Professional Documents
Culture Documents
Esda2006 95479
Esda2006 95479
net/publication/256352653
CITATIONS READS
4 1,192
4 authors, including:
All content following this page was uploaded by Riccardo Amirante on 01 March 2014.
ESDA2006-95479
ESDA2006-95479
15
δ pipe thickness
10
Subscripts
i step index 5
Superscripts 0
n time level 0 200 400 600 800 1000 1200 1400
Time (µs)
C1 1 µF (J100)
C2 C5 C7
10 nF
C9
C3 2.2 nF (J400)
0.015 µF
C4
(J100)
CAPACITORS C6 C8 3.3 nF (J100)
C12 100 nF
C13 0.33 µf
Tab 1 – ECU components list. Fig 4 - Current profile during a decreasing PWM duty cycle
phase.
In order to obtain the dynamic behaviour of a commercial
injection system and the current waveform shown in fig 2 in As an example, Figure 5 provides a diagram of the injection
the solenoid-valve, a suitable electronic control unit ECU pressure versus the time, measured by means of folded
(15) has been developed. Figure 3 and table 1 show the strain gages. An unrealistic pressure increase is visible in
circuit which controls the solenoid current by switching a Figure 5, approximately between t=570 and t=770
power mosfet with a waveform produced by proper Pulse microseconds; in this period, the injection has not started
Width Modulation (PWM) drivers. This equipment supplies yet: in fact, the pin lift is measured as the displacement of
the solenoid with a pulsating voltage with a squared the top of the control piston rather than of the needle. Thus,
waveform at 30 kHz frequency. The current value is a small pin lift is measured even when the compression on
modulated by changing the duty cycle of the waveform. In the pin is reducing, but the needle is still on its sealing seat.
particular, the ECU PWM generator employs two NE555 The measured pressure increase can be explained as follows:
timers to set up the injection duration (1) and the peak when the control piston presses the needle on its sealing
duration (2); other two similar timers are employed to seat, the injector body is stretched; the folded strain gages
generate the PWM signals which realize the peak (3) and are obviously calibrated for this stress distribution, which
hold (4) current values. The peek and hold signals are mixed occurs when there is no injection. On the contrary, when the
1400
1350
1300
1250
Pressure (bar)
1200
1150
1100
1050
1000
950
900
0 500 1000 1500 2000 2500 3000
Time [µs]
∂ρ 1 ∂p
∂ t = 2 ∂ t
c
(5)
∂ ρ 1 ∂p
= 2
∂ x c ∂x
By substituting the above expressions in the continuity
equation, and after rearranging the momentum equation, the
following set of governing equations is obtained:
∂p ∂p 2 ∂u
∂t + u ∂x + ρc ∂x = 0
∂u (6)
+u
∂u ∂p λu 3
Fig. 10 - The test bench + + =0
∂t ∂x ρ∂x 2 u D
pin+1 − pin+1/ 2 ∗ ∗
∗ pi − pi−1
∗ ∗
2 ui − ui−1 1250
+ u i + ρ c
i i =0
∆t ∆x ∆x 1200
2
(10)
pressure (bar)
n+1 n+1/ 2 1150
ui − ui
+ n
( ) ( )
+
2
1 pi∗ − pi∗−1 ui∗ − ui∗−1
2
( )
+
λ ui∗
3
=0
1100
∆t ∆x 2∆x
ρi 2D ui∗ 1050
2 1000
without sensor
950
The values of p and u at the time level n+1/2 are evaluated with sensor
as: 900
600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000
n+1 / 2 pi∗ + pin time (ms)
pi =
2 Fig. 11 - Comparison between the numerical pressure trend
(11) measured in the nozzle with and without the volume
∗ n
n +1 / 2 u i + u i necessary to assembly the sensor on the nozzle.
u i =
2 1300
Further models and equations are used for sudden section
1250
variations, pipe bifurcations and concentrated volumes
[15,16]. The corresponding expressions are omitted for 1200
conciseness.
pressure (bar)
1150
The code has been applied to the simulation of the unsteady 1100
flow determined by one injection in a common-rail system
equipped with one injector only. The resulting time diagram 1050
injection starts. Then, as usual (see also Figure 6), the with sensor
900
pressure decreases due to the flow acceleration; the system 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000
of pressure waves causes the following increase of the time (ms)
pressure, which reaches or even overcomes the rail pressure. Fig. 12 - Comparison between the numerical pressure trend
Higher frequency pressure oscillations are due to the measured in the pipe with and without the volume necessary
pressure waves reflected by the concentrated volumes. to assembly the sensor on the pipe
More importantly, the pipe lay-out has been then modified 1300
by properly inserting the pipe and the volume of the clamp
1250
adapter, to investigate the intrusivity of the employed
pressure sensor. The time diagram of the injection pressure, 1200
1150
injection nozzle, is provided in Figure 11 as a grey line. As
1100
clearly seen, the two curves of the injection pressure are
almost superposed: the major differences, of about 4 bar, 1050
Pressure (bar)
at t=45800 µs are determined by the wave reflections and by
Voltage (V)
15
1300
the injector closing, respectively. The pressure overcomes
1280
the rail pressure, reaching a maximum value of 1360 bar, 10
1260 Pipe pressure
and then decreases again below its initial value. After the Rail pressure
end of the injection phase, the pressure diagram presents 1240 Injector voltage waveform 5
large oscillations. 1220
1380 25 1200 0
44500 54500 64500 74500 84500
1360
Time (µs)
1340 20
1320 Fig. 15 - Pressure measured in the pipe, in the rail and the
Pipe pressure (bar)
1280
1260 10 The voltage profile has been finally set up so as to obtain
1240 two very close injections, in order to simulate the operating
1220 5 conditions of a multijet system. The aim is to verify that the
1200 circuit proposed in substitution of the standard automotive
1180 0 ECU is capable of realizing very close injections, and thus
44500 45000 45500 46000 46500 47000 47500 to analyze the pressure variations occurring in the multijet
Time (µs) systems. Experimental tests have been performed with rail
Fig. 14 - Time diagram of the pressure measured by piezo- pressure set up at 1320 bar, excitation time of 750 µs and
resistive pressure sensor in the pipe. dwell time of 1.5 ms, instead of the previous value of 40 ms.
The two injections can be approximately localized in time
Figure 15 provides the pressure diagram for a complete by looking at the voltage profile shown in Figure 17. As in
injection cycle. After reaching the lowest value of 1280 bar the previous case, an injection pressure drop of about 150
(pressure drop of 40 bar), approximately 1700 µs after the bar is measured during the first injection.
injector opening, the rail pressure progressively increases up 1380 25
to the value set up by the regulation valve, which is reached 1360
about 30 ms after the injector closing, namely close to the 1340
20
1320
following injection. The high-frequency oscillations of the
1300
pipe pressure are progressively damped: in particular, these
Pressure (bar)
Voltage (V)
1280 15
oscillations are significant for a period of about 8 ms, 1260
whereas low frequency oscillations are measured afterwards, 1240 10
which almost follow the rail pressure variation. Almost 1220
1200
steady conditions are achieved about 10 ms before the 5
1180 Nozzle pressure
following injection. 1160 Injector voltage waveform
The clamp adapter has been then mounted on the injector, so 1140 0
as to connect the pressure sensor to a section of the fuel 44500 45000 45500 46000 46500 47000 47500 48000
supply system very close to the injection nozzle, as shown in Time (µs)
Figure 8. Experimental tests have been performed with rail Fig. 16 - Pressure trend measured by piezo-resistive
pressure set up at 1315 bar, excitation time of 750 µs and pressure sensor in the nozzle
dwell time of 40 ms between two consecutive injections.
1280 15
1260
1240 Common-Rail Injection Systems: The Effect of the
1220 10 Multiple Injections on Emissions” SAE Paper 2001-01-
1200
1180 1068, 2001.
1160 5 [7] Bianchi, G.M., Pelloni, P., Filicori, F., and Vannini,G.,
1140 Nozzle pressure
1120 Injector voltage waveform “Optimization of the Solenoid valve Behavior in
1100 0 Common-Rail Injection Systems” SAE Paper 2000-01-
4000 6000 8000 10000 12000
2042, 2000.
Time (µs)
[8] Bianchi, G.M., Falfari, S., Pelloni P., Filicori,
Fig. 17 – Pressure diagrams for two close injections. F.,Milani, M., “Development of a Dynamic Model for
Studying the 1st Generation of Common Rail Injectors
for HSDI Diesel Engines” ICE, 2001, Capri 2001.
4. Conclusions [9] Beierer, P., Huhtala, K., Lehto, E, and Vilenius, M.,
The influence of the internal pipes and volumes of the clamp “Study of the Impact of System Characteristics on
adapter and of the pressure transducer has been evaluated Pressure Oscillations in a Common rail Diesel Fuel
numerically. The comparison between the time diagrams Injection System”, SAE Paper No. 2005-01-0910, 2005
without and with the sensor has assessed the reliability of [10] Catania, A. E., Ferrari, A., Manno, M., and Spessa, E.,
the measurement system, except when measuring the “Experimental Investigation of Dynamic Effects on
pressure in the supply pipe, where larger errors occur during Multiple-Injection Common Rail System
the increasing pressure phase. Performance”, Proceedings, Spring Technical
The developed test bench and the injectors driver system Conference of the ASME ICED, Chicago, IL, April 5-
allows to realize both single and multiple injections, thanks 7, 2005
to a suitable electronic circuit, which controls the current [11] L. A. Catalano, V. A. Tondolo, A. Dadone “Dynamic
profile in the solenoid-valve, by means of PWM drivers. It Rise of Pressure in the Common-Rail Fuel Injection
also allows to perform direct pressure measurements in System” SAE Paper 2002-01-0210, SAE Transactions
different sections and, in particular, near the injection – Journal of Engines, 2002
nozzle, thanks to the use of a newly-developed fast-response [12] Shames I.H.,“Mechanics of fluids”, McGraw-Hill,1992
pressure transducer. Experimental results obtained with both [13] Bosch, Private Communications, 2002
single and multiple injections have been provided to [14] Anderson J.D., “Modern compressible flow with
demonstrate the capability of the test bench of analyzing the historical perspective”, McGraw-Hill Book Company,
dynamical effects which occur in the pipes of a Common- 1982
Rail Diesel injection system, and of optimizing the [15] Roache P.J., “Computational fluid dynamics”, Hermosa
synchronism of multiple injections. In particular, the second Publishers, Albuquerque, 1972
injection should start before the injection pressure reaches [16] Benedict R.P., “Fundamentals of pipe flow”, John
its peak, so as to exploit the positive effects of the Wiley & Sons, New York, 1980
compression wave reaching the nozzle.