Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

JSAE 20159061
SAE 2015-01-1906

Numerical Investigation of GDI Injector Nozzle

Geometry on Spray Characteristics

Po-Wen Tu
University of Birmingham

Li Cao
Jaguar Land Rover

Hongming Xu
University of Birmingham

Adam Weall
Jaguar Land Rover

Jens Krueger Venus

Shell

Dhananjay Kumar Srivastava, Karl Dean and Daliang Jing

University of Birmingham

Copyright © 2015 SAE Japan and Copyright © 2015 SAE International

ABSTRACT increased counter-bore diameters for any given L/D

ratio. The positive recirculation inside the small
The large eddy simulation (LES) with Volume of Fluid diameter counter-bores improves the mass flow rate
(VOF) interface tracking method in Ansys-FLUENT from the inner hole. Inside the counter-bore cavity and
has been used to study the effects of nozzle hole further downstream from it, large interconnected
geometrical parameters on gasoline direct injection ligaments are present which undergo deformation and
(GDI) fuel injectors, namely the effect of inner hole secondary breakup.
length/diameter (L/D) ratio and counter-bore
diameters on near field spray characteristics. Using INTRODUCTION
iso-octane as a model fuel at the fuel injection
pressure of 200 bar, the results showed that the L/D In order to improve combustion efficiency and reduce
ratio variation of the inner hole has a more significant engine emissions there is an increasing need to
influence on the spray characteristics than the understand the fluid dynamics of fuel sprays and
counter-bore diameter variation. Reducing the L/D develop appropriate models to complement
ratio effectively increases the mass flow rate, velocity, experimental activities. In GDI engines it is well known
spray angle and reduces the droplet size and breakup that the fuel injector nozzle design can strongly
length. The increased spray angle results in wall influence spray characteristics such as breakup length,
impingements inside the counter-bore cavity, initial droplet size, spray angle and exit velocity [1, 2].
particularly for L/D=1 which can potentially lead to In order to optimize the spray pattern of multi-hole GDI
increased deposit accumulation inside fuel injectors. injectors, several factors must be examined: (1) fuel
The influences of the counter-bore diameter become flow rate distribution, (2) detailed nozzle geometry,
more obvious with decreased L/D ratio. For the lower and (3) initial exit boundary conditions such as exit
L/D ratio, the sprays with large included angles are velocity and initial droplet size.
more vulnerable to the air entrained into the
counter-bore cavity and the recirculation flow inside it. The influences of different nozzle geometry have been
Increasing the counter-bore diameter has an overall studied experimentally by several researchers. Su et
adverse effect on the spray characteristics as the al. [3] investigated two different types of diesel
spray exits the counter-bore. Flow rates and spray injectors with rounded inlets (RI) and sharp-edge
angles decrease and droplet sizes are increased with inlets (SEI) and they observed that spray generated
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

by the SEI injector had a higher velocity, longer spray plume angle.
penetration length, smaller droplet size and larger
spray angle. This is because the sharp-edge inlet Many fuel injector nozzle designs incorporate a
resulted in more flow contraction which reduced the counter-bore feature for alleviating the effect of
effective flow area at the nozzle exit. Von Kuensberg deposits in the nozzle. Dearn et al. [19] investigated
Sarre et al. [4] developed a phenomenological nozzle GDI nozzle deposits experimentally using advanced
flow model based on experimental data, to examine analytical techniques applied to a counter-bore design.
the effect of different hole diameters, L/D ratios and The counter-bore can also affect spray characteristics.
R/D ratios. Serras-Pereira et al. [5] also compared Kazour et al [20] reported that the counter-bore of the
two different hole diameters of 0.2 and 0.5 mm, GDI injector affects the spray penetration length. The
reporting that a small diameter injector has a higher spray characteristics of GDI injectors with and without
Reynolds Number (Re= Inertia forces/viscous forces) a counter-bore was also compared in the work of
at the start of injection. Befrui et al [17] who found that the counter-bore
introduces more vortices into the spray flow pattern.
Despite these results indicating the influence of
different nozzle geometries, the detailed flow patterns This paper examines the specific influences from
in the nozzle and at the exit plane are difficult to different L/D ratios and counter-bore sizes on the near
observe experimentally. Computational fluid dynamics flow field of a production GDI injector by using a
(CFD) offers an effective method to study the nozzle VOF-LES model. Furthermore, the initial jet breakup
flow and near field flow. Salvador et al [6,7] and Payri characteristics such as droplet size, spray angle and
et al. [8] validated their CFD models for real-size velocity are compared between different injector
diesel injectors with experimental data by comparing designs in order to investigate the detailed influences
mass flow rate, momentum flux and exit velocity. In of the different nozzle geometries. Recently, Xue [21]
order to understand turbulence and cavitation also coupled the nozzle flow data to spray models by
behavior in the injector, Payri et al. [9] and Salvador et using Lagrangian-Eulerian approach. Therefore, the
al. [10] also used a large eddy simulation (LES) model other objective of the study was to provide accurate
to investigate the distribution of the cavitation zone initial boundary conditions to lagrangian models to
inside a diesel injector. The influences of different improve the accuracy of fuel spray simulations
nozzle geometries such as entrance curvature radius underway at the University of Birmingham.
and orifice inclination angle have also been studied by
Payri et al. [11] and He et al. [12]. They recognized THE COMPUTATIONAL METHOD
that the changes of curvature radius and inclination
angle strongly influenced the distribution of pressure CONSERVATION EQUATION OF MULTI-PHASE
and cavitation inside the nozzle. The effect of wall SYSTEMS
roughness in cavitation flow was studied by
Echouchene et al. [13] and it was shown that the The main underlying theory of the LES model is to
effects of wall roughness are significant with low separate and calculate large eddy and small eddy
injection pressures. scales individually. The establishment of governing
equations for LES is done by filtering the
Volume of Fluid Large-Eddy-Simulation (VOF-LES) time-dependent Navier-Stokes equation in the
has been widely applied in simulations of jet breakup physical space. The ability of the equation to
and nozzle flow. The capability of this method determine the eddy scale size is related to the mesh
includes modeling the effects from liquid surface size. When the eddy scale size is larger than mesh
tension, turbulence and aerodynamics, which are the size, it will be resolved directly by Navier-Stokes
key issues influencing fuel atomization. Delteil et al. equations. However, for the small eddy scale size,
[14] and Fuster et al. [15] used VOF-LES to simulate they will be modeled by using a sub-grid scale stress
primary atomization and Rayleigh breakup of a liquid (SGS) model [22].
jet. They reported that the method can provide
accurate prediction of breakup length and droplet size. The description of the conservation equations of mass
The VOF-LES method has also been used to study and momentum are given by equations (1) and (2).
fuel injectors in order to predict initial droplet size and
velocity distributions, e.g. by E.de Villiers et al. [16] ߲ߩ ߲
who investigated the mechanism of liquid ൅ ሺߩ‫ݑ‬௜ ሻ ൌ Ͳ (1)
߲‫ݔ߲ ݐ‬௜
disintegration in a diesel injector. They found that the
initial perturbation is driven by the Kelvin-Helmholtz
instability mechanism. Befrui et al. [17,19]
߲ ߲ ߲ߪ௜௝ ߲ܲ ߲߬௜௝
investigated internal flow and spray characteristics in ሺߩ‫ݑ‬௜ ሻ ൅ ሺߩ‫ݑ‬௜ ‫ݑ‬௝ ሻ ൌ െ െ ൅ ܵெǡ௜ (2)
GDI injectors using the VOF-LES method; the ߲‫ݐ‬ ߲‫ݔ‬௝ ߲‫ݔ‬௝ ߲‫ݔ‬௜ ߲‫ݔ‬௝
influences of different nozzle L/D ratios, counter-bores
and nozzle tapered geometries were examined [1].
The results show that tapered nozzles increase the where ɏ represents the density, —୧ represents the
breakup length due to smoothing of KH instabilities velocity component in the ‹ th direction,  is the
and nozzles with a smaller L/D ratio result in a shorter pressure, and ୑ǡ୧ represents the gravitational body
breakup length due to increased level of vortices. force. The gravitational body force is approximated by
Shost et al [2] compared two nozzle geometries with using the Boussinesq approximation, so its definition is
different L/D ratios and it was suggested that the ୑ǡ୧ ൌ ሺɏ െ ɏ଴ ሻ‰ ୧ where ɏ଴ is the reference density
nozzles with a smaller L/D ratio have an increased and ‰ ୧ represents the component of gravitational
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018
th
acceleration in the ‹ direction. ɐ୧୨ is the stress The principle of the VOF model is that two or more
tensor due to molecular viscosity defined by equation immiscible fluids can be modeled by solving a single
(3) and ߬௜௝ is the subgrid-scale stress defined by set of momentum equations; the volume fraction of
equation (4). the fluid in the domain is tracked separately. The
tracking of the interface between each phase is based
on the solution of a continuity equation for the volume
߲‫ݑ‬௜ ߲‫ݑ‬௝ ʹ ‫ݑ‬௟ fraction of each phase. For the “୲୦ phase, the volume
ߪ௜௝ ‫ ؠ‬ቈߤ ቆ ൅ ቇ቉ െ ߤ ߜ (3) fraction equation for incompressible fluid is:
߲‫ݔ‬௝ ߲‫ݔ‬௜ ͵ ߲‫ݔ‬௝ ௜௝

୬
ͳ μ
൥ ൫Ƚ ɏ ൯ ൅ ‫ ׏‬ή ൫Ƚ୯ ɏ୯ ˜୯ ൯ ൌ ෍൫ሶ୮୯ െ ሶ୯୮ ൯൩ (9)
߬௜௝ ‫ݑߩ ؠ‬௜ ‫ݑ‬௝ െ ߩ‫ݑ‬௜ ‫ݑ‬௝ (4) ’୯ μ– ୯ ୯
୴ୀଵ

where ሶ୯୮ represents the mass transfer from

LARGE-EDDY SIMULATION METHOD phase“ to phase ’ and ሶ୮୯ is the mass transfer
from phase ’ to phase“. The primary-phase volume
Turbulent flows are characterized by eddies with a fraction is computed based on the following constraint:
wide range of lengths. The large scale eddies typically
represent the characteristic length of the mean flow ୬
and the small scales eddies are responsible for the ෍ Ƚ୯ ൌ ͳ (10)
dissipation of turbulence kinetic energy. In the LES ୯ୀଵ
model, as mentioned before, the large scale eddies
are directly resolved, while the small scales eddies The properties in the transport equation are calculated
are modeled. The momentum, mass and energy, are according to the presence of each component phase
transported by large scale eddies, and these large in each control volume. The definitions of the mixture
scale eddies are more problem-dependent. They can thermo-physical properties are as below:
reflect the influence of different geometries and
boundary conditions. ɏ ൌ Ƚ୯ ɏ୯ ൅ ൫ͳ െ Ƚ୯ ൯ɏ୮ (11)
The sub-grid scale stress turbulence model employs a
In general, for an  -phase system, the
Boussinesq hypothesis computing sub-grid scale
volume-fraction-average density takes the following
stress term,߬௜௝ , which can be expressed as:
form:
ͳ
߬௜௝ െ ߬௞௞ ߜ௜௝ ൌ െʹߤ௧ ܵ௜௝ (5) ɏ ൌ ෍ Ƚ୯ ɏ୯ (12)
͵

where ߤ௧ is the sub-grid scale turbulent viscosity. The

isotropic part of the sub-grid scale stresses ߬௞௞ is not THE NUMERICAL SOLUTION METHOD
modeled in this equation, but it is calculated to filter
the pressure term. The ܵ௜௝ term is the rate-of-strain All computations were done using ANSYS FLUENT to
tensor for the resolved scale which is defined by solve the LES-VOF conservation equations and they
equation (6). combine the special treatment for phase fraction
equations. The numerical method used a
ͳ ߲‫ݑ‬௜ ߲‫ݑ‬௝ second-order spatial and temporal discretization
ܵ௜௝ ‫ؠ‬ ቈ ൅ ቉ (6)
ʹ ߲‫ݔ‬௝ ߲‫ݔ‬௜ scheme but preserved the proper limits on physical
fluid dynamic variables.
The SGS model in this paper is a dynamic kinetic
energy sub-grid scale model. The sub-grid scale The Pressure-Implicit with Splitting Operators (PISO)
turbulence can be modeled by accounting for the pressure-velocity coupling scheme is based on the
transport of sub-grid scale turbulent kinetic energy. higher degree of the approximate relation between the
corrections for pressure and velocity.
The definition of sub-grid scale kinetic energy can be
expressed as below:
COMPUTATIONAL DOMAIN AND MESH
ͳ ଶ
݇௦௚௦ ൌ ቀ‫ݑ‬௞ଶ െ ‫ݑ‬௞ ቁ (7) A range of different nozzle geometries were
ʹ
investigated using a sensitivity analysis approach. In
The sub-grid scale eddy viscosity, ߤ௧ , is computed total 9 cases were modeled, with different L/D ratios
using ݇௦௚௦ as follows: and d/D ratios given in Figure 1. In order to investigate
the influences from different L/D ratios and
ଵ ଶ Τ counter-bore sizes, three different L/D ratios are
ߤ௧ ൌ ‫ܥ‬௞ ߩ݇௦௚௦ ο݂ܵ௜௝ (8)
discussed separately 1.0, 1.3 and 1.6, with different
d/D ratios of 1.85, 2.0 and 2.15.
VOF-BASED INTERFACE-TRACKING METHOD
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

Table 1. Nozzle flow turbulence length scales and time scale

Injection pressure 150bar 100bar 50bar

1
§ v3 · 4
K Kolmogorov[m] ¨¨ ¸¸ 1.42E-07ġġ 1.66E-07ġ 2.16E-07ġ
©H ¹
1
§ 15vu 2 · 2
OTaylor[m] ¨¨ ¸¸ 1.46E-05ġ 1.68E-05ġ 2.02E-05ġ
© H ¹
3

Figure 1 Case distribution of different L/D and d/D Lint egral[m] K Re 4 0.000353ġ 0.000363ġ 0.000365ġ

Figure 2 shows the three dimensional GDI geometry 1

and computational mesh used in the CFD simulation §v ·2
which includes the valve-group flow, counter-bore and tK [ s ] ¨¨ ¸¸ 2.51E-08ġ 3.46E-08ġ 5.86E-08ġ
near-field region. The computational mesh consisted ©H ¹
of approximately 3 million cells. It is acknowledged
that some of relevant physics such as the influence of
grooves upstream of the inner hole on the flow filed
were sacrefised to have reasonable computation time
for the modelTo increase numerical accuracy and BOUNDARY CONDITIONS
computational efficiency, the mesh size was varied in
the range of 2-5 Pm in the research zone (internal Without considering needle motion, the initial
area, counter-bore area and the near field area at a condition of liquid distribution was that the liquid filled
distance 5L from the counter-bore exit) and up to the domain up to the internal nozzle hole entrance.
10Pm in the remaining computational domain. The nozzle hole volume and pressure outlet volume
domains were filled with air. Dimensions of the fuel
Table1 presents the range of length scales (eddy cavity between the needle and above the inner hole
scale) under different injection pressures which can were much larger than the dimensions of the inner
provide a guide to establish mesh resolution in order hole. Thus, flow velocities were small in the cavity
to capture turbulent effects and small eddies [23]. compared to the inner hole. Flow entrance into the
From table 1, the smallest eddy size could be estimate inner hole produces significant vortices and eddies
as Ʉǡ called Kolmogorov length, and the largest eddy produced entrain upstream liquids into the inner hole.
size was estimated as L . Another turbulent length In order to include the effects of these vortices, the
inner hole had to be assumed to be empty at the initial
scale, the Taylor length,ɉ, offered a calculation for the
state. By assuming a full inner hole for the initial
fluctuating strain rate field. Also, the time scale for the
state, the full effects of these vortices could not be
smallest eddy was estimated as ‫ݐ‬ఎ indicating the
captured.
duration of eddy structure. According to ‫ݐ‬ఎ from Table
1, the time step size of the CFD model was set as The boundary conditions applied in this paper are
1e-8. described as follows:

In this paper, the movement of the injector needle was x Inlet: The fuel injection pressure was set to
not considered due to the high computational time and 200bar for the sensitivity study.
high mesh quality required for the moving mesh. The
pintle motion would change the aspect ratio of grid x Outlet: At ambient pressure, the length of the
which could cause numerical divergence during ambient domain was designed to be 10
calculation. diameters which would provide enough
distance for jet development and avoid the
reflection of pressure disturbance from the
wall as much as possible

x Test fuel: Iso-octane was used in all

simulations at ambient temperature. Its
physical properties are listed in Table 2.

Figure 2 GDI geometry and computational mesh

 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

Table 2. Fuel properties of Iso-octane ANALASYS METHODOLOGY OF FUEL SPRAY

CHARACTERISTICS
Name(s) Iso-Octane
This section introduces the methodologies used to
Molecular analyze the data from the flow field simulation. Figure
C8H18
Formula 4 presents the method used to measure the spray
angle and breakup length. These two parameters can
Molar Mass 114.2 g/mol
be used to analyze the intensity of atomization. Spray
-3 angle was used to define the droplet spatial
Density 702.6 kg/m
distribution in the combustion chamber. Breakup
length represents the liquid core length with which
Vapor Pressure 5500 Pa (at 21°C)
the liquid jet remains intact before it begins to break
-10 up into ligaments and droplets.
6.089x10 k Pa s
Viscosity
For spray angle, it is defined as the liquid dispersed
angle which is measured from the nozzle entrance
Surface Tension 0.01816 N/m and it includes the downstream liquid distribution. For
breakup length, it is the measured length of liquid jet
from nozzle inner hole exit to the end of liquid jet
which can be seen on the right side of Figure 4. The
METHODOLOGY OF DATA ANALYSIS liquid jet which was presented by VOF=0.5 for which
0.5 is the common value to represent liquid was for
VALIDATION OF THE CFD CODE measurement.

It is extremely difficult to directly measure the flow

characteristics inside a real-size nozzle hole and in
the spray near field region. Experimental studies using
optical nozzle technology to evaluate the internal flow
field have been limited due to the difficulty of tolerating
real injector fuel pressures and nozzle geometries
using transparent materials. Therefore, in this paper,
the injector static flow rate measured from a flow test
bench is used for comparison with the predicted flow
rate from the CFD model. Because the movement of
the injector needle has been ignored, the flow rate test
from the test bench used a pulse time of 3 ms for all
injection pressures to obtain average static flow data
in order to validate the CFD model. Figure 3 shows Figure 4 Methodologies to analyze spray angle and breakup
length
the comparison between experimental data and CFD
results at different injection pressures (50 bar, 100bar
The near field spray characteristics are very important
and 150 bar). From this comparison, the errors
for gaining insight into the primary breakup
between experiment and simulation are 6% for 50bar,
mechanism. In this analysis, 4 measuring planes were
7.2% for 100 bar and 2% for 150bar. The discrepancy
set up to analyze droplet size near the nozzle in order
between measurements and calculated results at
to better understand the near field spray.
different injection pressures was within the error range
of experimental measurements. From these results,
the Ansys-Fluent CFD model is shown to have the The modeling approach provides complementary
capability to simulate the nozzle flow in terms of mass insight into the spray, overcoming the limitations of
flow rate. experimental measurement approaches. For spray
near the nozzle hole, a Phase Doppler Particle
Analyzer (PDPA) cannot accurately measure the
3.0
Experimental data fuel flow characteristics such as droplet size and
CFD simulation droplet velocity due to the high liquid density. The
2.5 recommended reliable measuring distance is 30
mm from nozzle tip in the example of PDPA.
Mass flow rate (g/s)

2.0

1.5 Figure 5 is a cross plane image for VOF=0.5 which

indicates the red region as liquid and the blue region
1.0 as air. The calculation of droplet size is as below:

0.5

0.0
50 bar 100 bar 150 bar

Figure 3 Validation of mass flow rate

 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

RESULTS AND DISCUSSION

The distributions of VOF at 30 Ps for the 9 cases

investigated are presented in Figure 7. These images
were used to provide a better understanding of flow
patterns inside the counter-bore. It is shown that the

distribution of VOF=0.5 occurs at the counter-bore exit

plane (B-B) 30 Ps from the start of fuel entering the
Figure 5 Methodologies to analyze droplet size nozzle. With increased L/D ratio of the inner hole
(from case 1 to case 3, case 4 to case 6, and case 7
The dimensional effects of the two main parts of a
stepped hole GDI injector, namely the inner hole and to case 9 as shown in table 1), the liquid wall
counter-bore, on the injection characteristics such as impingement and wall attachment were decreased as
the mass flow rate, velocity, spray angle, breakup shown in Figure 8. It can be observed that the nozzle
length and droplet size are presented and discussed larger L/D ratio had less liquid on the nozzle wall. This
in this section. For the purpose of simulations, the is because of smaller spray angle and steady velocity
dimensional parameters of the inner hole and development inside the liquid. The more detail
counter-bore were considered as non-dimensional discussed were explain in the following section.
parameters of L/D ratios for the inner hole and d/D for
the counter bore, as shown in Figure 1. However, the
inner hole diameter and nozzle length (internal length
+counter-bore length) were kept constant. In the
analysis of the results the two parameters used are
the L/D ratio and counter-bore diameter. Figure 6
shows the locations of measuring planes in the
computational domain used to analyze the simulated
flow field data:

x A-A: the internal hole exit plane,

x B-B: the counter-bore exit plane,

x C-C: the cross plane at 0.2 mm from the

counter-bore exit plane,

x D-D: the cross plane at 0.4 mm from the

counter-bore exit plane.

Figure 8 Comparison of liquid distribution used for droplet size

calculation at the B-B plane at 30 Ps from the start of fluid flow into
the nozzle

Figure 6 Location of measuring planes used for data analysis

 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

Figure 7 VOF distribution at 30 Ps for different cases

INFLUENCE OF INNER HOLE L/D RATIO L/D ratios are presented in Figure 10. The nozzle
holes with smaller L/D ratios produced higher mass
Mass flow rate variations for Cases 1, 2 and 3 with flow rates, indicating a smaller restrictive effect on the
respect to time were calculated at the A-A plane and average mass flow rate as L/D decreased. The small
they are presented in Figure 9 . The mass flow rate of discrepancies in effect of L/D on the averaged
Case 1 appears more unsteady compared to the other simulated mass flow rates for different d/D values is
two cases with larger L/D ratios. This is most likely within the range of LES error.
due to the combined effects of numerical instability
caused by using LES model which produces slightly Cell based averaging of the liquid velocity at the A-A
different results each time it is executed, and the plane for the same three cases are presented in
physical nature of the in-nozzle reattachment flow Figure 11. The flow rate consistently reduced as L/D
downstream of the recirculation region. This flow increased. For L/D=1.3 and 1.6, the exit velocities
instability is reduced with increasing inner hole length, were more stable with exit velocities of 180 m/s for
since the reattached flow becomes more stable further L/D=1.3 and 172 m/s for L/D=1.6. The exit velocity for
away from the recirculation zone, caused by the sharp L/D=1.0 was unstable and it varied from 185 to 194
entrance edge rim. m/s reflecting its three dimensional nature.

4.5 L/D=1.0
L/D=1.3
4.0
L/D=1.6
3.5
Mass Flow rate (g/s)

3.0
2.5
2.0
1.5
1.0
0.5
0.0
d/D=1.85 d/D=2.0 d/D=2.15

Figure 10 Average simulated mass flow rates for different L/D and
Figure 9 Mass flow rates with respect to time for cases 7, 8 and 9 d/D ratios

In order to compare the differences between the mass

flow rates for different L/D ratios, average mass flow
rates were calculated over the time interval of 20 to 35
Ps which was after the initial transition time, as shown
in Figure 9. Simulated mass flow rates for the different
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

reviewing all the trend, the results indicate that smaller

250 L/D=1.0 L/D ratios give rise to improved jet atomization with
L/D=1.3 decreased breakup lengths.
L/D=1.6
200

Velocity (m/s) 150 0.60

L/D=1.0
0.55
L/D=1.3
0.50
100 L/D=1.6

Breakup length (mm)

0.45
0.40
50 0.35
0.30
0.25
0
d/D=2.15 0.20
d/D=1.85 d/D=2.0
0.15
0.10
Figure 11 Exit velocities at A-A plane for different L/D and d/D 0.05
ratios 0.00
d/D=1.85 d/D=2.0 d/D=2.15
Effects of L/D ratio on spray angle are presented in
Figure 12. Spray angles were measured and
averaged at five different times after the mass flow Figure 13 Breakup lengths for different L/D and d/D ratios
had become steady. The smallest L/D ratio resulted in
the largest spray angle and the spray angle tended to
decrease as L/D ratio increased. Higher exit velocities The method used to determine nominal droplet sizes
at the A-A plane associated with smaller L/D ratios is based on calculating the ratio of the liquid area to its
lead to more intensive interactions between air and circumference along any cross-section as described
the liquid in the counter-bore region, resulting in better earlier, although it cannot distinguish between
breakup of the liquid jet into droplets and wider spray droplets and ligament. However, it is a good indicator
angles. In shorter inner holes, the reattached flow of atomization quality. The estimated droplet size at
could not be fully developed into one directional flow the internal hole exit (plane A-A) for different L/D
when it exited the inner hole, and its original three ratios are presented in Figure 14. The initial droplet
dimensional nature lead to wider spray angles. size for L/D= 1.0 is the smallest (84 Pm to 96Pm). The
upstream geometry of the injector (see Figure 4)
L/D=1.0
forces the flow to make a sharp right turn to enter the
40
L/D=1.3 internal hole, resulting in a region prone to flow
Average spray angle (Degree)

35 L/D=1.6 detaching on the upper right side of the internal hole

wall. For small L/D ratios, the liquid did not reattach to
30
the wall before it exited the internal hole (plane A-A),
25 resulting in the reduction of the effective exit area and
20 thus higher exit velocities and smaller droplet sizes.
15
For the longer nozzle length (L/D=1.6), the initial
droplet size is almost equal to the hole diameter
10
indicating full attachment of the liquid jet to the internal
5 hole wall.
0
d/D=1.85 d/D=2.0 d/D=2.15 L/D=1.0
120 L/D=1.3
Figure 12 Average spray angles different L/D and d/D ratios L/D=1.6
100
Droplet diameter (Pm)

Figure 13 presents the comparison of breakup lengths

80
for different L/D ratios. The VOF-LES model was used
to simulate the transient flow field and the average 60
breakup lengths were calculated for different times. It
is shown that the averaged breakup length tended to 40
increase with the L/D ratio except for Case 5 (L/D=1.3,
d/D=2.0) which did not follow the same trend. This 20

could be due to the competing effects of liquid jet’s

0
turbulent intensity verses the damping of aerodynamic
d/D=1.85 d/D=2.0 d/D=2.15
force by the doughnut shape recirculation zone.
Higher liquid jet turbulent level (low L/D) results in
shorter breakup length. Stronger recirculation zone
Figure 14 Droplet sizes at plane A-A for different L/D and d/D
(larger d/D) reduces the velocity difference of the ratios
ambient and the liquid jet which reduces the
aerodynamic force and increases the breakup length. Figure 15, 16 and 17 present droplet sizes at different
Case 5 seems to be the condition where the damping measuring planes. Droplet sizes all decreased when
effect of the recirculation zone have had a dominant the measuring plane was moved away from the inner
effect compared to other cases. However, by hole exit plane. Comparing Figure 15 and 16, the
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

droplet size at counter-bore exit at plane B-B was

L/D=1.0
bigger than that at plane C-C, 0.2 mm downstream of 25 L/D=1.3
the counter-bore exit. This observation is consistent L/D=1.6
with previous experimental results obtained by other 20

Droplet diameter (Pm)

researchers [24] and is explained by further
atomization caused by aerodynamic and turbulence 15
effects resulting in Kelvin-Helmholtz (KH) instabilities.
10
For case 1 with L/D=1.0, its droplet sizes were larger
than those for cases 2 and 3. However, the droplet
5
sizes for case 1 had a larger size reduction (40%) as
they moved from plane B-B to plane C-C than for
0
cases 2 and 3 which were reduced by only 25%. This
d/D=1.85 d/D=2.0 d/D=2.15
is another indication of stronger atomization for
shorter L/D ratios caused by higher jet velocity, air
entrainment and possible cavitation effects. For the
Figure 17 Comparison of droplet size at measuring plane D-D for
cases with L/D=1.3 and 1.6, trends of droplet size different L/D and d/D ratios
increases were observed when L/D ratio was
increased. The comparison of droplet sizes at different
measuring planes indicated that the inner hole length INFLUENCE OF COUNTER-BORE DIMENSIONS
directly affected the droplet size; smaller L/D ratios
had smaller droplet sizes. This is because of the The influences of different counter-bore dimensions
shorter nozzle length could not provide enough length on spray characteristics such as droplet velocities,
for the three dimensional flow to fully develop resulting diameters, spray angle and breakup length are
in higher liquid turbulent level at the exit plane and presented and analyzed in this section. The
caused stronger liquid atomization. comparison of mass flow rates for different
counter-bore dimensions have been presented in
L/D=1.0 Figure 10. Injectors with smaller counter-bore
60 L/D=1.3 diameters have higher mass flow rates. This
L/D=1.6 observation can be explained by the fact that a liquid
jet flowing through the counter-bore produces a
Droplet diameter (Pm)

40 positive circulating air flow inside the counter-bore

cavity which reduces the flow resistance and
increases the total mass flow rate from the inner hole ,
in Figure 18. However, when the counter-bore
20 diameter is increased, air entrainment into the
counter-bore cavity is increased which reduces the
recirculation velocity and its positive effect on the flow
0 rate of the liquid jet.
d/D=1.85 d/D=2.0 d/D=2.15
The velocities at the inner hole exit (plane A-A) for
different counter-bore diameters are presented and
Figure 15 Comparison of droplet size at measuring plane B-B for compared in Figure 11. The effects of the
different L/D and d/D ratios counter-bore diameter on the inner hole exit velocity
L/D=1.0
at plane A-A are less pronounced for larger L/D ratios.
40 L/D=1.3 A clear reduction of the exit velocity is observed for
L/D=1.6 L/D=1 with increased counter-bore diameter. This
reduction can be explained by the fact that for larger
Droplet diameter (Pm)

counter-bore sizes more air is entrained into the

counter-bore cavity and it is thus harder to generate a
20 positive circulation. However, the influences from
different counter-bore diameters are negligible for the
L/D=1.3 and 1.6 cases. Since the recirculation zone
is induced by the velocity difference of the liquid jet
and the ambient, for longer L/D with lower jet velocity
the induced recirculation zone became week and
0
consequently it had less effect on the exiting jet.
d/D=1.85 d/D=2.0 d/D=2.15
Therefore, the exit jet velocity and the recirculation
zone feed each other and certain level of jet velocity is
required to induce strong enough recirculation zone
Figure 16 Comparison of droplet size at measuring plane C-C for
different L/D and d/D ratios and result in measurable differences in both mass
flow rates and atomization quality.

Exit velocities at the counter-bore exit plane B-B are

presented in Figure 19. The exit velocity at plane B-B
decreased with decreased counter-bore diameter for
L/D=1, because the corresponding spray angles were
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

larger, and for small counter-bore diameters, sprays which has an adverse effect on the droplet velocities.
impinged on the counter-bore wall resulting in loss of These are why the velocity at B-B was decreased with
momentum and exit velocity at plane B-B. However, increasing d/D as shown in Figure 19. For L/D=1.3,
for L/D=1.6 the spray angles were smaller and they spray angles are believed to be in a range that would
could pass through the counter-bore without result in intermittent impingements and thus there is
impinging on the wall. Larger counter-bore diameters no clear trend observed for the exit velocity with
have more entrained air and weaker recirculation different counter bore sizes.

Figure 18 Distribution of streamline at different d/D ratios

250
d/D=1.85
d/D=2.0
200 d/D=2.15
Velocity (m/s)

150

100

50

Figure 21 Schematic of air entrainment and recirculation zone in

0
the counter-bore
L/D=1.0 L/D=1.3 L/D=1.6

A comparison of velocity reduction as fluid passes

Figure19 Comparison of inner hole velocity at B-B for different
d/D and L/D ratios through the counter-bore with different diameters has
been made and the results are presented in Figure 20.
In accordance with the previous observation for L/D
40
=1 which produced large spray angles, the velocity
d/D=1.85 reduction through the counter-bore with the small
35 d/D=2.0 diameter (d/D <2) was very high (>30%) indicating the
d/D=2.15
Veolcity reduction (%)

30 significant effect of wall impingement on both velocity

25 and momentum. For the same case of L/D=1.0,
increasing the counter-bore diameter (d/D=1.85 to
20
d/D=2.15) substantially reduced the velocity loss by
15 as much as 90%, which indicates that for this case
10 there was little wall impingement. For the highest L/D
ratio (L/D=1.6) which produced the smallest spray
5
angles, increasing the counter bore diameter resulted
0 in increased velocity drop for the droplets passing
L/D=1.0 L/D=1.3 L/D=1.6 through the counter-bore, indicating the adverse effect
that the air entrained into the counter-bore cavity had
on the spray flow rate. Velocity drops observed for the
Figure 20 Velocity reduction from the inner hole exit A-A to the case of L/D=1.3 were small but had a decreasing
counter-bore exit B-B
trend with increased counter-bore diameter, which
indicated the possibility of slight wall impingement, but
with a substantially smaller effect on velocity loss.

Effects of the counter bore diameter on the spray

angle variation for different L/D ratios can be observed
from Figure 12. For all cases of L/D ratios investigated,
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

the spray angles were initially reduced when the B-B, the droplet size variation pattern with changing
counter-bore diameter was increased (from d/D=1.85 counter-bore diameter were similar to plane B-B,
to d/D=2.0) because of the spray velocity reduction except droplet sizes were reduced due to further
caused by the entrained air and establishment of a breakup. Downstream at the D-D plane, droplet sizes
torus shape recirculation zone as shown in Figure 21. were further reduced. The cases with the largest
However, a further increase of the counter-bore droplet size at plane C-C had the most size reduction
diameter (from d/D=2.0 to d/D=2.15) did not affect the at plane D-D. The relative distance from D-D to the
spray angles which indicated that larger counter-bore inner hole exit plane is more than 3 inner hole
did not increase the strength of the recirculation zone diameter which made ligament breakup (secondary
and its influence on the spray. breakup), a justifiable reason for the observed size
reductions, Figure 7.
Another important spray parameter to consider in
investigation of the counter-bore effects on spray is
the breakup length, as shown in Figure 13. For all d/D=1.85
cases considered, breakup lengths increased with 120 d/D=2.0
increased counter-bore diameter (d/D=1.85, 2.0 and d/D=2.15
2.15). Increasing the counter-bore diameter 100

Droplet diameter (Pm)

increased the amount of air entrained into the
80
counter-bore cavity resulting in liquid flow velocity
reduction which was accompanied by weakening of 60
the atomization process. A weak atomization process
produced larger droplets and increased the volume 40
ratio of ligaments to droplets. Many of these ligaments
were interconnected and went through deformation 20
and further breakup (secondary breakup) with an
0
overall effect of increased breakup length.
L/D=1.0 L/D=1.3 L/D=1.6

Figure 22 to 25 provide the results for the comparison

of droplet sizes for different counter-bore diameters at
Figure 22 Comparison of droplet sizes at measuring plane A-A
different measuring planes. From Figure 22, it can be
seen that increased counter-bore diameter had very
little effect on the droplet sizes at plane A-A which was
expected since the counter-bore diameter has a
downstream effect on the exiting jet at plane A-A and d/D=1.85
60 d/D=2.0
upstream of this plane was mostly liquid
(incompressible) whereas the flow downstream was d/D=2.15
mostly gaseous (compressible). Careful analysis of
Droplet diameter (Pm)

Figure 22 reveals that the increased counter-bore 40

diameter had a stronger effect on reducing the droplet
size for L/D=1 than for the other two cases of L/D=1.3
and 1.6, which was due to the increased resistance
20
caused by the increased air entrained into the counter
bore.

At the counter-bore exit plane B-B, droplet sizes were 0

decreased with increasing counter-bore diameter for L/D=1.0 L/D=1.3 L/D=1.6
L/D=1, as shown in Figure 23. This is due to the fact
for the short L/D ratio; the exiting liquid jet from the
inner hole was less developed and had a stronger Figure 23 Comparison of droplet sizes at measuring plane B-B
three dimensional nature with short breakup length
which made it more vulnerable to the resistance
d/D=1.85
caused by the increased entrained air. For L/D=1.3, 40 d/D=2.0
the more developed exiting jet from the inner hole d/D=2.15
made the liquid body in the counter-bore cavity more
Droplet diameter (Pm)

resistant to the entrained air and almost no changes in

the droplet size diameter were observed with the
increased counter-bore diameter. On the other hand,
20
droplet sizes for the larger L/D=1.6 with lower
velocities (Figure 17) increased with increasing
counter-bore diameter, because the increased
entrained air further reduced the jet velocity and
weakened the atomization.
0
L/D=1.0 L/D=1.3 L/D=1.6
Droplet sizes at two locations plane C-C and D-D
(Figure 7) downstream of the B-B plane are presented
in Figure 24 and 25 respectively. Since plane C-C was
Figure 24 Comparison of droplet sizes at measuring plane C-C
only 0.20 mm away from the counter-bore exit plane
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

d/D=1.85
Overall the results will provide the much needed initial
25 d/D=2.0 conditions for the Lagrangian modeling of the spray,
d/D=2.15 vaporization and combustion downstream. This study
20 lays down the foundation for investigating the effects

Droplet diameter (Pm)

of injection pressure, fluid properties and geometry
15 upstream of the inner hole on spray patterns.

10
A summary of the influence of different design
parameters used in the sensitivity study are shown in
table 3
5

Table 3 Summary of influence of different L/D ratio and CB

0 diameter
L/D=1.0 L/D=1.3 L/D=1.6

L/D increase CB increase

Figure 25 Comparison of droplet size at measuring plane D-D
Mass flow rate

Exit velocity Similar

CONCLUSION Breakup length

The sensitivity of main spray characteristics to the Spray angle Similar
internal nozzle geometry of a GDI fuel injector has
been investigated numerically with CFD using a
Droplet size Discuss
VOF-LES model. Influences of different L/D ratios and
counter-bore diameters on spray characteristics have
been analyzed. Key findings which have added value
and importance for future applications in injector flow
simulations and its geometrical design are:

1. The L/D ratio variation of the inner hole has a REFERENCES

more significant influence on the spray
characteristics than the counter-bore diameter 1. %HIUXL % &RUELQHOOL * 6SLHNHUPDQQ 3 6KRVW 0
variation. Reducing the L/D ratio effectively HW DO /DUJH (GG\ 6LPXODWLRQ RI *', 6LQJOH+ROH
increases the mass flow rate, velocity, spray )ORZ DQG 1HDU)LHOG 6SUD\ 6$( ,QW - )XHOV /XEU
angle and reduces the droplet size and breakup   GRL
length. The increased spray angle results in wall 2. Shost, M. A., Lai, M.-C., Befrui, B., Spiekermann,
impingements inside the counter-bore cavity, P., et al. "Gdi Nozzle Parameter Studies Using
particularly for L/D=1, which will potentially lead LES and Spray Imaging Methods." SAE
to increased injector deposit accumulation. International, 2014-01-1434, 2014,
doi:10.4271/2014-01-1434
2. The influence of the counter-bore diameter 3. Su, T.F., Farrell, P.V. and Nagarajan, R.T.
becomes more obvious with decreased L/D ratio. "Nozzle Effect on High Pressure Diesel Injection,"
For the lower L/D ratio, the sprays with large SAE 952360, 1995
angles are more vulnerable to the air entrained 4. Von Kuensberg Sarre, C., Kong, S.-C., and Reitz,
into the counter-bore cavity and the recirculation R. D. "Modeling the Effects of Injector Nozzle
flow inside it. Geometry on Diesel Sprays." SAE International,
1999-01-0912, 1999, doi:10.4271/1999-01-0912
3. Increasing the counter-bore diameter has an 5. Serras-Pereira, J., van Romunde, Z., Aleiferis, P.
overall adverse effect on the characteristics of G., Richardson, D., et al. "Cavitation, Primary
the spray exiting the counter-bore and further
Break-up and Flash Boiling of Gasoline,
downstream from it. Flow rates and spray angles
Iso-Octane and N-Pentane with a Real-Size
decreased and droplet sizes increased with
Optical Direct-Injection Nozzle." Fuel 89 (2010),
counter-bore diameters for different L/D ratios.
But the positive recirculation inside the pp. 2592-607
counter-bores with small diameters increased 6. Salvador, F. J., Romero, J. V., Roselló, M. D., and
the mass flow rate from the inner hole. Martínez-López, J. "Validation of a Code for
Modeling Cavitation Phenomena in Diesel Injector
4. Analysis of the simulation results reveals that Nozzles." Mathematical and Computer Modelling
inside the counter-bore cavity and further 52 (2010), pp. 1123-32
downstream from it, large interconnected 7. Salvador, F. J., Hoyas, S., Novella, R., and
ligaments are present which go through Martínez-López, J. "Numerical Simulation and
subsequent deformation and secondary Extended Validation of Two-Phase Compressible
breakup. Flow in Diesel Injector Nozzles." Proceedings of
the Institution of Mechanical Engineers, Part D:
 Downloaded from SAE International by University of British Columbia, Wednesday, August 01, 2018

Journal of Automobile Engineering 225 (2011), pp. 15. Fuster, D., Bagué, A., Boeck, T., Le Moyne, L., et
545-63 al. "Simulation of Primary Atomization with an
8. Payri, F., Payri, R., Salvador, F. J., and Octree Adaptive Mesh Refinement and Vof
Martínez-López, J. "A Contribution to the Method." International Journal of Multiphase Flow
Understanding of Cavitation Effects in Diesel 35 (2009), pp. 550-65
Injector Nozzles through a Combined 16. De Villiers, E., Gosman, A. D., and Weller, H. G.
Experimental and Computational Investigation." "Large Eddy Simulation of Primary Diesel Spray
Computers & Fluids 58 (2012), pp. 88-101 Atomization." SAE International, 2004-01-0100,
9. Payri, R., Gimeno, J., Marti-Aldaravi, P., and 2004, doi:10.4271/2004-01-0100
Bracho, G. "Study of the Influence of the Inlet 17. Befrui, B., and D'Onofrio, M. "Primary Atomization
Boundary Conditions in a LES Simulation of of a GDI Multi-Hole Plume Using Vof-Les
Internal Flow in a Diesel Injector." Mathematical Method." SAE International, 2014-01-1125, 2014,
and Computer Modelling 57 (2013), pp. 1709-15 doi:10.4271/2014-01-1125
10. Salvador, F. J., Martínez-López, J., Romero, J. V., 18. Befrui, B., Corbinelli, G., D'Onofrio, M., and Varble,
and Roselló, M. D. "Computational Study of the D. "GDI Multi-Hole Injector Internal Flow and
Cavitation Phenomenon and Its Interaction with Spray Analysis." SAE International, 2011-01-1211,
the Turbulence Developed in Diesel Injector 2011, doi:10.4271/2011-01-1211
Nozzles by Large Eddy Simulation (Les)." 19. Dearn, K., Xu, J., Ding, H., Xu, H. et al., "An
Mathematical and Computer Modelling 57 (2013), Investigation into the Characteristics of DISI
pp. 1656-62 Injector Deposits Using Advanced Analytical
11. Payri, R., Margot, X., and Salvador, F. J. "A Methods," 6$( ,QW - )XHOV /XEU   ,
Numerical Study of the Influence of Diesel Nozzle , doi..
Geometry on the Inner Cavitating Flow." SAE 20. Kazour, J., Befrui, B., Husted, H., Raney, M., et al.
International, 2002-01-0215, 2002, "Innovative Sprays and Particulate Reduction with
doi:10.4271/2002-01-0215 GDI Injectors." 2014, doi:10.4271/2014-01-1441
12. He, Z., Zhong, W., Wang, Q., Jiang, Z., et al. 21. Xue, 4., %attistoni, M., SoP, S., 4uan, S. et al.,
"Effect of Nozzle Geometrical and Dynamic "(ulerian C)D Modeling of Coupled 1o]]le )loZ
Factors on Cavitating and Turbulent Flow in a and Spray Zith 9alidation Against X5ay
Diesel Multi-Hole Injector Nozzle." International 5adiography Data," 6$( ,QW - (QJLQHV
Journal of Thermal Sciences 70 (2013), pp.   , , doi..
132-43 22. Ansys-Fluent 14.5 document
13. Echouchene, F., Belmabrouk, H., Le Penven, L., 23. Befrui, B. and D'Onofrio, M., "Primary Atomization
and Buffat, M. "Numerical Simulation of Wall of a GDi Multi-Hole Plume Using VOF-LES
Roughness Effects in Cavitating Flow." Method," SAE Technical Paper 2014-01-1125,
International Journal of Heat and Fluid Flow 32 2014, doi:10.4271/2014-01-1125.
(2011), pp. 1068-75 24. Tu, P., Jiang C., Ding, H., Li, C. et al.,
14. Delteil, J., Vincent, S., Erriguible, A., and "Investigation on the Spray Characteristics of
Subra-Paternault, P. "Numerical Investigations in DMF-Isooctane Blends using PDPA," SAE
Rayleigh Breakup of Round Liquid Jets with Vof Technical Paper 2014-01-1408, 2014,
Methods." Computers & Fluids 50 (2011), pp. doi:10.4271/2014-01-1408.
10-23

The Engineering Meetings Board has approved this paper for publication. It has successfully completed SAE's peer review process under the supervision of the session organizer.
This process requires a minimum of three (3) reviews by industry experts.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written permission of SAE International.
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the content of the
paper.
ISSN 0148-7191