COMBUSTION AND EMISSIONS OF A DIRECT INJECTION

GASOLINE ENGINE USING EGR

by

Thomas Lattimore

A thesis submitted to

The University of Birmingham

for the degree of

DOCTOR OF PHILOSOPHY

The University of Birmingham

School of Engineering

December 2015

i
 University of Birmingham Research Archive
e-theses repository

This unpublished thesis/dissertation is copyright of the author and/or third

parties. The intellectual property rights of the author or third parties in respect
of this work are as defined by The Copyright Designs and Patents Act 1988 or
as modified by any successor legislation.

Any use made of information contained in this thesis/dissertation must be in

accordance with that legislation and must be properly acknowledged. Further
distribution or reproduction in any format is prohibited without the permission
of the copyright holder.
 Abstract

This research has examined the combustion and emissions of a spray-guided direct-injection

spark-ignition (DISI) engine using exhaust gas recirculation (EGR). The impact of EGR type,

swirl and tumble intake airflows, compression ratio and fuel type were also investigated.

EGR addition resulted in significant fuel consumption improvements and differing particulate

matter (PM) behaviour depending on the knock limited maximum brake torque (KLMBT)

spark advance achieved. When comparing EGR types, cooled EGR achieved the best fuel

consumption and cooled EGR after three-way catalyst (TWC) achieved the best gaseous

emissions (NOx and HC).

Swirl and tumble intake airflows significantly increased fuel consumption. However, these

increases could be minimized with EGR addition. Swirl significantly reduced the

accumulation mode particulate emissions, providing a potential solution for PM reduction.

EGR addition did not significantly affect PM for the swirl and tumble intake airflow

conditions.

20%vol 1-butanol addition to gasoline fuel (Bu20) resulted in significant PM reductions at

8.5 bar IMEP. At 7.0 bar IMEP, EGR addition allowed the KLMBT spark timing to be

advanced, as the compression ratio was increased. Fuel consumption was improved by 0.4%

due to the spark advance and reduced pumping losses, and PM improved because the

formation of primary particles was reduced.

ii
 Acknowledgements

First and above all, I praise God, the most beneficent, the most merciful, for providing me

with this opportunity and for granting me the capability to succeed. This thesis would not

have been possible without the assistance and guidance of several people. I would therefore

like to offer my sincere gratitude and thanks to them all.

I would like to give my warmest and most sincere gratitude to Professor Hongming Xu who

throughout the duration of my PhD program has provided me with invaluable guidance,

experience and support to enable me to successfully complete my PhD. I would like to thank

my second supervisor, Professor Miroslaw L. Wyszynski for his useful comments and

feedback for my research work, and I would like to thank Professor Akbar Ghafourian for his

guidance, help and encouragement in my annual reviews. I would also like to thank the

Engineering and Physical Sciences Research Council for providing the funding for my PhD

scholarship, and I would like to thank the University Of Birmingham for facilitating the

scholarship and my PhD studies, as well as Professor Kyle Jiang for awarding the scholarship

to me.

Great thanks go to my PhD candidate and research assistant colleagues who have provided

me with technical knowledge and a good working environment to complete my PhD studies.

I would like to thank Dr. Chongming Wang, Dr. Jose Herreros, Soheil Rezaei, Dr. Changzhao

Jiang, Tawfik Badawy, Dr. Xaio Ma, Dr. Dhananjay Srivastava, Dr. Isaline Lefort, Dr.

Arumugam Sakunthalai Ramadhas, Dr. Mohammadreza Attar, Dr. He Ma and Maria Macias,

in particular, for their technical support at various times during my studies. I would also like

to thank Yasser Al Qahtani, Rafiu Olalere, Ricky Creegan and Lucas Polglase for providing a

friendly productive working environment at the University.

iii
I am grateful to Navin Kalian from Jaguar Land Rover for his technical support during the

PhD and to Shell Global Solutions UK for their supply of fuels to the engine labs. I am also

grateful to the Automotive Laboratory staff who have provided significant technical support,

knowledge and expertise. This is in addition to manufacturing and installing new components

for the engine test cell, and maintaining the Future Engines and Fuels Laboratory at the

University Of Birmingham. They are Carl Hingley, Peter Thornton, Lee Gauntlett and Jack

Garrod.

Finally, I would like to thank my wife, Mariam Kamel for her endless love, support and care

during my PhD studies, and our families, in particular our parents, for their love and

encouragement throughout my studies.

iv
 Contents

Abstract ................................................................................................................................ ii
Acknowledgements .............................................................................................................. iii
Contents ................................................................................................................................ v
List of Figures ..................................................................................................................... vii
List of Tables ........................................................................................................................ x
List of Notations .................................................................................................................. xi
List of Publications ............................................................................................................. xv
Chapter 1 Introduction .......................................................................................................... 1
1.1 Overview ..................................................................................................................... 1
1.2 Objectives and Approaches.......................................................................................... 3
1.3 Research Outline ......................................................................................................... 4
1.4 Thesis Outline ............................................................................................................. 5
Chapter 2 Literature Review.................................................................................................. 7
2.1 Introduction ................................................................................................................. 7
2.2 History of IC Engines .................................................................................................. 8
2.3 SI Combustion ............................................................................................................. 9
2.4 Spark Ignition Fuels ................................................................................................... 22
2.5 Regulated Engine-Out Emissions ............................................................................... 29
2.6 Summary ................................................................................................................... 37
Chapter 3 Experimental Setup and Techniques .................................................................... 39
3.1 Introduction ............................................................................................................... 39
3.2 Overview of Single-Cylinder Engine Test Cell .......................................................... 40
3.3 Control Systems ......................................................................................................... 57
3.4 Instrumentation .......................................................................................................... 61
3.5 Emissions Measurement ............................................................................................ 66
3.6 Data Acquisition and Processing ................................................................................ 73
3.7 Analysis of the Uncertainties in the Recorded Data .................................................... 91
3.8 Fuel Properties ........................................................................................................... 92
3.9 Experimental Test Procedures .................................................................................... 93
3.10 Summary ................................................................................................................. 94

v
Chapter 4 The Effect of EGR and its Type on Engine Combustion and Emissions ............... 95
4.1 Introduction ............................................................................................................... 95
4.2 Experimental Procedure ............................................................................................. 96
4.3 Results and Discussion .............................................................................................. 98
4.3.1 Hot EGR Effect on Combustion and Emissions in a DISI Engine ......................................... 98
4.3.2 Effect of Different EGR Types on Combustion and Gaseous Emissions in a DISI Engine .. 124
4.4 Data Continuity ....................................................................................................... 147
4.5 Conclusions ............................................................................................................. 148
Chapter 5 The Effect of Intake Airflow and Hot EGR on Engine Combustion and PM
Emissions .......................................................................................................................... 151
5.1 Introduction ............................................................................................................. 151
5.2 Experimental Procedure........................................................................................... 152
5.3 Results and Discussion ............................................................................................ 154
5.4 Data Continuity ....................................................................................................... 175
5.5 Conclusions ............................................................................................................. 175
Chapter 6 The Effect of Compression Ratio, Fuel and EGR on Engine Combustion and
Emissions .......................................................................................................................... 177
6.1 Introduction ............................................................................................................. 177
6.2 Experimental Procedure ........................................................................................... 179
6.3 Results and Discussion ............................................................................................ 181
6.3.1 Investigation of Compression Ratio and Fuel Effect on Combustion and Emissions ........ 181
6.3.2 Investigation of the Compression Ratio and Hot EGR Effect on Combustion and PM
Emissions............................................................................................................................... 204
6.4 Data Continuity ....................................................................................................... 217
6.5 Conclusions ............................................................................................................. 218
Chapter 7 Summary, Conclusions and Suggestions for Future Work ................................. 221
7.1 Summary and Conclusions ....................................................................................... 221
7.2 Suggestions for Future Work ................................................................................... 225
Appendix A1: Analysis of the Uncertainties in the Recorded Data .................................... 228
Appendix B1: Data Continuity (Chapter 4)........................................................................ 230
Appendix B2: Data Continuity (Chapter 5)........................................................................ 232
Appendix B3: Data Continuity (Chapter 6)........................................................................ 234
Appendix C1: Detailed Specific Further Investigation for Each Research Area ................. 236
List of References ............................................................................................................. 240

vi
 List of Figures

Figure 3-1 Schematic of single-cylinder engine test facility and instrumentation setup .... 40
Figure 3-2 Schematic of water cooling and oil lubrication circuits ................................... 44
Figure 3-3 Schematic of fuel supply systems ................................................................... 45
Figure 3-4 Relative positions of the direct injector along with the injector spray plume ... 47
Figure 3-5 Cooled EGR schematic .................................................................................. 49
Figure 3-6 Hot EGR after TWC schematic ...................................................................... 50
Figure 3-7 Cooled EGR after TWC schematic ................................................................. 52
Figure 3-8 a) Low tumble baffle plate, b) swirl baffle plate, c) high tumble baffle plate .. 52
Figure 3-9 ETCS LabView program front-panel (Luszcz, 2009)...................................... 57
Figure 3-10 VVT LabView program front-panel (Luszcz, 2009) ....................................... 59
Figure 3-11 Intake and exhaust camshaft profiles .............................................................. 60
Figure 3-12 Engine ETCS and VVT control systems; their arrangement and signal flow ... 61
Figure 3-13 Calibration results of the VAF Meter ............................................................. 65
Figure 3-14 Cambustion DMS 500 analyser operating principle ........................................ 68
Figure 3-15 Schematic of the DMS 500 two-stage dilution system .................................... 69
Figure 3-16 HSDAQ LabView program front-panel (Luszcz, 2009) .................................. 74
Figure 3-17 LSDAQ LabView program front-panel (Luszcz, 2009) .................................. 75
Figure 3-18 Ricardo WAVE simulation model .................................................................. 76
Figure 3-19 Ricardo WAVE model calibration procedure ................................................. 78
Figure 3-20 In-cylinder pressure versus CAD data comparisons for Ricardo WAVE model
calibration: a) case 1, b) case 2, c) case 3, d) case 4 and e) case 5 ........................................ 80
Figure 3-21 Engine data comparison for Ricardo WAVE model calibration: a) IMEP (bar),
b) Pmax (bar), c) indicated efficiency and d) VE ................................................................... 81
Figure 3-22 Engine data percentage difference comparison for Ricardo WAVE model
calibration: a) IMEP, b) Pmax, c) indicated efficiency and d) VE .......................................... 82
Figure 3-23 In-cylinder pressure versus CAD data comparisons for Ricardo WAVE model
verification: a) case 1, b) case 2, c) case 3, d) case 4 and e) case 5 ....................................... 84
Figure 3-24 Engine data comparison for Ricardo WAVE model verification: a) IMEP (bar),
b) Pmax (bar), c) indicated efficiency and d) VE ................................................................... 85
Figure 3-25 Engine data percentage difference comparison for Ricardo WAVE model
verification: a) IMEP, b) Pmax, c) indicated efficiency and d) VE ......................................... 86
Figure 3-26 Modified HSDAQ LabView program displaying the on-line knock
amplitudes…… ................................................................................................................... 90
Figure 4-1 In-cylinder pressures versus CAD at KLMBT spark timings for a) 5.5 bar
IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP ..................................................................... 103
Figure 4-2 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) 5.5 bar IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP .................................... 105
Figure 4-3 MFB versus CAD at KLMBT spark timings for a) 5.5 bar IMEP, b) 7.0 bar
IMEP and c) 8.5 bar IMEP ................................................................................................ 109
Figure 4-4 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings .................. 112

vii
Figure 4-5 EGT versus EGR ratio at KLMBT spark timings ......................................... 114
Figure 4-6 PN emissions versus particle diameter at KLMBT spark timings at a) 5.5 bar
IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP ..................................................................... 119
Figure 4-7 NOx emissions versus EGR ratio at KLMBT spark timings .......................... 121
Figure 4-8 HC emissions versus EGR ratio at KLMBT spark timings ........................... 123
Figure 4-9 In-cylinder pressures versus CAD at KLMBT spark timings for a) hot EGR, b)
cooled EGR, c) hot EGR after TWC and d) cooled EGR after TWC.................................. 127
Figure 4-10 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) hot EGR, b) cooled EGR, c) hot EGR after TWC and d) cooled EGR after
TWC…………………..………………………………………………………………...…..130
Figure 4-11 MFB versus CAD at KLMBT spark timings for a) hot EGR, b) cooled EGR,
c) hot EGR after TWC and d) cooled EGR after TWC ...................................................... 133
Figure 4-12 MFB50 versus EGR ratio at KLMBT spark timings ................................. 134
Figure 4-13 MFB10-90 versus EGR ratio at KLMBT spark timings ............................ 135
Figure 4-14 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings............... 138
Figure 4-15 PMEP versus EGR ratio at KLMBT spark timings ................................... 140
Figure 4-16 EGT versus EGR ratio at KLMBT spark timings ...................................... 141
Figure 4-17 COVIMEP versus EGR ratio at KLMBT spark timings ............................... 142
Figure 4-18 Intake plenum temperature versus EGR ratio at KLMBT spark timings .... 143
Figure 4-19 NOx emissions versus EGR ratio at KLMBT spark timings ...................... 145
Figure 4-20 HC emissions versus EGR ratio at KLMBT spark timings ........................ 147
Figure 5-1 In-cylinder pressure verses CAD at KLMBT spark timings for a) the three
intake airflow conditions at 0% EGR, b) low tumble, c) swirl, d) high tumble ................... 159
Figure 5-2 Calculated average in-cylinder temperature verses CAD at KLMBT spark
timings for a) the three intake airflow conditions at 0% EGR, b) low tumble, c) swirl, d) high
tumble………… ............................................................................................................... 162
Figure 5-3 MFB verses CAD at KLMBT spark timings for a) the three intake airflow
conditions at 0% EGR, b) low tumble, c) swirl, d) high tumble ......................................... 166
Figure 5-4 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings .................. 168
Figure 5-5 EGT versus EGR ratio at KLMBT spark timings ......................................... 169
Figure 5-6 PN emissions versus particle diameter at KLMBT spark timings for a) the three
intake airflow conditions at 0% EGR, b) low tumble, c) swirl, d) high tumble ................... 173
Figure 6-1 In-cylinder pressure versus CAD for a) Bu20, b) E20 and c) ULG95 at KLMBT
spark timings.. ................................................................................................................... 185
Figure 6-2 Calculated average in-cylinder temperature versus CAD at KLMBT spark
timings for a) Bu20, b) E20 and c) ULG95 ........................................................................ 187
Figure 6-3 MFB versus CAD at KLMBT spark timings for a) Bu20, b) E20 and c)
ULG95……….. ................................................................................................................ 189
Figure 6-4 MFB50 versus compression ratio at KLMBT spark timings ......................... 190
Figure 6-5 MFB10-90 versus compression ratio at KLMBT spark timings .................... 191
Figure 6-6 Indicated efficiency versus compression ratio at KLMBT spark timings....... 193
Figure 6-7 EGT versus compression ratio at KLMBT spark timings .............................. 194
Figure 6-8 PN emissions versus particle diameter at KLMBT spark timings for a) Bu20, b)
E20 and c) ULG95 ............................................................................................................ 197

viii
Figure 6-9 NOx emissions versus compression ratio at KLMBT spark timings .............. 201
Figure 6-10 HC emissions versus compression ratio at KLMBT spark timings ............ 202
Figure 6-11 Overall effect of compression ratio and fuel on gaseous emissions, indicated
efficiency and total PN (integrated across 10-289 nm range) at KLMBT spark timings ..... 203
Figure 6-12 In-cylinder pressure versus CAD at KLMBT spark timings for a) 0% EGR, b)
7% EGR and c) 14% EGR................................................................................................. 207
Figure 6-13 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) 0% EGR, b) 7% EGR and c) 14% EGR ....................................................... 209
Figure 6-14 MFB versus CAD at KLMBT spark timings for a) 0% EGR, b) 7% EGR and
c) 14% EGR….. ................................................................................................................ 212
Figure 6-15 Gravimetric ISFCnet versus compression ratio at KLMBT spark timings ... 213
Figure 6-16 PN emissions versus particle diameter at KLMBT spark timings for a) 0%
EGR, b) 7% EGR and c) 14% EGR ................................................................................... 217

ix
 List of Tables

Table 2-1 EU (EN 228-2012) and US (ASTM D4814 Rev B-2011) Fuel Specification
(Analiit AA, 2015, Chevron, 2009) ..................................................................................... 23
Table 2-2 Operating conditions for research and motor octane number test methods
(Bradley, 2004, Heywood, 1988)......................................................................................... 24
Table 2-3 European Union emission legislation for DISI petrol cars (Delphi, 2012) .......... 35
Table 2-4 United States Federal emission legislation for Light Duty SI Vehicles <80,000 km
(Delphi, 2012) ..................................................................................................................... 36
Table 3-1 Experimental single-cylinder engine specification ............................................. 41
Table 3-2 Estimated swirl and tumble ratios for the three intake airflow conditions ........... 53
Table 3-3 Summary of other researchers findings for swirl and tumble ratios .................... 54
Table 3-4 Single-cylinder engine camshaft geometry......................................................... 59
Table 3-5 SMPS 3936 system settings ............................................................................... 71
Table 3-6 Horiba MEXA-7100DEGR specification ........................................................... 72
Table 3-7 Test cases for Ricardo WAVE model calibration ............................................... 79
Table 3-8 Test cases for Ricardo WAVE model verification .............................................. 82
Table 3-9 Test Fuel Properties ........................................................................................... 92
Table 3-10 Gasoline GC Analysis ....................................................................................... 93
Table 4-1 Experiment test matrix (EGR addition & engine load) ....................................... 98
Table 4-2 Experiment test matrix (EGR addition & EGR type) .......................................... 98
Table 4-3 KLMBT spark timings (˚bTDC) (* = not knock limited) (EGR addition & engine
load)……… ...................................................................................................................... 100
Table 4-4 KLMBT spark timings (˚bTDC) (EGR addition & EGR type) ......................... 125
Table 4-5 Results summary (EGR addition & engine load) (highlighted=improvement,
underlined=worsening)...................................................................................................... 149
Table 4-6 Results summary (EGR addition & EGR type) (highlighted=improvement,
underlined=worsening)...................................................................................................... 150
Table 5-1 Experiment test matrix (intake airflow & EGR addition) ................................. 154
Table 5-2 KLMBT spark timings (˚bTDC) (intake airflow & EGR addition) ................... 156
Table 5-3 Results summary (intake airflow & EGR addition) (highlighted=improvement,
underlined=worsening)...................................................................................................... 176
Table 6-1 Experiment test matrix (compression ratio & fuel) ........................................... 181
Table 6-2 Experiment test matrix (compression ratio & EGR addition)............................ 181
Table 6-3 KLMBT spark timings (˚bTDC) (compression ratio & fuel) ............................ 183
Table 6-4 KLMBT spark timings (˚bTDC) (compression ratio & EGR addition) ............. 205
Table 6-5 Results summary (compression ratio & fuel) (highlighted=improvement,
underlined=worsening)...................................................................................................... 219
Table 6-6 Results summary (compression ratio & EGR addition)
(highlighted=improvement, underlined=worsening) .......................................................... 220

x
 List of Notations

AFR Air Flow Rate

aTDC After Top Dead Centre

BDC Bottom Dead Centre

BOC British Oxygen Company

bTDC Before Top Dead Centre

Bu20 20%vol 1-butanol in gasoline

CAD Crank Angle Degrees

CC Cubic Capacity

CCD Charge Coupled Device

CE Cooled EGR

CFR Cooperative Fuel Research

CI Confidence Interval

CID Combustion Initiation Duration (Crank angle from spark discharge to MFB5)

CLD Chemiluminescence Detector

COVIMEP Coefficient of Variation of IMEP

CO Carbon Monoxide

Comb. Combustion

CO2 Carbon Dioxide

DC Direct Current

DI Direct-Injection

DISI Direct-Injection Spark-Ignition

(dQ/dθ) Net Heat Release Rate

EC European Commission

EGR Exhaust Gas Recirculation

EGT Exhaust Gas Temperature

xi
EPSRC Engineering and Physical Sciences Research Council

ETCS Engine Timing Control System

EU European Union

EVC Exhaust Valve Closing

E20 20%vol ethanol in gasoline

FID Flame Ionization Detector

FTIR Fourier Transform Infra-Red

FTP Federal Test Procedure

GC Gravimetric Content

HC Unburned Hydrocarbon

HE Hot EGR

HoV Heat Of Vaporisation

HSDAQ High Speed Data Acquisition

IC Internal Combustion

ICE Internal Combustion Engines

IMEP Indicated Mean Effective Pressure

IN Inputs

ISFCnet Net Indicated Specific Fuel Consumption

IVO Intake Valve Opening

KLMBT Knock Limited Maximum Brake Torque

LHV Lower Heating Value

LSDAQ Low Speed Data Acquisition

MBT Maximum Brake Torque

MF Methylfuran

MFB Mass Fraction Burned

MFB10 Crank Angle at 10% Mass Fraction Burned

MFB10-90 Combustion Duration (Crank Angle Degrees from 10 to 90% MFB)

xii
MFB5 Crank Angle at 5% Mass Fraction Burned

MFB50 Crank Angle at 50% Mass Fraction Burned

MFB90 Crank Angle at 90% Mass Fraction Burned

MON Motor Octane Number

Net Mass Flow Rate of Fuel into the Combustion Chamber

NA Naturally Aspirated

NDIR Non-Dispersive Infra-Red

Nd:YAG Neodynium-Doped Yttrium Aluminium Garnet

NEDC New European Driving Cycle

NI National Instruments

NO Nitrous Oxide

NOx Nitrous Oxides

η Thermal Efficiency

OUT Outputs

PAH Polycyclic Aromatic Hydrocarbon

PAI Polyamid-Imide

PFI Port Fuel Injection

PID Proportional Integral Controller

PIV Particle Image Velocimetry

PLIF Planar Laser Induced Florescence

PM Particulate Matter

PN Particulate Number

PMEP Pumping Mean Effective Pressure

Pmax Maximum Pressure

PRF Primary Reference Fuel

Rc Compression Ratio

RDE Real Driving Emission

xiii
ROHR Rate Of Heat Release

RON Research Octane Number

RPM Revolutions Per Minute

SI Spark-Ignition

Sig. Significant

SMPS Scanning Mobility Particle Sizer

TDC Top Dead Centre

TWC Three-Way Catalyst

T90 90% Distillation Temperature

ULG Unleaded Gasoline

US United States

VAF Volumetric Airflow

Vc Clearance Volume

VVT Variable Valve Timing

Vd Swept Volume

VE Volumetric Efficiency

VVT Variable Valve Timing

WOT Wide Open Throttle

γ Heat Capacity Ratio

∆vapH Heat Of Vaporization

λ Air-Fuel Ratio

xiv
 List of Publications

1. Investigation of EGR Effect on Combustion and PM Emissions in a DISI Engine,

Thomas Lattimore, Chongming Wang, Hongming Xu, Miroslaw L. Wyszynski, Shijin Shuai,

2016, Applied Energy, 161, 256-267, doi:10.1016/j.apenergy.2015.09.080.

2. Investigation of Compression Ratio and Fuel Effect on Combustion and PM

Emissions in a DISI Engine, Thomas Lattimore, José M. Herreros, Hongming Xu, Shijin

Shuai, Fuel, 169, 68-78, doi: 10.1016/j.fuel.2015.10.044.

3. Impacts of Low-level MF content in gasoline on DISI engine combustion

behaviour and emissions, Chongming Wang, Hongming Xu and Thomas Lattimore. 2013.

SAE technical paper 2013-01-131, doi:10.4271/2013-01-1317.

4. Fuel Effect on Particulate Matter Composition and Soot Oxidation in a Direct-

Injection Spark Ignition (DISI) Engine, Chongming Wang, Hongming Xu, Jose Martin

Herreros, Thomas Lattimore and Shijin Shuai. 2014. Energy & Fuels, 28, 2003-2012, doi:

10.1021/ef402234z.

Awards:

TI Group Scholarship (PG) for outstanding academic performance

Thomas Lattimore, School of Mechanical Engineering, University Of Birmingham, March

2013

xv
 Chapter 1

Introduction

The aim of this chapter is to provide an overview of the PhD research investigation

conducted by the author. The study of fuel consumption reducing techniques of EGR (also

used to reduce NOx), swirl and tumble intake airflows, geometric compression ratio increase

as well as the study of biofuels is driven by the demand for reduced fuel consumption and

hence reduced CO2 emissions from vehicles. This is in addition to, in the case of biofuels, the

demand for renewable energy supply and increased energy supply security. The overall

research outline, objectives and investigation approach are presented briefly, after which the

outline for the thesis follows.

1.1 Overview

While many of us dream of a future where electric cars charged by renewable means are

commonplace and the small demand for a traditional engine sound is met by clean hydrogen

combustion, this dream is still a long way away. Crude oil supplies have at least 50 years

remaining (BP, 2015), if not more, especially because of the ongoing discovery and

exploitation of shale oil reserves, which has been a significant factor in the oil price slump we

are experiencing in 2015-16. Therefore, there is not a strong economic incentive for industry

to invest in the technology and infrastructure required for electric and hydrogen powered

vehicles. There are, however, legislative factors that are driving the research and

development in the automotive industry; most significantly, emissions legislation and tax

incentives. The progress though is not towards wide-scale electric vehicle use. It is towards

increased engine and vehicle efficiency through engine downsizing and vehicle hybridization,

amongst other methods, as well as reduced vehicle emissions produced by the efficiency

1
improvements and improved emissions after-treatment. Thus, for the meantime, conventional

spark-ignited and compression-ignited internal combustion (IC) engines are here to stay.

However, they will be increasingly downsized with increasing amounts of hybridization to

recover waste power from the engine and the vehicle to improve their overall efficiencies.

In spark-ignition (SI) engines, many technologies are being widely utilized to improve fuel

economy and to reduce emissions. These technologies include EGR, improved combustion

using swirl and tumble generated intake airflows, optimised compression ratio, DI,

turbocharging, variable valve lift and timing, cam profile switching and lean-burn stratified

charge combustion. These improve the fuel economy of the engine itself and reduce its

emissions, although PM and some gaseous emissions can potentially be increased. More

efficient catalytic converters and improved particulate filters are being utilized to reduce the

emissions of the vehicle as a whole, by converting the emissions from the engine into less

harmful substances.

Biofuels, particularly bio-ethanol, are also believed to be part of the solution to reduce

vehicle emissions. This is due to their oxygenated nature which reduces combustion

temperatures, providing decreased NOx emissions, along with reduced HC and PM emissions

through improved post-combustion oxidisation. The ability to produce bio-ethanol and

similar biofuels using renewable methods greatly reduces their net CO2 output; the CO2

produced during combustion is absorbed from the atmosphere when the raw plant material

used to produce the fuel is grown.

While EGR, swirl and tumble intake airflows, compression ratio and biofuels have been

studied in detail previously by other researchers, their effect on combustion in DISI engines

and their subsequent emissions has been much less thoroughly investigated, particularly

2
regarding their effect on PM emissions. Therefore, their impact on DISI engine combustion

and emissions needs to be assessed.

1.2 Objectives and Approaches

The main research objective of this investigation was to study the impact of EGR on the

combustion and emissions, in particular the PM emissions, of a DISI engine. A spray-guided

single-cylinder DISI thermal research engine was used for the engine experiments. PM

emissions were measured using a Cambustion DMS 500 and a TSi SMPS 3936. A Horiba

MEXA-7100DEGR gas analyser was used to measure the gaseous emissions of nitrous

oxides (NOx) and hydrocarbons (HC).

The following are the areas that were investigated throughout the author’s PhD study:

 The effect of EGR and its type on engine combustion and emissions

 The effect of intake airflow and EGR on engine combustion and PM emissions

 The effect of compression ratio, fuel and EGR on engine combustion and emissions

The novel ideas investigated in this thesis include:

 Investigation of the effect of different types of EGR (hot EGR, cooled EGR, hot EGR

after TWC, cooled EGR after TWC) on engine combustion and gaseous emissions at

the same engine load using a simulated TWC (this removes the effect of varying

TWC conversion efficiencies which result in varying EGR ratios)

 Investigation of the effect of 1-butanol blending into gasoline on engine combustion

and emissions (particulate and gaseous) at different compression ratios

 Unlike most other investigation, the PM analysis in this thesis has classified the

particles into nucleation and accumulation modes, which have been studied

individually, where appropriate

3
1.3 Research Outline

The research presented in this thesis was conducted by the author at the Future Engines and

Fuels research laboratory at the University of Birmingham with help and support from Jaguar

Land Rover and Shell Global Solutions UK. The research was mainly conducted to

investigate the effects of EGR on the combustion and emissions of a DISI engine.

The study of engine combustion includes an analysis of the KLMBT spark timing achieved at

different operating conditions with the engine. It includes a detailed and comprehensive

analysis of the in-cylinder pressure trace and mass fraction burned (MFB) profiles. A Ricardo

WAVE engine model has been utilized to calculate the in-cylinder temperature in order to

assist in the engine combustion study. Fuel consumption behaviour has been analysed in

detail to quantify the effect of the combustion at different engine conditions on this

parameter. Pumping mean effective pressure (PMEP), exhaust gas temperature (EGT) and

coefficient of variation of IMEP (COVIMEP) have been analysed too in order to further

quantify the combustion effects. Regulated emissions of PM, NOx and HC were then

examined in order to study the effect of the engine combustion on these critical parameters,

which are the most important regarding legislative requirements.

In order for biofuels to be commercialised, their combustion and emissions must be

quantified and subsequently shown to be competitive with fuels currently on the market (i.e.

gasoline). Therefore, this thesis also presents research comparing gasoline and oxygenated

fuels blended into gasoline.

4
1.4 Thesis Outline

This thesis is composed of seven chapters. A brief outline of each chapter is presented below.

Chapter 2- Literature Review

This chapter reviews the literature that is relevant to this investigation. DI and spark timing

are talked about in detail as well as the phenomenon of engine knock. The current trend of

engine downsizing is introduced along with fuel consumption improvement and CO2

emission reduction, with EGR, swirl and tumble intake airflows and compression ratio

explored in detail. After this, spark ignition fuels including oxygenated alternatives are

discussed. Finally, a discussion into the regulated engine-out emissions and related legislation

is provided.

Chapter 3- Experimental Setup and Techniques

This chapter gives detailed information of the engine and instrumentation setup, as well as

detailed information of the data acquisition and recording systems. The emission analysers

are introduced after which key calculations for engine parameters used in this thesis are

presented. Finally, the details of the Ricardo WAVE engine model are provided along with

the details of the fuels used in this investigation.

Chapter 4- The Effect of EGR and its Type on Engine Combustion and Emissions

This chapter provides details of the combustion and emissions characteristics of a DISI

engine operated with hot EGR at different engine loads of 5.5, 7.0 and 8.5 bar IMEP, and

with different EGR types of hot EGR, cooled EGR, hot EGR after TWC and cooled EGR

after TWC. In the first part, the hot EGR addition was increased across the different engine

loads, and in the second part the EGR addition was increased at a single engine load of 7.0

5
bar IMEP, and the EGR type is changed. The combustion and emission parameters at the

different conditions are compared to quantify the effect of EGR and its type on them.

Chapter 5- The Effect of Intake Airflow and Hot EGR on Engine Combustion and PM
Emissions

This chapter provides details of the combustion and PM emission characteristics of a DISI

engine operated with swirl and high tumble intake airflows, and with hot EGR addition. The

EGR addition was increased across the two intake airflow conditions along with a baseline

low tumble condition at a single engine load of 7.0 bar IMEP. The combustion and PM

emission parameters at the different conditions are compared to quantify the effect of intake

airflow and hot EGR on them.

Chapter 6- The Effect of Compression Ratio, Fuel and Hot EGR on Engine Combustion
and Emissions

This chapter provides details of the combustion and emissions characteristics of a DISI

engine operated with different compression ratios and different fuels, and with hot EGR

addition. In the first part, the two fuel splash blends of Bu20 and E20 were tested along with

a baseline of gasoline at a single engine load of 8.5 bar IMEP, with the compression ratio

increased after each of the fuel blends and the reference fuel were tested. In the second part,

gasoline fuel was tested at a single engine load of 7.0 bar IMEP, with the hot EGR addition

increased along with the compression ratio. The combustion and emission parameters at the

different conditions are compared to quantify the effect of compression ratio, fuel type and

hot EGR on them.

Chapter 7- Summary and Conclusions, and Recommendations for Future Work

This chapter summarizes the research conducted in this investigation and provides the key

conclusions of this thesis. Recommendations for future work are then given.

6
 Chapter 2

Literature Review

The aim of this chapter is to review the literature that is relevant to this investigation. The

main areas of discussion are SI engine combustion and emissions.

2.1 Introduction
Firstly, this chapter provides a brief introduction regarding the history of IC engines and

explains the key fundamental differences between spark and compression ignition engines.

Spark-ignited combustion is then further discussed, after which DI and spark-timing which

result in the combustion and emissions from the engine are talked about in detail. The

phenomenon of knock is then discussed with the factors affecting knocking tendency

explained.

Secondly, the current and ever ongoing trend of improved vehicle fuel economy and reduced

CO2 emissions is introduced followed by detailed discussion of the potential of EGR, swirl

and tumble intake airflows and geometric compression ratio increase to enable these

improvements to be achieved. SI fuels are then explored due to their significant effect on

combustion and emissions, including oxygenated alternatives, due to their increased use in

motor vehicles and due to their predicted increased future usage.

Finally, regulated engine-out emissions are discussed, including an overview of the current

legislation in both Europe and the US, due to their strong influence on the research and

development of automotive engines. TWCs are then briefly discussed due to their strong

influence on engine-out emissions.

7
2.2 History of IC Engines
The concept of the IC engine can be traced back to 1680 when Dutch Physicist Christian

Huygens designed a gunpowder fuelled IC engine (Stone, 1999, The Institute of Historical

Research, 2004). During the next 200 years further designs were proposed and built to

advance the concept further. The most significant breakthrough came with Nicolaus Otto who

built the first practical four-stroke IC engine in 1876 (Wu, 2007). The Otto cycle formed the

underlying working principle for the engine operation and it became the universally adopted

cycle for all liquid fuelled spark-ignited IC engines to this day. Later in 1893, Rudolph Diesel

achieved compression ignition in an engine operated using the Diesel cycle which is utilized

to this day in diesel compression ignition engines (Bennett, 2013).

The fundamental difference between gasoline and diesel engines is not the type of ignition

but the type of combustion that occurs, with spark-ignited engines typically burning with a

pre-mixed flame and compression-ignited engines typically burning with a diffusion flame

(Stone, 1999). In order to obtain a reliable ignition and combustion with a pre-mixed flame,

the fuel-air mixture of a SI engine must always be closed to stoichiometric, which

necessitates the use of a throttle valve to control the amount of air entering the combustion

chamber, which in return increases engine pumping work and thus reduces cycle efficiency

(Stone, 1999). SI engines also suffer from knock, which limits their compression ratio thus

limiting their thermal efficiency (Gupta, 2013). However, in diffusion flames, the fuel-air

mixture is only required to be approximately stoichiometric at the flame front, which allows

the engine to be de-throttled and the fuel injection to be controlled to produce the required

power output (Stone, 1999). Thus, there are no throttling losses, significantly increasing the

cycle efficiency as compared to a gasoline engine, resulting in improved part-load fuel

economy (Stone, 1999). Also, knock is not a limitation in CI engines, thus, they can be

8
operated with higher compression ratios, significantly improving their thermal efficiency as

compared to SI engines (Reitze Jr., 2001). However, PM is a significant problem in CI

engines due to the diffusion flame, necessitating the use of a diesel particulate filter, which

increases pumping losses (Bennett, 2010). The pre-mixed flame produces little PM meaning

the vehicle does not require a particulate trap. Thus, it does not suffer from any additional

pumping losses as a result of the higher exhaust back-pressure (Agarwal et al., 2014).

Today the main goal of engine researchers is to improve engine efficiency in order to reduce

engine CO2 emissions through reduced fuel consumption, while complying with ever tighter

emissions regulations (Zaccardi et al., 2009). This has led to the increased use of DI,

turbocharging and increased compression ratios (Heywood and Welling, 2009, Zaccardi et

al., 2009).

2.3 SI Combustion
During the compression stroke of the SI engine cycle, the piston causes the fuel-air mixture to

become highly compressed before a spark ignites the mixture causing it to combust. The

spark leaves a small nucleus of the flame that propagates into the unburned gas, and it is not

until that nucleus grows into a size which is the same order as that of the turbulence scale that

the flame propagation can be enhanced by in-cylinder turbulence (Stone, 1999). This period

is often referred to as the early burn, delay period or combustion initiation duration (CID) and

it comprises the initial laminar combustion and the transition to fully turbulent combustion,

typically lasting 1-2ms or 9-18 CAD at 1500 rpm (Stone, 1999). The crank angle interval

between the spark discharge and the MFB5 or MFB10 parameters are typically used to define

this period (Stone, 1999). The duration of this period depends on the temperature, pressure

and composition of the fuel-air mixture (Stone, 1999).

9
Between MFB10 and MFB90 is the main combustion period which is dominated by turbulent

combustion, the second stage of which occurs shortly after the peak pressure and it is affected

by the same parameters as the early burn period along with turbulence (Stone, 1999). The

flame in the final combustion stage makes contact with more of the combustion chamber

walls and consequently it makes less contact with the remaining unburned fuel-air mixture,

thus resulting in a significantly slower combustion (Stone, 1999).

2.3.1 Direct-Injection
DI is a technology that was initially developed for diesel engines but it was not until

relatively recently that it was incorporated into gasoline production engines, with the modern

DISI engine being first produced in 1996 by Mitsubishi, although the concept was initially

invented by Frenchman Léon Levavasseur in 1902 (Institution of Mechanical Engineers,

2013, Mitsubishi Motors, 1999). It has an advantage of improved fuel economy and power

output in comparison to PFI because the fuel can be injected into the combustion chamber at

the best time; it is not dictated by the intake valve opening time as with PFI (Harada et al.,

1997). DI benefits from charge cooling; the fuel cools as it is injected into the combustion

chamber, causing the knocking tendency to reduce, allowing the spark timing to be

subsequently advanced, as well as improving volumetric efficiency (Shimotani et al., 1995,

Yang and Anderson, 1998). This contributes to fuel economy and power output

improvements at medium-high loads (Alger et al., 2000, Daniel et al., 2012b, Harada et al.,

1997). Both DI and PFI provide an advantage compared to carburettor type fuel injection

because much more precise fuel metering is achieved which improves fuel economy and

reduces engine emissions (Institution of Mechanical Engineers, 2013, Robert Bosch GmbH,

2006). Spray-guided DI first produced by Mercedes in 2006 is the most widely used DI

method due to its improved thermodynamic efficiency and fuel economy compared to other

10
DI systems, such as wall-guided and air-guided DI, in addition to improved HC and PM

emission behaviour (Green Car Congress, 2006).

DI can be used to realize either homogenous or stratified combustion. Traditionally

homogenous combustion has been used to refer to that produced by PFI and stratified

combustion has been used to refer to that produced by DI. However, within DI itself there are

homogenous and stratified concepts (Harada et al., 1997, Takagi and Skippon, 1998). With an

early injection of fuel directly into the cylinder during the induction stroke, a more

homogenous mixture of approximately λ=1 can be achieved (Harada et al., 1997, Takagi and

Skippon, 1998). When the fuel is injected late into the cylinder during the compression stroke

a more stratified mixture is formed with an overall lambda of ~20 and a combustible mixture

of λ~1 around the spark plug (Harada, 1997, Takagi and Skippon, 1998). With homogenous

mixture formation, the mixture preparation is assisted by the high flow velocities and their

aerodynamic forces in the area of the opening and closing intake valves, helping the fuel to

evaporate quickly, ensuring it is well homogenized (Robert Bosch GmbH, 2006). With

stratified mixture formation, a mixture cloud is created which can then be transported into the

area around the spark plug by the air flows in the combustion chamber and by the upward

stroke of the piston (Robert Bosch GmbH, 2006). The mixture formation benefits from the

higher pressures and temperatures in the combustion chamber during the compression stroke

of the engine cycle (Robert Bosch GmbH, 2006).

2.3.2 Spark-Timing
Spark timing has a significant effect on engine combustion and emissions. Regarding engine

efficiency and thus fuel economy, greater expansion work can be produced on the piston as

the spark is advanced and the pressure is consequently increased at TDC. However,

advancing the spark also increases the compression work of the engine as the in-cylinder

11
pressure is increased during the compression stroke. Thus, there is a trade-off between the

compression and expansion work in the IC engine to achieve the best engine efficiency, and

this is controlled by the spark timing (Heywood, 1988). An optimum MFB50 phase of 8-

10˚aTDC is required to produce best engine efficiency and thus the best fuel economy (De O.

Carvalho et al., 2012). At low to medium engine loads, the spark timing can be chosen in

order to produce the optimal combustion phasing. However, at high engine loads, knock

limits the spark advance that can be achieved, meaning that more retarded spark timings are

required (Checkel and Dale, 1989). Emissions are also significantly affected by the engine

spark timing (Heywood, 1988); something that will be discussed in detail in a later section.

2.3.3 Knocking Phenomena

2.3.3.1 Introduction to Knocking Phenomena

The phenomenon of knock has existed for the majority of the history of IC engines, with the

most widely publicised solution to the problem being high octane fuels (Checkel and Dale,

1989, Kalghatgi, 2013). These increase the knock tolerance of the engine which in turn

allows the spark timing to be advanced for fuel economy and torque benefits. However, they

also increase the cost of the fuel because high octane fuel fractions are more expensive than

those with a low octane number (Checkel and Dale, 1989). The fuel economy benefits also

reduce the engine CO2 output, which along with fuel consumption is at the forefront of the

public perception (Gov UK, 2011, Kalghatgi, 2013). Therefore, it is anticipated that demand

for higher octane fuels will only increase in the future (Green Car Congress, 2013, Kalghatgi,

2014, Pawlowski and Splitter, 2015).

Knock itself is a shock wave generated by the auto-ignition of the fuel-air mixture in the end-

zone of the combustion chamber once it has exceeded a critical temperature of approximately

625-725˚C (Kalghatgi, 2013). This most commonly occurs when the spark timing is

12
advanced to an extent that the pressure and temperature in the engine cylinder are increased

to a critical level where the flame front generated from the spark discharge radiates

significant heat to the fuel-air mixture in the end-zone, causing its pressure and temperature

to increase through compression (Heywood, 1988). This occurs because the density of the

burned mixture increases by approximately a factor of four causing not only the unburned

mixture to be compressed but also compressing those parts of the charge that have already

burned and displacing them back towards the spark plug (Heywood, 1988). This causes the

temperature of the fuel-air mixture in the end-zone to rise beyond 625-725˚C resulting in

auto-ignition (Kalghatgi, 2013).

The auto-ignition produces two or more flame fronts in the combustion chamber which

rapidly radiate heat to the remaining unburned fuel-air mixture, causing it to combust more

rapidly than in normal combustion (Rothe et al., 2006). When its resulting shock wave makes

contact with the piston face, it produces structural damage rather than exerting a pressure to

push it (Oppenheim, 2004). The high pressure and temperature rise rates produced by the

knock produce very high mechanical and thermal loads in the pistons, bearings, cylinder head

and the cylinder head gasket, causing damage and even failure to these components (Amann

and Alger, 2012, Robert Bosch GmbH, 2006). The flame front of this shock wave can

propagate at a rate of 10 to 100 times that of normal combustion and it resonates on the

combustion chamber walls producing the audible “ping” which is characteristic of engine

knock (Hamilton et al., 2008).

Whether knock occurs or not reflects the outcome of a race between the advancing flame

front and the pre-combustion reactions in the unburned end gas. Knock will not occur if the

flame front consumes the end gas before these reactions have time to cause the fuel-air

13
mixture to auto-ignite (Heywood, 1988). However, if these reactions are given sufficient

time, they will produce auto-ignition before the flame front arrives (Heywood, 1988).

2.3.3.2 Factors affecting Knocking Tendency

There are several factors which affect the tendency of an SI engine to knock. The most

common is substantial spark advance which produces higher in-cylinder temperatures which

leads to higher temperatures in the end-zone of the combustion chamber, resulting in ever

greater levels of knock as the spark is continually advanced (Ulsoy et al., 2012). As the

torque and load demand from the engine increase, the charge density is increased which leads

to higher temperatures during the compression stroke, causing the engine knock tendency to

increase (Robert Bosch GmbH, 2006). By definition, lower octane number fuels will knock

more than those with a higher octane rating, thus, knock tendency increases with their use.

Increasing the compression ratio increases knock tendency because the pressures and

temperatures within the fuel-air charge are increased during the compression stroke (Robert

Bosch GmbH, 2006). Deposits in the combustion chamber also increase the effective

compression ratio, increasing knock tendency (Rajput, 2005, Robert Bosch GmbH, 2006).

Ineffective cooling within the engine can lead to hot spots forming within the combustion

chamber and it can lead to overall temperature increases; both of which increases knock

tendency (Rajput, 2005). Poor turbulence and swirl characteristics of an engine, produced by

the relevant engine geometries can increase knock tendency (Robert Bosch GmbH, 2006).

The use of homogenous fuel injection increases knock tendency because the mixture in the

end-zone of the combustion chamber is ignitable whereas with stratified injection, only the

mixture around the spark plug is rich enough to ignite (Robert Bosch GmbH, 2006). NO

introduced into the intake gas by EGR increases knock tendency (Hoffmeyer et al., 2009,

Lewis et al., 2014, Roth et al., 2010, Takaki et al., 2014, Yuan et al., 2015), but this is

14
overcome by the higher heat capacity of the exhaust gas compared to fresh air (Francqueville

and Michel, 2014), resulting in knock tendency reducing overall. Finally, the use of PFI

injection increases knock tendency in comparison to DI because it does not benefit from the

charge cooling effect that occurs with DI (Alger et al., 2000, Harada et al., 1997).

2.3.4 Downsizing, Fuel Consumption Improvement and CO 2 Emission

Reduction
The current trend in gasoline engine technology is aggressive downsizing which aims to

decrease the engine capacity while maintaining or improving power output, in order to

improve fuel economy and reduce emissions, through reduced pumping losses and other

efficiency gains (Cairns et al., 2006, Castagné et al., 2003, Lecointe and Monnier, 2003).

Technologies such as turbocharging and DI are used in order to achieve these aims (Castagné

et al., 2003, Lecointe and Monnier, 2003, Song et al., 2014). In addition, techniques such as

EGR, swirl and tumble intake airflows and increased compression ratio are being used to

improve the fuel economy and thus reduce the CO2 output of modern engines. Therefore,

EGR, swirl and tumble intake airflows and compression ratio increase have been researched

further in this investigation to help develop the knowledge of engine researchers in these

areas. These are discussed further in the following sections.

2.3.5 EGR
EGR, both internal and external, as well as lean mixtures, can be used to reduce the power

density and thus reduce the energy deposition rate of the mixture to improve engine knock

resistance (Francqueville and Michel, 2014, Galloni et al., 2013, Nishida et al., 1988),

allowing increased boosting in turbocharged engines as well as increased compression ratios

(Galloni et al., 2013). This however reduces the maximum engine load and torque, as well as

the maximum power density, meaning that it can only be utilized at part-load conditions. It

can most effectively be used in fast-burn engine designs such as the widely used four-valve

15
pentroof combustion chamber; the design produces significant tumble which leads to a rapid

burn (Stone, 1999, Urushihara et al., 1995). It is most effective in these designs because EGR

reduces the laminar flame speed of the fuel-air mixture, thus, a fast-burn balances the EGR

well (Endres et al., 1992, Heywood, 1988). This allows EGR to be utilised to reduce NOx

emissions, which is particularly important for stratified combustion at low loads, or it allows

it to be utilised to improve part-load fuel economy (Davis and Borgnakke, 1982, Heywood,

1988). However, EGR potentially increases particulate and HC emissions, thus, this limits its

use in modern DISI engines (Alger et al., 2009a, Diana et al., 1996, Wei et al., 2012, Zhong

et al., 2013).

Part-load fuel economy is improved because the throttle has to be more opened to achieve a

particular load, due to the EGR gases replacing some of the fresh air volume in the

combustion chamber (Stone, 1999). This reduces the pumping work required of the engine,

thus improving fuel economy (Abd-Alla, 2002, Grandin et al., 1998, Robert Bosch GmbH,

2006, Wei et al., 2012). This is important at high loads because stratification for the purpose

of reducing pumping losses is not possible. Thermal efficiency is also improved with EGR

addition due to reduced combustion temperatures (Ratnak et al., 2015, Siokos et al., 2015),

contributing to fuel economy improvements. As mentioned, the spark timing can be advanced

due to the improved knock resistance with EGR addition, allowing the optimum 8-10˚aTDC

MFB50 phasing to be achieved, also improving the fuel economy (De O. Carvalho et al.,

2012). EGTs are also reduced with EGR which can be useful for protecting the inlet turbine

of turbocharged engines (Alger et al., 2009a). However, the EGR system must be sufficiently

robust to withstand the deposits that accumulate in the system due to the lower EGTs

(Heywood, 1988, Robert Bosch GmbH, 2006).

16
Another drawback of EGR is that COVIMEP is increased, and eventually some cycles become

sufficiently slow burning with high EGR ratios that combustion is not completed by the time

the exhaust valve opens, eventually causing engine misfire (Matsushita et al., 1985, Siokos et

al., 2015, Wei et al., 2012). Thus, it is necessary to advance the spark in order to ensure that

the combustion is sufficiently completed before the exhaust valve opens. Depending on the

engine operating condition, the optimum 8-10˚aTDC MFB50 phasing can either be achieved

at a certain spark timing and EGR addition, or the combustion can degrade so much that

despite advancing the spark timing, the combustion duration is too long to achieve a

sufficiently advanced MFB50 combustion phasing, resulting in partial burns and misfire (Pan

et al., 2014). Therefore, there is a maximum rate of EGR that can be achieved while keeping

COVIMEP below a 5% maximum (Dunn-Rankin, 2008). Furthermore, combustion efficiency

is reduced at high EGR levels due to the increased combustion duration (Siokos et al., 2015,

Wei et al., 2012).

Typically higher engine loads are more tolerant to EGR addition but again, there is a trade-off

because as mentioned, the diluting effects of EGR reduce the maximum load that the engine

can achieve, and no EGR addition can be achieved at full-load without using an external

pump due to the equivalent intake and exhaust pressures. Thus, the maximum EGR addition

that can be achieved is limited (Institution of Mechanical Engineers, 2013).

2.3.5.1 EGR Type

Cooled EGR has been shown to be more effective in suppressing engine knock than hot EGR

due to reduced temperatures in the end-zone of the combustion chamber (Alger et al., 2009a),

while EGR after a TWC (catalysed EGR) further reduces knocking tendency because NO

present in EGR gases before a TWC induces knock (Hoffmeyer et al., 2009, Lewis et al.,

2014, Parsons et al., 2015). The reduced knock allows the spark timing to be advanced

17
resulting in fuel economy improvements. It also allows the compression ratio to be increased,

also resulting in fuel economy improvements (Su et al., 2014). Fuel economy is also

improved as a result of the lower in-cylinder temperatures and thus improved thermal and

volumetric efficiencies achieved by cooling the EGR gases (Su et al., 2014, Wei et al., 2012).

However, the maximum EGR addition achievable with cooled EGR may be limited because

of water droplet formation (Siokos et al., 2015) and it has been observed to increase COV IMEP

and HC emissions (Wei et al., 2012). Despite the work that has been done, much more

investigation needs to be conducted into the effect of different EGR types (hot EGR, cooled

EGR, hot EGR after TWC, cooled EGR after TWC) on engine combustion and emissions.

2.3.5.2 Stratified EGR

Most commonly, EGR is used in its homogenous form where the intake air is well-mixed

with the EGR gases. However, some researchers have investigated stratified EGR to help

overcome the reduction of burning speed, the HC emission increase and the increase in

COVIMEP (Wei et al., 2012). This is where the fresh air and EGR gases are separated in the

combustion chamber, which improves flame propagation at the spark plug region (Wei et al.,

2012). However, there are significant challenges in implementing stratified EGR. Due to the

flow structure in the combustion chamber being extremely complex, it is very difficult to

achieve complete separation of the fresh air and EGR gases, and it is challenging to maintain

the stratification throughout the intake and compression strokes prior to ignition (Wei et al.,

2012). Therefore, this is why stratified EGR is not pursued in this investigation.

2.3.6 Swirl and Tumble Intake Air Flows

Turbulence, commonly quantified as swirl and tumble, is generated as a result of the

induction and compression strokes of the engine cycle, along with the geometry of the

combustion chamber (Dunn-Rankin, 2008). Swirl is defined as organized rotation of the fuel-

18
air charge about the cylinder axis, while tumble is defined as organized rotation of the fuel-air

charge around an axis perpendicular to that of the cylinder axis, at half the height of the

combustion chamber (Matsushita et al., 1985, Stone and Ladommatos, 1992). While the swirl

motion will undergo some decay during the engine cycle due to friction, intake swirl usually

persists through the compression and combustion, and it is converted into general turbulence

during the expansion processes (Lee et al., 2000, Li et al., 2001).

Swirl is typically used to improve fuel air mixing (improved mixture homogeneity) and to

increase the speed of the combustion process, particularly with air-guided fuel injection in

order to transport the fuel-air mixture towards the spark plug (Agarwal et al., 2014,

Nagayama et al., 1977). For pre-mixed combustion in SI engines, the effect of increased in-

cylinder turbulence breaks-up or wrinkles the flame front, which has the effect of increasing

the area of the flame front increasing the combustion speed (Lee, 2007, Stone, 1999).

Increasing the in-cylinder turbulence allows leaner mixtures to be burned which are less

prone to knock, due to the reduced combustion temperatures; improving fuel economy and

affecting engine emissions (Nagayama et al., 1977, Stone, 1999). PM emissions, as well as

NOx and HC emissions will be affected, even with stoichiometric conditions (Kim and Kim,

1995, Mikulec et al., 1988, Urushihara et al., 1995). Indeed, (Mehta et al., 2001) reported that

the use of swirl valves resulted in significant improvements in PM mass as compared to low

tumble conditions. Furthermore, swirl motion improves COVIMEP, particularly as engine

speed is increased by creating consistent in-cylinder flow conditions (Zeng et al., 2015).

It is commonly thought that knock is reduced as a result of the increased in-cylinder

turbulence due to the faster combustion; the fuel-air charge in the end-zone of the combustion

chamber is consumed before it has an opportunity to auto-ignite (Hirooka et al., 2004).

However, it must be noted that a faster flame will also increase in-cylinder pressures which

19
will make auto-ignition more likely. In general, faster burning cycles are more prone to knock

(Kalghatgi, 2013, Leppard, 1982). Furthermore, swirl allows EGR, as discussed previously,

to be used at high ratios (Heywood, 1988, Nagayama et al., 1977).

A disadvantage of generating engine swirl and tumble intake airflows is that the volumetric

efficiency (VE) is reduced which can cause the fuel economy to worsen (Arcoumanis and

Kamimoto, 2009). This also limits the EGR addition that can be achieved, since there is a

smaller pressure difference between the exhaust and intake. Although the use of lean

mixtures will reduce the efficiency of a TWC causing emissions to increase, the emissions

improvements produced by the lean burn combustion may be sufficient to offset this loss

(Stone, 1999). Finally, excessive tumble can lead to an increase in HC emissions and an

increase in fuel consumption (Stone, 1999).

2.3.6.1 Swirl and Tumble with EGR Addition

Using swirl and tumble along with EGR addition allows the engine to operate under learn

burn conditions in order to improve fuel economy and NO x emissions, while maintaining HC

emissions at the same level (Endres et al., 1992). In addition, high tumble airflows can be

used to achieve high EGR additions of up to 30-40% which can be used to improve fuel

economy (Ikeya et al., 2015, Takahashi et al., 2015, Zhang et al., 2014a). High swirl airflows

could also be used for the same purpose. COVIMEP can also be reduced with a combination of

tumble and EGR addition as compared to using EGR alone, due to enhanced turbulent flow in

the cylinder (Zhang et al., 2014a). In particular, because EGR allows the spark timing to be

advanced, it allows the point of ignition to occur earlier in the engine cycle where the

turbulence is stronger. This increases the laminar flame speed allowing the combustion

slowing effects of EGR to be overcome.

20
2.3.7 Compression Ratio
Increased compression ratios can lead to significant fuel economy improvements (Gumbleton

et al., 1976, Heywood, 1988). In an engine in which the compression ratio was raised from

9.7:1 to 14.6:1, an efficiency gain of up to 15% was obtained with larger gains at part throttle,

due to the improved thermal efficiency (Stone, 1999). The EGT is decreased when

compression ratio is increased, due to more efficient conversion of the fuel energy into work

on the piston (Heywood, 1988).

However, there are also disadvantages to increased compression ratios. The fuel

consumption can worsen under low-speed, high-load conditions at high compression ratios

due to spark retardation caused by heavy knocking (Okamoto et al., 2003). NOx emissions

increase due to higher in-cylinder temperatures. Large squish areas and poor surface-to-

volume ratios increase the hydrocarbon emissions (Stone, 1999). This is in addition to the

increased hydrocarbons resulting from the higher in-cylinder pressures which cause more

HCs to be stored in the piston crevices (Heywood, 1988). When released during the

expansion stroke these do not burn fully, thus contributing to HC emission increases with

increased compression ratios (Heywood, 1988, Zhang et al., 2014b). They also result from

the lower EGTs which reduce the rate of HC oxidation (Heywood, 1988). Finally,

combustion deposits become more significant as the compression ratio is increased because

the ratio of their volume compared to the volume of the combustion chamber increases (Liiva

et al., 1992, Stone, 1999).

2.3.7.1 Compression Ratio and EGR

Compression ratio increase with EGR addition has been shown to significantly improve fuel

economy while reducing the sum of the NOx and HC emissions (Diana et al., 1996). The

21
combination of the two has also been shown to produce a relatively low COV IMEP due to

increased turbulence intensity and reduced cyclic variations (Pan et al., 2014).

2.4 Spark Ignition Fuels

Fuels are typically mixtures of hydrocarbons whose carbon and hydrogen bonds are broken in

order to form new bonds with oxygen atoms when the fuel is burned in air. This process

produces a significant amount of energy which is converted into work on the piston which

then undergoes a number of conversions to produce power at the vehicle wheels.

2.4.1 Gasoline
The size and geometry of the hydrocarbon molecule has a significant effect on its chemical

properties, providing a wide-range of potential fuels that can be utilized in IC engines (Stone,

1999). Indeed, gasoline is a mixture of many different sized alkane and alkene hydrocarbon

molecules with varying chain lengths and isomerisation. Hydrocarbons can also form

aromatic and cyclo-alkane ring structures such as benzene, which are also present in gasoline

fuel in varying quantities and types.

Overall, the two most important characteristics of gasoline are its volatility and octane

number (Stone, 1999). Gasoline fuel is required to vaporize well within a given temperature

range that is season dependent in most countries so that it can produce an ignitable fuel-air

mixture when injected into the combustion chamber. If it vaporizes too easily then it can

result in problems such as vapour lock and excessive evaporative emissions (Srinivasan,

2001). However, if it is not volatile enough it will be difficult to produce an ignitable fuel-air

mixture in cold conditions, leading to poor cold startability (Kalghatgi, 2013, Robert Bosch

GmbH, 2006). If the fuel does not vaporize well enough it can also pass into the engine oil on

the combustion chamber walls, causing it to become diluted (Kalghatgi, 2013, Robert Bosch

22
GmbH, 2006). Finally, poorly vaporized fuel can lead to excessive deposit formation (Liiva

et al., 1992). Fuel octane number limits the spark timing advance that can be achieved with

the engine, and thus the fuel economy and torque, because of engine knock, as discussed.

Thus, fuel octane number requirements have increased and will continue to increase in the

coming years, in order to achieve further fuel economy improvements (Green Car Congress,

2013). The fuel specifications for gasoline fuel in the EU and US are provided in Table 2-1.

Table 2-1 EU (EN 228-2012) and US (ASTM D4814 Rev B-2011) Fuel Specification
(Analiit AA, 2015, Chevron, 2009)

Parameter EU US
Density (kg/m³) 720-775 (at 15°C) -
1
RON 95 -
MON 85 -

Distillation Properties
T10 (°C) - ≤70
T50 (°C) - ≤77-121
T90 (°C) - ≤190
End of distillation point (°C) - ≤205
Evaporation Properties
Percentage evaporated at 70˚C
20-48 -
summer (volume %)2
Percentage evaporated at 70˚C
22-50 -
Winter (volume %)2
Percentage evaporated at 100˚C
46-71 -
(volume %)
Percentage evaporated at 150˚C
≥75 -
(volume %)
Oxygenates
Aromatics (m/m) ≤35% ≤20.9%
Olefins (%m/m) ≤18% ≤11.9%
Sulphur (ppm) ≤10 ≤80
1
Member States may decide to continue to permit the placing on the market of unleaded regular grade petrol with a
minimum motor octane number (MON) of 81 and a minimum research octane number (RON) of 91.
2
Depends on volatility class of region, country and season

The octane number scale is used to define the octane number for a given fuel; n-heptane is

assigned an octane number of 0 and isooctane is assigned an octane number of 100. The

octane number of a given fuel is thus defined as the mixture of n-heptane and isooctane

which produces the equivalent knocking tendency of the given fuel. For example, if the fuel

23
has the same knocking tendency as a fuel mixture of 95% octane and 5% heptane, then it is

given an octane number of 95 (Burton et al., 2000). For octane numbers above 100, a

tetraethyl lead knock additive is added to the isooctane fuel in order to allow the octane

number to be quantified (Heywood, 1988). This is effectively just an extrapolation of the 0-

100 octane scale.

The two most widely used tests for defining the knock tendency of a given fuel are the

Research Method (ASTM D-2699) and the Motor Method (ASTM D-2700), both of which

are conducted in a standardized single-cylinder engine (Heywood, 1988, Kalghatgi, 2013).

The test for MON differs from that used for the RON test by using pre-heated mixtures,

higher engine speeds and variable ignition timing. Thus, the fuel being examined is subjected

to more stringent thermal demands, which is why MON figures are typically lower than those

for RON (Bradley et al., 2004, Robert Bosch GmbH, 2006). The engine itself was developed

by the Cooperative Fuel Research (CFR) Committee. Thus, it is commonly referred to as a

CFR engine (Bock et al., 2013). The CFR engine is a robust four-stroke overhead valve

engine with a compression ratio that can be varied from 3 to 30 while operating (Heywood,

1988). The operating conditions for the two tests are shown in Table 2-2.

Table 2-2 Operating conditions for research and motor octane number test methods
(Bradley, 2004, Heywood, 1988)

Research Method Motor Method

Inlet Temperature 52˚C 149˚C

Engine Speed 600 rpm 900 rpm

Spark Advance 13˚bTDC 19-26˚bTDC (varies

with compression ratio)

Air/fuel ratio Adjusted for maximum knock

24
Overall, the test conditions are selected to represent the engine operating range where knock

is most severe. The air-fuel ratio is adjusted to achieve the most severe knock, then the

compression ratio is increased to produce knock of a standardized intensity as measured by a

knock detector (Heywood, 1988).

However, there are some limitations of the standard tests conducted in the CFR engine; both

RON and MON are evaluated at low engine speeds and the air/fuel ratio is adjusted to

produce the maximum knock (Stone, 1999). All practical gasolines are mixtures of aromatics,

olefins, naphthenes, oxygenates and paraffins which have auto-ignition chemistries different

from those of n-heptane and isooctane (Heywood, 1988). Compared to primary reference

fuels (PRF’s), real fuels become more resistant to knock as pressure is increased for the same

temperature, thus, the MON method of increasing the intake temperature is limited.

Furthermore, because alcohols burn more rapidly than gasoline, their ignition timings in the

RON and MON tests are over-advanced, leading to an underestimate of their knock

performance (Heywood, 1988). Therefore, engines can behave rather differently during their

daily use from what these standard tests would otherwise suggest. Due to these limitations, an

average of the RON and MON values is typically taken in order to produce an antiknock

index which offers an improved characterization of the antiknock properties of the fuel

(Robert Bosch GmbH, 2006).

Fuel volatility is significantly affected by the chemical structure of the fuel, with shorter

carbon chains having a higher volatility than longer chains, and branched carbon chains

having a higher volatility than straight chains with the same number of carbon atoms (Burton

et al., 2000). The chemical structure of the fuel also significantly affects its RON and MON.

Fuel octane number decreases with increasing chain length, it increases with increasing

numbers of side chains for the same number of carbon atoms and it increases with ring

25
structures such as those found in cyclo-alkanes and aromatics (Kalghatgi, 2013). This has

significant effects on achieving ever higher octane number fuels, because of concerns over

the carcinogenic properties of aromatics (Golding and Watson, 1999).

2.4.2 Oxygenated Alternatives

In recent years, interest in alcohol fuels has grown significantly due to their renewable nature

(although many are still made from crude-oil stocks), and along with other oxygenated

molecules, they are important components of modern day gasoline fuel (Hamilton et al.,

2008, Heywood, 1988). The main drawback of oxygenated fuels is their lower gravimetric

calorific values as compared to pure hydrocarbons, due to their partially oxygenated nature

(Daniel et al., 2012a, Kalghatgi, 2013). This typically results in higher engine-out CO2

emissions (Al-Hasan, 2003, Hsieh et al., 2002, Masum et al., 2014, Varol et al., 2014)

because more fuel needs to be burned in order to achieve the same power output. The reduced

power output can be overcome with compression ratio increases however, with 40-60%vol

ethanol in gasoline blends being shown to have a similar performance to gasoline (Goswami

et al., 2015), thus reducing the CO2 increases. Furthermore, alcohols such as ethanol have a

high octane number and high enthalpy of vaporization which allows the spark timing to be

advanced as compared to regular gasoline fuel, enabling the fuel consumption penalty to be

reduced and torque to be improved (Al-Hasan, 2003, Deng et al., 2013a, Irimescu et al.,

2015, Poitras et al., 2015). The high enthalpy of vaporization also reduces the charge

temperature which improves VE and the lower air/fuel ratios mean that more energy is

released per kilogram of stoichiometric mixture burned; both of these lead to overall

efficiency improvements (Ramadhas, 2011).

Oxygenated molecules such as ethanol and butanol burn more quickly than gasoline, leading

to enhanced combustion stability and improved engine efficiency (Deng et al., 2013b, Fu et

26
al., 2014, Masum et al., 2013, Turner et al., 2011). CO emissions are typically reduced due to

the greater oxidation resulting from the fuel oxygen content (Al-Hasan, 2003, He et al., 2003,

Karavalakis et al., 2014, Varol et al., 2014, Yejian et al., 2007). Reduced combustion

temperatures with oxygenated fuels reduce the NOx emissions and EGT (Elfasakhany, 2014)

and their oxygenated nature increases the rate of oxidation of PM and HC emissions (Chen et

al., 2015, Goswami et al., 2015, Gu et al., 2012, Irimescu et al., 2015, Mattimaricq, 2012,

Wang et al., 2014a, Zhang et al., 2015). Furthermore, (Wang et al., 2014c) observed that soot

produced from the combustion of oxygenated fuels has lower activation energies than that

produced by gasoline, resulting in increased oxidation rates in the exhaust. Some researchers

have reported increased NOx emissions with oxygenated fuels due to their increased fuel

oxygen content (Bayraktar, 2005, Feng et al., 2013, Graham et al., 2008, Keskin et al., 2011,

Masum et al., 2014) which is important to note. However, this may result from the air/fuel

ratio in their engines not being changed from that of gasoline when the oxygenated fuel was

used, resulting in an excess of oxygen in combustion and hence increased NO x emissions.

On the other hand, despite the improved knock properties resulting from the higher heat of

vaporization, the cold start performance can worsen because the fuel becomes harder to

vaporize (Aleiferis et al., 2013). The higher heat of vaporization combined with the typically

reduced vapour pressures and boiling points of oxygenated fuels in comparison to gasoline

means that they have to be injected earlier in the engine cycle to form a homogenous mixture,

reducing flexibility in engine calibration. Oxygenated fuels can absorb moisture which can

lead to corrosion and fuel separation problems (Hu et al., 2012, Stone, 1999). Alcohols can

also damage materials used in the fuel injection equipment and they may cause elastomer

swelling and corrosion (Heywood, 1988, Robert Bosch GmbH, 2006).

27
Hydrous ethanol provides benefits over conventional anhydrous ethanol such as lower cost

due to the simpler manufacturing process (final ethanol dehydration step is removed) and

reduced NOx emissions resulting from lower peak in-cylinder temperatures (its water content

absorbs heat) (Masum et al., 2013). However, problems arise when the hydrous ethanol is

blended into gasoline fuel because of the immiscibility of water and gasoline, limiting the

amount of water that can be used in such mixtures (Stone et al., 2012).

1-butanol while not widely used in 1-butanol-gasoline blends has the potential to be used in

the future with similar blend ratios as ethanol-gasoline blending; therefore, 1-butanol has

been researched in this investigation (Fountain, 2012). This is because its energy content is

higher than ethanol, butanol is less prone to water contamination and it is less corrosive than

ethanol (Szwaja and Naber, 2010). Therefore, this will allow it to be distributed in the same

infrastructure that is currently used for gasoline, significantly reducing the infrastructure costs

as compared to ethanol (Szwaja and Naber, 2010). It must be noted however that 1-butanol

has been observed to have poorer cold startability as compared to ethanol, because it is harder

to ignite at cold conditions (20˚C) (Serras-Pereira et al., 2013), possibly limiting the blend

ratio of 1-butanol in gasoline. Its production costs are also higher (Masum et al., 2013)

providing a further drawback. Although 2-butanol (iso-butanol) has potential advantages over

1-butanol (higher octane number of 101 compared to 96 for 1-butanol, lower boiling point of

99.5˚C compared to 117.7˚C for 1-butanol resulting in better cold startability) (Liu et al.,

2013), it was not investigated because of the greater availability of 1-butanol in the

marketplace.

Ethanol is used on a wide scale only in its blended forms of up to 10%vol and in the near

future this trend is likely to grow with ethanol-gasoline blends between 20-40%vol (Hamilton

28
et al., 2008, Stein et al., 2013, Wallner et al., 2013). This is being driven by a 10% minimum

target for the use of alternative fuels in transportation in the EU legislated by Directive

2009/28/EC in 2009, and by the Renewable Fuel Standard in the Energy Independence and

Security Act of 2007 in the USA (Daniel, 2012d, Directive 2009/28/EC, 2009, Splitter and

Szybist, 2014). Therefore, ethanol fuel blending in gasoline has also been researched in this

investigation. In order to improve the usefulness of the investigation, the blended form is

tested which not only allows the effect of ethanol addition to gasoline on DISI engine

combustion and emissions to be quantified; it also allows the precise effects of one of the

most relevant ethanol-gasoline blends on DISI engine combustion and emissions to be

quantified.

2.5 Regulated Engine-Out Emissions

2.5.1 PM Emissions
Solid particles created as a result of incomplete combustion are known as particulates which

consist of chains of carbon particles, i.e. soot, onto which uncombusted or partially

combusted hydrocarbons form deposits (Robert Bosch GmbH, 2006). According to (Zhang et

al., 2014b), the nucleation mode particles mainly result from droplets formed by hydrocarbon

condensation, as approximately measured by the HC emission, and the accumulation mode

particles are mainly composed of carbonaceous agglomerates formed in local rich-fuel zones

(Burtscher, 2005, Kittelson, 1998). Furthermore, there is a positive correlation between

polycyclic aromatic hydrocarbon (PAHs) and accumulation mode particles (Zhang et al.,

2014b).

The Euro VI (2014) emission legislation which has for the first time limited particulate

number (PN) for DISI engines, in addition to the mass limit already imposed by Euro V

29
(2009), poses a significant challenge for engine researchers; thus, it is an important area of

research (Hedge et al., 2011). The PN emissions are higher than those produced by a diesel

engine with a particulate filter; therefore, this is one reason why the legislation has become

stricter for DISI engines (Mathis et al., 2005, Wang, 2014b).

Due to the longer period of time they have been regulated for, PM mass emissions have

become relatively well controlled in comparison to PN emissions. In addition to this, ultra-

fine PM particles, in particular, have a negative impact on human health (Anderson et al.,

2012b, Hull et al., 2005, United States Environmental Protection Agency, 2014). Therefore,

there is a greater need to reduce PN which is why it is studied in detail in this investigation,

rather than PM mass. This is despite the PN being less consistent than PM mass (Eastwood,

2008). The smallest particles are at the limit of detection of PM measurement instruments,

causing the numbers of smaller particles recorded to vary more significantly than larger

particles which are more easily detectable (Eastwood, 2008). These smaller sized particles are

often the most numerous in the exhaust emission resulting in the more significant variation in

PN (Eastwood, 2008).

PN formation is worse with homogeneous combustion (early fuel injection) as compared to

stratified combustion (late fuel injection) due to less nuclei adsorption onto accumulation

mode particles occurring, because a lower number of accumulation mode particles are

produced in homogenous combustion as compared to stratified (Eastwood, 2008). This is a

result of reduced piston wetting and improved mixture formation (Eastwood, 2008). For this

reason, homogenous combustion has been studied in this investigation (start of injection

(SOI) =280˚bTDC) at medium to high engine loads (5.5-8.5 bar IMEP) where homogenous

combustion would be used (Eastwood, 2008).

30
Overall, the regulations are dictating the current and future direction of engine development

with growing emphasis being placed upon the reduction of PM mass and number based

emissions from DISI engines (Delphi, 2012). PFI produces less PM than DI because there is

typically more time for fuel-air mixing before combustion (Alger et al., 2000, Zhang et al.,

2014b) and there is no wall wetting and fuel impingement onto the piston crown as there is

with DI. Thus, the PN emissions of different vehicle types can be summarized as:

Diesel > DISI > Diesel + Particulate Filter > PFI

Particle formation in DISI engines can be classified into two stages. The first stage is the

formation of the particles; the first condensed phase material arises from the fuel molecules

via their oxidation and/or pyrolysis products which typically include various unsaturated

hydrocarbons (Heywood, 1988). The condensation reactions of gas-phase species such as

these result in the production of nucleation particles. The second stage of particle growth,

which includes both surface growth and aggregation, is where the bulk of the solid-phase

material is generated. This involves the bonding of gas-phase species to the surface of the

particles and their incorporation into the particulate phase (Heywood, 1988). Gaseous

particles also adsorb onto the particles in the engine exhaust (Heywood, 1988).

Overall, the formation mechanism of particulates in DISI engines is much the same as in

Diesel engines in the sense that it is highly dependent on local fuel equivalence ratio and

charge temperature (Piock et al., 2011). The greatest formation rates are thus typically found

in areas of the combustion chamber with a high local fuel equivalence ratio (rich) and a low

local charge temperature. However, it is important to note that lower charge temperatures can

also decrease PM formation through reducing the numbers of primary carbon particles

formed by the thermal pyrolysis and dehydrogenation reaction of fuel vapour/droplets (Zhang

et al., 2014b), thus complicating PM behaviour. Oxidation of the particulates occurs post-

31
combustion in the expansion and exhaust strokes, and in the exhaust system of the engine,

converting them to CO and CO2 as well as less complex particles (Heywood, 1988, Piock et

al., 2011).

EGR is generally considered to have a negative effect on engine particulates, however, not

many research studies have been published investigating this issue (Alger et al., 2009a,

Zhong et al., 2013). From the limited research available, it can be concluded that EGR

reduces the need for fuel enrichment to reduce EGTs in turbocharged engines, causing PM

mass and number to reduce significantly (Alger et al., 2010). However, much more research

needs to be conducted into the effect EGR has on engine particulates.

2.5.2 NOx and HC Emissions

Typically over 90% of the NO x produced in combustion is NO, however, this is readily

oxidize once it is released from the engine and it can go on to react with unburned non-

methane hydrocarbons in the presence of ultraviolet light to form photochemical smog

(Stone, 1999). Therefore, there is little benefit in measuring just NO or NO2 output from an

engine; rather it is more useful to measure NOx. NO forms in the high temperature burned

gas, where there is sufficient temperature for nitrogen in the air to oxidise, which typically

occurs at temperatures above 1525˚C (Heywood, 1988, Nieman et al., 2012). Pure NO2 is a

toxic reddish-brown gas with strong odour which can cause irritation of mucous membranes

when present in high concentrations, while some hydrocarbons are considered to be

carcinogenic when people are exposed to them for a long time (Robert Bosch GmbH, 2006).

In addition, some of their products produced when they are exposed to sunlight are also

considered to be carcinogenic (Robert Bosch GmbH, 2006).

32
Emissions of NOx and HC vary between different engines and are affected by ignition timing,

load, EGR ratio, speed and air/fuel ratio (Stone, 1999). Regarding spark timing, typically

advancing the spark timing will increase in-cylinder pressures and temperatures, increasing

the formation of NOx and increasing the oxidation of HCs in the combustion chamber,

resulting in increased NOx and decreased HC. Retarding the spark timing will typically

reduce the in-cylinder pressures and temperatures, causing the NOx emissions to decrease and

the HC emissions to increase. Their real-life behaviour however, is more complicated than

this simplification. More advanced spark timings will lead to poorer mixture preparation,

which alone would lead to higher HC emissions. However, as the spark becomes more

advanced, the time available for post-combustion oxidization is increased, which alone would

lead to lower HC emissions. These factors interact with each other, which can sometimes lead

to unexpected PM emission behaviour. Increasing the engine load will increase the in-

cylinder pressure and temperatures, which will typically increase NO x emissions and reduce

HC emissions. As mentioned, EGR addition significantly reduces NOx emissions; however,

HC emissions are typically increased (Alger et al., 2009a, Davis and Borgnakke, 1982, Zhong

et al., 2013).

Overall, HC emissions in SI engines originate from several sources. Firstly, unburned fuel-air

escapes into the exhaust during the valve overlap period (Stone, 1999). Secondly, misfires or

partial burns in which the mixture fails to ignite or the combustion was too slow or started too

late, respectively, causes HC emissions (Heywood, 1988, Robert Bosch GmbH, 2006).

Thirdly, absorption/desorption from oil films, combustion deposits and crevices occurs as the

pressure is rising and falling within the engine cylinder (Heywood, 1988, Tomita et al.,

1996). Fourthly, a quench layer of unburned and partially burned fuel is left at the

combustion chamber wall when the flame is extinguished as it approaches it, which can be

33
released during the exhaust stroke (Tomita et al., 1996). Finally, poor mixture preparation,

particularly during transients and cold start, leads to the partial oxidization of HCs in the

engine cylinder, causing HC emissions (Stone, 1999).

In this investigation the NO x emission from the engine was studied because as shown in

Table 2-3, the legislation regarding NO x has become stricter from 2009 (Euro V), making

these emissions important to investigate to help further control them. Although the THC

emission regulation has remained consistent since 2005 (Euro IV), they are still useful to

study because of their relationship with the NOx and nucleation mode particulate matter (PM)

emissions (Zhang et al., 2014b). CO emissions are also harmful to human health. The CO

molecule attaches itself to haemoglobin in the bloodstream, preventing oxygen from being

transported by red blood cells, making it harder for the body to circulate oxygen to vital

organs, resulting in nausea, headaches and other health problems. Despite this, CO emissions

were not studied because their regulated numbers have been consistent from 2005 (Euro IV),

thus, it was considered more useful to study other engine combustion and emission

parameters.

2.5.3 CO2 Emissions

CO2 emissions are at the forefront of public perception due to their greenhouse gas effect and

subsequent effect on global temperature and climate (Gov UK, 2011). The transportation

sector as a whole is a significant contributor to global CO 2 emissions; therefore, there is a

growing political drive to reduce CO2 emissions from automobiles (Johnson, 2010, Zaccardi

et al., 2009). While oxygenated fuels typically produce increased engine-out CO2 emissions

as compared to gasoline because of their reduced lower heating value (LHV) which increases

fuel consumption, as discussed, they are typically produced using renewable methods.

Therefore, net CO2 output from the use of these fuels is reduced (Daniel, 2012c et al., Owen

34
and Coley, 1995, Parag and Raghavan, 2009, Turner et al., 2011). Improvements in fuel

consumption also reduce CO2 output from the transportation sector (Zaccardi et al., 2009).

Therefore, oxygenated fuels and fuel consumption have been researched further in this

investigation.

2.5.4 Legislation
The evolution of European Union emission legislation for DISI petrol cars with a mass of less

than 1.305 kg is shown in Table 2-3.

Table 2-3 European Union emission legislation for DISI petrol cars (Delphi, 2012)

Tier Date Unit CO THC NMHC NOx HC + PM P (#/Km)

NOx
Euro I 1992 2.72 - - - 0.97 - -

Euro II 1996 2.2 - - - 0.5 - -

Euro III 2000 2.3 0.20 - 0.15 - - -

mg/km
Euro IV 2005 1.0 0.10 - 0.08 - - -

Euro V 2009 1.0 0.10 0.068 0.060 - 0.005 -

Euro VI 2014 1.0 0.10 0.068 0.060 - 0.005 6 x 1011*

* 6 x 1012/km within first three years from Euro 6 effective dates

The emissions limit themselves were defined firstly in 1970 (Directive (70/156/EC)) and they

are tested over the New European Drive Cycle (NEDC); a synthetically generated test cycle

consisting of phases of constant cruising speeds and constant rates of acceleration (Robert

Bosch GmbH, 2005). In the near future, they will be tested over a Real Driving Emission

(RDE) test cycle to more accurately measure the pollutant emissions by using real world

driving conditions rather than a synthetically generated test cycle, which bears little relevance

to real world driving conditions (European Council, 2016). US emission legislation is

considered to be more stringent than the European emission legislation, with the Federal Test

Procedure (FTP) drive cycle mirroring actual highway driving conditions which are harder to

35
specifically calibrate an engine for. This legislation is summarized in Table 2-4; it is

applicable for vehicles up to 6,356 Kg (Robert Bosch GmbH, 2005).

Table 2-4 United States Federal emission legislation for Light Duty SI Vehicles <80,000
km (Delphi, 2012)

Emission Unit Tier I Tier II

1994 2004
NMHC 160 65

NOx 250 90
mg/km
CO 2100 1050

PM (mass) 50 15

PM (number) #/km - -

There is a certain degree of flexibility with the US legislation because each vehicle

manufacturer must only ensure that the exhaust emissions for its total vehicle fleet do not

exceed the specified average; not every vehicle from the fleet must meet the regulations

(United States Environmental Protection Agency, 2010). Only a certain percentage of the

manufacturer’s fleet must comply with the legislation too, but the percentage is increasing

year-on-year in order to reduce the emissions across the vehicle fleets of manufacturers

(United States Environmental Protection Agency, 2010).

For both legislations, vehicles are randomly tested during their lifetime in order to ensure that

their emissions have not become too degraded, thus adding an extra layer of difficulty for

vehicle manufacturers (Robert Bosch GmbH, 2005).

36
2.5.4 TWCs
Despite the ability of TWCs to reduce the emissions from a gasoline vehicle significantly,

particularly once it is activated (warmed-up), it is still important to reduce the emissions

produced by the engine itself. This is because emissions increase significantly during cold-

start when the catalyst is not at its optimum operating temperature, causing significant

emissions to be produced by the vehicle (Frennet and Bastin, 1995). The NEDC begins from

when the engine is started so if the emissions produced by the engine alone are not below a

certain limit, the emissions produced during cold-start could potentially cause the vehicle to

fail the emissions test (Isermann, 2014). In addition, when the engine undergoes acceleration,

the air/fuel charge is typically enriched as well as during cold-start, meaning that the TWC

conversion efficiency is reduced because it is not operating at its optimum stoichiometric

condition (Alger and Mangold, 2009). Finally, there is a strong relationship between the

nucleation mode particles and HC emission (Zhang et al., 2014b); therefore, it is still useful

to study the HC emission to aid the study of PM emissions.

2.6 Summary
Overall, this literature review discussed the development of modern DISI technology,

particularly in relation to fuel consumption improvement and emissions reduction. The major

areas include discussion of the knocking phenomena, EGR and relevant emission regulations.

Engine downsizing was introduced after which fuel economy improving and thus CO2

reducing techniques of EGR (also used to reduce NOx emissions), swirl and tumble intake

airflows and compression ratio increase were discussed. SI fuels and emission regulations

then follow due to their strong influence, particularly the latter, on the development of IC

engines.

37
The main focus of the review is to provide a background to the fuel economy improving and

thus CO2 reducing techniques which are researched further in this investigation. Compression

ratio increase helps to accomplish the objectives of engine researchers to increase engine

efficiency and thus reduce fuel consumption, producing subsequent CO2 emissions benefits.

However, knock, which has been discussed in detail, is a limiting factor regarding the

maximum compression ratio that can be achieved, limiting engine efficiency. Therefore, EGR

is used to enable higher compression ratios to be achieved while reducing the engine NOx

emission output. Although the compression ratio of modern turbocharged SI engines is

reduced compared to their equivalent NA engines (Su et al., 2014), in order to achieve a

higher IMEP, this reduction can be reduced with EGR addition. Swirl and tumble intake

airflows are used to improve fuel-air mixing in the combustion chamber for improved

emissions, as well as increasing the combustion speed, to enable higher EGR levels and thus

higher compression ratios to be used.

Spark ignition fuels were discussed in detail due to their significant effect on the engine

emissions, with oxygenated fuels explored because of their potential to reduce net CO 2 output

from the transportation sector, if produced by renewable means. The relevant emissions

regulations have been discussed at length due to their significant influence on the

development of IC engines and fuels, with the new Euro VI emission regulation, in particular,

driving the reductions in PN emissions from DISI engines.

Ultimately, this literature review introduced the main motivation of this thesis which is to

investigate the effect of the fuel economy improving and thus CO 2 emission reducing

techniques of EGR (also used to reduce NOx), swirl and tumble intake airflows, increased

compression ratio, as well as the effect of oxygenated fuel components of 1-butanol and

ethanol, on the combustion and emissions behaviour of a modern DISI engine.

38
 Chapter 3

Experimental Setup and Techniques

The aim of this chapter is to present the experimental test facilities and the data acquisition

systems and analysis methods used in this study.

3.1 Introduction
This chapter describes all of the test equipment used throughout this study. This includes the

details of the single cylinder engine hardware and instrumentation, as well as the details of

the emissions equipment. The engine control methodology, data acquisition systems and

analysis methods are described in the following subsections.

The engine test facilities were created and developed by Jaguar Land Rover and the previous

research students at the University in their PhD studies. During the investigation, the author

was responsible for the development and maintenance of the single-cylinder thermal engine

test facility. Developing the engine test cell facility involved improving the fuel and variable

valve timing systems. Regarding maintenance of the facilities, this included changing minor

components such as fuel pumps, water pumps, water heaters, etc., when they failed, and three

engine rebuilds.

Overall, this chapter provides an overview of the current test facility. Further detailed

descriptions can be found in (Wang, 2014b), (Daniel, 2012d), (Turner 2010) and (Luszcz,

2009).

39
3.2 Overview of Single-Cylinder Engine Test Cell
The schematic of the experimental test facility is shown in Figure 3-1.

Figure 3-1 Schematic of single-cylinder engine test facility and instrumentation setup

The figure shows the 4-stroke single-cylinder spray-guided DISI engine used in this

investigation. The cylinder head is a single-cylinder version of that used in the Jaguar AJ133

(2010) 5.0 litre V8 production engine. This was attached to a balancing unit based on a

Ricardo Hydra engine which was redesigned by the Internal Combustion Engines (ICE)

group at the University of Oxford. The engine cylinder head provides the combustion

chamber with 4-valves and spray guided DI; both of which are found in modern DISI

engines.

The modified single cylinder research engine was not designed to be very resistant to knock

(high compression ratio of 11.5, large bore diameter of 90 mm). In addition, 95 RON

gasoline was used throughout this research which is of a lower octane number than the 98

40
RON gasoline specified for the engine. Therefore, engine knock occurred at loads of 6.0 bar

IMEP and above, which is somewhat lower than what can be expected with the state of the

art aggressively downsized engines of modern cars on sale today. Furthermore, audible knock

was observed to start occurring with 97 RON gasoline fuel at engine loads between 4.5 and

6.0 bar IMEP by previous research students using this research engine (Wang, 2014b, Daniel,

2012d). Therefore, the occurrence of knock at loads of 6.0 bar IMEP and above with 95 RON

gasoline fuel is consistent with these previous investigations. This allowed the effect of EGR

on knock to be studied at reduced loads to prevent any significant engine damage. The engine

also features a variable valve timing (VVT) system which will be discussed in detail in a later

section.

A full specification of the single-cylinder engine is shown in Table 3-1. The overall design of

the engine provides it with an over-square configuration resulting in a high compression ratio

and high swept volume, maximizing efficiency and output power.

Table 3-1 Experimental single-cylinder engine specification

Parameter

Engine Type 4-Stroke, 4-Valve

Combustion System Spray Guided DISI

Swept Volume 565.6 cc

Bore x Stroke 90 x 88.9 mm

Compression Ratio 10.7:1-11.5:1

Max. Torque @1500 rpm 44.2 Nm (~9.5 bar IMEP)

The balancing unit was attached to a test bed and the ancillary equipment used to operate the

engine and to record the data was attached onto the engine and balancing unit. This will be

discussed in detail in the following sections.

41
Before entering the engine, the intake air flow was filtered after which it passed through the

volumetric airflow (VAF) meter, as shown in Figure 3-2, which recorded the rate of air flow

through the engine. It then passed through a 100 litre damper tank which reduced the pressure

fluctuations through the VAF after which it passed through the throttle and the intake plenum

chamber; the chamber reduced the pressure fluctuations after the throttle. The air then went

into the engine before mixing with and undergoing combustion with the fuel, after which the

combustion products were ejected into the exhaust system. The exhaust gases passed through

the exhaust plenum chamber which reduced the pressure fluctuations before being analysed

by the Lambda meter to measure the air/fuel ratio. The gases were then analysed by the

particulate and gaseous emission measurement systems, after which they were pumped out to

the atmosphere via the engine laboratory ventilation system. An EGR line was connected just

after the exhaust plenum chamber to the intake plenum chamber to achieve external EGR. To

achieve swirl and tumble intake airflows in the engine, swirl and tumble inducing baffle

plates were fitted between the intake runner and cylinder head. These systems are discussed

in further detail in later sections.

3.2.1 Dynamometer and Electric Motor

In order to motor the engine, the crankshaft was attached to a dynamometer manufactured by

Mawdsley’s LTD. It used 450 volt electric power and it could maintain a constant speed or

constant engine load with motoring or firing conditions. The speed was manually controlled

by a dial in the engine control room and it was displayed on a digital indicator visible from

the control room.

3.2.2 Intake and Exhaust System

As shown in Figure 3-2 and discussed, the 100 litre intake damper tank along with the intake

and exhaust plenum chambers were used to reduce the pressure pulsations through the intake

42
and exhaust systems in order to stabilize the airflow through the engine. Based on the work of

a previous student (Daniel, 2012d), a 100 litre intake damper was used in order to supply the

engine with stable airflow at wide open throttle (WOT). Smaller plenums wee required just

before the intake and just after the exhaust because the 100 litre intake damper already

effectively reduced the pulsations in the intake and exhaust systems; they were required only

to reduce the pulsations caused by the throttle and engine respectively.

3.2.3 Heating and Cooling Circuits

Heating and cooling circuits for the engine water and oil were incorporated into the engine

test facility and are shown schematically in Figure 3-2. Each circuit was pumped and heated

individually through water and oil pumps and heaters, respectively. In addition, a large water-

water heat exchanger was used to transfer heat between the warm engine coolant and cold

mains water and another smaller water-oil heat exchanger was used to transfer heat between

the water and oil; both heat exchangers were manufactured by Bowman Ltd. Overall, the

pumping, heating and heat exchange allowed the temperature of the water and oil circuits to

be maintained within a reasonable range. During the experiments, the water and oil

temperatures were maintained at and respectively, using a proportional

integral differential (PID) controller. Once the temperature exceeded the set limit on the PID

controller, it sent a signal which, via a relay, opened a solenoid valve, allowing cold mains

water into the water-water heat exchanger to cool the engine coolant. This then cooled the

engine oil through the second water-oil heat exchanger.

43
 Figure 3-2 Schematic of water cooling and oil lubrication circuits

Tap water was used to cool the engine because the engine does not have any specialist

cooling requirements. 10W40 oil was used to lubricate the engine, as recommended by

Jaguar Land Rover. The water and oil levels were monitored and regularly topped up to

ensure there was enough water and oil available to meet the cooling and lubricant

requirements of the engine when motoring and firing. The oil was regularly replaced because

EGR led to degradation of the oil due to an increased number of particulates condensing into

the oil film on the combustion chamber walls.

3.2.4 Fuel Supply Systems

3.2.4.1 DI

The DI fuel system, as shown schematically in Figure 3-3, applied a constant pressure to the

fuel through the use of a free piston accumulator and a compressed nitrogen bottle supplied

by British Oxygen Company (BOC), which was controlled using a high pressure gas

regulator manufactured by the company. The pressure (150 bar) supplied is more consistent

44
than that of a high pressure fuel pump. However, only 2 litres of fuel could be supplied to the

accumulator with each charge, necessitating the engine to be stopped at regular intervals in

order to refill the system.

Figure 3-3 Schematic of fuel supply systems

The fuel required for the test was pumped using an electric out of tank pump from the main

fuel tank in the test cell, through a three-way valve (L) directly into the accumulator. The

three-way valve allowed the fuel to be pumped from the tank around the PFI system when it

45
was turned to the other position. A smaller auxiliary tank was used when a non-gasoline fuel

was tested, which allowed the gasoline to be constantly stored in the main fuel tank, rather

than having to be drained each time a non-gasoline fuel was used. The partially separate

systems also allowed contamination between the fuels to be eliminated to ensure that the test

data is reliable. The system operates via another three-way valve (M), so when the tank was

used, the valve was turned to its other position to allow fuel to be pumped from the auxiliary

fuel tank into the accumulator. Once the DI fuel system was charged, the pressurized fuel was

transported to the DI injector using a variety of rigid and flexible tubing, and stainless steel

compression fittings. The flow of fuel and nitrogen gas through the system was controlled

using various manually operated valves. When refilling the system and changing fuels, the

system was purged with nitrogen gas. If a non-gasoline fuel is used during the tests, the

auxiliary fuel tank and the system were cleaned using a small amount of the non-gasoline fuel

to ensure that there was no gasoline contamination in the non-gasoline fuel, when the fuel

tank was subsequently filled with it. Similarly, at the end of the day, the system was cleaned

with gasoline to avoid unnecessary damage to the injector and accumulator seals.

3.2.4.2 Port Fuel Injection (PFI)

A low pressure (3 bar) PFI system, also shown schematically in Figure 3-4, was used for the

sole purpose of warming the engine prior to DI testing. As mentioned, the out of tank fuel

pump was operated to pump fuel from the gasoline fuel tank and the three-way valve (L) was

changed so that the fuel was pumped to the PFI injector. After the fuel exits the PFI injector,

it passed through a water cooling line in order to avoid excessive heating of the fuel and

excessive consequential vapour loss from the system, as shown in the figure.

46
3.2.5 DI Fuel Injector and Spark Plug Configuration
The combustion chamber of the engine has a flat top piston and a centrally mounted six-hole

direct injector along with a side-mounted spark plug located close to the injector. Figure 3-4

shows the relative positions of the injector along with the injector spray plume.

Figure 3-4 Relative positions of the direct injector along with the injector spray plume

The spark plug was located between fuel spray plumes 1 and 6, and it was inclined at an

angle of 18˚ to the cylinder axis. As shown in the figure, the injector, which was

manufactured by Bosch Ltd., had two symmetric groups of holes with three holes on each

side of the injector, enabling the injector to deliver a hollow-cone spray.

Single-pulse injections were used throughout the experiments with an injection pulse width

ranging from between ~1.5 and ~3ms for the load range of 5.5-8.5 bar IMEP studied. The

injection event was triggered by a 12 volt DC electrical signal at the SOI of 280˚bTDC used

47
throughout the experiments. More information about the SOI and other electrical engine

control signals can be found in Section 3.3.

3.2.6 EGR

3.2.6.1 Hot EGR

A pipe with a valve was connected just after the exhaust plenum chamber to the intake

plenum chamber in order to implement and control hot EGR in the engine, as shown in

Figure 3-2. The Horiba MEXA-7100DEGR, discussed in detail in Section 3.5.2, had an in-

built function which allowed a sample line to be connected to it, enabling the system to

calculate the EGR addition (%) to the engine (accurate to ±0.02%). The sample line was

connected to the intake and the CO2 percentage was compared to the CO2 percentage in the

exhaust in order to calculate the EGR ratio using Equation 3-1.

(%) (3-1)

The position of the EGR valve (±0.25% EGR accuracy) was then changed manually along

with the throttle angle and injection pulse width to achieve the desired load and EGR

addition. Although the intake manifold temperature was not controlled in the engine, the

effects of the increased intake manifold temperature with increased EGR addition have been

taken into account in the results. (Please note that whenever EGR is discussed in this thesis,

hot EGR is being referred to, unless stated otherwise).

3.2.6.2 Cooled EGR

In order to achieve cooled EGR in the engine, the setup for the hot EGR was modified

slightly to incorporate a 1.1 metre long heat exchanger to cool the EGR gases before they

entered the engine, as shown in Figure 3-5. Air was blown over the heat exchanger to cool

the EGR gases as they passed through and the overall length of the EGR line was increased

48
slightly in order to incorporate the heat exchanger. Although some water vapour was found to

condense in the heat exchanger, it is believed that the vast majority did not condense. Thus,

the water vapour content of the EGR gases was not significantly changed. The same

procedure was followed as with the hot EGR to achieve the desired engine load and EGR

addition.

Figure 3-5 Cooled EGR schematic

3.2.6.3 Hot EGR after TWC

In order to simulate the effect of EGR after a TWC, a gas bottle supplied by BOC with a

mixture of 86% nitrogen and 14% CO2 was utilized. The mixture represents the composition

of gases that would be present after a TWC assuming that 100% of the NOx and CO from the

engine has been converted to N2 and CO2 respectively, and assuming that all of the water

vapour has been condensed and removed. It also assumes that all of the HCs and PM have

been successfully oxidized in the exhaust system. This removed the effect of varying TWC

exhaust gas conversion efficiencies which have been shown to vary the EGR ratio

(Hoffmeyer et al., 2009, Lewis et al., 2014). This is due to changes in the CO conversion

efficiency which affects the CO2 percentage of the exhaust gas, affecting calculation of the

EGR ratio (Equation 3-1).

49
Furthermore, it reduces the calorific value of the intake charge since energy cannot be

extracted by oxidizing the HCs and PM in the engine cylinder and it reduces the heat capacity

of the EGR gases because the water content of the EGR gases is removed. By using a

simulated TWC, the back-pressure effect of a real TWC was eliminated allowing the effects

of the post-TWC EGR gases on the engine combustion and emissions to be studied alone

without the back-pressure effects. Fuel consumption would be expected to increase if the

back-pressure effects were present because of increased pumping work.

The gas bottle was controlled using a regulator manufactured by BOC and a flow rate

controller, after which a one-way valve manufactured by Swagelok was used to prevent any

possible backflow into the gas bottle. The gas then passed through a 5 metre heated line

maintained at temperatures between 75-200˚C, depending on the flow rate of gas through the

line, to simulate the high temperature of the EGR gases after a TWC. Finally, the heated line

was connected to the intake plenum on the engine to allow the EGR gases to mix sufficiently

with the intake gases before they enter the engine. This setup is shown schematically in

Figure 3-6.

Figure 3-6 Hot EGR after TWC schematic

50
In order to achieve the desired engine load and EGR addition, the same procedure was

followed as with the hot and cooled EGR. However, rather than the EGR valve being

adjusted along with the throttle and injection pulse width, the flow rate controller was

adjusted along with the throttle and injection pulse width and the temperature of the heated

line was adjusted accordingly. Higher temperatures were used for higher flow rates and lower

temperatures were used for lower flow rates, to achieve approximately the same temperature

of heated gas for all flow rate conditions tested, as measured by the intake plenum

temperature thermocouple. This was done to accurately reflect the similar EGTs that would

be experienced when testing at a single engine load for the different EGR types in this

investigation.

Although the EGT at a particular engine load will decrease with EGR addition, it is expected

that the differences would become negligible after the TWC in a real vehicle. This is because

sufficient heat transfer between the exhaust system and the atmosphere would be expected to

occur by the time the exhaust gases passed through the TWC, cancelling out any relatively

small EGT differences that would be otherwise observed at different EGR additions.

Therefore, these differences have not been simulated in this investigation.

3.2.6.4 Cooled EGR after TWC

The same setup and procedure was used for the cooled EGR after TWC, except without the

heated line, as shown in Figure 3-7.

51
 Figure 3-7 Cooled EGR after TWC schematic

3.2.7 Swirl and Tumble Intake Airflows

Baffle plates, as shown with drawings in Figures 3-9b and 3-9c, were installed on the engine

between the intake runner and cylinder head in order to achieve swirl and high tumble airflow

conditions, respectively, within the engine cylinder. The plates worked by blocking part of

the intake flow into the engine to generate the swirl and high tumble. The swirl and tumble

ratios were estimated using findings from other researchers, the results of which are presented

in Section 3.2.7.1. As well as these two plates, another baffle plate shown in Figure 3-8a was

made with the same profile as the intake runner, so that the intake geometry was the same for

the reference low tumble condition as it was for the swirl and high tumble conditions. The

tumble for this condition was generated by the vertical inclination of the intake runner alone.

Low tumble baffle plate Swirl baffle plate High tumble baffle plate

(a) (b) (c)

(hatched area indicates

the baffle plate)

Figure 3-8 a) Low tumble baffle plate, b) swirl baffle plate, c) high tumble baffle plate

52
3.2.7.1 Swirl and Tumble Ratio Estimation

For this investigation, the swirl and tumble ratios of the single-cylinder engine with the baffle

plates fitted (Figure 3-8) were estimated based on data from previous investigations by other

researchers. Several attempts were made to measure the ratios using PIV in a single-cylinder

optical engine; however, they were unsuccessful due to technical difficulties which could not

be overcome within the time available. Therefore, it became necessary to estimate the ratios

rather than measure them.

Based on these findings of other researchers, the estimated swirl and tumble ratios of the

baffle plates, as shown in Figure 3-8, have been estimated as shown in Table 3-2. A summary

of the equipment and methods used by the other researchers to obtain their swirl/tumble ratios

along with their findings has been provided in Table 3-3.

Table 3-2 Estimated swirl and tumble ratios for the three intake airflow conditions

Intake Airflow Condition Swirl Ratio Tumble Ratio

Low Tumble 0 0.7

Swirl 1.5 0.7

High Tumble 0 1.8

53
 Table 3-3 Summary of other researchers findings for swirl and tumble ratios

Researcher Engine Cylinder Number Bore Intake Piston Engine Swirl/Tumble Swirl/Tumble Swirl Ratios Tumble
Type Head of Valves (mm) Runner Shape Speed Generation Measurement Ratios
Design Inclination (rpm) Method Method
Angle to
horizontal (˚)
AJ133 Engine Single- Pentroof 4 89 45 Flat Piston 1500 Baffle Plates - - -
cylinder

(Floch et al., Single- Pentroof 4 88 45 Flat Piston - Baffle Plates Steady-flow test rig 0 (no baffle 0.7 (no baffle
1995) cylinder plate), 1.5 (swirl plate), 1.8
generating (tumble
baffle plate) generating
baffle plate)

(Teraji et al., Single- Pentroof 4 93 30 Flat Piston 1500 Baffle Plates CFD (STAR-CD) 0 (no baffle 1.5 (no baffle
2009) cylinder plate), 1.5 (swirl plate), 1.5
generating (swirl
baffle plate), 0 generating
(tumble baffle plate),
generating 2.1 (tumble
baffle plate) generating
baffle plate)

(Fujimoto and 4-cylinder Pentroof 4 78 n/a (only swirl Flat Piston - Throttle plate in Steady-flow test rig 0.4 (slightly -
Tabata, 1993) (modified ratios measured, intake runner closed plate)-3.8
to operate and this only passage of one (fully-closed)
as single- affects tumble intake valve
cylinder) ratio)

(Li et al., 2001) 4-cylinder Pentroof 4 80 n/a (only swirl Flat Piston 1200 Blockage of intake PIV 2.30 -
(modified ratios measured, port into one intake (180˚bTDCcomb)
to operate and only this valve - 1.90
as single- affects tumble (60˚bTDCcomb)
cylinder) ratio)

54
Overall, it can be seen that the swirl/tumble ratios of other researchers have been observed

with equivalent or similar methods for generating the swirl and tumble intake airflows, as

compared to those used in this investigation, and they have been observed with similar 4-

valve pentroof combustion chamber designs. The results observed by other researchers are

similar to one another across similar engine designs (pentroof combustion chamber), with

different swirl/tumble generation methods and different intake runner inclinations to the

horizontal, accounting for the differences in swirl/tumble ratios observed. Therefore, while

not being 100% accurate, it is reasonable to expect that the swirl and tumble ratios in Table 3-

2 will provide a good estimation of those used in this investigation.

Furthermore, tumble ratios have been shown to vary depending on what equipment is used to

measure them (Kim et al., 2006). Therefore, it is believed that the estimation of the swirl and

tumble ratios obtained will not affect the interpretation of the findings of this investigation

more significantly than the use of different swirl/tumble measurement equipment. It is further

believed that the relative differences in swirl and tumble ratios between the baffle plate

designs have been accurately estimated, although the absolute magnitudes may differ slightly

from the true values. Thus, the interpretation of the findings of this investigation will not be

significantly affected.

55
3.2.8 Variable Compression Ratio
Although the single-cylinder engine had a geometric compression ratio of 11.5:1, the

compression ratio could be modified by adjusting the number and size of metal inserts which

are placed beneath the cylinder head. These acted to adjust the height of the cylinder head in

relation to the piston BDC, allowing the compression ratio to be changed. The maximum

compression ratio was however limited; beyond this limit the intake valves would have hit

the piston crown. The real dynamic compression ratio of the engine could be adjusted by

changing the valve timings; however, this was not pursued in this investigation. The

compression ratio of the engine is thus calculated using Equation 3-2 (Heywood, 1988).

Compression Ratio (Rc) = (3-2)

where Vd is the swept volume which remains constant and Vc is the clearance volume which

is altered with the metal inserts. To achieve the desired compression ratio, the equation was

solved for Vc which could then be related to the new size of the metal inserts (h) using

Equation 3-3. This allowed the desired compression ratio to be achieved by inserting the

metal inserts with the correct size.

(3-3)

56
3.3 Control Systems
The spark timing and injection pulse width along with the start of injection and coil charge

duration were controlled using the engine timing control system (ETCS); a program written

using LABVIEW and operated using a NI card (model 6602). This program was initially

developed by the ICE group at the University of Oxford but subsequent modifications were

made by a previous research student in the engine group (Luszcz, 2009). It allowed the

engine parameters to be modified in real-time while the engine is running. A screenshot of

the program front-panel is shown in Figure 3-9.

Figure 3-9 ETCS LabView program front-panel (Luszcz, 2009)

In order to achieve a particular engine load, fuel was injected and adjusted using the ETCS

software and the airflow into the engine was controlled using a butterfly valve that was

adjusted from the engine control room. The fuel and airflow were modified until the desired

57
engine load was achieved. When EGR was used, the EGR valve was also adjusted along with

the injection pulse width and throttle position, as discussed, in order to achieve the desired

engine load and EGR addition.

3.3.1 Crankshaft Encoder Setup

The crankshaft encoder was set up to correspond to the engine CAD in order to accurately

control injection, ignition and the variable valve timings. This was achieved using a timing

light which was attached to the ignition coil of the engine, enabling it to produce a

stroboscopic light signal when the ignition coil was activated by the ETCS system at a

specified ignition timing during low speed engine cranking. The light was shone onto the

flywheel which had CAD markings; the reading achieved when the light came on was

recorded and the encoder position was subsequently adjusted until the system was

synchronized.

3.3.2 Variable Valve Timing System

The VVT system operated using crankcase oil pressure (3 bar) in order to control the

positional offset of the intake and exhaust camshafts, relative to the position of the camshaft

pulley ( ). In order to control the system, a LABVIEW program developed by a

previous research student in the engine group (Luszcz, 2009) which controls the valve timing

using a NI card (model 6202), was used. A screenshot of the program front-panel is shown in

Figure 3-10. A Hall effect type sensor was used to detect the relative position of the camshaft

and this position was controlled using hydraulic solenoids mounted in the cylinder head,

which adjusted the oil pressure and thus changed the relative camshaft positions. The

program constantly adjusted the oil pressure in order to maintain the cam timing within a

range of .

58
 Figure 3-10 VVT LabView program front-panel (Luszcz, 2009)

Table 3-4 shows the geometry of the intake and exhaust camshafts and Figure 3-11 shows the

intake and exhaust valve lift profiles along with that of a typical in-cylinder pressure trace.

The camshaft itself had a high-lift profile typically used in SI engines and the valve timings

were chosen in order to minimise the residual gas fraction to increase the effect that EGR has

on the combustion and emissions. The engine was designed with an intake valve of a larger

inner seat diameter and higher maximum lift than the exhaust valve, in order to improve VE.

Table 3-4 Single-cylinder engine camshaft geometry

Valve and Camshaft Geometry

Intake Valve Lift 10 mm

Intake Valve Inner Seat Diameter 36 mm

Exhaust Valve Lift 9.3 mm

Exhaust Valve Inner Seat Diameter 33 mm

Intake Valve Duration 250 CAD

Exhaust Valve Duration 250 CAD

59
 30
Intake Valve Lift Profile

Valve Lift (mm)/In-cylinder pressure (bar)

25 Exhaust Valve Lift Profile

Typical In-Cylinder
20 Pressure Profile

15

10

0
0 180 360 540 720
TDCIntake TDCCombustion
CAD

Figure 3-11 Intake and exhaust camshaft profiles

All of the signals sent from the control software to operate the engine as well as those

produced by the sensors on the engine were sent to the control tower which consisted of the

ETCS and VVT interface boxes which were used for the ETCS and VVT systems,

respectively. Figure 3-12 summarizes the arrangement and flow for the various signals.

60
 Figure 3-12 Engine ETCS and VVT control systems; their arrangement and signal flow

3.3.2.1 Variable Valve Timing Setup

Detailed instructions on how to set up the variable valve timing system can be found in

(Wang, 2014b) and (Daniel, 2012d). These instructions were followed in order to set up the

system on the engine after each rebuild.

3.4 Instrumentation
Various ancillary equipment was fitted onto the research engine and balancing unit in order to

provide the measurements which were recorded and later analysed in this investigation.

These mainly consisted of pressure, temperature and engine speed data, each of which is

discussed further in the following sections.

61
3.4.1 Engine Speed and Torque
The engine was coupled via a universal joint to a DC dynamometer in order to maintain a

constant speed of 1500 rpm ( rpm), regardless of the engine torque output. The speed was

chosen because it represents a speed used frequently in emission test cycles and it is

commonly used by researchers (Alger et al., 2009a, Heffel, 2003, Institution of Mechanical

Engineers, 2013). This along with the medium-high engine loads (5.5-8.5 bar IMEP) studied

mimics vehicle acceleration or driving on an incline, which is important to study because it is

one of the worst conditions for fuel consumption and emission formation. A torque meter on

the engine dynamometer was used to provide an instantaneous torque reading, which is useful

in finding the MBT spark timings at a particular operating point.

3.4.2 Pressure and Temperature Measurements

In-cylinder pressure measurements in the engine were made using a Kistler 6041A water

cooled piezoelectric pressure transducer along with a Kistler 5011B charge amplifier, which

was used to amplify the signal from the transducer. The transducer itself was fitted at the side

of the engine cylinder head between the intake and exhaust valves, to ensure that it had direct

contact with the gases in the combustion chamber for accurate in-cylinder pressure

measurement (Hountalas and Anestis, 1998). The electric charges induced by the

piezoelectric material were amplified by the charge amplifier and then outputted as a

modulated voltage. The pressure transducer was calibrated by Kistler using an oil weight

bench machine in the range of 1 MPa/1 V to 10 MPa/10 V, and this calibration was then

input into the charge amplifier as a tuning coefficient. Overall, the system was capable of

absolute pressure measurements in the range of 0.08-10 MPa. Once the signals were

amplified, they were sent to the data acquisition system.

62
In order to measure the intake and exhaust manifold pressures, EPT 3100 media isolated

pressure transmitters manufactured by EuroSensor were used. K-type stainless steel

thermocouples supplied by RS were used for the temperature measurements. The signals for

these temperature sensors were sent to a TCK-4 thermocouple amplifier before being

sampled by the NI data acquisition card.

3.4.3 VAF Meter

A Romet (model number: G25) VAF meter was used to measure the rate of airflow into the

engine for fuel consumption calculations. The flow meter was designed to rotate one

revolution for a fixed volume of air, allowing the airflow rate to be quantified. The meter

itself was attached via a coupling to an encoder, with its speed recorded by the data

acquisition system. In order to relate the frequency of the encoder to the volumetric airflow

rate in the engine, a calibration was conducted at different throttle angles and thus different

engine volumetric airflow rates. The airflow in the engine was measured using an orifice

plate connected after the 100 litre intake damper and before the throttle, which created a unit

pressure difference for a unit flow across it. This pressure difference was measured using an

inclined u-tube manometer and the reading was used to calculate the actual flow rate in the

engine using Equation 3-4.

(3-4) (Eastop and McConkey, 1993)

Cd is a dimensionless number representing the orifice plate discharge coefficient and is the

ratio of the orifice flow plate areas. Cd can be obtained by dividing the actual flow rate with

the theoretical flow rate. This actual flow rate was calculated at a known orifice flow rate

calibration where a 4 kPa pressure drop is equivalent to an actual flow rate of 21

litres/second. The theoretical flow rate in this case was calculated using Equation 3-4 by

63
assuming that Cd is 1. The moisture pressure was not taken into account when calculating the

air density, limiting the accuracy of this method. Along with the u-tube monometer readings,

the frequency of the encoder was recorded, therefore allowing a correlation between the two

to be calculated. The calibration equation was then input into the low speed data acquisition

(LSDAQ) software, which is discussed further in Section 3.6.2, in order to calculate the

volumetric airflow rate at any given engine condition. The u-tube manometer and encoder

frequency readings were repeated three times at each throttle position in order to improve the

accuracy of the calculation.

Figure 3-13 shows the calibration between the calculated and actual airflow rates. It is clear

that the calibration achieved is good. However, even with a R2 value of 0.9966, there is an

average percentage error of 1.6% between the calculated and actual airflow rates. This was

obtained by individually calculating the percentage error for each of the data points on Figure

3-13, and then taking the mean average of these. The error is largely due to the limited

resolution of the u-tube manometer device and the limited unit pressure difference produced

across the orifice plate.

64
 300

y = 1.0001x + 0.3181
250 R² = 0.9966

Calculated Air Flow Rate (l/min)

200

150

100

50

0
0 50 100 150 200 250 300
Actual Air Flow Rate (l/min)

Figure 3-13 Calibration results of the VAF Meter

The minimum and maximum VAF rates measured in this study were 200 and 300 l/min

respectively, which is within the calibration range shown in Figure 3-13. The maximum

airflow rate in the engine if the VE was 80% would be 339 l/min according to Equation 3-5.

(3-5)

Therefore, the maximum airflow rate used in the experiment of 300 l/min at an engine load of

8.5 bar is approximately correct, considering the engine can reach a maximum load of

approximately 9.5 bar.

3.4.4 Lambda Meter

For all of the tests in the engine, the dynamic Lambda value was measured using a ETAS

LA4 Lambda meter and a Bosch heated wideband oxygen sensor. Fuel-specific curves were

used in the Lambda meter to interpret the dynamic (actual) AFR using the oxygen content in

the exhaust. Before a new fuel is tested in the engine, its hydrogen-to-carbon (H/C) and

oxygen-to-carbon (O/C) ratios were input into the system along with its stoichiometric

65
air/fuel ratio. This was then used by the Lambda meter to characterise the fuel curves to

calculate the dynamic AFR. As discussed, in order to achieve a specific engine load, the

throttle and injector pulse width were modified along with the EGR valve, if EGR was used.

Not only was the target load achieved using this procedure but the target Lambda value too.

Overall, this system was open loop meaning that the injection pulse width required regular

manual modifications to ensure Lambda remains within target.

3.5 Emissions Measurement

In this study the legislated European and United States emissions of PM, NOx and HC were

measured. The devices used to measure these emissions are described in the following sub-

sections.

3.5.1 Particulate Emissions

The PM emissions were sampled 0.3m from the exhaust valve, downstream of the exhaust

plenum, and they were measured using two devices; firstly the differential mobility

spectrometer (DMS) 500 manufactured by Cambustion Ltd. and secondly the scanning

mobility particle sizer (SMPS) 3936, manufactured by TSi. Through experience, the DMS500

provided more consistent readings, resulting from the equipment being newer. However, due

to limited equipment availability, the SMPS 3936 was used in these experiments too.

In order to improve the consistency of the PN data, particles smaller than 10 nm which are

detectable by the PM measurement equipment used in this investigation were removed from

the range of particles studied. The particles have been characterized in this investigation by

their diameter in order to distinguish between ultra-fine particles and fine particles, since the

ultra-fine particles are the most significant contribution to PN and they are the most harmful

to human health (Anderson et al., 2012b, Hull et al., 2005).

66
In this investigation, the nucleation mode particles have been classified as those within the

10-30 nm range and the accumulation mode particles have been classified as those within the

30-500 nm range for data from the DMS 500. For the results from the SMPS 3936, the

accumulation mode particle range was 30-300 nm, due to the lower maximum particle size it

can measure. This is based on the PM data recorded which can be seen to form two distinct

peaks within the stated ranges. (Zhang et al., 2014b) also classified nucleation mode particles

as those below 30 nm and accumulation mode particles as those above 30 nm in their SI

engine investigation using gasoline fuel along with butanol and ethanol splash blends.

It must be noted that the separation between the two modes is ill-defined (Kittleson, 1998)

and that in this investigation; the PM accumulation mode was shifted towards a smaller size

for the data collected using the SMPS 3936, because of the use of the thermodenuder, as will

be explained in a later section. It must also be noted that with the DMS 500 when no

thermodenuder was used and with the SMPS 3936 despite the use of the thermodenuder,

some nucleation mode particles still remained in the accumulation mode peak. This is

because they were adsorbed onto the accumulation mode particles. Also, despite the use of

the thermodenuder with the SMPS 3936, a nucleation mode particle peak was still observed.

Therefore, this method of PM mode separation has its limitations.

3.5.1.1 DMS 500

The DMS 500 classified particles according to their size by separating them in a parallel

differential electrical mobility analyser, as shown in Figure 3-14.

67
 Figure 3-14 Cambustion DMS 500 analyser operating principle

Firstly, the particulates were pumped from the exhaust into the two stage dilution system

built into the instrument, as shown in Figure 3-15. The sample line from the exhaust to the

dilution system was maintained at 150˚C in order to avoid condensation of the particulates.

The particulates were then diluted in the primary dilution stage at a fixed dilution ratio of 5:1

for gasoline engine measurement, in which the mass flow rate is controlled by HEPA filtered

compressed air. A rotating disc was then used to perform the secondary high ratio dilution

with a dilution ratio of 16:1 used for this investigation; the system allowed a maximum

secondary dilution ratio of 500:1. This resulted in an overall dilution ratio of 45.2:1. The

choice of dilution ratios ensured that a good signal to noise ratio was maintained while

extending instrument cleaning intervals, as recommended by the manufacturer (Cambustion,

2011). Samples were collected at a rate of 1 Hz for approximately 2 minutes for each reading

to improve the consistency of the data collected.

68
 Figure 3-15 Schematic of the DMS 500 two-stage dilution system

Once the particles were diluted, they were passed through a unipolar corona charger. This

charged the particles with a positive electric charge which was proportional to the particle

size. The filtered sheath flow then pumped the charged particles into the particle classifier.

The high voltage electrode as shown in Figure 3-14 then attracted the positively charged

particles towards the electrometer detector. The distance travelled by the particles along the

classifier was proportional to their size and inversely proportional to their mass. The classifier

which consisted of 22 electrometer detectors measured the charge quantity and distance

travelled by the particles, thus allowing the particle size, mass and number data to be

quantified. Finally, the Cambustion software recorded the particulate measurements.

69
3.5.1.2 SMPS 3936

The SMPS 3936 system was comprised of three different units. Firstly, a TSi rotating disk

thermodiluter (model 379020A) maintained at 150˚C was used to dilute the exhaust sample

before it was pumped through a Topas TDD 590 thermodenuder at a temperature of 400˚C, to

remove most of the volatile nucleation mode particles in order to make it easier to distinguish

the nucleation and accumulation mode particles from one another. PM emissions from DISI

engines are mostly composed of volatile particles; they are fundamentally different from

those produced by diesel engines. Volatilities not only exist in the nucleation mode, but they

are also adsorbed onto accumulation mode particles. This makes it difficult to measure the

PM size distribution accurately. Therefore, the use of the thermodenuder device enabled the

particle size and number distribution to be accurately measured (Wang, 2014b).

Secondly, the particles were processed in the electrostatic classifier (model 3080) which had

a differential mobility analyser (model 3081) attached to it, in order to separate the particles

according to size. Thirdly, the particles were counted in the condensing particle counter

(model 3775). The settings inputted into the SMPS system for this study are summarized in

Table 3-5. The overall dilution ratio used in this study for the SMPS is 45.72:1, based on the

work of a previous research student in the engine group (Daniel, 2012d).

70
 Table 3-5 SMPS 3936 system settings

SMPS 3936 System Settings

Sample Flow Rate (l/min) 1

Sheath Flow Rate (l/min) 10

Scan Up Time (s) 90

Retrace Time (s) 30

Minimum Particle Diameter (nm) 7.23

Maximum Particle Diameter (nm) 294.3

In addition to making it easier to distinguish the nucleation and accumulation mode particles

from one another, the use of the thermodenuder also resulted in a significant PN reduction

and the PM distributions being shifted towards a smaller size. As discussed, both of these

were clearer at higher engine loads where there was more soot formation. This is due to the

non-adsorbed nucleation mode particles and those nucleation mode particles adsorbed onto

accumulation mode particles, respectively for the nucleation and accumulation mode

particles, being partially removed by the thermodenuder. This was a compromise that the

author and the previous research student (Wang, 2014b) made in order to improve the ability

to distinguish between the nucleation and accumulation modes, as mentioned. Overall, what

is most important is the relative change between different engine conditions rather than

absolute numbers, since these can vary significantly between different engines (Wang,

2014b).

3.5.2 Gaseous Emissions

The gaseous emissions were sampled 0.3 metres from the exhaust valve on the opposite side

to the PM sampling point, downstream of the exhaust plenum, before being pumped through

a pre-filter and heated line, both of which were maintained at 190˚C to avoid condensation of

the emissions. They were then subsequently analysed by the Horiba MEXA-7100DEGR. The

specification of the system is shown in Table 3-6.

71
 Table 3-6 Horiba MEXA-7100DEGR specification

Horiba MEXA-7100DEGR Specification

Dry CLD

NOx Analyser (min. ppm=0-10, max. ppm = 0-10,000)

Zero gas = N2, span gas = NO/NO2

Hot-wet FID

HC Analyser (min. ppm=0-10, max. ppm = 0-50,000)

Zero gas = N2/air, span gas = C3H8

NDIR (dry)

(min. ppm=0-100, max. ppm = 0-12%)

CO Analyser
Zero gas = N2, span gas = CO/N2

NDIR (dry)

(min. ppm=0-5000, max. ppm = 0-20%)

CO2 Analyser
Zero gas = N2, span gas = CO/N2

Accuracy 1%

Nitrogen oxides in the Horiba system were measured using a dry chemiluminescence detector

(CLD) and total hydrocarbons, HC, were analysed using a hot-wet flame ionization detector

(FID) (Robert Bosch GmbH, 2006). Prior to the NO x being analysed, the NO2 was converted

into NO to enable it to be measured in the CLD. Once in the CLD, ozone was used to

produce a reaction with the NO, and as a result a photon was emitted. This produced a current

which was converted to a voltage output which was a function of the NO and thus NO x

concentration in the sample. For HC analysis, the sample gas was pumped into a nozzle

charged with a high voltage, after which it was analysed using a hydrogen flame. In the high

temperature hydrogen flame, some of the hydrocarbon molecules were ionized, resulting in a

flow of current between the positively charged nozzle and a collector. The current was then

converted into a voltage output which was a function of the HCs in the sample. The voltage

outputs created by the respected CLD and FID detectors were used to calculate the relative

concentration of NOx and HCs in the exhaust sample.

72
Oxides of carbon (CO and CO2) were measured using a non-dispersive infra-red (NDIR)

detector. The device measured the change in light intensity as an infra-red beam was passed

through the sample gas chamber. Oxides of carbon absorb infra-red light so the change in

light intensity could be used to calculate the relative concentrations of CO and CO 2 in the

sample gas.

Prior to each test, the Horiba was heated to its pre-set temperatures and the device was

calibrated using zero calibration gases. The test points were typically recorded one after

another during which the system continuously measured the gaseous emissions. When the

engine was stopped, the system was purged with compressed air to avoid build-up of

condensates in the system, which could have affected future measurements.

3.6 Data Acquisition and Processing

High speed crank-angle resolved engine data was recorded using a NI card (model 6251) and

the low speed, time resolved engine data was also recorded using a NI card (model 6602).

This is explained in detail in the following sections.

3.6.1 High Speed Data Acquisition

In order to acquire detailed in-cylinder pressure data with a good resolution, the in-cylinder

pressure data was recorded at high speed with a resolution of 0.5 CAD for 300 consecutive

cycles, using the high speed data acquisition (HSDAQ) program written in LabView by a

previous research student in the engine group (Luszcz, 2009). The screenshot of the front

panel of the program is shown in Figure 3-16. With an engine speed of 1500 rpm, this

equates to data being collected every 80 ms (12.5 Hz). In order to ensure the correct CAD

was recorded for the in-cylinder pressure data, the intake pressure was also recorded and the

two were “pegged” to each other, as described in Section 3.6.4. An average was taken for

73
these data over 300 engine cycles to reduce the effect of small fluctuations in the

measurements on the data. The stability of the VVT system was measured by recording the

intake and exhaust cam position data from their respected sensors.

Figure 3-16 HSDAQ LabView program front-panel (Luszcz, 2009)

3.6.2 Low Speed Data Acquisition

Due to the lower variation in the remaining engine data such as the engine temperatures

(intake, exhaust, cylinder wall, coolant, oil), ambient conditions (pressure, temperature), VAF

rate, throttle position, Lambda and the gaseous emissions data (NOx, HC), the data was

collected on a time resolved basis rather than a crank resolved basis. As with the high speed

data, an average was taken for these data, with an average taken over 200 samples. The data

was acquired using the LSDAQ program, again written in LabView by a previous research

student (Luszcz, 2009). The screenshot of the front-panel of the program is shown in Figure

3-17.

74
 Figure 3-17 LSDAQ LabView program front-panel (Luszcz, 2009)

3.6.3 Initial Data Processing

A custom MatLab script written by the ICE group at the University of Oxford and developed

by previous research students at the University was used to analyse the data from the high

and low speed data acquisition systems. The method described by Stone (Stone, 1999) was

used to calculate the net IMEP from each engine cycle, and this was averaged across 300

consecutive cycles.

3.6.4 In-cylinder Pressure Interpretation

As mentioned, the intake pressure data was collected along with the in-cylinder pressure data

in order to ensure the correct CAD was recorded for the in-cylinder pressure data. The two

were “pegged” to one another at the same CAD, (BDCintake), where the two pressures were

75
equal, to ensure that the correct CAD was recorded for the in-cylinder pressure data. Overall,

this helped to overcome the effects of drift which would otherwise have made the

piezoelectric pressure transducer used in these experiments unsuitable (PCB Piezeotronics,

2015). A smoothing function was programmed into the MatLab script in order to reduce the

fluctuations in the recorded pressure trace, to make it simpler to analyse.

3.6.5 In-cylinder Gas Temperature Calculation

PM emissions are strongly dependent on charge temperature as are NOx and HC emissions.

Therefore, a Ricardo WAVE model was created to simulate the in-cylinder conditions

recorded in the experiments, in order to calculate the average in-cylinder temperatures. The

model was then calibrated and verified using five cases for each, to ensure the model was

accurate for the engine conditions tested in this investigation. This is described in the

following sections.

3.6.5.1 Ricardo WAVE Combustion Model

Figure 3-18 shows the Ricardo WAVE model used in this investigation.

Figure 3-18 Ricardo WAVE simulation model

Ricardo WAVE is a one-dimensional engine and gas dynamics simulation tool created by

Ricardo PLC, and it is widely used in engine research and development by Universities and

the automotive industry. In order to simplify the analysis, the software uses the ideal gas law

combined with a prediction of trapped residuals and fuel vaporization behaviour, in order to

76
estimate the average in-cylinder gas temperature. This is much simpler than detailed chemical

kinetics, reducing the computational requirements and calculation times significantly.

Therefore, while the average in-cylinder temperature data in this investigation does not

represent actual measurements, it does provide some insight into the global average gas

temperatures. The fuel properties of indolene were used to simulate the combustion of

gasoline in order to further simplify the analysis. When simulating the combustion of the fuel

blends used in this investigation, the known properties were inputted but some unknown

properties such as viscosity-temperature behaviour were taken from indolene. A primary

combustion sub-model based on the recorded MFB profile was used along with a SI Wiebe

combustion sub-model, which required the input of MFB50 and MFB10-90, in order to

simulate the in-cylinder temperature conditions.

In order to produce the model, the following steps were taken:

1. Intake and exhaust systems and components were measured

2. Engine geometry such as the bore, stroke, valve diameters and cam profiles were

measured

3. The information was then used to construct the Ricardo WAVE model geometry

4. The model self-check function was used to ensure correct matching of inlet and outlet

geometries between the components in the model, to ensure it could be solved

After the geometry had been produced, the other parameters were input into the software

after which it was calibrated and verified, as described in the following sections.

77
3.6.5.2 Ricardo WAVE Combustion Model Calibration

Five test points were used to calibrate the Ricardo WAVE model and the procedure shown in

Figure 3-19 was followed in order to perform the calibration.

Figure 3-19 Ricardo WAVE model calibration procedure

The valve timings used for the model calibration were IVO=16˚bTDC and EVC=36˚aTDC,

and the IMEPs for the give cases are shown in Table 3-7.

78
 Table 3-7 Test cases for Ricardo WAVE model calibration

IMEP (bar)
Case 1 3.63

Case 2 4.86

Case 3 6.19

Case 4 7.43

Case 5 8.68

The actual and calculated in-cylinder pressure data is shown in Figures 3-20a, 3-20b, 3-20c,

3-20d and 3-20e for the five respected cases. Overall, there is a good agreement between the

data, showing that the model has been calibrated well.

Fig. 3-21a: In-cylinder Pressure (case 1) Fig. 3-21b: In-cylinder Pressure (case 2)

40 40
In-cylinder Pressure (bar)
In-cylinder Pressure (bar)

35 35 Actual
Actual

30 30
Calculated Calculated
25 25

20 20

15 15

10 10

5 5

0 0
-50 0 CAD 50 100 -50 0 CAD 50 100

Fig. 3-21c: In-cylinder Pressure (case 3) Fig. 3-21d: In-cylinder Pressure (case 4)
40 40
In-cylinder Pressure (bar)
In-cylinder Pressure (bar)

35 35
Actual Actual
30 30
Calculated Calculated
25 25

20 20

15 15

10 10

5 5

0 0
-50 0 CAD 50 100 -50 0 CAD 50 100

79
 Fig. 3-21e: In-cylinder Pressure (case 5)
40

In-cylinder Pressure (bar) 35 Actual

30 Calculated

25

20

15

10

0
-50 0 CAD 50 100

Figure 3-20 In-cylinder pressure versus CAD data comparisons for Ricardo WAVE model
calibration: a) case 1, b) case 2, c) case 3, d) case 4 and e) case 5

The actual and calculated IMEP, Pmax, indicated efficiency and VE for the five cases are

respectively shown in Figures 3-21a, 3-21b, 3-21c and 3-21d.

Fig. 3-22b: IMEP (bar) Fig. 3-22b: Pmax (bar)

9 40
Actual
8 35
Actual Calculated
7 30
Calculated
6
25
5
20
4
15
3
10
2

1 5

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

80
 Fig. 3-22c: indicated efficiency Fig. 3-22d: VE
0.50 0.7
Actual Actual
0.45
0.6
Calculated Calculated
0.40

0.35 0.5

0.30
0.4
0.25
0.3
0.20

0.15 0.2
0.10
0.1
0.05

0.00 0.0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

Figure 3-21 Engine data comparison for Ricardo WAVE model calibration: a) IMEP (bar),
b) Pmax (bar), c) indicated efficiency and d) VE

The percentage difference for these parameters is shown respectively in Figures 3-22a, 3-22b,

3-22c and 3-22d for IMEP, Pmax, indicated efficiency and VE. As with the in-cylinder

pressure data, there is good agreement, showing that the model has been calibrated well. The

confidence intervals have been calculated and analyzed for the parameters above; these show

that the random error in the data was very low (less than 0.1% in most cases). Therefore,

random error is not thought to have significantly contributed to the differences observed

between the actual and calculated data, and because it is so small, it has not been displayed

with confidence intervals on the figures.

Fig. 3-23a: IMEP (% difference) Fig. 3-23b: Pmax (% difference)

6 6

5 5

4 4

3 3

2 2

1 1

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

81
 Fig. 3-23c: indicated efficiency (% difference) Fig. 3-23d: VE (% difference)
6 6

5 5

4 4

3 3

2 2

1 1

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

Figure 3-22 Engine data percentage difference comparison for Ricardo WAVE model
calibration: a) IMEP, b) Pmax, c) indicated efficiency and d) VE

Overall, the VEs have all been calibrated to within 5% of each other at all of the engine loads,

and the other parameters correlate well with each other. Therefore, the calibration steps

described in Figure 3-19 have been completed.

3.6.5.3 Ricardo WAVE Combustion Model Verification

Five further test points were used to verify the Ricardo WAVE model. The valve timings

used were IVO=6˚aTDC and EVC=61˚aTDC, and the IMEPs for the cases are shown in

Table 3-7.

Table 3-8 Test cases for Ricardo WAVE model verification

IMEP (bar)
Case 1 3.74

Case 2 5.16

Case 3 6.47

Case 4 7.87

Case 5 8.90

The actual and calculated in-cylinder pressure data is shown in Figures 3-23a, 3-23b, 3-23c,

3-23d and 3-23e for the five respected cases. Overall, there is a good agreement between the

data, providing sufficient verification for the model.

82
 Fig. 3-24a: In-cylinder Pressure (case 1) Fig. 3-24b: In-cylinder Pressure (case 2)
35 35

In-cylinder Pressure (bar)

In-cylinder Pressure (bar)

30 Actual 30 Actual

25 Calculated
Calculated 25

20 20

15 15

10 10

5 5

0 0
-50 0 CAD 50 100 -50 0 CAD 50 100

Fig. 3-24c: In-cylinder Pressure (case 3) Fig. 3-24d: In-cylinder Pressure (case 4)
35 35
In-cylinder Pressure (bar)
In-cylinder Pressure (bar)

30 30
Actual Actual

25 25 Calculated
Calculated

20 20

15 15

10 10

5 5

0 0
-50 0 CAD 50 100 -50 0 CAD 50 100

83
 Fig. 3-24e: In-cylinder Pressure (case 5)
35

In-cylinder Pressure (bar) 30 Actual

25 Calculated

20

15

10

0
-50 0 CAD 50 100

Figure 3-23 In-cylinder pressure versus CAD data comparisons for Ricardo WAVE model
verification: a) case 1, b) case 2, c) case 3, d) case 4 and e) case 5

The actual and calculated IMEP, Pmax, indicated efficiency and VE for the five cases are

respectively shown in Figures 3-24a, 3-24b, 3-24c and 3-24d.

Fig. 3-25a: IMEP (bar) Fig. 3-25b: Pmax (bar)

9 40
Actual
Actual
8 35
Calculated
7 Calculated
30
6
25
5
20
4
15
3

2 10

1 5

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

84
 Fig. 3-25c: indicated efficiency Fig. 3-25d: VE
0.5 0.7
Actual Calculated Actual
0.4
0.6
Calculated
0.4
0.5
0.3

0.3 0.4

0.2 0.3
0.2
0.2
0.1
0.1
0.1

0.0 0.0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

Figure 3-24 Engine data comparison for Ricardo WAVE model verification: a) IMEP
(bar), b) Pmax (bar), c) indicated efficiency and d) VE

The percentage difference for these parameters is shown respectively in Figures 3-25a, 3-25b,

3-25c and 3-25d for IMEP, Pmax, indicated efficiency and VE. As with the in-cylinder

pressure data, there is good agreement, showing that the model has been sufficiently verified.

However, there is an exception with cases 3 and 5 having a percentage difference larger than

5% for between the actual and calculated indicated efficiencies. Despite this, the difference

was still below 10% and the other parameters had a percentage difference less than 5%.

Therefore, it is believed that the model has been sufficiently verified.

Again, the confidence intervals have been calculated and analyzed for the parameters above.

As with the calibration results, random error was found to be less than 0.1% in most cases.

Therefore, random error is not thought to have significantly contributed to the differences

observed between the actual and calculated data, and because it is so small, it has not been

displayed with confidence intervals on the figures.

85
 Fig. 3-26a: IMEP (% difference) Fig. 3-26b: Pmax (% difference)
8 8
7 7
6 6
5 5
4 4

3 3

2 2

1 1

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

Fig. 3-26c: indicated efficiency (% difference) Fig. 3-26d: VE (% difference)

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1

0 0
Case 1 Case 2 Case 3 Case 4 Case 5 Case 1 Case 2 Case 3 Case 4 Case 5

Figure 3-25 Engine data percentage difference comparison for Ricardo WAVE model
verification: a) IMEP, b) Pmax, c) indicated efficiency and d) VE

Overall, the VEs have all been verified to within 5% of each other at all of the engine loads,

and the other parameters correlate well with each other, therefore, the model has been

sufficiently verified for the load conditions (5.5-8.5 bar IMEP) used in this investigation. This

will enable it to be used to accurately calculate the average in-cylinder temperatures for the

recorded measurements in this investigation. Furthermore, the ability to change the intake air

temperature has been utilized in the study of EGR, particularly with hot and cooled EGR, in

order to incorporate the effect of the intake air temperature on the average in-cylinder

temperatures. Despite the model being accurate, the author acknowledges that more work

86
could be done in order to improve the accuracy of the modelling, such as creating and

developing a detailed 3D engine model.

3.6.6 Heat Release Analysis

The net heat release rate (dQ/dθ) was calculated from the in-cylinder pressure data in order to

calculate MFB, using Equation 3-6 (Stone, 1999).

(3-6)

where (heat capacity ratio) is the ratio of specific heats which is approximated for different

parts of the engine cycle, (dV/dθ) is the rate of change of in-cylinder volume with respect to

CAD, (dP/dθ) is the rate of change of in-cylinder pressure with respect to CAD and (dQw/dθ)

is the cumulative net heat release rate. The heat release rate was then integrated in order to

obtain the MFB, as shown in Equation 3-7 (Stone, 1999).

(3-7)

The numerator of the equation is the integral of the net heat release rate between a specified

intake crank angle before the combustion process (i.e. CAD at MFB0) and the crank angle of

interest (e.g. CAD at MFB50), and the numerator of the equation is the integral of the net

heat release rate across the complete combustion window (i.e. between the CAD at MFB0

and MFB100). Key parameters were then obtained from the MFB curve; mainly the MFB50

along with MFB10-90, which are important in the interpretation of the combustion data.

3.6.7 Thermal (Indicated) Efficiency

In order to calculate the indicated efficiency (η) for this study, Equation 3-8 was used

(Cengel and Boles, 2007).

87
 (3-8)

where is the net work output from the engine, LHV is the lower heating value of the fuel

(assumes that all water from the combustion remains as vapour) and is the net mass

flow rate of fuel into the combustion chamber.

3.6.8 Fuel Consumption

The gravimetric net indicated specific fuel consumption (ISFCnet) was calculated using the

aforementioned VAF readings (converted into a mass flow rate, ) and the Lambda value

(λ), using Equation 3-9.

(3-9)

where Air/Fuel ratiostoichiometric is the stoichiometric air/fuel ratio and P i is the indicated power.

The intake temperature was also used to calculate the air density in order to improve the

accuracy of the VAF rate reading (converted into ), before it was used to calculate the

fuel consumption.

3.6.9 Knock Analysis

A MatLab script was developed by the author in order to remove the unwanted noise from the

pressure trace, to calculate the knocking amplitude. The script read the pressure data and

applied a Butterworth 2nd order type filter to isolate the frequency range of 4-12 kHz, which

ensured that the first and second harmonic knocking frequencies from the engine remained

after the low and high frequency engine generated signal noise had been removed. It then

calculated the knocking amplitude from the amplitude of the filtered pressure trace. This

provided the knocking amplitudes from the pressure data, which allowed the KLMBT spark

timing to be quantified for each engine condition. The KLMBT was defined as the most

88
advanced spark timing in which 97% or less of the knock amplitudes were below 2 bar. The

maximum acceptable knock amplitude of 2 bar was chosen based on the work of (Mittal et

al., 2007). If the maximum brake torque (MBT) was reached before the KLMBT spark

timing, then this spark timing was defined as the KLMBT; an asterisk has been added to

denote that the spark timing was not knock limited. Using this method of finding the KLMBT

spark timing resulted in a variation margin of up to 10% for lower load (5.5 bar IMEP) and

5% for higher load (7.0 and 8.5 bar IMEP) conditions when the spark-timing was adjusted at

each testing point for the engine tests in Chapters 4 and 5 of this thesis. It is believed that this

should not affect interpretation of the data because the difference made by the margin in load

can be neglected.

All of the results in this investigation are reported at the KLMBT spark timings, or at the

MBT spark timing if the knock limit was not reached, because these are the optimum engine

operating points in order to achieve best fuel consumption and engine power (Zhu et al.,

2007). The engine of a vehicle will always be calibrated at the KLMBT/MBT (or at a set

spark retard from the KLMBT/MBT) for steady state warm running, if emissions are not a

concern, because customers typically purchase a vehicle based on its rated fuel consumption

and output power. Therefore, the author considered them to be the most appropriate operating

points at which to record the data, in order to maximize the impact of this research.

The HSDAQ LabView program discussed in Section 3.6.1 was later modified by the author,

as shown in Figure 3-26, to implement on-line filtering of the pressure trace, in order to

quickly identify the KLMBT spark timing before the engine data was recorded. This enabled

the tests in Chapter 6 of this thesis to be completed in a significantly shorter time than the

method described in the previous paragraph. It also enabled the discussed load variation to be

eliminated.

89
Figure 3-26 Modified HSDAQ LabView program displaying the on-line knock amplitudes

3.6.10 Gravimetric Gaseous Emissions Analysis

In order to calculate the indicated specific emissions (g/kWh) for this study, Equation 3-10

was used (Heywood, 1988).

(3-10)

where ρemission represents the density of the gaseous emission, Emissionppm is the reading of

the emission from the Horiba system which was recorded by the data acquisition system,

is the volumetric flow rate of the exhaust stream and P i is the indicated power.

90
The molar mass of the exhaust stream was calculated based on the fraction of each

component (NOx, HC, CO, CO2, N2) and it was subsequently used to calculate the density of

the emission (ρemission) using Equation 3-11.

(3-11)

where MEmission represents the molar mass of the emission (e.g. NO), P and T are the exhaust

stream pressure and temperature, respectively, and R is the gas constant. The calculation was

conducted separately for the NOx and HC emissions to provide the data presented in the

thesis.

3.6.11 Confidence Intervals

For the data presented in this thesis, the averaged data from the 3 readings for each

measurement was plotted along with the 95% confidence intervals, where appropriate, in

order to enable the significant effects on the data to be identified. The confidence intervals

were calculated using Equation 3-12.

(3-12)

where CI = confidence interval, = mean, = factor based on the desired confidence

interval of 95%, which is 1.96, = standard deviation and n = sample size.

These confidence intervals only address the random errors that occur in the measurements;

system errors are not included. However, these are discussed further in the following section.

3.7 Analysis of the Uncertainties in the Recorded Data

For a detailed analysis of the analysis of the uncertainties in the data recorded for this thesis,

please refer to Appendix A1.

91
3.8 Fuel Properties
The properties of the fuels used in this investigation are shown in Table 3-9. Both gasoline

and ethanol were supplied by Shell Global Solutions, UK. The 1-butanol and heptane were

supplied by Fisher Scientific UK Ltd at a purity of 99.5%.

Table 3-9 Test Fuel Properties

ULG95 1-Butanol Ethanol Bu20 E20

Chemical Formula C2-C14 C4H10O C2H6O C2-C14 C2-C14

H/C Ratio 1.922 2.5 3 2.038 2.084

O/C Ratio 0.021 0.25 0.5 0.067 0.093

Gravimetric oxygen content (%) 2.36 21.6 34.78 6.21 8.84

Density @ 20˚C (kg/m3) 743.9 811 790.9 757.3 753.3

RON 95 98 106 - 1021, 2

Stoichiometric air–fuel ratio 14.15 11.2 8.95 13.71 13.78

LHV (MJ/kg) 42.22 32.71 26.93 39.73 37.76

Heat of Vaporization ∆vapH (25˚C) (KJ/mol) ~37.34 52.05 42.36 34.6 34.6

Initial boiling point (˚C) 34.6 118 78.4 - -

1
(Anderson et al., 2012a)
2
(Foong et al., 2014)
3
Measured at the University of Birmingham.
4
(Balabin et al., 2007)
5
(National Institute of Standards and Technology, 2011a)
6
(National Institute of Standards and Technology, 2011b)

Overall, the properties of the gasoline used represent the majority of gasoline fuel sold in

Europe and America, making it suitable for this investigation. Although a commercial

gasoline fuel was used for part of this investigation (first part of Chapter 4), it matched the

RON and ethanol content of the ULG95 fuel used in the remainder of the investigation, so it

is assumed that the fuel specification did not vary significantly from the one shown in the

table. Oxygenate content is thought to have varied, however, based on the results observed

and this is discussed further in a later section. The gravimetric content (GC) analysis of the

fuel is shown in Table 3-10.

92
 Table 3-10 Gasoline GC Analysis

Component % volume
Paraffin 43.9

Olefin 11.7

Naphthene 6.2

Naphthene, 1.6

Polynaphthene <0.1

Aromatics 26.9

Saturated C11+Hydrocarbons 2.0

Oxygen Compounds 7.7

Saturated Hydrocarbons 50.1

Unsaturated Hydrocarbons 13.3

3.9 Experimental Test Procedures

Before any experimental data was collected for the engine tests, a baseline test was conducted

to ensure that the data recorded from the engine was approximately the same as a previously

recorded accurate baseline. If there was a significant variation between the two then the cause

was identified and resolved before the baseline test was repeated. The procedure was repeated

until the engine results on a particular day conformed to the previously determined baseline

conditions. Before recording data, the engine cylinder block was warmed to 95˚C as

measured by a thermocouple embedded 5 mm inside the block, to ensure consistent knock

behaviour. Once the engine was recording reliable data, the test point was conducted and it

was repeated several times until 3 consistent sets of data were recorded, in order to ensure

that the data recorded was an accurate representation of the engine combustion and emissions

and to reduce the impact of day-to-day fluctuations on the results. This was then repeated to

collect all of the test points, which were required on a particular day.

93
3.10 Summary
Overall, this chapter has described the experimental test facilities used in this work. All of the

tests were conducted in a 4-stroke spray-guided DISI single-cylinder thermal engine. The

engine is typical of modern, while not state of the art, gasoline DISI engines. Finally, the data

collection and analysis procedures were discussed.

Although the author benefitted from the engine setup created and developed by the various

previous research students at the University, several important improvements were made in

order to further develop and improve the setup. Firstly, the connection of the gasoline fuel

line directly to the fuel tank along with the incorporation of a three-way valve in the PFI fuel

line, in order to fill the DI accumulator directly from the gasoline fuel tank, improved safety

in the engine test cell and made fuel refilling easier. Secondly, the connection of the fuel

tank, previously used to fill up the DI system, to the main gasoline fuel tank, allowed residue

fuel from the DI system to be easily pumped out of the tank. This greatly minimized fuel

vapours from entering the atmosphere when the DI system was subsequently purged into the

tank. Thirdly, the creation of separate fuel tank for non-gasoline fuels made it easier to

prevent contamination of the gasoline and non-gasoline fuels. Finally, interference between

signal lines for the VVT system and water temperature control system were eliminated.

Previously when the mains water control valve was triggered open to cool the engine, the

VVT control system was sometimes affected, causing the valve timings to reset, meaning

measurements had to be repeated regularly. These modifications are in addition to three

engine rebuilds, VAF meter calibrations and ongoing work to maintain the engine test

facility.

94
 Chapter 4

The Effect of EGR and its Type on Engine Combustion and Emissions

The aim of this chapter is to provide details of the combustion and emissions characteristics

of a DISI engine operated with EGR addition at different engine loads of 5.5, 7.0 and 8.5 bar

IMEP, and with different EGR types of hot EGR, cooled EGR, hot EGR after TWC and

cooled EGR after TWC at a single engine load of 7.0 bar IMEP.

4.1 Introduction
The main combustion parameters investigated in this chapter are the KLMBT spark timing,

in-cylinder pressure, calculated average in-cylinder temperature, MFB, fuel consumption and

EGT. PM, NOx and HC are the main emission parameters investigated.

There is an ever growing demand for reduction of NOx emissions from engines and for

improvements in fuel economy, as emissions regulations become stricter and CO 2 emissions

are at the forefront of public perception. One way to achieve these demands is to use external

EGR in engines to suppress the temperature rise in the combustion chamber, reducing NO x

emissions and allowing the KLMBT spark timing to be advanced, as well as enabling the

throttle to be more opened, providing significant fuel economy benefits. Furthermore the

Euro VI emissions regulations which limit PN for the first time have increased interest in the

effect of EGR addition on engine particulates, with the potential of EGR coming under

question because it has been shown to have a negative effect on particulates (Gill et al., 2011,

Ma et al., 2014).

Overall, the author recognizes that the Euro VI emissions regulations have led many to

question the potential of EGR, noting that it is generally considered to increase engine

95
particulates. Thus, this is an area that has been researched further in this chapter. There is also

a lack of work regarding the effect of different types of EGR (hot EGR, cooled EGR, hot

EGR after TWC, and cooled EGR after TWC) on engine combustion and gaseous emissions,

on the same engine at the same load condition, thus, this was also investigated further in this

chapter.

4.2 Experimental Procedure

The experimental test procedure outlined in Section 3.9 of this thesis was followed in order to

conduct the engine tests, to collect the data presented and discussed in this chapter. Relative

air-fuel ratio λ was maintained at 1 during the experiments and a COV IMEP of 5% was not

exceeded. Valve timings were set at IVO=16˚bTDC, EVC=36˚aTDC. KLMBT spark timings

and a fixed geometric compression ratio of 11.5 were used. In order to achieve EGR in the

engine, and in order to achieve the desired engine load and EGR condition, the equipment

setup and test procedure outlined in Section 3.2.6.1 was followed for the hot EGR condition,

and those outlined in Sections 3.2.6.2, 3.2.6.3 and 3.2.6.4 were followed for the cooled EGR,

hot EGR after TWC and cooled EGR after TWC conditions, respectively.

The load range of 5.5-8.5 bar IMEP was chosen to study in the first part of this chapter

because it represents the medium-high load range of this engine where EGR should be of

most benefit to reduce engine knock, thus making the results more useful. In addition, the use

of DI in the engine is most relevant to this load range for knock suppression; at lower loads

the reduced in-cylinder pressure prior to ignition resulting from the charge cooling is not

necessary and it reduces indicated efficiency. Furthermore, in a real engine, a DI fuel pump

will consume more energy than a PFI fuel pump. Therefore, there must be a fuel consumption

benefit of using DI; i.e. charge cooling, in order to compensate for the additional energy the

DI pump would consume. The load of 7.0 bar IMEP was chosen to study in the second part of

96
this chapter because allowed a reasonable, while not a high EGR addition of 14% to be

achieved. A higher engine load would have meant that the throttle would have needed to be

more opened, thus reducing the maximum level of EGR that would have been achieved. One

load was chosen to study so that an in-depth understanding could be formed, which would not

have been possible if multiple loads were studied with the multiple EGR types.

The spark was swept for all engine loads and all EGR types in order to find the KLMBT

spark timing using the technique outlined in Section 3.6.9, or by using the torque calculated

from the engine data if the MBT was reached before the knock limit. A mixture of pump and

research grade gasoline was used in this research with the properties outlined in Section 3.8.

The pump grade gasoline was used in the experiments for the first part of this chapter and the

research grade gasoline was used in the experiments for the second part of this chapter. The

pump grade gasoline specification did not differ significantly from the research grade

gasoline specification; the RON and ethanol content were the same. Therefore, it is believed

that the use of both gasoline types did not significantly affect the interpretation of the results.

Oxygenate content is thought to have varied, however, based on the results observed and this

is discussed further in a later section. The DMS 500 was used to measure the PM emission

and the Horiba MEXA-7100DEGR was used to measure the gaseous emissions of NOx and

HC. PM was not measured in the second part of this chapter because of limited equipment

availability.

The test matrix provided in Table 4-1 was carried out in order to investigate the effect of hot

EGR at different engine loads on the engine combustion and emissions. The results and

discussion of these tests can also be found in (Lattimore, 2016a).

97
 Table 4-1 Experiment test matrix (EGR addition & engine load)

EGR Addition (%) 0 2 3 4 8 12 13

IMEP (bar)
5.5 1 2 3 4 5
7.0 6 7 8 9
8.5 10 11 12

The test matrix provided in Table 4-2 was carried out in order to investigate the effect of

different EGR types on the engine combustion and gaseous emissions at the single load of 7.0

bar IMEP.

Table 4-2 Experiment test matrix (EGR addition & EGR type)

EGR Addition (%) 0 7 10 14

EGR Type
Hot EGR 1 2 3
Cooled EGR 4 5
Hot EGR after TWC 6 7
Cooled EGR after TWC 8 9

4.3 Results and Discussion

4.3.1 Hot EGR Effect on Combustion and Emissions in a DISI Engine

4.3.1.1 KLMBT Spark Timing

The KLMBT spark timings for the tested engine conditions are presented in Table 4-3. EGR

enabled the spark to be advanced significantly, through the suppression of temperature rises

in the end-zone. It can be seen in Figures 4-2a, 4-2b and 4-2c that EGR addition led to

calculated average in-cylinder temperature reductions even with spark advances for the

engine loads tested. However, the spark advances did enable the combustion temperature

reducing effect of EGR to be mitigated, as can be most clearly seen in Figure 4-2b for the

engine load of 7.0 bar IMEP. It must be noted that knocking could have been avoided if the

compression ratio was fully optimized for the fuel used. However, this was not pursued in

98
order to investigate the effect of the parameter changes on engine knock limit and the

consequent combustion and emissions at this limit.

For the load of 5.5 bar IMEP, the spark timings were not knock limited due to the low

combustion pressure and temperature as shown in Figures 4-1a and 4-2a respectively,

allowing them to be advanced by 12 CAD to maintain their optimum MBT phase across the 0

to 8% EGR range. However, the laminar flame speed reducing effects of EGR (Rhodes and

Keck, 1985) resulted in the combustion phasing becoming retarded from the optimum with

further EGR additions of 12 and 13% (Figure 4-3a), despite the further 6 CAD spark advance

that was implemented.

The spark timings at the other engine loads however were knock-limited so they could only

be advanced by 8 and 4 CAD respectively for the engine loads of 7.0 and 8.5 bar IMEP. EGR

was limited to 3% for the engine load of 8.5 bar IMEP because the throttle valve was more

open (~15˚ throttle angle) compared to the other engine loads where the throttle valve was

more closed (~7˚ at 7.0 bar IMEP, ~5˚ at 5.5 bar IMEP), which increased the intake manifold

pressure and subsequently reduced the pressure difference between the exhaust and the

intake. This meant that the EGR gas flow rate in the EGR line was reduced, which reduced

the EGR ratio compared to the other engine loads. The relatively small throttle angles

resulted from the oversized throttle; it was designed for a 3.0 litre engine, meaning that it did

not need to be opened fully in order to achieve the maximum engine load, as the case would

have been with a smaller throttle valve. Despite the limited EGR that was achieved at the

highest considered load, it was still important to study this engine load to quantify what

improvements could be made with the limited EGR available. This condition is used to

represent the high-end load of this engine. The highest load of ~9.5 bar IMEP which can be

99
achieved with this engine was not pursued because no EGR addition can be achieved at this

condition, because the engine is operating at WOT.

It is anticipated that the KLMBT spark timing advances would have been greater for the

engine loads of 7.0 and 8.5 bar IMEP if the intake manifold temperature increases resulting

from the EGR addition could have been eliminated. This is because the lower intake

temperature would have decreased the engine knock tendency. For 5.5 bar IMEP, the flame

would be expected to travel slower in the cooler fuel-air mixture resulting from the cooler

intake manifold, thus necessitating greater MBT spark timing advances to attempt to maintain

the optimum combustion phasing.

Table 4-3 KLMBT spark timings (˚bTDC) (* = not knock limited) (EGR addition &
engine load)

EGR Addition (%) 0 2 3 4 8 12 13

IMEP (bar)
5.5 27* 33* 39* 45* 45*
7.0 14 14 18 22
8.5 10 12 14

4.3.1.2 In-cylinder Pressures and Temperatures

The in-cylinder pressures for the engine load of 5.5 bar IMEP are presented in Figure 4-1a.

They show that in-cylinder pressure reduced as the EGR addition increased. This is because

the EGR gases had a higher heat capacity than fresh air and the flame travelled at a slower

speed (Rhodes and Keck, 1985), reducing the temperature and subsequent pressure rises in

the combustion chamber, causing the overall in-cylinder pressures to reduce. In addition,

there was up to a 10% load variation at the lower engine loads when the spark-timing was

adjusted at each testing point, as discussed, accounting for some of the observed in-cylinder

100
pressure reduction as EGR was increased; IMEP reduced by 0.15 bar across the EGR range at

the engine load of 5.5 bar IMEP.

The in-cylinder pressures for the load of 7.0 bar IMEP in Figure 4-1b show a different

behaviour to those at 5.5 bar IMEP, with slight increases observed, despite similar ratios of

EGR being achieved. This occurred because the spark timings for the engine load of 5.5 bar

IMEP were not knock limited, however, at 7.0 bar IMEP they were. Therefore, the spark

timings at 5.5 bar IMEP could always be advanced to attempt to continue delivering their

optimum MFB50 phase of 8-10˚aTDC, which resulted in the optimum in-cylinder pressure

and temperature at the 0% EGR condition (optimum balance between compression and

expansion work produced on the piston). With the EGR addition, the combustion became

slower, and this could not be compensated for by advancing the MFB50. This meant that the

EGR addition resulted in the overall combustion pressures being reduced. The spark timing at

7.0 bar IMEP however, being knock limited, could always be advanced towards its optimum

MFB50 phasing with EGR addition, causing the in-cylinder pressure to be maintained and

even increased slightly. Despite this, it appears that the average in-cylinder temperature could

not be maintained causing it to reduce across the EGR range as shown in Figure 4-2b.

Similar in-cylinder pressure behaviour was observed at the engine load of 8.5 bar IMEP in

Figure 4-1c along with similar in-cylinder temperature decreases across the EGR range as

shown in Figure 4-2c. Therefore, it is again concluded, as with 7.0 bar IMEP, that although

the KLMBT spark timings advances (towards their optimum MFB50 phasing) were not

sufficient to maintain or increase the average in-cylinder temperature, they were sufficient to

increase the in-cylinder pressures. Thus, EGR addition appears to reduce the calculated

average in-cylinder temperature more than it reduces in-cylinder pressures.

101
At 5.5 bar IMEP, the in-cylinder pressure increases would likely have been reduced, due to

the lower intake temperatures, if the intake manifold temperature increases could have been

eliminated. It would be anticipated that the in-cylinder pressure increases would have been

greater for the engine loads of 7.0 and 8.5 bar IMEP if the intake manifold temperature

increases resulting from the EGR addition could have been eliminated. This is because the

KLMBT spark timing advances would have been greater, as discussed.

Fig. 4-1a: In-cylinder Pressure (5.5 bar)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR
12% EGR 13% EGR

Fig. 4-1b: In-cylinder Pressure (7.0 bar)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR 12% EGR

102
 Fig. 4-1c: In-cylinder Pressure (8.5 bar)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 2% EGR 3% EGR

Figure 4-1 In-cylinder pressures versus CAD at KLMBT spark timings for a) 5.5 bar
IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP

The calculated average in-cylinder temperatures decreased across the EGR range for the three

tested engine loads of 5.5, 7.0 and 8.5 bar IMEP, as shown in Figures 4-2a, 4-2b and 4-2c. As

explained, despite the in-cylinder pressures increasing across the EGR range for the engine

loads of 7.0 and 8.5 bar IMEP, the calculated average in-cylinder temperatures decreased, as

they did at 5.5 bar IMEP (where in-cylinder pressure decreased across the EGR range), due to

the temperature suppression effect of the EGR gases, resulting from their higher heat capacity

as compared to that of fresh air.

At 5.5 bar IMEP, it would be expected that the in-cylinder temperature decreases would have

been greater if the intake manifold temperature increases could have been eliminated, due to

the lower intake gas temperature. However, for the engine loads of 7.0 and 8.5 bar IMEP it is

believed that the increased intake manifold temperatures resulting from the EGR addition

were compensated for by the less optimized combustion phasing resulting from the

103
consequent reduced KLMBT spark timing advance, compared to the maximum spark

advance possible. This was a result of those higher intake manifold temperatures; lower

temperatures would have enabled a greater spark advance before significant levels of knock

occurred. The gas temperature during the intake stroke would have been reduced slightly, if

the intake manifold temperature increases could have been eliminated. However, it is thought

that the improved KLMBT spark timing would have quickly compensated for the decrease by

the time temperatures reached the beginning of the range reported (500˚C), due to the

combustion happening in a smaller volume; thus, the effect would have been minimal.

Fig. 4-2a: In-cylinder Temperature (5.5 bar)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR
12% EGR 13% EGR

104
 Fig. 4-2b: In-cylinder Temperature (7.0 bar)
2100

Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR 12% EGR

Fig. 4-2c: In-cylinder Temperature (8.5 bar)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 2% EGR 3% EGR

Figure 4-2 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) 5.5 bar IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP

105
4.3.1.3 MFB

From Figure 4-3a it can be seen that the MFB profile for the engine load of 5.5 bar IMEP

became smoother and elongated with EGR addition, as the laminar flame speed was reduced

by the EGR addition (Rhodes and Keck, 1985), increasing the combustion duration. This

decreased the rate at which the fuel was burned, resulting in the observed behaviour. This is

reflected in the calculated average in-cylinder temperature data discussed in the previous

section which showed that the overall combustion temperatures reduced with EGR addition,

not only at 5.5 bar IMEP, but at the other engine loads too. If the compression ratio was

optimised for the fuel used, the combustion phasing would have been optimised for each test

point. Therefore, it would be expected that the combustion of gasoline would have been

quicker at the knock limited spark timings due to higher in-cylinder turbulence at the time of

ignition.

The MFB for the load of 7.0 bar IMEP, as shown in Figure 4-3b, followed the same trend as

that for 5.5 bar IMEP between the EGR ratios of 0 and 4%; the profile became smoother and

elongated, as the combustion duration was increased. However, the MFB profile for the EGR

ratio of 8% became more like that observed at 0% EGR, and the combustion duration became

slightly longer. Despite this, the differences between the MFB data at the EGR additions of 4

and 8% are likely due to experimental uncertainties. The KLMBT spark timing was advanced

by 4 CAD between the EGR ratios of 8 and 12% resulting in the combustion beginning

earlier, allowing a more advanced MFB profile to be achieved as compared to that observed

at 0 and 8% EGR addition. However, the profile became more elongated because the slower

flame speed reduced the rate at which the fuel was burned. These results suggest that the

KLMBT spark timing advance achieved with EGR addition at 7.0 bar IMEP was more able to

overcome the reduced laminar flame speed caused by the EGR addition, but at 5.5 bar IMEP,

the spark advance was less able to overcome it. Overall, the gradients were much more

106
consistent for 7.0 bar IMEP (Figure 4-3b) as compared to 5.5 bar IMEP (Figure 4-3a)

because of this.

The MFB data in Figure 4-3c for the load of 8.5 bar IMEP showed a similar trend to the data

at 7.0 bar IMEP after the initial EGR addition and spark advance. However, as the spark was

advanced further with more EGR addition, the MFB profile became significantly advanced

suggesting that the increased spark advance at 3% EGR was better able to overcome the

combustion slowing effects of the EGR gases than was the case with the spark advance

achieved at 2% EGR.

Taking the MFB data for the three engine loads into consideration, the KLMBT spark

advances achieved with EGR addition could overcome the resulting slower flame speeds in

some cases, but in other cases, the advances were not sufficient, causing the MFB profiles to

become smoother and more elongated. This may be further related to the observed

improvements in fuel economy and the reduction in EGT, along with the engine emissions;

all of which will be discussed further in the following sections. Overall, these relationships

should be studied further to find an optimized MFB for gasoline combustion.

For 5.5 bar IMEP, the MFB profile would be expected to become more elongated if the

intake manifold temperature increases resulting from the EGR addition could have been

eliminated, due to the slower flame speed in the cooler fuel-air charge. This would be

expected to occur at the other engine loads too, but much less significantly, because the

temperature decreases with the cooler intake gases would be expected to be quickly cancelled

out by the effect of the greater KLMBT spark timing advances, as discussed. Furthermore,

the MFB profile for 5.5 bar IMEP between the EGR ratios of 0 and 8% would not become

more advanced or retarded, because the optimum combustion phasing would be expected to

107
be maintained. However, it would still be expected to become retarded at the EGR additions

of 12 and 13%. It would be anticipated that the MFB profile advances would have been

greater for the engine loads of 7.0 and 8.5 bar IMEP. This is because the KLMBT spark

timing advances would have been greater, as discussed.

Fig. 4-3a: MFB (5.5 bar)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR
12% EGR 13% EGR

Fig. 4-3b: MFB (7.0 bar)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR 12% EGR

108
 Fig. 4-3c: MFB (8.5 bar)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 2% EGR 3% EGR

Figure 4-3 MFB versus CAD at KLMBT spark timings for a) 5.5 bar IMEP, b) 7.0 bar
IMEP and c) 8.5 bar IMEP

4.3.1.4 Fuel Consumption

Gravimetric ISFCnet at the KLMBT spark timings for the three engine load conditions at the

different EGR ratios tested is shown in Figure 4-4. There is a clear trend of decreasing fuel

consumption as the EGR addition was increased because, as explained previously, the

KLMBT could be advanced significantly, as the increasing EGR addition suppressed

temperature rises in the end-zone of the combustion chamber, inhibiting knock that would

otherwise have been produced. This allowed the MFB50 phasing of the engine to be

advanced towards its optimum phase for the engine loads of 7.0 and 8.5 bar IMEP, increasing

the engine efficiency and consequently reducing the fuel consumption (de O. Carvalho et al.,

2012). This occurred because more of the combustion process occurred at a lower in-cylinder

volume and thus, it generated higher combustion pressures. This improved the rate of

combustion and it increased the expansion of the combustion products into useful energy.

109
Also, because the gas from EGR takes a part of the intake air volume, the throttle valve had

to be opened more to provide the same amount of fresh air as in the non-EGR case, or in

other words, the same amount of oxygen for combustion. This reduced the pumping losses

across the EGR range and thus the fuel consumption at all engine loads; PMEP reduced from

0.45 to 0.38 bar for the load of 5.5 bar IMEP, 0.33 to 0.25 bar for the load of 7.0 bar IMEP,

and 0.24 to 0.22 bar for the load of 8.5 bar IMEP, across their respected EGR ranges. For the

loads of 5.5, 7.0, and 8.5 bar IMEP, the fuel consumption reduced by 2.2, 4.1 and 1.0%

respectively across the EGR range. Furthermore, the reduced combustion temperatures with

EGR addition improved thermal efficiency (Ratnak et al., 2015, Siokos et al., 2015),

contributing to the fuel economy improvements observed.

The rate of fuel consumption decrease at 5.5 bar reduced at the higher EGR ratios because the

spark timings were not knock limited for this relatively low load. Thus, between 0 and 8%

EGR addition, the spark timing was at its optimum phase, and it became retarded slightly

from its optimum phase as the EGR addition was increased to 12 and 13%. This left little

opportunity for fuel consumption improvement through pumping loss reduction.

Furthermore, because the MFB50 became later at 12 and 13% EGR addition (Figure 4-3a),

the rate of fuel consumption decrease was reduced at these conditions. It is proposed that the

increased rate of fuel consumption decrease observed at the engine load of 7.0 bar IMEP

when the EGR ratio was increased from 8 to 12% occurred because the in-cylinder pressure

increased significantly in comparison to that observed with the previous EGR increase from 4

to 8% (Figure 4-1b). This was possibly due to the earlier start of heat release (Figure 4-3b).

The in-cylinder pressure at 8.5 bar IMEP only increased slightly between the EGR ratios of 0

and 2%, despite the EGR addition and consequent spark advance, leading to the relatively

small improvement in fuel economy (Figure 4-4). However, the in-cylinder pressures

110
increased significantly as the EGR addition was increased further to 3%, leading to the

significant improvement in fuel economy observed.

For the load of 5.5 bar IMEP, the fuel consumption improvements would have worsened if

the intake manifold temperature increases could have been eliminated, because the

combustion duration would have increased, as discussed, meaning that the conversion

efficiency of the pressure and heat from the combustion would have reduced. Fuel

consumption improvements for the loads of 7.0 and 8.5 bar IMEP would have been greater,

due to the greater KLMBT spark timing advances that would be expected. This effect would

be expected to overcome the slightly increased combustion duration that would also be

expected at these engine loads.

Despite the fuel consumption reduction observed and discussed, it must be noted that the

VAF had an average error of 1.6%; therefore, the reduction of 1.0% observed across the EGR

range at the engine load of 8.5 bar IMEP is limited, because it could have resulted from this

error. The reductions observed for 5.5 and 7.0 bar IMEP are however above 1.6%, thus, they

are not limited. The error mainly resulted from the limited resolution of the u-tube

manometer device used to calibrate the air flow meter. Despite the resolution being 1 mm, it

still remained a limiting factor in achieving better calibration between the VAF and u-tube

manometer reading, thus limiting the VAF accuracy. The error also resulted from the orifice

flow plate used to calibrate the flow meter.

111
 267
265
263
ISFCnet (g/kWh) 261
259
257
255
253
251
249
247
245
0 2 4 6 8 10 12 14
EGR (%)
5.5 bar 7.0 bar 8.5 bar

Figure 4-4 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings

4.3.1.5 EGT

The EGT decreased significantly across the EGR range for the three tested engine loads, as

shown in Figure 4-5. Overall, it decreased by 21.9, 32.7, and 7.6˚C for the loads of 5.5, 7.0

and 8.5 bar IMEP, respectively. As previously discussed, the calculated average in-cylinder

temperatures significantly reduced with EGR addition for the engine load of 5.5 and 8.5 bar

IMEP, and they reduced less significantly for the engine load of 7.0 bar IMEP, contributing

to some of the reductions in EGT across the EGR range. However, it is thought that the main

contribution to the reduced EGTs was the advanced combustion phasing enabled by the EGR

addition. It is proposed that this led to a more efficient conversion of the pressure and heat

from the combustion into piston work, causing the gases to cool more in the engine cylinder

before they were ejected into the exhaust. With the combustion starting earlier as the EGR

addition was increased, this effect became greater, causing the EGTs to reduce across the

EGR range. For turbocharged engines with EGR, this effect is very useful for protecting the

turbocharger turbine.

112
The temperature decrease is lower for 5.5 bar IMEP compared to 7.0 bar IMEP despite the

maximum level of EGR increasing. This is because the conversion efficiency of the pressure

and heat from the combustion into piston work could not be improved due to the combustion

already being at its optimum phasing. Furthermore, the optimum combustion phasing could

not be maintained at 12 and 13% EGR addition, as discussed, resulting in the EGTs not

decreasing further at these conditions. The EGT at 7.0 bar IMEP increased slightly by 6˚C at

the EGR ratio of 4%, whereas one would expect it to have decreased. Another slight increase

was observed between the EGR ratios of 0 and 2% at 8.5 bar IMEP. Despite these observed

increases, they are not considered to be significant; therefore, they will not be commented on

further.

For the load of 5.5 bar IMEP, the EGT decreases would have likely reduced if the intake

manifold temperature increase could have been eliminated, due to the aforementioned

reduced conversion efficiency of the pressure and heat from the combustion, due to an

expected increased combustion duration. The EGT decreases would likely have been greater

for the engine loads of 7.0 and 8.5 bar IMEP, due to an increased conversion efficiency of the

pressure and heat from the combustion, resulting from the greater KLMBT spark timing

advances that would be expected. This effect would be expected to overcome the slightly

increased combustion duration that would also be expected at these engine loads.

113
 680

660

640
EGT (°C)

620

600

580

560
0 2 4 6 8 10 12 14
EGR (%)
5.5 bar 7.0 bar 8.5 bar

Figure 4-5 EGT versus EGR ratio at KLMBT spark timings

4.3.1.6 PM Emissions

The nucleation mode of the PM emissions decreased and the accumulation mode increased as

EGR ratio was increased from 0 to 12% for the engine load of 5.5 bar IMEP, as shown in

Figure 4-6a. The two modes could be distinguished from one another because they formed

two distinct peaks; the nucleation mode peak is comprised of the smaller particles on the left-

hand side of the plot (10-30 nm) (Figure 4-6a) and the accumulation mode peak is comprised

of the larger particles on the right-hand side of the plot (30-500 nm). According to (Zhang et

al., 2014b) the nucleation mode particles mainly result from droplets formed by hydrocarbon

condensation and the accumulation mode particles are mainly composed of carbonaceous

agglomerates formed in local rich-fuel zones (Burtscher, 2005, Kittelson, 1998).

At the engine load of 5.5 bar IMEP, the increase in EGR suppressed the temperature rise in

the combustion chamber, reducing the oxidation rate of soot particles, leading to the

increased accumulation mode particles observed between 0 and 12% EGR addition. This

appears to have been more significant than the decreases in primary carbon particles formed

114
by the thermal pyrolysis and dehydrogenation reaction of fuel vapour/droplets, which would

have occurred with the reduced in-cylinder temperatures (Zhang et al., 2014b). It is proposed

that the observed reductions in nucleation mode particles occurred because they were

adsorbed onto the increasing amounts of soot accumulation particles. Thus, as the soot

accumulation mode particles increased across the EGR range, the numbers of nucleation

mode particles decreased. In addition to the reduced oxidation rate of soot particles across the

EGR range caused by the reduced combustion temperatures, it is proposed that the spark

timing advances reduced the time available for fuel-air mixing. This led to poorer mixture

preparation and thus more pockets of high local equivalence ratio which could not burn

completely contributing to the higher numbers of soot accumulation mode particles observed

across the EGR range. The EGT decreases across the EGR range also contributed to the

accumulation mode PM increases observed.

Increases in nucleation and accumulation mode particles were observed across the EGR range

for 7.0 bar IMEP as shown in Figure 4-6b. However, the increase in the nucleation mode was

not consistent, therefore, it will not be commented upon further. It is thought that the

calculated average in-cylinder temperature decreases led to the accumulation mode PM

increases. Again, this appears to have been more significant than the decreases in primary

carbon particle formation (Zhang et al., 2014b). In addition, due to the spark timing advances

achieved, it is proposed that the mixture preparation would have been poorer, contributing

further to the increased soot accumulation mode particles observed across the EGR range.

This is reflected in the MFB data (Figure 4-3b) which shows that the MFB50 was advanced

significantly at the EGR ratio of 12% when compared to the MFB at 0% EGR. Again, the

EGT decreases across the EGR range also contributed to the accumulation mode PM

increases observed.

115
The observed behaviour at 5.5 and 7.0 bar IMEP was not observed at 8.5 bar IMEP in Figure

4-6c, despite the calculated average combustion temperature decreasing (Figure 4-2c). At 8.5

bar IMEP, there was no significant change in the nucleation and accumulation mode particles

observed. This is due to the low EGR addition achieved at this engine load.

Overall, the contribution of nucleation mode particles became weaker as the engine loads

were increased. This agrees with the reduced HC emission observed as the engine load was

increased (Figure 4-8); the HC emission is mainly composed of the volatile particulate

fraction included in the nucleation mode (Costagliola et al., 2013). Thus, this explains the

strong relationship between the HC emission and nucleation mode particles observed in the

data. The accumulation mode particles increased with engine load due to the higher rate of

soot production and growth at higher temperatures in the combustion chamber (Zhang et al.,

2014b). In addition, as the load is increased, more fuel is injected which results in more

locally rich areas in the combustion chamber, further increasing soot formation (Wang,

2014b). (Wang et al., 2014c) observed the activation energies and temperatures required to

oxidise soot increased with engine load, further explaining the increased accumulation mode

particles as the engine load was increased. Furthermore, there is increased fuel impingement

at higher loads resulting from the longer injection pulse width and increased fuel mass

injected, resulting in yet more increases in soot formation (Wang, 2014b).

The PM behaviour at 5.5 and 7.0 bar IMEP is different from that observed by (Alger et al.,

2010) and (Hedge et al., 2011) where significant soot reductions across the EGR range was

reported, with soot levels being represented by the accumulation mode. It is likely however

that because the engines in their investigations were turbocharged, the EGR addition reduced

the need to enrich the fuel charge for the purpose of reducing turbocharger inlet temperatures,

consequently reducing their soot emission. Clearly there are complex behaviours regarding

116
the formation of PM emissions in engines which need to be investigated in detail on a

fundamental level.

It is proposed that the decrease in accumulation mode particles at 5.5 bar IMEP as the EGR

addition was increased to 13% occurred because the spark timing (Table 4-3) was not

advanced as compared to that at 12% EGR addition. This means that while the mixture

preparation was equivalent, the combustion duration was increased (Figure 4-3a), providing

more time for the particles to oxidize in the hot flames. Additionally, there is a decrease in

the accumulation mode particles for the load of 7.0 bar IMEP at 4% EGR addition (Figure 4-

6b). It is believed that the increased exhaust temperatures (Figure 4-5) increased the rate of

post-combustion oxidation of the accumulation mode particles, resulting in the observed

decrease. Such behaviour is reasonable to expect considering soot accumulation mode

particles from a DISI engine can be effectively oxidised between 400-600˚C (Wang et al.,

2014c).

If the intake manifold temperature increases could have been eliminated, the accumulation

mode particles would have been increased more significantly at 5.5 bar IMEP. This is due to

the expected increased MBT spark timing advances further worsening mixture preparation,

and the more significant in-cylinder temperature decreases further reducing the oxidation rate

of the particulates. Nucleation mode particles would consequently be reduced because they

would be adsorbed onto the greater number of accumulation mode particles. At 7.0 bar

IMEP, it would be expected that the accumulation mode particles would be increased too,

resulting, again, in nucleation mode particle decreases, due to poorer mixture preparation

resulting from the greater KLMBT spark timing advances. At 8.5 bar IMEP, it is expected

that there would be no significant effect of the lower intake manifold temperature on the PM

behaviour, due to the low 3% EGR addition achieved.

117
 Fig. 4-6a: PM (5.5 bar)
1.E+09

1.E+08

dN/dlogDp (#/cm3)
1.E+07

1.E+06

1.E+05
10 100 1000
Particle Diameter (nm)
0% EGR 4% EGR 8% EGR
12% EGR 13% EGR

Fig. 4-6b: PM (7.0 bar)

1.E+09

1.E+08
dN/dlogDp (#/cm3)

1.E+07

1.E+06

1.E+05
10 100 1000
Particle Diameter (nm)
0% EGR 4% EGR 8% EGR 12% EGR

118
 Fig. 4-6c: PM (8.5 bar)
1.E+09

1.E+08

dN/dlogDp (#/cm3)

1.E+07

1.E+06

1.E+05
10 100 1000
Particle Diameter (nm)
0% EGR 2% EGR 3% EGR

Figure 4-6 PN emissions versus particle diameter at KLMBT spark timings at a) 5.5 bar
IMEP, b) 7.0 bar IMEP and c) 8.5 bar IMEP

4.3.1.6 Gaseous Emissions (NOx and HC)

The NOx emission at the KLMBT spark timings for the three engine load conditions at the

tested EGR ratios is presented in Figure 4-7. The NOx emission increased significantly as the

engine load was increased due to the higher calculated average in-cylinder temperatures

which increased the formation of NO x in the hot flames. There is a clear trend of the NO x

decreasing as the EGR percentage was increased. EGR reduced the calculated average in-

cylinder temperatures (Figures 4-2a, 4-2b and 4-2c) for the three engine loads tested,

resulting in the NOx reductions observed. The overall reduction for the loads of 5.5, 7.0 and

8.5 bar IMEP are 64.3, 50.9 and 12.2%, respectively, showing the significant potential EGR

has on NOx emission reduction.

The rate of decrease of the NO x emission with EGR addition became greater as the engine

load was increased; however, the NOx decrease in terms of percent, over the achievable EGR

range for the respective loads, reduced as the engine load was increased. The rate of decrease

119
became greater as the engine load was increased because EGR addition was more effective at

reducing the average in-cylinder temperature as the engine load was increased. This can be

seen in Figures 4-2a, 4-2b and 4-2c, with similar calculated average in-cylinder temperature

reductions achieved for the three engine loads tested, despite the maximum EGR addition

reducing as the engine load was increased. It is thought that the greater mass of EGR gases as

the engine load was increased; at equivalent EGR additions in terms of percent, due to the

greater mass of intake air, caused the in-cylinder temperature to be suppressed more

significantly, resulting in the observed behaviour. It is thought that the NO x decrease in terms

of percent reduced as the engine load was increased, because of the reduced maximum EGR

addition that could be achieved.

At 5.5 bar IMEP, the NOx emission would have been reduced more significantly if the intake

manifold temperature increases could have been eliminated, due to the lower expected

combustion temperatures. At 7.0 and 8.5 bar IMEP, there would be no significant change

expected because, as discussed, the peak of the calculated average in-cylinder temperatures

would not have changed significantly. Again, it is thought that the improved KLMBT spark

timing would have quickly compensated for the decreased intake gas temperature by the time

temperatures reached the beginning of the range reported (500˚C). Thus, the effect on the

NOx emission which occurs at temperatures much higher than 500˚C would not have been

significant.

120
 11

10

8
NOx (g/kWh)

2
0 2 4 6 8 10 12 14
EGR (%)
5.5 bar 7.0 bar 8.5 bar

Figure 4-7 NOx emissions versus EGR ratio at KLMBT spark timings

The HC emission data for the three tested engine loads across their respective EGR ranges is

shown in Figure 4-8. At 0% EGR, lower HCs were observed as the engine load was increased

because the calculated average in-cylinder temperatures were higher which increased the

oxidation rate of the HCs, reducing the numbers observed. The reduced calculated average

in-cylinder (Figures 4-2a-c) and exhaust (Figure 4-5) gas temperatures across the EGR range

for the three tested engine loads resulted in the HC emission increases observed across the

ranges for the respective loads. They increased by 48.3, 19.0 and 6.8% across the EGR range

for the engine loads of 5.5, 7.0 and 8.5 bar IMEP, respectively. It is proposed that the reduced

in-cylinder temperatures increased the thickness of the quench layers covering the

combustion chamber surface and slowed down and reduced the evaporation of impinged fuel,

resulting in the HC emission increase. The reduced exhaust temperatures reduced the

oxidation rate of HCs contributing to the increases observed. Some researchers have observed

similar overall HC increases with EGR addition (Alger et al., 2009a, Diana et al., 1996,

Takaki et al., 2014, Zhang et al., 2014b), but others have observed unchanged HC (Grandin et

al., 1998, Kumano and Yamaoka, 2014, Song et al., 2014).

121
The reduced time available for the fuel-air mixture preparation due to the KLMBT spark

advances across the EGR range will have resulted in poorer charge preparation, leading to

more pockets of a high local equivalence ratio, also contributing to the HC emission increases

observed. The same reasoning can also help explain why the HC emissions reduced as the

engine load was increased and the spark timing became more retarded.

There was a small but significant reduction in the rate of HC emission increase at 8% EGR

for 5.5 bar IMEP. This is mirrored by the slight decrease in the rate of reduction in NO x

emissions at the same condition (Figure 4-7). It is believed that the relatively smaller change

in the MFB profile (Figure 4-3a) between the EGR ratios of 4 and 8%, as compared to that

observed between the EGR ratios of 0 and 4%, led to a smaller reduction in the ROHR,

resulting in this behaviour. The HC emission decreased slightly for 5.5 bar IMEP between the

EGR additions of 12 and 13%. Although a decrease was observed it is not considered to be

significant, as can be seen from the confidence intervals on the figure. Therefore, it will not

be commented on further.

If the intake manifold temperature increases could have been eliminated, it is expected that

the HC emission reduction at 5.5 bar IMEP would have been increased, due to the higher

EGTs which could be expected in such a case. At 7.0 and 8.5 bar IMEP, the effect of the

lower EGTs would be expected to be cancalled out by the increased post-combustion time

provided by the greater KLMBT spark timing advance. Thus, it is not thought that the intake

manifold temperature increases with EGR additon had a significant effect on the HC

emission at these engine loads.

Despite the increases in HCs observed across the EGR range for the three engine loads tested,

it would be expected that a traditional TWC would be able to handle these increases to

maintain the vehicle-out HC emissions at non-EGR levels. This is based on the work of

122
(Hoffmeyer et al., 2009) who found that HCs are almost completely reduced by the catalyst,

even as the EGR rate is increased, due to an increased conversion efficiency of the catalyst.

However, more research would be required to confirm whether it can be achieved in this

engine.

11.5

10.5

9.5
HC (g/kWh)

8.5

7.5

6.5

5.5

4.5
0 2 4 6 8 10 12 14
EGR (%)
5.5 bar 7.0 bar 8.5 bar

Figure 4-8 HC emissions versus EGR ratio at KLMBT spark timings

123
4.3.2 Effect of Different EGR Types on Combustion and Gaseous Emissions
in a DISI Engine

4.3.2.1 KLMBT Spark Timing

The KLMBT spark timings, as shown in Table 4-4, could be advanced significantly with

EGR addition for all the EGR types tested. They were advanced by 10, 10, 12 and 14 CAD

for the hot EGR, cooled EGR, hot EGR after TWC and cooled EGR after TWC EGR types,

respectively. The EGR addition enabled the spark to be advanced significantly, through the

suppression of temperature rises in the end-zone. The KLMBT spark timing was most

advanced for the cooled EGR after TWC condition across the EGR range because the

removal of NO from the EGR gases reduced the knocking tendency of the engine (Kawabata

et al., 1999) and the cooler EGR gases reduced the temperature in the end-zone of the

combustion chamber. These reasons also explain the greater KLMBT spark timing advances

for hot EGR after TWC and cooled EGR, respectively, as compared to the hot EGR

condition.

Despite this, KLMBT spark timing advances would likely have been greater if intake

manifold temperature increases resulting from the EGR addition could have been eliminated,

due to the reduced end-zone temperatures. It must be noted that knocking could have been

avoided if the compression ratio was fully optimized for the fuel used, as mentioned

previously.

The EGR addition could only be increased to 10% for the cooled EGR condition because of

the pressure loss in the EGR heat exchanger. This means that a pump would be required to

achieve higher levels of EGR or a more efficient heat changer in terms of cooling power vs.

pressure loss would be required. A pump would reduce any fuel consumption benefits of the

cooled EGR system, therefore reducing any advantages it may have over the hot EGR

124
system. Despite the pressure loss in the heated line for the hot EGR after TWC condition, this

could be compensated for by increasing the gas flow rate of the compressed gas bottle,

allowing the high EGR addition to be achieved.

Table 4-4 KLMBT spark timings (˚bTDC) (EGR addition & EGR type)

EGR Addition (%) 0 7 10 14

EGR Type
Hot EGR 14 18 24
Cooled EGR 20 24
HE after TWC 20 26
CE after TWC 20 28

4.3.2.2 In-cylinder Pressures and Temperatures

The in-cylinder pressures for the four EGR types at the KLMBT spark timings, as shown in

Figures 4-9a, 4-9b, 4-9c and 4-9d, generally increased with EGR addition and the subsequent

spark advances which could be achieved, and the maximum in-cylinder pressure became

slightly earlier due to the more advanced spark timings. This was because the KLMBT spark

timing advances resulted in the heat being released from the fuel at an earlier CAD, enabling

higher in-cylinder pressures to be achieved before the downwards piston motion caused the

in-cylinder pressure to decrease again. However, the maximum in-cylinder pressure remained

approximately the same across the EGR range for hot EGR (Figure 4-9a), with the peak

pressure becoming less advanced than with the other EGR types. This was due to the lower

spark advance achieved at equivalent EGR ratios (Table 4-4) and the resulting retardation of

the MFB50 across the EGR range (Figure 4-12).

Greater in-cylinder pressure increases would have been likely if the intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated. This is

because it would likely have led to greater KLMBT spark timing advances, as discussed.

125
 Fig. 4-9a: In-cylinder pressure (hot EGR)
40

In-cylinder Pressure (bar)

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

Fig. 4-9b: In-cylinder Pressure (cooled EGR)

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 7% EGR 10% EGR

126
 Fig. 4-9c: In-cylinder Pressure (Hot EGR after TWC)

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

Fig. 4-9d: In-cylinder Pressure (Cooled EGR after TWC)

40
In-cylinder Pressure (bar)

35

30

25

20

15

10

0
-50 0 CAD 50 100

0% EGR 7% EGR 14% EGR

Figure 4-9 In-cylinder pressures versus CAD at KLMBT spark timings for a) hot EGR,
b) cooled EGR, c) hot EGR after TWC and d) cooled EGR after TWC

The calculated average in-cylinder temperatures, as shown in Figures 4-10a, 4-10b, 4-10c and

4-10d at the KLMBT spark timings, decreased across the EGR range for the different EGR

types due to the temperature suppression effect of the EGR gases. The temperatures decreases

across the EGR range were greater for the cooled EGR and cooled EGR after TWC EGR

127
types as compared to the hot EGR and hot EGR after TWC EGR types due to their lower

intake plenum temperatures in comparison (Figure 4-18).

It is believed that the increased intake manifold temperatures resulting from the EGR addition

were compensated for by the less optimized combustion phasing resulting from the

consequently reduced KLMBT spark timing advance (Table 4-4), compared to the maximum

spark advance possible, as explained previously. It is thought that the improved KLMBT

spark timing would have quickly compensated for the intake gas temperature decrease by the

time temperatures reached the beginning of the range reported (500˚C); thus, the effect would

have been minimal.

Fig. 4-10a: In-cylinder Temperature (hot EGR)

1900
Calculated In-cylinder Temperature (˚C)

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

128
 Fig. 4-10b: In-cylinder Temperature (cooled EGR)
1900

Calculated In-cylinder Temperature (˚C)

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 7% EGR 10% EGR

Fig. 4-10c: In-cylinder Temperature (hot EGR after TWC)

1900
Calculated In-cylinder Temperature (˚C)

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

129
 Fig. 4-10d: In-cylinder Temperature (cooled EGR after TWC)
1900

Calculated In-cylinder Temperature (˚C)

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

Figure 4-10 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) hot EGR, b) cooled EGR, c) hot EGR after TWC and d) cooled
EGR after TWC

4.3.2.3 MFB

The MFB profiles at the KLMBT spark timings, as shown in Figures 4-11a, 4-11b, 4-11c and

4-11d for the four EGR types, started earlier and became elongated as the EGR ratio was

increased, due to the spark timing becoming more advanced and due to the laminar flame

speed reducing effects of the EGR gases, respectively (Rhodes and Keck, 1985). The MFB

profile for the cooled EGR condition (Figure 4-11b) was approximately the same at the EGR

ratios of 7 and 10%. This is because the rate of spark advance between 7 and 10% EGR was

1.33 CAD/EGR (%), which was higher than the 0.86 CAD/EGR (%) rate of spark advance

achieved between the EGR ratios of 0 and 7%. This allowed the MFB profile to be

maintained with little change between 7 and 10% EGR addition whereas between 0 and 7%

EGR addition, the less significant spark advance caused the MFB profile to become more

elongated, producing the observed behaviour. For a quantitative analysis of the combustion

130
speed, the MFB50 and MFB10-90 have been calculated from the MFB profiles (please see

next section).

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated; thus, the

MFB profiles would likely have been advanced slightly in comparison to those in Figures 4-

11a-d. The profile would also become slightly more elongated because the laminar flame

speed would be reduced in the cooler fuel-air charge.

Fig. 4-11a: MFB (hot EGR)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

131
 Fig. 4-11b: MFB (cooled EGR)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 7% EGR 10% EGR

Fig. 4-11c: MFB (hot EGR after TWC)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

132
 Fig. 4-11d: MFB (cooled EGR after TWC)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
0% EGR 7% EGR 14% EGR

Figure 4-11 MFB versus CAD at KLMBT spark timings for a) hot EGR, b) cooled EGR,
c) hot EGR after TWC and d) cooled EGR after TWC

4.3.2.4 MFB50 and MFB10-90

The MFB50 at the KLMBT spark timings became earlier across the EGR range for the three

conditions of cooled EGR, hot EGR after TWC and cooled EGR after TWC as shown in

Figure 4-12. This is because, as discussed, the KLMBT spark timings (Table 4-4) could be

advanced significantly with the EGR addition. However, for the hot EGR condition, the

MFB50 became later. It seems therefore that the KLMBT spark timing advance achieved was

not sufficient to overcome the laminar flame speed slowing effects of the EGR addition,

whereas for the other EGR types, it was. This is because the spark timing advance was less

significant for the hot EGR condition as compared to the other EGR conditions, because it

did not benefit from the cooled EGR gas or from the removed NO; the other three EGR types

benefited from at least one of these things. Therefore, the lesser spark advance was not able

to overcome the combustion slowing effects of the EGR gases, but the greater spark advances

achieved with the other EGR types were sufficient to overcome them.

133
Despite the same KLMBT spark timings being achieved for the cooled EGR, hot EGR after

TWC and cooled EGR after TWC at 7% EGR addition, the MFB50 data is not the same. It is

proposed that this is due to the different intake plenum temperatures (Figure 4-18) which

meant that the fuel-air charges were of different temperatures before combustion, resulting in

a variation in flame speeds and thus slightly different MFB50 data, despite the same spark

timings.

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated. Thus,

the MFB50 would likely have been slightly advanced in comparison.

18.5

18.0

17.5
MFB50 (˚aTDC)

17.0

16.5

16.0

15.5

15.0

14.5
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-12 MFB50 versus EGR ratio at KLMBT spark timings

The combustion duration (MFB10-90) at the KLMBT spark timings, as shown in Figure 4-

13, increased significantly for all of the EGR types across the EGR range because of the

aforementioned laminar flame speed reducing effects of the EGR gases (Rhodes and Keck,

1985). The cooled EGR and cooled EGR after TWC EGR types were more significantly

affected than the hot EGR and hot EGR after TWC EGR types. It is proposed that this is

134
because the flame travelled at a slower speed in the cooler fuel-air charge. The cases with the

simulated TWC were more greatly affected then their corresponding hot and cooled

conditions respectively. It is proposed that this was because of the slightly reduced intake

plenum temperatures (Figure 4-18) for the TWC conditions as compared to their

corresponding hot and cooled conditions. This is thought to have reduced the laminar flame

speed which increased the combustion duration for these in respect to their corresponding

non-TWC conditions.

As explained, if the intake manifold temperature increases could have been eliminated, it is

believed that the flame would travel more slowly in the resulting cooler fuel-air charge,

increasing the combustion duration slightly. If the compression ratio was optimised for the

fuel used, the combustion phasing would have been optimised for each test point, therefore, it

would be expected that the combustion at all of the conditions tested would have been

quicker due to higher in-cylinder turbulence at the time of ignition.

31

30

29

28
MFB10-90 (CAD)

27

26

25

24

23

22

21
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-13 MFB10-90 versus EGR ratio at KLMBT spark timings

135
4.3.2.5 Fuel Consumption

Gravimetric ISFCnet at the KLMBT spark timings decreased for all EGR types as shown in

Figure 4-14. The fuel consumption reductions were due to MFB50 advances for the cooled

EGR, hot EGR after TWC and cooled EGR after TWC EGR types, along with pumping loss

reductions and thermal efficiency improvements (reduced in-cylinder temperatures), as

explained in the first part of Chapter 4. It must be noted that if the hot and cooled EGR after

TWC EGR types had used a real TWC rather than a simulated TWC then their fuel

consumption would have been higher across the EGR range due to the increased back-

pressure resulting from a real TWC increasing the pumping losses.

Overall, the fuel consumption decreased by 1.8, 3.8, 1.6 and 2.9%, respectively, across the

EGR range, for the hot EGR, cooled EGR, hot EGR after TWC and cooled EGR after TWC

EGR types. Despite these fuel consumption reductions observed, it must be noted that the

VAF had an average error of 1.6%, as discussed, therefore, the reduction of 1.6% observed

across the EGR range for the hot EGR after TWC condition is limited, because it could have

resulted from this error. The reductions for the other EGR types are however not considered

to be limited. Again, KLMBT spark timing advances would likely have been greater if intake

manifold temperature increases, resulting from the EGR addition, could have been

eliminated. Thus, the fuel consumption improvements would likely have been greater.

The fuel consumption reduction was greatest for the cooled EGR condition despite the

limited EGR that could be achieved, because its MFB50 (Figure 4-12) was further advanced

by the EGR addition and consequent spark advances than that of the other EGR types. In

addition, its combustion duration (Figure 4-13) did not increase as significantly as that of the

cooled EGR after TWC EGR type, improving the conversion efficiency of heat and pressure

into piston work through improved combustion phasing. The next greatest fuel consumption

136
reduction achieved was for cooled EGR after TWC. This is because the KLMBT spark

timing advance (Table 4-4) was the greatest, therefore allowing the MFB50 to be advanced

closer to its optimum phase (Figure 4-12). Fuel consumption reductions were achieved with

hot EGR despite its MFB50 becoming retarded with EGR addition (Figure 4-12). This is

because its PMEP was significantly reduced across the EGR range (Figure 4-15) and it

reduced in-cylinder temperatures significantly (Figure 4-10a) which improved thermal

efficiency.

The hot EGR after TWC EGR type achieved fuel consumption improvements of lesser

magnitude than the cooled EGR and cooled EGR after TWC conditions because its KLMBT

spark timing could not be advanced as much, meaning its MFB50 (Figure 4-12) was more

retarded than those of the cooled EGR and cooled EGR after TWC EGR types. At 7% EGR,

its MFB50 was more advanced than that of the cooled EGR after TWC EGR type (Figure 4-

12) and its PMEP was lower (Figure 4-15), but its fuel consumption was still higher. It is

proposed that this was due to the higher peak of the calculated average in-cylinder

temperature (1844˚C) compared to that of the cooled EGR after TWC EGR type (1823˚C), as

shown in Figures 4-10c and 4-10d, respectively. It is thought that this increased heat transfer

to the combustion chamber walls, reducing the thermal efficiency, thus causing the fuel

consumption to be higher.

Hot EGR had lower fuel consumption at the maximum EGR addition of 14% than the hot

EGR after TWC EGR type. Despite this observation, it is not considered to be significant

considering it is within the margin of error, as indicated by the relevant confidence intervals

(Figure 4-14). For the cooled EGR and cooled EGR after TWC EGR types, the fuel

consumption reduction became less significant as the EGR addition was increased from 7%

to its maximum, as compared to that achieved as the EGR addition was increased from 0 to

137
7%. This observation is also not considered to be significant because it is within the margins

of error, as indicated on the figure.

If the intake plenum temperature increases (Figure 4-18) could have been eliminated, it is

believed that the fuel consumption improvements would have been greater, especially for hot

EGR, due to the greater KLMBT spark timing advances and thus more optimized MFB50

that would be expected.

260

258

256
ISFCnet (g/kWh)

254

252

250

248

246
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-14 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings

4.3.2.6 PMEP

The PMEP at the KLMBT spark timings, as shown in Figure 4-15, decreased for all of the

EGR types across the EGR range, because as mentioned, the throttle valve had to be opened

more to provide the same amount of fresh air as in the non-EGR case, or in other words, the

same amount of oxygen for combustion.

It is proposed that the greatest reduction across the EGR range was achieved with the hot

EGR condition because the intake plenum temperature increased the most (Figure 4-18). This

138
caused the intake air to become less dense meaning that the throttle had to be more opened in

order to maintain a stoichiometric mixture than it did with the other EGR types, thus causing

the pumping losses to reduce more significantly. For hot and cooled EGR, the reduction was

greatest as the EGR was increased from 7% to its maximum, whereas the greatest reduction

for hot and cooled EGR after TWC occurred as the EGR was increased from 0 to 7%. It is

thought that the more significant increase in intake plenum temperature (Figure 4-18) for hot

EGR, as the EGR addition was increased from 7 to 14%, meant that the throttle had to be

opened more than when the EGR addition was increased from 0 to 7%, resulting in the

observed behaviour. The observed behaviour for the cooled EGR condition is within the

margin of error as indicated by the confidence intervals; therefore, it is not considered to be

significant.

Pumping work was greater for the cooled EGR after TWC EGR type than the hot EGR after

TWC EGR type. It is proposed that this is due to its lower intake plenum temperature (Figure

4-18). This meant that the intake air was denser so the throttle had to be opened less to allow

the same amount of fresh air into the combustion chamber as with the hot EGR after TWC

EGR type, thus increasing the pumping work in comparison. It must be noted that if the hot

and cooled EGR after TWC EGR types had used a real TWC rather than a simulated TWC,

their PMEP would have been greater across the EGR range due to the increased back-

pressure resulting from a real TWC increasing the pumping losses.

The PMEP reduction would likely have been less significant if intake manifold temperature

increases, resulting from the EGR addition, could have been eliminated. This is because the

intake air would have become denser, necessitating the throttle to be more closed to allow the

same amount of fresh air into the combustion chamber, thus increasing the pumping work.

139
 0.35

0.33

0.31
PMEP (bar)

0.29

0.27

0.25

0.23
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-15 PMEP versus EGR ratio at KLMBT spark timings

4.3.2.7 EGT

The EGT at the KLMBT spark timings decreased significantly for all of the EGR types

across the EGR range, as shown in Figure 4-16. Overall, it decreased by 40.4, 41.6, 46.9 and

52.1˚C for the EGR types of hot EGR, cooled EGR, hot EGR after TWC and cooled EGR

after TWC, respectively. This is because the KLMBT spark timing advances (Table 4-4)

resulted in the pressure and heat from combustion being more efficiently converted into

piston work, as discussed.

The temperatures reduced for hot EGR despite its MFB50 becoming later. It is proposed that

the earlier KLMBT spark timing (Table 4-4) resulted in the energy from the fuel being

released earlier. This resulted in energy being more efficiently converted from the

combustion in piston work, despite the slightly later MFB50 (Figure 4-12) and the longer

combustion duration (Figure 4-13) across the EGR range, resulting in reduced EGTs across

the EGR range.

140
At 10% EGR addition, the greatest reductions were observed for cooled EGR. It is thought

that this is because its MFB50 (Figure 4-12) was the most advanced at 10% EGR addition,

meaning that the pressure and heat from the combustion was more efficiently converted into

piston work. This caused the EGT to decrease more than with the other EGR types which had

a later MFB50 at 10% EGR addition. Cooled EGR after TWC then hot EGR after TWC had

the next greatest rate of reduction across their tested EGR ranges. Again, this can be

explained by the more advanced MFB50 for the cooled EGR after TWC EGR type as

compared to the hot EGR after TWC EGR type (Figure 4-12).

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated. This

would have likely resulted in greater EGTs decreases across the EGR range, as discussed

previously.

660

650

640
EGT (˚C)

630

620

610

600
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-16 EGT versus EGR ratio at KLMBT spark timings

141
4.3.2.8 COVIMEP

The COVIMEP at the KLMBT spark timings increased significantly across the EGR range for

all of the EGR types, as shown in Figure 4-17. It is proposed that this is because the

combustion speed was reduced across the EGR range (Figure 4-13) which meant that the

flame was more affected by the airflow in the combustion chamber, thus increasing the

COVIMEP observed. As can be seen from the confidence intervals indicated on the figures, the

differences between the different EGR types could have resulted from experimental error,

therefore, they will not be commented upon further.

As discussed, it is believed that the combustion duration would have been slightly increased

if the intake manifold temperature increases could have been eliminated. This would be

expected to increase the COVIMEP slightly because the flame would be more affected by the

airflow in the combustion chamber.

4.4

3.9
COVIMEP (%)

3.4

2.9

2.4

1.9
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-17 COVIMEP versus EGR ratio at KLMBT spark timings

142
4.3.2.9 Intake Plenum Temperature

The intake plenum temperature at the KLMBT spark timings, as shown in Figure 4-18,

increased significantly across the EGR range for all of the EGR types. This was expected for

the hot EGR and hot EGR after TWC EGR types but it was less expected for the cooled EGR

and cooled EGR after TWC EGR types. It is proposed that the cooled EGR gases were not as

cold as the engine fresh intake air, therefore causing the intake plenum temperatures to

increase. The intake temperatures are still lower for the cooled EGR types as compared to the

hot EGR types, as is to be expected. Any further relative differences between the different

EGR types are not considered to be significant as can be seen from the confidence intervals

displayed on the figure.

38

36
Intake Plenum Temperature (˚C)

34

32

30

28

26
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-18 Intake plenum temperature versus EGR ratio at KLMBT spark timings

143
4.3.2.10 Gaseous Emissions (NOx and HC)

The NOx emission at the KLMBT spark timings, as shown in Figure 4-19, decreased

significantly by 50.5, 31.3, 43.7 and 45.3% across the EGR range for the hot EGR, cooled

EGR, hot EGR after TWC and cooled EGR after TWC EGR types, respectively. It is

proposed that the reduced calculated average in-cylinder temperatures across the EGR range

(Figures 4-10a-d) resulted in a reduced oxidation rate of in-cylinder nitrogen gas in the hot

flames, resulting in the observed behaviour.

The reduction was greatest for hot EGR, similar for hot and cooled EGR after TWC, and it

was the least for cooled EGR. It is thought that this is because the MFB50 (Figure 4-12)

became later for hot EGR across the EGR range. The MFB50 for cooled and hot EGR after

TWC were similar across the EGR range therefore explaining why their NOx emission was

similar, while the MFB50 was the most advanced for the cooled EGR condition, resulting in

the highest NOx emission across the EGR range. With an advanced MFB50, the fuel-air

mixture burns at an earlier point in the engine cycle where the in-cylinder turbulence is

stronger, causing the combustion to proceed more quickly. It is thought that this resulted in

higher local temperatures which resulted in the differences in NOx emissions between the

different EGR types across the EGR range. While there were differences in the peaks of the

calculated average in-cylinder temperatures for the different EGR types (Figure 4-10a-d), it is

thought that the differences in MFB50 and their subsequent effects on local in-cylinder

temperatures were more significant regarding the NOx behaviour. This is because the average

temperature differences were relatively small (~40˚C at maximum EGR additions).

As discussed, it is not believed that the intake manifold temperature increases with EGR

addition had a significant effect on the peak calculated average in-cylinder temperatures.

144
Therefore, it is not thought that the NOx emission was significantly affected by the intake

manifold temperature increases.

8.5

8.0

7.5

7.0
NOx (g/kWh)

6.5

6.0

5.5

5.0

4.5

4.0

3.5
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-19 NOx emissions versus EGR ratio at KLMBT spark timings

The HC emission at the KLMBT spark timings significantly increased across the EGR range

for all of the EGR types tested, as shown in Figure 4-20. It was increased by 14.7, 11.0, 12.9

and 10.1% for the four respected EGR types of hot EGR, cooled EGR, hot EGR after TWC

and cooled EGR after TWC. It is thought that the in-cylinder temperature (Figures 4-10a-d)

and EGT decreases across the EGR range (Figure 4-16) reduced the oxidation of HCs for all

of the EGR types in the engine cylinder and exhaust, respectively, leading to the HC

increases across the EGR range. Also, the increased combustion duration (Figure 4-13)

reduced the time available for post-combustion oxidization of HCs for all of the EGR types,

contributing to their increases across the EGR range.

The increase was greatest for hot EGR, followed by hot EGR after TWC, then cooled EGR,

then cooled EGR after TWC. It is proposed that the later MFB50 for hot EGR and hot EGR

after TWC across the EGR range (Figure 4-12) reduced the time for post-combustion

145
oxidation, causing their HC emission to increase the most. Although the MFB50 for the

cooled EGR condition was more advanced than that of the hot and cooled EGR after TWC

conditions, its HC emission across its EGR range of 0 to 10% was similar to that of the hot

EGR after TWC condition and higher than that of the cooled EGR after TWC condition

within the same 0 to 10% EGR range. This is because it introduced additional HCs into the

combustion chamber while the hot and cooled EGR after TWC conditions did not, because

the simulated EGR gas was taken from the gas bottle. Therefore, the additional HCs reduced

the oxidation of the HCs in the combustion chamber producing the higher than expected HC

emission. With a real TWC, most but not all of the HCs would have been removed from the

EGR gas, resulting in a similar effect, although its significance would be reduced. It is

proposed that the longer combustion duration for the cooled EGR after TWC condition as

compared to the hot EGR after TWC condition (Figure 4-13) resulted in an improvement in

HC oxidization because of the slower flame, reducing the emission in comparison.

As discussed in the first part of this chapter, it is not believed that the intake manifold

temperature increases had a significant effect on the HC emissions observed. Again, despite

the increases in HCs observed across the EGR range for the four EGR types tested, it would

be expected that a traditional TWC would be able to handle this increase to maintain the

vehicle-out HC emissions at non-EGR levels.

146
 7.4

7.2

7.0
HC (g/kWh)

6.8

6.6

6.4

6.2
0 2 4 6 8 10 12 14 16
EGR (%)
Hot EGR Cooled EGR Hot EGR after TWC Cooled EGR after TWC

Figure 4-20 HC emissions versus EGR ratio at KLMBT spark timings

4.4 Data Continuity

For a detailed analysis regarding the continuity of the data between the first and second parts

of this chapter, please refer to Appendix B1.

147
4.5 Conclusions
The effect of hot EGR on the combustion and emissions of a single cylinder DISI research

engine was investigated in this chapter and the following conclusions have been made.

1. KLMBT spark timing was advanced significantly by 18, 8, and 4 CAD for the three

respective engine loads of 5.5, 7.0 and 8.5 bar IMEP, due to EGR suppressing the

temperature rises in the end-zone of the combustion chamber.

2. Fuel consumption was reduced by 2.2, 4.1, and 1.0% for the three respective engine loads

with EGR addition, because the pumping work was reduced for all engine loads and the

KLMBT could be advanced towards its optimum phasing for the engine loads of 7.0 and

8.5 bar IMEP.

3. EGR addition increased the accumulation mode particles and reduced the nucleation

mode particles for the loads of 5.5 bar IMEP (from 0 to 12% EGR addition) and 7.0 bar

IMEP, because of the combustion temperature suppression effect of the EGR, and

subsequent nuclei adsorption by the increased numbers of accumulation mode particles.

However, the nucleation mode particle reduction for 7.0 bar IMEP was not significant

and the accumulation mode particles reduced as the EGR was increased from 12 to 13%

for the load of 5.5 bar IMEP. In addition, there was not a significant change in nucleation

and accumulation mode particles for the load of 8.5 bar IMEP because of the low 3%

maximum level of EGR achieved.

These main findings from this section are also summarized in Table 4-5.

148
 Table 4-5 Results summary (EGR addition & engine load) (highlighted=improvement,
underlined=worsening)

Result KLMBT ISFCnet EGT PM NOx HC

Summary
Nucleation Accumulation
Engine Load (bar)
5.5 18 CAD 2.2% 21.9˚C Reduction Increase 64.3% 48.3%

7.0 8 CAD 4.1% 32.7˚C Reduction Increase 50.9% 19.0%

8.5 4 CAD 1.0% 7.6˚C No sig. change No sig. change 12.2% 6.8%

The effect of different EGR types of hot EGR, cooled EGR, HE after TWC and CE after

TWC on the combustion and gaseous emissions of a single cylinder DISI research engine was

also investigated in this chapter and the following conclusions have been made.

4. KLMBT spark timing was advanced significantly by 10, 10, 12 and 14 CAD,

respectively, for the four EGR types, due to EGR suppressing the temperature rises in the

combustion chamber end-zone; cooler EGR gases and the removal the NO from the EGR

gases accounted for the differences between the different EGR types.

5. Fuel consumption was reduced by 1.8, 3.8, 1.6 and 2.9% for the four respective EGR

types, because pumping losses were reduced for all EGR types and the MFB50 could be

advanced towards its optimum phasing for most EGR types.

6. The NOx emission was reduced significantly by 50.5, 31.3, 43.7 and 45.3% across the

EGR range for the four respective EGR types, due to the temperature suppression effect

of the EGR gases. MFB50 combustion phasing differences resulting from the different

KLMBT spark timing advances achieved accounted for the differences between the

different EGR types.

149
7. Overall, if fuel consumption is the priority than the cooled EGR type is the best EGR

type, while if emission reduction (NOx and HC) is the priority than cooled EGR after

TWC is the best EGR type.

The main findings from this section are also summarized in Table 4-6.

Table 4-6 Results summary (EGR addition & EGR type) (highlighted=improvement,
underlined=worsening)

Result KLMBT ISFCnet EGT NOx HC

Summary

EGR Type
Hot EGR 10 CAD 1.8% 40.4˚C 50.5% 14.7%

Cooled EGR 10 CAD 3.8% 41.6˚C 31.3% 11.0%

HE after TWC 12 CAD 1.6% 46.9˚C 43.7% 12.9%

CE after TWC 14 CAD 2.9% 52.1˚C 45.3% 10.1%

150
 Chapter 5

The Effect of Intake Airflow and Hot EGR on Engine Combustion and PM
Emissions

The aim of this chapter is to provide details of the combustion and PM emissions

characteristics of a DISI engine operated with swirl and tumble intake airflows, and with

EGR addition, at a single engine load of 7.0 bar IMEP.

5.1 Introduction
The main combustion parameters investigated in this chapter are KLMBT spark timing, in-

cylinder pressure, calculated average in-cylinder temperature, MFB, fuel consumption and

exhaust temperature. PM is the emission parameter investigated.

There is an ever growing demand for reduction of PM emissions from engines, as emission

regulations become stricter, particularly with the Euro VI regulation which has limited both

particulate mass and size for the first time in DISI engines. It is well known that swirl and

tumble have a significant effect on air-fuel mixture formation as well as on the combustion

and emissions of IC engines. However, what is not well known is the effect they have on the

formation of particulates in DISI engines, and what effect combining swirl and tumble with

EGR has on the combustion and PM emissions. EGR suppresses the temperature rise in the

combustion chamber, reducing NO x emissions and allowing the KLMBT spark timing to be

advanced, as well as enabling the throttle to be more opened, providing significant fuel

economy benefits. However, swirl and tumble affect the speed of the combustion, which

affects the laminar speed slowing effect of the EGR gases. Therefore, it is important to

understand the influence of swirl and tumble, combined with EGR, on combustion and PM

emissions. The Euro VI emissions regulation has increased interest in the role of EGR

151
addition on engine particulates, with the potential of EGR coming under question because of

it has been observed to have a negative effect on particulates (Gill et al., 2011, Ma et al.,

2014). Therefore, it is an important area to study.

Overall, the author recognizes an opportunity to extend the research of swirl and tumble in

DISI engines further by investigating the effect of swirl and tumble, and EGR addition on PM

emissions in this chapter. By doing this, it is hoped that the effect can be quantified in order

to inform the engine design process in terms of the selection of swirl and tumble ratios, and

the possible incorporation of a swirl/tumble valve into DISI engines to help achieve these at

specific engine speeds and loads. The author also recognizes that the Euro VI emissions

regulation has led many to question the potential of EGR, considering it is generally

considered to increase engine particulates. Thus, this is another area researched further in this

chapter.

5.2 Experimental Procedure

The experimental test procedure outlined in Section 3.9 of this thesis was followed in order to

conduct the engine tests in the single-cylinder engine to collect the data presented and

discussed in this chapter. Relative air-fuel ratio λ was maintained at 1 during the experiments

and a COVIMEP of 5% was not exceeded. Valve timings were set at IVO=16˚bTDC and

EVC=36˚aTDC. KLMBT spark timings and a fixed geometric compression ratio of 11.5

were used. The baffle plates as shown and discussed in Section 3.2.7 were used to create the

swirl and high tumble intake airflows in the engine, in addition to what is already produced

by the relevant engine geometry, such as the design of the intake runner. As well as these two

plates, another baffle plate was made with the same profile as the intake runner, so that a

reference low tumble condition could be tested with the same intake geometry as the swirl

152
and high tumble conditions. The estimated swirl/tumble ratios for the three baffle plates are

provided in Table 3-2. In order to achieve hot EGR in the engine, and in order to achieve the

desired engine load and EGR addition, the equipment setup and test procedure outlined in

Section 3.2.6.1 was followed.

The load of 7.0 bar IMEP was chosen to study because it represents a medium-high load in

this engine, and while it produces knock, the knock produced is of a lower magnitude than

that at higher engine loads, allowing swirl and tumble to be studied without causing any

significant engine damage. It also allowed a reasonable, while not a high EGR addition of

14% to be achieved. A higher engine load would have meant that the throttle would have

needed to be more opened, thus reducing the maximum level of EGR that could have been

achieved. One load was chosen to study so that an in-depth understanding could be formed,

which would not be possible if multiple loads were studied with the different intake airflows.

The spark was swept for each intake airflow condition and EGR ratio in order to find the

KLMBT spark timing using the MatLab script pressure trace analysis technique outlined in

Section 3.6.9. A research grade gasoline was used to obtain the results in this chapter with the

properties outlined in Section 3.8. The SMPS 3936 was used to obtain the PM emission

results in this chapter. Gaseous emissions were not measured in this chapter because it is

believed that the literature sufficiently covers this area.

The test matrix provided in Table 5-1 was carried out in order to investigate the intake

airflow and EGR effect on engine combustion and PM emissions at the single engine load of

7.0 bar IMEP.

153
 Table 5-1 Experiment test matrix (intake airflow & EGR addition)

EGR Addition (%) 0 4 7 8 10 11

Intake
Air-Flow
Low Tumble 1 2 3 4
Swirl 5 6 7 8
High Tumble 9 10 11

5.3 Results and Discussion

5.3.1 KLMBT Spark Timings

From the KLMBT spark timings, as shown in Table 5-2 for the different intake airflow

conditions, it can be seen that the spark was retarded for the swirl and high tumble conditions,

as compared to the low tumble condition; the high tumble condition being the most retarded.

Overall, the increased turbulence produced in the swirl and high tumble conditions (CID was

reduced by ~1 CAD with greater reductions observed with EGR addition) as compared to the

baseline low tumble condition meant that the spark timing had to be retarded in order for the

turbulence and thus combustion speed to decay in order to avoid knock. Although the

additional turbulence was allowed to dissipate somewhat by retarding the spark timing, it is

still believed to have improved mixture formation more significantly than spark retardation

alone would have done, which will be explored further in the emissions section.

From the MFB data (Figure 5-3a) it can be seen that the combustion duration was longer for

the swirl condition than it was for the low tumble condition (MFB was ~24.0 CAD for the

swirl condition and it was ~22.5 CAD for the low tumble condition, at 0% EGR addition).

Therefore, it is thought that this provided more time for the temperature to rise in the end-

zone of the combustion chamber, producing auto-ignition, resulting in the retarded KLMBT

154
spark timing for the swirl condition. It must be noted that knocking could have been avoided

if the compression ratio was fully optimized for the fuel used, as discussed previously.

For the high tumble condition, it is proposed that the velocity of the intake air was increased

significantly, causing the fuel-air mixture to burn much more quickly (MFB10-90 was ~21.0

CAD, compared to ~22.5 and ~24.0 CAD for the low tumble and swirl conditions,

respectively, at 0% EGR addition). It is believed that this increased the amount of heat

radiated to the fuel-air mixture in the end-zone of the combustion chamber which caused

knock to be initiated at an earlier spark timing than for the other two intake airflow

conditions, necessitating the KLMBT spark timing to be retarded in comparison. It is

proposed that this effect was more significant than the faster flame speed, which does not

appear to have been fast enough to consume the auto-ignition site before it auto-ignited. This

evident relationship between flame speed and knock tendency should be investigated further

to find the optimum flame speed for minimum knock.

The end-zone temperatures were reduced (allowing the calculated average in-cylinder

temperatures to be maintained or their reductions minimized (Figures 5-2b, 5-2c and 5-2d))

by the combustion slowing effects of the EGR addition (Rhodes and Keck, 1985). This

allowed the spark timing to be advanced to an earlier point in the engine cycle where the in-

cylinder turbulence was stronger, cancelling out the effect EGR had on the laminar flame

speed, preventing engine knock that otherwise would have occurred.

At the high tumble condition with 0% EGR addition, the combustion was faster than with the

other two intake airflow conditions, as mentioned. This meant that good KLMBT spark

timing improvements could be made with EGR addition. Although EGR addition will have

allowed the KLMBT spark timing to be advanced at the low tumble and swirl conditions too,

155
it is proposed that the effect was greater with the high tumble condition. This is because its

KLMBT spark timings started further away from the optimal condition which meant that

greater (or equal) improvements in the KLMBT spark timing could be made at equivalent

EGR additions.

KLMBT spark timing advances would have been greater if the intake manifold temperature

increases resulting from the EGR addition could have been eliminated, as discussed

previously.

Table 5-2 KLMBT spark timings (˚bTDC) (intake airflow & EGR addition)

EGR Addition (%) 0 4 7 8 11

Intake
Air-Flow
Low Tumble 16 20 24 26
Swirl 8 10 12 14
High Tumble -2 2 6

5.3.2 In-cylinder Pressure and Temperature

The in-cylinder pressure data at 0% EGR addition presented in Figure 5-1a shows that the in-

cylinder pressure significantly decreased with the swirl and high tumble conditions, as

compared to the low tumble condition. This was due to the significant spark retard required

(Table 5-2), causing more energy to be released from the fuel during the expansion stroke,

once the in-cylinder pressure was already reducing. Therefore, this energy did not contribute

to the in-cylinder pressure increases, resulting in the in-cylinder pressure decreases observed.

The in-cylinder pressures across the EGR ranges, as presented in Figures 5-1b, 5-1c and 5-1d,

for the low tumble, swirl and high tumble conditions, respectively, generally increased across

the EGR range. The reason for the increases observed are the advances in the KLMBT spark

156
timing achieved with the EGR addition, releasing more of the energy from the fuel at an

earlier point in the engine cycle where it could contribute to the in-cylinder pressure

increases. However, for the maximum EGR addition of 11% at the low tumble and swirl

conditions, the in-cylinder pressure reduced. This is because the KLMBT spark timing

advance achieved with EGR addition was not sufficient at these conditions to overcome the

combusiton slowing effects of the additional EGR gases (Rhodes and Keck, 1985).

In-cylinder pressure increases with EGR addition will likely have been greater if the intake

manifold temperature increases could have been eliminated, due to the greater KLMBT spark

timing advances that would be expected.

Fig. 5-1a: In-cylinder Pressure (three airflow conditions)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
Low Tumble Swirl High Tumble

157
 Fig. 5-1b: In-cylinder Pressure (low tumble)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100

0% EGR 4% EGR 8% EGR 11% EGR

Fig. 5-1c: In-cylinder Pressure (swirl)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR 11% EGR

158
 Fig. 5-1d: In-cylinder Pressure (high tumble)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100

0% EGR 4% EGR 7% EGR

Figure 5-1 In-cylinder pressure verses CAD at KLMBT spark timings for a) the three
intake airflow conditions at 0% EGR, b) low tumble, c) swirl, d) high tumble

The calculated average in-cylinder temperature profile at 0% EGR, as presented in Figure 5-

2a, was approximately the same for the swirl condition as it was for the low tumble condition,

although it was slightly retarded in comparison. The KLMBT spark timing had to be retarded

for the swirl condition as compared to the low tumble condition (Table 5-2), resulting in the

observed behaviour. For the high tumble condition, the calculated average in-cylinder

temperature significantly increased compared to the other two intake airflow conditions, due

to the faster combustion, as discussed previously. This significantly increased the rate that

heat was released from the fuel, resulting in the increased calculated average in-cylinder

temperature. The calculated average in-cylinder temperature profile for the high tumble

condition was also retarded in comparison, because of its KLMBT spark timing retard (Table

5-2).

There was no significant change in the magnitude of the calculated average in-cylinder

temperatures across the EGR range for the low and high tumble conditions as shown in

159
Figures 5-2b and 5-2d. As discussed, the EGR addition allowed the KLMBT spark timing to

be advanced through suppression of end-zone temperatures, which appears to have overcome

the combustion temperature reducing effects of the EGR gases. They have a higher heat

capacity than the fresh air they replaced in the engine cylinder thus without advancing the

spark timing, the calculated average in-cylinder temperatures would have decreased.

However, the spark advance allowed them to be maintained across the EGR range by

releasing the heat earlier from the fuel. At 4 and 8% EGR addition for the low tumble intake

airflow it appears that the KLMBT spark timing advance was sufficient to increase the

calculated average in-cylinder temperatures causing the increases observed at these

conditions. For the swirl condition, as shown in Figure 5-2c, the KLMBT spark timing

advances were not sufficient to maintain the calculated average in-cylinder temperatures, thus

resulting in the decreases observed across the EGR range.

It is believed that the increased intake manifold temperatures resulting from the EGR addition

were compensated for by the less optimized combustion phasing resulting from the

consequently reduced KLMBT spark timing advance (Table 5-2), compared to the maximum

spark advance possible, as discussed in the first part of Chapter 4. Thus, it is thought the two

effects of increased intake manifold temperature and reduced subsequent KLMBT spark

timing advances cancelled one another out, meaning there was no overall effect on the peak

of the calculated average in-cylinder temperatures.

160
 Fig. 5-2a: In-cylinder Temperature (three airflow conditions)
2100

Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

Low Tumble Swirl High Tumble

Fig. 5-2b: In-cylinder Temperature (low tumble)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

0% EGR 4% EGR 8% EGR 11% EGR

161
 Fig. 5-2c: In-cylinder Temperature (swirl)
2100

Calculated In-Cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

0% EGR 4% EGR 8% EGR 11% EGR

Fig. 5-2d: In-cylinder Temperature (high tumble)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

0% EGR 4% EGR 7% EGR

Figure 5-2 Calculated average in-cylinder temperature verses CAD at KLMBT spark
timings for a) the three intake airflow conditions at 0% EGR, b) low tumble,
c) swirl, d) high tumble

162
5.3.3 MFB

The low tumble and swirl intake airflow conditions had similar MFB profiles, as shown in

Figure 5-3a at 0% EGR addition, despite the retarded spark timing for the swirl condition as

compared to the low tumble condition. The swirl condition increased the speed of the initial

flame propagation compared to the low tumble condition by approximately 1 CAD, causing

the MFB50s at 0% EGR to be very similar. (Nagayama et al., 1977) also observed a similar

effect. However, the later stage of combustion appears to have been longer for the swirl

condition. This is most likely due to the breakdown of the swirl flow into weak general in-

cylinder turbulence. The high tumble condition in comparison had a significantly retarded

MFB profile compared to the low tumble and swirl conditions, despite its initial flame

propagation also being quicker. This was due to its retarded KLMBT spark timing (Table 5-

2). Unlike the swirl condition, it does not appear to suffer from the longer late stage of

combustion.

The MFB data, presented in Figures 5-3b, 5-3c and 5-3d for the three respected intake airflow

conditions of low tumble, swirl and high tumble, shows that the MFB profile generally

became advanced with EGR addition. This is because the KLMBT spark timing could be

advanced (Table 5-2), as explained in the first part of Chapter 4. It can be seen in Figure 5-3b

for the low tumble condition that 11% EGR addition resulted in a significantly increased

combustion duration. Therefore, it appears that the KLMBT spark advance achieved was not

sufficient to overcome the combustion slowing effects of the additional EGR, at this

condition. A similar effect occurred with the swirl intake airflow condition, but rather than

increasing the overall combustion duration, it produced a significantly slower earlier stage of

combustion than observed at lower EGR additions and a similar overall combustion duration.

163
Therefore, the additional EGR appears to have had a significant effect on this early stage of

combustion, while not affecting the overall combustion duration.

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated. Thus,

the MFB profiles would likely have been advanced slightly in comparison to those shown in

Fig. 5-3b-d. In addition, it is expected that the flame would spread slightly more slowly in the

cooler charge, thus, the MFB profile would become slightly more elongated. If the

compression ratio was optimised for the fuel used, the combustion phasing would have been

optimised for each test point, therefore, it would be expected that the combustion of gasoline

would have been quicker due to higher in-cylinder turbulence at the time of ignition.

Fig. 5-3a: MFB (three airflow conditions)

1

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0
-50 0 CAD 50 100
Low Tumble Swirl High Tumble

164
 Fig. 5-3b: MFB (low tumble)
1

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0
-50 0 CAD 50 100

0% EGR 4% EGR 8% EGR 11% EGR

Fig. 5-3c: MFB (swirl)

1

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0
-50 0 CAD 50 100
0% EGR 4% EGR 8% EGR 11% EGR

165
 Fig. 5-3d: MFB (high tumble)
1

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0
-50 0 CAD 50 100
0% EGR 4% EGR 7% EGR

Figure 5-3 MFB verses CAD at KLMBT spark timings for a) the three intake airflow
conditions at 0% EGR, b) low tumble, c) swirl, d) high tumble

5.3.4 Fuel Consumption

Gravimetric ISFCnet increased significantly for the swirl and high tumble conditions, as

shown in Figure 5-4; it was increased by 6.4 and 15.5%, respectively, as compared to the low

tumble condition. This is because the combustion duration was increased for the swirl

condition (Figure 5-3a), and the KLMBT spark timing retard caused the MFB50 (Figure 5-

3a) to be retarded from its optimum 8-10˚aTDC phasing for the high tumble condition (de O.

Carvalho et al., 2012). Thus, it is thought that these resulted in a reduced conversion

efficiency of the pressure and heat from the combustion into piston work, resulting in the fuel

consumption increases. The MFB50 retard for the high tumble condition was more

significant than the combustion duration increase for the swirl condition, explaining why its

fuel consumption was higher in comparison. Although the baffle plates for the swirl and high

tumble conditions did partially block the intake runner which alone would have increased

fuel consumption through increased pumping losses, the greater throttle angle required to

166
burn the additional injected fuel required for the swirl and high tumble conditions effectively

compensated for this; pumping work was only increased by 0.01 bar for the swirl condition

and it was decreased by 0.06 bar for the high tumble condition.

The fuel consumption generally decreased across the EGR range by 2.4, 3.9 and 10.2% for

the three engine conditions of low tumble, swirl and high tumble respectively. This is because

the KLMBT spark timings could be advanced towards their optimum phasing with the EGR

addition (de O. Carvalho et al., 2012). PMEP reductions (from 0.36 to 0.28 bar for the low

tumble condition, 0.37 to 0.30 bar for the swirl condition, and 0.29 to 0.27 bar for the high

tumble condition), and the reduced combustion temperatures which improved thermal

efficiency (Ratnak et al., 2015, Siokos et al., 2015) also contributed to the fuel consumption

improvements. All of these were explained previously in the first part of Chapter 4. The VAF

had an average error of 1.6%, as discussed in the previous chapter; therefore, the fuel

consumption decreases observed are not limited.

The fuel consumption increased slightly with 11% EGR addition at the low tumble condition

because the rate of KLMBT spark timing advance was reduced, when compared with the

advances achieved with other increases in EGR addition. The KLMBT spark timing was only

advanced by 2 CAD compared to 4 CAD with previous EGR additions, so the MFB50

became retarded as the KLMBT spark advance achieved was not sufficient to overcome the

laminar flame speed slowing effects of the additional EGR (Rhodes and Keck, 2012). This

caused its MFB50 to become retarded away from the optimum 8-10˚aTDC phasing (Figure 5-

3b), increasing its fuel consumption.

167
Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated; thus, the

fuel consumption improvements would likely have been greater.

300

290

280
ISFCnet (g/kWh)

270

260

250

240
0 2 4 6 8 10 12
EGR (%)
Low Tumble Swirl High Tumble

Figure 5-4 Gravimetric ISFCnet versus EGR ratio at KLMBT spark timings

5.3.5 EGT

From Figure 5-5 it can be seen that the EGT increased significantly by 16.0 and 68.2˚C at 0%

EGR addition, respectively, for the swirl and high tumble conditions, as compared to the low

tumble condition. It is proposed that the longer combustion duration for the swirl condition at

0% EGR addition (~24.0 CAD) as compared to the low tumble condition (~22.5 CAD),

resulted in the pressure and heat from the combustion being less efficiently converted into

piston work. This caused the gas temperature to become higher in the expansion and exhaust

strokes, resulting in the EGT increase observed. It is proposed that the significantly later

MFB50 for the high tumble condition at 0% EGR resulted in the same effect, producing the

observed behaviour. The greater MFB50 retardation for the high tumble condition as

168
compared to the other two intake airflow conditions (Figure 5-3a) was more significant than

the increased combustion duration for the swirl condition as compared to the low tumble

condition. Therefore, this resulted in the more significant increase in EGT.

As to be expected, the EGT decreased with EGR addition for all of the intake airflow

conditions, because the MFB50 was more advanced (Figures 5-3b, 5-3c and 5-3d), as

explained in the first part of Chapter 4.

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated. This

would likely have resulted in greater improvements in EGTs across the EGR range, as

explained in the first part of Chapter 4.

740

720

700
EGT (°C)

680

660

640

620

600
0 2 4 6 8 10 12
EGR (%)
Low Tumble Swirl High Tumble

Figure 5-5 EGT versus EGR ratio at KLMBT spark timings

169
5.3.6 PM Emissions

5.3.6.1 Intake Airflow Effect on PN Emission

The nucleation mode particles, identified as the smaller diameter particles forming the peak

on the left-hand side of the plot (10-30 nm) shown in Figure 5-6a, decreased significantly

with the swirl and high tumble conditions, as compared to the low tumble condition. The soot

accumulation particles, identified as the larger diameter particles forming the peak on the

right-hand side of the plot (30-300 nm), decreased too for the swirl condition, while they

remained approximately the same for the high tumble condition, as compared to the low

tumble condition. These significant changes have been confirmed by analysing the 95%

confidence intervals, which could not be displayed in the figures because the figures

themselves would have become unclear.

It is proposed that the significant reduction in nucleation and accumulation particles observed

with the swirl condition was a result of not just the improved air-fuel mixture preparation due

to the KLMBT spark timing retard required (Table 5-2), but more importantly the increased

turbulence resulting from the swirl airflow. This provided more time for mixture preparation

as compared to the low tumble condition and it increased the rate at which the fuel and air

mixed, respectively. It is thought that this reduced the number of pockets of a high local

equivalence ratio meaning that the particulates oxidized more fully in the combustion,

resulting in the behaviour observed. The calculated average in-cylinder temperature profiles

for the swirl condition as compared to the low tumble condition (Figure 5-2a), at 0% EGR

addition, were similar with similar peaks; therefore, combustion temperature is not thought to

have caused the observed behaviour.

170
It is proposed that the significantly higher calculated average in-cylinder temperature for the

high tumble condition at 0% EGR addition (Figure 5-2a) resulted in the substantial reduction

of nucleation mode particles observed, along with the increased time available for air-fuel

mixture preparation resulting from the spark retard and the improved mixture formation

resulting from the high tumble airflow. However, the later MFB combustion phasing (Figure

5-3a) is thought to have reduced the time available for post-combustion oxidization, resulting

in similar numbers of accumulation mode particles as compared to the low tumble condition.

Overall, swirl had the greatest benefit regarding PN emissions, which means it has potential

to be used to reduce PN emissions to ensure compliance with the Euro VI emission

regulations.

Fig. 5-6a: PM (three airflow conditions)

1.E+06
dN/dlogDp (#/cm3)

1.E+05

1.E+04

1.E+03
10 100 1000
Particle Diameter (nm)
Low Tumble Swirl High Tumble

171
 Fig. 5-6b: PM (low tumble)
1.E+06

dN/dlogDp (#/cm3)
1.E+05

1.E+04

1.E+03
10 100 1000
Particle Diameter (nm)
0% EGR 4% EGR 8% EGR 11% EGR

Fig. 5-6c: PM (swirl)

1.E+06
dN/dlogDp (#/cm3)

1.E+05

1.E+04

1.E+03
10 100 1000
Particle Diameter (nm)
0% EGR 4% EGR 8% EGR 11% EGR

172
 Fig. 5-6d: PM (high tumble)
1.E+06

dN/dlogDp (#/cm3)
1.E+05

1.E+04

1.E+03
10 100 1000
Particle Diameter (nm)
0% EGR 4% EGR 7% EGR

Figure 5-6 PN emissions versus particle diameter at KLMBT spark timings for a) the
three intake airflow conditions at 0% EGR, b) low tumble, c) swirl, d) high
tumble

5.3.6.2 EGR Effect on PN Emission

The PN emissions for the low tumble, swirl and high tumble intake airflow conditions across

the EGR range are shown in Figures 5-6b, 5-6c and 5-6d, respectively. For the low tumble

condition, there was a significant decrease of nucleation and accumulation mode particles

across the EGR range. The significance of this and other PN changes discussed have been

confirmed by analysing the 95% confidence intervals, which could not be displayed in the

figures because the figures themselves would have become unclear. It is proposed that the

increased post-combustion oxidization time resulted in the observed behaviour. It is thought

that this was more significant than the reduced time available for fuel-air mixing and thus

poorer mixture preparation with the KLMBT spark advances. The peak calculated average in-

cylinder temperature across the EGR range when comparing 0 and 11% EGR addition

(Figure 5-2b) remained approximately the same so it is not thought that this had an effect.

173
For the swirl condition, the PN did not change significantly across the EGR range, although

the accumulation mode particles did significantly increase as the EGR addition was increased

from 4 to 8%. It is proposed that the reduced calculated average in-cylinder temperatures

across the EGR range (Figure 5-2c) and the poorer air-fuel mixture preparation due to the

KLMBT spark timing advances (Table 5-2) across the range cancelled out the effect of the

increased post-combustion oxidization time and the effect of decreased primary carbon

particle formation with the reduced in-cylinder temperatures (Zhang et al., 2014b). This

caused the PN to remain approximately the same across the range, when comparing those

observed at 0 and 11% EGR addition.

The PM for the high tumble intake airflow condition did not change significantly across the

EGR range. Despite the KLMBT spark timing being advanced as the EGR addition was

increased (Table 5-2), providing less time for fuel-air mixing, it is proposed that the improved

post-combustion oxidization time was sufficient to maintain the particles at similar levels

with the EGR addition.

For the swirl condition, as mentioned, the accumulation mode particles increased

significantly as the EGR addition was increased from 4 to 8%. It is proposed that this was

caused by the increased calculated average in-cylinder temperatures (Figure 5-2c) which

increased primary carbon particle formation by the thermal pyrolysis and dehydrogenation

reaction of fuel vapour/droplets (Zhang et al., 2014b). As the EGR addition was further

increased to 11%, the calculated average in-cylinder temperature decreased (Figure 5-2c),

which resulted in the PM reducing due to the decreased primary particle formation.

If the intake manifold temperature increases with EGR addition could have been eliminated,

then it is expected that the PM emissions would have been reduced for the low tumble

174
condition due to an increased post-combustion oxidization time resulting from the expected

KLMBT spark timing improvements (Daniel et al., 2012e). However, the reduced fuel-air

mixing time could cancel out some of these improvements. This behaviour would also be

expected with the swirl and high tumble conditions, because similar behaviour was observed

in the experimental data when the KLMBT spark timing was advanced.

Overall, these results agree with the research of (Alger et al., 2010) and (Hedge et al., 2011)

who observed PN reductions with EGR addition, although it must be noted that their results

were both from a turbocharged engine; the varying enrichment of the fuel-air charge will

have significantly affected their results.

5.4 Data Continuity

For a detailed analysis regarding the continuity of the data between this chapter and Chapter

4, please refer to Appendix B2.

5.5 Conclusions
The effect of swirl and tumble intake airflows and hot EGR on the combustion and PM

emissions of the single-cylinder DISI research engine was investigated in this chapter and the

following conclusions have been made.

1. Swirl and high tumble significantly increased the fuel consumption with respect to the

low tumble condition by 6.4 and 15.5% respectively, but EGR addition minimized these

increases. The fuel consumption increase was 4.8 and 6.2%, respectively, at the

maximum EGR additions for swirl and high tumble conditions, compared to that

observed at the maximum EGR addition for the low tumble condition.

2. At 0% EGR addition, swirl significantly reduced the nucleation and accumulation mode

particles, with the high tumble intake airflow condition only significantly reducing the

175
 nucleation mode particles. It is thought that the KLMBT spark timing retard required with

the swirl and high tumble intake airflow conditions increased the mixture preparation

time, resulting in the observed behaviour. Also, the additional turbulence produced by the

airflows improved mixture formation, reducing the number of pockets of a high local

equivalence ratio, resulting in the reduced particle numbers. A reduced post-combustion

oxidization time prevented the high tumble condition from reducing the accumulation

mode particles.

3. The PN emission was decreased significantly with EGR addition for the low tumble

condition, due to an increased post-combustion oxidization time, while there was no

significant change for the swirl and high tumble conditions. This is mainly because poorer

fuel-air mixture preparation caused by KLMBT spark timing advances cancelled out the

benefit produced by the increased post-combustion oxidization time.

The main findings from this chapter are also summarized in Table 5-3.

Table 5-3 Results summary (intake airflow & EGR addition) (highlighted=improvement,
underlined=worsening)

Result KLMBT ISFCnet EGT PM

Summary
Nucleation Accumulation
Intake
Airflow/EGR
Condition
Swirl (compared to 8 CAD 6.4% 16.0˚C Reduction Reduction
reference condition at 0%
EGR addition)

High tumble (compared 18 CAD 15.5% 68.2˚C Reduction No sig. change

to reference condition at
0% EGR addition)

Low tumble [max. EGR] 10 CAD 2.4% 41.1˚C Reduction Reduction

(compared to 0% EGR)

Swirl [max. EGR] 6 CAD 3.9% 23.7˚C No sig. change No sig. change
(compared to 0% EGR)

High tumble [max. EGR] 8 CAD 10.2% 71.8˚C No sig. change No sig. change
(compared to 0% EGR)

176
 Chapter 6

The Effect of Compression Ratio, Fuel and EGR on Engine Combustion

and Emissions

The aim of this chapter is to provide details of the combustion and emissions characteristics

of a DISI engine operated at different compression ratios with different fuels (8.5 bar IMEP),

and at different compression ratios with EGR addition (7.0 bar IMEP).

6.1 Introduction

The main combustion parameters investigated in this chapter are KLMBT spark timing, in-

cylinder pressure, calculated average in-cylinder temperature, MFB, fuel consumption and

exhaust temperature. PM, NOx and HC are the main emission parameters investigated.

There is an ever growing demand for reduction of PM, NO x and HC from engines as

emissions regulations become stricter, as well as the growing demand to reduce net CO2

emissions from the transportation sector, which is at the forefront of public perception. One

way to achieve these demands is to increase compression ratio; this improves thermal

efficiency causing fuel consumption and thus CO 2 output to reduce. Another way is to use

renewable oxygenated fuels in engines to increase the oxidation of particulates in the

combustion chamber to reduce their numbers, as well as potentially reducing NO x and HC

emissions. In addition, biomass is converted to make the renewable oxygenated fuel, which

reduces transportation’s net CO2 output (Wang et al., 2013). Furthermore, the Euro VI

emissions regulations which for the first time limit the PN from DISI engines have increased

interest in the effect of oxygenated fuels on engine particulates. They have the potential to

significantly reduce particulate emissions having health benefits, particularly for people

177
living in urban areas (Anderson et al., 2012b, United States Environmental Protection

Agency, 2014). The regulations have also increased interest in the effect of EGR addition on

engine particulates, with the potential of EGR coming under question because of it has been

observed to have a negative effect on them (Gill et al., 2011, Ma et al., 2014).

Overall, despite the amount of research that has been conducted into 1-butanol-gasoline and

ethanol-gasoline blended fuels, there appears to be lack of agreement in terms of the effect

these fuel blends on the combustion and emissions of gasoline engines. In addition, little

work has been conducted regarding the effect of these fuel blends on the combustion and

emissions of DISI engines, especially regarding the effect of 1-butanol-gasoline blends on

PM emissions. This is because the majority of the research conducted has been on PFI

engines with ethanol-gasoline fuel blends. Finally, 1-butanol-gasoline and ethanol-gasoline

fuel blends have not been studied well with each other along with a reference of gasoline fuel

at different compression ratios. Therefore, the research in this chapter has been conducted to

provide deeper knowledge about the effect of 1-butanol-gasoline and ethanol-gasoline

blended fuels on the combustion and emissions of DISI engines.

In addition, despite the amount of work that has been conducted into compression ratio and

EGR separately, little work has been conducted studying their combined effect on engine

combustion and PM emissions. Especially, it is important to know the combined effect of

these two fuel consumption (and NOx in the case of EGR) improving techniques on engine

particulates because of the limitations that the Euro VI emissions regulation has on these.

Therefore, these areas have also been investigated further in this chapter.

178
6.2 Experimental Procedure

The experimental test procedure outlined in Section 3.9 of this thesis was followed in order to

conduct the engine tests on the single-cylinder engine to collect the data presented and

discussed in this chapter. Relative air-fuel ratio λ was maintained at 1 during the experiments

and a COVIMEP of 5% was not exceeded. Valve timings were set at IVO=16˚bTDC and

EVC=36˚aTDC. KLMBT spark timings were used and the geometric compression ratio was

varied between 10.7 and 11.5 in the experiments. To change the compression ratio in the

engine the procedure outlined in Section 3.2.8 was followed. When using the non-gasoline

fuel splash blends the procedure discussed in Section 3.2.4.1 was followed to prevent

contamination between the fuels. In order to achieve hot EGR, and in order to achieve the

desired engine load and EGR addition, the equipment setup and experimental test procedure

outlined in Section 3.2.6.1 was followed.

The load of 8.5 bar IMEP was chosen to study for the investigation of the compression ratio

and fuel effect on engine combustion and emissions because it represents one of the worst

conditions for knock in this engine, as well as being an engine load that is highly relevant for

both NA and turbocharged DISI engines, increasing the usefulness of the data produced. The

load of 7.0 bar IMEP was chosen to study for the investigation of the compression ratio and

EGR effect on engine combustion and PM emissions because it represents a medium-high

load in this engine where knock is present, and it allowed a reasonable, while not a high EGR

addition of 14% to be achieved. A higher engine load would mean that the throttle would

have needed to be more opened thus reducing the maximum level of EGR that could have

been achieved. One load was chosen to study for both parts of the investigation so that an in-

depth understanding could be formed; this would not be possible if multiple loads were

studied.

179
The spark was swept for all compression ratios, fuels and EGR additions in order to find the

KLMBT spark timing using the LabView on-line pressure trace analysis technique outlined

in Section 3.6.9. This allowed the KLMBT spark timing to be found much more quickly than

the MatLab post-experiment processing method. This is because it enabled the KLMBT spark

timings to be found while the engine was operating, thus allowing the measurements at the

KLMBT spark timings to be recorded alone, saving significant amounts of time. The SMPS

3936 was used to measure the PM emission for both parts of this chapter and the Horiba

MEXA-7100DEGR was used to measure the NOx and HC gaseous emissions for the first part

of this chapter. Gaseous emissions were not measured in the second part of this chapter

because it is believed that the literature sufficiently covers this area, and the EGT data was

omitted for the same reason.

A research grade gasoline was used in this research. This was used in addition to 1-butanol

and ethanol for the results in the first part of this chapter. All of the fuel properties as well as

those of the two tested fuel blends of Bu20 and E20 are outlined in Section 3.8. The ULG95

was used in its supplied form, while the 1-butanol and ethanol fuels were mixed with the

ULG95 fuel to form the Bu20 and E20 fuel splash blends with each containing 20%vol 1-

butanol and 20%vol ethanol respectively. The ULG95 fuel was supplied with 5%vol ethanol

pre-mixed in it, so the 20%vol 1-butanol blend and ULG95 fuel also had 5%vol ethanol in

them too, while the E20 blend had no ethanol in addition to the 20%vol. In total the Bu20

blend had 20%vol 1-butanol with 5%vol ethanol, the E20 blend had 20%vol ethanol and the

ULG95 fuel had 5%vol ethanol.

The test matrix provided in Table 6-1 was carried out in order to investigate the compression

ratio and fuel effect on engine combustion and emissions at the single engine load of 8.5 bar

IMEP. The results and discussion of these tests can also be found in (Lattimore, 2016b).

180
 Table 6-1 Experiment test matrix (compression ratio & fuel)

Compression 10.7 10.9 11.2 11.5

Ratio

Fuel
Bu20 1 2 3 4
E20 5 6 7 8
ULG95 9 10 11 12

For the investigation of the compression ratio and EGR effect on engine combustion and PM

emissions at the single engine load of 7.0 bar IMEP, the test matrix provided in Table 6-2

was carried out.

Table 6-2 Experiment test matrix (compression ratio & EGR addition)

Compression 10.7 10.9 11.2 11.5

Ratio

EGR Addition (%)

0 1 2 3 4
7 5 6 7 8
14 9 10 11 12

6.3 Results and Discussion

6.3.1 Investigation of Compression Ratio and Fuel Effect on Combustion

and Emissions

6.3.1.1 KLMBT Spark Timing

From the KLMBT spark timings shown in Table 6-3, it can be seen that in the case of

gasoline, an increase in the compression ratio had no significant effect on KLMBT. The same

trend was also observed for the butanol blend (similar octane rating as gasoline) and even for

the ethanol blend, despite the high octane rating of ethanol. This is because at the engine load

of 8.5 bar IMEP, the engine was very prone to knock, even in the case of alcohols, due to the

high low temperature reactivity of alcohols (He et al., 2015) and the higher amount of fuel

181
being injected into the combustion chamber (i.e. ethanol has a lower calorific value than

butanol and gasoline). Thus, despite the compression ratio changing, no change in the

KLMBT spark timing could be realized. It must be noted that knocking could have been

avoided if the compression ratio was fully optimized for the fuels used. However, this was

not pursued in order to investigate the effect of the parameter changes on engine knock limit

and the consequent combustion and emission parameters at this limit.

It can also be seen that more advanced KLMBT spark timings could be achieved with Bu20

and E20 as compared to ULG95, with the most advanced spark timings being achieved with

Bu20. This is due to their higher octane numbers and the superior charge cooling effect of

alcohols compared to gasoline. Despite ethanol having a higher octane number than 1-butanol

and an increased cooling effect (in terms of mass), more advanced KLMBT spark timings

could be achieved with Bu20. It is believed that the 5%vol ethanol content in the Bu20 blend

(20%vol 1-butanol with 5%vol ethanol) was sufficient to compensate for the reduced charge

cooling effect and for the lower octane number of 1-butanol as compared to ethanol. It is also

thought that the higher chemical reactivity (He et al., 2015), faster laminar flame speeds

(Figure 6-5) and shorter fuel injection duration (less fuel quantity is required for the same

engine output power due to the higher heating value compared to ethanol) for the 1-butanol

blend with respect to the ethanol blend meant that the end-zone auto-ignition sites were

consumed before they had an opportunity to auto-ignite. Thus, these factors contributed to the

KLMBT spark timing advances.

Overall, it is believed that E20 would perform better over a larger compression ratio range

than Bu20; particularly at compression ratios higher than 11.5, due to its higher octane

number and increased cooling effect. Thus, the small compression ratio range is a limitation

of this experiment.

182
 Table 6-3 KLMBT spark timings (˚bTDC) (compression ratio & fuel)

Compression
Ratio 10.7 10.9 11.2 11.5

Fuel
Bu20 14 14 14 14
E20 12 12 12 12
ULG95 10 10 10 10

6.3.1.2 In-cylinder Pressures and Temperatures

The in-cylinder pressure traces for the two fuels blends of Bu20 and E20 along with that for

the ULG95 reference fuel are shown in Figures 6-1a, 6-1b and 6-1c, respectively. It is clear

that as the compression ratio was increased, the maximum in-cylinder pressure increased, for

the two fuel blends and the reference fuel tested. This is because the more compact

combustion chamber achieved through the compression ratio increase, reduced the heat losses

to the surroundings, resulting in the in-cylinder pressure increases. The in-cylinder pressures

were highest for Bu20, followed by E20, then ULG95. This is due to the more advanced

KLMBT spark timings (Table 6-3) which could be achieved with Bu20 and E20 as compared

to those achieved with ULG95, with the most advanced spark timings being achieved for

Bu20. This resulted in heat being released from the fuel at an earlier CAD, enabling higher

in-cylinder pressures to be achieved before the downwards piston motion caused the in-

cylinder pressure to decrease again.

183
 Fig. 6-1a: In-cylinder Pressure (Bu20)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-1b: In-cylinder Pressure (E20)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

184
 Fig. 6-1c: In-cylinder Pressure (ULG95)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100

CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-1 In-cylinder pressure versus CAD for a) Bu20, b) E20 and c) ULG95 at
KLMBT spark timings

Figures 6-2a, 6-2b and 6-2c show the calculated average in-cylinder temperatures for the two

fuel blends of Bu20 and E20 and for the reference fuel of ULG95, respectively. Overall, the

calculated average in-cylinder temperature increased as the compression ratio was increased.

This is because of the aforementioned more compact combustion chamber achieved with the

compression ratio increase reducing heat losses to the surroundings. The calculated average

in-cylinder temperatures were highest for ULG95, with Bu20 and E20 having lower but

similar calculated average in-cylinder temperatures across the compression ratio range. It is

proposed that this is due to the higher heat of vaporization of 1-butanol and ethanol as

compared to ULG95 (Table 3-9). This meant that more energy was required to vaporize these

fuels, causing the average in-cylinder temperatures to reduce. The reduced fuel calorific

values of Bu20 and E20 compared to ULG95 also contributed to the lower average in-

cylinder temperatures (longer injection pulse was required).

185
 Fig. 6-2a: In-cylinder Temperature (Bu20)
2100

Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-2b: In-cylinder Temperature (E20)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

CR=10.7 CR=10.9 CR=11.2 CR=11.5

186
 Fig. 6-2c: In-cylinder Temperature (ULG95)
2100

Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100

CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-2 Calculated average in-cylinder temperature versus CAD at KLMBT spark
timings for a) Bu20, b) E20 and c) ULG95

6.3.1.3 MFB

The MFB profiles for the two tested fuel blends of Bu20 and E20, and the reference fuel of

ULG95 are shown in Figures 6-3a, 6-3b and 6-3c, respectively. For E20, there are no

significant differences between the profiles at the different compression ratios while Bu20

and ULG95 show a slightly advanced combustion as the compression ratio was increased. It

is proposed that the more highly compressed fuel-air mixtures at the higher compression ratio

burned more quickly than the less highly compressed mixtures at the lower compression

ratios, causing the combustion to proceed more quickly. Despite this, it appears that the last

stage of combustion (less than 10% of the fuel mass remaining) was faster at lower

compression ratios for all three fuels. For a quantitative analysis of the combustion speed, the

MFB50 and MFB10-90 have been calculated from the MFB profiles (please see next

section).

187
 Fig. 6-3a: MFB (Bu20)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-3b: MFB (E20)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100

CR=10.7 CR=10.9 CR=11.2 CR=11.5

188
 Fig. 6-3c: MFB (ULG95)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-3 MFB versus CAD at KLMBT spark timings for a) Bu20, b) E20 and c)
ULG95

6.3.1.4 MFB50 and MFB10-90

Figure 6-4 shows the MFB50 data for the two tested fuel blends of Bu20 and E20, and the

tested reference fuel of ULG95, across the compression ratio range. As discussed and

explained previously, the KLMBT spark timings were most advanced for Bu20, with E20

second and ULG95 third, thus leading to the most advanced MFB50 of Bu20 across the

compression ratio range, followed by E20 and ULG95. The MFB50 remained almost

constant across the compression ratio range for E20; this is reflected in the MFB profile for

E20 (Figure 6-3b). It is thought that experimental error resulted in the observed trend, as

indicated by the confidence intervals. However, for the other two fuels of Bu20 and ULG95,

there was a significant reduction in the MFB50 across the compression ratio range. Again, it

is proposed that the more highly compressed fuel-air mixtures at the higher compression ratio

burned more quickly than the less highly compressed mixtures at the lower compression

ratios, causing the combustion to proceed more quickly, thus resulting in the MFB50

becoming advanced.
189
 23

22

21
MFB50 (˚aTDC)
20

19

18

17

16

15
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-4 MFB50 versus compression ratio at KLMBT spark timings

Figure 6-5 shows the combustion duration (MFB10-90) data for the two tested fuel blends of

Bu20 and E20, and the tested reference fuel of ULG95, across the compression ratio range. 1-

butanol and ethanol addition to gasoline reduced the combustion duration of the fuel; it is

proposed that 1-butanol and ethanol increased the laminar flame speed, due to the oxygen in

their molecules. The higher chemical reactivity of 1-butanol as compared to ethanol and the

shorter injection duration of Bu20 with respect to E20 explain its shorter combustion duration

in comparison. The shorter combustion duration of Bu20 and E20 as compared to ULG95 is

also due to enhanced in-cylinder turbulence (Stone, 1999) during their combustion which

results from their advanced spark timings. Furthermore, as a result of their advanced spark

timings, the combustion occurs in a smaller volume which increases the in-cylinder

temperature that further promotes combustion. It has to be also noted that the combustion

duration of Bu20 reduced significantly across the compression ratio range; this continues the

trend in Figure 6-4 which shows that the first half of the combustion process also proceeded

more quickly across the range.

190
If the compression ratio was fully optimised for each fuel tested, the combustion phasing

would have been optimised for each fuel. Therefore, it would be expected that the

combustion of each fuel, particularly that of ULG95, would have been quicker due to higher

in-cylinder turbulence at the time of ignition.

24.5

23.5
MFB10-90 (CAD)

22.5

21.5

20.5

19.5

18.5
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-5 MFB10-90 versus compression ratio at KLMBT spark timings

6.3.1.5 Indicated Efficiency

The indicated efficiency for the two tested fuel blends of Bu20 and E20, and the tested

reference fuel of ULG95, across the compression ratio range, is shown in Figure 6-6. They

increased by 1.3, 1.3 and 1.1% for Bu20, E20 and ULG95, respectively. This compares to a

maximum theoretical thermal efficiency increase of 1.8% which can be obtained from

Equation 6-1 by assuming γ=1.4 and solving for the minimum and maximum respected

compression ratios of 10.7 and 11.5.

(6-1)

191
Therefore, the thermal efficiency increase observed is realistic. It is well known that the

volumetric fuel consumption of oxygenated fuels and their low percentage blends in gasoline

is significantly higher than that of gasoline due to their lower calorific value. Therefore, this

is why the indicated efficiency rather than the volumetric fuel consumption has been analysed

in this section.

As the compression ratio is increased, indicated (thermal) efficiency increases, thus

producing the observed behaviour. Bu20 had the highest indicated efficiency, followed by

E20 then ULG95, due to their respected KLMBT spark timings (Table 6-3) and their

respected combustion durations (Figure 6-5). The more advanced the spark timing and the

faster the combustion, the more efficiently the fuel was converted into engine power, thus

resulting in the indicated efficiency increases observed. Their higher indicated efficiencies

also resulted from their lower combustion temperatures (Figures 6-2a, 6-2b and 6-2c) which

reduced the heat loss through the combustion chamber walls.

It should be noted that the compression ratio of this engine is not optimised. Within the

compression ratio range tested, the knock limit was reached before the optimal MFB50

phasing was achieved (Figure 6-4). If the fuels were tested across a range of lower

compression ratios, it is expected that knock may not occur even at WOT. Thus, at the tested

engine load of 8.5 bar IMEP, the engine efficiency may be improved for the fuels tested

because the combustion phasing will be optimized and there will be a higher in-cylinder

turbulence at the time of ignition. This would be expected to reduce the combustion duration

and thus improve thermal efficiency further. However, it is important to note that there is a

trade-off because thermal efficiency would be reduced at lower compression ratios. It must

also be noted that because the injected fuel mass is increased with the Bu20 and E20 fuel

blends as compared to ULG95, a fuel pump in a real engine would consume more power

192
which would reduce the indicated efficiency in comparison to that observed in this

investigation.

0.350

0.345
Indicated Efficiency

0.340

0.335

0.330

0.325

0.320
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-6 Indicated efficiency versus compression ratio at KLMBT spark timings

6.3.1.6 EGT

Figure 6-7 shows the EGTs for the two tested fuel blends of Bu20 and E20, and the tested

reference fuel of ULG95 across the compression ratio range. It is clear to see that there is

general small decrease in EGTs across the compression ratio range. It is proposed that as the

compression ratio was increased and the MFB50 became advanced, the pressure and heat was

more efficiently converted into piston work leading to the EGT decreases across the

compression ratio range. (Kramer et al., 2000) also observed EGT reductions as compression

ratio was increased.

The results also show that ULG95 had the highest EGT for all compression ratios, followed

by E20, then Bu20. It is proposed that the more advanced MFB50 of Bu20 as compared to

ULG95 and E20 (Figure 6-4) resulted in more efficient conversion of the pressure and heat

into work on the piston, resulting in the reduced EGTs in comparison. The MFB50 phasing
193
was more advanced for E20 than ULG95 for all compression ratios (Figure 6-4) leading to

lower EGTs in comparison. The lower calculated average in-cylinder temperatures for the

Bu20 and E20 fuel blends due to their higher heat of vaporization as compared to ULG95,

will have also contributed to their lower EGTs.

695
690
685
680
675
EGT (˚C)

670
665
660
655
650
645
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-7 EGT versus compression ratio at KLMBT spark timings

6.3.1.7 PM Emissions

6.3.1.7.1 Compression Ratio effect on PN Emission

The PN emissions for the two tested fuel blends of Bu20 and E20, along with the tested

reference fuel of ULG95 are shown in Figures 6-8a, 6-8b and 6-8c, respectively. It is clear to

see from Figure 6-8a that the compression ratio increase reduced the smaller nucleation mode

particles on the left-hand side of the plot (10-30 nm) for the Bu20 blend. It is proposed the

observed reduction was due to the increased calculated average in-cylinder temperatures

across the compression ratio range which increased the oxidation of the particles in the

combustion chamber. The KLMBT spark timing was unchanged across the compression ratio

194
range, therefore, mixture preparation was not considered to have had an effect on the

observed behaviour. Although the in-cylinder temperature increases would have increased

primary carbon particle formation by the thermal pyrolysis and dehydrogenation reaction of

fuel vapour/droplets (Zhang et al., 2014b), the increased oxidation of the particles also

resulting from the in-cylinder temperature increases was clearly the stronger effect.

E20 showed a similar trend to Bu20 but it was much weaker; the nucleation mode particles

decreased as the compression ratio was increased. Again, it is proposed that the higher

calculated average in-cylinder temperatures (Figure 6-2b) increased the rate of oxidation of

these particles in the combustion chamber, leading to the observed trend. For both Bu20 and

E20, no significant changes in accumulation mode particle numbers on the right hand side of

the plot (30-300 nm) were observed. It is believed that the increased oxidization of particles

resulting from the increased calculated average in-cylinder temperatures across the

compression ratio range was cancelled out by increased rate of particle formation caused by

the increased primary carbon particle formation (Zhang et al., 2014b), also resulting from the

increased in-cylinder temperatures.

The data for ULG95 shows a completely uni-modal distribution with no significant

nucleation mode particles being recorded. As the compression ratio was increased, the

formation of accumulation mode particles increased. It is proposed that the accumulation

mode particles increased across the compression ration range for ULG95 because the

increased calculated average in-cylinder temperatures increased primary carbon particle

formation (Zhang et al., 2014b), as with E20. This appears to have overcome the effect of

increased particle oxidization resulting from the higher calculated average in-cylinder

temperatures. Again, the KLMBT spark timing was unchanged across the compression ratio

range, thus mixture preparation is not thought to have had an effect on the observations.

195
For the two tested fuel blends Bu20 and E20, along with the tested reference fuel ULG95, it

is proposed that significant nuclei adsorption of nucleation mode particles onto the

accumulation mode particles occurred and this along with the thermodenuder, which removed

many of the nucleation particles before they could be measured, led to the mostly uni-modal

behaviour observed.

Overall, the effect of compression ratio increase on PM emissions is not significant when the

95% confidence intervals are taken into consideration, due to the compression ratio increase

being relatively small. The 95% confidence intervals themselves could not be displayed on

the figures because they would have become unclear.

Fig. 6-8a: PM (Bu20)

1.E+06
dN/dlogDp (#/cm3)

1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)
CR=10.7 CR=10.9 CR=11.2 CR=11.5

196
 Fig. 6-8b: PM (E20)
1.E+06

dN/dlogDp (#/cm3)
1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-9c: PM (ULG95)

1.E+06
dN/dlogDp (#/cm3)

1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)

CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-8 PN emissions versus particle diameter at KLMBT spark timings for a) Bu20,
b) E20 and c) ULG95

197
6.3.1.7.2 Fuel Effect on PN Emission

Comparing the behaviours of the different fuels in Figures 6-8a, 6-8b and 6-8c, 1-butanol

significantly reduced the PN when added to the gasoline fuel, whereas ethanol had little or no

effect. It is proposed that the significantly earlier MFB50 phasing (Figure 6-4) and shorter

combustion duration of Bu20 (Figure 6-5), as compared to the other two fuels, provided more

time for oxidation of the particulates after the combustion process, leading to the significant

PN reduction. This appears to have overcome effect of the advanced KLMBT spark timing

(Table 6-3), which would have reduced the fuel-air mixing time, as well as the effect of the

reduced calculated average in-cylinder temperatures (Figure 6-2a). These would have

resulted in more pockets of a high local equivalence ratio and a reduced oxidation rate in the

combustion chamber, respectively, which alone would have led to an increase in

accumulation mode particles.

Also, it is important to note that reduced calculated average in-cylinder temperatures will

have also reduced the soot formation rate through reducing the formation of primary carbon

particles (Zhang et al., 2014b), thus contributing to the reductions observed. In addition, it is

thought that because the gasoline already had 5%vol ethanol content, the increase in ethanol

content to 20%vol made little difference to the PN behaviour of E20. Furthermore, E20 may

not have reduced PM because of increased piston/wall wetting resulting from the increased

injection duration required to maintain the same engine load, due to its reduced calorific

value. The calorific value of Bu20 is higher so this effect was not as significant as with E20.

Overall, there is no significant effect of fuel type on the particles average size with all

distributions peaking at around 60 nm.

198
(Gu et al., 2012, He et al., 2010, Karavalakis et al., 2013, Niass et al., 2012, Zhang et al.,

2014b) also reported that 1-butanol addition to gasoline fuel reduced the PN concentration,

and (Bielaczyc et al., 2014, Catapano et al., 2013, Catapano et al., 2014, Costagliola et al.,

2013, Di Iorio et al., 2011, Ojapah et al., 2014, Storey et al., 2010, Vuk and Vander Griend,

2013) also observed the same for ethanol addition to gasoline fuel.

There are further reasons as to why the accumulation mode particles decreased with 1-

butanol addition to gasoline fuel. There is a positive correlation between the accumulation

mode particles and the polycyclic aromatic hydrocarbons (PAHs); the addition of alcohol to

gasoline reduces the aromatic content of the fuel, thus it also caused the accumulation mode

particles to decrease (De Abrantes et al., 2009, Zhang et al., 2014b). Furthermore, the oxygen

content in the fuel blend leads to a lower formation rate of soot and also to a higher oxidation

rate of soot (Zhang et al., 2014b). Biofuels tend to have a lower droplet velocity and

relatively low mean droplet diameter, which reduces piston impingement resulting in higher

combustion efficiency (Figure 6-6) and lower accumulation mode particle formation (Tian et

al., 2010). Furthermore, biofuels burn at higher pressures (Figures 6-1a and 6-1b) which help

to promote pyrolysis, reducing solid carbonaceous emissions (Daniel, 2012d).

Lastly, Bu20 had a noticeably higher number of nucleation mode particles than the other two

fuels tested. It is thought that this was due to the lower soot accumulation mode particles

observed, which meant less adsorption of the nucleation mode particles onto the

accumulation mode particle surfaces occurred, leading to higher numbers being observed.

Overall, the effect of 1-butanol addition to gasoline on PM emissions is significant when the

95% confidence intervals are taken into consideration, while ethanol addition to gasoline has

no significant effect at the blend ratio tested.

199
6.3.1.8 Gaseous Emissions (NOx and HC)

Figure 6-9 presents the NOx emission data for the two tested fuel blends and tested reference

fuel. Overall, there is a significant increase in NOx emissions across the compression ratio

range; they increased by 17.9% for Bu20, 21.7% for E20 and 23.5% for ULG95. These

increases occurred because of the aforementioned increase in calculated average in-cylinder

temperatures across the compression ratio range, which caused more NO x to be formed. It is

also clear that ULG95 had the highest NOx emission, followed by Bu20 then E20. It is

proposed that the lower calculated average combustion temperatures of Bu20 (Figure 6-2a)

and E20 (Figure 6-2b) reduced the formation of NOx emissions in comparison to ULG95.

Despite the calculated average in-cylinder temperatures being similar for Bu20 and E20

across the compression ratio change and ethanol having a higher oxygen to carbon ratio than

1-butanol, Bu20 produced more NOx emissions than E20. It is proposed that the earlier

MFB50 of Bu20 as compared to E20 (Figure 6-4) resulted in the observed behaviour. With an

advanced MFB50, the fuel-air mixture burns at an earlier point in the engine cycle where the

in-cylinder turbulence is stronger, causing the combustion to proceed more quickly. It is

thought that this resulted in higher local temperatures which resulted in the increased NO x

emissions for Bu20. Furthermore, the NOx emissions can also be attributed to the H/C ratio.

E20 had the highest H/C ratio, followed by Bu20 then ULG95. As reported by (Harrington

and Shishu, 1973), the NOx emission has an inverse relationship to H/C. The combustion of

fuels with a higher H/C ratio tends to have a lower adiabatic flame temperature because water

has a higher specific heat capacity than CO2 (Harrington and Shishu, 1973, Daniel et al.,

2011), thus further explaining why the NOx emission formation was lower for E20 as

compared to Bu20.

200
 11.0

10.5

10.0

9.5
NOx (g/kWh)

9.0

8.5

8.0

7.5

7.0

6.5
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-9 NOx emissions versus compression ratio at KLMBT spark timings

Figure 6-10 presents the HC emissions data for the two tested fuel blends and the tested

reference fuel. It is clear to see that the HC emissions increased significantly across the

compression ratio range; they increased by 20.9% for Bu20, 20.8% for E20 and 26.2% for

ULG95. It is believed that the increased surface to volume ratio of the combustion chamber

and the higher relative influence of the crevice volume as compared to the whole volume of

the combustion chamber resulted in the observed HC emission increases (Kramer et al., 2000,

Maji et al., 2001).

The HC emissions were lower for Bu20 and E20 as compared to ULG95 because their

oxygen content was higher, which promoted the oxidation of HC in the combustion chamber.

This appears to have overcome the reduced fuel-air mixing time caused by the more

advanced KLMBT spark timings and the reduced combustion temperatures, which alone

would have caused the HC emissions to increase. Ethanol has a higher oxygen to carbon ratio

than 1-butanol, thus, there was a higher HC oxidation rate with E20 as compared to Bu20,

leading to lower HC emissions in comparison. Also, the KLMBT spark timing was more
201
advanced for the Bu20 fuel blend in comparison to E20, resulting in poorer mixture

preparation and thus higher HC emissions. Despite the differences observed in the HC (and

NOx) emission between the different fuels, this would be expected to diminish in a vehicle

when a TWC is used, due to their high conversion efficiencies.

It must be noted that these HC emissions are uncorrected; the sensitivity of the FID analyser

to oxygenated compounds has not been taken into account. Research has shown that FID

analysers have reduced sensitivity to oxygenated fuels (Cheng et al., 1998, Wallner and

Miers, 2008), suggesting that the HC emissions for Bu20 and E20 are higher than those

observed in these results, necessitating detailed HC emission speciation for reliable analysis.

The ULG95 fuel also contains oxygenated components (5%vol ethanol content) so its HC

emissions are likely to be higher too.

5.1

4.9

4.7

4.5
HC (g/kWh)

4.3

4.1

3.9

3.7

3.5
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
Bu20 E20 ULG95

Figure 6-10 HC emissions versus compression ratio at KLMBT spark timings

202
6.3.1.9 Big Picture

Figure 6-11 shows the overall effect of compression ratio and fuel on the gaseous emissions,

indicated efficiency and total PN. It is clear to see that for ULG95, the gaseous emissions of

NOx and HC increased with increased compression ratio, along with the indicated efficiency

and total PN. However, when 1-butanol and ethanol are blended into the ULG95 fuel, the

gaseous emissions of NOx and HC are reduced, along with total PN, and the indicated

efficiency is increased. Ethanol is most effective to reduce the gaseous emissions (NOx and

HC) of the ULG95 fuel and 1-butanol is most effective to reduce the total PN emission.

Figure 6-11 Overall effect of compression ratio and fuel on gaseous emissions, indicated
efficiency and total PN (integrated across 10-289 nm range) at KLMBT spark
timings

203
6.3.2 Investigation of the Compression Ratio and Hot EGR Effect on
Combustion and PM Emissions

6.3.2.1 KLMBT Spark Timing

The KLMBT spark timings for the three EGR conditions are shown in Table 6-4. As the

compression ratio was increased, the KLMBT spark timing was retarded because the fuel-air

mixture was more compressed at TDC. This increased the pressure and temperature in the

combustion chamber, leading to increases in the end-zone temperatures, which would have

resulted in engine knock had the spark timing not been retarded. The research engine began

to knock at a load of 6.0 bar IMEP, thus, the compression ratio increase made the engine

more prone to knock at the tested load of 7.0 bar IMEP, resulting in the spark retard required.

The KLMBT spark timing was advanced as the EGR addition was increased because the

temperatures in the end-zone of the combustion chamber were suppressed by the EGR gases.

Despite this, KLMBT spark timing advances would likely have been greater if intake

manifold temperature increases, resulting from the EGR addition, could have been

eliminated. The overall effect of the compression ratio increase and EGR addition was to

advance the spark timing by 2 CAD, due to the knock suppression ability of the EGR gases.

The 4 CAD retard in KLMBT spark timing at 7% EGR addition as the compression ratio was

increased from 11.2 to 11.5, larger than the 2 CAD spark retard required with the other

compression ratio increases, is not considered to be significant. Therefore, it will not be

commented upon further.

204
 Table 6-4 KLMBT spark timings (˚bTDC) (compression ratio & EGR addition)

Compression 10.7 10.9 11.2 11.5

Ratio

EGR Addition (%)

0 22 20 18 16
7 26 24 22 18
14 30 28 26 24

6.3.2.2 In-cylinder Pressures and Temperatures

The in-cylinder pressures at the KLMBT spark timings, as shown in Figures 6-12a, 6-12b and

6-12c for the three EGR conditions, generally reduced as the compression ratio was increased

and the KLMBT spark timings (Table 6-4) had to be consequently retarded to avoid engine

knock. It must be noted that knocking could have been avoided if the compression ratio was

fully optimized for the fuel used. However, this was not pursued in order to investigate the

effect of the parameter changes on engine knock limit and the consequent combustion and

emission parameters at this limit.

At the compression ratio of 10.9, the in-cylinder pressure at 0 and 7% EGR addition did not

reduce as compared to that at the compression ratio of 10.7, and it increased at 14% EGR

addition. It is proposed that although the KLMBT spark timing had to be retarded with the

compression ratio increase, the rate of heat release (ROHR) increased due to the more highly

compressed fuel-air charge. At 14% EGR addition, this was also sufficient to overcome the

retarded KLMBT spark timing, as the compression ratio was increased to 10.9. However,

further increases in the ROHR produced by compression ratio increases beyond 10.9 were not

sufficient to overcome the retarded combustion phasing resulting from the KLMBT spark

timing retards, causing the in-cylinder pressure to reduce as the compression ratio was further

increased.

205
The overall effect of compression ratio increase and EGR addition was reduced in-cylinder

pressures as can be seen by comparing Figures 6-12a and 6-12c. It would appear the reduced

ROHR produced by the EGR addition was more significant than the advanced combustion

phasing resulting from the 2 CAD KLMBT spark timing advance (Table 6-4). Similarly

(Alger et al., 2015) observed that in-cylinder pressure increases resulting from an increased

compression ratio were cancelled out by the consequent spark retard required to avoid engine

knock.

Smaller in-cylinder pressure decreases with EGR addition would have been likely if intake

manifold temperature increases, resulting from the EGR addition, could have been

eliminated. This is because, as mentioned, it would likely have led to a greater KLMBT spark

timing advance and thus more of the energy released from the fuel would have contributed to

the in-cylinder pressure increases.

Fig. 6-13a: In-cylinder Pressure (0% EGR)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

206
 Fig. 6-13b: In-cylinder Pressure (~7% EGR)
45

In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-13c: In-cylinder Pressure (~14% EGR)

45
In-cylinder Pressure (bar)

40

35

30

25

20

15

10

0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-12 In-cylinder pressure versus CAD at KLMBT spark timings for a) 0% EGR, b)
7% EGR, c) 14% EGR

The calculated average in-cylinder temperatures at the KLMBT spark timings for the three

EGR conditions, as shown in Figures 6-13a, 6-13b and 6-13c, generally reduced with

compression ratio increases and consequent spark timing retards. This is again due to the

KLMBT spark timing retards required with the compression ratio increases, as with the in-

cylinder pressure reductions. The overall effect of increased compression ratio and EGR

207
addition was to reduce the average calculated in-cylinder temperatures as can be seen by

comparing Figures 6-13a and 6-13c. This is the result of the KLMBT spark timing retards

required to avoid engine knock at the increased compression ratios and the temperature

suppression ability of the EGR gases due to their higher heat capacity as compared to air.

It is believed that the increased intake manifold temperatures resulting from the EGR addition

were compensated for by the less optimized combustion phasing resulting from the

consequently reduced KLMBT spark timing advance, compared to the maximum spark

advance possible, as explained in the first part of Chapter 4. Thus, it is thought the two effects

of increased intake manifold temperature and reduced subsequent KLMBT spark timing

advances cancelled one another out, meaning there was no overall effect on the peak of the

calculated average in-cylinder temperatures. Again, it is thought that the improved KLMBT

spark timing would have quickly compensated for the intake gas temperature decrease by the

time temperatures reached the beginning of the range reported (500˚C); thus, the effect would

have been minimal.

Fig. 6-14a: In-cylinder Temperature (0% EGR)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

208
 Fig. 6-14b: In-cylinder Temperature (~7% EGR)
2100

Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-14c: In-cylinder Temperature (~14% EGR)

2100
Calculated In-cylinder Temperature (˚C)

1900

1700

1500

1300

1100

900

700

500
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-13 Calculated average in-cylinder temperatures versus CAD at KLMBT spark
timings for a) 0% EGR, b) 7% EGR and c) 14% EGR

6.3.2.3 MFB

The MFB profiles at the KLMBT spark timings for the three EGR conditions are shown in

Figures 6-14a, 6-14b and 6-14c. The profiles became retarded as the compression ratio was

increased because of the KLMBT spark retard that was required in order to avoid engine

knock. From the MFB profile for 14% EGR addition (Figure 6-14c) it can be seen that there

209
was a less significant difference between the profiles at different compression ratios as

compared to the MFB profiles for 0% and 7% EGR addition. It is thought that the knock

suppression ability of the EGR gases allowed the MFB50s at the different compression ratios

to be more advanced towards their optimum phasing, resulting in the MFB profiles becoming

more similar to each other at the highest EGR addition.

From comparing Figures 6-14a and 6-14c, it can be seen that the overall effect of the

compression ratio increase and EGR addition was to retard and elongate the MFB profile, due

to the KLMBT spark timing retard required with the compression ratio increase and the

combustion slowing effects of the EGR gases (Rhodes and Keck, 1985).

Again, KLMBT spark timing advances would likely have been greater if intake manifold

temperature increases, resulting from the EGR addition, could have been eliminated; thus, the

retardation of the MFB profiles would likely have been reduced. The MFB profiles will have

become slightly more elongated too because the slightly cooler fuel-air charge will have

burned more slowly. If the compression ratio was fully optimised for the fuel used, the

combustion phasing would have been optimised for each EGR ratio tested, therefore, it would

be expected that the combustion would have been quicker at each condition due to higher in-

cylinder turbulence at the time of ignition.

210
 Fig. 6-15a: MFB (0% EGR)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-15b: MFB (~7% EGR)

1.0

0.9

0.8

0.7

0.6
MFB

0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

211
 Fig. 6-15c: MFB (~14% EGR)
1.0

0.9

0.8

0.7

0.6

MFB
0.5

0.4

0.3

0.2

0.1

0.0
-50 0 CAD 50 100
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-14 MFB versus CAD at KLMBT spark timings for a) 0% EGR, b) 7% EGR and
c) 14% EGR

6.3.2.4 Fuel Consumption

Gravimetric ISFCnet at the KLMBT spark timings, as shown in Figure 6-15, increased across

the compression ratio range due to the KLMBT spark timing retards that were required. This

caused the MFB50 (Figures 6-14a-c) to become retarded away from its optimum 8-10˚aTDC

phase (de O. Carvalho et al., 2012), thus causing the fuel consumption increase observed.

Fuel consumption increased by 4.4, 4.5 and 4.4% for the 0, 7 and 14% EGR additions,

respectively. Therefore, increasing the compression ratio at 7.0 bar IMEP resulted in worse

fuel consumption behavior.

Across the EGR range the fuel consumption decreased significantly, due to the KLMBT

spark timing advances and PMEP reductions that could be achieved. It decreased by 4.6, 4.6,

4.3 and 4.6% for the compression ratios of 10.7, 10.9, 11.2 and 11.5 respectively. While the

KLMBT spark timing advances were not sufficient to advance the MFB profile (Figures 6-

14a-c), the PMEP was reduced significantly across the EGR range. The PMEP reduced from

212
0.36 to 0.28 bar, 0.36 to 0.27 bar, 0.38 to 0.26 bar, and 0.34 to 0.25 bar for the compression

ratios of 10.7, 10.9, 11.2 and 11.5, respectively. Furthermore, the reduced combustion

temperatures with EGR addition improved thermal efficiency (Ratnak et al., 2015, Siokos et

al., 2015), contributing to the fuel economy improvements observed.

Overall, the indicated fuel consumption decreased by 0.4% across the compression ratio and

EGR range; i.e. the fuel consumption was decreased by 0.4% at the compression ratio of 11.5

and EGR addition of 14% as compared to that observed at the compression ratio of 10.7 and

EGR addition of 0%. This is because of the reduction in pumping losses and the 2 CAD

KLMBT spark timing advance. Again, KLMBT spark timing advances would likely have

been greater if intake manifold temperature increases resulting from the EGR addition could

have been eliminated; thus, the fuel consumption reductions would likely have been greater.

Although the fuel consumption reduction observed is considered to be significant, it must be

noted that the VAF had an average error of 1.6%. Thus, the reduction of 0.4% could have

resulted from this error, meaning that the result observed is limited.

265

260

255
ISFCnet (g/kWh)

250

245

240

235

230
10.6 10.8 11.0 11.2 11.4 11.6
Compression Ratio
0% EGR 7% EGR 14% EGR

Figure 6-15 Gravimetric ISFCnet versus compression ratio at KLMBT spark timings

213
6.3.2.5 PM Emissions

The PN emissions at the KLMBT spark timings for the three tested EGR conditions are

shown in Figures 6-16a, 6-16b and 6-16c. For the 0 and 7% EGR additions, there was a

significant increase in nucleation mode particles across the compression ratio range. There

was also a small but significant increase in accumulation mode particles across the

compression ratio range at 0% EGR addition, while at 7% EGR addition, there was no

significant change in the accumulation mode particles across the range. Furthermore, the

accumulation mode particles significantly increased across the compression ratio range with

14% EGR addition, but there was no significant change in the nucleation mode particles at

this EGR addition.

It is proposed that the reduced calculated average in-cylinder temperatures (Figures 6-13a-c)

caused by the KLMBT spark timing retards (Table 6-4) reduced the oxidization of particles in

the combustion chamber in the expansion and exhaust strokes, resulting in the observed

behavior. This appears to have been more significant than the reduced primary carbon

particle formation, resulting from the reduced in-cylinder temperatures (Zhang et al., 2014b).

In addition, the time for post-combustion oxidization was reduced (Figures 6-14a-c) due to

the MFB profile becoming more retarded, contributing to the increased PNs observed. It is

believed that the accumulation mode particles did not significantly increase at 7% EGR

addition because the KLMBT spark timing became more retarded across the compression

ratio range (Table 6-4) as compared to the other EGR additions. This provided more time for

fuel-air mixing in comparison which reduced the number of pockets of a high local

equivalence ratio, meaning the fuel was burned more completely, cancelling out the effect of

the reduced oxidization produced by the average in-cylinder temperature reductions.

214
Across the EGR range there was a small but significant decrease in the nucleation and

accumulation particles observed for the compression ratios tested. It is proposed that the

calculated average in-cylinder temperature decreases across the EGR range (Figures 6-13a-c)

reduced primary carbon particle formation by the thermal pyrolysis and dehydrogenation

reaction of fuel vapour/droplets, resulting in the PN decreases observed (Zhang et al., 2014b).

In addition, the slower combustion produced by the EGR addition (Figures 6-14a-c) provided

more time for the particulates to oxidize in the hot flames, contributing to the reductions

observed. These effects appear to have been more significant than the reduced fuel-air mixing

time due to the KLMBT spark timing advances (Table 6-4). This alone would have created

more pockets of a high local equivalence ratio which would have produced higher PNs.

The overall effect of the compression ratio increase and EGR addition was a small but

significant increase in the nucleation mode particle number and a significant reduction in the

accumulation mode PN. The KLMBT spark timing was only advanced by 2 CAD (Table 6-4)

therefore, it is not thought that the relative homogeneity of the two mixtures was significantly

different. However, the primary carbon particle formation was reduced (Zhang et al., 2014b),

due to the in-cylinder temperature decreases across the EGR range. This resulted in the

reduced PN observed. In addition, the slower combustion produced by the EGR addition

provided more time for the particulates to oxidize in the hot flames, contributing to the PN

reductions. It is thought that due to the lower numbers of accumulation mode particles across

the compression ratio range, less nuclei adsorption happened, which caused the observed

increases in the nucleation mode particles.

If the intake manifold temperature increases with EGR addition could have been eliminated,

then it is expected that the PM emissions would have been further reduced with EGR addition

due to the greater post-combustion oxidation time available with the expected greater

215
expected KLMBT spark timing advances (Daniel et al., 2012e). The slightly longer expected

combustion duration would also decrease the PM observed. However, poorer mixture

preparation resulting from the greater KLMBT spark timing advances would be expected to

limit the PM reductions, especially since this effect would be expected to result in an overall

PM increase in the results from the first part of Chapter 4. As discussed, the intake manifold

temperature effect on the calculated average in-cylinder temperature is believed to have been

minimal; therefore, this is not thought to have affected the PM.

Fig. 6-17a: PM (0% EGR)

1.E+07
dN/dlogDp (#/cm3)

1.E+06

1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)

CR=10.7 CR=10.9 CR=11.2 CR=11.5

216
 Fig. 6-17b: PM (~7% EGR)
1.E+07

dN/dlogDp (#/cm3)
1.E+06

1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)
CR=10.7 CR=10.9 CR=11.2 CR=11.5

Fig. 6-17c: PM (~14% EGR)

1.E+07
dN/dlogDp (#/cm3)

1.E+06

1.E+05

1.E+04
10 100 1000
Particle Diameter (nm)

CR=10.7 CR=10.9 CR=11.2 CR=11.5

Figure 6-16 PN emissions versus particle diameter at KLMBT spark timings for a) 0%
EGR, b) 7% EGR and c) 14% EGR

6.4 Data Continuity

For a detailed analysis regarding the continuity of the data between this chapter and Chapters

4 and 5, please refer to Appendix B3.

217
6.5 Conclusions
The effect of compression ratio and fuel on the combustion and emissions of a single cylinder

DISI research engine was investigated in this chapter and the following conclusions have

been made.

1. 1-butanol and ethanol addition to gasoline advanced the MFB50 phasing as well as

reducing the overall combustion duration across the compression ratio range; 1-

butanol had the greatest effect on these parameters.

2. 1-butanol addition to gasoline significantly reduced the accumulation mode PN

emission, mainly due to the earlier combustion phasing and thus increased post-

combustion oxidization time; ethanol addition to gasoline had little effect on the

emission.

3. 1-butanol and ethanol addition to gasoline significantly reduced the NO x and HC

emission across the compression ratio range, with ethanol being the most effective.

4. Overall, if combustion and PN emission parameters are the priority, then the Bu20

fuel blend has the most potential, while if NOx and HC emission parameters are the

priority, then the E20 fuel blend has the most potential. Synergies between

compression ratio increase and alcohol addition to gasoline enable PM and gaseous

emissions to be simultaneously controlled while increasing indicated efficiency with

respect to standard gasoline combustion.

The main findings from this section are also summarized in Table 6-5.

218
 Table 6-5 Results summary (compression ratio & fuel) (highlighted=improvement,
underlined=worsening)

Result KLMBT Indicated EGT PM NOx HC

Summary Efficiency
Nucleation Accumulation
Compression
Ratio/Fuel
Bu20 (at compression 0 CAD 1.3% 21.0˚C No sig. change No sig. change 17.9% 20.9%
ratio of 11.5, compared to
10.7)

E20 (at compression ratio 0 CAD 1.3% 16.2˚C No sig. change No sig. change 21.7% 20.8%
of 11.5, compared to 10.7)

ULG95 (at compression 0 CAD 1.1% 13.4˚C No sig. change No sig. change 23.5% 26.2%
ratio of 11.5, compared to
10.7)

Bu20 (compared to 4 CAD 4.5% 21.3˚C Reduction Reduction 10.0% 7.7%

ULG95 at compression
ratio of 11.5)

E20 (compared to ULG95 2 CAD 2.4% 14.6˚C No sig. change No sig. change 20.9% 12.1%
at compression ratio of
11.5)

The effect of compression ratio and hot EGR on the combustion and PM emissions of a

single cylinder DISI research engine was also investigated in this chapter and the following

conclusions have been made.

5. EGR addition of 14% overcame the increased knock produced as the compression

ratio was increased from 10.7 to 11.5, resulting in an overall KLMBT spark timing

advance of 2 CAD with the compression ratio increases and EGR addition.

6. Fuel consumption was reduced significantly by 0.4% as the compression ratio was

increased from 10.7 to 11.5 and the EGR addition was increased from 0 to 14%;

reduced pumping losses, lower in-cylinder temperatures (improved thermal

efficiency) and the KLMBT spark timing advance resulted in the decrease observed.

7. PM was significantly reduced with compression ratio increase and EGR addition due

to the reduced primary carbon particle formation resulting from the in-cylinder

temperature decreases across the EGR range. In addition, the increased time available

219
 for PM to oxidize in the hot flames, due to the slower combustion which resulted from

EGR addition, contributed to the PM decreases observed.

The main findings from this section are also summarized in Table 6-6.

Table 6-6 Results summary (compression ratio & EGR addition)

(highlighted=improvement, underlined=worsening)

Result KLMBT ISFCnet PM

Summary
Nucleation Accumulation
Compression
Ratio/EGR
Compression ratio of 6 CAD 4.4% Increase Increase
11.5, 0% EGR addition
(compared to 10.7, 0%
EGR addition)

Compression ratio of 8 CAD 4.6% Reduction Reduction

11.5, 14% EGR addition
(compared to 11.5, 0%
EGR addition)

Compression ratio of 2 CAD 0.4% Increase Reduction

11.5, 14% EGR addition
(compared to 10.7, 0%
EGR addition)

220
 Chapter 7

Summary, Conclusions and Suggestions for Future Work

The main aim of this thesis was to study the effect of EGR on the combustion and emissions

of a DISI engine. In particular, the PM emissions were studied, and the results and discussion

from this work are shown in the preceding chapters. The simplicity of EGR should allow for

its adoption in all automotive gasoline engines rather than the partial adoption we see today,

allowing increased compression ratios to be used as well as increased swirl and tumble ratios.

However, PM still remains a challenge which could be overcome with increased swirl ratios

and by blending 1-butanol into gasoline fuel. The aim of this chapter is to present the

summary and conclusions of this current investigation and to make suggestions for future

work.

7.1 Summary and Conclusions

This thesis contains numerous results from three chapters of experimental data from a 4-

valve, 4-stroke, single-cylinder spray-guided DISI research engine, with a variable

compression ratio of 10.7-11.5:1. The most significant findings of the investigation are

presented below according to the area of study they came from.

It must be noted that the compression ratio was not optimized for each fuel used and for each

engine condition tested. For the conditions which did not knock (e.g. 5.5 bar IMEP) the

compression ratio could have been increased to improve thermal efficiency and the optimum

combustion phasing would not have been affected. For the conditions which did knock, the

compression ratio could have been reduced in order to obtain the optimum combustion

phasing to improve thermal efficiency, although this would have been limited by the

221
reduction in thermal efficiency resulting from the compression ratio reduction. The least

optimized load was 8.5 bar IMEP when gasoline fuel was used with no EGR addition since

this required significant spark retards in order to avoid engine knock.

The Effect of EGR and its Type on Engine Combustion and Emissions

EGR addition of up to 13% enabled the KLMBT spark timing to be significantly advanced

due to the suppression of end-zone temperatures, which was sufficient to advance the MFB

profiles at 7.0 and 8.5 bar IMEP, but because the load of 5.5 bar IMEP was not knock

limited, the MFB profile was already optimized so improvements could not be made. The

more optimum MFB50 combustion phasing at 7.0 and 8.5 bar IMEP along with pumping

reductions at all tested loads due to the throttle being more opened reduced the fuel

consumption across the EGR ranges (2.2% at 5.5 bar IMEP, 4.1% at 7.0 bar IMEP, and 1.0%

at 8.5 bar IMEP). PM accumulation mode emissions generally increased at 5.5 and 7.0 bar

IMEP due to the lower in-cylinder and exhaust temperatures while they reduced in the

nucleation mode through increased nuclei adsorption due to the greater number of

accumulation mode particles present. The low EGR addition of 3% at 8.5 bar IMEP resulted

in no significant changes in the PM. Finally, NOx emissions decreased significantly due to the

lower in-cylinder temperatures and HC emissions increased significantly due to the lower in-

cylinder and EGTs.

Extending the work further allowed the hot EGR system from the first experiment to be

compared against a cooled EGR system, along with a hot and cooled EGR system with a

simulated TWC, at 7.0 bar IMEP, with EGR addition of up to 14%. The KLMBT spark

timing could be most advanced with the cooled EGR after TWC EGR type due to the cooler

gases suppressing end-zone temperatures and because the EGR gases did not contain NO

which induces knock. The fuel consumption improvement (3.8%) was greatest for cooled

222
EGR, mostly due to its most improved MFB50 combustion phasing as compared to the other

EGR types. Gaseous emissions (NOx and HC) were best optimized with the cooled EGR after

TWC EGR type (45.3% NOx decrease and 10.1% HC increase). This is mainly because its

MFB50 combustion phasing did not advance across the EGR range as significantly as the

cooled EGR type. This resulted in less NOx formation in comparison, because with the less

advanced MFB50, the fuel-air mixture burned at a later point in the engine cycle where the

in-cylinder turbulence was weaker, causing the combustion to proceed less quickly. This is

believed to have resulted in lower local temperatures in comparison to the cooled EGR

condition which resulted in the lower NO x and thus greater NOx reductions across the EGR

range. Also, because the simulated gas came from a gas bottle, it did not introduce additional

HCs into the combustion chamber (reflecting the behaviour of a TWC with 100% efficiency)

as the hot and cooled EGR types did, reducing the HC increases observed.

The Effect of Intake Airflow and EGR on Engine Combustion and PM Emissions

Swirl and high tumble intake airflows significantly retarded the KLMBT spark timing at 7.0

bar IMEP. However, significant improvements could be made for all the intake airflow

conditions with EGR addition of up to 11%. Fuel consumption increased significantly with

swirl (6.4%) and high tumble (15.5%) intake airflows as compared to the reference low

tumble condition because of the KLMBT spark timing retards required. However, the

improvements made with EGR addition reduced the fuel consumption penalty significantly

compared to the reference condition (4.8% and 6.2% increase, respectively, at maximum

EGR additions).

PM was significantly reduced with the swirl intake airflow due to improved mixture

formation resulting from the KLMBT spark timing retard providing more fuel-air mixing

time and from the increased in-cylinder turbulence produced by the airflow. The high tumble

223
condition resulted in reduced nucleation but similar accumulation mode particles compared to

the reference low tumble condition. This is because the post-combustion oxidization time was

significantly reduced by the KLMBT spark timing retard and the improved mixture formation

was not sufficient to overcome this. EGR addition reduced PM for the reference low tumble

condition due to an increased post-combustion oxidization time resulting from the KLMBT

spark timing advances. However, it remained approximately the same across the EGR range

for the swirl and high tumble conditions. It is thought that the KLMBT spark timing advances

resulted in poorer fuel-air mixture preparation due to reduced mixture preparation time

available, which in addition to the reduced calculated average in-cylinder temperatures for

the swirl condition across the EGR range, cancelled out the effect of the increased post-

combustion oxidization time. It also cancelled out the effect of decreased primary carbon

particle formation resulting from the in-cylinder temperature reductions for the swirl

condition.

The Effect of Compression Ratio, Fuel and EGR on Engine Combustion and Emissions

At 8.5 bar IMEP, compression ratio increases from 10.7 to 11.5 did not result in KLMBT

spark timing changes because the load condition of 8.5 bar IMEP was already prone to

knock, so it did not worsen despite increasing the compression ratio. Bu20 had the most

advanced KLMBT spark timing, followed by E20, then ULG95, mainly due to the 5%vol

ethanol pre-mixed in it, giving it a 25%vol total alcohol content, as compared to only 20%vol

for E20. Indicated efficiency was most improved for Bu20 and E20 due to their greater

KLMBT spark timing advances as compared to ULG95. Compression ratio had little effect

on the PM emission, with the most significant effect being the Bu20 fuel blend which

reduced PN. This was mainly due to its most advanced spark timing and shorter combustion

duration significantly increasing the post-combustion oxidation time available. Increasing the

224
ethanol addition from 5%vol (ULG95) to 20%vol (E20) had no significant effect on the PM

emission.

At 7.0 bar IMEP, compression ratio increases from 10.7 to 11.5 resulted in the KLMBT spark

timing becoming more retarded. This is because the engine was less prone to knock at this

load, so small changes in the compression ratio made a significant difference to the knock

tendency. EGR addition of 14% however enabled this effect to be overcome, enabling the

KLMBT spark timing to be advanced overall by 2 CAD with compression ratio increases and

EGR addition. Fuel consumption was significantly improved by 0.4% with the compression

ratio increases and EGR addition due to the reduced pumping losses with EGR addition and

the KLMBT spark timing advances. Finally, PM emissions were reduced overall with the

compression ratio increases and EGR addition, because of the improved oxidation of

particulates in the slower flame, and the reduced primary carbon particle formation resulting

from in-cylinder temperatures decreases.

7.2 Suggestions for Future Work

This study has explored a wide range of combustion and emission aspects of EGR use in

DISI engines, along with that of swirl and tumble intake airflows, compression ratio increase

and oxygenated fuels. However, before the approaches outlined in this thesis can be fully

implemented in commercial DISI engines, further research is required to better understand

the behaviour of these parameters and to tackle the main challenges that have been identified

in their use. The author has thus outlined some of the outstanding issues below to serve as

recommendations for future work.

In general, the investigation should be extended to quantify the effects of EGR and its type,

swirl and tumble intake airflows, compression ratio and oxygenated fuels over the entire

225
engine load and speed range. Split-injection strategies, different fuel injection timings and

different valve timings/lifts should be studied to optimize the combustion and emissions.

Synergies between these different techniques (e.g. swirl & tumble and EGR along with

increased compression ratio) should be investigated once they have all been explored

individually on a deep level. Results should be obtained not only at KLMBT spark timings

but at a 10 CAD retard from the KLMBT too, because these are used in cold-start for catalyst

warming and in turbocharged engines for knock reduction.

The compression ratio should be optimized for each fuel to ensure that the MFB50 does not

become too retarded from its optimum phasing. However, it should be noted that any

decrease in the compression ratio will lead to lower indicated efficiencies at the engine loads

that did not knock while potentially increasing the indicated efficiencies at the higher engine

loads which did knock. This will only happen if the benefits of increased burning speed due

to the increased in-cylinder turbulence during the combustion are greater than the reduced

thermal efficiency resulting from the lower compression ratio.

This investigation should be done ideally in a modern prototype development engine with

turbocharging to increase the usefulness of the data for current engine development. It should

also be conducted with more accurate fuel consumption measurement equipment to

accurately quantify the fuel consumption improvements observed in this investigation.

Transient tests should be conducted including NEDC and the upcoming RDE tests both in

engine and vehicle test rigs to understand how the different techniques affect real-world

emissions.

Tests should be conducted to investigate whether the use of the thermodenuder makes the

nucleation mode particles recorded more consistent or not, since it is designed to remove

226
them. Furthermore, the tests should be conducted on 3 consecutive days in varying order to

take day-to-day variability, resulting from changing ambient conditions and engine drift

(Beck et al., 2006) into account, since this can potentially affect the data; especially PM.

Specific further investigation for each particular research area is detailed in Appendix C1.

227
 Appendix A1: Analysis of the Uncertainties in the Recorded
Data

The data collected in this investigation is accurate; however, there will always be a limit to

how accurate the data is. Therefore, the accuracy of the main components of the engine test

rig and emissions measurements systems have been collected from the relevant manufacturer

instruction manuals and they are discussed in the following section. This builds on the work

of a previous research student in the engine group who conducted the initial analysis (Turner,

2010).

In order to control the engine dynamometer, a Mentor II Digital DC drive manufactured by

Control Technologies was used. The encoder had an accuracy which is better than 0.01% and

the drive can maintain the dynamometer speed to an accuracy of 0.1%. The VVT system had

a steady state positional accuracy of as shown on the LabView control panel for

the system. The Kistler 6041A water cooled pressure transducer had a linearity error of less

than % of the full scale. Its short term drift, as measured by a reference sensor in the

engine, when it is operating at 1500 rpm and at an engine load of 9 bar IMEP, quantified as a

pressure and as a percentage, were bar and %, respectively. The Kistler 5011B

charge amplifier used to amplify the pressure signal had a linearity error of less than %

of the full scale.

The VAF meter produced AFR readings which had an average error of 1.6% for the AFRs

used in this investigation. The inaccuracies resulted mainly from the limited resolution of the

inclined u-tube manometer (1 mm) and the limited pressure difference produced by a given

flow rate for the orifice plate, as discussed previously.

228
Regarding the PM emission measurement, the DMS500 manufactured by Cambustion Ltd.

had a particle count accuracy of . For the SMPS system, the condensing particle

counter (model 3775) manufactured by TSi had a particle count accuracy of % for

particle concentrations lower than 5x104 particles/cm3 and % for particle concentrations

larger than 5x104 but smaller than 1x107 particles/cm3. The gaseous emission measurements

obtained with the Horiba MEXA-7100DEGR were accurate to within % of the full scale.

Overall, this data highlights the limited accuracy of the equipment used in this investigation;

particularly that of the particulate emission measurements. This should be taken into

consideration in regards to the repeatability of the results reported in this thesis.

229
 Appendix B1: Data Continuity (Chapter 4)

The results from the 7.0 bar IMEP condition in the first part of the chapter are directly

comparable with the results from the hot EGR condition in the second half of the chapter,

because they were recorded at the same engine condition (7.0 bar IMEP, hot EGR). As can be

expected, there are some differences between the data. 14% EGR addition was achieved with

the hot EGR condition while only 12% EGR addition was achieved at the 7.0 bar IMEP

condition because two EGR lines rather than one were used, which enabled the flow rate of

EGR gases to be increased. As a result, the KLMBT spark timing advances (Table 4-4 vs.

Table 4-3) were greater and the EGT reductions were also greater (Figure 4-16 vs. Figure 4-

5), which resulted from the increased EGR addition and the subsequently improved KLMBT

spark timing advances.

Despite this, the MFB50 was retarded across the EGR range for the hot EGR condition in the

second part of this chapter (Figure 4-12) while it was advanced across the EGR range at the

7.0 bar IMEP condition (Figure 4-3b). This was due to the flame speed reducing in the final

10% MFB as can be seen by comparing the profiles. It is believed that the more advanced

KLMBT spark timing resulted in poorer mixture preparation, producing the observed

difference. The fuel consumption improvements were greater in the first part of the chapter

(Figure 4-4) with the 7.0 bar IMEP condition than they were in the second part of the chapter

with the hot EGR condition (Figure 4-14), despite the more advanced spark timing. It is

thought that the 1.6% error in fuel consumption measurements discussed previously resulted

in the difference observed.

The HC emission increase was greater in the first part of the chapter with the 7.0 bar IMEP

condition (Figure 4-8 vs. Figure 4-20), despite the lower EGR addition that was achieved for

230
the hot EGR condition in the second part of the chapter. It is believed that the research fuel

used in the second part of the chapter had a higher oxygen content than the pump fuel used in

the first part of the chapter, helping to suppress the HC emission increases more effectively

through improved HC oxidation. Experimental error may also account for some of the

difference observed.

231
 Appendix B2: Data Continuity (Chapter 5)

The results from the 7.0 bar IMEP condition in the first part of Chapter 4 are directly

comparable with the results from the low tumble condition in this chapter, because they were

recorded at the same engine condition (7.0 bar IMEP, low tumble intake airflow). As can be

expected, there are some differences between the data.

11% EGR addition was achieved in this chapter even though 12% EGR addition was

achieved in the first part of Chapter 4. This was deliberately done in order to achieve the

same maximum EGR addition as that for the swirl intake airflow condition in this chapter, to

enable a direct comparison to be made. The KLMBT spark timings (Table 5-2) were more

advanced than they were in the results from the 7.0 bar IMEP condition in the first part of

Chapter 4 (Table 4-3), most likely because of experimental error. The in-cylinder pressures

were increased more significantly across the EGR range (Figure 5-1b) than they were in the

first part of Chapter 4 (Figure 4-1b), because of the greater spark advance achieved (2 CAD

greater with 1% less EGR), in comparison. This is also why the calculated average in-

cylinder temperatures (Figure 5-2b) could be increased across the EGR range, while only

maintained across the EGR range in the results from the first part of Chapter 4 (Figure 4-2b).

The fuel consumption improvement (Figure 5-4) was less than that achieved in the first part

of Chapter 4 (Figure 4-4) because of the 1.6% error in fuel consumption measurements. The

EGT reductions (Figure 5-5) were greater than those achieved in the first part of Chapter 4

(Figure 4-5) because of the improved combustion phasing with EGR addition, resulting from

the greater KLMBT spark timing advances achieved. This improved the efficiency of

pressure and heat transfer into piston work in comparison, resulting in the greater EGT

reductions.

232
PM was reduced with EGR addition (Figure 5-6b) while it was increased with EGR addition

in the first part of Chapter 4 (Figure 4-6b), because of the calculated average in-cylinder

temperature increases achieved with EGR addition, while they could only be maintained

across the EGR range with EGR addition in the first part of Chapter 4. This improved the

oxidation of PM across the EGR range, resulting in the observed differences. The PN was

significantly lower than that observed in the first part of Chapter 4 mainly because of the

effect of the thermodenuder which was used with the SMPS 3936, while for the DMS 500 it

was not used. Differences in the equipment design and slight differences in the measuring

principle (i.e. differential mobility sizer method vs. scanning mobility particle sizer method)

are also likely to have contributed to the lower PN.

Indeed, (Abdul-Khalek et al., 1999) found that the number count of particles less than 50 nm

in diameter are heavily influenced by the dilution conditions such as dilution ratio,

temperature, relative humidity and residence time, which would vary slightly between the

SMPS 3936 and DMS500. While those particles larger than 50 nm are not heavily influenced

by the dilution conditions, it is believed they would still be influenced by them to a certain

degree, thus helping to explain the observed differences between the data.

233
 Appendix B3: Data Continuity (Chapter 6)

The results from the 7.0 bar IMEP condition in the first part of Chapter 4 are directly

comparable with the results from the compression ratio of 11.5 in the second half of this

chapter, because they were recorded at the same engine condition (7.0 bar IMEP,

compression ratio=11.5). As can be expected, there are some differences between the data.

The KLMBT spark timing advance with EGR addition in the second part of this chapter

(Table 6-4) is less than that achieved in the first part of Chapter 4 (Table 4-3) even though the

EGR addition was increased (2% increase). It is believed that experimental error accounts for

some of the differences observed. Also, the EGR line insulation was improved for the

experiments in this chapter, which increased the intake manifold temperature, contributing to

the decreased spark advance achieved, in comparison. The in-cylinder pressure decreased in

the results from the second part of this chapter (Figures 6-12a-c), while it increased slightly in

the first part of Chapter 4 (Figure 4-1b), with EGR addition, due to the reduced KLMBT

spark timing advances, in comparison. The increased intake manifold temperatures also

decreased the calculated average in-cylinder temperature reductions achieved with EGR

addition in the second part of this chapter, as compared to those achieved in the first part of

Chapter 4.

The PM accumulation mode decreased with EGR addition in the second part of this chapter

(Figures 6-16a-c) while it increased with EGR addition in the first part of Chapter 4 (Figure

4-6b), because of the smaller calculated average in-cylinder temperature reductions achieved

which resulted in improved PM oxidation across the EGR range. The reduced KLMBT spark

timing advance with EGR addition compared to that achieved in the first part of Chapter 4

improved mixture formation, also contributing to the observed differences in PM behaviour.

234
The PN was significantly lower than that observed in the first part of Chapter 4 mainly

because of the effect of the thermodenuder which was used with the SMPS, while for the

DMS500 it was not used. Differences in the equipment design and slight differences in the

measuring principle (i.e. differential mobility sizer method vs. scanning mobility particle

sizer method) are also likely to have contributed to the lower PN.

Again, (Abdul-Khalek et al., 1999) found that the number count of particles less than 50 nm

in diameter are heavily influenced by the dilution conditions such as dilution ratio,

temperature, relative humidity and residence time, which would vary slightly between the

SMPS and DMS500. While those particles larger than 50 nm are not heavily influenced by

the dilution conditions, it is believed they would still be influenced by them to a certain

degree, thus helping to explain the observed differences between the data.

235
 Appendix C1: Detailed Specific Further Investigation for
Each Research Area

EGR and EGR Type

A real TWC should be used in the suggested extended investigation because the simulated

TWC does not introduce any additional HCs into the engine, which has an effect on the

results. While a real TWC would convert most of the HCs, it would not be 100% efficient,

thus it would introduce some HCs back into the engine, affecting the combustion and

emissions. However, the EGR ratio variations resulting from the varying conversion

efficiency of the TWC would have to be taken into consideration in the results. A real TWC

also increases the exhaust back-pressure resulting in increased pumping losses which is

another reason why it should be used in the suggested extended investigation.

The EGR addition should also be increased using an exhaust back pressure valve, for

example, in order to quantify the effect of higher EGR ratios on the engine combustion and

emissions. The trade-off between increased exhaust back-pressure resulting from the valve

and higher EGR ratios should be studied to investigate whether it would be beneficial to

increase exhaust back-pressure in engines for the purpose of achieving increased EGR ratios.

This trade-off should also be investigated to see whether it is beneficial to remove the NO

from the EGR gases for knock reduction, considering any improvements in fuel economy

from subsequent KLMBT spark timing advances may be cancelled out by the effects of the

increased back-pressure from the TWC. EGR should be studied at a range of temperatures to

study the effect EGR temperature has on engine combustion and emissions in more detail.

The PM emissions should also be studied fully with the different EGR types; limited

equipment availability prevented this from being done in this investigation. The effect of the

236
thermodenuder on the PM emission should be more deeply investigated as well as the

differences between the DMS 500 and SMPS 3936 PM measurement equipment. Work

should also be extended to optical investigations, particularly with Planar Laser Induced

Fluorescence (PLIF), to investigate further the effect EGR has on soot formation.

Swirl and Tumble Intake Airflows

Different swirl and tumble ratios should be tested in the suggested extended investigation in

order to optimize the ratios to achieve improved engine combustion and PM emissions

behaviour. Particularly, they should be further investigated in the optical DISI engine, in

order to fully quantify the effect of swirl and tumble ratios on the initial flame propagation

and the last remaining 10% MFB, in order to optimize the combustion process. The swirl and

tumble ratios should be quantified in the optical engine at 60 CAD intervals during the intake,

compression and ignition strokes in order to fully quantify the effect of the intake airflow on

the mixture preparation and combustion. A comparison should be made between the

estimated and actual swirl and tumble ratios to quantify the accuracy of the estimation

process undertaken in this investigation.

The PM change resulting from the retarded spark timings required with the additional intake

airflows of swirl and tumble and the PM change resulting from the increased in-cylinder

swirl/tumble ratios themselves should be quantified separately in order to investigate the

effect of these parameters on PM individually. Further to this, the relationship between flame

speed and knock tendency should be investigated further to find the optimum flame speed for

minimum knock.

An exhaust back pressure valve should be used on the thermal engine to enable higher EGR

ratios to be achieved to study the effect of the laminar flame speed increases resulting from

237
the increased swirl and tumble ratios being fully suppressed by the EGR gases. Finally,

butanol and ethanol fuels should be studied with swirl and tumble intake airflows since they

have the potential to compensate for the increased engine knock resulting from the faster

laminar flame speeds.

Compression Ratio

The engine should be tested with a greater compression ratio range because some of the

effects observed, particularly regarding PM behaviour, were not significant due to the

relatively small range tested in this investigation. In addition, it is believed that Bu20

performed better than E20 in this investigation because they were only tested over a small

compression ratio range. E20 would be expected to perform better than Bu20 over a larger

compression ratio range because of its superior properties (higher octane number, higher

HoV), particularly at compression ratios higher than 11.5. The larger compression ratio range

should also be investigated in a turbocharged DISI engine along with EGR, to quantify the

improvement in compression ratio that can be achieved with the knock suppression effect of

the EGR gases. This is because compression ratios in turbocharged engines are reduced in

comparison to their equivalent NA engines (Su et al., 2014), in order to achieve a high IMEP.

Thus, EGR has the potential to enable the compression ratio to be maintained or increased

when turbocharging an engine.

Oxygenated Fuels

1-butanol should be tested with and without 5%vol ethanol pre-mixed in the gasoline fuel to

quantify exactly what effect the pre-mixed ethanol has on the engine combustion and

emissions. In addition, 1-butanol and ethanol should be tested at different blend ratios in

gasoline, including blends with both 1-butanol and ethanol in gasoline, in order to find which

blends provide the best combustion and emissions characteristics. This is important because

238
rather than competing with ethanol, 1-butanol and ethanol could be blended together in

gasoline to provide potential improvements over using just one of these fuels blended in

gasoline. The work should also be extended to optical investigations; particularly that of

laminar flame speed measurement, to quantify the effect of fuel blending on this parameter.

This is because laminar flame speed significantly influences KLMBT which consequently

affects indicated efficiency, exhaust temperature, and PM and gaseous emissions. A suitable

FTIR detector should be used when testing the oxygenated fuels to ensure that the

oxygenated hydrocarbons can successfully be measured.

239
 List of References

ABD-ALLA, G. H. 2002. Using exhaust gas recirculation in internal combustion engines: a

review. Energy Conversion and Management. 43 (8), 1027–1042, doi:10.1016/S0196-
8904(01)00091-7.

ABDUL-KHALEK, I., KITTELSON, D. AND BREAR, F. 1999. The Influence of Dilution

Conditions on Diesel Exhaust Particle Size Distribution Measurements. SAE Paper 1999-01-
1142, doi:10.4271/1999-01-1142.

AGARWAL, A. K., PANDEY, A., GUPTA, A. K., AGGARWAL, S. K. AND KUSHARI,

A. 2014. Novel Combustion Concepts for Sustainable Energy Development. New Delhi:
Springer India.

ALEIFERIS, P. G., SERRAS-PEREIRA, J. AND RICHARDSON, D. 2013. Characterisation

of flame development with ethanol, butanol, iso-octane, gasoline and methane in adirect-
injection spark-ignition engine. Fuel, 109, 256-278,
doi:http://dx.doi.org/10.1016/j.fuel.2012.12.088.

ALGER, T., HALL, M. AND MATTHEWS, R. 2000. Effects of Swirl and Tumble on In-
Cylinder Fuel Distribution in a Central Injected DISI Engine. SAE Technical Paper 2000-01-
0533, doi:10.4271/2000-01-0533.

ALGER, T., CHAUVET, T. AND DIMITROVA, Z. 2009a. Synergies between High EGR
Operation and GDI Systems. SAE Int. J. Engines 1(1):101-114, doi:10.4271/2008-01-0134.

ALGER, T. AND MANGOLD, B. 2009. Dedicated EGR: A New Concept in High Efficiency
Engines. SAE Int. J. Engines 2(1):620-631, doi:10.4271/2009-01-0694.

ALGER, T., GINGRICH, J., KHALEK, I., AND MANGOLD, B. 2010. The Role of EGR in
PM Emissions from Gasoline Engines. SAE Int. J. Fuels Lubr. 3(1):85-98, doi:10.4271/2010-
01-0353.

ALGER, T., GUKELBERGER, R., GINGRICH, J. AND MANGOLD, B. 2015. The Impact
of Cooled EGR on Peak Cylinder Pressure in a Turbocharged, Spark Ignited Engine. SAE
Int. J. Engines 8(2):455-463, doi:10.4271/2015-01-0744.

AL-HASAN, M. 2003. Effect of ethanol–unleaded gasoline blends on engine performance

and exhaust emission. Energy Conversion and Management. 44 (9), 1547–1561,
doi:10.1016/S0196-8904(02)00166-8.

AMANN, M. AND ALGER, T. 2012. Lubricant Reactivity Effects on Gasoline Spark

Ignition Engine Knock. SAE Int. J. Fuels Lubr. 5(2):760-771, doi:10.4271/2012-01-1140.

ANALIIT AA. 2015. Gasoline EN 228. Available:

http://www.analiit.ee/index.php?option=com_content&view=article&id=48&lang=en. Last
accessed 27th February 2015.

240
ANDERSON, J., LEONE, T., SHELBY, M., WALLINGTON, T., BIZUB, J. J. AND
FOSTER, M. 2012a. Octane Numbers of Ethanol-Gasoline Blends: Measurements and Novel
Estimation Method from Molar Composition. SAE Technical Paper 2012-01-1274,
doi:10.4271/2012-01-1274.

ANDERSON, J. O., THUNDIYIL, J. G. AND STOLBACH, A. 2012b. Clearing the Air: A

Review of the Effects of Particulate Matter Air Pollution on Human Health. Journal of
Medical Toxicology, 8, 166-175, doi:10.1007/s13181-011-0203-1.

ARCOUMANIS, C. AND KAMIMOTO, T. 2009. Flow and Combusiton in Reciprocating

Engines. Berlin: Springer-Verlag.

BALABIN, R. M., SYUNYAEV, R. Z. AND KARPOV, S. A. 2007. Molar enthalpy of

vaporization of ethanol–gasoline mixtures and their colloid state. Fuel, 86, 323-327,
doi:10.1016/j.fuel.2006.08.008.

BAYRAKTAR, H. 2005. Experimental and theoretical investigation of using gasoline–

ethanol blends in spark-ignition engines. Renewable Energy. 30 (11), 1733–1747,
doi:10.1016/j.renene.2005.01.006.

BECK, C., STEVENSON, P. AND ZIMAN, P. 2006. The Impact of Gasoline Octane on Fuel
Economy in Modern Vehicles. SAE Technical Paper 2006-01-3407, doi:10.4271/2006-01-
3407.

BENNETT, S. 2010. Diesel Engines- Modern Diesel Technology. New York: Delmar.

BENNETT, S. 2013. Medium/Heavy Duty Truck Engines, Fuel & Computerized

Management Systems. New York: Delmar.

BIELACZYC, P., SZCZOTKA, A. AND WOODBURN, J. 2014. The Impact of Fuel Ethanol
Content on Particulate Emissions from Light-Duty Vehicles Featuring Spark Ignition
Engines. SAE Int. J. Fuels Lubr. 7(1):224-235, doi:10.4271/2014-01-1463.

BOCK, B., BELL, A. AND FLOWEDAY, G. 2013. Investigation into the Influence of
Charge Cooling and Autoignition Chemistry on the Greater Knock Resistance of Ethanol
over Iso-octane. SAE Technical Paper 2013-01-2615, doi:10.4271/2013-01-2615.

BP. 2015. Oil reserves. Available: http://www.bp.com/en/global/corporate/about-bp/energy-

economics/statistical-review-of-world-energy/review-by-energy-type/oil/oil-reserves.html.
Last accessed 15th Sep 2015.

BRADLEY, D., MORLEY, C. AND WALMSLEY, H. 2004. Relevance of Research and

Motor Octane Numbers to the Prediction of Engine Autoignition. SAE Technical Paper 2004-
01-1970, doi:10.4271/2004-01-1970.

BRUSIANI, F., FALFARI, S. AND CAZZOLI, G. 2014. Tumble Motion Generation in

Small Gasoline Engines: A New Methodological Approach for the Analysis of the Influence
of the Intake Duct Geometrical Parameters. Energy Procedia. 45, 997–1006,
doi:10.1016/j.egypro.2014.01.105.

241
BURTON, G., HOLMAN, J., LAZONBY, J., PILLING, G. AND WADDINGTON, D. 2000.
Chemical Storylines. Oxford: Heinemann.

BURTSCHER, H. 2005. Physical characterization of particulate emissions from diesel

engines: a review. Journal of Aerosol Science, 36, 896-932,
doi:10.1016/j.jaerosci.2004.12.001.

CAIRNS, A., BLAXILL, H. AND IRLAM, G. 2006. Exhaust Gas Recirculation for
Improved Part and Full Load Fuel Economy in a Turbocharged Gasoline Engine. SAE
Technical Paper 2006-01-0047, doi:10.4271/2006-01-0047.

CAMBUSTION LTD. 2011. DMS500 Fast Particulate Spectrometer with Heated Sample
Line High Ratio Diluter User Manual Version 3.5 Software Version UI v4.11.

CASTAGNÉ, M., DUMAS, J., HENRIOT, S. AND LAFOSSAS, F. A. 2003. New Knock
Localization Methodology for SI Engines. SAE Technical Paper 2003-01-1118,
doi:10.4271/2003-01-1118.

CATAPANO, F., DI IORIO, S., SEMENTA, P. AND VAGLIECO, B. 2014.

Characterization of Ethanol-Gasoline Blends Combustion processes and Particle Emissions in
a GDI/PFI Small Engine. SAE Technical Paper 2014-01-1382, doi:10.4271/2014-01-1382.

CENGEL, Y. A. AND BOLES, M. A. 2007. Thermodynamics An Engineering Approach.

New York: McGraw-Hill.

CHECKEL, M. AND DALE, J. 1989. Pressure Trace Knock Measurement in a Current S.I.
Production Engine. SAE Technical Paper 890243, doi:10.4271/890243.

CHEN, R., OKAZUMI, R., NISHIDA, K. AND OGATA, Y. 2015. Effect of Ethanol Ratio
on Ignition and Combustion of Ethanol-Gasoline Blend Spray in DISI Engine-Like
Condition. SAE Int. J. Fuels Lubr. 8(2):264-276, 2015, doi:10.4271/2015-01-0774.

CHENG, W. K., SUMMER, T AND COLLINGS, N. 1998. The Fast-response Flame

Ionization Detector. Progress in Energy and Combustion Science, 24, 89-124,
doi:10.1016/S0360-1285(97)00025-7.

CHEVRON. 2009. Motor Gasolines Technical Review. Available:

http://www.chevron.com/documents/pdf/MotorGasTechReview.pdf. Last accessed 27th
February 2015.

COSTAGLIOLA, M. A., DE SIMIO, L., IANNACCONE, S. AND PRATI, M. V. 2013.

Combustion efficiency and engine out emissions of a S.I. engine fueled with alcohol/gasoline
blends. Applied Energy, 111, 1162-1171,
doi:http://dx.doi.org/10.1016/j.apenergy.2012.09.042.

DANIEL, R., TIAN, G., XU, H., WYSZYNSKI, M.L., WU, X. AND HUANG, Z. 2011.
Effect of spark timing and load on a DISI engine fuelled with 2,5-dimethylfuran. Fuel, 90,
449-458, doi:10.1016/j.fuel.2010.10.008.

242
DANIEL, R., XU, H., WANG, C., RICHARDSON, D. AND SHUAI, S. 2012a. Combustion
performance of 2,5-dimethylfuran blends using dual-injection compared to direct-injection in
a SI engine. Applied Energy, 98, 59-68, doi:http://dx.doi.org/10.1016/j.apenergy.2012.02.073

DANIEL, R., WANG, C., XU, H., TIAN, G. AND RICHARDSON, D. 2012b. Dual-
Injection as a Knock Mitigation Strategy Using Pure Ethanol and Methanol. SAE Int. J. Fuels
Lubr. 5(2):772-784, doi:10.4271/2012-01-1152.

DANIEL, R., WANG, C., XU, H. AND TIAN, G. 2012c. Effects of Combustion Phasing,
Injection Timing, Relative Air-Fuel Ratio and Variable Valve Timing on SI Engine
Performance and Emissions using 2,5-Dimethylfuran. SAE Int. J. Fuels Lubr. 5(2):855-866,
doi:10.4271/2012-01-1285.

DANIEL, R. 2012d. Combustion and emissions performance of oxygenated fuels in a

modern spark ignition engine. PhD Thesis, University of Birmingham.

DANIEL, R., TIAN, G., XU, H. AND SHUAI, S. 2012e. Ignition timing sensitivities of
oxygenated biofuels compared to gasoline in a direct-injection SI engine. Fuel. 99, 72-82,
doi:http://dx.doi.org/10.1016/j.fuel.2012.01.053.

DAVIS, G. AND BORGNAKKE, C. 1982. The Effect of In-Cylinder Flow Processes (Swirl,
Squish and Turbulence Intensity) on Engine Efficiency — Model Predictions. SAE Technical
Paper 820045, doi:10.4271/820045.

DE ABRANTES, R., DE ASSUNÇÃO, J. V., PESQUERO, C. R., BRUNS, R. E. AND

NÓBREGA, R. P. 2009. Emission of polycyclic aromatic hydrocarbons from gasohol and
ethanol vehicles. Atmospheric Environment. 43 (3), 648–654,
doi:10.1016/j.atmosenv.2008.10.014.

DE O. CARVALHO, L., DE MELO, T. AND DE AZEVEDO CRUZ NETO, R. 2012.

Investigation on the Fuel and Engine Parameters that Affect the Half Mass Fraction Burned
(CA50) Optimum Crank Angle. SAE Technical Paper 2012-36-0498, doi:10.4271/2012-36-
0498.

DELPHI 2012. Worldwide Emissions Standards - Passenger Cars and Light Duty Vehicles.

DENG, B., YANG, J. AND ZHANG, D. 2013a. The challenges and strategies of butanol
application in conventional engines: The sensitivity study of ignition and valve timing.
Applied Energy, 108, 248-260, doi:http://dx.doi.org/10.1016/j.apenergy.2013.03.018.

DENG, B., FU, J., ZHANG, D., YANG, J., FENG, R., LIU, J., LI, K. AND LIU, X. 2013b.
The heat release analysis of bio-butanol/gasoline blends on a high speed SI (spark ignition)
engine. Energy. 60, 230–241, doi:10.1016/j.energy.2013.07.055.

DI IORIO, S., LAZZARO, M., SEMENTA, P., VAGLIECO, B. AND CATAPANO, F.

2011. Particle Size Distributions from a DI High Performance SI Engine Fuelled with
Gasoline-Ethanol Blended Fuels. SAE Technical Paper 2011-24-0211, doi:10.4271/2011-24-
0211.

243
DIANA, S., GIGLIO, V., IORIO, B. AND POLICE, G. 1996. A Strategy to Improve the
Efficiency of Stoichiometric Spark Ignition Engines. SAE Technical Paper 961953,
doi:10.4271/961953.

DIRECTIVE 2009/28/EC. 2009. On the promotion of the use of energy from renewable
sources. Official Journal of the European Union; 2009.

DUNN-RANKIN, D. 2008. Lean Combustion- Technology and Control. Burlington, MA:

Elsevier.

EASTOP, T. D. AND MCCONKEY, A. 1993. Applied Thermodynamics for Engineering

Technologists, London: Longman.

EASTWOOD, P. 2008. Particulate Emissions from Vehicles, Chichester, John Wiley & Sons
Ltd.

ENDRES, H., NEUΒER, H. AND WURMS, R. 1992. Influence of Swirl and Tumble on
Economy and Emissions of Multi Valve SI Engines. SAE Technical Paper 920516,
doi:10.4271/920516.

ELFASAKHANY, A. 2014. Experimental study on emissions and performance of an internal

combustion engine fueled with gasoline and gasoline/n-butanol blends. Energy Conversion
and Management. 88, 277–283, doi:10.1016/j.enconman.2014.08.031.

EUROPEAN COUNCIL. 2016. Vehicle emissions in real driving conditions: Council gives
green light to second package. Available: http://www.consilium.europa.eu/en/press/press-
releases/2016/02/12-vehicle-emissions-in-real-driving-conditions-2nd-package/. Last
accessed 6th Apr 2016.

FENG, R., YANG, J., ZHANG, D., DENG, B., FU, J., LIU, J. AND LIU, X. 2013.
Experimental study on SI engine fuelled with butanol–gasoline blend and H2O addition.
Energy Conversion and Management. 74, 192–200, doi:10.1016/j.enconman.2013.05.021.

FLOCH, A., VAN FRANK, J. AND AHMED, A. 1995. Comparison of the Effects of Intake-
Generated Swirl and Tumble on Turbulence Characteristics in a 4-Valve Engine. SAE
Technical Paper 952457, doi:10.4271/952457.

DENG, B., FU, J., ZHANG, D., YANG, J., FENG, R., LIU, J., LI, K. AND LIU, X. 2013b.
The heat release analysis of bio-butanol/gasoline blends on a high speed SI (spark ignition)
engine. Energy. 60, 230–241, doi:10.1016/j.energy.2013.07.055.

FOUNTAIN, H. 2012. Corn Ethanol Makers Weigh Switch to Butanol. Available:

http://www.nytimes.com/2012/10/24/business/energy-environment/weighing-butanol-as-an-
alternative-to-ethanol.html?_r=0. Last accessed 27th February 2015.

FRANCQUEVILLE, L. AND MICHEL, J. 2014. On the Effects of EGR on Spark-Ignited

Gasoline Combustion at High Load. SAE Int. J. Engines 7(4):1808-1823, doi:10.4271/2014-
01-2628.

244
FRENNET, A. AND BASTIN, J.-M. 1995. Catalysis and Automotive Pollution Control III.
Amsterdam: Elsevier.

FU, J., DENG, B., WANG, Y., YANG, J., ZHANG, D., XU, Z. AND LIU, J. 2014.
Numerical study and correlation development on laminar burning velocities of n-butanol, iso-
octane and their blends: Focusing on diluent and blend ratio effects. Fuel. 124, 102–112,
doi:10.1016/j.fuel.2014.01.092.

FUJIMOTO, M. AND TABATA, M. 1993. Effect of Swirl Rate on Mixture Formation in a

Spark Ignition Engine Based on Laser 2-D Visualization Techniques. SAE Technical Paper
931905, doi:10.4271/931905.

GILL, S., TURNER, D., TSOLAKIS, A. AND YORK, A. 2011. Understanding the Role of
Filtered EGR on PM Emissions. SAE Technical Paper 2011-01-2080, doi:10.4271/2011-01-
2080.

GALLONI, E., FONTANA, G. AND PALMACCIO, R. 2013. Effects of exhaust gas recycle
in a downsized gasoline engine. Applied Energy. 105, 99–107,
doi:10.1016/j.apenergy.2012.12.046.

GOLDING, B. T. AND WATSON, W. P. 1999. Possible mechanisms of carcinogenesis after

exposure to benzene. IARC scientific publications, 150, 75-88, doi: unavailable.

GOSWAMI, A., VASHIST, S. AND NAYYAR, A. 2015. Effect of Compression Ratio on

the Performance Characteristics of Spark Ignition Engine Fueled with Alternative Fuels: A
Review. SAE Technical Paper 2015-01-0766, doi:10.4271/2015-01-0766.

GOV UK. 2011. Public attitudes towards climate change and the impact of transport: 2010.
Available:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/8916/climate-
change-2010-report.pdf. Last accessed 27th February 2015.

GRAHAM, L. A., BELISLE, S. L. AND BAAS, C.-L. 2008. Emissions from light duty
gasoline vehicles operating on low blend ethanol gasoline and E85. Atmospheric
Environment. 42 (19), 4498–4516, doi:10.1016/j.atmosenv.2008.01.061.

GRANDIN, B., ÅNGSTRÖM, H., STÅLHAMMAR, P. AND OLOFSSON, E. 1998. Knock

Suppression in a Turbocharged SI Engine by Using Cooled EGR. SAE Technical Paper
982476, doi:10.4271/982476.

GREEN CAR CONGRESS. 2006. Mercedes-Benz Premiers New Gasoline Direct Injection
System for More Power and Lower Fuel Consumption. Available:
http://www.greencarcongress.com/2006/02/mercedesbenz_pr.html. Last accessed 26th Feb
2015.

GREEN CAR CONGRESS. 2013. 2013 SAE International High Octane Fuels Symposium:
the potential for high octane fuels (Part 1). Available:
http://www.greencarcongress.com/2013/01/hofs-20130131.html. Last accessed 25th Feb
2015.

245
GU, X., HUANG, Z., CAI, J., GONG, J., WU, X. AND LEE, C.-F. 2012. Emission
characteristics of a spark-ignition engine fuelled with gasoline-n-butanol blends in
combination with EGR. Fuel, 93, 611-617, doi:10.1016/j.fuel.2011.11.040.

GUMBLETON, J., NIEPOTH, G. AND CURRIE, J. 1976. Effect of Energy and Emission
Constraints on Compression Ratio. SAE Technical Paper 760826, doi:10.4271/760826.

GUPTA, H. N. 2013. Fundamentals of Internal Combustion Engines. 2nd ed. Delhi, India:
Phi Learning.

HAMILTON, L., ROSTEDT, M., CATON, P. AND COWART, J. 2008. Pre-Ignition

Characteristics of Ethanol and E85 in a Spark Ignition Engine. SAE Int. J. Fuels Lubr.
1(1):145-154, doi:10.4271/2008-01-0321.

HARADA, J., TOMITA, T., MIZUNO, H., MASHIKI, Z. AND ITO, Y. 1997. Development
of Direct Injection Gasoline Engine. SAE Technical Paper 970540, doi:10.4271/970540.

HARRINGTON, J. A. AND SHISHU, R.C. 1973. A Single-Cylinder Engine Study of the

Effects of Fuel Type, Fuel Stoichiometry, and Hydrogen-to-Carbon Ratio and CO, NO, and
HC Exhaust Emissions. SAE 730476, doi:10.4271/730476.

HE, B.-Q., WANG, J.-X., HAO, J.-M., YAN, X.-G. AND XIAO, J.-H. 2003. A study on
emission characteristics of an EFI engine with ethanol blended gasoline fuels. Atmospheric
Environment. 37 (7), 949–957, doi:10.1016/S1352-2310(02)00973-1.

HE, X., IRELAND, J., ZIGLER, B., RATCLIFF, M. A., KNOLL, K. E., ALLEMAN, T. L.
AND TESTER, J. T. 2010. The Impacts of Mid-level Biofuel Content in Gasoline on SIDI
Engine-out and Tailpipe Particulate Matter Emissions. SAE Technical Paper 2010-01-2125,
doi:10.4271/2010-01-2125.

HE, B.-Q., LIU, M.-B. AND ZHAO, H. 2015. Comparison of combustion characteristics of
n-butanol/ ethanol–gasoline blends in a HCCI engine. Energy Conversion and Management.
95, 101–109, doi:10.1016/j.enconman.2015.02.019.

HEDGE, M., WEBER, P., GINGRICH, J., ALGER, T. AND KHALEK, I. 2011. Effect of
EGR on Particle Emissions from a GDI Engine. SAE Int. J. Engines 4(1):650-666,
doi:10.4271/2011-01-0636.

HEFFEL, J. W. 2003. NOx emission and performance data for a hydrogen fueled internal
combustion engine at using exhaust gas recirculation. International Journal of Hydrogen
Energy. 28, 901–908, doi:10.1016/S0360-3199(02)00157-X.

HEYWOOD, J. 1988, Internal Combustion Engine Fundamentals, Singapore, McGraw-Hill.

HEYWOOD, J. AND WELLING, O. 2009. Trends in Performance Characteristics of Modern

Automobile SI and Diesel Engines. SAE Int. J. Engines 2(1):1650-1662, doi:10.4271/2009-
01-1892.

HIROOKA, H., MORI, S. AND SHIMIZU, R. 2004. Effects of High Turbulence Flow on
Knock Characteristics. SAE Technical Paper 2004-01-0977, doi:10.4271/2004-01-0977.

246
HOFFMEYER, H., MONTEFRANCESCO, E., BECK, L., WILLAND, J., ZIEBART, F.
AND MAUSS, F. 2009. CARE – CAtalytic Reformated Exhaust Gases in Turbocharged
DISI-Engines, SAE Int. J. Fuels Lubr. 2(1):139-148, doi:10.4271/2009-01-0503.

HOUNTALAS, D. T. AND ANESTIS, A. 1998. Effect of pressure transducer position on

measured cylinder pressure diagram of high speed diesel engines. Energy Conversion and
Management, 39, 589–607, doi:10.1016/S0196-8904(97)10009-7.

HSIEH, W.-D., CHEN, R.-H., WU, T.-L. AND LIN, T.-H. 2002. Engine performance and
pollutant emission of an SI engine using ethanol–gasoline blended fuels. Atmospheric
Environment. 36 (3), 403–410, doi:10.1016/S1352-2310(01)00508-8.

HU, E., HU, X., WANG, X., XU, Y., DEARN, K. D. AND XU, H. 2012. On the
fundamental lubricity of 2,5-dimethylfuran as a synthetic engine fuel. Tribology
International, 55, 119–125, doi:http://dx.doi.org/10.1016/j.triboint.2012.06.005

HULL, A., GOLUBKOV, I., KRONBERG, B., MARANDZHEVA, T. AND VAN STAM, J.
2005. An alternative fuel for spark ignition engines. International Journal of Engine
Research, 7 , 203-214, doi:10.1243/14680874JER02504.

IKEYA, K., TAKAZAWA, M., YAMADA, T., PARK, S. AND TAGISHI, R. 2015. Thermal
Efficiency Enhancement of a Gasoline Engine. SAE Int. J. Engines 8(4), doi:10.4271/2015-
01-1263.

ISERMANN, R. 2014. Engine Modeling and Control- Modeling and Electronic Management
of Internal Combustion Engines. New York: Springer-Verlag.

INSTITUTION OF MECHANICAL ENGINEERS. 2013. Internal Combustion Engines:

Performance, Fuel Economy and Emissions. Cambridge: Woodhead Publishing.

IRIMESCU, A., MARCHITTO, L., MEROLA, S. S., TORNATORE, C. AND

VALENTINO, G. 2015. Combustion process investigations in an optically accessible DISI
engine fuelled with n-butanol during part load operation. Renewable Energy. 77, 363–376,
doi:10.1016/j.renene.2014.12.029.

JOHNSON, T. 2010. Review of CO2 Emissions and Technologies in the Road Transportation
Sector. SAE Int. J. Engines 3(1):1079-1098, doi:10.4271/2010-01-1276.

KALGHATGI, G. 2013. Fuel/Engine Interactions, Warrendale (USA), SAE International.

KALGHATGI, G. 2014. The outlook for fuels for internal combustion engines. International
Journal of Engine Research. Special Article, 1-17, doi: 10.1177/1468087414526189.

KARAVALAKIS, G., SHORT, D., HAJBABAEI, M., VU, D., VILLELA, M., RUSSELL,
R., DURBIN, T. AND ASA-AWUKU, A. 2013. Criteria Emissions, Particle Number
Emissions, Size Distributions, and Black Carbon Measurements from PFI Gasoline Vehicles
Fuelled with Different Ethanol and Butanol Blends. SAE Technical Paper 2013-01-1147,
doi:10.4271/2013-01-1147.

247
KARAVALAKIS, G., SHORT, D., VU, D., VILLELA, M., ASA-AWUKU, A. AND
DURBIN, T. D. 2014. Evaluating the regulated emissions, air toxics, ultrafine particles, and
black carbon from SI-PFI and SI-DI vehicles operating on different ethanol and iso-butanol
blends. Fuel. 128, 410–421, doi:10.1016/j.fuel.2014.03.016.

KAWABATA, Y., SAKONJI, T. AND AMANO, T. 1999. The Effect of NOx on Knock in
Spark-ignition Engines. SAE Technical Paper 1999-01-0572, doi:10.4271/1999-01-0572.

KESKIN, A. AND GÜRÜ, M. 2011. The Effects of Ethanol and Propanol Additions Into
Unleaded Gasoline on Exhaust and Noise Emissions of a Spark Ignition Engine. Energy
Sources, Part A: Recovery, Utilization, and Environmental Effects. 33 (23), 2194-2205, doi:
10.1080/15567030903530558.

KIM, S. AND KIM, S. 1995. Effects of Swirl and Spark Plug Shape on Combustion
Characteristic in a High Speed Single-Shot Visualized SI Engine. SAE Technical Paper
951003, doi:10.4271/951003.

KIM, M., LEE, S. AND KIM, W. 2006. Tumble Flow Measurements Using Three Different
Methods and its Effects on Fuel Economy and Emissions. SAE Technical Paper 2006-01-
3345, 2006, doi:10.4271/2006-01-3345.

KITTELSON, D. B. 1998. Engines and nanoparticles: a review. Journal of Aerosol Science,

29, 575–588, doi:10.1016/S0021-8502(97)10037-4.

KRAMER, F., SCHWARZ, C. AND WITT, A. 2000. Effect of Compression Ratio on the
Combustion of a Pressure Charged Gasoline Direct Injection Engine. SAE Technical Paper
2000-01-0250, doi:10.4271/2000-01-0250.

KUMANO, K. AND YAMAOKA, S. 2014. Analysis of Knocking Suppression Effect of

Cooled EGR in Turbo-Charged Gasoline Engine. SAE Technical Paper 2014-01-1217,
doi:10.4271/2014-01-1217.

LATTIMORE, T., WANG, C., XU, H., WYSZYNSKI, M. L. AND SHUAI, S. 2016a.
Investigation of EGR Effect on Combustion and PM Emissions in a DISI Engine, Applied
Energy, 161, 2016, 256-267, doi:10.1016/j.apenergy.2015.09.080.

LATTIMORE, T., HERREROS, J. M., XU, H. AND SHUAI, S. 2016b. Investigation of

Compression Ratio and Fuel Effect on Combustion and PM Emissions in a DISI Engine,
Fuel, 169, 68-78, doi: 10.1016/j.fuel.2015.10.044.

LECOINTE, B. AND MONNIER, G. 2003. Downsizing a Gasoline Engine Using

Turbocharging with Direct Injection. SAE Technical Paper 2003-01-0542, doi:10.4271/2003-
01-0542.

LEE, S., TONG, K., QUAY, B., ZELLO, J. AND SANTAVICCA, D. A. 2000. Effects of
Swirl and Tumble on Mixture Preparation During Cold Start of a Gasoline Direct-Injection
Engine. SAE Technical Paper 2000-01-1900, doi:10.4271/2000-01-1900.

248
LEE, K., BAE, C. AND KANG, K. 2007. The effects of tumble and swirl flows on flame
propagation in a four-valve S.I. engine. Applied Thermal Engineering. 27 (11-12), 2122–
2130, doi:10.1016/j.applthermaleng.2006.11.011.

LEPPARD, W. 1982. Individual-Cylinder Knock Occurence and Intensity in Multicylinder

Engines. SAE Technical Paper 820074, doi:10.4271/820074.

LEWIS, A., AKEHURST, S., TURNER, J., PATEL, R. AND POPPLEWELL, A. 2014.
Observations on the Measurement and Performance Impact of Catalyzed vs. Non Catalyzed
EGR on a Heavily Downsized DISI Engine. SAE Int. J. Engines 7(1):458-467,
doi:10.4271/2014-01-1196.

LI, Y., ZHAO, H., PENG, Z., AND LADOMMATOS, N. 2001. Analysis of Tumble and
Swirl Motions in a Four-Valve SI Engine. SAE Technical Paper 2001-01-3555,
doi:10.4271/2001-01-3555.

LIIVA, P., VALENTINE, J., COBB, J., AND ACKER, W. 1992. Use of Multiple Pressure
Transducers to Find In-Cylinder Knock Location. SAE Technical Paper 922368,
doi:10.4271/922368.

LIU, H., LI, S., ZHENG, Z., XU, J. AND YAO, M. 2013. Effects of n -butanol, 2-butanol,
and methyl octynoate addition to diesel fuel on combustion and emissions over a wide range
of exhaust gas recirculation (EGR) rates. Applied Energy. 112, 246-256,
doi:10.1016/j.apenergy.2013.06.023.

LUSZCZ, P. 2009. Combustion diagnostics in Homogeneous Charge Compression Ignition

optical and thermal single cylinder engines. PhD Thesis, University of Birmingham.

MA, X., ZHANG, F., XU, H. AND SHUAI, S. 2014. Throttleless and EGR-controlled
stoichiometric combustion in a diesel–gasoline dual-fuel compression ignition engine, Fuel,
115, 765–777, doi:10.1016/j.fuel.2013.07.052.

MAJI, S., BABU, M. AND GUPTA, N. 2001. A Single Cylinder Engine Study of Power,
Fuel Consumption and Exhaust Emissions with Ethanol. SAE Technical Paper 2001-28-0029,
doi:10.4271/2001-28-0029.

MASUM, B. M., MASJUKI, H. H., KALAM, M.A., RIZWANUL FATTAH, I.M.,

PALASH, S.M. AND ABEDIN, M.J. 2013. Effect of ethanol–gasoline blend on NOx
emission in SI engine. Renewable and Sustainable Energy Reviews. 24, 209–222,
doi:10.1016/j.rser.2013.03.046.

MASUM, B. M., KALAM, M. A., MASJUKI, H. H., PALASH, S. M. AND FATTAH, M.

R. 2014. Performance and emission analysis of a multi cylinder gasoline engine operating at
different alcohol–gasoline blends. RSC Advances. 4, 27898-27904, doi:10.1039/c4ra04580g.

MATHIS, U., MOHR, M. AND FORSS, A. M. 2005. Comprehensive particle

characterization of modern gasoline and diesel passenger cars at low ambient temperatures.
Atmospheric Environment, 39, 107-117, doi:10.1016/j.atmosenv.2004.09.029.

249
MATSUSHITA, S., INOUE, T., NAKANISHI, K., OKUMURA, T. AND ISOGAI, K. 1985.
Effects of Helical Port with Swirl Control Valve on the Combustion and Performance of S. I.
Engine. SAE Technical Paper 850046, doi:10.4271/850046.

MATTIMARICQ, M. 2012. Soot formation in ethanol/gasoline fuel blend diffusion flames.

Combustion and Flame, 159, 170-180, doi:10.1016/j.combustflame.2011.07.010.

MEHTA, D., ALGER, T., HALL, M., MATTHEWS, R. D., AND NG, H. 2001. Particulate
Characterization of a DISI Research Engine using a Nephelometer and In-Cylinder
Visualization‖. SAE Technical Paper 2001-01-1976, doi:10.4271/2001-01-1976.

MIKULEC, A., KENT, J. AND TABACZYNSKI, R. 1988. The Effect of Swirl on

Combustion in a Pancake Chamber Spark Ignition Engine: The Case of Constant Inducted
Kinetic Energy. SAE Technical Paper 880200, doi:10.4271/880200.

MITSUBISHI MOTORS. 1999. GDI Engine Production Tops 400,000. Available:

http://www.mitsubishi-motors.com/en/corporate/pressrelease/corporate/detail791.html. Last
accessed 26th Feb 2015.

MITTAL, V., REVIER, B. AND HEYWOOD, J. 2007. Phenomena that Determine Knock
Onset in Spark-Ignition Engines. SAE Technical Paper 2007-01-0007, doi:10.4271/2007-01-
0007.

NAGAYAMA, I., ARAKI, Y. AND IIOKA, Y. 1977. Effects of Swirl and Squish on S.I.
Engine Combustion and Emission. SAE Technical Paper 770217, doi:10.4271/770217.

NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY. 2011a. 1-Butanol.

Available: http://webbook.nist.gov/cgi/cbook.cgi?ID=C71363&Mask=4. Last accessed 16th
Jan 2015.

NATIONAL INSTITUTE OF STANDARDS AND TECHNOLOGY. 2011b. 1-Ethanol.

Available: http://webbook.nist.gov/cgi/cbook.cgi?ID=C64175&Mask=4. Last accessed 16th
Jan 2015.

NIASS, T., AMER, A., XU, W., VOGEL, S., HORTMANN, K. K., ADOMEIT, P. AND
BRASSAT, A. 2012. Butanol Blending - a Promising Approach to Enhance the
Thermodynamic Potential of Gasoline - Part 1. SAE Int. J. Fuels Lubr. 5(1):265-273,
doi:10.4271/2011-01-1990.

NIEMAN, D., DEMPSEY, A., AND REITZ, R. 2012. Heavy-Duty RCCI Operation Using
Natural Gas and Diesel. SAE Int. J. Engines 5(2):270-285, doi:10.4271/2012-01-0379.

NISHIDA, M., INOUE, N., SUZUKI, H., AND KUMAGAI, S. 1988. Closed Loop Control
of the EGR Rate Using the Oxygen Sensor. SAE Technical Paper 880133,
doi:10.4271/880133.

OJAPAH, M., ZHAO, H. AND ZHANG, Y. 2014. Effects of Ethanol on Performance and
Exhaust Emissions from a DI Spark Ignition Engine with Throttled and Unthrottled
Operations. SAE Technical Paper 2014-01-1393, doi:10.4271/2014-01-1393.

250
OKAMOTO, K., ICHIKAWA, T., SAITOH, K., OYAMA, K., HIRAYA, K. AND
URUSHIHARA, T. 2003. Study of Antiknock Performance Under Various Octane Numbers
and Compression Ratios in a DISI Engine. SAE Technical Paper 2003-01-1804,
doi:10.4271/2003-01-1804.

OPPENHEIM, A. K. 2004. Combustion in Piston Engines, Berlin, Springer-Verlag.

OWEN, K. AND COLEY, T. 1995. Automotive Fuels Reference Book, Warrendale, Society
of Automotive Engineers.

PAN, M., SHU, G., WEI, H., ZHU, T., LIANG, Y. AND LIU, C. 2014. Effects of EGR,
compression ratio and boost pressure on cyclic variation of PFI gasoline engine at WOT
operation. Applied Thermal Engineering, 64, 491-498,
doi:10.1016/j.applthermaleng.2013.11.013.

PARAG, S. AND RAGHAVAN, V. 2009. Experimental investigation of burning rates of

pure ethanol and ethanol blended fuels. Combustion and Flame, 156, 997-1005,
doi:10.1016/j.combustflame.2008.10.011.

PARSONS, D., AKEHURST, S. AND BRACE, C. 2015. The potential of catalysed exhaust
gas recirculation to improve high-load operation in spark ignition engines. International
Journal of Engine Research. 16 (4), 592-605, doi:10.1177/1468087414554628.

PAWLOWSKI, A. AND SPLITTER, D. 2015. SI Engine Trends: A Historical Analysis with

Future Projections. SAE Technical Paper 2015-01-0972, doi:10.4271/2015-01-0972.

PCB Piezeotronics. 2013. Introduction to Piezoelectric Pressure Sensors. Available:

http://www.pcb.com/TechSupport/Tech_Pres.aspx. Last accessed 4th Mar 2015.

PIOCK, W., HOFFMANN, G., BERNDORFER, A., SALEMI, P. AND FUSSHOELLER, B.

2011. Strategies Towards Meeting Future Particulate Matter Emission Requirements in
Homogeneous Gasoline Direct Injection Engines. SAE Int. J. Engines 4(1):1455-1468,
doi:10.4271/2011-01-1212.

POITRAS, M., ROSENBLATT, D. AND GOODMAN, J. 2015. Impact of Ethanol and

Isobutanol Gasoline Blends on Emissions from a Closed-Loop Small Spark-Ignited Engine.
SAE Technical Paper 2015-01-1732, doi:10.4271/2015-01-1732.

RAJPUT, R. K. 2005. A Textbook of Internal Combustion Engines. New Delhi: Laxmi

Publications.

RAMADHAS, A. S. 2011. Alternative Fuels for Transportation. Boca Raton, FL: CRC Press.

RATNAK, S., KATORI, K., KUSAKA, J., DAISHO, Y., YOSHIMURA, K. AND
KENJIRO, N. 2015. Computational Study to Improve Thermal Efficiency of Spark Ignition
Engine. SAE Technical Paper 2015-01-0011, doi:10.4271/2015-01-0011.

REITZE JR., A. W. 2001. Air Pollution Control Law: Compliance & Enforcement.
Washington D. C.: Environmental Law Institute.

251
RHODES, D. AND KECK, J. 1985. Laminar Burning Speed Measurements of Indolene-Air-
Diluent Mixtures at High Pressures and Temperatures. SAE Technical Paper 850047,
doi:10.4271/850047.

ROBERT BOSCH GMBH. 2006. Gasoline-Engine Management, 3rd edition, Chichester,

John Wiley & Sons Ltd.

ROTH, D., KELLER, P. AND BECKER, M. 2010. Requirements of External EGR Systems
for Dual Cam Phaser Turbo GDI Engines. SAE Technical Paper 2010-01-0588,
doi:10.4271/2010-01-0588.

ROTHE, M., HEIDENREICH, T., SPICHER, U. AND SCHUBERT, A. 2006. Knock

Behavior of SI-Engines: Thermodynamic Analysis of Knock Onset Locations and Knock
Intensities. SAE Technical Paper 2006-01-0225, doi:10.4271/2006-01-0225.

SERRAS-PEREIRA, J., ALEIFERIS, P. G. AND RICHARDSON, D. 2013. An Analysis of

the Combustion Behavior of Ethanol, Butanol, Iso-Octane, Gasoline, and Methane in a
Direct-Injection Spark-Ignition Research Engine. Combustion Science and Technology. 185
(3), 484-513, doi:10.1080/00102202.2012.728650.

SHIMOTANI, K., OIKAWA, K., HORADA, O. AND KAGAWA, Y. 1995. Characteristics

of gasoline in-cylinder direct injection engine. JSAE Review. 17 (3), 267-272,
doi:10.1016/0389-4304(96)00021-5.

SIOKOS, K., KOLI, R., PRUCKA, R., SCHWANKE, J. AND MIERSCH, J. 2015.
Assessment of Cooled Low Pressure EGR in a Turbocharged Direct Injection Gasoline
Engine. SAE Int. J. Engines 8(4), doi:10.4271/2015-01-1253.

SONG, D., JIA, N., GUO, X., MA, X., MA, Z., GAO, D., LI, K., LAI, H. AND ZHANG, C.
2014. Low Pressure Cooled EGR for Improved Fuel Economy on a Turbocharged PFI
Gasoline Engine. SAE Technical Paper 2014-01-1240, doi:10.4271/2014-01-1240.

SPLITTER, D. AND SZYBIST, J. 2014. Intermediate Alcohol-Gasoline Blends, Fuels for

Enabling Increased Engine Efficiency and Powertrain Possibilities. SAE Int. J. Fuels Lubr.
7(1):29-47, doi:10.4271/2014-01-1231.

SRINIVASAN, S. 2001. Automotive Engines. New Delhi: Tata McGraw-Hill.

STEIN, R., ANDERSON, J. AND WALLINGTON, T. 2013. An Overview of the Effects of
Ethanol-Gasoline Blends on SI Engine Performance, Fuel Efficiency, and Emissions. SAE
Int. J. Engines 6(1), doi:10.4271/2013-01-1635

STONE, C. AND LADOMMATOS, N. 1992. The Measurement and Analysis of Swirl in

Steady Flow. SAE Technical Paper 921642, doi:10.4271/921642.

STONE, R. 1999. Introduction to Internal Combustion Engines, 3rd edition, Basingstoke:

Macmillan Press Ltd.

252
STONE, R., CHEN, L., HINTON, N. LEACH, F. AND XU, F. 2012. GDI Engine Operation
with Ethanol/Gasoline Blends and Aqueous Ethanol. Journal of Automotive Safety and
Energy. 3 (3), 257-264, doi:10.3969/j.issn.1674-8484.2012.03.009.

STOREY, J., BARONE, T., NORMAN, K. AND LEWIS, S. 2010. Ethanol Blend Effects On
Direct Injection Spark-Ignition Gasoline Vehicle Particulate Matter Emissions. SAE Int. J.
Fuels Lubr. 3(2):650-659, doi:10.4271/2010-01-2129.

STROJARSKI FAKULTET. 2015. Properties of some common fuels and hydrocarbons.

Available:
https://www.sfsb.hr/test/testhome/Test/solve/basics/tables/tablesComb/trash/hhv.html. Last
accessed 10th Mar 2015.

SU, J., XU, M., LI, T., GAO, Y. AND WANG, J. 2014. Combined effects of cooled EGR
and a higher geometric compression ratio on thermal efficiency improvement of a downsized
boosted spark-ignition direct-injection engine. Energy Conversion and Management. 78, 65–
73, doi:10.1016/j.enconman.2013.10.041.

SZWAJA, S. AND NABER, J. D. 2010. Combustion of n-butanol in a spark-ignition IC

engine. Fuel, 89, 1573–1582, doi:10.1016/j.fuel.2009.08.043

TAKAGI, Y. AND SKIPPON, S. 1998. Effects of In-Cylinder Fuel Spray Formation on

Emissions and Cyclic Variability in a Lean-Burn Engine. Part 1: Background and
Methodology. SAE Technical Paper 982618, doi:10.4271/982618.

TAKAHASHI, D., NAKATA, K., YOSHIHARA, Y., OHTA, Y. AND NISHIURA, H. 2015.
Combustion Development to Achieve Engine Thermal Efficiency of 40% for Hybrid
Vehicles. SAE Technical Paper 2015-01-1254, doi:10.4271/2015-01-1254.

TAKAKI, D., TSUCHIDA, H., KOBARA, T., AKAGI, M., TSUYUKI, T. AND
NAGAMINE, M. 2014. Study of an EGR System for Downsizing Turbocharged Gasoline
Engine to Improve Fuel Economy. SAE Technical Paper 2014-01-1199, doi:10.4271/2014-
01-1199.

TERAJI, A., KAKUHO, A., TSUDA, T. AND HASHIZUME, Y. 2009. A Study of the
Knocking Mechanism in Terms of Flame Propagation Behavior Based on 3D Numerical
Simulations. SAE Int. J. Engines 2(1):666-673, doi:10.4271/2009-01-0699.

THE INSTITUTE OF HISTORICAL RESEARCH, UNIVERSITY OF LONDON 2004.

History of Technology (Volume Twenty-Five). London: Thoemmes Continuum.

TIAN, G., LI, H, XU, H., LI, Y, AND SATISH M.R. 2010. Spray Characteristics Study of
DMF Using Phase Doppler Particle Analyzer. SAE 2010-01-1505, doi:10.4271/2010-01-
1505.

TOMITA, M., IWAKIRI, Y., SAKAI, E., URUSHIHARA, T., INOUE, R. AND KOJIMA,
K. 1996. Effects of Gas Flow and Mixture Properties on Engine-Out HC Emissions. SAE
Technical Paper 961952, doi:10.4271/961952.

253
TURNER, D. 2010. The combustion and emissions performance of fuel blends in modern
combustion systems. PhD Thesis, University of Birmingham.

TURNER, D., XU, H., CRACKNELL, R. F., NATARAJAN, V. AND CHEN, X. 2011.
Combustion performance of bio-ethanol at various blend ratios in a gasoline direct injection
engine. Fuel. 90, 1999-2006, doi:10.1016/j.fuel.2010.12.025.

ULSOY, A. G., PENG, H. AND ÇAKMAKCI, M. 2012. Automotive Control Systems.

Cambridge: Cambridge University Press.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY. 2010. Light-Duty

Vehicle Greenhouse Gas Emission Standards and Corporate Average Fuel Economy
Standards; Final Rule. Available:
http://www.nhtsa.gov/staticfiles/rulemaking/pdf/cafe/CAFE-GHG_MY_2012-
2016_Final_Rule_FR.pdf. Last accessed 27th February 2015.

UNITED STATES ENVIRONMENTAL PROTECTION AGENCY. 2014. Health.

Available: http://www.epa.gov/pm/health.html. Last accessed 16th Jan 2015.

URUSHIHARA, T., MURAYAMA, T., TAKAGI, Y. AND LEE, K. 1995. Turbulence and
Cycle-by-Cycle Variation of Mean Velocity Generated by Swirl and Tumble Flow and Their
Effects on Combustion. SAE Technical Paper 950813, doi:10.4271/950813.

VAROL, Y., ÖNER, C., ÖZTOP, H. F. AND ALTUN, Ş. 2014. Comparison of Methanol,
Ethanol, or n-Butanol Blending with Unleaded Gasoline on Exhaust Emissions of an SI
Engine. Energy Sources, Part A: Recovery, Utilization, and Environmental Effects. 36 (9),
938-948, doi:10.1080/15567036.2011.572141.

VUK, C. AND VANDER GRIEND, S. 2013. Fuel Property Effects on Particulates In Spark
Ignition Engines. SAE Technical Paper 2013-01-1124, doi:10.4271/2013-01-1124.

WALLNER, T. AND MIERS, S.A 2008. Combustion Behavior of Gasoline and

Gasoline/Ethanol Blends in a Modern Direct-Injection 4-Cylinder Engine. SAE 2008-01-
0077, doi:10.4271/2008-01-0077.

WALLNER, T., ICKES, A., LAWYER, K., MIERS, S., NABER, J., ERTL, D.,
WILLIAMSON, R., MIERS, S. A. AND NABER, J. D. 2013. Blend Ratio Optimization of
Fuels Containing Gasoline Blendstock, Ethanol, and Higher Alcohols (C3-C6): Part II -
Blend Properties and Target Value Sensitivity. SAE International 2013-01-1126,
doi:10.4271/2013-01-1126.

WANG, C., XU, H., DANIEL, R., GHAFOURIAN, A., HERREROS, J. M., SHUAI, S.
AND MA, X. 2013. Combustion characteristics and emissions of 2-methylfuran compared to
2,5-dimethylfuran, gasoline and ethanol in a DISI engine. Fuel, 103, 200-211,
doi:http://dx.doi.org/10.1016/j.fuel.2012.05.043.

WANG, C., XU, H., HERREROS, J. M., WANG, J. AND CRACKNELL, R. 2014a. Impact
of fuel and injection system on particle emissions from a GDI engine. Applied Energy, 132,
178–191, doi:http://dx.doi.org/10.1016/j.apenergy.2014.06.012.

254
WANG, C. 2014b. Combustion and emissions of a direct injection gasoline engine using
biofuels. PhD Thesis, University of Birmingham.

WANG, C., XU, H., HERREROS, J. M., LATTIMORE, T. AND SHUAI, S. 2014c. Fuel
Effect on Particulate Matter Composition and Soot Oxidation in a Direct-Injection Spark
Ignition (DISI) Engine. Energy & Fuels, 28, 2003-2012, doi:10.1021/ef402234z.

WEI, H., ZHU, T., SHU, G., TAN, L. AND WANG, Y. 2012. Gasoline engine exhaust gas
recirculation – A review. Applied Energy. 99, 534–544, doi:10.1016/j.apenergy.2012.05.011.

WILLIAMSON, R. 2013. Meeting RFS2 Targets with an E10/E15-like Fuel - Experimental

and Analytical Assessment of Higher Alcohols in Multi-component Blends with Gasoline.
SAE Int. J. Fuels Lubr. 6(3):691-701, doi:10.4271/2013-01-2612.

WU, C. 2007. Thermodynamics and Heat Powered Cycles: A Cognitive Engineering

Approach. New York: Nova Science.

YANG, J. AND ANDERSON, R.W. 1998. Fuel Injection Strategies to Increase Full-Load
Torque Output of a Direct-Injection SI Engine. SAE Technical Paper 980495,
doi:10.4271/980495.

YEJIAN, Q., CHENGJI, Z., HONGMING, X., JIAN, T. AND QIN, T. 2007. Experimental
Investigation of Using Ethanol-gasoline in Electronic Control Gasoline-injection Engine.
HKIE Transactions. 14 (2), 26-30, doi:10.1080/1023697X.2007.10668074.

YUAN, H., FOONG, T., CHEN, Z., YANG, Y., BREAR, M., LEONE, T. AND
ANDERSON, J. E. 2015. Modeling of Trace Knock in a Modern SI Engine Fuelled by
Ethanol/Gasoline Blends. SAE Technical Paper 2015-01-1242, doi:10.4271/2015-01-1242.

ZACCARDI, J., DUVAL, L. AND PAGOT, A. 2009. Development of Specific Tools for
Analysis and Quantification of Pre-ignition in a Boosted SI Engine. SAE Int. J. Engines
2(1):1587-1600, doi:10.4271/2009-01-1795.

ZENG, W., SJOBERG, M. AND REUSS, D. L. 2015. PIV examination of spray-enhanced

swirl flow for combustion stabilization in a spray-guided stratified-charge direct-injection
spark-ignition engine. International Journal of Engine Research. 16 (3), 306–322,
doi:10.1177/1468087414564605.

ZHANG, Z., WANG, T., JIA, M., WEI, Q., MENG, X. AND SHU, G. 2014a. Combustion
and particle number emissions of a direct injection spark ignition engine operating on
ethanol/gasoline and n-butanol/gasoline blends with exhaust gas recirculation. Fuel, 130,
177-188, doi:http://dx.doi.org/10.1016/j.fuel.2014.04.052.

ZHANG, Z., ZHANG, H., WANG, T. AND JIA, M. 2014b. Effects of tumble combined with
EGR (exhaust gas recirculation) on the combustion and emissions in a spark ignition engine
at part loads. Energy. 65, 18–24, doi:10.1016/j.energy.2013.11.062.

ZHANG, J., NITHYANANDAN, K., LI, Y., LEE, C.-F. AND HUANG, Z. 2015.
Comparative Study of High-Alcohol-Content Gasoline Blends in an SI Engine. SAE
Technical Paper 2015-01-0891, doi:10.4271/2015-01-0891.
255
ZHONG, L., MUSIAL, M., REESE, R. AND BLACK, G. 2013. EGR Systems Evaluation in
Turbocharged Engines. SAE Technical Paper 2013-01-0936, doi:10.4271/2013-01-0936.

ZHU, G. G., HASKARA, I. AND WINKELMAN, J. 2007. Closed-Loop Ignition Timing

Control for SI Engines Using Ionization Current Feedback. EEE Transactions on Control
Systems Technology, 15 (3), 416-427, doi:10.1109/TCST.2007.894634.

256