Downsized-Boosted-Gasoline-Engines Working-Paper ICCT 27102016 1

INVESTIGATION OF ENGINE DESIGN PARAMETERS ON
THE EFFICIENCY AND PERFORMANCE OF THE HIGH

SPECIFIC POWER DOWNSIZED SI ENGINE
A thesis submitted for the degree of Doctor of Philosophy
By
Barnaby Coates
School of Engineering and Design
Brunel University
United Kingdom
November 2012
Abstract
This study investigates the impact of employing the Miller cycle on a high specific
power downsized gasoline engine by means of Early Intake Valve Closing (EIVC) and
Late Intake Valve Closing (LIVC). This investigation assesses the potential for the Miller
cycle to improve fuel economy at part load points, as well as high load points with
significantly elevated boost pressures (Deep Miller) of up to 4 bar abs. The impact of
geometric Compression Ratio (CR) and Exhaust Back Pressure (EBP) has also been
investigated. The knock mitigating qualities of Deep Miller have been assessed, and its
ability to increase maximum engine load explored. Low Speed Pre-ignition (LSPI) and
autoignition tendencies with reduced coolant flow rates and with aged and new fuels
have also been studied.
This study comprises both experimental and analytical studies. A Ricardo Hydra single
cylinder thermodynamic engine was developed and used for the experimental
component of the study. This engine features a high specific power output (120kW/l)
cylinder head from the Mahle 1.2l 3 cylinder aggressively downsized engine. The
analytical component was carried out using a 1-dimensional GT-Power model based on
the Ricardo Hydra experimental engine. A Design of Experiments (DoE) based test plan
was adopted for this analytical study.
The experimental study found that EIVC was the optimal strategy for improving fuel
economy at both part-load and high-load conditions. LIVC yielded a fuel economy
penalty at part-load operations and a fuel economy improvement at high-loads. The
unexpected part-load LIVC result was attributed to the engine breathing dynamics of
the experimental engine. The analytical study found moderate LIVC to be the optimal
strategy at lower speeds, unless compensation for the increased degree of scavenging
experienced with EIVC was compensated for, in which case EIVC was optimum. At
higher speeds EIVC was found to be optimum regardless of whether or not
compensation for scavenging was employed. It was generally found that less sensitivity
to EBP was exhibited the more extreme the EIVC and LIVC. It was also found that a
higher geometric CR could be tolerated with extreme EIVC and LIVC, and a fuel
economy benefit could be obtained through the elevation of Geometric CR.
i
Acknowledgements
I would firstly like to express my thanks and utmost respect to my supervisor Professor
Hua Zhao for offering me this PhD opportunity in his world renowned Centre for
Advanced Powertrain and Fuels Research, and for his guidance and support over the
last 4 years.
I would also like to express my thanks to Neil Fraser and Paul Freeland of Mahle
Powertrain for giving me this opportunity, and for their patience. Without their
support and assistance, and their donation of numerous engine components and
instrumentation, this project could not have happened.
I also owe a great deal of thanks to other staff at Brunel University and Mahle
Powertrain who have given me guidance, advice and support on many occasions
including Alan Todd, Andrew Jaimeson, Carol Jackson, Dave Hancock, Dave
OudeNijeweme, Greg Taylor, Ian Reynolds, James Taylor, Phil Toon, Rob Beevers,
Cheryl Cramer, Chris Alan, Clive Barrett, John Langdon, Michael Low, Ray Kirby and
Reza Herfatmanesh. I would like to extend special thanks to Jens Neumeister, Justin
Mape, Tony Cains, Al Cairns and Yan Zhang for going out of their way to help me out
and assist me on a number occasions.
I also wish to express my gratitude to the engine lab technicians Andy Selway and Ken
Anstiss, with thanks in particular to Andy who has helped me out so greatly over the
last 4 years.
Finally I would like to thank my family and friends without whom I simply could not
have coped for the first 2 and a half years whilst the test cell was being constructed,
and during the various occasions the engine was not working and no results could be
obtained. I owe you all so much.
ii
Abbreviations, Symbols and Chemical Symbols
General Abbreviations
ABDC After Bottom Dead Centre

AC Alternating Current
AFR Air Fuel Ratio
ATDC After Top Dead Centre
BBDC Before Bottom Dead Centre
BDC Bottom Dead Centre
BEV Battery Electric Vehicle
BLD Borderline Detonation
BMEP Brake Mean Effective Pressure
BSFC Brake Specific Fuel Consumption
BTDC Before Top Dead Centre
BTDCF Before Top Dead Centre Firing
CA Crank Angle
CAAA Clean Air Act Amendments
CAD Crank Angle Degrees
CAFE Corporate Average Fuel Economy
CAI Controlled Autoignition
CARB Californian Air Resources Board
CAT Charge Air Temperature
CFD Computational Fluid Dynamics
CGI Compacted Graphite Iron
CI Compression Ignition
COST Changing One Separate factor at a Time
COV Coefficient Of Variation
CR Compression Ratio
DC Direct Current
iii
DI Direct Injection
DOC Diesel Oxidation Catalyst
DoE Design of Experiment
DPF Diesel Particulate Filter
eBoost Electric Boosting
ECU Electronic Control Unit
EE Extended Expansion
EGR Exhaust Gas Recirculation
EGT Exhaust Gas Temperature
EIVC Early Intake Valve Closing
EMOP Exhaust Maximum Opening Point
EMS Engine Management System
EOC End Of Compression
EOR End Of Ramp
EPA Environmental Protection Agency
E-REV Extended-Range Electric Vehicle
EU European Union
EVC Exhaust Valve Closing
FCEV Fuel Cell Electric Vehicle
FID Flame Ionization Detector
GDI Gasoline Direct Injection
GE General Electric
GIMEP Gross indicated Mean Effective Pressure
GVW Gross Vehicle Weight
HC Hydrocarbons
HCCI Homogeneous Charge Compression Ignition
HEV Hybrid Electric Vehicle
ICE Internal Combustion Engine
IMEP Indicated Mean Effective Pressure
IMOP Intake Maximum Opening Point
IMPR Intake Manifold Pressure Referencing
iv
ISFC Indicated Specific Fuel Consumption
IVC Intake Valve Closing
IVO Intake Valve Opening
KI Knock Intensity
LBDI Lean Boost Direct Injection
LEV Low Emission Vehicle
LF Low-pass Filter
LHR Low Heat Rejection
LIVC Late Intake Valve Closing
LIVO Late Intake Valve Open
LNT Lean NOx Trap
LSPI Low Speed Pre-ignition
LUSIE Leeds University Spark Ignition Engine
MAF Mass Air Flow
MAP Manifold Absolute Pressure
MBT Minimum spark advance for Best Torque
MFB Mass Fraction burn
NA Naturally Aspirated
NEDC New European Driving Cycle
NI National Instruments
NIMEP Net Indicated Mean Effective Pressure
NLT New Long Term
NMHC Non-methane Hydrocarbons
NMOG Non-methane Organic Gasses
NOx Nitrogen Oxides
NST New Short Term
NVH Noise, Vibration and Harshness
PC Passenger Car
PFI Port Fuel Injection
PID Proportional Integral Derivative
PIPR Polytropic Index Pressure Referencing
v
PM Particulate Matter
PMEP Pumping Mean Effective Pressure
PN Particulate Number
PNLT Post New Long Term
PRT Platinum Resistance Thermometer
RBF Radial Base Function
RM Reference Mass
RON Research Octane Number
rpm revolutions per minute
SCR Selective Catalytic Reduction
SI Spark Ignition
SOI Start Of Injection
SULEV Super Ultra Low Emission Vehicle
TC Time Constant
TDC Top Dead Centre
TKE Turbulent Kinetic Energy
UEGO Universal Exhaust Gas Oxygen
UGMF Unburned Gas Mass Fraction
ULEV Ultra Low Emission Vehicle
ULG Unleaded Gasoline
US United States
VVA Variable Valve Actuation
VVT Variable Valve Timing
WOT Wide Open Throttle
ZEV Zero Emissions Vehicle
Symbols
a Wiebe model constant
A Knock index multiplier, Activation energy multiplier, Flow area
vi
As Heat transfer surface area
b Bore
Cf Skin friction coefficient
Specific heat capacity at constant pressure, Pressure loss

Cp
coefficient
CL Connecting rod length
D Equivalent diameter
e Total internal energy per unit mass
h Heat transfer coefficient
H Total enthalpy
ISem Indicated specific emission
Km Percentage cylinder unburned at knock initiation
m Wiebe model constant, mass
ṁ Boundary mass flux into the volume
M Molecular weight
n Polytropic exponent/specific heat ratio
ON Octane number
p In-cylinder pressure, pressure
pf Pressure feedback
pn Predicted pressure
∆p Change in pressure
∆pc Change in pressure due to combustion
∆pv Change in pressure due to volume change
P Precursor reaction rate multiplier
Q Heat input
rV Geometric compression ratio
R Specific gas constant
s Stroke
vii
t Time
thkn Start of compression angle
T Livengood-Wu integral, Temperature
Ta Activation temperature
Tfluid Fluid temperature
Tu Instantaneous unburned gas temperature
Twall Wall temperature
∆T Temperature difference
u Velocity at the boundary
V Volume
VC Volumetric concentration
VI Cylinder volume at knock initiation
VTDC Volume at TDC
x MFB, Crank-pin offset
xb Mass fraction burned
ρ Density
θ Crank angle
θ0 Crank angle at combustion start
θb Combustion duration
ŋ Thermodynamic efficiency
ŋ Otto Otto cycle efficiency
φ Equivalence ratio
τ Douaud and Eyzat component
Chemical symbols
CHO Carbohydrate
CO Carbon Monoxide
viii
CO2 Carbon dioxide
H Hydrogen atom
H2SO4 Sulphuric acid
HCHO Formaldehyde
N Nitrogen atom
N2 Nitrogen
NO Nitric oxide
NO2 Nitrogen dioxide
O Oxygen atom
O2 Oxygen
O3 Ozone
OH Hydroxyl radical
SO2 Sulphur dioxide
SO3 Sulphur trioxide
ix
Contents
Page
Abstract
Acknowledgements
Abbreviations, Symbols and Chemical Symbols
Chapter 1: Introduction 1
1.1 Introduction 1
1.2 Objectives 5
1.3 Outline of Thesis 5
Chapter 2: Literature Review 8
2.1 Introduction 8
2.2 Gasoline Engine Emissions 10
2.2.1 Current Worldwide Emission Legislation 10
2.2.1.1 Carbon Monoxide (CO) 16
2.2.1.2 Unburned Hydrocarbons (HC) 17
2.2.1.3 Nitrogen Oxides (NOx) 18
2.2.1.4 Particulate Matter (PM) 19
2.3 Current Automotive Engines, Technologies and Trends 19
2.3.1 Hybrid and Electric Powertrains 20
2.3.2 Fossil Fuel Powertrains 21
2.3.3 Gasoline Engine Developments 22
2.3.4 Downsizing and Turbocharging 25
2.4 Atkinson/Miller Cycle 28
2.4.1 Introduction 28
x
2.4.2 History 29
2.4.3 Pumping Losses 33
2.4.4 Combustion Stability 36
2.4.5 Engine Breathing 39
2.4.6 Emissions 40
2.4.7 Knock Mitigation and Deep Miller 43
2.5 Summary 44
Chapter 3: Experimental Test Facility 46
3.1 Introduction 46
3.2 Experimental Setup 46
3.2.1 Engine Description 47
3.2.1.1 Single Cylinder Engine Description 49
3.2.1.2 Cylinder Block Design 51
3.2.2 Oil System 53
3.2.3 Coolant System 55
3.2.4 Fuel System 59
3.2.5 Intake System 62
3.2.5.1 Pipes 62
3.2.5.2 System Details 63
3.2.6 Exhaust System 67
3.2.6.1 Design 67
3.2.6.2 System Details 67
3.2.6.3 Conversion of Emissions to Specific Emissions 69
3.2.7 Dynamometer and EMS 70
3.3 Data Acquisition and Instrumentation 75
xi
3.3.1 Data Acquisition Software 77
3.3.2 Calibration of Sensors 82
3.3.3 Accuracy of Data and Known Measurement Errors 84
3.4 Daily Test Point 86
Chapter 4: Development of a Single Cylinder Engine GT-Power Model 87
4.1 Introduction 87
4.2 Development of the Single Cylinder Engine Model 88
4.2.1 Single Cylinder Engine Model Details 89
4.2.1.1 Flow Modelling 89
4.2.1.2 Combustion Modelling 91
4.2.1.3 Modelling Limitations 92
4.2.1.4 Other Modelling Parameters 93
4.3 Single Cylinder Engine Model 1: DoE Based Test Plan 93
4.3.1 Single Cylinder Engine Knock Model 96
4.3.1.1 Functionality of the Knock Model 96
4.3.1.2 Irresolvable Issues with the Knock Model 101
4.4 Single Cylinder Engine Model 2: Correlation of Thermodynamic Data 103
4.5 Mahle Powertrain 3 Cylinder Engine Model 104
Chapter 5: Application of the Deep Miller Cycle to Improve Fuel Economy and 106
Increase Maximum Load
5.1 Introduction 106
5.2 Development of a DoE Based Test Plan 107
5.2.2 Screening 107
xii
5.2.3 DoE Type 110
5.2.4 Limitations of the DoE 111
5.2.5 Test Point Description 114
5.2.5.1 1000rpm, 14 bar BMEP 116
5.2.5.2 2000rpm, 24 bar BMEP 117
5.2.5.3 5000rpm, 24 bar BMEP 117
5.3 ISFC optimisation 118
5.3.2 ISFC Trends 119
5.3.2.1 1000rpm, 14 bar BMEP 119
5.3.2.1.1 EBP Effects 119
5.3.2.1.2 Geometric CR Effects 123
5.3.2.2 2000rpm, 24 bar BMEP 128
5.3.2.3 5000rpm, 24 bar BMEP 137
5.3.3 Cam Timing Trends 146
5.3.4 Scavenging Effects 150
5.3.4.1 Scavenging 151
5.3.4.2 Charge Scavenging 159
5.3.5 Engine Breathing 161
5.3.6 General Compression Ratio Effects 169
5.4 Maximum Load Comparison 174
5.5 Summary 177
xiii
Chapter 6: Single Cylinder Engine Performance Evaluation 179
6.2 Comparison of the Single Cylinder Engine to the 3 Cylinder Engine 179
6.2.2 Description of Test Points 180
6.2.3 Results 182
6.2.3.1 ISFC 182
6.2.3.2 Engine Breathing 184
6.2.3.3 Knocking Tendency 192
6.2.3.4 Correct EBP Determination 195
6.3 Impact of Coolant Flow Rate on Autoignition Tendency 198
6.3.2 The Effect of Coolant Temperature Gradient 199
6.3.2.1 Results 202
6.3.3 Effect of Coolant Flow Rate at Higher Load Operations 210
6.3.3.1 Results 211
6.4 Impact of Cam Timing on Autoignition Tendency 213
6.4.2 Test Plan 213
6.4.3 Results 214
6.4.4 GT-Power Analysis 223
6.4.4.1 Introduction 223
6.4.4.2 Results 224
6.5 Effects of Fuel Age 228
xiv
6.5.2 Test Plan 229
6.5.3 Results 229
Chapter 7: Experimental Studies of Miller Cycle on Engine Performance and Fuel 232
Consumption
7.2 Achieving EIVC and LIVC 233
7.3 Description of Test Points 234
7.4 Low Load Operations with EIVC and LIVC (Region 1) 237
7.4.2 Results 238
7.4.2.1 Discrepancy at 1000rpm 244
7.4.2.2 Emissions 248
7.4.2.3 Analysis of the Individual Components of EIVC and LIVC 253
7.4.2.4 Optimum Cam Timing Observations 260
7.4.2.5 Combustion Phasing Observations 261
7.4.3 Summary 264
7.5 High Load Operations with EIVC and LIVC (Deep Miller) 265
7.5.2 Results 266
7.5.2.1 EBP Effects 273
7.5.2.2 Analysis of the Individual Components of EIVC and LIVC 274
7.5.2.3 Tubocharging/Supercharging Feasibility for Deep Miller 280
7.5.3 Summary 282
Chapter 8: Summary and Conclusions 284
xv
8.2 Analytical Study of Deep Miller 285
8.3 Experimental Engine Study of Miller Cycle 287
8.4 Recommendations for Future Work 289
References 290
Appendix 300
xvi
List of Figures
Figure 3.1 The Mahle 1.2l 3 Cylinder Advanced Downsizing Engine 47
Figure 3.2 The Mahle Downsized Demonstrator Vehicle 48
Figure 3.3 The Brunel Single Cylinder Downsized Engine 49
Figure 3.4 Cylinder Block Plumbing 51
Figure 3.5 Ansys Plot of Factor of Safety 52
Figure 3.6 Oil System Schematic Diagram 54
Figure 3.7 Coolant Jacket Height Relative to Coolant Tank Height 56
Figure 3.8 Coolant System Schematic Diagram 58
Figure 3.9 Fuel System Schematic Diagram 61
Figure 3.10 Intake System Schematic Diagram 66
Figure 3.11 Exhaust System Schematic Diagram 69
Figure 3.12 Dynamometer Torque Curve and Engine Torque Curve Comparison 71
Figure 3.13 Dynamometer Schematic Diagram 72
Figure 3.14 ECU Loom Connection List 73
Figure 3.15 Transient Combustion Analyzer Data Acquisition Software Interface 77
Figure 3.16 Graph Window Layout 78
Figure 3.17 Band-pass Filtering Calculation 79
Figure 3.18 Accuracy of Fuel Rail Pressure Control at 3000rpm and 4 bar BMEP 85
Figure 4.1 Staggered Grid Arrangement 90
Figure 4.2 Single Cylinder Engine Model 1 94
Figure 4.3 TDC Determination with a Hold Object 99

xvii
Figure 4.4 Livengood-Wu Integral Accuracy at 1 CAD Resolution 100
Figure 4.5 Mahle Powertrain 3 Cylinder Engine Model 105
Figure 5.1 Latin Hypercube (left) and D-Optimum (right) Quadratic Point
Distribution Comparison 110
Figure 5.2 Latin Hypercube (left) and D-Optimum (right) Quadratic Curve Fitting
Comparison with Lips Circled 111
Figure 5.3 Comparison of GT-Power Calculated Intake Cam Profiles for 152, 196,
240, 276 and 312 CAD Durations Cams 113
Figure 5.4 Impact of EBP on ISFC at a Geometric CR of 9 120
Figure 5.5 Comparison of Effective CR and ISFC for Each EBP at a Geometric CR
of 9 123
Figure 5.6 Comparison of Different Geometric CRs with an EBP of 1.5 bar 125
Figure 5.7 Comparison of Different Geometric CRs with an EBP of 2 bar 125
Figure 5.8 Comparison of Different Geometric CRs with an EBP of 2.5 bar 126
Figure 5.9 Impact of EBP on ISFC at a Geometric CR of 9 128
Figure 5.10 Optimum ISFC Cam IMOPs for an EBP of 2 bar at a Geometric CR of 9 129
Figure 5.11 Optimum ISFC cam IMOPs for an EBP of 3 bar at a Geometric CR of 9 129
Figure 5.12 Optimum ISFC cam IMOPs for an EBP of 4 bar at a Geometric CR of 9 130
Figure 5.13 Optimum IVO Cam Timing Points for EBPs of 2 and 3 bar Compared
to the IVO Limit for a Geometric CR of 9 131
Figure 5.14 Log P-V IVC Sweep for 196 CAD Duration Intake Cam at a Geometric
CR of 9 and an EBP of 2 bar 132
Figure 5.15 Log Temperature Vs. Volume Ratio IVC Sweep for 196 CAD Duration
Intake Cam at a Geometric CR of 9 and an EBP of 2 bar 133
xviii
Figure 5.19 Comparison of 2, 3 and 4 bar EBPs at a Geometric CR of 9 138
Figure 5.20 Effective CR and Lambda Plots at a Geometric CR of 9, EBP of 2 bar

and a Cam Duration of 276 CAD 141
Figure 5.24 The Theoretical Gain in Thermodynamic Efficiency Through the

Adoption of Optimum Cam Positioning 147
Figure 5.25 Number of Points where a Theoretically Non-optimal Cam Timing

was found to be Optimal in the Test Work 148
Figure 5.26 Maximum Benefit over Baseline Case (IVC at BDC) with each Cam at
each Geometric CR 149
Figure 5.27 Demonstration of Optimum Cam Timings for EIVC and LIVC 151
Figure 5.28 3 Cylinder Engine Speed and Load Map Showing where Exhaust
Lambda Enrichment and Leaning is Used 152
Figure 5.29 Scavenging Compensation (%) for each of the Different Cams at each
EBP and CR for 1000rpm and 14 bar BMEP 153
Figure 5.32 Comparison of In-cylinder Lambdas for the 152 CAD Duration Cam at
2 bar EBP and a Geometric CR of 9 for all Speeds 156
xix
Figure 5.33 1000rpm, 14 bar BMEP, EBP of 1.5 bar, Geometric CR of 9. With and
without Scavenge Compensation 157
Figure 5.34 2000rpm, 24 bar BMEP, EBP of 2 bar, Geometric CR of 9. With and
Figure 5.35 5000rpm, 24 bar BMEP, EBP of 2 bar, Geometric CR of 9. With and
Figure 5.36 Estimated Charge Scavenging at 1000rpm and 14 bar BMEP for 1.5,
2 and 2.5 bar EBP 160
Figure 5.37 Charge Scavenging at 2000rpm and 24 bar BMEP with an EBP of 2
bar and a Geometric CR of 9, 10.75 and 12.5 161
Figure 5.38 EBP Resolved on a CAD Basis for an EBP of 2 bar, a Cam Duration of
152 CAD for Speeds of 1000, 2000 and 5000rpm 162
Figure 5.39 Log P-V Overlay at an EBP of 2 bar, a Cam Duration of 152 CAD, and
for Speeds of 1000, 2000 and 5000rpm 163
Figure 5.40 Intake Mass Flow Rate Overlay for 240 CAD Cam Duration at an EBP
of 2.5 bar, and a Speed and Load of 1000rpm and 14 bar BMEP 164
Figure 5.41 2000rpm, 24 bar BMEP, 196 CAD Duration Cam, EBP 4 bar,
Geometric CR of 9. Intake Valve Mass Flow Rate Profiles from an IMOP Sweep 165
Figure 5.42 2000rpm, 24 bar BMEP, 196 CAD Duration Cam, EBP 4 bar,
Geometric CR of 9. MAP Requirements for an IMOP Sweep 166
Figure 5.43 Influence of Exhaust Gas Temperature (EGT) on Exhaust

Pressure/Rarefaction Wave at 5000rpm, 24 bar, 2 bar EBP and a Geometric CR
of 9 167
Figure 5.44 Comparison of Single Cylinder and 3 Cylinder EBP Profiles Resolved
on CAD Basis at 5000rpm and 24 bar BMEP 169
Figure 5.45 Log P-V Diagram at 2000rpm, 24 bar BMEP, 312 CAD Duration Cam,
EBP of 2 bar. Comparison of Different Geometric CRs 170
xx
Figure 5.46 1000rpm, 14 bar BMEP 50% MFB Points for EBPs of 1.5, 2 and 2.5
bar at Geometric CRs of 9, 10.75 and 12.5 171
Figure 5.47 2000rpm, 24 bar BMEP 50% MFB Points for EBPs of 2, 3 and 4 bar at
Geometric CRs of 9, 10.75 and 12.5 172
Figure 5.48 5000rpm, 24 bar BMEP 50% MFB Points for EBPs of 2, 3 and 4 bar at
Geometric CRs of 9, 10.75 and 12.5 172
Figure 5.49 Peak Load (NIMEP) at 1000, 2000 and 5000rpm for each Cam Profile
for an EBP of 2 bar 175
Figure 5.50 Peak Load (NIMEP) at 1000, 2000 and 5000rpm for each Cam Profile
for an EBP of 3 bar at 2000 and 5000 rpm, and an EBP of 2.5 bar at 1000rpm 175
Figure 6.1 Baseline Data Test Points 181
Figure 6.2 ISFC Comparison of 3 Cylinder and Single Cylinder Engines 183
Figure 6.3 Single Cylinder and 3 Cylinder Spark Timing Comparison 185
Figure 6.4 Single Cylinder and 3 Cylinder 50% MFB Point Comparison 186
Figure 6.5 Single Cylinder and 3 Cylinder 10-90% MFB Duration Comparison 186
Figure 6.6 Single Cylinder and 3 Cylinder MAP Requirement Comparison 187
Figure 6.7 Single Cylinder and 3 Cylinder EBP Comparison 187
Figure 6.8 Comparison of Intake Manifold Pressures and Intake Valve Mass Flow
Rate at 5000rpm and 24 bar BMEP 189
Figure 6.9 5000rpm 24bar BMEP logP-V Comparison 190
Figure 6.10 2000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation In-
cylinder Pressure Comparison 193
Figure 6.11 2000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation EBP
Comparison 193
Figure 6.12 3000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation In-
cylinder Pressure Comparison 194
Figure 6.13 3000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation EBP
Comparison 195
xxi
Figure 6.14 2000rpm EBP Comparison at Various Loads with the EBP Valve
Applied 197
Figure 6.15 2000rpm, Various Loads Single Cylinder EBP Overlay Vs. Valve
Overlap Period 198
Figure 6.16 Severe Autoignition Frequency 202
Figure 6.17 Peak Autoignition Cylinder Pressure Vs. Total Engine Coolant Flow
Rate 203
Figure 6.18 Peak Autoignition Cylinder Pressure Vs. Block Coolant Flow Rate 203
Figure 6.19 Head Coolant Delta T (deg C) 205
Figure 6.20 Block Coolant Delta T (deg C) 205
Figure 6.21 Head Coolant Flow Rate (l/min) 207
Figure 6.22 Block Coolant Flow Rate (l/min) 208
Figure 6.23 Cylinder Head Coolant Jacket Geometry 208
Figure 6.24 Average Band-pass Filtered KI (bar) 209
Figure 6.25 BLD 50% MFB (CAD ATDCF) 210
Figure 6.26 Autoignition Frequency in Percentage of Autoignition Cycles at

Different Coolant Flow Points 212
Figure 6.27 Peak Autoignition Cylinder Pressure 215
Figure 6.28 Frequency of Autoignition Events at Specific Cam Timings 215
Figure 6.29 Ridge 1 and Ridge 2 Definitions 216
Figure 6.30 Comparison of 2 Superknock Events, 1 Occurring Under Ridge 1, the

Other Under Ridge 2 217
Figure 6.31 Comparison of Pressure Histories for 2 Superknock Events, 1

Occurring Under Ridge 1, the Other Under Ridge 2 218
Figure 6.32 Polytropic Exponents 219
Figure 6.33 50% MFB Angle (CAD ATDCF) 220
Figure 6.34 EMOP Sweep at Constant Load and IMOP Demonstrating the Impact
of EMOP Timing on GIMEP 220
xxii
Figure 6.35 Autoignitions Under Ridge 1 (Units of IMOP and EMOP are CAD
ATDC and CAD BTDC Respectively) 221
Figure 6.36 Autoignitions Under Ridge 2 (Units of IMOP and EMOP are CAD
ATDC and CAD BTDC Respectively) 222
Figure 6.37 Exhaust Port Temperatures (0C) 223
Figure 6.38 Experimental (left) and Simulated (right) ISFC Comparison 225
Figure 6.39 Equivalent AFR at Cycle Start 226
Figure 6.40 Residual Gas Concentration at Cycle Start 226
Figure 6.41 EOC Pressure (bar) 227
Figure 6.42 EOC Temperature (0C) 227
Figure 6.43 Autoignitions with Aged ULG95 230
Figure 6.44 Autoignitions with Fresh ULG95 230
Figure 7.1 Cam Profile Comparison (as Shown in Their Maximum Valve Overlap
Positions) 233
Figure 7.2 Cam Phasing Ranges 234
Figure 7.3 Regions of Miller Cycle Studied 235
Figure 7.4 ISFC Comparison of Short, Standard and Long Cams at Low Loads 238
Figure 7.5 1000rpm, 8 bar BMEP Short, Standard and Long Cam Overlay 240
Figure 7.6 PMEP Comparison of Short, Standard and Long Cams at Low Loads 241
Figure 7.7 4000rpm, 4 bar BMEP Short, Standard and Long Cam Overlay 242
Figure 7.8 Intake Temperature Comparison of Short, Standard and Long Cams at
Low Loads 244
Figure 7.9 Comparison of the Intake Valve Mass Flow Rates for the 3 Different
Duration Cams at 1000rpm and 4 bar BMEP 245
Figure 7.10 MFB Profiles for the 3 Different Cams at 1000rpm and 4 bar BMEP 246
Figure 7.11 Comparison of COV for 1000rpm and 4 bar BMEP for Short (left),
Standard (middle) and Long (right) Cams 247
xxiii
Figure 7.12 HC Emissions Comparison of Short, Standard and Long Cams at Low
Loads 249
Figure 7.13 NOx Emissions Comparison of Short, Standard and Long Cams at
Low Loads 250
Figure 7.14 HC and NOx Emissions as Obtained from Test Work Carried out by
Mahle Powertrain 251
Figure 7.15 MAP Comparison of Short, Standard and Long Cams at Low Loads 252
Figure 7.16 Theoretical Temperatures at Various Points in the 4-Stroke Atkinson

and Otto Cycles 254
Figure 7.17 Comparison of 2-Stroke (left) and 4-Stroke (right) Theoretical

Atkinson Cycles 255
Figure 7.18 Theoretical Benefit with EIVC at 3000rpm and 4 bar BMEP 256
Figure 7.19 Actual Benefit with EIVC at 3000rpm and 4 bar BMEP 256
Figure 7.20 Theoretical Benefit with LIVC at 3000rpm and 4 bar BMEP 258
Figure 7.21 Actual Benefit with LIVC at 3000rpm and 4 bar BMEP 259
Figure 7.22 Map of 10-90% MFB Rates for an IMOP-EMOP Sweep with the Short
Cam at 3000rpm and 4 bar BMEP 261
Figure 7.23 MBT Comparison of Short, Standard and Long Cams at Low Loads 262
Figure 7.24 Spark Timing Comparison of Short, Standard and Long Cams at Low
Loads 263
Figure 7.25 Optimum IMOP Timing Comparison of Short, Standard and Long
Cams at Low Loads 264
Figure 7.26 2000rpm, 24 bar BMEP ISFC Comparison for Short, Standard and
Long Cams at 2 EBPs 266
Figure 7.27 HC Emissions for Different IVC Timings and 2 Different EBPs with the
Short Cam at 2000rpm and 24 bar BMEP with an EMOP of 100 CAD BTDC
(Maximum Valve Overlap Position) 267
Figure 7.28 COV for Short, Standard and Long Cams at 2000rpm, High Load with
and without EBP 269
Figure 7.29 10-90% MFB Duration for Short, Standard and Long Cams at
2000rpm, High Load with and without EBP 270
xxiv
Figure 7.30 50% MFB Angle for Short, Standard and Long Cams at 2000rpm,
High Load with and without EBP 271
Figure 7.31 Spark Timing for Short, Standard and Long Cams at 2000rpm, High
Load with and without EBP 272
Figure 7.32 2000rpm, 15 bar BMEP ISFC Comparison for Short, Standard and
Long Cams at 2 EBPs 272
Figure 7.33 EBP Profile Comparison between Short, Standard and Long Cams
with 2 bar Average EBP at 2000rpm and 24 bar BMEP 274
Figure 7.34 Theoretical Benefit with EIVC at 2000rpm and 24 bar BMEP with no
EBP 275
Figure 7.35 Theoretical Benefit with EIVC at 2000rpm and 24 bar BMEP with an
EBP of 2 bar 276
Figure 7.36 Actual Benefit with EIVC at 2000rpm and 24 bar BMEP with no EBP 277
Figure 7.37 Actual Benefit with EIVC at 2000rpm and 24 bar BMEP and 2 bar EBP 277
Figure 7.38 Theoretical Benefit with LIVC at 2000rpm and 24 bar BMEP with an
EBP of 2 bar 278
Figure 7.39 Actual Benefit with LIVC at 2000rpm and 24 bar BMEP with no EBP 279
Figure 7.40 Actual Benefit with LIVC at 2000rpm and 24 bar BMEP with an EBP
of 2 bar 279
Figure 7.41 MAP Requirement Vs. BMEP for Short, Standard and Long Cams at
an Engine Speed of 2000rpm 281
xxv
List of Tables
Table 2.1 Euro Standards Implementation Map 12
Table 2.2 Vehicle Categories Applicable to Euro Standards 12
Table 2.3 Current Worldwide Emissions Standards 15
Table 3.1 Daily Test Point Test Variables and Control Criteria 86
Table 4.1 Description of the Various Crank Angle Sensors in the Knock Model 98
Table 4.2 MAP Correction Factors 102
Table 4.3 EBP Correction Factors 102
Table 4.4 Geometric CR Correction Factors 102
Table 4.5 IMOP Correction Factors 102
Table 4.6 MAP and IMOP 3 Cylinder Map of BLD 50% MFB Points at a Speed and
Load of 5000rpm and 24 bar BMEP Respectively 103
Table 5.1 DoE Ranges, Levels and Resolution (1000rpm) 109
Table 5.2 Intake Valve IMOP Ranges for Different Duration Cams 109
Table 5.3 SOI Timings for Each of the Speed and Load Points 116
Table 5.4 1000rpm, 14 bar BMEP, Geometric CR of 9, EBP of 1.5 bar. With and
without Scavenge Compensation Effects on ISFC for an IMOP Sweep with the
152 CAD Duration Cam 158
xxvi
Table 6.1 Baseline Data Test Points 181
Table 6.2 Test Variables and Control Criteria 182
Table 6.3 Table of Average EBP during the Exhaust Valve Open Phase 196
Table 6.4 Test Plan Showing Desired ∆Ts and the Corresponding Flow Rate
Required to Achieve that ∆T 201
Table 6.6 Test Plan Showing Desired ∆Ts and the Corresponding Flow Rate
Required to Achieve that ∆T 211
Table 7.1 Test Points for De-throttling Studies (BMEPs are the 3 Cylinder
Equivalent Values) 236
Table 7.2 Test Points for Region 2 (BMEPs are the 3 Cylinder Equivalent Values) 236
Table 7.4 MOPs and Corresponding Valve Opening/Closing Points for Each of the
Cams 237
xxvii
1
Chapter 1
Introduction
1.1 Introduction
The advent of the Internal Combustion Engine (ICE) represents a huge

milestone in human progress, and has dramatically changed the world we live in today.
ICEs were made possible by an engineer called Nikolaus Otto who gave his name to the
famous Otto cycle process that allows energy and mechanical power to be extracted
from the combustion of air and fuel, more often than not by means of the
reciprocating motion of a piston in a cylinder.
ICEs represented a great leap forward in terms of power creation, providing a

smaller, neater, and much more diverse solution to the steam engine. ICEs are now
used extensively in a wide variety of applications. The ICE is now the most common
form of prime mover in the world today, and is regarded as a key enabler of
globalisation (Smil [1]).
Since a very early stage efforts were made to refine and optimise the ICE. Oil
reserves initially represented the greatest incentive to optimise the ICE and the
greatest hindrance to their widespread adoption. As better techniques were
developed to extract oil so the ICE gained popularity.
The main focus of early research and development was on improvement of fuel
economy and increasing power. The main motivation for this was for aircraft
1
applications which were responsible for technologies such as supercharging,
turbocharging and Direct Injection (DI) (Zhao [2]). These technologies are being
reincarnated today to improve fuel economy and increase power output.
After World War II emissions also became the focus of ICE research and
development. In the UK this was prompted by a new act of parliament called the
“Clean Air Act 1956” [3] which was introduced to combat increasing levels of air
pollution and photochemical smog.
In more recent history as the popularity of ICEs and the automotive sector have
increased drastically due to emerging global markets such as China and India, the
demand for fossil fuels has also increased massively, to the extent that relatively minor
conflicts in the world can see the price of fuel fluctuate considerably. This coupled with
growing concerns over the impact of carbon dioxide (CO2) on the environment and the
introduction of a new fining system for cars that produce more than a certain quantity
of CO2 has prompted a new flurry of research into cleaner and more fuel efficient ICEs.
Manufacturers facing the proposition of having to “pay” to be in-efficient and

polluting are now resorting to technologies that hitherto have been too expensive or
too advanced to justify, such as Gasoline Direct Injection (GDI), turbocharging for
gasoline engines, 2-stage turbocharging, Electrical Boosting (eBoost) systems, fully
Variable Valve Trains (VVT), Selective Catalytic Reduction (SCR) systems, and many
more technologies besides.
These new technologies allow far greater freedom, and real time variation of
engine parameters that would historically have been fixed, such as valve timing, valve
duration, Intake Manifold Absolute Pressure (MAP), fuel injection pressure, fuel
injection duration, fuel injection timing etc. all of which can be used to improve fuel
economy and reduce emissions. This in turn results in different engine concepts.
One such engine concept that has great synergies with the above technologies
is engine downsizing. A great deal of work has been invested in the concept of engine
downsizing over the last decade and has shown great potential to reduce engine out
emissions and CO2 whilst also providing a significant improvement in fuel economy.
2
The concept of turbocharging an engine is nothing new, and it has been carried
out successfully on many occasions throughout history, but it is important to
understand that downsizing is not simply turbocharging an engine. Turbocharging is
usually employed on premium “sporty” vehicles in an effort to boost performance
rather than to reduce emissions or improve fuel economy. For downsizing the
turbocharger is not there for “bolt on” performance, it is integral to the engine and is
required to compensate for the reduction in swept cylinder capacity. Turbocharged
engines in “sporty” vehicles still have the engine capacity to accelerate adequately at
low engine speeds, so the turbocharger performance at low engine speeds is usually
not of great importance to driver satisfaction. For a downsized engine though, the
turbocharger must be employed to accelerate adequately, and is therefore required to
function well at low speed and with very little lag time. What is more it must also
function well at every other engine speed and load point without compromising
reliability and/or Noise Vibration and Harshness (NVH).
Turbocharged engines are traditionally poor at low engine speeds in

comparison to their performance at higher engine speeds for 2 main reasons. Firstly
the turbocharger is receiving insufficient energy to compress the intake air and
secondly the low engine speed means the compression process takes a long time, and
the charge motion in the cylinder is usually quite slow resulting in a comparatively long
combustion duration. This makes knocking combustion much more likely. This problem
is further exacerbated in downsized engines because the MAP will need to be higher
for an equivalent load site in comparison to a larger naturally aspirated engine. This
has the effect of increasing the End Of Compression (EOC) pressure and
correspondingly temperature thus making knocking combustion even more probable.
A great deal of research effort is being invested in mitigating this problem,

including investigating 2 techniques known as Early Intake Valve Closing (EIVC) and
Late Intake Valve Closing (LIVC). EIVC refers to the technique of closing the intake valve
early of Bottom Dead Centre (BDC) and depressurising the charge in the cylinder
slightly, lowering its temperature which results in lower EOC pressures and
temperatures. LIVC involves closing the intake valve after BDC thus filling the cylinder
completely but expelling some air from the cylinder to the intake. EIVC and LIVC both
3
operate by reducing the effective Compression Ratio (CR), thus allowing greater scope
to increase the geometric CR and improving fuel economy at lower speeds and loads.
Both EIVC and LIVC strategies lower the volumetric efficiency of an engine
considerably and will therefore require an elevated MAP relative to running with more
“typical” intake cam durations and valve closing points. This can be advantageous at
low engine loads due to the reduction of pumping losses, however, at high engine
loads and with boost (“Deep Miller”) the success of EIVC and LIVC is heavily dependent
on the boosting strategy employed.
In this thesis the viability of EIVC and LIVC and its synergies have been assessed
with the following future technologies/techniques:
• Downsizing
• Downspeeding
• Turbocharging
• Supercharging/eboosting
• Increased geometric CR
• Internal Exhaust Gas Recirculation (EGR)
By means of varying the following:
• Intake cam duration
• Intake Maximum Opening Point (IMOP) timing
• Exhaust Maximum Opening Point (EMOP) timing
• MAP
• Exhaust Back Pressure (EBP)
• Geometric CR
4
Low load and speed EIVC and LIVC have been studied before, however,
high/maximum load EIVC and LIVC is still the subject of investigation. It was not until
GDI was considered a mainstream technology that EIVC and LIVC operation became
feasible.
1.2 Objectives
The objective of this study is to ascertain the benefits of EIVC and LIVC on a
downsized engine with the intention of using the de-throttling effect at low engine
load to optimise fuel economy, and the over-expansion effect at high load to mitigate
knock at high engine loads.
EIVC and LIVC shall be employed by means of cam phasing as well as the
employment of different duration intake cam profiles to achieve different effective
CRs. Another objective of this test work is to determine the maximum load that can be
achieved through the use of EIVC and LIVC to test its synergies with engine
downspeeding. The impact of other variables on the successful operation of EIVC and
LIVC shall also be assessed, including speed, load, EBP and geometric CR.
A single cylinder GT-Power model was developed during the course of this
project in an effort to better understand the differences in engine breathing when
EIVC and LIVC are employed. This model will also be used to determine the differences
in performance between the single cylinder engine and the 3 cylinder engine.
1.3 Outline of Thesis
Immediately following this brief introduction is a literature review which

summarises all of the relevant literature in the field of downsizing and effective CR
5
reduction through both EIVC and LIVC techniques. The current and future emissions
legislation shall also be discussed, as well as the details of the current and future laws
regarding CO2 emissions and other emissions too. An overview of the new
technologies that can now justifiably be used on current ICEs will also be discussed.
The third chapter will cover the experimental facility that has been built at
Brunel University for this project, as well as detailing all of the instrumentation and
sensors that have been installed to the engine. An overview of the data acquisition
software used will also be given in this chapter, as well as all of the equations used to
calculate various engine operating parameters. This chapter will also detail all of the
techniques used to setup and install the engine instrumentation accurately.
Chapter 4 will comprise the development of a one-dimensional single cylinder

analytical engine model, as developed in GT-Power [4]. It includes an explanation as to
why it was developed, details of the gas dynamic and combustion models used, and
also explains how knocking combustion has been predicted and how this was
implemented into a closed loop control system of combustion phasing. The techniques
used to calibrate this knock model will also be described. It also contains a brief
description of the 3 cylinder engine GT-Power model developed by Mahle Powertrain.
Chapter 5 comprises an analytical study of the Miller cycle based on the single
cylinder GT-Power model described in chapter 4. It will describe how a Design of
Experiments (DoE) based test plan was implemented to simultaneously analyse the
impact of several different variables. The results of 2 studies into the use of Deep
Miller to optimize fuel economy and the use of Deep Miller to maximize engine load
are also presented.
Chapter 6 consists of a performance evaluation of the single cylinder

experimental engine. This comprises test plan descriptions and results for studies on
the impact of fuel age on knocking characteristics and autoignition tendency, and the
impact of coolant flow rate on engine knocking and autoignition tendency.
Comparisons between the single cylinder experimental engine and the Mahle 3
cylinder 1.2l downsized engine on which the single cylinder experimental engine is
based will also be made in this chapter.
6
Chapter 7 consists of an experimental investigation into the Miller cycle and its
impact on engine performance and fuel consumption. This chapter contains 2 studies,
the first on the de-throttling potential of the Miller cycle and the second on the low
speed high load knock mitigating properties of Deep Miller. The results of both of
these studies will be presented, and their potential to improve fuel economy assessed.
Summary and conclusions from this study are given in chapter 8.

Recommendations for future work will also be given in this chapter. Please note that
all values of EBP given in this thesis are in absolute pressure.
7
2
Chapter 2
Literature Review
2.1. Introduction
Ever increasing fuel prices and climate change have shifted the focus of
powertrain development to the reduction of fuel consumption and emissions rather
than focusing on outright performance. The transport sector has been found to be a
high contributor to global emissions (approximately 21% according to Fuglestvedt, et al
[5]) of CO2 (with road vehicles representing 70% of all transport sector CO2 emissions
in the EU in 2008 [6]), and is also responsible for 55% of the world’s oil consumption
[7]. Reduction of emissions is also very important given concerns over climate change
and an increase in respiratory diseases in humans.
The key transport sector emissions that have been targeted as having the
biggest impact on the environment are carbon monoxide (CO), nitrogen oxides (NOx),
unburned hydrocarbons (HC) and particulates (PM). Recent attention has also turned
to emissions of CO2 although there is – at this moment in time - no legislation
stipulating maximum permissible quantities of CO2 for a single vehicle, but, future EU
legislation (EC 443/2009 [8]) dictates fleet average emissions of CO2 for all new cars
from 2015 onwards should be no more than 130g/km, and any cars in contravention of
this will result in the manufacturer of that car being fined. This limit could be further
reduced to 95g/km from 2020 onwards, and an even heavier fining structure imposed
for new vehicles in excess of this [9].
8
Numerous technologies that will reduce emissions and fuel consumption have
been investigated and researched with mixed results. Many of the technologies involve
optimisation of existing fossil fuel burning powertrains, whereas other systems utilize
different sources of energy such as electrical, hydrogen or fuel cell. The current fuel
infrastructure favours the optimisation of existing fossil fuel combustion technologies.
Hydrogen or fuel cell systems would require a significant fuel infrastructure
reconstruction. Extensive research is also being carried out into hybrid powertrain
designs. The success of a powertrain design also depends heavily on the additional
cost.
Some of these alternative powertrain designs have been developed to the

point where mass production is feasible and, crucially, existing fuel infrastructure has
not needed to be modified. Hybrid powertrains have been the most successful to date
with several vehicles from several big brand names utilising this technology and
showing great commercial success. Other powertrain technologies that have proven to
be successful are downsizing and boosting, which is another area that has been
researched heavily in recent years. Downspeeding is also seen as a way of improving
fuel consumption and is used in conjunction with downsizing. Downspeeding reduces
frictional losses and also permits higher geometric CRs (greater stroke lengths) to be
used.
Downsizing of an engine is the most promising near term means of improving

fuel consumption. Downsizing refers to the process of replacing a large Naturally
Aspirated (NA) engine with a smaller turbocharged engine of the same, or similar,
power output. Downsizing of an engine offers the greatest performance benefit at part
load operating points. Part load operating points are shifted to higher, more efficient
Brake Mean Effective Pressures (BMEP) in the engine speed load map thus reducing
the frictional and pumping losses for the engine. The emissions of downsized engines
generally speaking are also reduced, this depends greatly on the fuel injection and
exhaust after treatment strategies employed on the vehicle though.
9
2.2 Gasoline Engine Emissions
2.2.1 Current Worldwide Emission Legislation
There are numerous emission standards enforced around the world (an
overview of all of them is given in [10]), the majority of which concern the following
pollutants:
• Carbon monoxide (CO)
• Nitrogen oxides (NOX)
• Unburned hydrocarbons (HCs)
• Particulate matter (PM)
• Particulate number (PN)
At this moment in time there is no legislation on the horizon that dictates

maximum permissible quantities of CO2. This is because CO2 is not regarded as a
pollutant, however, measures are in place to reduce the emissions of CO2.
There are currently 4 major types of pollutant legislation enforced around the
world; European standards (Euro, EC), American standards (EPA, CARB, CAAA),
Japanese and Brazilian standards. The applicability of these standards depends on the
type of vehicle in question.
10
European standards:
The current standards applicable to European vehicles are the Euro standards
[11]. There are different categories of Euro standards denoted by number (Euro 1/2/3
etc.) with the higher number corresponding to the more rigorous standards. The date
of adoption of these mandatory standards is as shown in Table 2.1. The dates of
adoption depend on the vehicle type, Gross Vehicle Weight (GVW) and Reference
Mass (RM). Vehicle type descriptions are shown in Table 2.2.
American standards:
The US federal standards [12] are currently enforced in America (with the
exception of Brazil who have their own standards). There are at present 2 sets of
standards denoted tier I and tier II, tier II is what is currently being enforced for light
duty road vehicles. The tier II standards can be split up into sub categories denoted bin
1 to bin 8. Different emissions criteria apply to each bin, the lower the bin number the
more stringent the criteria, hence, the cleaner the vehicle. Newly registered cars can
fall into any 1 of the 8 bins as long as the vehicle manufacturer still meets a fleet
average of bin 5.
A tier III [13] has also been introduced recently stipulating that fleet average
fuel economy for cars must be 42mpg, and 26mpg for light duty trucks with a GVW not
exceeding 8,500lbs. The fleet average fuel economy for all light duty vehicles (GVW not
exceeding 10,500lbs) for a manufacturer must be 35.5mpg. This is due to be
implemented in 2016. Failure to achieve this fuel economy will result in vehicle
manufacturers incurring fines.
California has additional, more stringent, emissions criteria that must be

satisfied. This criterion was stipulated by the California Air Resources Board (CARB)
[14]. As of 2008 the CARB standards had also been adopted in 12 other American
states (Lyons, et al [15]).
11
Table 2.1 Euro Standards Implementation Map [10]
Table 2.2 Vehicle Categories Applicable to Euro Standards [10]
12
Japanese standards:
The current Japanese standards [16] are the Post New Long Term (PNLT) and
have been enforced since October 2009. These standards supersede the New Long
Term (NLT) standards which in turn superseded the New Short Term (NST) standards.
Brazilian standards:
The Brazilian standards [17] are a mix of the European and American standards
in that the FTP-75 cycle is adopted whilst the emissions themselves are based on EURO
standards. The reason why Brazil has adopted different standards is because the fuel in
Brazil contains high quantities of ethanol and the standards must be altered
accordingly.
These 4 standards represent what is currently adopted throughout most of the

world. Many countries adopt different levels of this criteria adopted some years ago by
the region of the emission standards origin (China for instance have adopted Euro 4
standards whereas the EU currently conforms to Euro 5b standards). These emissions
standards are comparable with each other and impose similar restrictions on the same
emissions. An overview of the main criteria of these standards (applicable to light duty
vehicles) is shown in Table 2.3.
The largest difference between them exist between the European (and
Japanese) and American standards. The American standards show a greater focus on
Non-Methane Organic Gasses (NMOG) and formaldehyde (HCHO) than HC and Non-
Methane Hydrocarbons (NMHC).
The CO emissions are also much less rigorous in the American standards than in
the Japanese and European standards. Although downsizing of engines in American
light duty vehicles is beginning to occur, the extent of downsizing being applied is not
13
as considerable as that being applied to European and Japanese engines. The reason
for this is fuel is considerably cheaper in America (58.3 pence per litre in America [18]
compared to 132 pence per litre in the UK [19]), larger engines are therefore more
affordable and commonplace in comparison to European and Japanese engines, and
the less rigorous American CO emissions requirements reflect this fact.
There are various reasons why the emissions shown in Table 2.3 require close
regulation. Sher [20] has documented extensively the reasons why these emissions are
monitored. What follows is a brief explanation of what the emission is and what its
impact is on the environment and general health and wellbeing, as well as an
explanation of the processes that bring about that particular emission. Heywood [21]
provides a very detailed review of emissions, and Cole, et al [22] provides a review of
the differences in emission characteristics between GDI and Port Fuel Injection (PFI)
applications.
14
Table 2.3 Current Worldwide Emissions Standards
Standard Standard Date of HC + CAFE CO2

Standard level type Introduction Unit CO HC NMHC NOx NOx PM PN HCHO NMOG (mpg) (g/km)
SI 09/2011 g/km 1 0.1 0.068 0.06 0.0045
5b
CI 09/2011 g/km 0.5 0.18 0.23 0.0045 6e11
SI 09/2014 g/km 1 0.1 0.068 0.06 0.0045 6e11
6-1
European CI 09/2014 g/km 0.5 0.08 0.17 0.0045 6e11
SI 09/2017 g/km 1 0.1 0.068 0.06 0.0045 6e11
6-2
CI 09/2017 g/km 0.5 0.08 0.17 0.0045 6e11
EC 443/2009 2015 130
Bin 6 2004 g/km 2.1131 0.0497 0.0062 0.0093 0.0466
Tier II Bin 5 2004 g/km 2.1131 0.0311 0.0062 0.0093 0.0466
Bin 4 2004 g/km 1.3052 0.0249 0.0062 0.0068 0.0435
American LEV 2004 g/km 2.1131 0.0311 0.0062 0.0093 0.0466
CARB ULEV 2004 g/km 1.0566 0.0311 0.0062 0.0050 0.0249
SULEV 2004 g/km 0.6215 0.0124 0.0062 0.0025 0.0062
EPA 2016 35.5 156
PC (SI) 09/2009 g/km 1.15 0.05 0.05 0.005
Japanese PNLT
PC (CI) 09/2009 g/km 0.63 0.024 0.08 0.005
Brazilian PROCONVE L6 Otto 01/2014 g/km 1.3 0.3* 0.05 0.08 0.02
Diesel 01/2013 g/km 1.3 0.05 0.08 0.025
* Applies to natural gas vehicles only
15
2.2.1.1 Carbon Monoxide (CO)
CO is toxic to humans. It combines with haemoglobin in the blood stream to

form CO2, thus considerably reducing the oxygen carrying capacity of blood and
depriving tissues of O2. CO can result in dizziness/light headedness, and large
quantities of it can lead to loss of consciousness and eventually death (Sher [20]).
CO production is influenced heavily by the Air Fuel Ratio (AFR). Harrington, et al

[23] found that when running with an equivalence ratio > 0.95 CO production can be
seen to increase dramatically. Conversely, mixture inhomogeneity may give rise to high
rates of CO production due to locally rich areas. The oxidation rate of CO is influenced
by burned gas temperatures and temperature gradients during the expansion and
exhaust strokes. Shimotani, et al [24] found that GDI offers no significant benefit over
PFI with regard to CO production at normal running conditions.
Due to the fact diesel combustion is always lean, CO production is minimal and
does not impose as significant an operating constraint as it does on gasoline engines.
Bowman [25] derived the chemical kinetic process responsible for the
production of CO which is as shown below:
RH → R → RO2 → RCHO → RCO → CO
where R is a hydrocarbon radical. The CO is slowly oxidised to CO2 in the exhaust by

the following reaction:
CO + OH = CO2 + H
Johnson, et al [26] found that the concentration of CO present at the end of the
exhaust stroke, although lower than in the cylinder, is still significantly greater than at
equilibrium conditions.
16
2.2.1.2 Unburned Hydrocarbons (HC)
HC emissions typically result in respiratory issues caused by

bronchoconstriction, this results in coughing, dry throat and wheezing. They can also
result in eye irritation. HC emissions can react with oxides of nitrogen to form ozone
(O3) which is detrimental to human health (Sher [20]).
HC is a very general term that applies to many different types of organic

hydrocarbon compounds that are emitted from vehicles. The term HC applies to
paraffins, olefins, acetylene and aromatics, and other types of hydrocarbon, all of
which can be found in exhaust gasses. Various HC emissions can contribute to
photochemical smog production and are considered quite harmful. Hydrocarbons are
therefore classed by their reactivity with highly reactive HC emissions receiving the
strictest regulation in the case of the American standards (formaldehyde receiving its
own emissions classification due to its propensity to cause respiratory and eye
irritation).
The main reason for HC emission production is usually incomplete combustion

or through the use of fuel enrichment. There are numerous other reasons for HC
emissions as well. Impingement on cylinder surfaces has been shown to contribute to
HC emissions. Zhao, et al [27] found that this is highly dependent on the injection
strategy employed (wall guided, spray guided etc.) and injector orientation relative to
the cylinder axis and the surfaces of the cylinder (particularly the piston crown). Daniel
[28] determined that flame quenching also contributes to HC emissions on cylinder
surfaces (due to the incomplete oxidisation of fuel). A study by Wentworth [29] found
that HC emissions are also produced from crevices where the flame front cannot
propagate to oxidise the fuel. Lubricating oil is also a known source of HC emissions as
it can be partially oxidised along with the charge.
HC emissions will increase significantly when fuel enrichment is employed, due

to the fact unburned fuel is unavoidable. However, due to the major drive cycles
17
requiring the engine to run at just low load points, enrichment has a negligible impact
on overall HC emissions.
2.2.1.3 Nitrogen Oxides (NOx)
NOx refers to 2 specific emissions; nitric oxide (NO) and nitrogen dioxide (NO2).
NO has been found to react with haemoglobin having potentially lethal consequences.
NO2 is an irritant and reacts with the lungs and, if inhaled in a certain concentration,
can cause severe damage to them (Sher [20]).
Stone [30] reports that for gasoline engines NO is produced in a far greater
quantity than NO2 in the cylinder itself through a basic chemical equilibrium analysis.
Miller, et al [31] confirmed that NO is principally formed through oxidation of
atmospheric nitrogen, but can sometimes form through oxidation of the nitrogen in
the fuel as well, this is in addition to that which is formed by atmospheric nitrogen. The
NOx is formulated by means of the well established Zeldovich mechanism (Zeldovich
[32] with an addition by Lavoie, et al [33]) which is as follows:
O + N2 = NO + N
N + O2 = NO + O
N + OH = NO + H
Burned gas temperature is a very large contributory factor in NOx formation.

Higher burned gas temperatures usually result in greater NOx concentration. EGR is
employed to suppress NOx and is a particularly powerful NOx suppressant, as it greatly
decreases the charge temperature. Komiyama, et al [34] determined that employing
15 - 25% EGR results in a significant reduction of NOx concentration. Spark timing was
found to have a very dramatic affect on the burned gas temperature in the cylinder
and has a great influence on NOx production.
18
2.2.1.4 Particulate Matter (PM)
PM consists of organic particles (including soot) and sulphates. The danger PM

imposes on human health greatly depends on the type of PM and the size. PM has
been linked to respiratory diseases and, in the case of very fine particles (≤10nm)
contamination of the lungs which the body itself cannot clear due to the particle being
too small (Sher [20]).
Organic particles (soot) occur in greater quantities in diesel engines than they
do in gasoline engines, depending on how the engine has been configured. Therefore,
in some circumstances, diesel engines require Diesel Particulate Filters (DPFs) to
reduce them.
Sulphide emissions are caused by sulphur present in the fuel. Kummer [35]
found that three way catalytic converters can increase the sulphide emissions by
oxidising sulphur dioxide (SO2) to sulphur trioxide (SO3). When emitted from the
exhaust and into the atmosphere this can create aerosol sulphuric acid (H2SO4).
Sulphuric acid is a major constituent part to acid rain.
The majority of PM emissions are simply by-products of a normal combustion

process and therefore impossible to eradicate completely. Black smoke can however
be produced in a gasoline engine when the charge is too rich.
2.3 Current Automotive Engines, Technologies and Trends
Although emissions standards around the world are striving toward the Zero
Emissions Vehicle (ZEV), scientists and engineers are still someway short of achieving
this on a global or even a fleet-wide scale. For now and the medium to long term
future this is not achievable, but new technologies and techniques have allowed
emissions to be dramatically reduced. What follows is a summary of these
technologies and techniques.
19
2.3.1 Hybrid and Electric Powertrains
At present there are 3 main categories of hybrid and electric vehicles; Hybrid
Electric Vehicle (HEV), Fuel Cell Electric Vehicle (FCEV) and Battery Electric Vehicle
(BEV). FCEVs can BEVs are similar in that if the energy they use comes from a
renewable source, and if one ignores the CO2 cost of manufacturing the vehicle, some
vehicle manufacturers can claim to have produced a true ZEV.
The Nissan Leaf, Peugeot iOn and Mitsubishi i-MiEV are examples of BEVs that
have been put into production, and if powered from a solar or nuclear energy source
they can be considered ZEVs. Owing to the fully electric powertrain BEVs can be
incredibly efficient (up to 80% according to Helms, et al [36]) BEVs do suffer some
considerable drawbacks however, such as a very limited range (a constraint imposed
by the current generation of lithium ion battery technology) and a very long charging
time using a typical domestic power supply.
The Honda FCX Clarity and Hyundai ix35 FCEV are examples of FCEVs. Like BEVs
they have the potential to be classed as true ZEVs. Thomas [37] documents that owing
to the much higher power density of compressed hydrogen FCEVs typically have a
much greater range than BEVs, they also have a refill time comparable to fossil fuel
powertrains. They are however less efficient than BEVs (48% efficient) and face
additional logistical constraints such as how to safely transport highly pressurised
hydrogen.
Both BEVs and FCEVs are lacking the energy infrastructure required to be
successful in the short term and are both prohibitively expensive to justify over a fossil
fuel powertrain at this moment in time. HEVs, though not ZEVs face no such constraint
and can considerably improve emissions over fossil fuel only powertrains. The Toyota
Prius is one such example of a HEV that has seen great commercial success, which has
prompted an influx of other HEVs from large manufacturers such as Honda, Mercedes,
Lexus, Ford, BMW and Volkswagen to name just a few. Plotkin, et al [38] determined
that the efficiency benefit offered by hybrids was approximately 25% in 2005, but this
20
is declining due to advancements in fossil fuel powertrain technologies (such as engine
downsizing) closing the gap.
Other hybrid technologies are also being researched such as range extending
ICEs designed to be employed on BEVs to generate electricity, resulting in a new
classification of vehicle known as the Extended-Range Electric Vehicle (E-REV). One
such example of this is the Mahle range extender engine (Basset, et al [39]). Another
novel concept is the air hybrid system as developed at Brunel University which
regenerates approximately 15% of a vehicles kinetic energy that would have been lost
under braking (Lee, et al [40]).
2.3.2 Fossil Fuel Powertrains
While BEVs and FCEVs offer great potential in terms of reduced emissions there
is still a great deal of work that needs to be done to make them a true global transport
solution. Until such a time as these technologies are ready, fossil fuel ICEs will still be
the dominant powertrain for automotive applications.
Ever tighter emissions standards and ever higher fuel prices are forcing fossil
fuel powertrains to become more versatile however, for instance, they are now
applied in HEVs and E-REVs, and will continue to be applied in far different applications
for the rest of their lifespan. This means that both gasoline and diesel ICEs will play a
pivotal role in automotive applications for many years to come, and both will need to
continue to evolve to satisfy the emissions legislation.
The key to this (particularly for diesel applications) is exhaust aftertreatment.

Gasoline ICEs require a 3-way catalytic converter in their exhausts to reduce emissions
of CO, NOx and HCs, whereas diesel ICEs require a Diesel Oxidation Catalyst (DOC)
because they run lean. Unlike 3-way catalytic converters, DOCs do not actively reduce
NOx emissions, so EGR is typically used instead to replace un-needed air with exhaust
gasses to eliminate any unnecessary NOx forming reactions.
21
Diesel ICEs can also suffer from increased soot emissions in comparison to
gasoline, so in some circumstances a DPF is required as well as a DOC. Euro 6 emissions
standards will require additional equipment still to reduce NOx emissions such as SCR
or Lean NOx Traps (LNT). This will invariably make diesel ICEs more expensive than
gasoline ICEs.
This will likely have a detrimental effect on the popularity of diesels in the
future and could see a rise in the popularity of gasoline ICEs. Gasoline ICEs also have
the added advantage of many new cost effective technologies that can be used to
greatly improve performance and emissions. These technologies are described in
section 2.3.3
2.3.3 Gasoline Engine Developments
In previous years, improvements in fuel economy for gasoline fuelled engines

have been achieved in the following ways:
• Downsizing and downspeeding (reducing gear ratios)
• Developing GDI fuel injection technology
• Increased engine specific power output (turbocharging/supercharging/eboosting)
• Optimising In-cylinder flow (allowing increased homogeneity or stratified charge)
• Using lighter materials (aluminium or Compacted Graphite Iron (CGI))
• Introducing VVT systems
• Using EGR at part and full load with boost
• Using alternative thermodynamic operating cycles (Miller/Atkinson, HCCI/CAI)
• Increasing geometric CR
22
This list is not exhaustive and it has not been possible to develop each of the
techniques listed in isolation. Many of the technologies above have synergies with
each other.
Extensive use of aluminium is now made in engine manufacture. Alloys

(particularly those of magnesium) are also used more for component design because
of their high strength properties. Austenitic steel has also received application in
exhaust manifolds where high temperature endurance, low weight and good thermal
expansion properties to prevent catalyst blow by are required. CGI could also find
application in block design for downsized gasoline engines in the near future, due to
increased cylinder pressures and loads.
Stratified charge has already been employed on production engines and is now
in its second generation (stratified charge with spray guided direct injection) and has
been documented by Schwarz, et al [41] to result in improvements of Brake Specific
Fuel Consumption (BSFC) of up to 20% on the New European Driving Cycle (NEDC).
Stratified charge shows great promise for reducing BSFC but can only be employed at
low - medium loads (as documented by Missy, et al [42]), and requires a DeNOx
catalyst or LNT to pass emissions regulations. Ricardo’s Lean Boost Direct Injection
(LBDI) (Lake, et al [43]) is an alternative concept that also allows the engine to run
homogeneous lean at higher loads and with boost. This is better suited to downsized
applications and can also result in a BSFC benefit of up to 20% when applied to a
downsized turbocharged engine and compared to a larger NA engine of equivalent
power and torque.
Homogeneous Charge Compression Ignition (HCCI) or Controlled Autoignition

(CAI) is a medium term technology that enables an engine to run homogeneous lean.
Unlike stratified charge and LBDI, combustion is not initiated by a spark but by
compression ignition. The mechanism of combustion is also different. Spark initiated
combustion relies on a flame front propagating through the end gas which is
susceptible to quenching, flame stretch and turbulence. HCCI results in the entire end
gas region undergoing simultaneous reactive envelopment. In spite of the fact HCCI is
a lean burn technology Zhao, et al [44] documented a significant reduction of NOx
23
emissions to the extent that a LNT is not required, which gives it a large advantage
over stratified charge and LBDI operating regimes. CO emissions are also massively
improved in comparison to stratified charge and LBDI, while HC emissions are similar.
Osborne, et al [45] found that HCCI yielded an 8% improvement in ISFC. However, HCCI
is restricted to low load operation only at this moment in time.
EGR is a well established means of improving fuel economy and dramatically

reducing exhaust emissions. At low load EGR can be used to partially fill the cylinder
with exhaust gas, thus reducing the size of the pumping loop. EGR can also be used at
high load Wide Open Throttle (WOT) conditions too. Cooled exhaust gasses have the
effect of reducing peak combustion temperatures which allows the combustion to be
advanced considerably. Cairns, et al [46] documents reductions of CO and HC
emissions of up to 70% and 80% respectively. In regions where enrichment is required
WOT EGR can yield an improvement in BSFC of up to 17%. Significant NOx reductions
have been seen when comparing WOT EGR to excess air too, but not so much
compared to stoichiometric operation.
The Atkinson/Miller cycle process is a near term means of improving fuel

economy and has already seen application on some production engines. Both Atkinson
and Miller cycles work on the same principle of changing the compression ratio
relative to the expansion ratio so the expansion ratio is greater than the compression
ratio. This technique is known as over-expansion. This technique has only become
feasible relatively recently due to the mechanical complexity of making the expansion
ratio longer than the compression ratio. This has been circumvented by the use of fully
variable valve trains and cam phasing devices by closing the intake valve slightly earlier
or later than BDC reducing the effective compression ratio relative to the expansion
ratio.
Although the full benefits of the Atkinson/Miller cycle process can only be
realised with a fully variable valve train, a benefit to BSFC can be obtained simply by
changing the intake cam profile, with no further work required (depending on the
application and the type of engine). Intake cam phasing (as is employed on many
automotive engines) can increase this BSFC benefit further still, while circumventing
24
the issues associated with reduced volumetric efficiency for higher load cases. Of all of
the technologies above it is also the cheapest to incorporate, with no need for any
additional exhaust after treatment systems beyond that required for a standard Otto
cycle engine. The Atkinson/Miller cycle process also offers synergies with all of the
technologies mentioned above (with the arguable exception of EGR).
The Atkinson/Miller cycle process has only really been automotive production
engine feasible since the introduction of VVT and GDI. Perhaps the most promising
aspect of the Atkinson/Miller cycle is its knock mitigating properties at high load.
Of all of the technologies discussed in this section it is downsizing that is of

greatest interest at this moment in time for the reasons described in section 2.3.4.
2.3.4 Downsizing and Turbocharging
Downsizing is already seeing widespread interest with many downsized engines

already in production (Volkswagen 1.4 R4 TSI, Bentley 4.0L V8, BMW 2.0L TVDI). At the
time of writing the most aggressively downsized production engine is the Ford
Ecoboost with a peak BMEP of over 25 bar (Friedfeldt, et al [47]). Many prototype
aggressively downsized engines have been designed such as the Mahle 1.2l downsized
engine (Hancock, et al [48]), the MCE-5 VCR engine [49] and the Jaguar Land Rover
Ultraboost engine (Turner, et al [50]), all of which surpass the extent of downsizing
demonstrated by the Ford Ecoboost by a significant margin.
The Mahle 1.2l downsized engine was designed to replace a 2.4l V6 NA PFI
engine and has achieved the same power output as the V6 and yielded an increase in
efficiency of 30%. It can also run at 30 bar BMEP, and in a study by Lumsden, et al [51]
it was run in excess of this. Turner, et al [50] expects similar benefits (35%) with the
Jaguar Land Rover Ultraboost engine, which is downsized to a greater extent still
(replacing a 5.0l V8 with a 2.0l I4 engine) and is required to run at a peak load of 35 bar
BMEP, and 25 bar BMEP at 1000rpm.
25
Downsized engines reduce the number of cylinders on an engine and the
overall engine displacement, this results in a considerable reduction of frictional losses.
An additional benefit is obtained through the concept of load point shifting too (a
smaller engine must produce a higher BMEP than a larger engine for the same power
output) which dramatically reduces the part load pumping losses.
In order to compensate for a lack of displacement the engine must be boosted

(usually turbocharged) to increase the specific power output. Unlike most conventional
turbocharged engines the turbocharger must provide sufficient boost from a very low
engine speed because a reduction of swept cylinder capacity will impact all speeds, not
just high. The selection of boosting system is therefore of great importance, as it not
only has to provide sufficient boost from low speeds but it also has to offer very little
transient delay.
Turbocharger technology has also come a long way in recent years. The
introduction of novel boosting strategies such as a 2-stage system (known as
“twincharging”) and wastegate boost pressure regulation are examples of 2
technologies that have made downsizing possible. Twincharging involves
complementing either a small turbocharger or supercharger with a large turbocharger
or supercharger either in series or in parallel. If a small supercharger is used it is
usually electrical or connected to the crankshaft by means of an electromagnetic
clutch so it can be disengaged at higher speeds to prevent high parasitic losses
occurring. This system is capable of providing high quantities of boost across the entire
engine operating envelope with minimal lag time. This concept first entered
production in 1985 on the Lancia Delta S4 Stradale [52]. Copeland, et al [53] detail a
similar twincharging system that was determined to be optimal on the Jaguar Land
Rover Ultraboost engine, for this engine a turbocharger has been employed for the low
pressure stage and a supercharger employed for the high pressure stage.
Wastegate boost pressure regulation refers to the process of using a wastegate

to divert flow away from a turbocharger turbine when it is not required, thus reducing
the EBP. Conversely it can also be used to limit the maximum pressure ratio and speed
of a turbocharger by purging exhaust gas that would ordinarily go through the turbine.
26
This permits the use of a smaller diameter turbine (good for transient response)
without sacrificing the overall pressure ratio. One such wastegate boost pressure
regulation system is the Mahle E-Actuator (Fraser, et al [54]) which offers improved
transient response and reduced pumping losses at part load conditions.
Both of the above systems have made turbocharging a much more viable
option, even for low engine speeds, and have improved it to the point where
conventional supercharging offers little to no advantage any more. It may even be
considered inferior to modern turbocharging due to the fact that superchargers
typically demonstrate lower adiabatic efficiency than turbochargers and therefore
require bigger intercoolers.
Fraser, et al [55] of Mahle powertrain investigated the impact of several

different boosting systems on the Mahle 1.2l downsized engine and found that a
twincharger system provided highest low speed load, while the turbocharger with
electric supercharger system gave the best transient response.
Overcoming the boosting issues at low speed is not the only problem with
downsizing however. Running at high load and low speed greatly increases the
propensity for an engine to experience autoignition in both knocking combustion and
Low Speed Pre-ignition (LSPI) form. Amann, et al [56] provides a definition of LSPI and
the speeds at which it applies. LSPI is basically pre-ignition that only occurs at engine
speeds of 3000rpm or less, and completely absent above 3000rpm. According to the
description in Heywood [21] this is more likely at lower speeds due to reduced charge
motion and also due to the elevated pressure and temperature histories during the
compression process. This phenomenon has the effect of greatly reducing the
maximum load at which the engine can run.
Anderson, et al [57] found that the Atkinson/Miller cycle process exhibits end
gas temperature reducing properties which may reduce the propensity for LSPI and
knocking combustion. The Atkinson/Miller cycle will also theoretically improve
efficiency at higher load points too. This is due to the fact lower end gas temperatures
will result in more spark advance relative to a baseline Otto case, which will
accordingly result in lower exhaust gas temperatures. This will reduce the amount of
27
enrichment required for component protection which will improve fuel economy and
HC emissions markedly at high speed and high load.
In spite of a great deal of interest very little research has been carried out on
this concept known as Deep Miller. For this study “Deep Miller” is defined by using the
Miller cycle at engine loads of at least 15 bar BMEP. One of the aims of this study is to
ascertain exactly what benefit can be obtained by running with the Miller cycle at high
load and low engine speed.
2.4 Atkinson/Miller Cycle
2.4.1 Introduction
With the rate at which the global transport sector is growing the automotive
industry and ICEs have been forced to evolve in a very short space of time. Engine
downsizing is the key to this rapid evolutionary process, as the infrastructure and
technology required already exists to make it a reality. GDI is also a key to this
evolution as it affords a great deal of freedom over how an engine can be run.
Downsizing, though relatively easy to implement, has its own set of unique problems.
High specific power outputs are what downsizing relies upon in order to be successful,
but knocking combustion is a side effect of running at high specific power, so in order
to elevate specific power, techniques for increasing the knock limit or Borderline
Detonation (BLD) limit must be researched.
The Miller cycle process is one near term solution to increasing the BLD limit of
an engine. It also has synergies with many other techniques to increase the BLD limit of
an engine. It does face challenges though such as whether or not in-cylinder charge
motion is affected by closing the intake valve either before or after BDC and what
impact this has on combustion stability and Coefficient Of Variation (COV) of the
Indicated Mean Effective Pressure (IMEP). There are also many questions left to be
answered on the concept of Deep Miller and whether or not it benefits the engine, and
28
whether or not an increase in the geometric CR can be utilised by running with a
reduced effective CR.
2.4.2 History
The Miller cycle was patented by an American engineer called Ralph Miller in
1947 (Miller [58]). It is very similar to the Atkinson cycle with the biggest difference
between them being the Miller cycle requires a supercharger to compensate for the
power lost by closing the intake valve away from BDC. There are however points
where the definition of the 2 become blurred.
There is a great deal of literature on the benefits offered by the Atkinson cycle,
and the benefits offered by a reduced pumping loop size. The Atkinson cycle engine in
spite of its relatively complicated geometry received a great deal of attention from a
very early stage. The British gas company built over 1000 2-stroke Atkinson cycle
engines between 1886 and 1893.
Various attempts were made to implement an Atkinson cycle in an automotive

engine with varying degrees of success. The most recent attempt was carried out by
Honda with the extended expansion linkage engine (EXlink) [59]. There are 2 different
types of Atkinson cycle engine, the original style Atkinson engine that achieves
different compression to expansion ratios by employing linkages to give different
geometric compression to expansion ratios, and the modern type which uses intake
valve closure timing to provide a different effective compression ratio and the
expansion ratio is still dictated by engine geometry. It is the second, modern, type of
Atkinson cycle that blurs the boundary with Miller cycle. Miller cycle actually stipulates
that the effective compression ratio is controlled by the Intake Valve Closing (IVC)
timing, but it also stipulates that a supercharger must be present.
In the majority of the literature (particularly in recent years [57][60-74]) the

Atkinson process referred to is the modern process, it is sometimes referred to as
Miller cycle and in other papers it is referred to as the Atkinson cycle. It can also be
29
referred to as EIVC, LIVC and Extended Expansion (EE). In some cases it can be referred
to as Deep Miller, this definitely applies to the Miller cycle only and the deep part
refers to the fact that very high levels of boost pressure are employed rather than
boosting simply to make up for the loss of volumetric efficiency incurred by closing the
intake valve away from BDC. For the sake of clarity any system that achieves effective
CR reduction by means of phasing the valves relative to BDC, whether NA or under
forced induction, shall henceforth be referred to as the Miller cycle.
The Miller cycle was first utilised in large engines for marine or power
generation applications. Very little regard for emissions was given then, the Miller
cycle was employed purely as a means of increasing power output for greater fuel
economy. In later years it gained the interest of manufacturers of smaller, more
mobile applications such as earth movers and other large off-road vehicles. Miller cycle
was also considered for use with Low Heat Rejection (LHR) engines. Ishizuki, et al [75]
of Komatsu started researching diesel engines that employed the Miller cycle (or
“extended expansion”) in the mid 1980s.
With the technologies available in those days Miller cycle was not feasible due
to problems with excessive smoke emissions at part load conditions associated with
reduced EOC temperatures. This was mainly due to the fact there was no convenient
way to change the intake valve timing for varying degrees of Miller cycle in real time
with engine load. Also EGR was found to give a similar performance benefit without
the need for novel cam phasing technologies.
The US military also showed some interest in Miller cycle LHRs for tank
applications slightly later in 1995 (Kamo, et al [76]). LHRs are of great importance to
military vehicles owing to the fact they operate in such hostile environments and are
constantly subject to attack. The majority of ballistic missile weaponry is heat seeking,
lowering the thermal footprint of a vehicle is therefore of great importance and
massively desirable.
The greatest commercial success of the Miller cycle has been in cogeneration
applications (engines generating both power and thermal energy). The Miller cycle has
been employed in natural gas cogeneration systems since 2000 in the Mitsubishi
30
GS16R. This particular series of engines represented a milestone in cogeneration
technology in that it was the first engine to exceed a thermal efficiency of 40% for that
class of generator at that time according to Kado, et al [77].
In 2010 a Miller cycle engine was released for commercial sale designed for use
in locomotive applications. It was developed by General Electric (GE) and is called the
powerhaul, this engines achieves 22 bar BMEP at 1500rpm with a BSFC of just 192
g/kWhr, a 9% reduction over any equivalent engine at that moment in time [78]. Due
to the fact transient response is not such an issue, this engine was turbocharged which
greatly improves the BSFC.
Engines of this type benefit from having a very small range of speeds and loads
to operate at, and can therefore adopt fixed valve timings for Miller, this reduces the
complexity of the engine massively and means the valve timings can be finely tuned for
just a few speeds and loads.
Its adoption on smaller vehicles such as passenger cars and gasoline

applications was slower due to the fact a supercharger was required. The cost of
supercharging an engine proved too great in comparison to the amount of fuel saved
for typical passenger cars. There is the added challenge of making the vehicle run at a
variety of different speed and load points, and a requirement for good transient
response.
Miller cycle for passenger vehicle type applications was still the subject of some
research work at a relatively early stage however. There is not a great deal on Miller
cycle work specifically due to the fact it is basically an extension of the Atkinson cycle.
Research commenced in the mid 80’s with a study from Sakai, et al [79] on the
prospect of using the Miller cycle on gasoline engines.
For all applications the Miller cycle engine was found to improve the thermal
efficiency greatly (in the region of 7–35% over baseline Otto cases depending on the
degree of Miller, and whether or not a turbocharger or supercharger was employed).
The reduction in EOC pressure and temperature was also found to greatly reduce NOx
emissions.
31
The Miller cycle has been employed on some passenger car engines for
production, such as the Mazda KJ-ZEM engine (Goto, et al [80]). The KJ-ZEM engine is a
2.3l V6 with a geometric CR of 10. This engine was used in the prestige Mazda model
of the time, the Mazda Millenia. The variant with the KJ-ZEM Miller cycle engine was
$3000 more than the equivalent Otto cycle engined model, but developed 24% more
power for the same fuel economy [81]. The KJ-ZEM engine was not a great commercial
success due mainly to the fact that maintaining the engine was very expensive, and the
new Lysholm supercharger on the engine was very expensive to replace too.
The Lysholm compressor was developed purely for this application and its
development has been documented by Takabe, et al [82]. It is a supercharger because
no turbochargers of that day could provide the transient response required for
automotive applications. It exhibits a very high adiabatic efficiency at very high
pressure ratios but ultimately still impacts heavily on the overall efficiency benefit in
comparison to a turbocharger.
The majority of work in the 80’s and 90’s was carried out by the Japanese
researchers such as Hiroshi Sakai and his team at the University of Tokyo in
conjunction with mainly Japanese industry [79][83-84]. One of the reasons for this
great Japanese interest is due to concerns over the quantity of NOx emitted in large
cities such as Tokyo. In 1992 a new law was introduced called the “The Law Concerning
Special Measures for Total Emission Reduction of Nitrogen Oxides from Automobiles in
Specified Areas” [85] aimed at radically reducing NOx emissions. Power plants were
also under greater pressure to become cleaner and more efficient, which is the reason
why the Japanese also invested very heavily in the Miller cycle for LHR and power
generation systems too.
In the case of passenger vehicles the price of fuel has risen to the extent that
the technology required for implementation of the Miller cycle is now justifiable for
the gain in efficiency obtained. The widespread introduction of VVT systems is also a
key enabler to the successful implementation of the Miller cycle on passenger vehicles.
With the growing number of downsized and turbocharged engines, boosting is
32
becoming cheaper and is now cheap enough to become viable for typical vehicle
models, not just prestige models.
The trend for engines to become smaller and turbocharged also means
knocking combustion poses a far greater constraint on engine performance than it has
ever done before. The EOC pressure and temperature reducing qualities of the Miller
cycle helps to inhibit knock which enables peak loads to be elevated.
2.4.3 Pumping Losses
One of the biggest reasons for adopting the Miller cycle is the reduction of
pumping losses. The benefit obtained through this is particularly marked at very low
load conditions where the pumping losses are similar in magnitude to the power
output of the engine. Depending on the speed and load condition, and the type of
boosting system employed, the pumping losses can actually become inverted and
produce a Pumping Mean Effective Pressure (PMEP) of less than zero.
In earlier work the maximum reduction in PMEP was very significant and of the
order of 80% in some cases, although this did not always give acceptable performance
[86-89]. The main reason for this is earlier work was mainly focussed on running the
engine completely unthrottled with symmetrical valve timing rather than simply
reducing the throttling. The potential benefit of running completely unthrottled was
very inviting and Theobald, et al [56] estimated it could return a theoretical fuel
economy benefit of up to 24% in the then current driving schedule. It should be noted
that this theoretical fuel economy improvement is not accurate for future engines. The
widespread use of downsized engines has resulted in load point shifting, and as a
result of this the required MAP in a downsized engine for a specific load is much higher
than that for an engine from when the Theobald, et al [90] study was conducted. The
maximum fuel economy benefit for today’s or even tomorrow’s downsized engines can
be estimated but is not actually known to the author’s best knowledge. One of the
aims of this study is to investigate this.
33
In subsequent years the actual PMEP benefit obtained varied significantly
depending on the type of valve actuation system employed and whether or not EIVC or
LIVC was being used. In the work of Tuttle [91-92] EIVC was found to be the optimum
strategy for reducing pumping losses, and the maximum PMEP benefit obtained was
approximately 40%. Other works on EIVC with symmetrical valve timing also confirm
that 40 - 60% is the maximum PMEP contribution reduction that can be obtained for
typical gasoline engine CRs [93-94]. Symmetrical LIVC typically demonstrates a greater
tolerance to effective CR reduction, this enables greater extents of LIVC to be attained
and pumping loop reductions of the order of 80% [86-87]. Although pumping loop
reductions of this size could be attained this yielded no benefit to BSFC, and in some
cases actually returned worse BSFC than the baseline engine/setup.
In 2007 Honda developed a prototype genuine Atkinson cycle engine (Koga, et

al [95]). This engine has fixed valve geometry. The reduction in PMEP peaked at
approximately 37% which is another indicator that this is the maximum extent the
PMEP can be reduced by with fixed geometry in order to still make an engine usable
and perform acceptably.
Another major problem with extreme EIVC and LIVC cases yielding PMEP
reductions of the order of 80% is that the cam profiles required to achieve this would
achieve only very low maximum loads, therefore PMEP benefits of this magnitude
were reserved only for costly Variable Valve Actuation (VVA) and VVT systems. It must
also be taken into consideration that extreme LIVC is much easier to achieve
logistically than extreme EIVC with conventional valves due to the fact that EIVC
requires very short valve open durations and correspondingly only very low valve lifts
can be obtained. This means choked flow through the valves is more likely to occur
thus restricting the maximum PMEP benefit that can be obtained with EIVC.
Many techniques and systems have been employed for increasing the
effectiveness of EIVC and LIVC strategies and for increasing the degree of Miller effect.
These techniques include increasing the geometric CR to compensate for low effective
CRs, employing tumble flaps to induce more turbulence, reducing valve lift and lean
34
combustion. The most commonly employed technique in the literature is asymmetric
valve timing.
Asymmetric valve timing and valve deactivation have been studied extensively
over the years [49][60-66][89]. Valve asymmetry and deactivation are both beneficial
to EIVC and LIVC in most cases, however, the benefit is not always obtained through
PMEP reduction. In most cases valve deactivation and asymmetric valve timing have
been combined with low valve lift. This usually has the effect of reducing the Indicated
Specific Fuel Consumption (ISFC) due to increased charge motion, however, for some
engine/load/valve lift combinations, this can actually have the effect of increasing the
PMEP contribution but still resulting in an improvement in BSFC due to better
combustion [64][67], although the benefit is usually marginal. Cleary, et al [67]
discovered that if the valve lift is too low and the engine load too high it can be
possible for “sonic throttling” to occur. This describes the case when the flow is choked
by the intake valve. This massively increases the pumping work contribution to the
point where running unthrottled is no longer beneficial.
Work on other strategies to improve the effectiveness of EIVC and LIVC with
tumble flaps is also underway with the Mahle cam in cam system (Taylor, et al [64]).
These technologies are not necessarily intended to reduce PMEP directly but mostly
through increasing charge motion. Tumble flaps have the effect of increasing tumble
and in-cylinder charge motion at the expense of pumping losses. Kapus, et al [60]
documented that the intake port needed to be blocked by up to 70% in-order to
induce adequate amounts of tumble. Late Intake Valve Opening (LIVO) is another
example of another strategy that achieves a benefit in economy through charge
motion rather than pumping loss minimisation [67-69].
LIVO is also the subject of research at this moment in time. This involves
phasing a short low lift intake cam toward BDC rather than Top Dead Centre (TDC).
This however offers no benefit to PMEP in comparison to the throttled case.
35
2.4.4 Combustion Stability
One of the drawbacks of the Miller cycle, particularly of EIVC, is the reduced
charge motion in the cylinder leads to a relatively slow flame front speed and a higher
COV of IMEP. This problem is particularly pronounced at very low load conditions
where the quality of combustion is very degraded and HC emissions increase rapidly.
Almost all of the literature available on EIVC and LIVC report this issue.
Combustion degradation is particularly pronounced at low load and with very extreme
degrees of EIVC and LIVC. All of the available literature is unanimous in the fact it is
caused by low EOC pressure and temperature making the charge harder to ignite and
burn in its entirety.
EIVC has the added drawback of exhibiting very low levels of tumble/bulk flow
due to the fact the valve is closed very early on in the cycle giving the charge a very
long time in which to dissipate its motion. This greatly reduces the speed of the
combustion process [63][70] and greatly restricts the extent of EIVC that can be
employed for gasoline engines with typical geometric CRs. Sellnau, et al [72] reported
that the tumble motion can be reduced by up to 70% over baseline cases.
Zhao, et al [96] documents that tumble motion is very beneficial to combustion

as it encourages mixing and is converted to turbulent motion near TDC. The conversion
to turbulence at TDC is crucial as this is what increases the rate of entrainment and
combustion, and increases the combustion stability. Swirl is another form of bulk
motion but this is less useful to homogeneous charge GDI engines due to the fact it is
not as effective at creating turbulence as tumble, and the combustion stability is not
improved markedly as a result of this.
Many attempts have been made to increase the degree of bulk flow with EIVC.
The most successful method of achieving this is through the use of valve asymmetry.
Battistoni, et al [63] discovered that valve asymmetry can greatly increase the swirl
component of motion but not so much tumble, which can actually decrease with valve
asymmetry and reduce the amount of turbulence at TDC. Sellnau, et al [68] found that
36
tumble is dependent on valve lift and the tumble index shows a strong, almost linear,
correlation with valve lift. But this is highly dependent on the engine (particularly
piston) geometry (as was found by Akihisa, et al [71]), and whether or not a tumble
flap or any form of masking has been incorporated [64][68].
Lee, et al [73] discovered that swirl motion can improve combustion stability
and improve the Mass Fraction Burn (MFB) duration to an extent, but if the swirl
component is too elevated it can have a detrimental effect. However, swirl is not best
suited to this application (it is better suited to stratified applications as described by
Zhao, et al [96]). Battistoni, et al [63] discovered that the swirl component is greatly
reduced for symmetrical EIVC.
With tumble motion being very dependent on valve lift, EIVC has an inherent
disadvantage over LIVC due to the fact that the intake open duration is very short, and
this greatly reduces the peak valve lift. Urata, et al [88] stated that even fully variable
valve train systems at low speed and low load struggle to open the intake cam by a
large extent.
Low valve lift does have some advantages over high lift though. Although the
tumble motion is reduced the initial Turbulent Kinetic Energy (TKE) can be seen to be
very high with low valve lift due to the fact the flow velocity through the intake valve is
very elevated [63][66]. The elevated TKE is very short lived however and typically
reduces very rapidly once the intake valve has closed. Asymmetric valve timing and
valve deactivation increase the TKE even more still (by up to 250% compared to
symmetric timing according to Matsumoto, et al [66]), this is a likely result of increased
swirl motion.
Valve asymmetry has the effect of increasing tumble by a small amount too but
this tumble component is increased by a greater extent still when masking/tumble
flaps are employed. This therefore provides a cross tumble (or “Swumble”) type
motion within the cylinder.
The effectiveness of this elevated charge motion is highly questionable. It can

be seen to improve the MFB duration slightly and improve combustion stability but
37
ultimately with EIVC the majority of the charge motion is lost at TDC firing regardless
of what charge motion stimulus has been employed on the engine. The increase of
charge motion comes at the expense of increased PMEP losses in most cases. In some
cases the increase of charge motion and the corresponding increase of PMEP losses
required return no significant improvement over the throttled cases [67][94].
In the majority of LIVC cases (with both symmetric and asymmetric timing) the
bulk flow is improved at TDC owing to the fact the valve closes later in the cycle. The
flow behaviour and components of flow depend on the engine being tested.
Söderberg, et al [94] reported that the components of tumble and swirl with LIVC were
similar to the baseline valve timings, this means that LIVC typically exhibits greater
turbulence at TDC which gives it a much improved MFB duration. However, the TKE
obtained with symmetrical LIVC is typically much lower which means the mixing is
inferior to the EIVC case.
Valve asymmetry and deactivation have been employed to remedy this issue,
as well as exploring the effect of low lift cams. Matsumoto, et al [66] found valve
deactivation to be a very effective method of increasing the TKE whilst maintaining
good bulk motion. A similar strategy has been employed in another study by
Söderberg, et al [94] and it was found that this was an effective method for increasing
the TKE, but, like the EIVC case, this was at the expense of increased PMEP losses. The
same study also confirmed that the maximum lift symmetrical LIVC case was the
optimum for net efficiency for this reason, in spite of the fact that 10-90% MFB
duration was one of the highest of all the LIVC strategies tested at this condition.
LIVO is another strategy for boosting charge motion. Cleary, et al [67] has
concluded that it is an unsuitable replacement for throttled operation due to the fact
the PMEP is so high. It does however result in similar levels of TKE as seen for EIVC only
much later in the cycle. It also gives a very high effective CR close to the geometric CR
which results in elevated EOC temperatures. This coupled with the increased TKE gives
very fast burn durations which is useful for cold start applications.
38
2.4.5. Engine Breathing
Both EIVC and LIVC have the potential to change engine breathing dynamics
considerably. Both strategies require the MAP to be elevated above what it would be
for conventional Otto operation. This will drastically change the quantity of residuals in
the cylinder, and the temperature.
EIVC is relatively insensitive to engine breathing due to the fact no charge is

purged back into the intake manifold at any point during the cycle. With LIVC however
a great deal of charge will be purged into the intake manifold which can have a
dramatic effect on engine performance.
One of the biggest problems with EIVC operation is charge heating during de-
pressurisation, this has been reported by Hitomi, et al [74]. LIVC has been found to
achieve reduced EOC temperatures compared to EIVC due to the fact the charge is de-
pressurised with EIVC, this causes the charge temperature to drop dramatically. The
charge is then heated up by the cylinder walls thus elevating the overall charge
temperature.
Blakey, et al [86] carried out an investigation on LIVC and found that the
resulting PMEP reduction did not correspond to a BSFC benefit compared to the
throttled case, it was in fact worse. A very high degree of LIVC was employed in order
to run at part load completely unthrottled and the higher the degree of LIVC the worse
this discrepancy became. This problem was traced back to the 8 branch intake
manifold installed on the engine at the time. The fuel rejected through LIVC running
was returning to the manifold plenum and was being distributed to other cylinders.
Taylor, et al [64] observed an increase in the intake port temperature during

testing with LIVC. The temperature reached up to 600C due to residuals being purged
to the intake manifold. This required the combustion phasing to be retarded
significantly due to the excessive temperature of the incoming air, thus impacting
heavily on the perceived benefit of running with LIVC. A similar observation was made
by Anderson, et al [57].
39
Residuals are effective at reducing NOx and CO emissions but other than this
they are generally detrimental to EIVC and LIVC combustion stability, particularly at
very low loads.
The general consensus from the literature is that LIVC is harder to execute
successfully than EIVC as it relies heavily on the engine tuning, particularly the intake
system.
2.4.6 Emissions
Both EIVC and LIVC are inherently effective at reducing NOx emissions due to
the lower cycle temperatures. The literature states this benefit is typically in the region
of 30 – 60% reduction in comparison to the baseline Otto cases [57][61][64][67-68][89-
90][97]. In certain circumstances, particularly with LIVC, the reduction in NOx can be
attributed to increased residual gas concentration. It is very difficult to gauge from the
literature which technique is optimum for reducing NOx emissions, the closest is the
work of Moore, et al [65] who compared NOx emissions for EIVC and LIVC in the same
study but failed to compare them at the same speed and load point. Gould, et al [98]
found that for the same extent of EIVC and LIVC in a 2 valve engine, EIVC returned
higher NOx emissions (approximately 10% higher) than LIVC, but that both were
considerably reduced below the Otto cycle baseline case. It was also found that EGR
could reduce NOx emissions further still but at the expense of thermal efficiency, CO
emissions and HC emissions.
Internal residuals as well as external EGR have the effect of reducing NOx and
CO emissions, however, Cleary, et al [67] attempted to implement EGR in a Miller cycle
engine and it resulted in a reduction in efficiency compared to just Miller on its own.
This was attributed to a reduction in combustion stability.
Asymmetric valve timing and valve deactivation generally show no strong

trends with regard to NOx emissions [61-62][65]. Stansfield, et al [61-62] have run 2
separate studies on the impact of Miller cycle on NOx emissions. In one study [61] the
40
NOx emissions were greater with valve deactivation in comparison to symmetrical
valve actuation, although this has been attributed in part to poor engine running and
misfiring. The engine also demonstrated a sensitivity to which intake cam was
deactivated, with one giving slightly different NOx emissions to the other.
In a later study by Stansfield, et al [62] valve deactivation and symmetrical

valve timing were analysed once again, only in this study several different valve lifts
were tested as well. The impact of valve deactivation on NOx emissions was found to
be dependent on speed and load in this study, with the low speed and low load point
generally favouring valve deactivation and low valve lift. The advantage of valve
deactivation was lost with increasing speed however. For higher load points valve
deactivation with low lift valves was not feasible without boost, and only 1 valve
deactivation point could be tested and this yielded no benefit over symmetrical valve
timing.
In most cases EIVC and LIVC result in a slight increase in HC emissions due to
lower EOC temperatures reducing the rate of HC oxidisation [67][86], however Urata,
et al [88] found little to no difference in HC emissions and may even be marginally
improved. This reveals that the HC emissions are very dependent on the engine
breathing dynamics.
HC emissions can generally be seen to decrease with increasing speed with

both EIVC and LIVC in a similar manner to Otto cycle engines [62][86], and they can
also be seen to decline with load (Boggs, et al [89]). Whether or not the Miller cycle HC
profile intersects with the Otto HC emission profile depends on the engine. In the case
of Blakey, et al [86] LIVC produces fewer HC emissions than the Otto baseline at low
speed but produces more HC emissions at high speed than the baseline. Boggs, et al
[89] on the other hand show elevated HC emissions at low loads (up to 94% higher
than baseline at idle) but HC emissions near enough the same as the baseline at higher
loads.
Cleary, et al [67] investigated numerous valve profiles for EIVC and showed that
as the degree of EIVC increased (the cam duration reduced) the HC emissions
41
increased almost linearly to give a HC emission increase of 20% relative to the baseline
to the most extreme EIVC profile.
The impact of valve deactivation can be seen to have very little effect on HC
emissions. Moore, et al [65] saw a slight reduction (6%) in HC emissions at low speed
and high load LIVC with valve deactivation compared to symmetric LIVC, but no gain
was obtained with EIVC with valve deactivation compared to symmetric EIVC.
Stansfield, et al [62] recorded a similar result with valve deactivation and low lift, in
that study it made very little difference to overall HC emissions.
CO emissions with Miller cycle are generally reduced compared to baseline

Otto cases with the notable exception of Sellnau, et al [68] which reports an increase
of 123% compared to the baseline case. This however was a result of the engine being
tuned to reduce NOx emissions for an EPA cycle rather than a reflection of the Miller
cycle process as a whole. In the literature the CO emissions are typically reduced by 30
– 60% over baseline with asymmetric valve timing/valve deactivation [61-62][89].
CO is dependent upon how homogenous the charge is at the time of ignition.

Of the limited data available on CO emissions and the Miller cycle, symmetrical EIVC
was found to massively increase CO emissions compared to baseline for some speeds
and loads. Stansfield, et al [61-62] found that both EIVC and LIVC benefit CO emissions
compared to the baseline points with asymmetric valve timing though. CO emissions
are largely insensitive to valve lift with EIVC which is somewhat unexpected, this would
imply that the TKE generated through asymmetric valve timing/valve deactivation is
sufficient for mixing the charge.
In a study by Sellnau, et al [68] LIVO was found to improve the cold start
performance of an engine. In a later study by Turner, et al [69] a CFD study revealed
that LIVO greatly improved the charge motion in the cylinder. When employed for just
the first 20 cycles this system can achieve a reduction in HC emissions of 35% over the
baseline case and prevent the use of accidental enrichment.
42
2.4.7 Knock Mitigation and Deep Miller
There is a wealth of literature on genuine Miller cycle engines for heavy diesel
applications but there is unfortunately very little on Miller cycle gasoline engines. The
majority of the papers are on Atkinson cycle only, or employing EIVC and LIVC without
forced induction.
An early study of the knock mitigating effects of the Miller cycle was carried out
by Ke, et al [93]. This was an analytical study of using EIVC to mitigate knock and
dramatically increase maximum load. This is the first study (to the authors best
knowledge) to focus on increasing engine load to what would be considered Deep
Miller levels of IMEP and predicted a maximum load with EIVC of almost 22 bar IMEP.
This study predicted that the EOC temperature reducing effects of EIVC could
yield a thermal efficiency benefit of approximately 7%, furthermore, it also predicted a
torque increase of 7.8%. This torque benefit was maximum at approximately 3000rpm
and the thermal efficiency benefit was predicted to be maximum for approximately
4250rpm and above where the need for enrichment was circumvented.
Another noteworthy study of knock mitigation was carried out by Sakata, et al

[99]. This study focussed on the effects of increased geometric CR on both knock and
thermal efficiency. This study demonstrates the impact of high geometric CR and LIVC
on spark timing. It was discovered that the combustion had to be retarded significantly
for the high geometric CR cases and the maximum knock limited load was reduced
significantly as a consequence of this. However, even with very retarded combustion
phasing the highest geometric CR case tested was still the optimum for ISFC. The peak
load was however obtained at a much lower geometric CR.
The most recent study to encroach on what would be considered Deep Miller is
that of Taylor, et al [64] who tested at a peak load of 15 bar BMEP with the Mahle cam
in cam system. The increased load on the intercooler, runner temps, 50% MFB point
and various other attributes were monitored to see what impact high load LIVC had on
them.
43
The intake manifold temperature was observed to increase significantly due to
the back flow of residual gasses and charge that had been purged from the cylinder as
part of the LIVC process. This elevated the charge temperature markedly. In spite of
this the 50% MFB point could still be advanced by 4 Crank Angle Degrees (CAD). With
the intake manifold temperature elevating effects of LIVC cancelled out, the 50% MFB
point could be advanced by up to 12 CAD. This yielded a thermal efficiency benefit of
up to 3.2%.
Turner, et al [50] carried out an analytical investigation into the use of EIVC at
high loads of up to 35 bar BMEP on the Jaguar Land Rover Ultraboost engine. EIVC and
low peak valve lift was found to offer the greatest improvement at low engine speeds
(approximately 6%). The BSFC improvement was lost very rapidly with increasing
speed however, with there being negligible difference in BSFC between the shorter
214 CAD duration cam and the standard 250 CAD duration cam above a speed of
approximately 2500rpm.
This area is of great research interest at this moment in time due to the need
for ever higher engine loads to be achieved at lower engine speeds. Current projects
such as the Jaguar Land Rover Ultraboost project and the likes of Ricardo have
expressed a great deal of interest in Deep Miller.
2.5 Summary
This chapter contains all of the past work and results from all of the major
publications in the history of the Miller/Atkinson cycle. It has also explained the future
direction of Miller cycle research such as exploring Deep Miller operation. It also
reflects how the Miller cycle has great potential to be employed on future gasoline
ICEs as both a short term and long term technology for reducing emissions and
improving fuel economy. The current and future emission limits that all ICEs must
adhere to has also been explained and outlined.
44
This chapter explains all of the relevant literature for the Miller/Atkinson cycle
work carried out in this thesis. It also helps to explain what has been discovered in this
thesis and whether or not a similar discovery has been obtained before and whether
or not the result obtained contradicts any of the results obtained in previous test work.
It has also demonstrated where the gaps in knowledge with regard to the Miller cycle
exist and how best to research this subject further.
45
3
Chapter 3
Experimental Test Facility
3.1 Introduction
This chapter describes the experimental test facility used to obtain all the
experimental data contained in this thesis. It also describes the data acquisition system
used and how the raw in-cylinder pressure data was analysed to retrieve key
experimental data.
The test facility used for this test work was designed from new by the author.
This includes installation of a new dynamometer, AC drive, dynamometer control unit,
boost rig, exhaust system, extractor fan, gas sensor, fuel supply, data acquisition
system as well as 3-phase and single phase power.
The engine used is a Mahle downsized 3 cylinder head mated to a single

cylinder Ricardo Hydra bottom end. A cylinder block therefore had to be designed, an
additional plate also had to be designed to accommodate the cam drive system.
3.2 Experimental Setup
46
3.2.1 Engine Description
The crankcase used for all test work was from a Ricardo Hydra single cylinder
engine. It is a single cylinder crankcase with a speed range of 600 – 5400rpm and
capable of withstanding peak cylinder pressures of up to 140 bar. The single cylinder
engine is unique in that the cylinder heads are interchangeable. The mode of operation
is therefore dictated by the head and it can potentially be run as a 2-stroke or 4-stroke
and with either gasoline SI or diesel CI.
The cylinder head used for all test work is the standard 3 cylinder head as
found on the Mahle 1.2l 3 cylinder downsized engine (Figure 3.1). A cylinder block has
been designed to mate the cylinder head to the crankcase. This cylinder block has been
designed to provide the cylinder head with the required coolant and oil supplies with
as little modification to the cylinder head as possible. The only modification that had to
be done to the cylinder head was the drilling of a hole which serves to act as an
emergency oil drain should the head overflow with oil.
Figure 3.1 The Mahle 1.2l 3 Cylinder Advanced Downsizing Engine (Hancock, et al [48])
47
The Mahle 1.2l 3 cylinder downsized engine is an advanced downsizing
technology demonstrator engine designed by Mahle (Hancock, et al [48]). It has been
designed to replace a 2.4l V6 PFI engine that one would typically find in a class C or D
European vehicle with a weight of around 1600kg. It has also been designed to achieve
a fuel consumption reduction of around 30% over the V6 over the NEDC cycle, whilst
simultaneously complying with Euro 5 and ULEV2 emissions standards. The Mahle 1.2l
3 cylinder downsized engine makes use of central GDI with a piezoelectric fuel injector
and injection pressures of up to 200 bar.
The 3 cylinder engine exists in 3 different guises, a single turbo version capable
of a peak power and peak torque of 90kW and 210Nm respectively, and a twin turbo
version capable of a peak power and peak torque of 144kW and 286Nm respectively.
This corresponds to peak BMEP values of 22 bar and 30 bar for the single turbo engine
and the twin turbo engine respectively. Both of these engines have a geometric CR of
9.75. A third single turbo variant has also been made, this has a maximum power
output of 120kW and a geometric CR of 9.3.
The third variant of this engine (or the second generation engine) has been
fitted to a Volkswagen Passat (Figure 3.2). This “mule” test vehicle has achieved an
NEDC fuel consumption figure of 5.8l/100km and an NEDC CO2 output figure of 135
g/km. The engine also achieved a 80-120 km/h fifth gear acceleration time of 8.9
seconds (Fraser, et al [100]).
Figure 3.2 The Mahle Downsized Demonstrator Vehicle

48
3.2.1.1 Single Cylinder Engine Description
The single cylinder engine (Figure 3.3) has a bore of 83mm and a stroke of
73.9mm, the same as that for the Mahle 3 cylinder engine, and a displacement of
400cc. The single cylinder engine uses the same connecting rod and similar piston as
on the 3 cylinder engine, although the piston on the single cylinder engine gives a
geometric CR of 9.25. It is possible to alter the geometric CR slightly with the same
piston by adding or subtracting shims that are placed between the crank case and the
cylinder block. Without changing the piston, this is the only variation in geometric CR
that can be obtained.
Figure 3.3 The Brunel Single Cylinder Downsized Engine
49
The engine has a 70lb flywheel. It is this component that imposes the 140 bar
average peak cylinder pressure constraint. It is also this component that limits the
maximum speed of the engine to 5400rpm.
In total 3 different intake cam profiles (152 CAD short profile, 240 CAD standard
profile and 292 CAD long profile) and 1 exhaust cam profile (276 CAD standard profile)
were used for all test work (profiles given in appendix A.1). Both the intake and
exhaust valves can be phased within ± 20 CAD of a set point.
The single cylinder engine consists of the 3 cylinder head from the standard
Mahle 1.2l 3 cylinder downsized engine mated to a single cylinder Ricardo Hydra
crankcase by means of a bespoke cylinder block (see section 3.2.1.2).
The middle cylinder of the cylinder head is used as the firing cylinder for this
engine, the other 2 cylinders have been blanked off. The valve holes for cylinders 1 and
3 have also been blanked off, as well as the oil gallery for the hydraulic lash adjusters,
to ensure that the cavity between the top of the cylinder and the cylinder block does
not fill with oil.
The middle cylinder was used because of geometrical constraints imposed by

the flywheel. In order to maintain the same geometric CR and stroke of the 3 cylinder
engine whilst using the same components as the 3 cylinder engine, the cylinder head
had to be mounted just 54mm above the top of the crankcase. The flywheel however
projects over the top of the crankcase by 41mm leaving very little clearance between
the head and the flywheel, whereas over the front of the engine there is no such
constraint. This meant that the cam drive sprockets were projected someway forward
of the front of the crankcase cam drive system. In order to compensate for this offset a
plate had to be designed to house the cam drive system. This plate was mounted on
the front of the crankcase and also housed the cam chain tensioner.
50
3.2.1.2 Cylinder Block Design
The cylinder block had to be designed specifically for this application owing to
very tight geometric constraints and the need for a very robust design which can
withstand very severe autoignition or “superknock” cylinder pressures of up to 250
bar. The block has a total height of 54mm at its thickest point and incorporates
galleries for head coolant, block coolant, oil supplies, oil draining and fixings (Figure
3.4).
Figure 3.4 Cylinder Block Plumbing
The head is designed for cross flow type coolant flow. This has been maintained
for the most part in the cylinder block design however the flow will be more fan like
than straight across owing to the fact that there is just 1 orifice on cylinder 2 for head
coolant feed and holes for all cylinders on the head coolant return side. This has been
done to prevent flow stagnation in the proximity of cylinder 2 while ensuring the
majority of the flow goes around cylinder 2.
Figure 3.4 also shows the positions of the cylinder head mounting bolts. There
are 8 in total which are located the cylinder side of the 2 main head coolant galleries
and running parallel to them. These bolt holes have been designed to provide the
51
required gasket clamping force of 100kN. A simplified version of this block has been
analysed in ANSYS V11.0 [101] in an effort to ascertain where the greatest stresses
occur and what material is best suited to its construction (Figure 3.5). Figure 3.5 shows
the factor of safety when grey cast iron is used, and also shows that the parts of the
structure exposed to the greatest stresses are the fixing holes. In this scenario the
entire length of the holes are threaded, however, it was later found that employing a
small counter thread of 10mm and a chamfer of 0.5mm reduced the stress
concentrations to such an extent that aluminium could be used in the blocks
construction instead.
Figure 3.5 Ansys Plot of Factor of Safety
Aluminium is the preferable material to construct the cylinder block out of

because of its corrosion inhibiting properties, ease of machining and the fact it has the
same coefficient of thermal expansion as the cylinder head (which is also aluminium),
which will reduce warping effects and damage to the head gasket.
The block coolant jacket geometry has been designed to accommodate an

aluminium liner with Nikasil coating as developed for a previous single cylinder Mahle
project and found to be robust.
There is room for minor adjustment of geometric CR by the fitment of shims

between the cylinder block and the crankcase. The block also features many dowels
52
which are used to correctly locate the other components such as cylinder head,
crankcase, head gasket and the cam drive plate.
3.2.2. Oil System
The hydra crankcase features a wet sump oil lubrication system, standard
Mobil1 0W-40 oil was used for all test work. The crankcase also contains 2 1kW electric
oil heaters in the sump to preheat the oil before running. The oil is circulated
externally to the engine by means of a single speed 3-phase electric oil pump located
at the back of the test cell. This pump imposes no parasitic loss on the engine. The oil is
filtered by means of an AC Delco X19 equivalent oil filter mounted on the test bed next
to an oil heat exchanger. The maximum heat rejection rate of this heat exchanger is
4kW. The rate of heat rejection is controlled by a Spirax Sarco capillary actuator which
is used to control the flow rate of raw cooling water through the heat exchanger.
Oil temperature and pressure are measured at multiple points in the oil system.
The pressure is monitored by means of 3 different sensors, 1 is an analogue gauge,
another sensor is a Druck 4.20mA feedback pressure sensor which is connected to a
low speed data acquisition system, the third sensor is a pressure switch sensor for the
dynamometer emergency stop system. Temperature is monitored in 3 different
positions. It is measured by 2 Platinum Resistance Thermometers (PRTs) located about
500mm upstream of the crankcase oil gallery, one for the low speed data acquisition
system, and the other for the dynamometer emergency stop system. Another PRT is
located around 200mm downstream of the oil sump, this is connected to the low
speed data acquisition system. The temperature is monitored both before and after
the engine so an idea of the rate of heat addition to the oil across the engine can be
obtained.
The oil system has a nominal flow rate of 9.1 l/min which is fed to a single oil
gallery in the crankcase. This oil gallery distributes oil through 3 different hoses, 1 for
lubricating the cams, 1 for the hydraulic lash adjusters and the final one for the cam
53
drive system (chain tensioner, chain lubricant and cam phasers), there is also a fourth
outlet for the oil gallery which is for the piston cooling jet located inside the crankcase.
Depending on which cam was employed, there was also a fifth oil feed for the high
pressure fuel pump. The oil heaters have no temperature ceiling and are controlled by
means of a manually operated instrument panel located on the test bench.
The oil pressure varies greatly depending on the oil temperature, typical
pressures are 6.5 bar abs when the oil is at room temperature and 4.25 bar abs when
the oil is at 900C. The oil pressure cannot be regulated. A maximum oil temperature of
900C was implemented in order to stop the oil pressure dropping below a minimum of
4.25 bar abs.
A schematic diagram of the oil system is shown below (Figure 3.6):
HW
5 5 5
7
3
1
63A
3ph
2
Figure 3.6 Oil System Schematic Diagram
Where the components are numbered as follows:
1. Oil pump 3-phase 9.1l/min

54
2. Oil heat exchanger (4kW heat rejection)
3. Oil Filter (AC Delco X19 or equivalent)
4. 2 x Inlet PRTs
5. 3 x Pressure sensors (Analogue gauge, pressure switch and Druck PTX 1400 0-
10 bar)
6. Oil heaters (2 x 1kW immersion in crankcase)
7. Outlet PRT
8. Power/control cabinet
The cabinet in Figure 3.6 with the red lines leading from it is a 3-phase power
supply. Figure 3.6 is slightly inaccurate in that it shows the oil being drained from the
cylinder head directly, it is actually drained from the head via the block. It also does
not show the oil supply to the high pressure fuel pump.
3.2.3 Coolant System
The coolant system is also external to the engine (3-phase single speed electric
pump located at the back of the test cell), and imposes no parasitic losses on the
engine. The coolant used for all test work was a mix of 50% de-ionised water and 50%
ethylene glycol. The coolant system also features a 3kW immersion heater to preheat
the coolant before engine running. The rate of coolant heat rejection is controlled by a
capillary actuator which regulates the raw water flow through a coolant heat
exchanger. The maximum rate of heat rejection of this heat exchanger is
approximately 53kW.
The cooling system for this engine is quite unique in that it has split cylinder
head and cylinder block cooling. The test bed has been set up so the flow rate of
coolant through the cylinder head and block can be varied independently of each
other. There is also a coolant bypass which allows the flow rate of coolant for the
entire engine to be varied. To an extent the pressure can also be varied, but not
independently of flow.
55
To ensure that the cylinder head is full of coolant, and that no part of the
cooling jacket is not sufficiently submerged in coolant, the coolant tank has been
located 121mm (H) higher than the highest point of the coolant jacket (Figure 3.7).
Figure 3.7 Coolant Jacket Height Relative to Coolant Tank Height
The flow rate through each section of the coolant system is controlled by
means of 3 ball valves, one for the head coolant, one for the block coolant, and
another for the bypass circuit, as shown in Figure 3.8. The 2 coolant flow meters have
been located upstream of the ball valves and in straight sections of pipe to ensure the
turbine type flow meters give as accurate a reading as possible.
The coolant pressure is monitored by means of a Druck 4.20mA pressure

sensor which is located approximately 200mm upstream of the inlet side of the
cylinder block at a tee where the flow splits. The feed for both the coolant and the
block is common up until this tee. The head and block coolant outlets are isolated from
each other from this tee up until the coolant tank where they are recombined. The
main purpose of this pressure sensor is to allow a means of monitoring coolant pump
health, but also to ensure constant pressure when varying the coolant flow rates. The
coolant pressure is typically between 1.2 and 1.5 bar abs depending on the
temperature and running condition.
56
The coolant temperature is monitored in 5 locations by 4 PRTs and 1
thermocouple. One PRT is located approximately 200mm upstream of the inlet of the
cylinder block in the same tee as the pressure sensor. This is to monitor the coolant
inlet temperature. Another 2 PRTs are located approximately 500mm downstream of
the outlet side of the cylinder block and head, this is to monitor the coolant outlet
temperatures. The final PRT is located in the coolant tank is used for the dynamometer
emergency stop system. A K-type thermocouple is located in the coolant heater
reservoir tank to ensure that the coolant heater does not exceed its maximum
temperature and that no boiling occurs.
The coolant pump is a 3-phase Beresford pump and has a nominal flow rate of
32 l/min with no ability to change this flow rate. This flow rate was reduced somewhat
with the pump in situ on the test bed to a maximum of 11 l/min. The coolant heater
has no temperature ceiling although it is fitted with a thermostat that will disable the
heating element once the temperature exceeds 800C. The heaters are controlled by
means of a manually operated instrument panel located on the test bench.
The relatively low maximum coolant flow rate imposed a significant constraint
on the peak speed and load the engine could be run at. Although the maximum
coolant temperature could be regulated with relative ease due to the high heat
rejecting capacity of the heat exchanger, the temperature difference across the engine
(between the coolant inlet and outlet) is dictated purely by coolant flow rate and no
other component. A typical ∆T between the inlet and outlet for a production engine is
2 - 100C, with poorly designed engines peaking at about 150C. The engine was run at a
maximum ∆T of 140C for baseline work, but a severe engine failure followed soon after
which resulted in a blown head gasket and softening of the cylinder head. After this a
∆T maximum limit of 60C was imposed.
An additional 12V DC automotive style coolant pump was later employed in an

attempt to boost the maximum coolant flow rate. Due to the high resistance nature of
the coolant system plumbing, the effectiveness of this pump was compromised and
delivered no more than the Beresford pump. For some test work both pumps were
57
employed and between them managed to deliver a maximum coolant flow rate of 13.5
l/min.
12
11
230V 0
0
0
13A
0 0
8
9 10 10 8
10
6
4 5
HW 0
0
0
230V
0 0
13A
3 12
Figure 3.8 Coolant System Schematic Diagram
1. Coolant pump 3-phase 32l/min

2. Coolant heat exchanger (heat rejecting capacity in the region of 53kW)
3. Coolant heater with thermocouple (3kW Redring immersion heater in a
reservoir)
4. Pressure sensor (Druck PTX 1400 0-4 bar)
5. Inlet PRT
6. Tee (0.5”)
7. Power/control cabinet
8. 2 x Outlet PRTs
9. 2 x Outlet flow meters (Apollo RN3/15 flow meter and Apollo RN3/20 flow
meter)
10. 3 x Ball valves
58
11. Coolant tank
12. 2 x Flow meter pulse counters (Apollo WFC/5 frequency to current converter)
The cabinet in Figure 3.8 with the red lines leading from it is a 3-phase power
supply. Figure 3.8 is slightly inaccurate in that it shows separate coolant flows from the
head and the block, in actual fact the flow from the head is output via a completely
different gallery within the block.
3.2.4 Fuel System
The fuel used for all test work is standard pump Unleaded Gasoline (ULG) with
a Research Octane Number (RON) of 95. The gasoline is circulated around the Brunel
University engine laboratory by means of a pump. All of the fuel lines around the
laboratory are of stainless steel construction to avoid the contamination effects
associated with the use of copper pipes as documented by Pereira, et al [102].
A schematic diagram of the fuel system is shown below (Figure 3.9). The
temperature of the fuel was controlled by means of a heat exchanger. Manual control
of the fuel temperature was employed throughout all test work. The temperature of
the fuel is measured in 2 locations, the first point being a PRT sensor located
immediately downstream of the low pressure fuel pumps. The second point where the
fuel temperature was measured was in the coriolis flow meter employed to measure
the fuel flow rate, although this temperature was not recorded for any test work.
The fuel pressure is monitored in 2 locations, the first point being immediately
downstream of the low pressure pumps by means of a 4.20mA Druck pressure sensor.
The second pressure sensor is an automotive type pressure sensor located in the
common rail. This pressure sensor is connected to the Engine Management System
(EMS) only, it was recorded for all test work.
The fuel flow rate was measured with an ABB FCM2000 coriolis flow meter with
a DN1.5 size sensor. This flow meter can also output the fuel density and temperature
although these values were not recorded for any part of the test work.
59
A port for the purging of fuel vapours is located in the low pressure circulation
part of the fuel system to avoid potential vapour lock issues. This has been installed in
case of any problems, but the fuel system is actually void less so this is not anticipated
to be an issue and is for redundancy only.
The engine has a PFI system as well as a centrally mounted DI system, although
only the DI system was used for this test work.
The fuel is fed from the laboratory fuel supply circuit to the low pressure
pumps under the force of gravity. These low pressure pumps are powered by an
external 12V supply and impose no parasitic loss on the engine. The fuel pressure is
regulated by means of a fuel pressure regulator located approximately 300mm
downstream of the pumps. The regulator regulates the pressure to approximately 4.8
bar abs. The fuel that has been purged from the regulator forms a local low pressure
fuel circulation system. This low pressure fuel circuit incorporates the heat exchanger
and a filter.
The fuel that is not circulated is fed to a high pressure pump via the coriolis
flow meter. The high pressure pump is powered by the engine intake camshaft. Its
power requirement is unknown but it is expected to be quite large due to the fact it
has been designed for much larger multi cylinder engines. This is expected to be one of
the largest parasitic losses imposed on the engine. The high pressure pump is a lossless
type pump and requires no bypass back to the low pressure system.
The common rail used for all test work is the original common rail used on the
3 cylinder engine with 2 of the ports blanked off. The injector is a piezoelectric type
direct injection injector with a maximum design pressure of 200 bar. It is an outward
opening type.
60
RON 95 ULG
(PUMP)
2 12V DC
70A
3 AFT/Injector Driver
5
3
12
7
9
5
10 11
8
6
Figure 3.9 Fuel System Schematic Diagram
1. RON 95 ULG fuel supply (regular pump fuel)

2. Solenoid ball valve
3. 2 x Fuel filters (Delphi)
4. Fuel heat exchanger
5. Low pressure fuel pumps
6. Pressure regulator
61
7. PRT
8. Low pressure sensor (Druck PTX1400 0-10 bar)

9. Fuel flow meter (ABB FCM2000)
10. High pressure fuel pump
11. Common rail
12. Rail pressure sensor (automotive type)
3.2.5 Intake System
3.2.5.1 Pipes
The intake system is constructed from a mixture of materials including rubber

hose (from the compressor to the pressure regulator) which has been rated at
pressures of up to 8 bar abs (the automotive style sensors in the system restrict this to
4 bar abs however). This also allows flexibility in terms of the compressor and pressure
regulator placement, and has been employed with the intention of reducing the
number of pressure reducing elbows in the intake system. The pipe work between the
pressure regulator and the plenum is mostly stainless steel construction to support the
weight of the charge air heater.
The pipes between the compressor and the plenum are all 1” diameter and
have been sized to eliminate the possibility of choking at any point in the system. The
pipes downstream of the plenum are all 1.5” and they are a mixture of stainless steel,
aluminium and silicon tubing construction. Owing to the vacuous environment
downstream of the throttle the use of silicon tubing has been minimised to make
“pinching” less likely. The silicon tubing has been designed to be the weak point of the
system and is intended to simply blow apart if the pressure in the plenum becomes too
high as the result of a pressure regulator fault.
The intake system post throttle has been designed to have as few interruptions
in the general airflow as possible in an effort to make the airflow through the intake
system more predictable and smooth.
62
3.2.5.2 System Details
Owing to the fact a single cylinder engine was used for all test work, a
turbocharger could not be used. Compressed air was provided by an industrial
compressor instead. A schematic diagram of this system (also known as the boost rig)
is shown in Figure 3.10. The compressor used was a Compair HV22RS AERD hydrovane
type compressor. It consists of a 22kW motor with a hydrovane compressor, an
intercooler, a refrigerant drier unit, a 5 micron oil filter and a 250l receiver.
The refrigerant drier was employed for all test work to provide consistent air as
humidity has been found to have a minor effect on the properties of the air (Taylor
[103]) and will have an impact on the propensity for an engine to start knocking. The
refrigerant drier dries the air to <3% humidity, no hygrometer was employed to check
the charge air humidity at any point during the test work. Dry air has the highest ratio
of specific heats and is therefore detrimental to turbocharged engine efficiency.
Therefore the data obtained from all test work will be slightly pessimistic in
comparison to data obtained from the 3 cylinder engine running in a typical northern
European climate.
This compressor delivers a maximum flow rate of 3.88m3/min effective at 1 bar

abs (4.61kg/min) and delivers this at a minimum pressure of 7 bar abs. The pressure is
regulated down to more suitable pressures by means of a Parker Hannifin EPDN4
closed loop control pressure regulator located downstream of the compressor in the
test cell. This pressure regulator is controlled by the EMS directly. The EMS pressure
sensor used to regulate pressure is a standard Bosch automotive
pressure/temperature sensor located downstream of the plenum and upstream of the
throttle.
The degree of pressure control achievable through this pressure regulator is

not very precise due to the fact it was not designed for precise pressure regulation but
for regulating large amounts of flow from very high pressures. Pressure control is
typically ± 0.15 bar of the requested pressure, therefore where possible the throttle
63
was used to achieve more refined control of the pressure so that the variability of
boost pressure is less than ± 0.08 bar.
A 3kW Secomak 632 charge air heater is located downstream of the pressure
regulator to elevate the charge air temperature to a more representative post
compressor temperature. A heater is required to heat of the charge air because the
compressor intercooler has a very high rate of heat rejection such that the charge air
temperature is typically just 80C above the ambient temperature. The Secomak 632 is
not designed as a charge air heater, it therefore had to be modified somewhat by
sealing it up with a non-silicone based sealant (because silicone has the possibility of
contaminating the air supply). The heater is controlled by means of closed loop control
based on the feedback from a K-type thermocouple located in the “plenum” pressure
vessel. The precision of control is ± 10C.
The next component in the system is the plenum which is a 40l stainless steel
cylindrical pressure vessel. It has been sized on what experience of previous single
cylinder engines has deemed the optimum size for a plenum for an engine of this size.
The pipe downstream of the plenum has also been sized depending on what previous
experience has deemed most appropriate. Three cylinder engines benefit from the fact
that the cylinders are 240 CAD out of phase which is almost exactly the duration of
typical intake and exhaust valve events, this means there are very few cylinder to
cylinder interaction effects and the flow of air through the engine is fairly constant.
This breathing effect is very difficult to duplicate with other engine configurations,
particularly on a single cylinder engine. A very large diameter intake manifold was
used, as well as a very large plenum volume to try and reduce pressure and rarefaction
waves as much as possible.
Located downstream of the plenum is a tee. This tee connects the plenum to
the throttle primarily, with the third junction being a ball valve which allows the engine
to be run in NA mode. This ball valve can also be left slightly open to enhance boost
pressure control at low flow conditions by bleeding some air off thus artificially
increasing the flow rate through the compressor and pressure regulator.
64
The throttle is placed post plenum which is unconventional for most
automotive engines, however, given the size of the plenum and the relatively unstable
control of the pressure regulator the throttle was located post plenum in an effort to
improve control. One and a half inch pipe was also employed post throttle in an effort
to create as large a volume as possible. The throttle is a standard automotive
electrically actuated throttle (Bosch DV-E5 40mm) with integrated position sensor. The
throttle is controlled by the EMS, although the throttle position is dictated by the
dynamometer, allowing closed loop control of load if required.
Automotive type pressure and temperature sensors are located both upstream
and downstream of the throttle to give the EMS pre and post throttle temperatures
and pressures. The post throttle sensor was used to control boost pressure and
temperature.
An automotive type Mass Air Flow (MAF) sensor was employed slightly
downstream of the post throttle pressure and temperature sensor. This is a 1-way hot
wire type sensor and is therefore of limited effectiveness as the air flow in this region
of the intake is pulsing rather than steady. The sensor feedback was treated as an
indication only. The sensor was located in straight unobstructed section of
approximately 4 pipe diameters upstream and downstream of the pipe to ensure the
flow was as uniform, and the flow measurement was as accurate, as possible. Location
of this sensor in a position of steady flow was not possible because the only section
long and straight enough would have been the feed to the pressure regulator which is
never below 7 bar abs pressure, which is too great for the sensor.
Approximately 80mm upstream of the intake port is a Kistler 4005BA10FA0

piezoresistive pressure transducer, details of which can be found in the data
acquisition section of this chapter (Section 3.3). The final component in the intake
system is a PRT temperature sensor which is located directly in the intake port to give
the charge temperature going into the engine, or to give an indication of engine
breathing.
65
1
2 3 5
63A 272l
3ph
6
230V
13A
8
14
13
4
12V DC 10
12
125A 40l
11
7
9
ASAP3
AFT
230V
13A
Figure 3.10 Intake System Schematic Diagram
1. Compressor (Compair HV22RS)

2. Filter
3. Air refrigerator dryer (cools to 30C)
4. Drain
5. Receiver 272l
6. Pressure regulator (Parker electronic pressure regulator EPDN4)
7. Intake heater (Secomak 632)
8. K-type thermocouple (RS 397-1539)
9. Analogue pressure gauge
10. Accumulator (40l)
11. Throttle (Bosch DV-E5 40mm)
12. MAF sensor
13. Pressure transducer (Kistler 4005BA10FA0)
14. Temperature sensor (PRT RS 611-8264)
66
3.2.6 Exhaust System
3.2.6.1 Design
The exhaust system is a mixture of stainless steel and austenitic stainless steel
construction and has been designed to withstand exhaust gas temperatures of up to
9800C at pressures of up to 4 bar abs average.
Like the intake the exhaust has been designed to have no sudden/step changes
in pipe diameter purely in the interest of keeping the exhaust flow as predictable as
possible. Also like the intake system, the single cylinder nature of the engine will result
in drastically different breathing dynamics compared to the 3 cylinder engine.
Although the breathing dynamics can be changed by installing a surge tank, the
breathing is still not going to resemble that of the 3 cylinder engine so a surge tank has
not been installed.
Where possible the exhaust components have been shrouded in insulation to

minimise the possibility of injury. The exhaust system has also been mounted as high
as possible in the cell, with as minimal an area possible at operator height.
3.2.6.2 System Details
A schematic diagram of the exhaust system is shown in Figure 3.11. The first
component in the exhaust system is a K-type thermocouple that is located in the
exhaust port to give the exhaust port temperature. A more precise PRT type
temperature sensor would have been used for this application but PRTs cannot
withstand temperatures of up to 9800C and could therefore not be used.
The cross sectional profile of the exhaust runner changes at this point from the
straight oval shaped exhaust port geometry to a 1” circle. The transition between
these shapes is achieved by means of a 50mm long variable geometry section. This has
67
been done to ensure no abnormal flow properties such as any considerable turbulence
occur in or close to the exhaust port.
A Kistler 4005BA10FA0 piezoresistive pressure transducer (details of which can
be found in section 3.3) with water cooling module is located approximately 100mm
downstream of the exhaust port. Approximately 50mm downstream of this is a
condensate drain where any condensation that has formed in the exhaust can be
drained away. The drain has been sited at the lowest point in the exhaust system. This
drain is emptied every day.
Approximately 350mm downstream of this is a low speed Druck PTX1400
pressure sensor (details have been given in section 3.3) to measure the average EBP.
The main reason for this is to set the EBP valve correctly. Although the main purpose of
this pressure sensor is to control the EBP valve, it is in fact connected to the data
acquisition system. Approximately 50mm downstream of this is another K-type
thermocouple which has been installed in this location to ensure that the lambda
sensor is not being subjected to temperatures out of its design range.
Two hundred millimetres downstream of this is an automotive type Universal
Exhaust Gas Oxygen (UEGO) lambda sensor. This is connected directly to the EMS
where the lambda is controlled by means of closed loop control as detailed in section
3.2.7. One hundred millimetres downstream of this is a boss for the exhaust gas
analysers which have been described in section 3.3.
The next component in the system is an EBP regulating valve. This valve is a
butterfly type valve with no plastic components/washers. It seals by means of a gasket
and clamps instead of a washer. Owing to design constraints this valve is never in the
fully open position, however, the pressure increase caused by this is negligible as the
butterfly is so wide open that the restrictive area imposed by the butterfly itself is
encompassed entirely in the butterfly shaft area.
This valve is actuated by means of a servo motor connected to the valve by
means of a rod. The servo motor is operated remotely through the dynamometer test
bed. The pressure control is open loop requiring the user to input a throttle open angle
(in %) to give the required EBP.
The exhaust system then exits the cell and exhausts to atmosphere by way of
an automotive type muffler and silencer.
68
9
AFT
5
7 6 2 1
Chemilumine
FID HC 4
scence Nox
Analyser
Analyser
Figure 3.11 Exhaust System Schematic Diagram

2. Pressure transducer (Kistler 4005BA10FA0)
3. Condensate Drain
5. UEGO AFT Lambda sensor
6. Rotork Analysis Model 443 Chemiluminescence NOx Analyser
7. Rotork Analysis Model 523 FID HC Analyser
8. CPEngineering EBP regulating butterfly valve
9. Silencer (outside building)
3.2.6.3 Conversion of Emissions to Specific Emissions
All emissions were recorded as volume concentrations in parts per million

(ppm) initially. These were later converted to indicated specific emissions. This was
achieved by firstly converting volume concentrations of the various emissions to mass
concentrations, which can then be multiplied by the mass of exhaust gas required to
generate 1 kWhr of energy.
69
Converting volume concentration to mass concentration requires calculation of
the ratio of molecular weight of the emission being investigated to the molecular
weight of air. Multiplying this by the volume concentration of the emission will give a
mass concentration. The molecular weight of NO is 30g/mole. The molecular weight of
HCs is the same as that of the span gas used to calibrate the FID analyzer (propane gas)
which is 44g/mole. The molecular weight of air is 29 g/mole. The ratio of molecular
weights must also be divided by the AFR due to the fact there are many more moles of
air per mole of fuel.
This value can be converted from ppm to percentage by firstly dividing by
1,000,000 to obtain a parts per kilogram fraction and then multiplying this by 100 to
get a percentage figure. This percentage figure is then multiplied by the amount of fuel
required to generate 1 kWhr of power, or the specific fuel consumption. For this work
indicated values were used and calculated.
The equation for this conversion (Eq.3.1) is given as,
3.1
where ISem is the indicated specific emissions of emission em, VCem is the volumetric
concentration of emission em in ppm, and Mem and Mair is the molecular weight of the
emission em and air respectively.
3.2.7 Dynamometer and EMS
The dynamometer used for all test work is a CP Engineering 48kW AC motor
with a 4 quadrant AC regenerative inverter drive and an operating envelope as shown
in Figure 3.12. The dynamometer is capable of a maximum speed of 6000rpm. The
dynamometer is capable of motoring and absorbing 48kW and has a maximum torque
rating of 140Nm. The dynamometer is also capable of transient operation although
this capability was not used for this test work.
70
Figure 3.12 Dynamometer Torque Curve and Engine Torque Curve Comparison
Although the CADET V14 light software has limited capability for logging data, it
was not used for logging any data for this test work. A dedicated logging system was
employed for this purpose instead (see section 3.3). Just one output from the
dynamometer was logged and that was the torque reading which was output from the
CADET V14 cabinet into the instrumentation system.
A schematic of the dynamometer system is given in Figure 3.13.
71
Figure 3.13 Dynamometer Schematic Diagram
The GDI injector is driven by a Vemac injector driver coupled to the Mahle
flexible Electronic Control Unit (ECU) based on the AFT PROtroniC platform. Where
possible the automotive sensors required for the ECU have been installed. The ECU has
2 looms leading from it, Figure 3.14 shows which connectors are on which loom, any
connectors that have been left disconnected for this work are highlighted in red.
72
Loom 1 Loom 2
Manifold pressure/temp Connectors:
Fuel rail pressure (DI Only) Ex cam sensor
Ex cam VCT Supecharger out pressure
Hot EGR valve Exhaust pressure
LP fuel pressure (PFI Only) HP fuel pump (DI Only)
Throttle Airflow meter
Boost pressure/temp EGR in temp
Supercharger bypass valve Supercharger in temp
Coolant temp Loom 1/loom 2 interface
Exhaust lambda Power ground
Inlet lambda Test bed loom interface
Port injector (PFI Only) Ground
Ignition coil CAN 1
In cam sensor CAN 2
In cam VCT Accelerator pedal
EGR valve Cylinder pressure
Crank sensor WO connectors:
Knock sensor Backpressure valve
Exhaust temp Injection timing
Cold EGR valve Ex cam sensor
Throttle position Error signal
Coil Gnd engine EGR pump drive
Fuel rail pressure Ignition
DI INJ (DI Only) Boost valve drive
Loom 2/loom 1 interface EGR rate
Back pressure valve
Injection enable
Injection pulse width
+5V input
In cam
Figure 3.14 ECU Loom Connection List
The ECU is controlled remotely via a Controller Area Network (CAN) connection
which is connected to an ETAS 571.3 interface card which in turn is connected to a
computer running ETAS INCA V7.0 [104]. This is the only means of connecting to the
ECU with the exception of the ignition and the accelerator pedal inputs which are
controlled by the CADET V14 dynamometer control system. The main reason for this is
to ensure that some control over the engine is still possible in the event that the
computer running the ECU should malfunction. The low pressure fuel pumps are also
controlled by the dynamometer and can be used as yet another failsafe in the event of
the ECU computer malfunctioning. Besides very primitive control of the engine
73
through the dynamometer, the only means of controlling the engine is through ETAS
INCA V7.0.
Most of the sensor feedbacks can be displayed and logged through ETAS INCA
V7.0, shown below is a list of sensor feedbacks that were used for data analysis:
• Manifold pressure
• Throttle position
• Rail pressure
• Injection timing
• Spark timing
• Relative air charge
• Intake cam timing
• Exhaust cam timing
• Exhaust lambda
• Boost pressure
The ECU offers closed loop control of lambda and boost pressure and can be
configured to give closed loop control of EBP too, although this feature was not used
for this test work. The spark timing is dictated by a map. For this work the ECU was
initially configured to use a map of relative air flow and speed to give the spark timing,
however, due to the fact that a 1-way flow sensor was used and that the throttle was
never truly wide open, the relative air flow value was found to be unreliable so the
software was later reconfigured to use a map of MAP and speed instead.
Although the knock sensor was connected it was not used to control spark
timing, it was instead connected to an oscilloscope to give an indication to the
operator whether or not the engine was knocking. The cylinder pressure was however
used almost exclusively to give an indication of whether or not knock was occurring.
The main reason for this is because doubts were raised over the effectiveness of the
knock sensor given its location at the end of the cylinder head, and the feedback was
therefore never fully trusted.
74
3.3 Data Acquisition and Instrumentation
A list of sensors and their descriptions is given in Appendix A.2.
There are 3 logging systems used for this test work, a device for high speed
acquisition, a second one for low speed acquisition and a third for the PRTs. The high
speed system is based on a National Instruments (NI) USB-6353, which is capable of
logging up to 32 channels (16 differential) and has a maximum sampling speed of
1MS/s when used in a multi-channel configuration. This device logs the following
channels:
• Cylinder pressure
• Intake pressure
• Exhaust pressure
• Fuel flow rate
• Shaft encoder clock
• Shaft encoder reference
• Torque
The shaft encoder used with this device is an Encoder Technology EB58-
204040. This gives 0.25 CAD resolution which is required for knock analysis, however,
the NI USB-6353 uses multiplexing which means all the channels have to be logged at
that resolution. At low speeds this is not a problem, but at speeds of up to 5400rpm
this card is overwhelmed with data and is not suitable for logging all of the data
channels. At 5200rpm this card can handle a maximum of 8 channels. These channels
have been configured to run differentially, attempts were made to run them single-
ended in an effort to increase the channel count but the noise was found to have too
great an impact on the in-cylinder pressure transducer signal.
75
The low speed system consists of a NI USB 6210. This device is capable of
logging up to 16 channels single ended (8 differential). This device logs the following
channels:
• Exhaust temperature
• Exhaust manifold temperature
• Total coolant flow
• Block coolant flow
• Oil pressure
• Low pressure fuel pressure
• Coolant pressure
• Average exhaust pressure
The sampling frequency for all test work was 0.5Hz. These channels have also
been configured to run differentially.
The PRT acquisition system consists of an eDAM-9015 acquisition card. This

card supports up to 7 PRT inputs and connects to the computer by means of a serial
connection.
Where possible (and applicable) signal amplifiers have been placed as close as
possible to their corresponding sensors in an effort to reduce noise. All sensor cables
are also screened and ground to a common ground. This common ground has been
used for all instrumentation and is located at the data acquisition card end of the
cables. A common ground or “star earth” configuration was used in-order to avoid
“ground loops” in the instrumentation. The instrumentation cables have also been
placed as far away from power cables as possible, and efforts were made to ensure
that power cables never ran parallel to instrumentation cables.
In spite of these efforts the cylinder pressure transducer still exhibited a great
deal of noise, so further screening was used on other non-instrumentation cables (the
EBP servo cable was found to be a source of considerable noise). The cylinder pressure
signal exhibited a maximum of 3mV of noise for this test work. As cylinder pressure is
76
the most important feedback (noise is indistinguishable from knock and knock cannot
be calculated using average values) for this work a disproportionate amount of effort
went into ensuring this sensor was noise free in comparison to the other sensors.
Other sensors such as the intake and exhaust pressure transducers exhibited
some noise as well, however, these feedbacks can be averaged which cancels out the
vast majority of the noise.
3.3.1 Data Acquisition Software
All 3 data acquisition devices pass data to a bespoke software package known
as the “Transient Combustion Analyzer” developed at Brunel University by Dr Yan
Zhang. The interface of this software package is as shown in Figure 3.15.
Figure 3.15 Transient Combustion Analyzer Data Acquisition Software Interface
77
This software gives real time feedback of all low speed channels as well as
some of the high speed channels. There are 4 graph windows in the user interface
(Figure 3.16) which display the following:
Figure 3.16 Graph Window Layout
Window 1: Cylinder Pressure and knock intensity.
Both of these are updated on a cycle to cycle basis. The cylinder pressure
feedback comes directly from the in-cylinder pressure transducer. The Knock Intensity
(KI) is a measure of the difference in the actual pressure as reported by the in-cylinder
pressure transducer, and the “predicted” pressure. The predicted pressure is
calculated by averaging the pressure across the 10 points before and 10 points after
the point in question (Eq.3.2). For instance, the predicted pressure (pn) at a crank angle
n is calculated by taking the average of the points from (n – 2.5) up to (n + 2.5) at 0.25
CAD resolution as shown graphically in Figure 3.17. Both cylinder pressure and KI are
plotted against the engine crank angle as determined by the shaft encoder clock signal.
78
Figure 3.17 Band-pass Filtering Calculation
3.2
The knock intensity is then calculated using the following formula (Eq.3.3):
3.3
where pf is the pressure feedback from the in-cylinder pressure transducer. This
technique for measuring knock intensity was used because it is computationally much
less expensive than using a band-pass filter. This is also an established technique for
measuring KI and is used in industry and has been found to be very effective.
This reading has been found to be very sensitive to noise, hence the reason
why so much attention was paid to noise reduction for the in-cylinder pressure
transducer. The peak rate of cylinder pressure change per CAD (dp/dθ) has not been
used to measure KI specifically (although it is logged) because the dp/dθ in knock free
running has been found to be close to a value that would be considered knock. The
dp/dθ method is also a lot more susceptible to noise.
79
Window 2: Mass Fraction Burn
This is updated on a cycle to cycle basis. The mass fraction burn rate is
calculated using the Rassweiler and Withrow method (Rassweiler, et al [105]) which
states that there are 2 mechanisms that can bring about a change in cylinder pressure
(∆p). They are A) a change in volume (∆pV), and B) the presence of combustion (∆pC).
This can be expressed as follows (Eq.3.4):
3.4
The change in pressure brought about by a change in volume (V) is easily

calculated from the relationship based on the polytropic equation (Eq.3.5):
3.5
where i represents a crank angle and i+1 represents an incremental difference in

crank angle. Combining these 2 equations results in the following (Eq.3.6):
3.6
The combustion process takes place over a finite period of time and therefore
cannot be said to be a constant volume process, and although the above equation will
indicate when combustion has finished it is not possible to determine the quantity of
fuel burnt at this point. It is therefore necessary to take the sum of the ∆pC up to the
point it becomes zero, the crank angle at which this occurs shall be called j, this will
then indicate that all the fuel that is going to burn has been burnt by crank angle j. The
MFB can then be determined by dividing the sum of the ∆pC up to a crank angle of i by
the sum of the ∆pC up to a crank angle of j (Eq.3.7):
3.7
80
where x is the MFB. It is also necessary to calculate ∆pC relative to a reference
position. This reference position can be anywhere but for the purposes of this study
the TDC (or the point of minimum volume) was used.
There are a great many assumptions made with this model. The biggest
assumption is the value of the polytropic index n which is assumed to be constant at all
times during the compression and expansion strokes. This is not accurate as the value
can be expected to change slightly. The value of n is determined from the log P-V
during the compression process.
Window 3: P-V (or log P-V)
This is updated on a cycle to cycle basis. The cylinder and intake pressures are
fed back directly from the in-cylinder and intake pressure transducers respectively. The
volume is calculated using the crank angle signal form the shaft encoder using the
following formula (Eq.3.8):
3.8
where V is the volume, s is the stroke, b is the bore, CR is the compression ratio, CL is
the connecting rod length, θ is the crank angle in radians, and x is the crank-pin offset.
Window 4: Cylinder IMEP Vs. cycle
This is updated on a cycle to cycle basis. This window displays the Net IMEP
(NIMEP) values obtained from the previous 100 cycles and gives an indication of how
stable the engine is at that particular running condition akin to a graphical
representation of the COV of IMEP.
The NIMEP is calculated from the cylinder pressure data. Owing to the nature
of the test work, where one can expect great variations in the size of the pumping
loop, the NIMEP value was used as opposed to Gross IMEP (GIMEP) value.
81
Maintaining a constant NIMEP means any increase in the pumping loop size
due to higher or lower EBPs is offset by an increase in the gross work. There is no
significant advantage to using NIMEP as opposed to GIMEP, the biggest reason is for
convenience in comparing the single cylinder data to the 3 cylinder data. It is also
easier to calculate the power “cost” of the turbocharger using NIMEP.
In-order to maintain accuracy of in-cylinder pressure data, a Polytropic Index

Pressure Referencing (PIPR) system was used to “peg” the data as described by Brunt,
et al [106]. This has been used instead of an Intake Manifold Pressure Referencing
(IMPR) system because of noise errors associated with the intake pressure transducer,
and the fact that for certain IMOPs and cam durations the intake valves may still be
open at BDC with pressure waves present in the intake manifold compromising
accuracy. Attempts were made to implement a BDC IMPR system but this technique
was found to present an issue with the polytropic exponent which sometimes reached
as high as 1.38 which is not accurate for GDI applications, even with dry air. A
polytropic exponent of 1.32 was assumed for the PIPR pegging technique between 100
and 65 CAD Before TDC (BTDC). This value was used because it was in closest
agreement with the predicted GT-Power value and made comparing the 2 datasets
easier.
There are also a great deal of boxes that feedback numerical data in real time, a
description of these is given in Appendix A.3.
3.3.2 Calibration of Sensors
Flow meters
The Apollo flow meters are turbine type flow meters and were calibrated April
2011 immediately before being installed to the test rig. These flow meters have been
calibrated over their respective ranges of 2-20l/min and 5-50l/min.
The ABB FCM2000 coriolis flow meter was last calibrated in February 2011
across its full flow range.
82
Pressure sensors
The Druck pressure sensors were tested across their respective ranges using a
dead weight tester. The sensors were only tested at their maximum pressures and at
atmospheric pressure. No offsetting was required for any of these sensors.
The Kistler in-cylinder pressure transducer was also dead weight tested. Owing
to the nature of piezoelectric pressure transducers the sensor cannot be calibrated
using a steady state load. In-order to calibrate this sensor the settings on the Kistler
5011B charge amplifier had to be adjusted. The high-pass filter or Time Constant (TC)
was set to long and the Low-pass Filter (LF) switched on to allow the ∆p to be recorded
correctly and to be displayed for such a period so as to ascertain if it was correctly
calibrated or not. The transducer was calibrated and the calibration values given by
Kistler were found to be correct and no additional corrections were required.
The intake and exhaust pressure transducers were not dead weight tested
initially because the pressure transducers and charge amplifiers were brand new and
freshly installed immediately before test work commenced.
Temperature sensors
Both the PRTs and the thermocouples were calibrated using water in the
vicinity of 900C. The hot water would be placed in a container and the temperature
sensor was then immersed in the water along with an RS 206-3722 temperature probe
and the feedbacks from both sensors were compared. This procedure was carried out
once before the test work commenced and all instrumentation temperature sensors
were found to be in good agreement with the RS 206-3722, with the maximum error
being 20C.
Shaft encoder
83
The TDC position was determined using a Kistler 2629C capacitance probe TDC
sensor which can determine the dynamic TDC accurate to within 0.1 CAD. The
thermodynamic loss angle was calculated at 1200rpm to be 0.9 CAD using the
horizontal cut principle (by recording the crank angle at the same capacitance either
side of the peak capacitance point and interpolating between those 2 points to find
TDC). It was assumed to be constant throughout the speed range of the engine. The
reason for this assumption is the shaft encoder resolution is relatively course, and the
software cannot be adjusted to less than 0.25 CAD resolution. The engine was
equipped with the TDC sensor and motored over the full engine speed range. Using the
horizontal cut principle there was a difference in thermodynamic loss angle, however,
this difference was too fine to input into the software, so the same value was used for
all speeds.
Exhaust gas analyzers
Both the Rotork Analysis model 443 Chemiluminescence analyzer and the
Rotork Analysis model 523 Flame Ionization Detector (FID) analyzer were analyzed
whenever used using span gasses. The span gas used for the Chemiluminescence
analyzer was 500ppm (nominal) nitric oxide, and the span gas used for the FID analyzer
was 500ppm (nominal) propane.
3.3.3 Accuracy of Data and Known Measurement Errors
In order to maintain as high a degree of accuracy as possible 300 cycles were

logged at each of the test points. This was done in order to obtain as accurate a value
of IMEP and standard deviation as possible as described by Brunt, et al [107].
The long and short cam profile camshafts differed from the standard profile
camshaft in geometry, this meant that 2 different types of high pressure fuel pump
had to employed for this test work. The 2 different pumps had very different degrees
84
of control, with the pump used for the short and long duration cam profiles having a
much higher degree of control. A comparison of the 2 pumps is shown in Figure 3.18.
Rail pressure control with different high pressure fuel pumps

110
108
106
104
Rail Pressure (bar)
102 Standard profile cam pump

long and short profile cam pump
100 Target rail pressure
98
96
94
0 5 10 15 20 25 30 35
Time (seconds)
Figure 3.18 Accuracy of Fuel Rail Pressure Control at 3000rpm and 4 bar BMEP
The average pressure with the standard cam high pressure fuel pump can still
be seen to average 102 bar which will have negligible impact on the engine HC
emissions and burn durations, but it can be found to have a significant impact on the
ISFC. The output from the coriolis fuel flow meter can be found to echo this sinusoidal
trend, which means that for a constant IMEP the ISFC can also be found to echo this
trend.
The degree of control can be found to decline greatly with lower fuel pressures
and medium speeds. Figure 3.18 shows the worst case scenario as experienced over
the course of 30 seconds at 3000rpm and 4 bar BMEP. To keep the error associated
with this control problem to a minimum logs were taken when the fuel pressure
equalled the requested pressure. The maximum error associated with this
phenomenon was approximately 8 g/kWhr in ISFC at 3000rpm and 4 bar BMEP.
85
3.4 Daily Test Point
In order to assess the performance and repeatability of the engine and the
instrumentation a test point was selected where the engine would run every day in
order to see if there was any difference in performance. This test point is as shown in
Table 3.1:
Table 3.1 Daily Test Point Test Variables and Control Criteria
Speed 2000rpm
Load 4 bar NIMEP
Spark timing 50%MFB at 8 CAD ATDC
IMOP 100 CAD ATDC
EMOP 120 CAD BTDC
Boost 1.35 bar
Air humidity dry air
Air temperature 40 deg C
Coolant temperature 80 deg C
Oil temperature 80 deg C
SOI 319CAD BTDC
Rail pressure 134 bar
Throttle position 14.8% open
Emissions HCs only
Owing to logistical constraints only data from the instrumentation system was
saved, the data from the ECU was not saved. Only HCs were recorded because the
Chemiluminescence NOx analyzer was not used for all test work.
86
4
Chapter 4
Development of a Single Cylinder Engine GT-Power

Model
4.1 Introduction
A single cylinder engine GT-Power [4] model has been developed for assessing
the benefits of EIVC and LIVC strategies. The main reason for developing a single
cylinder engine model is to test the effectiveness of Deep Miller (EIVC and LIVC with
very high MAPs) at high engine loads, different geometric CRs and very high EBPs to
both complement and compare to the data obtained from the single cylinder
experimental engine. An analytical approach has been adopted for the majority of the
Deep Miller work due to constraints imposed by the single cylinder experimental
engine (such as insufficient ignition energy to ignite the charge with high levels of
boost, and the inability to run with fuel enrichment) that would make an equivalent
experimental analysis impossible at this moment in time.
A secondary reason is to assist in understanding what is happening in the test

results obtained from the single cylinder experimental engine, and to help explain
precisely what impact the breathing dynamics of the single cylinder experimental
engine is having on the overall results. A second more basic single cylinder engine
model was created for this purpose. Although the 3 cylinder engine model had been
setup and validated at Mahle Powertrain, it has proven necessary to develop a single
87
cylinder engine model because the single cylinder engine will have very different
breathing dynamics to the 3 cylinder engine, which renders a direct comparison
between the 3 cylinder engine model and single cylinder engine experimental results
impossible. The single cylinder engine is also a great deal more versatile than the 3
cylinder engine, and the MAP, Charge Air Temperature (CAT) and EBP can be anything
the user specifies, rather than being dictated by a turbocharger and intercooler.
The efficacy of the EIVC and LIVC processes is dictated heavily by the MAP/EBP
ratio, particularly during the gas exchange period, so introducing independent control
of each will allow a more thorough investigation of both processes. One of the main
benefits offered by the EIVC and LIVC processes is the possibility of running with a
higher geometric CR than would ordinarily be achieved with the standard Otto cycle.
This model will therefore be used to analyse and quantify this benefit and to offer a
greater deal of understanding of the mechanisms that bring about any performance
benefits.
A brief overview of the Mahle Powertrain 3 cylinder engine GT-Power model is

also given at the end of this chapter.
4.2 Development of the Single Cylinder Engine Model
There are 2 main variants of the single cylinder engine model that have been
developed for this test work. The similarities between both models will be described in
section 4.2.1. Of the 2 models, 1 of them is almost entirely automated and features
closed loop control of numerous systems, this is described in section 4.3. The second
variant contains very little closed loop control and is almost entirely manual and is
described in section 4.4.
Both models are geometrically identical with the biggest difference between
them being the presence of a knock model in model 1 and numerous other closed loop
control systems. Given the similarities of both of these models, with the exception of
sections 4.2.1, 4.3 and 4.4 where the differences between both models are described,
88
both single cylinder models shall henceforth both referred to as the singular “single
cylinder engine model”.
The single cylinder engine model has been designed to resemble the single
cylinder experimental engine as closely as possible. However, many parts of the intake
system have been left out of the model in an effort to reduce the complexity, and
because certain aspects of the system introduce anomalies and strange phenomena to
the model without contributing to the accuracy.
4.2.1 Single Cylinder Engine Model Details
This section describes the commonalities of the 2 models and discusses their
limitations. The flow modelling has been described in section 4.2.1.1 while the
combustion modelling has been described in section 4.2.1.2. The limitations of both
models have been described in section 4.2.1.3. A summary of the other parameters
used in this model is given in section 4.2.1.4.
4.2.1.1 Flow Modelling
The flow of gasses has been modelled one-dimensionally. This approach

involves calculating the flow properties by solving the Navier-Stokes equations for the
conservation of energy (Eq.4.1), continuity (Eq.4.2) and momentum (Eq.4.3) in one-
dimension. This requires the whole system (intake and exhaust) to be divided (or
discretized) into sub-volumes. The Navier-Stokes equations are solved over the
boundaries of these sub-volumes where scalar values (such as pressure, density and
temperature) are averaged across the entire sub-volume, and vector quantities (such
as mass flux and velocity) are calculated at the boundaries of the sub-volume. This
approach is described as a staggered grid and is shown graphically in Figure 4.1.
89
Figure 4.1 Staggered Grid Arrangement
4.1
4.2
4.3
where m is the mass of the volume, e is the total internal energy per unit mass, t is
time, p is pressure, V is volume, ṁ is the boundary mass flux into the volume, H is total
enthalpy, h is heat transfer coefficient, As is heat transfer surface area, Tfluid is the fluid
temperature, Twall is the wall temperature, dp is the pressure differential acting across
dx, A is the cross-sectional flow area, Cf is skin friction coefficient, u is the velocity at
the boundary, D is equivalent diameter, Cp is pressure loss coefficient, ρ is density and
dx is the discretization length.
An explicit solver has been used for all test work in an effort to model the
pressure pulsation effects in the intake and exhaust systems (caused by exhaust
blowdown pulses) as accurately as possible.
90
4.2.1.2 Combustion Modelling
A single zone model has been adopted to model the cylinder of the engine for
all test work. This assumes that the pressure, temperature and composition of the
cylinder gasses are uniform and homogeneous. This model is unable to differentiate
between burned and unburned gasses and models combustion as a simple heat
addition process. The limitations of this approach have been outlined in section
4.2.1.3. Due to the fact that combustion is not modelled burn rates must be
approximated rather than calculated.
The burn rate was approximated using a method developed by Wiebe [108]
known as the Wiebe function (Eq.4.4), given by:
4.4
where a and m are constants (5 and 2 respectively), xb is the MFB, and θ, θ0 and ∆θb
denote instantaneous crank angle, crank angle at start of combustion and total
combustion duration in CAD respectively.
Although this model gives a very good estimation of the MFB rate profile, it
requires the user to input a value for the 10-90% MFB duration first, as this model
cannot predict burn duration. The 10-90% MFB duration was taken from 3 cylinder
experimental data (obtained by Mahle Powertrain) recorded at a similar speed and
load point to the test point being tested in the single cylinder engine model.
The values of a and m used are based on previous experiments by Heywood, et

al [109] where the optimum values were determined as 5 and 2 respectively. These
values could have been recalculated using the same 3 cylinder data used to obtain the
10-90% MFB duration, however, they were not obtained from this because the MFB
duration profile can change dramatically depending on what cam, cam timing, EBP and
MAP combination is being tested. Also at the time this model was developed the single
cylinder experimental engine was not operational and no directly comparable data was
available. At a later stage when the single cylinder experimental engine was
91
operational, a comparison between the single cylinder experimental engine and the
single cylinder engine model was made. With the standard duration cam using values
of 5 and 2 for a and m respectively the burn duration profile obtained from the Wiebe
function was found to be in good agreement with the burn duration profile from the
single cylinder experimental engine.
4.2.1.3 Modelling Limitations
The greatest limitation of the single cylinder engine model is presented by the
combustion modelling approach adopted. The single-zone model is the simplest
method of modelling combustion, which can be advantageous in that it affords the
user the greatest degree of control over the combustion event. However, this requires
the user to know key combustion criteria prior to the simulation being carried out.
Multi-zone models go some way to remedying this by modelling a flame front

and calculating burn durations, as well as factoring in other phenomena such as
crevices and flame quenching, however, most are still incapable of modelling
combustion. Some models such as the Blizard and Keck turbulent entrainment model
(Blizard, et al [110]) improve upon this by calculating burn durations, however, this
model still requires some assumptions and simplifications to be made.
The Leeds University Spark Ignition Engine (LUSIE) model (Merdjani, et al [111])
is more advanced and accurate still, however, the code still requires work and
incorporating this code into this single cylinder engine model is quite difficult and time
consuming.
For semi-quantitative analysis of the effect of EIVC and LIVC on the engine’s
performance, the standard single-zone GT-Power code was deemed adequate, with
the understanding that all data obtained from this test work will be indicative only.
92
4.2.1.4 Other Modelling Parameters
An engine is replicated in GT-Power through the use of objects which are linked
in sequences to form intricate engine geometry such as the intake and exhaust
manifolds for instance. The 2 most commonly used objects in this model are Pipe and
PipeRoundBend which are used to create cylindrical pipes and pipe elbows
respectively.
Besides the very important geometrical parameters such as length and

diameter of the object, GT-Power allows the user to input other parameters for each of
these objects such as surface roughness and wall temperature. Where applicable these
values were determined by following the guidance given in the GT-Power help
navigator [112]. For other more important parameters specific to the engine, such as
intake and exhaust port geometry, intake and exhaust valve geometry and port flow
coefficients, values from the Mahle Powertrain 3 cylinder engine GT-Power model
(section 4.5) were used. For components specific to the single cylinder engine such as
intake and exhaust manifolds, the geometry of these components was recorded as
accurately as possible from the single cylinder experimental engine.
The way in which combustion timing is altered in GT-Power with the Wiebe
function is by changing the 50% MFB point instead of the spark timing. Therefore for
this chapter and chapter 5 the combustion timing will be referred to as either
combustion phasing or the 50% MFB point and not spark timing. Due to the fact the
combustion duration and Wiebe function variables were constant for all test work, the
spark timing is directly correlated to the 50% MFB point however.
4.3 Single Cylinder Engine Model 1: DoE Based Test Plan
Single cylinder engine model 1 was developed purely to study Deep Miller, and
was not employed for any low load work. As mentioned briefly in section 4.1, EIVC and
93
LIVC processes are dependent upon many variables including MAP, CAT, EBP, CR,
Intake Valve Open (IVO) time and intake cam duration. For satisfactory analysis of EIVC
and LIVC processes it is necessary to determine the effect of each of the above
variables. Testing all of these variables by hand is a very long winded process, which is
the reason why a DoE must be employed and why this model requires closed loop
control of many different systems.
Model 1 has been developed purely for this application and can run almost
entirely independently of any input from the user, which makes it highly dynamic and
the perfect tool for executing large DoEs with.
A diagram of model 1 is shown in Figure 4.2.
Figure 4.2 Single Cylinder Engine Model 1
This model has closed loop control of the following:
• Intake pressure (by controlling to a requested MAP)
• Intake temperature (by controlling atmospheric temperature)
• Knock intensity (by controlling combustion phasing)
• Exhaust temperature (by controlling lambda)
• Exhaust pressure (by controlling to a requested EBP)

94
Although the model does feature closed loop control of intake pressure, this
can only control to a requested intake pressure. This may not seem necessary as the
operator could simply request an intake pressure instead of having to control it, but
this was not possible in practice because the pressure at the source of the intake
system did not necessarily correspond to the pressure in the intake runner, hence the
reason why a throttle with closed loop control was implemented. The same problem
also applied to the intake temperature.
Attempts were also made to implement closed loop control of BMEP by

regulating intake pressure. This could not be achieved as it was found to interfere with
the knock model. The knock model and BMEP control were found to interact
negatively and would generate the following scenario. If the combustion phasing was
retarded slightly by the knock model the BMEP control would increase the boost
pressure to compensate for the resulting loss of BMEP, this in turn would lead to a
greater degree of combustion phasing retard which would require yet another rise in
boost pressure and the boost pressure was found to progressively increase while the
combustion phasing would become more and more retarded.
The exhaust temperature is regulated by changing the degree of enrichment in

the cylinder. The temperature was regulated according to the exhaust gas temperature
in the exhaust runner. The maximum degree of enrichment was fixed to a lambda of
0.85.
For this model the EBP was regulated by means of a variable diameter orifice
rather than a conventional butterfly valve as implemented on the test engine, simply
because the variable diameter orifice offered a greater deal of control with the
Proportional Integral Derivative (PID) control system. The location of the variable
orifice diameter was dictated by the single cylinder experimental engine, hence the
reason why it is located so far downstream of the exhaust port. The EBP was regulated
according to the average pressure in the exhaust runner over an entire engine cycle.
The knock model is the most complex closed loop system and has been
described in section 4.3.1.
95
4.3.1 Single Cylinder Engine Knock Model
4.3.1.1 Functionality of the Knock Model
The single cylinder engine knock model (henceforth known as just the knock
model) has been applied to the engine itself and is calculated, and controlled, in real
time. Please note this is a knock model only and not to be confused with an
autoignition model. Autoignition encompasses both knocking combustion and pre-
ignition, and this model is unable to predict pre-ignition reliably (see section 6.4.4).
This knock model is based on 3 key equations that are fundamental to its
operation. The first is the induction time integral described by Livengood, et al [113]
known as the Livengood-Wu formula (Eq.4.5). The second is the combustion model
developed by Douaud, et al [114] known as the Douaud and Eyzat combustion model
(Eq.4.6). The third equation is the knock intensity formula developed by Gamma
Technologies [112] (Eq.4.7). The first component is the Livengood-Wu integral which is
given by:
4.5
where the integral limits -100 and thkn equal the angle in CAD BTDC where
compression is assumed to begin and the angle at which knocking is assumed to have
started respectively. Once T is equal to 1 knocking combustion has commenced. The
symbol τ represents the Douaud and Eyzat component, given by:
4.6
where P is the precursor reaction rate multiplier, ON is the fuel octane number, p is
the instantaneous cylinder pressure, A is the activation energy multiplier and Tu is the
instantaneous unburned gas temperature. The precursor reaction rate multiplier was
96
adjusted for each individual engine speed to increase knock model stability and to
ensure knock was occurring at the right moment in the cycle.
The start angle of the Livengood-Wu integral is usually the start of

compression. The start of compression is however very difficult to determine as it does
not necessarily correspond to intake valve closure. The knock model was initially set to
start integrating the moment the flow through the intake valves stopped, however,
this resulted in added complication to the knock model and introduced another
potential issue with knock model functionality. This angle was therefore fixed at 100
CAD BTDC because this was found to have negligible impact on the functionality of the
knock prediction point and intensity. The knock model still has the facility to revert
back to using the flow through the intake valves however.
As soon as the Livengood-Wu integral equals 1 knocking combustion is present

and the magnitude of this knocking combustion is calculated using the following
formula:
4.7
where A is the knock index multiplier, km is the percentage of cylinder mass unburned
at knock initiation, VTDC is the cylinder volume at TDC, VI is the cylinder volume at
knock initiation, Ta is the activation temperature (which has been assumed to be 6000
K) and φ is the equivalence ratio.
In order to solve these equations in real time (as is necessary for the closed
loop control of combustion phasing), numerous outputs from the engine must be fed
into the knock model. These required outputs are as follows:
97
• Instantaneous in-cylinder pressure
• Instantaneous in-cylinder unburned zone temperature
• Instantaneous in-cylinder lambda
• Instantaneous in-cylinder unburned gas mass fraction
• Instantaneous intake valve lift
• Instantaneous in-cylinder volume
• Crank angle
The intake valve lift does not appear in any of the knock model equations
(Eq.4.5-4.7) but is still required for the satisfactory operation of the knock model. The
intake valve lift is fed into the knock model to give the user the option of assuming the
start of compression occurs when the intake valve closes instead of having to provide
an arbitrary start of compression point.
Of all of the engine outputs the only one required numerous times is the crank
angle. This is output from the engine a total of 5 times by 5 different sensors (denoted
CAx where x equals 1 – 5) for 5 different purposes as shown in Table 4.1.
Table 4.1 Description of the Various Crank Angle Sensors in the Knock Model
Crank angle sensor Description

CA1 TDC determination
CA2 BDC determination
CA3 Value of Livengood-Wu integral at the last point before knocking combustion
CA4 Last point before knocking combustion determination
CA5 Knocking combustion point determination
TDC and BDC determination though rather rudimentary still needs to be

calculated. As mentioned previously, one of the factors of the DoE is geometric CR, and
there is no way to feed back the geometric CR directly in GT-Power, therefore it must
be calculated by dividing the volume at BDC by the volume at TDC.
Each of the crank angle sensors is used in the same way and all of them are
used as a control input for a hold object. A hold object requires an input, a control
input and a trigger point. The output of the hold object is fixed at the value of the input
98
at the trigger point which is triggered by the control input. This has been
demonstrated in Figure 4.3 which shows the TDC determination. For TDC the input is
cylinder volume, the control input is the crank angle and the trigger point is 0 CAD.
TDC Determination with a Hold Object

2.50E+05
2.00E+05
Cylinder volume (mm^3)/Input
1.50E+05
Input
Output
1.00E+05
Trigger
5.00E+04
0.00E+00
-120 -60 0 60 120 180 240 300 360 420 480 540 600
Engine crank angle (CAD ATDC)/Control input
Figure 4.3 TDC Determination with a Hold Object
Figure 4.3 depicts the first cycle of a simulation, for every cycle after this one
the output of the hold object will not until change again until 0 CAD is reached again.
This technique has been employed heavily throughout the knock model, it is
used to hold values of pressure, temperature, unburned gas mass fraction and volume
at the point where knocking combustion occurs. In these cases the Livengood-Wu
component of the knock model was used as the control input and the trigger was set
to 1. This ensured the Douaud and Eyzat and knock intensity components were
calculated just once per cycle and that they were calculated at the correct time (at the
point where knocking combustion occurs).
One of the problems experienced with this system was that the Livengood-Wu
component never equalled exactly 1, this is due to the fact that the resolution of the
simulation was 1 CAD. Obtaining a Livengood-Wu integral of exactly 1 is not possible
due to the fact the resolution of the simulation would have to be infinitely small. This
99
problem has been illustrated in Figure 4.4 where the accuracy of the point where
knocking combustion occurs can be seen.
In this instance the point where knocking combustion occurs would be

recorded as 23 CAD After TDC Firing (ATDCF) by the knock model, whereas in actual
fact the point where knocking combustion actually occurs is closer to 22.5 CAD ATDCF.
This error can be up to almost 1 CAD. This error in itself is acceptable as most ECUs are
accurate to 1 CAD, however, this error led to convergence issues within the knock
model. An interpolation system was therefore incorporated to interpolate between
the points both before and after the knocking combustion line to obtain a very precise
crank angle where knocking combustion occurred.
Livengood-Wu Integral Accuracy at 1 CAD Resolution

1.3
1.25
1.2
1.15
Livengood-Wu integral value
1.1
1.05
Livengood-Wu
1 integral
Knocking
0.95 combustion
0.9
0.85
0.8
0.75
0.7
18 19 20 21 22 23 24 25 26
Engine crank angle (CAD ATDCF)
Figure 4.4 Livengood-Wu Integral Accuracy at 1 CAD Resolution
This interpolation system was applied to each of the required inputs for the
knock intensity (Eq.4.7) component of the knock model, including τ. The incorporation
of this interpolation technique increased the accuracy of the knock model by up to
271%. This example also highlights the difference between crank angle sensors CA3
and CA4 and crank angle sensor CA5. CA3 and CA4 are used to solve the knock model
100
equations at the last point before knocking combustion commences, and CA5 is used
to solve the knock model equations after knocking combustion has begun.
Once the interpolation has taken place a knock intensity number is obtained
from Eq.4.7. This number is divided by a target knock intensity as input by the user. It
is this number that the PID controller is trying to target, all the time this number is not
equal to 1 the combustion phasing will be changed.
4.3.1.2 Irresolvable Issues with the Knock Model
One of the biggest problems encountered with the knock model was the fact
that the 50% MFB point was so variable from case to case (which is to be expected
when running a large DoE). The closed loop control system for the combustion phasing
on its own is incapable of dramatically changing the 50% MFB point in a short space of
time due to the fact there are so many other closed loop systems at work on the
model. This therefore requires the 50% MFB point to be moved incrementally along
with each of the other closed loop systems, else all the closed loop control systems
would become unstable.
The speed of convergence was found to dramatically improve when tables

were incorporated to offset the default 50% MFB point according to engine operating
conditions at that particular test point. These tables increase or decrease the degree of
combustion phasing retard automatically before simulation has even commenced, thus
allowing the 50% MFB point to converge within a shorter space of time (250 cycles).
There are a total of 4 tables correcting the 50% MFB point. These tables change
the 50% MFB point phasing based on:
• MAP (Table 4.2)
• EBP (Table 4.3)
• Geometric CR (Table 4.4)
• IMOP (Table 4.5)

101
The values in these tables are constant for all speed and load cases.
Table 4.2 MAP Correction Factors
MAP (bar) Correction factor

2 13.7
4 24.4
Table 4.3 EBP Correction Factors
EBP (bar) Correction factor

2 6.2
4 12.4
Table 4.4 Geometric CR Correction Factors
Geometric CR Correction factor

9 27.51428571
12.5 38.21428571
Table 4.5 IMOP Correction Factors
IMOP (CAD ATDC) Correction factor

-282 -39.41428571
-240 -28.21428571
A general rule of thumb with these tables is the larger the correction factor the
more retarded the initial combustion phasing would be.
The values in the these tables have been approximated based on test work
carried out with the 3 cylinder engine model involving changing MAP and EBP (Table
4.6), they were not controlled independently, and geometric CR was not changed at all
at any point in the test work. It is for these reasons that just one map incorporating
this data on its own was insufficient in giving suitable 50% MFB point approximations
that were close enough to the actual values to give convergence.
102
Table 4.6 MAP and IMOP 3 Cylinder Map of BLD 50% MFB Points at a Speed and Load of 5000rpm and
24 bar BMEP Respectively
IMOP (CAD ATDC)

-220 -230 -240 -250 -260 -270 -280 -290 -300
177.31 29.7638 27.67556 25.99715 24.7249 23.75866 23.25416 23.154 23.46551 24.07751
179.28 29.75526 27.6562 25.96599 24.68133 23.69995 23.17142 23.05351 23.33914 23.88449
181.25 29.74672 27.63685 25.93484 24.63776 23.64124 23.08868 22.95302 23.21277 23.69147
183.22 29.73818 27.61749 25.90368 24.59419 23.58253 23.00594 22.85253 23.0864 23.49845
185.19 29.72964 27.59813 25.87253 24.55062 23.52382 22.9232 22.75204 22.96002 23.30542
187.16 29.7211 27.57878 25.84137 24.50705 23.46511 22.84045 22.65155 22.83365 23.1124
189.13 29.71256 27.55942 25.81022 24.46348 23.4064 22.75771 22.55106 22.70728 22.91938
191.1 29.70402 27.54007 25.77906 24.41991 23.34769 22.67497 22.45057 22.58091 22.72636
193.07 29.69548 27.52071 25.74791 24.37634 23.28898 22.59223 22.35008 22.45454 22.53334
195.04 29.68694 27.50136 25.71675 24.33277 23.23027 22.50949 22.24959 22.32817 22.34032
Cam
duration 197.01 29.6784 27.482 25.6856 24.2892 23.17156 22.42674 22.1491 22.2018 22.1473
(CAD) 199.9901 29.2393 27.14799 25.4565 24.18046 23.2353 22.67515 22.56563 22.81887 22.8773
202.9702 28.8002 26.81398 25.2274 24.07172 23.29904 22.92356 22.98216 23.43594 23.6073
205.9504 28.3611 26.47997 24.9983 23.96298 23.36279 23.17196 23.39869 24.05301 24.3373
208.9305 27.922 26.14596 24.7692 23.85424 23.42653 23.42037 23.81522 24.67008 25.0673
211.9106 27.4829 25.81195 24.5401 23.7455 23.49028 23.66877 24.23175 25.28715 25.7973
214.8907 27.0438 25.47794 24.311 23.63676 23.55402 23.91718 24.64828 25.90422 26.5273
217.8708 26.6047 25.14393 24.0819 23.52802 23.61777 24.16558 25.06481 26.52129 27.2573
220.851 26.1656 24.80992 23.8528 23.41928 23.68151 24.41399 25.48134 27.13836 27.9873
223.8311 25.7265 24.47591 23.6237 23.31054 23.74526 24.66239 25.89787 27.75543 28.7173
226.8112 25.2874 24.1419 23.3946 23.2018 23.809 24.9108 26.3144 28.3725 29.4473
250 24 23 23 24.5 24.5 19.9 20.4 20.9 30
The data available for populating the intake cam duration and IMOP map was
quite extensive but only over a relatively short cam duration window, which is another
reason why it could not be used directly.
4.4 Single Cylinder Engine Model 2: Correlation of Thermodynamic Data
Single cylinder engine model 2 very closely resembles single cylinder engine
model 1 (Figure 4.2), with the only difference being the actuators have been removed
from the closed loop systems in model 2. Single cylinder engine model 2 was modified
in some instances to provide some closed loop control (most commonly fuel
enrichment control was re-employed because it was a very long process to determine
the correct AFR manually).
It was used for both low and high load work and was employed heavily for
deducing whether or not charge scavenging was occurring with LIVC. It was also used
to extract P-V data from GT-Power as this cannot be done from the DoE directly. This
103
provided a good opportunity to check the integrity of the data obtained from the DoE
too.
4.5 Mahle Powertrain 3 Cylinder Engine Model
A 3 cylinder engine GT-Power model has been developed by Mahle Powertrain

(Figure 4.5). It is based on the first generation 1.2l 3 cylinder downsized engine with a
single turbocharger and a geometric CR of 9.25. The geometry of the engine and
cylinder have been replicated very accurately in the model, and the port flow
coefficients of the intake and exhaust valves were measured on a flow bench.
This model has been fully correlated by Rothmaier [115] and was found to be a
very good correlation to the engine, particularly at part load. This model has also been
found to predict the breathing dynamics of the engine very accurately.
This model has been employed mostly for comparing the breathing dynamics of
the single cylinder engine to the 3 cylinder engine where no test data from Mahle
Powertrain exists or could be obtained. The results of this can be found in section 6.2.
It has also been used to help correlate the knock model (Table 4.6).
104
Figure 4.5 Mahle Powertrain 3 Cylinder Engine Model
105
5
Chapter 5
Application of the Deep Miller Cycle to Improve Fuel

Economy and Increase Maximum Load
5.1 Introduction
This chapter describes the development of a DoE based test plan and presents
the results of an analytical study into the Deep Miller cycle process (EIVC and LIVC with
very high boost pressures) executed in a single cylinder engine GT-Power model (as
described in chapter 4).
The intention of this test work is to ascertain the effects of EBP and geometric
CR on the effectiveness of Deep Miller at key speeds and loads that should
theoretically yield the most significant fuel economy benefit for a downsized engine.
This chapter has been arranged into 3 major sections, 1 section on describing
the DoE based test plan as well as test points, and the other 2 sections showing results.
The first section of this chapter explains why, and how, a DoE approach has been
adopted. It also explains what type of DoE has been employed and details the
screening process used to validate the model. This section also explains the choice of
test points and why they have been selected.
The first results section is focussed on the use of EIVC and LIVC to optimize fuel
consumption (or ISFC). It has been arranged to show ISFC trends with increasing EBP
and geometric CR, followed by an analysis of the cam timings. Following this is an
106
explanation of the phenomena that impacted very heavily on these results such as cam
timing, scavenging and engine breathing. This is followed by a section explaining the
positive and negative aspects of increasing geometric CR.
The second section is a study of the maximum loads that can be obtained using
Deep Miller. This section studies the impact of EBP on maximum load and explains how
the engine is restricted from producing higher loads.
5.2 Development of a DoE Based Test Plan
5.2.1 Introduction
The intention of this test work is to assess the impact of geometric CR and EBP
on Deep Miller effects. In order to do this many different cam durations must be
tested at many different IMOPs. These can be assessed by Changing One Separate
factor at a Time (COST), however, this is a very long winded approach, particularly
when several variables will need to be assessed, and it has also been found to give
misleading results (in some cases) in comparison to a fractional factorial design
(Eriksson, et al [116]). Therefore a fractional factorial DoE approach will be adopted
instead.
The DoE approach is used to assess the impact of many different variables in as
few experiments as possible. It is widely adopted in many different industries as time is
usually a tight constraint and it is usually very important that selected experiments are
maximally informative.
5.2.2 Screening
A program of screening tests was constructed to determine the most accurate

fractional factorial regime for this particular test work. The screening process in DoE
107
terms usually refers to a series of tests carried out to determine which variables can be
discounted from the DoE because their impact on the overall result is negligible.
However, for this work this was not carried out as each of the 5 variables have been
well documented to have a significant effect on the ISFC of an engine. For the purposes
of this study the screening process refers more to the process of determining a
suitable size for the DoE and optimum configuration. Additional screening was also
carried out to determine optimum gain values for the various PID control systems.
The majority of the screening process was carried out at the 5000rpm, 24 bar
BMEP speed and load point, with only a relatively small amount carried out at other
speeds. This point was selected for many reasons, chief among which is because the
Miller cycle benefits at high speed and high load are of the greatest interest in this
study (for its synergies with downspeeding). A high load point is also necessary to test
the knock model and a high speed is also necessary to test the closed loop enrichment
control and to gauge its influence on ISFC.
There are 3 main studies that make up the back bone of this screening process,
the first of which was a very lengthy process of ensuring the knock model functioned
correctly. The impact of this process can be seen in section 4.3.1 where the knock
model has been fully described. The other closed loop control systems were also
checked for correct functionality during this phase.
The second was to assess which factorial design would be optimum for this
application. The initial approach to this was to determine how many levels there
should be for each of the variables. Very coarse preliminary Box and Draper DoEs were
carried out to determine which of the variables would have the greatest impact on
ISFC. CR was found to have a more subtle and predictable impact on ISFC so this had
the fewest levels, while cam duration (the focus of this study) required the greatest
number of levels. The number of levels selected for each of the factors is as shown in
Table 5.1 (bracketed values represent the ranges and resolutions associated with the
1000rpm DoE):
108
Table 5.1 DoE Ranges, Levels and Resolution (1000rpm)
Table 5.2 Intake Valve IMOP Ranges for Different Duration Cams
Cam duration (CAD) IMOP range (CAD ATDC) Levels

152 50 - 120 8
196 60 - 120 7
240 80 - 120 5
276 100 - 140 5
312 110 - 150 5
Speed is another factor that has been investigated but it has not been included
in the above list because the knock model must be reconfigured for each speed, also
closed loop control of enrichment is not required at lower speeds. Therefore, for the
purposes of this study, different speeds shall be considered as different studies rather
than a variable.
Owing to the number of factors and the number of levels involved a full
factorial COST approach DoE is immediately unfeasible as it will consist of 6000 points
per speed and a DoE of this size will require computer resources that are, at this
moment in time, unavailable. A fractional factorial model was therefore adopted.
Test work from the experimental engine suggests a quadratic model is best
suited to capture the correct response. Quadratic models are also required to locate
optimal points. The intention of this study is not to capture the optimal points with a
great deal of precision because the knock model and combustion models cannot
permit this (due to their inherent weaknesses). The intention of this DoE test plan was
therefore not to determine with great accuracy where the optimal points arise but to
get an idea of trends across the entire range of all of the variables.
The third stage of this screening process was to determine what type of
fractional factorial DoE is best suited to the test range. This has been described in
section 5.2.3.
109
5.2.3 DoE Type
A Latin hypercube (McKay, et al [117]) design was adopted at first but this was
later abandoned due to problems experienced with the extreme points of the
experimental region, such as the corners of the hypercube where “lips” would form
due to the curve fitting algorithm. This is due to the fact that the Latin cube design will
never test the same value of a particular variable twice and is therefore very reliant on
the curve fitting algorithm. It was found to be very reliant on the knock model
functioning correctly over the entire range of the experimental region, which it rarely
did. A D-optimum (described by Eriksson, et al [116]) approach was then adopted
instead.
Figure 5.1 gives an example of a Latin cube quadratic point distribution and a
D-Optimum quadratic point distribution over the same range with the same number of
test points with 2 variables. The Latin hypercube points are noticeably more clustered
toward the centre of the experimental region and the D-Optimum points are more
uniformly distributed and cover the extreme regions of the experimental region.
Figure 5.1 Latin Hypercube (left) and D-Optimum (right) Quadratic Point Distribution Comparison
A D-optimum design was found to be more effective at reducing the “lipping”

effect at the extreme points of the hypercube (Figure 5.2), and is also better suited to
testing asymmetric experimental regions which this is due to the fact the IMOP range
changes with cam duration. The experimental region can be found to be highly
110
asymmetric for this test work, not only because of the IMOP limit but also because
certain MAP, EBP, IMOP and cam duration combinations would lead to unfeasibly high
volumetric efficiencies and loads. Due to logistical constraints an asymmetric
experimental region could not be used and it was instead broken down into regions of
different IMOPs, so this advantage of the D-optimum design could not be utilised in
practice.
Figure 5.2 Latin Hypercube (left) and D-Optimum (right) Quadratic Curve Fitting Comparison with Lips
Circled
A further advantage of the D-optimum design is the ability to extract and add
data to the experimental region after it has been carried out. For areas where data
integrity was quite low and the knock model was malfunctioning, data could be
obtained manually using single cylinder engine model 2 which could be used to
manually populate the experimental region.
5.2.4 Limitations of the DoE
Strictly speaking there are many variables besides the 5 tested that can have an
impact on ISFC, such as assumed exhaust temperature limit, intake temperature,
combustion phasing, fuel quality and age, and, in the case of the single cylinder engine,
exhaust and intake geometry and runner lengths. However, some of these variables
are not easily controlled such as the exhaust and intake manifold tuning. Fuel age and
111
fuel quality, which can be shown to have a profound effect on engine autoignition
performance, is another variable. Unfortunately it is not possible to replicate different
fuels reliably in GT-Power.
The exhaust temperature limit for most turbocharged engines is dictated by the
turbocharger. The temperature ceiling for all test work was set to 9300C. This is beyond
the maximum temperature limit imposed for all experimental test work which was
9000C. The reason for setting different temperature limits is because the physical
engine was constrained by the lambda sensor to a low maximum exhaust gas
temperature, and the majority of the GT-Power work is fundamentally theoretical so a
slightly higher exhaust temperature limit was imposed. Modern materials such as
austenitic steel and nickel-chromium alloys have raised the allowable exhaust
temperature limit, modern technologies such as Water Cooled Exhaust Manifold
(WCEM) also increase this limit dramatically (Taylor, et al [118]). The imposed
temperature ceiling of 9300C is therefore relatively conservative.
The number of experiments used was the maximum number the computational
resources would permit in an effort to define the regions where EIVC and LIVC were
not feasible, and the optimum points with as high a degree of accuracy as possible.
This number is approximately 150 experimental points. For an experimental region of
this size and for cam duration to be captured at a resolution of 10 CAD, 150 points is
still insufficient so the experimental region was broken down into sub experimental
regions each of 150 experimental points. Due to the fact the IMOP ranges changed
with cam duration, and the fact the cam duration had to be assessed at such a high
resolution, the 4 DoEs were for 4 different cam duration ranges which were as follows:
• 152 – 192 CAD duration
112
An attempt was made to split this into finer cam duration ranges still but this
was later abandoned because it was much more difficult to analyse the data and
yielded no improvement in the accuracy of the results.
The cam profiles were calculated by inputting a scale factor for the standard
cam, for instance the 152 CAD duration cam profile was determined by scaling the
standard cam down a factor of 1.58. The valve lift was kept constant however at 11mm
for all cam profiles. Figure 5.3 shows 5 valve profiles of 152, 196, 240, 276 and 312 CAD
duration.
12
11
10
8
152 CAD
7
Valve lift (mm)
196 CAD
6
240 CAD
5
(standard)
4 276 CAD
3 312 CAD
0
-250 -200 -150 -100 -50 0 50 100 150 200 250
Duration (CAD)
Figure 5.3 Comparison of GT-Power Calculated Intake Cam Profiles for 152, 196, 240, 276 and 312 CAD
Durations Cams
A Radial Base Function (RBF) was found to give results most like that acquired
from the test work and this was therefore used for curve fitting. The design
optimization tool in AutoDOE was used to extract raw data from the DoEs where
variables such as EBP, geometric CR, cam duration and IMOP could be fixed and a
target power could be requested. The software could interpolate to give the correct
MAP for that condition as well as interpolated values of ISFC and numerous other
variables, this was found to be the most effective way to analyse the data.
113
Data was not necessarily obtained at the resolution tested at in the DoE. For
instance, cam duration was not obtained at 10 CAD increments but for 5 different cam
durations of 152, 196, 240, 276 and 312 CAD. The reason for this is the malfunctioning
of the knock model meant trend lines were inaccurate, so the decision was made to
test as close as possible to set cam durations so no curve fitting was required due to
the fact that the extremities of the DoE are relatively well populated with raw data.
Owing to the great number of closed loop control systems the number of runs
per case was set to 250 to allow the numerous closed loop control systems to
converge. Any points that had not converged in this number of cycles were removed
from the DoE manually and omitted from the analysis. This many runs may alter the
result slightly but all cases were run 250 times, so if the results were distorted at all
they would all be distorted equally and still give a reliable representation of trends.
5.2.5 Test Point Description
The aim of this test work is to assess the performance benefits that can be
realised through the use of EIVC and LIVC at high load across a different range of
speeds. The DoE nature of this work means the data can also be used to get a better
understanding of which strategy is best for increasing peak load too.
At this moment in time emissions are not key concerns because generally
speaking loads as high as those tested are not part of the NEDC cycle and are therefore
not of direct interest in this simulated portion of the work. All work is also steady state
as there is no turbocharger present so transients are not of great interest or relevance.
The decision was made to adopt a DoE test plan at 3 different speeds and loads
where downsized engines are most heavily constrained, and/or where adoption of the
Miller cycle could greatly improve performance. The scope of the DoEs is outlined in
section 5.2.2 and the speed and load points tested are as follows:
• 1000rpm, 14 bar BMEP (15.14 bar NIMEP)
114
Please note that all BMEP values given are 3 cylinder equivalent loads. The
corresponding NIMEP values have been given in brackets. These points can all be seen
to be relatively high in load but the phenomena preventing higher loads from being
obtained can be seen to be quite different. One thousand and 2000rpm were tested
because autoignition is a particular problem at low speed. The final speed of 5000rpm
was tested due to the fact that fuel enrichment is required at this speed at high load. A
load of 14 bar BMEP at 1000rpm has been selected due to the fact that given the
current state of turbocharging/eboosting technology, this is likely to be the limit of
what can be achieved using extreme EIVC and LIVC at such a low speed. The 24 bar
BMEP load points have been selected so as to enable a direct comparison with the
experimental engine. An identical MAP limit of 4 bar abs has also been imposed to
make results comparable to the experimental engine. Additional reasons for running at
these speeds and loads are explained in the sections 5.2.5.1, 5.2.5.2 and 5.2.5.3 for
1000, 2000 and 5000rpm respectively.
No low load DoEs were carried out because they are out of the scope of this
portion of the test work, this is due to the fact that combustion is particularly affected
at low loads and the GT-Power model cannot predict COV values, nor can it factor in
combustion degradation due to too low EOC temperatures.
For all test work the EMOP was set to the maximum valve overlap position. This
was fixed to allow consistency across all test points and to provide the maximum
expansion ratio. This may be detrimental with higher EBPs, however, test work carried
out on the single cylinder experimental engine suggests that the maximum valve
overlap position is optimum for the exhaust valve with the baseline standard cam. EIVC
and LIVC work with the 3 cylinder GT-Power model also suggests the maximum valve
overlap position is optimum for the exhaust valve.
Although the single cylinder experimental rig has CAT control, the vast majority
of engines have very limited CAT control. Accurate determination of the CAT is made
115
even more difficult when EIVC and LIVC strategies are employed because the charge
temperature can be reduced by over pressurising/under pressuring the charge. It is for
this reason that the CAT will not be varied during this test work, it will simply be set to
a level that can be achieved by the intercooler on the 3-cylinder engine (400C).
The Start Of Injection (SOI) timing has been derived from data obtained from
Mahle Powertrain. The SOI timings are different depending on speed and load and are
shown in Table 5.3.
Table 5.3 SOI Timings for Each of the Speed and Load Points
Speed (rpm) BMEP (bar) SOI (CAD BTDCF)

1000 14 246
2000 24 259
5000 24 349
5.2.5.1 1000rpm, 14 bar BMEP
One thousand revolutions per minute is a very low speed to be running at high
load, even for downsizing and downspeeding concepts, however, the requirement for
performance is still there at these speeds (Turner, et al [50]) and this is still a very valid
speed point for high load.
Although the load is relatively low at this test point, performance can still be
seen to be autoignition (specifically knock) limited. In most cases though the
combustion phasing has only needed to be retarded a few CAD away from MBT. The 3
cylinder engine is run in the full valve overlap position at this speed and load to allow
for as much scavenging as possible in the interest of mitigating knock. Scavenging is
very effective at mitigating knock by evacuating the cylinder of as much hot residual
gas as possible from the previous cycle. However, allowing fresh air to flow through
the cylinder can be detrimental as it has the effect of diluting the exhaust gasses,
causing the engine to appear to be running lean at the lambda sensor. As a
consequence of this the EMS will inject more fuel to restore the engine to
“stoichiometric” when in actual fact it is actually enriching the mixture as seen from
the cylinder. This phenomenon has been explained in more detail in section 5.3.4.
116
In the case of the 3 cylinder engine a map exists that allows the EMS to
compensate for this accidental enrichment by running the engine slightly lean
(according to the lambda sensor). No such correction has been applied for this DoE.
5.2.5.2 2000rpm, 24 bar BMEP
This is arguably the most important speed to optimise as its one of the most
common operating speeds for automotive engines. At 2000rpm knocking combustion
still imposes a significant constraint on maximum load but much higher loads can be
reached than at 1000rpm due to the higher engine speeds and the faster burn rates
associated with greater degrees of in-cylinder turbulence.
For most downsized engines peak (or at least near peak) load can be achieved
at 2000rpm. It can also be achieved without the need for enrichment which makes it a
very desirable speed to run at.
In the case of the 3-cylinder engine, the MAP exceeds the EBP at this speed and
load, the engine is also run at the full valve overlap condition. This means the cylinder
will be very well scavenged at this speed, to the point that the EMS is set to run slightly
lean to compensate for this scavenging effect. For this DoE analysis the engine was run
at λ = 1 without any compensation for scavenging.
5.2.5.3 5000rpm, 24 bar BMEP
Five thousand revolutions per minute and 24 bar BMEP represents the
maximum speed and load point for this test work. At this speed and load the exhaust
temperatures are going to be very elevated and fuel enrichment will be required in
most cases to regulate this. Although the objective of this work is to optimise ISFC (as
with the other 2 speeds and loads), the mechanism by which this is achieved is slightly
117
different. An improvement in ISFC is yielded here through reduced fuel enrichment
rather than through mitigating knock directly.
Knock still imposes a significant constraint at this speed and load but the
propensity for the engine to experience pre-ignition is greatly reduced (due to an
absence of LSPI). The reason why 5000rpm is being treated as the maximum speed is
to ensure consistency with the downspeeding concept. This means that although the
engine is seldom run at this speed there is a great need for high peak load at this point
because the headline power output of an engine is very important to consumers and
vital to the commercial success of an engine.
At this speed and load the 3 cylinder engine would be in the minimum valve
overlap position due to the fact the EBP is much greater than the MAP. The 3 cylinder
engine also runs with the maximum degree of fuel enrichment at this speed and load,
with a lambda of 0.85. As a consequence of this, this speed and load point has the
highest BSFC of the entire operating envelope (with the exception of the very low load
points) and this speed and load point therefore has the greatest scope for
improvement, another reason why a speed of 5000rpm was selected.
5.3 ISFC Optimisation
5.3.1 Introduction
This section contains an analysis of the results from the DoE test plan described
in section 5.2. The objective of this section is to focus on the ISFC benefit that can be
obtained through the EOC pressure and temperature reducing qualities of the Deep
Miller process. The initial portion of this section contains a very detailed analysis of the
ISFC trends with varying EBP and geometric CR. This portion highlights particular
nuances seen in the data, and highlights several different phenomena that have
affected the outcome of this test work. This has been broken down into specific speed
118
and load sections. Following this is an analysis of the cam timings required for the
respective cams.
The second portion of this section is an overview of these phenomena and a

more general overview of the impact they have had on the test work as a whole. This
includes the impact of scavenging, engine breathing and geometric CR considerations.
5.3.2 ISFC Trends
5.3.2.1 1000rpm, 14 bar BMEP
5.3.2.1.1 EBP Effects
In this analysis the EBP of 1.5 and 2 bar are of much more importance than 2.5
bar because this is the most representative level of EBP that is likely to be required
with a typical turbocharger at this speed and load point (based on baseline data
obtained by Mahle Powertrain for the 3 cylinder engine). Results with 3 bar EBP have
been excluded from this analysis because of data integrity issues. All the data given in
this section is for a CR of 9, CR effects are discussed in section 5.3.2.1.2.
Figure 5.4 shows an overlay of the optimum IMOPs for ISFC for each of the
different cams and EBPs. The optimum IMOP for ISFC was determined from IMOP cam
timing sweeps across the full extent of the cam phasing range, and then selecting the
IMOP timing that gave the lowest ISFC.
From a purely theoretical perspective the optimum cam timing and the
optimum cam duration will depend heavily on the EBP and the difference in MAP and
EBP. For a low EBP extreme EIVC and LIVC are preferable as the more extreme the IVC
offset from BDC, the higher the required MAP. The combination of low EBP and high
MAP is preferable because this will invert the pumping loop so it contributes work to
the cycle, and the low EBP will maximise the energy that can be extracted from the
119
gross work contribution. However, if the MAP is elevated too high over the EBP too
much scavenging will take place during the valve overlap phase which will cause the
fuel efficiency to fall away fast (see section 5.3.4).
Figure 5.4 Impact of EBP on ISFC at a Geometric CR of 9
The influence of EBP can be seen to be greater for some cams than for others,
with the 2 shortest cams showing the greatest sensitivity to EBP. The 276 CAD duration
cam shows the least sensitivity to EBP and an apparent point of inflection, as well as
the transposition of the 1.5 bar and 2.5 bar EBP lines. The 312 CAD duration cam also
exhibits relatively little sensitivity to EBP but all points can be seen to be noticeably
higher in ISFC.
The decline in ISFC with the 312 CAD duration cam can be attributed to the fact
that the valve is still open so late into the compression stroke. This is advantageous in
that it allows the maximum possible over-expansion and EOC temperature reduction
benefit, but leaving the valve open during the compression stroke allows charge to
escape the cylinder into the intake manifold. It is no longer just air as fuel will be
diluted into it at that point due to the fact the SOI is at 246 CAD BTDC Firing (BTDCF),
and fuel as well as air will be exhaled from the cylinder into the intake manifold.
120
Although the fuel is effectively still in the intake manifold and has the potential
to be burned with no great impact on the ISFC, a lot of this fuel is lost during the valve
overlap period as it is scavenged straight through the cylinder into the exhaust. This
happens regardless of IMOP timing, and, to an extent, regardless of EBP to MAP ratio
because there is always a small amount of valve overlap with the exhaust valve in its
maximum overlap position (which was fixed due to the fact it gives the maximum
expansion ratio and optimum ISFCs in this position) and still a certain degree of
scavenging. This charge scavenging effect was witnessed using the animation feature
in GT-Power, stills from this animation can be found in Appendix A.4.
This effect is similar in appearance to the excess scavenging experienced with

the short cam profiles, with both scenarios ultimately culminating in unburned fuel
being present in the exhaust. However, with charge scavenging there is the potential
to use the unburned fuel in the exhaust as it will be mixed with air (not necessarily
stoichiometrically) and can therefore be burned providing more energy to the
turbocharger turbine, this may be required for high MAPs. It will also likely cool the
cylinder down more than just air during the scavenging process.
It is due to charge scavenging that the EBP of 1.5 bar and 2.5 bar lines have
become inverted at the 312 CAD duration cam position. An EBP of 2.5 bar will
obviously impede air flow from the intake and inhibit charge scavenging more than an
EBP of 1.5 bar. The 2.5 bar EBP in this case is keeping more of the charge in the
cylinder at the expense of raised EOC pressure and temperature.
The disadvantage of raised EOC pressure and temperature can be seen to be

outweighed by the advantage of reduced charge scavenging at this speed and load
condition, which leads to a reduced ISFC for the higher EBP condition.
The charge scavenging effect can be seen to occur with the 276 CAD duration
cam as well, however, due to the fact IVC occurs earlier in the compression process
and the fact the MAP does not have to be as high as with the 312 CAD duration cam,
very little charge completely short circuits the cylinder and goes into the exhaust
unburned. It is for this reason that the ISFC values for the 276 CAD duration cam can
be seen to be the lowest, because the effective CR is at a very low value while the cycle
121
is not hindered by charge scavenging or fresh air scavenging issues like the 312 CAD
duration cam and the 152 CAD duration cam.
For the 152, 196 and 240 CAD duration cams the 2.5 bar EBP line can be seen to
diverge quite sharply from the 1.5 and 2 bar EBP lines. This is due to the fact that the
EOC pressures and temperatures are elevated for the high EBP cases (which require
the combustion phasing to be retarded away from MBT). The combustion phasing has
to retarded from MBT in every case for the 2.5 bar EBP points for all cam durations
(relative to the 1.5 and 2 bar EBP points). This has not manifested itself for the 276
CAD and 312 CAD duration cams in Figure 5.4 because the retarded combustion
phasing is compensated by the lack of charge scavenging which has led to points for all
3 EBPs converging together and almost overlaying.
The difference in EOC pressure and temperature between each of the EBP
cases is much greater for cams running at a relatively high effective CR, and the
combustion phasing has to be retarded to a greater extent at these points as a
consequence. Retarding the combustion phasing will result in a lower peak combustion
temperature which can be found to be detrimental to ISFC (as dictated by the second
law of thermodynamics).
This is the reason why the ISFC profile shapes can be seen to show such a close
resemblance to the effective CR profile for each of the cams, as can be seen in Figure
5.5. The 312 CAD duration cam data has been omitted from Figure 5.5 because charge
scavenging has elevated ISFC and distorted the trend.
122
Figure 5.5 Comparison of Effective CR and ISFC for Each EBP at a Geometric CR of 9
5.3.2.1.2 Geometric CR Effects
One of the key advantages of using cam phasing and different duration cams is
it affords the ability to operate the engine at different effective CRs in real time. This
allows an increase of the geometric CR which in turn yields a theoretical benefit in fuel
economy. With this test work the engine will very seldom be run with the effective CR
equal to the geometric CR, it is not actually possible to run with the full geometric CR
with the 152, 276 and 312 CAD duration cams.
Although theoretically a higher geometric CR should benefit fuel economy, at

this speed and load it can actually be found to be detrimental to fuel economy because
a high CR (whether effective or geometric) will result in higher EOC pressures and
temperatures, this will require the combustion phasing to be retarded which will
further exacerbate the high EOC pressure and temperature problem in a vicious circle.
This is why typically turbocharged and downsized engines use a relatively low
123
geometric CR. Lowering the effective CR may circumvent this problem and allow the
use of higher geometric CRs.
Owing to data integrity issues no reliable 312 CAD duration cam data at
geometric CRs of 10.75 and 12.5 could be obtained and this cam has therefore been
omitted from this analysis. The reason for these data integrity issues was the elevated
EOC pressure and temperatures encountered with higher geometric CRs. This required
the combustion to be extraordinarily retarded to satisfy the knock model. The knock
model becomes unstable in these scenarios and ceases to function. Attempts were
made to re-acquire the data with the manual single cylinder GT model (model 2) but
similar issues were encountered as with model 1 in that no acceptable MAP, IMOP and
combustion phasing could be found that gave an acceptable KI value. It should be
stressed that the extent to which combustion phasing had to be retarded was beyond
the functional limit of the experimental engine, and any data that would have been
obtained would have been of no practicable use and purely hypothetical anyway (see
section 5.3.6).
The impact of geometric CR increase with constant EBP can be seen in Figures
5.6 to 5.8. These Figures reveal the optimum geometric CR and EBP combinations for
each of the different cam profiles. With 2 exceptions (which will be addressed later)
the lines peak at roughly the same cam duration for each geometric CR case. The peak
ISFC can be also be seen to correspond to the peak effective CR very closely, revealing
the correct trend of the higher the effective CR the lower the cycle efficiency.
124
300
290
280
ISFC (g/kWhr)
270 CR 9 EBP 1.5

bar
CR 10.75 EBP
260 1.5 bar
CR 12.5 EBP
1.5 bar
250
240
230
150 160 170 180 190 200 210 220 230 240 250 260 270 280
Intake cam duration (CAD)
Figure 5.6 Comparison of Different Geometric CRs with an EBP of 1.5 bar
300
290
280
ISFC (g/kWhr)
270 CR 9 EBP 2
bar
CR 10.75
260 EBP 2 bar
CR 12.5 EBP
2 bar
250
240
230
150 160 170 180 190 200 210 220 230 240 250 260 270 280
Figure 5.7 Comparison of Different Geometric CRs with an EBP of 2 bar
125
300
290
280
ISFC (g/kWhr)
270 CR 9 EBP 2.5

bar
CR 10.75
260 EBP 2.5 bar
CR 12.5 EBP
2.5 bar
250
240
230
150 160 170 180 190 200 210 220 230 240 250 260 270 280
Figure 5.8 Comparison of Different Geometric CRs with an EBP of 2.5 bar
The geometric CR lines can also be observed to diverge the higher the effective
CR. For the low effective CR cases, they can be seen to converge or in some cases
change positions. The effectiveness of geometric CR increase is almost entirely
dependent on EOC pressure and temperature at this running point, which is partly
reflected by the fact the red geometric CR of 10.75 line is usually between the
geometric CR of 9 and geometric CR of 12.5 lines, and why when going from a cam
duration with a high ISFC to a cam duration with a low ISFC the geometric CR of 10.75
can always be seen to intersect the geometric CR of 9 line first.
These results show that a high geometric CR is very rarely optimum for a cam
of any duration unless it is either extremely short or extremely long. The reason why
extremely short and extremely long cam durations are required for high geometric CR
is because they can reduce the effective CR to such a point that the combustion can be
phased close enough to the Minimum spark advance for Best Torque (MBT) to give an
advantage. There is some sensitivity to EBP with some strong evidence that higher
geometric CRs can be employed with lower EBPs.
126
Observing the geometric CR lines in Figures 5.6 to 5.8 reveals that with low
EBPs the benefits of extreme EIVC and LIVC are limited and minimal (the lines exhibit a
very shallow gradient). The geometric CR of 10.75 and geometric CR of 12.5 lines
suggest that higher geometric CRs are required to obtain considerable benefit from
EIVC and LIVC strategies.
The main reason for the minimal performance benefit is the fact that while this
is a comparatively high load for 1000rpm, it is overall just a medium load site. The MAP
level is relatively modest and for some points is not much above NA. As a result of this
EOC pressures and temperatures are moderately low and although EIVC and LIVC can
reduce these, the effect is not as pronounced as if they were higher. The variation in
ISFC for the 1.5 bar EBP and geometric CR of 9 case is the lowest with a variation of
just 3.3 g/kWhr across the entire range of cam durations. This variation changes
dramatically with increasing EBP. The variation in ISFC for 2 bar EBP and geometric CR
of 9 case is 9.2 g/kWhr and 27.8 g/kWhr with the 2.5 bar EBP geometric CR of 9 case.
The variations for the geometric CR of 10.75 and the geometric CR of 12.5 cases
can be seen to be a lot higher than these due to the increased EOC pressure and
temperature variations associated with these higher geometric CR cases.
For the most part the peaks in ISFC can be seen to correspond very well with
the peak effective CR. There are however 2 notable exceptions to that, both shown in
Figure 5.8 which shows the peak ISFC values to occur at the second highest effective
CR point for the 10.75 and 12.5 geometric CR cases. This offset appears to be due in
part to the pressure/rarefaction wave caused by the blowdown pulse in the exhaust
interacting with the intake manifold breathing (see section 5.3.5 for more details).
Although a high geometric CR offers an ISFC benefit, the combustion phasing is

in general much more retarded for high geometric CRs. This may be feasible for an
engine simulation but for the experimental engine this could lead to combustion
instability and a very high COV (as explained in section 5.3.6).
127
5.3.2.2 2000rpm, 24 bar BMEP
An ISFC comparison at a geometric CR of 9 and EBPs of 2, 3 and 4 bar can be

seen in Figure 5.9 (unfortunately no data is available for the 312 CAD duration cam at
an EBP of 4 bar due to data integrity issues). The profiles of the lines are broadly
similar but can be seen to change slightly (particularly between the 196 CAD duration
cam and the 240 CAD duration cam).
Figure 5.9 Impact of EBP on ISFC at a Geometric CR of 9
The most likely explanation for this is that each cam had a different optimum
cam timing and a different optimum effective CR. Figures 5.10, 5.11 and 5.12 show the
IVC, IVO and effective CR for the optimum cam position for each cam duration at a
geometric CR of 9 and an EBP of 2, 3 and 4 bar respectively.
128
Figure 5.10 Optimum ISFC Cam IMOPs for an EBP of 2 bar at a Geometric CR of 9
Figure 5.11 Optimum ISFC cam IMOPs for an EBP of 3 bar at a Geometric CR of 9
129
Figure 5.12 Optimum ISFC cam IMOPs for an EBP of 4 bar at a Geometric CR of 9
There is a clear trend for each of these points, they all favour a very small
degree of valve overlap (with the exception of the 196 CAD duration cam at 4 bar EBP),
the long cam profiles in particular. Figure 5.13 reveals where the optimum IVO opening
is relative to the IVO limit for this test work. In all cases the IVC can be seen to be close
to the limit but for durations of 240 CAD and above, the optimum IVC can be seen to
be on the very limit of testing.
130
Figure 5.13 Optimum IVO Cam Timing Points for EBPs of 2 and 3 bar Compared to the IVO Limit for a
Geometric CR of 9
The IVO time can be delayed until much later with the short cams owing to the
fact that even with an IVO of nearly 60 CAD After TDC (ATDC) the IVC point still occurs
at just 3 CAD After BDC (ABDC). With the longer cams however, even with a relatively
conservative IVO of 20 CAD ATDC, IVC occurs at 100 CAD ABDC which requires a very
high MAP. Going beyond 20 CAD ATDC IVO with the longest 312 CAD duration cam by
just 10 CAD will require an MAP of 4.84 bar which logistically is not feasible with
current technology and falls out of the scope of this project. The 276 CAD duration cam
has similar issues although the IVO time can be delayed slightly beyond the 312 CAD
duration cam but only by 6 CAD. The 240 CAD duration cam IVO time could be delayed
until later but was not for this test work as the intention was to use the standard cam
for baseline comparison purposes only. Further work has revealed that the ISFC can be
improved by 2 g/kWhr if the 240 CAD duration cam IVO time is delayed by a further 10
CAD, if it is delayed any more than this ISFC begins to increase. The reason for this
increase in ISFC beyond a certain IVO delay time can be attributed to the increase of
residuals in the cylinder.
131
All of this test work was carried out with the exhaust valve in the maximum
valve overlap position and the Exhaust Valve Closing (EVC) point fixed at 16 CAD ATDC.
Therefore between TDC and the EVC point the piston will begin to draw hot exhaust
gasses back into the cylinder. If there is no valve overlap period at all these gasses will
remain in the cylinder for the next cycle. Cool residual gasses have the effect of
reducing the propensity to knock, however, in this scenario they are hot and if they are
present in the cylinder in sufficient quantities they will considerably elevate the charge
temperature above the manifold temperature of 400C. This in turn will lead to elevated
EOC pressure and temperature and potentially an increased propensity to knock.
This phenomenon is particularly present for the 196 CAD duration cam where a
reduction in effective CR can be seen to result in an increase in EOC pressures and
temperatures (Figures 5.14 and 5.15 respectively). This can be attributed to the
breathing dynamics of the engine which have been explained in section 5.3.5.
Figure 5.14 Log P-V IVC Sweep for 196 CAD Duration Intake Cam at a Geometric CR of 9 and an EBP of
2 bar
132
400
In-cylinder temperature (deg C)
-2.47 CAD ATDC IVC

7.53 CAD ATDC IVC
17.53 CAD ATDC IVC
27.53 CAD ATDC IVC
37.53 CAD ATDC IVC
40
0 0.2 0.4 0.6 0.8 1 1.2
Volume ratio
Figure 5.15 Log Temperature Vs. Volume Ratio IVC Sweep for 196 CAD Duration Intake Cam at a
Geometric CR of 9 and an EBP of 2 bar
The 4 bar EBP points are of limited relevance for this speed and load condition
because for most of the optimum IMOP timings the MAP was not so high as to warrant
an EBP of 4 bar, unless a very poorly specified turbocharger was employed. The EBP
points of 2 and 3 bar were of greater relevance, however, the 152 CAD and 312 CAD
duration cams required very high MAPs, typically in excess of 3 bar, and the 4 bar EBP
points may be relevant to them.
The integrity of the data at 4 bar EBP is noticeably worse than at 2 and 3 bar,
particularly for points with a very late IVC. The EOC pressures and temperatures and
quantity of residuals in the cylinder were sufficiently high to cause the knock model to
malfunction and predict what is effectively (although not actually because the knock
model cannot predict pre-ignition) very severe pre-ignition. Even with IVCs that
allowed a high degree of valve overlap the cylinder would not scavenge because the
EBP was almost always elevated above the MAP. This also resulted in malfunctioning
of the knock model in some instances.
133
A comparison of the optimum IMOP timings for each of the cams at EBPs of 2, 3
and 4 bar can be found in Figure 5.13. The optimum IMOP timings can be seen to be
much more variable at 4 bar EBP than for the 2 and 3 bar EBP cases due to the fact
that scavenging no longer occurs during valve overlap periods. The biggest difference
between the 2 and 3 bar EBP cases and the 4 bar EBP case is with the 196 CAD
duration cam. The optimum IMOP has shifted from a very little valve overlap IMOP
timing to the maximum valve overlap IMOP timing.
The reason for this shift is the engine breathing dynamics and the presence of
pressure/rarefaction waves in the exhaust system (see section 5.3.5).
The EBP can be seen to have a different impact on the engine depending on
cam duration. Figure 5.9 shows the sensitivity with the 152 CAD duration cam to be
very low at 4 g/kWhr variation between the 3 different EBPs, whereas the variation
with the 240 CAD duration cam is very high at 13 g/kWhr. An insensitivity to EBP is
obviously desirable because increasing EBP is always detrimental to engine efficiency
due to the inability to scavenge and inability to the keep the cylinder cool.
The burn duration can be expected to increase in duration with rising EBPs
because of the greater quantities of residuals in the cylinder at cycle start, this makes
the ISFC values given in Figure 5.9 slightly optimistic for the 3 and 4 bar EBP cases in
comparison to the 2 bar EBP case.
Figures 5.16, 5.17 and 5.18 show overlays of the optimum ISFC points for each
of the different cam durations for 3 different geometric CRs of 9, 10.75 and 12.5
respectively. Due to data integrity issues Figure 5.18 had to be truncated.
134
135
The data reveals that the long cams show the greatest potential for running at
high geometric CRs. At this load point, where the engine is particularly susceptible to
knock, a higher geometric CR is actually beneficial to engine running. The very short
cam with a high geometric CR can be beneficial to ISFC too, but the results show a very
high sensitivity to EBP at this condition. The 196 CAD and 240 CAD duration cams
generally have a detrimental effect on ISFC with increasing geometric CR.
Generally speaking the cam profiles that offer the lowest effective CRs show
the greatest synergy with geometric CR increase. This is to be expected because
increasing geometric CR has the effect of undoing any effective CR reduction that has
already been applied. If the cam duration is such that not much effective CR reduction
can be applied, a geometric CR increase of 1.75 will increase the effective CR beyond
the previous geometric CR. With very long or very short cam durations this does not
occur, and even with an increase in geometric CR of 3.5, the effective CR can still be
seen to be far below the initial geometric CR.
The engine breathing at geometric CRs of 10.75 and 12.5 is very similar to the
engine breathing behaviour at a geometric CR of 9. The optimum load points occur at
136
very similar cam timings as the geometric CR of 9 line, however, there is a slight offset
in some cases. For low EBPs the optimum cam timing can be observed to be offset
slightly toward the minimum valve overlap point. The reason for this is a high
geometric CR results in a reduced clearance volume. The rate of scavenging can be
seen to be near enough constant for different geometric CRs therefore a low clearance
volume will be scavenged more rapidly than a larger clearance volume and more fresh
air will be scavenged into the exhaust. This results in more accidental fuel enrichment
which is detrimental to ISFC. For large degrees of valve overlap this makes ISFC
markedly worse for high geometric CRs compared to low geometric CRs.
In Figure 5.16 the geometric CR of 10.75 line can be seen to overlay the
geometric CR of 9 line very closely. With rising EBP the 2 lines can be seen to diverge
slightly, particularly for the 196 CAD and 240 CAD duration cams. This can be
attributed in part to the extended pumping loop afforded by the increase in geometric
CR.
5.3.2.3 5000rpm, 24 bar BMEP
Figure 5.19 shows an EBP sweep at a geometric CR of 9. The profile of the lines
can be seen to be drastically different to the lower speed and load profiles (Figures 5.4
and 5.9). This is attributed almost wholly to the requirement for enrichment to reduce
exhaust gas temperatures.
137
Figure 5.19 Comparison of 2, 3 and 4 bar EBPs at a Geometric CR of 9
Almost the exact opposite trend can be seen compared to the lower speeds
and loads where enrichment was not necessary. The short cams are required for
optimum ISFC for all EBP cases at this speed and load. At the lower speed points
excessive scavenging was identified as the cause for high ISFCs. This could have
potentially been offset by mapping the required lambda at the speed and load
condition but was not, and therefore represented one of the worse cam durations for
these speeds.
Running at this high a speed gives the engine less time to induct and exhaust
gasses, and as a consequence of this the MAP was typically a lot higher at this speed
than at the equivalent point at 2000rpm. Several points were excluded from this
analysis in spite of the fact they were the optimum for that particular cam duration,
geometric CR and EBP combination. The reason for this is that the engine would fail to
achieve the required load of 24 bar BMEP even with the MAP set to the maximum of 4
bar. The points most afflicted by this problem were the points with a very low effective
CR, where typically a lot of boost would be required. This has had a notable impact on
some trend lines obtained and these points will be identified throughout the course of
this section.
138
The breathing dynamics of the engine are very similar to the 1000rpm and
2000rpm cases in that the MAP to EBP ratio and valve overlap timings still dictate how
much scavenging or backflow takes place. At 5000rpm there is far less time to
scavenge the cylinder or for backflow to propagate into the intake manifold, so
although scavenging still takes place there is not as much as there would be at 1000
and 2000rpm. For the 152 CAD duration cam scavenging can still be seen to take place,
this results in accidental over enrichment of cylinder charge, however, at this speed
fuel enrichment would be required to regulate the exhaust temperature for some EBPs
anyway.
The degree of accidental over enrichment at a common running point of

geometric CR of 9, IMOP of 50 CAD ATDC and an EBP of 2 bar is 23.32% at 2000rpm as
opposed to just 6.81% at 5000rpm. With an over enrichment percentage this low the
152 CAD duration cam begins to provide an ISFC similar in magnitude to that predicted
by the use of lambda mapping at lower speeds (see section 5.3.4), and represents a
much more viable cam duration choice at this speed.
Another major advantage offered by the 152 CAD duration cam is that
scavenging air through the cylinder will have the effect of diluting the exhaust gasses,
thus reducing the exhaust gas temperature. This phenomena as well as the fact the
effective CR ratio is very low at this IMOP with this cam results in an exhaust gas
temperature of 787.80C, well below the maximum limit of 9300C. Accidental over
enrichment is partly responsible for such a low exhaust gas temperature. Removing
the accidental over enrichment results in an exhaust gas temperature of 827.670C
which is still well below the maximum temperature limit thus rendering any quantity of
enrichment unnecessary, however, a lower exhaust gas temperature does have a small
impact on the EOC pressure and temperature for the subsequent cycle. Over
enrichment also has the effect of lowering the polytropic exponent which serves to
reduce EOC pressure and temperature through charge cooling. This will influence the
combustion phasing and permits it to be moved closer to MBT for the accidental over
enrichment case. It is for this reason that a 6.81% reduction in fuel flow rate yields a
benefit in ISFC of 6.62% and not 6.81%, and the combustion must be retarded 0.266
CAD further away from MBT for the compensated case.
139
The 152 CAD duration cam exhibits the least sensitivity to EBP at this speed and
load condition. All of the EBP conditions can be seen to benefit from the reduced
scavenging, and for the 3 and 4 bar EBP cases the optimum cam timing is the
maximum valve overlap condition. The optimum cam timing for 4 bar EBP is also the
maximum valve overlap condition, however, this point was excluded from this analysis
because even with an MAP of 4 bar the engine could not reach a BMEP of 24 bar,
therefore the IMOP with the second largest degree of valve overlap, and the second
lowest ISFC, was selected instead.
As with the other cases the general profile of the lines in Figure 5.19 can be
seen to mimic the effective CR quite closely. The degree of divergence in each of the
lines can also be seen to grow as effective CR is increased too, reflecting the increasing
EOC pressures and temperatures associated with higher EBPs and higher effective CRs.
The peak ISFC does not occur at peak effective CR though, it is instead offset slightly
toward the longer duration cams, occurring at 240 CAD duration cam for the 4 bar EBP
case and 276 CAD duration cam for the 2 and 3 bar EBP cases.
Analysis of the 276 CAD duration cam data has revealed that the optimum
IMOP timing for an EBP of 2 bar is the maximum valve overlap position, whereas it is
the minimum valve overlap position for 3 and 4 bar EBP cases. The reason for this is
the pressure difference between MAP and EBP. For the 2 bar EBP case the MAP is
greater than the EBP hence scavenging occurs during the valve overlap phase, for the 3
and 4 bar cases EBP exceeds MAP leading to back flow through to the intake.
Maximum valve overlap is optimum for the 276 CAD duration cam with 2 bar
EBP case, however, the difference in ISFC across the entire IMOP sweep can be seen to
be very small in comparison to most of the other IMOP sweeps for other
speed/load/EBP/geometric CR combinations, showing a variation of just 7.84 g/kWhr
in ISFC across the entire sweep. This shows that the balance between scavenging and
effective CR reduction is very fine at this condition. The maximum valve overlap point
corresponds to the highest effective CR point for the 276 CAD duration cam whereas
the minimum valve overlap corresponds to the lowest effective CR point. Enrichment is
required at every point in the IMOP sweep to regulate exhaust gas temperature. Less
140
enrichment is required at the maximum valve overlap point due to the fresh air
scavenging that is taking place whereas more is required at the minimum valve overlap
point due to the presence of hot exhaust gasses in the cylinder during compression.
Very small amounts of charge can be seen to propagate into the intake manifold at the
minimum valve overlap point, but there is no evidence of charge scavenging.
Figure 5.20 shows the lambda, effective CR and ISFC for the geometric CR of 9,
EBP 2 bar IMOP sweep. The lambda line can be seen to resemble a V shape, the first
peak at an IMOP of 100 CAD ATDC (maximum valve overlap) is a result of fresh air
scavenging diluting the exhaust gasses and resulting in a reduction of exhaust gas
temperature. The first peak is the highest, however, the value given is exhaust lambda
which, as discussed before, is slightly lean of the cylinder lambda due to accidental
over enrichment. The second peak occurring at an IMOP of 140 CAD ATDC is as a result
of the very low effective CR reducing the EOC pressure and temperature.
Figure 5.20 Effective CR and Lambda Plots at a Geometric CR of 9, EBP of 2 bar and a Cam Duration of
276 CAD
For this cam duration this is the key determinant in what the optimum cam
timing is, and it is for this reason that the optimum cam timing shifts directly from a
141
maximum valve overlap position to a minimum valve overlap position with increasing
EBP (when fresh air scavenging is no longer possible).
Unlike the 1000rpm and 2000rpm points, at 5000rpm and 24 bar BMEP on a
turbocharged engine the EBP is almost inevitably above the MAP and 3 bar. Therefore
running in the maximum valve overlap position is not going to be a possibility and
effective CR reduction is the only means of yielding an ISFC improvement. Something
else that must be taken into consideration is that the air and fuel that has been
scavenged could react when in the exhaust system thus raising exhaust gas
temperature considerably. GT-Power is not able to model secondary combustion
outside the cylinder and this effect has therefore not been factored in.
At higher EBPs the lambda line can still be seen to resemble a V shape as shown
in Figure 5.20. The peak in lambda at the maximum valve overlap position is attributed
to accidental over enrichment giving the impression the engine is running leaner than
it actually is. The minimum valve overlap point (IMOP of 140 CAD ATDC) can be
observed to go from the optimum IMOP timing to the worst IMOP timing with
increasing EBP. The transition between optimum and worst is moderately slow
occurring over the course of a change of EBP of 1.3 bar, however, the transition from
worst to best for the minimum valve overlap point (IMOP of 100 CAD ATDC) is very
fast, occurring over the course of a change of EBP of just 0.2 bar.
This same trend can also be observed for the 312 CAD duration cam profile. In
spite of the fact the IVC point does not occur until well into the compression stroke for
all IMOP cases, very little charge scavenging can be seen to take place in comparison to
the equivalent 2000rpm point. Although due to the fact over enrichment is used
charge scavenging cannot be easily observed in the GT-Power data owing to the fact
the vapour fuel fraction in the exhaust is not zero or almost zero even during normal
running. As with the 276 CAD duration cam, the ISFC figures suggest that there is not a
considerable amount of charge scavenging otherwise it would not be the optimum
IMOP timing for ISFC.
Although the highest effective CR occurs at a cam duration of 240 CAD, the ISFC
peaks there for just the 4 bar EBP case. The reason why it does not peak there for the 2
142
and 3 bar EBP cases is because of the IMOP timing. For both the 240 and 276 CAD
duration cams at 2 bar EBP the optimum IMOP timing for ISFC is the maximum valve
overlap point. The extent of the maximum valve overlap differs for each cam however.
The IVO angle for the 240 CAD duration case is 21 CAD BTDC gas exchange, for the 276
CAD duration cam it is 16.15 CAD BTDC gas exchange. This allows a considerable
amount more scavenging to take place for the 240 CAD duration cam in spite of the
fact the MAP is lower. This has the effect of reducing the exhaust gas temperature. For
the 240 CAD duration cam this lowers the temperature enough to not warrant the use
of enrichment. The reduced exhaust gas temperature has a lower speed of sound
meaning the pressure/rarefaction wave caused by the EBP valve is slightly later in the
cycle, the significance of this is covered in section 5.3.5.
Figures 5.21, 5.22 and 5.23 show optimum ISFC points for each of the different
geometric CRs at EBPs of 2, 3 and 4 bar respectively.
360
340
320
300
ISFC (g/kWhr)
CR 9 EBP 2
280 bar
CR 10.75
EBP 2 bar
260
CR 12.5 EBP
2 bar
240
220
200
140 160 180 200 220 240 260 280 300 320
143
360
340
320
300
ISFC (g/kWhr)
CR 9 EBP 3
280 bar
CR 10.75
EBP 3 bar
260
CR 12.5 EBP
3 bar
240
220
200
140 160 180 200 220 240 260 280 300 320
360
340
320
300
ISFC (g/kWhr)
CR 9 EBP 4
280 bar
CR 10.75
EBP 4 bar
260
CR 12.5 EBP
4 bar
240
220
200
140 160 180 200 220 240 260 280 300 320
In Figures 5.21 to 5.23 all the geometric CR lines can be seen to show the same
trend and all peak at the same positions. Of the 3 EBPs only the 4 bar EBP case can be
seen to follow the effective CR trend where the peak ISFC occurs at the peak effective
144
CR. The reason for the offset for the 2 and 3 bar EBP cases can be attributed to
scavenging. The optimum cam IMOP can also be seen to change between the
maximum valve overlap point for the lower EBP cases to the minimum valve overlap
point for the 4 bar EBP case. As the difference in MAP and EBP is reduced the engine
breathing will be altered accordingly as described in section 5.3.2.3.1.
Figures 5.21 to 5.23 also reveal significantly less sensitivity to EBP for the 152
CAD duration cam than for any of the other cam durations for all geometric CRs. This
can be attributed in part to the reduced EOC pressures and temperatures which
greatly increases the tolerance to EBP, and also to the fact higher EBPs inhibit
scavenging and accidental over enrichment. The benefit offered by higher geometric
CRs can be seen to become less and less with increasing EBP due to the fact the EOC
pressure and temperature is elevated by higher EBPs.
With the exception of the 2 bar EBP case the 312 CAD duration cam can be
seen to break the trend of decreasing ISFC with decreasing effective CR. The reason for
this is the choice of IMOP points is simply very restricted for this long cam because of
the imposed 4 bar MAP limit. Therefore the chosen IMOP in this case does not
necessarily mean it is the optimum, just that it is the only point that achieved the load
with an MAP of 4 bar. There was just 1 IMOP timing that resulted in adequate load
being achieved for the geometric CR of 12.5 and 4 bar EBP point, and that was the
maximum valve overlap position because of the high effective CR it affords. Although
this point has been displayed in Figure 5.23 as being valid, the exhaust gas
temperature (even with the full degree of enrichment) is well above the temperature
ceiling at 1013.60C which actually renders it an invalid point too. The same applies to
the 4 bar EBP and geometric CR of 10.75 case too, which is MAP restricted for the 312
CAD duration cam. The point shown in Figure 5.23 for this is also the maximum valve
overlap point and it is also not strictly valid because the exhaust gas temperature
exceeds the temperature limit. In both of the above cases the minimum valve overlap
point would be optimum if a higher MAP limit was instated (because of the reduced
effective CR).
145
The geometric CR of 9 point is also dictated by the MAP restriction, however,
the IMOP point that gave acceptable load at this condition was the one that gave the
second lowest degree of valve overlap, which is much closer to the optimum cam
timing. The reason why adequate load could be achieved with the lower geometric CR
is because the EOC pressure and temperature is much lower which allows the
combustion phasing to be advanced slightly more than the geometric CR of 10.75 and
12.5 cases, the cycle efficiency is improved by this and the MAP therefore does not
have to be as high. The exhaust gas temperature will also be reduced by the increased
degree of combustion phasing advance.
The MAP limit has also manifested itself in Figure 5.22 (3 bar EBP) for the 312
CAD duration cam and has resulted in a point being selected that is not the optimum,
hence the reason why it does not follow the reduced ISFC with reduced effective CR
trend as closely as the 2 and 3 bar EBP lines.
Although not as visible as for the 312 CAD duration cam, the 152 CAD duration
cam is also restricted in places by the MAP limit. As with the 312 CAD duration cam the
MAP limit is only a problem at the higher EBPs where scavenging is impeded. The
IMOP must be shifted to a lower valve overlap point when maximum valve overlap is
optimum for this cam. Shifting the IMOP to a lower valve overlap point increases the
effective CR which increases ISFC. The effect is far less marked though due to the fact
less valve overlap results in less scavenging and less accidental over enrichment, this
helps to offset the increase in EOC pressure and temperature.
5.3.3 Cam Timing Trends
Section 5.3.2 highlighted the impact of scavenging and various other

phenomena on the ISFC performance of the single cylinder engine model. The purpose
of this section is to highlight the impact of these phenomena on the optimum cam
timing position for ISFC, and how far away this is from the theoretical optimum
position for the cam in terms of effective CR reduction.
146
Figure 5.24 gives a general indication of the theoretical percentage gain in
efficiency that can be obtained through using the theoretically optimum cam timing
compared to what was found to be the optimum cam timing in the test work. This is
shown for each cam. The frequency of the use of a theoretically non-optimal cam
timing is shown in Figure 5.25.
6
Theoretical gain in efficiency (%)
5
1000rpm, 14 bar
BMEP
4
2000rpm, 24 bar
BMEP
3
5000rpm, 24 bar
2 BMEP
0
142 162 182 202 222 242 262 282 302 322
Figure 5.24 The Theoretical Gain in Thermodynamic Efficiency Through the Adoption of Optimum Cam
Positioning
147
Figure 5.25 Number of Points where a Theoretically Non-optimal Cam Timing was found to be Optimal
in the Test Work
Although the number of theoretically non-optimal cam timings with the 152
CAD duration cam is far greater than for any other cam, the theoretical loss of
efficiency is not as considerable as it is for the 296 and 312 CAD durations cams. This is
due to the fact that a higher degree of effective CR reduction can be adopted with the
276 and 312 CAD duration cams. Due to a hard point limitation with the 152 CAD
duration cam, the effective CR can never be as low as with the longer cams. The impact
of this on theoretical efficiency benefit is shown in Figure 5.26. Please note, the
profiles of the lines are not perfectly parabolic due to the fact slightly more cam
phasing was used with the 276 CAD duration cam.
148
10
Maximum benefit over baseline cam with IVC at BDC (%)

8
gCR of 9
5
gCR of 10.75
4 gCR of 12.5
0
142 162 182 202 222 242 262 282 302 322
Figure 5.26 Maximum Benefit over Baseline Case (IVC at BDC) with each Cam at each Geometric CR
This means that it is not as critical to run at the optimum cam timing with the
152 CAD duration cam as it is with the 312 CAD duration cam.
Figure 5.24 reveals a clear trend with regard to engine speed. Of the 1000rpm
14 bar BMEP points, the ones that are not optimally timed all occur at short cam
durations where scavenging was found to have a detrimental impact on ISFC. The
5000rpm 24 bar points generally did not experience the same issue with scavenging
(due to the fact that the optimal timing is the minimum valve overlap position) so the
cam could be placed in the optimal position, unlike at low speed. It was however
heavily affected for the longer cam durations. The reason for this is the cam timing had
to be reduced because the volumetric efficiency was so low with the cam in the
optimum position that even with an MAP of 4 bar a load of 24 bar BMEP was
unattainable.
The impact of geometric CR increase has very little effect on the cam timing
with the exception of 5000rpm, this is due to the fact that the MAP limit was reached
earlier with higher geometric CRs. This is likely due to the fact that a higher geometric
149
CR results in a lower overall cylinder volume for a constant swept cylinder capacity,
hence a higher charge density is required to compensate for this.
EBP also shows no strong trends. Although elevated EBPs were found to inhibit
scavenging this effect was not significant enough to shift the IMOP a whole 10 CAD
(which was the resolution tested at for this work). It may be possible to observe cam
timing effects at finer IMOP resolutions.
5.3.4 Scavenging Effects
Two different forms of scavenging were observed to take place during this
study, one form is the scavenging of fresh air (called simply “scavenging” for this study)
and the other is the scavenging of charge (called “charge scavenging”). The effects of
these 2 forms of scavenging are particularly marked for the 152 CAD duration cam and
the 312 CAD duration cam. Although both EIVC and LIVC are both very susceptible to
the effects of scavenging, the effects are usually less noticeable with LIVC due to the
fact the optimum cam timing is the minimum valve overlap point. Figure 5.27
demonstrates this.
150
Figure 5.27 Demonstration of Optimum Cam Timings for EIVC and LIVC
5.3.4.1 Scavenging
Although scavenging is generally beneficial to cycle efficiency, it can have the

effect of diluting the exhaust gases with fresh air and causing the exhaust gas to
appear lean. As a consequence of this the EMS will inject more fuel than is necessary
to restore the engine to “stoichiometric” when in actual fact in-cylinder lambda is too
rich (accidental over enrichment).
Scavenging (the blue region of Figure 5.27) was found to be particularly

dependent on speed, load, EBP and geometric CR. Scavenging is particularly marked at
151
low speed and high load where the MAP is usually greater than the EBP, and the valve
overlap period is usually quite long in terms of a period of time (no different when
measured in CAD). This affect has been mapped on the engine operating envelope and
is denoted by the blue region in Figure 5.28, which was obtained from the 3 cylinder
engine at Mahle Powertrain.
Figure 5.28 3 Cylinder Engine Speed and Load Map Showing where Exhaust Lambda Enrichment and
Leaning is Used
This map could not be used at any point in this test work though, not even for
the 240 CAD duration standard cam. This is because the engine breathing is very
different for the single cylinder engine, as well as the MAP and EBP combinations
which, for the 3 cylinder engine, are dictated by the turbocharger and fixed. The huge
degree of variation available for the single cylinder engine, as well as the different cam
durations, made creating maps like that in Figure 5.28 a huge task. This could
theoretically be carried out in GT-Power by incorporating closed loop control of
exhaust lambda based on in-cylinder lambda but the complexity of the model would
be increased significantly and places even more onus on the in-cylinder combustion
modelling for accurate results.
152
Instead the decision was made to run the engine at lambda 1 for all test work
and to compensate for the accidental over enrichment at a later date. The method
employed for compensating for accidental over enrichment was to extract the in-
cylinder lambda from the DoE data and multiply it by the ISFC. This is a very simplistic
approach and fails to take into account the added charge cooling effect of the fuel.
This makes the ISFC after this calculation slightly optimistic.
The accidental over enrichment has had the effect of skewing the results
slightly, and is the reason why the optimum IMOP position for ISFC recorded for the
standard 240 CAD duration cam for speeds of 1000rpm and 2000rpm is not the
maximum valve overlap position as is the case with the 3 cylinder engine. The decision
to not use lambda correction for accidental over enrichment has the biggest impact at
low EBP points with high MAPs and with large degrees of valve overlap. The impact on
the quoted ISFCs for the worst case speed and load point of 1000rpm, 14 bar BMEP in
percentage terms can be seen in Figure 5.29. The tables show the maximum
percentage difference between the compensated and uncompensated values for each
cam, EBP and CR combination. The percentage values refer to how much larger the
ISFC is for the uncompensated value in comparison to the compensated value.
Geometric CR Geometric CR
9 10.75 12.5 9 10.75 12.5
1.5 14.91 21.95 24.74 1.5 4.95 24.63 25.40
EBP (bar) 2 13.67 17.97 24.72 EBP (bar) 2 10.69 18.83 17.38
2.5 8.27 8.77 19.60 2.5 10.99 11.83 10.51
152 CAD duration cam 196 CAD duration cam
9 10.75 12.5 9 10.75 12.5
1.5 5.38 19.32 29.34 1.5 4.54 15.33 11.10
EBP (bar) 2 6.87 16.42 26.41 EBP (bar) 2 10.13 15.27 12.94
2.5 7.64 15.93 24.46 2.5 8.02 10.54 9.80
Geometric CR
9 10.75 12.5
1.5 11.53 16.66 10.43
EBP (bar) 2 2.08 11.83 10.91
2.5 1.46 6.90 6.79
312 CAD duration cam
Figure 5.29 Scavenging Compensation (%) for each of the Different Cams at each EBP and CR for
1000rpm and 14 bar BMEP
153
The cams that are most affected by accidental over enrichment are the shorter
cams. The main reason is the longer duration cams gain efficiency through the
reduction of effective CR, which is generally not achieved with high degrees of valve
overlap. The values are generally larger for the higher CR points because the smaller
clearance volume for higher CRs is quicker to scavenge than for lower CRs, therefore
more fresh air short circuits to the exhaust than with lower CRs. The shorter cam also
requires the highest MAP at the maximum valve overlap point of all of the cams being
tested, which will increase scavenging.
Accidental over enrichment is still present at 2000rpm (Figure 5.30) and is

similar in magnitude to the 1000rpm case, in spite of the fact the cylinder has the half
the amount of time to scavenge at this speed. The reason for this is the load is much
higher at the 2000rpm point and the MAP is much greater as a consequence of this.
9 10.75 12.5 9 10.75 12.5
2 18.20 29.64 30.69 2 12.54 11.87 12.47
EBP (bar) 3 13.47 18.93 23.00 EBP (bar) 3 7.17 7.78 7.35
4 11.05 14.28 24.56 4 2.46 2.43 2.94
9 10.75 12.5 9 10.75 12.5
2 6.12 6.05 8.51 2 21.55 19.38 24.68
EBP (bar) 3 2.89 2.27 7.30 EBP (bar) 3 9.05 10.01 18.26
4 1.22 1.89 5.48 4 2.21 4.47 9.20
Geometric CR
9 10.75 12.5
2 34.91 29.55 32.66
EBP (bar) 3 11.21 10.57 17.85
4 3.31 4.90 8.73
The extent of the scavenging at 5000rpm cannot be easily quantified from this
data due to the fact fuel enrichment was used, which has distorted the in-cylinder
lambda values. The data that could be extracted is shown in Figure 5.31. The only data
shown is that obtained at a requested lambda of 1, or in other words where the
154
exhaust temperature is below the temperature limit without the need for fuel
enrichment. An overlay of the in-cylinder lambdas (Figure 5.32) demonstrates the
impact of speed on scavenging at a point where the ISFC is not influenced by the use of
fuel enrichment to give an idea of the different effect of scavenging at different speeds
and loads.
9 10.75 12.5 9 10.75 12.5
2 1.59 5.72 8.31 2 2.77 2.88 4.25
EBP (bar) 3 1.30 1.55 3.88 EBP (bar) 3 2.81 4.04 N/A
4 0.72 3.12 2.18 4 N/A N/A N/A
9 10.75 12.5 9 10.75 12.5
2 3.07 5.37 5.00 2 N/A N/A N/A
EBP (bar) 3 N/A N/A N/A EBP (bar) 3 N/A N/A N/A
4 N/A N/A N/A 4 N/A N/A N/A
Geometric CR
9 10.75 12.5
2 2.77 5.56 4.62
EBP (bar) 3 N/A N/A N/A
4 N/A N/A N/A
155
Figure 5.32 Comparison of In-cylinder Lambdas for the 152 CAD Duration Cam at 2 bar EBP and a
Geometric CR of 9 for All Speeds
Even at a reduced load of 14 bar BMEP the extent of the scavenging is still
relatively high for the 1000rpm case. In spite of the far greater MAP to EBP ratio, the
scavenging at 5000rpm is barely noticeable, showing a peak difference of just 8.31%. It
is for this reason that the maximum valve overlap condition was the optimum for the
152 CAD duration cam for 5000rpm and not for 1000 and 2000rpm. The optimum
running point for the 152 CAD duration cam theoretically is the maximum valve
overlap point because of the reduced effective CR.
The with and without compensation ISFC curves for each of the different speed
and load points can be seen in Figures 5.33, 5.34 and 5.35 for speeds of 1000rpm,
2000rpm and 5000rpm respectively. All of these figures are for an EBP of 2 bar (1.5 bar
EBP for the 1000rpm case) and a geometric CR of 9.
156
Figure 5.33 1000rpm, 14 bar BMEP, EBP of 1.5 bar, Geometric CR of 9. With and without Scavenge
Compensation
Figure 5.34 2000rpm, 24 bar BMEP, EBP of 2 bar, Geometric CR of 9. With and without Scavenge
Compensation
157
Figure 5.35 5000rpm, 24 bar BMEP, EBP of 2 bar, Geometric CR of 9. With and without Scavenge
Compensation
The actual benefit to the cycle is much less than Figures 5.29 to 5.31 would
imply. The reason for this is Figures 5.29 to 5.31 show the difference between exhaust
lambda and in-cylinder lambda at the IMOP where the maximum difference was
found, not the difference between the IMOP with the highest degree of scavenging
and the optimum from the plot without compensation as can be seen in Figures 5.33
to 5.35. Table 5.4 demonstrates this for the 152 CAD duration cam IMOP sweep.
Table 5.4 1000rpm, 14 bar BMEP, Geometric CR of 9, EBP of 1.5 bar. With and without Scavenge
Compensation Effects on ISFC for an IMOP Sweep with the 152 CAD Duration Cam
IMOP ISFC (w/o compensation) ISFC (with compensation) % Difference between points
50 281.704307 239.6956692 14.91231648
60 265.072522 240.8037164 9.155534273
70 252.091628 241.8045279 4.080698824
80 247.161895 243.1013907 1.642852074
90 246.775231 244.9863621 0.724898081
100 249.81143 248.0040081 0.723514481
110 256.110358 254.1515305 0.764837273
120 264.871743 263.4493316 0.537018947
Figures 5.29 to 5.31 refer to the “% Difference between points” column

whereas Figures 5.33 to 5.35 refer to the difference between the optimum points
158
(highlighted in green). This reveals that scavenging, and compensating for scavenging,
though producing a significant difference at the maximum valve overlap point
ultimately has a relatively minor effect on the ISFC in this case.
This effect is more pronounced at other running points, particularly at high CRs
and EBPs where running with a very low effective CR is crucial to efficiency. This has
the tendency of pushing the optimum IMOP back toward the maximum valve overlap
point, even though scavenging is taking place. This is more a problem for the medium
length cams of 196 and 240 CAD duration where the effective CR is very high
regardless of IMOP. The peak difference in optimum ISFC with and without
compensation occurs at a geometric CR of 12.5, an EBP of 2.5 bar and a cam duration
of 196 CAD. The optimum IMOP in both cases is the maximum valve overlap position,
which equates to a difference in ISFC of 10.22% higher for the without compensation
case in comparison to with compensation.
The 2000rpm case exhibits the greatest sensitivity to scavenging (maximum of

6.3%), with the 1000rpm and 5000rpm cases showing much reduced levels of
sensitivity (maximum of 3.1% and 1.8% respectively). The reason for the greater
sensitivity at 2000rpm is the fact the speed is relatively low and the load is relatively
high.
5.3.4.2 Charge Scavenging
Charge scavenging (the red region of Figure 5.27) occurs only with LIVC. It has
been found to be similar in magnitude to scavenging in some low speed cases, but the
exact amount depends heavily on the ratio of MAP to EBP. It is only a problem with
LIVC due to the fact that LIVC purges a great deal of charge into the intake manifold
during the compression process. If the MAP is quite high and the EBP quite low
scavenging of this purged charge can take place and result in unburned charge being
short circuited straight through the cylinder and into the exhaust.
159
Calculation of the impact of this on ISFC is not possible with GT-Power V6.1. If
the scavenged charge is stoichiometric it will appear invisible to both the in-cylinder
lambda and exhaust lambda feedbacks. It is possible to estimate it using the animation
feature of GT-Power, this is achieved through observation of the vapour fuel fraction in
the exhaust. This technique does not work with 5000rpm though, due to the fact fuel
enrichment is used as a matter of course.
The volume of the exhaust port on the cylinder head and the exhaust manifold
was estimated using the geometry of the GT-Power model and an estimation of the
volume of unburned charge was obtained by observing how far downstream the
vapour fuel fraction propagated through the exhaust system after EVC. It was assumed
that this vapour fuel fraction cloud was a stoichiometric mix of air and fuel, and from
this assumption an estimate of the amount of scavenging can be obtained.
Figure 5.36 shows the effect of EBP on charge scavenging at the 1000rpm and
14 bar BMEP point. The charge scavenging is reduced by 60.3% by running with an EBP
of 2.5 bar compared to 1.5 bar.
Estimated Charge Scavenging at 1000rpm and 14 bar BMEP for 1.5, 2 and 2.5 bar
EBP
25
20
15
Charge
scavenging (%)
10
0
1.5 2 2.5
EBP (bar)
Figure 5.36 Estimated Charge Scavenging at 1000rpm and 14 bar BMEP for 1.5, 2 and 2.5 bar EBP
Figure 5.37 shows the effects of charge scavenging at the higher speed and
load point of 2000rpm and 24 bar BMEP. The degree of charge scavenging has dropped
160
considerably in comparison to the 1000rpm case in spite of the fact the MAP is much
higher. Geometric CR has been found to increase the degree of charge scavenging at
this speed and load, likely due to the fact that the MAP is slightly higher for higher
geometric CRs and because the clearance volume is slightly reduced. In this instance
the degree of charge scavenging has increased by 30% through an increase in
geometric CR of 39%.
No data could be obtained at higher EBPs due to the fact that not enough
charge scavenging was occurring to become visible in the animation window. A similar
problem was encountered at 5000rpm, even with low EBPs where fuel enrichment was
not applied.
Charge Scavenging at 2000rpm and 24 bar BMEP with an EBP of 2 bar and a
Geometric CR of 9, 10.75 and 12.5
7
3 Charge
scavenging (%)
0
9 10.75 12.5
Geometric CR
Figure 5.37 Charge Scavenging at 2000rpm and 24 bar BMEP with an EBP of 2 bar and a Geometric CR
of 9, 10.75 and 12.5
5.3.5 Engine Breathing
The single cylinder exhaust system geometry exhibits some very pronounced
tuning effects, particularly when EBP is applied. This is noticeable with both this single
cylinder model and the single cylinder experimental engine. A very large pressure wave
161
is present in the system at all speeds and can be found to originate from the exhaust
back pressure regulating valve itself (see section 6.2.3.4). This pressure wave is a
reflection of the exhaust blowdown pulse and similar in magnitude to it. The effect of
this pressure wave is different depending on the speed and load condition. In the case
of 2000rpm and 24 bar BMEP the initial pressure wave reflection occurs during the
exhaust stroke when only the exhaust valve is open so is of limited significance. The
second reflection however peaks at almost exactly TDC during the gas exchange
process and has a huge impact on the residuals in the cylinder for the subsequent
cycle.
Overlays of the EBP on a crank angle degree basis and log P-V diagram effects
have been shown in Figures 5.38 and 5.39 respectively. These figures show the
optimum IMOP points at a geometric CR of 9 and an EBP of 2 bar.
Figure 5.38 EBP Resolved on a CAD Basis for an EBP of 2 bar, a Cam Duration of 152 CAD for Speeds of
1000, 2000 and 5000rpm
162
Figure 5.39 Log P-V Overlay at an EBP of 2 bar, a Cam Duration of 152 CAD, and for Speeds of 1000,
2000 and 5000rpm
Even at this relatively low EBP condition the peak magnitude of the pressure
waves in Figure 5.38 can be seen to have a profound effect on the P-V diagram (Figure
5.39). As can be expected the pressure waves occur at a higher frequency at the lower
speed condition because the speed of sound in the exhaust is comparable for all 3
speeds whilst the engine speed is dramatically different.
For a supposedly constant EBP the pressure peaks can be seen to be

dramatically different in magnitude indicating that the 2000rpm case is more heavily
affected than the 1000rpm case. The period of greatest importance for predictable
exhaust flow is the valve overlap period. Any pressure waves that are present during
the valve overlap period can be seen to propagate through to the intake manifold and
drastically alter the engines breathing performance.
The influence of EBP and pressure waves present in the system has been
analysed at the 1000rpm condition where it can be seen to result in unpredictable
performance with the 240 CAD duration cam. The intake manifold mass flow rates for
the 3 240 CAD duration cam cases in Figure 5.8 are shown in Figure 5.40.
163
Figure 5.40 Intake Mass Flow Rate Overlay for 240 CAD Cam Duration at an EBP of 2.5 bar, and a
Speed and Load of 1000rpm and 14 bar BMEP
The pressure waves are very evident in Figure 5.40 and manifest themselves as
very large mass flow spikes. Notice the phasing of the pressure waves is different for
the geometric CR of 9 case and the geometric CR of 10.75 and geometric CR of 12.5
cases. The phasing for the geometric CR of 10.75 and geometric CR of 12.5 cases is
identical, and both profiles exhibit very large negative mass flow rates immediately
before IVC. This will result in charge being drawn into the intake manifold. Notice too
the large mass flow spike shortly after IVO, particularly for the geometric CR of 12.5
case. This is caused by a rarefaction wave in the exhaust system drawing air straight
through the cylinder. This is scavenging occurring which is having a detrimental effect
on the ISFC for these 2 cases.
At higher EBPs the peak pressure spikes can be seen to increase in magnitude
considerably which can heavily influence the MAP required to maintain a certain load.
This occurs at all engine speeds but is most noticeable at the 2000rpm, 24 bar BMEP
point with a 196 CAD duration cam, a geometric CR of 9 and an EBP of 4 bar.
164
For the 60 CAD ATDC IMOP point the pressure wave in the exhaust can be seen
to travel through the cylinder and into the intake manifold. Because the intake
manifold is considerably shorter than the exhaust manifold this pressure wave travels
at a much higher frequency through the intake and causes a considerable peak in
intake valve flow rate whilst the intake valve is still open. This is visible at
approximately 440 CAD ATDCF (Figure 5.41). This air ramming effect means the MAP
can be reduced significantly, in spite of the fact the effective CR is comparatively low
(Figure 5.42).
Figure 5.41 2000rpm, 24 bar BMEP, 196 CAD Duration Cam, EBP 4 bar, Geometric CR of 9. Intake Valve
Mass Flow Rate Profiles from an IMOP Sweep
165
Figure 5.42 2000rpm, 24 bar BMEP, 196 CAD Duration Cam, EBP 4 bar, Geometric CR of 9. MAP
Requirements for an IMOP Sweep
The result of this pressure wave propagation through to the intake results in a
great deal of exhaust gas entering the intake manifold and mixing with the fresh intake
air, as can be seen by the briefly negative mass flow rate through the intake valve in
Figure 5.41. This has the effect of raising the intake port temperature approximately
190C in comparison to the minimum valve overlap case, but ultimately the EOC
temperature and pressure is much lower for the maximum valve overlap case.
This phenomenon is present at lower EBPs too, its impact is much less
noticeable at these conditions though. It is also noticeable with the shorter 152 CAD
duration cam too, although it is most pronounced with the 196 CAD duration cam due
to the fact the pressure wave in the intake occurs almost precisely at the MOP point of
the cam. It occurs slightly too late with the 152 CAD duration cam which greatly
reduces the peak amplitude (0.13 kg/s peak mass flow rate for the 152 CAD duration
cam Vs. 0.17 kg/s peak mass flow rate for the 196 CAD duration cam) of the pressure
spike in the intake manifold, and slightly too early for the 240 CAD duration cam,
again, making its impact on the cycle almost negligible. The MAP is also much higher
for the 152 CAD duration cam which impedes reverse flow through the intake valve
166
(0.03 kg/s peak mass flow rate for the 152 CAD duration cam Vs. 0.04 kg/s peak mass
flow rate for the 196 CAD duration profile).
The impact of the variability in the speed of sound on the exhaust pressure
wave at the 5000rpm, 24 bar BMEP, 2 bar EBP and geometric CR of 9 point is shown in
Figure 5.43. Figure 5.43 shows the 5 lines corresponding to each of the different
duration cams. In brackets is the exhaust gas temperature in 0C. The exhaust gas
temperature is the main factor influencing the speed of sound, however, residuals and
enrichment will have the effect of changing the exhaust gas density somewhat, but not
considerably.
Figure 5.43 Influence of Exhaust Gas Temperature (EGT) on Exhaust Pressure/Rarefaction Wave at
5000rpm, 24 bar, 2 bar EBP and a Geometric CR of 9
For lower exhaust gas temperatures the pressure wave can be seen arrive at
shortly after TDC gas exchange (this pressure data was obtained at the exhaust
pressure transducer site, the pressure wave will not reach the cylinder until
approximately 5 CAD later than shown in Figure 5.43). The fact that the pressure wave
arrives later for the 240 CAD duration cam coupled with the fact IVC occurs at an
167
earlier point is the reason why the scavenging is much improved at 240 CAD duration
site compared to the 276 CAD duration site.
Similar engine behaviour can be seen at an EBP of 3 bar also. At this point the
276 CAD duration cam optimum point has shifted from the maximum valve overlap
position to a minimum valve overlap position. The 240 CAD duration cam exhibits
significant scavenging akin to the 2 bar EBP case, the 276 CAD duration point exhibits
little to no perceivable scavenging due to the IMOP being shifted to the minimum
valve overlap position. At this IMOP the efficiency of the cycle is dictated by the
effective CR which is not sufficiently low enough to be more efficient than the 240 CAD
duration cam point.
At an EBP of 4 bar though the scavenging can be seen to be greatly reduced for
the 240 CAD duration cam (although the maximum valve overlap position is still
optimum for this cam) which requires more fuel enrichment to regulate the
temperature as opposed to fresh air. With both the 240 CAD duration cam and 276
CAD duration cam now requiring enrichment, the benefit afforded by the reduction in
effective CR by the 276 CAD duration cam has reduced the ISFC at this point to less
than that at the 240 CAD duration cam point.
Determining the precise effect of the pressure waves in the exhaust system is
impossible without modifying the exhaust geometry and re-doing all test work. It has a
different and unpredictable effect on engine breathing at almost every single test
point. The most noticeable effect of this exhaust tuning is an offset in the worst ISFC
point with decreasing EBP or geometric CR changes. The point of optimum ISFC can be
seen to be relatively insensitive to EBP however, this is likely to be because of the fact
that in most cases the 276 CAD duration cam was optimum, and in all cases this was
with a very small degree of valve overlap so any exhaust tuning effects will likely stay
in the exhaust and not manifest themselves in the intake in a manner as can be seen in
Figure 5.41.
Charge scavenging may be influenced to a small degree by exhaust manifold

tuning but the impact will be minimal due to the fact charge scavenging happens with
the long duration cams only which also run with a small degree of valve overlap.
168
The exhaust manifold tuning for a speed of 5000rpm is quite closely matched to
that of the 3 cylinder engine (Figure 5.44), with the pressure peaks occurring close to
240 CAD apart and the pressure peaks being similar in magnitude, so for this speed the
single cylinder engine would be quite comparable to the 3 cylinder engine if it were
not for the fact the blowdown pulse changes depending on the engine running
conditions and the local speed of sound at 5000rpm (Figure 5.43).
Figure 5.44 Comparison of Single Cylinder and 3 Cylinder EBP Profiles Resolved on CAD Basis at
5.3.6 General Compression Ratio Effects
Increasing geometric CR has the theoretical effect of increasing the cycle

efficiency as can be determined from the Otto cycle efficiency formula (Eq.5.1).
5.1
where ŋOtto is the Otto cycle efficiency, rv is the geometric CR and n is the polytropic
exponent.
169
It can also be seen to extend the pumping loop. Whether or not this is
beneficial or detrimental to the overall cycle efficiency is dictated by the MAP to EBP
ratio
Figure 5.45 reveals the extent of this increase in pumping loop at the 2000rpm,
24 bar BMEP point with a cam duration of 312 CAD.
Figure 5.45 Log P-V Diagram at 2000rpm, 24 bar BMEP, 312 CAD Duration Cam, EBP of 2 bar.
Comparison of Different Geometric CRs
At 2 bar EBP the pumping loop contribution is positive and can yield an ISFC
improvement of 0.5 g/kWhr or 2.3 J/cycle. Extra work can be extracted from the gross
work too which is much more significant and results in an extra energy yield of 119
J/cycle. It should be noted though that for the geometric CR of 12.5 case the EOC
pressure is considerably higher. For the case demonstrated in Figure 5.45 the
temperature and pressure history is such that the combustion phasing can be
advanced and this extra portion of gross work can be extracted. Extracting this extra
portion of power from the cycle is difficult when running at BLD and in most cases it
was not possible with a geometric CR of 12.5. Under certain circumstances an
advantage in ISFC could be yielded with a geometric CR of 10.75.
170
Evidence of this can be seen in the Figures 5.16 to 5.18 which show all but 1 of
the geometric CR of 12.5 points to be far less efficient than the geometric CR of 9 and
geometric CR of 10.75 points. In every one of the cases where the geometric CR of 12.5
point is higher than the geometric CR of 9 point the EOC pressure can be seen to be at
least 8 bar higher. The 50% MFB points have also had to retarded dramatically for BLD
running with the geometric CR of 12.5 points as can be seen.
For all speeds the 50% MFB point had to be retarded with higher geometric
CRs. Figures 5.46 to 5.48 show the 50% MFB for each of the cam durations, geometric
CRs and EBPs.
50
45
CR 9 EBP 2.5 bar
40
CR 9 EBP 2 bar
50% MFB point (CAD ATDCF)
35 CR 9 EBP 1.5 bar
CR 10.75 EBP 2.5

30 bar
CR 10.75 EBP 2
bar
25 CR 10.75 EBP 1.5
bar
CR 12.5 EBP 2.5
20 bar
CR 12.5 EBP 2
bar
15 CR 12.5 EBP 1.5
bar
Combustion
10 stability limit
5
140 160 180 200 220 240 260 280 300 320
Figure 5.46 1000rpm, 14 bar BMEP 50% MFB Points for EBPs of 1.5, 2 and 2.5 bar at Geometric CRs of
9, 10.75 and 12.5
171
50
45
CR 9 EBP 4 bar
40
CR 9 EBP 3 bar
35 CR 9 EBP 2 bar
CR 10.75 EBP 4
30 bar
CR 10.75 EBP 3
bar
25 CR 10.75 EBP 2
bar
CR 12.5 EBP 4
20 bar
CR 12.5 EBP 3
bar
15 CR 12.5 EBP 2
bar
Combustion
10 stability limit
5
140 160 180 200 220 240 260 280 300 320
Figure 5.47 2000rpm, 24 bar BMEP 50% MFB Points for EBPs of 2, 3 and 4 bar at Geometric CRs of 9,
10.75 and 12.5
50
45
CR 9 EBP 4 bar
40
CR 9 EBP 3 bar
35 CR 9 EBP 2 bar
CR 10.75 EBP 4
30 bar
CR 10.75 EBP 3
bar
25 CR 10.75 EBP 2
bar
CR 12.5 EBP 4
20 bar
CR 12.5 EBP 3
bar
15 CR 12.5 EBP 2
bar
Combustion
10 stability limit
5
140 160 180 200 220 240 260 280 300 320
Figure 5.48 5000rpm, 24 bar BMEP 50% MFB Points for EBPs of 2, 3 and 4 bar at Geometric CRs of 9,
10.75 and 12.5
Figures 5.46 to 5.48 clearly show very strong correlations with their respective
ISFC plots in Figures 5.6, 5.9 and 5.19. Both geometric CR and EBP can be observed to
172
have a significant effect on the combustion phasing at all speeds. Some cam durations
arguably exhibit less sensitivity to EBP than others. Generally the longer cams are less
sensitive to EBP than the shorter ones, although this is not visible for the 5000rpm case
(Figure 5.48), this can be attributed to the fact that the cam IMOP timing was not
optimised due to the MAP limit restricting it. The shorter cam durations can be
expected to be more sensitive to EBP due to the scavenging inhibiting properties of
higher EBPs and the fact that the valve overlap period is generally a lot longer with the
shorter cams.
Figure 5.46 to 5.48 highlight the extent to which combustion must be retarded
to run at the BLD condition. It should be noted that it may not be possible to retard the
50% MFB point to this extent for the high geometric CR and high EBP points. There are
2 reasons for this, the ignition energy to ignite the charge may be insufficient with an
EOC pressure above approximately 50 bar. This constraint can be rectified with a
higher energy ignition system. The second reason (which is far bigger constraint) is the
fact that the experimental engine has been found to misfire and experience abnormal
combustion with 50% MFB points later than 32 CAD ATDC at all speeds. This limit is
shown in Figures 5.46 to 5.48 as a bright red dashed line.
The trends shown in Figures 5.46 to 5.48 reveal that high geometric CRs can
only be employed effectively when the effective CR is low enough, or in other words,
only for the extremely short or long cam durations. Even when effective CRs are low it
still may not be possible to run at high geometric CRs depending on the cam timing.
The extent of the benefit afforded by operating at a geometric CR of 12.5 and

10.75 is generally greatest for low EBPs. As can be expected, the greatest advantage of
running at a higher geometric CR can be shown at the lower speed and load case of
1000rpm and 14 bar BMEP. At this point the geometric CR of 10.75 benefits the ISFC
for each of the cam duration points in comparison to a geometric CR of 9. The load is
however still slightly too high to be of considerable benefit for a geometric CR of 12.5
for all but the very long 312 CAD duration cam. Although the engine can still run with
the 50% MFB phased to 32 CAD ATDCF it is not advisable due to the fact LSPI has a
much higher propensity to occur at these speeds.
173
5.4 Maximum Load Comparisons
For the purposes of this section the focus shall be shifted away from
optimisation of ISFC and shall be shifted to which cam profile is best suited for
maximum load generation. The qualities to look out for when determining the
maximum load point are similar to those for minimum ISFC. A low EOC pressure and
temperature for as high a mass of air in the cylinder will provide the highest load, a
relatively high degree of combustion phasing advance will also be required to keep the
exhaust gas temperature as low as possible.
Mechanical constraints may also be a limiting factor in the pursuit for

maximum load, for this work the maximum acceptable cylinder pressure will be 140
bar. A maximum degree of combustion phasing retard will also be imposed to ensure
the maximum load outputs are realistic in the real world. The limit is a maximum 50%
MFB anchor point phasing of 32 CAD ATDCF.
Figures 5.49 and 5.50 show the theoretical maximum load outputs for each of
the different cams at an EBP of 2 and 3 bar respectively (the dashed lines between the
points represent trend lines for each of the 5 cams).
174
46
44
42
40
38
36 152 CAD duration

NIMEP (bar)
34 196 CAD duration

240 CAD duration
32 276 CAD duration
30 312 CAD duration
28
26
24
22
0 1000 2000 3000 4000 5000 6000
Engine speed (rpm)
Figure 5.49 Peak Load (NIMEP) at 1000, 2000 and 5000rpm for each Cam Profile for an EBP of 2 bar
46
44
42
40
38
36 152 CAD duration

NIMEP (bar)
34 196 CAD duration

240 CAD duration
32 276 CAD duration
30 312 CAD duration
28
26
24
22
0 1000 2000 3000 4000 5000 6000
Engine speed (rpm)
Figure 5.50 Peak Load (NIMEP) at 1000, 2000 and 5000rpm for each Cam Profile for an EBP of 3 bar at
2000 and 5000 rpm, and an EBP of 2.5 bar at 1000rpm
The 2 bar EBP case NIMEP figures can be seen to be very high in comparison to
the 3 bar/2.5 bar figures which are more like what can be seen on the 3 cylinder
engine. The reason for this is the lack of EBP leading to very low EOC pressure and
175
temperature. In almost all cases the MAP limit of 4 bar (3 bar for 1000rpm) can be
seen to be the limiting factor on peak power with the exception of the 2000rpm point
with the 276 CAD duration cam which had to be reduced slightly because of the 140
bar cylinder pressure limit. The only 2 points in Figure 5.49 that were not MAP limited
were the 2000rpm 196 CAD duration cam and 240 CAD duration cam points which
were limited by the elevated EOC pressures and temperatures at these points.
The MAP limit manifests itself most clearly at the 5000rpm cases in both
Figures 5.49 and 5.50 where the short 152 CAD duration cam and long 312 CAD
duration cam can be seen to be down in peak load in spite of the fact the EOC
pressures and temperatures are lowest at these points. This is due to the very low
volumetric efficiency afforded by these cams.
The 3 bar/2.5 bar EBP point (Figure 5.50) showed a similar MAP limit constraint
at 1000 and 5000rpm but the 2000rpm points are very much knock limited with the
full MAP being required for just the 152 CAD duration cam point. These points can be
seen to follow the logical trend of the lower effective CR resulting in the highest load
being achieved for the most part, with the notable exception of the 196 and 240 CAD
duration cam points that appear to be inverted. One of the possible explanations for
this is the tuning phenomena that affects the 196 CAD duration cam (see section 5.3.5)
which resulted in the anomaly seen in Figure 5.12.
The 276 CAD duration cam is optimum for maximum load for 4 of the 6 speed
and EBP combinations because of good low effective CR and high mass trapping
qualities. Although the 152 CAD duration cam offers similar effective CRs to the 312
CAD duration cam (slightly greater even) it can be seen to be below the 312 CAD
duration cam at every point. The reason for this is that although the effective CR is
comparable the trapped charge mass is less than with the 312 CAD duration cam. This
due to the fact that IVC is classed as the 1mm valve lift point and not the actual valve
closing point. It is also dependent on the direction of piston motion when the valve
closes. With EIVC the piston is moving downwards, therefore flow is restricted into the
cylinder, whereas the piston motion is upwards with LIVC and flow is being restricted
176
out of the cylinder. This means that the mass of air trapped in the cylinder will always
be greater with LIVC than EIVC for equivalent effective CRs.
5.5 Summary
The effectiveness of EIVC and LIVC changes dramatically depending on the

speed, load, EBP and geometric CR combination. In most cases LIVC offers the greatest
benefit but only at low speeds. This was not always achieved with the maximum
degree of LIVC (312 CAD duration cam), in some cases moderate LIVC was more
efficient (276 CAD duration cam), particularly at low speeds and low EBPs where
charge scavenging was observed to take place.
EIVC was also observed to offer a benefit but only at high speeds where the
degree of scavenging was reduced to the extent that accidental over enrichment no
longer had a significant impact on the ISFC. EIVC can be seen to offer greater benefits
at lower speeds if the EMS is somehow able to compensate for the scavenging. If
compensation is used EIVC is generally more effective at reducing ISFC than LIVC.
The efficiency benefits offered by EIVC and LIVC are heavily dependent on EBP.
Increasing EBP generally yielded a greater performance benefit for both EIVC and LIVC,
however, there were some notable exceptions to this trend, particularly for EIVC at
very high EBPs (4 bar).
Geometric CR increase was found to benefit extreme cases of EIVC and LIVC for
almost every speed, load and geometric CR case, and lesser degrees of EIVC and LIVC
for low EBP cases. It was found to be detrimental to ISFC for the standard duration cam
and detrimental to lesser degrees of EIVC and LIVC in some cases.
The results from this test work were heavily influenced by the exhaust manifold
tuning. The influence of the exhaust manifold was found to be particularly great for
higher EBP cases where the blowdown pulse (and the corresponding
pressure/rarefaction wave) were greater in magnitude than for lower EBP cases. For
177
high degrees of valve overlap the exhaust tuning could be found to influence the
intake system and distort the required MAPs slightly. This was found to be detrimental
to cam profiles that favoured a high degree of valve overlap, in most cases.
Both EIVC and LIVC show the potential to increase the maximum load (relative
to the standard cam profile) but the requirement for high MAPs is much greater. At a
speed of 2000rpm the MAP requirement is typically the maximum limit of 4 bar for the
extreme EIVC and LIVC cases, this results in higher loads being reached for these cases,
but at a speed of 5000rpm, where the MAP requirement for the standard cam profile
is also 4 bar, the load can be seen to be reduced relative to the standard cam profile. A
similar effect can be observed at 1000rpm too where the maximum MAP was limited
to 3 bar.
178
6
Chapter 6
Single Cylinder Engine Performance Evaluation
6.1 Introduction
The objective of the first part of this chapter is to assess how the single cylinder
experimental engine performs in comparison to the 3 cylinder engine and the single
cylinder engine model, in an attempt to better ascertain what impact engine breathing
will have on the successful implementation of EIVC and LIVC strategies.
Autoignition has also been studied extensively. The single cylinder

experimental engine has shown a greater propensity for severe autoignition events
(severe meaning cylinder pressures in excess of 30% higher than the average peak
cylinder pressure) in comparison to the 3 cylinder engine, particularly at low engine
speeds. The source of this severe autoignition behaviour has been investigated,
including whether or not it is caused by LSPI or simply by knocking combustion. The
impact of coolant flow rate effects and fuel age effects on autoignition tendency has
also been investigated.
6.2 Comparison of the Single Cylinder Experimental Engine to the 3 Cylinder Engine
179
6.2.1 Introduction
The purpose of this investigation is to see how the single cylinder experimental
engine compares to the 3 cylinder engine. Ultimately the intention of this study is to
apply what has been learned here to the 3 cylinder engine, and the objective of this
test work is to ascertain whether or not the data and the trends seen on the single
cylinder experimental engine are going to be a good representation of the behaviour
of the 3 cylinder engine.
The engines will be compared at equivalent speed and load sites across the
engine operating range. All of the other criteria such as injection timing, rail pressure,
coolant and oil temperatures and CAT as used on the 3 cylinder engine will be
duplicated on the single cylinder engine as closely as possible.
6.2.2 Description of Test Points
Cam timing sweeps were carried out in various regions of the operating map to
both obtain some baseline data (which can be compared to the GT-Power model of
both the single cylinder and 3 cylinder engines) and also to compare the different
profile cams to one another. The test points are shown in Figure 6.1 and are given in
Table 6.1. This baseline data was obtained with the standard 240 CAD duration 3
cylinder engine intake cam.
180
Figure 6.1 Baseline Data Test Points
Table 6.1 Baseline Data Test Points
Speed (rpm) BMEP (bar) NIMEP (bar) EBP valve position (% closed) SOI (CAD BTDC) Rail pressure (bar)
1000 4 4.56 0 320 85.93
1000 8 8.83 0 276 132.63
2000 4 4.64 0 320 91.93
2000 8 8.90 0 319 131.06
3000 4 4.80 0 320 101.83
3000 8 8.98 0 319 134.45
4000 4 4.98 0 291 110.81
4000 8 9.05 0 297 140.56
2000 15 16.04 0 290 168.54
2000 15 16.04 45 290 168.54
3000 15 16.44 0 303 167.8
3000 15 16.44 45 303 167.8
2000 22 23.48 0 262 169.36
2000 22 23.48 40 262 169.36
2000 24 25.58 0 259 169.26
2000 24 25.58 * 2 bar EBP 259 169.26
3000 24 25.49 0 300 169.2
3000 24 25.49 35 300 169.2
4000 24 25.84 0 326 169.3
5000 24 26.02 0 349 170.01
* - EBP in bar, not valve position
The maximum load tested is relatively conservative in comparison to the 3

cylinder engine BMEP curve. The 3 cylinder comparison was obtained from baseline
data carried out by Mahle Powertrain on the second generation Mahle 1.2l 3 cylinder
downsized engine with Bosch-Mahle TurboSystems turbocharger and a CR of 9.3. A
181
maximum BMEP of 24 bar was imposed because of mechanical constraints. The single
cylinder test points in Figure 6.1 correspond to 3 cylinder engine effective BMEP
values, for example, 24 bar BMEP at 5000rpm for the 3 cylinder engine requires a
NIMEP of 26.02 bar, the single cylinder engine was therefore run at an NIMEP of 26.02
bar at this point. The single cylinder brake values will not be used in this chapter
because they are artificially low due to the fuel pump and 3 cylinder camshaft (as
opposed to a single cylinder specific camshaft) parasitic losses rendering them
unrepresentative.
A list of the test variables and control criteria are given in Table 6.2.
Table 6.2 Test Variables and Control Criteria
Variable Control criteria

Air <3% humidity, temperature maintained at +/- 3 degC of 40 degC intake port temp
Coolant Temperature maintained at +/- 3 degC of 80 degC
Oil Temperature maintained at +/- 3 degC of 90 degC
Fuel Fresh ULG RON 95, supply temperature regulated to +/- 5 degC of 25 degC
Speed See Table 6.1
Loads See Table 6.1
Intake cam Standard 240 CAD profile
Exhaust cam Standard 278 CAD profile
IMOP 80 - 120 CAD ATDC (swept in 10 CAD increments)
EMOP 100 - 140 CAD BTDC (swept in 10 CAD increments)
Spark timing BLD or MBT depnding on load
Exhaust For EBPs see Table 6.1. Temperature control not required
Data acquisition 300 cycles logged
6.2.3 Results
6.2.3.1 ISFC
A direct comparison of ISFC (Figure 6.2) shows that the single cylinder engine is
higher in ISFC at every point in comparison to the 3 cylinder engine, with the exception
of 5000rpm and 24 bar BMEP, where the ISFC on the single cylinder can be seen to be
below that of the 3 cylinder. The reason for this is likely to be the employment of fuel
enrichment.
182
Enrichment could not be applied reliably at this point on the single cylinder
engine, which imposed an EBP constraint at high speed and load. The reason for this is
the lambda sensor was found to malfunction at very high temperatures. Attempts
were made to re-site the lambda sensor and to cool the sensor down but the problem
persisted. Different types of lambda sensor were used but exhibited similar issues.
Attempts were made to run the engine with open loop control of lambda but due to
boost pressure fluctuations from the boost rig, stable lambda values were
unobtainable. This fault is the reason why the 4000 and 5000rpm 24 bar BMEP points
“with EBP” and the “no EBP” points align perfectly, because no EBP could be
employed.
320
310
300
290
3 cylinder
280
ISFC (g/kWhr)
270
Single
cylinder
260 (with EBP)
250
Single
cylinder (no
240 EBP)
230
220
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
4 bar 4 bar 4 bar 4 bar 8 bar 8 bar 8 bar 8 bar 15 bar 15 bar 22 bar 24 bar 24 bar 24 bar 24 bar
Engine speed (rpm)
Load (BMEP)
Figure 6.2 ISFC Comparison of 3 Cylinder and Single Cylinder Engines
There are many possible reasons why the ISFC may be higher on the single
cylinder engine including; inaccurate in-cylinder pressure reading, inaccurate fuel flow
rate reading, incorrect TDC position, incorrect lambda sensor feedback, elevated heat
losses in the cylinder, large cycle to cycle COV, MFB rate, intake and exhaust manifold
tuning, combustion phasing, residuals in the cylinder and air humidity. It is unlikely
that the increase in ISFC for the single cylinder engine can be attributed to just one of
these things. The in-cylinder pressure transducer and fuel flow meter are both
183
calibrated and accurate. A capacitive TDC sensor is used to determine dynamic TDC
(see section 3.3.2). This TDC probe is applied at one speed only (1200rpm) when this is
not ideal as the thermodynamic loss angle can change dramatically with engine speed,
however, the biggest difference in ISFC occurs at low engine speed and reduces as
engine speed rises which demonstrates the incorrect trend to make incorrect TDC
measurement the cause of the discrepancy seen in Figure 6.2.
The difference in ISFC between the 2 engines can be seen to be, for the most
part, insensitive to either engine speed or load. This trend rules out elevated heat
losses because although the surface area of the single cylinder engine is relatively large
in comparison to the 3 cylinder engine (possibly even greater than the actual 3 cylinder
engine) provided the engine is kept at the same temperature the heat flux through it
(although far greater than the 3 cylinder engine when calculated on a per-cylinder
basis) should be constant regardless of engine speed and load.
6.2.3.2 Engine Breathing
A more viable explanation for the difference in ISFC between the 2 engines
could be engine breathing and in-cylinder flow structure. The impact of the difference
in engine breathing dynamics is very clear at low speed, particularly at low load. The
largest discrepancy between the single cylinder engine and 3 cylinder engine is at the
1000rpm, 4 bar BMEP condition. The reason for this can be partly attributed to the
difference in MFB durations between the 2 different engines, which shows the single
cylinder engine is characterised with a much longer combustion duration as defined by
the 10-90% MFB duration than the 3 cylinder engine (3 cylinder data obtained from
Mahle Powertrain). The 10-90% MFB duration and spark timing for the single cylinder
is 20.9 CAD and 28.6 CAD BTDC respectively, compared to 17.2 CAD and 19.4 CAD
BTDC respectively for the 3 cylinder engine.
Given the different intake and exhaust manifold configurations and flow
properties the engine breathing between the 2 engines can be expected to be
significantly different. The difference in the spark timing and combustion duration
184
would suggest that the single cylinder engine either has a greater residual mass
fraction at cycle start than the 3 cylinder engine and/or that there is less turbulence in
the cylinder meaning the flame front propagates through the end gas at a lower speed.
Figure 6.3 shows a spark timing comparison between the single cylinder and 3
cylinder engines. It shows that at low speed the spark timing had to be advanced to
achieve the same 50% MFB timing of approximately 8 CAD ATDC (MBT point). This
could be indicative of reduced charge motion or increased residual gas concentration
with the single cylinder engine. At higher loads this trend can be seen to reverse at
2000rpm and 8 bar BMEP. Figure 6.4 shows a 50% MFB point comparison of the 3
cylinder and single cylinder engine with and without EBP. The fact that there is no
considerable difference in the MBT timings at low loads means the MFB duration is
longer at 1000rpm and low loads, and faster at medium loads. This is confirmed by a
comparison of 10-90% MFB durations too (Figure 6.5). All of these signs point strongly
to different breathing behaviour.
40
35
30
25
Spark timing (CAD BTDC)
3 cylinder
20
Single
15 cylinder
(with EBP)
10
Single
cylinder (no
5
EBP)
-5
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
Engine speed (rpm)
Load (BMEP)
Figure 6.3 Single Cylinder and 3 Cylinder Spark Timing Comparison
185
30
25
50% MFB point (CAD ATDC)
20
3 cylinder
15
Single
cylinder
(with EBP)
10
Single
cylinder (no
EBP)
5
0
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
Engine speed (rpm)

Load (BMEP)
Figure 6.4 Single Cylinder and 3 Cylinder 50% MFB Point Comparison
26
24
22
10-90% MFB duration (CAD)
20 3 cylinder
18
Single
cylinder
16 (with EBP)
Single
14 cylinder (no
EBP)
12
10
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
Engine speed (rpm)
Load (BMEP)
Figure 6.5 Single Cylinder and 3 Cylinder 10-90% MFB Duration Comparison
The difference in the breathing performance is perhaps best illustrated by the

difference in boost pressure requirement for the equivalent speed and load points for
both engines (Figure 6.6) and by the EBP (Figure 6.7).
186
3
2.5
2
3 cylinder
MAP (bar)
1.5
Single
cylinder
(with EBP)
1
Single
cylinder (no
0.5 EBP)
0
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
Engine speed (rpm)
Load (BMEP)
Figure 6.6 Single Cylinder and 3 Cylinder MAP Requirement Comparison
3.5
2.5
3 cylinder
EBP (bar)
2
Single
cylinder
(with EBP)
1.5
Single
cylinder (no
1 EBP)
0.5
1000 2000 3000 4000 1000 2000 3000 4000 2000 3000 2000 2000 3000 4000 5000
Engine speed (rpm)
Load (BMEP)
Figure 6.7 Single Cylinder and 3 Cylinder EBP Comparison
In spite of the fact the single cylinder engine is run with a similar EBP to that of
the 3 cylinder, it requires a much higher MAP in comparison to the 3 cylinder engine.
187
One possible explanation for this massive increase in boost requirement is the rate at
which air flows through the intake manifold. In the case of the 3 cylinder engine the
phase difference between the cylinders and the cam durations result in a near
continuous flow through the intake system, whereas for the single cylinder engine the
air in the intake manifold is continuously stopping and starting through the entire
intake system.
The position of the plenum relative to the intake port will influence engine
breathing and is supposed to help the engine maintain a constant MAP. The influence
of the plenum position can be seen in Figure 6.8. Pressure waves can be seen for each
of the 3 solid lines, but the waves can be seen to be vastly different for the single and 3
cylinder engines. These waves are caused by the blowdown pulse that occurs when a
valve is initially opened and are inevitable regardless of engine configuration, but the
magnitude of the waves across the entire cycle can be seen to be lower for the 3
cylinder engine simulation than for the single cylinder simulation. It must also be noted
that for the single cylinder engine the throttle has been located downstream of the
plenum whereas on the 3 cylinder engine the throttle is located upstream of the
plenum.
188
3 0.125
2.5 0.1
Single cylinder
simulation: pressure
2 0.075 3 cylinder simulation:
Mass flow rate (kg/s)

pressure
MAP (bar)
Single cylinder
1.5 0.05 experiment: pressure
Single cylinder
simulation: mass
1 0.025 flow rate
3 cylinder simulation:
mass flow rate
0.5 0
0 -0.025
-80 -20 40 100 160 220 280 340 400 460 520 580 640
Figure 6.8 Comparison of Intake Manifold Pressures and Intake Valve Mass Flow Rate at 5000rpm and
24 bar BMEP
The blowdown pulse itself is roughly equal in magnitude between both single
cylinder scenarios and the 3 cylinder simulation, however, the instantaneous MAP for
the 3 cylinder engine can be found to recover much more quickly in comparison to the
single cylinder engine, this results in a greatly increased flow rate through the valve in
spite of the greatly elevated EBP. The MAP for the 3 cylinder engine can also be found
to be much more consistent throughout the cycle, and the average MAP is a good
indication of what the cylinder actually sees when the intake valve is opened. In the
case of the single cylinder engine the MAP whilst the intake valve is open can be seen
to be more approximately 0.4 bar down on the average MAP across the entire cycle.
This can be found to have an impact on ISFC in the form of a slightly larger
pumping loop (Figure 6.9) by forming a slight downward lip as highlighted by the circle.
In total though the 3 cylinder engine can be found to have a much larger pumping loop
in this instance, owing to the fact the EBP has been massively elevated by the
turbocharger. The positive contribution of the pumping loop for the single cylinder
engine is very small in comparison, and is confined to a very small area by a volume
189
ratio of approximately 0.8 to 1. The rest of the pumping loop is actually negative which
means positive work is applied to the piston during the gas exchange process.
This positive work contribution during the pumping loop is provided solely by
the boost rig in this instance.
100
Single cylinder simulation

In-cylinder pressure (bar)
3 cylinder simulation
10 Single cylinder
experiment
1
0 0.2 0.4 0.6 0.8 1 1.2
Volume ratio
Figure 6.9 5000rpm 24bar BMEP logP-V Comparison
The positive pumping loop work contribution can be found to impact the rest of
the cycle too. As mentioned previously, the BMEP figures quoted in this chapter are 3
cylinder effective BMEPs, in order to obtain these effective BMEPs a NIMEP figure was
targeted. NIMEP encompasses all 4 strokes of the thermodynamic cycle, therefore if a
constant NIMEP is to be maintained, a different sized pumping loop work contribution
must be compensated for by a different sized compression and expansion (gross) work
contribution.
At this test point the extra work required to drive the turbocharger is 1.428 bar
Mean Effective Pressure (MEP) which equates to 2.38kW per cylinder. The actual cost
however is more than this as boosting the GIMEP 1.428 bar requires the combustion
phasing to be retarded to a greater extent (which is visible in Figure 6.9) and a higher
190
peak cylinder pressure. The elevated EBP has also resulted in a higher EOC pressure
which will require the combustion phasing to be retarded to a greater extent still.
The impact of this phenomenon on ISFC is pronounced, and is best

demonstrated by Figure 6.2. A comparison of the single cylinder engine with and
without EBP lines shows this effect and reveals a reduction in ISFC of up to 3.2% when
turbocharger representative levels of EBP are employed.
The single cylinder engine can be seen to benefit from this lack of considerable
EBP, not only through a reduction in the pumping loop losses but also through the
lower end of induction pressure and temperature. However, this beneficial effect of
lower pressure and temperature has been lost during the compression process. The
exact reason for this is likely to be down to different residual gas concentrations at the
beginning of compression, or perhaps a difference in the way the fuel evaporation is
modelled in GT-Power and how the fuel evaporates in the physical engine.
GT-Power simulations of the single cylinder and 3 cylinder engines reveal that
at the 5000rpm, 24 bar BMEP point the single cylinder will have 1.57% residual burned
gas at cycle start, and the 3 cylinder engine will have 2.76% residual burned gas at
cycle start. Exhaust gasses can be found to have a higher heat capacity than fresh air
and fuel mixture because exhaust gas contains more large molecules than fresh air
does. Sonntag, et al [119] documents that large molecules (such as typical exhaust gas
molecules) have greater capacity for storing heat and can store energy in more ways
such as vibrational and rotational energy in comparison to smaller molecules (such as
air molecules).
This means that residuals have the effect of reducing the polytropic exponent
during the compression phase leading to lower EOC pressures and temperatures. This
theory does not explain what is apparent in Figure 6.9 however. Although the
increased residuals will have the effect of reducing the polytropic exponent slightly,
the influence of the raised EBP in the case of the 3 cylinder engine will result in an EOC
pressure and temperature far higher than that of the single cylinder engine. This can
be seen by the EOC pressure offset between the single cylinder simulation and the 3
cylinder simulation in Figure 6.9.
191
6.2.3.3 Knocking Tendency
In terms of knocking performance the single cylinder engine appears to be in

very good agreement with the 3 cylinder engine for the most part. Figure 6.4 shows
the differences in 50% MFB points between both engines.
The most noticeable difference between the engines is the 2000 and 3000rpm
15 bar BMEP points. Without EBP the single cylinder engine is in very good agreement
with the 3 cylinder engine, however, even with a modest amount of EBP, combustion
has to be retarded significantly.
The reason for this is the tuning state of the exhaust manifold of the single
cylinder engine. The rarefaction/pressure wave caused by the blowdown pulse is
sufficiently large enough to greatly impede the flow of burned gasses out of the
cylinder, thus leading to a greater residual gas concentration at the start of cycle. This
has the effect of elevating the EOC pressure and temperature in comparison to the 3
cylinder engine, and also serves to reduce the MFB duration. A comparison of the
cylinder pressures and EBPs at 2000rpm and 15 bar BMEP is given in Figures 6.10 and
6.11 respectively.
192
70
Single cylinder
test (with EBP)
7 3 cylinder
simulation
0.7 0 0.2 0.4 0.6 0.8 1 1.2

Volume ratio
Figure 6.10 2000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation In-cylinder Pressure
Comparison
2.2 12
2
10
1.8
8 3 cylinder
1.6 simulation
EBP (bar)
Single cylinder
1.4 6
Lift (mm)
test (no EBP)
1.2 Single cylinder

4 test (with EBP)
1 Exhaust valve
lift
2
0.8
0.6 0
-180 -120 -60 0 60 120 180 240 300 360 420 480 540
Crank Angle (CAD ATDCF)
Figure 6.11 2000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation EBP Comparison
193
A similar phenomenon can be seen at the 3000rpm, 15 bar BMEP condition
also. Figure 6.12 shows the impact of the EBP on in-cylinder pressure and Figure 6.13
shows the actual EBP data.
70
Single cylinder
test (with EBP)
7
3 cylinder
simulation
0.7
0 0.2 0.4 0.6 0.8 1 1.2
Volume ratio
Figure 6.12 3000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation In-cylinder Pressure
Comparison
194
2.4 12
2.2
10
2
1.8 8 3 cylinder
simulation
1.6
EBP (bar)
6 Single cylinder
Lift (mm)
test (no EBP)
1.4
Single cylinder
1.2 4 test (with EBP)
1 Exhaust valve
lift
2
0.8
0.6 0
-180 -120 -60 0 60 120 180 240 300 360 420 480 540
Figure 6.13 3000rpm, 15 bar BMEP Single Cylinder and 3 Cylinder Simulation EBP Comparison
6.2.3.4 Correct EBP Determination
Accurately simulating a turbocharger on a single cylinder engine is very difficult

and virtually impossible with just a butterfly valve. Setting the butterfly valve to an
angle that gives the same average pressure as a turbocharger can result in the sort of
phenomena that is apparent in Figures 6.10 to 6.13 where the flow is choking through
the partially closed orifice leading to very severe pressure waves.
The method used for setting the appropriate EBP was to adjust the valve so the
average EBP for the single cylinder engine across the entire cycle was equal, or close to
the EBP for the 3 cylinder engine across the entire cycle. Figures 6.11 and 6.13 reveal
that for these speed and load conditions this was not the correct approach, and that
the average EBP while the exhaust valve is open is very different to the average EBP
across the entire cycle.
195
Table 6.3 reveals that the average EBP across while the exhaust valve was open
is higher for the single cylinder engine than it is for the 3 cylinder engine with
supposedly the same EBP.
Table 6.3 Table of Average EBP during the Exhaust Valve Open Phase
Case Average EBP (bar)

2000rpm, 15 bar single cylinder (no EBP) 1.136
2000rpm, 15 bar single cylinder (with EBP) 1.275
2000rpm, 15 bar 3 cylinder simulation 1.268
3000rpm, 15 bar single cylinder (no EBP) 1.192
3000rpm, 15 bar single cylinder (with EBP) 1.526
3000rpm, 15 bar 3 cylinder simulation 1.330
For both speeds the average EBP is raised by the pressure/rarefaction wave and
for both speeds, from a profile point of view, and from a residuals point of view, the
single cylinder without EBP data is a more reliable representation of the 3 cylinder
engine than the single cylinder with EBP data.
This pressure/rarefaction wave is common to all 2000 and 3000rpm test points
where the EBP valve has been applied, as can be seen from Figure 6.14 for the
2000rpm cases. Frequency analysis confirms that the CAD interval between peaks
corresponds to a distance of about 4m when approximating the speed of sound in the
exhaust based on exhaust port temperature. This is almost exactly twice the distance
between the exhaust pressure transducer and the EBP valve.
196
3.5 12
3
10
2.5
8
2 2000rpm, 15
bar
6 2000rpm, 22
Lift (mm)
EBP (bar)
bar
1.5 2000rpm, 24
bar
4
1
2
0.5
0 0
-180 -120 -60 0 60 120 180 240 300 360 420 480 540
Figure 6.14 2000rpm EBP Comparison at Various Loads with the EBP Valve Applied
The presence of this pressure wave during the exhaust stroke will undoubtedly
have an impact on engine performance and on the residuals present in the cylinder for
the subsequent cycle. However due to its position in the cycle its influence will be
minimised, the third pressure wave (in the region of 340 to 440 CAD ATDCF) will have a
far greater impact on residuals as this coincides exactly with the valve overlap point.
Figure 6.15 shows where the exhaust pressure waves fall in comparison with the valve
overlap position.
197
3.5 7
EBP profiles
3 6
Exhaust valve lift
(EMOP 100 CAD
BTDC)
2.5 5
Intake valve lift
(IMOP 120 CAD
2 4 ATDC)
Intake valve lift
Lift (mm)
EBP (bar)
(IMOP 110 CAD

1.5 3 ATDC)
Intake valve lift
(IMOP 100 CAD
1 2 ATDC)
Intake valve lift
(IMOP 90 CAD ATDC)
0.5 1
Intake valve lift

0 0 (IMOP 80 CAD ATDC)
300 310 320 330 340 350 360 370 380 390 400 410 420
Crank Angle (CAD ATDCF)
Figure 6.15 2000rpm, Various Loads Single Cylinder EBP Overlay Vs. Valve Overlap Period
This will greatly limit the scavenging ability of the engine and will distort the
result regardless of what cam timing combination is being tested.
6.3 Impact of Coolant Flow Rate on Autoignition Tendency
6.3.1 Introduction
Owing to the intricate geometry that the COSCAST process allows (Smith, et al
[120]), and the fact the coolant jacket has been designed for high specific power
outputs, the coolant jacket geometry on the Mahle 1.2l 3 cylinder downsized head is
capable of dissipating heat very effectively from all parts of the cylinder head and
greatly reduces the likelihood of any hot spots occurring.
The coolant jacket geometry was dictated by the results of a Computational

Fluid Dynamics (CFD) study, and has been designed to create flow of a 3D nature. Fins
198
within the coolant jacket are used to guide coolant flow to further ensure optimum
cooling is taking place.
A standard 12V DC coolant pump was initially installed to the Mahle 1.2l 3
cylinder downsized engine, which is capable of pumping coolant at a rate of up to 100
l/min. With a coolant jacket this sophisticated and effective, the very valid question of
whether or not the coolant pump needs to be this large and whether or not the head
can function correctly, particularly at low speed, with a lower coolant flow rate is
raised.
In order to establish whether or not this was the case a test plan was devised
which involved regulating the coolant flow rate to the head, the block, and to the
entire engine independently of one another to determine which component would
have the biggest effect (if any at all) on autoignition tendency.
The work was split into 2 phases. In the first phase of test work the coolant
flow rate has been regulated to give a certain coolant temperature difference between
the engine coolant inlet side and the engine coolant outlet side which will henceforth
be called the ∆T. Because this engine has a split cooling system and a certain hardware
setup it is possible to regulate the coolant flow for the cylinder head and the coolant
flow for the cylinder block independently of one another. It is also possible to measure
the coolant flows for the head and the block separately by means of 2 turbine type
flow meters.
6.3.2 The Effect of Coolant Temperature Gradient
The objective of this first phase of test work was to try and regulate the flow to
produce a predetermined ∆T across the head and the block. Owing to the dangers
involved in going in to the test cell whilst the engine is running and adjusting the
coolant flow rates, the engine was run with full coolant flow rate initially and run at
the 2000rpm, 20 bar NIMEP running condition and left to stabilize for 10 minutes. A
199
log was then taken of the coolant flow rates and ∆Ts across the head and block, as well
as spark timing and manifold pressure data.
The ∆Ts and corresponding flow rates are shown below:
Component Coolant flow rate (l/min) Delta T (deg C)

Head 5.23 6.4
Block 5.37 6.7
Using this data flow rates for ∆Ts of 9, 12 and 150C were calculated using the
following relationship (Eq.6.1):
6.1
where the Q is heat rejection, m denotes mass (coolant mass flow rate in this instance)
and Cp is specific heat capacity at constant pressure. For this test work the value of Q
is constant (because the speed and load are constant) and Cp is constant. With this in
mind Eq.6.1 can be rearranged to give a relationship of the variation of ∆T with
coolant mass flow rate m (Eq.6.2).
6.2
Using this formula it is then possible to derive a table of flow rates for the
desired ∆Ts as shown in Table 6.4. Please note that at the head and block ∆T of 150C
point (the lowest total coolant flow rate point) the flow rate is so low that it is out of
the calibrated range of one of the flow meters. This point was still tested but may have
been compromised by this.
200
Table 6.4 Test Plan Showing Desired ∆Ts and the Corresponding Flow Rate Required to Achieve that
∆T
Block Coolant Delta T (deg C)

6.7 9 12 15
5.369581 3.997355 2.998016 2.398413
6.4 5.230148 5.230148 5.230148 5.230148
5.369581 3.997355 2.998016 2.398413
9 3.719216 3.719216 3.719216 3.719216
Head Coolant Delta T (deg C)
5.369581 3.997355 2.998016 2.398413
12 2.789412 2.789412 2.789412 2.789412
5.369581 3.997355 2.998016 2.398413
15 2.23153 2.23153 2.23153 2.23153
Block Coolant Flow Rate (l/min)

Head Coolant Flow Rate (l/min)
The engine was then run at each of these points for 3 minutes with 100 cycles
being logged every minute unless a severe autoignition was witnessed, in which case
that was logged as well. The total number of autoignition events was recorded by
means of a counter which is triggered every time the peak cylinder pressure of a cycle
goes above a certain threshold. The threshold for this work was set to 100 bar.
Other test variables and their control criteria are shown in Table 6.5.
201
6.3.2.1 Results
In total 9 severe autoignition events were experienced during this test work.
Figure 6.16 shows the distribution of these events across the range of coolant flow
rates tested, the z-axis represents the number of severe autoignition cycles at each
coolant flow rate combination over the period of 3 minutes. This data showed a vague
trend but no statistically significant results were obtained.
Figure 6.16 Severe Autoignition Frequency
The autoignition events were slightly different in character depending on the

ΔT condition. The autoignition frequency showed a vague trend with coolant flow rate
(Figure 6.17) and showed that block coolant temperature is perhaps of greater
importance than head coolant temperature in the prevention of autoignition (Figure
6.18).
202
200
190
Peak in-cylinder pressure (bar) 180
170
160
Peak cylinder
150 pressure Vs Total
coolant flow rate
140
Linear (Peak cylinder
pressure Vs Total
130
coolant flow rate)
120
110
100
4 4.5 5 5.5 6 6.5 7 7.5 8
Total coolant flow rate (l/min)
Figure 6.17 Peak Autoignition Cylinder Pressure Vs. Total Engine Coolant Flow Rate
200
190
180
Peak in-cylinder pressure (bar)
170
160
Peak cylinder pressure Vs
150 Actual delta T block
140
Linear (Peak cylinder
pressure Vs Actual delta
130 T block)
120
110
100
6 6.5 7 7.5 8 8.5 9 9.5
Block del T (deg C)
Figure 6.18 Peak Autoignition Cylinder Pressure Vs. Block Coolant Flow Rate
Trend lines have been added to Figures 6.17 and 6.18, although the trend line
in Figure 6.18 is not as close a fit as that in Figure 6.17. Figure 6.17 shows that the
203
lower the coolant flow rate the more severe the autoignition event is likely to be.
However, the manner in which autoignition events occur also changes with coolant
flow rate. The 128 bar, 102 bar, 105 bar and 167 bar autoignition events were all
isolated events, whereas the 156 bar and 198 bar events were not isolated but
occurred as a cluster of 2 or 3 autoignition events.
The presence of more severe autoignitions at lower coolant flow rates suggest
that perhaps hot spots are forming in the cylinder. There is also an unusual variation in
coolant flow rate with lower head coolant flow rates, implying that perhaps the
method of cooling is changing from forced convection to nucleate boiling.
The trend line in Figure 6.18 is made even more questionable by the fact the
coolant flow for the head also passes through the block both on its way to and from
the head. The coolant gallery for the head extends along a significant length of the
block (Figure 3.4) and at low head coolant flow rates will give sufficient time for the
head coolant to reject a great deal of heat to the block. This will also result in some
uncertainty of the true ΔTs obtained from the head and block respectively with the
most likely result being that the actual head ΔT is going to be much greater than the
measured head ΔT, and the actual block ΔT being less than the measured block ΔT.
The ΔTs for the head and block are shown in Figures 6.19 and 6.20 respectively.
204
Figure 6.19 Head Coolant Delta T (deg C)
Figure 6.20 Block Coolant Delta T (deg C)
205
As can be seen from both Figures 6.19 and 6.20, the actual ΔT is far less than
the predicted ΔT. There are many potential reasons for this, the most likely of which is
that as the coolant flow rate is reduced more heat is lost through the plumbing
because the coolant is taking longer to reach the temperature measuring PRTs. This
will therefore mean the coolant is cooled down to a greater extent at low flow rates.
Another reason is that the way in which the coolant is flowing may result in a greater
heat flux for the coolant, for example if nucleate boiling is occurring the rate of heat
rejection will increase dramatically compared to film boiling and forced convection, as
has been documented by Robertson, et al [121].
Figure 6.20 is a map of block coolant temperature for each of the test points. It
can be seen that the block temperature is largely insensitive to head temperature,
however, Figure 6.19, a map of head temperatures at each of the test points, can be
found to show a large dependence on the block temperature, particularly at large head
coolant ΔT predictions (very low head coolant flow rates). This dependence is so large
in fact that at a predicted head ΔT of 150C, when the block predicted ΔT is swept from
60C to 150C the actual block ΔT is the same in magnitude as the actual head ΔT.
This could be attributed partly to a discrepancy in head coolant flow rate

(Figure 6.21) at high predicted head ΔTs. The isolines in Figure 6.21 should be
horizontal, the fact that they are not means that something is changing the flow rate.
This discrepancy can be attributed to user error to an extent, but all flows were set at a
2000rpm, 4 bar NIMEP condition to within a maximum error of ±0.1 l/min. The extent
of this user error is perhaps most visible in Figure 6.22 which shows a map of block
coolant flow rate, and can be seen to be relatively insignificant.
206
Figure 6.21 Head Coolant Flow Rate (l/min)
One possible reason for this discrepancy in head coolant flow rate is that as the
flow rate is reduced, the flow behaviour through the head coolant jacket is changing.
The coolant jacket features complex 3 dimensional flow geometry (Figure 6.23) which
has not been designed for flows as low as these. However, if the flow was drastically
different it would manifest itself at the 2000rpm 4 bar NIMEP condition too, implying
that the discrepancy in flow would more likely be caused by increased load and heat
rejection.
207
Figure 6.22 Block Coolant Flow Rate (l/min)
Figure 6.23 Cylinder Head Coolant Jacket Geometry (Hancock, et al [48])
Although the maximum temperature was always controlled to 800C for this test
work it is highly likely that the local temperatures in the head (and block) will go far in
excess of this, (particularly if the intended coolant jacket 3D flow has been destroyed
by insufficient flow rate) resulting in stagnating pools of coolant in the coolant jacket
where nucleate boiling or even film boiling is occurring. This will likely change the
coolant pressure in the head and reduce/increase the coolant flow rate
correspondingly. To an extent this is supported by Figure 6.19 which shows greatest
heat rejection in the top portion of the map where the isolines can be found to be
208
more vertical than in the bottom portion of the map. This implies that more of the
block’s heat is being absorbed through the head, a side effect of the coolant in the
head having a higher heat flux. As mentioned previously coolant in a nucleate boiling
condition has a higher heat flux than coolant in a forced convection condition.
There was no perceivable increase in KI (Figure 6.24) across the entire range of
the coolant flow rates tested. For each test point the BLD point was located and for
each test point it was almost exactly the same (Figure 6.25) with slight differences
between test points but no obvious trends.
Figure 6.24 Average Band-pass Filtered KI (bar)
209
Figure 6.25 BLD 50% MFB (CAD ATDCF)
6.3.3 Effect of Coolant Flow Rate at Higher Load Operations
The second phase of this test work involved running at a higher load for a
longer duration to get some statistically significant autoignition data following the
inconclusive results obtained from the first phase.
The load was increased from 20 bar NIMEP to 24 bar NIMEP and the duration
the engine was held there for increased from 3 minutes to 15 minutes, or limited by
the time it took 6 autoignitions to occur, whichever came first. Other than that this
test is identical to the test plan in the first phase and the criteria outlined in Table 6.5.
Owing to the elevated load the coolant flow rates required for specific ΔTs had
to be recalculated. The same technique for determining the correct flow rates as in the
first phase of the test work was adopted for the second phase, which was firstly
running the engine at the test condition with maximum coolant flow rate and then
210
using Eq.6.2 to calculate a table of corresponding theoretical coolant flow rates to give
a required ΔT.
The ∆Ts and corresponding maximum flow rates are shown below:
Component Coolant flow rate (l/min) Delta T (deg C)

Head 4.90 6.9
Block 5.40 3.4
From this a map of test points was produced as shown in Table 6.6.
Table 6.6 Test Plan Showing Desired ∆Ts and the Corresponding Flow Rate Required to Achieve that
∆T
Block Coolant Delta T (deg C)

3.4 6 9 12 15
5.396921 3.05825546 2.038837 1.529128 1.223302
6.9 4.904733 4.9047326 4.904733 4.904733 4.904733
5.369581 3.05825546 2.038837 1.529128 1.223302
9 3.760295 3.760294993 3.760295 3.760295 3.760295
Head Coolant Delta T (deg C)
5.369581 3.05825546 2.038837 1.529128 1.223302
12 2.820221 2.820221245 2.820221 2.820221 2.820221
5.369581 3.05825546 2.038837 1.529128 1.223302
15 2.256177 2.256176996 2.256177 2.256177 2.256177
Block Coolant Flow Rate (l/min)

Head Coolant Flow Rate (l/min)
Some of these values have been highlighted in red. These represent points that
are below the calibrated range of the flow meters and were therefore not tested. One
hundred cycles were logged every 5 minutes for this test work, unless the autoignition
was so frequent that the engine could not be run at that test point for the full 15
minutes, in which case they were logged at more regular intervals. As with the first
phase of this work a threshold and counter were used to log the number of
autoignitions, and every time an autoignition event that exceeded this threshold
occurred it was logged. The peak cylinder pressure threshold for this was 120 bar.
6.3.3.1 Results
The results showed no real trend for autoignition but it did show a great deal of
difference in frequency, as can be seen from Figure 6.26. The reason for this is
211
unknown but could be attributed to the fact that in some instances 2 autoignitions
would happen within a few cycles of each other (with 1 occurring as a consequence of
the other), and this elevates the frequency somewhat. The decision was made to count
all of these autoignitions separately. This decision was made because ultimately each
severe autoignition event is damaging to the engine, regardless of whether or not they
occur relative to each other or if one occurs as a consequence of another. The
distribution of these “run-on” autoignition events is also random and would therefore
have no influence on the trend.
Figure 6.26 Autoignition Frequency in Percentage of Autoignition Cycles at Different Coolant Flow
Points
This phase of the test work also revealed no correlation between coolant flow
rate and the BLD point, no correlation between coolant flow rate and autoignition
frequency. It has also shown that there is no correlation between peak autoignition
cylinder pressure and coolant flow rate, contrary to what Figure 6.17 might have
suggested.
212
From this test work it would appear as though this cylinder head can be run at
a significantly lower flow rate than the 3 cylinder coolant pump currently delivers,
however, the cylinder liner was found to have cracked soon after completing this test
work. Although there is no conclusive evidence this failure was caused by this test
work, least of all by the reduction in coolant flow rate, it should be taken into
consideration for all future test work.
6.4 Impact of Cam Timing on Autoignition Tendency
6.4.1 Introduction
The single cylinder engine has been found to be much more prone to severe
autoignition than the 3 cylinder engine. This issue of severe autoignition has imposed a
heavy constraint on the maximum permissible load the single cylinder engine can
operate at, and has forced the implementation of a 26 bar NIMEP load limit when the
3 cylinder engine can achieve almost 32 bar NIMEP.
The fact that this issue is only present at low engine speeds suggests that LSPI
could be the cause of this issue. One of the intentions of this study is to determine
whether or not these severe autoignitions are the result of LSPI.
6.4.2 Test Plan
The main objective of this test work was to try and find the optimum cam
timing from an autoignition perspective, and whether or not there was a difference in
the type of autoignition event depending on the cam timing. This work has also been
carried out in a GT-Power simulation in an effort to better understand the mechanisms
behind this increased tendency to autoignite (section 6.4.4).
213
In order to find an optimum cam timing a cam timing sweep was carried out
which involved sweeping both inlet and exhaust cams through their full operable
ranges (IMOP 80–120 CAD ATDC, EMOP 100–135 CAD BTDC) in 5 CAD increments. This
was done to ascertain what impact valve overlap had on autoignition tendency and
whether or not the autoignition was being caused by residuals in the cylinder or some
other sort of deposit from a previous cycle.
The other test variables and their control criteria can be found in Table 6.7.
6.4.3 Results
The test work revealed potential hotspots for autoignition in terms of cam
timing combinations. Very severe autoignition or “superknock” (peak cylinder pressure
above 200 bar, knock intensity above 100 bar) cycles were encountered twice during
this phase of the test work. Plots of autoignition locations, peak magnitudes and
frequency are shown in Figures 6.27 and 6.28 respectively.
214
135
130
125
233-250
216-233
120 199-216
EMOP (CAD BTDC)

182-199
165-182
115 148-165
131-148
114-131
110
97-114
80-97
105
100
80 85 90 95 100 105 110 115 120
IMOP (CAD ATDC)
Figure 6.27 Peak Autoignition Cylinder Pressure
135
130
125
EMOP (CAD BTDC)

120 4-5
3-4
115 2-3
1-2
0-1
110
105
100
80 85 90 95 100 105 110 115 120
IMOP (CAD ATDC)
Figure 6.28 Frequency of Autoignition Events at Specific Cam Timings
Figures 6.27 and 6.28 reveal 2 potential autoignition hotspots, for the purposes
of this test work these hotspots will be broken down into “ridges” named as shown in
Figure 6.29.
215
Figure 6.29 Ridge 1 and Ridge 2 Definitions
These ridges encompass 87% of the autoignition events witnessed during this
test work, what is more they are also lines of constant valve overlap indicating that
valve overlap is of great importance when it comes to autoignition, and where this
overlap occurs with respect to piston TDC is of less importance.
A well known source of autoignition is residuals, therefore it may be logical to

assume that valve timings that offer less valve overlap will be more prone to
autoignition. This can be seen to an extent in Figure 6.29 as ridge 2 highlights an area
of very little valve overlap, and an autoignition event that falls just outside the
boundary of ridge 2 with an even lesser degree of valve overlap also shows a relatively
light autoignition event. Ridge 1 however highlights an area of large valve overlap
where the pressure differential between intake and exhaust will strongly encourage
scavenging. A GT-Power simulation was carried out to determine the impact on engine
breathing and knocking performance. The results of this study are presented in section
6.4.4.
216
As can be seen from Figure 6.27 superknock events can be found to occur
under both ridge 1 and ridge 2. Although they are both very severe autoignition
superknock events they can be found to be different when analysing the in-cylinder
pressure traces (Figure 6.30).
250
225
200
175
150
85 CAD ATDC IMOP,
125 115 CAD BTDC EMOP
(Ridge 1)
100 115 CAD ATDC IMOP,

125 CAD BTDC EMOP
(Ridge 2)
75
50
25
0
-15 -10 -5 0 5 10 15 20 25 30 35 40
Figure 6.30 Comparison of 2 Superknock Events, 1 Occurring Under Ridge 1, the Other Under Ridge 2
One of the superknock events occurs at a much earlier crank angle than the
other. In the case of Figure 6.30 this offset in superknock angle can be attributed to
the pressure history of the cylinder prior to autoignition (Figure 6.31). In both of these
cases the superknock cycles were caused by LSPI, as can be seen clearly from Figure
6.31. The ridge 2 case is very clearly caused by LSPI, while the ridge 1 case is still LSPI it
does not occur until later into the cycle.
217
100
90
80
70
85 CAD ATDC IMOP, 115

CAD BTDC EMOP (Ridge 1)
60
50 CAD BTDC EMOP (Ridge 2)
40 85 CAD ATDC IMOP, 115

CAD BTDC EMOP Spark
30
CAD BTDC EMOP Spark
20
10
0
-60 -50 -40 -30 -20 -10 0 10
Figure 6.31 Comparison of Pressure Histories for 2 Superknock Events, 1 Occurring Under Ridge 1, the
Other Under Ridge 2
There are a few potential explanations for the greater EOC pressure for the
cases encompassed by ridge 2. These include the fact that no EBP was applied at any
point during this test work, but the MAP was still relatively high. This means that
during the valve overlap phase only scavenging will take place, and with high degrees
of valve overlap (such as under ridge 1) the scavenging will be much greater than with
smaller degrees of valve overlap (such as under ridge 2). This scavenging will be
beneficial to the engine but will also result in slight accidental over enrichment, and
the engine running slightly rich under ridge 1, whereas the AFR will be much closer to
stoichiometric under ridge 2.
Fuel enrichment is well known to result in a reduction in polytropic exponent,

so for this test work the PIPR pegging technique was abandoned in favour of pegging
using the MAP instead (as described in section 3.3.1) to try and ascertain the extent of
this scavenging effect. The polytropic exponent can be seen to be lower for the high
valve overlap cases (Figure 6.32) implying that accidental over enrichment is occurring
as a result of the scavenging. The polytropic exponent values obtained from this are
somewhat higher than expected, it is suspected that the reason for this is the lack of
218
an intake pressure transducer when this work was carried out, the automotive style
MAP sensor had to be used instead. Although this is not ideal or particularly accurate
each of the cases were pegged using the same method and the same sensor, so the
trend shown should be accurate.
Figure 6.32 Polytropic Exponents
Lower EOC pressures will reduce the knocking propensity and the spark can be
advanced several CAD as a result of this. Figure 6.33 shows that the 50% MFB angle for
the points under ridge 1 is much more advanced than for the points under ridge 2. The
lower EOC pressure is the most probable explanation for this but the spark timing will
also be influenced by the EMOP timing. Late EMOP timings will reduce the degree of
valve overlap but they will also reduce the effective expansion ratio, this will have the
effect of reducing the size of the gross work contribution. This reduction of expansion
ratio will need to be compensated for by an increase in boost pressure thus effectively
shifting the engine to a higher load point (Figure 6.34). This will consequently require
the spark timing to be retarded.
219
Figure 6.33 50% MFB Angle (CAD ATDCF)
Figure 6.34 EMOP Sweep at Constant Load and IMOP Demonstrating the Impact of EMOP Timing on
GIMEP
220
Strangely the standard autoignition events under ridge 1 and ridge 2 echo this
trend (Figures 6.35 and 6.36) when perhaps the opposite trend would be expected and
might suggest that the autoignitions under ridge 1 and ridge 2 could be caused by
different mechanisms. Five of the autoignition events that occurred under ridge 1 were
very clearly caused by LSPI when spark timings are compared to in-cylinder pressure,
however, for the rest of the autoigniting cycles it was not possible to determine with
certainty the cause of autoigntion.
200 80 IMOP,
125 EMOP
180 80 IMOP,
135 EMOP
160 85 IMOP,
115 EMOP
85 IMOP,
140
115 EMOP
In-cylinder Pressure (bar)
85 IMOP,
120 115 EMOP
85 IMOP,
100 115 EMOP
95 IMOP,
80 105 EMOP
95 IMOP,
60 105 EMOP
95 IMOP,
105 EMOP
40
95 IMOP,
105 EMOP
20
100 IMOP,
110 EMOP
0
100 IMOP,
-15 -10 -5 0 5 10 15 20 25 30 35 40 110 EMOP
Figure 6.35 Autoignitions Under Ridge 1 (Units of IMOP and EMOP are CAD ATDC and CAD BTDC
Respectively)
221
200
180
160
100 IMOP,
135 EMOP
140
110 IMOP,
In-cylinder Pressure (bar)
130 EMOP
120
110 IMOP,
130 EMOP
100 115 IMOP,
125 EMOP
80 115 IMOP,
125 EMOP
60 120 IMOP,
120 EMOP
40
20
0
-15 -10 -5 0 5 10 15 20 25 30 35 40
Figure 6.36 Autoignitions Under Ridge 2 (Units of IMOP and EMOP are CAD ATDC and CAD BTDC
Respectively)
Zahdeh, et al [122] discovered that Fuel-Oil interactive effects were a cause of

LSPI, however, fuel enrichment had also been found to eliminate LSPI in the same
study. The exact degree of in-cylinder fuel enrichment was unknown in the study by
Zahdeh, et al, as it is unknown here, which may explain why it is still autoigniting under
ridge 1 but the autoignitions have stopped at the very corner point (IMOP 80 CAD
ATDC, EMOP 100 CAD BTDC). Figure 6.32 would imply that the charge is significantly
richer at the bottom left hand corner than under ridge 1, in spite of their close
proximity. The difference in fuel enrichment from the bottom left hand corner and
ridge 1 is also very apparent from a plot of exhaust gas temperatures (Figure 6.37).
Unfortunately no physical effective AFR data is available from this test work due to the
use of a 1-way air flow meter which gives unreliable readings (see section 3.2.5.2 for
details). The effective AFR has been analysed in the GT-Power work as shown in
section 6.4.4.
222
0
Figure 6.37 Exhaust Port Temperatures ( C)
6.4.4 GT-Power Analysis
6.4.4.1 Introduction
In order to better gauge what is happening in the engine a GT-Power analysis

was carried out using a modified version of single cylinder model 2 (see section 4.4)
which was modified to give closed loop control of NIMEP.
The test plan adopted for the GT-Power work was the same as for the
experimental test work. The IMOP and EMOP were changed in 5 CAD increments
between their respective limits (80–120 CAD ATDC IMOP and 100–135 CAD BTDC
EMOP). A full factorial DoE was used after disappointing results were obtained with a
Latin hypercube type DoE. Once the test had been carried out the response fitting
algorithm employed was a complex RBF emulator. This was found to give a response
most like that obtained in the physical test work, superior to all the polynomial models
223
(the polynomial models were found to give “lips” at the extremities of the maps,
whereas RBF had a lesser propensity to do this).
To compensate for the lack of closed loop combustion phasing control the 50%
MFB points obtained from the physical test work were used. These were mapped
against IMOP and EMOP angles. To further increase the accuracy of the results the 10-
90% MFB duration was also mapped with respect to IMOP and EMOP. At first the
average 10-90% MFB duration was used for all test points, however, it was later found
that the 10-90% MFB duration had a large impact on the quality of results, so it was
later mapped.
6.4.4.2 Results
The general pattern of the GT-Power data is fairly similar to the experimental
test results, however, it predicts the minimum ISFC region very accurately and clearly
has similar trends to the experimental data, as can be seen in Figure 6.38. As this ISFC
map corresponds so closely to the test rig data it has been assumed that the data from
the model will give a strong indication of the performance and trends of the
experimental engine, but absolute values will not be trusted.
The data obtained from this test work is therefore of use as the main purpose
of it is to ascertain what is happening in the cylinder in terms of residual gasses and
scavenging. It is also hoped that this model will be able to give some indication as to
what is causing the autoignition events.
224
Figure 6.38 Experimental (left) and Simulated (right) ISFC Comparison
Figure 6.39 is a plot of equivalent cylinder AFR which also factors in the air used
to scavenge the cylinder, and it confirms to an extent what was seen in the physical
test work. It clearly shows that the effective AFR under ridge 1 is a lot richer than
under ridge 2, although ridge 2 is still slightly rich of stoichiometric. Figure 6.40 is a plot
of residual gas concentration at cycle start. It confirms that the residual gas
concentration at cycle start is less for cycles that fall under ridge 1 as opposed to those
that fall under ridge 2.
Although this simulation is not reliable in terms of absolute values, it provides a

strong indication that the effective AFR is rich under ridge 1, and that there is a strong
correlation between ridge 1 and ridge 2 in terms of AFR. However, it provides no
obvious reason why the maximum valve overlap point should not also fall under ridge
1. The only reasonable explanation is that the charge cooling effect of the extra fuel
will reduce the EOC pressure and temperature considerably, which in turn will reduce
the tendency for autoignition to occur. The minimum difference in the 50% MFB point
between the maximum valve overlap point and ridge 1 (as can be seen in Figure 6.33)
is 1 CAD. The corresponding EOC pressure and temperature benefit afforded by this
enrichment/scavenging is shown in Figures 6.41 and 6.42 respectively.
225
Figure 6.39 Equivalent AFR at Cycle Start
Figure 6.40 Residual Gas Concentration at Cycle Start
226
Figure 6.41 EOC Pressure (bar)
0
Figure 6.42 EOC Temperature ( C)
227
An attempt was made to predict pre-ignition with the single cylinder knock
model. The reason why this was attempted is because the predicted KI is quite high in
the vicinity of ridge 1. There is also a point under ridge 2 where the intensity can be
seen to be very high. In order to test the models ability to predict pre-ignition all spark
values in the map were retarded 5 CAD, the plot of knock intensity was then analysed
to see if any of the points (particularly under ridge 2) had become incredibly high.
Previous experience with the single cylinder engine model has revealed that if the EOC
pressures and temperatures are too high the engine will “knock” before any of the
cylinder charge has actually been burned, this leads to an incredibly high knock
intensity (as knock intensity is calculated on the percentage of fuel unburned at the
time of knock). Results from this experiment revealed that the knock models ability to
predict pre-ignition is not effective at this running condition.
6.5 Effects of Fuel Age
6.5.1 Introduction
One of the biggest disadvantages for downsized engines and their worldwide
success is the requirement for very high RON fuels. In Europe gasoline with a RON of
95 and 98 and even higher still are widely available, however, the downsized engine
must also cater to the likes of large emerging markets such as China and India where
the RON values are generally lower and high RON fuels, though available, are generally
less commonplace.
With China now having the largest automotive market in the world it is of great
importance that all future downsized engines can run on Chinese grade fuels so as not
to compromise the downsized engine’s global success.
In an effort to try and assess the sensitivity of downsized engines to different

fuel types, the single cylinder engine was run with 2 samples of pump ULG RON 95.
One of these samples was a year old, the other was fresh.
228
Over time the lighter components can be found to evaporate which generally
has the effect of decreasing RON. Unfortunately the facility to analyse both the aged
and fresh fuel chemically was unavailable at the time this test work was carried out.
6.5.2 Test Plan
This test work consisted of running the engine at steady load for 15 minutes
with the 2 different fuels, one being aged (12 months) ULG RON 95 and the other
being fresh (<1 month old) ULG RON 95, and observing the autoignition frequency. The
test criteria was as shown in Table 6.8.

Fuel Aged and fresh ULG RON 95, no temperature control
Speed 2000rpm
Loads 24 bar NIMEP
Intake cam Standard 240 CAD profile
IMOP 100 CAD ATDC
EMOP 120 CAD BTDC
Spark timing BLD
Exhaust Pressure not regulated, temperature control not required
Test duration 15 minutes
6.5.3 Results
The fresh fuel resulted in a considerable decline in autoignition frequency. Over

the course of 15 minutes 6 autoignition cycles were observed with the aged fuel and 3
were observed with the fresh fuel. A comparison of the autoignition cycles obtained
with the aged and fresh fuels is shown in Figures 6.43 and 6.44 respectively. Six cycles
were obtained with the fresh fuel because the test was run twice.
229
300
250
200
1
2
3
150
4
5
6
100
7
50
0
-15 -10 -5 0 5 10 15 20 25 30 35 40
Figure 6.43 Autoignitions with Aged ULG95
300
250
200
1
2
150 3
4
5
100 6
50
0
-15 -10 -5 0 5 10 15 20 25 30 35 40
Figure 6.44 Autoignitions with Fresh ULG95
The maximum pressure and intensity can be seen to be fairly similar for both
fuels (with the exception of 1 autoignition cycle with the aged fuel). However, the peak
cylinder pressure is reached earlier with the aged fuel from the autoignition events
230
observed. This suggests that the aged fuel had a greater propensity for LSPI in
comparison to the fresh fuel. The very high peak cylinder pressures appear to be
caused by a different mechanism with the fresh fuel which was very severe knock,
although from the data obtained it is not possible to say with certainty whether LSPI or
knock was the cause of the autoignition events. One potential reason for the later peak
cylinder pressures with the fresh fuel is that over time fuel can react with oxygen to
create long chain olefins and diolefins. Sasaki, et al [123] documents that owing to
their size, long chain olefins and diolefins are more likely to autoignite as they have an
overall larger and correspondingly weaker structure.
Other than the considerable reduction in autoignition frequency there were no

considerable differences between the different fuels. The 50% MFB point could be
advanced 0.4 CAD with the fresh fuel but there was no difference with the 10-90%
MFB duration. The aged fuel exhibited a 0.81% improvement in ISFC, in spite of the
fact the BLD point was retarded in comparison to the fresh fuel. The exhaust
temperature was also 140C lower for the aged fuel, which is where the improvement in
ISFC might have come from.
The COV of IMEP was slightly higher for the aged fuel but not significantly.
Unfortunately no emissions data was obtained from this test work.
231
7
Chapter 7
Experimental Studies of Miller Cycle on Engine

Performance and Fuel Consumption
7.1 Introduction
The purpose of this chapter is to explore the benefits that can be yielded
through the adoption of the Miller cycle on the single cylinder experimental engine.
This chapter has been arranged into 2 major sections. The first section is a
study of the benefits that can be yielded through the de-throttling effect of the Miller
cycle at low loads. The second section of this chapter is an exploration of the
autoignition mitigating effects of Deep Miller at low speed and high load.
The components that bring about a benefit over the standard Otto
configuration of the engine have also been explored, as well as other phenomena that
have been found to influence engine performance. These phenomena include engine
breathing, cam timings and combustion phasing observations. The feasibility of
supercharging/turbocharging with Deep Miller has also been investigated.
232
7.2 Achieving EIVC and LIVC
In order to carry out this work the experimental engine (described in section
3.2.1) had to be fitted with different geometry intake cams. In total 3 different
duration intake cams were tested. The cam durations were defined as the number of
CAD between the End Of Ramps (EORs). The first was a short low-lift cam with an EOR
to EOR duration of 152 CAD and a maximum lift of 5mm. The second was the standard
Mahle intake cam with an EOR to EOR duration of 240 CAD and a maximum lift of
11mm. The third was a standard-lift long cam with an EOR to EOR duration of 292 CAD
and a maximum lift of 11mm. Figure 7.1 shows an overlay of these 3 cams in their
maximum overlap positions.
12
10
8 Short
(Max
overlap)
Valve lift (mm)
Standard
6 (Max
overlap)
Long
4 (Max
overlap)
0
-60 0 60 120 180 240 300
Engine crank angle (CAD ATDC)
Figure 7.1 Cam Profile Comparison (as Shown in Their Maximum Valve Overlap Positions)
These were all tested with the standard Mahle exhaust cam with an EOR to
EOR duration of 278 CAD and a maximum lift of 11mm. The geometry of all of the
cams is given in Appendix A.1.
The cams can also be phased by up to 40 CAD to reduce the valve overlap.
Figure 7.2 shows the minimum and maximum limits of the cam phasing.
233
12
10 Short
(Max overlap)
Standard
8 (Max overlap)
Long
Valve lift (mm)
(Max overlap)
6 Exhaust
(Max overlap)
Short
4 (Min overlap)
Standard
(Min overlap)
2 Long
(Min overlap)
Exhaust
0 (Min overlap)
-300 -240 -180 -120 -60 0 60 120 180 240 300
Engine crank angle (CAD ATDC)
Figure 7.2 Cam Phasing Ranges
Using the cam phasers the effective CR can be reduced to as low as 5.57 and
6.16 (as calculated from the 1mm lift point) with the 152 CAD duration cam and 292
CAD duration cam respectively. All cams were swept across their respective ranges in
10 CAD increments for all test work.
7.3 Description of Test Points
The Miller cycle performance benefits were assessed in the following 2 regions
(Figure 7.3)
234
Figure 7.3 Regions of Miller Cycle Studied
These regions can be broken down as follows:
Region 1: De-throttling effect at part-load conditions
De-throttling is the technique of using the intake valves to throttle the air flow
to the engine instead of a conventional throttle. The advantage of this is the pumping
losses can be reduced by elevating the manifold pressure and then implementing
either EIVC (stopping the air from going into the cylinder during the induction stroke)
or (LIVC exhaling charge from the cylinder during the compression stroke) to regulate
the mass of air in the cylinder.
The test points in Table 7.1 were selected to study the de-throttling effect on
part-load fuel economy. The minimum load is constrained by engine misfire below
certain loads.
235
Table 7.1 Test Points for De-throttling Studies (BMEPs are the 3 Cylinder Equivalent Values)
Speed (rpm) BMEP (bar) NIMEP (bar) EBP valve position (% closed)
1000 4 4.56 0
1000 8 8.83 0
2000 4 4.64 0
2000 8 8.90 0
3000 4 4.80 0
3000 8 8.98 0
4000 4 4.98 0
Region 2: Effective CR reduction for autoignition mitigation at full-load and low speed
operations
In this region EIVC and LIVC will be employed to reduce the effective CR, this
will allow the combustion to be phased closer to MBT without autoignition occurring.
In this region even at maximum load no fuel enrichment will be required, but owing to
the low speed of the engine there will be a greater propensity for autoignition to take
place. It is in this region that the effectiveness of EIVC is questioned as it has been
found to reduce in-cylinder charge motion considerably at lower loads. Table 7.2 gives
the test points for region 2.
Table 7.2 Test Points for Region 2 (BMEPs are the 3 Cylinder Equivalent Values)
Speed (rpm) BMEP (bar) NIMEP (bar) EBP valve position (% closed)
2000 15 16.04 0
2000 15 16.04 45
2000 24 25.58 0
2000 24 25.58 * 2 bar EBP
* - EBP in bar, not valve position
Due to rig constraints this test plan had to be truncated considerably from the
original intended test plan that included some high speed and high load sites. The
engine was found to be incapable of running at higher speeds and loads due to the
insufficient coolant flow rate of the coolant pump.
A list of the test variables and control criteria are given in Table 7.3.
236

Fuel Fresh ULG RON 95, supply temperature regulated to +/- 5 degC of 25 degC
Speed See Tables 7.1 and 7.2
Loads See Tables 7.1 and 7.2
Intake cam Short 152 CAD profile, standard 240 CAD profile and long 292 CAD profile
IMOP See Table 7.4 (swept in 10 CAD increments)
EMOP 100 - 140 CAD BTDC (swept in 10 CAD increments)
Spark timing BLD or MBT depnding on load
Exhaust For EBPs see Tables 7.1 and 7.2. Temperature control not required
The IMOP timing is dependent on the cam being tested. Table 7.4 gives the
ranges of IMOPs for the 3 different cams and their respective IVO and IVC points. Also
shown are the exhaust EMOP ranges and corresponding EVO and EVC points.
Table 7.4 MOPs and Corresponding Valve Opening/Closing Points for Each of the Cams
Maximum overlap phasing Minimum overlap phasing

Cam MOP IVO/EVO (1mm lift) IVC/EVC (1mm lift) MOP IVO/EVO (1mm lift) IVC/EVC (1mm lift)
152 CAD Intake 35 CAD ATDC -18 CAD ATDC 87 CAD ATDC 75 CAD ATDC 22 CAD ATDC 127 CAD ATDC
278 CAD Exhaust 100 CAD BTDC 215 CAD BTDC 16 CAD ATDC 140 CAD BTDC 255 CAD BTDC -24 CAD ATDC
7.4 Low Load Operations with EIVC and LIVC (Region 1)
7.4.1 Introduction
Low load points can typically be found to be less efficient than medium to high
load points. This is due to the problem of pumping losses at low load being of a similar
magnitude to the gross power output. Downsizing an engine has the effect of reducing
these pumping losses relative to the gross power by running at what is effectively a
higher load point (load point shifting). There is still scope for improving this further still
by throttling with the valve instead of a throttle. The scope of this benefit has not been
investigated before with an aggressively downsized engine with a very high degree of
load point shifting.
237
The objective of this portion of the work is to assess the extent of the benefit of
throttling with the valve at low load, and what impact this has on the size of the
pumping loop.
The single cylinder engine is relatively limited in how low a load it can run at
satisfactorily, and the minimum load it can sustain for all IMOP and EMOP
combinations was 4 bar BMEP. The test work involved carrying out IMOP and EMOP
sweeps at speeds of 1000, 2000 and 3000rpm at loads of 4 and 8 bar BMEP, and
4000rpm and 4 bar BMEP, as shown in Table 7.1.
7.4.2 Results
Comparisons of the optimum cam timing combinations at each speed and load
site for each cam are shown in Figure 7.4. As stated previously, BMEPs are the 3
cylinder equivalent values.
320
310
300
290
Short cam
(152 CAD)
280
ISFC (g/kWhr)
270 Standard cam

(240 CAD)
260
250 Long cam

(292 CAD)
240
230
220
1000 1000 2000 2000 3000 3000 4000
4 bar 8 bar 4 bar 8 bar 4 bar 8 bar 4 bar
Engine speed (rpm)
Load (BMEP)
Figure 7.4 ISFC Comparison of Short, Standard and Long Cams at Low Loads
238
The short cam (152 CAD duration) exhibits the lowest ISFC values for every
speed and load combination, and the long cam (292 CAD duration) showed the highest
values typically. There is a slight break from this trend at 1000rpm where the standard
cam (240 CAD duration) exhibits the highest ISFCs.
One possible explanation for this is the fact that the effective CR and the
volumetric efficiency are highest for the standard cam. This therefore results in the
standard cam requiring the lowest MAP of the 3 different cams, and in a GT-Power
simulation carried out reveals that this results in the largest back flow of exhaust gas to
the intake of the 3 cams during the valve overlap period. As a consequence of this the
residual gas concentration at cycle start for the standard cam can be seen to be the
largest of the 3 cams.
This is not detrimental in itself but very high residual gas concentration can
have the effect of reducing the flame front speed (at this speed and load) which will
make cycles less repeatable. As a consequence of this lambda control is also made
more difficult, particularly for the single cylinder engine where the lambda sensor is
located 800mm downstream of the exhaust port.
The residual gas concentration at cycle start according to the GT-Power model
is more elevated for the standard cam than for the other 2 cams. However, the short
and long cams still exhibit elevated levels of residuals in the cylinder and still suffer
from very high COVs, this is therefore not (on its own) a suitable explanation for the
discrepancy seen with the standard cam at these 1000rpm points.
It should be noted that of the standard cam tests the 1000rpm points were the
last to be done before a blown head gasket was observed. The head gasket failure
allowed combustion gasses to leak past the gasket and pressurise the coolant circuit.
This manifested itself on the instrumentation as a low coolant flow rate. The coolant
flow rate for these points was approximately 4% lower than for other test points,
however, gas leakage would manifest itself on a P-V diagram by way of a reduced
polytropic exponent during the expansion stroke. No appreciable difference can be
seen between the short, long and standard cams in polytropic exponent during the
expansion stroke (Figure 7.5), and the head gasket issue was rectified for the short and
239
long cams. This is also not a satisfactory explanation on its own for the break in ISFC
trend at 1000rpm.
Figure 7.5 1000rpm, 8 bar BMEP Short, Standard and Long Cam Overlay
There are several reasons why the short cam exhibits lower ISFCs at every
single point. The short cam gave the minimum pumping loop size of the 3 cams for
every single point, as can be seen in Figure 7.6 (short cam PMEP values are closest to
zero).
240
Figure 7.6 PMEP Comparison of Short, Standard and Long Cams at Low Loads
The minimisation of the pumping loop is achieved by the effective elimination

of part of the pumping loop as can be seen in Figure 7.7. The in-cylinder pressure for
the short cam can be seen to start reducing (through polytropic expansion) at a
volume ratio of approximately 0.6. From this point on therefore the cylinder effectively
becomes a closed system and the mass of air trapped in the cylinder is fixed. The rest
of the intake stroke is therefore polytropic expansion and is “cancelled out” of the
pumping loop by the polytropic compression that occurs during the compression
stroke (although not quite due to charge heating).
241
Figure 7.7 4000rpm, 4 bar BMEP Short, Standard and Long Cam Overlay
In Figure 7.6 the pumping loop size was determined from BDC to TDC to BDC
rather than calculating the size of just the pumping loop itself, between the
intersections of the lines (as can be seen in Figure 7.7).
Although both EIVC and LIVC reduce the pumping loop size, EIVC reduces it to a
greater extent than LIVC, even for the same effective CR. It must be stressed that
reducing the pumping loop size does not necessarily mean reducing pumping losses.
Reducing effective CR requires a higher MAP to ensure the correct mass of air is
trapped in the cylinder to make the required load, it is this principle that (at low loads)
reduces the vertical size of the pumping loop. EIVC offers a further decrease in
pumping loop size by reducing the horizontal size of the pumping loop and in the case
of Figure 7.7 leaving the horizontal portion of the pumping loop between a volume
ratio of 0.6 and 1 almost entirely unpopulated. It is for this reason that the short cam
line in Figure 7.6 is significantly flatter than the other 2 cams. While LIVC reduces the
horizontal size of the pumping loop somewhat, the portion between a volume ratio of
0.6 and 1 is still partially populated.
Restricting the geometric size of the pumping loop is not always optimum for
higher load cases where the MAP is above the EBP. In these cases the pumping loop
242
contribution becomes negative and is beneficial to the cycle. It would therefore be
logical to assume that the larger the pumping loop the better in these cases. Although
a larger negative pumping loop is theoretically beneficial to the cycle (because the
gross power contribution can be made smaller as a consequence of this), in the cases
where the negative pumping loop is very large for the long cam in comparison to the
short and standard cams (1000rpm, 8 bar BMEP and 2000rpm, 8 bar BMEP for
example), this does not translate to a lower ISFC. Potential reasons for this are given in
section 7.4.2.1.
Although the pumping loop is comparatively large at these speed and load
points its impact on ISFC is quite conservative. The peaks and troughs in Figure 7.6
correspond quite well to the profiles of the lines in Figure 7.4, however the difference
in ISFC brought about by the change in pumping loop size is much smaller than the
theoretical gain obtained from the change in pumping loop size (see section 7.4.2.3).
The most likely explanation for this is that at these low load points the efficiency
gained through the decrease in power loop size (to maintain constant NIMEP) is offset
by a low EOC temperature and pressure and a corresponding degradation of
combustion efficiency.
The long cam demonstrating the highest ISFCs is an unexpected result. Some
literature suggests that LIVC is not as effective at reducing ISFC as EIVC at low load, but
cases of LIVC performing this much worse than the standard cam are unheard of. This
is suspected to be a result of differences in charge motion exhibited by the different
cams (see section 7.4.2.1) leading to incomplete combustion for the long cam.
According to GT-Power studies no charge scavenging will take place at any of the
speed and loads tested for this portion of the test work. Charge will however still be
exhaled to the intake manifold during the compression process, this will likely contain
some fuel. Figure 7.8 is a plot of intake port temperatures for each of the cams and
speed and load points, and demonstrates the presence of residual gasses and charge in
the intake manifold for the long cam by elevated temperatures.
243
90
85
Intake port temeprature (deg C) 80

Short cam
(152 CAD)
75
70
Standard cam
(240 CAD)
65
Long cam
60
(292 CAD)
55
50
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.8 Intake Temperature Comparison of Short, Standard and Long Cams at Low Loads
7.4.2.1 Discrepancy at 1000rpm
The 1000rpm 4 bar BMEP test point (with all 3 cams) has been repeated in a
GT-Power simulation to give a direct comparison to the single cylinder experimental
engine. The GT-Power simulation was found to be very accurate at predicting engine
breathing dynamics but very inaccurate at predicting the ISFC trends observed.
One potential reason for this was the difference in effective CRs. In order to
better gauge what the theoretical thermal efficiency should be for each of the cams an
approximation was made based on the data retrieved from the GT-Power model.
Relative to the standard cam the long cam (with the second highest degree of effective
CR reduction) is 1.90% more efficient, and the short cam is 6.06% more efficient. This
trend is consistent for both of the 1000rpm cases with the experimental test data,
however, the actual improvement in efficiency was far greater than the theory
stipulates. The long cam returned an ISFC benefit of 3.73% relative to the standard
cam, and the short cam returned an ISFC benefit of 12.55% relative to the standard
244
cam. The GT-Power model presented the greatest efficiency disparity by predicting a
0.08% decline in efficiency for the long cam relative to the standard cam, and a 3.14%
decline in efficiency for the short cam relative to the standard cam.
Another notable inaccuracy is the required MAP for the long cam. GT-Power
predicts an MAP of 0.9 bar whereas the physical engine required an MAP of 1.096 bar.
This discrepancy is considerable, as well as the discrepancy between the ISFCs for the
long cam (247 g/kWhr in GT-Power Vs. 292 g/kWhr for the experimental engine).
The suspected reason for all of these discrepancies is the in-cylinder charge
motion during the compression stroke, and the difference in the quantity of residuals
present. The impact of neither charge motion nor residuals can be simulated in GT-
Power, changing the properties of the Wiebe combustion model is as close as it is
possible to get.
The charge motion can be expected to differ for each of the different cams. The
short cam for instance combines a very low lift with a very short duration whereas the
long cam exhibits a high degree of lift for a longer duration. The impact of this on the
mass flow rate through the intake valves is considerable (Figure 7.9).
Figure 7.9 Comparison of the Intake Valve Mass Flow Rates for the 3 Different Duration Cams at
245
The short cam can be found to exhibit peak mass flow rates greater than both
the standard and the long cams, even with a reduced valve lift. This will result in the
flow velocity through the valves being far greater for the short cam than for the other
2 cams. High flow velocity can be found to result in very high levels of TKE but
relatively low levels of bulk flow such as tumble (Matsumoto, et al [66]). The higher
TKE will result in a more homogeneous charge but the lack of bulk flow will reduce
overall charge motion and increase the burn duration. The standard and long cams
however have high levels of lift which will promote bulk flow but drastically reduce the
TKE. As a result of this the overall charge motion will be relatively high but the charge
itself will be relatively in-homogeneous. Figure 7.10 supports this theory and shows
the difference in the MFB duration for the 3 different cams.
Figure 7.10 MFB Profiles for the 3 Different Cams at 1000rpm and 4 bar BMEP
Both the standard and long cams exhibit very similar profiles which would
imply the in-cylinder flow is very similar for both cams. Both the standard and long
cams have very similar 10-90% MFB durations too, with durations of 25.2 CAD, 20.9
CAD and 20.4 CAD for the short, standard and long cams respectively. The only point
where the profiles can be seen to differ is at the very high MFB points (~0.95 MFB)
very late into the expansion stroke where the total fuel mass burned for the long cam
246
fails to reach the full 100%. In reality 100% MFB cannot be achieved but the short and
standard profile get much closer to full 100% MFB than the long cam. The most likely
reason for this is that the majority of the charge expelled to the intake during the
compression stroke with the long cam is air and that it contains relatively little fuel,
this would lead to enrichment in the cylinder and an excess of fuel.
In order for this to occur the HC emissions must be elevated, which they are
not markedly for the long cam, however, the high quantity of residuals present at cycle
start will contain the majority of the unburned HCs from the previous cycle, which will
be burned during the current cycle. This will cause the HC emissions to appear lower
than they effectively are.
The COV of IMEP can also be observed to be higher for the standard and long
cams (Figure 7.11). Higher COV values can be indicative of increased residuals in the
cylinder, they can also be indicative of higher degrees of tumble motion in some
circumstances (Zhao, et al [96]). Another potential reason is the reduced EOC
temperature associated with the use of EIVC and LIVC. This is a valid explanation for
increased COV and decreased combustion stability, but if low EOC temperature was
the main reason for it it would have manifested itself to a greater extent with the short
cam than it has.
Figure 7.11 Comparison of COV for 1000rpm and 4 bar BMEP for Short (left), Standard (middle) and
Long (right) Cams
Most of the above factors are present for all engines running with EIVC and
LIVC, however, this is the first time on average that LIVC has returned a considerably
lower thermal efficiency than the standard cam to the author’s best knowledge. This
would suggest that it is one of the above issues or a separate issue that is present on
247
this single cylinder engine. The fact that this engine is a single cylinder as opposed to a
multi-cylinder engine (as used for the majority of the other studies) could change the
behaviour slightly, as well as the intake and exhaust tuning dynamics that had such a
great impact on the results shown in chapter 5.
The disparity mentioned earlier about the MAP requirement for the single
cylinder model being so different to that required for the single cylinder engine
indicates that the engine breathing is having an effect, even at low speeds. It has also
been found that the minimum load achievable with the single cylinder engine is 4 bar
BMEP because below this misfiring occurs.
The reason for this was investigated, and a back to back GT-Power study of the
single cylinder and 3 cylinder engines revealed that residuals are massively elevated as
a result of the engine breathing differences. For the same degree of valve overlap at
the 1000rpm, 4 bar BMEP load point the residuals were found to be 49.5% higher for
the single cylinder engine. At 2000rpm and 4 bar this can be seen to drop to 34.1%
higher for the single cylinder engine. As the standard cam exhibits the greatest
quantity of residuals this is the most likely reason why the single cylinder engine
performed so poorly with the standard cam at 1000rpm in comparison to the other 2
cams.
7.4.2.2 Emissions
Data has been obtained for both HCs and NOx emissions at all test points. The
main objective of recording this emissions data is to assess the differences in
combustion and using it as an additional tool to determine what is happening in the
engine rather than having a mind to reduce them. For this test work the injection
timing and pressure was fixed to the standard 3 cylinder values at the equivalent
running conditions. Analysis of more emissions would have been required if this had
been altered but, as it was not, HCs and NOx were deemed adequate. Data obtained
from the 3 cylinder also suggests that other emissions such as CO, O2, CO2 and smoke
number were all fairly constant over this speed and load region. Smoke emissions may
248
have proven useful, and attempts were made to monitor smoke number, however, the
smoke emissions were so low that an AVL 415 smoke meter was unable to reliably
collect data from the engine. Smoke emission monitoring was therefore abandoned.
HC emissions and NOx emissions are shown in Figures 7.12 and 7.13
respectively.
0.05
0.045
0.04
Short cam
HC emissions (g/kWhr)
(152 CAD)
0.035
0.03
Standard cam
(240 CAD)
0.025
0.02 Long cam

(292 CAD)
0.015
0.01
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.12 HC Emissions Comparison of Short, Standard and Long Cams at Low Loads
249
0.07
0.06
0.05
Short cam
NOx emissions (g/kWhr)
(152 CAD)
0.04
Standard cam
0.03 (240 CAD)
0.02
Long cam
(292 CAD)
0.01
0
1000 1000 2000 2000 3000 3000
4 bar 8 bar 4 bar 8 bar 4 bar 8 bar
Engine speed (rpm)
Load (BMEP)
Figure 7.13 NOx Emissions Comparison of Short, Standard and Long Cams at Low Loads
Figure 7.12 shows a considerable increase in HC emissions for the 1000rpm

points for all cams, which is caused by a decline in combustion stability at this speed.
With the exception of 1000rpm the HC emissions stabilize for all other speeds with the
standard cam in all cases returning the fewest HC emissions. This is most likely due to
the fact that the standard cam exhibited the highest quantity of residual gasses at
cycle start of the 3 cams, thus providing the greatest opportunity to burn off any HC
emissions from the previous cycle. The shortest cam would be expected to have the
highest HC emissions because it has consistently the lowest quantity of residual gasses
at cycle start, a reduced EOC temperature in comparison to the standard cam, and it
has the least opportunity to burn off any HC emissions from previous cycles.
Studies with the 3 cylinder engine show that HC emissions should generally
decline with higher loads (Figure 7.14). This trend cannot be seen in Figure 7.12, in fact
the long cam actually demonstrates the opposite trend to this. This is likely to be due
to the engine breathing and the fact that at higher loads there is considerably less
residual gas concentration at cycle start due to the higher MAP inhibiting back flow.
250
0.07
0.06
0.05
Emissions (g/kWhr)
0.04
3 cylinder HC
emissions
0.03
3 cylinder N0x
0.02 emissions
0.01
0
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.14 HC and NOx Emissions as Obtained from Test Work Carried out by Mahle Powertrain
The HC emissions can also be observed to be low in comparison to the 3

cylinder engine. The exact reason for this is unknown but is most likely to be due to
higher rates of internal EGR for the single cylinder engine. The absolute numbers
output from the FID analyser were not trusted for this reason, the values relative to
each other and trends observed are likely to be accurate however.
The NOx emissions provide a better indication of the residual gas concentration
at cycle start with the short cam being consistently higher than the other 2 cams. This
is an unexpected result as the literature suggests NOx emissions should be reduced
with lower EOC temperatures, however, the residual gas concentration at cycle start
with the standard cam and long cam will inhibit NOx formation considerably, and this
is the most likely explanation for this unexpected result.
The overall trends are also more like what was observed with the 3 cylinder
engine in that higher load increases NOx emissions drastically. The reason for this is a
reduction in residual gas concentration at cycle start in combination with higher EOC
temperatures leading to higher combustion temperatures.
251
With such an elevated MAP requirement (Figure 7.15) for the long cam the NOx
emissions could be expected to be slightly higher than they are (due to a higher EOC
temperature) but it is important to remember that in the case of the long cam the
MAP can be quite misleading because the values given in Figure 7.15 correspond to the
average pressure across the 4-stroke cycle, not the pressure whilst the valve is open
which is of greatest importance, the same applies to the short cam MAP requirement
too. The flow of charge from the cylinder to the intake will also distort the MAP
slightly. The theoretical MAP requirement for the long cam is approximately 15%
greater than what is required for the standard cam, so the NOx emissions should be
15% higher but the peak combustion temperature is lower for the long cam due to the
lower effective CR, so the fact that both the standard cam and long cam overlay each
other closely is coincidental and not to be unexpected.
1.6
1.4
1.2
Short cam
(152 CAD)
MAP (bar)
1
Standard cam
(240 CAD)
0.8
Long cam
(292 CAD)
0.6
0.4
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.15 MAP Comparison of Short, Standard and Long Cams at Low Loads
252
7.4.2.3 Analysis of the Individual Components of EIVC and LIVC
This section contains a breakdown of the individual components that cause an

increase or decrease in the overall efficiency for both EIVC and LIVC in comparison to
the standard cam. The 3000rpm and 4 bar BMEP point was analyzed in this section
because it is arguably the most typical point recorded without any nuances.
The theoretical efficiency was calculated on a theoretical cycle basis rather

than through the use of a formula. The reason for this is the formulas quoted for the
Atkinson cycle process such as those given in Stone [30] and Heywood [21] calculate
the thermodynamic efficiency for a 2-stroke cycle and not a 4-stroke cycle.
A major limitation of the equations given in Stone and Heywood is that the
system is closed throughout the entire cycle, otherwise by definition it cannot be
considered a cycle. This means that the isobaric processes must result in a change of
temperature (according to the perfect gas law), this change of temperature is directly
proportional to the volume ratio swept. When the cycle is considered open for the
duration of the isobaric processes (as is the case with a 4-stroke engine) no
temperature change takes place.
What this means is the thermal efficiency of a realistic Atkinson cycle can be
calculated using the same formula used to calculate Otto cycle efficiency (Eq.7.1) given
by,
7.1
where T1, T2, T3, and T4 are temperatures at the respective positions shown in Figure
7.16.
253
Figure 7.16 Theoretical Temperatures at Various Points in the 4-Stroke Atkinson and Otto Cycles
For a constant heat input (as is the case in Figure 7.16) the Atkinson cycle can
be seen to yield a benefit by virtue of the fact T4 is reduced.
Stone and Heywood both neglected the pumping loop contributions in their
derivations, this limits the efficiency calculation to the gross work of the cycle only. At
low loads the pumping contributions will be considerable and they cannot be
neglected.
The only way to incorporate the pumping losses is to calculate the entire cycle
manually. The magnitude of the pumping losses will also affect the size of the gross
contribution of the cycle (if the net work is kept constant).
A comparison of the 2-stroke and 4-stroke Miller/Atkinson cycles is shown in

Figure 7.17. Both cycles depict the same theoretical gross power output with realistic
MAP.
254
Figure 7.17 Comparison of 2-Stroke (left) and 4-Stroke (right) Theoretical Atkinson Cycles
The biggest difference between the 2-stroke and 4-stroke processes is the
isobaric processes. The isobaric processes refer to the processes that take place at
constant pressure. In the 2-stroke cycle there is just 1 very small isobaric process that
occurs before the polytropic compression process. In the 4-stroke cycle there are 3
isobaric processes at 2 different in-cylinder pressures, this creates a work contribution
that is not present in the 2-stroke case.
In the following section the net work of the cycle was kept constant which
required the gross work to change depending on the size of the pumping loop. A
reduction of the gross work was instigated by changing the theoretical mass of air
trapped in the cylinder at cycle start according to Eq.6.1. This was done by changing
the MAP. The ∆T was associated with the heat addition process of combustion was
kept constant for all theoretical work, so the only change in heat input was through a
change of mass alone.
The theoretical benefits offered by the use of EIVC are shown in Figure 7.18,
these can be compared to the actual benefits observed from the test work (Figure
7.19).
255
Theoretical benefit with EIVC (%)
Total
Over-
expansion Theoretical
benefit (%)
Pumping
loop
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Figure 7.18 Theoretical Benefit with EIVC at 3000rpm and 4 bar BMEP
Actual benefit with EIVC (%)
Total
Breathing/combustion
Reduction in bulk flow Actual

benefit (%)
Over-expansion
Pumping loop
-6 -4 -2 0 2 4 6
Figure 7.19 Actual Benefit with EIVC at 3000rpm and 4 bar BMEP
The mechanisms that have influenced the theoretical efficiency have been
listed in Figure 7.18 where “Pumping loop” corresponds to the benefit associated with
256
a reduction in pumping loop size. “Over-expansion” refers to the benefit associated
with running with over-expansion as opposed to the standard Otto cycle. The over-
expansion value is influenced by the pumping loop size because constant NIMEP was
maintained. In order to factor the influence of the pumping loop out of the gross work
component, the influence of over-expansion was calculated at a constant GIMEP point,
or by subtracting the “Pumping loop” component from the “Total” component. The
“Total” value is calculated based on the thermal efficiency as calculated from the
theoretical open system 4-stroke Atkinson cycle, or the 4-stroke Otto cycle efficiency.
For the actual data (Figure 7.19) “Pumping loop” refers to the difference in
PMEP between the standard and the short cam as derived from the experimental data.
“Over-expansion” was calculated by subtracting the experimental pumping loop value
from the theoretical thermal efficiency as calculated from the theoretical open system
4-stroke Atkinson cycle (4-stroke Otto cycle) analysis. The “Reduction in bulk flow”
refers to the influence of the increased burn duration associated with EIVC. It has been
called the reduction of bulk flow because that is the theorized cause for the slow burn
(see section 7.4.2.1). It has been calculated using GT-Power. The
“Breathing/combustion” component is the efficiency benefit associated with
combustion deterioration caused by either low EOC pressure and temperature,
residuals, low in-cylinder charge motion, or a high COV. This has been calculated by
adding the other components and subtracting this from the “Total” which has been
calculated as the difference in ISFC for the standard and short cam derived from the
experimental data.
The influence of pumping loop size and over-expansion can be seen to be

roughly equal in size. The over-expansion component is of the same magnitude as the
pumping loop contribution due to the fact that the effective CR is so low. The over-
expansion component can be expected to reduce in size with higher effective CRs
approaching the geometric CR.
The pumping loop component is entirely dependent on load and will not
necessarily decrease with increasing effective CR, as can be seen in Figure 7.6. In this
257
example the pumping loop component is very large for the standard cam due to the
fact the engine is being run at very low load.
The reduction in bulk flow can be seen to be quite inconsiderable in

comparison to the loss associated with the engine breathing/combustion deterioration
according to Figure 7.19. However, it was not possible to determine precisely how
much of the combustion deterioration was caused by the reduction of bulk flow,
likewise, it was not possible to determine accurately what effect an increase in TKE had
on the overall efficiency either.
The theoretical gain in efficiency can be seen to match the actual gain in
efficiency quite closely. This is likely to be more coincidence than proof that the 4-
stroke Atkinson cycle analysis is an accurate measure of the efficiency of EIVC and
LIVC, as the components of pumping and over-expansion can be expected to be far
different in reality due to engine breathing behaviour.
Figures 7.20 and 7.21 show the theoretical and actual advantages of LIVC
(compared to the standard cam).
Theoretical benefit with LIVC (%)
Total
Over-
expansion Theoretical
benefit (%)
Pumping
loop
0.00 0.50 1.00 1.50 2.00 2.50 3.00
Figure 7.20 Theoretical Benefit with LIVC at 3000rpm and 4 bar BMEP
258
Actual benefit with LIVC (%)
Total
Actual
benefit (%)
Over-expansion
Pumping loop
-10 -8 -6 -4 -2 0 2 4
Figure 7.21 Actual Benefit with LIVC at 3000rpm and 4 bar BMEP
The pumping loop contributions can be seen to be markedly different for LIVC.
The predicted contribution is a lot higher than the actual contribution. This is due to
the fact that the long cam at this point requires an MAP similar to the standard cam,
thus resulting in similar pumping loop sizes, whereas the MAP requirement in theory
should be 0.1 bar lower for the standard cam.
The combustion deterioration that affects LIVC can also be seen more clearly
here and is accountable for an ISFC increase of 9%. This could likely be improved to a
level similar to the level obtained with EIVC if the maximum valve lift with the long cam
were reduced to encourage more turbulent mixing, although the lack of bulk flow
could impact on this benefit somewhat.
For both EIVC and LIVC cases the losses of efficiency associated with
combustion deterioration could potentially be reduced by adopting a higher geometric
CR. A higher geometric CR will increase the degree of in-cylinder turbulence toward
TDC and will also elevate the EOC temperature and pressure which will further aid
combustion stability. If combustion deterioration issues could be eliminated
completely this would yield an ISFC improvement of 5% with EIVC and 9% with LIVC.
259
7.4.2.4 Optimum Cam Timing Observations
The optimum cam timings make it very clear that the full potential of the short
and long cams could not be yielded at speeds and loads as low as these with this
particular engine. The optimum cam timing for the short cam was the maximum valve
overlap point (the point for minimum effective CR) for 3 of the 7 test points. The
optimum cam timing for the long cam was the minimum valve overlap point (the point
for minimum effective CR) for 3 of the 7 test points as well.
Analysis of the results shows that on almost every occasion the cam timing that
gave the highest EOC pressure was the optimum for the 4 bar BMEP cases. Eight bar
BMEP proved to be sufficient to run with the full extent of effective CR reduction with
the long cam, however, the optimum cam timing for the short cam at 8 bar BMEP was
typically in the middle of the cam phase range, offering neither optimum effective CR
nor a great deal of scavenging. Generally speaking the favoured cam timing was that
which offered the highest EOC pressure again.
The cam sweeps carried out show quite clearly the impact of effective CR
(IMOP sweeping) and residuals (EMOP sweeping) on the combustion stability and
impact on burn rate, however, the impact on ISFC could not be isolated due to the fact
that effective CR and effective expansion ratio change when IMOP and EMOP
(respectively) are adjusted. Figure 7.22 is an example of a 10-90% MFB IMOP-EMOP
sweep map extracted from the test data. The sensitivity to EMOP differed across the
speed and load range with the 8 bar BMEP points typically showing a reduced
sensitivity to EMOP and residuals. The most likely explanation for this is the elevated
MAP is resulting in scavenging of the cylinder rather than allowing back flow.
260
Figure 7.22 Map of 10-90% MFB Rates for an IMOP-EMOP Sweep with the Short Cam at 3000rpm and
4 bar BMEP
The sensitivity of 10-90% MFB duration to IMOP (or effective CR) can be seen
to be similar in magnitude for all speed and load points. The COV however is not
particularly sensitive to the degree of IMOP or EMOP but is typically greater with
extremes of IMOP, so either with maximum valve overlap or minimum valve overlap
for both the short and long cams. The COV typically increases with later IMOPs
(tending toward minimum valve overlap) because the effective CR is increasing.
Although the quantity of residuals will reduce with later IMOPs, the MAP will also
reduce allowing more back flow of exhaust gasses to the intake, so the quantity of
residuals cannot be expected to decline that predictably with increasing later IMOPs.
7.4.2.5 Combustion Phasing Observations
The MBT timing exhibited very little sensitivity to cam duration and fell
between 6 and 9 CAD ATDC for all test work (Figure 7.23).
261
9.5
8.5
MBT timing (CAD ATDC)
Short cam
8 (152 CAD)
7.5 Standard cam

(240 CAD)
7
Long cam
(292 CAD)
6.5
6
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.23 MBT Comparison of Short, Standard and Long Cams at Low Loads
Figure 7.24 reveals the spark timings to be dramatically different with the short
cam requiring a great deal more spark advance than the other 2 cams, which can be
seen to be broadly the same. The short cam can be seen to exhibit a slight upward
trend with increasing speed, which would imply that increasing engine speed does not
address the problem of reduced charge motion to the same extent as it does with the
longer cam.
The spark timing can be seen to be highly dependent on load with the short
and standard cams, with the higher load points requiring the spark to be retarded
significantly as can be expected. The long cam however exhibits a much smoother
profile and shows a far reduced sensitivity to load in comparison to the other 2 cams.
Figure 7.8 illustrates that intake port temperatures are very elevated at certain speeds
and loads in the case of the standard cam. This indicates a large amount of back flow
through the intake valves. The standard cam line can be seen to have a very erratic
profile. This is explained by the optimum IMOP timing for each of the different speed
and load points. For every speed and load point the optimum EMOP was the maximum
valve overlap position, therefore the quantity of residuals in the cylinder is influenced
purely by the IMOP timing. The optimum IMOP timing for the 4 bar BMEP cases (with
262
the exception of 3000rpm) was either the maximum valve overlap position or the
second highest degree of valve overlap position (Figure 7.25). This allowed a great deal
of back flow from the cylinder into the intake manifold during the valve overlap period
which elevated the intake port temperatures significantly. The 8 bar BMEP cases
generally favoured significantly lower degrees of valve overlap, and the MAP was
higher for these points. Both of these result in greatly reduced backflow of
residual/exhaust gasses to the intake thus resulting in a considerably reduced intake
port temperature.
Figures 7.12 and 7.13 also suggest that the residual gas concentration with the
standard cam at the 1000 and 2000rpm 4 bar BMEP points is particularly high, this will
likely result in the very pronounced variations in spark timing seen with the standard
cam. The standard cam would be expected to exhibit the highest MFB durations due to
the fact the EOC temperatures are inherently higher than with the short and long
cams.
60
50
Short cam
40
(152 CAD)
30 Standard cam
(240 CAD)
20
Long cam
(292 CAD)
10
0
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.24 Spark Timing Comparison of Short, Standard and Long Cams at Low Loads
The reason for the very erratic profile of the short cam line can likely be
attributed to the cam timing (Figure 7.25). The cam timings for the short cam at the
263
low load points tends to be toward the maximum valve overlap position, which for the
short cam is the position that inherently produces the lower EOC temperatures due to
the low effective CR. At higher loads the short cam favours the reduced valve overlap
positions, this will result in a higher EOC temperature.
160
140
120
IMOP (CAD ATDC)
100 Short cam

(152 CAD)
80 Standard cam
(240 CAD)
60
Long cam
(292 CAD)
40
20
1000 1000 2000 2000 3000 3000 4000
Engine speed (rpm)
Load (BMEP)
Figure 7.25 Optimum IMOP Timing Comparison of Short, Standard and Long Cams at Low Loads
7.4.3 Summary
EIVC proved to be the optimum strategy for reducing ISFC at low load. The
reason for this is a reduction in effective CR and a consequent increase in MAP. The
increase in MAP ensured back flow of exhaust gasses into the intake manifold and the
consequent residual gas concentration at cycle start was kept to a minimum. The
increase in MAP also resulted in a reduction of the pumping loop size.
LIVC was found to be hindered by combustion degradation issues. The cause of

this is unknown but is suspected to be due to the fact charge is exhaled into the intake
manifold where it may react and spoil before it is readmitted to the cylinder for the
264
next cycle. Both GT-Power and HC emissions analysis confirm that no significant
quantities of charge scavenging are taking place.
It is suspected that both EIVC and LIVC strategies at these low load conditions
would benefit from an increase of geometric CR to help raise the EOC pressure and
temperature and to ensure higher levels of turbulence are generated in the cylinder to
allow thorough mixing.
7.5 High Load Operations with EIVC and LIVC (Deep Miller)
7.5.1 Introduction
High load EIVC and LIVC are arguably of greatest importance to the downsizing
concept. The mechanism by which an improvement in fuel economy and performance
is theorized to take place is as described in section 7.3. The reduction of EOC pressure
and temperature is of great importance because it will advance the BLD limit (in
comparison to the standard cam).
Unfortunately the test plan for this work had to be truncated to just the
2000rpm, 15 bar BMEP and 24 bar BMEP points because of a severe failure of the
engine believed to have been brought about by running at high speeds and loads.
The focus of this work was on reducing ISFC rather than increasing maximum
load for the reasons given above. The 24 bar BMEP points were carried out at 2
different EBPs, one scenario was with no EBP and the other with an EBP of 2 bar. The
15 bar BMEP points were also carried out at 2 different EBPs, one scenario was with no
EBP the other was setting the EBP valve to 45% closed. This was determined as the
correct EBP valve position to give turbocharger representative levels of EBP at this
speed and load.
No emissions were logged for this test work except HC. The reason for not
logging NOx is that this no longer falls in the remit of the NEDC cycle.
265
The test points are shown in Table 7.2, and the test variables and control
criteria are defined in Table 7.3.
7.5.2 Results
A comparison of all 3 cams with no EBP and 2 bar EBP is shown in Figure 7.26.
270
265
260
ISFC (g/kWhr)
255 No EBP
2 bar EBP
250
245
240
Short Standard Long
Figure 7.26 2000rpm, 24 bar BMEP ISFC Comparison for Short, Standard and Long Cams at 2 EBPs
The increase in load has been found to greatly increase the combustion stability
for both the short and long cam, but particularly for the long cam which, for this work,
was phased to its maximum extent (minimum valve overlap). The short cam could not
be phased to provide the minimum effective CR however, due to the fact that at
maximum valve overlap (the optimum timing for low effective CR for the short cam)
scavenging was resulting in accidental enrichment of the charge as evidenced by the
profile of HC emissions Vs. IMOP angle (Figure 7.27).
266
Figure 7.27 HC Emissions for Different IVC Timings and 2 Different EBPs with the Short Cam at
2000rpm and 24 bar BMEP with an EMOP of 100 CAD BTDC (Maximum Valve Overlap Position)
The accidental over enrichment is reduced in magnitude for the 2 bar EBP case
but can still be seen to occur owing to the low effective CR requiring an MAP well in
excess of the EBP. EBP effects have been analysed in greater depth in section 7.5.2.1.
In spite of the fact the short cam could not be phased to provide the maximum
degree of effective CR reduction it still offers the greatest improvement in ISFC in
comparison to the long cam. One of the reasons for this is that the EOC pressure was
lower for the short cam. This can likely be attributed (to an extent) to the fact that the
degree of effective CR reduction obtainable with the short cam is slightly greater than
that with the long cam.
The degree of effective CR reduction obtainable with the short cam is even
greater still due to fact the effective CR is calculated from the 1mm lift position. In
reality the flow through the intake valve will begin to reduce at valve lifts of more than
1 mm which in the case of EIVC means cylinder depressurisation will occur before the
1mm lift point is reached, and for LIVC cylinder pressurisation will occur before the
1mm lift point. The difference between EIVC and LIVC is that the flow into the cylinder
will begin to reduce earlier for EIVC and the flow out of the cylinder will reduce earlier
for LIVC. This means that the effective mass of air trapped in the cylinder with EIVC will
267
be less than the effective CR will imply, while the mass of air trapped in the cylinder
with LIVC will be more than the effective CR will imply.
Another one of the reasons why the ISFC improvement is not as marked for the
long cam as it is for the short cam is the fact that charge scavenging is taking place.
This has been confirmed by a GT-Power simulation and is accountable for a 1.91% drop
in ISFC (this is less than the values quoted in section 5.3.4.2 due to the fact that a
stoichiometric charge was assumed to be scavenged in that instance). The presence of
charge scavenging has also manifested itself in the HC emissions results obtained from
the experimental engine (With HC emissions of 0.0225, 0.0243 and 0.039 g/kWhr for
the short, standard and long cams respectively). There is no appreciable rise in exhaust
gas temperature which indicates that the charge that has been scavenged has not
been burning in the exhaust gasses, however, this is the temperature as measured at
the exhaust port so this is not necessarily a good indicator. There is a small increase in
EBP for the no EBP case (please note: no EBP means the EBP valve was set to 0%
closed, not necessarily a complete absence of EBP and/or pressure/rarefaction waves)
for the 24 bar BMEP point, and also a small increase for both 15 bar BMEP with and
without EBP points. This could indicate that some combustion is occurring in the
exhaust manifold, but not enough data has been obtained to determine with any
degree of certainty whether there is any combustion at any point in the exhaust
system.
Although the combustion stability is considerably improved (relative to the low

load performance) the COV for the short cam is markedly worse than for the other 2
cams, but only at the 24 bar BMEP points (Figure 7.28).
268
4
3.5
3
Short cam
2.5 (152 CAD)
COV (%IMEP)
2 Standard cam
(240 CAD)
1.5
Long cam
(292 CAD)
1
0.5
0
2000rpm 2000rpm 2000rpm 2000rpm
24 bar BMEP 24 bar BMEP 15 bar BMEP 15 bar BMEP
no EBP 2 bar EBP no EBP 45% EBP
Figure 7.28 COV for Short, Standard and Long Cams at 2000rpm, High Load with and without EBP
The lack of combustion stability is also visible (and understandable) when

observing the 10-90% MFB duration results (Figure 7.29). At every point the 10-90%
MFB duration is markedly longer for the short and long cams in comparison to the
standard cam. The burn rate is particularly long for the short cam which, as with the
low load points, is probably due to a lack of bulk flow in the cylinder as well as lower
EOC pressure and temperature. The 10-90% MFB durations are still long for the long
cam but considerably better than for the short cam. The reason for this is the in-
cylinder charge motion is likely to be comparable to the standard cam and consist of a
great deal of bulk fluid motion, but the effective CR reduction will result in lower EOC
pressures and temperatures which will reduce flame front speed.
269
26
24
22
10-90% MFB (CAD)
Short cam
20
(152 CAD)
18 Standard cam
(240 CAD)
16 Long cam
(292 CAD)
14
12
2000rpm 2000rpm 2000rpm 2000rpm
Figure 7.29 10-90% MFB Duration for Short, Standard and Long Cams at 2000rpm, High Load with and
without EBP
The increase in 10-90% MFB duration is highly unlikely to be caused by

residuals because a GT-Power simulation has revealed there to be no considerable
residual gas concentration at cycle start with all 3 cams. The very long 10-90% MFB
duration has not had a great impact on the BLD point at the 24 bar BMEP point, with
no strong trends to suggest the short cam requires a greater deal of combustion
phasing retard as a consequence of a longer burn duration (Figure 7.30). However, the
impact of a longer burn duration on the 50% MFB point is more noticeable at 15 bar
BMEP.
270
25
23
21
50% MFB angle (CAD ATDCF)
19
Short cam
(152 CAD)
17
Standard cam
15 (240 CAD)
13 Long cam
(292 CAD)
11
7
2000rpm 2000rpm 2000rpm 2000rpm
Figure 7.30 50% MFB Angle for Short, Standard and Long Cams at 2000rpm, High Load with and
without EBP
The short cam also demonstrates the lowest EOC pressure at the 15 bar BMEP
site too. This can be found to manifest itself by the degree of spark advance that can
be achieved with the short cam (Figure 7.31). Owing to a lack of charge motion
associated with low lift, this is not corresponding to a 50% MFB benefit. A similar trend
can be seen with the long cam too where the spark can be considerably advanced due
to reduced EOC temperature but this is not yielding a benefit in the 50% MFB due to
reduced charge motion.
271
26
24
22
20
18
16
14 Short cam
(152 CAD)
12
10 Standard cam
(240 CAD)
8
6 Long cam
(292 CAD)
4
2
0
-2
2000rpm 2000rpm 2000rpm 2000rpm
Figure 7.31 Spark Timing for Short, Standard and Long Cams at 2000rpm, High Load with and without
EBP
The 15 bar BMEP ISFC trends can be seen to match the 24 bar BMEP trends
quite closely (Figure 7.32).
260
255
250
ISFC (g/kWhr)
245 No EBP
45% EBP
240
235
230
Short Standard Long
Figure 7.32 2000rpm, 15 bar BMEP ISFC Comparison for Short, Standard and Long Cams at 2 EBPs
272
Although the absolute ISFC benefit afforded by the long cam is similar in
magnitude at 15 bar BMEP and 24 bar BMEP, relative to the short cam it appears less
effective at this load. Although 15 bar BMEP is quite a way down in load compared to
24 bar BMEP, it is still high enough for the test results to indicate the theoretical
optimum cam timing is the actual optimum cam timing, which shows that the benefits
offered by over-expansion are not outweighed by the problems associated with
combustion instability.
The EBP can be seen to have very little effect here, the reason for this is 45%
EBP valve closure corresponds to an EBP of approximately 1.2 bar abs which is
considerably less than the 2 bar EBP applied to the 24 bar BMEP point.
A GT-Power analysis has confirmed that a small amount of charge scavenging

still takes place at this load, however, the experimental test data shows no strong
indication of this with only a minor increase in the HC emissions for the long cam in
comparison to the short and standard cams.
7.5.2.1 EBP Effects
The presence of EBP has only a slight effect on the ISFC (Figure 7.26) but it can
be seen to have a significant effect on the COV which drastically increases with EBP.
This is most likely caused by an increase of MFB duration associated with increased
residuals (Figure 7.29). The long cam is the only cam to exhibit this trend of longer
combustion duration with increasing EBP, with the other 2 cams exhibiting faster
combustion. Figure 7.27 shows that higher EBPs can afford the use of lower effective
CRs with the short cam by inhibiting scavenging, however the optimum cam timings
for ISFC for the short cam still do not match the theoretical optimum cam timings.
One such point where the increase in EBP has permitted the use of a lower
effective CR is the 2000rpm, 24 bar BMEP point with the short cam. In spite of the
increased in-cylinder pressure and temperature associated with higher EBPs the
combustion phasing could be advanced to closer to MBT with an EBP of 2 bar than it
273
could with no EBP (Figure 7.30). This has yielded no benefit to ISFC however because
of the reduced combustion stability associated with higher residual gas concentration.
Employing the EBP valve has also produced an effect similar to that shown in
chapter 5 in that pressure waves in the exhaust can be seen to influence the in-
cylinder pressure profile (on a CAD basis) as seen in Figure 7.33.
Figure 7.33 EBP Profile Comparison between Short, Standard and Long Cams with 2 bar Average EBP
at 2000rpm and 24 bar BMEP
This has the effect of either reducing or increasing the pumping loop size
depending on the MAP. This will have the most significant effect on the short cam
because that is the only 1 of the 3 cams that is running with the maximum degree of
valve overlap. The impact of this on the thermodynamic cycle is very hard to
determine but an attempt has been made in the following section (section 7.5.2.2).
7.5.2.2 Analysis of the Individual Components of EIVC and LIVC
This section contains a breakdown of the individual components that cause an

increase or decrease in the overall efficiency for both EIVC and LIVC in comparison to
274
the standard cam. The 24 bar BMEP point has been selected for analysis both with and
without EBP.
The theoretical analysis is as shown in Figures 7.34 and 7.35 with no EBP and 2
bar EBP respectively.
Theoretical benefit with EIVC with no EBP (%)
Total
Over- Theoretical
expansion benefit (%)
Pumping loop
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
Figure 7.34 Theoretical Benefit with EIVC at 2000rpm and 24 bar BMEP with no EBP
275
Theoretical benefit with EIVC with 2 bar EBP (%)
Total
Over- Theoretical
expansion benefit (%)
Pumping loop
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Figure 7.35 Theoretical Benefit with EIVC at 2000rpm and 24 bar BMEP with an EBP of 2 bar
The gain in efficiency predicted with the 2 bar EBP case over the no EBP case
can be attributed to the fact the short cam can be phased 10 CAD closer to maximum
valve overlap position with an EBP of 2 bar, as discussed in section 7.5.2.1. In the case
of the experimental engine an efficiency benefit (relative to the standard cam) can be
yielded by increasing the EBP. Figures 7.36 and 7.37 show the actual components and
their impact on the cycle efficiency for the no EBP and 2 bar EBP cases respectively.
276
Actual benefit with EIVC and no EBP (%)
Total

benefit (%)
Over-expansion
Pumping loop
-2 -1 0 1 2 3 4 5 6
Figure 7.36 Actual Benefit with EIVC at 2000rpm and 24 bar BMEP with no EBP
Actual benefit with EIVC and 2 bar EBP (%)
Total

benefit (%)
Over-expansion
Pumping loop
-2 -1 0 1 2 3 4 5 6
Figure 7.37 Actual Benefit with EIVC at 2000rpm and 24 bar BMEP and 2 bar EBP
The pumping loop and breathing/combustion components offer the majority of

the improvement in performance when an EBP of 2 bar has been employed. The
277
breathing/combustion component refers to the influence of EOC pressure and
temperature leading to a greater deal of spark advance, this was found to be of more
importance than the reduction of bulk flow/increased MFB duration.
The over-expansion component can be seen to decrease in size slightly with

increasing EBP. This is an indication that the exhaust temperature was elevated to a
greater extent than expected by the increase in EBP.
Due to the tuning effects discussed in section 7.5.2.1 the theoretical benefit
afforded by the reduced pumping loop size component with no EBP is not realised in
practise. The pumping loop contribution can actually be seen to decrease with
decreasing EBP.
The theoretical benefit offered by LIVC with an EBP of 2 bar is shown in Figure
7.38. The no EBP case has not been shown due to the fact the effective CR does not
change, this therefore results in a negligible difference between both scenarios.
Theoretical benefit with LIVC with 2 bar EBP (%)
Total
Over-expansion Theoretical
benefit (%)
Pumping loop
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Figure 7.38 Theoretical Benefit with LIVC at 2000rpm and 24 bar BMEP with an EBP of 2 bar
This is not much different to the LIVC case due to the fact the effective CRs are
almost the same for both cases. The actual benefits for the no EBP and the 2 bar EBP
cases are shown in Figures 7.39 and 7.40.
278
Figure 7.39 Actual Benefit with LIVC at 2000rpm and 24 bar BMEP with no EBP
Figure 7.40 Actual Benefit with LIVC at 2000rpm and 24 bar BMEP with an EBP of 2 bar
The most noticeable difference between the 2 EBP cases is the size of the
pumping loop which increases drastically with EBP. The breathing/combustion
contribution can also be seen to increase dramatically with EBP. The
279
breathing/combustion issues worsen with increasing EBP which is somewhat
contradictory. The reason why the breathing/combustion contribution is so large is
because of the charge scavenging that takes place. At an EBP of 2 bar the ratio of MAP
to EBP is lower so it would be logical to assume less charge scavenging will be taking
place. The HC emissions obtained from this test work suggest that the degree of
charge scavenging at 2 bar EBP is reduced compared to the no EBP case, GT-Power also
suggests that less charge scavenging is taking place at 2 bar EBP.
The only other possible explanation is that an exhaust rarefaction wave during
the brief valve overlap period is resulting in charge scavenging, and that the elevated
exhaust gas temperature is causing the air/fuel mixture to combust before a sample
reaches the exhaust gas analyser. As was the case previously though, not enough
evidence exists to prove that this is happening.
7.5.2.3 Turbocharging/Supercharging Feasibility for Deep Miller
From the data obtained from both the GT-Power work and the experimental
test work, the penalty to ISFC for running with turbocharger representative levels of
EBP is not sufficient to justify supercharging, as this will almost certainly be less
efficient.
There is however just argument for supercharging at some of the points tested
because of the need to run at high MAPs even at relatively conservative loads. For
instance Figure 7.15 reveals the MAPs required to run at the part load points. For some
speeds the required MAP for 4 bar BMEP is exceeding 1 bar for both the short and long
cams. Even at loads as low as this, at 4000rpm most turbochargers will be capable of
providing very light boost as required, but at 1000rpm the mass flow rate of air
through the turbine will be insufficient for even light boost.
There are turbochargers that can provide significant boost at 2000rpm and 24
bar BMEP however, it is unlikely that they will be able to achieve the required MAP of
3.5 bar abs, as is required for the long cam with 2 bar EBP.
280
The long cam generally had a greater need for high MAPs at low speeds and
loads. The short cam on the other hand required elevated MAPs in comparison to the
standard cam, but these generally did not exceed atmospheric pressure. For low
speeds therefore LIVC will likely require some sort of supercharging/e-boosting system
whereas EIVC is better suited to turbocharged applications.
LIVC also required higher MAPs than EIVC at higher loads too, as can be seen in
Figure 7.41 for an engine speed of 2000rpm.
Figure 7.41 MAP Requirement Vs. BMEP for Short, Standard and Long Cams at an Engine Speed of
2000rpm
The reason for the higher MAP requirement is as mentioned previously, the
definition of the intake valve close point is at 1mm lift, in reality the charge will not
stop escaping to the manifold at 1mm lift thus reducing the effective CR to a level
below that of the short cam.
281
7.5.3 Summary
EIVC was found to be the more effective strategy for reducing ISFC at every
load and EBP condition tested. This benefit amounted to between 4.9 and 9%
depending on the load point and EBP. LIVC also delivered a slight benefit at these
elevated loads but it was a lot more conservative than that offered by EIVC and
amounted to 1.1 to 2.9% depending on the load point and EBP.
The main reason for the modest increase in efficiency from LIVC is charge
scavenging which was found to occur with very high MAP to EBP ratios. The impact of
this on the cycle efficiency is very difficult to determine but ranged from 0.2% to 1.9%
according to GT-Power simulations.
It is also suspected that the valve lift for the long cam might have benefitted
from being reduced to promote more turbulence in the cylinder which would aid
mixing. There was evidence of insufficient mixing apparent from the MFB duration
data, the MFB duration was consistently longer than with the standard cam, this could
be due to the bulk flow in the cylinder being destroyed as charge is expelled back into
the intake during compression. There was also evidence of less mass being burned in
total with LIVC, even though the 10-90% MFB duration was quite reasonable. It is
suspected that this is due to in-cylinder accidental over enrichment caused by non-
homogeneous charge being expelled to the intake during the compression stroke.
EIVC exhibited the longest MFB durations which are suspected to be the result
of reduced bulk flow in the cylinder. The low lift valve is suspected to compensate for
this somewhat by increasing the flow velocity through the valve, which will be
converted into high amounts of TKE which will ensure good mixing.
Both EIVC and LIVC resulted in an increase in the COV of IMEP. At lower loads
LIVC generally exhibited greater COV of IMEP values than EIVC and the standard cam.
EIVC exhibited greater COV of IMEP values at higher loads than LIVC, LIVC was found to
be similar to the standard cam.
282
The EBP valve is suspected to influence the results, particularly at the 2 bar EBP
points where the blowdown pulse could be seen to bounce between the EBP valve and
the engine. The exact impact of this on engine performance is very difficult to gauge
but comparing the data with the EBP valve with the data with no EBP indicates that the
effect is not massively significant.
Although both the long and short cams were selected to give similar ranges of
effective CR reduction through equal degrees of LIVC and EIVC respectively. The
maximum degree of LIVC will always be slightly greater than the maximum degree of
EIVC because the effective CR was calculated at the 1mm lift point. In reality air/charge
will continue to flow beyond these points. The error this has resulted in could not be
calculated due to the presence of pressure waves and ram air effects distorting how
much air is actually trapped in the cylinder at the point of valve closure.
283
8
Chapter 8
Summary and Conclusions
8.1 Introduction
This thesis contains a thorough and in-depth exploration of both low and high
load Miller cycle processes on a modern aggressively downsized engine. A Ricardo
Hydra based single cylinder experimental engine with a 3 cylinder head from a Mahle
downsized DI gasoline engine was designed and commissioned. Its performance was
analysed both thermodynamically and analytically across the speed and load range of
the engine. The Miller cycle process was applied successfully and at loads that were
hitherto untested and/or unpublished in the public domain.
Both a thermodynamic and an analytical comparison were made between the

single cylinder experimental engine, the single cylinder model and the Mahle 3 cylinder
engine (using a 3 cylinder model developed by Mahle Powertrain and baseline data
acquired by Mahle Powertrain to assist). The differences in key performance criteria
such as efficiency, engine breathing, autoignition propensity and effective exhaust
back pressures were identified and analysed.
An analytical analysis of “Deep Miller” was carried out in a one-dimensional GT-

Power model, and the impact of EIVC and LIVC strategies at speeds of 1000, 2000 and
5000rpm and loads of 14, 24 and 24 bar BMEP respectively were assessed. A thorough
analysis of the trends in the results was also carried out. Analysis of the maximum load
284
that could be obtained was also carried out to obtain an estimate of the performance
benefit that could be expected when running with the Miller cycle at high load.
In engine experiments both EIVC (152 CAD duration cam) and LIVC (292 CAD
duration cam) operating regimes were tested with a VVT system applied to both intake
and exhaust cams, and the cams swept across their entire ranges of operation.
Extensive test work with the standard cam was carried out to provide a suitable
comparison. An exhaust back pressure valve was applied to both the experimental and
analytical engine to simulate a turbocharger, and an intake heater applied to ensure
the incoming air temperature was at turbocharger representative levels.
Further test work was carried out into autoignition and “superknock” tendency
in an effort to better understand the causes of autoignition and why it is so prevalent
on the single cylinder engine in comparison to the 3 cylinder engine.
8.2 Analytical Study of Deep Miller
The impact of Deep Miller was assessed using a DoE based test plan which
allowed the assessment of other variables such as intake cam timing, exhaust back
pressure and geometric compression ratio too. Three DoEs were carried out at
different speeds to assess the increase in boosting requirement at low speed, the
autoignition imposed limitations at low to medium speeds and the fuel enrichment
requirement at high speed.
Both EIVC and LIVC were observed to offer a benefit over the baseline case but
the extent of the EIVC and LIVC that could be employed were both constrained by
various factors. A great deal of scavenging was observed to take place with EIVC due to
the fact the intake manifold pressure had to be elevated considerably over the
baseline case. This was found to be an issue in particular for EIVC because the effective
compression ratio is lowest (the optimum position) when the short cam is in the
maximum valve overlap position. Scavenging, though good for exhaust gas
285
temperature reduction and purging the cylinder of residuals, resulted in accidental
over enrichment of the charge in the cylinder which elevated ISFC considerably.
It was discovered that if the engine management system is capable of

compensating for this scavenging and maintaining a stoichiometric charge in the
cylinder, EIVC resulted in a 5.3% improvement in thermal efficiency compared to a
maximum of 3% for LIVC at realistic exhaust back pressures. With no compensation
applied the optimum strategy for fuel economy varied depending on speed, load, and
very much so on exhaust back pressure. In almost all of the 1000rpm and 2000rpm
cases moderate LIVC was found to be the optimum strategy and yielded a maximum
benefit of up to 7% at very elevated exhaust back pressures. EIVC was hindered by
excessive scavenging leading to accidental enrichment.
At 5000rpm the maximum extent of EIVC was favoured due to the fact that fuel
enrichment was required to regulate the exhaust gas temperature, which makes the
fact it was occurring by accident academic. The maximum benefit achieved over the
standard baseline cam was 3.2%.
It was also found that the engine was highly sensitive to geometric
compression ratio, particularly for cams that exhibited very low effective compression
ratio reduction. For some exhaust back pressures an increase in geometric
compression ratio was beneficial to thermal efficiency, but only for extreme EIVC and
LIVC cases. This gain in efficiency peaked at 5.8% and was typically greater at lower
exhaust back pressures.
The potential to increase maximum engine load using Deep Miller was
explored. LIVC was found to offer the greatest increase in load. The greatest increase
at higher exhaust back pressures was obtained with the longest (most extreme LIVC)
cam but the moderate LIVC cam was found to be optimal for the lower EBP cases. The
most extreme EIVC and LIVC cams were greatly limited by the peak intake manifold
pressure however. This was not such a constraint at lower speeds but at 5000rpm the
maximum load obtained with extreme LIVC was lower than that obtained with the
moderate LIVC cam by virtue of the fact that the volumetric efficiency of the extreme
LIVC cam was the bottleneck. The maximum load increase at turbocharger
286
representative levels of exhaust back pressure was 1.3% at 1000rpm, 12.3% at
2000rpm and 1.7% at 5000rpm.
8.3 Experimental Engine Study of Miller Cycle
The effect of both EIVC and LIVC was assessed at numerous points over the
speed and load map, and hydrocarbon and NOx emissions were both recorded in order
to gain an understanding of the impact of Miller cycle on emissions performance.
At low load EIVC was found to give thermal efficiency benefits for every single
low speed point of up to 7.2% over the baseline cam. LIVC however exhibited lower
thermal efficiency than the standard baseline cam at all low load points with the
exception of the 1000rpm points. The reduction of thermal efficiency was found to be
up to 7.1% with LIVC.
It was found that the unexpected result with LIVC could be attributed to the
breathing behaviour of the single cylinder experimental engine. An analytical analysis
of the same test points revealed that the long cam (LIVC) had a considerably higher
residual gas concentration than the equivalent EIVC point. It is also suspected that
exhaling charge into the intake manifold had a detrimental effect on thermal
efficiency.
The PMEP component was found to be greatly reduced for both EIVC and LIVC
strategies. The PMEP component for EIVC was found to be minimised in size
considerably, however, the pumping component for LIVC was generally less than that
of the standard cam baseline, but its contribution to the overall cycle was much higher
than with EIVC. At low speeds the PMEP component for LIVC was found to be negative
and actually contributed work to the cycle from loads as low as 8 bar BMEP. This work
contribution was as high as 4.7% to the entire (net) cycle at low speed.
The effect of Miller cycle at low speed and high load operations was also
assessed and it was found that both EIVC and LIVC resulted in a thermal efficiency
increase over the standard cam baseline. EIVC was found to give a greater efficiency
287
benefit of 6% at 24 bar BMEP and 8.8% at 15 bar BMEP, while LIVC gave an efficiency
benefit of 1.3% at 24 bar BMEP and 2.8% at 15 bar BMEP, both were at turbocharger
representative levels of exhaust back pressure.
The limited benefit offered by LIVC was attributed to increased rates of charge
scavenging which were confirmed both by hydrocarbon emissions and in an analytical
study. It was discovered that the intake port temperature was massively elevated (up
to 900C) beyond anything reported in the literature.
Both EIVC and LIVC strategies exhibited greater COV of IMEP values in
comparison to the standard cam baseline point, even at loads as high as 24 bar BMEP.
The COV of IMEP showed a dramatic increase when exhaust back pressure was
applied. It was discovered that the exhaust back pressure valve created considerable
pressure waves within the exhaust system that resulted in a pressure wave reaching
the cylinder at the valve overlap point, it is suspected that this may have had an
adverse affect on the COV of IMEP values obtained.
The mass fraction burn rate for EIVC and LIVC were longer than with the
standard baseline cam. EIVC exhibited the longest mass fraction burn duration, which
indicates that reduced charge motion is still an issue, even at very high loads, but that
its effect on thermal efficiency was minimal.
A summary of the different components that led to an increase in thermal

efficiency was also carried out. Two studies were carried out, one assessing the
theoretical benefits that should be yielded and the other showing the actual benefits
that were yielded based on thermodynamic and analytical data. The loss of
thermodynamic efficiency due to combustion was found to be considerable for both
EIVC and LIVC (up to 5.1% and 9.2% respectively) while the loss of thermodynamic
efficiency due to a lack of bulk flow with EIVC was relatively small in comparison
(0.5%).
288
8.4 Recommendations for Future Work
This study has investigated the use of the Miller cycle at both low and high
loads, the first study of its kind. It has revealed that the Miller cycle, particularly EIVC
shows great high load potential. LIVC showed some potential but was hindered by the
engine breathing dynamics of the single cylinder experimental engine, and it would be
unfair to rule it out as a result of this study. It would be a worthwhile investment of
time to optimise the intake and exhaust tuning for LIVC for future work and perhaps
trying both strategies on the 3 cylinder engine.
Deep Miller appears to be a very viable future technology for increased fuel
efficiency from the analytical work and experimental work carried out in this study.
Whilst it is perhaps not a good idea to increase the boost pressure ceiling beyond 4 bar
abs for near term future work at this moment in time, it would be useful to assess the
experimental benefits of Deep Miller at more speeds such as at 1500rpm and
5000rpm.
It would also be beneficial to test different piston geometries in future work

such as different geometric compression ratios and piston bowls designed to give
more turbulence and tumble motion, mainly for the benefit of EIVC. Valve deactivation
almost certainly is not feasible with EIVC with Deep Miller, but may be feasible with
LIVC. This is another avenue that could potentially be explored in future work. With
the current state of tune of the single cylinder engine, full load EGR is likely not viable,
the residual gas concentration for LIVC has been seen to have a very detrimental
impact on thermal efficiency and EGR will likely exacerbate this problem.
289
References
[1] Smil, V., “The Two Prime Movers of Globalization: History and Impact of Diesel
Engines and Gas Turbines”, Journal of Global History 2, pp 373-394, 2007.
[2] Zhao, H., “Advanced Direct Injection Combustion Engine Technologies and
Development Volume 1: Gasoline and Gas Engines”, Woodhead Publishing Limited,
2009.
[3] Clean Air Act 1956, 4 & 5 Eliz.2 Ch. 52, Parliament of the United Kingdom, 1956.
[4] GT-SUITE V6.1.0, Gamma Technologies Inc, Westmont, Illinois, 2004.
[5] Fuglestvedt, J., Bernsten, T., Myhre, G., Rypdal, K., Skeie, R.B., “Climate Forcing
from the Transport Sectors”, Proceedings of the National Academy of Sciences, Vol.
105, No. 2, pp 454-458, 2008.
[6] “EU Transport in Figures: Statistical Pocketbook 2011”, European Commission,

2011.
[7] “World Oil Outlook 2011”, Organization of the Petroleum Exporting Countries
(OPEC), 2011.
[8] Regulation (EC) No 443/2009 of the European Parliament and of the Council of 23
April 2009 Setting Emission Performance Standards for New Passenger Cars as Part of
the Community’s Integrated Approach to Reduce CO2 Emissions from Light-Duty
Vehicles (Text with EEA Relevance), 2009.
[9] Proposal 2012/0190 (COD) for a Regulation of the European Parliament and of the
Council Amending Regulation (EC) No 443/2009 to Define the Modalities for Reaching
the 2020 Target to Reduce CO2 Emissions from New Passenger Cars, 2012.
[10] “Worldwide Emissions Standards: Passenger Cars and Light Duty Trucks”, Delphi,
2012.
[11] Regulation (EC) No 715/2007 of the European Parliament and of the Council of 20
June 2007 on Type Approval of Motor Vehicles with Respect to Emissions from Light
Passenger and Commercial Vehicles (Euro 5 and Euro 6) and on Access to Vehicle
Repair and Maintenance Information (Text with EEA Relevance), 2007.
[12] US EPA Clean Air Act Title II Part A – Motor Vehicle Emission and Fuel Standards,
1990.
290
[13] US EPA Control of Air Pollution from Motor Vehicles: Tier III Motor Vehicle
Emission and Fuel Standards, 2011.
[14] California Exhaust Emissions Standards and Test Procedures for 2001 Through
2014 Model Criteria Pollutant Exhaust Emissions Standards and Test Procedures and
2009 Through 2016 Model Greenhouse Gas Exhaust Emission Standards and Test
Procedures for Passenger Cars, Light-duty Trucks and Medium-duty Vehicles, California
Environmental Protection Agency Air Resources Board (CARB), 1999.
[15] Lyons, J. M., Carlson, T. R., Austin, T. C., “Evaluation of California Greenhouse Gas
Standards and Federal Energy Independence and Security Act – Part 1: Impacts on New
Vehicle Fuel Economy”, SAE Paper 2008-01-1852, 2008.
[16] Air Pollution Control Law Chapter III, Ministry of the Environment, Government of
Japan, 1996.
[17] Resolution No 415 of 24 September 2009, Conama 415/2009, Federal

Government of Brazil, 2009.
[18] “Gasoline and Diesel Fuel Update”, U.S. Energy Information Administration, 2012.
[19] Bolton, P., “Petrol and Diesel Prices”, House of Commons Library, 2012.
[20] Sher, E., “Handbook of Air Pollution from Internal Combustion Engines: Pollutant
Formation and Control”, Academic Press, 1998.
[21] Heywood, J. B., “Internal Combustion Engine Fundamentals”, McGraw Hill, 1988.
[22] Cole, R. L., Poola, R. B., Sekar, R., “Exhaust Emissions of a Vehicle with a Gasoline
Direct-injection Engine”, SAE Paper 982605, 1998.
[23] Harrington, J. A. and Shishu, R. C., “A Single Cylinder Study of the Effects of Fuel
Type, Fuel Stoichiometry, and Hydrogen to Carbon Ratio and CO, NO and HC Exhaust
Emissions”, SAE Paper 730476, 1973.
[24] Shimotani, K., Oikawa, K., Horada, O., Kagawa, Y., “Characteristics of Gasoline In-
cylinder Direct Injection Engine”, Proceedings of the Internal Combustion Engine
Symposium – Japan, 1995.
[25] Bowman, C. T., “Kinetics of Pollutant Formation and Destruction in Combustion”,

Progress in Energy and Combustion Science, Vol. 1, pp 33-45, 1975.
[26] Johnson, G. L., Caretto, L. S., Starkman, E. S., “The Kinetics of CO Oxidation in
Reciprocating Engines”, The Combustion Institute Spring Meeting, 1970.
[27] Zhao, F., Lai, M. C., Harrington, D. L., “The Spray Characteristics of Automotive
Port Fuel Injection – A Critical Review”, SAE Paper 950506, 1995.
291
[28] Daniel, W. A., “Flame Quenching at the Walls of an Internal Combustion Engine”,
Proceedings of the Sixth International Symposium on Combustion, 1957.
[29] Wentworth, J. T., “The Piston Crevice Volume Effect on Exhaust Hydrocarbon
Emission”, Combustion Science and Technology, Vol. 4, pp 97-100, 1971.
[30] Stone, R., “Introduction to Internal Combustion Engines”, Third Edition, Palgrave
Macmillan, 1999.
[31] Miller, J.A. and Bowman, C. T., “Mechanism and Modelling of Nitrogen Chemistry
in Combustion”, Progress in Energy and Combustion Science, Vol. 15, pp 287-338,
1989.
[32] Zeldovich, Y. B., “The Oxidation of Nitrogen in Combustion and Explosions”, Acta
Physiochim URRS, Vol. 21, pp 577-628, 1947.
[33] Lavoie, G. A., Heywood, J. B., Keck, J. C., “Experimental and Theoretical Study of
Nitric Oxide Formation in Internal Combustion Engines”, Combustion Science and
Technology, Vol. 1, pp 313-326, 1970.
[34] Komiyama, K. and Heywood, J. B., “Predicting NOx Emissions and Effects of
Exhaust Gas Recirculation in Spark-ignition Engines”, SAE Paper 730475, 1973.
[35] Kummer, J. T., “Catalysts for Automobile Emission Control”, Progress in Energy
and Combustion Science, Vol. 6, pp 177-199, 1981.
[36] Helms, H., Pehnt, M., Lambrecht, U., Liebich, A., “Electric Vehicle and Plug-in
Hybrid Energy Efficiency and Life Cycle Emissions”, Eighteenth International
Symposium Transport and Air Pollution, Session 3: Electro and Hybrid Vehicles, 2010.
[37] Thomas, C. E., “Fuel Cell and Battery Electric Vehicles Compared”, International
Journal of Hydrogen Energy, Vol. 34, pp 6005-6020, 2009.
[38] Plotkin, S., Santini, D., Vyas, A., Anderson, J., Wang, M., He, J., Bharathan, D.,
“Hybrid Electric Vehicle Technology Assessment: Methodology, Analytical Issues, and
Interim Results”, Centre for Transportation Research, Argonne National Laboratory,
ANL/ESD/02-2, 2001.
[39] Bassett, M. D., Hall, J., Darkes, D., Fraser, N. A. J., Warth, M., “Design and
Development of a Dedicated Range Extender Engine”, Internal Combustion Engines:
Improving Performance, Fuel Economy and Emissions, Institute of Mechanical
Engineers, 2011.
[40] Lee, C. Y., Zhao, H., Ma, T., “Analysis of a Cost Effective Air Hybrid Concept”, SAE
Paper 2009-01-1111, 2009.
292
[41] Schwarz, C., Schünemann, E., Durst, B., Fischer, J., Witt, A., “Potentials of the
Spray-guided BMW DI Combustion System”, SAE Paper 2006-01-1265, 2006.
[42] Missy, S., Schwarz, C., Schünemann, E., “Brennverfahrensentwicklung am Beispiel

der Neuen BMW Sechs – Und Vierzylinder Ottomotoren mit High Precision Injection
und Schichtbrennverfahren”, Direkteinspritzung im Ottomotor IV, 2007.
[43] Lake, T., Stoke, J., Murphy, R., Osborne, R., Schamel, A., “Turbocharging Concepts
for Downsized DI Gasoline Engines”, SAE Paper 2004-01-0036, 2004.
[44] Zhao, H., Li, J., Ma, T., Ladommatos, N., “Performance and Analysis of a 4 Stroke
Multi-cylinder Gasoline Engine with CAI Combustion”, SAE Paper 2002-01-0420, 2002.
[45] Osborne, R. J., Li, G., Sapsford, S. M., Stoke, J., Lake, T. H., Heikal, M. R.,
“Evaluation of HCCI for Future Gasoline Powertrains”, SAE Paper 2003-01-0750, 2003.
[46] Cairns, A., Blaxill, H., Irlam, G., “Exhaust Gas Recirculation for Improved Part and
Full Load Fuel Economy in a Turbocharged Gasoline Engine”, SAE Paper 2006-01-0047,
2006.
[47] Friedfeldt, R., Zenner, T., Ernst, R., Fraser, A., “Three-cylinder Gasoline Engine with
Direct Injection”, ATZ Autotechnology, Vol. 12, No. 2, 2012.
[48] Hancock, D., Fraser, N., Jeremy, M., Sykes, R., Blaxill, H., “A New 3 Cylinder 1.2l
Advanced Downsizing Technology Demonstrator Engine”, SAE Paper 2008-01-0611,
2008.
[49] http://www.mce-5.com/english/index.html accessed on 24/11/2012 (see

appendix A.5).
[50] Turner, J., Pearson, R., Luard, N., “Investigation of Appropriate Technologies for a
Highly-downsized DISI Engine”, Engine Downsizing Conference, Institute of Mechanical
Engineers, 2011.
[51] Lumsden, G., OudeNijeweme, D., Fraser, N., Blaxill, H., “Development of a
Turbocharged Direct Injection Downsizing Demonstrator Engine”, SAE Paper 2009-01-
1503, 2009.
[52] http://en.wikipedia.org/wiki/Twincharger/ accessed 23/11/2012 (see appendix

A.5).
[53] Copeland, C., Martinex-Botas, R., Turner, J., Pearson, R., Luard, N., Carey, C.,
Richardson, S., Di Martino, P., Chobola, P., “Boost System Selection for a Heavily
Downsized Spark Ignition Prototype Engine”, Tenth IMechE International Conference
on Turbochargers and Turbocharging, 2012.
293
[54] Fraser, N. and Bassett, M., “Extreme Engine Downsizing with a Single
Turbocharger – 100kW/l and 30 bar BMEP”, Engine Downsizing Conference, Institute
of Mechanical Engineers, 2011.
[55] Fraser, N., Blaxill, H., Lumsden, G., Bassett, M., “Challenges for Increased
Efficiency Through Gasoline Engine Downsizing”, SAE Paper 2009-01-1053, 2009.
[56] Amann, M., Alger, T., Mehta, D., “The Effect of EGR on Low-speed Pre-ignition in
Boosted SI Engines”, SAE Paper 2011-01-0339, 2011.
[57] Anderson, M. K., Assanis, D. N., Filipi, Z. S., “First and Second Law Analyses of a
Naturally-aspirated, Miller Cycle, SI Engine with Late Intake Valve Closure”, SAE Paper
980889, 1998.
[58] Miller, R., “Supercharging and Internal Cooling Cycle for High Output”, Oil and Gas
Power Division Proceedings of the National Conference, ASME 46-OGP-4, pp 1-5, 1946.
[59] http://world.honda.com/powerproducts-technology/exlink/ accessed 23/11/2012

(see appendix A.5).
[60] Kapus, P. E., Denger, D., Holland, T., “Intelligent Simplification – Ways Towards
Improved Fuel Economy”, SAE Paper 2002-01-0236, 2002.
[61] Stansfield, P. A., Wigley, G., Garner, C. P., Patel, R., Ladommatos, N., Pitcher, G.,
Turner, J. W. G., Nuglisch, H., Helie, J., “Unthrottled Engine Operation using Variable
Valve Actuation: The Impact on the Flow Field, Mixing and Combustion”, SAE Paper
2007-01-1414, 2007.
[62] Stansfield, P. A., Wigley, G., Garner, C. P., Patel, R., Ladommatos, N., Pitcher, G.,
Turner, J. W. G., Nuglisch, H., “Comparison Between Unthrottled, Single and Two-valve
Induction Strategies Utilising Direct Gasoline Injection: Emissions, Heat-release and
Fuel Consumption Analysis”, SAE Paper 2008-01-1626, 2008.
[63] Battistoni, M. and Mariani, F., “Fluid Dynamic Study of Unthrottled Part Load SI
Engine Operations with Asymmetric Valve Lifts”, SAE Paper 2009-24-0017, 2009.
[64] Taylor, J., Fraser, N., Dingelstadt, R., Hoffmann, H., “Benefits of Late Inlet Valve
Timing Strategies Afforded Through the Use of Intake Cam in Cam Applied to a
Gasoline Turbocharged Downsized Engine”, SAE Paper 2011-01-0360, 2011.
[65] Moore, W., Foster, M., Lai, M. C., Xie, X. B., Zheng, Y., Matsumoto, A., “Charge
Motion Benefits of Valve Deactivation to Reduce Fuel Consumption and Emissions in a
GDI, VVA Engine”, SAE Paper 2011-01-1221, 2011.
294
[66] Matsumoto, A., Zheng, Y., Xie, X., Lai, M. C., Moore, W., “Interactions of Multi-
hole DI Sprays with Charge Motion and Their Implications to Flexible Valve-trained
Engine Performance”, SAE Paper 2011-01-1883, 2011.
[67] Cleary, D. and Silvas, G., “Unthrottled Engine Operation with Variable Intake Valve
Lift, Duration, and Timing”, SAE Paper 2007-01-1282, 2007.
[68] Sellnau, M., Kunz, T., Sinnamon, J., Burkhard, J., “2-step Variable Valve Actuation:
System Optimization and Integration on an SI Engine”, SAE Paper 2006-01-0040, 2006.
[69] Turner, J. W. G., Pitcher, G., Burke, P., Garner, C. P., Wigley, G., Stansfield, P.,
Nuglisch, H., Ladommatos, N., Patel, E., Williams, P., “The HOTFIRE Homogeneous GDI
and Fully Variable Valve Train Project – An Initial Report”, SAE Paper 2006-01-1260,
2006.
[70] Hara, S., Nakajima, Y., Nagumo, S. I., “Effects of Intake-valve Closing Timing on
Spark-ignition Engine Combustion”, SAE Paper 850074, 1985.
[71] Akihisa, D. and Sawada, D., “Research on Improving Thermal Efficiency Through
Variable Super-high Expansion Ratio Cycle”, SAE Paper 2010-01-0174, 2010.
[72] Sellnau, M. and Rask, E., “Two-step Variable Valve Actuation for Fuel Economy,
Emissions, and Performance”, SAE Paper 2003-01-0029, 2003.
[73] Lee, K., Bae, C., Kang, K., “The Effects of Tumble and Swirl Flows on Flame
Propagation in a Four-valve S.I. Engine”, Applied Thermal Engineering, Vol. 27, pp
2122-2130, Elsevier, 2007.
[74] Hitomi, M., Sasaki, J., Hatamura, K., Yano, Y., “Mechanism of Improving Fuel
Efficiency by Miller Cycle and Its Future Prospect”, SAE Paper 950974, 1995.
[75] Ishizuki, Y., Shimizu, Y., Hikino, H., Kawashima, Y., “A New Type of Miller
Supercharging System for High Speed Engines Part 2 – Realization of High BMEP Diesel
Engines”, SAE Paper 851523, 1985.
[76] Kamo, R., Mavinahally, N. S., Bryzik, W., Reid, M., “Emission Characteristics of a
Turbocharged Multi-cylinder Miller Cycle Diesel Engine”, United States of America
Department of Energy, 1995.
[77] Kado, T., Osaka, Y., Noguchi, T., Horimoto, K., Ogawa, M., Narita, H., “The
Development of 1 MW Class Gas Engine and its Application to Cogeneration Systems”,
Technical Review, Vol. 45, No. 1, 2008.
[78] http://www.getransportation.com/rail/rail-products/locomotives/powerhaul-
series-locomotive.html accessed 23/11/2012 (see appendix A.5).
295
[79] Sakai, H., Noguchi, H., Kawauchi, M., and Kanesaka, H., “A New Type of Miller
Supercharging System for High-speed Engines - Part 1 Fundamental Considerations and
Application to Gasoline Engines”, SAE Paper 851522, 1985.
[80] Goto, T., Hatamura, K., Takizawa, S., Hayama, N., Abe, H., Kaneska, H.,
“Development of V6 Miller Cycle Gasoline Engine”, SAE Paper 940198, 1994.
[81] http://www.edmunds.com/mazda/millenia/2002/features-specs.html accessed

28/11/2012 (see appendix A.5).
[82] Takabe, S., Hatamura, K., Kanesaka, H., Kurata, H., Iguchi, Y., Matsubara, H.,
“Development of the High Performance Lysholm Compressor for Automotive Use”, SAE
Paper 940843, 1994.
[83] Ichimaru, K., Sakai, H., Kanesaka, H., Winterborne, D. E., “A Rotary Valve
Controlled High Expansion Ratio Gasoline Engine”, SAE Paper 940815, 1994.
[84] Ueda, N., Sakai, H., Iso, N., Sasaki, J., “A Naturally Aspirated Miller Cycle Gasoline
Engine – Its Capability of Emission, Power and Fuel Economy”, SAE Paper 960589,
1996.
[85] The Law Concerning Special Measures for Total Emission Reduction of Nitrogen
Oxides from Automobiles in Specified Areas: Law No. 70 of 1992, Ministry of the
Environment: Government of Japan, 1992.
[86] Blakey, S. C., Saunders, R. J., Ma, T. H., Chopra, A., “A Design and Experimental
Study of an Otto Atkinson Cycle Engine Using Late Intake Valve Closing”, SAE Paper
910451, 1991.
[87] Rabia, S. M. and Korah, N. S., “Knocking Phenomena in a Gasoline Engine with
Late-intake Valve Closing”, SAE Paper 920381, 1992.
[88] Urata, Y., Umlyama, H., Shimizu, K., Fujiyoshi, Y., Sono, H., Fukuo, K., “A Study of
Vehicle Equipped with Non-throttling S.I. Engine with Early Intake Valve Closing
Mechanism”, SAE Paper 930820, 1993.
[89] Boggs, D. L., Hilbert, H. S., Schechter, M. M., “The Otto-atkinson Cycle Engine -
Fuel Economy and Emissions Results and Hardware Design”, SAE Paper 950089, 1995.
[90] Theobald, M. A., Lequesne, B., Henry, R., “Control of Engine Loads via
Electromagnetic Valve Actuators”, SAE Paper 940816, 1994.
[91] Tuttle, H., “Controlling Engine Load by Means of Late Intake Valve Closing”, SAE
Paper 800794, 1980.
296
[92] Tuttle, H., “Controlling Engine Load by Means of Early Intake Valve Closing”, SAE
Paper 820408, 1982.
[93] Ke, Y. and Pulcher, H., “Controlling the Load and the Boost Pressure of a
Turbocharged SI Engine by Means of Early Intake-valve Closing”, SAE Paper 960588,
1996.
[94] Söderberg, F. and Johansson, B., “Fluid Flow, Combustion and Efficiency with Early
or Late Inlet Valve Closing”, SAE Paper 972937, 1997.
[95] Koga, H. and Watanabe, S., “Research on Extended Expansion General-purpose

Engine – Heat Release and Friction”, SAE Paper 2007-32-0003, 2007.
[96] Zhao, F., Lai, M. C., Harrington, D. L., “Automotive Spark-ignited Direct-injection
Gasoline Engines”, Progress in Energy and Combustion Science, Vol. 25, pp 437-562,
Pergamon, 1999.
[97] Wang, Y., Lin, L., Roskilly, A. P., Zeng, S., Huang, J., He, Y., Huang, X., Huang, H.,
Wei, H., Li, S., Yang, J., “An Analytic Study of Applying Miller Cycle to Reduce NOx
Emission from Petrol Engine”, Applied Thermal Engineering, Vol. 27, pp 1779-1789,
Elsevier, 2007.
[98] Gould, L. A., Richeson, W. E., Erickson, F. L., “Performance Evaluation of a Camless
Engine Using Valve Actuators with Programmable Timing”, SAE Paper 910450, 1991.
[99] Sakata, Y., Yamana, K., Nishida, K., Shimuzu, T., Shiga, S., Araki, M., Nakamura, H.,
Obokata, T., “A Study on Optimization of an Over-expansion Cycle Gasoline Engine
with Late-closing of Intake Valves”, SAE Paper 2007-24-0089, 2007.
[100] Fraser, N., Korte, V., Rückauf, J., Harms, K., Miersch, J., Brandt, M., Münz, S.,
Rauscher, M., “Das MAHLE-Bosch Downsizing Demonstrator FahrzeugI”, Aachener
colloquium Fahrzeug- und Motorentechnik 2010, 2010.
[101] ANSYS, Release 11.0, ANSYS Inc, Canonsburg, PA, 2007.
[102] Pereira, R. C. C. and Pasa, V. M. D., “Effect of Alcohol and Copper Content on the
Stability of Automotive Gasoline”, Energy Fuels, Vol. 19, No. 2, pp 426-432, 2005.
[103] Taylor, C. F., “The Internal Combustion Engine in Theory and Practice, Volume 1”,
MIT press, 1960.
[104] ETAS INCA V7.0, ETAS GmbH, Stuttgart, Germany, 2011.
[105] Rassweiler, G. M. and Withrow, L., “Motion Pictures of Engine Flames Correlated
with Pressure Cards”, SAE Transactions, Vol. 83, pp 185-204, 1938.
297
[106] Brunt, M. and Pond, C., “Evaluation of Techniques for Absolute Cylinder Pressure
Correction”, SAE Paper 970036, 1997.
[107] Brunt, M. and Emtage, A., “Evaluation of IMEP Routines and Analysis Errors”, SAE
Paper 960609, 1996.
[108] Wiebe, L., “Halbermpirische Formel für die Verbrennungsgeschwindigkeit”,

Verlag der Akademie der Wissenschaften der UdSSR, 1967.
[109] Heywood, J. B., Higgins, J. M., Watts, P. A., Tabaczynski, R. J., “Development and
Use of a Cycle Simulation to Predict SI Engine Efficiency and NOx Emissions”, SAE Paper
790291, 1979.
[110] Blizard, N. C. and Keck, C. J., “Experimental and Theoretical Investigation of

Turbulent Burning Model for Internal Combustion Engines”, SAE Paper 740191, 1974.
[111] Merdjani, S. and Sheppard, C. G. W., “Gasoline Engine Cycle Simulation using the
Leeds Turbulent Burning Velocity Correlations”, SAE Paper 932640, 1993.
[112] GT-Suite Help Navigator, Gamma Technologies Inc, Westmont, Illinois, 2004.
[113] Livengood, J. C. and Wu, P. C., “Correlation of Auto-ignition Phenomena in

Internal Combustion Engines and Rapid Compression Machines”, The Fifth
International Symposium on Combustion, 1955.
[114] Douaud, A. M. and Eyzat, P., “Four-octane Number Method for Predicting the
Anti-knock Behaviour of Fuels in Engines”, SAE Paper 780080, 1978.
[115] Rothmaier, A., “NGM 1.2 GT-Power Baseline Correlation”, Mahle Powertrain
Internal Report, 2010.
[116] Eriksson, L., Johansson, E., Kettaneh-Wold, N., Wilkström, C., Wold, S., “Design of
Experiments: Principles and Applications”, Umetrics, 2000.
[117] McKay, M. D., Beckman, R. J., Connover, W. J., “A Comparison of Three Methods
for Selecting Values of Input Variables in the Analysis of Output From a Computer
Code”, Technometrics, Vol. 42, pp 55-61, 2000.
[118] Taylor, J., Fraser, N., Wieske, P., “Water Cooled Exhaust Manifold and Full Load
EGR Technology Applied to a Downsized Direct Injection Spark Ignition Engine”. SAE
Paper 2010-01-0356, 2010.
[119] Sonntag, R. E. and Borgnakke, C., “Fundamentals of Thermodynamics”, Eighth

edition, Wiley, 2013.
[120] Smith, R. A. and Wilkins, P. S. A. W., “Low Pressure Sand Casting: Current
Experience with a New Process”, AFS Transactions, pp 785-792, 1986.
298
[121] Robertson, A., Hartland, J., Gil-Martinez, A., “Heat Flux Estimation and the
Control of Nucleate Boiling in a Laboratory Test Rig”, SAE Paper 2005-01-2007, 2005.
[122] Zahdeh, A., Rothenberger, P., Nguyen, W., Anbarasu, M., Schmuck-Soldan, S.,
Schaefer, J., Goebel, T., “Fundamental Approach to Investigate Pre-ignition in Boosted
SI Engines”, SAE Paper 2011-01-0340, 2011.
[123] Sasaki, N., Nakata, K., Kawatake, K., Sagawa, S., Watanabe, M., Sone, T., “The
Effect of Fuel Compounds on Pre-ignition Under High Temperature and High Pressure
Condition”, SAE Paper 2011-01-1984, 2011.
299
Appendix
A.1 Cam Profiles
Exhaust Cam
Intake Cam Profile Lift (mm) Profile Lift (mm)
Cam
Angle 152 CAD 240 CAD 292 CAD 276 CAD
-90 0 0 0 0
-89.5 0 0 0 0
-89 0 0 0 0
-88.5 0 0 0 0
-88 0 0 0 0
-87.5 0 0 0 0
-87 0 0 0 0
-86.5 0 0 0 0
-86 0 0 0 0
-85.5 0 0 0 0
-85 0 0 0 0
-84.5 0 0 0 0
-84 0 0 0 0
-83.5 0 0 0 0
-83 0 0 7.28E-07 0
-82.5 0 0 5.1E-06 0
-82 0 0 2.26E-05 0
-81.5 0 0 7.28E-05 0
-81 0 0 0.000195 0
-80.5 0 0 0.000454 0
-80 0 0 0.00095 0
-79.5 0 0 0.001808 0
-79 0 0 0.003166 0
300
-78.5 0 0 0.005128 0
-78 0 0 0.007755 0
-77.5 0 0 0.011043 0
-77 0 0 0.014955 0
-76.5 0 0 0.019426 0
-76 0 0 0.024384 0
-75.5 0 0 0.029749 0
-75 0 0 0.035443 0
-74.5 0 0 0.041391 0
-74 0 0 0.047542 0
-73.5 0 0 0.053873 0.00021603
-73 0 0 0.060426 4.32E-04
-72.5 0 0 0.067302 0.00129617
-72 0 0 0.074692 0.00216028
-71.5 0 0 0.082848 0.00518011
-71 0 0 0.092094 0.00819994
-70.5 0 0 0.102777 0.013412325
-70 0 0 0.115281 0.01862471
-69.5 0 0 0.12996 0.02581042
-69 0 0 0.147183 0.03299613
-68.5 0 0 0.167246 0.04163922
-68 0 0 0.190452 0.05028231
-67.5 0 0 0.217007 0.06046157
-67 0 0 0.247122 0.07064083
-66.5 0 0.000236 0.280906 0.08435545
-66 0 0.000471 0.318484 0.09807007
-65.5 0 0.001413 0.359884 0.11804255
-65 0 0.002356 0.405167 0.13801503
-64.5 0 0.005568 0.454308 0.16620477
-64 0 0.008781 0.507328 0.19439451
-63.5 0 0.014157 0.564174 0.231860675
-63 0 0.019532 0.624843 0.26932684
-62.5 0 0.026117 0.689265 0.3165987
301
-62 0 0.032702 0.757428 0.36387056
-61.5 0 0.039838 0.829249 0.42114724
-61 0 0.046974 0.904704 0.47842392
-60.5 0 0.055465 0.983702 0.54568351
-60 0 0.063955 1.066207 0.6129431
-59.5 0 0.075863 1.152118 0.6899893
-59 0 0.08777 1.241385 0.7670355
-58.5 0 0.105504 1.333898 0.85352975
-58 0 0.123237 1.429591 0.940024
-57.5 0 0.148944 1.528341 1.03550565
-57 0 0.17465 1.630068 1.1309873
-56.5 0 0.209846 1.734634 1.23489
-56 0 0.245043 1.841939 1.3387927
-55.5 0 0.290575 1.951832 1.45045805
-55 0 0.336108 2.064192 1.5621234
-54.5 0 0.392334 2.178848 1.68081145
-54 0 0.44856 2.295655 1.7994995
-53.5 0 0.515502 2.414421 1.9243951
-53 0 0.582444 2.534973 2.0492907
-52.5 0 0.659919 2.657098 2.1795061
-52 0 0.737394 2.780604 2.3097215
-51.5 0 0.825091 2.905278 2.4442979
-51 0 0.912789 3.03093 2.5788743
-50.5 0 1.010309 3.157371 2.7167828
-50 0 1.107829 3.284438 2.8546913
-49.5 0 1.214711 3.411976 2.99485325
-49 0 1.321592 3.539856 3.1350152
-48.5 0 1.437346 3.667959 3.276355
-48 7.28E-07 1.5531 3.796181 3.4176948
-47.5 5.1E-06 1.677199 3.924429 3.5592215
-47 2.26E-05 1.801298 4.052619 3.7007482
-46.5 7.28E-05 1.933107 4.180675 3.8416804
-46 0.000195 2.064917 4.308526 3.9826126
302
-45.5 0.000454 2.203632 4.436109 4.1224098
-45 0.00095 2.342347 4.563358 4.262207
-44.5 0.001808 2.486946 4.690221 4.400499
-44 0.003166 2.631545 4.816638 4.538791
-43.5 0.005129 2.780842 4.942562 4.6753072
-43 0.007756 2.930139 5.067938 4.8118234
-42.5 0.011044 3.082874 5.192724 4.94635705
-42 0.014957 3.235609 5.316869 5.0808907
-41.5 0.019428 3.390545 5.440334 5.21327685
-41 0.024387 3.54548 5.563071 5.345663
-40.5 0.029752 3.701464 5.685045 5.475767
-40 0.035447 3.857448 5.806209 5.605871
-39.5 0.041398 4.013478 5.926531 5.73358
-39 0.047559 4.169507 6.045966 5.861289
-38.5 0.053919 4.324772 6.164485 5.9865048
-38 0.06054 4.480037 6.282044 6.1117206
-37.5 0.067561 4.633911 6.398616 6.2343563
-37 0.075227 4.787785 6.514158 6.356992
-36.5 0.083857 4.93979 6.628646 6.4769705
-36 0.093864 5.091795 6.742039 6.596949
-35.5 0.105684 5.241558 6.854315 6.7142007
-35 0.119803 5.39132 6.965437 6.8314524
-34.5 0.136665 5.538542 7.075384 6.9459142
-34 0.156733 5.685763 7.18412 7.060376
-33.5 0.180381 5.8302 7.291629 7.17198935
-33 0.207987 5.974636 7.397877 7.2836027
-32.5 0.239806 6.116081 7.502851 7.39231485
-32 0.2761 6.257526 7.606518 7.501027
-31.5 0.317 6.395807 7.708867 7.6067888
-31 0.362651 6.534088 7.809868 7.7125506
-30.5 0.413074 6.66905 7.909513 7.8153162
-30 0.468323 6.804013 8.007772 7.9180818
-29.5 0.528334 6.935522 8.10464 8.0178084
303
-29 0.593093 7.067031 8.200088 8.117535
-28.5 0.66248 7.194965 8.294114 8.2141825
-28 0.736429 7.322899 8.38669 8.31083
-27.5 0.81478 7.447147 8.477815 8.40436
-27 0.897426 7.571394 8.567462 8.49789
-26.5 0.984174 7.691855 8.655631 8.5882685
-26 1.074882 7.812316 8.742297 8.678647
-25.5 1.169328 7.928897 8.827462 8.7658385
-25 1.267335 8.045478 8.911101 8.85303
-24.5 1.368651 8.158093 8.993216 8.9370035
-24 1.473052 8.270708 9.073784 9.020977
-23.5 1.580239 8.379279 9.152808 9.10170125
-23 1.689918 8.487849 9.230265 9.1824255
-22.5 1.801713 8.592301 9.30616 9.25987125
-22 1.915238 8.696754 9.380471 9.337317
-21.5 2.030051 8.797018 9.453201 9.4114565
-21 2.145719 8.897282 9.52433 9.485596
-20.5 2.261805 8.993294 9.593862 9.5564015
-20 2.377905 9.089305 9.661778 9.627207
-19.5 2.493649 9.181004 9.728082 9.694653
-19 2.608699 9.272702 9.792754 9.762099
-18.5 2.722758 9.360031 9.855801 9.82616125
-18 2.835545 9.447359 9.917203 9.8902235
-17.5 2.946824 9.530263 9.976966 9.95087875
-17 3.056356 9.613167 10.03507 10.011534
-16.5 3.163954 9.691595 10.09153 10.068761
-16 3.269416 9.770023 10.14631 10.125988
-15.5 3.372594 9.843927 10.19943 10.1797665
-15 3.47332 9.917831 10.25088 10.233545
-14.5 3.571479 9.987166 10.30065 10.28385625
-14 3.66693 10.0565 10.34872 10.3341675
-13.5 3.759588 10.12122 10.39512 10.38099275
-13 3.849334 10.18594 10.43982 10.427818
304
-12.5 3.936107 10.24601 10.48282 10.471142
-12 4.019804 10.30608 10.52412 10.514466
-11.5 4.100381 10.36146 10.56371 10.554274
-11 4.177749 10.41684 10.60159 10.594082
-10.5 4.251877 10.46749 10.63777 10.630359
-10 4.322687 10.51814 10.67222 10.666636
-9.5 4.390158 10.56404 10.70495 10.699372
-9 4.454219 10.60993 10.73595 10.732108
-8.5 4.514859 10.65104 10.76524 10.7612905
-8 4.572012 10.69215 10.79279 10.790473
-7.5 4.625675 10.72844 10.81861 10.8160935
-7 4.675787 10.76473 10.84269 10.841714
-6.5 4.722347 10.79619 10.86504 10.8637655
-6 4.765299 10.82765 10.88564 10.885817
-5.5 4.804646 10.85425 10.90451 10.904292
-5 4.840334 10.88085 10.92162 10.922767
-4.5 4.872369 10.90257 10.937 10.9376615
-4 4.900701 10.92429 10.95063 10.952556
-3.5 4.92534 10.94112 10.96251 10.963866
-3 4.946238 10.95795 10.97264 10.975176
-2.5 4.96341 10.96988 10.98102 10.9828995
-2 4.976813 10.9818 10.98765 10.990623
-1.5 4.986467 10.98881 10.99253 10.9947605
-1 4.992333 10.99582 10.99565 10.998898
-0.5 4.994435 10.99791 10.99702 10.999449
0 4.99274 11 10.99664 11
0.5 4.987277 10.99716 10.9945 10.9969685
1 4.978017 10.99433 10.99061 10.993937
1.5 4.964997 10.98656 10.98497 10.987326
2 4.948192 10.97879 10.97757 10.980715
2.5 4.927645 10.9661 10.96843 10.9705315
3 4.903336 10.9534 10.95753 10.960348
3.5 4.875313 10.93578 10.94488 10.9465985
305
4 4.843562 10.91817 10.93048 10.932849
4.5 4.808135 10.89563 10.91433 10.9155415
5 4.769022 10.87309 10.89643 10.898234
5.5 4.726279 10.84564 10.8768 10.8773805
6 4.679904 10.81819 10.85541 10.856527
6.5 4.629954 10.78585 10.83229 10.8321385
7 4.576432 10.7535 10.80742 10.80775
7.5 4.519404 10.71627 10.78081 10.7798395
8 4.458879 10.67904 10.75247 10.751929
8.5 4.394928 10.63693 10.7224 10.72051225
9 4.32757 10.59483 10.6906 10.6890955
9.5 4.256886 10.54788 10.65707 10.65418775
10 4.182905 10.50094 10.62181 10.61928
10.5 4.105717 10.44916 10.58484 10.580898
11 4.025365 10.39738 10.54614 10.542516
11.5 3.941951 10.3408 10.50573 10.5006785
12 3.855537 10.28423 10.46361 10.458841
12.5 3.766239 10.22287 10.41978 10.4135695
13 3.674139 10.16152 10.37425 10.368298
13.5 3.579375 10.09543 10.32702 10.31961175
14 3.482054 10.02934 10.2781 10.2709255
14.5 3.382338 9.958539 10.22749 10.21884975
15 3.280362 9.887739 10.17519 10.166774
15.5 3.176315 9.812269 10.12121 10.1113305
16 3.070364 9.736799 10.06556 10.055887
16.5 2.962725 9.656701 10.00824 9.997102
17 2.8536 9.576603 9.949247 9.938317
17.5 2.743234 9.491921 9.888605 9.876216
18 2.631865 9.407238 9.826306 9.814115
18.5 2.51977 9.318021 9.762367 9.7487245
19 2.407228 9.228803 9.696782 9.683334
19.5 2.294547 9.135103 9.629571 9.614683
20 2.182051 9.041402 9.560728 9.546032
306
20.5 2.070079 8.943275 9.490273 9.474149
21 1.958991 8.845148 9.418201 9.402266
21.5 1.84914 8.742655 9.344534 9.3271815
22 1.740899 8.640162 9.269265 9.252097
22.5 1.634603 8.533368 9.192417 9.173843
23 1.5306 8.426574 9.113985 9.095589
23.5 1.429179 8.315548 9.033993 9.0141985
24 1.33065 8.204522 8.952436 8.932808
24.5 1.235256 8.08934 8.869338 8.8483165
25 1.143272 7.974157 8.784696 8.763825
25.5 1.0549 7.854897 8.698535 8.676268
26 0.970387 7.735638 8.610853 8.588711
26.5 0.889897 7.612388 8.521675 8.498127
27 0.81364 7.489138 8.431001 8.407543
27.5 0.741731 7.36199 8.338857 8.3139715
28 0.674334 7.234842 8.245246 8.2204
28.5 0.611508 7.103897 8.150193 8.123884
29 0.553362 6.972952 8.053702 8.027368
29.5 0.499895 6.838317 7.955803 7.92794965
30 0.451149 6.703682 7.856499 7.8285313
30.5 0.40705 6.565477 7.755822 7.72625965
31 0.367557 6.427271 7.653779 7.623988
31.5 0.332503 6.285625 7.550402 7.5189113
32 0.301735 6.143978 7.445702 7.4138346
32.5 0.274972 5.999035 7.339711 7.3060055
33 0.251929 5.854093 7.232445 7.1981764
33.5 0.232207 5.706016 7.123937 7.0876512
34 0.215406 5.55794 7.014207 6.977126
34.5 0.201044 5.406918 6.903291 6.8639645
35 0.188664 5.255896 6.79121 6.750803
35.5 0.177783 5.102155 6.678004 6.63506985
36 0.167984 4.948414 6.563698 6.5193367
36.5 0.158889 4.792236 6.448335 6.4011011
307
37 0.15022 4.636057 6.331942 6.2828655
37.5 0.141774 4.477812 6.214564 6.16220355
38 0.133434 4.319567 6.096232 6.0415416
38.5 0.125144 4.159788 5.976992 5.9185368
39 0.116889 4.00001 5.85688 5.795532
39.5 0.108679 3.83952 5.735944 5.670278
40 0.100539 3.67903 5.61422 5.545024
40.5 0.092501 3.519037 5.491761 5.417627
41 0.084603 3.359044 5.368607 5.29023
41.5 0.076886 3.201033 5.244812 5.1608127
42 0.069394 3.043021 5.120422 5.0313954
42.5 0.062165 2.888521 4.995493 4.9001037
43 0.055243 2.734021 4.870076 4.768812
43.5 0.04866 2.584468 4.744232 4.635821
44 0.042455 2.434914 4.618018 4.50283
44.5 0.036651 2.291627 4.491499 4.368358
45 0.031278 2.148339 4.364738 4.233886
45.5 0.02635 2.01254 4.237808 4.09821475
46 0.021884 1.87674 4.110782 3.9625435
46.5 0.017883 1.749547 3.983739 3.8260541
47 0.014349 1.622355 3.856764 3.6895647
47.5 0.011273 1.504749 3.729945 3.55278765
48 0.008643 1.387143 3.603382 3.4160106
48.5 0.006438 1.27989 3.477176 3.27963065
49 0.004634 1.172637 3.351442 3.1432507
49.5 0.0032 1.076205 3.226293 3.00803805
50 0.002105 0.979773 3.101863 2.8728254
50.5 0.001306 0.894347 2.97828 2.7396078
51 0.000761 0.808921 2.855694 2.6063902
51.5 0.000414 0.734523 2.734244 2.47605895
52 0.000212 0.660125 2.614099 2.3457277
52.5 0.000102 0.596683 2.495407 2.2192391
53 4.53E-05 0.533241 2.378347 2.0927505
308
53.5 1.79E-05 0.480564 2.263072 1.9711282
54 5.89E-06 0.427888 2.149768 1.8495059
54.5 1.4E-06 0.385597 2.038583 1.73381305
55 2E-07 0.343307 1.929699 1.6181202
55.5 0 0.310758 1.823257 1.50939285
56 0 0.27821 1.719427 1.4006655
56.5 0 0.254335 1.618334 1.2998511
57 0 0.23046 1.520133 1.1990367
57.5 0 0.213599 1.424926 1.10694265
58 0 0.196737 1.332853 1.0148486
58.5 0 0.184664 1.243994 0.932099465
59 0 0.17259 1.158471 0.84935033
59.5 0 0.162851 1.076345 0.776365465
60 0 0.153112 0.997724 0.7033806
60.5 0 0.143965 0.922652 0.64036415
61 0 0.134817 0.851223 0.5773477
61.5 0 0.125685 0.783467 0.52427295
62 0 0.116553 0.719466 0.4711982
62.5 0 0.107416 0.659234 0.427776025
63 0 0.098279 0.602838 0.38435385
63.5 0 0.089141 0.550273 0.349973265
64 0 0.080004 0.501586 0.31559268
64.5 0 0.070922 0.456742 0.28923157
65 0 0.061839 0.41575 0.26287046
65.5 0 0.053082 0.378534 0.243068905
66 0 0.044324 0.345044 0.22326735
66.5 0 0.036485 0.315138 0.20827593
67 0 0.028646 0.288687 0.19328451
67.5 0 0.022349 0.265462 0.18130754
68 0 0.016052 0.245236 0.16933057
68.5 0 0.011613 0.227688 0.158891145
69 0 0.007174 0.212498 0.14845172
69.5 0 0.004595 0.199281 0.138611075
309
70 0 0.002016 0.187668 0.12877043
70.5 0 0.001296 0.177271 0.119067715
71 0 0.000576 0.167752 0.109365
71.5 0 0.000288 0.158806 0.09968151
72 0 0 0.150206 0.08999802
72.5 0 0 0.141787 0.08041351
73 0 0 0.133456 0.070829
73.5 0 0 0.125167 0.061529965
74 0 0 0.116911 0.05223093
74.5 0 0 0.1087 0.0436178
75 0 0 0.100558 0.03500467
75.5 0 0 0.092518 0.02770376
76 0 0 0.084619 0.02040285
76.5 0 0 0.0769 0.01494349
77 0 0 0.069406 0.00948413
77.5 0 0 0.062176 0.00612099
78 0 0 0.055252 0.00275785
78.5 0 0 0.048668 0.001772905
79 0 0 0.042461 7.88E-04
79.5 0 0 0.036657 0.00039398
80 0 0 0.031282 0
80.5 0 0 0.026354 0
81 0 0 0.021887 0
81.5 0 0 0.017884 0
82 0 0 0.01435 0
82.5 0 0 0.011273 0
83 0 0 0.008644 0
83.5 0 0 0.006438 0
84 0 0 0.004635 0
84.5 0 0 0.003201 0
85 0 0 0.002105 0
85.5 0 0 0.001306 0
86 0 0 0.000761 0
310
86.5 0 0 0.000415 0
87 0 0 0.000212 0
87.5 0 0 0.000102 0
88 0 0 4.54E-05 0
88.5 0 0 1.79E-05 0
89 0 0 5.91E-06 0
89.5 0 0 1.4E-06 0
90 0 0 2E-07 0
90.5 0 0 0 0
91 0 0 0 0
91.5 0 0 0 0
92 0 0 0 0
92.5 0 0 0 0
93 0 0 0 0
93.5 0 0 0 0
94 0 0 0 0
94.5 0 0 0 0
95 0 0 0 0
95.5 0 0 0 0
96 0 0 0 0
96.5 0 0 0 0
97 0 0 0 0
97.5 0 0 0 0
98 0 0 0 0
98.5 0 0 0 0
99 0 0 0 0
99.5 0 0 0 0
100 0 0 0 0
A.2 Sensor List
Thermocouples:
311
• Exhaust temp
• Exhaust manifold temp
• Plenum air temp
PRT's:
• Head and block coolant temp in

• Head coolant temp out
• Block coolant temp out
• Oil temp in
• Oil temp out
• Low pressure fuel temp
• Intake port temp
Voltage output:
• Flow meters
o Total coolant (Apollo RN3/15)

o Block coolant (Apollo RN3/10)
o Fuel (ABB FCM2000)
• Pressure sensors
o Oil (Druck PTX 1400 0-6 bar)

o Barometric (Mercury barometer)
o Low pressure fuel (Druck PTX 1400 0-10 bar)
o Coolant (Druck PTX 1400 0-4 bar)

o Average exhaust (Druck PTX 1400 0-10 bar)
o Intake (Kistler 4005BA10FA0)
o Exhaust (Kistler 4005BA10FA0)
o In-cylinder (Kistler 6041A0)
312
• Exhaust gas analysers
o NOx (Rotork Analysis Model 443)

o HC (Rotork Analysis Model 523)
Apollo RN3
Range 2 - 20 l/min and 5 - 50 l/min

Pressure range 1 - 36 bar abs
Temperature range -30 - 110 degC
Linearity ± 0.5%
Repeatability ± 0.1%
Size 0.5" BSP and 0.75" BSP
Construction 316 Stainless Steel
Warm up N/A
Output 0 - 10V
ABB FCM2000
Range 0 - 65 kg/hr
Linearity ± 0.15%
Size DN1.5
Warm up 30 mins
Output 0 - 10V
Druck PTX1400
313
Range 1 - 5 , 1 - 7 and 1 -11 bar abs
Pressure range 1 - 5 , 1 - 7 and 1 -11 bar abs
Linearity N/A
Size 0.25" BSP
Warm up N/A
Output 4 - 20mA
Kistler 4005BA10FA0
Range 0 - 10 bar abs

Temperature range -20 - 125 degC (1100deg C w/cooling adapter)
Linearity ± 0.2%
Size M5 (M10 w/cooling adapter)
Warm up 30 mins
Output 0 - 10V
Kistler 6041A0
Range 0 - 250 bar abs

Linearity ± 0.5%
Size M8
Warm up 30 mins
Output 0 - 10V
314
Rotork Analysis 443 FID analyser
Range 0 - 10000 ppm

Pressure range N/A, 1.5l/min sample fow
Temperature range 10 - 30 degC
Linearity ± 1%
Repeatability ± 1%
Size 3 x 19" rack mount
Warm up 40 mins
Output Analogue display
Rotork Analysis 523 Chemiluminescence analyser
Range 0 - 10000 ppm

Pressure range N/A, 3l/min sample fow
Temperature range 10 - 30 degC
Linearity ± 1%
Repeatability ± 1%
Size 19" rack mount
Warm up 40 mins
Output Analogue display
A.3 Feedback Descriptions
Label Description
Speed Engine speed. Input manually
Torque Engine brake torque. As fedback from the

dynamometer force transducer. High speed
315
feedback
Lambda Feedback from independent lambda sensor.

Not used for this test work. This data is saved
using INCA instead
Thrt. Opn. Throttle angle. Not used for this test work. This
data is saved using INCA instead
Oil Temp. Oil in temperature. Low speed feedback
Oil Pres. Oil pressure. Low speed feedback
Clnt. Temp. Coolant in temperature. Low speed feedback
COVimep The coefficient of variation in NIMEP.

Calculated using the following formula:
n = number of samples (calculated over 100

cycles)
Intake Temp. Intake port temperature. Low speed feedback
Intake Pres. Intake pressure as determined by taking the

average pressure from the high speed intake
pressure transducer from the entire cycle
T Exha. Exhaust port temperature. Low speed

feedback
P Exha. Exhaust pressure as determined by taking the

average pressure from the high speed exhaust
pressure transducer from the entire cycle
Spark Timing The spark timing. Not used for this test work.
This data is saved using INCA instead
316
Inj. Timing Injection timing. Not used for this test work.
Inj. Pulse Width Injection duration. Not used for this test work.
Inj. Pres. Fuel pressure going into the high pressure

pump. The main reason for this is to ensure
that the high pressure fuel pump has sufficient
pressure. Low speed feedback
Mode Allows the user to select how the engine is

being driven, whether it is being motored or
fired
State Allows the user to select how many cycles are

recorded. “Steady” records 100 cycles.
“Transient” records cycles until the user
requests it to stop
P Ambi Ambient pressure. Input manually
T Ambi Ambient Temperature. Input manually
Humidity % Air Humidity. Input manually. Set to 0
AFlow[g/cyc] Air Flow through the engine. Not used for this
test work
FFlowRate[g/s] Fuel flow rate through the engine. High speed

feedback
Clnt. Pres. Coolant pressure. Low speed feedback
IMEP NIMEP. Both the “Tra.” And “Ave.” boxes are

updated on a cycle to cycle basis. “Tra.” Gives
the NIMEP from the previous cycle. “Ave.”
gives the average NIMEP from the last 100
cycles
317
BMEP Brake Mean Effective Pressure (BMEP). Both
the “Tra.” And “Ave.” boxes are updated on a
cycle to cycle basis. “Tra.” Gives the BMEP
from the previous cycle. “Ave.” gives the
average BMEP from the last 100 cycles
PMEP Pumping Mean Effective Pressure (PMEP).

Both the “Tra.” And “Ave.” boxes are updated
on a cycle to cycle basis. “Tra.” Gives the PMEP
average PMEP from the last 100 cycles. PMEP
is calculated using the following formula and
the Gross IMEP (GIMEP):
FMEP Friction Mean Effective Pressure (FMEP). Both

the “Tra.” And “Ave.” boxes are updated on a
cycle to cycle basis. “Tra.” Gives the FMEP from
the previous cycle. “Ave.” gives the average
FMEP from the last 100 cycles. FMEP is
calculated using the following formula:
ISFC Indicated Specific Fuel Consumption (ISFC).

Updated on a cycle to cycle basis. Calculated
using the following formula:
F= Fuel flow, V= Swept volume, N= Engine

speed
BSFC Brake Specific Fuel Consumption (BSFC).

Updated on a cycle to cycle basis. Calculated
318
using the following formula:
F= Fuel flow, V= Swept volume, N= Engine

speed
Pi Indicated power. Updated on a cycle to cycle

basis. Calculated using the following formula:
V= Swept Volume, N= Engine speed
Pe Brake power. Updated on a cycle to cycle basis.

Calculated using the following formula:
V= Swept Volume, N= Engine speed
T Ex. M. Exhaust manifold temperature. Low speed

feedback
T Fuel Low pressure fuel temperature. Low speed

feedback
T Oil out Oil out temperature. Low speed feedback
T Head Coolant temperature out of the cylinder head.

Low speed feedback
T Block Coolant temperature out of the cylinder block.

Low speed feedback
B Clnt FR Coolant flow rate through the cylinder block.

Low speed feedback
H Clnt FR Coolant flow rate through the cylinder head.
319
Low speed feedback
IVO Intake Valve Opening (IVO) timing. Input

manually. Not used for this test work. This data
is saved using INCA instead
IVC Intake Valve Closing (IVC) timing. Input

IVL Maximum intake valve lift. Input manually. Not

used for this test work
EVO Exhaust Valve Opening (EVO) timing. Input

EVC Exhaust Valve Closing (EVC) timing. Input

EVL Maximum exhaust valve lift. Input manually.

Not used for this test work
RGF Residual gas fraction. Input manually. Not used

for this test work
ECR Effective CR. Not used for this test work
CO Carbon monoxide emissions. Not used for this

test work
CO2 Carbon dioxide emissions. Not used for this

test work
HC Hydrocarbon emissions. Not used for this test

work. This data was logged by hand instead
NOx Nitrogen oxides emissions. Not used for this

test work. This data was logged by hand
320
instead
O2 Oxygen emissions. Not used for this test work
EGR Exhaust Gas Recirculation (EGR) quantity. Not

used for this test work
Lambda Secondary lambda reading obtained from an

exhaust gas analyzer. Not used for this test
work
Pmax Peak cylinder pressure. Both the “Tra.” And

“Ave.” boxes are updated on a cycle to cycle
basis. “Tra.” Gives the peak cylinder pressure
average peak cylinder pressure from the last
100 cycles
CA Crank angle of peak cylinder pressure. Both the

“Tra.” And “Ave.” boxes are updated on a cycle
to cycle basis. “Tra.” Gives the CA of peak
cylinder pressure from the previous cycle.
“Ave.” gives the average CA of peak cylinder
pressure from the last 100 cycles
dPdCA Peak Dp/Dθ. Both the “Tra.” And “Ave.” boxes

are updated on a cycle to cycle basis. “Tra.”
Gives the peak Dp/Dθ from the previous cycle.
“Ave.” gives the average peak Dp/Dθ from the
last 100 cycles
CA Crank angle of peak Dp/Dθ. Both the “Tra.”

And “Ave.” boxes are updated on a cycle to
cycle basis. “Tra.” Gives the CA of peak Dp/Dθ
average CA of peak Dp/Dθ from the last 100
cycles
321
CA10 The crank angle at 10% MFB. Both the “Tra.”
cycle basis. “Tra.” Gives the 10% MFB angle
average 10% MFB angle from the last 100
cycles

cycles

cycles
Dura. The 10–90% MFB duration. Both the “Tra.” And

“Ave.” boxes are updated on a cycle to cycle
basis. “Tra.” Gives the 10–90% MFB duration
average 10–90% MFB duration from the last
100 cycles
I_Knock The peak KI value. Both the “Tra.” And “Ave.”

boxes are updated on a cycle to cycle basis.
“Tra.” Gives the peak KI value from the
previous cycle. “Ave.” gives the average peak
KI value from the last 100 cycles
Tmax Not used for this test work
Fuel Type Either Gasoline or Diesel. This was set to
322
Gasoline for all test work
Threshold Allows the user to specify a maximum cylinder

pressure above which it is assumed a pre-
ignition event has taken place. This was set to
20% above the maximum cylinder pressure
obtained during normal running at that
particular test condition
Number The number of cycles the cylinder pressure has

exceeded the threshold pre-ignition value
Ref. Position Allows the user to input a shaft encoder signal

offset to allow the setting of Top Dead Centre
(TDC) position using software as opposed to
hardware
Gamma Allows the user to specify a gamma value (ratio

of specific heats) for the air. This was set to
1.32 for all test work
d The cylinder bore measurement
s The stroke length
Conrod The connecting rod length
CR The engine geometric CR
Offse The crank pin offset as shown below:
A common practice for circumventing NVH
323
issues, the crank pin offset had to be factored
in because it will impact on the accuracy of
NIMEP measurements
SamplesPerRevolution The resolution of the shaft encoder. For all test

work a 0.25 CAD shaft encoder was used, this
was therefore set to 1440
RevolutionsPerCycle The number of engine crank revolutions per

cycle. This was set to 2 for all test work
NumberOfChannels The number of channels going into the data

acquisition system
Saved File Name The name of the folder containing the last
recorded set of data
A.4 GT-Power Animation Showing Charge Scavenging
The following Figures show freeze frames from various points on an animated 4
stroke cycle obtained from a GT-Power simulation. The test point is 1000rpm, 14 bar
BMEP, 2.5 bar EBP, CR of 9 with a cam duration of 312 CAD and an IMOP of 150 CAD
ATDC. At this running condition fuel vapours can be demonstrated to flow back into
the intake manifold and then scavenge straight through the cylinder during the gas
exchange process resulting in unburned charge being short circuited into the exhaust
manifold.
Unburned charge is shown on the Figures A.1 to A.7 as a scarlet coloured red
with areas without unburned charge being shown as green. Fuel vapours can be
demonstrated to flow as far upstream as the end of the variable section in the intake
manifold. Figure A.1 shows the cycle starting at BDC at the end of the intake stroke.
324
Figure A.1 Piston at End of Intake BDC
Figure A.2 Charge Beginning to Flow Back Up the Intake Manifold
325
Figure A.3 Charge Still Continuing to Flow Even at 90 CAD BTDCF
Figure A.4 Piston at TDCF. Charge Still Trapped in Intake Manifold
326
Figure A.5 Piston at End of Power BDC. Exhausting has Commenced, No Unburned Vapours Present in
Exhaust Gasses
Figure A.6 Piston at Gas Exchange TDC. No Scavenging has yet Begun Despite Both Valves Being Open
327
Figure A.7 Charge Scavenging Beginning at Approximately 15 CAD ATDC
A.5 References
328
Reference 49:
329
Reference 52:
330
331
332
333
Refernce 59:
334
335
336
Reference 78:
337
Reference 81:
338
339
340
341
342
343

Downsized-Boosted-Gasoline-Engines Working-Paper ICCT 27102016 1

Uploaded by

Copyright:

Available Formats

Downsized-Boosted-Gasoline-Engines Working-Paper ICCT 27102016 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Downsized-Boosted-Gasoline-Engines Working-Paper ICCT 27102016 1

Uploaded by

Copyright:

Available Formats

INVESTIGATION OF ENGINE DESIGN PARAMETERS ON

THE EFFICIENCY AND PERFORMANCE OF THE HIGH

A thesis submitted for the degree of Doctor of Philosophy

School of Engineering and Design

ABDC After Bottom Dead Centre

a Wiebe model constant

A Knock index multiplier, Activation energy multiplier, Flow area

Cf Skin friction coefficient

Specific heat capacity at constant pressure, Pressure loss

CL Connecting rod length

e Total internal energy per unit mass

h Heat transfer coefficient

ISem Indicated specific emission

Km Percentage cylinder unburned at knock initiation

m Wiebe model constant, mass

ṁ Boundary mass flux into the volume

n Polytropic exponent/specific heat ratio

p In-cylinder pressure, pressure

∆pc Change in pressure due to combustion

∆pv Change in pressure due to volume change

P Precursor reaction rate multiplier

rV Geometric compression ratio

R Specific gas constant

thkn Start of compression angle

T Livengood-Wu integral, Temperature

Tfluid Fluid temperature

Tu Instantaneous unburned gas temperature

Twall Wall temperature

u Velocity at the boundary

VI Cylinder volume at knock initiation

VTDC Volume at TDC

x MFB, Crank-pin offset

xb Mass fraction burned

θ0 Crank angle at combustion start

ŋ Otto Otto cycle efficiency

τ Douaud and Eyzat component

Abbreviations, Symbols and Chemical Symbols

1.3 Outline of Thesis 5

Chapter 2: Literature Review 8

2.2 Gasoline Engine Emissions 10

2.2.1 Current Worldwide Emission Legislation 10

2.2.1.1 Carbon Monoxide (CO) 16

2.2.1.2 Unburned Hydrocarbons (HC) 17

2.2.1.3 Nitrogen Oxides (NOx) 18

2.2.1.4 Particulate Matter (PM) 19

2.3 Current Automotive Engines, Technologies and Trends 19

2.3.1 Hybrid and Electric Powertrains 20

2.3.2 Fossil Fuel Powertrains 21

2.3.3 Gasoline Engine Developments 22

2.3.4 Downsizing and Turbocharging 25

2.4 Atkinson/Miller Cycle 28

2.4.3 Pumping Losses 33

2.4.4 Combustion Stability 36

2.4.5 Engine Breathing 39

2.4.7 Knock Mitigation and Deep Miller 43

Chapter 3: Experimental Test Facility 46

3.2 Experimental Setup 46

3.2.1 Engine Description 47

3.2.1.1 Single Cylinder Engine Description 49