You are on page 1of 14

INTRODUCTION

In modern diesel engines, especially in high-speed direct injection engines, the performance,
efficiency noise, and pollutant emissions have a strong dependency on the characteristics of the
fuel injection process. Nowadays, the amount of fuel injected is not the only relevant
characteristic of the injection process. The instantaneous fuel mass flowrate introduced into the
combustion chamber, the evolution of the spray, and its interaction with the air are also important
[1–3]. One important phenomenon in this process is the flow behaviour across the injector nozzle,
which is recognized as being influential on the atomization of the fuel spray (droplet formation)
and the mixing of fuel with air [4–10].

Since there is a strong relation between pollutant emissions, fuel injection features, and spray
formation, several studies have been carried out with the aim of characterizing the instantaneous
fuel injection process. However, owing to the small geometrical characteristic of the nozzles, it is
rather difficult to obtain the parameters inside the nozzle such as the pressure profile on the
needle tip and the needle dynamic. Several factors control the flow across the nozzle, and for this
reason it is advisable to sort them into two categories: operating conditions and geometrical
characteristics. The first category is defined by the flow type (steady or pulsating), the flow
régimen (Reynoldsnumber), and, finally, whether the flow is cavitating or non-cavitating
(cavitation number). The second category includes the nozzle seat type (with sac volume or
without sac), the inlet orifice diameter,theorificeconicity(convergentconicalhole, divergent conical
hole, or cylindrical hole), inlet orifice rounding, and hole lenght.

Several investigations have been carried out to understand the different behaviour of the Flow in
the most used nozzle types i diesel engines (microSAS and valves covered orifice (VCO). The aim of
the present investigation is to identify the relationship betweenthe injectordynamic(needle
liftbehaviour) injector nozzle seat type, instantaneous fuel mass flowrate, and spray behaviour
help of experimental information. The internal Flow across the nozzle has been experimentally
analysed under continuous and pulsating conditions. To avoid influences on the internal Flow
owing to other geometrical actors (inlet orifice diameter, orifice conicity, inlet orifice rounding,
and hole length), two fivehole nozzles with the same configuration, except the seat type, have
been used. The spray momentum flux has been measured for each orifice in both nozzles to obtain
the effective dischange velocity, and the nnitrogen test rig has been used in order to characterize
the macroscopic characteristics of the spray.

The paper is divided into four sections. The first section gives a brief description of the
experimental set-up. Next, the results obtained in the different test rigs for microSAC and VCO
ozzles are shown and compared with each other. The third section offers an analysis of the results
base don the spray theory, and finally a summary and the main connclusions of the paper are
presented.

2. EXPERIMENTAL SET-UP

Four different test rigs have been used in this research in order to obtain the spray momentum
flux, the fuel mass flowrate injected under steady Flow, the instantaneous injection rate under
pulsating flow, and the spray macroscopic characteristics. The details of these systems will be
described in the following sections.
The injection system used in the experiments is a standard common rail system composed of a
high-pressure pump supplying the required injection pressure at the rail, a fuelling system, a
control system, and an injector holder fitted with a needle lift sensor. The whole system is
controlled by a Genotec impulse generator. The injector is the typical one from a common rail, the
three main components of which are a control valve, an intensifier plunger and barrel, and a
nozzle. The valve is an electronically controlled solenoid-poppet valve whose purpose is
tostartandstoptheinjectionprocess.Thecharacteristics of the two nozzles used in the present study
are shown in Table 1. The fuel used for this experiment was Repsol CEC RF-06-99 with a density of
820.2 kg/m3 and a kinematic viscosity of 2.67 mm2/s, both at 40 °C.

2.1 Steady flow test rig


The object of the steady flow test rig (SFTR) is to determine the mass flowrate in the orifices
of an injector nozzle under steady flow conditions. For this purpose, the injector nozzle to be
tested is installed without the needle in an injector holder, which has been modified to feed the
nozzle where the needle was placed. The needle is removed to avoid the pressure loss
produced by the needle seat. This means that the discharge coefficient of the holes can be
obtained.
The injection pressure is controlled using a standard pressure regulator and can be
fixed at different values, up to a maximum depending on the maximum mass flow that the pump
is able to supply. The injector is mounted in a 1 l vessel filled with fuel, where pressure is
controlled, fixing the injector orifice outlet pressure. A sketch of the installation is shown in Fig.
1.

The pressure difference across the orifice can be kept constant, and, since the geometry is also
fixed, steady flow conditions are achieved. In order to carry out the measurements at different
pressure differences, the injection pressure, Pi, which is the pressure at the orifice inlet, is kept
constant, and the discharge pressure in the vessel, Pb, is varied.

In order to change the flow conditions in the injector orifice, tests were conducted at values of
injection pressure,i.e. orifice inlet pressure,of 10 and 20 MPa, while the pressure at the outlet was
varied between minimum values of 0.01 MPa up to the injection pressure. After a short
stabilization time, when steady flow conditions have been achieved, the mass flowrate across the
injector is measured.

The injection pressure is measured upstream with an AVL piezoresistive sensor. The discharge
pressure is measured with a PMA GmbH sensor (reference P40). Both sensors were calibrated
before the measurements were taken with a deadweight pressure tester. The errors of the dis
charge coefficient and theoretical Reynolds number measurements were estimated using a
standard procedure, as outlined by Holman [19]. The level of uncertainty obtained for the
discharge coefficient and theoretical Reynolds number were around 1.5 and 0.3 per cent
respectively.
2.2 Injection rate test rig

The test rig where the measurements of injection rate were carried out is a standard injection rate
discharge curve indicator (IRDCI) described in reference [20], and it enables the data describing
the chronological sequence of an individual fuel injection event to be displayed and recorded.

To obtain a good estimation of the experimental errors, several repetitive measurements were
carried out at the same test point (energizing time, rail pressure, and backpressure), and the
obtained dis- persion was about 0.6 per cent, with proper calibration of the equipment.

In order to determine the momentary mean quantity, an accumulative measurement process


must therefore take place downstream of the IRDCI. The mean quantity corresponds to the mean
surface area of the pressure integral for a recorded sequence of injection events.
Three injection pressure values, P , fixed at the common rail, were used: 30, 80, and
i
130 MPa.The backpressure or pressure at the discharge of the injector, P , was 4 MPa. For all
b
the measurements taken, the injector energizing time was set at 0.25, 0.5, 1, and 3 ms. The first
three are used to find out the injection rate shape in a typical injection event in that engine.
The last one can produce a stable injection process and thus obtain more consistent results
for injection rate characterization and analysis.
The injection rate, fuel injection pressure, needle lift, and energizing time intensity signals were
recorded using a data acquisition and display YOKOGAWA system and in-house developed
TRATASA software to obtain the injection rate curves.

2.3 Spray momentum test rig


With this experimental equipment, it is possible to determine the impact force of the spray,
which is equivalent to the momentum flux of each nozzle orifice [21]. Figure 2 shows a sketch
of the momentum test rig. The chamber can be pressurized with nitro- gen up to 8 MPa to simulate
discharge conditions in the engine.

The force is measured with a piezoelectric pressure sensor calibrated to measure the force. The
sensor is placed at 5 mm from the hole exit. The frontal area of the sensor is high enough
to catch all the frontal area of the spray. Owing to the conser- vation of momentum, it can
be assumed that the force measured by the sensor will be the same as the momentum flux
at the orifice outlet, since the pressure inside the chamber is constant and sur rounds the
entire spray. Taking this into account, it can be said that the density of the chamber fluid
does not affect the measurements, so that any chamber pressure can be used without disturbing
the results. Figure 3 shows a sketch of the spray and sensor used in the momentum test rig.

2.4 Nitrogen test rig

In order to visualize the sprays, the nitrogen test rig is used. The test rig basically consists of a
steel cube with a chamber and various connecting flanges machined into it. The design is
modular, and ancillaries can be added depending on the required experiment [22]. The rig and
ancillaries are designed for a maximum pressure of 8 MPa. The test rig has an internal volume
of 1 l.

It is necessary to circulate the nitrogen through the rig because otherwise the injected diesel
would obscure the windows and severely degrade the quality of the images. Furthermore,
it is important to keep rig pressure P and nitrogen temperatura constant during each
b
experiment. Two filters collect the fuel injected to keep the gas stream clean. The temperature
of the nitrogen can be fixed between 15 and 50 °C in order to obtain the desired density inside the
chamber. The test rig operates in cold conditions at all times, thus avoiding fuel evaporation. A
sketch of the high-pressure nitrogen test rig is shown in Fig. 4.
The images are taken with a 12-bit colour CCD camera with a spatial resolution of 1280 × 1024
pixels and a minimum exposure time of 10 ms with a jitter of ±5 ms. Illumination is created
with a conven- tional electronic flash of controllable duration (set for far longer than the
minimum camera gate) and intensity. In spite of the fact that a conventional flash is used to
present a shot-to-shot fluctuation that is higher than more sophisticated light sources, the
results obtained with the software have proved to be insensitive to fluctuations in the
illumination intensity within a reasonable range. All the experi- mental equipment (camera–
flash–injection) has been synchronized with a purpose-built electronic system, using the injector
trigger signal as a reference signal to take the image sequences. In Fig. 5, the experi- mental
set-up for the image acquisition is presented.

The images are digitally processed using purpose- developed software [23]. The segmentation
algorithm used, based on the log-likelihood ratio test (LRT), has the advantage of using the three
channels of RGB images for a proper determination of boundaries that are not well defined, as
in the case of sprays. This method proved to be almost completely insensitive to intensity
fluctuations between pictures for the tested cases, and provided better results than some other
algorithms checked. Prior to the systematic use of the algorithm for parametric studies, the
influence of the illumination quality on the results was evaluated in specific tests. Results
demonstrated that the algorithm properly detects the estimated spray boundaries even in the
case of reasonably poor illumination. Details of the image processing soft- ware are available
in reference [23]. Figure 6 shows the spray macroscopic characteristics obtained using this
installation and the notation employed.

3. RESULTS
3.1 Steady flow test rig

The final results applying this methodology to the two nozzle seat types, microSAC and VCO,
are pre- sented in Fig. 7, where the injection mass flowrate versus the square root of effective
pressure loss across the nozzle is plotted for two values of rail pressure of 10 and 20 MPa. As
explained before, these measurements were taken in steady conditions of inlet and outlet
pressure.

As can be seen, there are small differences between the amounts of fuel injected, the main
difference being the curve slope. If the expression for fuel mass flowrate injected by an injector
nozzle, equation (1), is analysed, this difference means that both nozzles have variations in
permeability value
where m is the mass flowrate, C is the nozzle dis- charge coefficient, A is the nozzle area, r is
N dN N f
the fluid density, P is the injection pressure, and P is the backpressure in the combustion
i b
chamber. The term of the square root is the theoretical velocity defined by Bernoulli, Uth .
To produce a better analysis, it is necessary to plot the discharge coefficient, CdN, against the
theoretical Reynolds number, l, as depicted in Fig. 8. The diameter used in the calculations is the
outlet orifice diameter, which is shown in Table 1.
In agreement with Fig. 7, it is possible to say that the differences found are due to the small variation
in the diameter. The discharge coefficients in both nozzles are different, showing that the microSAC
value is higher than the VCO value. It is important to point out that no influence of the needle seat
is present in these experimental results. Therefore, the flow is only restricted by the geometry of
the nozzle orifices, and in this case the orifice inclination angle, a, in the microSAC nozzle is more
favourable (less restriction) to the flow than in the VCO nozzle, as can be seen in Fig. 9, which can
justify the higher value of the discharge coefficient in this nozzle.
It is important to note that the fact of not having mass flowrate collapse for the more critical
pressure conditions (higher pressure differences) implies that a cavitation phenomenon does not
exist in the VCO nozzle or the microSAC nozzle. This is because of the high level of conicity of the
nozzle orifices [9, 10, 21]. A rigorous study of the influence of cavitation on the internal flow and
spray characteristics was performed by Payri et al. [9].
3.2 Injection rate test rig
Figure 10 shows the injection rate curve at three injection pressures fixed at the common rail, a
back- pressure of 4 MPa, and with an energizing time of 1 ms. It is of fundamental importance to
note that the same injector holder has been used in both nozzles to carry out the measurements;
therefore, they will have the same maximum needle lift, and any differences due to the injector
holder are avoided. At all pressures it is possible to notice the unlikely behaviour of the injection
rate curves in both nozzles, as the microSAC nozzle has a lower maximum value than the VCO
nozzle. This difference, as in the results of the steady flow test rig, are in part justified by the
differences in diameter between the two nozzles.
It is essential to analyse the needle lift behaviour for the injection rate curves to achieve a better
understanding of the nozzle flow behaviour before making conjectures, since the needle does not
reach the maximum lift with an energizing time of 1 ms. This means that the amount of fuel
injected depends on the lift that the needle can reach. Figure 11 shows the needle behaviour in
the microSAC and VCO nozzles.

It is feasible to discover that, when the injection pressure is increased, the differences in the
needle lift curves between microSAC and VCO increase, showing that the microSAC needle lift is
lower than the VCO needle lift. The same result is obtained for other energizing times. These
differences can be explained by variations in the pressure profile on the needle between the two
nozzles, which means that the losses along the nozzle seat are higher in the microSAC nozzle than
in the VCO. Consequently, the force generated by the pressure profile on the needle tip, which
pulls the needle up, will be smaller in the microSAC than in the VCO nozzle, causing a slower
needle lift and thus a smaller injection rate [14].

As the injection rate and the needle lift behaviour are known, it is possible to obtain the curve of
CdN against time as Fig. 12 shows. Small differences can be found in the discharge coefficient
values between the two nozzles, with the microSAC having a slightly higher value than the VCO
nozzle [14, 15]. These differences are more visible in the stationary part of the curves. Although
for a small needle lift the losses in the seat of the microSAC nozzle are bigger, this is in part
compensated for by the less restriction of the flow owing to the smaller inclination of the orifices
in this nozzle (Fig. 9), especially for higher needle lift (80 and 130 MPa). Thus, the differences in
the discharge coefficient are more important at high pressure (needle lift) in favour of the
microSAC nozzle.
3.3 Spray momentum test rig

Figure 13 shows the mean spray momentum flux (impact force) curves at three injection pressures
fixed at the common rail, a backpressure of 4 MPa, and with an energizing time of 1 ms.

For the three pressures tested, as for the injection rate test rig, the instantaneous value of
momentum flux is smaller in the microSAC than in the VCO. Since a turbulent flow without
cavitation is considered, the outlet velocity profile can be assumed to be uniform all along the
outlet area of the hole. With this assumption, it is possible to estimate the e ffective outlet
velocity, Uo [21], by dividing the momentum flux Mo, by the mass Flow mo.
Figure 14 shows the instantaneous outlet velocity, U , during the needle opening stage. It
o
can be observed that, in the case of 30 MPa, the value of the outlet velocity in the microSAC
nozzle is about 7 per cent smaller than in the VCO nozzle (considering the stationary part),
but nevertheless this tendency radically changes when the injection pressure increases, with an
effective velocity in the microSAC nozzle about 5 per cent higher than in the VCO nozzle
for an injection pressure of 130 MPa. The reason for this is related to the behaviour of the
discharge coefficient previously stated.

3.4 Nitrogen test rig


Figure 15 shows some samples of images obtained in the nitrogen test rig. The picture shows
the sprays from two nozzles at an injection pressure of 130 MPa, and for a time after injection
of 1000 ms. Figures 16 and 17 depict the values of the spray penetration and spray cone angle.
The results shown correspond to the mean value of five independent sprays. The standard
deviation (not represented in order to clarify the figures) is small: the penetration has a
maximum standard deviation of around ±0.7 mm, and the cone angle has a maximum
standard deviation of ±1°. Therefore, the error determining these macroscopic parameters
is very small. The measurements were obtained in the nitrogen test rig at three injection
pressures fixed at the common rail, 4 MPa of backpressure, and with 1 ms of energizing time. It
is only possible to visualize the spray up to a maximum value of 42 mm because of optical
Access window limitations.

From Figs 16 and 17 some conclusions can be drawn: the spray tip penetration for the
microSAC nozzle is higher than in the VCO nozzle, but the value of the spray angle, as with the spray
volume, is higher in the VCO nozzle than in the microSAC. In both cases, when the injection
pressure increases, the differences between these values becomes higher. The rise in the spray
angle is due to the fact that the spray momentum is more distributed in the radial direction
than in the axial direction, which leads to more important axial velocity decay. This smaller axial
velocity results in a decreased spray penetration.

On the other hand, a higher spray angle is related to an increase in air entrainment. Figure 18
shows the spray volume for each nozzle. As with the angle, the volume of the spray is a
representative parameter of the air entrainment process. This figure shows that the spray
volume of the VCO nozzle is bigger than that of the microSAC nozzle. Combining this
measurement with the fuel injection rate, it is possible to estimate the fuel/air ratio, W
and thus determine which nozzle behaves better in ter ms of air entrainment . In Fig.
19 it can be observed that the value of W in the VCO nozzlw, for given time, is
smaller tan in the microSAC nozzle.
4. ANALYSIS OF THE RESULTS

The studies dedicated, at least principally, to the experimental or theoretical examination of free
spray penetration are incredibly numerous [23–29]. Many conclude with the proposal of an
empirical or semiempirical equation that tries to estimate the instant
spraypenetrationrelatedtodiverseparameters(outlet velocity, environment density, orifice
geometry, etc.). Hay and Jones [24] carried out a critical revision of the proposed correlations in the
bibliography for diesel spray in the time period before 1972 and arrived at the conclusion that those
that best fit the experimental results were the prroposals by Wakuri et al. [25] and Dent [26]. In this
way, as demonstrated by Wakuri et al., by applying the conservation of the quantity of movement
in a constant cone angle th over a time period, where it is assumed that the air/fuel mixture is
homogeneous, an equation is obtained that according to Hay and Jones fits perfectly with the
experimental data. Therefore, as a reference in this analysis the expression proposed by them is
used [25]

Where S(t) is the penetration as a function of time, r is the air density, Do is the diameter at the
orifice outlet, DP is the pressure differential, t is the time from the start of injection, and theta is the
angle of the of the spray cone. The geometry of the orifices is taken into account through the
diameter of the orifice. As both nozzles hace the same orifice configuration, it is expected that this
constant will be the same for the both nozzles.
The penetration behaviour equation, wich is established by the theory of the gaseous sprays
[equation (4)], can be written as

where Mo, is the momentum flux at the orifice outlet, equal to the mass flowrate per orifice
multiplied by the velocity at the orifice outlet

Taking into account equation (¡), the mass flowrate can be written as

Where Ao is the outlet área of one orifice in the nozzle. In this equation, because all orifices hace
the same geometrical characteristics, it is assumed that the discharge coefficient of each orifice is
the same as the discharge coefficent of the ozzle, CdN. The los included in the discharge coefficient
could bbe divided into two parts [equation (8)]: the velocity coefficient, Cv, which takes into
account the los in terms of velocity, and the área coefficient, Ca, which incorporates the los of área
owing to Flow contraction and the non-uniform velocitu profiles at the outlet section of the hole.
[21,30]

The velocity coefficient, Cv, is defined as the relation ship between effective velocity and theoretical
velocity

In order to quantify the differences observed, and concerning the fuel-air mixing as well as the fuel-
air ratio, one representative parameter is the mixing time, tm()m which is defined as the time
passed from the beginning of the injection until a determined fuel/air ratio has been obtained in
the axis of the spray. An expression for the mixing time as a function of the spray and Flow
parameters can be deduced.

The equation of momentum conservation along the spray axis can be written as

With ma is the mass of the air entrained by the spray, and Uaxis is the velocity of the spray along
the spray axis, which is the derivation of the penetration as a function of time. Taking into account
aquations (5) to (9), the penetration of the spray can be written as

Considering stabilized conditions, tan(th/2) is constant, and then, deriving equation (11), the
expression for the velocity of the spray is obtained

Taking into account

and substituting equations (12) and (13) into equation (10), the following expression is obtained
Finding the value of t, and considering the fuel–air ratio, W, the following expression is
obtaineddivided

For a determined fuel/air ratio, the values obtained from equation (15), which are proportional to
theoutlet mixing time, tm(W), of the two nozzles at the three injection pressures tested, are
compared in Table 2.

For calculation of the parameters, the guidelines below were followed.

1. The effective velocity is obtained from Fig. 14, and, taking into account the theoretical velocity,
the coefficent Cv can be worked out.
2. The nozzle discharge coefficent can be obtained from Fig. 12, and, taking into account equation
(8), the value of Ca can be worked out. Because there is no cavitation, the values of Ca obtained
for both nozzles are very similar and near to ¡ [21].
3. The value od tan(th/2) can be obtained from Fig. 12 (steady conditions).

Substituting the values obtained in equation (15), values that are proportional to the mixing time
can be obtained. From this table it can be concluded that the VCOnozzle behaves betterfrom the
pointof view of the fuel–air mixing process on account of the reduction in the mixing time. This
reduction is around 18 per cent for low injection pressures and it is reduced to 14 per cent for the
highest pressure tested.The reduction in the mixing time with pressure is due to the fact that the
value of the velocity coefficient, Cv, becomes higher in the microSAC nozzle than in the VCO nozzle
for high injection pressures. Nevertheless, in spite of this increment in terms of velocity, the di
fferences found in terms of the spray cone angle result in better fuel–air mixing (a shorter mixing
time) for the VCO nozzle.

5. Resumen y conclusiones
In the present study, an experimental investigation wascarriedoutin
ordertocompareinjectioncharacteristics and the spray behaviour and to understand the differences
between two real five-hole vertical VCO and microSAC nozzles.

From this study it is possible to draw the following conclusions

1. Two di fferent installations were used to characterize the flow across the nozzles under
continuous and pulsating Flow. The discharge coefficient of the nozzles was depicted as a
function of the theoretical Reynolds number, l, under steady flow and without needle
influence, showing that the value of the microSAC CdN is greater than that of the VCO.
2. The combination of momentum flux measurements and those obtained from mass flowrate
characterization has permitted the determination of the spray outlet velocity, and the
observed tendency is that, for the lower injection pressure of 30 MPa, the microSAC nozzle has
a velocity around 7 per cent smaller than in the VCO nozzle. Nevertheless,this tendency
radically changes when the injection pressure increases, with an effective velocity in the
microSAC nozzle around 5 per cent higher than in the VCO nozzle for an injection pressure of
130 MPa. This change in tendency is due to the fact that, although for a small needle lift the
losses in the seat of the microSAC nozzle are greater, this is in part compensated for by the
smaller restriction of the inlet flow owing to the lower inclination of the orifices in this nozzle,
especially for higher needle lift.
3. As far as the influence on the macroscopic behaviour of the spray is concerned, it can be
concluded that the microSAC nozzle has a higherbe spray penetration than the VCO nozzle,
whereasthe the spray angle was higher in the VCO than in theof microSAC, which partially
explains the betterreduction fuel–air mixing process in the VCO nozzle.
4. In order to ascertain which flow and macroscopicis parameters control the mechanism of fuel–
airtested mixing, a theoretical analysis, with the limitationsis that it assumes, has been made.
In this study, an expression for the mixing time was obtained by combining the results of the
spray macroscopic study with the previously obtained nondimensional flow parameters. The
main conclusión of this analysis is that the VCO nozzle behaves better in the fuel mixing process.
The differences observed in terms of spray cone angle result in the VCO nozzle having lower
mixing times than the microSAC nozzle, despite having lower effective velocity values at high
injection pressure.

You might also like