Detailed Modeling of Common Rail Fuel Injection Process

X.L.1. SEYKENS
1
, L.M.T. SOMERS
1
, R.S.G. BAERT
1,2

1. Faculty oI Mechanical Engineering, Division Thermo Fluids Engineering (TFE), Eindhoven
University oI Technology, The Netherlands, P.O. Box 513, 5600 MB Eindhoven, E-mail:
X.L.J.Seykenstue.nl
2. TNO Automotive, P.O. Box 155, 2600 AD DelIt, The Netherlands

ABSTRACT

Modeling oI Iuel injection equipment is a tool that is used increasingly Ior explaining or predicting
the eIIect oI advanced diesel injection strategies on combustion and emissions. This paper reports on the
modeling oI the high-pressure part oI a research type Heavy Duty Common Rail (CR) Iuel injection
system. More speciIically, it reports on the observed dynamics oI the injection system and the capability
oI the model to capture this. For that reason, the total high-pressure part oI the injection system, i.e. the
Iuel pump, rail and injector, has been modeled using the AMESim code (Imagine S.A., 2004). The
reliability oI the resulting hydraulic model is tested through a comparison between numerical results and
actual injection measurements. This detailed comparison is based on measurements oI injected mass Ilow
rate, needle liIt and pressure oscillations in the injection duct Ior a series oI single injection events.
It is shown that the hydraulic model is able to accurately simulate the injection rate, needle liIt and
injection pressure Ior diIIerent rail pressure levels. For accurate numerical results, it is vital that the
stiIIness oI the injector needle assembly and the discharge coeIIicients oI the diIIerent Ilow restrictions in
the injector (e.g. nozzle holes) are correctly modeled. Assuming a rigid injector needle results in a too
early start oI injection. Discharge coeIIicient values Iound in literature shows a wide spread. This makes
it very diIIicult to simulate the injected mass Ilow rate accurately on the basis oI literature data. Using the
measured injected mass oI Iuel to tune the discharge coeIIicient, together with the inclusion oI the
elasticity oI the injector needle, results in a good approximation oI the injection rate.

KEYWORDS

Common Rail Iuel injection, Modeling, Discharge coeIIicient

1. INTRODUCTION

With modern diesel engines the injection process (i.e. the injection rate and injection pressure) has a
major impact on noise production, exhaust gas emissions, Iuel consumption and engine eIIiciency. In
view oI the ever-increasing demands on these engines, modeling oI the Iuel injection system has become
an essential step in the Iuel injection equipment design and optimization process. The large number oI
technical papers regarding the modeling oI the dynamic behavior oI the injection system (e.g. Borghi et
al. |1|, Catania et al. |2| and Cantore et al. |3|) conIirms this.
Detailed modeling oI the diesel spray development and combustion process inside the diesel engine
(e.g. phenomenological or 3D CFD modeling) requires the injection rate and velocity (i.e. momentum) to
be known as an input (e.g. Wickman et al. |4| Hiroyasu et al. |5|, Barba et al. |6|). ThereIore, correct
representation oI the injection rate (and velocity) Ior a given operating condition (rail pressure and
injection duration) is oI great importance. For this reason, it was decided to develop a one-dimensional
hydraulic model oI a Common Rail Iuel injection system that is capable oI correctly simulating the
injection rate as a Iunction oI rail pressure.
In this article the diIIerent steps towards realizing such model are presented. First, the CR test rig
that was developed will be described together with a general description oI the corresponding hydraulic
model. ThereaIter the model validation process is discussed. A comparison between measurements and
numerical results is presented. The validation oI the used injector nozzle hole discharge coeIIicient and
the inIluence oI cavitation on the actual value oI this discharge coeIIicient are also discussed in great
detail. Finally an overview oI the main conclusions is given.

2. COMMON RAIL TEST FACILITY

The Iuel injection system being analyzed is oI the common rail type. It has been designed Ior Iuel
delivery to a single cylinder 2.1-litres HD diesel research engine. In Iigure 1, this injection system is
shown schematically. The system uses a 2
nd
generation light-duty CR high-pressure Iuel pump with an
electronically controlled throttling valve to adjust the delivered mass Ilow rate. A Irequency-controlled
electrical engine drives the CR pump. The injector is connected close to the centre oI the rail. The injector
is an 8-hole HD diesel engine CR sac-hole nozzle injector. The nozzle holes have a nominal diameter oI
0.184 mm, a length/diameter ratio oI approx. 5 and an inlet rounding oI approx. 7. A heat exchanger is
included to control the temperature in the Iuel injection system. Maximum injection pressure is 1400 bar.
Single shot injections are perIormed into a closed chamber, the so-called Zeuch-chamber. This chamber is
Iilled with Iuel at a certain pressure (20 40 bar). From the measured pressure rise in the chamber during
injection, the injection rate can be calculated. This method Ior injection rate determination is an
alternative to the BOSCH method. The size oI the chamber is chosen such that it Iits HD like injection
events: the internal volume is equal to 1.8710
-4
m
3
.
A Micro-Epsilon eddy-current needle liIt sensor type U05 measures the injector`s needle liIt. The
pressure in the Zeuch-chamber is measured by a Kistler 7061 quartz pressure transducer and the pressure
in the injection duct between rail and injector is measured by a Kistler piezoresistive pressure sensor type
4076.
















Figure 1: Common Rail Iuel injection system.

3. COMMON RAIL IN1ECTION SYSTEM MODEL DESCRIPTION

Figure 2a shows a typical HD CR Iuel injector with electromagnetic Iuel injection control. In closed
position, the rail pressure is present in both the control chamber above the control plunger and the nozzle
chamber. Because the area oI the top oI the control plunger is larger than the area oI the needle-shoulder
in the nozzle chamber, a net closing Iorce is present. The needle tip is pushed on its seat and no injection
oI Iuel can take place. When the solenoid in the top oI the injector is energized, the resulting magnetic
Iorce liIts the ball valve Irom its seat. Because the Ilow rate through the Z-throttle is smaller than the Ilow
rate through the A-throttle (see Iigure 2a), the pressure in the control chamber drops. The rail pressure is
still present in the nozzle chamber and the needle is pushed upwards, starting the injection oI Iuel. As the
current through the solenoid is stopped, the solenoid spring Iorces the ball valve back on its seat. As a

result, the pressure in the control chamber increases again and the needle is pushed down on its seat, thus
stopping the injection.
The CR Iuel injector together with the total high-pressure part oI the CR injection system shown in
Iigure 1 (high-pressure pump, rail, ducts and Zeuch-chamber) is modeled using the AMESim code |7|
(Imagine S.A., 2004). In Iigure 2b a graphical presentation oI the CR injector is shown. In the AMESim
code each physical component oI the system is represented by an appropriate icon, and is associated to
one or more lumped parameter models (called submodels).
Principally, in AMESim, hydraulic systems can be modeled by isothermal and/or adiabatic
submodels that are either one-dimensional (Iuel lines) or zero-dimensional (restrictions, volumes).
Because time scales in the injection process are very short, adiabatic submodels have been used to model
the injection system.






















Figure 2a (left): Cross-section oI CR injector.
Figure 2b (right): AMESim CR injector model.

In the graphical representation oI the AMESim model in Iigure 2b, the physical elements oI the
injector can also be distinguished. The model can be seen as a chain oI restrictive and capacitive
elements. In the capacitive elements, such as the control chamber and nozzle chamber, Iluid pressure and
temperature are calculated. For the pressure variable, the continuity equation is Iormulated in terms oI
pressure as Iollows:

=
dt
dT
T p
J
m m
T p B
dt
dp
out in
) , (
1
) , (

& &


(1)

where
in
m& and
out
m& are respectively the ingoing and outgoing mass Ilow rates oI the volume J . Fluid
properties in equation (1) are bulk modulus B , density , and cubic expansion coeIIicient .
The temperature oI the Iluid in the volume is computed through solving the energy equation:

dt
dp
c
T
c J
Q h dt dm h m h m
dt
dT
p p
out in

+

+
=

&
& & ) ( ) ( ) (


(2)


with h the enthalpy,
p
c the speciIic heat at constant pressure and Q
&
the heat Ilow Irom the surroundings
into the volume.
In resistive elements, such as a throttle, only the mass Ilow rate is calculated, using the modiIied
Bernoulli equation:


where,
d
C , is the discharge coeIIicient. For non-cavitating Ilow,
d
C is dependent on Ilow velocity, Iluid
density and viscosity only. This dependency is taken into account by Iormulating the discharge coeIIicient
as a Iunction oI a dimensionless Ilow number, :

p
d
h

=
2


(4)

where
h
d is the hydraulic diameter and is the kinematic viscosity. The cross-sectional area, A, can be
constant or a Iunction oI liIt, as in the needle-tip component.
To correct Ior the occurrence oI cavitation, the so-called cavitation number CaN is used. In this
study, as in AMESim, this dimensionless number is deIined as the ratio oI the pressure diIIerence over the
restriction to the downstream pressure:

down
down up
p
p p
CaN

=

(5)

In case oI the injector nozzle holes,
up
p would be equal to the pressure in the sac-volume (see Iigure 2b)
and
down
p would be equal the pressure in the volume component that models the Zeuch-chamber volume
(not shown in Iigure 2b).
II the cavitation number is higher than a pre-deIined critical cavitation number,
critical
CaN , equation
(3) in AMESim changes to (see |7|,|8|):

CaN
CaN
p
A C m
c
d
+

=
1
2

&
critical
CaN CaN

(6)

where
c
d
C is the discharge coeIIicient oI the restrictive element Ior Iully cavitating Ilow, i.e. at the limit
oI cavitation (CaN ). In the injector model, cavitation is modeled in the nozzle holes and the A- and
Z-throttle.
Finally, all capacitive and restrictive components can be connected by long Iuel lines. In these
hydraulic lines pressure wave dynamics is taken into account. Assuming one-dimensional Ilow, the
equation oI motion oI a hydraulic line is:

0 ) ( ) sin( = + + + q h g A
dx
dq
u
dx
dp A
dt
dq

(7)

p
A C m
d

=
2
) ( &

(3)
with q the volumetric Ilow rate, the angle the line makes with the horizontal and ) (q h the viscous
Iriction term which is dependent on the relative roughness oI the pipe wall. The continuity equation is:

0
2
= + +
dx
dq
A
c
dx
dp
u
dt
dp


(8)

where c is the speed oI sound and A the cross sectional area oI the pipe.
Fluid pressure is transIormed` into a Iorce in so-called transIormation elements`, such as the piston
components in Iigure 2b. In the mass components the dynamics oI motion are evaluated. In Iigure 2b a
linear mechanical spring-damper element can be seen. This element models the elasticity oI the control
plunger / needle assembly in the injector. Other components in the injection system, such as the high-
pressure pump and the Zeuch-chamber, are modeled in a similar manner |7|.

4 MODEL VALIDATION

In the Iollowing paragraphs the model validation will be discussed. First the necessary data required
Ior the validation will be described Iollowed by a comparison between measurements and numerical
results. Additional measurements that were perIormed Ior determining the correct values oI important
model parameters are described in detail.

4.1 FUEL PROPERTIES AND EXPERIMENTAL DATA

In a Iuel injection system, pressures, and to a lesser extent temperatures, can vary considerably. This
results in a wide variety oI important Iuel properties such as density, bulk modulus oI elasticity and
kinematic viscosity. Having good Iluid property models over a large range oI pressures and temperatures
is thereIore essential Ior accurate modeling. Figure 3 gives an overview oI the important Iluid properties
oI the Iuel as used in the AMESim environment. The Iluid properties shown in Iigure 3 show good
correspondence with correlations presented by Rodrigez-Anton et al. |9| (max. deviation 0.3). For
ambient conditions, measured values Ior the density oI the actual test Iuel also match well (deviation
0.7) with the calculated values.




Figure 3: Density (leIt), bulk modulus oI elasticity (middle) and kinematic viscosity (right) oI used diesel Iuel as
Iunction oI pressure and temperature. Taken Irom AMESim |7|.

Besides the Iluid properties, also inIormation about the geometry oI the modeled components is
necessary. As can be seen in Iigure 2b, in AMESim the diIIerent components in the Iuel injection system
are built up using several elementary building blocks, such as restrictions, volumes, etc. Most oI these
0 50 100 150
810
820
830
840
850
860
870
880
890
900
Pressure [MPa]
D
e
n
s
i
t
y

[
k
g
/
m
3
]
0 50 100 150
1200
1400
1600
1800
2000
2200
2400
2600
2800
3000
Pressure [MPa]
B
u
l
k
m
o
d
u
l
u
s

[
M
P
a
]
0 50 100 150
2
4
6
8
10
12
14
16
18
20
22
Pressure [MPa]
K
i
n
e
n
a
m
t
i
c

V
i
s
c
o
s
i
t
y

[
c
S
t
]
20
o
C
30
o
C
40
o
C
20
o
C
30
o
C
40
o
C
20
o
C
30
o
C
40
o
C

building blocks require geometrical data. For example, modeling oI an injector nozzle hole requires the
knowledge oI the diameter oI the equivalent restriction. The necessary geometrical data has been obtained
Irom work drawings or by disassembly oI the component and perIorming actual measurements. For small
critical components such as the A- and Z-throttle and the injector nozzle holes, an electronic microscope
has been used.
An important element Ior injection timing is the electromagnetically liIted ball valve at the top oI the
injector. The maximum liIt and liIt velocity oI this ball valve is diIIicult to determine. ThereIore this
process was not modeled but additional literature data on a similar valve has been used as a reIerence
(|10|, |11|).
Finally, experimental data oI the injection system is required to validate and tune the model.
Injection measurements have been perIormed into the Zeuch-chamber. During these measurements three
signals were logged:
1) Control plunger movement. an eddy-current needle liIt sensor measures the movement oI the top oI
the control plunger. This signal is used to determine the injection timing and needle tip liIt as a
Iunction oI time.
2) Infection pressure: the pressure in the injection duct between rail and injector is measured at a
position oI 120 mm Irom the entrance oI the injector. Oscillations in this pressure will inIluence the
injection rate.
3) Pressure in the Zeuch-chamber: Irom the pressure rise in the Zeuch-chamber during the injection, the
injection rate can be determined by using the Iollowing equation |12|:

dt
dp
B
J
dp
dJ
dt
dm
infected

+ =

(9)

with J the volume and dp dJ the stiIIness oI the Zeuch-chamber. Injections are perIormed into a
closed chamber Iilled with Iuel at a certain pressure. The stiIIness oI the Zeuch-chamber, i.e. the
dp dJ Iactor, has been determined Irom Finite Element Method (FEM) calculations |13| to be
suIIiciently high to have a negligible inIluence. This method represents an indirect measurement oI
the injection rate. From equation (9) it is clear that the accuracy oI calculated mass Ilow rates is
mainly determined by the uncertainty in density and modulus oI elasticity, which are Iunctions oI
both pressure and temperature. During the determination oI the injection rate, the density and Iluid
modulus oI elasticity are taken as a Iunction oI the measured Zeuch-chamber pressure. Through this,
the increase in density and elasticity oI the Iluid inside the chamber, as a result oI the injected mass
into the closed volume, is taken into account. The temperature in the chamber is assumed to remain
constant during an injection. Analysis has shown that the inaccuracy Ior determined mass Ilow rates
is < 2.3. The inaccuracy is mainly caused by the uncertainty in temperature oI the Iluid inside the
Zeuch-chamber. This temperature is not directly measured.

In the Iollowing paragraphs the numerical results Ior the control plunger movement, injection
pressure and injection rate will be compared with measured data. During the injection measurements the
hold current oI the input signal to the injector is held constant Ior 3.0 ms. This results in injection
durations oI about 3.8 ms.

4.2 CONTROL PLUNGER MOVEMENT

Figure 4 shows the measured and simulated displacement oI the top oI the control plunger Ior rail
pressures oI 1400 bar and 800 bar. In this Iigure it can be clearly seen that the start oI Iuel injection into
the Zeuch-chamber (this is determined Irom the measured pressure rise in the chamber) does not
correspond to the start oI the control plunger movement. This is caused by the elastic deIormation oI the
control plunger and injector needle by the rail pressure prior to injection. Because oI this elasticity, the
measured movement oI the top oI the control plunger does not correspond to the actual movement oI the
needle tip.
As can be seen in Iigure 2b, the elasticity oI the control plunger/needle assembly is modeled using
one equivalently linear spring-damper element. The stiIIness oI this linear spring has been determined
Irom experimental data. In a series oI experiments, the displacement oI the top oI the control plunger was
measured with rail pressure increased Irom 0 to 1400 bar. No injections were perIormed during these
experiments. In Iigure 5 the elastic deIormation is shown as a Iunction oI rail pressure. It can be
concluded that this deIormation is indeed a linear Iunction oI the rail pressure. Note that at a rail pressure
oI 1400 bar the initial deIormation oI the control plunger and needle are about 36(!) oI the total
measured displacement.
In Iigure 5 also the measured displacement oI the control plunger until the measured start oI injection
(SOI) is depicted. Obviously, as a result oI the control plunger/needle elasticity, the actual start oI
injection is dependent on the rail pressure. However, because oI the rapid displacement oI the control
plunger the actual point in time oI start oI injection only varies slightly with rail pressure (order oI 0.01
ms).
During injection, the rail pressure also deIorms the control plunger/needle assembly Irom the bottom,
shortening the control plunger/needle. This results in a longer measured downward movement than the
measured upward control plunger movement Irom injection start to maximum needle liIt. This can also be
seen in Iigure 4 where the level oI injection end is lower than the deIormation level at injection start.
AIter the injection oI Iuel into the Zeuch-chamber has ended, oscillations in measured control
plunger movement can still be seen. They are the result oI oscillations in rail pressure, causing oscillating
elastic deIormation oI the control plunger/needle assembly. However, no post injections take place. From
Iigure 4 it can be concluded that the model is well capable oI capturing all the occurring phenomena.
Figure 4 (left): Measured and simulated control plunger movement.
Figure 5 (right): Elastic deIormation prior to injection as a Iunction oI rail pressure.

4.3 IN1ECTION PRESSURE

In Iigure 6 the measured and simulated injection pressures are shown Ior 4 diIIerent initial rail
pressures. In the injection pressure curves the eIIect oI the elasticity oI the control plunger/needle
assembly can be seen in the time delay between the start oI liIt oI the ball valve (SLB) and the actual start
oI injection (SOI). Prior to the pressure drop at the start oI Iuel injection, a small pressure drop can
already be observed. This pressure drop originates Irom the pressure drop in the control volume above the
control plunger when the ball valve liIts.
During the injection, pressure oscillations are present. Both the Irequency and the amplitude oI these
pressure waves are simulated correctly. The main Irequency oI the pressure waves, ~850 Hz, is equal to
the Irequency oI a standing wave in a one-sided open duct where the rail acts as the open end and the
injector nozzle holes as the closed end. Accurate simulation oI the Irequency oI the pressure waves is
0 1 2 3 4 5 6 7
-0.2
-0.1
0
0.1
0.2
0.3
Time [ms]
D
i
s
p
l
a
c
e
m
e
n
t

[
m
m
]
Meas. 1400 bar
Meas. 800 bar
Sim. 1400 bar
Sim. 800 bar
EIastic initiaI
deformation
Start of
Injection
End of
injection
EIastic deformation
during injection
700 800 900 1000 1100 1200 1300 1400 1500
0.07
0.08
0.09
0.1
0.11
0.12
0.13
0.14
0.15
0.16
Pressure [bar]
D
i
s
p
l
a
c
e
m
e
n
t

[
m
m
]
Measured
AMESim
1st order fit
Displacement for SO

mainly dependent on density and bulk modulus oI the Iluid (which determine the wave traveling
velocity).
The rapid closing oI the injector results in high-pressure oscillations known as the water hammer
eIIect`, which can be clearly seen in both the simulated and measured pressures. For the damping oI the
pressure waves caused by the closing oI the injector, not only the rail volume but also the Ilow restriction
Iormed by the edge Iilter (see Iigure 2a) was Iound to be oI importance. The loss coeIIicient oI this edge
Iilter was tuned to the available injection pressure data. Both amplitude and phase oI the pressure
oscillation are well captured. Overall it can be concluded that a good agreement between measured and
simulated injection pressures is present. Not only as a Iunction oI time but also as a Iunction oI initial rail
pressure level.

















Figure 6: Measured and simulated pressure in the injection duct Ior diIIerent rail pressures. The dashed vertical
lines indicate the Start LiIt Ball valve (SLB), Start OI Injection (SOI) and End OI Injection (EOI).

4.4 IN1ECTION RATE

Finally, Iigure 7 shows the simulated and measured mass Ilow rates corresponding to the Iour
injection pressure signals oI Iigure 6. The correct simulation oI the injector needle liIt can also be seen in
the mass Ilow rate signals. During needle liIt, good correspondence exists between simulations and
measurements. Figure 7 also clearly shows the inIluence oI the oscillations in injection pressure on the
injection rate: the main Irequency oI the Iluctuations in the mass Ilow rate corresponds to the observed
Irequency oI the oscillations in the injection pressure oI 850 Hz. However, higher Irequency oscillations
are more pronounced in the injection rate. Because it was not possible to correlate these higher
Irequencies to standing wave Irequencies in the injector, it is assumed that these Irequencies arise inside
the Zeuch-chamber as a result oI the injection event. The rapid opening and closing oI the injector Ior
instance induces pressure Iluctuations in the Zeuch-chamber. Also the implosion oI cavitation bubbles
inside the Zeuch-chamber is a possible explanation Ior the higher Irequency Iluctuations.

To obtain the simulated injected mass Ilow rates Iorm Iigure 7, the discharge coeIIicient oI the nozzle
holes at maximum cavitation
c
d
C ( CaN , see equation 6) was tuned to Iit the measured mass Ilow
rates. This Iit is based on the average mass Ilow rate during Iull needle liIt. The value oI the nozzle hole
discharge coeIIicient is essential Ior accurate modeling oI the injected mass Ilow rate during maximum
needle liIt. A value oI 0.765 is used Ior all simulations in Iigure 7. This tuned value is compared with
actual measurements oI the nozzle hole discharge coeIIicient. Preliminary experimental results conIirm
the value chosen Ior
c
d
C .
0 1 2 3 4 5 6 7
600
700
800
900
1000
1100
1200
1300
1400
1500
Time [ms]
P
r
e
s
s
u
r
e

[
b
a
r
]
Measured
Simulated
SLB
SOI EOI
"Water hammer"
Injection duration
















Figure 7: Measured and simulated injected mass Ilow rate Ior diIIerent rail pressures.


5 DISCUSSION

For accurate modeling oI the injection rate as a Iunction oI initial rail pressure and time, the elasticity
oI the control plunger/needle assembly and the value oI the nozzle hole discharge coeIIicients are very
important. In this paper, additional measurements have been perIormed to determine correct values Ior
the required model parameters. However, such experimental data is not always available. In this section it
will be analyzed what errors are made when this detailed inIormation is not present.
Usually, the measured liIt oI the top oI the control plunger ('needle liIt) and the injected Iuel mass
per injection are available Irom engine tests. Also, because the nozzle hole discharge coeIIicient is a
much-studied subject in literature, many papers can be Iound that can be used to estimate the value oI the
nozzle hole discharge coeIIicient. UnIortunately, values Iound in literature Ior the discharge coeIIicient oI
an injector nozzle hole show a wide spread. For the length/diameter ratio oI the used nozzle hole the
correlation oI Lichtarowicz |14|, which is valid Ior non-cavitating Ilow through a sharp-edged nozzle
hole, gives a discharge coeIIicient oI 0.78. Nurick |15| presents values Ior the discharge coeIIicients oI
sharp-edged nozzles and Iully cavitating Ilow in the order oI 0.62. Discharge coeIIicients in the range oI
0.69 0.73 are Iound Ior a Iully cavitating Ilow by several authors such as Arcoumanis et al. |16|,
Favennec et al. |8| and von Kuensberg Sarre et al. |17|. There are also authors who use signiIicantly
higher values. Ganippa et al. |18|, Ior instance, mentions discharge coeIIicients in the order oI 0.78-0.8
Ior cavitation numbers in the range oI 10 20. Even higher values are used by Goney et al. |19|. They use
values in the range oI 0.8 0.925 Ior a cavitating Ilow in a sharp-edged nozzle hole.
Six cases are evaluated during the analysis, as is summarized in table 1. Cases I to III represent the
situation when the elasticity oI the control plunger/needle assembly is not taken into account. Here, the
measured movement oI the top oI the control plunger is used to prescribe the motion oI the needle-tip. For
Cases IV to VI, the elasticity oI the control plunger/needle assembly is modeled using the measured value
Ior the spring stiIIness. Also a division in the nozzle hole discharge coeIIicient is made. Simulations are
perIormed using a value oI 0.7, which is obtained Irom literature. A discharge coeIIicient oI 0.75, which
is in line with the preliminary measurements, is used in cases II and V. Finally, the discharge coeIIicients
that give the best Iit between measured and simulated injection quantity are used Ior cases III and VI.






0 1 2 3 4 5 6 7
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Time [ms]

n
j
e
c
t
i
o
n

r
a
t
e

[
k
g
/
s
]
Measured
Simulated
800 bar
1000 bar
1200 bar
1400 bar

Needle/control plunger assembly
Rigid Elastic
Literature ( 7 . 0 =
c
d
C )
Case I Case IV
Measured ( 75 . 0 =
c
d
C )
Case II Case V


c
d
C
Fit on
inf
m
Case III Case VI
Table 1: Overview oI evaluated cases.

Figure 8 shows the diIIerence in simulated needle-tip movement Ior a rigid and elastic control
plunger/needle assembly (cases I and IV). Neglecting the elasticity oI the control plunger/needle
assembly causes a deviation oI about 0.2ms in the injection start. At an engine speed oI 2300 rpm, this
would represent an error oI 2.8 crank angles in the start oI injection. This is a signiIicant deviation since
the start oI injection is most Irequently given with a resolution oI 0.1 crank angles. Incorporation oI the
elasticity causes the control plunger to relax Iirst beIore it actually liIts. This also results in a lower
maximum liIt oI the needle-tip. Fluctuations in needle liIt during Iull opening are higher because the
oscillations in sac-volume pressure can deIorm the control plunger/needle assembly Irom below. The
higher maximum needle liIt does not have a signiIicant inIluence on the injected mass Ilow rate as can be
seen in Iigure 9. The higher needle liIt in the non-elastic case does not result in a signiIicant increase in
Ilow area. Also, the Ilow area oI the nozzle holes is rate limiting. As a result, the pressure at the entrance
oI the nozzle holes will be nearly identical and no real deviation in mass Ilow rate will be present.
In cases I to III, the oscillations in the measured control plunger movement aIter the start oI
injection, will directly result in a liIt oI the needle-tip. In Iigure 9 it can be seen that this causes post-
injections to taken place, which are not present in reality. In the Iollowing discussion, these post-
injections will be ignored and only the main injection pulse will be considered.
Figure 8 (left): Simulated needle-tip movement Ior rigid and elastic needle (cases I and IV).
Figure 9 (right): Comparison between measured and simulated mass Ilow rates Ior cases I, IV and V.

From Iigure 9 it can be concluded that the use oI the discharge coeIIicient value Irom literature
( 7 . 0 =
c
d
C ) results in a too low injection rate during maximum needle liIt. This causes the simulated
injected quantity too diIIer signiIicantly Irom the measured injected mass
inf
m oI
007 . 0
284 . 0

grams. It
has to be noted here, that the measured injected mass is derived Irom the injection measurements in the
Zeuch-chamber and not Irom actual Iuel consumption measurements on an engine.

0 2 4 6 8 10
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Time [ms]
N
e
e
d
l
e

l
i
f
t

[
m
m
]
Case
Case V

Inj

0 2 4 6 8 10
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
Time [ms]

n
j
e
c
t
i
o
n

r
a
t
e

[
k
g
/
s
]
Measured
Case
Case V
Case V








Table 2: Overview oI results oI evaluated cases.

Table 2 presents a summary oI the simulated injected mass Ior the diIIerent cases. Between brackets
the percentage deviation with the measured mass Ilow rate is given. For cases III and VI, the discharge
coeIIicients are given Ior which the simulated injection quantity is identical to the measured injected
mass. The earlier injection start oI case I compared to case IV explains the smaller deviation between the
measured and simulated injected mass. This also explains the over prediction oI the injected mass by case
II. Using the measured value Ior the discharge coeIIicient together with an elastic needle (case V), results
in a good agreement between measured and simulated injected mass. In Iigure 9 it can be seen that the
mass Ilow rate during maximum needle liIt is slightly under predicted Ior this case. ThereIore, the tuned
discharge coeIIicient Irom case VI is also slightly higher. It has to be noted that this discharge coeIIicient
is lower than the earlier mentioned tuned discharge coeIIicient oI 0.765 (see paragraph 4.4). The
diIIerence between these two tuned discharge coeIIicients is that the discharge coeIIicient Irom paragraph
4.4 is tuned on the Ilow rate during maximum needle liIt. The value Irom table 2 is tuned on the total
injection quantity. A discharge coeIIicient oI 0.765 results in an injected mass oI 0.286 grams, which
corresponds to an over-prediction oI 0.74. This is still well within the measurement uncertainty Ior the
injected mass. The Iact that the two discharge coeIIicients are diIIerent indicates that there are slight
deviations between simulated and measured injection rates during opening and closing oI the injector.
In general it can be concluded that inclusion oI the elasticity oI the control plunger/needle assembly
is essential Ior the simulation oI the injection timing. The importance oI taking the elasticity into account
is expected to increase with the increase in maximum common rail injection pressures in Iuture engines
(~ 200 MPa). The wide spread in discharge coeIIicient values Iound in literature makes it very diIIicult to
simulate the mass Ilow rate and injected quantity with the accuracy required Ior Iurther analysis, e.g. as
input Ior CFD computations. Tuning oI the discharge coeIIicient on actual measured injection quantities
is necessary. This already gives a good approximation oI the nozzle hole discharge coeIIicient and thus oI
the injection rate.

6 CONCLUSIONS

A hydraulic model oI a research type Heavy Duty Common Rail Iuel injection system has been
developed using the AMESim code. The model is capable oI accurately predicting the injection pressure,
needle liIt and injection rate Ior varying injection pressure.
For good modeling accuracy, the elastic deIormation oI the control plunger/ needle assembly has to
be taken into account. The initial elastic deIormation oI the control plunger, which shows a linear
dependency on rail pressure, results in a time delay between start oI control plunger liIt and actual needle-
tip liIt (i.e. start oI injection). For a rail pressure oI 1400 bar, the initial deIormation was Iound to be 36
oI the total movement oI the control plunger. However, because oI the rapid rise oI the control plunger,
the inIluence oI rail pressure on the time delay between start oI liIt and actual start oI injection is only
small (order oI 0.01 ms). With the increase in common rail injection pressures Ior Iuture engines, the
importance oI the inclusion oI the injector needle elasticity will also increase.
Equally important Ior accurate results is the value Ior the nozzle hole discharge coeIIicient. This
discharge coeIIicient is strongly inIluenced by cavitation, which decreases the value oI the discharge
coeIIicient. The tuned value Ior the nozzle hole discharge coeIIicient is relatively high, but is Iound to be
in good agreement with measured values.
The use oI the measured control plunger movement as a direct representation oI the injector needle-
tip motion, results in a signiIicant error in the start oI injection. The wide spread in values Ior nozzle hole
Needle/control plunger assembly
Rigid Elastic
Literature ( 7 . 0 =
c
d
C ) 277 . 0 =
inf
m g (-2.50) 265 . 0 =
inf
m g (-5.52)
Measured ( 75 . 0 =
c
d
C ) 294 . 0 =
inf
m g (3.52) 282 . 0 =
inf
m g (-0.95)


c
d
C
Fit on
inf
m 726 . 0 =
c
d
C 759 . 0 =
c
d
C
discharge coeIIicients Iound in literature makes them a inaccurate basis Ior injection rate simulations.
Using the injected mass oI Iuel to tune the discharge coeIIicient together with the inclusion oI the
elasticity oI the injector needle results in a good approximation oI the injection rate.

ACKNOWLEDGEMENTS

The authors grateIully acknowledge Imagine S.A. Ior the use oI AMESim.

REFERENCES

|1| Borghi M., Milani M., Piraccini M. (2001), Dynamic Analysis oI Diesel Engine Common Rail
Injection System. Part I: Inejctor Dynamics, In: Proceedings of 2001 ASME International
Mechanical Engineering Congress and Exposition, IMECE2001/FPST-25001
|2| Catania A.E., Dongiovanni C., Mittica A., Negri C., Spessa E. (1999), Simulation and Experimental
Analysis oI Diesel Fuel-Injection Systems With a Double-Stage Injector, In: Transactions of the
ASME - 1ournal of Engineering for Gas Turbines and Power, Vol. 121, pp. 186-196
|3| Contore G., Mattarelli E., Boretti A.A. (1999), Experimental and Theoretical Analysis oI a Diesel
Fuel Injection System, In: SAE International SP-1415, paper 1999-01-0199
|4| Wickman D.D., Senecal P.K. Reitz R.D. (2001), Diesel Engine Combustion Chamber Geometry
Optimization Using Genetic Algorithms and Multi-Dimensional Spray and Combustion Modeling,
In: Proc. of the SAE 2001 World Congress, paper 2001-01-0547
|5| Hiroyasu H., Kadota T., Arai M. (1983), Development and Use oI a Spray Combustion Modeling to
Predict Diesel Engine EIIiciency and Pollutant Emissions, In: Bulletin of the 1SME, Vol. 26, Nr.
214, pp. 569-575
|6| Barba C., Burkhardt C., Boulouchos K., Bargende M. (2000), A Phenomenological Combustion
Model Ior Heat Release Rate Prediction in High-Speed DI Diesel Engines with Common Rail
Injection, In: SAE International, paper 2000-01-2933
|7| Imagine S.A (2002). AMESim 4.0 User manual
|8| Favennec A., Lebrun M. (1999), Models Ior injector nozzles, In: The Sixth Scandinavian
Conference on fluid Power SICF`99 Tampere Finland
|9| Rodriguez-Anton L.M., Casanova-Kindelan J., Tardajos G. (2000), High Pressure Physical
Properties oI Fluids used in Diesel Injection Systems, In: Proc. of the SAE 2000 World Congress,
paper 2000-01-2046
|10| Bauer W. (1997), Berechnungsmodell zur Simulation von Common-Rail-Injektoren mit gesteuerter
Dsennadel Ir Dieselmotoren, PhD thesis, University oI Technology.
|11| Bianchi G.M., Pelloni P., Filicori F., Vannini G. (2000), Optimization oI Solenoid Valve Behavior
in Common-Rail Injection Systems, In: Proc. Of the SAE 2000 World Congress, paper 2000-01-
2042
|12| Zeuch W. (1961), Neue VerIahren zur Messung de Einspritzgesetzes und de Einspritz-
Regelmigkeit von Diesel-Einspritzpumpen, In: MTZ Motortechnische Zeitschrift, nr. 9
|13| Bruijstens J. (2000), Ontwerp van een meetopstelling ten behoeve van inspuitmetingen, Internal
Report WVM 20000.05, Combustion Technology, Eindhoven University oI Technology
|14| Lichtarowicz A., R.K. Duggings, E. Markland (1965), Discharge coeIIicients Ior incompressible
non-cavitating Ilow through long oriIices, In: 1ournal Mechanical Engineering Science, Vol. 7,
Nr. 3, pp. 210-219
|15| Nurick W.H. (1976), OriIice Cavitation and Its EIIect on Spray Mixing, In: 1ournal of Fluid
Engineering, December 1976, pp. 681-687
|16| Arcoumanis C., Flora H., Gavaises M., Badami M. (2000), Cavitation in Real-Size Multi-Hole
Diesel Injector Nozzles, In: Proc. Of the SAE 2000 World Congress, paper 2000-01-1249
|17| von Kuensberg Sarre C., Kong S., Reitz R.D. (1999), Modeling the EIIects oI Injector Nozzle
Geometry on Diesel Sprays, In: Proc. Of the SAE 1999 World Congress,paper 1999-01-0912
|18| Ganippa L. C., Andersson ., Chomiak J. (2000), Transient measurements oI discharge coeIIicients
oI diesel nozzles, In: Proc. Of the SAE 2000 World Congress, paper 2000-01-2788
|19| Goney K.H., Corradini M.L. (2000), Isolated eIIects oI ambient pressure, nozzle cavitation and hole
inlet geometry on diesel injection spray characteristics, In: Proc. Of the SAE 2000 World
Congress, paper 2000-01-2043