Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland

MODELLING OF COMMON RAIL FUEL INJECTION SYSTEM AND

INFLUENCE OF FLUID PROPERTIES ON INJECTION PROCESS
X.L.J. Seykens1, L.M.T. Somers2 and R.S.G. Baert3
1. Faculty of Mechanical Engineering, Division Thermo Fluids Engineering (TFE), Eindhoven
University of Technology, The Netherlands, email: [email protected].
2. Faculty of Mechanical Engineering, Division Thermo Fluids Engineering (TFE), Eindhoven
University of Technology, The Netherlands.
3. Faculty of Mechanical Engineering, Division Thermo Fluids Engineering (TFE), Eindhoven
University of Technology, The Netherlands.
ABSTRACT
This paper focuses on the modelling of a research type Heavy Duty Common Rail (CR) fuel injection
system. More specifically it reports on the observed interaction between fuel properties and injection and on
the capability to model this. For that reason a hydraulic model of the fuel injection system has been
developed using the AMESim code (Imagine S.A., 2003).
The reliability of the numerical results is tested through a comparison between numerical and experimental
results when using regular diesel fuel. Basis for this detailed comparison are measurements of injected mass
flow rate, needle lift and pressure oscillations in the injection duct for a single injection. Simulation results
for regular diesel show good agreement with measured data for pressure oscillations in the injection duct,
needle lift and injected fuel mass flow rate.
A comparison of experimental and simulated results for Rapeseed oil Methyl Ester (RME) also shows good
correspondence, which proves the capability of the model to capture the influence of different fuel
properties.
Keywords: Fuel injection, Modelling, Fluid properties
pressure are fluid density and fluid bulk modulus of
1 INTRODUCTION
With modern diesel engines the injection process elasticity. Again, fuel viscosity was not found to
(i.e. the injection rate and injection pressure) has a affect injection timing. All of the above analyzed
major impact on noise production, exhaust gas pump-in-line fuel injection systems.
emissions and fuel consumption. In view of the Ziejewski et al. [6] studied the discharge coefficient
ever-increasing demands on these engines, of a diesel injector nozzle for laminar and turbulent
modelling of the fuel injection system has become flows and for different (alternative) fuels. He found
an essential step in the fuel injection equipment that for a specific injector nozzle geometry a
design and optimisation process. relation between Reynolds number and discharge
As part of the move towards greenhouse gas coefficient could be established that was valid
reduction and diversification of energy supply there irrespective of fuel properties.
is a growing interest to test and enhance the ability The goal of the present study is to validate our
of diesel engines to run on alternative fuels. hydraulic, one-dimensional, model of a Common
Because fuel properties such as density, bulk Rail (CR) type injection system using both standard
modulus of elasticity and viscosity influence the diesel and RME and to analyze differences in the
injection behaviour, the use of an alternative fuel fuel injection behaviour related to the difference in
will affect the injection process. Szybist et al. [1] fluid properties. The capability of the model to
found that injection timing advanced and injection capture these differences is tested.
duration shortened with increased biodiesel content. 2 COMMON RAIL INJECTION SYSTEM
The higher viscosity of biodiesel is believed to be The studied fuel injection system is of the Common
responsible for the change in injection timing by Rail (CR) type and is designed for the fuel delivery
Choi et al. [2], but no mechanism is offered for this to a one-cylinder 2.1-litres Heavy Duty (HD)
conclusion. Arcoumanis et al. conclude that only research diesel engine. In figure 1, this injection
fluid bulk modulus of elasticity is responsible for a system is shown schematically.
change in injection timing. Rapokoulos and The system comprises a 2nd generation light-duty
Hountalas [5] state that the only fluid properties CR high-pressure pump with an electronically
that are of importance for modeling of fuel line controlled throttling valve to adjust the delivered
 Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland
Data acquisition A-throttle Control
1. Valve and injection
Zeuch chamber
2. Adjustable pressure control
chamber
limiter Solenoid spring
3. CR Throttling valve 5.
CR injector
4. Pressure/temperature with needle- Solenoid
sensor on injection duct lift sensor
·
.... ··· · ....
5. Pressure sensor in 4. Ball valve .. ..
Zeuch chamber Rail .. ..
Z-throttle
CR pump
2. Heat
1.
exchanger
Control plunger Edge-filter
3.
Fuel tank
Frequency controlled Filter Electrical Needle spring Needle-tip
electrical engine supply pump

Needle ·· ··
Figure 1. Common Rail fuel injection system.
mass flow rate. The injector is an 8-hole HD diesel
engine CR sac-hole nozzle injector. The nozzle
holes have a nominal diameter of 0.184 mm, a Sac-volume
Nozzle hole
length/diameter ratio of ≈5 and ≈7% inlet rounding.
Since in this study only stationary conditions are to
be tested, an adjustable mechanical pressure limiter Figure 2. Details of the CR injector and
is used to control the pressure in the system. A heat corresponding hydraulic AMESim model.
exchanger is included to control the temperature in injector can also be distinguished. The model can
the fuel injection system. Maximum injection be seen as a chain of restrictive and capacitive
pressure is 1400 bar. elements. In the capacitive elements, such as the
3 MODEL OF CR INJECTION SYSTEM control chamber and nozzle chamber, fluid pressure
The high-pressure part of the CR injection system and temperature are calculated.
depicted in figure 1 is modelled using the AMESim In the resistive elements, such as the A-throttle,
code [6] (Imagine S.A., 2004). In this code each only the mass flow rate, m , is calculated, based on
physical component of the system is represented by the Bernoulli equation:
an appropriate icon, and is associated to one or 2 ∆p
more lumped parameter models (called submodels). m = ρ ⋅ C d (λ ) ⋅ A (1)
Principally, in AMESim hydraulic systems can be ρ
modelled by isothermal and/or adiabatic submodels where, ρ is the fluid density, C d , is the discharge
that are either one-dimensional (fuel lines) or zero coefficient and A the (lift-dependent) cross-
dimensional (restrictions, volumes). Because of the sectional area. For non-cavitating flow, C d is
fast nature of the injection process, adiabatic dependent on flow velocity, fluid density and
submodels have been used in this study. viscosity only. This dependency is taken into
In figure 2 a schematic representation of the CR account by use of a flow number, λ .
injector is given together with details from the To correct for cavitation, a cavitation number CN
corresponding AMESim model. When the solenoid is introduced. This number is defined as the ratio of
in the top of the injector is energized, the resulting the pressure difference over the restriction to the
magnetic force lifts the ball valve from its seat. downstream pressure. If the cavitation number is
Because the flow rate through the Z-throttle is higher than a pre-defined critical cavitation number,
smaller than the flow rate through the A-throttle,
CN critical , equation (1) changes to [7]:
the pressure in the control chamber drops. Because
the rail pressure is still present at the needle tip, the 2 ∆p 1 + CN
needle is pushed upwards, starting the injection. As m = ρ ⋅ C dc (λ ) ⋅ A ⋅ ⋅ (2)
ρ CN
the current through the solenoid is stopped, the
solenoid spring forces the ball valve back on its with C dc the discharge coefficient at the limit of
seat. As a result, the pressure in the control cavitation ( CN → ∞ ). In the injector model,
chamber increases again and the needle is pushed cavitation is modelled in the nozzle holes and A-
down on its seat, thus stopping the injection. and Z-throttle.
In the graphical representation of the AMESim All capacitive and restrictive components can be
model in figure 2, the physical elements of the connected by long fuel lines. In these one-
 Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland
Dimensional hydraulic lines pressure wave during injection. A possible explanation for this is
dynamics is taken into account. the presence of cavitation inside the Zeuch
Dynamics of motion is evaluated in mass chamber. Also, rapid opening and closing of the
components. The spring-damper component in injector induces pressure waves inside the chamber.
figure 2 models the elasticity of the control plunger The model of the Zeuch chamber does not capture
and injector needle. In so-called ‘transformation these phenomena.
elements’, such as piston components, a pressure is
Measured
transformed into a force. Other components in the 0.09
1400 bar Simulated

injection system are modelled in an analogous 0.08

1200 bar
manner [6]. 0.07
1000 bar

Injection rate [kg/s]

0.06
800 bar
4 INJECTION MEASUREMENTS 0.05
Injection measurements with both standard diesel 0.04
and Rapeseed oil Methyl Ester (RME) have been 0.03
performed to validate the model. In table 1 the 0.02

relevant physical properties of both fuels are 0.01

presented. 0
-0.01
Table 1. Fluid properties of used fuels. 0 1 2 3 4
Time [ms]
5 6 7 8

Fluid property (20oC) Diesel RME Figure 3. Measured and simulated injection
Density [kg/m3] 823.4 882.5 rates for different rail pressures.
Bulk modulus [MPa] 1393.0 1522.5
Kinematic viscosity [cSt] 4.59 6.57 The lower order frequencies in the injection rate are
caused by fluctuations in injection pressure. The
The signals used to validate the model are the link between rail pressure and injection rate is of
measured pressure in the injection duct, at a course through the different obstructions and lines
position 120 mm from the injector entrance, the in the injector. Figure 4 shows the measured and
needle lift and the pressure inside the Zeuch simulated pressure in the injection duct.
chamber, see also figure 1. From the latter, the 1500 start of injection Measured
injected mass or volumetric flow rate can be 1400
Simulated

determined using the Zeuch method [8]. In this Lift ball valve
1300
method injections are performed into a pressurized end of injection "water hammer"
chamber (Zeuch-chamber) filled with fuel.
Pressure [bar]

1200

Since geometrical parameters are known, the only 1100

unknown parameters are discharge coefficients and
1000
loss coefficients of the different restrictive
components. These parameters are obtained by 900 Injection duration

tuning the results from the AMESim CR model to 800

the injection measurement data. 700
0 1 2 3 4 5 6 7 8
4.1 Results for standard diesel Time [ms]

In figure 3 the measured and simulated injection

rate is depicted for a range of rail pressures, typical Figure 4. Measured and simulated pressure in
for modern diesel engines. The hold current of the injection duct for different rail pressures.
input signal to the injector is held constant at 3ms The main frequency of the pressure waves, ~850
for all rail pressures, resulting in 3.8 ms injection Hz, is equal to the frequency of a standing wave in
duration. a one-sided open duct when the rail acts as the open
Both the simulated start and end of the injection end and the injector nozzle holes as the closed end.
correspond well to the measurements. Also, the As shown in figure 4, the end of injection is
main fluctuations during full needle lift are characterised by a fast pressure increase caused by
captured well by the model. the rapid closing of the injector (‘water hammer-
The higher order frequencies in the measured effect’). From figure 4 it can be concluded that
injection rates are caused mainly by pressure simulations and measurements correspond well.
fluctuations that arise inside the Zeuch chamber Both frequency and amplitude of the pressure
 Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland
waves are calculated well. Accurate modelling of control plunger is measured at the top of the
the frequency of the pressure waves is mainly plunger and does not correspond to the actual
dependent on the density and bulk modulus of the displacement of the needle tip.
fluid, which determine the wave travelling velocity. From figure 5 it can be concluded that the control
For the amplitude (i.e. damping) of the pressure plunger displacement is simulated well for 800 bar
waves between rail and nozzle exit, the loss rail pressure, but for the case of 1400 bar rail
coefficient of the edge filter (see figure 2) was pressure, the calculated lift is slightly slower.
found to have a large influence. Calculations at other rail pressures confirm a trend
The critical cavitation number is taken equal to 3, of greater differences in lift at higher rail pressures.
which is a common value for diesel injectors, see The slower calculated needle lift causes the
e.g. [7]. Mean cavitation numbers during the injection rate to increase slower than the measured
injection range from 13 for 800 bar rail pressure to rate during needle lift, as can be seen in figure 3.
20 for 1400 bar rail pressure (downstream pressure The important model parameters here are the
40 bar). These values are well above the critical discharge coefficients of the A- and Z-throttle.
cavitation number indicating that the flow is Calculated cavitation numbers indicate that the flow
cavitating. Therefore, for modern diesel engines the is non-cavitating in both throttles. Tuned discharge
operating range is well outside the range discussed coefficients, corresponding to fully turbulent flow,
by Ziejewski et al. [5] and discharge coefficients are 0.84 for the A- and 0.75 for the Z-throttle.
are not a function of Reynolds numbers only.
4.2 Results for RME
Injection start is determined by the elasticity of the
Injection measurements have also been performed
control plunger and injector needle (see figure 2).
using Rapeseed oil Methyl Ester (RME). The
As can be seen in figure 4, a slight decrease in
AMESim CR model is again tuned to these results,
pressure is already present before the actual
now using RME fluid properties.
injection starts. When the pressure in the control
chamber of the injector drops as a result of the 0.09
Measured
1400 bar Simulated
lifted ball valve, the control plunger will first 0.08
elongate before it actually lifts. 0.07
1200 bar
1000 bar
Injection rate [kg/s]

Measured 1400 bar

0.25 0.06
Measured 800 bar
Simulated 1400 bar 800 bar
0.2
0.05
Simulated 800 bar
0.04
0.15
0.03
Displacement [mm]

0.1
Injection start 0.02
1 = 800 bar
0.05 Elastic 0.01
2 = 1400 bar Injection end
deformation
0 0
1.
-0.05 2.
-0.01
0 1 2 3 4 5 6 7 8
-0.1 Time [ms]

-0.15 Elastic deformation

during injection Figure 6. Measured and simulated injection
-0.2
0 1 2 3 4 5 6 7 8 rate for RME at different rail pressures.
Time [ms]
Also for RME good correspondence between
Figure 5. Measured and simulated control measured and simulated injection rates is found as
plunger displacement. can be concluded from figure 6 and figure 7. Start
and end of injection match well. Amplitude and
Figure 5 shows the measured and calculated
main frequency of the fluctuations in the injection
displacement of the control plunger for two
rate are well captured by the model. This indicates
different rail pressures. At zero rail pressure the
that the injection pressure is simulated correctly.
maximum lift of the control plunger is 0.25 mm.
The amplitude of the higher order frequencies seem
During injection, the control plunger and needle are
be higher than for standard diesel. This can be an
still deformed by the rail pressure working at the
indication for more severe cavitation.
bottom of the injector needle. This compression
Control plunger displacement shows the same trend
causes the injection end to correspond with larger
as standard diesel fuel. For higher rail pressures the
displacements than injection start, as is shown in
calculated control plunger lift lags behind, but more
figure 5. This is because the displacement of the
than for standard diesel fuel.
 Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland
0.25
Measured 1400 bar
Measured 800 bar
somewhere in between, as do our tuned discharge
Simulated 1400 bar
Simulated 800 bar
coefficients. In contrast to many other authors
0.2
Goney et al. use actual sac-pressure measurements.
0.15
Ganippa et al. [11] found discharge coefficients
0.1 Injection start around 0.78-0.8 for the same range of cavitation
Displacement [mm]

1 = 800 bar numbers as calculated in our measurements. In

0.05 2 = 1400 bar
Elastic
deformation
Injection end view of the above and for lack of more data, tuned
0
1. discharge coefficients were considered to be
-0.05
2. acceptable and they were retained. Experiments are
-0.1 being conducted for further evaluation of the
-0.15 Elastic deformation capability of the AMESim code to model the
during injection injection system.
-0.2
0 1 2 3 4 5 6 7 8
Time [ms] 5 INFLUENCE OF FLUID PROPERTIES
The influence of the difference in fluid properties
Figure 7. Measured and simulated control
between standard diesel and RME are most evident
plunger displacement for RME.
when the volumetric injection rates are compared,
4.3 Discussion see figure 8. For RME a decrease in mean injection
For modelling of the injection rate, discharge rate of 6.9% with respect to diesel fuel is present.
coefficients of the nozzle holes and injector needle 4
x 10
10
tip passage are key parameters. Tuned discharge RME
9 Stand. diesel
coefficients for the needle tip and injector nozzle 1200 bar
8
holes are respectively 0.9 and 0.835. These values Volumetric flow rate [mm /s]
3

7
are relatively high. A critical cavitation number of 3
6
has been used, which means that the discharge
5
coefficient of the nozzle holes would be equal to
4
0.96 for fully turbulent and non-cavitating flow, see
equation (4). For the length/diameter ratio of the 3

used nozzle hole geometry the correlation of 2

Lichtarowicz [9], for non-cavitating flow and 1

sharp-edged nozzle holes, gives a discharge 0

0 1 2 3 4 5 6 7 8
coefficient of 0.78. For fully cavitating flows Time [ms]

discharge coefficients of ~0.73 are found, e.g. by

Favennec et al. [7]. Figure 8. Measured volumetric injection rate for
No real improvement could be made to lower the standard diesel and RME.
tuned values of the discharge coefficients. Since the The higher density of RME causes the volumetric
pressure at the injector entrance is simulated injection rate to be lower. From the lower flow
correctly, lower values for the discharge coefficient velocities due to the higher viscosity and density of
of the nozzle holes can only be obtained by a RME it is expected that cavitation is less severe and
smaller pressure drop inside the injector. However, discharge coefficients are higher. However, in the
even when no losses are assumed, a decrease of CR model, the discharge coefficient for maximum
only 3.6% in discharge coefficient is realised. cavitation of the injector nozzle holes had to be
Also the uncertainty in measured injection rate lowered by 3.2%. A possible explanation for this
(<3.8%), which is merely present through an decrease can be a higher air release during injection
uncertainty in used fluid bulk modulus of elasticity, because of more dissolved gas in the RME.
is too low to be significant. Because the results for In contrast to Szybist et al. [1] and Arcoumanis et
RME also give relatively high discharge al. [3], no difference in injection start can be
coefficients, the bulk modulus of the fluid is not distinguished between standard diesel and RME.
taken as the error source. On the other hand, Goney Because in CR systems the injector needle is not
et al. [8] also found high values for the discharge lifted through a pressure increase by the pump (as
coefficients (0.8 – 0.925) for cavitating flows in a in pump-in-line systems), injection timing is not
sharp-edged nozzle hole and 0.85 – 0.975 for non- influenced by the compressibility of the fuel.
cavitating flows through a nozzle hole with high The increase in flow rate during needle lift is
inlet rounding. Our nozzle geometry lies slightly slower for RME. This is the result of a
 Proceedings of VAFSEP2004, 6-9 July 2004, Dublin, Ireland
slower needle lift, as can be seen in figure 9. The This research will be continued for other alternative
slower lift is caused by lower volumetric flow rates fuels with a wider variation of fuel properties.
through the A- and Z-throttle as a result of the
higher density of RME. This results in a slower ACKNOWLEDGEMENTS
pressure drop in the control chamber of the injector. The authors greatly acknowledge Imagine S.A. for
Also, the flow rate into the sac-volume is lower. the use of AMESim.
The pressure rise in the sac-volume is therefore
slower and the needle is pushed upwards with less REFERENCES
force. Because the flow through the A- and Z- [1] James P. Szybist, André L. Boehman, Behavior
throttle is turbulent almost immediately, the of a diesel injection system with biodiesel fuel,
viscosity of RME only has a marginal influence. SAE International, SAE paper 2003-01-1039
(2003)
RME
0.35 Standaard diesel [2] C.Y. Choi, G.R. Bower, R.D. Reitz, Effect of
0.3
biodiesel blended fuels and multiple injections
on D.I. diesel engines, SAE International
0.25
Congress and Exposition, SAE 970218 (1997)
Lift [mm]

0.2 [3] C. Arcoumanis, et al., Application of FIE

0.15 computer model to an in-line pump-based
0.1
injection system for diesel engines, SAE
International, SAE paper 970348 (1997)
0.05
[4] C.D. Rapopoulos, D.T. Hountalas, A
1.4 1.6 1.8 2 2.2 2.4 simulation analysis of a DI diesel engine fuel
Time [ms]
injection system fitted with a constant pressure
Figure 9. Measured control plunger movement valve, Energy Conversion Management,
for standard diesel and RME. Vol.37(2) (1996), 135-150
[5] Mariusz Ziejewski, Hans J. Goettler, Discharge
In general, differences are small and it can be coefficients for multi-hole fuel injection nozzle
concluded that the use of RME does not have a for alternative fuels, SAE International, SAE
large effect on the injection behaviour. To observe paper 890448
more pronounced differences, a wider spread of [6] Imagine S.A., AMESim 4.0 – User manual,
fuel viscosity and density must be used. (2002)
6 CONCLUSIONS [7] Anne-gaël Favennec, Michel Lebrun, Models
The model of the CR injection system presented in for injector nozzles, The Sixth Scandinavian
this study gives good results for both standard Conference on fluid Power, SICF’99 (1999).
diesel fuel and Rapeseed oil Methyl Ester. Though, [8] W. Zeuch., Neue Verfahren zur Messung de
differences between RME and diesel fuel are small. Einspritzgesetzes und de Einspritz-
It is therefore expected that also blends of standard Regelmä igkeit von Diesel-Einspritzpumpen,
diesel and RME will be simulated correctly. MTZ, nr. 9 (1961)
Used discharge coefficients for the injector nozzle [9] A. Lichtarowitcz, R.K. Duggings, E. Markland,
holes and needle tip passage are found to be quite Discharge coefficients for incompressible non-
high, but resulted in acceptable calculation results. cavitating flow through long orifices, Journal
For an optimal simulation when using a different Mechanical Engineering Science, Vol. 7(3)
fuel, experimental results are necessary to (1965), 210-219
determine fluid properties and discharge [10] Kayhan H. Goney, Michael L. Corradini,
coefficients of the different components. Isolated effects of ambient pressure, nozzle
In CR systems injection timing is not influenced by cavitation and hole inlet geometry on diesel
fluid bulk modulus of elasticity and in the normal injection spray characteristics, SAE
operating range of diesel engines, fluid viscosity International, SAE paper 2000-01-2043 (2000)
only has a marginal influence on needle lift because [11] Lionel Christopher Ganippa, Sven Andersson,
the flow is turbulent almost immediately. Jerzy Chomiak, Transient measurements of
For research on the influence of fluid properties, a discharge coefficients of diesel nozzles, SAE
wider spread in fluid properties must be present. International, SAE paper 2000-01-2788 (2000)