Development and performance assessment of the new-generation CF fuel injection system for diesel passenger cars

Applied Energy 91 (2012) 483–495

Contents lists available at SciVerse ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Development and performance assessment of the new-generation CF fuel injection system for diesel passenger cars A.E. Catania, A. Ferrari ⇑ Dipartimento di Energetica, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy

a r t i c l e

i n f o

Article history: Received 20 June 2011 Received in revised form 23 August 2011 Accepted 25 August 2011 Available online 13 November 2011 Keywords: Fuel injection system Rail CF system

a b s t r a c t An innovative hydraulic layout for high-pressure fuel injection systems has been developed. The rail has been removed from the high-pressure circuit of standard Common Rail injection apparatus in order to have: reduced production costs, easy installation on the engine and a fast dynamic response of the injection system during engine transients. The innovative high-pressure pump has been machined with extra delivery ports in order to obtain direct connection with each of the injector feeding pipes. A special pump delivery-chamber, featuring an adequate volume, was designed to provide the minimum hydraulic capacitance required for a proper control of the high-pressure in the system. A prototype of the new-concept Common Feeding (CF) injection system was set up at the Politecnico di Torino laboratory and subjected to an extensive experimental campaign on an advanced hydraulic test rig. The hydraulic performance of the new apparatus was compared with that of standard Common Rail system, in terms of injected flow-rate time histories, cycle-to-cycle dispersion and multiple injection dependence on dwell time. Finally, the new-generation system was installed on a Euro 5 diesel engine and tested in a dynamometer-cell. Pollutant emissions were measured at some reference working conditions as well as along the NEDC and compared to analogous data obtained for the same engine, but set up applying the standard Common Rail. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction After approximately 10 years of production, the Common Rail (CR) fuel injection apparatus has reached a high level of sophistication that ensures accuracy and repeatability of the injected amounts [1–5]. Commercial CR systems are capable of applying up to five injection shots per engine cycle with close-to-zero dwell times and can control tiny injected quantities, even smaller than 0.7 mm3 in a precise manner [6]. Besides, the application of piezo-technology has recently allowed directly actuated injectors to be produced, with flow-rate shaping capability [7]. Nevertheless, injection system upgrading is still ongoing in order to cope with today’s market demand. The main drivers are: lower emissions along with reduced fuel consumption [8,9], robustness against different types of fuel quality [10,11], affordable technologies for emerging markets and low-priced vehicles [12]. An important branch of the current industrial research has focused on the development of innovative high-pressure layouts that are capable of preserving the injection system performance, but with a remarkable cost reduction [13].

⇑ Corresponding author. Tel.: +39 0110904426; fax: +39 0110904599. E-mail address: [email protected] (A. Ferrari). 0306-2619/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.apenergy.2011.08.047

A key point in the design of low-priced high-performance layouts pertains to the assessment of the rail volume effects on system dynamics [14]. An injector-upstream pressure level, which is always kept adequately close to the desired nominal value reported in the engine electronic control unit (ECU) maps, is essential to guarantee accuracy and repeatability of the injected fuel quantities. In commercial setups, two different strategies, which apply two distinct actuators, are used to control the high-pressure level [15]. In the first, a pressure control valve (PCV) discharges the excess pumped fuel, that is not required for the injectors [16]. In the other strategy, a fuel metering valve (FMV) controls the flow-rate at the pump inlet [17]. In both cases, a relatively large volume of the accumulator has traditionally been considered fundamental to dampen the pressure fluctuations caused by the fuel pulses delivered by the pump and the fuel-injection cycles [15]. Preliminary experimental analyses with lower accumulation volumes than typical production range ones were made in [18]. The progressive reduction in the accumulator volume, from 20 to 5 cm3, was shown to not affect the injector performance to any great extent. In fact, the high-pressure control capability resulted from the synergic action of both the system hydraulic capacitance and the pressure control device. Although the duty cycle of either the PCV or the FMV depended on the rail size, the high-pressure control system was capable of keeping the pressure level

484

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

Nomenclature a bmep bsfc d d1 DT ECU EMI EVI ET FMV G ICEAL iV K KMM l l1 M MFB50 MR NEDC p Pu PCV Q SOImain

sound speed brake mean effective pressure brake specific fuel consumption inner diameter of the injector feeding pipe inner diameter of the pump-to-rail pipe dwell time electronic control unit indicator of the injected volume injection-rate indicator energizing time fuel metering valve mass flow-rate internal combustion engine advanced laboratory engine total displacement geometrical term in the expression of the static leakage continuous measuring flow meter system length of the injector feeding pipe length of the pump-to-rail pipe fuel mass instant at which half of the injected mass is burned Minirail New European Driving Cycle pressure engine power pressure control valve volumetric flow-rate start angle of the main current pulse to the injector

adequately close to the nominal value in all the analyzed setups. From this result, two different approaches can be followed to develop innovative CR hydraulic layouts: one leads to a high-pressure circuit without a rail, the other to systems that apply larger rail volumes than those currently available in production, but without any electronic feedback control of the pressure [19]. As far as the possibility of removing the rail from the high-pressure circuit is concerned, the innovative layout, with a pipe junction, proposed and investigated in [20] should be regarded as an intermediate step. The junction (Minirail) had the main function of connecting the pump to the injectors and providing the hydraulic circuit with an accumulation volume of approximately 2 cm3. The fuelling capability of the injectors was not altered compared to the standard CR and the injector-to-injector fluid dynamics interference continued to be negligible. However, the Minirail system does not seem a radically innovative concept since its layout architecture is basically the same as that of the standard CR [20]: the major difference only concerns the accumulator dimensions, which are dramatically reduced. The removal of the rail from the piping system would lead to a highly innovative hydraulic layout and to further remarkable advantages, in terms of cost reduction and easy installation on engine, compared to the Minirail system, but it requires a suitable new location for both the PCV and the ECU pressure sensor. 2. Experimental facilities The hydraulic activity was carried out on the Moehwald-Bosch MEP2000-CA4000 test rig [21], installed in the IC Engine Advanced Laboratory (ICEAL) at the Politecnico di Torino. The bench, rating 35 kW of maximum power and 6100 rpm of maximum shaft speed, can simulate engine-like working conditions and is equipped with the following main measuring instruments: EMI to gauge the whole injected oil mass and, separately, the volume injected at

t T V / k

l r

time temperature; natural period of the pressure waves fuel volume restrictor diameter relative air-to-fuel ratio, normalized with respect to the stochiometric value fuel dynamical viscosity standard deviation measured by the EMI; percentage deviation in the injected volume fluctuations

Subscripts acc accumulator EMI measured by the indicator of the injected volume inj injected inj inlet at the injector inlet leak leakage main main injection max maximum min minimum nom nominal level in the rail PCV through the pressure control valve pil pilot injection post post injection pump at pump delivery rail rail tank in the tank

each shot, in multiple injections; a Bosch type injection-rate indicator, EVI; KMM flow-meters to continuously detect the flow-rate recirculated by each injector; piezoresistive sensors to monitor the pressures in the rail and at the injector inlet. The tests on the engine were carried out in the ICEAL high-performance dynamometer cell [22]. This cell has an ‘ELIN AVL APA 100’ cradle mounted AC dynamometer, with nominal torque and power of 525 Nm and 220 kW, respectively. The ‘AVL KMA 4000’ fuel metering system is capable of continuously and accurately measuring the engine fuel consumption over a 0.28–110 kg/h range. As far as the exhaust gas analysis is concerned, an AVL AMAi60 raw exhaust-gas multipurpose analyzer was employed to gauge the engine-out pollutant emission levels of NOx, HC and CO. In order to measure the soot from the diesel engine, the dynamic test bench was equipped with an AVL415S smoke meter. Furthermore, a high-frequency piezoelectric transducer was installed on the glow plug seat of one cylinder by means of a specific adapter in order to measure the pressure time-history of the gases in the combustion chamber and be able to perform a heat release analysis. 3. Capacitance and inertial effect of the accumulator Figs. 1 and 2 report the time histories of the experimental pressure at the injector inlet and in the rail (junction) as well as the injected flow-rate from the EVI profile [23], for both the Minirail system (Vacc = 2 cm3) and the standard Multijet CR apparatus (Vacc = 20 cm3). Fig. 1 refers to pnom = 500 bar, ET = 800 ls, whereas Fig. 2 refers to pnom = 1500 bar, ET = 600 ls. Injector feeding pipes with l = 200 mm and d = 3.0 mm as well as a pump-to rail duct with l1 = 300 mm and d1 = 3.0 mm were considered in the comparison. The impact of the accumulation volume on the system dynamics is shown through the pressure distributions in the rail (prail) and

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

485

Fig. 3. Natural period of the pressure waves. Fig. 1. Single injection: pnom = 500 bar, ET = 800 ls.

Fig. 2. Single injection: pnom = 1500 bar, ET = 600 ls.

in the junction (pjunct), which have been plotted with thin solid lines. The pressure drop detected at the end of injection in the accumulator (capacitance effect) significantly increased in the case of the Minirail (the difference is up to 35 bar in Fig. 1 and up to 55 bar in Fig. 2). From the continuity equation for the accumulator, following a lumped parameter approach, one obtains:

V acc dpacc ¼ Gpump  Ginj a2 dt

inlet

 GPCV

ð1Þ

where pacc is the mean pressure in the accumulator (equal to prail or pjunct, depending on the considered system), a is the liquid-flow sound speed, Vacc the accumulator volume, Gpump the mass flow-rate delivered by the pump, GPCV the mass flow-rate discharged through the PCV and Ginj inlet the mass flow-rate entering the injectors. Eq. (1) shows that the diminution rate of pacc, due to injection, increases as Vacc reduces. However, the remarkable disparities in the accumulation volume between the Minirail and standard CR resulted in reduced differences in the corresponding prail and pjunct drops. This was mainly due to the compensating action of the pressure control system. In fact, the duty cycle of the PCV changed with Vacc and so did GPCV(t), which remained lower in the case of the Minirail system. As a consequence, the injected flow rate time histories were very similar for the two pieces of apparatus. A smaller accumulator also induces a lower amplitude and a higher frequency of the free pressure oscillations that travel along the high-pressure circuit after nozzle closure; both these features are beneficial for a reduction in the pressure-wave induced distur-

bances on the injection system dynamics. This inertial effect of the accumulator is evident if one compares the injector inlet pressure time distributions (plotted with thick solid lines), which refer to the two distinct layouts. In fact, the energy stored in a wave train grows with both the amplitude and frequency of the waves [24], the value of this latter being closely related to the geometry of the high-pressure circuit. Since the energy stored in the pressure oscillations mainly depends on the values of ET and pnom [18], a layout modification that induces an increase in the wave frequency, without any appreciable change in viscous power dissipation, would lead to a diminution in their amplitude and vice versa. Fig. 3 reports the natural period (T) of the pressure fluctuations at pnom = 800 bar, Ttank = 40 °C for both the Minirail and the standard CR. The results refer to values of l that range from 100 to 275 mm and to two different values of d. The solid and dashed line distributions were obtained using a lumped parameter model of the high-pressure hydraulic circuit, which was developed and assessed in [18] and the symbols refer to experimental values plotted for comparison with the predictions. The most interesting outcome of the theoretical analysis is that, at fixed values of l and d, the natural period of the waves decreases when the accumulator volume is reduced, this also being consistent with what was previously observed in Figs. 1 and 2. Therefore, unexpectedly, the reduction in Vacc results to have a positive impact on the pressure oscillations.

4. Development of the new-generation CF system The previous experimental tests, together with the data reported in [20], showed that it is feasible to investigate the possibility of removing the rail from the CR layout. This would lead to a new-generation fuel injection system, that is, the CR without a rail. When this component is eliminated, a minimum accumulation volume DV (about one order of magnitude lower than the standard rail volume) should however be added to the high-pressure circuit, in order to avoid an excessive decrease in the pressure level during the injection event. Furthermore, the presence of such a volume DV, where the pressure can be efficiently monitored by a sensor, is required for stable working of the pressure control system. In principle, two locations are available for DV, either the pump or the injectors (Fig. 4). In the former case (Layout 1), a special pump delivery chamber featuring a volume of size DV should be designed, whereas in the latter case (Layout 2) a large chamber should be realized and located upstream from the injector nozzle. The effect that DV has on the system pressure drop that occurs during the injection (capacitance effect) is independent of the location of this volume in the high-pressure circuit. Instead, its impact on the injection dynamics (inertial effect) can change to a great extent, depending on whether DV is integrated in the pump or in the

486

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

Fig. 4. Possible layout solutions without the rail.

l, d

ΔV

φ

l, d Rail

PCV Pump

l1 d1

PCV

FMV

Fig. 5. Schematic of the new-generation CF system.

injectors. When a pressure wave triggered by the injection reaches a system component that can be regarded as a hydraulic-capacitance element, part of the incoming wave is transmitted downstream from the capacitance and part is reflected back (Fig. 4). As the volume of the capacitance element increases, the amplitude of the reflected wave becomes larger, whereas that of the transmitted wave becomes smaller. The disturbances induced on the nozzle dynamics by the DV-reflected pressure waves should be much lower for Layout 1 than for Layout 2. In the former solution both the incoming and the reflected waves are dampened by both the concentrated losses (restrictions) and the wall friction actions through the high-pressure circuit from the pump to the nozzle. In Layout 2 most of the unsteady waves travel forwards and backwards between the nozzle and the injector delivery chamber (DV in Fig. 4) and only a minor part of these waves is transmitted towards the pump. Although these pressure oscillations are characterized by a high frequency value (as already explained this has a positive effect on the wave amplitude), their impact on the injection dynamics can be relevant, due to the absence of any remarkable flow dissipation in the short path from the nozzle to the injector delivery chamber and back. The above considerations led to the definition of a new injection system, based on the Layout 1 solution. A modified version of the CR pump, featuring a specific delivery chamber with an enlarged volume was prototyped. The innovative pump was also equipped with a PCV and a pressure sensor, which were both placed in correspondence to the newly designed delivery chamber. An FMV was installed at the pump inlet in order to make the control of the highpressure level in the apparatus possible by applying either FMV- or PCV-based control strategies. A schematic of the new-generation Common Feeding (CF1) system is reported in Fig. 5.

FMV

Fig. 6. Schematic of the standard CR system.

to assess whether the removal of the rail would be critical in the presence of reduced leakages. Piezo-injectors are affected by small leakages, because the high-pressure force acting on the pilot valve tends to keep this latter closed [25]. When a pump delivery phase is in progress, some fuel is added to the high-pressure circuit, the mean pressure in this circuit tends to rapidly increase and a peak value is reached. The pressure rise is mitigated by both the static leakage at the injector pilot valve and the rail hydraulic capacitance: in principle, the larger the static leakage and the rail volume, the lower the pressure peak value should be. Therefore, the application of piezoelectric injectors to the CF system could be considered a difficult challenge to assess the robustness of the pressure control system. As far as the geometrical features of the high-pressure layout are concerned, injector feeding pipes of l = 200 mm and d = 3.0 mm were selected for both the systems, and a pump-to-rail duct of l1 = 275 mm and d1 = 3.0 mm was used in the case of the CR apparatus, all these sizes being typical of automobile applications. Initially, the CF system was not provided with a restrictor in the high-pressure circuit. On the other hand, calibrated orifices (/ = 1.7 mm) were present on the standard CR where the rail was connected to the injector inlet pipes, in order to damp pressure waves travelling along the high-pressure circuit. It still has to be verified whether the accumulation volume reduction, which is a an active damping strategy of the pressure oscillations, is capable, on its own, of compensating for the dissipative action of the orifices (passive damping) in the CR. 5.1. Single injection dynamics

5. CF vs. CR system: hydraulic-test results and discussion A comparison between the CF (Fig. 5) and CR (Fig. 6) systems was carried out on piezoelectric injectors. This choice was made 1 Catania and Ferrari ‘‘ Advanced High-Pressure Pump for Diesel Injection Systems’’, Patent Application No. TO2009A000715.

Fig. 7 reports the time-histories of the injected mass flow-rate (Ginj) from the EVI pressure profile and of the experimental pressure at the injector inlet (pinj inlet) as well as in the rail (prail), for the standard CR, or at the pump delivery chamber (ppump), for the CF, for a single injection event (pnom = 500 bar, ET = 200 ls, Minj  0.7 mg). Although the drop in the high-pressure level

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

487

Fig. 7. Single injection: pnom = 500 bar, ET = 200 ls.

Fig. 8. Single injection: pnom = 800 bar, ET = 600 ls.

subsequent to injection is significantly higher in the case of the CF system, the injected flow-rate time history remains virtually the same for both the pieces of apparatus. When injection occurs, a depression wave propagates from the nozzle along the injector feeding pipe and is then reflected, as a compression wave, at the first system component, which features an adequate hydraulic capacitance. This component is the rail or the pump delivery chamber for the CR and the CF systems, respectively. The injection dynamics starts to be affected by the high-pressure layout geometry from the instant the reflected compression wave reaches the nozzle. Thus, the minimum energizing time, which is required to make injector dynamics sensitive to the high-pressure layout geometry, is approximately given by 2l/a for both systems (in Figs. 5 and 6, l is the length of the injector feeding pipe). In the present analysis (l = 200 mm), such a threshold value ranges from 210 to 300 ls, depending on the values of pnom and of the mean fluid temperature in the high-pressure circuit. As a consequence, no appreciable changes were observed between the CF and CR injected flow-rates when low ETs, that are typical for pilot injections were applied. Fig. 8 plots the same variables as Fig. 7, but with reference to a medium injected quantity (Minj  21 mg, pnom = 800 bar and ET = 600 ls). After reaching a local minimum at t  1.1 ms, the injector inlet pressure was observed to rise. In fact, from this instant onwards the fuel amount ejected through the nozzle holes and at the pilot valve could no longer balance the flow-rate entering the injector from the feeding pipe as a result of the injection induced depression [3]. This determined the start of the pressure increase detectable in the injector inlet time distributions. The compression wave, which originated as a consequence of the illustrated events, propagated through the injector feeding pipe and reached the rail in the CR system as well as the pump delivery chamber in the CF system and mitigated the pressure drop due to injection in these accumulation volumes. Such a compensating action of the compression wave on the pressure drop was stronger for the smaller accumulation volume, that is, in the case of the pump delivery chamber (CF system), which, on the other hand, underwent the more intense depression event subsequent to injection. The described dynamics of the pressure transients reduced the difference between prail and ppump during the injection phase and thus could explain why the injected flow-rate time histories had almost the same values for the two apparatuses. The peak pressure value recorded at t  1.8 ms was due to the nozzle-closure-induced water hammer [21]. When large fuel injected quantities were considered, such as that in Fig. 9 (Minj  78 mg), slight differences were observed in the injected flow-rate between the CF and CR systems. The disparity in the hydraulic capacitance gave rise to significant differences

between prail(t) and ppump(t), which, however, only had a reduced impact on the time distributions of Ginj. The CF system was characterized by slightly lower values of the injected flow-rate during the 1.8 ms < t < 2.35 ms interval. In fact, unlike the cases at ET = 600 ls, the large pressure drop due to the reduced accumulation volume prevailed over any dynamic effects related to the unsteady pressure wave motion in the high-pressure circuit. The ppump(t) in Fig. 9 is an ‘‘average line’’ around which the injector inlet pressure fluctuates. From a mathematical point of view, ppump(t) is accurately approximated by the first low-frequency harmonic terms in the Fourier series of the corresponding injector inlet pressure time history. In the case of the CR system, prail(t) remains significantly higher than the ‘‘average line’’ of pinj inlet(t) during the entire injection phase. In fact, the calibrated orifice at the connection between the rail and the injector feeding pipe introduces a remarkable ‘‘static’’ pressure drop when the fuel is conveyed from the rail towards the nozzle, where it is then ejected. The presence of the calibrated orifice is also responsible for the fast damping of the free pressure oscillations after the end of injection as evidenced by all the pinj inlet time distributions reported in Figs. 8 and 9 for the CR system. 5.2. Injector characteristics and leakages Fig. 10 reports the injector characteristics, that is, the injected mass measured by means of the EMI as a function of ET for different pnom values, for both the CF (symbols) and the CR (solid lines). The tests refer to Ttank = 30 °C, but no important changes were observed in the measured quantities when the tank fuel temperature was varied in the range 30–60 °C.

Fig. 9. Single injection: pnom = 1500 bar, ET = 1000 ls.

488

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

90

MEMI [mg]

70 60 50 40

Symbols Symbols Solid Solid Lines Lines

500 bar 800 bar 1000 bar 1200 bar 1500 bar 1700 bar

CF System Innovative System Traditional System CR System

30

80 70

MEMI [mg]

80

90 500 bar 800 bar 1000 bar 1200 bar 1500 bar 1700 bar

60 50 40

500 bar 800 bar 1000 bar 1200 bar 1500 bar 1700 bar

500 bar 800 bar 1000 bar 1200 bar 1500 bar 1700 bar

Symbols FMV Solid Lines PCV

30

20

20

10

10

0 100 200 300 400 500 600 700 800 900 1000 1100

0 100 200 300 400 500 600 700 800 900 10001100

ET [μs]

ET [ μs]

Fig. 10. EMI injector characteristics.

40 35 30

V leak [mm 3 ]

The absolute differences in the EMI injected mass between the two systems were below 4 mg in all cases and the percentage differences, evaluated with respect to the CR quantities, were lower than 7% for higher injected masses than 10 mg (the percentage differences are not appropriate for lower injected masses than 10 mg due to the excessively small values of MEMI). In general, the EMI injected mass was slightly lower for the CF apparatus, especially at high pnom and large ET values. As already explained, at these working conditions, the pressure drop experienced by ppump was remarkably higher than that experienced by prail and this was the main reason for the difference in the MEMI quantities between the two pieces of apparatus. Instead, no appreciable differences were detected in [20] in any of the injector characteristics when the Minirail was compared to the CR. Nevertheless, the accumulation volume at the junction, in the Minirail, was comparable with that of the pump delivery chamber in the CF system. However, the Minirail had a 3.0 mm diameter pump-to-junction duct, which provided an extra accumulation volume to the injection apparatus compared to the CF system. The absence of this extra accumulation volume seemed to be the main responsible for the slight loss in fuel capability, at high pnom and large ET values, shown by the CF system compared to the Minirail. A threshold value exists for the hydraulic capacitance of the high-pressure circuit of the injection system, below which the general fuelling capability of the apparatus decreases. In general, the smaller the hydraulic capacitance of the high-pressure circuit with respect to this threshold value, the stronger the loss in fuelling capability of the apparatus and the larger the pnom/ET range in which such a loss can be observed. When the high-pressure circuit accumulation volume becomes too small, instability can occur in the control of the high-pressure level. The injector characteristics in Fig. 10 were obtained by controlling the high-pressure level in the systems using the PCV. It was interesting to assess, for the CF system, whether the adopted pressure control strategy could have an impact on the EMI injected masses. Fig. 11 shows a comparison of the injector characteristics obtained under PCV- and FMV-based control strategies. As can be inferred, the differences are negligible (lower than 1 mg), a result which is in line with what is usually observed in conventional CR systems. Fig. 12 shows a plot of the comparison of the injector leakages, which are reported as volumes (Vleak) vs. ET at different pnom values. The injector leakage is the sum of the static and the dynamic leakages. The former is the leakage that occurs through the pilot valve when this is closed, the latter is the leakage through the pilot valve when this is open by the piezo-actuator. Therefore, the static leakage is independent of ET, whereas the dynamic leakage is dependent on ET. In the graph, the dynamic leakage at each fixed pnom results to linearly increase with ET for both injection systems.

Fig. 11. Injector characteristics: PCV vs. FMV.

500 bar 800 bar 1000 bar 1200 bar 1500 bar

500 bar 800 bar 1000 bar 1200 bar 1500 bar

25 20 15 10 5

Symbols Symbols Solid Solid Lines Lines

Innovative System CF System Traditional System CR System

0 100 200 300 400 500 600 700 800 900 1000 1100

ET [ μs] Fig. 12. KMM injector leakages.

Instead, the static leakage is given by the intersection point of each distribution with the ordinate axis. The distribution marked with symbols (CF) for each considered pnom seems to be translated towards higher values than the corresponding one plotted with a solid line (CR). This means that the dynamic leakage is almost the same for the two systems, while the static leakage is rather higher for the CF apparatus. The absolute differences in the leakage are below 4.5 mm3 in all cases and the percentage differences range from 90% at pnom = 500 bar, ET = 200 ls to 10% at pnom = 1500 bar, ET = 1000 ls. The higher static leakage experienced by the CF is explained in what follows. Fig. 13 reports ppump (CF) and prail (CR)

Fig. 13. Comparison between ppump (CF) and prail (CR) at pnom = 1000 bar, ET = 600 ls.

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

as functions of the time for a complete engine cycle at pnom=1000 bar and ET=600 ls. The time average value of prail along the engine rail , is higher than p pump (p rail  950 bar, p pump  cycle, namely p 930 bar). Furthermore, the static leakage can be expressed as a stationary volumetric flow-rate according to the following Hagen–Pouseille type formula [26]:

 p V_ leak  K leak l

ð2Þ

) is a geometrical term that depends on the where Kleak = Kleak(p  is the  ¼ lðp ; TÞ clearances between the pilot valve and the seat, l  and T are the time averaged values of fuel dynamic viscosity, p the pressure and temperature in the high-pressure circuit ¼p pump and p ¼p rail for the CF and the CR, respectively). Finally, (p temperature T was verified to be much higher (up to some tens of  is sensitive to T to degrees) in the case of the CF system. Since l a great extent (and decreases with T), the static leakage was demonstrated to increase for the CF. In conclusion, a thermal effect seemed to be the main cause of the differences between the two injection systems, shown in Fig. 12. The higher injector leakage of the CF system had no influence on the on-engine behavior because the full load fuel request continued to be satisfied. This meant that the displacement and the transmission ratio adopted for the CF pumping unit, which were the same as those of the standard CR pump, were able to guarantee the extra fuel required in order to compensate for the higher injector leakage. Furthermore, the increased injector leakage of the CF system was verified to not have any appreciable effect on the engine brake specific fuel consumption (see bsfc data in subsection 6.1). 5.3. Cycle-to-cycle dispersion and minimum injectable quantity The graphs in Fig. 14 report the standard percentage deviation vs. the average injected mass measured over 100 cycles by the EMI device for the CF system. The standard deviation was related

to the cycle-to-cycle dispersion in the injection system performance. The data referred to the PCV controlled system and to different pnom values. The dispersion was experimentally verified to be within the limits prescribed by the injector manufacturer for the different pnom and injected quantity values. These limits are specified in the charts and can be checked by interpolating and extrapolating the injected masses in each diagram. The standard deviation at each pnom is shown to generally increase as the injected quantity reduces, reaching higher values than 3% for smaller injected masses than 2 mg. This is generally in agreement with what is usually observed in CR systems. One drawback of the CR system, and, in this case, of the CF, in fact involves the accurate control of tiny injected masses. As the nozzle opens, the fuel is injected at a high-pressure level, which is related to pnom, and thus it is difficult to accurately dose minute injected quantities. Furthermore, the needle is ballistic at short ET values, hence a noteworthy cycle-to-cycle dispersion is likely to occur in the injection temporal length, and this can have a significant impact on the repeatability of the injected quantity. All this is responsible for the trend observed in Fig. 14 at the lowest injected quantities. The minimum injected mass for which the standard deviation measured by EMI over 100 cycles is lower than 10% can be defined as the minimum injectable quantity. This mass was almost analogous for the two pieces of apparatus at the different pnom. The minimum injectable mass data did not show any definite trend with respect to pnom and varied in the 0.3–0.4 mg range for both systems (the measurements were controlled with an analytical balance providing very high accuracy). The value of the energizing time, for which the minimum injectable mass was obtained, decreased from 165 ls to 120 ls when pnom was varied from 400 bar to 1700 bar, for both systems. 5.4. Multiple injections Figs. 15 and 16 plot the time histories of prail(t), ppump(t), pinj inand Ginj(t) for the pilot-main injection events at the different pnom, ETpil, ETmain values which are quoted in the figure captions. Each figure has two graphs, namely (a) and (b), which refer to distinct values of the dwell time (DT). The pilot injected flow-rate time history in each graph maintains the same pattern when passing from the CF to the CR system, which is consistent with what occurred in the Section 5.1 (ETpil is lower or equal to 235 ls), whereas differences can occur in the main injected flow-rate time histories. These can be noted in Fig. 16, where the maximum value of Ginj(t) is higher for the CF system. The pressure drop induced by the double injection on ppump(t), in all the examined cases, is larger than the corresponding one induced on prail(t). However, as soon as the main current signal is switched on, prail(t) and ppump(t) exhibit values that are still very close to each other. The possible differences in the main injected flow-rate peak values between the two systems are determined by the nozzle pressure values during the main injection. These, in turn, depend on the pressure transients that are triggered by the pilot-injection induced water-hammer in the two systems and thus on the selected set of injection parameters (pnom, ETpil, ETmain, DT). In Fig. 15, the injector inlet pressure time distributions, which are expected to be almost phased to the nozzlepressure histories, are roughly the same for both systems during the main injected flow-rate growth. On the other hand, pinj inlet(t) in Fig. 16b remains slightly higher for the CF apparatus during the 2.6 ms < t < 2.8 ms time interval. All this is physically consistent with the trends shown by the corresponding flow-rates. In general, when pilot-main injections are performed, the system highpressure at the beginning of the main injection is still close to the nominal value, due to the small amount of fuel injected in the pilot let(t)

Limit: 1.48% at 10 mg

Limit: 0.85% at 15 mg

Limit: 8.33% at 1.6 mg

Fig. 14. Cycle-to-cycle dispersion at different pnom.

489

490

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

(a) - DT = 600 μs

(a) - DT = 800 μs

(b) - DT = 1000 μs

(b) - DT = 1000 μs

Fig. 15. Pilot-main injection: pnom = 500 bar, ETpil = 235 ls, ETmain = 536 ls.

Fig. 16. Pilot-main injection: pnom = 1000 bar, ETpil = 155 ls, ETmain = 505 ls.

shots, regardless of the accumulation volume. As a consequence, the main injected flow-rate time history is affected more by the accumulator inertial effect (pressure wave dynamics) than by its capacitance effect (size of the volume). The results and the conclusions pertaining to Figs. 15 and 16 are in agreement with those reported in [20], where both pilot-main and main-post injection events were analyzed for the Minirail system equipped with an enhanced ECU (see also Section 6). The amplitude of the free pressure waves that occurred after nozzle closure is larger for the CF system, due to the absence of any purposely designed damping restrictor in its high-pressure circuit. The pressure oscillations at the end of the pilot shot are responsible for the variations in the main injected volume as the DT between the pilot and the main injection is changed [3]. Fig. 17 shows the percentage deviation r of the measured main injected fuel-volume from its average value, in the 300–3500 ls DT range, for both the CF (a, b) and the CR (c, d) systems, with reference to a double injection (Vpil = 2 mm3, Vmain = 30 mm3) at pnom = 500 bar (a, c) and pnom = 1000 bar (b, d). The ordinate r is defined as:

where DTmin = 300 ls and DTmax = 3500 ls. Fig. 17 shows the r distributions for different d values (2.4, 3.0 and 3.5 mm): the frequency of the r fluctuations for d = 3.0 mm coincides with that of the corresponding pressure waves illustrated in Figs. 15 and 16. In general, the r fluctuations exhibited a stronger amplitude for the CF system, in a consistent way to what is shown in Figs. 15 and 16 for the pressure waves. The maximum absolute values of r were comparable for the two pieces of apparatus operating under the same pnom. However, the r oscillations were dampened to a great extent as DT increased in the CR apparatus, due to the action of the calibrated orifices on the pressure waves, at the connections between the rail and the injector feeding pipes, whereas they appeared to be only slightly dampened in the CF system. In this latter case, the viscous dissipation at the injector restrictions and at the boundary walls was the only mechanism that was involved in damping the pressure waves.

 main V ðDTÞ  V r ¼ main   100 V main

ð3Þ

where Vmain (DT) is the main injected volume for a given DT and  main is its average value, that is: V

 main ¼ V

1 DT max  DT min

Z

DT max

DT min

V main dt

ð4Þ

6. Upgrade of the CF system and tests on the engine On the basis of the results of the previous section, an important improvement was made on the design of the CF system which involved the introduction of a restrictor in the high-pressure circuit in order to damp pressure waves. Fig. 17 shows that the dynamic effect, due to the accumulation volume reduction, is not capable on its own of compensating for the dissipative action of the calibrated orifices in the CR system. With regard to the effect of DT on multiple injections, Figs. 18 reports another comparison between the CR apparatus and the enhanced version of the CF system

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

Fig. 17. Injected volume fluctuations with respect to DT: CF vs. CR (l = 200 mm; d = 2.4 mm, d = 3.0 mm and d = 3.5 mm).

Fig. 18. CF system with the pump integrated orifice (l = 200 mm, d = 3.0 mm).

491

492

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

with a pump integrated orifice. A damping restrictor (/ = 1.8 mm) was placed at the pump delivery chamber inlet. As can be inferred, the main injected volume fluctuations diminished for the CF injection system as DT increased. In the figures, it can be observed that the maximum absolute value of r (|rmax|) was higher for the CR, even though the calibrated orifices applied to this system had smaller diameters than the one used in the CF (/ = 1.7 mm instead of / = 1.8 mm). This apparent inconsistency can be explained by the already mentioned dynamic effect of the reduced accumulation volume on the modulation of the pressure wave kinetic energy. The introduction of a damping restrictor in the CF system has expected to reduce the system fuelling capability. Fig. 19 reports injector characteristics of the CF apparatus at some pnom values in the absence (solid line) and in the presence (symbols) of the orifice. The discrepancies in the injected masses between the two setups are appreciable, but not remarkable: the absolute differences are below 2.3 mg in all cases and the percentage differences lower than 5% for higher injected masses than 10 mg. It has also been verified that the differences in the injected masses between the CF system with the orifice and the CR system were below 6.3 mg in all cases and lower than 9% for higher injected masses than 10 mg. The reduced fuelling capability of the injection system, with respect to the CR is not a problem, provided the full-load fuel request continues to be satisfied. If this is the case, only a recalibration of the engine maps is required. However, it is necessary to assess whether the modified calibration deteriorates engine out pollutant emissions at part load. A possible solution to limit the remarkable pressure drop, detected in the CF as a consequence of injection, consists of the application of look ahead strategies in the control of ppump(t). The pressure at the start of injection is kept higher than pnom by the ECU, so that the time-average of ppump(t) during the whole injection event can be made equal to the desired value. This control approach could also be helpful when post injections are considered in the injection schedule. In the case of the main-post injections, the pressure drop in the high pressure circuit, induced by the main injection, can have a significant influence on the post injection for the CF system. In fact, an excessive difference can occur between the actual and nominal pressure levels after the end of the main injection (see in Figs. 2, 9 and 16), where post injections are feasible. The look-ahead strategy has already been applied successfully to the Minirail system in [20] and its implementation has resulted in efficient main-post injections, with a contained gap between the actual and nominal pressure levels at the end of the main injection shot. However, in the present investigation, the ECU was still at a preliminary development stage and

80 70

MEMI [mg]

60 50

500 bar 1000 bar 1500 bar

500 bar 1000 bar 1500 bar

Symbols With orifice Solid Lines Without orifice

40 30 20 10 0 100 200 300 400 500 600 700 800 900 1000 1100

ET [μs] Fig. 19. CF system: effect of the pump integrated orifice on the injected masses.

was therefore not capable of supporting such a sophisticated control strategy. 6.1. Pollutant emission performance The upgraded CF system was tested in the dynamometer cell on a Turbo Euro 5 engine (iV = 1950 cm3, Pu = 190 CV) and its performance was compared with that obtained using the standard CR. The experimental analysis was carried out on the piezoelectric injectors and the following key points were selected to characterize the engine operation at part load: N = 1500 rpm and bmep = 2 bar (1500  2), N = 2000 rpm and bmep = 5 bar (2000  5), N = 2500 rpm and bmep = 8 bar (2500  8). The same ECU calibration set was adopted to perform the tests on the two injection systems. In fact, the previous hydraulic characterization at the Moehwald Bosch test rig showed only minor differences in the injection performance at part load between the two injection systems. However, since the ECU maps were optimized for the engine configuration with the standard CR, the pollutant emission comparison was unfavorable for the CF system and thus constituted a tricky test for the latter system. Figs. 20 and 21 compare the Soot-NOx trade-off (a) and the corresponding bsfc–k curve (b) of the two injection systems, at 2500  8 and 2000  5, respectively. Each operative point of the Soot-NOx trade-off, and thus of the bsfc–k curve, was obtained varying only the EGR rate and keeping all the other working parameters constant. EGR is minimum at the symbol furthest on the right in each diagram and increases as NOx or k diminishes. The plotted quantities are dimensionless, as they (apart from k) are normalized with respect to the value of the considered variable obtained at the nominal EGR level in the case of the standard CR. The engine configuration with the CF shows better Soot-NOx trade-off and bsfc values in both the figures. In particular, the discrepancies in the Soot level between the two engine setups were significant, whereas those on bsfc were only minor (the maximum percentage difference was around 1.5%). The comprehensive analysis of the injector parameters pointed out the presence of shorter ETmain values for the CF system at all the operative points of the trade-offs. Due to the different dynamics of the pilot-injection triggered pressure waves in the two pieces of apparatus (see the Section 5.4), at the considered working conditions the CF was capable of injecting the same fuel amount as the CR, but for shorter ETmain values. As a consequence, since SOImain was the same for both systems, MFB50 would be slightly advanced in the CF system and this fact alone would give rise to a better bsfc [27]. This improvement in bsfc then led to a reduction in the fuel request (k increased slightly in the figures for the CF system) and thus to a further diminution of ETmain and a subsequent additional advance of MFB50. The heatrate analysis based on the in-cylinder experimental pressure trace [28] confirmed the presence of differences in MFB50 (up to 1.8 deg at 2000  5 and up to 1 deg at 2500  8) between the two systems. The reduced injection temporal length, caused by the shorter ETmain, also improved the interaction between the injected fuel and cylinder charge close to the injector tip during the last phase of injection, which in turn led to lower soot emissions. Furthermore, the increased k and the advanced MFB50 were the main reasons for the increase in the NOx emissions shown by the CF in Figs. 20 and 21. Fig. 22 plots the CO–NOx trade-off (a) and the corresponding bsfc–k curve (b) for the CF and the CR systems, with reference to key point 1500  2. At such a low-load engine condition, the soot levels were minute (FSN was lower than 0.4) and thus a comparison was more significant concerning the CO emissions. In fact, due to the large k values and to the reduced in-cylinder temperatures, the CO emission increased considerably. The CO–NOx trade off was shown to be virtually the same for the two engine setups. How-

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

493

Fig. 20. Emission performance at N = 2500 rpm, bmep = 8 bar.

Fig. 21. Emission performance at N = 2000 rpm, bmep = 5 bar.

ever, unlike what appeared in Figs. 20 and 21, the bsfc deteriorated in the CF (up to 3%). The explanation for this behavior is the same as that provided in Figs. 20b and 21b, but in this key point the dynamics of the pressure waves was more favorable for the CR, which could generally inject the same quantity at lower ETmain values. In conclusion, the emission performance of the two injection systems seems to be almost comparable when the engine is run at a steady point: the possible differences can vary according the considered working condition because they depend on the pressure-wave motion in each injection apparatus, which in turn also depends on the selected DT, ETpil, prail and SOImain values. Finally, in order to evaluate the emission performance on the engine transients, the Euro 5 engine equipped with the CF system was run according to the New European Driving Cycle (NEDC) load-speed profile. The NEDC consists of four repeated ECE-15 driving cycles and an Extra Urban Driving Cycle or EUDC [29]. Table 1 reports the cumulate emission values (g/km or mg/km) of the pollutants, which were measured downstream from the diesel oxygen catalyst at the end of each of the five main phases of the hot NEDC, as well as the fuel consumption; the test was conducted with no diesel particulate filter at the engine outlet. The results show considerably higher CO for the first ECE15 repetition as well as significant differences in NOx and soot emissions for the first ECE-15 compared to the levels obtained in the fourth repetition. In fact, after the engine had been warmed up (cooling-water temperature close to 95 °C at engine out), it was stopped and the automatic procedure required about 3 minutes in order to purge

and pre-sample the measuring instruments before the NEDC could be run. This soaking time and the absence of the glow-plug (the calibration was not available at this preliminary development stage) determined a temperature decrease during the first ECE15 that was the cause of the trend observed in the urban emissions. The data on pollutants, referring to the whole hot NEDC (penultimate row in Table 1), should be compared with those reported in the last row, as they represent the target values of the Euro 5 standard for diesel passenger cars [30]. In order to take into account both the aging effect and engine production dispersion, the emission values measured for the whole NEDC should be multiplied by a corrective factor (in the present case the value 1.15 was considered), before being compared with the limits of the Euro 5 standard. As can easily be verified, the thus obtained CO and THC emissions resulted to be much smaller than the corresponding Euro 5 limit values, because the cycle was executed after the engine had warmed up. Furthermore, a final quantity of 165.61.15  190.4 mg/km, which was 6% higher than the limit of 180 mg/km, was achieved for the NOx. Lastly, the Soot level resulted to be much higher (23.81.1527.4 mg/km) than the target value (5 mg/km). Since the efficiency of the DPF on the NEDC is usually at least 90%, a value of about 2.7 mg/km could, in principle, be achieved downstream from the DPF. All these results are particularly promising as far as the Euro 5 standard is concerned, and a sensible improvement can be expected if a calibration set that has been optimized for the CF apparatus is used.

494

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495

similar to that of the standard CR. The multiple injection performance was improved with respect to the CR, because of the reduced variation in the fuel injected volume as DT varied. The smaller accumulation volume of the CF system induced a lower amplitude and a higher frequency of the free pressure oscillations after nozzle closure. Both these features were beneficial in reducing the disturbances induced by pressure-waves on the multiple injections (the phenomenon is here referred to as accumulator inertial effect). In addition to this effect, a restrictor was integrated in the high-pressure pump circuit to damp the pressure waves and it played the same role as the calibrated orifices that are usually installed in the CR layout where the rail is connected to the injector feeding pipes. Tests on a Euro 5 engine were performed in the dynamometer cell to assess the performance of the CF system concerning emissions. Soot-NOx and CO–NOx trade-offs were made at different working conditions. The results have shown that the emission performance and the brake specific fuel consumption are comparable for the two engine configurations applying the CF and the CR systems. The small discrepancies that were detected were ascribed to minor differences in the main injection flow-rate pattern, which were induced by the characteristic dynamics of the pressure waves in the high-pressure circuit of each injection apparatus. Finally, a hot NEDC was performed on the engine equipped with the CF system. The pollutant emissions that were measured at the end of the cycle were promising to cope with the Euro 5 target values. In general, the results pertaining to the emissions at the key points and on the hot NEDC, which were obtained for the CF, are appealing because they were achieved using the available ECU calibration set, which had been optimized for the engine setup applying the standard CR. A future development of the present work could consist in comparing emission and fuel consumption results obtained on the NEDC for the standard and the new injection systems, each of them being optimized with a tailored engine calibration strategy. The optimal values of the calibration variables for the CF apparatus (ETs, DTs, pnom, SOI, EGR rate, boost pressure, swirl ratio, etc.) could be determined through a design-of-experiment procedure.

(a) - CO-NOx trade-off

(b) - bsfc-λ curve Fig. 22. Emission performance at N = 1500 rpm, bmep = 2 bar.

Table 1 Emissions along the hot NEDC. –

CO2 (g/ km)

CO (mg/ km)

THC (mg/ km)

NOx (mg/ km)

THC + NOx (mg/km)

SOOT (mg/ km)

FC (l)

ECE1 ECE2 ECE3 ECE4 EUDC NEDC Target

16.2 16.5 16.3 16.3 83.8 149.1 –

25.8 0.7 0.7 0.6 2.5 30.3 500

3.3 3.4 3.1 3.2 5.4 18.4 –

14.5 14.0 15.1 16.1 105.9 165.6 180

17.8 17.4 18.2 19.3 111.3 184 230

2.6 2.0 1.8 1.7 15.7 23.8 5

0.06 0.06 0.06 0.06 0.32 0.56 0.6

References

7. Conclusions The main feature of the innovative CF apparatus is the absence of the rail in the hydraulic layout. A small hydraulic capacitance was integrated in the pump housing as the delivery chamber was enlarged. The pump was connected directly to the injector feeding pipes through screwed ports, which were machined on its housing, and it was possible to control the high-pressure level in the system by either the PCV or the FMV. The CF injection system had low hydraulic inertia, which gave rise to a fast dynamic response to the transients, and reduced production costs. Moreover, this system matched the requirements of easy installation on the engine. The cycle-to-cycle dispersion of the CF system was verified to be within the acceptable tolerances for the injectors at different working conditions. Furthermore, the minimum controllable quantity that could be injected by the new-generation apparatus was

[1] Suh HK. Investigations of multiple injection strategies for the improvement of combustion and exhaust emissions characteristics in a low compression ratio (CR) engine. Appl Energy 2011;88(12):5013–9. [2] Chiodi M, Mack O, Bargende M, Paule K, Brandt von Fackh J, Wichelhaus D. Improvement of a high-performance diesel-engine by means of investigation on different injection strategies. SAE paper no. 2009-24-0008. [3] Catania AE, Ferrari A, Spessa E. Numerical–experimental study and solutions to reduce the dwell time threshold for fusion-free consecutive injections in a multijet solenoid-type C.R. system, ICE best paper award. J Eng Gas Turb Power 2009;131(1). 022804-1–022804-14. [4] Ghaffarpour M, Noorpoor AR. NOx reduction in diesel engines using rate shaping and pilot injection. Int J Automotive Technol Manage 2007;7(1):15. [5] Payri R, Salvador FJ, Gimeno J, De la Morena J. Influence of injector technology on injection and combustion development – Part 1: Hydraulic characterization. Appl Energy 2011;88:1068–74. [6] Shayler PJ, Brooks TD, Pugh GJ, Gambrill R. The influence of pilot and split-main injection parameters on diesel emissions and fuel consumption. SAE paper no. 2005-01-0365; 2005. [7] Atzler F, Kastner O, Rotondi R, Weigand A. Multiple injection and rate shaping. Part 1: Emissions reduction in passenger car diesel engines. SAE paper no. 2009-24-0004; 2009. [8] Tanabe K, Kohketsu S, Nakayama N. Effect of fuel injection rate control on reduction of emissions and fuel consumption in a heavy duty DI diesel engine. SAE paper no. 2005-01-0907; 2005. [9] Ismailov M, Blaisot JB, Dementhon JB. A novel diesel injection nozzle for future HCCI engines. SAE paper no. 2008-01-0940; 2008. [10] No YS. Inedible vegetable oils and their derivatives for alternative diesel fuels in CI engines: a review. Renew Sustain Energy Rev 2011;15:131–49. [11] Kim H, Choi B. The effect of biodiesel and bioethanol blended diesel fuel on nanoparticles and exhaust emissions from CRDI diesel engine. Renew Energy 2010;35:157–63.

A.E. Catania, A. Ferrari / Applied Energy 91 (2012) 483–495 [12] Natu M, Bhise V, Shulze R. Development of specifications for the UM-D’s low mass vehicle for China, India and United States. SAE paper no. 2005-01-1027; 2005. [13] Fan L, Long W, Zhu Y, Xue Y. A characteristic study of electronic in-line pump system for diesel engines. SAE paper no. 2008-01-0943; 2008. [14] Catania AE, Ferrari A, Manno M. Parametric analysis of layout effects on common rail multiple-injections. In: ASME ICED fall technical conference. Paper no. ICEF2005-1288; 2005. [15] R. Bosch, GmbH. Diesel engine management. Technical handbook, 4th ed. John Wiley & Sons; 2005. [16] Xiao W, Liang F, Tan W, Mao X, Yang L, Zhuo B. Analysis of common rail pressure build-up and assistant-establishment of engine phase position in starting process. SAE paper no. 2006-01-3525; 2006. [17] Catania AE, Ferrari A. Experimental analysis, modeling and control of volumetric radial-piston Pumps. ASME Transactions. J Fluids Eng 2011;133(8):081103. [18] Baratta M, Catania AE, Ferrari A. Hydraulic circuit design rules to eliminate the dependence of the injected fuel on dwell time in multijet CR systems. J Fluids Eng 2008;130. 121104-1–121104-13. [19] Spinnler F, Zanetti C. High-pressure injection system. Patent application no. US 2002/0017278A1; 2002. [20] Catania AE, Ferrari A, Mittica A, Spessa E. Common rail without accumulator: development, theoretical–experimental analysis and performance enhancement at DI-HCCI level of a new generation FIS. SAE paper no. 200701-1258; 2007.

495

[21] Catania AE, Ferrari A, Manno M, Spessa E. Experimental investigation of dynamic effects on multiple injection CR system performance. J Eng Gas Turb Power 2008;130(3):032806. [22] Catania AE, d’Ambrosio S, Finesso R, Spessa E. Effects of rail pressure, pilot scheduling and EGR rate on combustion and emissions in conventional and PCCI diesel engines. SAE J Eng 2010;3:773–87. [23] R. Bosch, GmbH. The Fuel rate indicator: a new measuring instrument for display of the characteristics of individual injection. SAE paper 660749; 1966. [24] Landau LD, Lifshitz EM. Course of theoretical physics. ButterworthHeinemann; 1987. [25] Arpaia A, Catania AE, Ferrari A, Spessa E. Development and application of an advanced numerical model CR piezo indirect acting injection systems. SAE paper no. 2010-01-1503; 2010. [26] Batchelor GK. An introduction to fluid dynamics. New-York: Cambridge University Press; 2009. [27] Catania AE, d’Ambrosio S, Ferrari A, Finesso R, Spessa E, Avolio G, et al. Experimental analysis of combustion processes and emissions in a 2.0L multicylinder diesel engine featuring a new generation piezo-driven injector. SAE paper no. 2009-24-0040; 2009. [28] Heywood JB. Internal combustion engine fundamentals. McGram-Hill; 1988. [29] R. Bosch, GmbH. Emissions-control technology for gasoline engines. Automotive technology handbook. Bosch Yellow Jackets; 2003. [30] Regulation (EC) No. 725/2007 of the European parliament and of the council of 20 June 2007 on emission standards and on board diagnostics.