Clemson University Clemson University

TigerPrints TigerPrints

All Dissertations Dissertations

August 2021

An Assessment of Thermal Barrier Coatings for The Low-An Assessment of Thermal Barrier Coatings for The Low-

Temperature Combustion Family: From HCCI To GCI Temperature Combustion Family: From HCCI To GCI

Ziming Yan Clemson University, yanziminga@gmail.com

Follow this and additional works at: https://tigerprints.clemson.edu/all_dissertations

Recommended Citation Recommended Citation Yan, Ziming, "An Assessment of Thermal Barrier Coatings for The Low-Temperature Combustion Family: From HCCI To GCI" (2021). All Dissertations. 2897. https://tigerprints.clemson.edu/all_dissertations/2897

This Dissertation is brought to you for free and open access by the Dissertations at TigerPrints. It has been accepted for inclusion in All Dissertations by an authorized administrator of TigerPrints. For more information, please contact kokeefe@clemson.edu.

i

AN ASSESSMENT OF THERMAL BARRIER COATINGS FOR THE LOW-TEMPERATURE COMBUSTION FAMILY: FROM HCCI TO GCI

A Thesis Presented to

the Graduate School of Clemson University

In Partial Fulfillment of the Requirements for the Degree

Doctor of Philosophy Automotive Engineering

by Ziming Yan August 2021

Accepted by: Dr. Benjamin Lawler, Committee Chair Dr. Zoran Filipi, Committee Co-Chair

Dr. Robert Prucka, Committee Member Dr. Qilun Zhu, Committee Member

ii

ABSTRACT

Thermal barrier coatings (TBCs) reduce in-cylinder heat transfer losses and increase

thermal efficiency. Beyond the efficiency improvement, many challenges associated with

low-temperature combustion (LTC) could be potentially improved with TBC. Therefore,

the investigation of the effects of TBC in LTC serves as the motivation of this thesis. The

thesis includes both experimental and computational investigations, which are mainly

divided four-fold.

First, this dissertation experimentally demonstrated the feasibility and

comprehensively investigated the effects of thick thermal barrier coatings on pure

Homogeneous Charge Compression Ignition (HCCI) combustion (i.e., low residual and

high compression ratio) with two different fuels (conventional gasoline and wet ethanol

80). A deterioration of the high load limit was not observed, which implies that the charge

heating penalty does not occur in pure-HCCI. Both combustion and thermal efficiency

increased for the thicker TBC with a reduced intake temperature requirement. It is also

observed that a dense top sealing layer results in a significant improvement to unburned

hydrocarbon (UHC) emissions.

Then, a parametric computational investigation into the effects of various coating

properties on pure-HCCI combustion was performed. A zero-dimensional (0D)

thermodynamic model of the engine cycle was established and coupled to a 1D transient

heat transfer model of the coating and piston. Three parameters were thoroughly

investigated independently: thermal conductivity (k), coating thickness, and volumetric

heat capacity (s). The results revealed that the volumetric efficiency actually increases with

iii

a thicker coating due to a reduction in heat transfer during the compression stroke, which

lowers the required intake temperature to reach autoignition. The results also indicate that

the optimal coating configuration for pure-HCCIis a combination of the lowest k, the

lowest s, and the thickest coating before reaching the charge heating limit.

Since LTC contains a big family tree, ranging from HCCI to stratified LTC such as

Gasoline Compression Ignition (GCI), it was desired to explore the effects of TBCs in GCI

combustion, firstly by understanding GCI combustion through the Partial Fuel

Stratification (PFS) combustion strategy. PFS was successfully enabled at a 1.6 bar boost

level. The peak pressure rise rate (PPRR) was successfully reduced by up to 30% with the

latest injection event and the lowest split fraction. However, a new double late injection

strategy was also proposed that enables another 27% reduction in PPRR, which indicates

that the φ distribution has been broadened dramatically, thereby unlocking further potential

for higher loads.

Last, this dissertation established a preliminary guideline for TBCs with GCI via

thermodynamic modeling. The coating performance was evaluated with two candidates.

The results show that increasing coating thickness increases the thermal efficiency of GCI

combustion with a trend of diminishing returns. Charge heating was much less than

expected due to the high level of intake boost. A study of the intake and exhaust valves

revealed an exhaust valve peak surface temperature of ~1000K, which could be a concern

for coating temperature durability. It was shown that coating the piston and firedeck was

very rewarding in terms of efficiency improvement with low charge heating. However, it

is not worth coating the liner clearance due to a minimal efficiency gain.

iv

ACKNOWLEDGMENTS

My deepest gratitude goes to Dr. Benjamin Lawler, as a mentor not only in academia but

also in my life. Thanks for patiently guiding me and sharing the excitement of five years

of discovery. Being an international student and thousands of miles away from home, you

are the person who always cheers me up and makes me feel at home. Your unwavering

passion and personal kindness helped me finding confidence, being professional, and made

my life delightful in the past few years.

I would also like to acknowledge Dr. Zoran Filipi as my co-advisor, who kept my

focus on the right track. Your top-notch spirit runs in the family, and I will carry it with me

in my future career. I would like to express my thanks to Dr. Robert Prucka and Dr. Qilun

Zhu as my committee member for your brilliant guidance and feedback. I would like to

thank Dr. Sanjay Sampath from Stony Brook University for his experimental support

associated with the thermal barrier coating.

I would like to express my appreciation to my colleagues and friends, especially to

Dr. Brian Gainey and Dr. Deivanayagam Hariharan, for the days of turning the wrench and

running the engine together. All my work would be impossible without your help.

My special thanks to my parents for raising, believing in, and supporting me

through all my life. To my wife, Minni, for being my strongest supporter and helper. To

Tao for bringing me so much happiness and laughter. To my best friends from Stony Brook

University, Guangyu and Fan, for encouragement and being there for me.

TABLE OF CONTENTS

Page

ABSTRACT ........................................................................................................................ ii

ACKNOWLEDGMENTS ................................................................................................. iv

TABLE OF CONTENTS .....................................................................................................v

LIST OF TABLES ............................................................................................................. ix

LIST OF FIGURES .............................................................................................................x

LIST OF ABBREVIATIONS .......................................................................................... xvi

CHAPTER 1. INTRODUCTION ........................................................................................1

1.1 The development of TBCs in internal combustion engines (ICEs) ............................................. 1

1.2 Low-temperature combustion (LTC) family .............................................................................. 4

1.2.1 Homogeneous Charge Compression Ignition (HCCI) ............................................................................ 5

1.2.2 Stratified Gasoline Compression Ignition (GCI) .................................................................................... 6

1.2.3 Challenges with Low-Temperature Combustion ................................................................................ 11

1.3 Objective of the Current Approach and Specific Tasks............................................................ 16

1.3.1 Demonstration and justification – Comprehensive experimental investigation of thick thermal

barrier coatings for HCCI (CHAPTER 2 & CHAPTER 3) ...................................................................................... 17

1.3.2 Deep dive into the fundamentals – A parametric computational investigation into the effects of

various coating properties for HCCI (CHAPTER 4 & CHAPTER 5) ..................................................................... 17

1.3.3 Clemson GCI test cell commissioning (CHAPTER 6)............................................................................ 17

1.3.4 Establishing the understanding of GCI - an experimental investigation of injection strategies for

gasoline PFS (CHAPTER 7) ............................................................................................................................... 18

1.3.5 A preliminary guideline for TBCs with GCI – A computational evaluation (CHAPTER 8) .................... 18

CHAPTER 2. EXPERIMENTAL SETUP AND METHODOLOGY ...............................19

2.1 Experimental engine test cell .................................................................................................. 19

2.1.1 Specific experimental setup for TBC-HCCI study ................................................................................ 21

2.1.2 Fuel delivery method for the gasoline PFS study ............................................................................... 26

v

vi

Table of Contents (Continued) Page

2.2 Data Acquisition and Analysis Methodology .......................................................................... 27

2.3 Application of the TBCs and Measurements of Their Thermophysical Properties .................. 29

CHAPTER 3. THICK THERMAL BARRIER COATINGS FOR HCCI -

EXPERIMENTAL RESULTS AND DISCUSSION ........................................................33

3.1 Objective and Experimental operating conditions for thermal barrier coating study ............ 33

3.2 Performance of TBC at different loads with conventional gasoline HCCI ............................... 35

3.2.1 Intake temperature requirement - Gasoline ...................................................................................... 36

3.2.2 The Heat Release Process and Load Range – Gasoline ...................................................................... 38

3.2.3 Efficiency and Energy Distribution – Gasoline ................................................................................... 42

3.2.4 Emissions – Gasoline .......................................................................................................................... 48

3.3 Performance of TBC at different loads with WE80 and compared with gasoline ................... 49

3.3.1 Load range, efficiencies, and energy distribution – WE80 & gasoline ............................................... 50

3.3.2 Intake temperature requirement – WE80 & gasoline ........................................................................ 55

3.4 Performance of TBC with WE80 with varied SOI timings ........................................................ 58

3.4.1 The effect of injection timing and TBC on heat release process ........................................................ 58

3.4.2 The effect of injection timing and TBC on efficiencies and emissions ............................................... 60

3.4.3 The intake and exhaust temperatures ............................................................................................... 67

CHAPTER 4. MODELING SETUP AND VALIDATION ..............................................70

4.1 0-D thermodynamic engine cycle modeling ............................................................................ 70

4.1.1 Conservation of mass & flow characterization .................................................................................. 70

4.1.2 Energy balance ................................................................................................................................... 71

4.1.3 Thermodynamic properties of the working fluid ............................................................................... 72

4.2 1D transient heat transfer modeling ...................................................................................... 75

4.3 Model validation ..................................................................................................................... 79

4.3.1 Validation against the experimental metal baseline cases ................................................................ 80

4.3.2 Validation against the experimental TBC cases ................................................................................. 82

vii

Table of Contents (Continued) Page

CHAPTER 5. THICK THERMAL BARRIER COATINGS FOR HCCI - MODELING

RESULTS AND DISCUSSIONS ......................................................................................86

5.1 Objective and simulation cases setup ..................................................................................... 86

5.2 The effects of thermal conductivity - 𝒌 ................................................................................... 88

5.3 The effect of TBC thickness at low-𝒌 ....................................................................................... 96

5.4 The effect of volumetric heat capacity - 𝒔, with the thick, low-𝒌 TBC .................................. 102

CHAPTER 6. GASOLINE COMPRESSION IGNITION TEST CELL

COMMISSIONING .........................................................................................................110

6.1 Single-cylinder light-duty GCI engine commissioning ........................................................... 110

CHAPTER 7. INVESTIGATION INTO THE INJECTION STRATEGIES FOR

GASOLINE PFS – EXPERIMENTAL RESULTS & DISCUSSION ............................116

7.1 Objective and Experimental operating conditions for GCI investigation .............................. 116

7.2 The effects of SOI timing on PFS combustion characteristics, efficiencies, and emissions ... 117

7.3 The effects of Split Fraction (SF) on PFS combustion characteristics, efficiencies, and

emissions ........................................................................................................................................... 124

7.4 The combined effects of SOI timings and split fraction ........................................................ 129

7.5 The Double Late Injection (DLI) strategy with comparisons to the Single Late Injection (SLI)

strategy ............................................................................................................................................. 132

CHAPTER 8. THERMAL BARRIER COATINGS FOR GCI - MODELING RESULTS

AND DISCUSSIONS ......................................................................................................140

8.1 Model upgrade and validation ............................................................................................. 141

8.2 Effects of coating thickness on gasoline compression ignition combustion ......................... 147

8.3 Effect of boost on the performance of “traditional” and “temperature swing” coatings .... 152

8.4 Valve heat transfer estimation and coating coverage investigation .................................... 157

CHAPTER 9. CONCLUSIONS AND UNIQUE CONTRIBUTIONS ...........................167

viii

Table of Contents (Continued) Page

9.1 Summary and conclusions .................................................................................................... 167

9.1.1 Demonstration and justification – Comprehensive experimental investigation of thick thermal

barrier coatings for pure-HCCI ...................................................................................................................... 168

9.1.2 Deep dive into the fundamentals – A parametric computational investigation into the effects of

various coating properties for pure-HCCI...................................................................................................... 170

9.1.3 Establishing the understanding of GCI - an experimental investigation of injection strategies for

gasoline PFS .................................................................................................................................................. 171

9.1.4 A preliminary guideline for TBCs with GCI - A computational evaluation ........................................ 174

9.2 Unique contributions ............................................................................................................ 176

9.3 Suggestion for future work ................................................................................................... 177

PUBLICATIONS .............................................................................................................179

Thesis publications ............................................................................................................................ 179

Other publications ............................................................................................................................. 179

REFERENCES ................................................................................................................181

APPENDICES A .............................................................................................................195

APPENDICES B ..............................................................................................................200

ix

LIST OF TABLES

Table Page

TABLE 1: ENGINE SPECIFICATIONS ......................................................................................................................... 20

TABLE 2: FUEL PROPERTIES .................................................................................................................................. 26

TABLE 3:COATING LAYER PROPERTIES .................................................................................................................... 31

TABLE 4: COMBINE LAYER PROPERTIES ................................................................................................................... 31

TABLE 5: APS CONFIGURATIONS .......................................................................................................................... 31

TABLE 6:ENGINE OPERATING CONDITIONS FOR THE TBC STUDIES ................................................................................ 35

TABLE 7: EVALUATION OF CRITICAL VALIDATION METRICS FOR THE METAL BASELINE CASES ............................................... 82

TABLE 8: EVALUATION OF THE CRITICAL VALIDATION METRICS FOR THE TBC CASES ......................................................... 84

TABLE 9: INVESTIGATED CASES AND SWEEP CONFIGURATIONS .................................................................................... 87

TABLE 10: OPERATING CONDITIONS ...................................................................................................................... 88

TABLE 11: ENGINE PERFORMANCE AT LOAD OF 4.5 BAR IMEPG ............................................................................... 107

TABLE 12: ENGINE PARAMETERS ........................................................................................................................ 112

TABLE 13: ENGINE OPERATING CONDITIONS IN PFS STUDY ...................................................................................... 117

TABLE 14: ENGINE SPECIFICATIONS FOR ARAMCO LIGHT-DUTY GCI ENGINE ................................................................ 145

TABLE 15:PPCI-DIFFUSION COMBUSTION OPERATING CONDITIONS ........................................................................... 146

TABLE 16: REAL-WORLD MATERIAL PROPERTIES FOR ENGINE COATING APPLICATION ..................................................... 148

x

LIST OF FIGURES

Page

FIGURE 1:EQUIVALENCE RATIO & TEMPERATURE MAP FOR SOOT AND NOX FORMATION [11] . .......................................... 5

FIGURE 2: GASOLINE COMPRESSION IGNITION WITH DIFFERENT INJECTION STRATEGY AND STRATIFICATION LEVEL [25] ............ 9

FIGURE 3: SIMULATED TEMPERATURE DISTRIBUTION AT TDC [57] .............................................................................. 13

FIGURE 4: ENGINE TEST CELL LAYOUT .................................................................................................................... 19

FIGURE 5: GEOMETRY OF THE COMBUSTION CHAMBER AT TDC .................................................................................. 20

FIGURE 6: TBC MACHINE & SPRAY PROCESS ........................................................................................................... 23

FIGURE 7: TBC CONDITIONS AFTER 20 HOURS OF TESTING ........................................................................................ 24

FIGURE 8: OPTICAL MICROGRAPH FOR UNSEALED TBC LAYERS ................................................................................... 32

FIGURE 9: OPTICAL MICROGRAPH FOR SEALED TBCS LAYERS ...................................................................................... 32

FIGURE 10: PEAK MOTORING PRESSURE VS. INTAKE TEMPERATURE AT NATURALLY ASPIRATED INTAKE ................................ 34

FIGURE 11: INTAKE AND EXHAUST TEMPERATURE VS. IMEPG AT 1200RPM WITH A CONSTANT CA50 FOR EACH PISTON, I.E.,

PHASE-MATCH .................................................................................................................................................. 38

FIGURE 12: GROSS HEAT RELEASE RATE (LEFT) & CYLINDER PRESSURE (RIGHT) VS. CRANK ANGLE ...................................... 39

FIGURE 13: (A) PPRR, (B) CA50 COMBUSTION PHASING, AND (C) 10-90% BURN DURATION VS. IMEPG AT 1200RPM WITH A

CONSTANT CA50 FOR EACH PISTON, I.E., PHASE-MATCH ........................................................................................... 41

FIGURE 14: (A) COMBUSTION EFFICIENCY, (B) GROSS INDICATED THERMAL EFFICIENCY, AND (C) GROSS INDICATED FUEL

CONVERSION EFFICIENCY VS. IMEPG AT 1200RPM WITH A CONSTANT CA50 FOR EACH PISTON, I.E., PHASE-MATCH ............. 45

FIGURE 15: ENERGY DISTRIBUTION CHART FOR GASOLINE WITH THE FIVE DIFFERENT PISTONS AT LOAD OF 4.6 BAR IMEPG.

METALH IS THE METAL PISTON WITH THE HIGHER COMPRESSION RATIO AND METALL IS THE METAL PISTON WITH THE LOWER

COMPRESSION RATIO. ......................................................................................................................................... 48

FIGURE 16: (A) UHC, (B) CO, AND (C) NOX EMISSIONS VS. IMEPG, AT 1200RPM WITH A CONSTANT CA50 FOR EACH PISTON,

I.E., PHASE-MATCH ............................................................................................................................................ 49

xi

Table of Figures (Continued) Page

FIGURE 17: (A) COMBUSTION EFFICIENCY, (B) GROSS INDICATED THERMAL EFFICIENCY, AND (C) GROSS INDICATED FUEL

CONVERSION EFFICIENCY VS. IMEPG FOR GASOLINE (SOLID LINES) AND WET ETHANOL (DASHED LINES) AT 1200RPM WITH A

CONSTANT CA50 FOR EACH PISTON, I.E., PHASE-MATCH ........................................................................................... 51

FIGURE 18: (A) UHC, (B) CO, AND (C) NOX EMISSIONS VS. IMEPG FOR GASOLINE (SOLID LINES) AND WET ETHANOL (DASHED

LINES) AT 1200RPM WITH A CONSTANT CA50 FOR EACH PISTON, I.E., PHASE-MATCH ..................................................... 53

FIGURE 19:ENERGY DISTRIBUTION CHART FOR WE80 AND GASOLINE .......................................................................... 55

FIGURE 20: INTAKE TEMPERATURE VS. IMEPG AT 1200RPM WITH A CONSTANT CA50 FOR EACH PISTON, I.E., PHASE-MATCH 56

FIGURE 21: COOLING POTENTIAL OF DIFFERENT FUELS .............................................................................................. 57

FIGURE 22: INTAKE AND EXHAUST TEMPERATURE VS. IMEPG (WET ETHANOL 80) AT 1200RPM WITH A CONSTANT CA50 FOR

EACH PISTON, I.E., PHASE-MATCH ......................................................................................................................... 57

FIGURE 23: GROSS HEAT RELEASE RATES (BOTTOM) AND PRESSURE TRACES (TOP) VS. CRANK ANGLE DEGREE AT SOI OF -350

DEG ATDC ....................................................................................................................................................... 58

FIGURE 24: (A) PPRR, (B) COMBUSTION PHASING, (C) BURN DURATION VS. SOI TIMING WITH DIFFERENT COATED PISTONS AND

THE METAL BASELINE CASES, CA50 = 6.8 DEG ATDC ............................................................................................... 60

FIGURE 25: EFFICIENCIES VS. SOI TIMING FOR THE DIFFERENT COATED AND METAL BASELINE PISTONS, CA50 = 6.8 DEG ATDC

...................................................................................................................................................................... 62

FIGURE 26: FUEL SPRAY VISUALIZATIONS AT INJECTION TIMINGS OF -330, -300, -270, -240, AND -210 DEGREES ATDC ..... 63

FIGURE 27: VISUALIZATION OF TWO INCLUDED ANGLES (150° AND 60°) AT SOI TIMING OF -330 DEG ATDC ..................... 64

FIGURE 28: COMBUSTION EFFICIENCY WITH TWO INJECTION ANGLES (150° AND 60°) .................................................... 65

FIGURE 29: EMISSIONS AND PEAK CYLINDER TEMPERATURE VS. SOI TIMING FOR THE DIFFERENT COATED AND METAL BASELINE

PISTONS, CA50 = 6.8 DEG ATDC ......................................................................................................................... 66

FIGURE 30: THE INTAKE TEMPERATURE REQUIREMENTS AND THE MEASURED EXHAUST TEMPERATURES VS. SOI TIMING FOR THE

DIFFERENT COATED AND METAL BASELINE PISTONS, CA50 = 6.8 DEG ATDC ................................................................. 67

FIGURE 31: INTAKE TEMPERATURE WITH TWO INJECTION ANGLES (150° AND 60°) ........................................................ 68

xii

Table of Figures (Continued) Page

FIGURE 32: ONE-DIMENSIONAL TRANSIENT FINITE ELEMENT HEAT TRANSFER SCHEMATIC ................................................ 76

FIGURE 33: DETERMINATION OF OIL COOLING CONVECTIVE HEAT TRANSFER COEFFICIENT FOR THE BACKSIDE BOUNDARY

CONDITION ....................................................................................................................................................... 78

FIGURE 34: CYLINDER PRESSURE TRACE OF METAL CASE VALIDATION AT LOADS OF 2.7 AND 4.6 BAR IMEPG ....................... 81

FIGURE 35: CYLINDER PRESSURE TRACE FOR THE 2MM TBC CASES AT LOADS OF 2.6 BAR AND 4.6 BAR IMEPG PROVIDING

VALIDATION OF THE MODEL’S ABILITY TO CAPTURE THE THERMODYNAMICS AS WELL AS THE PERFORMANCE OF THE TBC ........ 85

FIGURE 36: COATING NODE TEMPERATURES FOR THE FIRST 20 NODES WITH A ΔX SPACING OF 25 MICRONS BETWEEN NODES

(LEFT) AND BULK GAS TEMPERATURES (RIGHT) OVER THE ENGINE CYCLE FOR THE 4.6 BAR IMEPG LOAD CASE ...................... 85

FIGURE 37: (A) COATING SURFACE TEMPERATURE AND (B) BULK GAS TEMPERATURE VS. CRANK ANGLE AT A LOAD OF 3.5 BAR

IMEPG OVER THE 𝑘 SWEEP .................................................................................................................................. 90

FIGURE 38: TEMPERATURE SWING VS. THERMAL CONDUCTIVITY, 𝑘 ............................................................................. 92

FIGURE 39: (A) HEAT TRANSFER LOSSES AND (B) GROSS INDICATED THERMAL EFFICIENCY VS. THERMAL CONDUCTIVITY, 𝑘 ...... 93

FIGURE 40: INTAKE TEMPERATURE VS. THERMAL CONDUCTIVITY, 𝑘 ............................................................................. 94

FIGURE 41: TEMPERATURE AT EVO VS. THERMAL CONDUCTIVITY, 𝑘 ........................................................................... 95

FIGURE 42: (A) COATING SURFACE TEMPERATURE AND (B) BULK GAS TEMPERATURE VS. CRANK ANGLE AT THE LOAD OF 3.5 BAR

IMEPG THROUGH THICKNESS SWEEP ..................................................................................................................... 99

FIGURE 43: INTAKE TEMPERATURE AND 𝜙 VS. COATING THICKNESS ........................................................................... 100

FIGURE 44: HEAT TRANSFER LOSSES (RED, LEFT AXIS) & THERMAL EFFICIENCY (BLUE, RIGHT AXIS) VS. COATING THICKNESS AT THE

LOAD OF 3.5 BAR IMEP ................................................................................................................................... 102

FIGURE 45: (A) COATING SURFACE TEMPERATURE AND (B) BULK GAS TEMPERATURE VS. CRANK ANGLE AT THE LOAD CONDITION

OF 3.5 BAR IMEPG OVER THE VOLUMETRIC HEAT CAPACITY SWEEP ........................................................................... 105

FIGURE 46: OPTIMIZATION ROUTINE AND ENGINE PERFORMANCE ............................................................................. 109

FIGURE 47: GROSS INDICATED THERMAL EFFICIENCY VS. MAXIMUM SURFACE TEMPERATURE .......................................... 109

FIGURE 48: CLEMSON UNIVERSITY SINGLE-CYLINDER LIGHT-DUTY GCI TEST CELL OVERVIEW .......................................... 111

xiii

Table of Figures (Continued) Page

FIGURE 49: TOP: LAYOUT OF HIGH-PRESSURE DIRECT INJECTION FUEL SYSTEM; LEFT: LOW-PRESSURE LOOP CART; BOTTOM

RIGHT: HEAT EXCHANGERS FOR FUEL RETURN COOLING. .......................................................................................... 113

FIGURE 50: ECM AFR AND EGR MODULES IN SERIES BUS CONNECTION .................................................................... 115

FIGURE 51: INJECTION STRATEGY FOR SINGLE LATE INJECTION (SLI) WITH SF70........................................................... 118

FIGURE 52: GHRR FOR AN SOI SWEEP AT A SPLIT FRACTION OF 70% (SF70, I.E., 70% OF THE FUEL MASS WAS PORT FUEL

INJECTED AND 30% WAS DIRECT INJECTED AT THE TIMING INDICATED IN THE LEGEND) ................................................... 119

FIGURE 53: COMBUSTION CHARACTERISTICS AND EFFICIENCIES VS. INJECTION TIMING AT SF70 ...................................... 122

FIGURE 54: EMISSIONS VS. INJECTION TIMING AT A SPLIT FRACTION OF 70 .................................................................. 124

FIGURE 55: GHRR & PRESSURE TRACE FOR DIFFERENT SPLIT FRACTIONS AT AN SOI TIMING OF -50 DEG ATDC ................. 126

FIGURE 56: COMBUSTION CHARACTERISTICS AND EFFICIENCIES VS. SPLIT FRACTION AT AN SOI = -50 DEG ATDC ............... 127

FIGURE 57: EMISSIONS VS. SPLIT FRACTION AT AN SOI = -50 DEG ATDC ................................................................... 128

FIGURE 58: (LEFT) COMBUSTION CHARACTERISTICS AND EFFICIENCIES; (RIGHT) EMISSIONS VS. START OF INJECTION TIMING AT

DIFFERENT SPLIT FRACTIONS ............................................................................................................................... 130

FIGURE 59: SPRAY AT -140, -110, -80, AND -50 DEG ATDC .................................................................................. 130

FIGURE 60: INJECTION STRATEGY FOR DOUBLE LATE INJECTION (DLI) WITH SF70 ........................................................ 132

FIGURE 61: GHRR FOR HCCI, SF70 WITH A SINGLE COMPRESSION STROKE INJECTION, AND SF70 WITH A DOUBLE LATE

INJECTION (DLI) AT VARIOUS SPLIT INJECTION SPACINGS .......................................................................................... 134

FIGURE 62: COMBUSTION CHARACTERISTICS AND EFFICIENCIES VS. INJECTION SPACING AT DIFFERENT EQUIVALENT INJECTION

TIMINGS AT A SPLIT FRACTION OF 70 ................................................................................................................... 136

FIGURE 63: EMISSIONS VS. INJECTION SPACING AT DIFFERENT EQUIVALENT INJECTION TIMINGS AT A SPLIT FRACTION OF 70 . 138

FIGURE 64: PPCI-DIFFUSION COMBUSTION STRATEGY FROM YU ET AL. [111]. ............................................................ 141

FIGURE 65: HEAT FLUX TO DIFFERENT COMBUSTION CHAMBER COMPONENTS. ............................................................ 143

FIGURE 66: PISTON HEAT FLUX ESTIMATION FROM THREE DIFFERENT METHODS. BLUE: HOHENBERG CORRELATION; RED: CFD

RESULTS; DASHED BLACK LINE: HYBRID HEAT FLUX (0D DT + ℎ𝐶𝐹𝐷) ........................................................................ 145

xiv

Table of Figures (Continued) Page

FIGURE 67: CYLINDER PRESSURE AND BULK GAS TEMPERATURE COMPARISON BETWEEN CFD AND 0D MODELS ................. 146

FIGURE 68: SURFACE TEMPERATURES THAT RESULT FROM DIFFERENT THICKNESSES FOR TWO DIFFERENT CANDIDATE MATERIALS

AT 2500RPM, 23.5 BAR IMEP. ......................................................................................................................... 149

FIGURE 69: GROSS INDICATED THERMAL EFFICIENCY V.S. THICKNESS (LEFT) AND PEAK SURFACE TEMPERATURE (RIGHT) WITH

TWO CANDIDATE MATERIALS. ............................................................................................................................. 150

FIGURE 70: BULK GAS, COATING, AND METAL PISTON SURFACE TEMPERATURE AT 23.5 BAR IMEPG ................................ 151

FIGURE 71: BULK GAS AND PISTON SURFACE TEMPERATURE AT 15 BAR IMEPG AND 1.6 BAR INTAKE PRESSURE. ................ 153

FIGURE 72: IVC TEMPERATURE AND EQUIVALENCE RATIO AT DIFFERENT BOOST LEVELS ................................................. 154

FIGURE 73: GROSS INDICATED THERMAL EFFICIENCY AT DIFFERENT BOOST LEVELS ........................................................ 155

FIGURE 74: BOOST PRESSURE V.S. EQUIVALENCE RATIO. ......................................................................................... 156

FIGURE 75: ENGINE VALVES HEAT TRANSFER BREAKDOWN (LEFT) AND SCHEMATIC OF 1D MODELING (RIGHT) ................... 158

FIGURE 76: HEAT FLUXES FROM DIFFERENT BOUNDARY CONDITIONS AT 2500RPM, 23.5 BAR IMEP. ............................. 159

FIGURE 77: COATED AND BASELINE SURFACE TEMPERATURE OF PISTON AND VALVES AT 2500RPM, 23.5 BAR IMEP. ......... 160

FIGURE 78: POSSIBLE TBC COVERAGE IN THE COMBUSTION CHAMBER ....................................................................... 161

FIGURE 79: LEFT: GROSS INDICATED THERMAL EFFICIENCY FOR DIFFERENT COVERAGE @ 2500RPM, 23.5 BAR IMEPG RIGHT:

EFFICIENCY GAIN RATIO DISTRIBUTION ................................................................................................................. 162

FIGURE 80: RELATIVE TEMPERATURE SWING FOR PISTON, HEAD, AND VALVES WITH 200 MICRONS GDZR @ 2500 RPM, 23.5

BAR IMEPG .................................................................................................................................................... 163

FIGURE 81: THE EFFECT OF SUBSTRATE ON COATING PERFORMANCE .......................................................................... 164

FIGURE 82: EFFICIENCY GAIN DISTRIBUTION AT DIFFERENT LOADS. LEFT: 23.5 BAR IMEPG. RIGHT: 6 BAR IMEPG .............. 165

FIGURE 83: GROSS INDICATED THERMAL EFFICIENCY V.S. RELATIVE CHARGE HEATING .................................................... 166

FIGURE 84: FUEL CONVERSION EFFICIENCY VS. IMEPG FOR GASOLINE (SOLID LINES) AND WET ETHANOL (DASHED LINES) ..... 195

FIGURE 85: GROSS HEAT RELEASE RATE (BOTTOM) & PRESSURE TRACE (TOP) VS. CRANK ANGLE DEGREE AT DIFFERENT SOI

TIMINGS ........................................................................................................................................................ 196

xv

Table of Figures (Continued) Page

FIGURE 86: (A) VOLUMETRIC EFFICIENCY AND (B) EQUIVALENCE RATIO VS. COATING THICKNESS ..................................... 197

FIGURE 87: SWEEP THE PROPERTIES IN ORDER OF K, S, AND THICKNESS ..................................................................... 198

FIGURE 88: GHRR & CYLINDER PRESSURE FOR SOI SWEEP AT A SPLIT FRACTION OF 70 ................................................ 198

FIGURE 89: ESTIMATED CYLINDER BULK TEMPERATURE ........................................................................................... 199

FIGURE 90: NEAR-WALL AND BULK GAS TEMPERATURES ......................................................................................... 199

FIGURE 91: DETAILED DRAWING FOR KISTLER 6125C CYLINDER PRESSURE TRANSDUCER MOUNTING ADAPTER. ................. 200

xvi

LIST OF ABBREVIATIONS

aTDC after top dead center APS atmospheric plasma spray CAD crank angle degree CA50 crank angle of 50% mass fraction burned CCD combustion chamber deposit CDC conventional diesel combustion CO carbon monoxide COV coefficient of variation CR compression ratio CTE coefficient of thermal expansion EGR external-cooled exhaust gas recirculation GCI gasoline compression ignition GHRR gross heat release rate HCCI homogeneous charge compression ignition HOF latent heat of vaporization IMEPn net indicated effective mean pressure IVO intake valve open LTC low-temperature combustion LTHR low temperature heat release MFB mass fraction burned NHRR net heat release rate NOx nitrogen oxides OEM original equipment manufacturer PFS Partial fuel stratification PPRR peak pressure rise rate RCCI reactivity-controlled compression ignition SOI start of injection SI spark ignition TBCs thermal barrier coatings TSCI thermally stratified compression ignition UHC unburned hydrocarbon YSZ yttria-stabilized zirconia

1

CHAPTER 1. INTRODUCTION

1.1 The development of TBCs in internal combustion engines (ICEs)

Since the transportation sector consumes approximately one-third of the total energy used

by the U.S. and contributes a similar fraction of CO2 emissions, there is a strong motivation

for energy-efficient transportation solutions for automotive vehicles. Accordingly, the

CAFE standards for fuel economy have become more aggressive in recent years. Spark

ignition (SI) engines have cleaner tailpipe emissions due to homogeneous and

stoichiometric operation and use of a three-way catalytic converter, but SI engines are

limited due to their lower thermal efficiency. The newest regulations on greenhouse gases,

especially CO2 emissions, require a significant efficiency improvement over traditional

spark ignition (SI) engines. Meanwhile, the next EPA tailpipe emissions regulations have

nearly suspended the development of diesel engines for the light-duty market in the U.S.

due to the NOx and soot emissions, despite their higher efficiency compared to traditional

SI engines. All of these demands and regulations motivate a revolutionary advancement in

the traditional automotive powertrain. There are several promising technologies and

approaches to meet these new requirements; for example, the hybrid electric powertrain is

an attractive option for the light-duty market. However, improvements made to the engine

efficiency can benefit both conventional vehicles and hybrid electrics and there is still a

considerable amount of room for improvements to the efficiency of combustion engines

from a thermodynamic standpoint. Heat transfer losses, which account for ~30% of the

energy released by fuel [1], are one of the largest energy losses that lower engine efficiency.

Therefore, reducing heat transfer losses can directly increase thermal efficiency.

2

Furthermore, minimizing heat transfer losses can reduce the coolant load and coolant

capacity required for heat dissipation, which eventually could be used to improve the

vehicle's weight and aerodynamics, thereby indirectly benefitting fuel economy. Since heat

transfer losses are caused by the temperature difference between the hot gas and relatively

cold walls, one approach to prevent heat loss is to coat the walls with thermal insulation,

also known as Thermal Barrier Coatings (TBCs).

The concept of the "adiabatic engine" was first proposed in the early 1970s and

followed by simulation studies to predict the performance by Kamo et al. [2] and Sudhakar

[3]. Both studies predicted the significant potential to decrease fuel consumption and

hydrocarbon emissions. Further experimental investigations were fulfilled by the

Cummins/TACOM adiabatic engine program during the 1980s [4]. By the end of the

program, although the feasibility of utilizing TBCs on a turbocompound diesel engine was

demonstrated, there were several unforeseen problems. For example, one of the issues

discovered was that a large amount of heat transfer losses that were saved by the TBC were

converted into higher exhaust losses instead of into useful work. However, a more

significant concern was that the thick TBCs increased the wall temperatures throughout the

entire engine cycle, which led to a large amount of charge heating during the intake stroke.

This charge heating reduced the density of incoming charge and reduced the peak power

density of the engine [5]. Furthermore, there was also a significant challenge with SI

engines, where the hot walls with thick TBCs led to an increased propensity for end-gas

knock [6]. To resolve these issues, the community proposed a new direction in TBCs

research.

3

Contrary to the thick TBC approach which results in high temperatures throughout

the entire engine cycle, a new approach has been proposed that uses thin coatings to

produce a "temperature swing", where the temperature of the surface fluctuates over the

engine cycle, which enable the coating temperature to rapidly respond to changes in the

gas temperature [7][8]. During the intake stroke, the incoming charge sufficiently cools the

coating surface, where the surface temperature is targeted to be very close to that of an

uncoated metal wall. Then, during the compression and expansion stroke, the surface

temperature of temperature swing TBCs is designed to increase and follow the gas

temperature to reduce heat transfer losses. This temperature-following feature eliminates

the intake charge heating issue of thick TBCs to a large extent, which eliminates the

negative impacts on volumetric efficiency while improving thermal efficiency [9]. A 3D-

CFD study showed that with a proper combination of the thickness, thermal conductivity,

and heat capacity, the TBC surface could achieve even lower temperatures than uncoated

metal pistons, which theoretically would increase the volumetric efficiency and improve

engine knock [7]. Later, the same group proved the effectiveness of TBCs at reducing heat

transfer losses and increasing work extraction and exhaust energy on a diesel engine

architecture. Meanwhile, a surface temperature swing of 140 °C was measured by laser-

induced phosphorescence imaging using temperature swing materials for TBCs [9]. A 0D

thermodynamic cycle simulation study was conducted by Andruskiewicz et al. showing

both volumetric and brake efficiency improvements with temperature swing coatings in an

SI engine architecture [8]. However, a following experimental investigation by the same

group showed a performance deteriorated due to poor durability with that type of

4

temperature swing coating (i.e., relying heavily on porosity to provide the favorable TBC

properties for large temperature swings) [10]. This work revealed the significance of

surface sealing, finishing, and roughness. Overall, many of the previous research efforts on

conventional combustion modes have discovered that the state-of-the-art materials or

approaches for temperature swing coatings still require further improvements to the

thermophysical properties, such as significantly lower thermal conductivity, lower heat

capacity, and enhanced durability.

All of these trials were focused on reducing heat transfer losses and were based on

conventional combustion modes; however, the community has also made significant

progress researching and developing advanced combustion modes called low-temperature

combustion (LTC). LTC exhibits the merits of both SI and conventional diesel combustion,

namely, high thermal efficiency compared to SI and ultralow emissions compared to

conventional diesel combustion [11][12][13]. A thorough introduction to LTC is presented

below.

1.2 Low-temperature combustion (LTC) family

The concept of Low-Temperature Combustion (LTC) has been studied by engine research

community for more than a decade [11][12]. LTC is intended to create a premixed (or

partially premixed) lean mixture and operate in the regions that bypass the NOx and soot

islands as shown in Figure 1. This lean mixture also benefits the thermal efficiencies due

to the higher ratio of specific heats. The combustion process is comparatively clean with

near-zero particulate and NOx emissions, therefore only requiring an oxidation catalyst for

aftertreatment, while still achieving high thermal efficiencies similar to conventional diesel

5

combustion due to the high compression ratio, and high ratio of specific heats (γ = cp/cv)

associated with low temperatures and lean operation [1].

Figure 1:Equivalence ratio & temperature map for soot and NOx formation [11] .

1.2.1 Homogeneous Charge Compression Ignition (HCCI)

Several advanced LTC modes have been proposed; for example, the early attempts of

HCCI revealed its high efficiency and ultra-low emissions characteristics [14][15].

However, the lack of control over the start and the rate of combustion limit HCCI's

commercial potential [16]. This narrow operating range and rapid heat release are related.

Studies have been conducted to try to understand the fundamental behind the scenes. Some

early attempts were to delay the combustion even that shifts the combustion to later

expansion stroke [17]. This approach is effective, but the trade-off is also significant. The

retarded combustion phasing reduces the thermal efficiency from a thermodynamic

standpoint. Also, the amount of phasing retard is limited by high cycle-to-cycle variability

and eventually misfire [18]. However, early combustion phasing causes engine knock and

6

high heat transfer losses which is also detrimental to thermal efficiency. Therefore, the

optimal combustion phasing is determined by competing effects between heat transfer

losses and effective expansion work [19], which necessitates the requirement of

staged/staggered heat release via stratified combustion.

1.2.2 Stratified Gasoline Compression Ignition (GCI)

Studies have been conducted along the path of using fuels that perform ‘staged heat

release’, such as low-temperature heat release (LTHR) or intermediate-temperature heat

release (ITHR) followed by high-temperature heat release (HTHR). An early study from

Dec et al. [20] discovered that the autoignition tendency for iso-octane and conventional

gasoline is not affected by increasing charge equivalence ratio at naturally aspirated

conditions. Moreover, the cool-flame reactions, which is another term for LTHR, is not

measurable for gasoline at naturally aspirated conditions. He also indicated that the

autoignition tendency of Primary Reference Fuel 80 (PRF80), however, is strongly affected

by increasing φ even at lower intake pressures. Based on this φ-sensitivity, the stratified

charge of PRF80 significantly advances combustion phasing and improves low load HCCI

combustion efficiency. Based on these findings, Sjöberg et al. successfully implemented

partial fuel stratification (PFS) and stretched heat release process by using stratify charged

PRF80 and PRF83 at naturally aspirated condition [21]. It was evident that two-stage

autoignition fuels (i.e., fuels with either LTHR or ITHR in addition to their HTHR) are

required to successfully implement PFS. Also, Sjöberg mentioned that diesel could be a

potential substitute fuel for PFS since it exhibits substantial two-stage heat release,

provided that the issues with its low volatility can be resolved for the intake stroke

7

injection. When the results are compared with fully premixed charge HCCI, PFS had a

lower PPRR, and was able to effectively increase the load from 5.37 bar to 5.97 bar with

acceptable engine knock and emissions. After successfully implementing PFS with PRF

blends, new findings by Dec et al. showed that gasoline exhibits φ-sensitivity at

substantially higher boost levels [22]. Specifically, for 87 Anti-Knock Index (AKI)

gasoline, which is the same fuel that was chosen in this study, the φ-sensitivity at the

naturally aspirated conditions was not observed; however, 87-AKI gasoline starts to exhibit

φ-sensitivity at an intake boost level of 1.6 bar absolute. Although the φ-sensitivity was

noticeable around 1.6 bar, it was not very strong until the boost level reached 2 bar. It was

indicated that both delaying the SOI timing and decreasing the split fraction could be

effective at decreasing the PPRR and reducing the burn rate. The following equation shows

the mathematical definition of injection split fraction (SF) that will be used throughout this

study.

𝑆𝐹 =�̇�𝑃𝑟𝑒𝑚𝑖𝑥𝑒𝑑

�̇�𝑇𝑜𝑡𝑎𝑙× 100% =

�̇�1𝑠𝑡 𝑖𝑛𝑗.

�̇�𝑇𝑜𝑡𝑎𝑙× 100% =

�̇�𝑃𝐹𝐼

�̇�𝑃𝐹𝐼 + �̇�𝐷𝐼× 100 (1)

In low-temperature combustion, especially HCCI combustion, mid-range

compression ratios are used. This is often because the engine platform is based on an SI

engine, and sometimes there is a motivation to operate the engine in either SI or HCCI (i.e.,

dual-mode capabilities). Almost all of the findings discussed above were conducted on a

heavy-duty diesel engine with a relatively low compression ratio (around 12-14). Olsson

at al. have shown that high compression ratio benefits HCCI combustion and lowers NOx

emissions at high load with almost no effects on the heat release rate [23]. In one of the

8

most recent experimental studies [24], the authors concluded that the high load limit was

more successfully extended at a compression ratio of 16 than a compression ratio of 14. It

is encouraging that gasoline displays φ-sensitivity at boosted conditions, but there is strong

motivation to explore strategies that can enhance the effectiveness of PFS at lower and

more realistic light-duty automotive boost levels (e.g., 1.6 bar). Additionally, the majority

of the experiments in the literature (described above) have been conducted on a medium-

duty or heavy-duty engine at a mid-range compression ratio. Therefore, it is desired to

explore and optimize gasoline PFS combustion strategy on a light-duty diesel engine at a

relatively high compression ratio of 16:1 and at a more practical boost level.

The cylinder stratification level is the key to authorize control over the combustion

process, where the injection strategy is one of the most effective approaches to manipulate

the stratification level. As a result, even using the same fuel as gasoline, the naming of

combustion mode/process could be somewhat distinct based on different injection

strategies. Figure 2 shows the employment of different injection that results in different

level of fuel stratification.

In PFS, the majority of fuel, i.e., 70 - 90%, is premixed via either port fuel injection

or very early direct injection so that the fuel has enough time to mix with fresh incoming

charge. The direct injection occurs at the mid-to-late compression stroke, i.e., 60 to 40

CAD before TDC, to create fuel stratification. After compression, the richer regions ignite

first, and the ignition happens sequentially based on the local equivalence ratio (typically

the combustion happens from rich to lean). The overall premixed charge in PFS mitigates

the formation of NOx and soot emissions, while the fuel stratification provides some level

9

of control over the combustion phasine and burn rate; however, as mentioned above, the

low stratification level could still be one of the limiting factors for the high-load operation

associated with high noise/heat release rate.

Figure 2: Gasoline compression ignition with different injection strategy and stratification level [25]

As the injection events are closer to TDC, the domain falls into medium/heavy fuel

stratification, or sometimes referred as just Gasoline Compression Ignition, or gasoline

Partially Premixed Compression Ignition (PPCI) [26]. Gasoline PPCI employs multiple

mid-to-late compression stroke injections, which provides more cylinder stratification and

sufficiently authorizes more control over the heat release process. Due to gasoline’s long

ignition delay, there is more time for it to get entrained with air before ignition happens

(when compared to conventional diesel fuel), and that limits the formation of soot

emissions. However, too much stratification or inhomogeneity could still bring difficulties

10

meeting the stringent emissions regulations [27]. Therefore, there is a tradeoff between

control over the combustion process and emissions formation. Finding the optimal

injection and air entrainment strategies could be the key to unlocking robust and clean GCI

combustion [28][29][30].

Overall, GCI has achieved great success from many perspectives, including

excellent fuel compatibility with low octane gasoline [31][32], high thermal efficiencies

(similar to conventional diesel combustion) [33], lower aftertreatment devices cost (mostly

related to low NOx and low soot emissions, which are the two of the most challenging

emissions to treat in conventional diesel combustion), lower fuel delivery system cost (due

to low injection pressure) [34], potentially better fuel economy (no need for DPF

regeneration, reduced parasitic losses, great potential for downsizing and down speeding).

However, the commercialization challenges for GCI remain. To achieve a higher

load limit without excessive ringing/noise, the optimal octane number of the fuel would be

relatively high (RON > ~70 or CN < ~25). This introduces great difficulties for low load

operation in terms of cyclic variability and high CO and UHC emissions [31][35][36]; thus,

cold start and idle stability are also difficult [37]. These challenges at low load operation

require a high IVC temperature, as illustrated in the previous section, to enable fuel

chemistry and achieve autoignition. Rebreathe or recompression (i.e., Negative Valve

Overlap, NVO) valvetrains are great candidates for the low load operation. They both

introduce hot residual from the previous cycle and accelerate fuel chemistry; however, the

penalty associated with high pumping losses reduces fuel economy to some extent [38][39].

On the other hand, high load operation prefers lower IVC temperatures and a large amount

11

of external cooled EGR to elongate the heat release process which lowers ringing and

mitigates the formation of NOx emissions [40]. This contradicts the ideal valvetrain setup

for the low load operation, which is another obstacle from a commercialization perspective.

Fast response and high accuracy EGR control could be challenging as well [41][42].

Additionally, high EGR rates significantly lower exhaust enthalpy which introduces

difficulties with turbocharging and aftertreatment.

Many of these challenges mentioned above are commonly experienced in various

LTC modes. The following section will introduce more about the challenges associated

with LTC.

1.2.3 Challenges with Low-Temperature Combustion

Stratified low-temperature combustion modes like GCI [43][44][45], reactivity controlled

compression ignition (RCCI) [47][48][49], and thermally stratified compression ignition

(TSCI) [50][51] provide the means of stratifying either the equivalence ratio (ϕ), in-

cylinder reactivity, or temperature distributions to achieve staged heat release process.

Although these LTC strategies still may not be able to achieve loads as high as the state-

of-the-art conventional combustion modes yet, the community is still actively working to

further extend their load ranges, and the issues of poor control and narrow load range are

not as limiting as they previously were. Because of LTC’s unique combustion and charge

preparation process, it features better emissions and potentially better efficiency too;

however, some other challenges remain, independent of the load range and control issue

that is actively being mitigated by second-generation LTC concepts.

12

One of the commonly experienced challenges in LTCs is the lower combustion

efficiency (compared to traditional SI combustion or conventional diesel combustion). In

LTC, the charge is overall lean and the peak bulk temperature is low, especially at low load

conditions [52]. Flowers et al. have shown that combustion efficiency improves

substantially with intake charge heating, but the penalty is faster heat release rates and a

decreased high load limit [53]. Additionally, previous research has shown that applying a

certain amount of EGR helps improve combustion efficiency at low-to-medium loads but

could also cause deterioration in combustion if the EGR increases beyond a certain limit

[54][55]. The same studies also indicated that EGR changes the mixture's properties, such

as its γ, which potentially decreases the amount of work that could be extracted from the

engine cycle and decreases the indicated thermal efficiency. Moreover, with some second-

generation, controlled LTC modes such as PFS or GCI, EGR lowers the oxygen availability

and restricts the maximum load limit [56]. By understanding the sources of unburned

hydrocarbon (UHC), oxygenated hydrocarbon (OHC), and carbon monoxide (CO)

emissions, the combustion efficiency can be improved. Dronniou et al. employed Planar

Laser Induced Fluorescence (PLIF) imaging technique to shed light on the in-cylinder

temperature stratification [57]. Figure 3 shows the motoring in-cylinder temperature

distribution at the top dead center (TDC), demonstrating that the wall-affected regions

could be 100 K colder than the hot spots, and some cold pockets are observed in the bulk

gas. Since LTC heavily relies on fuel chemistry to start autoignition, some of these cold

regions may never reach a sufficient temperature for long enough to achieve autoignition

or complete the combustion reaction, which contributes to incomplete combustion

13

emissions. This incomplete combustion emissions can be improved by applying a layer of

thermal barrier coatings onto the combustion chamber surfaces to increase their

temperature and help oxidization at those regions. The coating manufacturing and

preparation process will be discussed in detail in Section 2.3.

Figure 3: Simulated temperature distribution at TDC [57]

Achieving sufficient ignition temperatures requires high intake valve closing (IVC)

temperatures, which is usually achieved by trapping hot residuals or high intake

temperatures. Kuo et al. showed the feasibility of using a flexible valvetrain with negative

valve overlap to increase internal residuals; however, the penalty associated with extra

pumping losses was also noticeable [59]. On the other hand, intake heating could

potentially increase parasitic losses and indirectly affect fuel economy. Another

disadvantage of LTC is its low exhaust enthalpy due to the relatively high compression

ratios, lean operation, and high thermal efficiency, which is a challenge for aftertreatment

and turbocharging.

All of these challenges with LTC mentioned above (i.e., low combustion efficiency,

high required IVC temperature, and low exhaust enthalpies) can be improved with thermal

barrier coatings (TBCs). TBCs can lower the required intake temperature, increase the

14

temperature in the near-wall areas that contribute to UHC and CO and improve combustion

efficiency, and increase exhaust temperature for LTC aftertreatment and turbocharging.

Therefore, there is a promising marriage between the TBCs and the LTC, which can result

in high brake efficiencies while mitigating the remaining issues of LTC.

Recently, studies of thin TBCs in LTC have shown that the hotter wall temperatures

enabled by TBCs not only benefit thermal efficiency but also improve combustion

efficiency. Powell et al. have performed sequential studies with an yttria-stabilized zirconia

(YSZ) coating that showed advanced combustion phasing and an operating region shifted

toward lower load when maintaining the same intake temperature, which lowered the HCCI

low load limit associated with incomplete combustion or misfire [60]. A following trial by

the same research group focused on optimizing the temperature swing by establishing

structured porosity into coating structure (YSZ-SP), which provides more favorable

thermophysical properties with almost half the thermal conductivity and heat capacity of a

fully dense layer. Tangible improvements in thermal efficiency were reported, where the

gains of the porous structure were more than doubled than that of the fully dense coating

[61]. The authors also noted that excessive porosity and roughness could negatively impact

HCCI combustion. To avoid the possible losses caused by porosity while maximizing the

benefits of low thermal conductivity material, Filipi et al. began a search for a new pathway

of using a naturally low-k material called Gadolinium Zirconate (GdZr) with modest

porosity of only ~10% [63]. Similar experimental investigations were implemented, and

the results showed an overall enhancement of combustion and thermal efficiencies,

operating range, as well as durability, compared to previous coating generations [63].

15

All of the HCCI trials mentioned above were conducted using thin temperature

swing coatings, where the intake temperature reported in the literature was ~90 °C, and a

rebreathe valvetrain was used to trap a considerable amount of hot residuals. This intake

temperature requirement is already above most other combustion concepts, and the hot

internal residuals further contribute to high intake valve closing (IVC) temperatures. If a

low internal residual strategy is employed, the intake temperature requirement would be

much higher than ~90 °C. In other words, intake charge heating is inevitable for kinetically

controlled combustion. However, we propose that applying a thick TBC to meet the

required ignition threshold could be beneficial for combustion efficiency and thermal

efficiency in HCCI, where the thick TBCs would not incur any additional intake charge

heating penalty beyond what is currently imposed by hot internal residuals and/or intake

charge heating. In fact, the high intake temperature requirement of HCCI is viewed as one

of the current limitations to commercialization. Therefore, thick TBCs in HCCI can lower

the required intake temperature while increasing combustion efficiency, thermal

efficiency, and exhaust enthalpy (which is a separate challenge for LTC). It is expected

that these benefits will not be associated with any drawbacks to end-gas knock or charge

heating in HCCI. However, as mentioned earlier, the community has developed a number

of second-generation LTC modes that are variations on HCCI, and the combustion

characteristics vary case-by-case. Thus, the “thick” coating may not be the panacea for all

LTC modes. This dissertation will help determine the impacts of various coating properties

on different advanced combustion modes.

16

1.3 Objective of the Current Approach and Specific Tasks

The common challenges associated with LTCs can be improved with thermal barrier

coatings on the combustion chamber. The majority of the recent literature is related to

temperature-swing TBCs for conventional combustion modes to avoid the issues of charge

heating. Recently, the assessment of thin, temperature-swing TBC on LTC has been

investigated. However, the application of thick or other forms of TBCs with LTC modes

have not been studied and tested in the past. More importantly, the LTC modes are

fundamentally different from conventional combustion modes. For example, the

combustion event in conventional combustion modes is triggered either by spark discharge

(e.g., in SI combustion), or by the direct injection timing in mixing controlled combustion

(i.e., conventional diesel combustion). Therefore, the mixture's temperature history in the

conventional combustion modes is not as critical as it is for kinetically-controlled LTC

modes. In LTC, autoignition occurs when the thermodynamic conditions are fulfilled, such

as the temperature, pressure, equivalence ratio, and time. Since TBCs can alter the

mixture's thermodynamic state by influencing the intake requirements and compression

heat transfer, and since the effects of TBCs and their independent properties on LTC have

not been systematically explored, it is necessary to perform a comprehensive study on the

effects of TBCs on LTCs. Meanwhile, the impact of different fuel properties, such as

cooling potential and autoignition resistance, are also in the scope of the current work. In

the end, the evaluation of the most favorable coatings properties for different LTC modes

(i.e., HCCI and GCI) will be discussed.

17

Beyond what has been discussed above, low-temperature GCI has shown encouraging

characteristics on high efficiency and ultra-low emissions engine operation. However,

significant efforts should still be paid on optimizing the injection strategy to enhance the

cylinder stratification level to authorize more control over the combustion process at a

practical boost level, i.e., 1.6 bar, which would be favorable in a commercialization goal.

The specific tasks of this dissertation are summarised below.

1.3.1 Demonstration and justification – Comprehensive experimental investigation of

thick thermal barrier coatings for HCCI (CHAPTER 2 & CHAPTER 3)

• Experimentally investigate the effects of thick TBCs on HCCI in terms of engine

efficiencies, emissions, load range, intake heating requirement, and exhaust

enthalpy

• Investigate the performance of thick TBCs in HCCI with different fuels, including

a conventional fuel (87-AKI gasoline) and wet ethanol

• Investigate the interaction between the high latent heat of vaporization fuel spray

and the coating surface

1.3.2 Deep dive into the fundamentals – A parametric computational investigation into

the effects of various coating properties for HCCI (CHAPTER 4 & CHAPTER 5)

• Develop a 0D thermodynamic model simulation coupled with a 1D TBC solver to

numerically investigate the TBCs’ thermophysical properties and their independent

influence on the combustion characteristics of HCCI

• Optimize the TBC parameters to maximize the benefits for HCCI

1.3.3 Clemson GCI test cell commissioning (CHAPTER 6)

18

Design and commissioning a state-of-the-art GCI engine

1.3.4 Establishing the understanding of GCI - an experimental investigation of injection

strategies for gasoline PFS (CHAPTER 7)

• Explore and optimize gasoline PFS combustion strategy on a light-duty diesel

engine at a more practical boost level (i.e., 1.6 bar)

1.3.5 A preliminary guideline for TBCs with GCI – A computational evaluation

(CHAPTER 8)

• Investigate the effect of TBCs on GCI through 0D thermodynamic modeling

coupled with 3D-CFD modeling

• Explore and provide preliminary guidance of optimal TBC configuration (i.e.,

material and thickness) for GCI combustion

19

CHAPTER 2. EXPERIMENTAL SETUP AND METHODOLOGY

2.1 Experimental engine test cell

The experiments were conducted on a 421.5 cc single-cylinder light-duty diesel engine.

Figure 4 shows the layout of the entire engine system. The Ricardo Hydra engine block

was coupled with the first cylinder of a production four-cylinder, 1.7-liter GM-Isuzu engine

head, and the other cylinders were deactivated.

Figure 4: Engine test cell layout

Since advanced combustion modes do not rely on in-cylinder turbulence and

mixing in a similar manner to the conventional combustion modes, the OEM piston with a

re-entrant bowl was replaced by a custom-designed shallow bowl piston (shown in Figure

5) to improve heat transfer characteristics, combustion efficiency, and UHC emissions

associated with the LTC. Figure 5 shows the geometry of the combustion chamber with

the custom-designed pistons at the top dead center (TDC). The custom piston was designed

to have the same compression ratio (CR) as the production re-entrant bowl piston. It can

20

be noted that the squish region has been considerably reduced with the shallow bowl

design, which helps minimize incomplete combustion. Moreover, the more favorable

surface-to-volume ratio could potentially reduce heat transfer losses [30]. Some of the

relevant engine specifications are shown in Table 1.

Figure 5: Geometry of the combustion chamber at TDC

Table 1: Engine specifications

Displacement [cc] 421.5

Bore [mm] 79

Stroke [mm] 86

Connecting Rod Length [mm] 160

Compression Ratio 16.0

Number of Valves 4

Intake & Exhaust Valve Lift [mm] 8.12

Intake valve open (IVO) [deg aTDC] 354

Intake valve close (IVC) [deg aTDC] -146

Exhaust valve open (EVO) [deg aTDC] 140

Exhaust valve close (EVC) [deg aTDC] 366

The air is boosted by an air compressor and then throttled to target boost value or

naturally aspirated condition by an Alicat Scientific MCRW-1000 flow controller to

21

simulate the boosted intake air. The intake controller also measures the mass flow rate of

the intake air. A 5-kW heater is downstream of the Alicat flow meter and is PID-controlled

by a custom Labview program to control the intake temperature. After the exhaust plenum,

a back-pressure valve is mounted to regulate the exhaust pressure to ensure that the

pressure in the exhaust plenum is higher than the intake which simulates a turbocharger

and enables external exhaust gas recirculation (EGR) back to the intake plenum. An

electronically controlled solenoid valve controls the EGR flow rate. The EGR percentage

is calculated based on the CO2 content in the intake plenum, which is measured by a Horiba

MEXA 7100 D-EGR emissions bench. K-type thermocouples are used to measure most

temperatures such as the intake and exhaust temperatures immediately before and after the

intake and exhaust ports, respectively, the EGR temperature in the external-cooled EGR

line, as well as the coolant and oil temperatures at the engine inlet and outlet.

The cylinder pressure is measured by a Kistler 6041A water-cooled pressure

transducer on a high-speed basis and pegged to the intake pressure around bottom dead

center (BDC). All of the high-speed measurements are trigger by the Kistler 2614C11

crank angle encoder coupled to a pulse-multiplier, whose final resolution is 0.1 crank angle

degrees (CAD). Both the intake and exhaust pressure are measured on a high-speed basis,

which captures the working fluid’s dynamics.

2.1.1 Specific experimental setup for TBC-HCCI study

In the process of machining several pistons to be tested, some with and some

without TBCs applied, there were slight errors in the clearance volume caused by the

machining process that caused the compression ratio to deviate from the targeted value of

22

15.8:1. The compression ratios of the coated pistons varied slightly from 14.7 to 15.2.

Figure 6 shows the piston preparation process from the unmachined blank piston to the

final coated piston with the desired shallow bowl shape. First, the blank piston was

machined down to the different levels with shallow bowl shape depending on the desired

TBC thickness. Then, the primary coating materials were plasma sprayed onto the top

surface of the piston, layer-by-layer, while masking the other piston surfaces such as the

ring pack area and the piston shirt. The last step is to create a dense sealing layer on the top

surface of the coating, if desired, which involved spraying a much denser layer of smaller

particles of the same material onto the surface of the coated piston. The TBC surface was

finished with some light polishing of the ceramic surface. A more elaborate description of

the spray and coating technique is provided in the following section. Figure 7 shows the

three coated pistons after ~15 hours HCCI operation.

23

Figure 6: TBC machine & spray process

24

Figure 7: TBC conditions after 20 hours of testing

The performance of three coated pistons is compared to two uncoated metal pistons.

The coated pistons include two pistons with a 1mm and a 2mm coating that are finished by

the surface sealing process, and a 2mm coated piston without the surface sealing.

Two fuels were tested, and the details about fuels’ properties are provided in Table

2. Gasoline was only injected into the intake port to create a homogeneous air-fuel mixture.

The gasoline was an EPA Tier III EEE certification gasoline with 10% ethanol from

Haltermann Solutions. In addition to gasoline, wet ethanol 80 was chosen as an alternative

fuel with a very high latent heat of vaporization, which is 80% ethanol and 20% water on

a mass basis. Compared with other fuels such as gasoline or diesel, ethanol itself has a very

high latent heat of vaporization. Adding water further increases the heat of vaporization of

the mixture, which presents unique opportunities but also challenges. Previous work found

that this extremely high cooling potential gives wet ethanol the ability to control the IVC

25

temperature by varying the injection timing during the intake stroke, which varied the

fraction of heat that was absorbed from the combustion chamber surfaces versus the

incoming charge [68]. However, this massive evaporative cooling capacity further

increases the intake temperature requirement of LTC to balance the cooling effect from

vaporization. Additionally, the wet ethanol can result in increased wall wetting because of

its cooling potential. By applying thick TBCs with LTC of wet ethanol, the charge heating

penalty incurred by the TBCs can be used to counteract the intake evaporative cooling, and

the TBC can provide hotter walls that ensure proper evaporation. Thus, it is mutually

beneficial to use thick TBCs with advanced LTC using wet ethanol.

The DI fuel system was from a conventional diesel engine fuel delivery system

using the high-pressure common rail with a Bosch CP3 pump and a Bosch solenoid-style

direct injector. Based on the combustion mode and start of injection (SOI) timings, a

centrally mounted six-hole injector with 150°/60° included angle was used. The 150°

injector tip was used in the load sweep at fixed SOI timing of -330 aTDC. This timing and

injection angle is intended to avoid the spray targeting to the piston surface, which

maximized the penetration length of the spray and ensured the mixture's homogeneity. The

60° injector tip was used in the SOI timing sweep from -350 to

-210 deg aTDC, and this setup is intended to investigate the interaction between the spray

and the coated piston surface.

Since wet ethanol does not have the same lubricative ability as diesel, 500 ppm of

Infineum R655 lubricity additive was premixed with the wet ethanol before adding the fuel

to the fuel system. Previous experimental test results showed that adding this amount of

26

lubricity additive does not have a noticeable impact on the fuel’s autoignition properties

nor the combustion process [69]. The gasoline PFS study uses the same fuel system but

with some minor differences such as the injector included angle, and a split fraction of port

fuel injection (PFI) and direct injection (DI). The following paragraph introduces the fuel

delivery method for the gasoline PFS study.

Table 2: Fuel properties

Fuel 87-AKI gasoline Ethanol WE 80

H/C Ratio 2.003 3 3.64 O/C Ratio 0.0333 0.5 0.82

Ethanol Content [%] 9.9 100 80 Water Content [%] 0 0 20

Lower Heating Valve [MJ/kg] 41.85 26.74 21.39 Anti-knock Index (R+M)/2 87.5 100 -

2.1.2 Fuel delivery method for the gasoline PFS study

In the PFS investigation, 87-AKI (antiknock index) gasoline was delivered via both

PFI and DI (with a 60-degree included angle injector) to enable cylinder fuel stratification.

Although researchers have shown that using a dual direct injection, where the first injection

occurs during the intake stroke to create a premixed mixture and the second injection

occurs during the late compression stroke to increase the φ stratification, is an effective

method to enable PFS in a production engine [67], in this study, the gasoline is both port

fuel injected to create the homogeneous background equivalence ratio and direct injected

during the compression stroke to create the desired φ stratification. While this approach is

further from production, it provides a more fundamental experiment because the PFI and

27

DI fuel flows can be measured accurately and controlled independently, which provides

straightforward control of the split fraction. Additionally, the experiment is more

fundamental because the PFI of the background gasoline avoids any possible

inhomogeneity created by an intake stroke DI on a light-duty engine.

2.2 Data Acquisition and Analysis Methodology

Three hundred consecutive cycles of both high- and low-speed data are recorded by a

custom Labview program, which also performs real-time monitoring of several combustion

performance metrics such as processed heat release rate, ringing intensity, efficiencies, etc.

as well as providing real-time engine controls of injection pressure, timing, and the number

of injections. The raw saved data are post-processed by a high-accuracy data analysis

routine developed in a custom Matlab script. The code uses higher-order accuracy

derivatives and integration techniques, uses the NASA polynomials for mixture properties

based on the temperature, pressure, and composition of the mixture at each time-step

throughout the engine cycle, and uses a heat transfer correlation with energy closure to

provide accurate results from a heat release analysis of the experimentally collected

pressure data. The heat release analysis is performed through the following procedure. The

determination of the net heat release rate (NHRR) is first, which is derived from the first

law of thermodynamics and is shown in the following equation:

𝑁𝐻𝑅𝑅 [𝐽

𝐶𝐴] =

𝛾 ∗ 𝑝𝑑𝑉

𝛾 − 1+

𝑉𝑑𝑃

𝛾 − 1 (2)

where P is the cylinder pressure, V represents the combustion chamber volume and 𝛾 is the

ratio of specific heats. This net heat release rate represents the heat release that can be

28

captured by the cylinder pressure and volume changes, but the heat transfer losses are not

included in this step.

Chang’s heat transfer correlation is used to estimate the heat transfer losses to the wall,

head, and the piston [70], shown in the following convective heat transfer model:

𝐻𝑇 = ℎ ∗ 𝐴𝑟𝑒𝑎 ∗ (𝑇𝑏𝑢𝑙𝑘 − 𝑇𝑤𝑎𝑙𝑙) (3)

where Tbulk is the cylinder bulk temperature calculated from cylinder pressure, volume, and

trapped mass, and Twall is the estimated cylinder wall temperature at 430 K. Although the

cylinder liner, head, valves, and piston are made of different materials, and they might have

different temperatures, compared to a constant assumed wall temperature, the difference

by accurately prescribing those temperatures individually is negligible. Thus, a constant

wall temperature is used here. The area is the surface area of the combustion chamber.

Moreover, h is the convective heat transfer coefficient and it is related to the displaced

volume, bulk temperature, cylinder pressure, etc. using the Chang correlation. The heat

transfer losses are scaled using the energy closure method (i.e., scaled to match the

difference between the cumulative net heat release and the amount of fuel energy that was

injected and release by combustion). Once the heat transfer losses are determined, the gross

heat release rate (GHRR) is determined by adding the heat transfer losses to the net heat

release rate. The mass fraction burned (MFB) curve can be determined by integrating the

GHRR with respect to the crank angle, and then normalizing by the total cumulative heat

release. The MFB provides the relationship between the fraction of fuel that has burned

and the piston position, which is indicated by the crank angle. In this study, CA10

represents the crank angle where ten percent of fuel has burned, which is considered as

29

combustion initiation; CA50 represents the crank angle where 50 percent of fuel has been

burned, which is denoted as the combustion phasing. The burn duration is the duration

between CA10 and CA90, which shows how long the combustion takes from start to

completion. An uncertainty analysis of the heat release process was conducted during post-

processing [71].

The data were recorded after reaching steady-state to ensure the data quality, and

the following metrics were indicative of steady-state operation: the variation of CA50 (i.e.,

the crank angle location where 50% of the mass has burned) was less than 0.5 CAD, the

intake temperature variation was less than 0.3 K, and the coefficient of variation (COV) of

net IMEP was less than 3%. For safety considerations, the PPRR is limited below 8.5 bar

per crank angle degree, which is the most knocking case for pure HCCI. Moreover, the

peak pressure is limited below 100 bar.

2.3 Application of the TBCs and Measurements of Their Thermophysical Properties

An argon-hydrogen atmospheric plasma spray (APS) process (Oerlikon Metco, F4MB)

configured with a 6mm nozzle and a 90° 1.8mm injector was used to fabricate the primarily

yttria-stabilized zirconia (YSZ, Saint Gobain, SG204) TBCs. The piston surfaces were

prepared for TBC application by grit blasting the surface at 60 psi from a 125mm distance

using 24 mesh alumina grit. The surfaces were then cleaned and dried and were ready for

the TBC application by APS. Previous attempts to apply thick TBCs (on the millimeter

scale) experienced issues of cracking, delamination, or other failures induced by thermal

stresses during the engine cycle [72]. It was determined that these issues were related to

the difference in the coefficient of thermal expansion (CTE) between the base piston

30

material and the coating which resulted in internal stresses [73]. Additionally, thicker

coatings posed a larger challenge because of the larger change in temperature; therefore,

the internal stresses increased with the coating thickness. This issue can be resolved by

functionally grading the CTE layer-by-layer to minimize the step changes in CTE and

reduce thermal stresses [74]. In the present study, four layers of varying composition were

applied to grade the CTE starting from pure Ni5Al (Oerlikon Metco, 480NS) as a bond

coat applied to the surface of the uncoated piston. This bond coat represented about 5% of

the total thickness of the layer and was used to increase adhesion strength and resist high-

temperature oxidation in addition to grading CTE. Following the bond coat, a 50-50% by

volume YSZ-Ni5Al layer was applied to constitute 10% of the total thickness of the

coating. Next, a 70-30% by volume layer was sprayed representing 20% of the total

thickness. Finally, the bulk (65%) of the thickness of the coating was a 95-5% mixture by

volume of YSZ and Ni5Al. If the top sealcoat was desired, then an additional 97-3% by

volume thin layer was applied at about 40 μm. The seal coat was comprised of a finer YSZ

feedstock (Saint Gobain, SG240F) and a finer Ni5Al (Orelikon Metco, Diamalloy 4008).

A TA Instruments DXF 3050 thermal flash method was used to determine the properties

of each layer. An optical micrograph of the unsealed and sealed TBCs layers are shown in

Figure 8 and Figure 9, respectively.

Table 3 shows the details of each coating layer, and Table 4 has the effective coating

properties. Table 5 has the carrier gas flow rates which were optimized based on principles

reported by Vasudevan et al. [75] and the plasma gas flow rates and power, which were

determined based on a design of experiment considering in-flight particle properties

31

(Tecnar Automation AccuraSpray 3G) as outlined by Vaidya et al. [76]. For more

information about the thermal spray process, please refer to [77].

Table 3:Coating layer properties

Layer L1 L2 L3 L4 L5

1mm Layer Δx [μm] 50 100 200 650 40

2mm Layer Δx [μm] 120 240 480 1560 40

2mm Unsealed Layer Δx [μm] 120 240 480 1560 -

k [W/m-K]

[Wm-1K-1]

14.2 7.57 4.48 0.93 1.74

ρ [kg/m3] 7511 5893 5577 4490 5706

c [J/kg-K] 410 309 319 363 442

α [mm2/s] 4.62 4.15 2.52 0.52 0.69

Table 4: Combine layer properties

Combined Layer 1 mm sealed 2 mm sealed 2 mm unsealed

Thickness [μm] 1040 2440 2400

k [W/m-K] 1.328 1.267 1.261

ρ [kg/m3] 5026 4895 4880

c [J/kg-K] 354.3 347.9 346.0

α [mm2/s] 0.7455 0.7442 0.7469

Table 5: APS configurations

Layer L1 L2 L3 L4 L5

Argon [NLPM] 45 45 45 45 47

Hydrogen [NLPM] 4 6 6 6 6

Current [A] 550 550 550 550 600

Carrier Gas [NLPM] 3.5 4 3 3.5 3.5

Spray Distance [mm] 100 150 150 150 100

32

Figure 8: Optical micrograph for unsealed TBC layers

Figure 9: Optical micrograph for sealed TBCs layers

33

CHAPTER 3. THICK THERMAL BARRIER COATINGS FOR HCCI -

EXPERIMENTAL RESULTS AND DISCUSSION

3.1 Objective and Experimental operating conditions for thermal barrier coating

study

The main objective of this experimental study was to investigate the effects of thick TBCs

on the combustion and emissions characteristics of LTC. In this study, five pistons were

tested, including three coated pistons and two uncoated metal pistons to serve as

baseline/reference values. For all of the coated pistons, the desired coating thickness was

pre-machined off of the surface of the piston to provide the desired compression ratio to

match the metal baseline piston; however, due to the inaccuracies in the machining process,

the actual compression ratio of the coated pistons was slightly different from the desired

value and ranged from 14.7 to 15.2.

In order to provide a thorough comparison, two metal baseline cases with two

compression ratios were tested (15.8: and 14.0:1). It can be seen in Figure 10 that three of

the coated pistons are grouped relatively close to each other at a compression ratio of

~15.0:1, which is almost perfectly in between the two metal baselines cases. When a trend

goes above/below the metal baselines, a strong conclusion can be made that the effect of

TBC overpowers the effect of the difference in the compression ratio. Additionally, since

the coated pistons are relatively close in compression ratio, direct comparisons can be fairly

made between the coated cases.

34

Figure 10: Peak motoring pressure vs. intake temperature at naturally aspirated intake

Table 6 shows the engine operating conditions for thermal barrier coating investigation.

The engine was operated naturally aspirated in HCCI with two fuels and five pistons. For

each combination of the piston and fuel, a load sweep was performed while maintaining

the peak pressure rise rate (PPRR) below ~5 bar/CAD. In addition, due to the unique

cooling potential of WE80, several start of injection (SOI) timings were employed to

investigate the effect of spray/piston surface impingement, as well as the cooling effect

from WE80. The engine was running naturally aspirated with no exhaust throttle, i.e., near-

zero ∆𝑃 across the engine and near-zero pumping losses. It was shown that the rebreathe

or re-compression valve events could help to provide a favorable thermal environment and

to achieve controlled auto-ignition (CAI). Thus, significant intake heating or high

compression ratios were not required with CAI, which is potentially more viable for

commercialization on a gasoline engine archiecture, especially with potential multi-mode

35

operation with spark ignition. However, this study was carried out on a high compression

ratio, diesel-based engine architecture, and residuals are kept low (e.g., ~6% residual gas

fraction) to achieve pure-HCCI operation. Thus, intake heating is employed to provide the

necessary thermal condition for autoignition, and the intake temperature was varied to

match the combustion phasing for each piston at different loads.

Table 6:Engine operating conditions for the TBC studies

Engine Speed [rpm] 1200

DI Fuel Wet Ethanol 80 DI SOI Timing with 150° injector [deg aTDC] -330 IMEPg with 150° injector [bar] Swept from 2 bar to 5 bar DI SOI Timing with 60° injector [deg aTDC] Swept from -350 to -210

IMEPg with 60° injector [bar] ~3.8

DI Pressure [bar] 500

PFI Fuel 87-AKI Gasoline

PFI SOI Timing [deg aTDC] -120 IMEPg with 150° injector [bar] Swept from 2 bar to 5 bar

PFI Pressure [psi] 28

Coolant Temperature [K] 370

Oil Temperature [K] 360

3.2 Performance of TBC at different loads with conventional gasoline HCCI

Two fuels were investigated: The gasoline experiments are intended to determine the

effects of thick TBC on HCCI combustion with a conventional fuel. Wet ethanol is

intended to determine the interactions between thick TBCs and fuels with a high

evaporative cooling effect. A load sweep was performed for each fuel and piston

36

combination. During the load sweep, the low load was limited by excessive incomplete

combustion and COVs of IMEPg, while the high load was limited by the peak pressure rise

rate. In the following sections, the effects of thick TBCs on gasoline HCCI will be

illustrated first; then, the assessment of thick TBCs using wet ethanol as fuel will be

presented and compared with conventional gasoline. After that, the effect of WE80 SOI

timings will be investigated.

3.2.1 Intake temperature requirement - Gasoline

Figure 11 (a) shows the intake temperature with different pistons at different load

conditions. For all cases, the temperature requirement reduces as the load increases. Since

gasoline is not φ-sensitive at naturally aspirated conditions, this reduced intake temperature

requirement is presumably because of the hotter walls and higher residual gas temperatures

as load increases. As can be seen, the TBC cases did not break from the brackets of two

metal baseline cases; however, the estimated intake temperature for the equivalent

compression ratio of 14.9 is shown in green dotted line by taking the average of the two

metal cases. Both 2mm cases are able to lower the intake temperature requirement by 15

degrees because the hotter piston surface heats the incoming charge to reach the same

temperature after compression to achieve autoignition. The 1mm TBC cases did not show

any reduction in intake temperature even compared with the equivalent case, which could

be due to the lower CR. The error bars were not applied on this plot because the error for

K-type thermocouple is constant at about ±0.4% of the reading value.

For combustion modes like SI, diesel, and CAI combustion with constant intake

temperatures, the application of thick TBCs would lead to an intake charge heating penalty

37

and reduced volumetric efficiency and power density. However, thick TBCs seem to be

well-suited to pure-HCCI, i.e., HCCI with low residuals and higher intake temperatures,

especially for fuels with a high octane number and/or evaporative charge cooling potential.

Previous studies conducted by Powell [60][61], O'Donnell [62], and Filipi et al. [63]

showed that thin temperature swing coatings are a promising approach for HCCI with a

constant intake temperature and high internal residuals. It is important to note that there

are many ways to achieve HCCI. These studies were conducted on a mid-compression ratio

gasoline-based engine, which utilized a rebreathe valvetrain and fell into the category of

CAI. The main idea of CAI is aiming for commercialization feasibility, which brings the

hot exhaust gas back into the cylinder during the intake stroke to assure the thermal

environment for auto-ignition. In other words, if the intake temperature of CAI is not

allowed to be varied, it would be mimic a production engine’s operation. The approach

then was to apply the external EGR or variable valvetrain to adjust the combustion phasing.

The approach for the current study was to vary the intake temperature to match combustion

phasing while operating on a low-residual, pure-HCCI regime. In a production engine,

actively varying the intake temperature is not practical; however, the ability to control

combustion phasing in LTC is a requirement. Therefore, it is presumed that the production

variant would have variable valve timing with NVO, which would achieve the same effect

as varying the intake temperature to match combustion phasing. Since the experimental

engine used in this research has fixed valve timings, the intake temperature was used to

match combustion phasing rather than variable valve timing with NVO. Due to the

difference in approach between the previous work of Powell [60][61], O'Donnell [62], and

38

Filipi et al. [63] and the current study, the results and conclusions may be distinct from the

pioneering studies mentioned above due to the difference in engine architecture and

operating strategy.

Figure 11: Intake and exhaust temperature vs. IMEPg at 1200rpm with a constant CA50 for each piston, i.e., phase-match

3.2.2 The Heat Release Process and Load Range – Gasoline

The gross heat release rate (GHRR) and cylinder pressure are shown in Figure 12. The

lowest load of approximately 2 bar IMEPg is shown on the top plot, and the highest load

of approximately 4.6 bar IMEPg is shown in the bottom plot. The pressures after

compression and the peak pressures during combustion vary by piston due to the

compression ratios differences. More importantly, at a certain load condition, regardless of

the coated or metal pistons, the combustion phasings are kept constant, and the heat release

39

profiles are very similar. At the high load condition in the bottom plot, the metal piston

with CR = 15.8 (shown in blue) has a lower peak heat release rate; this is presumably due

to slightly retarded combustion phasing.

Figure 12: Gross heat release rate (left) & cylinder pressure (right) vs. crank angle

Additional combustion characteristics are shown in Figure 13, including the PPRR,

the CA50 combustion phasing, and the 10-90% burn duration. Throughout the load sweep,

the combustion phasing of different pistons was kept fairly consistent, and the CA50 is

40

retarded as the load increases because of the threshold of PPRR of ~5 bar/CAD. The PPRR

plot in Figure 13(a) and the burn duration plot in Figure 13(c) show that the 2mm sealed

TBC piston (red) has the highest PPRR and the shortest burn duration among all of the

tested pistons; however, the various piston cases are sometimes mixed with each other and

the trends are therefore inconclusive. It was hypothesized that TBCs might reduce the high-

load limit of HCCI, since HCCI combustion is extremely sensitive to the wall temperatures

and the thermal stratification in the cylinder [78][79], which might indicate that the higher

surface temperatures with TBC would accelerate the combustion process. However, the

data collected here did not conclusively show any impact of the TBCs on the high-load

limit of HCCI, since the 1mm sealed TBC, 2mm unsealed TBC, and two metal baselines

show the same trend. Thus, it can be concluded that among all of the tested pistons, the

thick TBCs do not have a noticeable impact on the heat release process. This is an important

finding because TBCs were shown to affect the high-load limit of both SI and conventional

diesel combustion due to charge heating. These results show that this effect is not a

consideration for the high-load limit of HCCI.

41

Figure 13: (a) PPRR, (b) CA50 combustion phasing, and (c) 10-90% burn duration vs. IMEPg at 1200rpm with a constant CA50 for each piston, i.e., phase-match

42

Contrary to the high-load limit, the low-load limit is considerably extended by the

TBCs. In this study, the low-load limit is defined as the load when combustion efficiency

is excessively low (below 86%). As shown in Figure 14, the dashed line is the low-load

cutoff line. The metal baselines reach the low load limit of 2.35 bar, and the 2mm TBC

case reaches 2 bar which is a 14.8% extension on the low-load limit.

3.2.3 Efficiency and Energy Distribution – Gasoline

This study follows the efficiency terminology and definitions in Heywood [1]. The

combustion efficiency, 𝜂𝑐𝑜𝑚𝑏𝑢𝑠𝑡𝑖𝑜𝑛, is defined as the percentage of fuel that burned during

the combustion process, and it is calculated from emissions speciation as shown in equation

(4), where 𝑥𝑖 is the mass fraction of CO, H2, and UHC.

𝜂𝑐𝑜𝑚𝑏 = 1 −(𝑚𝑎𝑖𝑟 + 𝑚𝑓𝑢𝑒𝑙) ∑ 𝑥𝑖𝑄𝑙ℎ𝑣,𝑖𝑖

𝑚𝑓𝑢𝑒𝑙 ∗ 𝑄𝑙ℎ𝑣,𝑓𝑢𝑒𝑙

(4)

The gross indicated fuel conversion efficiency, 𝜂𝑖𝑔,𝑓, is the efficiency based on the

total fuel energy delivered to the engine. The mathematical definition is shown in the

following equation:

𝜂𝑖𝑔,𝑓 =𝑊𝑖𝑔

𝑚𝑓𝑢𝑒𝑙 ∗ 𝑄𝑙ℎ𝑣,𝑓𝑢𝑒𝑙 (5)

where the thermodynamic work in the numerator is calculated from the measured cylinder

pressure and volume. The total fuel energy is the denominator where 𝑚𝑓𝑢𝑒𝑙 is the fuel that

is delivered to the engine per cycle, and the 𝑄𝑙ℎ𝑣,𝑓𝑢𝑒𝑙 is fuel’s lower heating value. Since

the fuel that is delivered to the engine does not completely release its energy, the gross

indicated thermal efficiency, 𝜂𝑖𝑔,𝑡ℎ, is introduced to decouple the effects of unburned fuel

43

on thermodynamics. The derivation of gross indicated thermal efficiency and the

relationship between these three efficiencies is shown in the equation below:

𝜂𝑖𝑔,𝑡ℎ =𝑊𝑖𝑔

𝑚𝑓𝑢𝑒𝑙 ∗ 𝑄𝑙ℎ𝑣,𝑓𝑢𝑒𝑙 ∗ 𝜂𝑐𝑜𝑚𝑏𝑢𝑠𝑡𝑖𝑜𝑛=

𝜂𝑖𝑔,𝑓

𝜂𝑐𝑜𝑚𝑏𝑢𝑠𝑡𝑖𝑜𝑛 (6)

In this calculation, the denominator only includes the energy that is released from the

oxidized fuel because that is the heat that was added to the thermodynamic cycle (i.e., the

“thermal” efficiency).

Figure 14 shows all three efficiencies introduced above. It can be seen in Figure

14(a) that the combustion efficiency generally increases with the load due to higher bulk

temperatures. The two metal baseline cases are relatively close to each other, with the lower

compression ratio case having a slightly higher combustion efficiency. This is due to the

higher pressures before ignition for the higher compression ratio case, which stores more

unburned fuel in the crevices. This trend agrees well with the findings from Sjöberg et al.

[80]. The combustion efficiency of the TBC cases is generally higher than that of the metal

baseline cases, and the 2mm sealed TBC case has the highest gain of about 1.5 percentage

points. For the TBC cases with a top seal coat, the combustion efficiency increases with

TBC thickness. Since most of the incomplete combustion in LTCs occurs in the cold

regions near the combustion chamber walls, the increased TBC thickness elevates the

surface temperature and improves the combustion efficiency. However, the 2mm unsealed

case does not show a significant benefit to combustion efficiency, which indicates that the

dense sealing surface has a substantial impact on fuel oxidization by sealing open pores

which can store unburned fuel. Tree et al. have shown the impacts of piston surface

roughness and porosity on fuel consumption of a diesel architecture [81]. Additionally,

44

both Powell et al. and Andruskiewicz et al. have indicated that the porosity and surface

roughness could potentially cause fuel pooling and absorption for open surface pores

during HCCI operation [82].

45

Figure 14: (a) Combustion efficiency, (b) gross indicated thermal efficiency, and (c) gross indicated fuel conversion efficiency vs. IMEPg at 1200rpm with a constant CA50 for each piston,

i.e., phase-match

46

Figure 14(b) shows the gross indicated thermal efficiency. As expected, the thermal

efficiency of the TBC cases is generally higher than the metal baseline cases, and both two

2 mm TBCs are generally higher than all the other cases. Since the reduction of heat

transfer losses and increase in thermal efficiency are only related to the material’s

properties and the TBC thickness but not the surface porosity, the thicker coatings

generally have higher thermal efficiencies than the thinner TBC or metal baselines. The

differences between the two metal baseline cases are due to the different CRs, where the

higher CR case has a higher thermal efficiency.

In Equation 5, the indicated fuel conversion efficiency is the product of the

combustion efficiency and the indicated thermal efficiency. As a result, the fuel conversion

efficiency of the 2mm sealed TBC case achieves the highest efficiency gain by 1.5 to 2

percentage points, which is about a 4-5% increase due to better combustion efficiency and

higher thermal efficiency due to reduced heat transfer losses. The 1mm sealed and the 2mm

unsealed cases have approximately the same improvement because the former has higher

combustion efficiency and the latter has higher thermal efficiency. The fuel conversion

efficiency gain with TBCs appears to have a trend of diminishing returns with increasing

thickness, where the increase from metal to 1mm is about 1.5 percentage points, and the

increase from 1mm to 2mm is only about 0.5 percentage points. However, more thickness

trials are required to determine conclusively.

The energy distribution chart of the highest load (at 4.6 bar) is shown in Figure 15.

The entire column corresponds to the fuel energy that is delivered to the engine, which

includes four sections: 1) the gross work, 2) the combustion inefficiency, 3) the exhaust

47

waste heat, and 4) the heat transfer losses (from Chang’s heat transfer correlations [70]

with energy closure as described above). The heat transfer portions are generally lower for

TBC cases, which supports the trends in thermal efficiency and the analysis mentioned

above. The saved heat transfer losses and the reduced unburned fuel increase the useful

work and the exhaust gas enthalpy. Comparing two metal baselines, the heat transfer losses

are approximately the same, and the MetalL case (the thicker head gasket and lower

compression ratio) has higher exhaust enthalpy and lower work output because of the lower

compression ratio and lower thermal efficiency. Note that if the compression ratio of metal

baseline perfectly matched the TBC cases, the values of the baseline would be about the

average of the two metal cases, since the change in compression ratio is small and the trend

can be considered linear over that range.

48

Figure 15: Energy distribution chart for gasoline with the five different pistons at load of 4.6 bar IMEPg. MetalH is the metal piston with the higher compression ratio and MetalL is the metal

piston with the lower compression ratio.

3.2.4 Emissions – Gasoline

The emissions are shown in Figure 16. The 2mm sealed TBC case has the lowest UHC and

CO emissions due to higher surface temperatures which raise the gas temperature of the

cold regions near the piston surface. It is interesting to note that the TBCs improve both

the UHC and CO emissions, rather than only affecting the UHC emissions. For the 2mm

unsealed case, even though the surface temperature is higher than that of the 1mm sealed

surface, it has higher UHC emissions, which is assumed to be due to the porous surface

storing unburned fuel.

49

It was hypothesized that the NOx emissions might increase for the TBC cases;

however, the results did not break the metal baseline brackets. Therefore, it can be

concluded that thick TBC do not negatively impact NOx emissions.

Figure 16: (a) UHC, (b) CO, and (c) NOx emissions vs. IMEPg, at 1200rpm with a constant CA50 for each piston, i.e., phase-match

3.3 Performance of TBC at different loads with WE80 and compared with gasoline

In this section, similar experiments were conducted with wet ethanol 80 instead of gasoline.

The wet ethanol was direct injected at -330 CAD aTDC. The results from one of the metal

50

baseline cases (the higher compression ratio case) and the 2mm sealed TBC case will be

compared with the gasoline cases. The comparison is mainly focused on the load range,

efficiencies, and intake temperature requirement.

3.3.1 Load range, efficiencies, and energy distribution – WE80 & gasoline

Figure 17 shows the efficiencies versus the load range of both wet ethanol (dashed lines)

and gasoline (solid lines). Since ethanol has a higher octane number, it can be seen that the

load range for WE80 is slightly shifted toward higher loads compared to gasoline. For

WE80, the 2 mm sealed TBC improves the low-load limit from 2.6 bar IMEPg to 2.2 bar

IMEPg, which is an improvement of 15.4%, while the high load limit is approximately

constant (5.02 vs. 4.99 bar).

51

Figure 17: (a) Combustion efficiency, (b) gross indicated thermal efficiency, and (c) gross indicated fuel conversion efficiency vs. IMEPg for gasoline (solid lines) and wet ethanol (dashed

lines) at 1200rpm with a constant CA50 for each piston, i.e., phase-match

52

The combustion efficiency is shown in Figure 17(a). The WE80 metal baseline is

generally lower than that of the gasoline. One possible reason is that ethanol has a higher

autoignition resistance, which potentially leads to more incomplete combustion from the

cold regions. Otherwise, the high cooling potential of WE directly injected into the cylinder

and the spray targeting the piston crown during injection might cause poor evaporation and

wall wetting on the cold piston surface. The combustion efficiency of WE80 is significantly

improved with the 2mm sealed TBC on the piston. At lower loads, gasoline still exhibits

higher combustion efficiency because the surface temperature may not be high enough to

overcome the evaporative cooling of the WE. But, as the load increases, the surface

temperature of the piston increases, which aids evaporation and helps the combustion

efficiency of the WE80 case eventually catch up with the gasoline case. Overall, the

combustion efficiency is increased by up to 1.5 percentage points for WE. The gasoline

experiences most of the combustion efficiency benefits at low loads, while the WE

experiences most of its combustion efficiency benefits at mid-to-high load. The emissions

data for the WE and gasoline comparison are shown in Figure 18, which agrees well with

the discussion above. These combustion efficiency improvements provide evidence of why

thick TBCs are well suited to fuels with high evaporative cooling potential.

53

Figure 18: (a) UHC, (b) CO, and (c) NOx emissions vs. IMEPg for gasoline (solid lines) and wet ethanol (dashed lines) at 1200rpm with a constant CA50 for each piston, i.e., phase-match

54

Thick TBCs provide a similar gross indicated thermal efficiency gain for the WE

case as the gasoline case, and the increase in fuel conversion efficiency is higher for the

WE due to the larger improvement in combustion efficiency. The plots of thermal

efficiency and fuel conversion efficiency did not include error bars because of the clearness

of the figure. A version of the fuel conversion efficiency plot that includes error bars is

shown in the Appendix in Figure 84.

The energy distribution chart at the load of 4.6 bar for both WE (second highest

load for WE) and gasoline (highest load for gasoline) is shown in Figure 19. The

distribution structure is similar to Figure 15. For the metal baseline comparison, WE has a

slightly higher percentage of work output than gasoline. One of the possible explanations

for this is that the air-fuel mixture with WE has a slightly higher ratio of specific heats (𝛾)

than gasoline. Other than that, the distribution for both gasoline and WE are very similar.

Reduced heat transfer losses are tangible in both WE and gasoline cases, and the reduction

in heat transfer losses for the 2mm sealed piston is about 13.6%. It is important to note that

only the piston was coated due to ease in the coating process and the availability of spare

piston blanks. However, in a production variant of this concept, the piston, head, and valves

would all be coated which would amplify all of the trends shown in this study.

55

Figure 19:Energy distribution chart for WE80 and gasoline

3.3.2 Intake temperature requirement – WE80 & gasoline

The trends of intake temperature are shown in Figure 20. The intake temperature of all

cases decreases as load increases. Since both gasoline and wet ethanol do not exhibit φ-

sensitive at this intake pressure and compression ratio [22], this decrease in intake

temperature is most likely associated with increasing residual gas temperatures and

increasing wall temperature. Dec et al. have shown similar trends using iso-octane [83],

which is also a single-stage heat release fuel (that is not φ-sensitive) at naturally aspirated

conditions.

56

Figure 20: Intake temperature vs. IMEPg at 1200rpm with a constant CA50 for each piston, i.e., phase-match

Another interesting trend that can be seen in Figure 20 is that the temperature

decreases by about 40 K from the lowest load to the highest load for gasoline – however,

it only decreased by about 10 K for WE80. This is because the WE80 has an extremely

high evaporative cooling potential. As shown in Figure 21, the cooling potential of WE80

is significantly higher than gasoline (about 6.5 times), and this evaporative cooling

increases as the equivalence ratio increases. The high cooling potential needs to be

compensated for with intake heating to achieve autoignition. It is important to note that

these results were generated from steady-state operating points. The temperature

requirement of the transient operation would be different due to differences in transient

wall temperatures, residual properties, and combustion phasings. The 2mm sealed TBC

case is able to decrease the intake temperature requirement by about 10 degrees when

57

compared with the equivalent CR case, but the reduction is not as much as gasoline. The

plot that includes all of the tested pistons for WE80 is shown in Figure 22.

Figure 21: Cooling potential of different fuels

Figure 22: Intake and exhaust temperature vs. IMEPg (Wet Ethanol 80) at 1200rpm with a constant CA50 for each piston, i.e., phase-match

58

3.4 Performance of TBC with WE80 with varied SOI timings

3.4.1 The effect of injection timing and TBC on heat release process

Since wet ethanol has a high latent heat of vaporization, a sweep of SOI timing from -350

to -210 deg aTDC is used to help study the effects of thick TBC with a high heat of

vaporization fuel in advanced combustion. It can be observed in Figure 23 that all of the

cases have very similar heat release characteristics. Additional gross heat release rate

(GHRR) plots for other SOI timings are provided in the Appendix, Figure 85. Other heat

release indicators, such as peak pressure rise rate (PPRR), CA50, and burn duration for all

SOI timings, are shown in Figure 24.

Figure 23: Gross heat release rates (bottom) and pressure traces (top) vs. crank angle degree at SOI of -350 deg aTDC

The CA50s are targeted at a constant phasing around 6.8 deg aTDC, by adjusting

the intake temperature, which is shown in Figure 24(b). The metal piston with the thick

59

head gasket has a narrower SOI timing sweep (truncated at -250 deg aTDC), due to the

high intake temperature requirement for the low compression ratio case. The maximum

intake temperature supported by the experimental setup is 470 K, which was reached, and

the injection timing could not be retarded further. For all cases, the PPRR has a generally

consistent agreement. Although the 2mm sealed TBC case is usually slightly higher than

the others, the TBCs generally did not increase the pressure rise rate noticeably, despite the

higher piston surface temperature. By examining all of the cases in Figure 24 and the

60

GHRR figures in the Appendix, it can be concluded that TBC did not significantly impact

the heat release process and the knock intensity when matching CA50.

Figure 24: (a) PPRR, (b) combustion phasing, (c) burn duration vs. SOI timing with different coated pistons and the metal baseline cases, CA50 = 6.8 deg aTDC

3.4.2 The effect of injection timing and TBC on efficiencies and emissions

It can be observed from Figure 25 that the coated pistons and two metal baselines exhibit

the same trends that were shown in the previous section. Where the 2mm seal coated case

61

has the highest combustion efficiency, gross indicated thermal efficiency, and

consequently, fuel conversion efficiency. The combustion efficiency generally increases

with SOI timing before -270/-240 deg aTDC, then decreases. The reason for this trend is

that at early SOI timings such as -350 to -330 aTDC, the piston is very close to TDC, and

there is not much air in the cylinder yet. Therefore, the fuel impinges upon the piston

surface, which results in some mass in the wall film and lowers the combustion efficiency.

As the piston moves down at later SOI timings around -270/240 deg aTDC, a considerable

amount of hot air is inducted into the cylinder, which helps spray break up and provides

heat to compensate for the evaporative cooling. Therefore, the unburned fuel decreases due

to less mass in the wall film and the combustion efficiency increases. As the injection

timing is delayed further, the combustion efficiency starts to decrease after -240 deg aTDC

due to the spray targeting the crevice volume around -230 deg aTDC, where the unburned

fuel is stored and released as a significant portion of UHC emissions. Further delaying

injection timing leads to the fuel spray impinging upon the cylinder liner which causes wall

wetting and lowers combustion efficiency. Figure 26 includes a 3D CAD model that shows

the spray angle and piston position at different injection timings, which supports the

explanations above.

62

Figure 25: Efficiencies vs. SOI timing for the different coated and metal baseline pistons, CA50 = 6.8 deg aTDC

63

Figure 26: Fuel spray visualizations at injection timings of -330, -300, -270, -240, and -210 degrees aTDC

Since the trends of combustion efficiency remain mostly the same among the metal

baselines and TBC cases through all of the tested injection timings, the spray/coating

surface interaction is not conclusive at this point. In order to further investigate the

interaction between the fuel spray and hot coated piston, a subset was chosen and analyzed.

In this subset, two injection angles (150° and 60°) were enabled by switching the injector

tips, and both injectors used an SOI timing of -330 deg aTDC with the same fueling rate.

Figure 27 shows the spray direction and the piston position. It can be seen that the 150°

injection angle avoids the spray aiming at the piston surface, and the penetration length

goes all the way up to the liner. Thus the 150° injector case represents the low impingement

interaction case. On the contrary, the 60° injector represents the high spray/surface

interaction case, where the spray aims at the center of the piston crown, and there is not

much room for it to penetrate before impinging the piston.

64

Figure 27: Visualization of two included angles (150° and 60°) at SOI timing of -330 deg aTDC

The combustion efficiency of this subset comparison is shown in Figure 28. By

comparing the two metal baselines (in blue), it can be seen that the 150° case has higher

combustion efficiency than the 60° case, which is mostly due to the longer penetration

length that provides better evaporation and less wetting on piston surface. In addition, it

can be seen that the TBC improves the combustion efficiency of both injection angles.

However, it is interesting to learn that the improvement of the 60° injector is larger than

the 150° one (1.65% versus 1.02%). For the 150° case, the increase was mostly due to the

hotter near-wall regions, but wall wetting may not be significantly addressed since the

spray was targeting only the edge of the piston. The improvement of the 60° case was

associated with both fewer cold regions and better evaporation (reduced fuel film mass)

because the spray can evaporate off the hot coating surface when the spray impinges on the

piston in the 60° case. Although the TBC helps the evaporation and improves the

combustion efficiency at SOI of -330, Figure 25 shows that the optimal SOI timing in terms

65

of combustion efficiency is about -240 aTDC, which is to say that the TBC helped

addressed wall wetting to some extent, but the longer penetration length and hot air in the

cylinder are more essential in terms of spray breakup and reduced wall wetting. The

findings from other researchers also support this theory [102].

Figure 28: Combustion efficiency with two injection angles (150° and 60°)

Figure 29 shows the estimated peak cylinder temperature and emissions with

different coated pistons and the metal baseline cases. The UHC and CO emissions trends

agree well with the trend in combustion efficiency discussed above as well as the trends in

the previous section. Both the UHC and CO emissions decrease with TBC cases due to an

increased piston surface temperature, and the NOx emissions remain at a very low level.

66

Figure 29: Emissions and peak cylinder temperature vs. SOI timing for the different coated and metal baseline pistons, CA50 = 6.8 deg aTDC

67

3.4.3 The intake and exhaust temperatures

The intake temperature is shown in Figure 30 (a). The coating decreases the intake

temperature requirement, and the thicker coating results in lower intake temperature while

maintaining the same combustion phasing.

Figure 30: The intake temperature requirements and the measured exhaust temperatures vs. SOI timing for the different coated and metal baseline pistons, CA50 = 6.8 deg aTDC

The subset of data for the different injector included angles has been chosen to

evaluate the effects of spray/surface interaction on intake temperature, and the results are

shown in Figure 31. Due to more spray/coating surface interaction, the reduction in intake

68

temperature of 60° injector is almost twice as large as the 150° injector (19K versus 10K).

This is presumably because the fuel spray was directed at the hotter piston surface so that

the evaporation process absorbs some amount of heat from the coating surface instead of

the air/relatively cold liner; thus, less heat was needed from the intake charge to achieve a

similar IVC temperature. The trends in both Figure 28 and Figure 31 correlate very well to

each other, and it can be concluded that the spray/TBC impingement would amplify any

effect that was caused by switching from cold metal piston surface to hot coated surface in

terms of the fuel evaporation.

Figure 31: Intake temperature with two injection angles (150° and 60°)

The exhaust temperatures are shown in Figure 30(b). The exhaust temperatures for

the TBC cases were expected to be higher than the metal baseline cases due to the lower

heat transfer losses from the better-insulated combustion chamber. Some of the would-be

69

heat transfer losses were converted to work, which is why the thermal efficiency increased,

and some of that saved energy is rejected to the exhaust. However, there is no observable

effect of the TBCs on the exhaust gas temperature from this experimental study, possibly

due to the compression ratio differences or possibly due to experimental inaccuracies in

measuring the exhaust temperature.

Since TBCs can alter the mixture’s thermodynamic state by influencing the intake

requirements and compression heat transfer, and since the effects of TBCs and their

independent properties such as thermal conductivity, volumetric heat capacity, and coating

thickness on HCCI have not been systematically explored, it is necessary to use the

modeling techniques to perform a comprehensive study on the effects of TBCs on HCCI

from a fundamental thermodynamic perspective. Thus, a zero-dimensional thermodynamic

model was established to perform a systematic computational study of the effects of TBC

on HCCI. A detailed description of the model is provided in the next chapter.

70

CHAPTER 4. MODELING SETUP AND VALIDATION

4.1 0-D thermodynamic engine cycle modeling

A zero-dimensional, single-zone thermodynamic model was established based on the

experimental research engine described above to simulate the effects of TBCs on LTC

more systematically. The model was established and validated based on the engine

mentioned in the previous section. The initial state of the mixture is given as an initial

condition, and the model solves for the in-cylinder composition and states for the next step

based on mass flow, energy conservation, and the chamber volume change due to the piston

motion. The time resolution used throughout this work is 0.1 crank angle degrees (CAD).

4.1.1 Conservation of mass & flow characterization

The mass conservation equation includes the mass flows into the cylinder during the intake

stroke, �̇�𝑖𝑛𝑡, and the injection event, �̇�𝑓𝑢𝑒𝑙. Mass leaves the cylinder during the exhaust

stroke, �̇�𝑒𝑥ℎ, or due to blowby losses, �̇�𝑏𝑏. The subscript 𝑖 indicates the instantaneous

time step. The mass conservation equation is given as follow:

�̇�𝑐𝑦𝑙,𝑖 = �̇�𝑖𝑛𝑡,𝑖 + �̇�𝑓𝑢𝑒𝑙,𝑖 + �̇�𝑒𝑥ℎ,𝑖 + �̇�𝑏𝑏,𝑖 (7)

where the intake, exhaust, and blowby flows are derived from 1D isentropic flow analysis

through an orifice [1]. Equation 8 describes the unchoked flow, where 𝑃𝑑

𝑃𝑢> (

2

𝛾+1)

𝛾

𝛾−1 .

Otherwise, Equation 9 should be applied for the choked conditions.

�̇�𝑢𝑛𝑐ℎ𝑜𝑘𝑒𝑑 = 𝐶𝑑 ∗ 𝐴𝑟𝑒𝑎 ∗ 𝑃𝑢 (

𝑃𝑑

𝑃𝑢)

1𝛾 √

2𝛾 ∗1 − (

𝑃𝑑

𝑃𝑢)

𝛾−1𝛾

(𝛾 − 1) ∗ 𝑅 ∗ 𝑇𝑢

(8)

71

�̇�𝑐ℎ𝑜𝑘𝑒𝑑 = 𝐶𝑑 ∗ 𝐴𝑟𝑒𝑎 ∗ 𝑃𝑢

√2𝛾 ∗

(1

𝛾 + 1)

𝛾+1𝛾−1

(𝛾 − 1) ∗ 𝑅 ∗ 𝑇𝑢

(9)

where 𝐶𝑑 is the discharge coefficient, 𝐴 is the minimum of the curtain area or the valve

area, 𝑃𝑢 and 𝑃𝑑 are the upstream and downstream pressures that were determined by the

flow direction (e.g., if the working fluid flows through the intake port into the cylinder, the

upstream pressure 𝑃𝑢 would be the intake manifold pressure and the downstream pressure

𝑃𝑑 would be the pressure inside of the cylinder). The intake and exhaust conditions are

considered stagnation conditions, and the cylinder condition changes through the cycle.

Since pressure wave travels as the speed of the sound, the cylinder pressure is assumed to

be uniform in the combustion chamber.

Since this investigation is mostly related to the features of HCCI combustion, it is

not necessary to apply a spray model. Thus, the mixture is considered as homogeneous as

soon as the injection occurs.

4.1.2 Energy balance

According to the first law of thermodynamics, the energy balance for an open system is

shown in Equation 10:

�̇�𝑖 = −�̇�𝑖 − �̇�ℎ𝑡,𝑖 − �̇�𝑐𝑟,𝑖 + ∑ �̇�𝑖𝑛,𝑖𝑗 ∗ ℎ𝑖𝑛,𝑖𝑗 − ∑ �̇�𝑜𝑢𝑡,𝑖𝑗 ∗ ℎ𝑜𝑢𝑡,𝑖𝑗 (10)

where �̇� is the change in the internal energy, �̇� is the thermodynamic work extraction rate

that is derived from 𝑝�̇�. Equation 11 shows the mathematical definition of the convective

heat transfer rate �̇�ℎ𝑡, which uses Chang’s heat transfer correlations for HCCI combustion

72

[84], where ℎ𝑐𝑜𝑛𝑣 is the convective heat transfer coefficient determined from Chang’s

correlation, 𝑇𝑔𝑎𝑠 is the bulk gas temperature, and 𝑇𝑠𝑢𝑟𝑓𝑎𝑐𝑒 is the chamber surface

temperature.

�̇�ℎ𝑡,𝑖 = 𝐴 ∗ ℎ𝑐𝑜𝑛𝑣 ∗ (𝑇𝑔𝑎𝑠 − 𝑇𝑠𝑢𝑟𝑓𝑎𝑐𝑒) (11)

�̇�𝑐𝑟,𝑖 = �̇�𝑐𝑟,𝑖 ∗ ℎ𝑖,𝑐𝑟/𝑏𝑢𝑙𝑘 (12)

Equation 12 shows the modeling of the rate of crevice flow �̇�𝑐𝑟, where �̇�𝑐𝑟 is the

rate of mass flow into and out of the crevice volume due to the pressure difference and ℎ

is either the enthalpy of the cylinder bulk gas or the gas in the crevice depending on the

flow direction [85]. The subscript 𝑗 in the mass flow terms and enthalpy terms in Equation

10 indicates different flow types that were shown in Equation 7. Then, the specific internal

energy of formation and the specific volume at each step can be determined as the forms

below, where 𝑉𝑖 is the chamber volume at each step.

𝑢𝑖 =∑ �̇�𝑖

𝑛𝑖=1

𝑚𝑐𝑦𝑙,𝑖 (13)

𝑣𝑖 =𝑉𝑖

𝑚𝑐𝑦𝑙,𝑖 (14)

4.1.3 Thermodynamic properties of the working fluid

In the simulation, the working fluid is assumed to be an ideal gas that contains six species:

fuel, oxygen, nitrogen, CO2, argon, and H2O. Since experimental results showed that the

TBCs have no observable impacts on the rate and duration of the heat release process when

matching the combustion phasing and fueling and it has been documented in different

studies that the reduced intake temperature could compensate for the increased wall

73

temperature to match combustion phasing [86][87], the combustion process uses a

prescribed mass fraction burned (MFB) curve from the experimental data with an adjusted

intake temperature, and from the prescribed MFB curve, the changes in species fractions

can be determined at each time step. There are two parameters that characterize the HCCI

combustion process that need to be predicted by the model: the ignition timing and the burn

rate. The ignition timing is represented as CA0, which is adjusted by altering the intake

temperature. The determination of CA0 follows the Livengood-Wu autoignition integral

shown in Equation 15, where 𝜏 is the ignition delay time in second, and the 𝑡𝐶𝐴0 is time

when the integral equals 1 (i.e., when the autoignition occurs).

∫1

𝜏

𝑡𝐶𝐴0

𝑡0

𝑑𝑡 = 1 (15)

The autoignition integral requires an ignition delay correlation, which is shown in

Equation 16. It was modified from He’s iso-octane correlation by changing the activation

energy and the preexponential constant to best fit the experimental results (since gasoline

was used in this experiment rather than isooctane in He’s work). In Equation 16, 𝑃𝑖 is the

instantaneous cylinder pressure in bar, 𝜙 is the equivalence ratio, 𝑥𝑂2is the fraction of

oxygen on a molar basis, R is universal gas constant in 𝑐𝑎𝑙 ∙ 𝐾−1 ∙ 𝑚𝑜𝑙𝑒−1, and 𝑇𝑖𝑠𝑒𝑛,𝑖 is

isentropic unburned gas temperature proposed by Lawler et al. that was derived from the

isentropic ideal gas relationship. It provides a thermodynamic relation for the hottest

possible temperature (i.e. the adiabatic core) prior to autoignition. Equation 17 shows the

mathematical determination of the isentropic unburned temperature, where 𝑇𝐼𝑉𝐶 and 𝑃𝐼𝑉𝐶

are the temperature and pressure at IVC, respectively, and 𝛾𝑖 is the fluid’s ratio of the

74

specific heats. More introduction related to the isentropic temperature can be found in

[92][93].

𝜏 = 4.4 ∙ 10−7𝑃𝑖−1.05𝜙−0.77𝑥𝑜2

−1.41𝑒𝑥𝑝 (29970

𝑅𝑇𝑖𝑠𝑒𝑛,𝑖) (16)

𝑇𝑖𝑠𝑒𝑛,𝑖 = 𝑇𝐼𝑉𝐶 ∗ (𝑃𝑖

𝑃𝐼𝑉𝐶)

1−1𝛾𝑖

(17)

For the burn rate, a Wiebe function is a typical approach for modeling the burn rate

[101]; however, the heat release process varies engine-to-engine based on a variety of

factors. Since there is a vast amount of experimental data available for the modeled engine,

which primarily serves as a map of different heat release profiles associated with different

operating conditions, an interpolation-style combustion model is adopted to best reproduce

the combustion event on this engine platform. The previous results showed that the burn

rate was not significantly affected by the TBCs. Furthermore, previous work from Zhou et

al. demonstrated that the HCCI heat release process could be accurately defined by CA0

and equivalence ratio (𝜙) [91]. This is to say that as long as CA0 and 𝜙 are predetermined,

a valid MFB curve can be interpolated from the experimental data. Therefore, the

autoignition integral and ignition delay correlation in Equation 15 and 16were used to

determine CA0. Then, based on CA0 and 𝜙, a MFB curve was interpolated from the

experimental data set such that the model can still be considered validated and it will be

predictive in relation to the TBCs effects on combustion.

The combustion efficiency in HCCI is usually correlated with the peak cylinder

temperature in 0-1D modeling studies, which has been adopted and reported in the

75

literature [94][95]. However, the TBCs can affect the cold near-wall regions, which

contribute to incomplete combustion, thereby increasing combustion efficiency without a

significant effect on the peak bulk temperature. Therefore, the approach of correlating

combustion efficiency with peak bulk temperature is not viable with TBCs. As a result,

this study only focuses on the indicated thermal efficiency, which is independent of the

combustion efficiency (based on the efficiency definitions found in Heywood [1]). For the

trends in combustion efficiency, the experimental data in the previous section show how

TBCs affect combustion efficiency in LTC.

After defining the gas composition, the mixture’s thermodynamic properties are

determined through a Cantera-style subroutine that uses the NASA 9-coefficient

polynomial parameterization for each species. According to the properties determined from

Equations 13 and 14, the bulk gas temperature and pressure can be solved from the ideal

gas law. Then, with the new pressure and temperature for the next step (𝑖 + 1), the mass

and energy flows can be determined sequentially through the entire engine cycle.

4.2 1D transient heat transfer modeling

To determine how the coating properties affect the thermodynamic performance of the

engine cycle, a one-dimensional transient finite element heat transfer model was

established to simulate the resultant surface temperature and its effects on the

thermodynamics for different coating properties.

Figure 32 is a schematic of the 1D transient heat transfer model with boundary

conditions, which was inspired by the study from Güralp et al. [88]. The model includes

two layers, where the TBC layer is shown in red and subscripted by “c.” Below that is the

76

metal layer, which is shown in blue and subscripted with “M.” The horizontal axis is the

time domain through the engine cycle from -360 CAD to 360 CAD aTDC at a resolution

of 0.1 CAD (TDC is at CAD = 0). The depth/distance domain is on the vertical axis.

Figure 32: One-dimensional transient finite element heat transfer schematic

The boundary condition at the top surface uses the convective heat transfer from

the bulk gas into the wall surface and then conduction occurs through the coating. Equation

18 shows the derivation of the top surface boundary condition:

�̇� = �̇�ℎ𝑡,𝑖𝑛 − �̇�𝑜𝑢𝑡 = ℎ𝐴(𝑇𝑔𝑎𝑠 − 𝑇𝑛𝑥) − 𝑘𝑐𝐴𝜕𝑇

𝜕𝑥 (18)

�̇�ℎ𝑡,𝑖𝑛 is the heat transfer rate from the bulk gas, and �̇�𝑜𝑢𝑡 is the heat conduction

rate to the sub-layer elements. On the far left-hand side, the first term, �̇�, is the rate of

77

change of internal energy of the top row of elements: �̇� = 𝑚𝑐𝑝∆𝑇/∆𝑡. Substituting this

into Equation 18, the temperature profile for the first row can be determined as follow:

𝑇𝑛𝑥𝑖+1 = 𝑇𝑛𝑥

𝑖 +2∆𝑡{ℎ(𝑇𝑔𝑎𝑠

𝑖 − 𝑇𝑛𝑥𝑖 ) − 𝑘𝑐(𝑇𝑛𝑥

𝑖 − 𝑇𝑛𝑥−1𝑖 )}

𝜌𝑐𝑐𝑝,𝑐∆𝑥𝑐2

(19)

where ℎ is the convective heat transfer coefficient that was determined from the Chang

heat transfer correlation for HCCI combustion [84], 𝑘 is the thermal conductivity, 𝑐𝑝 is the

specific heat capacity, and 𝜌 is the density of the substance.

The boundary condition at the bottom layer, where 𝑗 = 1, is assumed to be a

forced convection heat transfer condition. In this study, TBCs will be simulated on the

piston, head, and valves. The back of the piston is cooled by an oil squirter whereas the

back of the cylinder head is cooled by coolant. The valves themselves have unique heat

transfer pathways including conduction through the valve seat or convection by the

surrounding flow during gas exchange. As an example of how the backside boundary

condition’s convective heat transfer coefficients are determined, the piston will be

discussed first. Since the oil temperature and pressure are maintained constant, the oil

cooling jet is relatively repeatable from cycle-to-cycle. Therefore, it is safe to assume that

the convective heat transfer coefficient for the oil cooling jet, ℎ𝑜𝑖𝑙 , is a constant value

during steady-state operation. This value is determined experimentally by using the

measured surface temperature of the metal piston and an effective heat conduction

thickness determined from a CAD model of the piston. For this engine architecture and

operating conditions, the ℎ𝑜𝑖𝑙 is determined to be 1335 W/m2K based on the best matching

of measured metal surface temperature from previous experimental results [86]. This value

78

is similar to the value that Assanis et al. used in their engine cycle simulation [89]. The

effective thickness of the piston is 16.4 mm, and for convenience, the cooling area uses the

bore area. Note that this coefficient would change with oil pressure, engine speed, oil

gallery geometry, and other factors related to backside piston cooling with engine oil. With

the same methodology, the convection coefficient of the backside of the firedeck is

determined to be 1880 W/m2K based on measured head temperatures. Due to the lack of

measured data of valves’ surface temperature, the backside convective coefficient for the

vales is assumed to be the same as the firedeck surface. In the future, additional

experimental data or computational fluid dynamics modeling would help determine the

convective heat transfer coefficient for the intake and exhaust valves.

Figure 33: Determination of oil cooling convective heat transfer coefficient for the backside boundary condition

79

After assigning the top and bottom boundary conditions, the remaining layers use

the 1D transient heat conduction equation shown in Equation 20:

𝜕𝑇

𝜕𝑡=

𝑘

𝜌𝑐𝑝×

𝜕2𝑇

𝜕𝑥2 (20)

Since the two materials have different properties, the heat flux at the contact surface

on both sides must equal (i.e., the heat conduction rate from 𝑛𝑐+1 to 𝑛𝑐 is the same as that

of 𝑛𝑐 to 𝑛𝑐−1), which serves as another boundary condition.

Once the surface temperature has been determined, it can be substituted back into

Equation 11 and thermodynamic energy flow can be re-simulated to calculate the new

cylinder temperature and pressure. Then, using that new gas temperature as the new topside

boundary conditions to the 1D transient heat conduction solver for the TBC layer, a new

surface temperature can be determined. The process iterates until the error converges,

which requires between 3 to 20 iterations depending on the coating properties.

4.3 Model validation

The model was validated against experimental data in two steps. First, the model

performance was compared with the metal baseline at different loads. Then, the model

performance was examined by validating against experimental data for the TBC case with

a 2 mm TBC-coated piston. During the validation process, there are three main metrics to

evaluate the model’s fidelity: the agreement of the cylinder pressure, the mass flow rate of

air, and the gross indicated mean effective pressure (IMEPg). The air mass flow rate and

pressure at IVC are two significant indicators to justify the trapped mass, composition, and

the thermodynamic state at the beginning of the closed cycle. Then, the cylinder pressure

80

values after compression and during combustion are good indicators for tuning the

compression ratio, blowby mass, and crevice flow. Most importantly, IMEPg is the metric

to evaluate the model’s accuracy at predicting the engine’s efficiency and power.

4.3.1 Validation against the experimental metal baseline cases

The purpose of metal case validation is to examine the model’s capability of precisely

capturing the thermodynamics, conservation of mass, and the energy balance. In this

baseline validation process, the model implements the exact same amount of fuel measured

from the experimental data, as well as the other operating conditions such as the engine

speed, intake and exhaust pressures, the injection strategies, etc.

The engine was operated at a constant speed of 1200 rpm, wide open throttle (WOT)

and without external EGR. The fuel was delivered through a port fuel injection (PFI)

system to the intake port at -120 deg aTDC. The fueling rate varied from 5.6 mg/cycle to

11 mg/cycle, which corresponded to an equivalence ratio of 0.25 to 0.46. Additional

information about the operating conditions can be found in Table 6. Two sets of

comparison at loads of 2.7 bar and 4.6 bar IMEPg were chosen and are shown in Figure 34.

It can be seen that the simulated pressure traces are very close to the experimental data.

Although there is a very small discrepancy in the peak pressure, the agreement between the

model and experiments is very good, which provides confidence in the ability of the model

to accurately predict the thermodynamics of the engine cycle.

81

Figure 34: Cylinder pressure trace of metal case validation at loads of 2.7 and 4.6 bar IMEPg

Detailed numerical comparisons of the mass flow rate of air and IMEPg at different

loads are listed in Table 7. The errors used the absolute error, where positive error means

the model predicts a higher value than experimental data. It can be observed that the model

has a very close prediction on both IMEPg and mass flow rate of air. Compared with the

experimental data, the errors were fairly small and were consistently less than 0.5% for

most cases. The surface temperature of the metal piston is shown too, and it will be

compared with the coated pistons in the following section.

82

Table 7: Evaluation of critical validation metrics for the metal baseline cases

𝐼𝑀𝐸𝑃𝑔,𝑒𝑥𝑝 [bar] 2.34 2.73 3.00 3.40 3.72 4.00 4.28 4.60

𝐼𝑀𝐸𝑃𝑔,𝑠𝑖𝑚 [bar] 2.32 2.73 3.01 3.41 3.72 4.00 4.28 4.61

𝐼𝑀𝐸𝑃𝑔,𝐸𝑟𝑟𝑜𝑟 [%] -0.75 -0.03 0.35 0.17 0.09 0.11 -0.05 0.22

�̇�𝑎𝑖𝑟,𝑒𝑥𝑝 [g/cyc] 0.340 0.342 0.344 0.349 0.352 0.353 0.355 0.356

�̇�𝑎𝑖𝑟,𝑠𝑖𝑚 [g/cyc] 0.341 0.343 0.344 0.349 0.351 0.353 0.354 0.356

�̇�𝑎𝑖𝑟,𝐸𝑟𝑟𝑜𝑟 [%] 0.16 0.15 -0.08 -0.11 -0.34 0.11 -0.15 0.04

𝑇 𝑠𝑢𝑟𝑓𝑎𝑐𝑒,𝑎𝑣𝑔 [K] 416 419 421 422 423 424 425 427

4.3.2 Validation against the experimental TBC cases

With the metal baseline validation concluded, the examination of the transient heat transfer

model can be performed next by validating the model with the experimental TBC results.

Instead of only using the metal layer, the TBC layer is enabled in the 1D transient heat

conduction model with material properties and a thickness value that correspond to the

experimentally tested TBC. The 1D transient heat transfer model solves the surface’s and

sublayer’s temperatures and iteratively integrates with the thermodynamic cycle simulation.

The primary validation metrics are still cylinder pressure, air mass flow rate, and

IMEPg. The ideal validation of the transient heat transfer model would be to compare the

model predictions of the TBC surface temperature with experimental measurements.

However, due to the significant challenge associated with fast response surface temperature

measurements, especially on the surface of TBCs, a viable approach to evaluate the

model’s performance is to instead compare the metrics that would change as a consequence

83

of the high surface temperature of the TBC, using metrics that can be easily measured. First,

the higher surface temperature reduces heat transfer losses which cause higher bulk gas

temperatures, and eventually, higher cylinder pressure and IMEP; additionally, the higher

surface temperature heats the incoming intake charge (i.e., charge heating), which affects

the air mass flow rate. Thus, these three indicators (pressure, IMEPg, and air mass flow

rate) still constitute a reasonable assessment of the model’s performance for TBCs. Figure

35 shows the pressure traces for the 2.6 bar and 4.6 bar IMEPg cases. Note that the peak

pressure of the 4.6 bar case is slightly higher than that of the metal baseline, which is mostly

caused by the advanced combustion phasing. Apart from that, the simulated cylinder

pressure including the effects of TBCs is as accurate as the metal baseline.

Table 8 shows the errors between the measured and predicted IMEPg and air

mass flow rate, as well as the model’s prediction of the piston surface temperatures at

different loads. Similar to the metal case validation, the errors are less than 0.5% for all

TBC cases, which provides confidence in the model’s ability to capture the effects of the

TBCs in advanced combustion. Since the TBC surface temperatures include more

variation over the engine cycle, both the average and the peak temperatures are shown in

Table 8. Compared with the temperatures in Table 7, the temperatures of the coated

pistons are generally 40 to 60 K higher than the metal pistons. In addition to the tabulated

results, the gas temperature for the 4.6 bar IMEPg load case and the coating temperature

of the first 20 layers (25 microns per layer) are shown in Figure 36. It can be seen that the

surface temperature changes dynamically throughout the cycle, where it is lower during

the intake and reaches a peak during combustion. The temperature swing (i.e., the

84

difference between the lowest and the highest TBC surface temperatures) is about 50 K.

A more comprehensive investigation is given in the results and discussion section below.

Table 8: Evaluation of the critical validation metrics for the TBC cases

𝐼𝑀𝐸𝑃𝑔,𝑒𝑥𝑝 [bar] 1.94 2.59 2.94 3.43 3.95 4.62

𝐼𝑀𝐸𝑃𝑔,𝑠𝑖𝑚 [bar] 1.95 2.59 2.95 3.43 3.95 4.62

𝐼𝑀𝐸𝑃𝑔,𝐸𝑟𝑟𝑜𝑟 [%] 0.38 -0.03 0.46 0.07 -0.12 0.02

�̇�𝑎𝑖𝑟,𝑒𝑥𝑝 [g/cyc] 0.334 0.337 0.342 0.346 0.350 0.354

�̇�𝑎𝑖𝑟,𝑠𝑖𝑚 [g/cyc] 0.333 0.337 0.342 0.346 0.350 0.355

�̇�𝑎𝑖𝑟,𝐸𝑟𝑟𝑜𝑟 [%] -0.19 -0.01 -0.07 0.01 -0.05 0.19

𝑇 𝑠𝑢𝑟𝑓𝑎𝑐𝑒,𝑎𝑣𝑔 [K] 455 464 467 470 474 482

𝑇 𝑠𝑢𝑟𝑓𝑎𝑐𝑒,𝑚𝑎𝑥 [K] 486 499 502 505 510 519

Based on the good agreement between the experimental data and the simulation

results, it can be concluded that the model is capable of precisely capturing the working

fluid’s thermodynamic state and properties, as well as the coating’s temperature and its

impact on the engine performance. Therefore, a parametric study was conducted to

investigate the effects of TBC on LTC exhaust temperatures presented in the next sections.

85

Figure 35: Cylinder pressure trace for the 2mm TBC cases at loads of 2.6 bar and 4.6 bar IMEPg providing validation of the model’s ability to capture the thermodynamics as well as the

performance of the TBC

Figure 36: Coating node temperatures for the first 20 nodes with a Δx spacing of 25 microns between nodes (left) and bulk gas temperatures (right) over the engine cycle for the 4.6 bar

IMEPg load case

86

CHAPTER 5. THICK THERMAL BARRIER COATINGS FOR HCCI -

MODELING RESULTS AND DISCUSSIONS

5.1 Objective and simulation cases setup

There are three essential coating properties from both a material selection and coating

manufacturing perspective: the thermal conductivity (𝑘), the coating thickness, and the

volumetric heat capacity (𝑠), which is the combination of the density and the specific heat

(𝑠 = 𝜌 ∗ 𝑐𝑝).

To systematically evaluate each TBC property’s effects on HCCI, simulations were

conducted by sweeping each parameter individually while maintaining the other two as

constant. This individual property setup is to better understand how each property affects

coating performance, and the overall coating recommendation for HCCI will be provided

at the end of this chapter. The engine environment mimics the operating conditions from

previous experimental data, and six load conditions were adopted based on the cases shown

in Table 8. The swept parameters are shown in Table 9, where the thickness limit is the

thickness that results in impractically low intake temperatures, which will be discussed in

the thickness section. The range of values in the matrix for each material property was

selected based on a realistic range to survive an engine environment [90]. For example, the

thickness ranges from 0.2 mm to 6mm, where 0.2mm is a typical value for thin temperature

swing TBC application to provide sufficient insulation while avoiding charge heating, and

6mm is somewhat thicker than previously tested thick TBCs but offers significantly better

thermal insulation. The thermal conductivity was swept from 0.3 W/mK to 10 W/mK,

where 0.3 W/mK is the lowest possible value that state-of-the-art coatings can achieve by

87

either introducing structured porosity or natively low-k materials. On the high end of the

conductivity range, 10 W/mK is a value that provides insufficient thermal insulation and

starts to approach the value of steel (𝑘 of steel is ~40 W/mK). Typically, the volumetric

heat capacity for plasma-sprayed ceramic material would be in the range of 0.5 – 3 MJ/m3K

[90], where the lower end could be achieved by increasing the porosity. However, the

amount of porosity is limited by coating strength and reliability. Recent research has

proposed a novel metallic hollow microsphere structure that can significantly reduce the

volumetric heat capacity [100], but further development is needed to improve the durability

of this approach in the harsh engine environment. In this volumetric heat capacity sweep,

the lower bound is chosen as an ideal scenario of 0.2 MJ/m3K, which is slightly beyond the

range of realistic values. It is still useful to simulate these values because it provides an

assessment of the possible gains that could be achieved with the idealized thermophysical

properties to determine if it is worth pursuing a lower heat capacity approach. For

reference, the properties of aluminum are also provided in the last column of Table 9. The

general operating conditions are shown in Table 10.

Table 9: Investigated cases and sweep configurations

Sweeps Property

Thickness Sweep

Thermal Conductivity

Sweep

Volumetric Heat Capacity Sweep Aluminum

Thickness [mm] 0.2 - 6 0.2 Thickness Limit -

𝑘 [W/mK] 0.3 0.3 - 10 0.3 240

𝑠 [MJ/m3K] 0.97 0.97 0.2 – 5 2.61

88

Table 10: Operating conditions

Engine Speed [rpm] 1200

Fuel Type 87-AKI Gasoline

Fuel Delivery PFI

Injection Timing [deg aTDC] -120 (21 degrees after IVC)

Intake Pressure [bar] 1

Exhaust Pressure [bar] 1.1

Oil Temperature [K] 358

Coolant Temperature [K] 368

EGR [%] 0

Intake Temperature [K] Varied to match CA0

IMEPg [bar] 2.0; 2.5; 3.0; 3.5; 4.0; 4.5

Equivalence ratio 𝜙 0.19-0.42 (Varied to match load)

5.2 The effects of thermal conductivity - 𝒌

In this section, the TBC’s thermal conductivity, 𝑘 , was swept while maintaining the

thickness at 200 microns and the volumetric heat capacity as ~1 MJ/m3K, which is a typical

value for plasma sprayed TBCs. The coating covers the piston crown, engine head facing

the combustion chamber, and valves. This coverage provides the best insulation that could

be applied to a practical in-cylinder situation without impacting sealing and lubrication (by

coating the liner). Figure 37 shows the coating surface temperature (top) and the bulk gas

temperature (bottom) throughout the engine cycle.

89

It can be seen in Figure 37(a) that the TBC’s 𝑘 has a significant impact on the

surface temperature. As the thermal conductivity decreases, the overall average surface

temperature increases due to a reduction in heat transfer losses (shown explicitly in Figure

39 below). Additionally, the temperature swing, which is the temperature difference

between the peak and minimum surface temperatures, is significantly increased as 𝑘

decreases. This temperature swing effect is very important for the conventional combustion

modes because it maintains the wall temperature at a relatively low level during the intake

stroke, which prevents intake charge heating and loss of power density. The goal of this

work is to evaluate the impacts of TBCs on LTC using the coupled 0D thermodynamic

engine model and 1D heat conduction model. Therefore, Figure 37(b) shows the impact of

the TBC conductivity sweep on the bulk gas temperatures.

By examining the bulk gas temperature in Figure 37(b), it can be seen that despite

the change of coating surface temperature, the bulk gas temperatures at the same load

condition were very similar with a difference in the peak temperature of only ~25K. The

three zoom-ins in Figure 37(b) show the differences in bulk temperature at three different

times during the cycle: at IVC, just before ignition, and the peak temperatures. The

temperatures at IVC are lower with decreasing conductivity. This result is caused by two

factors. First, the lower conductivity coatings have elevated surface temperatures and a

larger temperature swing (shown in Figure 37(a)), which reduces heat transfer during the

compression stroke, and second, LTC must reach a given temperature threshold to achieve

autoignition. Therefore, due to the lower heat transfer losses during compression, the IVC

temperature is actually slightly lower with the TBCs than the metal piston. Based on this

90

lower IVC temperature, the charge density is actually slightly higher with the TBCs than

the metal piston. This trend is counter to the conventional combustion modes, and these

results indicate that rather causing a volumetric efficiency penalty, TBCs cause a slight

increase in charge density in LTC.

Figure 37: (a) Coating surface temperature and (b) bulk gas temperature vs. crank angle at a load of 3.5 bar IMEPg over the 𝑘 sweep

91

As a result of the increased charge density, the mass flow rate with the TBCs is

slightly higher, and the charge is slightly leaner than the metal piston. Due to the slight

reduction in equivalence ratio with the TBCs, the temperature to reach ignition must

increase slightly, shown in the top left zoom-in in Figure 37(b). An additional consequence

of the changing equivalence ratio is that the peak temperature is slightly lower after

combustion with the TBCs, shown in the top right zoom-in in Figure 37(b). Lowering the

conductivity amplifies these effects in comparison to the metal piston. This result is due to

the requirement of premixed autoignition-driven combustion in LTC, which results in

differing effects of TBCs on LTC compared to the conventional combustion modes.

Figure 38 shows the relationship between thermal conductivity and temperature

swing. The swing increases exponentially as 𝑘 decreases. These results show that the most

optimistic temperature swing based on the estimates of current realistic materials is ~100

K at this load and speed condition. It is important to note that the temperature swing is also

a function of load and speed because the engine conditions affect the heat flux into the

coating increases with load.

92

Figure 38: Temperature swing vs. thermal conductivity, 𝑘

As a result of the temperature swing and the elevated temperature during

compression, expansion, and exhaust, the heat transfer losses are reduced with TBCs.

Figure 39(a) shows the heat transfer losses decrease as 𝑘 decreases. The percent reduction

is higher at lower loads (2.4 percentage points at 2 bar versus 1.7 percentage points at 4.5

bar), which is because the heat transfer losses constitute a larger percent at lower loads. As

a result of the lower heat transfer losses, the gross indicated thermal efficiency increases,

as shown in Figure 39(b). The efficiency increase with lower conductivity is mostly caused

by the reduction in heat transfer losses; however, the lower equivalence ratio with the TBCs

due to the improved charge density to maintain the same load (𝜙 reduced from 0.35 to 0.34)

is also partially responsible for the improved efficiency through the thermodynamic 𝛾

effect. Overall, even the most optimistic 𝑘 values for realistic TBC materials only improves

the efficiency by ~1 percentage point.

93

Figure 39: (a) Heat transfer losses and (b) gross indicated thermal efficiency vs. thermal conductivity, 𝑘

The main objective of applying TBCs is increased efficiency. However, in LTC,

there are other anticipated benefits, including a reduction in the required intake

temperatures, which are currently a production challenge. Since the heat transfer losses are

reduced during the compression stroke, and there is some charge heating during the intake

stroke with the slightly higher surface temperature with TBCs, the intake temperature

requirement to achieve autoignition in LTC reduces with TBCs, as shown in Figure 40.

94

The trend of required intake temperature versus 𝑘 at different loads compared to the dashed

metal reference lines shows that the potential intake temperature reduction is modest (~5

K) and only occurs for the lowest conductivity coatings.

Figure 40: Intake temperature vs. thermal conductivity, 𝑘

By comparing the metal baseline with the lowest 𝑘 case, it can be inferred that

about 50% of the recovered heat transfer losses are converted to useful work, while the

remaining 50% shifted to exhaust losses. In addition to improved efficiency and reduced

intake temperatures, a hypothesized benefit of TBCs in LTC is increased exhaust

temperatures due to shifting from heat transfer losses to exhaust losses. Higher exhaust

temperatures and enthalpies would benefit aftertreatment and turbocharging, which are

current challenges in LTC. However, as shown in Figure 41, the bulk gas temperatures at

EVO are surprisingly consistent throughout the sweep, and this trend is independent of

load. This finding was also observed experimentally in the previous section, but was

95

previously unexplained. The coupled thermodynamic-TBC model helps to illuminate the

factors that cause this unintuitive phenomenon. With low conductivity TBCs, the intake

temperature is ~6K lower. Then, as shown in Figure 37(b) and discussed above, the

temperature differences before ignition are within 3 degrees, where the lowest 𝑘 case has

the highest ignition temperature because it is slightly leaner. However, the leanest mixture

also results in the lowest peak temperature after combustion (by ~25 K). Although the

lowest conductivity case starts the expansion process from the lowest temperature, the

temperatures at EVO are almost exactly constant due to the reduced heat transfer losses

during expansion and the slightly higher ratio of specific heats of the lowest conductivity

(leanest) case. Overall, this finding was unintuitive and due to the nature of LTC where the

kinetics-driven autoignition heavily relies on the mixture’s thermodynamic state.

Figure 41: Temperature at EVO vs. thermal conductivity, 𝑘

From a tailpipe perspective, this uniform EVO temperature does not necessarily

mean that the exhaust enthalpies are identical. Due to improved volumetric efficiency, the

96

exhaust flow enthalpy increased by 0.8%. The simplified calculation for the relative change

in enthalpy at EVO is shown in Equation 21, where the subscript i represents the simulated

case, and mb represents the metal baseline. There is no pressure term because the

backpressures are constant, and the pressures at EVO are nearly identical.

𝛿𝐻𝐸𝑉𝑂 = (𝑚𝑖,𝑐𝑦𝑐𝑇𝑖,𝐸𝑉𝑂

𝑚𝑚𝑏,𝑐𝑦𝑐𝑇𝑚𝑏,𝐸𝑉𝑂− 1) ∗ 100% (21)

From a cycle energy distribution perspective, the energy portion of exhaust waste

is higher with the TBCs due to the higher temperature difference (∆𝑇) between the intake

and exhaust. In other words, the enthalpy delivered into the system is lower, while the

rejected enthalpy is higher.

5.3 The effect of TBC thickness at low-𝒌

The previous section only discussed the effects of 𝑘, based on a very thin layer. Since the

thermal conductivity is on a unit length basis, increasing the coating thickness would

amplify the insulating effects of low-𝑘 TBCs. This section investigates the impact of TBC

thickness with the lowest 𝑘 value from the previous section. The thickness was varied from

0.2 mm to 6 mm while holding 𝑘 at 0.3 W/m-K and volumetric heat capacity as ~1 MJ/m3K.

Similarly to the previous section, the coating only covers the piston crown, head, and valves,

but leaves the liner uncoated. Since the previous section found that the impact of TBCs is

similar at different load conditions, a representative load of 3.5 bar is selected for analysis

for most of the rest of this study.

Figure 42 is similar to Figure 37 where the surface temperatures are shown in the

top plot, and the bulk gas temperature is shown in the bottom plot. The surface temperatures

97

in Figure 42(a) are entirely shifted up with the increased coating thickness, and this is

distinct from lowering conductivity, where the temperature mostly swings up during

combustion but remains relatively close to the metal baseline during the intake stroke. Also,

the magnitude of the swing decreases somewhat with increasing thickness due to the

reduction in heat flux.

Due to the elevated surface temperatures, the thick TBCs have substantial impacts

on the bulk gas temperature. First, the thickest TBCs require the lowest IVC temperature

because they have the lowest heat transfer during compression. Therefore, they can start

from a lower temperature at IVC and still reach their autoignition threshold. Due to their

lower IVC temperature (which implies higher charge density and lower equivalence ratio),

the thick TBC requires the highest temperature to start autoignition to compensate for their

lower equivalence ratio. The peak bulk temperatures have a difference of more than 100 K

due to the dilution effect.

The intake temperature requirement to achieve autoignition is shown in Figure 43.

It is seen that the reduction in intake temperature is significant as thickness increases due

to both charge heating during the intake stroke and less heat transfer during the

compression stroke. Note that although “charge heating” is occurring, it is only used to

help lower the intake temperature requirement. Since the IVC temperature is lower with

the thick TBCs, the charge heating that occurs in LTC does not reduce the density of the

incoming charge compared to the uncoated metal baseline (like it does for the conventional

combustion modes).

98

At the load of 4.5 bar, the intake temperature falls below 320 K, which is an

approximate intake temperature that would occur in a production application. Since further

cooling the intake temperature is impractical, the thickness where the intake temperature

reaches 320 K is defined as the thickness limit, and this limit may change with different

fuels, engine specifications, and coating properties. Theoretically, this thickness ceiling

would be higher with fuels that have higher autoignition resistance (such as natural gas or

high octane gasoline) or high cooling potential (such as methanol or wet ethanol [97][98]).

In other words, these fuels would benefit more from the application of thick TBCs.

99

Figure 42: (a) Coating surface temperature and (b) bulk gas temperature vs. crank angle at the load of 3.5 bar IMEPg through thickness sweep

As a result of this significantly reduced intake temperature, the volumetric

efficiency is improved by 7.6% to 10% (5 to 7 percentage points) before reaching the

thickness limit. Hence, the mixture is much leaner while maintaining the same load (at the

100

load of 4.5 bar, 𝜙 decreased from 0.42 to 0.35). Plots of volumetric efficiency and

equivalence ratio are shown in the Appendix, Figure 86.

Figure 43: Intake temperature and 𝜙 vs. coating thickness

The heat transfer losses and thermal efficiency are shown in Figure 44, where the

plot is truncated at the thickness limit. With this low conductivity material, by increasing

the thickness from 0.2 mm to 3.2 mm, the heat transfer losses further reduce from 28.5%

to 21% (a 7.5 percentage point reduction). The efficiency increase is 3.4 percentage points,

which is slightly less than half of the recovered heat transfer losses. The thermal efficiency

benefited from a combined effect of lower heat transfer losses due to the coatings, and a

more favorable 𝛾 associated with the leaner mixture (0.6 percentage points can be

attributed to the leaner mixture alone). By subtracting the efficiency gain that can be

attributed to the leaner mixture, the results indicate that only about one-third of the

recovered heat transfer losses (2.8/7.5) were converted into work, while the remaining two-

101

thirds shifted from heat transfer losses to exhaust losses. The previous 𝑘 section showed a

work conversion rate of nearly 50%. This discrepancy is due to that the fact that the low-𝑘

conductivity coating’s mechanism to reduce heat transfer was primarily through

temperature swing which reduces heat losses from the highest temperature portion of the

cycle around the combustion process. Minimizing heat transfer around combustion leaves

most of the expansion stroke to convert the saved energy into work. However, the thicker

coating’s mechanism to reduce heat transfer is by elevating the surface temperature

throughout the entire cycle. It is more difficult to convert any saved heat transfer losses

from the early compression stroke or late expansion stroke into useful work because the

full expansion stroke cannot be leveraged for work extraction. Other researchers have also

documented similar trends [8][99]. In other words, the most efficient way of transforming

the recovered heat into work is to prevent heat losses around TDC. Otherwise, the saved

heat transfer losses contribute more to exhaust enthalpy. However, in LTC, reducing the

heat losses in favor of increasing exhaust enthalpy is also desirable because greater exhaust

enthalpies will aid aftertreatment and turbocharging, which are current challenges for LTC.

102

Figure 44: Heat transfer losses (red, left axis) & thermal efficiency (blue, right axis) vs. coating thickness at the load of 3.5 bar IMEP

5.4 The effect of volumetric heat capacity - 𝒔, with the thick, low-𝒌 TBC

Since it was shown that increased thickness provides significant efficiency benefits, the

coating thickness for the following results is chosen as the limit discussed in the previous

section (~3.2 mm), and the thermal conductivity was kept constant at 0.2 W/mK.

Figure 45(a) shows the surface temperature with the change of 𝑠 . The most

substantial effect of low volumetric heat capacity is that it increases the temperature swing

due to the reduced ability to store heat. In other words, it enables the surface temperature

to follow the bulk gas temperature more closely. It is seen that the temperature swing has

a very minimal effect on bulk gas temperature in Figure 45(b), where the ignition

temperatures are almost identical, and the difference at the peak is less than 5 K.

Interestingly, the peak temperature actually increases with lower 𝑠 due to a richer mixture,

which is contrary to the results in 𝑘 and thickness sweep. This reason for this opposite trend

is described in the following paragraph.

103

Comparing the two coating surface temperatures in Figure 45(b), the high swing

coating (red dashed line) has a consistently lower surface temperature through the intake

and compression stroke, which results in less charge heating during the intake stroke and

more heat transfer losses during the compression stroke. This is desirable in the

conventional combustion modes to minimize charge heating, improve volumetric

efficiency, and increase power density. However, reducing the charge heating and

increasing the compression stroke heat transfer are not desirable in LTC because the

ignition threshold must be met and by reducing charge heating and increasing compression

stroke heat transfer losses, a higher intake temperature is required to achieve autoignition

(7.4 K variation over the sweep). This results in a lower volumetric efficiency and a less

favorable ratio of specific heats (i.e., a richer mixture).

Despite the intake and compression stroke drawbacks of the higher temperature

swing coating, the combustion and expansion stroke heat transfer is reduced, where the

recovered heat can be converted into work most effectively. The overall thermal efficiency

increased by 0.2 percentage points. This efficiency gain is small compared to the increased

thickness approach, which is because the reduced heat transfer losses and more favorable

mixture properties worked together to increase thermal efficiency when the conductivity

and thickness were swept; however, when the volumetric heat capacity is swept, the

mixture properties are partially counteracting the benefit of reduced heat transfer losses.

From the thickness section, Figure 44 was truncated at the thickness limit, which

was imposed by considering a realistic lower limit on the intake temperatures that would

be possible in a production setting. If not for this intake temperature-imposed thickness

104

limit, thicker TBCs could be employed to further improve the thermal efficiency.

Coincidentally, the temperature swing not only increases efficiency but it also increases

the intake temperature requirement by 7.4K. Therefore, lower volumetric heat capacity

TBCs can be used to raise the thickness limit slightly and further improve efficiency. With

this in mind, an optimization routine is performed to determine the maximum potential

efficiency improvement.

105

Figure 45: (a) Coating surface temperature and (b) bulk gas temperature vs. crank angle at the load condition of 3.5 bar IMEPg over the volumetric heat capacity sweep

Figure 46 shows the optimization routine and the engine performance

characteristics in terms of (a) efficiency, (b) intake temperature, (c) cumulative heat

transfer by separating the charge heating effect (on the negative side) and the heat transfer

losses (on the positive side), and (d) coating surface temperatures. Initially, a very thin

106

coating is applied where minimizing 𝑘 (in blue) provides the most insulation. During that

process, the thermal efficiency increased by 0.8% with no tangible sign of charge heating.

The reduced heat transfer losses during the compression stroke were responsible for the

decreased intake temperature. Then, using that low-𝑘 TBC and increasing the thickness

shifted the surface temperature higher with a slightly reduced temperature swing (in red).

While thickness increases, the intake temperature at the highest load condition reaches the

lowest practical temperature for a production engine setting (320K), where the

corresponding thickness is defined as the thickness limit. Both intake charge heating and

less heat transfer during the compression stroke contribute to the lower intake temperature

requirement. As a result, the efficiency growth is truncated at the thickness limit, after

reaching an improvement of 4.1 percentage points compared to the metal baseline (3.3

percentage points compared to the thin, low-𝑘 TBC). At that thickness, reducing the

volumetric heat capacity (in purple) to a substantially lower value of 0.2 J/m3K increases

the temperature swing and leads to an additional 0.2 percentage point improvement to

efficiency. However, the intake temperature is increased due to the lower surface

temperature during the intake and compression stroke, which unleashes more room to

increase the thickness. Finally, by lowering volumetric heat capacity and thickening the

coating layer together (in green), the thermal efficiency reaches a 4.6 percentage points

improvement compared to the metal baseline, which is a 10% improvement on a relative

basis compared to the metal baseline. Figure 87 in the Appendix shows the sweep in order

of k, s, and then thickness. It can be seen that the efficiency gain from varying s has been

107

increased compared to the trends in Figure 46, but increase thickness with low k and low s

material still shows the most potential for efficiency improvement.

Table 11 shows the other performance measures at the load level of 4.5 bar. This

load was selected instead of 3.5 bar because the temperature limit occurs at 4.5 bar. Note

that although the EVO temperature is lower than the metal baseline, the exhaust flow

enthalpy is 5.7% higher due to higher volumetric efficiency with the TBC. This

improvement is smaller than that of the conventional combustion modes, where the intake

temperature is held constant to minimize loss of power density. However, due to the fact

that the intake temperature is 60 K lower, it is still surprising that the exhaust enthalpy with

the TBC is higher.

Overall, it can be concluded that most benefits came from the increased coating

thickness with low thermal conductivity. The low volumetric heat capacity is critical in

conventional combustion and CAI modes, but not as crucial in pure-HCCI combustion

modes.

Table 11: Engine performance at load of 4.5 bar IMEPg

Optimal configuration Metal baseline

𝜂𝑖𝑔,𝑡ℎ [%] 48.79 44.22

𝜂𝑣𝑜𝑙𝑢𝑚. [%] 78.35 70.97

𝑇𝑖𝑛𝑡 [K] 320 379

𝑇𝑒𝑥ℎ [K] 765 799

𝛿𝐻𝐸𝑉𝑂 [%] 105.7 100

𝜙 0.35 0.43

108

Another interesting finding is that the thermal efficiency in LTC is highly related

to the maximum surface temperature, as shown in Figure 47. Increasing thickness has the

steepest slope. This relationship reveals the importance of elevated surface temperature on

LTC efficiency; the most effective way to increase the surface temperature is to use thick

TBCs with a low-𝑘 material.

Finally, due to the 0D/1D nature of this model, combustion efficiency effects were

not studied. However, the previous experimental work found that the benefits to

combustion efficiency with thick TBCs in pure-HCCI were at least as significant as the

benefits to thermal efficiency. It is anticipated that elevated surface temperatures would

also significantly improve combustion efficiency, and those benefits would be additive

with the thermal efficiency improvements.

109

Figure 46: optimization routine and engine performance

Figure 47: Gross indicated thermal efficiency vs. maximum surface temperature

110

CHAPTER 6. GASOLINE COMPRESSION IGNITION TEST CELL

COMMISSIONING

The study of thick thermal barrier coatings for HCCI established a fundamental

understanding of how various coating properties affect LTC engine performances.

Additionally, it revealed that the optimal coating thickness for LTC could be distinct from

the one for conventional combustion modes such as SI and CDC. However, HCCI is still

impractical in terms of commercialization due to the lack of control over the combustion,

which necessitates research focused on the second generation LTC modes.

The following sections will focus on gasoline compression ignition (GCI) as a stratified

LTC mode that holds promise of nearer-term commercialization due to its improved

controllability and load range compared to HCCI. To gain a better understanding of the

GCI family of combustion concepts, first, gasoline partial fuel stratification (one of GCI

branches) will be experimentally investigated, where different injection strategies will be

employed and evaluated. Then, the thesis continues with exploring the optimal TBC

properties for GCI by quantitively determining the benefits and tradeoffs using the 0D/1D

model that was previous developed for investigating the coupling between TBCs and

thermodynamics.

6.1 Single-cylinder light-duty GCI engine commissioning

A GCI engine test cell was designed and commissioned in-house to provide a state-of-the-

art engine platform to operate GCI combustion at Clemson University.

111

Figure 48: Clemson University single-cylinder light-duty GCI test cell overview

The engine head was modified from a GM four-cylinder 1.6L production diesel

engine. Only the second cylinder was activated and coupled with an FEV single-cylinder

engine block. The specific engine geometry and specifications are listed in Table 12. There

are two bore options for this engine (79.7 mm and 82.0 mm), and the different bore can be

achieved by swapping the custom-made liner. The engine is equipped with a production

re-entrant bowl piston that is typically used for conventional diesel combustion. Recent

studies have shown that this type of bowl geometry could be favored for GCI high load

operation [103].

The engine is equipped with two fuel injection systems, a high-pressure fuel system

that was supplied by a Bosch CP3 pump (belt-driven from the crankshaft with a gear ratio

of 1:2), and a port fuel injection system to create the nearly homogeneous mixture at the

intake port.

112

Table 12: Engine parameters

Displacement 400 [cc]

Bore 79.7 / 82.0 [mm]

Stroke 80.1 [mm]

Number of Valves 4

IVO [deg ATDC] -360

IVC [deg ATDC] -160

EVO [deg ATDC] 145

EVC [deg ATDC] -345

Compression Ratio 16 - 17

Maximum speed 4000 [rpm]

Injection type DI & PFI

DI Injection pump Bosch CP3

Injector Denso DCRI300770 7-hole

The schematic of the high-pressure direct injection system is shown in Figure 49 –

top. The low-pressure loop has two main objectives. First, it serves as the feeder to the

high-pressure CP3 pump at constant pressure (~7 psi); second, it measures the fuel flow

rate by a MicroMotion Coriolis flowmeter. The check valve at the junction of low-pressure

and high-pressure loop prevents backflow to ensure accurate fuel flow measurement. The

two heat exchangers (as shown in Figure 49 - right) in the high-pressure loop are critical

because gasoline has a much lower boiling point than diesel fuel such that any heat from

the pump, metering unit, or pressure control valve could vaporize the gasoline, which could

potentially damage the pump or cause inaccurate fuel flow measurements. Fuel pressure is

managed by controlling the metering unit and pressure control valve (at the end of the high-

113

pressure rail). The rail pressure was measured by two Kistler 4067C3000S pressure

transducers at a distance of 50 mm and 300 mm from the rail to the injector.

Figure 49: Top: Layout of high-pressure direct injection fuel system; Left: Low-pressure loop cart; Bottom right: Heat exchangers for fuel return cooling.

The PFI system is designed in a similar manner but with much less complexity due to

the absence of the high-pressure loop. The PFI flow rate is also measured by another

MicroMotion Coriolis flow sensor, and the injection pressure was rated at 40 psi to the

intake port.

The intake air is boosted by the building compressor (up to 65 psi) and then throttled

by an Alicat flow controller to the desired boost level for engine intake. The air flow rate

to the engine is measured by a Fox Thermal FT2a flow meter and a second time by the

114

Alicat mass flow controller. There is a 3-kW intake heater (PID controlled) upstream of

the intake plenum to condition the intake temperature. Both intake temperature and

pressure were measured at the intake port via K-type thermocouple and a Kistler

4045A5V200S pressure transducer (high-speed), respectively. The intake port has a swirl

valve with six valve positions.

The cylinder pressure is measured by Kistler 6125C cylinder pressure transducer. The

glow-plug hole in the engine head was modified for cylinder pressure transducer

installation. Figure 91 (in Appendix B) shows the detailed dimensions for the in-house

designed and made cylinder pressure transducer mounting adapter. The Kistler type

2614A1 encoder mechanism and a 2614A2 signal conditioner measure the crank angle to

interpret piston’s position and combustion chamber’s volume. The high-speed

measurements, such as cylinder, intake, exhaust, and rail pressures, were triggered by crank

angle measurement at a resolution of 0.1 crank angle degrees.

Five types of emissions (UHC, CO, CO2, O2 NOx) were sampled at the exhaust

plenum and measured by Horiba Mexa 7100D-EGR analyzer bench. The same bench also

samples the intake charge to calculate the EGR rate based on CO2 fraction. The AFR and

a redundant EGR measurement are implemented by two sets of ECM oxygen sensors with

pressure compensation. A detailed bus series connection and sensor instrumentations for

ECM CANp modules and sensors are shown in Figure 50.

115

Figure 50: ECM AFR and EGR modules in series bus connection

A custom LabVIEW program monitors engine operating conditions such as cylinder

pressure, breathing conditions (intake/exhaust pressures and temperatures), fuel system

pressures, oil and coolant thermal management, etc. Several combustion performance

metrics are also real-time processed and monitored, such as processed heat release rate,

ringing intensity, efficiencies, etc. The program also provides real-time engine controls of

boosting pressure, EGR and exhaust valve positions, injection pressure, the number of

injections and timings, etc. The data is saved to local disks with 300 consecutive cycles

while reaching steady states.

At the time of this dissertation defense, this single cylinder light-duty GCI engine has

been motored and fired and final shakedown testing and debugging is ongoing. The GCI

project funded by Aramco Services Company that motivated this test cell commissioning

will continue for another ~2 years beyond the end of this dissertation and will be used for

the future experimental engine testing under that research contract.

116

CHAPTER 7. INVESTIGATION INTO THE INJECTION STRATEGIES FOR

GASOLINE PFS – EXPERIMENTAL RESULTS & DISCUSSION

7.1 Objective and Experimental operating conditions for GCI investigation

The experiments presented in this section were conducted on the single cylinder Ricard

Hydra engine described in Chapter 2. The objective was to develop an understanding of

the GCI combustion family by experimenting with the injection characteristics of PFS

combustion and its effects on burn rate, efficiencies, and emissions.

The engine was operated at a constant speed of 1200 rpm. The total fuel flow rate

for PFI and DI (both gasoline) was held fixed and targeted at 16 mg per cycle regardless

of the split fraction. Subsequently, the charge-mass equivalence ratio φ’ is kept constant at

0.35. Other engine operating conditions such as split fraction, intake temperature, the start

of injection (SOI) timing for single/double late injection, etc., are listed in Table 13.

Since EGR is applied, the charge mass equivalence ratio is used to identify the

overall cylinder fuel richness. Unlike the most common definition of equivalence ratio φ,

which was only based on the dilution of air, the charge mass equivalence ratio, sometimes

denoted by φ’, defines the overall dilution level in the cylinder by air and/or EGR. The

mathematical definition for φ’ is shown in the equation below:

𝜙′ = 𝜙 ∗ (1 −𝐸𝐺𝑅 + 𝑅𝐺𝐹

100 ) (22)

All data are saved after combustion reaches steady states and follows the saving

criterion that was mentioned in the previous sections.

117

Table 13: Engine operating conditions in PFS study

Engine Speed [rpm] 1200 PFI Fuel 87-AKI Gasoline DI Fuel 87-AKI Gasoline

PFI SOI Timing [deg aTDC] -120 PFI Pressure [psi] 28

DI ‘Single’ Late Injection Timing [deg aTDC] -140; -110; -80; -60; -50; -45; -40

DI ‘Double’ Late Injection Timing [deg aTDC]

[-70, -50]; [-80, -40]; [-90, -30]; ~ -60

[-60, -40]; [-70, -30]; [-80, -20]; ~ -50

[-50, -30]; [-60, -20]; ~ -40

DI Pressure [bar] 550 Split Fraction 90%; 80%; 70%

Total Fuel Rate [mg/cycle] 16 Charge-Mass Equivalence Ratio (φ’) 0.35

Intake Pressure [bar] 1.6 Intake Temperature [K] 340 Coolant Temperature [k] 373

Oil Temperature [K] 363 Combustion Phasing- CA50 [deg aTDC] 8.6

7.2 The effects of SOI timing on PFS combustion characteristics, efficiencies, and

emissions

In this section, the experiments are conducted while holding the amount of fuel flow per

cycle constant at 16mg/cycle. To isolate the effects of combustion phasing on the

combustion process and performance metrics, CA50 is targeted at 8.6 degrees aTDC for

all cases, which is controlled by varying the externally cooled EGR percentage into the

intake plenum. The intake pressure is boosted to a 1.6 bar. The intake temperature is

maintained at 340 K. The injection timing window is chosen from -140 to -40 deg aTDC

118

with a diminishing spacing between each SOI timing. The reason for this diminishing

arrangement is that the later injection event creates more fuel stratification in the cylinder

and the combustion process becomes increasingly sensitive. Other operating conditions,

including the injection strategies, are shown in Table 13. And the idea of injection strategy

is shown in

Figure 51: Injection strategy for single late injection (SLI) with SF70

Figure 52 only shows the heat release rate as a sweep of SOI timing at a split

fraction of 70%. The pressure traces are not shown in this figure to provide a clear

visualization of the heat release rate (HRR), but the combined HRR-pressure trace plot can

be found in the Appendix (Figure 88). From Figure 52, it can be seen that the earliest four

injection timing cases overlap with each other, indicating that delaying SOI from -140 to -

60 degrees aTDC does not result in a significant change to the combustion process.

However, delaying the injection timing further starts to have a more noticeable effect on

combustion in the window of -60 to -40 degrees aTDC.

119

Figure 52: GHRR for an SOI sweep at a split fraction of 70% (SF70, i.e., 70% of the fuel mass was port fuel injected and 30% was direct injected at the timing indicated in the legend)

With the delay of SOI timing, the peak heat release decreases and the heat release

process is elongated – this knock-reducing trend will be quantified in more detail in Figure

53. This trend is due to the in-cylinder φ stratification that is created by the later

compression stroke injection event. As the SOI timing retards, the time for air and fuel

mixing before autoignition is shorter, and this creates more fuel stratification in the

cylinder. As the cylinder temperature increases from compression, the rich regions auto-

ignite first and release their heat, which increases the cylinder pressure and temperature

further, contributing to the subsequent autoignition of progressively leaner regions.

The combustion characteristics and efficiencies are shown in Figure 53. Overall,

the load was kept at a relatively constant level. For the earlier injection timings, the SOI

120

timing has almost no effects on the peak pressure rise rate (PPRR); however, continuing to

delay SOI timing leads to a decrease in the PPRR starting at -60 degrees aTDC. This

indicates that at this boost level and this type of fuel, the φ-sensitivity is relatively weak,

and it is difficult to provide considerable control over the combustion process until a

relatively late injection event is used. Before -60 deg aTDC, the air and fuel are too

premixed to provide enough φ stratification. For SOI timing of -60 to -40, the PPRR

decreased from 7 to 5.8 bar/CAD, and the burn duration is prolonged from 7.5 to 8.1 CAD.

These outcomes agree with the heat release features in Figure 52. The combustion

efficiency, ηc, does not change significantly as the SOI timing is varied, although there is

a slightly decreasing trend for the latest injection timings. This is most likely due to the

incomplete combustion of either locally low-temperature regions or locally rich regions;

this trend agrees with the CO emissions shown in Figure 54. Overall, this φ stratification

level is not strong enough to deteriorate combustion, and the combustion efficiency was

maintained at around 95% through all the experiments.

It can be seen in Figure 53 that, in this SOI timing sweep, the gross indicated

thermal efficiency is decreased by 1.5 percentage points from 45% to 43.5%. There are

three possible explanations for this decrease. First, the burn duration is extended and that

will reduce the effective expansion work. Second, another possible consideration is the

increased portion of EGR from about 47% to 50%, which changes the properties of the

mixture and decreases the ratio of specific heats, γ. The reason for the EGR increase will

be discussed in the following paragraph in this section. The following equation shows the

ideal Otto cycle’s thermodynamic efficiency. Here CR is the geometric compression ratio.

121

𝜂𝑡ℎ = 1 −1

𝐶𝑅(𝛾−1) (23)

Therefore, the change in the mixture’s ratio of specific heats could be another

reason that thermodynamically supports the efficiency decrease which was also mentioned

in a previous EGR-focused study from Olsson et al. [104]. The third factor in the 1.5

percentage point decrease in ηth,ig as the injection timing is retarded is that the evaporation

of the fuel absorbs heat from the thermodynamic cycle, which, depending on when that

heat absorption occurs, can be beneficial or detrimental to thermal efficiency. The

hypothesis here is that as the injection occurs later in the compression stroke, there is a

larger penalty to thermal efficiency due to the heat absorption of the evaporating liquid,

which is likely that this thermodynamic factor is playing a role in the decrease in thermal

efficiency with retarded SOI timing. Yang et al. have shown a similar decreasing trend in

ηth,ig with delayed SOI timings [105]. Wissink et al. have also shown this trend from a

purely thermodynamic analysis [106].

122

Figure 53: Combustion characteristics and efficiencies vs. injection timing at SF70

The emissions data are shown in Figure 54. On the top subplot, it can be observed

that the EGR percentage requirement to match combustion phasing increases with the delay

of SOI timing. The shape of the trend is similar to the other combustion metrics in Figure

53, i.e., slowly elevates for the earlier injection timings, but become much steeper for SOI

timings after -60 deg aTDC. This increase in EGR percentage is related to the increased

fuel stratification. Compared with a fully premixed charge, the less well-mixed charge has

rich and lean regions in the combustion chamber. The rich regions will autoignite earlier

123

and release their heat, which results in an earlier combustion phasing (CA50). Since CA50

was kept constant using EGR percentage in this study, more EGR was needed to retard the

combustion phasing back to the target value of 8.6 deg aTDC.

The UHC emissions are relatively constant, matching the trend in the combustion

efficiency. The CO emissions are constant for the early SOI timings, but start to increase

linearly after -60 deg aTDC. This is potentially due to CO chemistry freezing as the burn

duration increases because CO to CO2 conversion strongly depends on the OH radicals,

which are insufficient at a lower temperature. After -60 deg aTDC, the reduced heat release

rate and slightly higher EGR fractions result in a lower bulk cylinder temperature; thus, the

CO emissions increase. The NOx emissions have a similar trend as CO emissions, but for

a very different reason. The NOx formation is closely coupled with high temperature due

to the dissociation of O2 and N2. Hence, the increase of NOx emissions is related to the

locally high-temperature regions, which result from the combustion of the locally near-

stoichiometric regions. There are a larger number of these regions as the SOI timing retards

beyond -60 deg aTDC, which causes the NOx emissions to increase.

124

Figure 54: Emissions vs. injection timing at a split fraction of 70

7.3 The effects of Split Fraction (SF) on PFS combustion characteristics, efficiencies,

and emissions

The effects of SOI timing on combustion performance and emissions have been

systematically studied and discussed above. For the purpose of having a better

understanding of the level of in-cylinder fuel stratification, as well as the φ-sensitivity of

87-AKI gasoline under a relatively low boosted condition, the combined effects of split

fraction and SOI timing will be discussed in this section. Here ‘split fraction’ is the ratio

of port injected fuel mass flow rate to the total fuel flow rate, which includes both PFI and

DI. For example, SF100 means that all of the fuel has been delivered by PFI, which could

125

be considered as HCCI, and SF70 means that 70% of the fuel has been injected into the

intake manifold as fully premixed charge and 30% of fuel has been directly injected into

the cylinder during the compression stroke. For consistency with the injection timing study

described above, the operating conditions for the split fraction study are identical to the

conditions listed in Table 13.

Figure 55 shows the heat release rate and pressure trace for the different split

fractions with the same SOI timing of -50 deg aTDC. A considerable reduction in the peak

heat release rate can be observed. This indicates that increasing the portion of directly

injected fuel increases the level of in-cylinder φ stratification. A higher fraction of directly

injected fuel mass will result in locally richer regions that autoignite earlier. This also

explains the EGR increase as split fraction decreases shown in Figure 57, because more

EGR is needed for those earlier autoigniting regions (due to the higher fuel stratification

level) to maintain constant combustion phasing. Thus, as shown in Figure 56, the burn rate

decreases, which lowers the pressure rise rate and the duration of heat release increases.

By close examination of Figure 55, it can also be seen that the peak pressure before

combustion generally decreases with a lower split fraction. This is due to the effect of

evaporative cooling of the fuel during the compression stroke. As more fuel is injected

during the compression stroke, more heat is absorbed from the cycle, and the pressure is

reduced. This phenomenon is responsible for the thermodynamic efficiency penalty

associated with compression stroke injections described above in the discussion

surrounding the efficiency decrease in Figure 53.

126

Figure 55: GHRR & pressure trace for different split fractions at an SOI timing of -50 deg aTDC

Figure 56 shows the IMEPg, PPRR, burn duration, combustion efficiency, and

gross thermal efficiency for the operating cases shown in Figure 55. Intuition might suggest

that the combustion efficiency would decrease as the split fraction decreases due to over-

rich regions with insufficient oxygen, longer burn durations that can potentially freeze late-

cycle CO, and the slightly increasing EGR fraction to maintain CA50 (Figure 57).

However, Figure 56 shows that the combustion efficiency is approximately constant, with

a slight increase as the split fraction is reduced. It is speculated that this indicates that the

unreacted premixed fuel in the crevices is responsible for the combustion inefficiency.

Therefore, since the total fuel in the cylinder is constant, as the split fraction decreases, the

background equivalence ratio decreases, which results in less unburned fuel being trapped

in the crevices at this SOI timing. This explains the slight decrease in UHC emissions in

Figure 57.

127

Figure 56: Combustion characteristics and efficiencies vs. split fraction at an SOI = -50 deg aTDC

The gross indicated thermal efficiency decreases from 46% to 44% as the split

fraction decreases. Both Yang et al. [105] and Dernotte et al. [67] reported similar

decreasing trends with an increase in the DI fuel portion. This decrease is most likely

caused by the same three factors as discussed above: 1) the longer burn duration, 2) the

higher required EGR fraction that changes the mixture properties, and 3) the increase in

the energy absorbed during the compression stroke due to the evaporative cooling of the

128

spray. Since the combustion efficiency only increases slightly, the overall fuel conversion

efficiency still has a decreasing trend as the DI portion increases.

The NOx emissions increase as the split fraction decreases due to existing of more

rich and near-stoichiometric regions. However, the NOx emissions are formed only above

a certain temperature threshold, and it was maintained at a very low range throughout all

of the experiments. This is presumably due to the high usage of EGR. As shown in Figure

57, the EGR percentage for all points is above 45%, and that is the primary reason for the

ultra-low NOx emissions in addition to the boost level. The boosting and EGR rate results

in a charge-mass equivalence ratio of 0.35, and consequently, near-zero engine-out NOx

levels despite the load being about a 6.5 bar.

Figure 57: Emissions vs. split fraction at an SOI = -50 deg aTDC

129

7.4 The combined effects of SOI timings and split fraction

Figure 58 shows the change of emissions, combustion characteristics, and efficiencies

versus various SOI timings and different split fractions which are labeled by three distinct

colors (red, yellow, and blue represent split fractions of 90%, 80%, and 70%, respectively).

The black dashed line represents the results of the fully premixed HCCI operation baseline.

The HCCI baseline had the same fueling rate and combustion phasing as the other cases.

In this section, the combined overall effects of SOI timing and the split fraction will be

studied. The benefits and drawbacks of a realistic boost condition will also be discussed.

In order to make a clear visualization, the transparent blue box in both figures is the data

set for the previous split fraction study.

There is an interesting trend for UHC emissions in Figure 58. As the injection

timing retarded from -140 to -40 degrees aTDC, the UHC emissions have an overall slight

decrease for the early injection timings, followed by an increase at the later injection

timings. The first decrease is most likely due to the injection angle of 60 degrees. As shown

in Figure 59, at an SOI timing of -140 deg aTDC, the spray is approximately aimed into

the cervices between the piston and liner. As the SOI timing is delayed, the spray is targeted

toward the top of the piston. Thus, the UHC emissions decrease at the beginning of the

sweep. Further delaying the SOI timing causes the UHC emissions to increase, which could

be due to the spray and wall impingement or higher fuel stratification resulting in rich

regions in the cylinder. A future CFD simulation can shed light on the level of stratification,

spray-wall impingement, and the sources of UHC and CO emissions. Even though NOx

130

emissions increase with later injection timings, they still remained remarkably low due to

the high EGR rate.

Figure 58: (Left) Combustion characteristics and efficiencies; (Right) Emissions vs. start of injection timing at different split fractions

Figure 59: Spray at -140, -110, -80, and -50 deg aTDC

Compared with the HCCI baseline, PFS with a single late injection (SLI) strategy

was able to reduce the peak pressure rise rate by up to 31% (from 8.5 to 5.85 bar/CAD) at

the lowest split fraction of 70% and the latest injection timing at -40 deg aTDC. It can be

131

observed that the gap between the three trend lines is smaller between 80% and 70% than

the gap between 90% and 80%. This indicates that the effects of the split fraction are

diminishing. Further increasing split fraction would cause increased UHC and CO

emissions and deteriorated combustion efficiency due to the inhomogeneities. This agrees

well with the results by Yang et al. showed that the optimal DI portion for a higher boost

level is around 15% [105]. Further delaying SOI timing would raise the UHC and CO

emissions again due to several possible factors; 1) the spray-wall impingement contributes

to the UHC emissions increase, 2) the lower heat release rate leads to a lower cylinder bulk

temperature which causes an increase of CO emissions, and 3) the higher fuel stratification

level might create some over-rich regions that do not burn completely. The NOx emissions

increase as the injection timing is delayed because there is less time for mixing before

autoignition, resulting in rich and near-stoichiometric regions, which causes locally high

temperatures that produce NOx. The combustion efficiency remains almost constant at

95%. The gross indicated thermal efficiency has a clear decrease as the fuel stratification

level increases.

From all of the discussion above, it can be concluded that both SOI timing and split

fraction have some effect on the combustion and heat release process in PFS. The peak

pressure rise rate and the maximum pressure rise rate are reduced with no observable

combustion and emissions deterioration. Therefore, to some extent, both the SOI timing

and split fraction could be used to reduce engine knock and expand the maximum load

limit. However, the impact is not significant at this boost level due to the relatively weak

φ-sensitivity of gasoline. In order to address this difficulty and have more controllability

132

on the heat release process, a double late injection (DLI) strategy (i.e., two compression

stroke injections) has been introduced and investigated in the following section.

7.5 The Double Late Injection (DLI) strategy with comparisons to the Single Late

Injection (SLI) strategy

As shown in the previous section, the fuel stratification level can be manipulated by varying

the SOI timing and the DI split fraction. However, in a high compression ratio engine

operating at a realistic boost level, the φ-sensitivity of 87-AKI gasoline is not inherently

strong enough to make a sufficient reduction on heat release rate and PPRR as the load

increases. That is to say, the stratification and the fuel’s φ-sensitivity are not enough to

further reduce engine knock while reaching the load limit. Therefore, we propose a possible

way to enhance the controllability of the heat release process at these realistic conditions –

by applying a double late injection (DLI) strategy (i.e., two compression stroke injections)

to develop a stronger φ stratification in the cylinder.

Figure 60: Injection strategy for double late injection (DLI) with SF70

Compared to a single compression stroke injection, two injection events with the

same duration are chosen to replace the single injection. In order to make a fair comparison,

the total amount of fuel being supplied during these two injections equals to that of a single

injection. Therefore, the split fraction of DI and PFI is not changed. As for the timings of

133

the two compression stroke injections, they are set equally at some crank angle spacing

before and after the original single SOI timing. For example, if the single SOI timing is -

60 deg aTDC, then the corresponding double SOI timings are set to be -70 and -50, or -80

and -40, or even -90 and -30 deg aTDC. In this way, the equivalent SOI timing is still -60

deg aTDC, but the effects on the combustion process for a single versus double

compression stroke injection can be studied with varying spacing between the two

compression stroke injections.

Figure 61 shows the heat release rate and pressure trace for fully premixed HCCI,

PFS with single late injection (SLI) at SOI = -60 deg aTDC, and three PFS with double

late injection (DLI) at an equivalent SOI timings of -60 deg aTDC, but with different split

injection spacings. It can be observed that compared to HCCI and the single injection

strategy, using a DLI strategy significantly elongates the heat release process and further

reduces the peak heat release rate. The dashed brown line in Figure 61 shows the peak heat

release rate at a split fraction of 70 and an SOI timing of -40 deg aTDC, which was the

lowest single injection peak heat release rate that was achieved by manipulating the SOI

timing and split fraction. This indicates that the DLI is able to stagger the ignition timing

of various regions more effectively than what can be achieved by an SLI strategy at the

same split fraction.

134

Figure 61: GHRR for HCCI, SF70 with a single compression stroke injection, and SF70 with a double late injection (DLI) at various split injection spacings

The combustion features and efficiencies are shown in Figure 62. Compared with

the HCCI baseline (8.46 bar/CAD), the PPRR reduces by 50% (down to 4.2 bar/CAD)

through the DLI strategy, which is an additional 25% more than that of the maximum SLI

point (5.85 bar/CAD). In other words, the prolongation of the burn duration from the HCCI

baseline condition to the DLI case is almost double that of the maximum SLI case. The

reason for the further reduction in the burn rate and PPRR is presumably due to the

increased φ stratification when applying the DLI strategy. The first late injection creates

some amount of φ stratification, but it gets mixed with the premixed charge somewhat due

to the additional time for mixing, and the turbulence created by the piston motion.

However, there is still some stratification before the second injection. When the second

injection happens, it further elevates the mixture’s inhomogeneity locally near the

135

penetration of the spray plumes and the evaporation areas. These results show that splitting

the compression stroke injection into a larger number of smaller injections creates a more

favorable φ distribution prior to ignition. This effect will be discussed more in the

following paragraph.

From Figure 62, it can be seen that at the same equivalence injection timing, the

PPRR reduces as the injection spacing increases. This indicates that the larger injection

spacing results in a higher fuel gradient primarily because the second injection becomes

later and results in a shorter autoignition delay of the richer regions. This is confirmed by

examining the EGR fraction shown in Figure 63. Higher EGR rates are required with the

DLI strategy to maintain the same combustion phasing. Delaying the equivalent SOI timing

from -60 to -40 deg aTDC further reduces the PPRR, but with diminishing returns and the

maximum stratification level merges to about the same level. The largest injection spacing

tested with a centroid of -60, -50, and -40 all have similar heat release processes, PPRRs,

and similar combustion characteristics.

136

Figure 62: Combustion characteristics and efficiencies vs. injection spacing at different equivalent injection timings at a split fraction of 70

The combustion efficiency decreases as the injection spacing increases. There are

no strong trends in the UHC emissions; although they all have a decreasing trend, the

magnitude for the reduction is less than 10%. Therefore, the decrease in combustion

efficiency is due to the significant increase in CO emissions (i.e., the CO more than doubles

137

as the injection spacing increases, see Figure 63). CO emissions are extremely sensitive to

the peak cylinder temperature, and because as the heat release is over-elongated, the peak

cylinder temperature is significantly reduced. That causes the CO – CO2 chemical reaction

to freeze, which increases CO emissions and subsequently decreases the combustion

efficiency from about 95% to 93.5%. The trends of combustion efficiency and CO

emissions agree well with previous research results by Dernotte et al. [67]. The estimated

cylinder bulk temperatures are shown in Figure 89 (in Appendix), and it can be observed

that the peak temperature difference from HCCI to the lowest peak heat release rate case

is almost 100 K (1800 K to 1715 K).

138

Figure 63: Emissions vs. injection spacing at different equivalent injection timings at a split fraction of 70

The change in gross indicated thermal efficiency is the result of competing effects;

longer burn durations and lower heat transfer losses. The former causes a reduction in the

effective expansion work. The latter reduces the energy loss to the environment resulting

in higher efficiencies. It can be concluded from the trends that the reduction in heat transfer

losses is dominant at these conditions. As a result, the gross indicated thermal efficiency

139

increases with the increasing injection spacing. The overall fuel conversion efficiency does

not significantly change because it is the product of combustion efficiency and gross

indicated thermal efficiency. However, the gross fuel conversion efficiency is still higher

for the DLI strategy than the single injection strategy by up to 1 percentage point while

additionally providing lower PPRR and the peak heat release rates.

Overall, it is encouraging to see that the double late injection could achieve a higher

stratification level that authorizes more control over the combustion process and enables a

higher load limit at a boost level of 1.6 bar. Since a good fundamental understanding of

GCI has been established, the following section will explore the effects of TBCs on GCI

combustion using the 0D/1D simulation code that was described in Chapter 4.

140

CHAPTER 8. THERMAL BARRIER COATINGS FOR GCI - MODELING

RESULTS AND DISCUSSIONS

The lesson learned from the previous HCCI - TBC modeling shows that the thicker

coatings could potentially be more favorable for LTC in terms of efficiency performance,

and the thickness limit is mostly dominated by the charge heating and the equivalence ratio

(usually at the highest load condition).

The previous chapter showed that the high load limit of PFS could be potentially

improved by applying a double late injection strategy. However, the load range for PFS is

still on the lower end when compared with the conventional combustion modes.

Additionally, as the load increases, the PFS strategy can be limited due to oxygen

availability, which eventually requires more boost which can be somewhat impractical or

cause an efficiency penalty. Therefore, a more stratified gasoline compression ignition

strategy is needed at high-load operating conditions where the injection window shifts to

the PPCI range [107][108]. However, challenges remain for pure PPCI modes at high load

because they rely on a significant amount of EGR dilution to maintain the NOx emissions

and combustion noise in an acceptable range, where accurate control of EGR rate and

combustion stability can become difficult [109]. Recent studies have proposed a hybrid

mode incorporating both PPCI and diffusion combustion [110] [111]. The PPCI-diffusion

strategy could be achieved by two/multiple injections during the late compression stroke

and followed by one main injection near TDC to promote gasoline diffusive combustion.

A CFD simulated heat release rate, cylinder pressure, and in-cylinder distributions are

shown in Figure 64 from [111]. It can be seen that during the compression stroke, the early

141

injection enables PPCI combustion which creates a desired in-cylinder thermodynamic

conditions for a gasoline diffusive combustion process driven by the second injection. This

hybrid combustion process has shown promising combustion performance with tolerable

emissions [111][112]. This PPCI-diffusion combustion strategy will be adopted for most

of the simulations in this GCI-TBC study in this chapter. The CFD models were provided

by Aramco Services Company, i.e., the model for generating the results in Figure 64 [111].

Figure 64: PPCI-diffusion combustion strategy from Yu et al. [111].

Left: heat release rate and cylinder pressure; Right: in-cylinder mixing and temperature distribution

8.1 Model upgrade and validation

A 0-dimensional thermodynamic model has been established for the GCI – TBC study.

The overall frame was similar to the model frame introduced in Chapter 4. However, a few

concerns have to be addressed due to the different in combustion modes and data

availability. Since the diffusion combustion phase of GCI is not very sensitive to the IVC

temperature, the intake temperature was left unchanged throughout these simulations. In

theory, the PPCI phase of combustion is expected to be influenced by the coating, but the

possible change of PPCI was excluded from the scope of this thesis because the portion of

142

heat release is relatively small compared to the diffusion phase, and the phasing could

potentially be adjusted by implementing a different injection strategy. As a result, the heat

release was prescribed from the CFD output. Additional attention should be paid to heat

transfer modeling. In Chapter 4 and 5, the research target was HCCI, whose mixture and

combustion process can be considered nearly homogeneous, which is to say that the heat

fluxes are relatively uninformed to each chamber components such as the piston, firedeck,

valves, and liner. Thus, a modified Wochini heat transfer correlation could be adopted to

capture heat flux change when coatings are applied. However, in this section, the research

target switches to GCI with PPCI-diffusion combustion strategy. It can be observed from

Figure 64 that the PPCI and diffusion combustion were extremely heterogeneous. The

injection plumes targeted the piston bowl to ensure the wall-guided air utilization, and

combustion happens much closer to the piston rather than the head and liner.

By examining the averaged heat flux to each independent chamber component (Figure

65, from CFD simulation results), it can be seen that the heat flux to piston was much

higher than the heat flux to the other components (the valves and firedeck were about the

same and the heat flux to liner was the lowest). This highly heterogeneous heat flux

distribution broke the uniformed heat flux assumption that was utilized in the HCCI

analysis. In other words, the traditional heat transfer correlation (i.e., Woschni, Hohenberg,

Modified Woschni) may be able to capture the average heat flux, but they will not be able

to determine the local heat fluxes which would dictate the highest possible heat fluxes and

temperatures experienced by different engine components. Since our goal is to evaluate

143

TBCs and their influence on each combustion chamber component, employing accurate

heat fluxes became critical.

Figure 65: Heat flux to different combustion chamber components.

Recall from Equation 11 that the heat flux consists of two parts: a convective heat

transfer coefficient ℎ𝑐𝑜𝑛𝑣 and a temperature difference between the gas and wall (𝑇𝑔𝑎𝑠 −

𝑇𝑠𝑢𝑟𝑓𝑎𝑐𝑒). The source of error came from vastly different ℎ𝑐𝑜𝑛𝑣 experienced by different

components, and the CFD result shows that the gas side temperatures were fairly similar

for each component (which is shown in the Appendix, Figure 90). Therefore, a hybrid heat

flux estimation approach was established. Instead of using a convective heat transfer

correlation, the model adopts the convective heat transfer coefficient from the spatially

averaged CFD result, ℎ𝑐𝑜𝑛𝑣,𝐶𝐹𝐷,𝑖𝑘 . Since the gas side temperatures for each component were

144

fairly close to the bulk gas temperature, the heat flux equation still uses the same terms for

gas and wall surface temperatures. The modified equation became:

�̇�ℎ𝑡,𝑖𝑘 = ℎ𝑐𝑜𝑛𝑣,𝐶𝐹𝐷,𝑖

𝑘 ∗ (𝑇𝑔𝑎𝑠 − 𝑇𝑠𝑢𝑟𝑓𝑎𝑐𝑒𝑘 ) (24)

where subscript 𝑖 indicates the time step and superscript 𝑘 indicates different chamber

components (e.g., piston, valves, head, and liner). ℎ𝑐𝑜𝑛𝑣,𝐶𝐹𝐷,𝑖𝑘 is the spatially averaged

convective heat transfer coefficient from CFD output. Since this coefficient is mainly

affected by turbulence and combustion, it is assumed to be unchanged for each operating

condition as long as the heat release process is replicable. Although the combustion process

may be affected by the coatings due to the hotter walls, a CFD result shows no noticeable

change to the diffusion phase, and altering the injection timing of the first injection could

help to compensate for the phasing for the PPCI combustion.

Figure 66 shows piston heat flux comparison among the hybrid method (in dashed

black), pure CFD estimation (in red), and Hohenberg heat transfer correlation (in blue).

Note that Hohenberg's correlation is chosen here (instead of Chang’s modified Woschni

correlation) because of the diffusive combustion. It can be observed that the Hohenberg

correlation has a considerably underpredicted piston heat flux during combustion.

Comparing the hybrid method and CFD prediction, the former has a slightly lower peak,

but the overall magnitude and trend agree very well with the CFD results.

145

Figure 66: Piston heat flux estimation from three different methods. Blue: Hohenberg correlation; Red: CFD results; Dashed black line: Hybrid heat flux (0D dT + ℎ𝐶𝐹𝐷)

The engine platform in this section adopts Aramco’s advanced light-duty GCI

engine. Detailed engine specifications are listed in Table 14. In addition, the model was set

to match the full load operating conditions at 23.5 bar IMEPg, and the overall engine

operating conditions are provided in Table 15.

Table 14: Engine specifications for Aramco light-duty GCI engine

Geometric Compression ratio 17

Stroke [mm] 123

Bore [mm] 82

Connecting rod length [mm] 205

Critical valve timings [aTDC] IVC: -164 EVO: 152

Head bore ratio 1.10

Piston bore ratio 1.16

146

Table 15:PPCI-diffusion combustion operating conditions

Engine load [IMEPg] 23.5

Intake pressure [bar] 3.3

Intake temperature [K] 330.6

Exhaust pressure [bar] 4.2

External EGR [%] 30

Engine speed [rpm] 2500

Combustion Efficiency [%] 99

Fueling rate [mg/cycle] 78

φ / φ’ 0.77 / 0.51

Since the actual engine is still under commissioning, the 0D model was validated

against the CFD simulation results. The CFD model with PPCI-diffusion combustion mode

was well-validated against experimental data from a heavy-duty GCI engine through

multiple metrics [111]. As a result, the two models agree reasonably well with each other,

building confidence in the current model’s predictivity. The modeling results of TBCs for

GCI will be discussed in the following section.

Figure 67: Cylinder pressure and bulk gas temperature comparison between CFD and 0D models

147

8.2 Effects of coating thickness on gasoline compression ignition combustion

From the study in Chapter 5, it can be concluded that low thermal conductivity (𝑘) and low

volumetric heat capacity (𝑠) are always favored for coating performance regardless of the

combustion strategy. For traditional SI and conventional diesel combustion, low 𝑘, low 𝑠,

and “thin” coating are preferred because of the existence of the charge heating penalty and

knock for SI. For LTC which requires intake thermal conditioning (e.g., HCCI with high

octane fuel, TSCI, or low load GCI), thicker coatings could be helpful in terms of

combustion, thermal, and volumetric efficiency. Thus, in this section, two sets of real-

world coating material properties will be chosen, and the effect of thickness will be

investigated.

A list of real-world material properties is provided in Table 16. Some engine-

relevant metal properties are also included for reference. The engine head adopts the

property of aluminum; valves (assumed not to be sodium-filled) and piston uses the

property of steel; the liner uses the property of iron. The highlighted materials are two

coating candidates selected in this section for investigation. Gadolinium Zirconate (GdZr)

has shown very encouraging potentials due to its originally low thermal conductivity, and

the performance has been experimentally demonstrated [63]. The other candidate, Gen.2

Candidate#2, was one of the recently found materials with superb thermophysical

properties for coating application. The name of the material is not provided due to a

potentially confidential disclosure for commercialization. These two candidates were

evaluated via four thicknesses (100, 200, 500, and 1000 microns) and compared with the

148

metal baseline. Only the piston was coated in this section, and the rest of the chamber

surface remained uncoated. A detailed coverage study will be conducted in a later section.

Table 16: Real-world material properties for engine coating application

Thermal

conductivity (𝒌) [W/mK]

Heat capacity (𝒄𝒑)

[J/kgK]

Density (𝝆) [kg/m3]

Volumetric heat capacity (𝒔)

[kJ/m3K]

Aluminum 240 921.1 2830 2606.71

Steel 43 502.4 7850 3943.84

Iron 69.4 460.5 7850 3614.93

YSZ 1.33 354 5026 1688.48

Gen. 1 GdZr 0.74 430 5850 2515.50

Gen 2 Candidate#1 0.6 500 2987 1493.50

Gen. 2 Candidate#2 0.32 665 3200 2128.00

Gen. 3 Candidate#1 0.3 415 2996 1243.34

Air 0.04 718 0.75 0.54

The simulated surface temperatures are shown in Figure 68. The values of

temperature swing were labeled purple on the left side, and the values of thickness were

labeled orange on the right. Similar to the results that were shown previously, increase the

coating thickness elevates the overall surface temperature and reduces the magnitude of

temperature swing. Better material properties combination (i.e., lower 𝑘 and 𝑠) lead to a

higher temperature swing, which has a lower average temperature and enables a similar

peak surface temperature with a thinner coating.

149

Figure 68: Surface temperatures that result from different thicknesses for two different candidate materials at 2500rpm, 23.5 bar IMEP.

The gross indicated thermal efficiency is shown in Figure 69. It is seen that both

coatings improve the thermal efficiency. The improvement was highly pronounced at the

beginning of the thickness sweep when 100 microns coatings are applied. As the coating

thickness increases, the improvement starts showing a trend of diminishing returns for three

main reasons. First, the temperature swing was reduced as the coating thickness increases,

which means that the increment of peak temperature is less than the elevation of

temperature everywhere else during the cycle. Since the heat transfer blocked near TDC is

most important for efficient thermodynamic conversion, the elevated temperature

everywhere else is not as rewarding as elevating the temperature in the near-TDC region,

i.e., where the peak occurs. This is mainly responsible for the trend of diminishing returns.

The second reason was that the charge heating occurs more with thicker coatings, which

results in a richer charge with a lower 𝛾. Finally, the bulk gas temperature increases as the

coating thickness increases, which promotes heat transfer to the uncoated firedeck and liner.

The figure on the right shows that the thermal efficiency was linearly correlated with the

150

peak surface temperature, which agrees well with the findings in the HCCI section (Figure

47). Since the improvement was achieved by only varying the thickness, both candidates

have indistinguishable slopes, and the small gap between these two was due to the different

material properties, i.e., 𝑘 and 𝑠.

Overall, the improvement value seems low because only the piston was covered.

Although the efficiency increment reduces as the coating thickness increases, the thermal

efficiency is still increasing at 1 mm. Charge heating is one of the potential reasons that

may prohibit the thickness from increasing further.

Figure 69: Gross indicated thermal efficiency v.s. thickness (left) and peak surface temperature (right) with two candidate materials.

Concurrently, CFD simulations are being performed on this project as part of a

separate thesis and the CFD simulations are best suited to understand the impacts of TBCs

on GCI combustion. The bulk gas temperature, the GdZr coated piston surface temperature,

and the temperature of the metal baseline are shown in Figure 70.

The IVC (at -164 CAD aTDC) and peak temperatures were shown in the zoomed-

in boxes on the left and right, respectively. It can be seen that the IVC temperature of 0.1

mm coated case was even lower than the metal baseline, which is because the coating

151

surface temperature during the intake stroke swings down due to the intake charge cooling.

Therefore, the charge density at IVC is actually higher with the coated piston than the metal

piston. The peak gas temperature has a similar trend due to one additional reason: the

charge is leaner which lowers the peak combustion temperature. It was very surprising to

see that the IVC temperature did not increase very significantly. It only increased by 4.2 K

from the metal baseline to the 1mm coated case (6 K with Gen.2 coating). This unexpected

low charge heating was not previously observed in the HCCI section, partially because for

these results, the intake temperature was held constant instead of varying it based on

ignition delay correlation of HCCI. In addition, the previous HCCI study was naturally

aspirated. In this case, the intake was boosted to 3.3 bar, which results in less temperature

increase due to the larger mass of air in the cylinder. A more detailed intake boost study

could be helpful to understand and quantify the boost level and coating interactions.

Figure 70: Bulk gas, coating, and metal piston surface temperature at 23.5 bar IMEPg

152

Overall, it was encouraging that the efficiency was predicted to be higher with

thicker coatings and that the charge heating was not as much as expected as a result of the

high level of intake boost.

8.3 Effect of boost on the performance of “traditional” and “temperature swing”

coatings

It was speculated that the charge heating could be improved by intake boosting. This

section will quantitively investigate the effects of intake boosting on TBC performance

thermodynamically. Meanwhile, the performance difference between traditional and

temperature swing coating will be organized and discussed.

The previous load condition was 23.5 bar with 3.3 bar intake pressure, and the

overall phi was ~ 0.77. Since this load is not achievable and realistic with a lower boost

level, the load was lowered to 15 bar IMEPg, and the intake pressure ranged from 1.6 to 3.5

bar, i.e., equivalence ratio of 0.79 to 0.34, respectively. Yttria-stabilized zirconia was

selected to represent traditional coating, and Gen.2 Candidate#2 was selected for the

temperature swing coating. The properties of both materials were provided in Table 16.

The thickness of Gen.2 coating was pre-selected as 1 mm, and both piston and firedeck

were covered in the simulation. The thickness of the traditional coating was determined by

matching the peak surface temperature to the Gen.2 coating (which was 6.8mm in this case)

since it was shown that the efficiency is highly correlated with the peak surface temperature.

Figure 71 shows the surface temperature of two coatings and metal baseline, with their

corresponding bulk gas temperatures. The performance is evaluated from three main

153

aspects: the IVC temperature, the gross indicated thermal efficiency, and the mixture’s

equivalence ratio.

Figure 71: Bulk gas and piston surface temperature at 15 bar IMEPg and 1.6 bar intake pressure.

Figure 72 shows the IVC temperatures (on the top row of subplots), which is the

metric to evaluate the charge heating and charge density at IVC. At a boost level of 1.6 bar

(the figure on the top left), the temperature swing coating has a 24 K charge heating penalty,

whereas the traditional coating incurs more (38 K) charge heating due to the higher

temperature throughout the engine cycle. As a consequence of charge heating, the

equivalence ratio increased (shown in the bottom left plot), which may cause deteriorated

combustion and emissions performance. Alkidas et al. have shown that the engine with

traditional coating had equal or superior fuel consumption compared to the metal baseline.

However, significantly higher NOx emissions were reported and speculated as a sign of

altered mixture’s property (i.e., a richer mixture) and a deteriorated combustion process

[113]. In addition, if the load is limited by the equivalence ratio, the power density would

154

be decreased by ~4% and ~8% for temperature swing coating and traditional coating,

respectively.

Figure 72: IVC temperature and equivalence ratio at different boost levels

The coatings’ performances were not ideal at a lower boost pressure, but the swing

coating still shows a 0.6% percentage points improvement in thermal efficiency (can be

observed from Figure 73). The traditional coating has less efficiency gain due to a richer

mixture (lower 𝛾 ) and possibly more heat transfer losses because of a higher gas

temperature. However, this improvement is purely thermodynamic, which means that this

model would not capture the possible deterioration of combustion due to altered air

utilization. Therefore, it is premature to confirm the validity of the efficiency improvement

with a higher equivalence ratio.

155

Figure 73: Gross indicated thermal efficiency at different boost levels

When looking at the 3.5 bar boost case, the charge heating has been significantly

reduced for both coatings, where the temperature swing coating resulted in only ~10 K

charge heating penalty, and the traditional coating has 22 K of charge heating. The

equivalence ratio for temperature swing coating is actually lower than the metal baseline

despite the charge heating due to higher efficiency; therefore, less fuel was required to

maintain the same load. However, the traditional coating has a higher φ than the metal

baseline. The 3.5 bar boost increases efficiency for all cases, including the metal baseline.

The improvement of coating cases was more pronounced (by an extra one percentage point

compared with the metal baseline) due to the additional charge heating reduction. The

improvement for temperature swing coating could be considered valid because the

equivalence ratio falls below the metal baseline, which is not likely to cause any

deterioration of combustion, and the thermal efficiency benefit became 1.6 percentage

points (3.5% relative) at this boost level. It is important to note that the combustion

efficiency is likely to increase with any type of TBC. Since the 0D single-zone model does

not accurately capture combustion efficiency changes, especially when the thermal

156

boundary layers are affected, future CFD or experimental works are expected to shed light

on potential combustion efficiency improvements with TBCs. Figure 74 shows that the

equivalence ratio transitions from increasing with the Gen.2 TBC compared to the metal

baseline to decreasing with the Gen.2 TBC compared to the metal baseline as intake

pressure increases. It can be observed that both coatings started with a higher ϕ than the

metal baseline at low boost levels. However, as the boost level rises, the swing coating and

metal baseline reach a breakeven point (at ~2.9 bar). Therefore, it can be implied that before

that level of boost, a charge heating penalty has occurred that reduced the power density.

When the pressure increases above 2.9 bar, although the IVC temperature could still be

higher than the metal baseline, the same engine load can be achieved with the same ϕ (i.e.,

less fuel) because the engine operates more efficiently. Note that if the improvement of

combustion efficiency is also considered, a slightly lower breakeven level is expected.

Figure 74: Boost pressure v.s. equivalence ratio.

A comparison among temperature swing coating, traditional coating, and steel baseline.

157

Overall, it is evident that the thick temperature swing coating outperformance the

traditional insulation in terms of thermal efficiency and managing charge heating. The

following section will provide more details about the impacts of coating coverage.

8.4 Valve heat transfer estimation and coating coverage investigation

A thorough coating coverage study necessitates an estimation of valve heat transfer.

Although engine valves have complex geometry and a 1D assumption may not be perfect,

it is still fairly reasonable when considering the front surface temperature prediction. The

computational study from Shojaefard et al. has shown that after reaching steady-state, the

temperature was distributed quite evenly in the horizontal plane despite a small area of

local inhomogeneity around the valve seat, i.e., the temperature changes mostly from

bottom to top but not horizontally [114]. Thus, a similar 1D transient heat transfer

framework was adopted again to study the engine valves, but with several additional

boundary conditions shown in the figure below. Since valve heat transfer consists of

multiple sources and types of interaction, an estimated valve heat transfer breakdown is

shown below on the left.

158

Figure 75: Engine valves heat transfer breakdown (left) and schematic of 1D modeling (right)

On one side of the valve (shown as the top of the modeling framework schematic

on the right side of Figure 75 or shown as the bottom of the valve front on the left side of

Figure 75), the convective heat transfer happens from combustion gas to the valve front

face, and the heat transfer coefficient (ℎ𝑓𝑟𝑜𝑛𝑡,𝐶𝐹𝐷) was adopted from CFD outputs (e.g.,

yellow and purple lines in Figure 65). Three heat transfer sources were considered on the

backside: conduction heat transfer to the valve seat, convective heat transfer from the back

surface to either the intake or exhaust gases, and heat transfer from the valve stem to the

oil film/stem guide. The values of heat transfer coefficient, seat, and oil film temperature

were adopted from the literature ([115] and [116]). The backside gas convective heat

transfer coefficient and gas side temperature were also imported from CFD outputs. The

heat fluxes of different boundary conditions mentioned above are shown in Figure 76.

159

Figure 76: Heat fluxes from different boundary conditions at 2500rpm, 23.5 bar IMEP.

Left: Intake valves; Right: Exhaust valves

It can be seen from the zoomed-in box that the heat flux to valve seats was

significantly reduced during the gas exchange period because the valves were not in contact

with the seats; therefore, the conduction heat transfer to the valve seat changes to

convective heat transfer from the flowing gas. The backside gas heat flux increases during

the same period due to the fluid flow. As a result, the simulated surface temperatures for

each chamber component are shown in Figure 77.

It was shown that the valves have a much higher temperature than the piston, mostly

because the piston and was cooled by oil squirter on the backside. For the metal baseline,

the intake valve was almost consistently ~60 K cooler than the exhaust valve at this load

condition because of the backside charge cooling by the fresh charge as opposed to exhaust

gases for the exhaust valve. This is a reasonably good estimation when comparing with the

boundary condition generated by GT-SUITE software [117]. The swing was lower on the

valve because of the high average temperature and reduced heat flux. It is important to note

160

that the maximum temperatures of either GdZr or the Gen.2 candidate were approaching

1000 K and 1100 K, and the temperature peaks locally could be much higher than these

averaged values, which raises the concern about coating durability since a temperature limit

of ~1300K for both of these materials was reported. A 3D FEA approach would be required

to evaluate the local peak temperature, but the coating durability and stress analysis are out

of the scope of the current dissertation. Overall, the surface temperature estimations have

been established for each chamber component. Next, a coating coverage investigation will

be performed.

Figure 77: Coated and baseline surface temperature of piston and valves at 2500rpm, 23.5 bar IMEP.

It is feasible to apply either thermal sprayed or solution sprayed coating onto the

piston surface, engine head, and valve surface. However, applying a TBC to the liner is

less straightforward, and realistically, the TBC could only be applied from the top to the

first ring land when the piston reaches the TDC. The rest of the liner can not be covered

because of the piston ring-liner interaction and oil lubrication. Figure 78 shows a coating

coverage schematic that was studied in this section. Although the coverage on liner could

161

be more difficult than other components, it is still investigated numerically to determine

the potential gains and tradeoffs.

Figure 78: Possible TBC coverage in the combustion chamber

The gross indicated thermal efficiency with different coverages and materials is

shown in Figure 79 on the left, and the gain ratios of each component are shown on the

right. It can be observed that increasing the TBC coverage increases thermal efficiency.

The number on each column portion represents the additional gain from the last coverage

case. For example, the +0.13 percentage points (pp) for “+ Cover Head” corresponds to

covering the head with GdZr in addition to the piston and means that the efficiency gain

was 0.13 percentage points compared to the only piston-covered case labeled “Cover

Piston”. The total gain compared to the metal baseline for covering the head and piston is

0.49 percentage points. In this way, the efficiency benefits of covering each individual

component can be compared. At this load, covering the piston has the most gain for both

coating materials (~ 60% of full coverage gain, shown in the pie chart on the right).

162

Although the head and valves have almost identical areas (Avalves : Ahead = 50.3% : 49.7%),

the benefit of covering the head (~22% of full coverage gain) was more pronounced than

covering all four valves (~15% of full coverage gain). This was because the valves already

have a high temperature and low heat flux; therefore, the relative temperature swing (i.e.,

the temperature difference between the coating surface and the original uncoated surface)

was lower than the head, resulting in a smaller efficiency improvement. Coating the liner

is an extreme approach that requires a significant amount of effort, and the modeling results

show that the payback was very small due to the small area and low heat flux. Therefore,

it is not worth coating the liner clearance. The total efficiency improvement, excluding the

liner coverage, is 0.56 percentage points (1.2% relative) and 0.92 percentage points (2.0%

relative) by GdZr and the Gen.2 candidate, respectively.

Figure 79: Left: Gross indicated thermal efficiency for different coverage @ 2500rpm, 23.5 bar

IMEPg Right: Efficiency gain ratio distribution

From Figure 80, it seems that the head has a higher relative swing than the piston.

This is because the head substrate is aluminum, which has six times the conductivity of

163

steel. In other words, the heat could be dissipated much faster by the coolant on the

backside, and therefore the temperature of the aluminum head was much lower than the

steel piston. Once the coating is applied, it blocks the heat transfer from both sides, i.e., the

hot gas and cold aluminum, and the relative swing is more pronounced. Despite the

existence of a higher relative temperature swing, Figure 79 shows the gain from the head

are less than that from the piston because the head has less area and combustion occurs

closer to the piston in GCI, which increases the local turbulence and results in a higher

convective heat transfer coefficient.

Figure 80: Relative temperature swing for piston, head, and valves with 200 microns GdZr @

2500 rpm, 23.5 bar IMEPg

An example of how different substrates affect the relative temperature swing is

shown in Figure 81. The same boundary conditions, effective metal thickness, and TBC

properties were applied to both aluminum and steel. It can be seen that without the coating,

the steel piston was ~140 K hotter than the aluminum piston, which infers that the steel

164

piston is likely to have a higher thermal efficiency due to the hotter surface temperature.

Schaedler et al. have tested the piston with steel and aluminum under the same operating

condition, and it was found that the steel piston had a ~5% (relative) improvement in fuel

consumption [118]. The improvement was reported as the result of decreased piston-bore

clearance and better dimensional stability of steel. However, this simulation result also

implies that it is likely that higher surface temperatures also played and important roles in

that efficiency improvement.

When a 200 microns GdZr was applied, the aluminum-based and steel-based piston

temperature differences were reduced to 60~70K. As a result, the aluminum piston has a

much higher relative temperature swing, which means that the potential TBC improvement

could be more pronounced when using an aluminum substrate rather than steel. This does

not explicitly mean that the aluminum piston with the coating will have higher efficiency

than that of steel piston, it only implies that the aluminum substrate could achieve a higher

efficiency gain by applying TBC on the surface.

Figure 81: The effect of substrate on coating performance

165

As learned from the above cases, the individual heat flux to each component and

substrate material governed the efficiency improvement. Thus, the gain ratio could be

different when operating condition changes. Figure 82 shows that the gain ratio shifted

toward the firedeck coverage at a lower operating condition. At 23 bar, coating the piston

gets more benefits because of the significantly higher heat flux to the piston. When

operating at 6 bar in GCI with more uniformed (HCCI-like) heat flux to each component,

the full coverage on firedeck became more rewarding due to the aluminum substrate (for

head) and increased heat flux for valves.

Figure 82: Efficiency gain distribution at different loads. Left: 23.5 bar IMEPg. Right: 6 bar IMEPg

166

Figure 83 shows the relationship between thermal efficiency and relative charge

heating (i.e., IVC temperature compared to metal baseline). At the same level of efficiency

gain, the full coverage case incurs much less charge heating than using a thicker coating

applied only on the piston. Therefore, coating the full piston and firedeck was

recommended regardless of the operating condition. Coating thickness for each component

was determined by local maximum temperature (for durability consideration), intake

charge heating, and the mixture's equivalence ratio. Although this work found that thick

TBCs do not incur as much charge heating penalty as the traditional combustion modes

because of the high intake boost and overall lean operation, it is still not recommended to

incur excessive charge heating, which may affect the mixture’s properties and air

utilization for diffusive combustion. It is important to note that the coating thickness on

individual chamber components can be different due to their own substrate material and

cooling strategies.

Figure 83: Gross indicated thermal efficiency v.s. relative charge heating

167

CHAPTER 9. CONCLUSIONS AND UNIQUE CONTRIBUTIONS

9.1 Summary and conclusions

In-cylinder thermal barrier coatings (TBCs) have great potential to reduce heat transfer

losses and increase thermal efficiency. It had been previously shown that thick TBCs

negatively impact the performance of conventional combustion modes by reducing

volumetric efficiency and increasing end-gas knock tendency. Due to the nature of kinetics-

driven combustion, low-temperature combustion (LTC) is fundamentally different from

the conventional combustion modes, where these issues may not apply. Therefore, it was

desired to explore the favorable coating configurations for LTC systematically. This thesis

dissertation explored and evaluated TBCs in LTC from a different perspective: thick,

temperature swing coatings. The key hypothesis is that kinetic-driven combustion requires

certain thermodynamic prerequisites to achieve autoignition, where the charge heating

associated with thick TBCs becomes beneficial for fulfilling these prerequisites along with

additional efficiency and emissions benefits.

This thesis was majorly four-fold: The first goal was to experimentally demonstrate

and comprehensively investigated the effects of thick thermal barrier coatings on pure-

HCCI combustion with two different fuels. Second, a parametric computational

investigation into the effects of various coating properties on pure-HCCI combustion was

performed. Since LTC contains a large family tree ranging from HCCI to stratified LTC

such as GCI, it was desired to explore TBC with GCI in addition to HCCI. The first step

was to understand the GCI combustion process with an experimental investigation into PFS

combustion and injection strategy. Finally, a goal of the dissertation was to establish a

168

preliminary guideline for TBCs with GCI via thermodynamic modeling. In general, it was

shown that thick thermal barrier coatings have great potential and benefits to work with

various low-temperature combustion modes. An overview of the key findings are listed as

follows:

9.1.1 Demonstration and justification – Comprehensive experimental investigation of

thick thermal barrier coatings for pure-HCCI

Experiments were conducted on a light-duty single-cylinder research engine to investigate

the effects of thick thermal barrier coatings on pure-HCCIwith different fuels. The study

mainly focuses on three aspects:

• The effects of TBC thickness and surface finish (with or without dense sealing

layer) on gasoline pure-HCCI

• The potential benefits of TBCs when using an alternative fuel with a high latent

heat of vaporization (wet ethanol 80)

• The interaction between the hot TBC surface and spray of WE80.

In order to achieve a comprehensive investigation, five pistons were tested including:

1. A metal piston baseline at a CR of 15.8;

2. A metal piston baseline at a CR of 14;

3. A 1mm TBC piston at a CR of 14.7 with a dense sealing layer applied to the coating

surface;

4. A 2mm TBC piston at a CR of 15.2 with a dense sealing layer applied to the coating

surface;

169

5. A 2mm TBC piston at a CR of 15.0 without the dense sealing layer applied to the

coating surface;

Two fuels were examined and compared for the pistons listed above: an 87-AKI gasoline

and an alternative biofuel with high cooling potential – Wet Ethanol 80. A full load sweep

was conducted with each fuel on each piston. The conclusions are as follow:

• Thick TBCs extend the low load limit by 14.8% with gasoline, and 15.4% with WE80

by improving combustion efficiency. No deterioration of the high load limit was

observed.

• There was no discernible impact by the TBCs on the burn duration or heat release

process.

• The combustion efficiency increases with TBC thickness. The increment is up to 1.5

percentage points with both gasoline and WE80. The gasoline cases experience the

most benefits at low load (2 to 3 bar IMEPg), while the WE experiences the most

benefits at mid-to-high load (3 to 4.5 bar IMEPg).

• Higher thermal efficiencies were achieved with TBCs. Increasing the TBC thickness

reduces heat transfer losses and improves thermal efficiency. Due to the improved

combustion and thermal efficiencies, the fuel conversion efficiency increased by up to

4.3% with WE 80, and 3.8% with gasoline.

• The dense sealing layer reduces surface porosity and improves UHC emissions and

combustion efficiency.

• With the 2mm TBCs, the intake temperature requirement was reduced by 15 K with

gasoline, and 10 K with WE80.

170

• Spray/TBC impingement amplifies any effect that was caused by switching from cold

metal piston surface to hot coated surface in terms of the fuel evaporation.

• In general, TBCs do not significantly affect the exhaust temperature in LTC. This is

due to a variety of competing thermodynamic factors.

In this study, only the piston was coated, which means that all of the potential benefits such

as the efficiency gains, emissions reduction, lower intake temperature requirements, etc.

could be amplified if the engine head and valves were also coated.

9.1.2 Deep dive into the fundamentals – A parametric computational investigation into

the effects of various coating properties for pure-HCCI

An 0D single-zone thermodynamic model was established and coupled to a 1D heat

conduction solver to explore TBC properties' effects on pure-HCCI fundamentally. The

model is first validated against experimental results with an aluminum piston, followed by

validation against experimental results with a 2mm TBC-coated piston. Three parameters

were then thoroughly investigated, including thermal conductivity, thickness, and

volumetric heat capacity. Six load conditions were selected and matched from 2.0 bar to

4.5 bar IMEPg in increments of 0.5 bar. The following conclusions can be drawn:

• The effects of TBCs on pure-HCCI are distinct from the conventional combustion

modes.

• Increasing thickness with a low-𝑘 material leads to considerable improvements to

efficiency, reductions in the required intake temperature, and an improvement in the

exhaust flow enthalpy. Low volumetric heat capacity is desired to promote temperature

swing, thus further increasing efficiency.

171

• In pure-HCCI, any coating that reduces compression stroke heat transfer allows for

lower IVC temperatures which actually increases charge density, which is opposite of

the trend with the conventional combustion modes. This effect was most noticeable

with thick TBCs, although low conductivity TBCs also reduce compression stroke heat

transfer.

• In general, TBCs do not significantly affect the exhaust temperature in pure-HCCI.

This is due to a variety of competing thermodynamic factors. The exhaust flow

enthalpy does increase due to the increased mass flow rates with TBCs.

• Increasing thickness elevates the average surface temperatures and slightly lowers the

temperature swing. Reductions in thermal conductivity elevate the average surface

temperature and increase the magnitude of temperature swing (mostly for very low

conductivity values). Reductions in volumetric heat capacity do not affect the average

surface temperature, but they do increase the temperature swing.

• In pure-HCCI, temperature swing is not an important factor to improve thermal

efficiency on its own; instead, elevating the surface temperature is the most important

factor to improve efficiency in pure-HCCI while resolving some of the other challenges.

The most direct mechanism to elevate the surface temperature is to apply thick TBCs

with a low-𝑘 material.

9.1.3 Establishing the understanding of GCI - an experimental investigation of injection

strategies for gasoline PFS

In order to understand the gasoline compression ignition fundamentals, an experimental

study using the PFS combustion strategy was conducted on a single-cylinder light-duty

172

diesel engine. This study aims to investigate the features of PFS combustion on a practical

modern light-duty diesel engine with a compression ratio of 16 and at a realistic boost level

of 1.6 bar. Additionally, this study determined the effects of a double late injection (DLI)

strategy on the heat release process and combustion characteristics. The results can be

divided into two sections. First, the effects of single late injection (SLI) timings and split

fractions on PFS were presented. Second, the effects of DLI on PFS combustion and the

comparison between DLI and SLI were presented. The following conclusions can be drawn

from the results:

I. At this boost level, the variation of SOI timings between -140 to -80 has almost no effect

on combustion. Injection timing begins to affect the combustion process at and after -60

deg aTDC. Further retarding SOI timing:

• Increases φ-stratification.

• Reduces the PPRR, peak heat release rate, and prolongs the burn duration.

• Decreases the gross indicated thermal efficiency and has almost no effect on

combustion efficiency.

• Increases the requirement of EGR to maintain combustion phasing constant.

• Increases CO and NOx emissions.

• Decreases the UHC emissions from -140 to -50 aTDC, followed by an increase

until -40 aTDC.

Decreasing the split fraction (SF), i.e., increasing the portion of DI fuel injected during the

compression stroke, results in:

• Higher φ-stratification.

173

• Lower PPRR and heat release rate, and an extended burn duration.

• Lower gross indicated thermal efficiency and slightly higher combustion efficiency.

• Higher required EGR rate to match combustion phasing.

• Lower UHC emissions but higher CO and NOx emissions.

• The effects of the split fraction are diminishing. SF80 and SF70 have almost

identical effects on PFS combustion.

II. Compared to the maximum stratification level that can be achieved by a single late

injection (SF = 70% & SOI = -40 deg aTDC), the double late injection strategy:

• Increases the φ-stratification and results in a more staggered autoignition event.

• Lowers the PPRR and the peak heat release rate, and elongates the combustion

process.

• Decreases the combustion efficiency, increases the gross indicated thermal

efficiency and the fuel conversion efficiency. However, the thermal and

combustion efficiencies could be improved with a more advanced combustion

phasing.

• Slightly increases the EGR requirement to match combustion phasing.

• Decreases the UHC and NOx emissions but increases the CO emissions.

Increasing the injection spacing and delaying the equivalent SOI timing for the double late

injection strategy both result in:

• Higher φ-stratification but with diminishing returns and the maximum stratification

level merges.

• Lower PPRR and HRR, and a longer burn duration.

174

• Lower combustion efficiency but a higher gross indicated thermal efficiency. The

fuel conversion efficiency is approximately constant.

• A slightly higher EGR requirement to match combustion phasing.

• Lower UHC and NOx emissions, and higher CO emissions.

The key conclusion of this work is that the results indicate that separating the compression

stroke injection into a larger number of smaller injections is beneficial for control and for

the global combustion characteristics of PFS.

9.1.4 A preliminary guideline for TBCs with GCI - A computational evaluation

The 0-1D thermodynamic model has been established and collaborated with CFD results

to investigate thermal barrier coatings for gasoline ignition combustion with PPCI-

diffusion combustion strategy primarily at a load of 23.5 bar IMEPg. Two real-world

coating materials were pre-selected as Gen.1 and Gen.2 candidates. The investigation

consists of three main topics: the effect of coating thickness on GCI, the effect of intake

boost on the performance of traditional and temperature swing coatings, and the effect of

combustion chamber coverage on engine performance. Some key findings are listed below:

• Increasing coating thickness increases the thermal efficiency for GCI combustion,

but the increment has a noticeable trend of diminishing return. The rate of

efficiency gain after 500 microns is very low.

• Even with two distinct material properties, it was found that the thermal efficiency

is still linearly correlated with the peak surface temperature at an almost identical

slope.

175

• The charge heating was much less than expected. The increased IVC temperatures

were 4.2 K and 6 K for Gen.1 and Gen.2 coating, respectively. The high-level

intake boost is mainly responsible for this low charge heating.

• At a boost level of 1.6 bar, 15 IMEPg, both the traditional and temperature swing

coatings incurred significant charge heating penalties, causing an increased

equivalence ratio, potentially lower power density, and potentially poor air

utilization.

• As boost level increases, the temperature swing coating reached an equivalence

ratio breakeven point at 2.9 bar compared with metal baseline, which indirectly

confirmed the validity of the efficiency gain. However, the traditional coating never

reached the breakeven point, even until 3.6 bar.

• A 1D transient valve heat transfer model was established. A bulk peak surface

temperature of ~1000K was noticed, which raises concerns about the coating

durability.

• Coating the piston and firedeck were very rewarding in terms of efficiency

improvement with low charge heating. However, it is not worth coating the liner

clearance due to a minimal efficiency gain.

• At 23.5 bar, covering the piston results in a larger efficiency gain than the firedeck,

but it was the opposite trend when at a load of 6 bar. This is because of a competing

effect between the high substrate thermal conductivity and heterogeneous heat flux.

• Comparing the steel and aluminum piston, the former is likely to have a higher

efficiency due to higher surface temperature. However, when the coatings are

176

applied to the piston, the aluminum piston is likely to experience a higher efficiency

gain due to a larger relative temperature swing.

In summary, low conductivity and low volumetric heat capacity materials are preferred for

any in-cylinder insulation with IC engines, i.e., both conventional and LTC. The optimal

thickness may be different based on the combustion strategy with advanced combustion

strategies generally benefiting from thicker coatings (200μm to ~2mm) and conventional

combustion modes requiring thinner coatings (50 to 200μm).

9.2 Unique contributions

In addition to the conclusions that were listed above, there were some unique contributions

of this thesis that results. These contributions include:

• Proposed and experimentally investigated the effects of thick TBCs on HCCI (with

two different fuels) in terms of engine efficiencies, emissions, load range, intake

heating requirement, and exhaust enthalpies

• Developed a 0D engine cycle thermodynamic model and established a 1D transient

heat conduction model. These two models were then coupled together to

numerically investigate the TBCs’ thermophysical properties and their independent

influence on the thermodynamics of HCCI

• Designed and commissioned a state-of-the-art GCI engine at Clemson University

for the ongoing GCI project with Aramco Services Company as well as future

projects

• Explored gasoline PFS combustion strategy on a light-duty diesel engine at a more

practical boost level (i.e., 1.6 bar) and proposed using a double late injection

177

strategy to successfully lower the peak pressure rise rate by ~50% compared to

HCCI operation.

• Investigated the effect of TBCs on GCI through 0D thermodynamic modeling

coupled with 1D wall heat conduction solver. Additionally, this dissertation

explored and provided preliminary guidance of optimal TBC configuration (i.e.,

material and thickness) for GCI combustion

9.3 Suggestion for future work

• The GCI single-cylinder research engine has been 99% commissioned at Clemson

University. It would be great to finalize the engine commissioning and conduct

experimental work of TBC with GCI combustion strategy.

• The optimal coating configuration, e.g., thickness, could be different for different

chamber components and operating conditions. This necessitates drive cycle level

optimization to determine the best coating candidates.

• It has been shown that the favorable coating properties are different from HCCI to

GCI, and there are many other types of LTC modes such as RCCI, TSCI, etc.

Therefore, new coating configurations will need to be explored with other low-

temperature combustion modes.

• The valve simulation study revealed the potential critical temperature for coating

durability. Thus, a detailed temperature and stress analysis would be required for

the local peak temperature and coating durability. A CFD-FEV collaboration

method can be a good approach to shed light on this area.

178

• A low conductivity and low volumetric heat capacity coatings are desired in any

engine application. The current state-of-the-are coatings are located on the start of

the line of the exponential growth area, where further coating property

improvement could be significantly rewarding.

• The PFS work shows that optimizing the injection strategy could authorize more

potential for controlling the combustion process. However, injection strategy is

only one of the aspects to optimize; therefore, thorough studies on injection angle,

pressure, piston geometry, swirl, etc., are desired to gain more control over the

combustion process.

179

PUBLICATIONS

Thesis publications

1. Yan, Z., Gainey, B., Gohn, J., Hariharan, D., Saputo, J., Schmidt, C., Caliari, F., Sampath, S. and Lawler, B., 2020. The Effects of Thick Thermal Barrier Coatings on Low-Temperature Combustion. SAE International Journal of Advances and Current Practices in Mobility, 2(2020-01-0275), pp.1786-1799.

2. Yan, Z., Gainey, B., Gohn, J., Hariharan, D., Saputo, J., Schmidt, C., Caliari, F., Sampath, S. and Lawler, B., 2021. A comprehensive experimental investigation of low-temperature combustion with thick thermal barrier coatings. Energy, 222, p.119954.

3. Yan, Z., Gainey, B. ., and Lawler, B., “A parametric modeling study of thermal barrier coatings in low-temperature combustion engines.” under review

4. Yan, Z., Gainey, B., Hariharan, D. and Lawler, B., 2020. Improving the controllability of partial fuel stratification at low boost levels by applying a double late injection strategy. International Journal of Engine Research, 22(4), pp.1101-1115.

Other publications

5. Yan, Z., Gainey, B., Hariharan, D., and Lawler, B., "Investigation into reactivity separation between direct injected and premixed fuels in RCCI combustion mode," In ASME 2019 Internal Combustion Engine Division Fall Technical Conference, ICEF2019-7130.

6. Gainey, B., Hariharan, D., Yan, Z., Zilg, S., Rahimi Boldaji, M. and Lawler, B., 2020. A split injection of wet ethanol to enable thermally stratified compression ignition. International Journal of Engine Research, 21(8), pp.1441-1453.

7. Hariharan, D., Gainey, B., Yan, Z., Mamalis, S., & Lawler, B. (2019, October). Experimental study of the effect of start of injection and blend ratio on single fuel reformate RCCI. In Internal Combustion Engine Division Fall Technical Conference (Vol. 59346, p. V001T03A011). American Society of Mechanical Engineers.

8. Gainey, B., Yan, Z., Gohn, J., Boldaji, M. R., & Lawler, B. (2019). TSCI with wet ethanol: an investigation of the effects of injection strategy on a diesel engine architecture (No. 2019-01-1146). SAE Technical Paper.

9. Gainey, B., Gohn, J., Yan, Z., Malik, K., Boldaji, M. R., & Lawler, B. (2019). HCCI with wet ethanol: investigating the charge cooling effect of a high latent heat of vaporization fuel in LTC (No. 2019-24-0024). SAE Technical Paper.

180

10. Hariharan, D., Boldaji, M. R., Yan, Z., Mamalis, S., & Lawler, B. (2020). Single-fuel reactivity controlled compression ignition through catalytic partial oxidation reformation of diesel fuel. Fuel, 264, 116815.

11. Gainey, B., Yan, Z., Rahimi-Boldaji, M., & Lawler, B. (2019, October). On the Effects of Injection Strategy, EGR, and Intake Boost on TSCI With Wet Ethanol. In Internal Combustion Engine Division Fall Technical Conference (Vol. 59346, p. V001T03A006). American Society of Mechanical Engineers.

12. Gainey, B., Yan, Z., Moser, S., Vorwerk, E. and Lawler, B., 2020. Tailoring thermal stratification to enable high load low temperature combustion with wet ethanol on a gasoline engine architecture. International Journal of Engine Research, p.1468087420945960.

13. Gainey, B., Yan, Z. and Lawler, B., 2021. Autoignition characterization of methanol, ethanol, propanol, and butanol over a wide range of operating conditions in LTC/HCCI. Fuel, 287, p.119495.

14. Gainey, B., Yan, Z., Moser, S. and Lawler, B., 2021. Lean flammability limit of high-dilution spark ignition with ethanol, propanol, and butanol. International Journal of Engine Research, p.1468087421993256.

181

REFERENCES

[1]. Heywood, John B. Internal Combustion Engine Fundamentals. New York: McGraw-

Hill Education, 2018.

[2]. Kamo, R. and Bryzik, W., "Adiabatic Turbocompound Engine Performance

Prediction," SAE Technical Paper 780068, 1978, https://doi.org/10.4271/780068.

[3]. Sudhakar, V., "Performance Analysis of Adiabatic Engine," SAE Technical Paper

840431, 1984, https://doi.org/10.4271/840431.

[4]. Kamo, R. and Bryzik, W., "Cummins/TACOM Advanced Adiabatic Engine," SAE

Technical Paper 840428, 1984, https://doi.org/10.4271/840428.

[5]. Wong, V., Bauer, W., Kamo, R., Bryzik, W. et al., "Assessment of Thin Thermal

Barrier Coatings for I.C. Engines," SAE Technical Paper 950980,

1995, https://doi.org/10.4271/950980.

[6]. Assanis, D. and Mathur, T., "The Effect of Thin Ceramic Coatings on Spark-Ignition

Engine Performance," SAE Technical Paper 900903,

1990, https://doi.org/10.4271/900903.

[7]. Kosaka, H., Wakisaka, Y., Nomura, Y., Hotta, Y. et al., "Concept of “Temperature

Swing Heat Insulation” in Combustion Chamber Walls, and Appropriate Thermo-

Physical Properties for Heat Insulation Coat," SAE Int. J. Engines 6(1):142-149,

[8]. Andruskiewicz, Peter, Paul Najt, Russell Durrett, Scott Biesboer, Tobias Schaedler,

and Raul Payri. "Analysis of the effects of wall temperature swing on reciprocating

internal combustion engine processes." International Journal of Engine Research 19,

no. 4 (2018): 461-473.

[9]. Wakisaka, Y., Inayoshi, M., Fukui, K., Kosaka, H. et al., "Reduction of Heat Loss and

Improvement of Thermal Efficiency by Application of “Temperature Swing”

Insulation to Direct-Injection Diesel Engines," SAE Int. J. Engines 9(3):1449-1459,

2016, https://doi.org/10.4271/2016-01-0661.

[10]. Andruskiewicz, Peter, Paul Najt, Russell Durrett, and Raul Payri. "Assessing the

capability of conventional in-cylinder insulation materials in achieving temperature

182

swing engine performance benefits." International Journal of Engine Research 19, no.

6 (2018): 599-612.

[11]. Dec, John E. "Advanced compression-ignition engines—understanding the in-

cylinder processes." Proceedings of the combustion institute 32, no. 2 (2009): 2727-

2742.

[12]. Reitz, Rolf D. "Directions in internal combustion engine research." Combustion

and Flame 1, no. 160 (2013): 1-8.

[13]. Dempsey, Adam B., N. Ryan Walker, Eric Gingrich, and Rolf D. Reitz.

"Comparison of low temperature combustion strategies for advanced compression

ignition engines with a focus on controllability." Combustion Science and

Technology 186, no. 2 (2014): 210-241

[14]. Thring, Robert H. Homogeneous-charge compression-ignition (HCCI) engines.

No. 892068. SAE Technical paper, 1989.

[15]. Eng, J.A., 2002. Characterization of pressure waves in HCCI combustion (No.

2002-01-2859). SAE Technical Paper.

[16]. Saxena, Samveg, and Iván D. Bedoya. "Fundamental phenomena affecting low

temperature combustion and HCCI engines, high load limits and strategies for

extending these limits." Progress in Energy and Combustion Science 39, no. 5 (2013):

457-488.

[17]. Zhao F, Asmus TN, Assanis DN, Dec JE, Eng JA, Najt PM. Homogeneous charge

compression ignition (HCCI) engines. SAE Technical Paper; 2003 Mar 3.

[18]. Haraldsson G, Tunestål P, Johansson B, Hyvönen J. HCCI combustion phasing in

a multi cylinder engine using variable compression ratio. SAE Transactions. 2002 Jan

1:2654-63.

[19]. Sjöberg M, Dec JE, Babajimopoulos A, Assanis D. Comparing enhanced natural

thermal stratification against retarded combustion phasing for smoothing of HCCI

heat-release rates. SAE transactions. 2004 Jan 1:1557-75.

183

[20]. Dec JE, Sjöberg M. Isolating the effects of fuel chemistry on combustion phasing

in an HCCI engine and the potential of fuel stratification for ignition control. SAE

transactions. 2004 Jan 1:239-57.

[21]. Sjöberg M, Dec JE. Smoothing HCCI heat-release rates using partial fuel

stratification with two-stage ignition fuels. SAE Technical Paper; 2006 Apr 3.

[22]. Dec, J., Yang, Y., and Dronniou, N., "Boosted HCCI - Controlling Pressure-Rise

Rates for Performance Improvements using Partial Fuel Stratification with

Conventional Gasoline," SAE Int. J. Engines 4(1):1169-1189,

2011, https://doi.org/10.4271/2011-01-0897.

[23]. Olsson JO, Tunestål P, Johansson B, Fiveland S, Agama R, Willi M, Assanis D.

Compression ratio influence on maximum load of a natural gas fueled HCCI engine.

Sae Transactions. 2002 Jan 1:442-58.

[24]. Dec JE, Dernotte J, Ji C. Increasing the load range, load-to-boost ratio, and

efficiency of low-temperature gasoline combustion (LTGC) engines. SAE

International Journal of Engines. 2017 Jun 1;10(3):1256-74.

[25]. Dempsey, Adam B., Scott J. Curran, and Robert M. Wagner. "A perspective on the

range of gasoline compression ignition combustion strategies for high engine

efficiency and low NOx and soot emissions: Effects of in-cylinder fuel stratification."

International Journal of Engine Research 17, no. 8 (2016): 897-917.

[26]. Yang, H., Shuai, S., Wang, Z. and Wang, J., 2013. Fuel octane effects on gasoline

multiple premixed compression ignition (MPCI) mode. Fuel, 103, pp.373-379.

[27]. Zhang, Y., Pei, Y., Engineer, N., Cho, K. et al., "CFD-Guided Combustion Strategy

Development for a Higher Reactivity Gasoline in a Light-Duty Gasoline Compression

Ignition Engine," SAE Technical Paper 2017-01-0740, 2017, doi:10.4271/2017-01-

0740.

[28]. Pei, Y., Pal, P., Zhang, Y., Traver, M. et al., "CFD-Guided Combustion System

Optimization of a Gasoline Range Fuel in a Heavy-Duty Compression Ignition Engine

Using Automatic Piston Geometry Generation and a Supercomputer," SAE Int. J.

184

Adv. & Curr. Prac. in Mobility 1(1):166-179, 2019, https://doi.org/10.4271/2019-01-

0001.

[29]. Liu, H., Mao, B., Liu, J., Zheng, Z. and Yao, M., 2018. Pilot injection strategy

management of gasoline compression ignition (GCI) combustion in a multi-cylinder

diesel engine. Fuel, 221, pp.116-127.

[30]. Liu, X., Goyal, H., Kook, S., and Ikeda, Y., "Triple Injection Strategies for Gasoline

Compression Ignition (GCI) Combustion in a Single-Cylinder Small-Bore Common-

Rail Diesel Engine," SAE Technical Paper 2019-01-1148, 2019,

https://doi.org/10.4271/2019-01-1148.

[31]. Kalghatgi, G. T., Leif Hildingsson, Andrew Harrison, and Bengt Johansson. "Low-

NOx, Low-Smoke Operation of a Diesel Engine Using ‘Premixed

Enough’Compression Ignition—Effects of Fuel Autoignition Quality, Volatility and

Aromatic Content." In THIESEL 2010 conference on thermo-and fluid dynamic

processes in diesel engines. 2010.

[32]. Cho, K., Latimer, E., Lorey, M., Cleary, D. et al., "Gasoline Fuels Assessment for

Delphi’s Second Generation Gasoline Direct-Injection Compression Ignition (GDCI)

Multi-Cylinder Engine," SAE Int. J. Engines 10(4):1430-1442, 2017,

https://doi.org/10.4271/2017-01-0743.

[33]. Manente, V., Zander, C., Johansson, B., Tunestal, P. et al., "An Advanced Internal

Combustion Engine Concept for Low Emissions and High Efficiency from Idle to

Max Load Using Gasoline Partially Premixed Combustion," SAE Technical Paper

2010-01-2198, 2010, https://doi.org/10.4271/2010-01-2198.

[34]. Sellnau, M., Foster, M., Moore, W., Sinnamon, J. et al., “Pathway to 50% Brake

Thermal Efficiency Using Gasoline Direct Injection Compression Ignition,” SAE Int.

J. Advances & Curr. Prac. in Mobility 1(4):1581-1603, 2019, doi:10.4271/2019-01-

1154.

[35]. Hildingsson, L., Johansson, B., Kalghatgi, G., and Harrison, A., "Some Effects of

Fuel Autoignition Quality and Volatility in Premixed Compression Ignition Engines,"

SAE Int. J. Engines 3(1):440-460, 2010, https://doi.org/10.4271/2010-01-0607.

185

[36]. Kolodziej, C., Sellnau, M., Cho, K., and Cleary, D., "Operation of a Gasoline Direct

Injection Compression Ignition Engine on Naphtha and E10 Gasoline Fuels," SAE Int.

J. Engines 9(2):979-1001, 2016, https://doi.org/10.4271/2016-01-0759.

[37]. Borgqvist, P., Tunestal, P., and Johansson, B., "Gasoline Partially Premixed

Combustion in a Light Duty Engine at Low Load and Idle Operating Conditions,"

SAE Technical Paper 2012-01-0687, 2012, https://doi.org/10.4271/2012-01-0687.

[38]. Lang, O., Salber, W., Hahn, J., Pischinger, S. et al., "Thermodynamical and

Mechanical Approach Towards a Variable Valve Train for the Controlled Auto

Ignition Combustion Process," SAE Technical Paper 2005-01-0762, 2005,

https://doi.org/10.4271/2005-01-0762.

[39]. Borgqvist, P., Tunestal, P., and Johansson, B., "Comparison of Negative Valve

Overlap (NVO) and Rebreathing Valve Strategies on a Gasoline PPC Engine at Low

Load and Idle Operating Conditions," SAE Int. J. Engines 6(1):366-378, 2013,

https://doi.org/10.4271/2013-01-0902.

[40]. Johansson, Bengt. "High-load partially premixed combustion in a heavy-duty diesel

engine." In Diesel engine emissions reduction (DEER) Conference. Chicago, Illinois,

pp. 21-25. 2005.

[41]. Sellnau, M., Foster, M., Moore, W., Sinnamon, J. et al., "Second Generation GDCI

Multi-Cylinder Engine for High Fuel Efficiency and US Tier 3 Emissions," SAE Int.

J. Engines 9(2):2016, doi:10.4271/2016-01-0760.

[42]. Sellnau, M., Hoyer, K., Moore, W., Foster, M. et al., “Advancement of GDCI

Engine Technology for US 2025 CAFE and Tier 3 Emissions,” SAE Technical Paper

2018-01-0901, 2018, doi:10.4271/2018-01-0901.

[43]. Wolk, B., Chen, J.Y. and Dec, J.E., 2015. Computational study of the pressure

dependence of sequential auto-ignition for partial fuel stratification with gasoline.

Proceedings of the Combustion Institute, 35(3), pp.2993-3000.

[44]. Sellnau, Mark, Matthew Foster, Kevin Hoyer, Wayne Moore, James Sinnamon, and

Harry Husted. "Development of a gasoline direct injection compression ignition

(GDCI) engine." SAE International Journal of Engines 7, no. 2 (2014): 835-851.

186

[45]. Kalghatgi, Gautam, and Bengt Johansson. "Gasoline compression ignition

approach to efficient, clean and affordable future engines." Proceedings of the

Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering 232,

no. 1 (2018): 118-138.

[46]. Wolk, Benjamin, and Jyh-Yuan Chen. "Computational study of partial fuel

stratification for HCCI engines using gasoline surrogate reduced mechanism."

Combustion Science and Technology 186, no. 3 (2014): 332-354.

[47]. Reitz, Rolf D., and Ganesh Duraisamy. "Review of high efficiency and clean

reactivity controlled compression ignition (RCCI) combustion in internal combustion

engines." Progress in Energy and Combustion Science 46 (2015): 12-71.

[48]. Kokjohn, Sage L., Reed M. Hanson, D. A. Splitter, and R. D. Reitz. "Fuel reactivity

controlled compression ignition (RCCI): a pathway to controlled high-efficiency clean

combustion." International Journal of Engine Research 12, no. 3 (2011): 209-226.

[49]. Dempsey, Adam B., Scott Curran, and Rolf D. Reitz. "Characterization of reactivity

controlled compression ignition (RCCI) using premixed gasoline and direct-injected

gasoline with a cetane improver on a multi-cylinder engine." SAE International

Journal of Engines 8, no. 2 (2015): 859-877.

[50]. Lawler, Benjamin, Derek Splitter, James Szybist, and Brian Kaul. "Thermally

Stratified Compression Ignition: A new advanced low temperature combustion mode

with load flexibility." Applied energy 189 (2017): 122-132.

[51]. Gainey, B., Yan, Z., Gohn, J., Rahimi Boldaji, M. et al., "TSCI with Wet Ethanol:

An Investigation of the Effects of Injection Strategy on a Diesel Engine Architecture,"

SAE Technical Paper 2019-01-1146, 2019, https://doi.org/10.4271/2019-01-1146.

[52]. Dec, J. and Sjöberg, M., "A Parametric Study of HCCI Combustion - the Sources

of Emissions at Low Loads and the Effects of GDI Fuel Injection," SAE Technical

Paper 2003-01-0752, 2003, https://doi.org/10.4271/2003-01-0752.

[53]. Flowers, D., Aceves, S., Martinez-Frias, J., Smith, J. et al., "Operation of a Four-

Cylinder 1.9L Propane Fueled Homogeneous Charge Compression Ignition Engine:

187

Basic Operating Characteristics and Cylinder-to-Cylinder Effects," SAE Technical

Paper 2001-01-1895, 2001, https://doi.org/10.4271/2001-01-1895.

[54]. Yan, Z., Gainey, B., Hariharan, D., and Lawler, B., "Investigation into reactivity

separation between direct injected and premixed fuels in RCCI combustion mode,"

In ASME 2019 Internal Combustion Engine Division Fall Technical Conference,

ICEF2019-7130.

[55]. Gainey, B., Yan, Z., Hariharan, D., and Lawler, B., "On the Effects of Injection

Strategy, EGR, and Intake Boost on TSCI with wet Ethanol," In ASME 2019 Internal

Combustion Engine Division Fall Technical Conference, ICEF2019-7164.

[56]. Dec, J., Yang, Y., Dernotte, J., and Ji, C., "Effects of Gasoline Reactivity and

Ethanol Content on Boosted, Premixed and Partially Stratified Low-Temperature

Gasoline Combustion (LTGC)," SAE Int. J. Engines 8(3):935-955,

2015, https://doi.org/10.4271/2015-01-0813.

[57]. Dronniou, N. and Dec, J., "Investigating the Development of Thermal Stratification

from the Near-Wall Regions to the Bulk-Gas in an HCCI Engine with Planar Imaging

Thermometry," SAE Int. J. Engines 5(3):1046-1074, 2012,

https://doi.org/10.4271/2012-01-1111.

[58]. Aceves, Salvador M., Daniel L. Flowers, Charles K. Westbrook, J. Ray Smith,

William Pitz, Robert Dibble, Magnus Christensen, and Bengt Johansson. "A multi-

zone model for prediction of HCCI combustion and emissions." SAE transactions

(2000): 431-441.

[59]. Kuo, Tang-Wei, Barry L. Brown, Paul M. Najt, and James A. Eng. "Valve and

fueling strategy for operating a controlled auto-ignition four-stroke internal

combustion engine." U.S. Patent 7,021,277, issued April 4, 2006.

[60]. Powell, Tommy, Ryan O'Donnell, Mark Hoffman, and Zoran Filipi. "Impact of a

yttria-stabilized zirconia thermal barrier coating on HCCI engine combustion,

emissions, and efficiency." Journal of Engineering for Gas Turbines and Power 139,

no. 11 (2017).

188

[61]. Powell, Tommy, Ryan O’Donnell, Mark Hoffman, Zoran Filipi, Eric H. Jordan,

Rishi Kumar, and Nick J. Killingsworth. "Experimental investigation of the

relationship between thermal barrier coating structured porosity and homogeneous

charge compression ignition engine combustion." International Journal of Engine

Research (2019): 1468087419843752.

[62]. O'Donnell, Ryan, Tommy Powell, Mark Hoffman, Eric Jordan, and Zoran Filipi.

"Inverse analysis of in-cylinder gas-wall boundary conditions: Investigation of a

yttria-stabilized zirconia thermal barrier coating for homogeneous charge compression

ignition." Journal of Enginaeering for Gas Turbines and Power 139, no. 10 (2017).

[63]. Filipi, Zoran, Mark Hoffman, Ryan O’Donnell, Tommy Powell, Eric Jordan, and

Rishi Kumar. "Enhancing the efficiency benefit of thermal barrier coatings for

homogeneous charge compression ignition engines through application of a low-k

oxide." International Journal of Engine Research (2020): 1468087420918406.

[64]. Farrell, Alexander E., Richard J. Plevin, Brian T. Turner, Andrew D. Jones,

Michael O'hare, and Daniel M. Kammen. "Ethanol can contribute to energy and

environmental goals." Science 311, no. 5760 (2006): 506-508.

[65]. Saffy, Howard A., William F. Northrop, David B. Kittelson, and Adam M. Boies.

"Energy, carbon dioxide and water use implications of hydrous ethanol

production." Energy conversion and management 105 (2015): 900-907.

[66]. Flowers, D., Aceves, S., and Frias, J., "Improving Ethanol Life Cycle Energy

Efficiency by Direct Utilization of Wet Ethanol in HCCI Engines," SAE Technical

Paper 2007-01-1867, 2007, https://doi.org/10.4271/2007-01-1867.

[67]. Dernotte J, Dec J, Ji C. Efficiency improvement of boosted low-temperature

gasoline combustion engines (LTGC) using a double direct-injection strategy. SAE

Technical Paper; 2017 Mar 28.

[68]. Benajes, Jesús, Antonio García, José Manuel Pastor, and Javier Monsalve-Serrano.

"Effects of piston bowl geometry on Reactivity Controlled Compression Ignition heat

transfer and combustion losses at different engine loads." Energy 98 (2016): 64-77.

189

[69]. Gainey, B., Hariharan, D., Yan, Z., Zilg, S., Rahimi Boldaji, M., & Lawler, B.

(2018). A split injection of wet ethanol to enable thermally stratified compression

ignition. International Journal of Engine Research, 1468087418810587.

[70]. Chang, J., Güralp, O., Filipi, Z., Assanis, D. et al., "New Heat Transfer Correlation

for an HCCI Engine Derived from Measurements of Instantaneous Surface Heat

Flux," SAE Technical Paper 2004-01-2996, 2004, https://doi.org/10.4271/2004-01-

2996.

[71]. Gainey, B., Longtin, J., and Lawler, B., "A Guide to Uncertainty Quantification for

Experimental Engine Research and Heat Release Analysis," SAE Int. J.

Engines 12(5):509-523, 2019, https://doi.org/10.4271/03-12-05-0033.

[72]. Beardsley, M. Brad, Darrell Socie, Ed Redja, and Christopher Berndt. Thick

Thermal Barrier Coatings (TTBCs) for Low Emission, High Efficiency Diesel Engine

Components. No. DOE/OR/22580-1. Caterpillar Inc., PO Box 1875, Peoria, IL 61656-

1875, 2006.

[73]. Hutchinson, J. W., and A. G. Evans. "On the delamination of thermal barrier

coatings in a thermal gradient." Surface and Coatings Technology 149, no. 2-3 (2002):

179-184.

[74]. Sampath, S., W. C. Smith, T. J. Jewett, and Hwan Kim. "Synthesis and

characterization of grading profiles in plasma sprayed NiCrAlY-zirconia FGMs."

In Materials science forum, vol. 308, pp. 383-388. Trans Tech Publications, 1999.

[75]. Srinivasan, V., M. Friis, A. Vaidya, T. Streibl, and S. Sampath. "Particle injection

in direct current air plasma spray: Salient observations and optimization

strategies." Plasma Chemistry and Plasma Processing 27, no. 5 (2007): 609-623.

[76]. Vaidya, A., V. Srinivasan, T. Streibl, M. Friis, W. Chi, and S. Sampath. "Process

maps for plasma spraying of yttria-stabilized zirconia: An integrated approach to

design, optimization and reliability." Materials Science and Engineering: A 497, no.

1-2 (2008): 239-253.

[77]. Finot, M., S. Suresh, C. Bull, and S. Sampath. "Curvature changes during thermal

cycling of a compositionally graded Ni Al2O3 multi-layered material." Materials

190

[78]. Lawler B, Mamalis S, Joshi S, Lacey J, Guralp O, Najt P and Filipi Z. Effect of

operating conditions on thermal stratification and heat release in a homogeneous

charge compression ignition engine. Appl Therm Eng 2017; 112: 392–402

[79]. Dronniou, N. and Dec, J., "Investigating the Development of Thermal Stratification

from the Near-Wall Regions to the Bulk-Gas in an HCCI Engine with Planar Imaging

Thermometry," SAE Int. J. Engines 5(3):1046-1074, 2012,

https://doi.org/10.4271/2012-01-1111.

[80]. Sjöberg, M. and Dec, J., "Combined Effects of Fuel-Type and Engine Speed on

Intake Temperature Requirements and Completeness of Bulk-Gas Reactions for HCCI

Combustion," SAE Technical Paper 2003-01-3173,

2003, https://doi.org/10.4271/2003-01-3173.

[81]. Tree, D., Wiczynski, P., and Yonushonis, T., "Experimental Results on the Effect

of Piston Surface Roughness and Porosity on Diesel Engine Combustion," SAE

Technical Paper 960036, 1996, https://doi.org/10.4271/960036 -Effects of Engine

Variables," SAE Technical Paper 2004-01-1971, 2004, https://doi.org/10.4271/2004-

01-1971.

[82]. Hoffman, Mark A., Benjamin J. Lawler, Orgun A. Güralp, Paul M. Najt, and Zoran

S. Filipi. "The impact of a magnesium zirconate thermal barrier coating on

homogeneous charge compression ignition operational variability and the formation

of combustion chamber deposits." International Journal of Engine Research 16, no. 8

(2015): 968-981.

[83]. Dec, J. and Sjöberg, M., "Isolating the Effects of Fuel Chemistry on Combustion

Phasing in an HCCI Engine and the Potential of Fuel Stratification for Ignition

Control," SAE Technical Paper 2004-01-0557, 2004, https://doi.org/10.4271/2004-

01-0557.

[84]. Chang, J., Güralp, O., Filipi, Z., Assanis, D. et al., "New Heat Transfer Correlation

for an HCCI Engine Derived from Measurements of Instantaneous Surface Heat

Flux," SAE Technical Paper 2004-01-2996, 2004, https://doi.org/10.4271/2004-01-

2996.

191

[85]. Gatowski, J., Balles, E., Chun, K., Nelson, F. et al., "Heat Release Analysis of

Engine Pressure Data," SAE Technical Paper 841359, 1984,

https://doi.org/10.4271/841359.

[86]. Chang, J., Z. Filipi, D. Assanis, T. W. Kuo, P. Najt, and R. Rask. "Characterizing

the thermal sensitivity of a gasoline homogeneous charge compression ignition engine

with measurements of instantaneous wall temperature and heat flux." International

Journal of Engine Research 6, no. 4 (2005): 289-310.

[87]. Lawler, Benjamin, Satyum Joshi, Joshua Lacey, Orgun Guralp, Paul Najt, and

Zoran Filipi. "Understanding the effect of wall conditions and engine geometry on

thermal stratification and HCCI combustion." In ASME 2014 Internal Combustion

Engine Division Fall Technical Conference. American Society of Mechanical

Engineers Digital Collection, 2014.

[88]. Güralp, Orgun, Paul Najt, and Zoran S. Filipi. "Method for determining

instantaneous temperature at the surface of combustion chamber deposits in an HCCI

engine." Journal of engineering for gas turbines and power 135, no. 8 (2013).

[89]. Assanis, D. and Badillo, E., "Transient Heat Conduction in Low-Heat-Rejection

Engine Combustion Chambers," SAE Technical Paper 870156, 1987,

https://doi.org/10.4271/870156.

[90]. Saputo, J., Smith, G., Lee, H.,Sampath, S., "Thermal Swing Evaluation of Thermal

Barrier Coatings for Diesel Engines, " Journal of Thermal Spray Technology, Under

Review.

[91]. Zhou, Yingcong, Deivanayagam Hariharan, Ruinan Yang, Sotirios Mamalis, and

Benjamin Lawler. "A predictive 0-D HCCI combustion model for ethanol, natural gas,

gasoline, and primary reference fuel blends." Fuel 237 (2019): 658-675.

[92]. Lawler, Benjamin, Derek Splitter, James Szybist, and Brian Kaul. "Thermally

Stratified Compression Ignition: A new advanced low temperature combustion mode

with load flexibility." Applied energy 189 (2017): 122-132.

[93]. Lawler, B., Lacey, J., Dronniou, N., Dernotte, J. et al., "Refinement and Validation

of the Thermal Stratification Analysis: A post-processing methodology for

192

determining temperature distributions in an experimental HCCI engine," SAE

Technical Paper 2014-01-1276, 2014, https://doi.org/10.4271/2014-01-1276.

[94]. Zhou, Yingcong, Aimilios Sofianopoulos, Brian Gainey, Benjamin Lawler, and

Sotirios Mamalis. "A system-level numerical study of a homogeneous charge

compression ignition spring-assisted free piston linear alternator with various piston

motion profiles." Applied Energy 239 (2019): 820-835.

[95]. Babajimopoulos, Aristotelis, V. S. S. Prasad Challa, George A. Lavoie, and Dennis

N. Assanis. "Model-based assessment of two variable cam timing strategies for HCCI

engines: recompression vs. rebreathing." In ASME 2009 Internal Combustion Engine

Division Spring Technical Conference, pp. 597-610. American Society of Mechanical

Engineers Digital Collection, 2009.

[96]. Eng, J., "Characterization of Pressure Waves in HCCI Combustion," SAE

Technical Paper 2002-01-2859, 2002, https://doi.org/10.4271/2002-01-2859.

[97]. Liu, Haifeng, Shanju Li, Zunqing Zheng, Jia Xu, and Mingfa Yao. "Effects of n-

butanol, 2-butanol, and methyl octynoate addition to diesel fuel on combustion and

emissions over a wide range of exhaust gas recirculation (EGR) rates." Applied energy

112 (2013): 246-256.

[98]. Gainey, B., Gohn, J., Yan, Z., Malik, K. et al., “HCCI with Wet Ethanol:

Investigating the Charge Cooling Effect of a High Latent Heat of Vaporization Fuel

in LTC,” SAE Technical Paper 2019-24-0024, 2019, doi:10.4271/2019-24-0024.

[99]. Morel, T., Fort, E., and Blumberg, P., "Effect of Insulation Strategy and Design

Parameters on Diesel Engine Heat Rejection and Performance," SAE Technical Paper

850506, 1985, https://doi.org/10.4271/850506.

[100]. Farzinazar, Shiva, Tobias Schaedler, Lorenzo Valdevit, and Jaeho Lee. "Thermal

transport in hollow metallic microlattices." APL Materials 7, no. 10 (2019): 101108.

[101]. Liu, Jinlong, and Cosmin E. Dumitrescu. "Single and double Wiebe function

combustion model for a heavy-duty diesel engine retrofitted to natural-gas spark-

ignition." Applied energy 248 (2019): 95-103.

193

[102]. Gainey, Brian, James Gohn, Deivanayagam Hariharan, Mozhgan Rahimi-Boldaji,

and Benjamin Lawler. "Assessing the impact of injector included angle and piston

geometry on thermally stratified compression ignition with wet ethanol." Applied

Energy 262 (2020): 114528.

[103]. Badra, J., khaled, F., Sim, J., Pei, Y. et al., "Combustion System Optimization of a

Light-Duty GCI Engine Using CFD and Machine Learning," SAE Technical Paper

2020-01-1313, 2020, https://doi.org/10.4271/2020-01-1313.

[104]. Olsson, J., Tunestål, P., Ulfvik, J., and Johansson, B., "The Effect of Cooled EGR

on Emissions and Performance of a Turbocharged HCCI Engine," SAE Technical

Paper 2003-01-0743, 2003, https://doi.org/10.4271/2003-01-0743.

[105]. Yang Y, Dec J, Dronniou N, Sjöberg M, Cannella W. Partial fuel stratification to

control HCCI heat release rates: fuel composition and other factors affecting pre-

ignition reactions of two-stage ignition fuels. SAE International Journal of Engines.

2011 Jan 1;4(1):1903-20.

[106]. Wissink ML, Splitter DA, Dempsey AB, Curran SJ, Kaul BC, Szybist JP. An

assessment of thermodynamic merits for current and potential future engine operating

strategies. International Journal of Engine Research. 2017 Feb;18(1-2):155-69.

[107]. Kalghatgi, G., Risberg, P., and Ångström, H., "Partially Pre-Mixed Auto-Ignition

of Gasoline to Attain Low Smoke and Low NOx at High Load in a Compression

Ignition Engine and Comparison with a Diesel Fuel," SAE Technical Paper 2007-01-

0006, 2007, https://doi.org/10.4271/2007-01-0006.

[108]. Hanson, R., Splitter, D., and Reitz, R., "Operating a Heavy-Duty Direct-Injection

Compression-Ignition Engine with Gasoline for Low Emissions," SAE Technical

Paper 2009-01-1442, 2009, https://doi.org/10.4271/2009-01-1442.

[109]. Manente, V., Johansson, B., and Tunestal, P., "Partially Premixed Combustion at

High Load using Gasoline and Ethanol, a Comparison with Diesel," SAE Technical

Paper 2009-01-0944, 2009, https://doi.org/10.4271/2009-01-0944.

194

[110]. Paz, J., Staaden, D., and Kokjohn, S., "Gasoline Compression Ignition Operation

of a Heavy-Duty Engine at High Load," SAE Technical Paper 2018-01-0898, 2018,

https://doi.org/10.4271/2018-01-0898.

[111]. Zhang, Y. and Sellnau, M., "A Computational Investigation of PPCI-Diffusion

Combustion Strategy at Full Load in a Light-Duty GCI Engine," SAE Technical Paper

2021-01-0514, 2021, https://doi.org/10.4271/2021-01-0514.

[112]. Cung, K, & Ciatti, S. "A Study of Injection Strategy to Achieve High Load Points

for Gasoline Compression Ignition (GCI) Operation." Proceedings of the ASME 2017

Internal Combustion Engine Division Fall Technical Conference. Volume 1: Large

Bore Engines; Fuels; Advanced Combustion. Seattle, Washington, USA. October 15–

18, 2017. V001T03A014. ASME. https://doi.org/10.1115/ICEF2017-3625.

[113]. Alkidas, A.C., 1989. Performance and emissions achievements with an uncooled

heavy-duty, single-cylinder diesel engine. SAE Transactions, pp.137-148.

[114]. Cerdoun, M., Carcasci, C. and Ghenaiet, A., 2018. Analysis of unsteady heat

transfer of internal combustion engines’ exhaust valves. International Journal of

Engine Research, 19(6), pp.613-630.

[115]. Shojaefard, M.H., Noorpoor, A.R., Bozchaloe, D.A. and Ghaffarpour, M., 2005.

Transient thermal analysis of engine exhaust valve. Numerical Heat Transfer, Part A:

Applications, 48(7), pp.627-644.

[116]. Cerdoun, M., Khalfallah, S., Beniaiche, A. and Carcasci, C., 2020. Investigations

on the heat transfer within intake and exhaust valves at various engine speeds.

International Journal of Heat and Mass Transfer, 147, p.119005.

[117]. “GT Power Software” Gamma Technologies, Inc., Westmont, IL, 2018.

[118]. Schaedler T, Webb-HRL P, Andruskiewicz P, Durrett R, Najt P, Potter-GM M.

Temperature-following thermal barrier coatings for high efficiency engines. Timeline.

2017 Jan 1;12(31):2020.

195

APPENDIX A

Figure 84: Fuel conversion efficiency vs. IMEPg for gasoline (solid lines) and wet ethanol (dashed lines)

196

Figure 85: Gross heat release rate (bottom) & pressure trace (top) vs. crank angle degree at different SOI timings

197

Figure 86: (a) Volumetric efficiency and (b) equivalence ratio vs. coating thickness

198

Figure 87: Sweep the properties in order of k, s, and thickness

Figure 88: GHRR & cylinder pressure for SOI sweep at a split fraction of 70

199

Figure 89: Estimated cylinder bulk temperature

Figure 90: Near-wall and bulk gas temperatures

200

APPENDIX B

Figure 91: Detailed drawing for Kistler 6125C cylinder pressure transducer mounting adapter.