverse Jets

Caricato da

Copyright:

Formati disponibili

J. Fluid Mech. (2013), vol. 714, pp. 393433.

c Cambridge University Press 2013

On steady and pulsed low-blowing-ratio

G. Bidan and D. E. Nikitopoulos

The CRVP is commonly considered to be the principal mixing structure in

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

TABLE 1. Inlet boundary-layer characteristics at Xj = 6, Yj = 0.

On steady and pulsed low-blowing-ratio transverse jets

profiles were obtained using single-component hot-wire anemometry, no distinction

G. Bidan and D. E. Nikitopoulos

F IGURE 3. (a) Reactive Mie scattering visualization in the plane Yj = 0 at BR = 0.188.

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

F IGURE 7. An XY view of the Laplacian of the pressure iso-surfaces at 1P = 1.2 kPa m2

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

F IGURE 9. Time-averaged streamwise vorticity (filled) and normalized temperature contours

found in figure 9 as the usual CRVP, although a few considerations lead us to do

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

F IGURE 14. Time-averaged streamwise vorticity (filled) and normalized temperature

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

F IGURE 17. Time-averaged streamwise vorticity (filled) and contours by increments of

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

TABLE 2. Forced case parameters: a no flow-meter record available; b no experiments at

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

F IGURE 25. Instantaneous pressure difference contours P Porigin and two-dimensional

On steady and pulsed low-blowing-ratio transverse jets

F IGURE 26. An XZ view of the Laplacian of the pressure iso-surfaces at 1P = 80 kPa m2

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

F IGURE 35. Ingestion. (a,b) Instantaneous three-dimensional streamlines for case 3 at

G. Bidan and D. E. Nikitopoulos

4.4. Characteristic frequencies

On steady and pulsed low-blowing-ratio transverse jets

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

rollups responsible for the tilt. A two-dimensional inviscid induction mechanism

G. Bidan and D. E. Nikitopoulos

On steady and pulsed low-blowing-ratio transverse jets

Potrebbero piacerti anche