Study of the compression ratio influence on the performance of an advanced automotive diesel engine operating in conventional and PCCI combustion mode

Transcription

1 THIESEL Conference on Thermo- and Fluid Dynamic Processes in Diesel Engines Study of the compression ratio influence on the performance of an advanced automotive diesel engine operating in conventional and PCCI combustion mode C. Beatrice, C. Guido, N. Del Giacomo and C. Bertoli Istituto Motori Consiglio Nazionale delle Ricerche (CNR), Viale Marconi, 15 Napoli, Italy. Telephone: +(39) Fax: +(39) Abstract. The present paper, carried out in a collaborative project between Istituto Motori and Centro Ricerche Fiat, describes a detailed experimental analysis on the effect of the compression ratio on the performance of a diesel engine operating both with an EURO (conventional combustion) and with an advanced low NOx, approaching the estimated EURO emission limits (PCCI combustion). The employed engine was the four-cylinder FIAT 1.9 Liter 1 Valve MultiJet. Starting from a reference engine configuration with a compression ratio of 1.5, the compression ratio was reduced in two steps to 15.5 and 1.5 respectively. Each compression ratio value was characterized in terms of thermodynamic parameters, emissions and fuel consumption in some operating test points representative of the engine behaviour running on the NEDC cycle. The results allowed the definition of the optimum range values of the compression ratio able to minimize the NOxparticulate trade-off coupled with acceptable increment in unburned compound emissions (HCs+CO) and specific fuel consumption, helping to the definition of the guide-lines for the future EURO engine design. Notation AR Aspect ratio. BSFC Brake specific fuel consumption. C.A. Crank angle. CR Compression Ratio. DOC Diesel Oxidation Catalyst. dp/dt max Maximum gradient value versus time of the cylinder pressure. dp/dθ max Maximum gradient value versus crank angle of the cylinder pressure. DPF Diesel Particulate Filter. EGR Exhaust Gas Recirculation. FC Fuel consumption. I z Moment of inertia with respect Z axis. K-factor K-factor. LD Light Duty Engine. MBF5% Angular position corresponding to the 5% of Burned Fuel Mass, with respect to the TDC. M z Momentum with respect Z axis. NEDC New European Driving Cycle PAH Polycyclic aromatic hydrocarbons. PCCI Premixed Combustion Compression Ignition. P rail Pressure in the common rail. Q pil Pilot injection quantity. RR Reentrat ratio. TDC Top Dead Center. VCR Variable Compression ratio.

2 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo SOI Start Of Injection timing. SOI main Start of main injection under conventional combustion and start of single shot injection with PCCI operating mode. SR Swirl Ratio. St% Percentage of the opening position of swirl control flap in the intake duct. T exh Engine exhaust temperature upstream turbine. τ id Ignition delay time. VGT Variable Geometry Turbine. VGT% Percentage of the opening position of VGT. 1. Introduction A large field of the internal combustion engine research is today dedicated to the investigation on the new concept of Premixed Combustion Compression Ignition (PCCI) operation mode. Despite the interesting advantages offered by PCCI condition in terms of very low NOx and soot emissions, several problems still remain for its large application to the modern diesel engines. At the moment, one of the main issues, is the limitation of its use in the low and medium load/speed range of the whole engine operating map, due to the difficulties in controlling contemporary both the premixed air-fuel charge and the ignition timing. The extension to higher speed and load ranges, in fact, is coupled with problems of combustion noise control and excessive fuel consumption increment (Beatrice et al., 7). At the same time, of course, it is essential to enlarge the PCCI applicability in the whole NEDC cycle, to take advantage of its well known potentiality in reaching the very low NOx emissions in view of the future EURO regulation. However even if a further development in PCCI combustion does not guarantee the fulfilment EURO limit, requiring a foreseeable application of complex after-treatment systems as DeNOx plus DPF, PCCI use will help to reduce the load factor on the exhaust catalytic systems, reducing the fuel consumption increment required for system regeneration. Many engine operating and geometrical parameters have a strong influence on both the quality of the premixed charge and ignition timing and as a consequence on combustion characteristics of PCCI in the diesel engines (Beatrice et al., 7). Some operating parameters are quickly and precisely controlled by the adoption of technologically advanced accessories as swirl throttle valves, sophisticated turbo and EGR systems, variable valve actuation system and so on. However, among all, the most effective parameter in controlling the ignition delay time, by means of the controlled pressure and temperature values at TDC, remains the compression ratio (CR) (Wagner et al., 3; Kitabatake et al., 7, Hamada et al., 5). A lot of technological solutions for the realization of the real variable compression engine (VCR) are under development, but up to now they are still in the prototype version; an example can be found in David at al. (). Therefore, in order to extend the PCCI applicability to the modern light duty engines to the greatest possible engine operating area, the compression ratio has to be chosen carefully. In fact, a too low CR value, if permits a long ignition delay period and so high premixing air-fuel level (Hamada et al., 5), it suffers for cold engine startability and reduced engine efficiency, mainly at high load. On the contrary, an high CR gives no problems in cold starting and permits good performance at high load but limits the premixing air-fuel time and the EGR tolerance thus reducing the benefits of PCCI combustion. To this aim, an extensive research program was planned between Istituto Motori and Centro Ricerche Fiat. The experimental program lasted two years including both simulation and experimental activities. At the end of the activities a complete database on the performance of the FIAT LD diesel engine versus compression ratio value was carried out. Some of these results were already presented in a previous paper (Imarisio et al., 7). The present paper describes a part of the research activities performed within the project and in particular the results collected testing three different compression ratio values in a reference four-cylinder engine running in conventional diesel and PCCI combustion mode. The explored CR values were chosen in cooperation with Centro Ricerche Fiat on the basis of a preliminary analysis in a range of possible real solutions for the next generation of LD engines, taking into account also the current worldwide trends in diesel engine CR reduction. In the following, the methodology used for the design of the piston geometries, the engine test bench characteristics, the adopted testing methodology and the results analysis are described in detail. The effect of compression ratio on LD diesel engines was explored in various previous experimental works; some of them are reported in literature (Hamada et al., 5, Kimura et al., 1, Miyamoto et al., 7, Walter and Gatellier, 3 Pedersen and Schramm, 7, Araki et al., 5, Laguitton et al., ) and regard studies mainly performed on single-cylinder research engines. However a critical approach to the exploitation of low CR values for emission reduction on both conventional and PCCI combustion is not present in literature until now. Following this aim, the peculiarity of the present paper is the analysis of the effect of CR reduction on the per-

3 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 3 formance of a four-cylinder engine, at the state of the art of the current technology, and with the imposition of realistic limits in smoke emission, combustion noise and fuel consumption increment with respect to the current standards. This testing methodology will be carefully described in a dedicated section.. Reference four-cylinder engine The engine selected for the research is the FIAT 1.9 MultiJet. Its displacement represent the most popular size of diesel engine for passenger car in the European market. The main characteristics of the engine in its EURO version are listed in Table 1. Table 1. FIAT 1.9 MultiJet specifications Engine type In-line cylinders Bore [mm]. Stroke [mm] 9. Compression Ratio 17.5 Displacement [cm3] 191 Valves per cylinder Injection system Bosch Common Rail nd generation (1 bar of maximum injection pressure) Injector Bosch CRIP., Centred 7 holes microsac Turbocharger Single stage VGT, model Garrett GT179MV Swirl control Throttle valve on intake duct electronically controlled Catalyst system Closed-couple DOC plus underfloor DOC Maximum Power 11 rpm Maximum Torque 3 rpm The engine is fully instrumented for indicated signal measurements in each cylinder (cylinder pressure, injection pressure, energizing injector current). At the engine exhaust, smoke was measured by a high-resolution (.1 FSN) smoke meter (AVL15S), while gaseous emissions were measured upstream and downstream the oxidation catalyst by means of an analysis test bench (AVL-CEB-). An overview of the engine layout scheme with the main sensors positions is displayed in Fig. 1. Fig. 1. FIAT 1.9 MultiJet engine layout. The fuel employed for the tests is a commercial low sulphur diesel fuel, sold in Italy by AGIP Petroli with the name of BluDiesel, whose characteristics, respecting the EN 59 standards, are listed in Table.

4 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo Table. Main specifications of the employed fuel Density [Kg/m 3 ] 3. Kinematic viscosity [mm /s] 3 Flash point [ C] 55 Water, max [mg/kg]] Distillation temperature [ C] (% of volume recovered) 5%> 5 5%< 35 Ash, max [% weight].1 Sulfur, max [mg/kg] 1 Cetane number 51 PAH content, max [% weight] 11 Net Heating Value [MJ/Kg] 3,1 C:H Ratio (H=1), 3. Compression ratio value selection and piston bowl design The current CR standard of most of the EURO engines is in the range The reference engine in its EURO production version is 17.5 (see Table 1). However a brief survey on the most advanced LD PCCI engines reported in literature, points out that the modified engines have a CR value in the range between 17 and 1 approximately (Imarisio et al.,, Hotta et al., 5, Kanda et al. 5, Kawamoto et al.,, Huang et al., 5, Araki et al., 5, Helmantel and Denbratt,, Duret et al.,, Vignaud et al.,, Cooper et al., ). Also the last trends in diesel engine development indicate a tendency in compression ratio reduction in a range between 1.5 and 15.5 (Imarisio et al.,, Hara et al., ). Looking at the results of the papers dedicated to the analysis of the CR effect on PCCI performance (Wagner et al., 3; Kitabatake et al., 7, Hamada et al., 5, Walter and Gatellier, 3 Pedersen and Schramm, 7, Araki et al., 5) it appears clear that the reduction of CR in a range of 1 1, with respect to the current value of EURO engine, permits the increment of upper engine load limit curve of PCCI applicability, without deterioration of BSFC. However, as a general overview and as clarified in the introduction, it is unclear which are the real limits in CR reduction, considering some practical constrains as BSFC and noise increment that have to be maintained within limit values. On the basis of the above considerations, three scaled CR values were selected for the experimental program: 1.5, 15.5 and 1.5. The designs of the three piston bowls, corresponding to the three CR values, were carefully performed with the 3D simulation by Centro Ricerche Fiat. Starting from the design of the EURO piston shape, the following guide-lines for piston shape design were met: To maintain the same air flow structure at the end of compression stroke in order to assure about the same swirl and turbulence characteristics versus CR; To avoid an increase in the squish height and keep constant the internal diameter, in order to increase the k-factor (see in Tab. 3 the definition) and to keep the same squish flow inside the bowl during compression stroke; To keep the same lip profile of the bowl in order to guarantee the same structural robustness at rated power. In this way, all the characteristics of air-flow mixing process (spray/wall interaction, spray dispersion etc.) and combustion products oxidation of the designed bowl shape of the production piston should be preserved. Furthermore, the difference carried out by experiments will be ascribed mainly to the CR variation. At the end of the work the three shapes were licensed as plotted on the left of Fig.. In the same figure, on the right, it is also reported the scheme of the piston geometry indicating some main geometrical parameters.

5 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 5 Fig.. Piston bowl shape for the three selected CR values. In the figure is reported also the Z axis assumed as the central axis of the cylinder. The well known geometrical factors of the bowl design, RR, K-factor and AR, corresponding to the three CR values, are reported in Table 3 together with the definition of each factor. The significance and the importance of each factor on the diesel combustion can be easily found in literature and also described by Cipolla et al. (7). Table 3. Geometrical factors of the bowl geometry for the three CR values Parameter Formula K-factor [%] Bowl volume/combustion chamber volume RR [%] (Dext-Dint)/Dint AR [%] Hmax/Dext Tab.3 indicates that all the factors increase with CR reduction. K-factor increases as desired, the others show a progressive increase versus CR precisely due to the bowl enlargement keeping the same geometrical structure. The air flow patterns for the three CRs crank angle degrees before the TDC is plotted in Fig. 3. The plots are the result of an engine cycle simulation without injection performed at Istituto Motori with KIVA3V code. Fig. 3. Air flow patterns for the three selected CR values crank angle degrees before the Top Dead Center. The arrows are plotted with the same scale for all CR. As expected, from the Figure the main air flow patterns appear very similar, with the main difference for CR 1.5 in the central part of the bowl. In terms of swirl ratio behaviour versus CR, Figure reports the three curves for the three CRs. The slight increase of swirl ratio versus CR decrease depends on the friction loss reduction due to the surface-volume ratio reduction as well as from the increase of moment of inertia. In fact, following the swirl ratio formula (Heywood, 199): SR = I z M z rpm π 3 Lowering the CR, the moment of inertia I z is reduced and the momentum M z decrement is less effective due to the lowered friction loss as can be observed in the two graphs in Fig. 5. (1)

6 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo 5. Swirl Ratio [a.u.] c.a. [ ] =.7 RC 1.5 RC 15.5 RC crank angle [ ] Fig.. Swirl ratio (SR) versus crank angle for the three CR values Iz [g*cm ] RC 1.5 RC 15.5 RC 1.5 Iz [g*cm /s] RC 1.5 RC 15.5 RC 1.5. SR@ -1 c.a. [ ] = crank angle [ ] crank angle [ ] Fig. 5. Moment of inertia (on the left) and Momentum (on the right) versus crank angle for the three CR values. The differences for the three bowls are small and so the differences in terms of ignition delay time as well as combustion evolution can be ascribed mainly to the thermodynamic status at the start of injection timing (SOI) derived by CR variation.. Test methodology In general, when experimental engine tests are performed in order to evaluate the influence of geometrical and/or operating parameter on combustion performance, the analyzed parameter is varied with respect to a reference well known condition. This is often observable in a lot of works on single cylinder research engines. However, with this methodology, even if it is possible to put into evidence the parameter influence, very important from a scientific and technical point of view, it is difficult to evaluate the transferability of the results or trends to a real four cylinder engine. The reason is that hardware or software constrains present in the real engine configuration are not carefully considered in the single cylinder experimental tests or are difficultly reproducible. So, the single cylinder research engine data are useful to analyze the effect of parameter variation but not for the quantitative evaluation of four cylinder engine improvement. Furthermore, often, the tests are carried out in some operating steady state points that are poorly representative of the engine performance in the whole NEDC cycle that includes also transient conditions. On the contrary, the analysis of the CR reduction effect on engine performance under NEDC cycle or transient conditions requires an ad hoc engine for each CR step.

7 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 7 On the basis of these considerations, in order to point out the CR influence on the engine performance, taking into account both the practical limits of the real engine configuration and a test point matrix reasonable representative of the engine behaviour during the NEDC procedure, the following test methodology was chosen. Starting from an engine configuration with CR equal to 1.5, the CR was progressively reduced to 15.5 and to 1.5. For each engine configuration, under conventional combustion, a first characterization with the standard EURO was performed in terms of trade-off NOx-PM, varying the EGR rate, in five steady state operating points as listed in Table. Table. Selected engine operating points with conventional combustion Notation Engine speed [rpm] Engine load BMEP [bar] C1 15 C 15 5 C3 C 5 C5 5 During the tests, in order to evaluate the CR effect on the engine performance with the lowest fuel consumption variation, for each CR value only the SOI of the main injection (SOI main ) and the throttle position of VGT (VGT%) were progressively adjusted to keep constant MBF5% value and boost pressure in the intake manifold. In the same way, to avoid the combustion noise deterioration caused by premixed fraction combustion increase with low CR, the pilot injection quantity (Q pil ) was adjusted for each CR value in order to maintain constant the maximum gradient of the cylinder pressure. As well known, in this way it is possible to put in evidence the CR effect on the combustion improvement without influence of EGR circuit functionality (hysteresis, temperature variation in EGR circuit due to CR change etc.), evaluating how much the CR reduction permits to lower the NOx-PM trade-off. For PCCI tests, starting from a first whole engine supplied by Centro Ricerche Fiat relative to a CR equal to 1.5, the engine operating points were tested for each CR value in a engine speed range from 1 rpm to 3 rpm and increasing the engine load from idle until one of the following limits were reached: 5% of fuel consumption increment with respect to the EURO value; bar/ms of maximum pressure rise (dp/dt max ) for each engine condition preserving the current comfort standards; 1.5 FSN of exhaust smoke value as estimated limit in order to avoid excessive DPF system loading, when adopted. The choice of 5% of maximum FC increment with respect to EURO standards can appear too high preliminarily considering the current trends strongly focused on FC reduction. However, its choice depends on the fact that the engine has an EURO configuration, so a further contribution to FC improvement by PCCI use are expected from EURO 5 and EURO technology. Foreseeing the PCCI application to EURO 5 and to EURO engines, a current 3% of FC increment could be recovered by technology upgrade. The difference of % of the accepted FC increment is due to the injection pressure control system that has the pressure regulator placed on the rail. Therefore, at same condition, the injection pump sends to the rail the total inlet fuel flow. On the contrary, the reference FC data of EURO refers to an engine scheme with a Bosch CP1h high-pressure injection pump that works only on the fuel flow injected in the cylinder (pressure regulator placed on the inlet channel of the pump) reducing the friction power required by the injection pump. The difference in FC penalty for the present injection system configuration as a whole with respect to EURO arrangement was estimated of about %. In this way, the exploration of PCCI combustion mode permits the identification of the maximum load curve versus rotating speed, and then of the relative applicable engine operating area, for each CR value. This also allows to the identification of the factors limiting or improving the PCCI application to diesel engines. Of course, as followed with conventional combustion test methodology, the SOI main of the single shot injection adopted in the PCCI was tuned varying the CR in order to respect the prefixed limits of FC increment and maximum cylinder pressure rise, while VGT aperture was adjusted, varying the CR to maintain constant the intake charge pressure. Any change on the EGR actuation map was actuated assuming that the engine out NOx emissions in the operating area with PCCI were preliminarily inside the post EURO 5 regulation, and taking into account that too high EGR level could affect the vehicle driveability. The adopted operating points of the PCCI are displayed in Fig., together with the markers of the test points chosen for conventional combustion analysis and the engine operating area of the vehicle, equipped with the same engine (Alfa Romeo 17 with an inertia mass of 135 kg), that runs in the NEDC cycle.

8 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo Engine load - BMEP [bar] 1 1 Steady state test points in conventional and PCCI combustion PCCI Conventional NEDC operating area Engine speed [rpm] Fig.. Engine operating points with conventional and PCCI combustion together with the operating area of the engine executing the NEDC cycle. From Fig. it is easily observable that the selected test points with both conventional and PCCI combustion are well representative of the engine performance in the NEDC operating area. 5. Effect of CR on engine performance with conventional combustion In the present paragraph the results carried out running the engine under conventional combustion for all CR configurations will be analyzed. Taking into account the large amount of data, only the most significant test points will be presented, also considering that the not reported test points have shown a behaviour very similar to their closer operating point analyzed in the paper. Fig. 7 shows the trade-off Soot-NOx (on the left) and CO-NOx (on the right) versus CR, performed varying the EGR rate in the first test point C1. Soot is converted in mass emission from smoke measurement by means of the following AVL correlation reported in (AVL application note AT17E, ): Soot[ g / m 3 ] 1.5 (.31 FSN ) = 5.3 FSN e () Soot [g/kwh] EGR 7% Test condition: C1 1 NOx [g/kwh] EGR % CO raw [g/kwh] Test condition: C1 EGR 7% 3 1 EGR 3% EGR % 1 NOx [g/kwh] Fig. 7. NOx-Soot (on the left) and NOx-CO trade-off (on the right) versus CR obtained varying the EGR rate in the test point C1. The NOx-soot trade-off shows that at low load it is very easy to reduce NOx with EGR, also keeping very low level of soot. With, only at highest EGR, a slightly increase of soot is detectable. Lowering CR to 15.5 and subsequently to 1.5, a drastic drop of smoke, and so soot, was revealed for all EGR rates. For the last two CRs, the classic trade-off was practically absent, and a progressive contemporary reduction of both NOx and

9 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 9 soot was measured. This is a typical result for the modern automotive engines with low CR running at low load and low speed, and it depends on the smoothed transition from the conventional diesel combustion to a PCCI combustion when EGR is increased at the maximum admissible value and CR is reduced. This explanation will be clearly depicted in the next where the indicated cycles will be analyzed. For this reason, at low load and low speed as the C1 condition, generally the engine is tuned optimizing the NOx-CO trade-off reported in the right plot of Fig. 7. From the right diagram it is notable the significant increase of CO emission lowering the CR value. The is particularly critical. HCs emissions follow about the same trend of CO, as reported in Fig.. HC raw [g/kwh] EGR 7% EGR 3% Test condition: C1 EGR % 1 NOx [g/kwh] Fig.. NOx-HC trade-off versus CR obtained varying the EGR rate in the test point C1. In C1 condition, the pressure rise and as a consequence, the combustion noise, is generally low and also reducing the CR any increment of dp/dθ max was detected. Therefore only SOI main and VGT% were adjusted in order to keep the same MBF5% and boost pressure increasing the EGR rate. This is shown in Fig. 9 where pilot quantity and SOI main versus the exhaust lambda are plotted and in Fig. 1 reporting the dp/dθ max (on the left) and MBF5% (on the right) versus lambda. The lambda is calculated for each step of EGR rate. As expected, to keep the same MBF5% value, SOI main has to be advanced due to both the EGR increase and the CR reduction. Of course, if MBF5% is fixed for all conditions, VGT% adjustment versus CR requires only small variation in order to maintain the same level of the intake boost pressure. Test condition: C1 1 Test condition: C1 Q pil [mm3/stroke] SOI main [c.a. BTDC] Fig. 9. Q pil (on the left) and SOI main (on the right) versus lambda in the test point C1 for all CR values.

10 1 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo dp/dθ dp/dteta max [bar/c.a. ] Test condition: C1 MBF5% [c.a. ATDC] Test condition: C Fig. 1. dp/dθ max (on the left) and MBF5% (on the right) versus lambda in the test point C1 for all CR values. The respect of a constant MBF5% versus CR permits to limit the FC deterioration reducing CR. Fig. 11 (left side), for all CR values, displays the BSFC for all CR values measured during trade-off execution. BSFC [g/kwh] Test condition: C1 T exh upstream Turbo [ C] Test condition: C Fig. 11. BSFC (on the left) and T exh (on the right) versus lambda in the test point C1 for all CR values. In the same plots, only for the variance bars calculated as average measure on three repetitions are reported. Fig. 11 reveals that if a FC penalty is expected with low CR due to a contemporary reduction of thermodynamic efficiency (expansion ratio decrement) and combustion efficiency (unburned emission increment), the SOI main adjustment with CR permits to lower the BSFC deterioration. This behaviour is mainly ascribable to the combustion efficiency decrement instead of the thermodynamic efficiency reduction. The low thermodynamic efficiency reduction versus CR with the tuning of combustion, it is explainable considering that for low CR the mechanical losses are reduced and with lean mixture, the expansion stroke reduction doesn t leads necessary to higher exhaust temperature linked with a thermodynamic efficiency decrement. This result is also confirmed by the very similar values, among all CRs, of the exhaust temperature (T exh ) measured upstream the turbine and plotted in the right diagram of Fig. 11. These results are in line with results of Laguitton et al. () relative to the effect of CR (in the range 1 1.5) on a single-cylinder PCCI engine performance. All above considerations on engine behaviour in C1 conditions for the three CR values are well evidenced by the pressure cycle analysis carried out from Fig. 1. The curves are relative to a lambda value roughly equal to the EURO one, for the same operating point.

11 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 11 Test condition: C Test condition: C1 -.1 Cylinder pressure [bar] 5 3 RC 1.5 RC 15.5 RC 1.5 ROHR [%/ ] RC 1.5 RC 15.5 RC C.A. [ ] C.A. [ ] Fig. 1. Cylinder pressure trace (on the left) and ROHR curve (on the right) versus CR in the test point C1 for a lambda value.9. As reported before, from Fig. 1 it clearly appears the tendency to shift towards a super diluted lean combustion (similar to PCCI operating mode) reducing the CR value. This behaviour explains the contemporary NOx and soot emission reduction as displayed in Fig. 7. Also the classical thermodynamic effect of pilot combustion that increases pressure and temperature before SOI main event, becomes imperceptible, and the pilot influence is limited to the effect of radicals, deriving from the partial oxidation of the pilot fuel. This consideration is derived not only by the ROHR analysis, but also from tests without pilot injection that confirm a pilot influence on ignition delay time of main injection. As a consequence, it is reliable the hypotesis that a large fraction of HC and CO emissions linked with the lowest CR, is imputable to the bad combustion of pilot injection due to excessive wall wetting of the fuel and its over-dispersion inside the combustion chamber. For these reasons, a pilot combustion efficiency improvement and an adequate control of HC and CO emissions could be performed adopting a different injection strategy based on the shift from a conventional pilot injection, generally phased 1 µs before the SOI main to a very close pilot injection phased about 3 µs before SOI main, avoiding the pilot fuel dispersion inside the bowl volume. Another contribution to unburned compound emissions could depend on the wall wetting of the main injection, considering the significant SOI main advancement reducing the CR (Fig. 9). In the operating point C3 ( rpm and bar of BMEP), the engine has shown trends practically equal to the test point C1 and so its analysis is omitted for brevity. Also at 5 bar of BMEP the engine has produced the same results between the two testing speed: 15 and rpm (C and C test points). Therefore, the engine response for these conditions can be made analyzing only the C point. Cylinder pressure [bar] Test condition: C -.5 RC Cyl.1 RC Cyl.1 RC Cyl.1 ROHR [%/ ] RC Cyl.1 RC Cyl.1 RC Cyl C.A.[ ] -5 Test condition: C -.5 Fig. 13. Cylinder pressure trace (on the left) and ROHR curve (on the right) versus CR in the test point C for a lambda value of.5. Fig. 13, put in evidence the change of engine behaviour versus CR increasing the engine load. In fact, in this case, due to the prominent tendency to reach a noisy combustion with low CR, in order to maintain the same values of dp/dθ max and MBF5% for all CRs, Q pil has to be increased while SOI main, especially for, has to C.A.[ ]

12 1 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo be retarded notwithstanding the low CR value, due to the strong effect of big Q pil. These trends are very well illustrated in the Figures 1 and 15 respectively. The Q pil increment is easy detectable also from Fig. 13 looking at the heat release rate of pilot fuel combustion. As a consequence, a progressive deterioration of FC was measured, mainly due to the increase of Q pil. This trend is relievable in Fig. 1 where BSFC versus lambda for all CR values is plotted. In the dp/dθ max diagram (left side of Fig. 1) is also displayed the values relative to two tests for and respectively, performed with a lambda equal to the nominal engine and the same Q pil of the case. The diagram put into evidence that a strong increment of the pressure rise, corresponding to very noisy combustion, occurs with low CR values when any action to compensate the ignition delay time increase is undertaken. dp/dθ dp/dteta max [bar/c.a. ] 7 5 (Qpil=) (Qpil=) 3 Test condition: C MBF5% [c.a. ATDC] Test condition: C Fig. 1. dp/dθ max (on the left) and MBF5% (on the right) versus lambda in the test point C for all CR values. Q pil [mm3/stroke] Test condition: C SOI main [c.a. BTDC] Test condition: C Fig. 15. Q pil (on the left) and SOI main (on the right) versus lambda in the test point C for all CR values. BSFC [g/kwh] 7 5 Test condition: C T exh upstream Turbo [ C] Test condition: C Fig. 1. BSFC (on the left) and T exh (on the right) versus lambda in the test point C for all CR values. In Figure 1 is also reported the variance bars for the measured as described before for C1 test conditions. Significant differences were observed only for, while in the case of, for all lambda values, the FC increment was at the limit of the test to test variation. However, the FC deterioration was mainly im-

13 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 13 putable to Q pil increment, while the thermodynamic efficiency reduction with low CR was almost insignificant as in C1 test case. In fact, it was measured that the percentage of FC increment was roughly equal to the Q pil increment and also the plot of T exh for the C test case (on the right of Fig. 1), very similar for all CR values, confirms a negligible thermodynamic efficiency variation versus CR as in C1 test case. The correspondent emission trends for the present test point C are displayed in Fig. 17 and 1 in terms of NOxsoot, NOx-CO and NOx-HC trade-off respectively. Soot [g/kwh] Test condition: C EGR 7% EGR % (Qpil=) (Qpil=) EGR % CO raw [g/kwh] Test condition: C NOx [g/kwh] NOx [g/kwh] Fig. 17. NOx-Soot (on the left) and NOx-CO trade-off (on the right) versus CR obtained varying the EGR rate in the test point C. HC raw [g/kwh] Test condition: C NOx [g/kwh] Fig. 1. NOx-HC trade-off versus CR obtained varying the EGR rate in the test point C. In terms of NOx and soot emissions, it is noticeable that NOx are always controlled by EGR level and differently by low load condition (C1 test case), the necessity to adequate the Q pil (in order to keep constant the dp/dθ max ) does not permit to lower the soot emission exploiting the longer ignition delay of the low CR values. In fact, in this test case the reduction of CR leads to not remarkable improvements in NOx-Soot trade-off. To confirm the limitation on NOx-soot trade-off improvement by Q pil adjustment, Fig. 17 also reports the results (single points) on NOx and soot emission of both and for the test conditions with the same Q pil of. Without Q pil adjustment a significant effect of CR on smoke emission becomes clear. The unburned compounds as HC and CO versus EGR follow the typical trend at medium load, where the reduction of the mass flow rate at the exhaust (due to the increase of the EGR) has a predominant effect on the increase of the concentration, showing a progressive decrement of HC and CO flow mass rate versus EGR. Regarding the CR effect on HC and CO, the engine follows the same trends of C1 test point, with a marked rise of both HC and CO. In the C5 operating point tested with conventional combustion (5 rpm and bar of BMEP), the engine has shown a behaviour very similar to C test point, with very small effect of CR on both NOx-soot trade-off and FC penalty. Therefore, omitting the description of this last test point the following main results can be summarized for the CR effect on the engine behaviour with conventional diesel combustion: At low engine load condition the lowering of CR tends to shift the combustion toward very lean PCCI conditions characterized by very low level of NOx and soot exhaust concentration and a large amount of CO and HC emissions with small effect on BSFC, mainly dependent on the combustion efficiency reduction;

14 1 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo At medium engine load, lowering the CR, notwithstanding the use of a pilot fuel injection, the combustion tends to very noisy conditions. If the noise level, and so dp/dθ max has to be taken constant, the Q pil has to be increased limiting the potentiality of low CR values on NO-soot trade-off. The BSFC variation depends mainly on the Q pil increase and the combustion efficiency decrement, while negligible variation of thermodynamic efficiency versus CR was detected. Also at medium engine load the increment of regulated unburned compounds was remarkable; Looking at the whole engine behaviour versus CR with conventional combustion, the CR value of 15.5 appears the best compromise between NOx-soot trade-off improvement, unburned emission increment and FC penalty.. Effect of CR on engine performance with PCCI combustion As introduced before, PCCI combustion was realized with a single shot injection, taking care to not overcome the prefixed limits on dp/dθ max, FC increment and smoke emission. For all explored engine speed ranges (1 3 rpm), with no changes in the engine and adjusting only the SOI main and the VGT%, the torque was continuously increased until one of the three limits was reached. With this test procedure the following maximum engine load curve versus speed with PCCI combustion mode for the three CR values was carried out: Engine load - BMEP [bar] RC 1.5 RC 15.5 RC Engine speed [rpm] FC increment w respect to E level [%] Engine speed [rpm] Fig. 19. On the left, maximum engine torque curve versus speed with PCCI combustion within the prefixed limits on dp/dθ max, FC increment and smoke emission versus speed for all CR values. On the right, FC increment with respect to EURO standards and smoke versus speed for all CR values. The left chart in Fig. 19 illustrates that reducing the CR only small increments in maximum engine torque are possible without changing the engine technology (EURO ) and. The SOI main was adjusted in order to reach the fixed limit on dp/dθ max value in each point and so the maximum torque was limited by the achievement of FC increment limit, as reported on the right of Fig. 19. As expected considering the PCCI characteristics, the higher FC with respect to conventional combustion depends on the low fuel conversion efficiency as well as on the thermodynamic efficiency reduction due to the needs to retard the MBF5% value. In the same diagram the smoke curve versus speed is also displayed. Smoke limit was never reached while a strong drop of smoke values were observed for and 1.5. The corresponding NOx emissions were very similar among all CRs since, under PCCI conditions, they are mainly dependent on EGR level. They are plotted in the left diagram of Fig., while on the right the corresponding emissions of HC and CO are displayed Smoke [FSN]

15 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 15 NOx [g/kwh] 3 1 RC 1.5 RC 15.5 RC 1.5 EURO level HC raw [g/kwh] RC 1.5 RC 15.5 RC CO raw [g/kwh] Engine speed [rpm] Engine speed [rpm] Fig.. NOx emission of the maximum engine load curve with PCCI combustion versus speed for all CR values. On the right, HC and CO emission of the maximum engine load curve with PCCI combustion versus speed for all CR values. As well known, the application of PCCI combustion generates a strong rise of unburned emissions and if PCCI engine management is coupled with the CR reduction, a further rise in HC and CO exhaust concentration is expected, as illustrated in the above diagram. The exploration of the maximum engine torque versus speed with PCCI combustion has put in evidence that negligible improvement can be achieved in the maximum attainable load with CR reduction without changing the engine hardware or its management. At the same time, a sharp rise in unburned emission was found, whereas a remarkable drop in exhaust smoke was observed. To investigate this behaviour, the analysis of indicated signal can give useful information. For this purpose, the test point at 1 rpm and bar of BMEP was selected for the comparison. It is one of the points in the middle of the maximum torque curve versus speed and all CRs have given the same maximum torque. The relative cylinder pressure, injector energizing current and ROHR curves are plotted in Fig PCCI mode - 1 bar BMEP 1 5 PCCI mode - 1 bar BMEP Cylinder pressure [bar] Injector energizing current [A]. ROHR [%/c.a. ] C.A.[ ] C.A.[ ] Fig. 1. Cylinder pressure traces and injector energizing current curves on the left, ROHR curves on the right, versus CR in the test point 1 rpm at bar of BMEP with PCCI combustion mode. The analysis of the curves shows the increase in the ignition delay time reducing CR. The ignition delay time (τ id ) increment was higher in the change from to 15.5 than from to 1.5. It is supposed that beyond other secondary factors that can affect the ignition delay time versus CR as fluid-dynamics and so on, the main reason could be the influence of the negative temperature coefficient of τ id. In fact, reducing the CR in the range , the final compression temperature at TDC, for the considered operating point, varies in the range from about 15K to 9K. In this range the typical τ id of diesel fuel as a function of pressure and temperature shows the classical negative temperature coefficient (NTC) for temperature below 1K (see Pfahl et al., 199 and Yanagihara, 1). Therefore if the τ id falls in the NTC range passing from to 1.5 it is expectable a reduced effect of CR lowering on τ id. This explains also the reduced difference on smoke between and with respect to the comparison between and. This trend was generally detected for all tested conditions except in some case. In particular at 1 rpm, the very noisy conditions at maximum engine torque (.3 bar of BMEP) with the lowest CR (1.5) required a later SOI main with respect to the other CR values.

16 1 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo Through the analysis of the data in all the tested points, two main factors were identified as limiters of the attainable engine torque with PCCI combustion: the fuel consumption and the smoke emission, since dp/dθ max was controlled by SOI main adjustment, it affects as a consequence FC and smoke. Among all the operating engine parameters influencing the three limiting factors, two parameters were recognized as the most effective: the swirl control flap position (St%) and the nominal injection pressure (P rail ). At the maximum attainable torque, to limit the smoke below the 1.5 FSN, the swirl ratio has been regulated at its medium-high operating range (.5 ) corresponding to a very closed position of the flap. This actions limit the permeability of the intake duct, thus reducing the volumetric efficiency and increasing both the pumping loss and the fuel dispersion under local over-leaning mixture within piston bowl. Furthermore, to control the smoke, P rail has been kept high, so increasing fuel impingement quantity on piston bowl, dp/dθ max and the power requirement from the high pressure injection pump. The VGT% was not very effective in increasing the engine load, as the closure VGT stator at a small boosting pressure rise an equivalent backpressure rise occurs without evident effect on engine load. For different turbocharger systems (for example two-stage turbo) an improvement could be expected. However, considering the strong advantage of the low CR in terms of smoke, an evaluation of the possibility to improve the maximum engine torque with PCCI combustion by means of a different engine management of the SR and P rail level was carried out. The injection strategy was not considered, as the analysis of sensibility requires the variation of too many parameters, like the number of injections, fuel distribution versus injection, timing of all etc. So to this aim, the maximum attainable engine torque with PCCI combustion was parameterized as function of St% and P rail. Both parameters were progressively lowered, always optimizing SOI main in each conditions, until the limits on FC and smoke were reached again. The new maximum PCCI engine torques for both and are reported in Fig.. 7 Engine load [bar] R.C. 1.5 R.C R.C Fig.. Maximum attainable PCCI engine torque versus speed. Engine speed [rpm] The maximum torque was increased from about bar to bar of BMEP. For the improvement was negligible because the smoke emissions were very close to the limit of the reference. To give a brief overview of the engine sensitivity to the swirl ratio and injection pressure parameterization, four diagrams are reported in Fig.3, displaying the comparison in terms of BSFC increment (upper-left), the smoke emissions (upper-right), the NOx emissions (lower-left) and HC+CO emissions (lower-right) with respect to EURO level for the reference PCCI and the SR%-P rail for the RC 1.5 in operating point rpm at bar of BMEP.

17 Compression Ratio Influence on an Advanced PCCI Diesel Engine Performance 17 NOx emission [g/kwh] BSFC increment w respect to E standard [%] bar of BMEP reference bar of BMEP Engine speed rpm.5 bar of BMEP bar of BMEP reference Engine speed rpm reference.5 bar of BMEP bar of BMEP Smoke [FSN] HC+CO emission [g/kwh] bar of BMEP reference bar of BMEP Engine speed rpm.5 bar of BMEP bar of BMEP reference Engine speed rpm reference.5 bar of BMEP bar of BMEP reference reference reference reference reference reference Fig. 3. BSFC (upper-left), smoke emissions (upper-right), NOx emissions (lower-left) and HC+CO emissions (lower-right) with respect to EURO level for the reference PCCI and the SR%-Prail for the RC 1.5 versus engine load at engine speed of rpm. It clearly shows the overall improvement of engine performance with acceptable increment of smoke, well within the fixed limit. NOx emissions tend to increase slightly with the reduction of the premixing level. This trend is supposed to depend on the progressive reduction of the leaning mixture zone for an increase of stoichiometric or rich mixture zone characterized by high flame temperature. This hypothesis is supported by the presence of diffusive combustion trace in the ROHR curve of the as shown in the right diagram of Fig.. The same figure also shows the strong advantage in terms of combustion phasing, with respect to the TDC, reducing swirl ratio and injection pressure. These, together with the fuel conversion efficiency improvement with the, are the main factors producing the BSFC reduction (upper-left diagram of fig.3). Fuel conversion efficiency is not reported for brevity but its trend can be estimated looking at the HC+CO diagram reported in Fig. 3. Cylinder pressure [bar] PCCI mode - bar BMEP reference Injector energizing current [A]. ROHR [%/c.a. ] PCCI mode - bar BMEP reference C.A.[ ] C.A.[ ] Fig.. Cylinder pressure trace and injector energizing current curve at left, and ROHR curve at right, versus engine in the test point rpm at.5 bar of BMEP for the with PCCI combustion mode.

18 1 C. Beatrice, C. Bertoli, C. Guido, N. Del Giacomo Similar trends were observed in all the tested engine speed range. As said before no change in engine hardware and injection strategy were applied. Only at 3 rpm, for all CR values, the exhaust smoke level showed a significant rise very close to the limit, due to the lower P rail, as reported in Fig. 5; thus any torque improvement with CR reduction was possible at the highest tested engine speed. This result depends on the short time available for air-fuel premixing during the τ id at high engine speed, and as a consequence the combustion evolution tends towards diffusive characteristics. This trend is notable in the plot on the right of Fig Engine speed 3 3. bar of BMEP 7 PCCI mode bar BMEP Smoke [FSN] 1.5 Cylinder pressure [bar] 5 3 reference 1 1 ROHR [%/c.a. ] 1 reference C.A.[ ] Fig. 5. Smoke emission (at left), cylinder pressure trace and ROHR curve (at right), versus CR in the test point 3 rpm at 3. bar of BMEP for the with PCCI combustion mode. Unburned compound emission remains at very high level at the engine exhaust and tend to increase reducing the CR despite the use of an re. Downstream the catalyst HC and CO were found at the same level of EURO standards when the catalyst operates beyond the light-off temperature. However, this last result is not very important because NEDC procedure foresees the cold start phase where the catalyst is not active. Therefore the high unburned emission penalty of low CR engines appears one of the main problem to their development, as also pointed out by Kitano et al. () in a study on cold starting performance of a low CR engine. The following main results can be summarized for the CR reduction on the engine behaviour with PCCI combustion: In every engine condition FC penalty versus CR reduction was very small and mainly dependent on combustion efficiency reduction. With the chosen operating constrains on cylinder pressure and smoke and without change the, the only CR reduction is not capable of increasing the maximum attainable PCCI engine load that for our engine was about bar of BMEP; The pollutants emissions are characterized by a marked drop in smoke and a drastic rise in HC and CO while NOx remain mainly sensible to the adopted EGR level; The drop in the smoke level at the exhaust can be easily exploited to improve the maximum torque. Working only on the reduction of both swirl ratio and injection pressure, pumping and friction loss can be lowered and intake duct permeability increased. This permitted to bring the maximum PCCI engine torque at about bar of BMEP, reducing at the same time the unburned emissions; It is confirmed that also for PCCI operating mode, the possibility to adjust the combustion phasing in the expansion stroke together with the friction losses reduction does not cause the FC deterioration for low CR engines and permits to lower the NOx emission well beyond the EURO 5 limit; One of the key issue with low CR PCCI engine appears the very high HC and CO emission, that cannot be controlled by after-treatment system during cold start. For this reason, looking at the overall engine performance, and taking into account the use of an engine with the current technology, a CR value around 15.5 appears the best compromise. Conclusions The paper presents the main results from an extensive research program performed at the Istituto Motori in cooperation with Centro Ricerche Fiat and based on the analysis of compression ratio influence on the current two liter class FIAT diesel engine operating with conventional as well as PCCI combustion.