Combustion and emission analysis of the direct DME injection enabled and controlled auto-ignition gasoline combustion engine operation

Fuel 107 (2013) 800–814

Contents lists available at SciVerse ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Combustion and emission analysis of the direct DME injection enabled and controlled auto-ignition gasoline combustion engine operation H.F. Zhang, K. Seo ⇑, H. Zhao Centre for Advanced Powertrain and Fuels, School of Engineering and Design, Brunel University London, UK

h i g h l i g h t s " Dual Fuel CAI combustion is achieved by PVO without intake heating. " In-cylinder fuel blending is used to vary the auto-ignition quality of the charge mixture. " DME injection timing and quantity affect the combustion characteristics. " The changed combustion characteristic will aid to extend CAI operating range. " Compare to SI engine, over 90% reduction of NOx emission is achieved.

a r t i c l e

i n f o

Article history: Received 16 November 2012 Received in revised form 28 January 2013 Accepted 30 January 2013 Available online 19 February 2013 Keywords: Controlled auto-ignition (CAI) Homogeneous Charge Compression Ignition (HCCI) Low temperature combustion (LTC) DME Positive Valve Overlap (PVO)

a b s t r a c t Controlled auto-ignition (CAI), also known as HCCI combustion has been researched extensively for gasoline engines due to their potential benefits of improved engine efficiency and low NOx emissions. One of the major challenges of typical CAI/HCCI combustion is to control the start and speed of combustion. In order to introduce some control over the heat release process of a premixed gasoline and air mixture, direct DME injection was exploited to enhance the mixture ignitability and control the combustion timing of the combustible mixture. The heat release rate and ignitability were also regulated by the internal exhaust gas recirculation (EGR). In this research, this was achieved and tested in a single cylinder gasoline port fuel injection engine by the use of Positive Valve Overlap (PVO) and direct DME injection via a commercial Gasoline Direct Injection (GDI) injector. Three DME injection strategies, early single injection, split injection and late single injection, were applied to implement pure auto-ignition, hybrid combustion of flame assisted auto-ignition, and flame propagation, respectively, as the engine load increases. Stable combustion was achieved from 1.3 bar net IMEP to 9.1 bar net IMEP with 90% reduction in NOx emission than the SI combustion operation. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Among the numerous research papers, controlled auto-ignition (CAI) combustion, also known as Homogeneous Charge Compression Ignition, has been shown to improve fuel economy and exhaust gas emissions of gasoline engines. The efficiency benefits are obtained by its high level dilution charge and lower peak in-cylinder temperature. What attracts the researchers most is that the PMs are at near zero level and the NOx emission is much lower than either conventional SI or CI engines. However, although the potential benefits of CAI combustion are great, there are some drawbacks that need to be considered, such as higher uHC emission levels, limit of operating region and difficulty in controlling the combustion phasing. In order to achieve CAI combustion, the engine requires ⇑ Corresponding author. E-mail address: [email protected] (K. Seo). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.01.067

high temperature intake charge and appropriate charge dilution so that the homogeneous or partially premixed mixture can reach the auto-ignition conditions. High in-cylinder temperature can initiate and maintain the chemical reactions inducing auto-ignition, and adequate charge dilution limits the peak heat release rate, both of which can be implemented by introducing exhaust gas recirculation (EGR) Zhao et al. [1] performed a comprehensive research on the effects of EGR on CAI combustion and presents five kinds of effects, which are: charge heating effect, dilution effect, heat capacity effect, chemical effect and stratification effects. When the high octane number fuel such as gasoline is used for CAI combustion, it requires high intake charge temperatures due to its poor auto-ignition characteristics. In order to satisfy the certain temperature requirement to initiate CAI combustion, several technologies such as inlet air heating, variable compression ratio (CR), EGR and dual fuel strategies have been applied and investigated by previous researches [2–9].

801

H.F. Zhang et al. / Fuel 107 (2013) 800–814

Nomenclature AI auto-ignition ATDC after top dead centre BTDC before top dead centre CA crank angle CAI controlled auto-ignition CN cetane number COVimep coefficient of variation in IMEP CR compression ratio DAQ data acquisition DI direct injection DME di-methyl ether EGR exhaust gas recirculation iEGR internal exhaust gas recirculation EVC exhaust valve closing EVO exhaust valve opening FuelMEP fuel mean effective pressure GDI Gasoline Direct Injection HC hydrocarbons IMEP indicated mean effective pressure ISFC indicated specific fuel consumption IVC inlet valve closing IVO inlet valve opening LHV lower heating value MBT minimum spark advance for best torque MFB mass fraction burned NVO Negative Valve Overlap ON octane number PCCI Premixed Charge Compression Ignition PFI port fuel injection PM particulate matter PMEP pumping mean effective pressure ppm parts per million PVO Positive Valve Overlap RCCI fuel reactivity controlled compression ignition rpm revolutions per minute

Internal EGR (iEGR) can provide a large amount of hot residual gas into the cylinder, where ‘large’ and ‘hot’ are considered as two key factors when EGR is used to initiate CAI combustion. Lavy et al. [10] adopted internal EGR using variable valve timing (VVT) to enhance CAI and achieve low NOx emissions in the 4-stroke powered gasoline auto-ignition controlled combustion engine (4-SPACE). Internal EGR can be obtained by either Negative Valve Overlap (NVO) or Positive Valve Overlap (PVO) or the exhaust throttling method. NVO is an effective and feasible way to obtain internal EGR, but the restriction of gas exchange limits the CAI operating range at high speed [5]. Yang et al. [2] has successfully overcome the gas exchange limit associated with NVO and achieved CAI combustion with PVO together with variable compression ratio and intake charge heating. To eliminate the requirement of intake charge heating, PVO together with exhaust throttling is applied to implement internal EGR in this research. Dual fuel CAI, also known as dual fuel PCCI (Premixed Charge Compression Ignition) or fuel RCCI (Reactivity Controlled Compression Ignition), applies in-cylinder fuel blending with port fuel injection (PFI) or early direct injection of high octane number (ON) fuel (such as gasoline) and utilizing various direct injection strategies of high cetane number (CN) fuel (such as diesel) to control combustion phasing in the whole CAI operation regime, which could achieve high engine efficiencies and low emissions at the same time. With this method, CAI combustion can be achieved and controlled by adjusting auto-ignition quality of the fuel rather than adjusting the in-cylinder condition by intake charge heating.

SI SOI1 SOI2 TDC VCR VVT WOT k

spark ignition first start of injection second start of injection top dead centre variable compression ratio variable valve timing wide open throttle lambda, relative air/fuel ratio

Definitions H IMEP IMEP ¼ V1d p:dV

100rIMEP COVimep COVimep ¼ P 100 1

IMEPi

where

rIMEP is the standard

deviation in IMEP ðmg LHVg þmD LHVD Þ Vd

FuelMEP FuelMEP ¼ QhMEP

QhMEP = gc  FuelMEP R expantion W

IMEPgross IMEPgross ¼ V dg ¼ V1d H 1 n IMEPnet W V d ¼ V d p:dV

compression

P  dV

gt

gross gt ¼ IMEP Q h MEP ;

gc

gc ¼ 1 

gg

gross gg ¼ IMEP ; FuelMEP IMEPnet gn ¼ FuelIMEP where mg is the mass of gasoline in one

gn

P

x LHVi ðmg þmD þma Þ i i mg LHVg þmD LHVD

cycle, mD mass of DME in one cycle, ma mass of intake air in one cycle, Wg gross work in one cycle, Wn net work in one cycle, T break torque, xi the mass frictions of CO and HC respectively (H2 and PM are not considered), LHVi, lower heating values for CO (10.1 MJ/kg) and HC (44 MJ/kg) respectively

Bessionette et al. [11] suggested that the ideal fuel for CAI combustion is CN 27 at high load (up to 16 bar BMEP) and CN 45 at low load (up to 2 bar BMEP). DME is a better auto-ignition fuel than diesel. The CN of DME is over 55 which is higher than diesel (CN 40–50) and its autoignition temperature of 508 K is lower than diesel. In addition, the boiling point of DME is only 248.1 K ensuring that DME will be vaporized immediately after being injected into the cylinder. DME molecule contains only C–O and C–H bands rather than C–C band with large fraction of oxygen atom, which leads to reduce in unburned hydrocarbon and PM emissions. In this research, therefore, DME is introduced and directly injected as an ignition promoter while gasoline is injected by PFI as the main fuel. The current study utilizes a single cylinder engine with dual fuel (gasoline and DME) strategy and internal EGR to investigate the effects of DME injection strategy in the control of CAI combustion and enlargement of its operating range. The experiments consisted three parts. Firstly, CAI combustion was achieved by DME split injection strategy only and the preliminary CAI operating range was identified. Secondly, three different injection strategies, early single injection, splits injections and late single injection, were then applied to enlarge the CAI combustion operating region. The combustion characteristics, performance and emission were investigated and analysed. Finally, the dual fuel CAI results were compared to conventional SI combustion results in order to determine the advantages and drawbacks.

802

H.F. Zhang et al. / Fuel 107 (2013) 800–814

Table 1 The Ricardo E6 engine test condition. Parameters

Value

Displacement Compression ratio

507.11 cc 13:1 (Geometric) 12.92:1 (Effective) 76.2  111.1 1500 rev/min 80 °C 55 °C 20–25 °C Naturally aspirated 2.5 bar 40 bar EVO = 205 ATDC EVC = 12 ATDC IVO = 12 ATDC IVC = 212 ATDC

Bore X stroke Engine speed Coolant temperature Oil temperature Intake temperature Intake pressure Gasoline injection pressure DME injection pressure Valve timing (63 °CA overlap)

2. Experimental setup A single cylinder, four stroke Ricardo E6 research engine was adopted and modified for the experimental studies. An adjustable single overhead camshaft was designed and implemented so that the cam phasing could be changed manually to realize PVO of different overlaps. The maximum valve lift was 8 mm and the intake valve duration and exhaust valve duration were fixed at 224 °CA and 229 °CA respectively. The valve open and closure timing of the intake and exhaust valves were measured by a dial test indicator and were adjusted by rotating the two cam lobes that were mounted and locked on the camshaft. The valve timings were firstly set as EVO = 187 °CA BTDC, EVC = 30 °CA ATDC IVO = 33 °CA BTDC and IVC = 191 °CA ATDC, which results in a 63 CA degree positive overlap duration. Further information of the engine is given in Table 1. Because of its ease of vaporization and flammable characteristics, care was taken in the design and implementation of the

DME supply and the DME direct injection system. Fig. 1 shows the schematic diagram of DME supply and injection system. The main parts of the DME supply system are DME and nitrogen cylinders, check valve, window chamber, filter, air-driven high pressure pump, common rail and injector. The two-stage pressure raising strategy is applied to increase DME injection pressure. Firstly, a 15 bar nitrogen gas is connected to the inlet valve of DME cylinder, which pressurizes DME to 15 bar, keeping it liquid in the entire supply pipe. Secondly, the air driven pump driven by nitrogen from another nitrogen bottle raises the DME pressure up to 150 bar. A diesel common rail injection system has been used for DME direct injection in order to keep the injection pressure constant at high pressure injections. In this study, the injection pressure was set to 40 bar by the pressure relieve valve, in order to minimum the DME injection quantity within the injector’s ability. The injector is a production solenoid actuated DI gasoline injector and it is controlled by a bespoke injector drive unit. The engine allows varying compression ratio from 4.5:1 to 20:1 by means of a worm gear, which adjusts the position of the cylinder head relative to the crankcase. To identify the optimal CR during the test, the CR was varied from 14.5 to 13 with the above valve timing. The DME split injection strategy was applied. The first DME injection (SOI1) started at 100 °CA BTDC with a quantity of 2.5 mg/ cycle. The start of second DME injection (SOI2) was varied around TDC and its quantity was fixed at 1.8 mg/cycle. The total lambda (k) was kept at 1.2 by controlling the amount of port fuel injected gasoline. Fig. 2 shows how net IMEP values vary with the SOI2 timing and the compression ratio. At CR 14.5 and 13.3, very early timing of SOI2 causes serious knocking combustion and limits the maximum net IMEP. However, at 13:1 compression ratio, knocking combustion is avoided. IMEP initially rises to a region of maximum values when the SOI2 timing is retarded. Retarded injection beyond 25 °CA BTDC causes lower net IMEP due to the reduction of expansion work. Further retarding the SOI2 timing results in unstable combustion and eventually misfire in some engine cycles. The

23 22

21

20 19 18

OUTDOOR 15 2

5

4

6 7

9

10

8

11

16

14

13

N2

DME

1

3

17

N2 12 1 Nitrogen Cylinder

9 Manual On-off Valve 3

2 Nitrogen Regulator

10 DME Filter (2.5 um)

3 DME Cylinder

11 Air Driven High Pressure Pump

4 DIN-1 Duo-pole Valve

12 Nitrogen Bomb

5 Manual On-off Valve 1

13 Air Driven Regulator (outlet 3 bar)

6 Manual On-off valve 2

14 Relieve Valve

7 Check Valve

15 Pressure Gauge (0-200 bar)

8 DME Chamber Setting

16 Common Rail

17 DME Injector 18 Pressure Gauge (0-200 bar) 19 Regulator (outlet 3 bar) 20 Pressure Gauge (0-80 bar) 21 Manual On-off Valve 4 22 Manual On-off Valve 5 23 DME Detector

Fig. 1. Diagram of DME supply and injection system.

803

H.F. Zhang et al. / Fuel 107 (2013) 800–814

6

Net IMEP (bar)

5

4

3

2 5

10

15

20

25

30

35

DME SOI2 (BTDC) CR=14.5

CR=14

CR=13.5

CR=13.3

CR=13

Fig. 2. Net IMEP as a function of the second DME injection timing at different compression ratios.

maximum net IMEP is obtained at 25 °CA BTDC under this condition. At CR 13, the SOI2 timing can be considered as an appropriate way to control combustion phasing. In addition, the engine with CR 13 will achieve higher efficiency than it would with CR < 13 at the same SOI2 timing. Therefore, CR 13 was chosen as the compression ratio for this research. In-cylinder pressure data was obtained by a water cooled piezoelectric pressure transducer (Kistler type 7061B). The in-cylinder data was recorded and calculated to obtain the heat release rate, IMEP and COVimep by a Labview DAQ (Data Acquisition) based system. The exhaust gas was continuously analyzed by Horiba Mexa 554JE, Signal 3000HM and Signal 4000VM, which indicate CO and CO2 emissions, uHC emissions, and NOx emissions respectively, shown in Fig. 3. The overall lambda (k) is determined from the exhaust gas components based on the Brettschneider Equation [12] shown in Appendix. In this study, a maximum rate of pressure rise (dP/dCA)max of 5 MPa/ms is applied on a cycle-by-cycle basis to determine the knocking combustion [13]. 5 MPa/ms is equivalent to 5 bar per one crank angle degree when the engine speed is 1500 rpm. Any individual cycle with (dP/dCA)max > 5 bar/deg is considered as a knocking combustion cycle and the percentage of (dP/dCA)max > 5 bar/deg cycles of 100 continuous cycles is used as a criterion to evaluate the extent of knocking combustion. If over 50% of 100 cycles with (dP/dCA)max > 5 bar, the test point is identified as unacceptable knocking combustion, which should be avoided. COVimep is used to indicate the combustion stability. Unstable combustion is recognized when COVimep is over 10%. Internal EGR is achieved by applying PVO and increasing exhaust back pressure if the EGR rate achieved by PVO is not sufficient. The concentrations of CO2 in the exhaust pipe and inside cylinder during compression are used to calculate the trapped residual rate. The in-cylinder sampling technique coupled with exhaust CO2 measurement is used to determine the trapped EGR rate. The internal EGR rate is defined as follows equation:

iEGR rate ðby volumeÞ ¼

CO2 % inside cylinder CO2 % in exhaust

The in-cylinder CO2 concentration was measured by a high speed intermittent sampling valve developed in the author’s laboratory [14], which was installed in one of the two spark plug holes on the cylinder head to extract in-cylinder gas samples. The desired sampling timing and opening duration were controlled by the driving current sent to the solenoid actuator in the sampling valve from a signal delay unit. The actual sampling valve opening time and duration were precisely measured by a co-axial displacement sensor and then adjusted through the driving current signal

Fig. 3. The schematic of the lab setup.

from the signal delay unit. The sampling valve was set to open at 160 °CA BTDC and closes at 120 °CA BTDC during the compression stroke. The sampled gas was collected in a sampling bag and then analyzed by the CO2 analyzer for cycle averaged CO2 and hence EGR concentration measurements.

3. Methodology In order to investigate the effect of DME injections in the dual fuel CAI combustion mode, three experiments were carried out on the single cylinder engine with PVO. The first experiment was to investigate the limits and range of CAI operation by split injection strategy only. The engine was operated at wide open throttle (WOT) and the load was controlled by the amount of iEGR for a given set of valve timings. The tests started at mid load with iEGR = 27%. Then the iEGR rate was reduced by lowering the exhaust back pressure while the load increased until knocking combustion occurred. In contrast, in order to find the low load boundary, the iEGR rate was raised by increasing the exhaust back pressure so that the load decreased until the mixture failed to ignite. Therefore, the operating range was determined by knocking at high load, misfire and partial burn limits and low load conditions. In these tests, uHC emission was examined to identify misfire and partial burn. PVO was fixed at 63 CA degree duration and iEGR rate was varied by adjusting the exhaust back pressure. Lambda is defined as the total related air fuel ratio of both gasoline and DME, and it was adjusted by varying the quantity of gasoline with fixed DME quantity. For the split injection strategy, the first DME injection (SOI1) was used to produce a more reactive premixed fuel/air mixture and the injection timing was fixed at 290 °CA. The second DME injection (SOI2) was injected near TDC and adjusted to control the combustion timing. Table 2 shows the DME injections quantity and timing. In the second experiment, three different injection strategies were tried and tested to complete the full map of CAI operation. The operation strategies are as follows:

804

H.F. Zhang et al. / Fuel 107 (2013) 800–814

Table 2 DME injection quantity and timing for split injection strategy. Parameters

Value

DME SOI1 timing (°CA) DME SOI duration (ms/°CA) Mass of DME SOI1 (mg/cycle) DME SOI2 timing (°CA, adjusted for each case) DME SOI2 duration (ms/°CA) Mass of DME SOI2 (mg/cycle)

290 1.12/10.8 2.55 324–349 0.98/9 2

In-cylinder Pressure (bar)

45 Late Single Injection 40 35 30 25 20 15 10 5 0 280 300 320 340 360

SOI Gasoline

380

400

420

440

460

Crank Angle (degree) Fig. 7. Knocking combustion determination by split injection strategy.

In-cylinder Pressure (bar)

Fig. 4. Late single DME injection strategy.

280

45 40 35 30 25 20 15 10 5 0

Split Injections

300

320

340

(3) To extend the lower load border and avoid partial burn or misfire, ‘Early Single DME Injection Strategy’ was applied in Fig. 6 when iEGR was higher than 37% with the load lower than 2.5 bar net IMEP as well as under leaner mixture when 27%
SOI1 SOI2 Gasoline

4. Results and discussions 360

380

400

420

440

460

4.1. CAI operating range with split injection strategy

Crank Angle (degree) Fig. 5. Split DME injection strategy.

(1) To extend the high load border and avoid knocking, ‘Late Single DME Injection Strategy’ in Fig. 4 was applied to engine operations with the near stoichiometric air fuel mixture and the lowest EGR concentration as indicated in the bottom left region in Fig. 7, where the load was higher than 6 bar of net IMEP, k = 1.0–1.2, and 13%
45 Early Single Injections 40 35 30 25 20 15 10 5 0 280 300 320 340 360

SOI

380

400

420

Crank Angle (degree) Fig. 6. Early single DME injection strategy.

440

460

The maximum load using the split injection strategy is limited by the appearance of knocking combustion as shown in Fig. 7, in which the number represents the percentage of knocking cycles. It shows that when the iEGR rate is less than 20% and lambda is less than 1.2, knocking combustion is unavoidable. It can be also seen that within the knocking combustion region, the extent of knocking increases as the total air/fuel ratio is reduced to near stoichiometry and the iEGR lowered. At any point of the knocking boundary defined by 50% knocking cycle, either increasing iEGR rate or reducing total k will reduce the knocking tendency of the mixture due to lower unburned mixture temperature as a result of lower EGR temperature and reduced compression effect from the burned gas region. Thus, the knocking boundary defines the highest load that the engine can possibly achieve with the split injection strategy. The low load boundary is determined by combustion stability due to partial burned combustion or misfire. Unstable combustion is identified when COVimep is over 10%. As shown in Fig. 8, the unstable combustion is observed when iEGR rate is greater than 37% regardless of the air to fuel ratio. When the mixture of lambda 1.0–1.4 with 37% EGR is present in the cylinder, it leads to misfire. In addition, when a relatively lean mixture, from lambda 1.4 to 1.7 is present with 23–37% EGR in the cylinder, it leads to partial burn. Although both partial burn and misfire are unstable combustion, the unburned HC emissions are of different results as shown in Fig. 9. On the left-hand side of the misfire limit, complete combustion takes place. In comparison, the uHC emission of the partial burn is much higher than that near the misfire limit. This phenomenon was also observed in Oakley’s research [15] which involved experiments with external EGR. However, the concentration of

H.F. Zhang et al. / Fuel 107 (2013) 800–814

805

Fig. 8. COVimep as a function of iEGR rate and lambda. Fig. 10. Operating range for split injection strategy.

Fig. 9. uHC emission as a function of iEGR rate and lambda.

uHC was higher than 5000 ppm in Oakley’s research. The lower uHC emissions in this research are thought to be a result of higher charge temperature of internal EGR and the more complete combustion characteristic of DME. Therefore, the working range map by split injection strategy is shown in Fig. 10. The maximum load of 6.8 bar net IMEP and the minimum load of 3.2 bar net IMEP can be achieved by split injection strategy without knocking and unstable combustion. 4.2. Enlargement of cai operating range and characteristics 4.2.1. Combustion characteristics The enlargement of CAI operating range was then achieved by various DME injection strategies shown in Figs. 4–6 as discussed above.

Fig. 11. Operation range by various injection strategies.

Fig. 11 presents the net IMEP values as a function of iEGR rate and total k. The operating range is split into three regions: early single injection strategy at high iEGR on the right, split injection strategy at moderate iEGR in the middle, and late single injection strategy with low iEGR on the left. As the throttle is fully open, the engine outputs are determined by the amount of iEGR and lambda. As expected, the load is reduced gradually as more iEGR is introduced. Furthermore, for a constant iEGR rate, the net IMEP increases when the lambda value is reduced in most of the operating region. However, at the operating conditions near the border of two strategies between the late single injection and split injection, the effect of lambda is less evident. This is because the later DME

806

H.F. Zhang et al. / Fuel 107 (2013) 800–814

injection can be more advanced with leaner mixture without the occurrence of knocking combustion. Thus, the net IMEP is seen to be hardly changed since the mixture goes lean for a constant iEGR rate under this condition. In contrast, as the lambda value decreases and the mixture become richer, the single DME injection timing is limited by the occurrence of knocking combustion. The minimum output is 1.35 bar net IMEP and it occurs when the engine is operated under the condition of maximum iEGR rate and the leanest mixture. This minimum value could be lower if the iEGR rate is further increased and DME quantity fraction of total injected fuel is raised. However, as the method to achieve iEGR is PVO combined with exhaust gas throttling, the exhaust back pressure for the maximum iEGR nearly reaches 1.1 bar (16 psi) gauge pressure. Due to the concern with the safety operation of the exhaust pipe, no further increase in the exhaust back pressure was made. It is thought that the minimum load border could be further extended if other methods are applied to realize greater amount of iEGR. The percentage of knocking cycles (max dP/dCA> 5 bar) of 100 cycles is shown in Fig. 12. Compared to Fig. 7 with the split injection alone, the high load boundary is extended to much lower iEGR concentrations without knocking combustion. It can be explained that with the split injection strategy, the end gas mixture contains the pre-mixed DME from the early injection and hence is more prone to auto-ignition than the gasoline/air mixture in the single late DME injection. The auto-ignition of the end gas region due to premixed DME then causes extremely fast combustion and rapid pressure rise of the characteristics of knocking combustion. Fig. 13 shows the combustion stability of CAI combustion achieved by three injection strategies. Compare to Fig. 8 with the split injection alone, early single injection strategy aids to extend both misfire boundary and partial burn boundary. When the iEGR rate is lower than 30%, the air/fuel ratio is the dominant factor for combustion stability. As the iEGR rate is higher than 30%, concentration of iEGR becomes the dominant factor affecting combustion stability. Late single injection and split injections lead to more stable combustion than the early single injection strategy. This relatively unstable combustion at low load condition can be understood by considering the effect of the cyclic variation of iEGR. In the case of early single injection strategy, iEGR becomes the

Fig. 12. Knocking combustion determination by various injection strategies.

Fig. 13. COVimep as a function of iEGR and total lambda.

dominant factor to control combustion timing and phasing as iEGR provides the thermal energy to initiate auto-ignition. Therefore, the higher COVimep value is due to the variation of iEGR rate. 4.2.2. Heat release analysis DME injection plays an important role in combustion characteristics of the three injection strategy. Since flame propagation is known to be characterized with a slower heat release rate, it has been shown that the inflexion point on the mass fraction burned (MFB) curve can be used to infer the transition between the slower flame propagation and faster auto-ignition combustion [16]. This is referred to as the T point as shown in Fig. 14b. In addition, the combustion process has been further analysed in terms of 10%, 50%, and 90% of MFB (CA10, CA50, and CA90, respectively). Combustion duration was calculated from 10% to 90% of MFB (CA10–CA90). Fig. 14 shows the in-cylinder pressure and heat release characteristics of the above three combustion modes. The ‘Early Single Injection’ mode is characterized with fast combustion and fast heat release rate. Since higher exhaust back pressure is applied to achieve high iEGR, the in-cylinder pressure is extremely high during the exhaust stroke. The T point of the MFB graph does not appear because there is only auto-ignition combustion process. In the ‘Split Injection’ mode, the heat release process is seen to take two forms, a slow heat release process before 365 °CA and a rapid heat release process afterwards, as shown in Fig. 14C. As a result, in-cylinder pressure suddenly increases in the middle of the combustion process resulting in greater peak pressure. The T point can be readily identified from the MFB curve. In the ‘Late Single Injection’ mode, slow combustion with a long heat release process is shown in the diagrams. Since both of intake and exhaust throttling are minimized in this case, the pumping work is minimum compared to the other modes. There is no apparent T point on the MFB curve as combustion takes the form of flame propagation. Fig. 15 shows the ignition timing, phasing and duration of combustion for the three injection strategies. In the late single injection region, the ignition timing only depends on the SOI2 DME timing as shown by the constant ignition delay in Fig. 15a. In the early single injection region, the ignition timing strongly depends on

H.F. Zhang et al. / Fuel 107 (2013) 800–814

45 In-cylinder Pressure (bar)

40 35 30 25 20 15

Early Single Late Single Split Injection

10 5 0 330

340

350

360

370

380

390

400

410

420

Crank Angle (degree)

(a) Pressure Traces of Three Injection Strategies 100

MFB (%)

80 60 40 20

Early Single Late Single Split Injection

T

0 330

340

350

360

370

380

390

400

410

420

Crank Angle (degree)

(b) MFB of Three Injection Strategies 100

Early Single Late Single Split Injection

HRR(J/deg)

80 60 40 20 0 330

340

350

360 370 380 390 Crank Angle (degree)

400

410

420

(c) Heat Release Rate of Three Injection Strategies Fig. 14. Combustion characteristic of the injection strategies.

the in-cylinder condition such as the air/fuel ratio, iEGR rate, incylinder temperature and pressure. In the Split Injection region, as the iEGR rate is increased, the ignition timing is firstly advanced until iEGR rate reaches 30% and then retarded as the iEGR rate is further increased even with the advanced SOI2 timing. This can be explained by the opposing effect of charge heating and dilation by iEGR [1]. As the iEGR rate increases to 30%, the ignition timing becomes earlier not only because the advanced DME SOI2 timing but also the increasing heating effect on auto-ignition timing by a rising EGR rate. However, as the iEGR rate is further increased, the heat capacity effect of the residual gas becomes the main factor for retarding the ignition timing rather than heating effect and earlier SOI2 timing to advance ignition timing. The combustion duration is determined by the speed of combustion process and affected directly by the combustion mode shown in Fig. 15c. Long combustion duration is achieved in the late single injection region, where slow flame propagation dominates. The combustion duration shows a strong relationship with the A/ F ratio and iEGR rate in the split injection region. The long combustion duration in the leaner mixture is caused by the late start of the auto-ignition process in the hybrid combustion mode. The fastest combustion takes place at moderate iEGR rate (28%). As the iEGR

807

rate is increased beyond 30% in the split injection region, combustion duration becomes almost fully dependent on the iEGR rate. This is thought to be caused by the increased concentration of inert EGR species such as CO2 and H2O which would hamper the chain propagating and degenerate-branching reactions [17]. Thus, the reaction rates of the auto-ignition process in the hybrid combustion are slowed down by increasing iEGR. With the early single injection, the fastest combustion happens in the middle of the region at lambda 1.3 and iEGR 42%, where the earliest CA50 is located. In this case, only auto-ignition combustion takes place. As shown in Fig. 15a, the auto-ignition timing is advanced as the relative air/fuel ratio increases. For a given iEGR ratio, the auto-ignition starts either too early with a leaner mixture or too late with a richer mixture than lambda 1.3 for combustion to take place near TDC. 4.2.3. Emissions and efficiencies Fig. 16 shows the distribution of indicated specific CO emission over the operating range. The maximum CO emission of 85 g/kW h is obtained under early single injection strategy with lambda 1% and 45% iEGR rate, and the minimum CO emission of 3.0 g/kW h is obtained in late single injection under lambda 1.6% and 10% iEGR rate. In general, CO emission is increased as lambda and iEGR are increased. CO emission with the early single injection strategy is sensitive to the air/fuel ratio variation because of the high iEGR rate. Since more iEGR replaces fresh charge air in the cylinder, there is insufficient oxygen present to fully oxidize the carbon atoms into carbon dioxide (CO2). As a result, the CO emission is rapidly increased as the mixture strength is approaching stoichiometric. Fig. 17 shows results of NOx emissions as a function of the iEGR rate and lambda value. As expected, NOx emissions are higher as the engine load is increased and peak at the highest load point (iEGR rate 5%, lambda 1.0). When the engine load decreases and more iEGR is introduced, the NOx emissions are dramatically reduced. The average NOx emissions are lower than 70 ppm when the iEGR rate is over 35%. Specific NOx emission is 0.25 g/KW h with iEGR 27% and lambda 1.6. It is interesting that with lambda of 1.0 and 1.1, the NOx emission rises slightly as the iEGR rate is increased from 10% to 18% and then drops rapidly once the iEGR rate is over 18%. In contrast, at lambda 1.5 and 1.6, NOx emissions are firstly reduced and then slightly increased between iEGR 35% and 43%. These two regions are the ones in which the combustion mode changes. Therefore, it can be concluded that the NOx emissions are also strongly affected by the change in combustion modes. As the fraction of auto-ignition is increased, the NOx emission is reduced due to higher iEGR. Fig. 18 shows the measured unburned HC emissions. It can be seen that uHC emissions only depend on the iEGR level and increase with the iEGR rate due to the presence of unstable combustion. The highest uHC emission is obtained when iEGR is about 30% with the leanest mixture, which is different from the condition for the maximum CO emission with the highest iEGR rate near the stoichiometric mixture. Occasional uHC peaks are caused by the unstable combustion as shown by the higher COVimep values at the same operating conditions. Furthermore, uHC emissions are found to be related to the exhaust temperature shown in Fig. 19. As the exhaust temperature is higher within the late single injection region, the uHC emissions is significantly reduced. Fig. 20 shows that the combustion efficiency decreases with the increasing concentration of iEGR. As more burned gas is presented, the combustion deteriorates resulting in high CO and uHC emissions. The highest combustion efficiency (98%) is achieved in the Late Single Injection region and the lowest value (91.5%) is obtained under maximum EGR rate in a stoichiometric mixture.

808

H.F. Zhang et al. / Fuel 107 (2013) 800–814

Fig. 15. Combustion phasing of three injection strategies.

The gross and net indicated efficiency are shown in Fig. 21a and b respectively. Gross indicated efficiency of 34% is as its maximum at the minimum iEGR rate with lambda 1.3 and 1.4. It is not the point that maximum load is achieved but when the most advanced MBT timing is applied. The peak is shifted from bottom to top-left corner in Fig. 20a. This is because in the bottom with the late single injection region, the combustion phasing was retarded to avoid knocking combustion. Thus, the peak of gross indicated efficiency is achieved under the lean mixture with minimum iEGR rate. At the highest efficiency point, CO and uHC emissions are approximately at their lowest in the entire region. The NOx emissions

are higher than most of the operation points but not the highest. It is possible to achieve higher efficiency under one iEGR rate with a leaner mixture. As the iEGR rate increases, the gross indicated efficiency is almost constant until the early single injection strategy is applied. Once the combustion mode is switched to pure CAI, the value is slightly reduced due to relatively lower combustion efficiency. However, net indicated efficiency is significantly reduced as iEGR rate increases due to a dramatic rise of pumping loss. The engine works at WOT of intake resulting in negligible pumping work. The pumping loss comes from exhaust throttling, which is used to

H.F. Zhang et al. / Fuel 107 (2013) 800–814

809

Fig. 19. Exhaust temperature by various injection strategies. Fig. 16. CO emissions by various injection strategies.

ISNOx (g/KWh)

Late Single

Split Inj.

Early Single Inj.

16 14 12 10 8 6 4 2 0 0

10 λ=1 λ=1.4

20 30 iEGR Rate (%) λ=1.1 λ=1.5

40

λ=1.2 λ=1.6

50 λ=1.3 λ=1.7

Fig. 17. NOx emissions by various injection strategy.

Fig. 20. Combustion efficiency by various injection strategies.

Fig. 18. uHC emissions by various injection strategies.

increase the exhaust pressure and then increase the iEGR rate with fixed valve overlap. However, in the Late Single region at lambda 1.0, 1.1 and 1.2, although the iEGR rate is increased, the net indicated efficiencies of these points are almost constant. This is because the iEGR rate is increased from 5% to 12% by means of increased Positive Valve Overlap (overlap changed from 24 °CA to 63 °CA by both retarding EVC timing and advancing IVO timing), without exhaust throttling. Therefore, it is believed that if other methods are used to realize large iEGR, the increasing pumping work at high iEGR conditions can be avoided. Re-breath strategy is one of such methods, which has been selected for further experiments.

810

H.F. Zhang et al. / Fuel 107 (2013) 800–814 Table 3 SI test conditions. Parameters

Value

Parameters

Value

Compression ratio Engine speed

13 1500 rpm

8 mm 20–25 °C

Coolant temperature Oil temperature

80 °C

Valve lift Intake temperature Gasoline

55 °C

Valve timing (63 °CA overlap)

IVO IVC EVO EVC

Unleaded 95 RON 2.5 bar

Gasoline inj. pressure Lambda

33 BTDC of intake 191 ATDC of intake 187 BTDC of intake 30 ATDC of intake

1.1

DME inj. pressure

40 bar

EGR

13%

Load controlled by throttle

10

COVimep (%)

8 6 4 2 0 1

2

3

4

5

6

7

8

9

Net IMEP (bar)

SI Lambda=1.1

DME Pilot Lambda=1.1

Fig. 22. COVimep as a function of load by SI and DME pilot combustion.

combustion which could be achieved with the same valve timings as shown in Table 3.

Fig. 21. Indicated efficiencies by various injection strategies.

4.3. Comparison with SI combustion The third part of experiments is to compare the results of various DME injection strategies combustion with conventional SI using the same engine. Thus, the advantages and drawbacks of the combustion achieved by various DME injection strategy can be discussed. The SI test conditions are shown in Table 3. Since knocking combustion occurred regularly in the single cylinder engine at the CR of 13:1 under high load operations at lambda 1 when the engine was operated in the SI mode, A leaner mixture of k = 1.1 was used to compare the SI and DME pilot

4.3.1. Operation range and combustion stability The COVimep of both CAI and SI combustions are shown in Fig. 22. It can be seen that the operation range, either the highest or the lowest load boundary, has been successfully extended by various DME injection strategy. Under CR 13 and lambda 1.1, the highest load of SI combustion is limited by knocking combustion while the lowest load is limited by misfire due to relatively high iEGR. The highest load of Various DME Injection combustion (net IMEP 8.5 bar under lambda 1.1; 9.13 bar under lambda 1.0) is nearly impossible to be achieved with SI combustion under CR 13 without air boosting. The lower border of various DME injection strategies combustion can be further extended, as mentioned if more iEGR is introduced. Compared with the SI mode, more stable combustion can be achieved with CAI combustion with DME injection. In addition, since flammable DME is used as the ignition source, the flame starts from multiple injection points of DME rather than a point in SI mode, resulting in better combustion stability. 4.3.2. Efficiencies The trends of gross indicated efficiency and net indicated efficiency are shown in Fig. 23. The gross indicated efficiency is generally higher, especially in the low load range in DME split injection combustion mode than it is in conventional SI mode. At higher load range (Net IMEP > 5 bar), the efficiency improvement is insignificant due to the flame

811

35

30

30

25

25

Gross 20

20

Net 15

50

2

4

6

8

40 30 20 10 0 1

15 0

ISCO (g/KWh)

35

Net Indicated Efficiency (%)

Gross Indicated Efficiency (%)

H.F. Zhang et al. / Fuel 107 (2013) 800–814

3

10

5 Net IMEP (bar)

7

9

Net IMEP (bar)

SI Lambda=1.1 SI Lambda 1.1

Fig. 23. Gross and net indicated efficiency (%) as a function of load by SI and DME pilot combustion.

CA at 50% MFB (CA)

400 390 380 370 360 350 1

3

5

7

9

Net IMEP (bar) SI Lambda=1.1

DME Pilot Lambda=1.1

DME Pilot Lamda=1.1

DME Pilot Lambda=1.1

Fig. 24. CA50 as a function of load with SI and DME pilot combustion.

Fig. 26. CO emissions as a function of load with SI and CAI controlled by DME combustion.

However, the net indicated efficiencies of DME pilot combustion is lower than the SI mode in the entire region except in the maximum load point. The difference between gross and net work is the pumping work. In the SI mode, the throttle is used to control the engine load so that the intake pumping loss is increased during part load operation. In the DME pilot mode, the engine is operated at WOT to minimize the intake pumping work, however, as the exhaust pipe is throttled to achieve high internal EGR rate, the exhaust pumping work is significantly increased in the low load operation. In Fig. 23, when the load is higher than 6.5 bar net IMEP, the net indicated efficiency of DME pilot mode is higher than SI mode due to better gross output and the intake pumping work in SI is higher than the exhaust pumping work in DME pilot mode. With the load reduced (net IMEP < 6.5 bar), the increased exhaust pumping work in DME pilot combustion is so large that it losses the benefits of both CAI combustion and WOT operation, resulting in low net indicated efficiencies.

CA10-90 (CA)

50 40 30 20 10 0 1

3

SI Lambda=1.1

5 Net IMEP (bar)

7

9

DME Pilot Lambda=1.1

Fig. 25. CA10–90 as a function of load with SI and DME pilot combustion.

propagation of the entire (SI and maximum load DME Split) or main combustion (mid load DME Split) process in both modes. The combustion phasing (CA50) and combustion duration (CA10– 90) of DME pilot mode are slightly earlier and shorter than SI mode as shown in Figs. 24 and 25 respectively, resulting in slightly higher engine output. However, as the loads decrease (Net IMEP < 5 bar), the improvement of gross indicated efficiency is noticeable with the DME Split injection combustion mode. This is because in the low load DME pilot mode, fast CAI combustion becomes the main combustion process and the benefits of CAI are apparent compared with SI combustion. The earlier and faster combustion of CAI in DME pilot mode can be clearly seen in CA50 and CA10–90 diagrams.

4.3.3. Heat release analysis and emissions Figs. 24 and 25 show the combustion phasing and combustion duration of both modes respectively. The combustion is advanced as load is reduced in both modes when the IMEP is above 4 bar. The CA50 of CAI combustion with DME injection is more advanced than the SI mode, due to fast auto-ignition combustion. The combustion durations in both modes are similar at highest loads. In the case of CAI combustion with DME injection, the combustion duration decreases rapidly as the load is reduced and reaches its minimum at 4 bar IMEP, due to the change in combustion modes and the effects of iEGR as explained above. In comparison, the combustion duration in the SI mode stayed fairly constant at high loads and then increases as the load decreases due to slower flame speed at lower pressure and temperature by the throttled intake. At the minimum load (2.7 bar net IMEP) the SI combustion is extremely long even with the most advanced spark timing. CO, uHC and NOx emissions of both modes are presented in Figs. 26–28. It can be seen that CO emissions are similar for both operations. The DME pilot combustion has the same uHC emissions with SI mode at high load but extremely higher uHC emissions than SI combustion when load reduces. As expected, large reduction in NOx emissions are realised in DME pilot combustion compared to SI combustion. In order to further understand both combustion processes, two specific load points with the same fuelling rate are selected and the results are shown in Table 3 and Figs. 29–32. The conclusions are as follows:

812

H.F. Zhang et al. / Fuel 107 (2013) 800–814

5 4.5

In-cylinder Pressure (bar)

45 40

ISHC (g/Kwh)

35 30 25 20 15 10 5

DME Split SI

4 3.5 3 2.5 2 1.5 1 0.5

0 1

3

5 Net IMEP (bar)

SI Lambda=1.1

7

9

0 0

200

DME Pilot Lambda=1.1

35

isNOx (g/Kwh)

10 8 6 4 2 0 3

5

7

9

In-cylinder Pressure (bar)

12

1

Net IMEP (bar)

SI Lambda=1.1

25 20 15 10 5

Fig. 28. NOx emissions as a function of load with SI and CAI controlled by DME combustion.

340

350

360 370 380 390 Crank Angle (degree)

400

410

420

Fig. 31. Pressure-°CA diagram at constant fuelling rate.

100

DME Split SI

30

DME Split SI

30

0 330

DME Pilot Lambda=1.1

35

600

Fig. 30. Partial P–V diagram at constant fuelling rate.

Fig. 27. uHC emissions as a function of load with SI and CAI controlled by DME combustion.

90 80 70

25

MFB%

In-cylinder Pressure (bar)

400 Volume(cm3)

20

60 50 40

15

30

10

20

DME Split SI

10

5

0 330

0 0

100

200

300

400

500

600

Volume(cm3) Fig. 29. Full P–V diagram at constant fuelling rate.

(1) Fast flame propagation and auto-ignition hybrid combustion is successfully achieved with the assistance of DME injection while the conventional SI combustion is characterised as slow flame propagation process. (2) The CAI combustion with DME injection is more close to constant volume combustion than the SI mode and able to achieve higher output and better efficiencies with a same fuelling rate (Fig. 29).

340

350

360 370 380 390 Crank Angle (degree)

400

410

420

Fig. 32. MFB-°CA diagram at constant fuelling rate.

(3) As the SI mode presented here is operated at 63 °CA valve overlap, the intake pressure is higher than nature aspired pressure and the exhaust gas are re-breathed from the exhaust pipe to intake pipe. Thus, iEGR are also included in SI mode and the drawbacks of intake throttle are not apparent. The exhaust throttling significantly increases the exhaust back pressure, which induces large exhaust pumping work during the exhaust stroke (Fig. 30). (4) In the CAI combustion mode with DME injection, the peak pressure is always higher than SI due higher peak heat release rate (Fig. 31).

813

H.F. Zhang et al. / Fuel 107 (2013) 800–814 Table 4 Operation conditions and result of DME split injection and gasoline SI combustion. Parameters

DME Split

SI

Geometric compression ratio Effective compression ratio Lambda iEGR (%) Intake throttle (%) SOI2 or SI timing (°CA) COVimep Fuel MEP (bar) Gross IMEP Net IMEP Pumping MEP Combustion efficiency (%) Gross indicated efficiency Net indicated efficiency Thermal efficiency CA 10 (CA°) CA 50 (CA°) CA 90 (CA°) CA 10–90 (CA°) CO (%) uHC (ppm) NOx (ppm)

13

13

12.92 1.1 31 WOT 347 2.12 15.3 4.55 3.47 1.09 95.24 29.7 22.6 31.19 363 372 379 16 0.27 2035 167

12.46 1.1 6 8 335 2.9 14.9 4.15 3.78 0.36 97.33 27.74 25.3 28.5 360 378 395 35 0.26 1185 1702

ISCO (g/KW h) ISHC (g/KW h) ISNOx (g/KW h)

14.8 17.6 0.98

sion in uHC. Relatively higher gross efficiency was obtained than SI combustion, but net indicated efficiency was lower than SI due to high pumping loss from exhaust throttle. Further research is currently carried out to obtain internal EGR with residual gas re-breathing method, which is expected to lead to dramatic increase of the net indicated efficiency as well as maintaining the low level of NOx emission. Appendix A Determination of overall AFR with measurement of exhaust components [12].

½CO2  þ

NO 2

 þ

Hcv 4



Kp ½CO

K p þ½CO



  ð½CO2  þ ½COÞ  Ocv 2

where [XX] is the gas concentration in volume, Hcv is the atomic ratio of hydrogen to carbon in the fuel, Ocv is the atomic ratio of oxygen to carbon in the fuel, K1 is the number of carbon atoms in each of the HC molecules being measured (here all uHC is considered as methane so that K1 = 1); Kp is the assumed relation of CO2, CO, H2O and H2 concentrations.

Kp ¼ (5) In Fig. 32, it can be seen that the heat release of SI starts earlier and last longer during the combustion process. However, the CAI combustion with DME injection is characterised with a hybrid mode of flame propagation and multiple auto-ignition combustion. The transition point, T, between the two combustion modes is at 369 °CA degree and the ratio of flame propagation and auto-ignition is 18.5%/81.5% (Fig. 32). (6) As shown in Table 4, the gross indicated efficiency of CAI combustion with DME injection is higher than that of SI mode. But the net indicated efficiency of CAI combustion with DME injection is lower than SI mode because of its greater pumping loss. (7) At the same fuelling rate, CO emissions are similar but uHC emissions from DME pilot mode are as twice as they are in SI mode, resulting in relatively low combustion efficiency. However, over 90% reduction of NOx emissions are achieved in DME pilot mode.

2

þ ½O2  þ

2   1 þ H4cv  O2cv  ð½CO2  þ ½CO þ K1  ½HCÞ

k¼

10.2 7.36 7.19

CO

½CO  ½H2 O ¼ 3:5 ½CO2   ½H2 

To specify Hcv and Ocv, the total intake fuel mixture including gasoline and DME is considered as a single hydrocarbon with balanced subscripts chemical formula CaHbOc. The value a, b and c vary with different fuel contents. Thus, the H/C ratio and O/C ratio of 0% (by mass) DME and 100% gasoline is 1.87 and 0 respectively. The 1% DME and 99% gasoline have the H/C and O/C ratio 1.88 and 0.003 etc. until 100% DME and 0% gasoline refer to 3 and 0.5. The stoichiometric AFR varies from 14.6 for a pure gasoline limit to 9 for a pure DME limit. Therefore, all 100 possible k values are calculated with 100 possible H/C and O/C input by spreadsheet. The k varies within a narrow range. For example, Exhaust

CO2

Concentration k Range

10.38% 0.24% 1.336–1.327

CO

uHC

NOx

O2

2257 ppm

49.1 ppm

5.84%

5. Conclusions In this paper, the Positive Valve Overlap together with exhaust throttling, and gasoline-DME duel fuel strategies were investigated in a single cylinder engine to investigate the effects of DME injection strategies in the enlargement of CAI combustion operating range and the control of combustion phasing. Three injection strategies, early single injection, split injection and late single injection are used to directly inject DME into cylinder. Direct injected DME is used as the ignition source or to increase the auto-ignition quality of charge mixture in the cylinder. It is demonstrated that the combustion characteristic is changed by DME injection strategy and it can be used to extend the CAI combustion operating range. Early single injection leads to pure CAI combustion, split injection results in hybrid combustion of initial flame propagation followed by faster CAI combustion, whilst only flame propagation is observed in the late single injection strategy. The engine operating range was extended and achieved from 1.3 bar IMEP to 9.1 bar IMEP at 1500 rpm engine speed without intake charge heating. At the same fuelling rate with SI combustion engine, over 90% reduction in NOx emission can be achieved with similar CO emissions but almost twice high emis-

The chemical formula of DME is C2H6O. If this one oxygen atom is assumed as being consumed by two hydrogen atoms, the H/C ratio of the hydrocarbon left is 2, which is similar as the H/C ratio of gasoline (1.876). In this example, the k value is determined as 1.3.Therefore, the final lambda value is identified by this method. References [1] Zhao H, Peng Z, Williams J, Ladommatos N. Understanding the effect of recycled burned gases on the controlled auto-ignition (CAI) combustion in four-stroke gasoline engines. SAE 2001-01-3607; 2001. [2] Yang C, Zhao H, Megaritis T. Investigation of CAI combustion with positive valve overlap and enlargement of CAI operating range. SAE 2009-01-1104; 2009. [3] Hyvonen J, Haraldsson G, Johansson B. Operating range in a multi cylinder HCCI engine using variable compressions ratio. SAE 2003-01-1829; 2003. [4] Inagaki K, Fuyuto T, Nishikawa K, Nakakita. Dual-fuel PCI combustion controlled by in-cylinder stratification of ignitability. SAE 2006-01-0028; 2006. [5] Li J, Zhao H, Ladommatos N, Ma T. Research and development of controlled auto-ignition combustion in a four-stroke multi-cylinder gasoline engine SAE 2001-01-3608; 2001. [6] Manente V. Gasoline partially premixed combustion, an advanced internal combustion engine concept aimed to high efficiency, low emissions and low acoustic noise in the whole load range, PhD thesis, division of combustion engines. Department of Energy Sciences, Lund Institute of Technology; 2010.

814

H.F. Zhang et al. / Fuel 107 (2013) 800–814

[7] Hanson R, Splitter D, Reitz R. Operating a heavy-duty direct-injection compression-ignition engine with gasoline for low emissions. SAE 2009-011442; 2009. [8] Hanson RM, Kokjhon SL, Splitter DA, Reitz Rolf D. An experimental investigation of fuel reactivity controlled PCCI combustion in a heavy-duty engine. SAE 2010-01-0864; 2010. [9] Yeom K, Jang J, Bae C. Gasoline HCCI engine with dme (di-methyl ether) as an ignition promoter. In: Proc 6th international conference on GDI engines ‘‘Direkteinspritzung im Ottomotor/Gasoline Direct Injection Engines’’, expert verlag; 2005. p. 229–39. [10] Lavy J, Dadadie JC, Angelberger C, Duret P, Willand J, Juretzka A, Schaflein J, Ma T, Lendresse Y, Satre A, Schulz C, Kramer H, Zhao H, Damiano L. Innovative ultra-low NOx controlled auto-ignition combustion process for gasoline engines: the 4-SPACE project. SAE 2000-01-1837; 2000. [11] Bessonette PW, Schleyer CH, Duffy KP, Hardy WL, Liechty MP. Effects of fuel property changes on heavy-duty HCCI combustion. SAE 2007-01-0191; 2007.

[12] Brettschneider, Johannes. Berechung des Luftverhaeltnisses k von LuftKraftstoff-Gemsichen und des Einflusses on Messfehlern auf k, Bosch Technische Berichte (1979), Band 6, Heft 4, Seite 177–186, Stuttgat. [13] Andreae MM, Cheng WK, Kenney T, Yang J. On HCCI engine knock. SAE 200701-1858; 2007. [14] Zhao H, Lowry G, Ladommatos N. Time-resolved measurements and analysis of in-cylinder gases and particulates in compression-ignition engines. SAE 961168; 1996. [15] Oakley A, Zhao H, Ladommatos N. Experimental studies on controlled autoignition (CAI) combustion of gasoline in a 4-stroke engine. SAE 2001-01-1030; 2001. [16] Zhang Y, Zhao H, Xie H, Hou S, Yang C, Characterization and heat release model development of SI-CAI hybrid combustion and its application to a 4-stroke gasoline engine operating with highly diluted mixture. IC engine and gas turbine combustion; 2010. [17] Silvis W. An Algorithm for calculation the air/fuel ratio exhaust emissions. SAE Paper 970514; 1997.