Assessing the Hydrocarbon Emissions in a Homogeneous ... Injection Spark Ignited Engine

Assessing the Hydrocarbon Emissions in a Homogeneous Direct

Injection Spark Ignited Engine

by

Michael S. Radovanovic

B.S., Mechanical Engineering

Purdue University, 2004

SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING IN

PARTIAL FULFULLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE IN MECHANICAL ENGINEERING

AT THE

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

August 2006

0 2006 Massachusetts Institute of Technology

All Rights Reserved

Signature of Author:

Certified by:

Accepted by:

Department of Mechanical Engineering

August 11, 2006

Wai K. Cheng

Professor of Mechanical Engineering

Thesis Supervisor

Lallit Anand

Chairman, Departmental Graduate Committee

MASSACHUSETTS IN$T1IUTE.

OF TECHNOLOGY

JAN 2 3

2007

LIBRARIES

2

Assessing the Hydrocarbon Emissions in a Homogeneous Direct

Injection Spark Ignited Engine

By

Michael S. Radovanovic

Submitted to the Department of Mechanical Engineering

On August 11, 2006 in Partial Fulfillment of the

Requirements for the Degree of Masters of Science in

Mechanical Engineering

Abstract

For the purpose of researching hydrocarbon (HC) emissions in a direct-injection spark ignited (DISI) engine, five experiments were performed. These experiments clarified the role of coolant temperature, injection pressure, and injection timing in HC emissions; the final two experiments illustrated the effect of coolant temperature and injection pressure on separate sweeps of injection timing and the subsequent HC levels. The first three experiments were performed with isopentane. All five of the experiments were repeated with two fuels: UTG 91, a typical research gasoline, and a fuel with a high driveability index (DI), i.e. a less volatile fuel.

The results showed less-than intuitive results for the response of HC to varying coolant temperature and varying injection pressure. For the coolant temperature data, the deviation from intuition is discussed and is probably due to vaporization problems. For the injection pressure results, the counterintuitive trend is expected to be the balance of two negative effects of high and low fuel pressure: high droplet velocities and large droplet diameters.

Finally, the injection timing results were more logical. The early injections are high for this engine due to late exhaust valve closing, and the late injections have high HC because of a decreasing time to vaporize and poor mixing caused by the lack of intake air motion.

Thesis Supervisor: Wai K. Cheng

Title: Professor of Mechanical Engineering

3

4

Acknowledgements

For the opportunity to learn at MIT, I'm thankful to the group of automotive and oil companies which funded my research and education. Raymond Phan and Thane DeWitt, on staff at the lab, were critical to setting up my project; without them, I could not have completed my project. In particular, I am appreciative of Professor Cheng. The last month has included enjoyable and humorous moments as we attempted to combine our intellects and understanding to explain the data. "Before you leave MIT you'll learn that everything is rational," as he rapped a check valve with a wrench to fix my fuel supply problem.

I have made good friends in the lab and having this close-knit group of people made my stay at MIT more bearable. Vince Costanzo, Alex Sappok, and Kevin Lang provided especially amiable company. Developing black and white photos in the SAA and making projects in Hobby Shop were wonderful outlets for me, and they played an important role in several pleasant memories at MIT.

Even more critical to my state-of-mind has been the support, prayers and social interaction

I've received from numerous people from Hope Fellowship Church. I've yet to meet a group of people who care so much for the well-being of others, for their temporary needs and their eternal ones. Leaving this community will definitely be the hardest part about leaving Boston. "And now, 0 Israel, what does the Lord your God ask of you but to fear the Lord your God, to walk in all his ways, to love him, to serve the Lord your God with all your heart and with all your soul, and to observe the Lord's commands..." Deuteronomy

10: 12-13a.

A few friends deserve special mention: Clay Edens, for weekly conversations, Jeffrey D.

Robinson, for funny voicemail messages, Chris Sullivan, for sharing food and guitar interests, Randy Ewoldt, for tremendously enjoyable lunches, and Mila K. Radovanovic, for her concern for me and for my fashion sense or lack thereof.

My parents are worthy of my most sincere thanks: for their sacrifices in raising their children, for disciplining me, for instilling the importance of family, for adjusting my oftwandering ego, and for their unconditional love.

5

6

Biographical Note

Michael S. Radovanovic accepted a position with Caterpillar Inc. in Mossville, IL in the

Engine Research Division after finishing his M.S. in mechanical engineering (2006) from the Massachusetts Institute of Technology. He was awarded honorable mention for the

National Science Foundation fellowship in 2004. His undergraduate degree in mechanical engineering was obtained with honors from Purdue University in 2004, and he was selected to participate in the Department of Mechanical Engineering's GEARE (Global Engineering

Alliance for Research and Education) program, through which he studied and worked in

Germany.

7

8

Table of Contents

Abstract..............................................................................................

Acknowledgem ents............................................................................

Biographical Note..............................................................................

Nom enclature.....................................................................................

3

5

List of Figures .....................................................................................

List of Tables.......................................................................................

List of Appendices ..............................................................................

1 Introduction ................................................................................

16

17

2

1.1 Background............................................................................18

1.1.1 Production H istory ....................................................... 19

1.1.2 Efficiency ....................................................................... 22

1.1.3 H ydrocarbon Em issions................................................ 23

1.1.4 V aporization...................................................................

Experim ental Setup.....................................................................

24

35

2.1 Instrum entation..................................................................

2.1.1 Air Flowm eter ................................................................

38

38

2.1.2 In-Cylinder Pressure Transducer................................38

2.1.3 Lam bda M eter ............................................................. 39

2.1.4 Fast Flame Ionization Detector ................................... 40

2.2 System s ..................................................................................

2.2.1 Intake System ................................................................

40

40

2.2.2 Fuel System .................................................................... 41

2.2.3 Coolant System ............................................................. 42

2.2.4 Exhaust System .............................................................. 44

2.2.5 Engine Control System .................................................. 44

3

2.2.6 D ata A cquisition System ............................................... 45

R esults........................................................................................... 51

7

11

12

15

9

3.1

Coolant Tem perature Experim ent..................................... 54

3.2

Injection Pressure Experim ent............................................

60

3.3

Injection Tim ing Experim ent .............................................

62

3.3.1

Injection Timing Experiment with Varying Coolant

Tem peratures..............................................................................

71

3.3.2

Injection Timing Experiment with Varying Injection

Pressures....................................................................................... 73

4 Sum m ary ....................................................................................

76

4.1

Conclusions............................................................................ 76

4.2 Future Im provem ents .........................................................

77

4.3 Future Experim ents..............................................................

78

References ...........................................................................................

80

Appendix .............................................................................................

82

10

MAP

MEP

MPI

NOx

PFI

PM

PROCO

RTD

DISI

EOI

EVC

EVO

FFID

GDI

HC

ISFC

IVC

IVO aTDC / aBDC

BDC

BSFC

bTDC / bBDC

CAD

DI

DI

SMD

SOI

TDC

Nomenclature

After top dead center / after bottom dead center

Bottom dead center

Brake specific fuel consumption

Before top dead center / before bottom dead center

Crank angle degrees

Direct injection

Driveability index

Direct injection spark ignited

End of injection

Exhaust valve closing

Exhaust valve opening

Fast flame ionization detector

Gasoline direct injection

Hydrocarbons

Indicated specific fuel consumption

Intake valve closing

Intake valve opening

Manifold absolute pressure

Mean Effective Pressure

Multiport fuel injection

Oxides of nitrogen

Port fuel injection

Particulate Matter

Ford programmed combustion control system

Resistance temperature detector

Sauter-mean diameter

Start of injection

Top dead center

11

List of Figures

Figure 1-1 Comparison of indicated specific fuel consumption (ISFC) between throttled and unthrottled operation as a function of engine load............................................. 20

Figure 1-2 Worldwide sulfur concentration in gasoline (data published in 1998)........... 21

Figure 1-3 Variation of air/fuel ratio during the 11th cycle of the LA-4 mod vehicle test.. 21

Figure 1-4 Comparison of knock-limited spark advance between the GDI and PFI en g in es.......................................................................................................................... 2 3

Figure 1-5 Droplet life time under varying conditions ................................................... 25

Figure 1-6 Atom ization characteristics ............................................................................ 25

Figure 1-7 The effect of spray-tip velocity on Sauter Mean Diameter ............................. 26

Figure 1-8 Comparison of drop size distributions for sprays from swirl-type and holetyp e injectors ................................................................................................................ 2 7

Figure 1-9 HC concentration trace for homogeneous charge and stratified charge operation with Japanese fuel (mean of 250 cycles)................................................. 27

Figure 1-10 HC concentration trace for homogeneous charge operation with 1) prevaporized iso-Pentane and, 2) in-cylinder injection of Japanese fuel (mean of 250 cy cles)........................................................................................................................... 2 8

Figure 1-11 Difference in fuel behavior between the PFI and DISI engine. ................... 29

Figure 1-12 Verification of the effectiveness in reducing HC .......................................... 30

Figure 1-13 Engine-out HC behavior............................................................................... 30

Figure 1-14 Effect of hot operation of a multihole injector for four levels of ambient back pressure; six-hole nozzle; 500 spray; fuel: indolene; fuel temperature: 90'C; fuel injection pressure: 11.0 Mpa; amount of fuel injected: 10 mg per injection. ....... 31

12

Figure 1-15 Effects of back pressure (Pamb) and fuel rail pressure (Pinj) on the crosssections of GDI sprays from a swirl injector........................................................... 32

Figure 1-16 Effect of swirl-injector operating temperature on spray development with iso-octane at 20'C, 75

0

C, and 100

0

C........................................................................ 33

Figure 1-17 Effect of swirl injector operating temperature on spray development with indolene at 20"C, 75 0

C, and 100

0

C. .........................................................................

Figure 1-18 Spray SMD profiles for three injector operating temperatures.....................33

33

Figure 2-1 1Combustion chamber; injector and spark plug location................................36

Figure 2-2 Piston shape.................................................................................................... 37

Figure 2-3 Fuel system diagram ..................................................................................... 41

Figure 2-4 Coolant system diagram ......................................................................................

Figure 2-5 D ata acquisition system .................................................................................

Figure 3-1 Coolant temperature sweep.............................................................................55

Figure 3-2 Coolant temperature sweep with percent reduction caused by isopentane.........56

43

45

Figure 3-3 Hot oil test with old and new oils ...................................................................

Figure 3-4 Injection Pressure sweep.................................................................................60

58

Figure 3-5 Spray penetration vs time after SOI, with injection pressures of 30, 70, and

120 bar C Siem ens V D O .......................................................................................... 61

Figure 3-6 End of injection hydrocarbon sweep...............................................................63

Figure 3-7 End of injection hydrocarbon sweep with piston position and velocity ...... 67

Figure 3-8 Experimental droplet penetration and model equation ................................... 68

Figure 3-9 Droplet CAD delay from leaving injector until piston impingement ............. 69

Figure 3-10 Dependence of full load performance on injection timing .......................... 70

13

Figure 3-11 Injection timing sweep with varying coolant temperatures and UTG 91 fuel .

71

Figure 3-12 Injection timing sweep with varying coolant temperatures and High DI fuel .

72

Figure 3-13 Injection timing sweep with varying fuel pressures and UTG 91 fuel...... 73

Figure 3-14 Injection timing sweep with varying injection pressures and High DI fuel..... 74

14

List of Tables

Table 1:1 Major factors contributing to the BSFC advantage of a G-DI engine as compared to that of a conventional PFI engine ........................................................ 22

Table 1:2 Negative factors that moderate the fuel economy potential of a stratifiedcharge G -D I engine.................................................................................................. 24

Table 2:1 Engine geometry details ................................................................................... 35

Table 2:2 Data acquisition components and functions ................................................... 46

Table 2:3 Fuel filling procedure ........................................................................................ 47

Table 2:4 Engine motoring procedure ...............................................................................

Table 2:5 Engine running procedure .................................................................................

48

Table 2:6 Engine shut-down procedure ............................................................................

Table 3:1 Baseline operating conditions..........................................................................

Table 3:2 Engine parameter experiments ..........................................................................

49

50

51

52

Table 3:3 Chevron Phillips UTG 91 fuel..........................................................................53

Table 3:4 Reasoning for coolant temperature experiment expectations...........................54

15

List of Appendices

Appendix- 1 Valve timing location in the pressure trace................................................. 82

Appendix- 2 Injector calibration based on the nitrogen regulator pressure value ....... 83

Appendix- 3 Exhaust runner time-resolved HC concentration measurements for spark timing of 150 bTDC at 3.0 bar Net-IMEP, X=1.0, CMCP, and 90'C fluids [18]. ........ 83

Appendix- 4 Comparison of engine operating conditions and data between PFI and DISI en g in es [18].................................................................................................................. 84

Appendix- 5 Drop vaporization lifetime versus surface temperature [23]....................... 84

Appendix- 6 Spray penetration picture at different injection pressures Siemens VDO... 84

Appendix- 7 Typical cylinder pressure bounding the range of injection pressures......... 85

Appendix- 8 Comparison of injection timing sweep data with two additional papers ........ 85

Appendix- 9 Operating conditions for Appendix- 8 ....................................................... 86

Appendix- 10 Volumetric efficiency and map changes with injection timing (taken with

UTG 91 fuel at baseline conditions) ....................................................................... 86

Appendix- 11 Injection timing sweep with low coolant temperatures and varying fuels.... 87

Appendix- 12 Injection timing sweep with high coolant temperatures and varying fuels... 87

Appendix- 13 Injection timing sweep with lower injection pressures and varying fuels .... 88

Appendix- 14 Injection timing sweep with higher injection pressures and varying fuels ... 88

16

1

Introduction

With 17 million passenger cars sold every year in the US and crude oil supplies expected to last about 40 more years, researchers explore every possible improvement for the current internal combustion engine. Meanwhile, record high prices of crude oil are above

$50/barrel for the first time since 1981. Thus, improving engine efficiency is essential to our accustomed culture of mobility. Research has directly shown air quality deterioration caused by transportation sources. The primary smog producing emissions that need to be reduced from vehicle emissions are hydrocarbons (HC) and oxides of nitrogen (NOx). So the automotive industry must solve the following conundrum: how to design a power plant which will increase fuel economy and reduce emissions? The gasoline direct injection (DI) engine is a viable option for fuel economy improvement in the near future. To assess its environmental impact, the HC emissions of a gasoline DI engine must be researched to continually improve emissions.

While stratified engines exhibit better fuel economy than port fuel injection (PFI) ones, they do not satisfy the US emissions regulations. On the other hand, homogeneous DI engines, which operate with a stoichiometric fuel mixture, can utilize three-way catalysts and hence, meet the US emissions requirements.

The potential benefits of gasoline DI engines over conventional PFI engines motivate this research. In current port fuel injected engines, the fuel is mostly vaporized by heat from the intake port and valve surfaces. However in gasoline DI engines, the heat from the intake

17

air vaporizes most of the fuel, thereby cooling the charge. This charge cooling enables various improvements such as higher compression ratios, undersizing engines, and turbocharging. Finally, by directly delivering the fuel into the cylinder, the precise fuel metering allows many possibilities for optimizing fuel economy and simultaneously reducing emissions.

Since this method of fuel delivery to the engine could produce substantial unburned fuel in the combustion process, the hydrocarbon emissions are a particular concern for gasoline DI engines. Conventional engines, with port fuel injection, have been researched more extensively. For example, Cheng et al. published which mechanisms affect the HC emission levels, and he expounded on their relative affect [1]. First, the crevices, located around the piston and ring pack, the cylinder head gasket, and spark plug, were estimated to contribute about 38% of the HC emissions. Second, the oil layers contribute about 16% of the HC emissions, because these layers can absorb hydrocarbons; after the combustion event, they are desorbed and transported into the exhaust. Furthermore, deposits contribute an additional 16% by shielding HC from the combustion in the deposit's porosity. Finally, the presence of liquid fuel in the cylinder adds 20% to HC emissions. Of course, these mechanisms will be different with direct injection.

1.1 Background

In light of these benefits and the potential for improvement, companies and researchers have completed numerous studies to validate and further understand direct injection spark ignited (DISI) engines. This section will highlight the history of production DISI engines and the most relevant research results for a broad base of understanding for the reader.

18

1.1.1 Production History

Like variable valve timing and several other concepts in the automotive industry, the idea of direct injection gasoline engines has existed for many years. As early as the 1940s, radial aircraft engines operated with direct injection [2]. The Benz 300SL, produced in

1954, was one of the earliest homogeneous DISI engines. Later in the 1960s and 1970s, companies became more interested in the fuel economy benefits of stratification with examples like the MAN-FM and Ford PROCO (Ford programmed combustion control system) systems [2]. In both systems, the specific power output was below the competing

PFI engines, and both engines had problems with higher HC emissions at light loads, as a result of the relatively poor air utilization.

The next significant wave of commercial DISI engines came in 1996 with Mitsubishi's 1.8

L Galant series engine, which was sold in Asia. Mitsubishi coined the term GDI, gasoline direct injection, and they operated the engine in multiple stages with varied injection timing and mixture air/fuel ratios. For the majority of its operating map, the engine burned stoichiometric air/fuel ratios, but it utilized homogenous, lean charges (air/fuel ratios between 20 and 25) during light load operation. In addition, Mitsubishi designed the system with two catalysts; the upstream iridium catalyst was effective for the lean mixtures while the typical platinum catalyst was placed downstream. Afterwards, Toyota, Nissan,

Ford, Opel, and Volkswagen among others, have developed and marketed stratified gasoline DI engines in Asia and Europe. The benefits of unthrottled operation is shown below in Figure 1-1[3].

19

450

400

31 350 v300

-Throtfld Opration

Unthr.UW4Opra I...

200

0 2 4 6

Wep bar

8 10 12

Figure 1-1 Comparison of indicated specific fuel consumption (ISFC) between throttled and unthrottled operation as a function of engine load.

Observers of the DISI trend in these continents will wonder why the trend has not swept into the US with similar speed. One of the reasons for this delay comes from the resulting exhaust of lean operation; the optimum three-way catalyst conversion efficiency of NO,

CO, and HC hovers around stoichiometric air/fuel ratios, +/- 0.1 air/fuel ratios [4]. This tight band of operation makes the three-way catalysts ineffective for the operating points that these Asian and European calibrations use. Consequently, the DISI engines in these markets take advantage of lean NOx catalysts, also known as selective reduction catalysts.

More important than the higher cost as a barrier to entering the American market, these catalysts are poisoned by fuels with higher sulfur content, which varies significantly with different countries as shown below in Figure 1-2[5].

20

1000

PetmCv~

SR ular Gaso**e

800-

II

0

CC

Figure 1-2 Worldwide sulfur concentration in gasoline (data published in 1998)

Another benefit of direct injection is the improved combustion stability as shown in reduced air/fuel ratio variation in Figure 1-3 [15].

0.9

0. 8

2

Ap IMd fitI fI

trrt

It

IF VII f

0. 9 nl

Time (se)

Figure 1-3 Variation of air/fuel ratio during the 11th cycle of the LA-4 mod vehicle test

This air/fuel ratio data was obtained with an Isuzu GDI for the LA-4 mode vehicle test, and the reduced variation is readily apparent as proof of the precise fuel metering capability provided by the DISI system.

21

1.1.2 Efficiency

While many publications tout the fuel consumption improvements with direct injection engines, the compromises, and hence efficiency decreases from the theoretical maximum, needed to meet the emissions standards cannot be overlooked. A part load fuel consumption reduction of 20-25% has been reported with charge stratification, depending on test conditions and optimization [2]. Zhao, et al. provided the factors contributing to this improvement in Table 1:1.

Table 1:1 Major factors contributing to the BSFC advantage of a G-DI engine as compared to that of a conventional PFI engine

Pumping Loss Reduced pumping losses due to unthrottled part-load operation using overall lean mixtures

In-Cylinder Charge Increased cooling of the intake charge due to in-cylinder injection

Cooling during induction

Compression Ratio Increased knock-limited compression ratio due to a lower end-gas temperature

Specific Heat

Ratio

Increased cycle efficiency due to incrementally higher specific heat ratio of lean mixtures

Heat Loss Reduced cylinder wall and combustion chamber heat losses due to stratified-charge combustion

Fuel Cut-Off Fuel cut-off during vehicle deceleration

As mentioned previously, the charge cooling provided by the direct injection is a significant means for improving the fuel consumption of combustion engines. Since a cooler charge can increase the knock-limited compression ratio by two full ratios [6], the subsequent effect is an increased fuel conversion efficiency of 2.7% for a compression ratio increase from 10 to 12. Furthermore, Andriesse et al. estimated the fuel economy benefits with a homogeneous DISI system to be 7-10% [7]. The largest component (3-4% improvement) is

22

the potential for downsizing the engine while providing the same power and torque. Other improvements could be made by higher EGR capability (1.5-2%) and optimized fuel/air metering during warm-up and transient operation.

Of course, the charge cooling effect is highly dependent on several parameters, one of which is the start of injection (SOI) timing. In general, as the injection is advanced, the effect is magnified since the cooler charge has less time to absorb heat from the cylinder walls. As a result, the potential for efficiency improvements due to charge cooling is proportional to the SOI advance, shown in Figure 1-4[8].

W6T. 150 rpM, AF 12-5. S1 RON.

j4

2

-Tl~os mmnstconn

82Clntak,) L

-360 -30 -300 -270 -240 -210 -180 -150 -120

SO timng (CAD)

Figure 1-4 Comparison of knock-limited spark advance between the GDI and PFI engines

As the SOI is advanced, the charge is increasingly cooler and more resistant to knocking.

Thus, the knock-limited spark can be advanced.

1.1.3 Hydrocarbon Emissions

In the compilation by Zhao et al. [2], the authors include Table 1:1.

23

Table 1:2 Negative factors that moderate the fuel economy potential of a stratified-charge G-DI engine

Emissions Emissions constraints that may not allow the engine to operate at the optimum brake specific fuel consumption (BSFC) point

Fuel Consumption Fuel consumption penalty due to high-pressure pump and/or air

Combustion compressor for air-assisted injection

Negative impact of increased surface-to-volume ratio of a stratified-

Chamber Design charge combustion system due to an increased compression ratio coupled with a more complex piston geometry

The compromise between fuel economy benefits and reduced emissions will affect the ability to reduce throttling losses. Throttling can be reduced by lean operation of the engine, but the stability of combustion and lower exhaust temperatures limit the range of effective lean operation. Zhao et al. cited that light throttling is an effective means for controlling HC emissions while slightly decreasing fuel economy. These issues begin to illustrate the disparity between the theoretical efficiency improvement of 20-25% and the realized improvement of 10-12% [2].

1.1.4 Vaporization

One of the concerns with HC emissions is that sufficient vaporization may not occur before combustion. Vaporization can be related to the temperature of the charge, demonstrated in

Figure 1-5 [8].

24

iOW Inte IroI

O.300 40 S 600

Air tempratur, at IVC P(

700 $00so

Figure 1-5 Droplet life time under varying conditions

Saturation vapor pressure curves of normal paraffins can also illustrate the ease with which most elements will be at boiling conditions at standard cylinder wall temperatures. Despite this boiling, total vaporization is not guaranteed as discussed in the Coolant Temperature

Experiment section.

Similarly to charge temperature, the fuel pressure is another factor affecting the potential for vaporization since the drop velocity and size are sensitive to this parameter, shown in

Figure 1-6 [9].

25

20

15

10 5

10

Fuel Pressure(MPa)

15

Figure 1-6 Atomization characteristics

25

This data was taken with a Toyota swirl injector so the absolute values for Sauter Mean

Diameter (SMD) and pressure may not be representative for a different design; logically, regardless of the particular injector design, the SMD will reach a lower bound despite further increases in fuel pressure. This general trend is illustrated with a large variety of injectors and fuel pressures in Figure 1-7 [11].

0 20 30 40 so 7

Figure 1-7 The effect of spray-tip velocity on Sauter Mean Diameter

Despite the common usage of SMD as a metric portraying the useful droplet size, other values such as DV90 prove more useful in quantifying the spray's likelihood to vaporize, which heavily affects the HC emissions. The DV90 value is a function of the statistical droplet size distribution; it is the droplet diameter under which 90% of the volume is comprised with smaller droplets. The reason the DV90 can be more appropriate for quantifying the fuel spray is illustrated in the following example: "Each 50-pm fuel droplet

in a spray size distribution having an SMD of 25-gm not only has eight times the fuel mass of the mean droplet, but also will remain as a liquid long after the 25-prm drop has evaporated. In fact, it is very informative to consider that when all of the 25-pm droplets are evaporated, the original 50-Rm droplets will still have a diameter of 47-pm. [2]"

26

Furthermore, injectors cannot be assumed to have similar distributions as illustrated below in Figure 1-8 [10]

-- of. nozzle SMDO-19.6 pM

0.05 ... Swi nozzlesmD-16s11m,

C

0.043.....

0.02

S0.00

0. 2O 40 80 SO 100

Droplet diameter (pim)

Figure 1-8 Comparison of drop size distributions for sprays from swirl-type and hole-type injectors

Another research institution tested a DISI system with a wall-guided combustion system, which uses the piston to direct the fuel towards the spark plug. Of course, the stratified experiment was expected to have high HC emissions with overmixing, i.e. too lean of a mixture, at the edge of the air/fuel cloud and additional problems with undermixing in the center of the spray jet and in the case of pool burning [11]. The results obtained comparing homogeneous and stratified emissions are shown in Figure 1-9.

4000

3000-

.2000

1000

TDC

EVO IVO VC

--

--

Homogeneous

Stratiled

-180 TDC 180

Crank position [CAD'ATDCJ

TDC

Figure 1-9 HC concentration trace for homogeneous charge and stratified charge operation with

Japanese fuel (mean of 250 cycles).

27

The explanation for the peak in homogeneous operation is the roll-up vortex as the piston scrapes the cylinder wall and transports the unburned HC, which escaped the combustion event through the crevice and oil desorption mechanisms. Another experiment was performed to assess the role of liquid fuel in the HC trace, Figure 1-10.

4000

3000

E

EL2000

1000

TDC

EVO ~ IVO

EVC

-

Japanese

[so-P entane

-180 TDC 180

Crank position [CAD ATDC

TDC

Figure 1-10 HC concentration trace for homogeneous charge operation with 1) pre-vaporized iso-

Pentane and, 2) in-cylinder injection of Japanese fuel (mean of 250 cycles).

In this case, the significant difference after exhaust valve opening (EVO) is credited to piston impingement due to an early injection time of 30 CAD aTDC.

Another contribution to the literature included a review and comparison of the HC capabilities between the modern PFI engine and the DISI engine. Because of the poor vaporization with cold port surfaces, the standard PFI engine injects much more fuel than would be needed for a stoichiometric mixture, knowing that only a small fraction will vaporize and be transported to the cylinder. A significant amount of fuel remains on the port surfaces, creating a puddle of fuel as shown by the large amount of "port wetting fuel

[12]." Note that multiport fuel injection (MPI) and port fuel injection (PFI) are used interchangeably in this text.

28

Engine-out HC

200

150 port wetig fuel

Burned fe

Cylinder wetting

0

DISI MPI

Figure 1-11 Difference in fuel behavior between the PFI and DISI engine.

As shown in Figure 1-11 above, despite injecting quite a bit less volume of fuel, the DISI still transports more HC emissions from the engine. Furthering the comparison, the HC emissions were plotted for the DISI engine, after being normalized to a value of 100. On the other hand, the PFI emissions level was approximately 60% of that value. In order to understand the difficulty in achieving lower emissions with a DISI engine, several methods were evaluated. The first method eliminated the injection pressure transience that normally enters the process because of the fuel pump's reliance on the mechanical speed of the engine. Another method involved a late opening of the intake valve, which increases turbulence and mixing of the fuel. The third method combined the first method with the addition of late injection, which aids the vaporization of the fuel because of the warmer air charge due to its compression. The final case included three advancements: a steady state fuel pressure, late injection, and heated fuel injection. The heated fuel injection aids the vaporization even more than the other methods, although its application in production automobiles is unlikely.

29

au

0

1 2

LConventional DISI

2.Engine Starting W/5MPa

3Late intake valve opening

3 4 5 6

4.2 + late injection

5: 4 + heated fuel in ection

6: MPI

400(

C

S

0.

0.

U

300(

2000

100

0

Figure 1-12 Verification of the effectiveness in reducing HC

As illustrated above in Figure 1-12, each case after the conventional DISI improves the HC emissions, but only the most complicated and expensive option can compete with the production PFI engine. As a result, manufacturers are interested in, if not skeptical of, the capability to reduce HC emissions in the DISI engine beyond the current levels achieved with the PFI engine.

Continuing the experiments above, Koga et al. reviewed the effect of early and late injection timing in a DISI engine. On a time scale, these two cases were examined, see

Figure 1-13, to reveal how the HC escapes the combustion event.

Exhaust streke aryInjec on

Late ection

90 ISO 2 /0

Injection start timing ( 0 BTDC)

30

Figure 1-13 Engine-out HC behavior

30

In particular, the black peak is most likely caused by the surface wetting of the fuel.

Surface wetting can cause diffusion burning of the fuel; plus, a concentration of fuel at the cylinder wall is unfavorable because of the amount that will remain unburned after the flame has been quenched at the wall. Moreover, after the fuel remains unburned near the wall, the piston scrapes the wall and transports a mixture of more highly concentrated HC into the exhaust.

Finally, the spray pattern of fuel substantially affects the vaporization and thus, the HC emissions of a DISI [2]. While it is more common to consider the injection pressure, the ambient pressure, into which the fuel is transported, is also a parameter which can be affected by the injection timing as the intake stroke progresses or especially if the compression stroke starts, see Figure 1-14.

Figure 1-14 Effect of hot operation of a multihole injector for four levels of ambient back pressure; sixhole nozzle; 50* spray; fuel: indolene; fuel temperature: 90*C; fuel injection pressure: 11.0 Mpa; amount of fuel injected: 10 mg per injection.

31

As the back pressure is decreased, the individual effect of each nozzle hole becomes less pronounced, and as expected, the fuel penetration increases because the deceleration from aerodynamic drag is smaller. Similarly, another comparison related the footprint of a spray at a location 40 mm from the injector tip, and the image was captured 4 ms after the SOI in

Figure 1-15.

PAeb I b ar 9 bar 14 bar

55lb bara

Figure 1-15 Effects of back pressure (Pamb) and fuel rail pressure (Pinj) on the cross-sections of GDI sprays from a swirl injector.

As expected, the increasing injection pressure causes a greater dispersion of the droplets and a smaller droplet size. Likewise, the increasing ambient pressure causes a higher concentration of droplets given the same injection pressure. Of course, the injection pressure increases by factors below two, but the ambient pressure is changed by an order of magnitude. Consequently, Figure 1-15 is not meant to be an exact comparison of the relative sensitivity to injection and ambient pressure. Interestingly, the injector temperature also affects the spray pattern as shown in Figure 1-16, Figure 1-17, and Figure 1-18 [2].

32

Figure 1-16 Effect of swirl-injector operating temperature on spray development with iso-octane at

20

0

C, 75"C, and 100'C.

Figure 1-17 Effect of swirl injector operating temperature on spray development with indolene at 20'C,

75*C, and 100*C.

Nmmt~U3~COB~ YiP ft~~PHEWs~. PA Ut,.

ws~suE ~ai tim.AJELPW.M

P4IECY1ONP.~O

111

IQOltP. am

14

12

,44

C

-20

DISTANCE FRM lNJECTOR AXI (mm}

20

$0

Figure 1-18 Spray SMD profiles for three injector operating temperatures.

In this case, the image was taken 1.5 ms after SOI, and the ambient pressure was 0.1 MPa.

The increased temperature causes a reduced effective spray cone angle and deeper penetration. It is notable that the temperature of the injector does not alter its operation;

33

rather, the injector temperature causes heat transfer to the fuel, until the fuel has reached an equal temperature, which alters the fuel's tendency to vaporize. More important than the pictures of penetration is the inverse relationship between temperature and the droplet

SMD.

34

2 Experimental Setup

In order to examine the hydrocarbon emissions from a DISI engine, a test cell was designed and fabricated for a production 2005 GM Ecotec engine, which was fully instrumented.

The engine has a four-cylinder inline layout, and it produces 155 hp at 5600 rpm.

Important geometric details can be seen in Table 2:1.

Table 2:1 Engine geometry details

Displacement, Vd

Bore, B

Stroke, L

Compression ratio, r,

2198 cm

3

86.0 mm

94.6 mm

12.0

00 aTDC* / 650 aBDC Intake Valve Opening/Closing

Exhaust Valve Opening/Closing 10.50 bBDC* / 44.50 aTDC*

*These values are taken from an internal GM-Opel paper with clarification from Arthur

Freeland [17] and have not been confirmed as factual for this particular engine. This paper lists the "duration" for the intake valve timing as 240 CAD and the "spread" as 120 CAD.

The spread is defined as the CAD between TDC and the max valve lift. Hence, the intake valve opening is estimated to be at TDC. The intake valve closing time has been experimentally verified, as it can be clearly seen in the pressure trace as the valve lands on its seat. Furthermore, the pressure traces are shown in Appendix- 1.

These valve timings can be compared to a similar PFI GM Ecotec engine in the Sloan Auto

Laboratory with IVO/IVC at 7 bTDC/56 aBDC and EVO/EVC at 68 bBDC/16 aTDC.

Since the engine was likely modified from a lean burn predecessor for the European

35

markets [17], the exhaust valve timings are quite different from the PFI engine. The exhaust valve opening (EVO) is supported by the pressure trace in Appendix- 1, because an earlier EVO would have produced a visible decline in pressure as the exhaust blows down from a higher pressure. Also, the EVC is supported by the results in the Injection Timing

Experiment section (p. 62).

In addition, the engine utilizes Siemens injectors with a variable stock operating pressure ranging from 40 to 120 bar, and the injector is located on the side of the chamber, between the intake valves as seen in Figure 2-1. The internal GM-Opel paper does not provide information about whether the injectors are swirl injectors or simply multihole; however, it does clarify that the SMD<16pm.

Figure 2-1 iCombustion chamber; injector and spark plug location

Even though the engine was produced in Europe, it was designed for homogeneous, stoichiometric operation as indicated by the lack of bowl in the piston in Figure 2-2. A

36

bowl shape in the piston typically denotes the potential for stratified operation by the sprayguided method of operation.

Figure 2-2 Piston shape

Since each cylinder would produce similar conclusions about the HC emissions mechanisms, all experiments have been performed on one running cylinder while the engine is motored, i.e. simply spun without fuel or combustion. As a result of this mode of operation, the intake and exhaust manifolds were separated to precisely measure and analyze the air entering and exiting the chosen cylinder. At the same time, the flanges of the manifolds were retained for a proper seal to the engine head. The flow characteristics, e.g. velocity field, of the intake air should be representative of the stock engine since the air is divided to the two intake runners with 15 cm of the stock manifold retained.

Typical engine operation consisted of using an electric motor to determine and stabilize the engine speed, while the eddy current dynamometer, produced by Froude Consine Limited model AG30, provided a constant torque on the engine. The manifold absolute pressure

(MAP) was controlled by two gate valves to inhibit the flow of the intake air.

37

Typical operating procedures for filling the fuel system, motoring the engine, starting the engine, and shutting down the engine can be found at the end of this chapter: Table 2:3,

Table 2.:4, Table 2:5, and Table 2:6.

2.1 Instrumentation

In order to assess the engine performance and control it to a high degree, several instruments were necessary.

2.1.1 Air Flowmeter

Since the intake manifold for the firing cylinder is separated from the other three cylinders, the air flowmeter measures only the air entering the cylinder of interest. The sensor is an

EPI model 8716MPNH with two-inch tube style connections on both ends; after selecting an operating range, the flowmeter has a sensitivity of 7.14 g/s / V with an accuracy of +/-

[1% Reading + (0.5% Full Scale + 0.02% / C)]. Two resistance temperature detectors

(RTDs) comprise the internal sensing elements. One RTD measures the ambient temperature, while the second RTD is heated to a higher temperature. As the flow of air increases, and hence the heat transfer from the heated RTD increases, the flowmeter measures the power needed to maintain the heated RTD's temperature. From this power consumption, the flow rate of air can be determined accurately.

2.1.2 In-Cylinder Pressure Transducer

In order to determine the engine's mean effective pressure (MEP) and other characteristics of each cycle's combustion, the engine has a Kistler 6125A pressure transducer. This

38

transducer was selected for the application because of its response and recommendations from industry researchers. GM performed the complicated machining into the engine head; the machining must reach a precise depth in order to minimize the volume between the combustion chamber and the tip of the transducer, without damaging the structural integrity of the combustion chamber's contour. Furthermore, this professional machining is recommended to achieve a properly sealing sleeve between the transducer and the oil and coolant galleries. In the event that a future project requires firing additional cylinders, they have been prepared to accept the same 6125A transducer installed.

The 6125A transducer has a sensitivity of approximately -16.3 pC/bar with a linearity of

+/- 0.5% of full scale. Finally, the pressure transducer is a relative measurement of pressure; as a result, the pressure signal must be "pegged" to a certain value in order to obtain meaningful results. For this application, the LabView code pegged the cylinder pressure to be equal to the MAP sensor pressure at BDC intake.

2.1.3 Lambda Meter

To acquire the air/fuel ratio information, an Etas LA4 Lambda meter is located about 17 cm from the exhaust port; it operates with a combination of a Nernst concentration cell and a pump cell to transport oxygen ions. The planar two-cell design of the lambda meter consists of a solid body electrolyte comprised of ceramic layers, which separate a reference gas and the exhaust. The voltage produced across this planar sensor is highly sensitive to the oxygen content in the exhaust, and hence, is sensitive to the air/fuel ratio.

39

2.1.4 Fast Flame Ionization Detector

The primary tool for exhaust analysis was a Cambustion HFR-400 Fast Flame Ionization

Detector (FFID) with a time response (10-90%) of approximately one millisecond [18].

The instrument was calibrated before and after each experiment using 1500 ppm C3 (4500 ppm C1) propane (C3H8) span gas and zeroed with ambient air. Time-averaged downstream HC emissions were sampled from a well-mixed, large volume, pulsedampening tank located 180 cm from the exhaust port.

2.2 Systems

Due to the operation of internal combustion engines, several parts of the experimental setup are best explained through a system view.

2.2.1 Intake System

The intake air enters the system through a filter and flows through the aforementioned air flowmeter. Next, a parallel arrangement of gate valves allows fine and broad control of the flow; this arrangement replaces the function of a throttle, which would be harder to control and unnecessarily complicate the system since the manifold absolute pressure (MAP) was seldom varied. Downstream of the valves, a tank is used to dampen the pressure oscillations created by the periodic vacuum of the cylinder. The volume of the tank is approximately ten liters, chosen specifically to be a factor of 20 increase of the displaced volume of the active cylinder. The hose connecting the damping tank to the intake manifold is rated to a full vacuum, i.e. 29.9 mm Hg, and it has a smooth inner contour. The hose is attached to the intake manifold through a wire clamp and further sealed with silicon

40

tape. Despite the large number of pipe and hose fittings, the quality of the intake system, and its lack of leaks, is shown by the 0.07 bar vacuum that the system can sustain with the gate valves completely closed. Thus, the intake airflow rate can be accurately determined

by the air flowmeter.

2.2.2 Fuel System

The stock engine employs a mechanical fuel pump driven by the camshaft, but for the purposes of controlling the injection pressure independently of speed, this setup utilizes the system shown in Figure 2-3.

Pressur e R egulator

Tank

Pressure

Gauge

Fuel

Pump

Fuel

Tank

Fuel Filter

Che ck Valve

Figure 2-3 Fuel system diagram

Hy draulic

Accumulator

Fuel Rail and Injectors

41

The primary component of the fuel system is the hydraulic accumulator, which is a highpressure vessel with an internal piston. One side of the piston is pressurized with nitrogen from a tank, which has a fully charged pressure of 160 bar. The pressure in the line and in the accumulator is controlled by the manual pressure regulator. The other side of the accumulator piston can be filled with fuel by the electrical pump. It is important to note the significance of the check valve in the system to ensure the highly pressurized fuel does not travel back to the fuel tank. The second connection on the pressurized side of the accumulator is a line supplying fuel to the common rail and operating injector. The fuel system is limited to 135 bar (2000 psi) by the strength of the stainless steel tubing used for the high-pressure lines. The fuel injectors were calibrated with a custom rig to capture the spray in a vial which sealed to the circumference of the injector. The calibration is shown in Appendix- 2.

2.2.3 Coolant System

For the purpose of investigating the vaporization achieved in this DISI engine, the coolant system was designed to reach a wide range of coolant temperatures. The two ball valves allow two coolant loops: one to heat and one to cool. The details and operation of the coolant system is shown in Figure 2-4.

42

City W ater

Solenoid Valve

(normally open)

Heat Exchanger

Engine Block

St

1

Tee Ball Valve

F Tee Ball Valve

IF

Coolant

Tank

Coolant Pump

Figure 2-4 Coolant system diagram

First, the heating loop is achieved with both handles of the valves aligned with the singular port of the pipe tee. For chiller usage, both handles should be parallel with the two ports of the pipe tee. The 10 KW heater provides a relatively quick ramp up in the coolant temperature, and 80*C was generally used as the maximum temperature. A simple controller was used to regulate the coolant temperature within a certain range. If the coolant temperature, measured entering the engine, dips below the lower set temperature, then the controller sends a signal to close the solenoid valve of the cooling water and to turn on the heater. Likewise, if the temperature increases beyond the high set point, the controller opens the solenoid valve to the heat exchanger water and simultaneously turns

43

off the heater. The lower bound of the coolant temperature for steady state operation is about 20'C, depending on the ambient temperatures.

2.2.4 Exhaust System

In a similar manner to the intake manifold, the exhaust manifold was separated so that measurement samples were only taken from the running cylinder. The lambda sensor and first thermocouple are located approximately 18 cm from the exhaust port. Since the first set of experiments were planned to examine steady state operation, a mixing tank, located about 180 cm from the exhaust port, was placed in the exhaust line so that the exhaust species would be spatially homogeneous. After mixing, the exhaust temperature and FFID ports were located in the exhaust stream.

2.2.5 Engine Control System

The engine control system consists of a master and slave computer. The master computer accepts the user inputs of engine speed (RPM), target lambda value, end of injection (EOI, in CAD aBDC compression), and spark timing (CAD aBDC compression). Its C program calculates the start of injection (SOI) and the dwell in crank angle degrees. The dwell is the amount of time that the spark plug needs to be charged before it emits its spark. The master program places this information into an array which is sent to the slave computer.

The slave computer must run its C program in DOS since the computations for small actions, like moving the mouse, have the potential to cause the outputs enough delay that the engine could misfire for a cycle. The slave program accepts the inputs of feedback gain

(typically 10) and injection duration (psec). Then it provides the digital signal for the

44

operation of the injector and the spark plug. A hardware switch is used to interrupt the engine's operation and ensure that the sequence of injecting and firing is stopped with the injector in the closed position.

2.2.6 Data Acquisition System

The data acquisition system is based upon a National Instruments platform, schematically shown in Figure 2-5; a description of its components and the respective purposes is given in

Table 2:2.

Chassis SCXI 100X

Tle oeouph Data

Master Campuder Tower

BNC 2095

SCXI 1112 1

S 1349P622

BDC aampress

Crank Augle signal sitinl syte

Figure 2-5 Data acquisition system

45

Table 2:2 Data acquisition components and functions

Item

BNC 2095

SCXI 1112

SCXI 1100

Function

Records up to 16 differential analog BNC signals

Records up to 8 thermocouple signals

Multiplexes the signals into a single programmable gain instrumentation amplifier (PGIA)

SCXI 1302

SCXI 1349

Allows digital signals to feed through to the master computer

Necessary adapter to include the feed through signals

PCI 6220 Data acquisition card

The data was acquired through a LabView program, and the majority of data was recorded on a crank angle basis, except temperature data, which cannot be recorded as quickly.

46

Table 2:3 Fuel filling procedure

1. Depressurize the nitrogen line by turning the pressure regulator several counterclockwise turns until there is only ambient pressure in the line and accumulator.

2. Ensure that sufficient fuel exists in the ATL tank by lifting the tank to gauge its weight

(if the fuel pump is activated without enough fuel in the tank, the line will be filled with air and will need to be purged).

3. Remove the blue cap to allow air flow into the tank.

4. Turn on the fuel pump at the control panel. The pressure gauge on the system should read about 25 psi while filling the accumulator, and it will rise to the relief valve threshold (80 psi) when the accumulator is full. Filling a completely empty accumulator should take approximately 75 seconds from the measured flow rate.

5. Turn off the fuel pump.

6. Replace the blue cap on the ATL tank; if the fuel is left open to atmosphere the vaporization is not only a danger, but it lowers the remaining fuel's volatility.

7. Allow flow from the nitrogen bottle by turning the knob counter clockwise.

8. Pressurize the nitrogen line by turning the pressure regulator knob clockwise.

47

Table 2:4 Engine motoring procedure

1. Turn on the lab fan and water pump.

2. Allow flow from the nitrogen bottle by turning the knob counter clockwise.

3. Pressurize the nitrogen line by turning the pressure regulator knob clockwise.

4. Ensure the National Instruments chassis is powered.

5. Flip the motor breaker to the "on" position, normally allowed 5 minutes to warm up.

6. Check the rotating parts for possible obstructions or wires in their path.

7. Check oil level with the dipstick and coolant level through the sight tube on the tank.

8. Check that the coolant valves on the test bed are in the proper position.

9. On the wall, adjust the ball valves directing water flow to the dynamometer, coolant heat exchanger, and oil heat exchanger. Unless specifically needed, the valves should be opened with a maximum of 300. If the valves are open too far, the water pressure will be insufficient for other test cells. Small flow rates for the coolant heat exchanger can help the stability of the engine coolant temperature.

10. Open main water valve to allow water flow.

11. Turn on the sensor switch, labeled Lambda and Mdot, at the control panel. Enter the test cell to press the "operate" button on the pressure transducer amplifier.

12. Turn on the dyno controller by the right button. (Do not press the left button, which engages the dyno. It will be engaged after starting and motoring the engine).

13. Turn on the coolant pump switch, at the control panel.

14. If experiments are highly temperature dependent, turn on the oil pump and chiller toggle switches. To engage the oil pump, push the start button on the oil pump controller located on the tower in the test cell. Note the heater controller should stay in stand-by

"Stby" mode until the engine is started, to avoid the potential to vaporize the coolant with the high powered heater.

15. Turn on the motor controller switch at the control panel. Shortly thereafter, the motor controller will make a higher pitched sound as it enters the overdrive mode.

16. After hearing this sound, briefly crank the engine with the key on the control panel.

The duration should be no longer than 1-2 seconds. If the crank time is appropriate, the motor will spin the engine as shown by the rotation of the crankshaft, which can be seen from the test cell window. If the crank time is too short, try a slightly longer duration.

17. The motor controller value corresponds to the engine speed, which can be read on the dyno controller. For example, a value of 25.2 corresponds to 1230 rpm, which is then reduced to 1200 rpm as the dyno is engaged.

48

Table 2:5 Engine running procedure

1. After following the Engine Motoring Procedure, turn the dyno load set point (LSP) to zero. Now press the left button on the dyno controller; the engine will slow down for a few seconds, but the motor will ramp the speed back up to its initial value.

2. Slowly increase the load to the desired set point, which will slow down the motor, for example from 1230 rpm to 1200 rpm.

3. Both the slave and master computers should already be on. The slave computer should be running in DOS mode, not simply a DOS command prompt; the operating file is located in C:\D\singleb. The master computer should use a command prompt to access its program located in C:\D\master b.

4. On the master computer, enter the rpm as read from the dynamometer controller.

5. When prompted, enter the target lambda value, the injection timing (e.g. 660 CAD aBDC), and the spark timing (e.g. 160 CAD aBDC).

6. When prompted, enter the feedback gain (e.g. 10) and the injection duration (e.g. 1700 psec), and press enter.

6. After hitting enter, the engine sound will change assuming good combustion is achieved as shown by the lambda meter readout.

7. If the lambda readout is extremely rich (below 0.7) or extremely lean (1.3), stop the engine running program with the hardware switch located in the desk. Problems starting combustion can be diagnosed by logging data after completing step 6, see step 10.

8. After motoring the engine for a minute, reattempt to get the engine running, ie step 4.

The motoring process alleviates the chances of backfiring the excess unburned fuel from the previous step.

9. Follow the Omega controller instructions to us the coolant temperature controller.

Essentially, the controller will attempt to keep the coolant inlet temperature between two set values. If the coolant becomes too warm, the controller will send a signal to open the solenoid to run water through the heat exchanger. If the coolant becomes too cool, the heater will turn on the heater located in the coolant tank.

10. On the master computer, open the master LabView program to record data.

49

Table 2:6 Engine shut-down procedure

1. Press Enter twice on the coolant temperature controller to put it into stand-by mode.

2. Flip the hardware switch to end the engine running program, which ends the program running on the slave computer. The sound of the engine should change frequency. The master computer can be exited with three zeros (0 0 0).

3. Turn the appropriate instruments (including coolant pump) off at the control panel.

4. Disengage the dyno by deactivating the left button.

5. Turn off the motor with the switch on the control panel, and the engine should reach zero rpm within a few seconds.

6. Turn off the dyno controller power with the right button.

7. Flip off the motor breaker inside the test cell.

8. The National Instruments chassis and control panel power supply can be left on.

9. Stop the flow of water by closing the main water supply valve. The other valves can be left in their position for convenience for the next experiment.

10. Ensure the blue fuel cap is finger-tight on the ATL tank.

11. Close the nitrogen tank by turning its knob clockwise.

12. Drain the nitrogen in the system, unless experiments will be run within one day. The extra pressure in the system is not ideal if something were to break or the fuel system leaks a few drops of fuel due to the high pressure, but draining and filling the lines every time hurts the usage time of each nitrogen tank.

50

3 Results

In order to achieve a common basis for comparison of the experiments, a baseline condition was determined. These conditions are listed in Table 3:1.

Table 3:1 Baseline operating conditions

Engine Speed 1200 rpm

MAP 0.46 bar

End of Injection Timing 120 CAD aTDC

Fuel Pressure 70 bar

Coolant Inlet Temperature 60

0

C

Fuel Chevron Phillips UTG 91

The engine speed and manifold absolute pressure (MAP) were selected to represent the fast idle period. The injection timing is always listed as the end of the injection (EOI). For comparison to papers listing SOI, the start of injection timing can be approximated from the pulse width of 15 CAD. While 15 CAD is easy to visualize on the graphs of data, when a more accurate value is needed, a more precise pulse width of 12.7 CAD was calculated from the mass flow rate of air, the lambda value, and the injector calibration. The fuel pressure baseline was selected as a pressure in the middle of the stock engine operation, which is variable from 40-120 bar. Finally, the coolant temperature was selected to be quite warm at 60'C, but 80 0

C was not chosen because of the additional time that every experiment would take to warm up to steady state.

As a result of the control systems installed, some of the baseline parameters had a slight variation. For example, the MAP experienced some fluctuation, but it was maintained

51

universally within +/- 0.015 bar. Similarly, the manual control of the nitrogen pressure regulator caused a difficulty in matching the exact desired fuel pressure. Across all experiments, the fuel rail pressure was kept within a +/- 1.5 bar tolerance. The coolant temperature was controlled within +0.51-1.0 C of the target value. In general, the experiments were repeated on two separate days, and each point within the experiments was logged two to four times with about 50 cycles worth of data recorded each time.

Hence, the plots generally represent at least 200 cycles of averaged HC data. The exceptions to this rule are the Injection Pressure Experiment with High DI fuel and all isopentane tests, which include 100 cycles of averaged HC data.

Given the capability of the experimental setup, five sets of experiments were performed to illustrate the underlying mechanisms of HC emissions as shown in Table 3:2. Each of these experiments was completed during steady state operation.

Table 3:2 Engine parameter experiments

1. Sweep of coolant temperatures at baseline conditions

2. Sweep of injection pressure at baseline conditions

3. Sweep of injection timing at baseline conditions

4. Sweep of injection timing at varying coolant temperatures

5. SwNeep of injection timing at varying injection pressures

The above experiments were also repeated with two types of fuel, Chevron Phillips UTG91 and a High DI fuel, with the properties of the former given in Table 3:3.

52

Table 3:3 Chevron Phillips UTG 91 fuel

Property

Specific Gravity at 60/60 OF

API Gravity

Copper Corrosion, 3 h at 50 *C

Existent Gum (washed), mg/100 mL

Suifur, ppm

Reid Vapor Pressure, psia

Lead, gfgal

Phosphorus, gigal

Hydrogen, wt

Carbon, wt

Carbun Density, gigal

Distillation Range at 760 mming, 'F

Initial

10%

Boling Point

50%

90%

End Point

Oxidation Stablity, min

Ileat of Conustion, Net, Btlb

Composition, vol %

Aromatics

Olefn

Saturates

Research Octane Number

Motor Octane Number

Anti-Knock Index, (R+M)2

Sensitivity

Oxygenates, vol %

Benrene, vol %

24.0

6.0

70.0

90.8

828

86.8

7.8

Typical Value

0.7350

61.0

1

2

130

9.0

0.0010

0.001

13.7

86.3

88

122

212

321

399

>1440

18500

Specification

0.7343 -0.7440

Report

I max

5 max

1000 max

8.8-9.2

0.0050 max

0.0D2 max

Report

Report

Report

75-95

200- 230

300- 325

415 max

1440 min

Report

35.0 max

10.0 max

Report

90.3-917

Report

87.0 max

7.5 min

0.0 max

Report

Test Method

ASTM D 4052

ASTM D 1250

ASTM D 130

ASTM D 381

ASTM D 5453

ASTM D 5191

ICP/OES

ICPIOES

ASTM D 5291

ASTM 0 5291

Calculated

ASTM D 86

ASTM D 525

ASTMI D 240

ASTM D 1319

ASTM 0 2699

ASTM D 2700

Calculated

Calculated

Chromatography

Chromatography

The second fuel had the following specifications written on the barrel: 1250 DI QF 1721

LS 04 Haltermann Gasoline 3 UN1203 II ERG 128.

The driveability index (DI) is calculated by the following formula:

DI= 1.5 * T10 + 3 *T50 + T90

DI= 1.5 * 122 + 3 * 212 + 321 =1140

(1)

(2) (for the UTG 91 example)

Where T10, T50, and T90 are the distillation temperatures (OF) at the 10%, 50%, and 90% distillation points; hence, UTG 91 has a DI of 1 140 0

F. On the other hand, the second fuel has a DI of 1250

0

F; this fuel will be referred to as "High-DI" fuel. Granted, not all effects on HC can be so easily linked with the lower volatility of the second fuel because of the complications of boiling heat transfer, but they are expected to be closely related.

53

3.1 Coolant Temperature Experiment

The primary purpose of this experiment was to illustrate and clarify the vaporization of fuel inside the cylinder, and the coolant temperature was varied between 20"C and 80'C. In the case of the UTG 91 and isopentane fuels, 20'C was not able to be reached because of the high ambient temperatures and the relatively high temperature of the city water flowing through the heat exchanger. The general expectation for this experiment was steadily lower

HC emissions as the coolant temperature increases; assuming the major HC mechanisms should be the same as in the PFI engine and compilation from Cheng et. al. [1], the reasons for this expectation are given below in Table 3:4.

Table 3:4 Reasoning for coolant temperature experiment expectations

Mechanism

Crevices

Deposits

Oil Layers

Expectation as Coolant

Temperature Increases

Decrease

Decrease

Decrease

Reasoning

Ideal gas law

Ideal gas law

Henry's law a temperature

Liquid Fuel Generally Decrease

The results of this experiment are shown in Figure 3-1.

Dependent on mode of boiling

54

5500

-*- Average - High DI MAP = 0.46-0.47 bar

-.Average UTG 91 RPM = 1200

Fuel Pressure = 70 bar

5000 ------- ------------- -A-Averae Isopentane------- E01 = 120 CAD aTDC -----------

E

C.

CL

4500

----

3000-

10 20 30 40 50 60

Coolant Temperature (C)

70 80 90

Figure 3-1 Coolant temperature sweep

Obviously, the general trend of decreasing HC with increasing coolant temperatures is shown. However, the particular behavior of the UTG 91 and High DI fuels is not nearly as orderly as the isopentane. As a reference for similar operation of PFI engines, information from the Hallgren PhD thesis [18] has been included in Appendix- 3 and Appendix- 4.

Visually averaged over the entire cycle, the steady state HC value is approximately 2250 ppm C1 to 2500 ppm Cl. Granted, there were some differences in the engine's operating condition versus the above test, but the comparison is nevertheless worthwhile. Hence, by comparison to the PFI engine in the Sloan Auto Laboratory, the HC levels were relatively high.

Further, the isopentane fuel shows an expected decrease in HC with warmer coolant temperatures. This trend is a combination of the effects listed in Table 3:4. Literature

55

research comparing toluene and iso-octane [19] with respective boiling points at 110.5

0

C and 99.5

0 C respectively, showed about a 30% reduction in HC for the same injection timing. Thus, it was expected that the isopentane would have produced a more striking difference in HC versus the UTG 91 fuel. The percent reduction found with the isopentane is shown below Figure 3-2.

4600

4400-

4200-

-------------------------- ---------------------------------------

-

-

4000 --------------

*

----------- ------------- ---------- ----------- - - - - ---

E

C.

3800

3600

3400

--

--

MAP = 0.46-0.47

RPM = 1200 bar ~ ~ --. ~~

Fuel Pressure = 70 bar

EOI = 120 CAD aTDC

3200

~~ ~ ~-~~--~~~

-'--Average - UTG 91

-

-A-Average - Isopentane

-c- Percent Reduction - UTG 91 to Isopentane

3000

1 0 20 30 40 50 60

Coolant Temperature (C)

70

12.5%

--------------------

10.0%

7.5%

5.0%

2.5%

0.0%

--

-2.5%

I e

E

C

-5.0%

80

-7.5%

90

-10.0%

Figure 3-2 Coolant temperature sweep with percent reduction caused by isopentane

Even the rough approximation from Cheng et al. [1], the elimination of liquid fuel in a PFI engine ought to reduce HC emissions about 20%, and the test with isopentane shown in

Figure 1-11 [29] yields about a 15% reduction in HC levels. Two explanations are possible. First, if fuel in the oil is significant, then the HC values for the experiments should be shifted equally. Thus, the optimization of this engine is spectacular to perform so closely with isopentane. Second, if the isopentane was contaminated by the previous fuel

56

through the pumping process, it would shift the isopentane curve up. Both of these explanations could be contributing to high levels for isopentane and small difference between the isopentane and the UTG 91.

Because the oil can retain fuel from one test or day to another, the duration between tests can become an important factor. The UTG 91 and High DI tests were performed over a span over 19 days, while the isopentane test was performed 40 days after those tests. Since the isopentane levels are relatively high, it is not expected that fuel was desorbed from the oil during that span of 40 days. Likewise, the engine was motored 8 days after the isopentane test to determine the steady state level of HC found in the exhaust. When motored, the fuel in the oil contributed 100 ppm CI to the exhaust. Thus, the data shown in this research is shifted 100 ppm Cl too high, but it can be concluded that the tests should be accurate relative to each other.

To examine the explanation of fuel in the oil, two procedures were performed. First, the oil used in all of the experiments in this thesis was fully warmed up through engine operation for over 70 minutes at 90'C. Running the engine hot is a common method for forcing any fuel to desorb from the oil. After this test was performed, the oil was drained without replacing the volume of oil in the filter. Again, the engine was warmed to a coolant temperature of 90

0

C, and the HC emissions were recorded. The results of these two experiments are shown below in

57

3500

3000 -------- ------- ------------

2000 --------------- ---------------------------------------- ------------- --------------------------- -------------

E a.

O 1500

- ---- - - - -- - - - - - -- - - - - -- - - - - - - - - - - - - - - - - - - - - - - - - - - -- - - - -

1000

- - ---- -- - - - - - - - -

500 ------------- ---

0 -

0 10 20

-oldoil

-

MAP = 0.45-0.47

RPM = 1200 bar new 1 OW-40 oil Coolant = 90 C

E01 = 120 CAD aTDC

30 40 50

Time Elapsed at 90 C (min)

60

---------- -------------

70 80

Figure 3-3 Hot oil test with old and new oils

First, it is evident that the HC emissions only decreased about 100 ppm Cl until a steady state value was reached after testing for 30 minutes. Secondly, the new oil reduced the HC emissions by a minimal amount. Through these tests, it has been shown that oil desorbed in the fuel is not a significant contributor to the relatively high HC emissions from this DISI engine. One curious point is the HC levels obtained when motoring the engine after shutting off the fuel. The oil fuel still produced about 800 ppm C1 after 2 minutes of motoring, while the new oil had significantly lower levels at 200 ppm Cl. Since the fuel would continually desorb from the oil at this temperature, the explanation is that the oil contributed about 600 ppm Cl to the exhaust. After desorbing, about 500 ppm Cl is burned during combustion or oxidized in the hot exhaust. The remaining 50-100 ppm C1 contributes to the higher level of measured HC.

58

Second, the UTG 91 and High DI data is relatively intuitive between 20'C and 40*C and

70'C and 80

0

C. However, the curious region is the region between 40'C and 70'C. From the mechanisms listed in Table 3:4, the top three mechanisms of crevices, deposits, and oil layers are relatively straightforward. The reason for the "generally decreases" expectation is the complicated dynamics of boiling heat transfer. Stanglmaier et al. [23] among others has shown the effect of film boilng in the vaporization process of the fuel. When the boiling mode switches from nucleate boiling to film boiling, the heat transfer rate from the piston to the fuel is decreased despite the larger temperature difference between the two masses. Clarifications on boiling heat transfer and the Leidenfrost point can be found in various textbooks. In addition, Stanglmaier et al. showed a vivid experimental illustration of this effect with isopentane among other fuels, displayed in Appendix- 5. It is possible that the complex nature of the boiling with the UTG 91 fuel and the High DI fuel caused the significant difference in HC at 60'C for the High DI fuel.

Moreover, the hindered vaporization affects other mechanisms like wall-quenching. The problem is not simply a larger amount of liquid in the cylinder, but the fuel that has managed to vaporize from the liquid film must still diffuse into the majority of the charge.

Stanglmaier et al. also explained the effect of too little diffusion into the charge; as a result of this lack of movement, the wall-quenching of the flame produces a compounding negative effect of the surface wetting. To summarize the explanation for the nondecreasing HC level between 400 and 70'C, the film boiling hurts the overall vaporization.

In turn, the relatively larger amount of liquid fuel and the wall-quenching of the flame both contribute to offsetting the other HC mechanisms which decrease in their contribution to

HC.

59

3.2 Injection Pressure Experiment

With a similar range to the stock engine's variable operating pressure, the effect of fuel pressure was researched on the setup. Unlike the expectation for the coolant temperature experiment, the results of this experiment were not intuitive or anticipated, as shown in

Figure 3-4.

5000

4800

4600

------------

MAP = 0.46-0.47

R M=1200 bar

~

--

-A rge-Hg D ------------ ---------------

Average - High DI-------

Coolant = 600 -- Average - UTG 91

-.----------- E01= 120 CAD aTDC ------- --- Avera-e---Isopentane

4400

4200

E

CL

4000

-A- ,tan vrq -----

---------------- -------------- ------------ -------------- ------------

-------------- - -- -----------

3800

------ ---------------

3600 ---------------- ------------- -------------- ------------------------------------------- ---------------

3400

3200

3000

0 20 40 60 80

Fuel Pressure (bar)

100 120 140

Figure 3-4 Injection Pressure sweep

The relatively high values of HC from the isopentane have already been discussed in the previous section.

At the same time, the isopentane trend versus fuel pressure is intuitive since the fuel pressure ought to have little effect on the emissions of a fuel which flash boils upon its travel from high pressure of 70 bar to a low pressure around 0.5 bar. Furthermore, the

60

variations within the isopentane data are difficult from which to draw conclusions because of the overall error in the test and instrumentation.

For the UTG 91 and High DI fuels, the fuel pressure will have an effect on the droplet size and velocity as shown in Figure 1-6 (p. 25) and Figure 1-7 (p. 26). For this particular injector, the penetration information is given below in Figure 3-5 and shown pictorially in

Appendix- 6. However, the droplet size distribution was not provided. Ideally, this droplet size distribution would significantly enhance the understanding of this experiment since any large drops have a significant effect as mentioned in the Vaporization section.

C

0

C

.5

0.

70.00

65.00

60.00

55.00

50.00

45.00

40.00

35.00

30.00

25.00

20.00

15.00

10.00

5.00

0.00

0

Bench: SB01

PW= 1 rS

FP tolerance: +/-1 bar

FT = 22"C +/- 2"

BP tolerance: +/-0.5bar

FT = 22"C +1- 2"

0.2 0.4 0.6 0.8 1 1.2 dely afler SUI [mm]

1.4 1.6 1.8 2

Figure 3-5 Spray penetration vs time after SOI, with injection pressures of 30, 70, and 120 bar D

Siemens VDO

At high pressures, the HC levels are higher because the droplets have higher velocities and less vaporization occurs in the flight time between leaving the nozzle and the impact on the cylinder surfaces. This data in Figure 3-4 is the first evidence of surface wetting affecting

61

the HC levels, and further support of this phenomenon is shown in the Injection Timing

Experiment.

On the other hand, the data displays an optimum fuel pressure at which the tradeoff between high velocity and large droplets is assumed to be the optimum for this injection timing and the in-cylinder charge motion. If the injector produces larger droplets at lower pressures as anticipated, they would be a significant source of liquid fuel. Thus, the higher

HC levels would follow from vaporization issues (boiling and wall-quenching) as explained in the Coolant Temperature Experiment section. Granted, the simple explanation of larger droplets with lower injection pressures does not explain the constant or decreasing HC levels from 50 to 20 bar for both UTG 91 and High DI fuels.

3.3 Injection Timing Experiment

As previously explained, the injection timing was varied from 30 CAD aTDC intake to 210

CAD aTDC with the remaining operating conditions at the baseline values. The results from this experiment are shown below in Figure 3-6. The typical pulse width is depicted to illustrate where the SOI can be approximated.. Even though the real-time value was not recorded for every data run, it is an accurate measure for this visual purpose.

62

9000

8000 ------------- ------------

-------------------------- ------------ ------------ --------------------------

7000 - ------------ -------------------------- ------------ ------------ --------------------------

6000 - ------------- --& ------- o -- -- .I -------------

E

0.

0.

5000 ------------- ------- -------------------------------- ------------- --- ----- -|--------------------------

2000 -------------- ------------ -----------------

-

-

Average High DI MAP = 0.46-0.47 bar

RPM = 1200

-s-Average

UTG 91 Coolant = 60 C

-A- Averaqe Isopentane Fuel Pressure 70 bar

1000 ------------- ------------ -------------------------- ---------------------

Typical Pulse Width ~15 CAD

0

0 30 60 90 120 150 180 210

----

240

Intake Stroke Start End of injection (CAD aTDC) + BDC

Figure 3-6 End of injection hydrocarbon sweep

First the general shape of the curves will be discussed. This trend, which displays decreasing HC emissions, a minimum and then increasing HC emissions as injection timing is retarded, has been shown in several pieces of literature including Price et. al. [19]and

Alger et. al.[20]. These studies both used an engine design with the injector position in the center of the bore, whereas the engine for this paper had its injector located on the outer radius of the bore between the intake valves. In addition, the results and test conditions from these studies are plotted with the above data in Appendix- 8 and Appendix- 9.

With the design of a DISI engine, it is possible that very early injections will place fuel directly into the exhaust stream. Hypothetically, if the EVC were 15 CAD aTDC and a typical pulse width is 15 CAD, the end of injection (EOI) would need to be later than 30

CAD aTDC to ensure hydrocarbons do not escape combustion directly through the exhaust

63

valve. However, if the EVO is 44.5 CAD aTDC for this engine, any injection before 60

CAD is liable to produce higher HC levels. This result is seen in the isopentane experiment. The only rationale for such high HC with isopentane is that the fuel has circumvented combustion in the cylinder by exiting the cylinder before the EVC.

Even though the exhaust valve is still open for the EOI data point at 50 CAD aTDC, the lift is very small and the distance that the isopentane must travel is too far to reach the exhaust valve before it closes. For instance, the typical liquid velocity from the injector is about 50

m/s, so this liquid fuel would take about 1.89 ms to cross the bore. In that flight time, about

14 CAD will have passed, and thus, the valve would be completely closed. In addition, the isopentane would quickly become a vapor due to flash boiling so the HC would be carried to the exhaust valve by the charge motion.

As shown in Figure 1-14, the back pressure can have an effect on the spray penetration and cone shape. The largest difference in back pressures is 0.28 bar for any 15 CAD segment of the pressure trace, given in Appendix- 7. Thus, the back pressure is not expected to heavily influence the injection timing sweep.

For early injection, Price explains that the higher HC levels are due to surface-wetting.

Throughout this paper, the term "surface-wetting" will include wetting of the cylinder walls and the piston crown. Various papers use the term "wall-wetting" which can become unclear whether the author writes specifically about the cylinder walls or all surfaces, including the piston. Despite the hot surfaces in the engine, it cannot be so quickly assumed that these surfaces will vaporize all of the fuel that impacts it because of the film boiling explained in the Coolant Temperature Experiment section. As a result of this

64

decreased heat transfer rate during film boiling, the surface-wetting of early injection can nevertheless allow liquid fuel to remain in the cylinder into the compression stroke. Alger explains the decrease in HC as a function of the decreasing surface-wetting as the piston moves further down in the stroke, and consequently, more fuel is able to vaporize before contacting a surface. Hence, the explanation is closely tied to the reasoning for the effect of higher velocities created by higher injection pressures. Another explanation by Price is that the fuel spray cone has not been allowed to develop for the early injection timings. As a result, the fuel has not been dispersed well in the air nor on the piston; consequently, the fuel that forms a liquid layer on the piston will have a larger height and less surface area.

One may think that the temperature distribution across the piston could account for the shape of Figure 3-6. For early injection, the fuel would land on the outer part of the piston.

For the middle injection timings, the spray would hit the hottest part of the piston, i.e. the center. For late injections, the fuel would hit the outer edges of the piston, which would be cooler. However, this logic does not explain other literature performed with centrally mounted and aimed injectors. Even with the injector aimed at the same location on the piston, the injection timing experiments show a similar concavity, shown plotted with the data from this research in Appendix- 8.

For late injection, Price explains the higher HC with two reasons: inhomogeneity of charge and vaporization problems. The inhomogeneity is supported by the high particulate matter

(PM) emissions from the engine. The poor vaporization reasoning is supported by the significantly lower HC levels for iso-octane, as mentioned in the Coolant Temperature

Experiment section. On the other hand Alger credits inhomogeneity and poor mixing with the intake air. The rationale for the inhomogeneity case is based upon variations in the

65

fiber optic probe measurements of the equivalence ratio at the spark plug gap. Meanwhile, the effect of the intake air motion is based upon an optical study showing a reduced disruption of the spray pattern for late injection timing.

For the DISI engine used in this experiment, the higher HC levels for isopentane illustrate that the fuel is escaping before the exhaust valve has been closed at 44.5 aTDC. Therefore, for all fuels, the major reason for the high HC for early timings before 50 CAD aTDC is the overlap with the SOI and the EVC. However, this explanation cannot cover the entire region of high HC for early injections. Consequently, the other influential mechanism is the surface wetting, which for early timings, occurs almost exclusively on the piston.

Meanwhile, the vaporization must be a primary contributor because of the stark contrast between the isopentane fuel and the UTG 91 and High DI fuels. Essentially, poor vaporization can account for the reasoning mentioned in the previous papers: the inhomogeneity of the charge and the compounded effect of poor interaction with the intake motion. Since the intake air motion scales with the piston speed [4], this nondimensionalized quantity has been overlaid on the previous data to yield Figure 3-7. It is apparent from the figure that the intake motion will be significantly less forceful for the late timings after 180 CAD aTDC compared with the motion before 120 CAD aTDC.

66

End of Injection Hydrocarbon Sweep

8000

8000 --- ------------ I------------

1

E

7000 - -- --- ------------ ------------ ------------- --------- ------------ L------------

- 0

3000 ------------------------------------------------------------------------------------

----

0.5

2000 -------------------------------- -

-&-Average- UTG 91

MAP = 046-047 bar

RPM = 1200

-&- Average - Isopentane Coolant = 60 C

1000 -------------- ----------------------N.D. Piston Position

-N.D. Piston Velocity

Fuel Pressure = 70 bar ----

Typical Pulse Width -15 CAD

0

0 30 60 90 120 150 180 210

--

240

-1.

+Intake Stroke Start End of Injection (CAD aTDC)

+

BDC

Figure 3-7 End of injection hydrocarbon sweep with piston position and velocity

While the dynamics of the intake flow are complicated and violent and the exact size of droplets is unknown, the trajectory of the droplets has been estimated at the baseline case with a fuel pressure of 70 bar. Due to limited knowledge of the exact injector angle, it was approximated as 45 degrees from the horizontal. Further, this analysis of the spray disregarded the smaller droplets which would be become entrained in the air, and it also disregarded the effect of the spray cone, instead focusing on the straight-shot direction out of the injector. The model information was gathered from Siemens experimental information given previously in Figure 3-5; from these points, the model equation was selected (from logarithmic fit, power fit, etc) to fit the experimental points the best while also being physically logical in shape.

67

Y = 51.829 * t

.7

588

(3)

Where y has the units of millimeters, and time has the units of milliseconds. The fit of the penetration data is shown in Figure 3-8.

Figure 3-8 Experimental droplet penetration and model equation

An additional assumption is that the clearance height at TDC is negligible compared to the stroke length. Given this equation and geometric details like the bore and stroke of the engine, the flight time until impingement is shown in Figure 3-9.

68

30

25-

---- ---- --- ---- --- 4i--- - - - - - -

0

E

* 0-

E

E 15

0

--- -- ----------- -- ---- ---- ---- -- I ---- ------- -- -- --- ------ --

E r-10-

--- ----- -- --

5-----------

L--------------J---------------

0

0 30 60 90 120

E01 (CAD aTDC)

150 180 210 240

Figure 3-9 Droplet CAD delay from leaving injector until piston impingement

For example, the droplets, which leave the injector at 30 CAD aTDC, almost instantaneously impact the piston. On the other hand, the baseline case of EOI at 120 CAD aTDC has a delay of 18 CAD before those droplets would impact the piston. The effect of this delay is that it allows more liquid to vaporize during flight so the liquid film on any surface will have less mass.

In summary, the HC emissions decrease with an EOI from 30 to 50 CAD aTDC because of the late exhaust valve closing at 44.5 CAD aTDC. At the same time, the HC emissions from 50 to 90 CAD aTDC decrease, because the increasing flight time allows more vaporization. For late injections, the film boiling inhibits vaporization from the engine surfaces.

69

Furthermore, experiments by Iwamoto et al. [21] have researched the effect of injection timing on the volumetric efficiency (77,) of the engine.

r7V =_

Pa,i *Vd* N

(4) where rha is the mass flow rate of air, Pai is the density of the inlet air, Vd is the displacement volume, and N is the rotation speed of the engine (1/s).

Essentially, the charge-cooling effect allows a higher flow rate of air, making the engine more volumetrically efficient. However, if the fuel impacts the piston, the effective chargecooling will have less of an effect since some heat has been transferred from the piston.

Iwamoto [21] achieved a 5% improvement in volumetric efficiency illustrated in Figure

3-10.

oF 14,7

Base: Port Injection (CA 12:1)

~F

V g 5d"~

I-

U intake Stwke

30 270 IO

End of Injection deg. BTDC s

Figure 3-10 Dependence of full load performance on injection timing

For the five tests performed for this research (two with UTG 91 and two with High DI fuels), the volumetric efficiency varied between -2% and 1% change over the course of the

70

injection timing sweep, for an example see Appendix- 10. Not only were these changes nominal, but the MAP also varied with a span of .03 bar over these individual tests. Given this variation in MAP, conclusions cannot be draw about the effect of injection timing on volumetric efficiency.

3.3.1 Injection Timing Experiment with Varying Coolant

Temperatures

To further gain fundamental understanding of the HC mechanisms, the injection timing was swept for different coolant temperatures, and the data is shown below in Figure 3-11 and

Figure 3-12. Additionally, the data was plotted to aid a comparison between the two fuels in Appendix- 11 and Appendix- 12. For instance, the injection timing data is shown for coolant temperatures of 20*C and 40C for both fuels.

8000

7000

--------------

------------- ------------- --------------

-------------- --------------- ------------- -------------

6000

-------------- ---- -- ----- ------------- -------------- -------------- --------- --- ------------- -------------

5000

E

4000

CL

3000

2000

1000

------------- ------------- -------------

-+

S lOBar -UTG 91

-t--O~a UT91

MAP =

RPM

0.46-0.47

=1200 bar

30 Bar -UTG 91 Coolant = 60 C

------------- ------------- ------------- ---------------------------- ---------------------------------------

Typical Pulse Width -15 CAD

0

30

4

Intake Stroke Start

60 s0 120 150

End of Injection (CAD aTDC)

180

4

BDC

210 240

Figure 3-11 Injection timing sweep with varying coolant temperatures and UTG 91 fuel

71

10000

9000

8000

7000

------------- ---- ------ --- ------ --------- -- ------- ------------- -------------

6000

E aL a.

5000

4000 ------------- -------------- --- - - ---------------- ------------- --------------

3000

2000 --------------------------- -- - -- ---

--- 20C -HighDI

--40 C High DI

-ypi-

60 C- High DI

1000 .... .... .... .... .... . -. .... .... . --80 -H

Typica1 Pulse Width -15 CAD n

0 30

4

Intake Stroke Start

60 90 120 150

End of Injection (CAD aTDC)

AP = 0.46-04 a

RPM = 1200

Pressure = 70 bar

- -

180 t BDC

210 240

Figure 3-12 Injection timing sweep with varying coolant temperatures and High DI fuel

In both figures, the trend for each coolant temperature is very well defined and separated for injection before 90 CAD aTDC. A possible explanation for this separation is that the coolant temperatures affect the fuel cone as shown in Figure 1-16 and Figure I-I7(p.33)

[2]. While the spray cone changes shape, the more important change with increasing temperature is the smaller droplet size as shown by the SMD data in Figure 1-18. Why does the coolant temperature have a stronger effect for the early injections? First, the warmer temperatures produce smaller droplets. While this effect is generally positive across the operating map, the droplet size is more critical in the early injections when the spray has shorter flight times in which to vaporize directly into the air charge. Secondly, the additional heat in the engine is especially helpful for reducing HC emissions when the footprint of the spray is small, i.e. during early injections. Conversely, when the footprint has a larger area, the temperatures are not so necessary for vaporization.

72

3.3.2 Injection Timing Experiment with Varying Injection Pressures

Similarly, the injection timing was varied with an array of injection pressures, and the results are shown below in Figure 3-13 and Figure 3-14. Again, the data is plotted for a direct comparison of the fuels Appendix- 13 and Appendix- 14. For example, the HC levels are plotted for lower pressures of 20 and 70 bar for the UTG 91 and High DI fuels.

8000

7000

.--------------

-------------

-------------

---------------

-----.--------- ------------- ------------- -------------

6000

5000

E

0. 4000

U

3000

.--....-- --------

- -- r--U G 1 ---- - - - - - --------

3 ------------------- UTG----- ----1---

2000

1000

-*

------------- -------------- I

120 Bar -UTG 91

-------------- -RPM

10BCoolant

MP06OTa

= 1200

=60 C

------- -----------------------------------------

Typical Pulse Width -15 CAD

0

30

Intake Stroke Start

60 so

120

End of Injection (CAD aTDC)

150

180

SB6DC

210 240

Figure 3-13 Injection timing sweep with varying fuel pressures and UTG 91 fuel

73

0000

High DI

E

7000 - -------------- ------------ -------------n

T- - -- -- - --- --- r- -- -- -- -- -- -- - --- -- -- - ---o

2000 -------------- ---------- ------------ - - -- r -- .- ------ 4 .7br -- ------

Fr 120 Bar d

2000-------------------------- -------------.-70 Bar-HigDIba

High

L--------------

RPM = 1200

-------------

300

3-6

Bar- High DI io 2 CA C

8

1 4

TYical Pulse Width -15 CAD

60 0 30 fIntake Stroke Start

90 120 150

End of Injction (CAD TDC) h

180 210 240

Figure 3-14 Injection timing sweep with varying injection pressures and High DI fuel

Two features of this data will be discussed. First, the high pressure injection at 120 bar produces significantly higher HC levels with an injection timing of 120 CAD aTDC.

Secondly, the three pressures have a distinct separation for the late injection at 210 CAD aTDC, especially for the UTG 91 fuel.

For the first issue of explaining the data at 120 CAD aTDC, the air flow past the intake valve and into the cylinder is difficult to assess without an experimental or computational quantification of its behavior. From DISI literature, it is known that liquid fuel is most harmful to HC emissions when it lands in an area that is close to the exhaust valves [24], i.e. on that side of the cylinder walls. As a result of this short path, the chances for oxidizing the fuel are lower. It is possible that the dynamics of the incoming air at 120

74

CAD aTDC combined with high droplet velocities, resulting from high pressure, create a higher amount of wetting on the cylinder liner beneath the exhaust valves.

There are a couple of explanations for the second issue of the effect of fuel pressure at late injection timings. It is not expected that the standard mechanisms of deposits, oil layers, or crevices are contributors to this separation of the data. Further, the piston position has barely changed from BDC to 30 CAD aBDC so the geometry and flow field within the cylinder is expected to be similar. As a result of these similarities, the additional time to vaporize is the main difference between injecting at BDC versus 30 CAD aBDC. As discussed in the Injection Pressure Experiment section, the high pressures seem to have a detrimental affect on HC emissions because of the resulting surface wetting. Similarly, the high pressures (high velocities) seem to dominate lower pressures (larger droplets) at this late injection timing. Hence, the 4.2 ms shorter time to vaporize combined with a larger effect of surface wetting combine to produce the higher HC emissions and the separation between pressures for late injection.

75

4 Summary

4.1 Conclusions

For the Coolant Temperature Experiment, it was confirmed that the warmer coolant temperatures produced a general decrease in the HC emissions. Also, it is postulated that the effect of film boiling hinders the reduction of HC emissions between coolant temperatures of 40'C and 70'C.

The relatively high levels of HC produced by the isopentane fuel are expected to be produced by fuel contamination by the previous fuel as the isopentane was pumped through the same filter, pump and line system. In addition, the theory of significant quantities of fuel in the oil has been shown to have a minimal influence.

Regarding the Injection Pressure Experiment, the overall effect is that lower pressures produced less HC emissions. The particular details of the data show an apparent optimization between high droplet velocities and large droplets, which can both produce excessive surface impingement and problems with vaporization.

The Injection Timing Experiment verified the relatively late closing of the exhaust valve, which allowed unburned fuel to enter the exhaust. Furthermore, late injection timing after

150 CAD aTDC caused higher HC levels because of a decreasing intake air motion and the shorter time for vaporization before the spark.

76

When varying the coolant temperature for different injection timings, it was shown that the coolant temperature is especially important for early injection timings. The primary reasons for this dependence are the effect of temperature on droplet size and the necessary heat transfer for small spray footprints created at early injections. Moreover, the high pressures for very late injections, after 180 CAD aTDC, are particularly detrimental to HC because of the combined effects of more surface wetting and less amount of time to vaporize.

4.2 Future Improvements

The LabView program should be corrected to account for the 4 CAD shift that is inherent in the way the shaft encoder was installed. Every crank angle measurement given in this thesis should be increased by about 4 CAD. For the purposes of researching HC mechanisms, this shift is negligible, but for further research and MEP data, the shift should be fixed.

The intake and exhaust valve timings should be verified by an additional means. It is possible they could be measured experimentally, although the resolution of measurement would be difficult. Another potential is graphing the pressure trace on a log P- log V graph.

The pressure transducer is fitted into a sleeve with a flame arrestor on the end. This arrestor has not been cleaned during the history of the research, and most likely, substantial deposits have accumulated, which may affect its response. Per instructions from industry, the transducer should be removed so the flame arrestor can be cleaned with a solvent.

77

4.3 Future Experiments

The following suggestions are listed in order of importance for improved explanations of the data.

1. As mentioned previously, the values for the isopentane tests seem especially high since the fuel should be completely vaporized. It would be valuable to repeat these experiments, despite the cost of isopentane. One possible explanation is that the residue from the High DI fuel contaminated the isopentane enough to produce higher HC. In the future, installing a new fuel pump and filter could reduce this contamination. A further improvement would be to bypass storing the isopentane in the fuel tank or pumping it through the fuel lines, but this task is challenging because of the difficulty to fill the accumulator by a new method. Finally, the contamination from the accumulator and line to the fuel rail would be the most difficult to achieve.

2. One of the largest steps to increase the understanding of HC mechanisms in a

DISI engine is the quantification of mixing. For all future experiments, a parameter relating to mixing should be recorded or calculated after the experiment; experimental values which indicate the mixedness of the charge include PM and CO. On the other hand, the bum rate can be calculated as an indicator of in-cylinder mixing. From the contact with engineers in the industry, measuring CO seems to be the recommendation to quantify mixing.

3. The engine should always be motored before and after a test so that there is a quantification of the role that fuel desorbed in the oil has played for that

78

4. experiment. In addition, some members from industry have recommended that oil samples be taken before, after, and possibly during the experiments to quantify how much fuel has come into the oil. Not only would this information be important to the HC emissions, but the fuel reduces the lubricity of the oil, which creates a valid concern for automakers.

With the intake port deactivation enabled by the swirl flap in one of the intake runners, another set of experiments can be performed. Changing the amount of swirl may be especially insightful for its effect on the Injection Pressure

Experiment and the Injection Timing Experiment. This valve was normally open (i.e. the position in which it is spring-loaded) for the testing performed in this paper; the valve could easily be closed fully, and incremental changes could also be built into experiments.

5.

6.

In order to research the effect of deposits, the remaining three cylinders, which are completely clean, would be used to repeat the coolant temperature, injection pressure, and injection timing experiments. The spark and injection would be easy to achieve from the experimental setup, and the pressure transducer could be transferred. However, the intake and exhaust systems would need modification to operate from a different cylinder. If the mass flow rate of air is not necessary, only the exhaust system would need to be modified.

Future experimenters may consider a baseline condition at 80'C or 90'C for an easier means of comparing with other literature.

79

References

1. Cheng W. K., Hamrin D., Heywood J. B., Hochgreb S., Min K., and Norris M., "An

Overview of Hydrocarbon Emissions Mechanisms in Spark-Ignited Engines," SAE

Paper 932708, 1993.

2. Zhao, F., Harrington, D. L., Lai, M., Automotive Gasoline Direct-Injection Engines,

Society of Automotive Engineers, Inc., Warrendale, PA, 2002.

3. Fraidl, G. K., Piock, W. F., Wirth, M., "Gasoline direct injection: actual trends and future strategies for injection and combustion systems," SAE Paper 960465, 1996.

4. Heywood, J. B., Internal Combustion Engine Fundamentals, McGraw-Hill, New York,

1998.

5. Noma, K., Iwamoto, Y., Murakami, N., lida, K., Nakayama, 0., "Optimized gasoline direct injection for the European market," SAE Paper 980150, 1998.

6. Zhao, F.-Q., M.-C. Lai, and D.L. Harrington, "A Review of Mixture Preparation and

Combustion Control Strategies for Spark-Ignited Direct-Injection Gasoline Engines,"

SAE paper 970627, 1997.

7. Andreisse, D., et al., "Assessment of stoichiometric GDI engine technology,"

Proceedings of AVL Engine and Environment Conference, 1997.

8. Anderson, R. W., Yang, J., Brehob, D. D., Vallance, J. K., Whiteaker, R. M.,

"Understanding the thermodynamics of direct-injection spark-ignition (DISI) combustion systems: an analytical and experimental investigation," SAE Paper 962018,

1996.

9. Harada, J., Tomita, T., Mizuno, H., Mashiki, Z., Ito, Y., "Development of directinjection gasoline engine," SAE Paper 970540, 1997.

10. Tomoda, T., Sasaki, S, Sawada, D, Saito, A, Sami, H., "Development of direct injection gasoline engine study of stratified mixture formation," SAE Paper 970539, 1997.

11. Fraidl, G. K., Piock, W. F., Wirth, M., "Gasoline direct injection: actual trends and future strategies for injection and combustion systems," SAE Paper 960465, 1996.

12. Sandquist, H., Lindgren, R., Denbratt, I., "Sources of hydrocarbon emissions from a direct injection stratified charge spark ignition engine." SAE Paper 2000-01-1906,

2000.

13. Koga, N., Miyashita, S., Takeda, K., Imatake, N., "An experimental study on fuel behavior during the cold-start period of a direct-injection spark-ignition engine," SAE

Paper 2000-01-0969, 2001.

14. Anderson, R. W., Brehob, D. D., Yang, J., Vallance, J. K., and Whiteaker, R. M., "A new direct injection spark ignition (DISI) combustion system for low emissions,"

FISITA-96 Technical Paper P0201, 1996.

80

15. Shimotani, K., Oikawa, K., Tashiro, Y., and Horada, 0., "Characteristics of exhaust emission on gasoline in-cylinder direct injection engine," Proceedings of the Internal

Combustion Engine Symposium Japan (in Japanese), 1996.

16. Yamada, T., Gardner, D. V., Bruno B. A., Zello, J. V., and Santavicca, D. A., "The effects of engine speed and injection pressure transients on gasoline direct injection engine cold start," SAE Paper 2002-01-2745, 2002.

17. Freeland, A. R., Multiple Personal Interviews, September 2005-2006.

18. Hallgren, B.E., "Impact of Retarded Spark Timing on Engine Combustino,

Hydrocarbon Emissions, and Fast Catalyst Light-Off," MIT PhD Thesis, February

2005.

19. Price, P., Stone, R., Collier, T., and Davies, M., "Particulate Matter and Hydrocarbon

Emissions Measurements: Comparing First and Second Generation DISI with PFI in

Single Cylinder Optical Engines," SAE Paper 2006-01-1263, 2006.

20. Alger, T., Hall, M., and Matthews, R., "The Effects of In-Cylinder Flow Fields and

Injection Timing on Time-Resolved Hydrocarbon Emissions in a 4-Vavle, DISI

Engine," SAE Paper 2000-01-1905, 2000.

21. Iwamoto, Y., Noma, K., Nakayama, 0., Yamauchi, T., Ando, H., "Development of

Gasoline Direct Injection Engine," SAE Paper 970541, 1997.

22. Williams, P. A., Davy, M. H., and Brehob, D. D, "Effects of Injection Timing on the

Exhaust Emissions of a Centrally-Injected Four-Valve Direct-Injection Spark-Ignition

Engine," SAE Paper 982700, 1998.

23. Stanglmaier, R. H., Roberts, C. E., and Moses, C. A, "Vaporization of Individual Fuel

Drops on a Heated Surface: A Study of Fuel-Wall Interactions within Direct-Injected

Gasoline (DIG) Engines," SAE Paper 2002-01-0838, 2002.

24. Stanglmaier, R. H., Li, J., Matthews, R. D., "The effect of In-Cylinder Wall Wetting

Location on the HC Emissions from SI engines," SAE Paper 199-01-0502, 1999.

81

Sample Pressure Trace - Firing

Appendix

-----------------...............

--------- T ---------------

I

---------------

I -----------------

0.9 ---------------- .........------ --------

*

0.8 --------------------------------

0.7 ------------

-------------------- -

-------

----------------------- -----------------------

0.6 -------------------------------- ,

0.5

500 520 540 560

CAD

580

O"iii, --

600 620

-------

640

Sample Pressure Trace - Motoring

2 .......

1.8 ---------------- ----------

.......... --- ...

---------------

---------------

---------------

I -------------------------

......

1.6 - ---------------- I --------------- T --------------- I ---------

1.4 -------------- --------------- --------------- -------

1.2 ---------------- --------------- ---------------

----------- I

---- --------------------------------

-------------------------------------

6.

1 - ----------- ------------ ---- -------------.... .......

......... ...... ------------- -- ------- --- ------

0.8 ------------- --------------------------------- -----------

--------------- --------------- ---------------

0. C) - ------------------------------- --------------- ---

0.4 -------------------------------- --------------- ---

0.2 ---------------------------------- ....... I ............... --------------- ---------------

0

0 20 40 60 80 100 120 140

CAD

Sample Pressure Trace - Firing

3 ..............

EVO--j

2.5 ----------------- . ........... ............... ---------------

I ---------------- --------------- ---------------

2 ---------------- ---------------

--------- ---------- -------

IL

1.5 ------------------------------------------------------------

--------------- -------------------------------

- ------------- ---------- ------------- -------------

. ...........................................................................

---------------------- ---------------

0.5 ---------------- ------------------------------------------------------------ --------------- ---------------

0

280 300 320 380 400 420 340

CAD

360

Appendix- 1 Valve timing location in the pressure trace

82

120 bar 90 bar 70 bar 50 bar 20 bar y

=

10.59879x

+

0.21510 y

=

9.19802x

+

0.21668 y

=

8.07961 x

+

1.09678 y

=

6.73386x

+

1.04272 y = 4.30790x + 1.97892

1 gJ=

0.99978 R2 = 0.99968 R2 =

0.99997 R2= 0.99993 R2

=

0.99910

110 --------- ----------- J ----- ----- ---------. --- ----- ----

100 --------------- --------- ------- ------ --- ------- L------ -------- ------- I ...--- ..

90

----------------------- ------ ---- --- -------- -- -------------- ------- --.--- ------.

80

------- T----- --

IM

E

7 0

- -

---

-- - -J-- - -L

------ --------------------

- - -- J - - -L - -

------ ------- .----

-- - --

_

60 -.-------- --------- - -- - ------- - -------

A 120 bar

+

7 a

S 50 -.-------- --------- -- --------I--- ------- -------- 2 0 a

x 50 bar

S 40 --------- ------ -. -- ------- - ---- I---------------- m 20 bar

30 -.------- --- - -- ----- I - ----- L ------- ------- 1 ------- L-..----- -

Linear (50 bar)

Li e r( b r

20 ------------------------------------------------ -------

10 --- -------- -------

T-------

--..----- 1------------- -----

Linear (70 bar)

-- Linear (90 bar)

Linear (120 bar) --

0 2 4 6 8 10 12 14

Pulse Width (ms)

16 18 20 22 24

Appendix- 2 Injector calibration based on the nitrogen regulator pressure value

0

0 01203040 0 0 0

Appendix- 3 Exhaust runner time-resolved HC concentration measurements for spark timing of 150 bTDC at 3.9 bar Net-IMEP, X=1.O, CMCP, and 90"C fluids 1181.

83

RPM Spark IMEP Fluids Probe location HC ppm Cl

Timing (bar) Temp

PFI 1500 150 bTDC 3.0 90

0

C 37 cm from exhaust valve 2000-3000

DISI 1200 200 bTDC ~4.5 80

0

C 180 cm from exhaust valve 3400

Appendix- 4 Comparison of engine operating conditions and data between PFI and DISI engines 1181.

.. n.

i

9

'-1

0

0

0 50 100 150 200

Sudftc* Tmporatur. (d". C)

250 300

Appendix- 5 Drop vaporization lifetime versus surface temperature [23]

30 bar 70 bar

I.*ft.

120 bar

25 mm

Appendix- 6 Spray penetration picture at different injection pressures 0 Siemens VDO

84

0.9

0.8

0.7

I-

(U

.0

I..

U) a,

0~

0.6

-------- ------------ ---------------------- --------------------- -------- --

0.5

-------------------- ------------------- - --------------- ------------------- -----------------

0.4

- ------------------------------------------- -- - - - - ----------------- ---

0.3

0.2

0.1

0

0 50 100 150 200 250

CAD

Appendix- 7 Typical cylinder pressure bounding the range of injection pressures

-

-

-

Average - UTG 91

Average Isopentane

-2006-01-1263 Toluene

-2006-01-1263 Iso-Octane

Average - High DI

- - 2000-01-1905 Standard Valves

- - 2000-01-1905 Reverse Tumble Valves

- - 2000-01-1905 High Tumble Valves

9000

8000 ------------ ------------ -------- ---------------- r- -------------- ----------- ------------- --- .-

7000

6000

------------ ---------- -----------

----

---- l-- ---- ------------- ------------ ------..----- ----- --- - -.-

.--------------- - -- ----- ------------- ----------- ------- -- ....--- --------- .---..--- --

5000

------------- -------- ------- -- ----- --- ----- - -- ---- .------------- -

CL 4000

-- -----

3000

.-- - -------------- - ----- -- ----- -- - - - ----------- ------------- ------------

2000

1000

Typical Pulse Width ~15 CAD

0

0 rI

30

+Intake Stroke Start

60 90 120 150

End of Injection (CAD after TDC)

180

+

BDC

210 240

Appendix- 8 Comparison of injection timing sweep data with two additional papers

85

Engine Injector Comp. Speed MAP Injection Coolant Ignition IMEP

Location .

Ratio

(rpm) pressure

(bar) (bar)

Temp.

("C) Timing

GM DISI

2000-01-

1905

Side

Center

12.0

8.5

1200

750

0.46

1.0

70

50

60

?

20 bTDC -4.6

Simulate

d MBT

?

2006-01-

1263

Center ? 1500 ? ? ? 35bTDC

Appendix- 9 Operating conditions for Appendix- 8

0.35-

0.5

0.34

049

I -

0.33

----------------------

0.32

-047

0.31

0.3

E 0.29

>0

0.28-

---------4-------

------------

-0-46

-------- --------------------------------------

-- --

-- -- -- -- -

-- -- -

-- -- -- -- -- -

I -- --

-- -

-- -------- ---------- --------- I ----4----

045

:E

-0.44

-0.43

0.27

0.42

-0.41

0.26-

0.25

0 30 60 90 i

120 150

End of Injection (CAD aTDC)

180 210 2

40

0.4

Appendix- 10 Volumetric efficiency and map changes with injection timing (taken with UTG 91 fuel at baseline conditions)

86

10000

9000 - -

- -- - -- --- - -- -- - -- -- -- -- - - -- -- -- -- -- --

------------- ----- --- -- - --... -- -- - -. - - .

.........- .....-

8 00 0 ------- ----- ----- -T-------------r

7000 - -----------------------

8000.---------

000

----------------

---------------------

------------------

E a 5000

- ----------- a.

--------

X 4000 - ----------- ----------------------------- ---- -

-----------

-- ------ -- -

----

...... --.... -- ..

2000 -------------------------------

1000 Pulse- h-

Typical Pulse Width

0

0 30

+ Intake Stroke Start

60

------- .-

-

--

20 C High DI

20 C - UTG 91

40 C - High DI

-m-40 C UTG 91

90 120 150

End of Injection (CAD aTDC)

MAP= 0.46-0.47

RPM = 1200 bar ----

Fuel Pressure 70 bar

180

+

BDC

210

Appendix- 11 Injection timing sweep with low coolant temperatures and varying fuels

10000

240

8000 -.-------------- -------------- ------------- ------------- I------------- -- -1 ---........-. ----

7000 -.------------- ------------- ------------- -- ----------- ..-

8000

-- ---------------- - ---- -------.. ---- -....---..--..-

7000

--

E

.

X

-

4000 -------------------- ------------------- ---------------------------

3000

2000

1000

-------------- ------------- -------------------- --- 6..-- --H g .D-I.. .- .AP

-t-

60 C -High DI

-- 60C-UTG91

RPM= 1200

Typical Pulse Width ~15 CAD

- -P-s-

80 C -High DI

-

~UG9

Fuel Pressure;i 70 bar ----

0

0

30

+

Intake Stroke Start

60 90 120 150

End of Injection (CAD aTDC)

180 f BDC

210 240

Appendix- 12 Injection timing sweep with high coolant temperatures and varying fuels

87

8000

5000-

-- - - - - - - - - .----- -- ---- - - ------------- --------- ---

- - - ---- - -- - - - - - - - - - - ------------

E

C. 4000

C, -

3000

2000

-- -ar-H-ighDI-

-a--70 Bar- High DI

----- -------- ---------------- -- 30 Bar - High DI --------~- RPM = 1200

---------------------------

--

70 Bar UTG 91

S-------- UTG 91

1000

Typical Pulse Width -15 CAD

0

0 30 t

Intake Stroke Start

60

0

90

12

120

1

150

End of Injection (CAD aTDC)

MAP = 0.46-0.47 bar

Coolant = 60 C

------------ ----------- ------------

180

BDC

210 240

Appendix- 13 Injection timing sweep with lower injection pressures and varying fuels

8000

7000-

4

6000 -- - -- - -------------

- ---------------

5000

C,

E a. 4000

C,

------------

----- --:-------------

---------- --- --- --

3000

--------------- ---------------T -- -- r-------------

2000

------------- -------------

1000

-

Typical Pulse Width -15 CAD

0

0 30

tIntake

Stroke Start

60

-- -

-

70 Bar - High DI ----

-120 Bar UTG 91

--

4

OBar -UTG 91

-~--------

90 120 150

End of Injection (CAD aTDC)

MAP = 0.46-0.47 bar

RPM

= 1200

Coolant = 60 C

180

+BDC

210 240

Appendix- 14 Injection timing sweep with higher injection pressures and varying fuels

88