International Communications in Heat and Mass Transfer 76 (2016) 108–116

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer

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

Experimental visualization and LES investigations on cloud cavitation

shedding in a rectangular nozzle orifice☆
Zhixia He a,b, Yuhang Chen a,⁎, Xianyin Leng b, Qian Wang a, Genmiao Guo a
a
School of Energy and Power Engineering, Jiangsu University, Zhenjiang, China
b
Institute for Energy Research, Jiangsu University, Zhenjiang, China

a r t i c l e i n f o a b s t r a c t

Available online 24 May 2016 Shedding of cloud cavitation in a nozzle orifice is a common phenomenon of cavitation instability which has a
great influence on erosion damage and spray atomization process, but it hasn't been studied in details and the
Keywords: mechanism remains unclear. In this paper, visualization experiments and Large Eddy Simulation (LES) were car-
Nozzle ried out to study the process of periodic cloud cavitation shedding and re-entrant jet happened in a 2 mm-width
avitation nozzle with fuel of diesel. Numerical and experimental images provide a better insight into this physical mech-
Experimental visualization
anism. The results show that cavitation inception is formed at the edge of inlet and the recirculation zone
LES
Shedding of cloud cavitation
which below the shear layer. Then the attached cavitation grows and shedding of cloud cavitation occurs due
to the re-entrant jet located between cavity and wall. Moreover, the dynamic mechanism causing shedding of
cloud cavitation is a combination of shedding vortex and pressure fluctuations. Cavitation shedding frequency
can also be measured by experiment and numerical simulation. As a result, inception of cavitation, re-entrant
jet and cavitation cloud shedding are accurately predicted by LES in accordance with the quantitative images
observed in the experiment.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction
The fuel atomization process and spray characteristics directly affect
the mixture process of fuel and air. Then further the combustion process
is extremely important in reducing fuel consumption and exhaust emis-
sions in direct injection diesel engines. While the cavitating flow in die-
sel injector nozzles has a great influence on the fuel spray atomization, it
is essential to make clear of this nozzle cavitating flow characteristics for
better understanding of the spray primary atomization mechanisms.
Bergwerk [1] firstly reported the influence of cavitation on spray for-
mation, and its connection with nozzle flow which was observed exper-
imentally, and also found that a decrease in the discharge coefficient of
the nozzle under cavitating conditions. However, the strong transient
characteristics of the actual fuel injection process, small scales of the
fuel injector nozzle and the lack of optical access make the experimental
observation of the flow inside the injector nozzles in realistic conditions
extremely hard. Therefore, most investigations have been carried out in
a transparent scaled-up model at the aim of visualizing the cavity for-
mation, growth and collapse process when occurrence of cavitation in
diesel nozzles [2–3]. Payri et al. [4–5] reported that cavitation leads to
an increase of the spray cone angle as well as flow outlet velocity, and
measured the spray momentum in order to explain the effects of nozzle
☆ Communicated by W.J. Minkowycz.
⁎ Corresponding author.
E-mail address: cyh5830739@163.com (Y. Chen).
geometry on the cavitating flow. However, few attempts have been
done to investigate the shedding of cloud cavitation in a diesel nozzle
orifice.
The study of the unsteady cavitating flow which named cloud cavi-
tation occurring on hydrofoils has reported that “re-entrant” jet is cen-
tral to the process of cloud cavitation shedding which is a motion of a
liquid jet beneath the attached cavity in the direction opposite to the
main flow [6]. Therefore, for the diesel injector nozzle, studying the cav-
itating flow, especially the shedding of cloud cavitation in it will be cru-
cially significant for better understanding of the motion of re-entrant jet
and the increase of spray angle induced by this nozzle cavitating flow.
Recently, a number of studies [7–9] have tried to explore the mech-
anism of the cloud cavitation instability, both numerically and experi-
mentally. It is typically thought that the re-entrant jet is created by
the flow expanding in the closure region behind the cavity, impinging
with the wall and establishing a local stagnation point [10]. Sato and
Saito [11] performed visualization of periodic cloud shedding in large
scale cylindrical orifices. Stanley and Barber [12] investigated the re-
entrant jet mechanism for periodic cloud shedding cavitation experi-
mentally, which showed a constant presence of a liquid sub-layer
between the cloud cavitation and the nozzle wall, and it is also found
that traveling pressure wave driven by the previous bubble collapse.
Therefore, with the rapid development of computing abilities and
limitations of measurement techniques, a number of computational
studies [13–16] on cavitation have been reported over the years. In
most previous simulations, the Reynolds Averaged Navier–Stokes

http://dx.doi.org/10.1016/j.icheatmasstransfer.2016.05.033
0735-1933/© 2016 Elsevier Ltd. All rights reserved.
 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
(RANS) approach like κ-ε models were used to treat turbulence.
Actually, they do not address the influence of turbulent fluctuations
on the onset and development of cavitation, and RANS models led
cavitation to a steady-state solution. It seemed to be unable to predict
the shedding of cloud cavitation and re-entrant jet. Accordingly, it is
obvious that some more advanced turbulent models should be used to
simulate this turbulent two-phase flow for capture of more details.
At present, Large Eddy Simulation (LES) has become a great alterna-
tive to advance the knowledge of the internal turbulent flow [17], as it is
often more capable of reproducing large unsteady motion of the flow
field. LES with the WALE SGS stress model was adopted to calculate
the cavitation shedding and horse-shoe structures by Ji et al. [18].What's
more , Ji also analyzed the pressure fluctuations induced by cloud cavi-
tation around a hydrofoil. Sou et al. [19] proposed a new combination of
Large Eddy Simulation (LES), Eulerian–Lagrangian Bubble Tracking
Method (BTM), and the Rayleigh–Plesset (RP) equation to simulate an
incipient cavitation in a scaled-up nozzle of fuel injector. The result
showed that incipient cavitation flow can be predicted by LES using a
fine grid. Egerer et al. [20] presented large-eddy simulations (LES) of
cavitating flow of a diesel-fuel-like fluid in a generic throttle
geometry, and the LES with the employed cavitation modeling predicts
relevant flow and cavitation features accurately within the uncertainty
range of the experiment. Fuchs et al. [21] performed Large Eddy Simula-
tions which not only are able to reproduce vortex cavitation, but also
give further insight into the complex interaction between cavitation
and turbulence.
In this study, we aim to investigate the shedding of cloud cavitation
both in experimental and numerical methods. Additionally, induced
pressure oscillations and turbulent vortex shedding were also analyzed
with LES model to find out the relationship between cloud cavitation
shedding and re-entrant jet. Given the importance of these phenomena
on the subsequent mixing and spray processes, so the results hereby
presented may be of interest for diesel nozzle orifice designers. To do
so, a visual experimental system was set up to capture the images of
the process of cloud cavitation in a 2 mm-wide rectangular orifice by a
high speed CCD camera and supply the experimental data for
verifications of numerical models.
2. Experimental setup
Aimed to further investigate the cloud cavitation shedding, a
visualization experimental bench was set up. The schematic diagram
of the visual experiment setup was shown in Fig. 1. Diesel was pumped
by a gear pump to supply a working liquid to the rectangular nozzle. An
electromagnetic flow meter and a pressure gauge were installed to
measure the flow rate and static upstream injection pressure,
Fig. 1. Experimental setup.
109
respectively. About 0.8 m length of pipe before the nozzle inlet was
used to reduce the upstream instabilities. It is noted that, nozzle flow in-
ject into air so we keep the back pressure 0.1 MPa. The tested rectangu-
lar nozzle was made of transparent acrylic that has almost the same
index of refraction as diesel fuel. The refractive index of acrylic is 1.49.
Geometry of the rectangular nozzle was illustrated in Fig. 2. The width
W and length L of the nozzle were about 2 and 4 mm, respectively.
The nozzle had a W/L ratio of 2 and a nominally sharp entrance. This
structure was representative for cavitating flow in fuel injectors, since
the velocities and cavitation obtained are similar to the large-scaled
fuel injector nozzle and it is easy for us to investigate cloud cavitation.
For verification of numerical simulation results, a high-speed CCD
camera (MotionPro-TM10000, image size of 512 × 512 pixels at
10,000 fps) connected with a long distance microscope (Questar QM-
1) was mounted on a tripod to capture the periodic behavior of cloud
cavitation in the nozzle. It is clear in magnification and resolution
when your specimen lies between 56 cm and1.5 m. The nozzle was
illuminated by a high power LED light (99 LED bulbs) installed at the
opposite side of the camera across the nozzle [22]. The cavitation
number σ was defined as:
p0 −pv
σ¼ ð1Þ
0:5ρv2
where P0, Pv and v were the back pressure, the saturated vapor pressure
and the mean liquid velocity in the nozzle, respectively. Here, we keep
the back pressure around 0.1 MPa. The non-dimensional parameter
can be well used for description of the cavitation conditions in nozzles.
3. Numerical models
The described case was simulated by using CFD software ANSYS
Fluent 14.5. The spray cone angle and the breakup of fuel spray just
out of the nozzle orifice are affected by the combination of turbulence
and cavitation. The largest eddies are responsible for a good mixing
and, consequently, for a good performance of the combustion reaction
[23]. In addition, the shedding of cloud cavitation and re-entrant jet
are complicated to simulate due to the highly transient behavior and
the RANS turbulence model was not enough to capture transient cavi-
tating flow. Thus, in this study, a combination of Large Eddy Simulation
(treat the turbulent flow), a homogeneous multiphase mixture flow
model (handling the interactions between the cavitation bubbles and
the diesel flow field) and Schnerr–Saurer Cavitation model (capturing
the formation of cavitation bubbles) was proposed to simulate the
transient behavior of cloud cavitation.
110 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116

continuity equations to derive the exact expression for the net mass
transfer from liquid to vapor:

∂  !
ðαρv Þ þ ∇  αρv V ¼ Re −Rc ð6Þ
∂t

where ρv and ρl are the density of vapor and liquid, respectively. Re and
Rc are the source terms representing the rates of growth and collapse of
bubbles, respectively. Also α is the vapor volume fraction, which can be
written as:

n  4πRB 3 =3
α¼ : ð7Þ
Fig. 2. Transparent nozzle geometry. 1 þ n  4πRB 3 =3

This expression connects the vapor volume fraction to the number of

bubbles per volume of liquid, where RB is the radius of bubbles, and n is
the number density of bubbles per volume of liquid. Here n is defined as
3.1. Large Eddy Simulation and cavitation model
1013 according to Schnerr and Sauer [26].
Growth and collapse of cavitation bubbles and nuclei are computed
Germano et al. and subsequently Lilly [24–25] conceived and
by solving the following Rayleigh–Plesset equation [28]:
developed a Dynamic Smagorinsky–Lilly Model (DSM) which was
used to compute the sub-grid scale stress τij, and the equation can be sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ρ ρ 3 2 ðpV −pÞ
used as follows: Re ¼ v l α ð1−α Þ pV ≥p ð8Þ
ρ RB 3 ρl
1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
τij ¼ −2us Sij þ τkk δij : ð3Þ
3 ρ ρ 3 2 ðp−pV Þ
Rc ¼ v l α ð1−α Þ pV ≤p ð9Þ
ρ RB 3 ρl
The sub-grid eddy viscosity coefficient μs and Smagorinsky model
constant Cs can be expressed as:
where pV is the saturated vapor pressure, and p is usually taken to be the


same as the cell center pressure.
μ s ¼ ρC s Δ Sij

2

ð4Þ
3.2. Computational grid and boundary conditions
Lij Mij
Cs ¼ ð5Þ
M ij M ij
The meshing module ICEM within ANSYS software was used. In
order to save the computational resources and time, it is not necessary
where Sij is the average rate of strain tensor. Δ is the local grid scale. Lij to simulate the whole experimental flow geometry. Therefore, only
can be computed from the resolved large eddy field. the upper part of the rectangular nozzle is carried out as the simulation
In this paper, Schnerr and Sauer model [26] was used since it is domain which we consider the orifice as the most interesting part.
compatible with the LES turbulence model and suitable for complex Because mesh density and size have a great influence on the accura-
orifice flow conditions [27]. The model uses the following two-phase cy of numerical simulation, the model is calculated with four different

Fig. 3. Effect of mesh densities on volume flow rate.

 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
Fig. 4. Computational mesh.
Table 1
Boundary conditions and physical parameters.
Boundaries and initial condition Value
Pressure inlet 0.2–0.3 MPa
Pressure outlet 0.1 MPa
Liquid phase T = 23 °C
ρl = 830 kg/m3; νl = 0.0024 kg/m
Vapor phase ρv = 0.029 kg/m3, pv = 1200 Pa
densities of the grid partition. As shown in Fig. 3, the relationship of
volume flow rate between the simulated and experimental results is
given. It can be seen that with the increasing of the mesh densities,
the difference between the simulated and experimental results is tiny.
Therefore, the mesh number of 1.4 million was selected in this study.
The mesh was refined near the walls and the inlet sharp edge as
shown in Fig. 4. The minimum cell size in the near wall vicinity is
3 μm and the maximum dimensionless wall distance (y+) in the area
of interest is about 1.
The solution employs the PISO algorithm for the pressure velocity
coupling and for the pressure discretization, the PRESTO! scheme is
used which is very effective in complex cavitation flows. The time step
is set as 10−6 s which is enough to capture the dynamic characteristic
of cloud cavitation shedding and
measure the pressure fluctuations. The boundary conditions consist
of a fixed pressure at inlet and the fluid properties are listed in Table 1,
and the wall is defined as non-slip wall condition.
As cavitation occurs mostly in the four corners of the rectangular
orifice, we can clearly see the process of cloud cavitation shedding in
Fig. 5. Simulated slice and the pressure measure point #1.
111
comparison to the shedding in cylindrical orifice. In order to represent
a more detailed flow sight in the test nozzle, a slice at x = 0.75 mm
was established as shown in Fig. 5 which is more appropriate for
representing the location of cavitation. It is noted that 0.75 mm is the
distance between the middle slice of the simulated geometry and the
monitored slice. We defined the coordinate system at the middle of
the inlet cross profile. Point #1 is near the wall and 3 mm distance
from the inlet where the end of the shedding path can be found. This
point is used to monitor the instantaneous pressure fluctuations accord-
ing to the unsteady behavior of cavitation.
4. Results and discussion
4.1. The inception of cavitation
A typical flow structure and possible cavitation locations at slice of
x = 0.75 mm are shown in Fig. 6(a), and the dark iso-surface
corresponds to cavitation volume fraction of 0.1. The experimental
and simulated results can be seen in Fig. 6(b) at Pin = 0.2 MPa (v =
12.2 m/s, σ = 1.58) which shows the location of the inception
cavitation. The image was taken with the high-speed camera's exposure
time of 100 μs. The dark region in the picture corresponds to the
cavitating area. It is noted that the inception of cavitation induced in a
rectangular nozzle can be classified into three types according to the
incipient position, which are cavitation inception at inlet, cavitation
inception at recirculation zone and vortex cavitation bubble due to the
vortices.
As shown in Fig. 6(b), primary cavitation inception is located at the
sharp inlet edge of the nozzle due to the reduction of cross-section
and it collapsed quickly. In addition, flow separation occurs from the
sharp edge corner leading to the formation of a layer of high shear. It
can be seen from the simulation image of Fig. 6(b) that the separated
shear layer is inherently unstable and below it a recirculation zone
which called vena contracta is formed. Compared with the observed
and simulation images, cavitation inception at recirculation zone is
lying about 1 mm distance from the nozzle inlet along the flow
direction. In addition, it occurs at the recirculation zone under the
local pressure which is below the separated layer. Moreover, with the
higher velocity gradient and oscillation at the separated layer, a large
quantity of small vortices formed and shed from it. The cores of vortices
for the pressure drop cause vortex cavitation bubble. Then these
bubbles stretch in the shear layer with the high shear and collapse or
pair into the fixed cavity. However, the cavitation observed inside the
nozzle orifice in experiments is larger than those from the simulations
and some vortex cavitation bubbles are hard to reflect in simulation.
112 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
Fig. 6. Observed and calculated inception cavitation flows.
It is highlighted that the attached cavity is unstable which can be
seen from the enlarge view (which can be seen in Fig. 6(b)) of the
recirculation zone. Reattachment of the flow establishes the re-entrant
jet which seems to form at the closure region of the attached cavity
and move towards the entrance which is opposite to the main stream.
4.2. Shedding frequency of cloud cavitation
With the decreasing of cavitation number, the inception cavities
grow and become unsteady. A sequence of unsteady shedding behavior
of cloud cavitation is particularly observed at σ = 1.008 with injection
pressure of 0.26 MPa from Fig. 7. It is noted that, only the right half of
the cavitating flow in the nozzle is shown in Fig. 6.
It can be seen from the picture that cloud cavitation occurs at the
recirculation zone where we consider that the secondary inception
cavitation is generated. Simultaneously, it seems that development
and shedding of cloud cavitation mainly exist in the zone which is
below the red solid line in Fig. 7. In addition, shedding of cloud cavita-
tion and collapse sequentially make the length of cavitation fluctuates
so that cloud cavitation sees a certain degree of extension and tends to
be considerably unstable. The length of cavitation is defined from the
inlet to the most distant location where cloud cavitation can reach
(see, red arrows in the Fig. 7). When the cloud cavitation shed
downstream before it collapses, the length of cavitation reaches the
maximum. Then the shedding of cloud cavitation collapses due to the
pressure recovery and the length becomes minimum.
Fig. 8 presents the time evolution of cavitation length l in the nozzle
(at Pin = 0.26 MPa, v = 15.3 m/s, σ = 1.008), which is obtained and
measured from the experimental images of the nozzle flow. The
Fig. 7. Development and shedding of cavitation.
sampling frequency is 10,000 Hz. As shown in Fig. 8, the maximum cav-
itation length is about l/L = 0.9. Then cloud cavitation collapses to
micro-bubbles and the length of cavitation has a dramatic decrease
which can be even below l/L = 0.45. The non-dimensional cavitation
length l/L fluctuates between 0.44 to 0.90 due to the flash shedding
and collapse of cloud cavitation. Moreover, the average length l/L is
0.68. From these sampling, it is possible to evaluate the frequency of
the cloud cavitation shed and collapse and it can be concluded that
the shedding frequency of cavitation is about 2800 Hz.
4.3. LES analysis of transient dynamic characteristic of cloud cavitation
A typical sequence of the cloud cavitation shedding and collapse
with Pin of 0.26 MPa can be seen in Fig. 8. The images in Fig. 9(a) are
taken using the high-speed camera with a time interval of 10 μs.
Furthermore, the CFD results provide an indication of the driving
mechanism of the cavity shedding. In Fig. 9(b), CFD simulations present
the instantaneous pressure distribution at the slice of x = 0.75 mm and
the resulting cavitation volume fraction iso-surface of 0.1 colored by
black in the front view of the nozzle. Obviously, areas where the
pressure drops below saturation pressure correspond to the presence
of cavitation. From the experimental and simulation results, some
small differences between experiment and simulation can be observed.
Fig. 8. Time evolution of cavitation cloud length.
 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
Fig. 9. Transient shedding of cloud cavitation (Pin = 0.26 MPa, results are shown every 0.1 ms).
For example, some small cloud cavities collapse and stretch to filament-
like cavitation at downstream can be observed in the experiment.
However these phenomena are hardly predicted by the numerical
simulation.
Images 1 and I show that the attached cavitation, labeled A, grow to
the maximum and begin to oscillate more like cloud. Note that, in image
1 some shed cloud cavitation collapse into micro-bubbles which cannot
be seen in the simulation image, it is believed the pressure is higher than
saturation pressure in these zones. The flow reattachment and pressure
pulses due to the cloud cavitation collapse established the re-entrant jet
(labeled arrow C). However, identifying the exact location of the re-
entrant jet is very difficult and it is more likely to locate at the cavity
closure region. After 0.1 ms, the motion of re-entrant moves upstream
and causes the shedding of cloud cavitation as shown in images 2 and
II. At this stage, a larger-scale attached cavitation (labeled B) appears
to be totally separated from the adjacent wall. In succeeding images
the new attached cavitation A continues to grow and the cloud cavita-
tion B entrained by the flow and begins to move downstream. With
the high shear and accompany with the vortex shedding, cloud
cavitation B rotates rapidly downstream and then collapses to small-
scale ring-like vortex cavities due to pressure recovery. It is highlighted
that when the cloud cavitation B continues to move to the outlet, it
collapses violently into micro-scale bubbles as shown in images 4
and IV, thus causing erosion and enhancing the spray cone angle.
Simultaneously, a new re-entrant jet C is formed and the shedding
process is repeated.
The cavitation structures show a good agreement between the LES
calculations and the experiments, which indicates the applicability of
the LES turbulence model for simulation and prediction of the process
of cloud cavitation shed. Although the captured images of the flow
113
field are not as cyclic as the numerical results, it can indicate the
whole process of the shedding, from growth to collapse.
The formation and shedding of cloud cavitation have enormous
influence on the pressure fluctuations in the nozzle orifices. Additional-
ly, pressure fluctuations and flow attachment create the re-entrant jet
which seems to be the reason of cloud cavitation formation and
shedding. Therefore, it is significant for us to investigate the pressure
fluctuations in the nozzle orifices. In this section, the relationship
between unsteady cavitation behaviors and pressure fluctuations is
discussed with the numerical results. The output signal point #1 is
chosen to monitor the fluctuations of pressure signal shape along the
process of cloud cavitation shedding. The point is located at the rear of
attached cavitation where some cloud cavitation will exist or collapse.
Consequently, the pressure fluctuations are affected by cavitation
shedding and collapse. Measurement of the pressure fluctuations in
the experiment is not included here because it is hard to measure the
pressure in a very small and enclosed nozzle orifice. Thus, in this
paper only the numerical results are presented and compared with
the transient cavitation patterns.
Fig. 10 shows the simulated absolute pressure fluctuations at point
#1 around about 1.5 cycle. The point #1 which we have mentioned pre-
viously is also marked in the simulation slice in Fig. 10. The labels A, B, C,
and D are corresponding to the typical pressure values within a cycle of
cavitation evolution, and for better understanding, the experimental
cavitation distributions and the simulated cavitation volume fraction
are also presented. From Fig. 10, it can be seen that the attached
cavitation length reaches to the maximum
after about 1/4 cycle which is shown in image A. Next, with the effect
of re-entrant jet, cloud cavitation formed and became to shed from the
attached cavitation which we can be seen from image B. It is highlighted
114 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116

Fig. 10. Relationship between pressure and unsteady cavitation behaviors.

that during the period before the pressure decreases dramatically, small
local pressure fluctuations induced by the small detached cavities (black
arrows indicate bubbles shed here and red arrows indicate bubbles
collapse or move downstream). Subsequently, after a certain point,
large cloud cavitation shed accompanied the extremely pressure
decrease and then pressure reaches a peak with the collapse of cloud
cavitation near point #1.
From looking at the temporal progress of the generation and collaps-
ing of cavitation, the unsteady behavior is obvious. But it is still difficult
to identify the frequency of cavitation shedding just by one single cycle.
For this reason, Fig. 11 shows the time evolution of the pressure
fluctuations (left) and the fast Fourier transform (FFT) of this progress
(right). Nevertheless, two dominant frequencies can be found. Within
a range of 2650–2700 Hz dominant frequencies are related to the
shedding and collapse of vapor. Further dominance, with higher ampli-
tudes, can be found at frequencies of 5500–6000 Hz which may be
caused by small local bubbles shed and rebound. The small local bubbles
shed and collapse will be investigated in the next step work. Here the
simulated shedding frequency of cloud cavitation is about 2680 Hz,
which is in good agreement with the one identified by the experimental

Fig. 11. Time evolution of the pressure fluctuation (left) and FFT (right) at point #1.
 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
Fig. 12. Velocity of vector during the attached cavitation (t = t0 + 0.1 ms at the slice of
x = 0.75 mm).
measurement. This shows the ability of the numerical method, which is
used in the present work, to simulate cavitating flows and estimate the
shedding frequency by measuring the pressure fluctuations.
It can be concluded that the pressure fluctuations produced by the
cavitation are mostly due to the shedding of cloud cavitation. Moreover,
a sharp crease is found when the cavitation collapse near the point #1.
However, due to the limitation of the present numerical model and
time scale, more detailed information with small bubbles shed and
collapse or cavitation recovery is not predicted. On the other hand, it
still can give a certain prediction of pressure fluctuations where
pressure is hard to measure in such a small nozzle orifice.
As the re-entrant jet is hard to capture and measure in experiments,
CFD simulations provide a more detailed insight. Indeed, detailed
velocity vector distribution during the detachment of cavitation and
effect of the re-entrant jet which is visible in the zoomed region at the
slice of x = 0.75 are shown in Fig. 12. Here the contour denotes
formation between the attached cavitation and the adjacent wall.
Fig. 13. Vapor volume fraction distribution and criterion Q (t = t0 + 0.1 ms at the slice of
x = 0.75 mm).
115
In particular, the re-entrant jet first starts from the cavity closure lo-
cation and due to the adverse pressure gradient it “filling” the cavity
voids and moves towards the nozzle entrance which forces the cavity
to separate from the wall and move downstream rotating. During the
cavity separation, significant vorticity is generated due to the opposing
directions of the bulk of the flow and the re-entrant jet.
One of the main goals of LES is to capture the vortical structures in
the flow. In this study, it is found that vortex shed is always accompa-
nied with shedding of cloud cavitation and cloud cavitation finally
collapses into vortex structure form Fig. 13. For the representation, the
second invariant of the velocity vector Q was used. It is also observed
that the vorticity is primarily deposited locally on the cavity interface
and the sharp edge entrance during the cavitation shedding process
and reaching the maximum values at the interphase of liquid–vapor
region. From that, it is induced that cavitation can enhance the
turbulence in the diesel nozzle which may have a great influence on
the subsequent spray.
In order to regard the effect of cavitation on the velocity field, Fig. 14
shows the velocity profiles in different positions along the nozzle orifice
at the slice of x = 0.75 mm. The dashed line shows the velocity profiles
at the condition of cloud cavitation initiate to form (t = t0) and the solid
line shows the velocity profiles at the condition of cloud cavitation col-
lapsed (t = t0 + 0.3 ms) at the same locations away from the nozzle
inlet. It can be seen that the velocity is relatively low near the nozzle
wall which corresponds to the recirculation zone. Further downstream
the flow reattaches and it has a wider relatively high velocity distribu-
tion and smooth velocity field, as shown at cross-section S3 especially.
The effect of cloud cavitation on the velocity field can be revealed ac-
cording to comparing the dashed line with the solid line. The velocity
profiles are significantly different due to shedding of cloud cavitation.
From the solid line, it can be seen that cloud cavitation shed and collapse
have the fluid velocity more complex. It can be seen that the collapse of
cavitation has led to the appearance of several inflection points in the
velocity field, which are known to be a possible source of instability.
At the outlet section S4,
Fig. 14. Velocity profiles at different positions along the nozzle orifice at the slice of x =
0.75 mm.
116 Z. He et al. / International Communications in Heat and Mass Transfer 76 (2016) 108–116
more oscillation points are found and it also has a wider relatively
high velocity, therefore, could be looked as a possible reason for increas-
ing of the spray cone angle subsequently.
5. Conclusions
In this paper, experimental visualization and LES results for the
cavitation inception and cloud cavitation shedding in a 2-mm-wide
diesel nozzle orifice have been presented. The unsteady cloud cavitation
shedding is studied and the numerical results are compared with the
images of the cavitation captured by the high speed video camera.
Finally, cavitation-induced pressure fluctuations and re-entrant jet are
also analyzed. Based on the present study, the following conclusions
have be drawn:
1) Cavitation inception is formed at the sharp edge of inlet and in the
recirculation zone which is below the shear layer. Moreover, some
vortices shed from the shear layer induced the vortex cavitation.
2) The length of cavitation varies with the shedding of cloud cavitation.
The variation frequency of cavitation lengths is similar to the
frequency of cloud cavitation shedding.
3) Shedding of cloud cavitation in the rectangular nozzle is well
predicted by LES and the motion of re-entrant jet is the main reason
causing cloud shedding.
4) The shedding cavitation generates significant pressure oscillations in
the flow field. Shedding of cloud cavitation induces vortex shedding,
since the highest values of vorticity in the nozzle are found at the
liquid–vapor interphase.
5) Finally, it has been observed that the velocity is more complex when
cloud cavitation shed and collapsed. The oscillation points are most
likely the source of instability after cavitation shed and collapsed
which leading to the observed breakup in cavitating spray.
Acknowledgements
This research was supported by the National Natural Science
Foundation of China (No. 51176066, No. 51276084), a Project Funded
by the Priority Academic Program Development of Jiangsu High
Education Institutions, Six Talent Peaks Project of Jiangsu Province
(2013-JNHB-017) and Specialized Research Fund for the Doctoral
Program of Higher Education (No. 21033227110007).
References
[1] W. Bergwerk, Flow pattern in diesel nozzle spray holes, Proc. Inst. Mech. Engrs, vol.
173, 1959.
[2] Z. He, W. Zhong, Q. Wang, Z. Jiang, An investigation of transient nature of the
cavitating flow in injector nozzles, Appl. Therm. Eng. 54 (1) (May 14 2013) 56–64.
[3] H. Chaves, M. Knapp, A. Kubitzek, et al., Experimental study of cavitation in the noz-
zle hole of diesel injectors using transparent nozzles, SAE Paper No. 950290, 1995.
[4] F. Payri, J. Bermudez, R. Payri, The influence of cavitation on the internal flow and
the spray characteristics in diesel injection nozzles, Fuel 83 (2004) 419–431.
[5] R. Payri, J.M. Garcya, F.J. Salvador, Using spray momentum flux measurements to un-
derstand the influence of diesel nozzle geometry on spray characteristics, Fuel 84
(2005) 551–561.
[6] M. Callenaere, J.P. Franc, J. Michel, M. Riondet, The cavitation instability induced by
the development of a re-entrant jet, J. Fluid Mech. 444 (2001) 223–256.
[7] D.F.D. Lange, G.J.D. Bruin, Sheet cavitation and cloud cavitation, re-entrant jet and
three-dimensionality, Appl. Sci. Res. 58 (1998) 91–114.
[8] K. Hayashi, A. Sou, A. Tomiyama, A volume tracking method based on nonuniform
subcells and continuum surface force model using a local level set function, Comput.
Fluid Dyn. J. 15 (2006) 225–232.
[9] G. Wang, I. Senocak, W. Shyy, T. Ikohagi, S. Cao, Dynamics of attached turbulent cav-
itating flows, Prog. Aerosp. Sci. 37 (2001) 551–581.
[10] A. Kubota, H. Kato, H. Yamaguchi, M. Maeda, Unsteady structure measurement of
cloud cavitation on a foil section using conditional sampling technique, J. Fluids
Eng. 111 (1989) 204–210.
[11] Y. Sugimoto, K. Sato, Visualization of unsteady behavior of cavitation in circular
cylindrical orifice with abruptly expanding part, The 13th International Topical
Meeting on Nuclear Reactor Thermal Hydraulics, 2009.
[12] T.R.G. Stanley, C. Barber, Re-entrant jet mechanism for periodic cavitation shedding
in a cylindrical orifice, Int. J. Heat Fluid Flow 50 (0) (2014) 169–176.
[13] E. Giannadakis, M. Gavaises, C. Arcoumanis, Modelling of cavitation in diesel injector
nozzles, J. Fluid Mech. 616 (2008) 153–193.
[14] A. Alajbegovic, D. Greif, B. Basara, U. Iben, Cavitation calculation with the twofluid
model, 3rd European Japanese Two-phase Flow Group Meeting 21–27 September,
2003.
[15] S. Molina, F.J. Salvador, M. Carreres, D. Jaramillo, A computational investigation on
the influence of the use of elliptical orifices on the inner nozzle flow and cavitation
development in diesel injector nozzles, Energy Convers. Manag. 79 (2014) 114–127.
[16] X. Pan, G. Wan, Z. Lu, Flow field simulation and a flow model of servo-valve spool
valve orifice, Energy Convers. Manag. 52 (2011) 3249–3256.
[17] X. Margot, S. Hoyas, A. Gil, S. Patouna, Numerical modelling of cavitation: validation
and parametric studies, Eng. Appl. Comput. Fluid Mech. 6 (2012) 15–24.
[18] B. Ji, X.W. Luo, R.E.A. Arndt, X.X. Peng, Y.L. Wu, Large eddy simulation and theoretical
investigations of the transient cavitating vortical flow structure around a NACA66
hydrofoil, Int. J. Multiphase Flow 68 (2015) 121–134.
[19] A. Sou, B. Biçer, A. Tomiyama, Numerical simulation of incipient cavitation flow in a
nozzle of fuel injector, Comput. Fluids 103 (2014) 42–48.
[20] C.P. Egerer, S. Hickel, S.J. Schmidt, N.A. Adams, Large-eddy simulation of turbulent
cavitating flow in a micro channel, Phys. Fluids 26 (2014) 085102.
[21] M. Altimira, L. Fuchs, Numerical investigation of throttle flow under cavitating con-
ditions, Int. J. Multiphase Flow 75 (2015) 124–136.
[22] Z. He, Z. Shao, Q. Wang, W. Zhong, X. Tao, Experimental study of cavitating flow in-
side vertical multi-hole nozzles with different length-diameter ratios using diesel
and biodiesel, Exp. Thermal Fluid Sci. 60 (2015) 252–262.
[23] C. Arcoumanis, M. Badami, H. Flora, M. Gavaises, Cavitation in real-size multi-hole
diesel injector nozzles, SAE Trans. J. Eng. (2000) (2000–01-1249 109–3).
[24] M. Germano, A dynamic subgrid-scale eddy viscosity model, Phys. Fluids A 3 (7)
(1991) 1760–1765.
[25] D. Lilly, A proposed modification of the Germano subgrid-scale closure method,
Phys. Fluids A 4 (3) (1992) 633–635.
[26] G.H. Schnerr, J. Sauer, Physical and numerical modeling of unsteady cavitation dy-
namics, Fourth International Conference on Multiphase Flow, New Orleans, USA
2001, p. 1.
[27] W. Yuan, J. Sauer, G.H. Schnerr, Modeling and computation of unsteady cavitation
flows in injection nozzles, Méc. Ind. 2 (2001) 383–394.
[28] M.S. Plesset, A. Prosperetti, Bubble dynamics and cavitation, Annu. Rev. Fluid Mech.
9 (1977) 145–185.