Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Injection study of a diesel engine fueled with pilot-ignited, directly-injected natural gas Larson, Chadwick R. 2003

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2003-0258.pdf [ 9.79MB ]
Metadata
JSON: 831-1.0080985.json
JSON-LD: 831-1.0080985-ld.json
RDF/XML (Pretty): 831-1.0080985-rdf.xml
RDF/JSON: 831-1.0080985-rdf.json
Turtle: 831-1.0080985-turtle.txt
N-Triples: 831-1.0080985-rdf-ntriples.txt
Original Record: 831-1.0080985-source.json
Full Text
831-1.0080985-fulltext.txt
Citation
831-1.0080985.ris

Full Text

INJECTION  S T U D Y O F A DIESEL E N G I N E F U E L E D  PILOT-IGNITED,  W I T H  D I R E C T L Y - I N J E C T E D N A T U R A L G A S  By  C H A D W I C K R. L A R S O N B.Sc. in Mechanical Engineering, University o f Alberta, Canada, 1999  A THESIS S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F THE REQUIREMENTS FORTHE DEGREE OF M A S T E R OF APPLIED SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES D E P A R T M E N T OF M E C H A N I C A L ENGINEERING  We accept this thesis as conforming to the required standard  T H E U N I V E R S I T Y OF BRITISH C O L U M B I A January 2003 © Chadwick R. Larson, 2003  In presenting degree freely  this  thesis  in partial  fulfilment  of the requirements  for an advanced  at the University of British Columbia, I agree that the Library shall make it available for reference  copying  of this  department  or  and study. I further  agree that permission for extensive  thesis for scholarly purposes may be granted by his or  her representatives.  It  is  by the head of my  understood  that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department  of  /fl£cLa*;c*(  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  £*fjin*<*k  ABSTRACT The method of fueling diesel engines with high-pressure direct injection (FfPDI) of natural gas reduces regulated emissions while maintaining diesel-cycle efficiency. The relative injection timing between the pilot and natural gas, the absolute injection timing, and the injection pressure were varied in operation of a modified heavy-duty single-cylinder diesel engine. The emissions studied included nitrogen oxides (NOx), carbon monoxide (CO), particulate matter (PM) and total hydrocarbons ( T H C ) . The engine was operated at three speeds and three equivalence ratios. The effects of changing exhaust pressure and boost pressure were also examined. Exhaust back pressure significantly affects all emissions, but not gross efficiency. For consistent effect on emissions, an absolute back pressure of 150 k P a was selected for the standard test procedure. The relative injection timing (RIT) significantly affects most emissions, but does not appear to affect P M or specific fuel consumption. With respect to RIT, burn duration correlates well with emissions. A l l pollutant emissions are at a minimum when the R I T is set to 1.8 ms for the conditions tested. The timing of the 50% cumulative heat release (HR50) is a good variable for relating N O  x  and efficiency to injection timing. The reduction of specific N O x emissions as a function of FIR50 with retarded injection is approximately independent of equivalence ratio. Specific fuel consumption as a function H R 5 0 is nearly independent of load and speed; the best fuel consumption occurs when H R 5 0 is approximately 5° after top dead center. The effects of retarding absolute timing on C O depend upon the equivalence ratio. Total hydrocarbon emissions increase with excessively retarded absolute timing. The effects of timing on P M depend on speed and equivalence ratio. The relationship between N O and efficency and ITR50 does not change with injection x  pressure. Increasing injection pressure appears to increase mixing, which generally improves C O emissions at high equivalence ratio, but worsens T H C and C O emissions at a low equivalence ratio, late timing. Moderately changing engine air flow-rates with a constant equivalence ratio does not significantly affect emissions.  ii  TABLE OF CONTENTS ABSTRACT  1  2  TABLE OF CONTENTS  iii  LIST OF T A B L E S  vi  LIST OF FIGURES  vii  LIST OF S Y M B O L S A N D ABBREVIATIONS  xiii  ACKNOWLEDGEMENTS  xv  INTRODUCTION  1  1.1 1.2 1.3 1.4  1 1 2 2  Preliminary Remarks Motivation for Research Methodology Thesis Overview  BACKGROUND AND OBJECTIVES 2.1 2.2 2.3  2.4 2.5 3  ii  Methane and Diesel Combustion of Pilot-Ignited Direct Injection of Natural Gas Pollution Formation Nitrogen Oxides Total Hydrocarbons ( T H C ) Carbon Monoxide Particulate Matter Engine Variables that Affect Performance and Emissions Previous Studies Objectives  4 4 4 5 6 6 7 7 7 8 11  EXPERIMENTAL METHOD  12  3.1  12 14 17 18 19 21 22 26 26  3.2 3.3 3.4 3.5 3.6 3.7  Experimental Apparatus Auxiliary Systems System Controls Instrumentation Data Acquisition and Analysis Test Matrix Data Processing Calculations Error Analysis Experimental Uncertainty  in  4  5  EXHAUST BACK-PRESSURE  28  4.1 4.2 4.3 4.4 4.5  28 31 34 35 46 46  ABSOLUTE INJECTION TIMING  47  5.1 5.2 5.3 5.4  Performance Nitrogen Oxides Total Hydrocarbons Carbon Monoxide Particulate Matter Summary  48 55 58 60 63 65  RELATIVE INJECTION TIMING  67  6.1 6.2  67 74 75 80 86  5.5  6  6.3  7  Performance Variable Pilot and Natural Gas Injection Performance Emissions Summary  INJECTION PRESSURE  7.1 7.2  7.3  8  Non-Combustion Engine Effects Intake Manifold Pressure Heat Release Emissions Summary Test Procedure  87  Gas Injection Mechanics Combustion Effects Varying Charge A i r Mass Performance Emissions Nitrogen Oxides Total Hydrocarbons Carbon Monoxide Particulate Matter Summary  87 88 89 90 93 93 95 98 100 103  CONCLUSIONS  104  8.1 8.2  104 104 105 105 106 107  8.3  Introduction Conclusions Relative Injection Timing Absolute Injection Timing Injection Pressure and Supercharging Recommendations for Future Study  iv  REFERENCES  109  A  CALIBRATION  112  B  TEST P R O C E D U R E  117  C  LIST OF A C Q U I R E D P A R A M E T E R S  118  D  N A T U R A L G A S PROPERTIES  119  E  CHAUVENET'S CRITERION  121  F  SMOOTHED AND UNSMOOTHED HEAT RELEASE  122  G  BACKPRESSURE  123  H  F U E L S P E C I F I C N O X EMISSIONS  125  I  R E L A T I V E INJECTION TIMING  126  Ll  126  Variable Pilot Injection Timing  J  INJECTION PRESSURE  129  K  M E T H A N E A N D N O N - M E T H A N E EMISSIONS  131  L  C R A N K A N G L E OFFSET EFFECTS  133  M  IMPINGEMENT SIMULATION  136  M.l M.2 M.3 M.4  136 136 136 137  Introduction Initial Condiditon Validation Settings Discussion of Results  v  LIST OF TABLES Table 3.1  Engine Specifications  13  Table 3.2  J-31 Injector Geometric Dimensions  14  Table 3.3  Emission Gas Analyzer Specifications  18  Table 3.4  Engine Instrumentation Specifications  19  Table 3.6  Test Modes: Operating Conditions  22  Table 3.5  Test Matrix  22  Table 3.7  Uncertainty Summary  27  Table 6.1:  Burn Duration Correlations  85  Table D.1  Properties of the Components of B . C . Natural Gas  119  Tabled  Instantaneous and Average Intake Manifold Pressures  123  Table L . l  Measurement Offset Sensitivity for 1200 rpm, +10o G S O I .  135  vi  LIST OF FIGURES Figure 2-1  Typical H D P I heat release rate diagram.  4  Figure 3-1  Westport H P D I T M Injector Schematic  13  Figure 3-2  Engine map of the single cylinder engine  14  Figure 3-3  Engine A i r - F l o w Schematic  16  Figure 3-4  Injection Control Scheme  17  Figure 4-1  Pressure-volume diagram for two back-pressures 1200 rpm, 4g/cycle of air, GSOI +15oATDC.  29  Figure 4-2  Effect of back pressure on engine brake power for different speeds. The power difference is referenced to brake power at minimum back pressure for each speed.  30  Figure 4-3  Exhaust manifold temperature versus back pressure at 1200 rpm, air=4g/cyele, G S O I = - 5 ° .  30  Figure 4-4  Volumetric efficiency for various back pressures at 1200 rpm, 4g/cycle air.  31  Figure 4-5  Intake manifold pressure for 4 back pressures at 1200rpm, 4 g/cycle of air.  32  Figure 4-6  Early stroke intake manifold pressure for 4 back pressures at 1200rpm, 4 g/ cycle of air.  33  Figure 4-7  Intake manifold pressure for 3 speeds, 4g/cycle, back pressure > 150 kPag.  34  Figure 4-8  Comparison of apparent heat release rate for high and low back pressure at 1200 rpm, <b=0.4, G S O I -5°, 4 g/cycle air.  34  Figure 4-9  Heat release for 3 equivalence ratios at 1200 rpm, 4 g/cycle air, G S O I =0°.  36  Figure 4-10  In-cylinder pressure trace for 2 equivalence ratios at 1200 rpm, air, G S O I + 1 5 ° A T D C .  4 g/cycle  36  Figure 4-11  N O x vs. back pressure for 3 equivalence ratios at 1200 rpm, G S O I = - 5 ° , 4 g/ cycle air.  37  Figure 4-12  N O x emissions vs. back pressure for 2 timings, 1200 rpm, <b=0.4, 4 g/cycle air.  38  vii  Figure 4-13  N O x emissions vs. back pressure for 3 speeds, G S O I = - 5 ° , <b=0.4, 4 g/cycle air.  38  Figure 4-14  Total hydrocarbons vs. back pressure for 3 equivalence ratios at 1200 rpm, GSOI=-5o, 4 g/cycle air.  39  Figure 4-15  Total hydrocarbon emissions vs. back pressure for 2 timings, 1200 rpm, <))=0.4, 4 g/cycle air.  40  Figure 4-16  Total hydrocarbon emissions vs. back pressure for 3 speeds, GSOI=-5o, <b=0.4, 4 g/cycle air.  41  Figure 4-17  Carbon monoxide emissions vs. back pressure for 3 equivalence ratios at 1200 rpm, GSOI=-5°, 4 g/cycle air.  42  Figure 4-18  Carbon monoxide emissions vs. back pressure for 2 timings, 1200 rpm, <b=0.4, 4 g/cycle air.  42  Figure 4-19  Carbon monoxide emissions vs. back pressure for 3 speeds, GSOI=-5°, (j)=0.4, 4 g/cycle air.  43  Figure 4-20  Particulate matter emissions vs. back pressure for 3 equivalence ratios at 1200 rpm, G S O I = - 5 ° , 4 g/cycle air.  44  Figure 4-21  Particulate emissions vs. back pressure for 2 timings, 1200 rpm, <b=0.4, 4 gl cycle air.  44  Figure 4-22  Particulate matter emissions vs. back pressure for 3 speeds, GSOI=-5o, ()>=0.4, 4 g/cycle air.  45  Figure 5-1  Comparison of I M E P for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  48  Figure 5-2  Apparent heat release rate for various injection timings at 1200 rpm, 4 g/cycle air, (j>=0.4.  49  Figure 5-3  In-cylinder pressure for various injection timings at 1200 rpm, 4 g/cycle air, (j)=0.4.  49  Figure 5-4  Burn duration for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  50  Figure 5-5  Standard deviation of 50% heat release for various injection timings at 1200 rpm, 144 kg/hr air and 3 equivalence ratios.  51  Figure 5-6  Coefficient of variation of I M E P for various injection timings at 1200 rpm,  51  Vlll  4 g/cycle air and 3 equivalence ratios. Figure 5-7  Apparent heat release rate for various injection timings and 3 speeds at 0=0.4, with 50% Heat Release- + 1 0 ° A T D C .  52  Figure 5-8  Indicated specific fuel consumption for various injection timings at <b=0.4, 4 g/cycle air and 3 speeds, plotted against a) G S O I and b) 50% Heat Release Crank Angle.  53  Figure 5-9  Comparison of I S F C for various injection timings and 3 equivalence ratios, 1200 rpm, 4 g/cycle air.  54  Figure 5-10  Nitrogen oxides emissions for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  56  Figure 5-11  Nitrogen oxides production for various timings and 3 speeds, (j)=0.4, 4 g/cyele air.  57  Figure 5-12  Total Hydrocarbon production for various timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  59  Figure 5-13  Total hydrocarbon emissions for various timings and 3 speeds at <b=0.4, 4 g/ cycle of air.  60  Figure 5-14  Carbon monoxide emissions for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  62  Figure 5-15  Carbon monoxide emissions for various timings and 3 speeds at (J>=0.4, 4 g/ cycle of air.  63  Figure 5-16  Particulate matter emissions for various timings and 3 equivalence ratios at 1200 rpm, 4g/cycle air.  64  Figure 5-17  Particulate matter emissions for various injection timings and 3 speeds at <b=0.4, 4g/cycle air.  64  Figure 6-1  Comparison of apparent heat release rate for various short relative injection timings at 0=0.4, PSOI=-5° A T D C , 1200 rpm.  68  Figure 6-2  Comparison of apparent heat release rate for various extended relative injection timings at 0=0.4, 1200 rpm, G S O I = - 5 ° .  69  Figure 6-3  Comparison of combustion events at 0=0.4, G S O I = - 5 ° A T D C , 1200 rpm at various R I T .  71  Figure 6-4  Burn duration various relative injection timings at 0=0.4, GSOI=-5°, 1200  72  IX  Figure 6-5  Coefficient of variation of I M E P for various relative injection timings at <M).4, G S O I = - 5 ° A T D C , 1200 rpm.  73  Figure 6-6  Standard deviation of combustion events for various relative injection timings at <J)=0.4, GSOI=-5° A T D C , 1200 rpm.  73  Figure 6-7  Illustration of the determination of the variable relative injection timing settings.  74  Figure 6-8  Efficiency for various relative injection timings at <|>=0.4, 1200 rpm, with H R 5 0 timings of + 5 ° and + 1 5 ° A T D C .  75  Figure 6-9  Comparison of burn duration for various relative injection timings at ())=0.4, 1200 rpm, with H R 5 0 timings of + 5 ° and +15° A T D C . 77  77  Figure 6-10  Burn duration for various relative injection timings at two speeds, <b=0.4, HR50=+15°  78  Figure 6-11  Standard deviation of 50% heat release for various relative injection timings at two absolute timings, (b=0.4, 1200 rpm.  79  Figure 6-12  Standard deviation of 50% heat release for various relative injection timings at two speeds, <))=0.4, H R 5 0 = + 1 5 ° A T D C .  79  Figure 6-13  Comparison of N O x production for various relative injection timings at two absolute timings, cb=0.4, 1200 rpm.  81  Figure 6-14  N O x emissions for various relative injection timings at two speeds, (J)=0.4, HR50=+15°ATDC.  81  Figure 6-15  Carbon monoxide and total hydrocarbon emissions for various relative i n jection timings at two absolute timings, (b=0.4, 1200 rpm.  83  Figure 6-16  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two speeds, <b=0.4, H R 5 0 = + 1 5 ° A T D C  83  Figure 7-1  Apparent heat release rate for two injection pressures at, 1200 rpm, 0=0.5, G S O I 0 ° and 4 g/cycle of air.  88  Figure 7-2  Burn duration for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  89  Figure 7-3  Efficiency for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  90  Figure 7-4  Efficiency for two injection pressures at two speeds, <b=0.4, 1200 rpm, 4g/  91  X  cycle of air and various timings. Figure 7-5  Efficiency for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  92  Figure 7-6  Oxides of nitrogen emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  94  Figure 7-7  Nitrogen Oxides emissions for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  95  Figure 7-8  Total hydrocarbon emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  96  Figure 7-9  Total hydrocarbons emissions for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  97  Figure 7-10  Total hydrocarbons at 0=0.4, for two injection pressures, air flowrates, 1200 rpm.  97  Figure 7-11  C O emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  99  Figure 7-12  Carbon Monoxide emissions for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  100  Figure 7-13  Carbon monoxide emissions for two injection pressures at two air flowrates, 0=0.4, 1200 rpm, and various timings.  100  Figure 7-14  Particulate matter emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  101  Figure 7-15  Particulate matter for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  102  Figure 7-16  Particulate matter emissions for two injection pressures at two air flowrates, 0=0.4, 1200 rpm, and various timings.  102  Figure F-1  Unsmoothed heat release trace for mode 7  122  Figure F-2  Smoothed heat release trace for mode 7  122  Figure G - l  N O x emissions vs. backpressure at 800 rpm, 0 0.4 and 2 timings  123  Figure G-2  C O emissions vs. backpressure at 800 rpm, 0 0.6, G S O I - 5 °  124  xi  Figure G-3  Intake manifold pressure versus time for three speeds.  124  Figure H - l  Nitrogen oxides emissions normalized with fuel for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  125  Figure 1-1  Comparison of emissions production at 0=0.4, 1200 at various R I T , G S O I held constant at -5o A T D C .  126  Figure 1-2  Standard Deviation of S O C , 0= 0.4, two speeds  Figure 1-3  Particulate matter emissions at various relative timings at 0=0.4, for 1200 rpm  Figure J-1  N O x emissions for two injection pressures, two speeds and various timings.  Figure J-2  In-cylinder pressure trace for GSOI=0, 6.5 g/cycle air, phi=0.4.  Figure K - l  Methane emissions for various timings, two injection pressures, two equivalence ratios at 1200 rpm.  131  Figure K - 2  Non-methane emissions for various timings, two injection pressures, two equivalence ratios at 1200 rpm.  132  Figure L - l  Apparent heat release near T D C for various offsets at 1200 rpm, P S O I -2o, G S O I +10°, 0 = 0.4, air = 4g/cycle, 7 = 1.32  133  Figure L - 2  Apparent heat release for various offsets at 1200 rpm, P S O I -2o, G S O I +10°, 0 = 0.4, air = 4g/cycle, 7 = 1.35  134  Figure L - 3  Apparent heat release for various offsets at 1200 rpm, P S O I -2o, G S O I +10°, 0 = 0.4, air = 4g/cycle, 7 = 1.32  135  Figure M - l  Contour plots of methane concentration, 15°c.a.after injection at two timings for 0=0.5.  138  Figure M - 2  Contour plots of methane concentration, 15°c.a.after injection at G S O I of 0 ° for 0=0.4.  138  127  128 129 130  xii  LIST OF SYMBOLS AND ABBREVIATIONS SYMBOLS 0  Equivalence Ratio  A/F  A i r to Fuel Ratio (mass)  ABBREVIATIONS BD  Burn Duration  °c.a.  Degrees Crank Angle  CI  Compression Ignition  CO  Carbon Monoxide  C0  2  Carbon Dioxide  COV  Coefficient of Variation  DAQ  Data Acquisition System  EOC  E n d of Combustion  EOI  E n d of Injection  EVC, E V O  Exhaust Valve Closing, Opening,  GPW  Gas Pulse Width  GSOI  Gas Start of Injection  HPDI  H i g h Pressure Direct Injection  HRR  Apparent Heat Release Rate  HR50  50% Heat Release (ca.)  IMEP  Indicated Mean Effective Pressure  ISFC  Indicated Specific Fuel Consumption  IVC, IVO  Intake Valve Closing, Opening  MBT  Timing For M a x i m u m Brake Torque  NO  Nitrogen Oxides  x  OEM  Original Equipment Manufacturer  P j in  Injection Pressure  P  c y l  Cylinder Pressure  PM  Particulate Matter  PPW  Diesel Pilot Pulse Width  PSOI  Pilot Start of Injection  RIT  Relative Injection Timing  SCRE  Single Cylinder Research Engine  SOC  Start of Combustion  SOI  Start of Injection  T D C , A T D C , B T D C - top dead center crank position, After T D C , Before T D C TEOM  Tapered Element Oscillating Microbalance  THC  Total Hydrocarbons  XIV  ACKNOWLEDGEMENTS I first and foremost would like to express my profound gratitude to my supervisors Dr. Philip H i l l and Dr. Kendal Bushe for their time, support, and guidance. Most of what I learned from them extends far beyond this thesis or even engineering. I wish to thank Dr.Sandeep Munshi for his support and assistance throughout the course of this project. I would like to thank the U B C staff of Gordon McTaggart-Cowan and Richard Van Dolder who commissioned and nursed the experimental apparatus. A l s o , Thomas Brakel for commissioning the particulate matter sampling apparatus. I wish to thank Dr. Guowei L i for assistance in my forays i n the simulation realm. I want to express thanks for commraderie, advice, and innumerable contributions of fellow and former students C o l i n Blair, Jorge Lozada, Ray Grout, Conor Reynolds, and Greg Brown. A l s o Katja Lee, who assisted with non-technical aspects of this thesis. I would like to acknowledge N S E R C and Westport Innovations for the financial support which allowed me to pursue a project with positive contributions to the environment. I greatly appreciate the involvement of Dr. Patric Ouellette, who facilitated my research with Westport at U B C . Finally I would like to thank my family for their encouragement and support over the course of many years.  XV  1. INTRODUCTION 1.1  Preliminary Remarks The diesel engine is known for its high efficiency, high torque capabilities, and durability.  For these qualities, diesel engines have become the most prevalent engine type for medium and heavy-duty applications including most commercial ground transportation such as tractor trucks and locomotives. A recent EPA[1] air pollution study in the United States however, found that diesel engines contribute 33% of national nitrogen oxides ( N O ) emissions, and 70% of mobile x  sub-2.5um particulate matter ( P M ) emissions. Studies have shown that diesel particulate emissions are toxic and N O x emissions have adverse health effects in addition to creating photochemical smog and acid rain[l,2]. These emissions are of considerable concern in urban areas with high concentration of traffic. Globally, governments are recognizing the health risks associated with diesel emissions and legislating severe reductions from current diesel emission levels [3].  Methods are under  development for reducing these emissions include exhaust gas recirculation, exhaust aftertreatment, fuel additives, and fuel substitution. Westport Innovations has developed a fueling system that employs pilot-ignited, late cycle, high-pressure direct injection ( H P D I ™ ) of natural gas. It has been shown that Westport's natural gas H P D I ™ system reduces N O , P M , and C 0 x  2  emissions while preserving the thermal efficiency of the diesel engine [4,5,6,7]. The emission reductions in a U S F T P cycle test were 45%, 85%, and 7 1 % for N O x , n m H C , and P M respectively as compared to 1998 E P A emission requirements [7]. These emissions have been shown to be affected by altering injection parameters such as injection pressure and timing[6,7,8].  1.2  Motivation for Research A s diesel engine emissions are known to be detrimental to health and the environment, it  is imperative that engine injection parameters are tuned to minimize emissions. Engine injection parameters that are adjustable without modification are pilot injection timing, gas injection timing, and injection pressure. It is well-known that with injection timing, the trade-off for N O x reduction is decreased efficiency; however, the effects of the timing between the pilot and natural gas are not well understood. Changing the injection pressure w i l l affect the rate of fuel injection and turbulent mixing i n the cylinder and, as the combustion is mostly mixing-limited, this is  1  expected to influence the engine performance and emissions. Many of the previous H P D I studies were conducted with 2-stroke diesel engines and proof-of concept prototype injectors. The capabilities of new H P D I ™ injectors have overcome the limitations of the earlier prototype injectors, where the injector used for this study uses less pilot and provides more control of injection. Supercharging has been shown to improve emissions in diesel engines. However, supercharging C I engines fuelled with directly injected natural gas has been insufficiently explored to date. The general focus of this study is to experimentally improve the fundamental understanding of changing injection parameters on engine performance. 1.3 Methodology The measurements of performance and emissions were conducted with U B C ' s single cylinder research engine ( S C R E ) fueled with pilot ignited H P D I ™ of natural gas. The engine is a 4-stroke, modified Cummins I S X heavy-duty diesel engine, configured to run with Westport H P D I ™ technology, and commissioned to serve as a flexible platform to perform fundamental studies[9].  Using a single-cylinder engine provides a better opportunity than a multi-cylinder  engine to study the effects of changing injection parameters. A single injector simplifies control, and avoids the extra maintenance costs and experimental noise of multiple injectors. The facility is equipped with auxiliary systems for independent control of air supply, exhaust back-pressure, and fuel pressure. The engine system includes instrumentation of all flow rates, temperatures, emissions, and high-speed in-cylinder pressures. This experimental system provides more control over operating parameters than a standard engine with the advantage that controlled experiments can be used to fundamentally investigate the effects of injection parameters on emissions. 1.4  Thesis O v e r v i e w The combustion mechanism for pilot-ignited H P D I engines is explained in chapter 2.  Chapter 2 also outlines pollutant formation and summarizes previous research, which leads into the specific objectives of this study. The experimental apparatus and experimental methodology are explained in chapter 3. The effects of exhaust back-pressure are established in chapter 4 to support a reliable testing procedure. The experimental results of changing injection parameters are presented in the order of absolute injection timing, relative injection timing, and, finally, injection pressure in consecutive chapters. Conclusions and recommendations are offered in chapter 8. The emissions measured include N O , P M , carbon monoxide and total hydrocarbon x  2  emissions. The performance measures are in terms of efficiency and qualitative observations of heat release data.  3  2. BACKGROUND AND OBJECTIVES This chapter provides background knowledge regarding high-pressure direct injection (HPDI) of natural gas in compression ignition engines. The differences between natural gas and diesel fuels are highlighted and the use of pilot-ignition for natural gas is introduced. The combustion mechanics of the H P D I process are subsequently considered. The formation of major pollutant emissions from compression ignition (CI) engines are characterized. A brief literature review is discussed with respect to parametric studies that have been conducted with H P D I . Finally, the research objectives of this study are explained.  2.1  Methane and Diesel The natural gas used in these experiments is composed of 96% methane, 2% ethane, 1 %  nitrogen, 0.5% propane, and smaller amounts of higher hydrocarbons [10]; therefore the properties of methane dominate combustion. Methane has a lower adiabatic flame temperature than diesel; the lower combustion temperatures reduce N O x formation[5].  The stoichiometric air-to-fuel ( A /  F) ratio of this natural gas is 16.8[10] and the stoichiometric ratio for the #2 diesel fuel is approximately 14.4[12]. However, the natural gas has a lower heating value of approximately 49.1 MJ/kg[10], which is higher than the heating value for diesel at approximately 42.8 M J / kg[12]. This means that when using the equivalent of 1 g of diesel energy content, natural gas requires only 14.6 g of air, where diesel requires 14.4 g of air. A s a result, the breathing requirements of a diesel engine fueled with natural gas are very similar. However, methane w i l l auto-ignite within an engine timescale of 2ms only when it is at temperatures of at least 1035 K[13], as compared to 650K for diesel[14]. The maximum in-cylinder temperature for a 19:1 compression ratio engine reaches approximately 900 K , which is why diesel pilot ignition can be used for natural gas H P D I ™ . The diesel pilot w i l l hereafter be referred to as 'pilot'.  2.2  Combustion of Pilot-Ignited Direct Injection of Natural Gas Dumitrescu [8] describes the H P D I combustion mechanism as being the same as the three  phases of diesel combustion: premixed, mixing-controlled and late combustion. A s shown in Figure 2-1, this is an insufficient description for H P D I combustion as the pilot ignition and subsequent delay before gas injection are a notable events in the combustion process. It w i l l be shown that the relative timing of these events is a significant factor in H P D I combustion. These  4  extra processes essentially form 5 major components of the pilot-ignited H P D I combustion process. The first (ab) and second stages (be) of the diesel pilot combustion are identical to normal diesel combustion of  ignition delay and premixed combustion phase as described by  Hey wood [14]. The small quantity of diesel burns rapidly and likely proceeds directly into a  late  combustion phase (c) and mixing of burned products. A n ignition delay of the natural gas occurs as it mixes sufficiently with the air and burned pilot gases, after which the mixture ignites (d). The ignition delay of the natural gas is much shorter than the diesel auto-ignition delay and is not noted in the illustration. Ignition of natural gas transpires v i a local interaction of high temperatures and radicals resulting from the diesel combustion. From this second ignition, the H P D I combustion process mimics diesel process, progressing from pre-mixed combustion (de) to  mixing-controlled combustion (ef) to late burning combustion (fg) phases. The late burning phases occur when partially burned combustion products are relatively slowly oxidized as they mix with oxygen. 3.0  (TDC) Crank angle [°]  PEOI  Figure 2-1  GSOI  GEOI  Typical H D P I heat release rate diagram. ' P ' and ' G ' denote the pilot and natural gas fuels. ' S O I ' and ' E O F refer to 'start' and 'end' of injection.  2.3 Pollution F o r m a t i o n The formation mechanisms of individual emission species in compression ignition engines are essential to describing the effects of changing various injection parameters and operating conditions.  5  2.3.1 Nitrogen Oxides Nitrogen oxides (NOx) are one of the contributors, along with S 0 , to acid rain[2]. 2  Locally, they are toxic compounds and can also photochemically react to form highly toxic smog. A s detailed in Warnatz et al. [15], the four accepted modes of N O formation in C I engines, some of which oxidizes to N 0 . The first and most dominant production mode is a thermal mechanism 2  where an equilibrium system relationships favors N O at high temperatures above 1700K. These reactions are kinetically limited in engines by the short time spent at high temperature conditions. The radicals within the flame provide another mechanism for N O formation, known as the 'prompt' or Fenimore mechanism. The prompt mechanism is less temperature sensitive than the thermal mechanism and involves C H radicals, which occur only in the flame itself. The N 0 2  mechanism for N O formation is also an equilibrium relationship, similar to the thermal mechanism. However, the limiting chemistry is a third body reaction and as such is highly pressure sensitive, but as the activation energy requirements are lower than thermal mechanism, this progresses N O x production at lower temperatures than via a non-thermal mechanism.  The  least dominant mechanism for N O production in C I engines is fuel-borne nitrogen, which is chemically bound in the fuel and is released i n combustion reactions. This mode is the least significant in H P D I fueling as the small amount of nitrogen in the diesel is a negligible component of the total fuel and the nitrogen content in natural gas is not chemically bound, but a molecular constituent^]. The amount to which each mechanism contributes to total N O x production is dependent upon in-cylinder conditions such as temperature and engine speed. Not all N O created is emitted in the exhaust as some N O decomposes after peak cylinder temperature via a reverse of the thermal mechanism [16]. 2.3.2 Total H y d r o c a r b o n s ( T H C ) There are many different hydrocarbons that form in C I engine combustion, some of which are nearly inert, others which react readily to form smog, and others that are carcinogens[14]. Most of the hydrocarbons are from fuel that does not completely. Essentially two major mechanisms are responsible for fuel escaping combustion: over-lean or over-rich at time of flame propagation[14]. In either case, the fuel residuals w i l l oxidize throughout the expansion stroke. Over rich and over-lean conditions can both occur in the combustion zone, however over-leaning is generally the most prevalent in C I engines as the combustion is mostly non-premixed and in overall lean conditions [14]. Other possible sources of unburned hydrocarbons include wall  6  quenching of the flame, and also fuel escaping the injector. Wall temperature can significantly affect T H C concentration, with increased temperatures promoting more oxidation. Unburned hydrocarbons are most prevalent in C I engines at idle, over fueling (acceleration), and at retarded timings with high cyclic variability. [14]  2.3.3 Carbon Monoxide The health risks associated with carbon monoxide (CO) inhalation arise from the preferred attraction of C O to hemoglobin over oxygen, which can cause asphyxiation. The dominant factor of C O production is equivalence ratio (0, defined as the ratio between actual air mass and stoichiometric air mass) regardless of which fuel is burned. A s equivalence ratio is increased, C O production is nominal until 0 approaches unity and then dramatically increases beyond this point[14]. A s such, C O production i n compression ignition engines is primarily produced in the rich core of the jet as fuel is injected[14]. It can also be formed under ultra-lean conditions when the C O oxidation is limited by low temperatures [15].  2.3.4 Particulate Matter Soot, or particulate matter ( P M ) , is considered a respiratory health risk, particularly a lung irritant, and, recently, small particles have also been associated with increased mortality rates [2,17]. Soot is different from unburned hydrocarbons in that it precipitates once exhausted and has a very high carbon-to-hydrogen ratio. It is generally accepted that soot formation occurs in two basic steps of particle formation and particle growth, as summarized by Hey wood [14]. Particle formation results from incomplete combustion as some fuel is pyrolyzed i n the fuel-rich core of a fuel jet, forming soot precursors. These precursors are the nuclei of particle growth. Natural gas is less likely to form soot than diesel in a C I engine[4]. Most of the soot formed during combustion is later oxidized when mixed with adequate air during the expansion stroke, before the exhaust valve opens. Once the burnt gas is exhausted, the soot precipitates as the exhaust stream is mixed with and cooled by ambient air. Some fraction of the particulate matter is attributable to unburned lubricating oil, which is adsorbed by the soot.  2.3.5 Engine Variables that Affect Performance and Emissions There are several parameters that can be adjusted to affect pollution formation from engines: speed, air flow-rate, overall equivalence ratio, and load of the engine, all of which influence combustion and thereby emissions. The fuel injection rate and timing also affect the efficiency and emissions from C I engines[14]. The injection timing affects the conditions in  7  which combustion occurs. The rate of fuel injection influences the mixing rates in the engine and thus affects performance and emissions. A n unusual characteristic of this engine system is adjustable back-pressure, which w i l l be shown to considerably affect emissions. 2.4  Previous Studies M u c h research has been conducted on proof of concept for diesel pilot-ignited, late-cycle  direct injection of natural gas in diesel engines[18,19,20]. This technique is the only dual fuel system which retains the overall performance and thermal efficiencies of a diesel cycle engine[19,21]. Two basic methods have been used to introduce both fuels including mixing the fuels [20, 22], and sequential injection of the pilot[6,20]. Early work using two injectors [18,20] found problems associated with this configuration at low loads, although at high loads the ratio of pilot could be reduced to 2%. Further developments of the injection system to a sequential injection through a single injector improved stability and reduced the amount of diesel [18]. There has been a drive to minimize the pilot in all cases and different configurations influence the amounts of minimum pilot for stable combustion. Ouellette [23] investigated the fundamentals of natural gas injection and combustion interaction between diesel and natural gas with experiments and modelling. Although most of the jet behavior study was conducted with conical injection, the combustion modelling is qualitatively applicable. The simulations showed that optimal combustion occurred when the natural gas jet was delivered through burning diesel (i.e. after diesel auto-ignition), indicating that both timing and injection configuration is important. Detailed simulations by L i et al. [24] and experiments by Dumitrescu [8] confirmed that the configuration of the diesel pilot and gas jet is important for performance. This led to the conclusion that a difference in number of holes between pilot and diesel is required for consistent combustion. A new injector, J-31, was designed by Westport to incorporate electronic control of both pilot and gas injection. The J-31 injector was studied in a six-cylinder, turbocharged Cummins I S X engine for soot investigations by Baribeau [17] and injection schemes by Harrington et al..[7]. McTaggart-Cowan [9] commissioned a single-cylinder research engine ( S C R E ) at U B C i n the interest of studying H P D I of natural gas under more control and with exhaust gas recirculation. The U B C research engine, a modified 6-cylinder Cummins I S X , also employed the J-31 injector.  8  Diesel Injection Studies Extensive timing and pressure sweeps conducted by Stumpp et al. [25] on a turbocharged, 6-cylinder diesel engine showed that N O x increased with increased diesel injection pressure for all timings. There was a N O x versus P M and efficiency trade-off with respect to changing the injection timing. The study presented data showing a lower limit to N O x reduction by retarding timing. The minimum N O x level was increased by increasing injection pressure. However, increasing pressure decreased the P M for all injection timings. A s the engine was turbocharged, air-to-fuel ratios were changing with timing. Studies have shown that supercharging a diesel engine enhances mixing and reduces ignition delay, and the combination of these effects increase the relative amount of diffusive (non-premixed) burning which decreases the N O x production [26,27].  HPDI Absolute Injection Timing Absolute timing studies of an H P D I system vary the pilot and natural gas injections together such that the relative time between them is fixed. Douville [28] conducted timing studies with an injector that mixed diesel and natural gas in a naturally aspirated, 2-stroke, Detroit Diesel 1-71 single cylinder engine. The study found that with retarding injection at low load increased T H C and C O emissions and decreased N O x emissions. A t high load, retarded injection caused decreases in T H C , and N O x with no effect on C O emissions. Dumitrescu [8] investigated timing using a single injector with separate holes for diesel and natural gas and the same engine as Douville [28]. Again, N O x decreased with retarded injection. The study indicated that T H C and C O were almost unaffected by retarding injection. In both studies, the timing was based upon the diesel pilot, which constituted 25-50%, and 15-50%, for Douville [28] and Dumitrescu [8] respectively. Baribeau [17] studied three injection timings for effect on C O and N O x emissions on a turbocharged six-cylinder Cummins I S X engine with a sequential injector. The N O x emissions were consistently reduced by retarding injection, but effects on C O emissions depended on the operating conditions.  HPDI Relative Injection Timing The relative timing between the pilot and natural gas injections is important to combustion in an H P D I engine. Relative injection timing of diesel and natural gas in a conical injection pattern was simulated by Ouellette [23] with K I V A 2 code using simplified combustion chemistry. The results showed that some delay of the natural gas after the pilot injection produced 9  the most favorable combustion for T H C and C O emissions and the highest burn rate. Very limited experimental studies by Wakenell et. al [18] wereconducted with a multi-injector system on the effects of relative injection timing. A s the injector configuration was found to be important for combustion however, the results are not directly relavent to a single injector design. Relative injection timing using a single injector with fuel-specific holes has only been investigated by Dumitrescu [8] at 1200 rpm. Two relative injection timings were attained by changing injector pre-load springs and the change in relative timing was inferred from controller information. The shorter R I T was the minimum possible relative timing for stable operation of the engine. The estimated relative times between the gas and pilot start of injections were 3.7 and 5.9 °c.a.. The shorter R I T was found to adversely affect efficiency and all emissions under all engine conditions tested. Limitations of the injector were acknowledged and further study with more relative injection flexibility was suggested.  H P D I Injection Pressure The effects of injection pressure on engine emissions and performance were also studied by Dumitrescu [8] and Douville [28]. Both studies were conducted with the same single-cylinder engine, operating at one speed. Both studies found increasing injection pressure increased N O x emissions. N o consideration however, was given for combustion occurring earlier due to increased injection rate. The study by Douville [28] showed little change i n efficiency with varying injection pressure between 100 and 140 bar except at high load, where an intermediate pressure (120 bar) gave best thermal efficiency. Dumitrescu [8] found similar results with a sequential fuel H P D I injector when pressure was varied from 100 to 160 bar, where the optimal pressure was found to be 130 bar. Baribeau [17] studied injection pressure with a variable pressure scheme set according to operating conditions, which varied between 14 and 25 M P a . The results showed that the effect of injection pressure on C O emissions was dependent upon engine conditions. Harrington et al. [7] remarked that a variable injection pressure/timing scheme may result in the best possible emissions across different operating conditions. The study also noted that high injection pressure may be required at high speed operation to ensure combustion does not extend too late in the cycle.  Particulate Matter A n injection pressure and timing study of P M was conducted by Baribeau [17]. The test points were taken from the A V L 8-mode test cycle. The results of timing were inconclusive as  10  timing had different effects on P M at different test points.  Increasing injection pressure reduced  P M at low loads, but did not affect P M at high loads. Preliminary P M studies were conducted by Brakel [30] on the S C R E engine that verified the validity of T E O M measurements against gravimetric samples. The investigation also included C O emissions and a brief study of backpressure influence where increasing back-pressure was found to affect both P M and C O emissions. The C O and P M emissions were found to correlate only at some operating conditions. N o other literature was found on P M from direct injection of natural gas in C I engine. A preliminary study of back-pressure on the S C R E by McTaggart-Cowan [9] found no measurable effect of back-pressure on indicated power or N O x emissions. There was no other literature regarding details on effects of back-pressure on emissions or engine operation. Summary The effects of back-pressure warrant further investigation as it is crucial to establishing a reliable test procedure and valid comparison with turbocharged engines. This review of literature indicates that a comprehensive timing/pressure study is required to determine the actual effects of increasing injection pressure, rather than indirect effects arising from changes in injection rate. It is also apparent that P M emissions are poorly understood in natural gas H P D I engines, and as such P M emissions should be included in the timing/pressure study. 2.5 Objectives The objectives of this investigation are summarized as follows: •  Establish the effects of back-pressure as a component of a reliable testing procedure for examining injection parameters.  •  Study the effects of changing the absolute injection timing.  •  Study the effects of changing the relative injection timing to deterimine i f there is an optimum relative timing.  •  Study the effects of changing injection pressure, while accounting for corresponding changes in timing of the combustion event  •  Study the effects of changing the supercharging rates. The performance measures that w i l l be used are the N O x , C O , T H C , and P M emissions, as  well as indicated specific fuel consumption. These measures w i l l be correlated with heat release and in-cylinder pressure data for explanation of effects.  11  3. EXPERIMENTAL METHOD The experiments i n this study were conducted i n the Department o f Mechanical Engineering at the University of British Columbia using a single-cylinder modified Cummins I S X 400 diesel engine, fueled by Westport Innovations, Inc. H P D I technology[9]. Test operating parameters of the engine were computer-controlled through custom-built Westport control system The test engine is instrumented for high-speed in-cylinder data acquisition in.addition to emissions and performance measurements. Measurements were recorded using a computer controlled data acquisition system and the data was analyzed with spreadsheets and Matlab programs developed b y Westport and U B C specifically for this engine.  3.1 Experimental Apparatus The experiments were conducted on the apparatus described i n the Master's theses o f McTaggart-Cowan [9] and Brakel [30]. A new component i n this system is a reciprocating compressor plumbed into the air intake system. The engine is a modified Cummins I S X engine and the original specifications for the engine are listed i n Table 3.1. The engine is factory-altered such that only one cylinder fires. The other five cylinders have blocked valves and holes i n the pistons. The flywheel on the research engine is the largest available to maximize inertia. The air intake manifold, fuel rails and functioning piston are a l l from a production 6-cylinder engine. Alterations were also performed by Westport to enable the engine to run on H P D I . This requires a custom Westport control system that replaces the Cummins electronic controls. A separate fuel rail i n the I S X engine is used for the gas supply. The diesel circulates continually, while the gas rail terminates at the injectors. The dual fuel system requires custom dummy injectors for the non-firing cylinders to prevent diesel from leaking into either the cylinders or the gas rail.  The operating diesel injector was replaced with a Westport H P D I J-31 injector. A  conceptual injector schematic shown i n Figure 3-1. The injector is orientated vertically i n the cylinder and the relative position o f diesel and natural gas jets. In the I S X engine, the injector is centered i n the cylinder and the piston b o w l is toroidal shaped. The geometric characteristics o f the injector are detailed i n Table 3.2.  11  Table 3 . 1 : Engine Specifications Single-cylinder  2.5 L  6-cylinder Rated Power  Displacement Volume  300 kW  (1800 rpm)  Compression Ratio  19:1  Bore  137 mm  Stroke  4  Stroke  169 mm  Valves/  4  6-cylinder Rated Torque  1966 N-m  (1200 rpm)  Cylinder Piston Bowl  Toroidal  (combustion chamber)  Bowl  IVO  Connecting Rod EVO  -3° ATDC  IVC  EVC  ATDC  722° ATDC  f  1  m  ml  Diesel Pilot Figure 3-1  502° ATDC  188°  1  262 mm  Westport HPDI™ Injector Schematic  12  Table 3.2: J-31 Injector Geometric Dimensions Angle from Firedeck  Hole Diameter (mm) +/-0.01  Number of Holes  Total Flow Area (mm )  Diesel  18°  0.12  7  0.079  Natural Gas  18°  0.71  8  3.2  2  3.1.1 Auxiliary Systems To start the engine and to overcome the friction from the other 5 cylinders during low-load conditions, a Baldor Z D M 4110T-5 35kW electric motor provides supplemental torque. A map of the approximate engine operating capabilities is shown i n Figure 3-2. The lower load limit that the S C R E can operate at increases with speed, and high speed, low load operation is not possible. A t higher loads excess power is absorbed by an inductor-type, 150 k W General Electric T G dynamometer. The electric motor drives the engine through a toothed belt attached to the rear shaft o f the dynamometer. Torque is transferred between the engine and the dynamometer through a spider coupling.  20 18 16 14  ro  n_ d.  LU  2  Ingersoll-Rand Compressor  ».«-<P5^  \  • •  12 10  Screw Compres sor  8 6 4 2 0 400  600  800  1000  1200  1400  1600  1800  2000  S p e e d (rpm)  Figure 3-2  Engine map o f the single cylinder engine  The intake air-flow rate is supplied independent o f engine operation with a choice o f two compressors: a Lysholm Technologies 1600AX supercharger screw compressor or an IngersollRand reciprocating compressor. The supercharger is driven by a speed-controlled 35 k W Baldor electric motor. The Ingersoll-Rand has the capacity to supply more air at higher pressures than  13  the screw compressor and an approximation of these capabilities are shown i n Figure 3-2. A schematic o f the air-exchange system is detailed in Figure 3-3. When the intake air is routed through the supercharger, the air passes through a filter before compression, after which it is then cooled by a water-cooled heat exchanger. When the air is compressed by the Ingersoll-Rand, the air is condensed and sent to a storage drum before reaching the test cell. Once the air reaches the test cell, the air is filtered and then the flow rate is controlled with a manual regulator. The condenser after the Ingersoll-Rand reduces the relative humidity o f the intake air from ambient conditions and, as such, emissions cannot be directly compared with screw-compressor engine operation. A three-way valve isolates the compressor air systems from each other. To shield the compressors from pulsations from the engine, a 132L surge tank is located between the three-way valve and the engine intake. The air system is also piped to optionally allow exhaust gas to recirculate into the engine intake flow, but this option was not used for this study. Exhaust gas from the engine is routed through a manifold and then passes through another 132L surge tank which isolates the back-pressure valve and E G R system from the pulsating effects o f the engine. To simulate the back-pressure effects o f a turbocharger, a manually controlled Bernard Electric type O A 8 electric actuator is coupled to a Fisher Posi Seal type A 4 1 , 2" butterfly valve situated downstream o f the surge tank.  14  txhrut TD Emissions Bench  Supffchjrjr  Prcsaic Rclcf VbslEDDiE  tt  45 psD burst dsk  Figure 3-3  Engine A i r - F l o w Schematic  The diesel supply is stored at ambient pressure before it is pressurized by a Dynex PF1001 displacement pump. A mixing valve controls the diesel temperature by mixing hot return diesel with diesel flowed through a heat exchanger and mixing valve. This ensures constant diesel fuel temperatures throughout testing. A custom dome-loaded, self venting G o regulator is used to maintain the gas rail pressure and a +50 psi differential diesel rail pressure. The engine coolant is plumbed for both heating and cooling. Cooling is employed during engine operation and flow is channeled through an external water/coolant heat exchanger for heat dissipation. This system is regulated by an engine thermostat to maintain coolant temperature o f 80°C. Before start-up, coolant is pumped through a heating loop including a 1.5 k W process heater for starting purposes and to reduce warm-up times. T w o other heaters are attached to the engine; a 1.5 k W block heater and a 1.5 k W immersion oil heater. A t lower speeds and loads the engine temperature is slightly lower than 80 °C as not enough heat is generated by the single operating cylinder at these conditions.  15  3.1.2 System C o n t r o l s The experimental apparatus is controlled through a combination o f manual, automatic, and computer controls. The most important controller is the Westport H P D I injection controller. This controller provides independent control o f both pilot and gas injection. The parameters are set manually on a computer interface, timed by milliseconds rather than crank angle. A schematic of the different timings relative to top dead center (TDC)  o f the compression stroke is depicted in  Figure 3-4. The pilot pulse width (PPW) is the commanded time in ms for a cavity to fill with diesel inside the injector. The pilot is then injected into the engine at the commanded pilot start o f injection (PSOI), with respect to T D C in ms. The commanded relative injection timing (RIT) setting determines the time between the start of pilot injection and the start o f gas injection. After the RIT, the gas pulse width (GPW)  is the commanded times for the pulse o f natural gas. The  start o f gas injection (GSOI) is calculated by adding the R I T to the PSOI. A s there are hydraulic effects, the exact start o f injection is slightly delayed after the command. A l l timings in this study are reported relative to top dead center (TDC)  o f the compression stroke, in terms o f absolute time  (ms) or crank angle (ca.).  RIT  GPW  TD c  PPW  PSOI  GS 01*  Control Timing in ms * Calculated Parameter  Figure 3-4  Injection Control Scheme  The electric motor driving the engine is torque-controlled with a potentiometer connected to a Baldor series 18H vector drive. The dynamometer is speed-controlled with a Digalog 1022A PID controller. Supercharger throughput is controlled by a potentiometer connected to a Baldor variable speed drive series 15H. The exhaust back-pressure is set using a potentiometer wired directly to the butterfly-valve actuator. The inlet air temperature is regulated by setting an Omega 77000 series P I D controller to the desired temperature. This controller mitigates the water flow through the intake air heat exchanger.  16  3.2  Instrumentation The new components i n this system include a more sensitive diesel-flow rate  measurement, an intake-manifold pressure transducer, and additional heating elements for the particulate measurement system. A s the critical focus o f this study was emissions, much attention was given to the exhaust sampling and the emission bench. The exhaust gas was sampled just after the exhaust manifold and routed through a 7m stainless steel heated line to the emission bench containing the gas analyzers. The sampling line is heated to prevent water from condensing and particulate matter from precipitating and adsorbing on the sample line walls. The exhaust sample passes through 2 sizes o f heated filters and a heated sample pump before being split. One hot stream is sent directly to a flame ionization detector, the other stream is routed through a condenser and then into the other analyzers. A s such, measurements for the second stream require a conversion from dry-based emissions based on calculated water content. The gases measured and their respective analyzers are summarized i n Table 3.3. The high range C 0  2  analyzer is used for exhaust measurements and the low range C 0 analyzer is used for P M 2  dilution ratio measurements. The C O , C 0 , and N O emission measurements were measured and 2  x  reported i n accordance with S A E recommended practice [32]. Table 3.3: Emission Gas Analyzer Specifications Uncertainty Gas Analyzed CO (Low range^  Principle Non-Dispersive InfraRed  Make California  Model Model  Range 0-2 %  (+/-) 0.04 %  Analytical  100  0 - 10 %  0.2 %  2 (High range) C 0  NDIR  Beckman  880  0 - 20 %  0.2 %  02  Paramagnetic  Siemens  Ultramat 22P  0 - 21 %  0.1 %  CO  NDIR  Siemens  Ultramat 21P  0- 10 000 ppm  20 (0-2000) ppm 100 (2000 - FS)  tHC  Flame ionization detector  Ratfisch  RS 55  0-1000  10  ppm 0- 10 000  ppm 100  Chemiluminescent  API  0 - 3000  0.5 %  ppm  Reading  NOx  200 AH  The physical operating parameters o f the engine are captured by a collection o f thermocouples for temperature, pressure transducers for process pressure measurements, and an eddy-current probe for speed. The intake air-flow rate is determined with either a turbine meter 17  installed before the supercharger or a venturi from the Ingersoll-Rand. High-speed instrumentation include an intake manifold piezo-resistive pressure transducer and an in-cylinder, flush-mounted, piezo-electric pressure transducer, which are referenced against rotary position via an optical shaft encoder. The torque absorbed by the dynamometer is measured through a load cell. The gas-flow rate is determined with a coriolis meter and the diesel consumption is determined through a regression fit of fuel reservoir mass readings. A summary of the nonanalyzer instrumentation employed is shown in Table 3.4. T a b l e 3.4:  Engine Instrumentation Specifications  Instrument  Make  Model  Sensitivity  Range  Uncertainty  Thermocouples  Omega  type-K  0.7 C  0-1370 C  1.1 C  Process  EnergyKinetics  209  n/a  0 - 345 kPa  0.85 kPa  In-cylinder Pressure transducer  AVL  QC33C  < 1.5 kPa  20 M P a  1 % IMEP  Speed (RPM)  Digilog  Incremental Optical Shaft Encoder  BEI  H25D  0.5°  8000 rpm  0.25 degree  Intake  PCB  1501  0.06 kPa  0 - 600 kPa  <0.9 k P a  Dynamometer Load Cell  Artech Industries  20210  0.1 N  0-1.1 kN  N/A  Diesel Mass Scale  Tara  SECount  0.9 mN  44.5 N  *see section 4.2.2  Turbine Air Flowmeter  Superflow  6"  n/a  n/a  1%  Coriolis G a s Flow-meter  Micromotion  custom  n/a  0-15 kg/ hr  2.46%  Venturi pressure transducer  Autotran  Pressure transducers  2 rpm  manifold pressure  3.3  Systems  0.125%  reading -10%  Data A c q u i s i t i o n and Analysis  A computerized data acquisition system (DAQ) captured and recorded all the instrumentation signals using a National Instruments Labview 6i and NiDaq software platforms. The Pentium III 533MHz computer employs a National Instruments 64 channel, 1.25 MS/s, 12-bit  18  P C I - M I 0 - 1 6 E - 1 D A Q card that collects the analogue signals routed through a S C X I 1001 chassis. Two collection processes were employed using the D A Q , one being 45, 720° cycles (4 strokes) o f high-speed acquisition o f the in-cylinder and intake manifold pressures, which were referenced with respect to crank angle. The low-speed acquisition process logged all other analogue instruments at 1 H z . To achieve sufficient accuracy for all measurements, the low-speed data sampling period was 5 minutes for each test point. The associated high-speed data was taken at the beginning and end o f the slow-speed sampling period. Due to o i l temperature variations and torque fluctuations caused by the electric drive system, indicated mean effective pressure ( I M E P ) instead o f brake torque must be used to determine engine load. The I M E P values were obtained from the pre and post-test high-speed data. If these values were within 1% o f each other, only the first sample was retained. Otherwise, the I M E P values were averaged. The D A Q computer also had the capability to calculate real-time heat release based on 20-cycle averages o f in-cylinder pressure data. When this option was enabled, no data could be logged. Diesel flow rate and P M emissions rate were regression-fit calculated from the ' s l o w ' data. A l l other signals are instantaneous measurements which were checked against Chauvenet's Criterion and then averaged. The criteria involves an assumption that the distribution is gaussian and a reading is rejected i f the probability o f obtaining a particular deviation from the mean is less than 0.5*n, where n is the number o f samples. Details are included in Appendix E .  Particulate Matter To measure particulate matter ( P M ) , a portion o f the exhaust stream is routed through a mini-dilution tunnel. The diluted exhaust is sampled at a constant flow rate with a Rupprecht & Patashnick series 1105 Tapered Element Oscillating Microbalance ( T E O M ) . The commissioning o f this instrument is detailed by Brakel [30]. The sample air is drawn through a continuouslyweighed filter which measures real-time mass accumulation from which instantaneous mass rates and concentrations may be derived. The filter is a hydrophobic P F A (teflon)-coated borosilicate glass which, in conjunction with sample filter temperature o f 5 0 ° C (well above ambient conditions), reduces the impact o f humidity changes. The 1.4 factor difference between filter samples and the T E O M reading found by Brakel [30] appears to have been improved to 1.1 by fully heating the sample line to the T E O M instrument [31].  19  Diesel F l o w Rate  The diesel fuel system is a recirculating loop which is characterized by slight fluctuations with time and as such direct measurement of the small net diesel consumption is not possible. Diesel reservoir mass measurements are taken for several minutes and then the data is fit by linear regression to determine diesel consumption. The diesel error component of the total fuel measurement is relatively insignificant when at least 100 samples is taken[29]. By using longer data sets, or at higher diesel fueling rate, the relative uncertainty is reduced. This system uses a scale with finer resolution than used by McTaggart-Cowan [9] and Brakel [30]. 3.4  Test M a t r i x  The tests were divided into four subsets: exhaust back-pressure; absolute injection timing; relative injection timing between the pilot and natural gas; and injection pressure. The parameters that were varied for each subset of test is listed in Table 3.5, including the different engine conditions, called the test modes. The different engine conditions are detailed in Table 3.6. For each test mode, the air and fuel flow rates were held constant. The speeds were varied between 800 and 1600 rpm, and equivalence ratio was varied at 1200 rpm between 0.3 and 0.5. The equivalence ratios are represent most of the range of typical turbo-charged 6-cylinder operation at 1200 rpm. The maximum speed tested, 1600 rpm, is the fastest that the SCRE engine can be ran at medium load. The 23 MPa maximum injection pressure is the maximum attainable pressure for a continuous supply. A l l tests were conducted with a PPW of 0.65 ms which (in theory) allows the same volume of diesel to be injected every cycle regardless of speed/load. The RIT setting of 1.8 ms is a recommended Westport value used for general engine operation. The inlet air temperature at the aftercooler was set to 30°C. Increasing the equivalence ratio for a constant air-flow rate corresponds to an increase in engine load. Holding the charge-air mass per cycle constant at different speeds provides a good comparison of performance and emissions as the in-cylinder conditions are identical between different speeds. A n air-flow rate of 4 g/cycle was selected from the approximate mid-load point of 6-cylinder operation at 1200 rpm. This was extended to other speeds to negate effects due to volumetric efficiency.  20  Table 3.5: Test Matrix  Parameter  Exhaust Back-pressure  Absolute Injection Timing  Relative Injection Timing  Injection Pressure  Test Modes  1-5  1-5  2,3  1-7  PPW (ms)  0.65  0.65  0,65  0.65  RIT (ms)  1.8  1.8  0-5.7  1.8  -5,+10  -10 to+10  -5,+5  -10 to+10  (details: Table 6)  GSOI  (°ATDC)  (approximate) Injection Pressure  19  19,23  19  19  10-190  50  50  50  (MPa) Back Pressure (kPa)  Table 3.6: Test Modes: Operating Conditions Mode Number  Air Flow (g/cycle)  Speed  Equivalence Ratio  1  4  1200  0.3  2  4  800  0.4  3  4  1200  0.4  4  4  1600  0.4  5  4  1200  0.5  6  3  1200  0.4  7  6.5  1200  0.4  3.5 Data Processing Calculations Equivalence Ratio ichiometric ^  "  i i\  (factual  Where A / F is the mass-based air-to-fuel ratio IMEP Due to variation i n oil temperature which affects brake torque, and brake loads that are sometimes negative, I M E P is used to determine engine load. The gross I M E P is used for analysis,  21  which excludes the work in the pumping loop (only considers compression and power strokes). IMEP is determined from: IMEP  jpdV  (Eq. 3.2)  Where p and V are the in-cylinder pressure and volume. In-cylinder pressure was pegged at 180° using intake manifold pressure as suggested by Randolph [33]. The IMEP is calculated at discrete 0.5° intervals of P and dV. A second order discretization was used as follows: IMEP =  "-^  +  (Eq. 3.3)  V  The coefficient of variation (COV) is used to determine the relative variation of cycle to cycle combustion. For example, the C O V of IMEP is calculated as follows: COV {%)  = (o /IMEP)-  IMEP  Where o  1 E M P  100  IMEp  (Eq. 3.4)  is the standard deviation of IMEP over the number of cycles of data.  Heat Release Rate The apparent net heat release rate (HRR) of the in-cylinder gas is the difference between chemical energy released and heat transfer from the cylinder. The net heat release rate is less than the gross heat release rate due to crevice region effects and losses due to heat transfer. The HRR is inferred from the in-cylinder pressure and volume, with ideal gas assumptions, as follows [14]:  HRR = -J- W p  7 - r d0  + _ L _ V^P ( v - l ) d0  (Eq. 3.5)  Where y is the specific heat ratio of the in-cylinder gas and dB is the change in crank angle. A seven point, third order smoothing algorithm was applied for one iteration to the incylinder pressure trace as follows[39]: PrV =  ^ ( - ^ - 3  +  3  ft-2  +  6  /V-l  +  7  /^  +  6  A+l  %>n + 2 - r i + 3)  +  2  (Eq. 3.6)  Where i denotes the iteration. This technique has been shown to give consistent heat release rate[40], and a smoothed and unsmoothed heat release curve are included in Appendix F. The HRR was calculated with a second order discretization using the in-cylinder pressure data and a constant y of 1.30 that approximates the value of the hot combustion gases. B y integrating  22  the H R R over a period from -30° to + 7 0 ° A T D C , the percentage o f heat release, start o f combustion (10% heat release), and burn duration-(10-90% heat release) can also be examined.  Fuel Equivalence For consistency, the mass flow rate o f the natural gas is converted to an equivalent mass flow o f diesel on an energy basis and added to the diesel pilot flow as follows:  m  = m ., +rh.. (  ]  LHVcNG  f  fuel  pilot  r  NG\ j fjy v  diesel'  I  (Eq. 3.7)  \  H  /  Where m is the mass flow and L H V is the lower heating value for the corresponding fuels.  Efficiency The measure for efficiency used i n this discussion is the Indicated Specific Fuel Consumption (ISFC) which essentially only considers in-cylinder combustion/heat transfer effects and neglects mechanical friction. A s such, this is a good measure o f combustion efficiency for comparison o f different speeds. The I S F C is calculated as follows:  isfc = Where F  ind  (Eq. 3.8)  is the indicated power calculated as: IMEP Pind =  -V.-N ZT*—  (Eq- 3.9)  R  n  Where V is the displacement volume o f the cylinder, N is the engine speed and d  is the  number o f crank revolutions per power stroke.  Emissions Most emissions reported i n this study are standardized against indicated power i n indicated specific emissions. The term indicated specific refers to the mass o f pollutant produced per a fixed amount o f indicated energy (kW-hr), which is calculated from the in-cylinder pressure. This standardizes the results so that comparisons can be made at different speeds, loads, and also against different engines. For example, the indicated specific carbon monoxide emissions are calculated as:  23  coU  %¥i^) - Km  --  c  (Eq  As the CO and N O exhaust sampling systems pass through a drier before the analyzers, a x  wet/dry correction factor must be employed as follows: [*/,coJ = IXdryK  1  ( q . 3-1 1) E  - H o) X  2  To correct for humidity changes, the measured NOx value was divided by a correction factor, where the correction factor K is defined as[32]: K =1 + 7 - ( a 0 4 . ^ - 0 . 0 0 4 J  -((0- 1 0 . 7 ) + 1 . 8 ^ - 0 . 1 1 6 - ^ + 0 . 0 0 5 j • ( T  i n t a k e  - 29.4)  (Eq.  3.12)  where F/A is the overall fuel/air ratio and w is the specific humidity of the intake air. Particulate M a t t e r  The rate of particulate matter emitted from the engine is calculated as following:  pm,ex  = -  m  T  J  exhaus,  (Eq. 3.13)  m  dil  m  Where m  pm t e o m  is the measured P M accumulation rate in the T E O M and the mass flow of  raw exhaust through the T E O M ,  m  d i l  is given by:  PQteom dil  Where p and  Q  t e o m  _  = 7^  m  „ , „.  (Eq.  3.14)  are the density and sample flow rate through the T E O M , and the wet  dilution ratio is given by[30]:  Jf^  ^  DR„,. =  1  —  (Eq. 3.15)  ^J -LWf)- 2^  [C0  l  [C0  v  dry'  Where ex, dil, and tot denote the concentrations in exhaust, dilution air, and diluted exhaust respectively[30]. The dry dilution ratio is given by:  ^  DR  =  \co i  -\co\  24  -'  (Eq  3 16)  3  10)  3.6 Error Analysis  The absolute measurement error based on instrumentation uncertainty was calculated for reported values. For a function where the value R respective uncertainties for x x^  =  R(x  h  x% .., x ) n  and w,, w , w 2  n  are the  x , then the global uncertainty for R is determined as  v  n  follows: 0.5 (Eq.3.17)  (0  R  n  L  For example, based on equations the uncertainty in measurement of V is derived from ind  (Eq. 3.9) and (Eq. 3.17) as: Vd  to  = ind  . { 2 • IMEP + u} 2  W  • N} 2  MEp  05  (Eq. 3.18)  ft  r  The maximum measurement error for P  ind  is calculated to "7- 1.1 kW.  3.7 Experimental Uncertainty  To determined the experimental uncertainty for this study, a repeatability test was conducted by repeating two set-points over several test days. The repeatability test was conducted at two operating conditions, 15 times over 5 test days to determine the uncertainty of the measurement due to calibration, experimental procedure, and engine variability. The engine operated for both test points at 1200 rpm, 144 kg/hr of intake air, with 50 kPag exhaust backpressure. One repeatability point was at<b of 0.5 with GSOI of +5°c.a., and the other repeatability point was at <> j of 0.3 and GSOI of -5°c.a. The maximum standard deviation between the two repeatability points is used for the experimental uncertainty. The error is assumed to be normally distributed and a 95% confidence interval (1.96 standard deviations) is used. The maximum measurement and experimental uncertainties of the performance measures are reported in Table 3.7. The repeatability analysis estimates the experimental uncertainty, includes random errors introduced by experimental method, but excludes systematic errors. The experimental uncertainty is more useful when comparing measurements from the same system and, as such, is used for errors in plots. The repeatability uncertainty is generally lower than the instrument uncertainty for these experiments. 25  T a b l e 3.7: Uncertainty S u m m a r y  Parameter  Measurement Uncertainty  Repeatability Uncertainty  NOx (g/kW-hr)  1.1  0.54  tHC (g/kW-hr)  0.12  0.063  tHC (g/hr)  0.85  0.93  CO (g/kW-hr)  0.33  0.22  CO (g/hr)  2.7  6.1  PM (mg/kW-hr)  N/A  2.9  ISFC (g/kW-hr)  15.7  3.1  26  4. EXHAUST BACK-PRESSURE The first set of experimental tests presented are the effects of back pressure on performance and emissions. The effects of exhaust back-pressure on the S C R E intake manifold pressure, heat release, and exhaust emissions were examined to establish a reliable testing procedure for later tests. Back-pressure testing was carried out at each speed and equivalence ratio in the test matrix. The effect of back pressure at different timing was also investigated. The sampling requirements include a minimum back pressure to provide sufficient flow to the emissions instrumentation. The objective of this portion of the study was to find a single exhaust pressure that results in a reliable, repeatable testing state. As exhaust back-pressure is raised, the pressure of the gases in the cylinder at exhaust valve closure (EVC)  is higher and therefore a greater residual exhaust mass is retained by the  engine. This is similar to applying exhaust gas recirculation to the engine. Increasing the exhaust residuals increases the specific heat of the charge air, dilutes the oxygen concentration, and alters the chemical kinetics by changing the concentrations of H 0 and C 0 species[14,34]. These 2  2  changes lessen the amount of nitrogen oxides (NOx) emitted, increase particulate matter ( P M ) formation, as well as limit the amount of oxidation of fuel and carbon monoxide (CO)[34]. A study by Ladommatos[34] isolated chemical, thermal, and dilution effects to determine that the most significant mechanism in P M formation and N O x reduction is the dilution of oxygen concentration. The preliminary study of back pressure by McTaggart-Cowan [9] indicated little effect on N O x or indicated power output within experimental error. It w i l l be shown that adding exhaust back-pressure can induce appreciable changes in emissions. 4.1 N o n - C o m b u s t i o n E n g i n e Effects The mechanical effects of raising back pressure are examined for an operating condition of 1200 rpm, 4g/cycle air, (j) of 0.4, with a G S O I of +15°.  H o w back pressure affects in-cylinder  pressures, exhaust temperatures, efficiencies, and brake power is discussed. A six-cylinder version of this engine usually runs with a turbo-charger, with induced back pressures between 60 and 90% of intake manifold pressure[35]. However, with exhaust gas recirculation ( E G R ) , exhaust back-pressures may be much higher than in a n o n - E G R engine. For these experiments, back pressure was varied from 10 kPag to approximately 180 kPag in small increments below 60  28  kPag and larger increments above 60 kPag. Here and throughout this chapter, the pressures described are gauge pressures unless otherwise indicated. The in-cylinder pressure-volume trace for high and low back pressure is shown in Figure 4-1.  While not visible, the peak cylinder pressure was 3% higher with 180 kPag back pressure  than with 24 kPag back pressure. A s combustion occurs well after top dead center, this likely indicates more mass in the cylinder and, as the fresh air-flow rate is the same, the additional mass must be residual exhaust gas. The indicated power output value is the same within error and, as such, gross efficiency is not affected, which concurs with results of McTaggart-Cowan [9]. Increasing back pressure however, does increase the amount of work done to pump gas out of the engine. A s a result, more work is done by the engine and engine power output is negatively affected as shown in Figure 4-2. The amount of work done by the engine to pump exhaust out is dictated by the characteristics of the both intake and exhaust manifold at different speeds. It appears that the effect of back pressure on power consumption is quite similar between different speeds, except at high back pressures when more power is lost at high speed. It is expected that more power is lost at high speed, based on the same back pressure. The engine lost power because increased work is done to expel the exhaust gas. The work is absorbed by the exhaust gas, resulting in higher exhaust temperatures. A s shown in Figure 4-3, the temperature increase is linear with back pressure increasing, with R value of 0.99 for 0 of 0.4. Changing back pressure 2  also affects the volumetric efficiency of the engine, which decreases slightly with increasing back pressure as shown in Figure 4-4. Qualitatively, the engine was found to run more quietly with the initial application of back pressure. 10000 24 k P a 180 k P a  100 -I 1.000E-04  !  1  1.000E-03 Volume (m )  1.000E-02  3  Figure 4-1  Pressure-volume diagram for two back-pressures 1200 rpm, 4g/ cycle of air, G S O I +15°ATDC.  29  rpm + 800  <P t  o  1200  •  1600  CO  o  CL CO  ±c CO CO  50 Figure 4-2  100  150  Backpressure (kPa-g)  Effect of back pressure on engine brake power for different speeds. The power difference is referenced to brake power at minimum back pressure for each speed.  800 750  Bit  -650 2  600  I  550  o  ffl  ffl  700 0  m  D  ffl  TO m  •  I  I  I  i 3 ' CO  w 450 400 350 300 Figure 4-3  0  50  i  100 Backpressure (kPa-g)  +  0.3  0  0.4  •  0.5  150  Exhaust manifold temperature versus back pressure at 1200 rpm, air=4g/cycle, GSOI=-5°.  30  0.8 0.75 o § 0.7 .o LU  § 0.65 E  I  0.6 0.55 0.5  Figure 4-4  50  100  Backpressure (kPa-g)  150  Volumetric efficiency for various back pressures at 1200 rpm, cycle air.  4g/  4.2 Intake Manifold Pressure Examining the instantaneous intake manifold pressure gives some indication of how back pressure is affecting the engine. Intake airflow was kept constant at 144 kg/hr, injection settings were held constant and only back pressure valve settings were varied. The intake manifold pressure was measured approximately 45 cm from the intake valve at high speed with respect to crank angle as shown i n Figure 4-5. The pressure traces indicate that there are significant sinusoidal oscillations in the intake manifold pressure occurring at the same frequency, regardless of engine back-pressure. Pulsations are expected in the intake manifold due to single-cylinder operation; however, they are exacerbated by a flexible connector near the intake which allows the manifold volume to change. The intake manifold is designed for a 6-cylinder engine and not for single-cylinder operation, and this probably adds a minor effect to the pressure oscillation. The intake-manifold pressure rose with increasing back pressure and the supercharger speed was increased to maintain a consistent airflow.  This implies increased mass in the cylinder with the  same net flow through the cylinder. There is a non-linear increase of manifold pressure with change in back pressure, with an increasing effect at higher back pressure. A s the back pressure is measured downstream from the exhaust manifold, past the surge tank and after several meters of piping, the actual pressure at the exhaust manifold is unknown. To compound this, there are likely fluctuations in the exhaust manifold, which are not measured  31  due to high exhaust temperature: piezo-transducers are not able to withstand the constant high temperatures found in engine exhaust.  Backpressure 11 kPa 55 kPa  - 9 0 kPa 157 kPa  0° is TDC intake stroke 0 Figure 4-5  100  200  300  400  500  Crankshaft Crank Angle (°)  600  700  Intake manifold pressure for 4 back pressures at 1200rpm, 4 g/cycle of air.  There is apparently a perturbation in the intake manifold pressure shortly after top dead center (TDC)  between the intake and exhaust strokes, and another at approximately 140 °c.a.  The  earlier perturbation is larger and is magnified in Figure 4-6 and it is presumed that the second perturbation is a reflection of the first. The perturbation is apparent in all cases except 55 kPag back pressure at approximately 20°c.a. after T D C .  A t back pressures above 55 kPag, the  pulsation increases the intake manifold pressure at +20°c.a.. A possible explanation for the positive pulsation i n the intake manifold is back flow from the cylinder to the intake manifold transducer. With 15 kPag of back pressure, the perturbation is a reduction of intake manifold pressure at the same point, which may correspond to a surge of air into the cylinder from the intake manifold. The perturbations suggest a rapid air exchange between the cylinder and intake manifold at intake valve opening (IVO). A s shown in Figure 4-1, there is relative pressure difference between I V O and the intake stroke (approximately intake manifold pressure) for high and low back pressures. This pressure differential likely causes a pulse of mass transfer of air, or pressure wave, as the intake valve is opened. After some time (approximately 20° at 1200 rpm) this pulsation reaches the intake manifold transducer. A s there is a 5 ° overlap between I V O and  32  exhaust valve closure ( E V C ) , there may also be a direct exchange of mass between intake and exhaust manifolds.  190  Backpressure • 11 kPa 55 kPa 90 kPa 157 kPa 0° is  TDC intake stroke 10 Figure 4-6  20  30  40  Crankshaft Crank Angle (°)  50  60  Early stroke intake manifold pressure for 4 back pressures at 1200rpm, 4 g/cycle of air.  Intake manifold pressure is compared at different speeds in Figure 4-7. The pressure at I V O (near 0°) for 1600 rpm is considerably different from the average pressure in the intake manifold. A t 800 rpm, the difference between average manifold pressure and pressure at I V O is smaller. Based on the manifold pressures at I V O for all speeds, approximating a turbocharger would induce back pressure between 40 and 70 kPag. Because the average manifold pressure is dependent on speed and the instantaneous exhaust pressure is not known, a meaningful differential measurement could not be obtained. The difference between the average manifold pressure and the pressure at I V O would induce a bias i f attempting to set the back pressure based on a difference of average measurements.  Therefore, all effects of back pressure are compared at  gauge values instead of differential values between intake and exhaust.  33  Crank Angle (°)  Figure 4-7  4.3  Intake manifold pressure for 3 speeds, 4g/cycle, back pressure > 150 kPag.  Heat Release  A n examination of heat release at the same baseline case of 1200 rpm, ty 0.4, G S O I -5° A T D C was conducted to determine effects on combustion rate. The heat release rates are shown for maximum and minimum back pressure in Figure 4-8 and no difference is apparent. N o drastic differences were expected as the I M E P is the same for both back pressures. The implication is that differences in emissions are not due to changes in combustion rate.  -0.5  1  -10  Figure 4-8  J 0  •  1  10 20 Crank angle [°]  30  40  I 50  Comparison of apparent heat release rate for high and low back pressure at 1200 rpm, 0=0.4, G S O I -5°, 4 g/cycle air.  34  4.4  Emissions The pollutant presented are the indicated specific emissions of N O x , t H C , C O and P M .  There are more exhaust residuals retained in the cylinder as back pressure is increased. Three sets of data were compared to determine the effects of back pressure on emissions. The back pressure was varied at operating conditions of 1200 rpm, G S O I -5° A T D C , 4g/cycle of air, and equivalence ratios of 0.3, 0.4 and 0.5. The effect of back pressure at different timings was also investigated at 1200 rpm, with G S O I of -5 and +15°c.a., 4g/cycle of air and (j) of 0.4. The relationship of back pressure to speed was investigated with tests at 800, 1200 and 1600 rpm with <J> of 0.4 and G S O I  of -5° A T D C . The effects of speed and timing on emissions are discussed in the next chapter.  Equivalence R a t i o A s injection timing was varied at o) of 0.3, 0.4, and 0.5, it is important to first understand the effects of changing equivalence ratio. A heat release rate (HRR) diagram shown in Figure 4-9 shows the effects of increasing equivalence ratio on combustion. A s <t) was increased, the first 'peak' of combustion, which corresponds to premixed combustion, was not drastically affected. However, as <>\ was increased, the second 'peak' of combustion, which corresponds to mixinglimited combustion, increased in magnitude. This was due to the longer injection of natural gas, which increases the total injection momentum, and thereby total mixing rates. Higher mixing rates and more fuel to burn results in more and stronger mixing-limited combustion, and the premixed combustion does not change. A comparison of the corresponding in-cylinder pressure for <J) of 0.3 and 0.5 is shown in Figure 4-10. The pressure was identical until later in the combustion process, where the pressure for a (J) of 0.5 continues to rise. This is due to greater energy released to the cylinder contents, which contain the energy at higher pressures and temperatures.  35  CO  CD i  ZJ cn  -140.0  -90.0  -40.0  10.0  60.0  110.0  Crank Angle (°CA)  Figure 4-10  In-cylinder pressure trace for 2 equivalence ratios at 1200 rpm, 4 g/cycle air, G S O I + 1 5 ° A T D C .  Nitrogen Oxides The reported N O x emissions are adjusted by an empirical correction factor as noted in section 3.5. This correction factor is based on diesel fuel and incorporate relative humidity and the air-to-fuel ratio. A s natural gas has different stoichiometry than diesel, the correction factor may cause inaccuracies. Increasing back pressure can effect a reduction i n N O x production for equivalence ratios of 0.3 and 0.4 as shown in Figure 4-11. There is some scatter in the data, but a definite downward trend in N O x production is evident as back pressure is increased. However, it appears that there is no effect at an equivalence ratio of 0.5 as the N O x emissions are not affected significantly. The effect of back pressure at different timings is shown i n Figure 4-12. back  36  pressure reduces N O x emissions more for the earlier injection timing. This trend is also confirmed at 800 rpm, found in appendix G . The effects of back pressure on N O x are much smaller than the effects of timings tested. The N O x emissions as a function of back pressure for different speeds are shown in Figure 4-13. N O x is virtually unaffected by increases in back pressure at 1600 rpm; however, a modest effect is seen at 1200 rpm and the effect is more prominent when speed is reduced to 800 rpm. The production of N O x appears to be more affected by back pressure at lower speeds. A possible cause of the reduced effect at higher speeds is due to the reduced time of the valve overlap, which would result in less exhaust residuals retained by the engine. However, noting that back pressure has little effect on low specific emissions of N O x (at cj>=0.5), the reduced effect with speed may simply be consistent for lower levels of specific N O x . Further investigation may truly isolate the effect of back pressure on N O x emissions with respect to speed and timing.  50  Figure 4-11  100  Backpressure (kPa-g)  N O x vs. back pressure for 3 equivalence ratios at 1200 rpm, G S O I = - 5 ° , 4 g/cycle air.  37  12  v_  <t  10  S-  o  f  <> t  GSOI -5°  * +10°  ^ r 6  ><  o  •z.  4[  0  Figure 4-12  i  0  ,  50  ,  100 Backpressure (kPa-g)  ,  150  N O x emissions vs. back pressure for 2 timings, 1200 rpm, <b=0.4, 4 g/cycle air.  30  25  20 rpm + 800  Ix  15  O  o  1200  •  1600  10  o'  0  Figure 4-13  '  50  '  100 Backpressure (kPa-g)  '  1  150  N O x emissions vs. back pressure for 3 speeds, G S O I = - 5 ° , (j)=0.4, 4 g/cycle air.  38  Total H y d r o c a r b o n s  Total hydrocarbon emissions are affected very differently by back pressure as shown in Figure 4-14. Until back pressure exceeds a threshold for each equivalence ratio, the unburned hydrocarbons appear relatively unaffected by back pressure. Once back pressure exceeds this critical point, the amount of unburned hydrocarbons emitted increases dramatically. The effect of different equivalence ratio is similar to NOx, where there is a more pronounced effect at lower equivalence ratios. If a similar amount of excess hydrocarbons escape combustion at each equivalence ratio, the higher exhaust temperature for higher equivalence ratios is expected to oxidize more of the hydrocarbons. As such, the greater effects of back pressure on THC emissions at lower equivalence ratios is not surprising. There also appears to be a leveling off of THC emissions at <J) of 0.3, above 110 kPag back pressure. This levelling event is unexplained. The effect of back pressure on THC emissions at different timings is shown in Figure 415. There is an offset between the emissions at different timings, but the slopes of THC versus back pressure appear somewhat parallel. Therefore, it is assumed the effect of back pressure on THC emissions is similar at different injection timings.  0.8 o  5 0.6 O X  0.4 +  0.2  50  Figure 4-14  100 Backpressure (kPa-g)  4  0.3  o  0.4  •  0.5  150  Total hydrocarbons vs. back pressure for 3 equivalence ratios at 1200 rpm, GSOI=-5°, 4 g/cycle air.  39  1.6 1.4 1.2 q  1  §0.8h ^0.6 0.4 GSOI  + -5°  0.2  o +10° 0  Figure 4-15  50  100 Backpressure (kPa-g)  150  Total hydrocarbon emissions vs. back pressure for 2 timings, 1200 rpm, 0=0.4, 4 g/cycle air.  The effects of back pressure on T H C emissions at different speeds is shown in Figure 416. The response of T H C emissions to back pressure is very similar between 800 and 1200 rpm. However, the back-pressure threshold at which T H C emissions begin to increase appears to be slightly higher for 1600 rpm than for 800. The higher threshold may be due to higher surface temperatures in the cylinder, or perhaps less exhaust residuals in the engine at higher speeds. The T H C emissions rise i n similar fashion in response to increasing back pressure for all speeds. For all conditions tested, back pressure values below 50 kPag do not appear to significantly affect T H C emissions. The increased temperatures associated with higher back pressures should increase oxidation rates, but the exhaust manifold temperatures may be too low to affect T H C emissions at that point. The increases in T H C may be caused by the additional exhaust residuals which affect oxidation as noted earlier. It is not obvious what is causing the increases in T H C emissions.  40  1 _0.8 q |0.6  o I  ~ 0.4 +  0.2  "0 Figure 4-16  50  100 Backpressure (kPa-g)  rpm 800  0  1200  •  1600  150  Total hydrocarbon emissions vs. back pressure for 3 speeds, GSOI=-5°, 0=0.4, 4 g/cycle air.  Carbon Monoxide  a  The C O emissions shown i n Figure 4-17 show a radically different response to back pressure as compared to N O x or T H C emissions. A t 0 of 0.3 and 0.4, the C O emissions are identical and show no significant change with increasing back pressure. However at 0 of 0.5, increased back pressure resulted in a reduction of C O to a level statistically the same as at lower 0. A s this result was unusual, it was repeated and two sets of data are shown in Figure 4-17. Further testing that confirms this effect at low speed and high equivalence ratios is found in appendix G . This effect was also evident at some conditions tested by Brakel [30]. The higher exhaust temperatures corresponding to increased back pressure likely increase C O oxidation. The specific C O emissions appear to approach a similar value at higher back pressures, regardless of equivalence ratio. A s shown in Figure 4-18, there is no discernible effect on C O emissions of changing back pressure at the different timings tested. The effect of back pressure at different speeds on C O emissions is shown in Figure 4-19. There appears to be no effect of back pressure at 800 and 1200 rpm. A t 1600 rpm, C O emissions are higher at low back pressure compared to other speeds, but decline as back pressure is increased. The specific C O emissions appear to approach a similar  41  value at higher back pressures, regardless of speed. It appears that back pressure w i l l decrease C O emissions i f they are above approximately lg/kW-hr for both cases of C O reduction.  50 Figure 4-17  150  Carbon monoxide emissions vs. back pressure for 3 equivalence ratios at 1200 rpm, G S O I = - 5 ° , 4 g/cycle air.  50 Figure 4-18  100 Backpressure (kPa-g)  100 Backpressure (kPa-g)  150  Carbon monoxide emissions vs. back pressure for 2 timings, 1200 rpm, (j)=0.4, 4 g/cycle air.  4 2  1.6 1.4 1.2 1  S 0.8 O O 0.6 rpm 800  0.4 0.2 0  Figure 4-19  0  50  100 Backpressure (kPa-g)  o  1200  •  1600  150  Carbon monoxide emissions vs. back pressure for 3 speeds, G S O I = - 5 ° , cb=0.4, 4 g/cycle air.  Particulate Matter There is no consistent effect of back pressure on P M emissions at any equivalence ratio at 1200 rpm as shown i n Figure 4-20. There is also no discernible effect at equivalence ratios of 0.3 and 0.4. For <]> of 0.5, it appears that back pressure decreases P M emissions with the exception of the highest back pressure tested. There is no significant effect of back pressure on P M emissions by changing the timing at 1200 rpm,  of 0.4 as shown in Figure 4-21. The effect of changing  back pressure at different speeds is shown in Figure 4-22. There does not appear to be any effect of changing back pressure on P M emissions at 800 rpm or 1200 rpm. However, P M emissions at 1600 R P M show differences at 4 back pressures. The P M emissions at 13 and 75 kPag of back pressure are higher than P M emissions at back pressures above 145 kPag. It seems that high amounts of back pressure at high speed can reduce P M emissions, although the trend is not consistent for this testing procedure.  43  20 18  0.3  16  o 0.4 •  14  0.5  ?12 "5)10  1  8  6 a  :>4L  9  4>  4 2 0  Figure 4-20  50  100  Backpressure (kPa-g)  150  Particulate matter emissions vs. back pressure for 3 equivalence ratios at 1200 rpm, GSOI=-5°, 4 g/cycle air.  12  10  "5) 6 E  50  Figure 4-21  100  Backpressure (kPa-g)  150  Particulate emissions vs. back pressure for 2 timings, 1200 rpm, 0=0.4, 4 g/cycle air.  44  25 —  20  rpm 800  o  1200  •  1600  .E 15  "5)  2  10  50  Figure 4-22  100 Backpressure (kPa-g)  150  Particulate matter emissions vs. back pressure for 3 speeds, GSOI= 5°, 0=0.4, 4 g/cycle air.  45  4.5  Summary Increasing back pressure: a) increases exhaust residuals retained by the engine; b) does not affect gross power output, which concurs with McTaggart-Cowan [9]; c) decreases brake power output; d) increases exhaust temperature; e) decreases volumetric efficiency; f) can decrease N O x emissions, most strongly at earlier timings, lower equivalence ratios and lower speeds, with no significant effect at 1600 rpm, <p=0.4; or 1200 rpm, <j)=0.5; g) increases T H C ' s above 50 kPag, and the effect appears to be consistent for different speeds and timings; the effect is reduced at higher equivalence ratios; h) appears to decrease C O i f C O emissions are greater than 1 g/kW-hr (e.g. 1200 rpm, <b=0.5 and 1600 rpm, <))=0.4); i) causes no significant effect on C O i f C O emissions are less than 1 g/kW-hr for all speeds, loads, and timings tested; j) generally does not affect P M , but appears to decrease P M at some conditions, which coincide with back pressure induced C O reductions. The N O x results verify that back pressure can also induce no significant effect as found by  McTaggart-Cowan [9]. 4.5.1 Test Procedure A n objective of this chapter was to establish a back pressure for the test procedure. For emission sampling system requirements and simulation of a turbocharger, exhaust back-pressure is necessary to the system. A s turbocharged engines typically run with a back pressure that can approach 90% of the intake manifold pressure, this is a logical maximum. A turbocharger would in a six-cylinder engine would cause back pressures between 40 and 70 kPag for the conditions tested. It is impossible to choose a back pressure such that all emissions w i l l remain unaffected at all conditions. The effects of back pressure on emissions is the smallest for most conditions when back pressure is less than 50 kPag. With that rationale, the test back pressure was chosen to be 50 kPag, which is between 15 and 40 kPag lower than intake manifold pressure at I V O depending on the speed.  46  5. ABSOLUTE INJECTION TIMING This chapter examines the effects on performance and emissions of changing the absolute timing of the injection. Altering the absolute injection timing was accomplished by shifting the natural gas injection while maintaining a constant relative timing between the pilot and natural gas. The injection timing was primarily retarded past timing of best efficiency to examine nitrogen oxides ( N O ) reduction with corresponding performance and emission effects. The x  effects of injection timing were studied at a constant speed with 3 equivalence ratios (<))), and a constant ty across 3 speeds. Results are compared with experimental findings of Dumitrescu [8]. The experiments were conducted in two sets of tests while gas injection was changed in 5° increments. One set of tests varied timing for ty of 0.3,0.4, and 0.5 at 1200 rpm. The other set of tests varied the timing at ty of 0.4 for 800, 1200, and 1600 rpm. A l l tests presented were conducted with 4 g/cycle of air, an injection pressure 19 M P a , and a setting of 1.8 ms relative timing (RIT) between pilot and gas injection. A s the R I T was fixed i n time, it varied in terms of crank angle with speed. A s the natural gas constitutes between 93-97% of the total energy content of the fuel, the timing of the start of gas injection (GSOI) is more critical than the timing of the diesel, and the timings reported in this chapter are referenced to G S O I . The absolute timing is hereafter simply referred to as the injection timing. The criterion used for efficiency comparison is the Indicated Specific Fuel Consumption (ISFC). The emissions reported are nitrogen oxides ( N O ) , total hydrocarbons (THCs), carbon monoxide (CO), and particulate matter x  (PM).  Comparisons with earlier H P D I timing studies and general diesel timing studies are included for comparison of effects on emissions. A s injection is retarded past optimum efficiency a combination of effects occur, including reduced thermodynamic efficiency and lower maximum in-cylinder pressures. A reduction in maximum in-cylinder pressure corresponds to a reduction in maximum in-cylinder temperatures. The lower temperatures are due to combustion later in the expansion stroke, which allows the burned gases to expand in a greater volume. The more retarded the timing, the further the decrease in efficiency. The late combustion reduces the time between combustion and exhaust valve opening ( E V O ) , decreasing time available for kinetics of oxidation to occur. Late combustion in diesel engines generally decreases N O , but increases unburned fuel, intermediate x  species such as C O and P M , and exhaust temperatures[14].  47  5.1  Performance A s timing of the gas injection is changed, there are changes in the combustion and power  output. A s shown in Figure 5-1, as timing is retarded for a constant fuelling rate, there is a decrease in engine output. This is expected as burning the fuel later in the cycle reduces the peak pressure and the expanding gas only does work to the piston through part of the stroke. The variation in output at a given injection timing is due to differences in the fuelling rate, as the instrument error for I M E P is less than 1%. Through study of the apparent heat release rate (FIRR) shown in Figure 5-2, it is observed that heat release rates increase and burn durations decrease with retarded timings. The corresponding in-cylinder pressure curves are depicted in Figure 5-3. A comparison of burn duration at different equivalence ratios and injection timings shown in Figure 5-4 confirms the general trend of shorter burn durations at later timings for all equivalence ratios. The burn duration is the timing between 10% of total heat release and 90% of total heat release. The greater pressure difference between the injector and in-cylinder at late timings may result in higher mixing rates. However, the gas injection duration is actually longer at late timings, which implies slower injection and lower mixing rates. The explanation for the increased heat release rate with shorter burn duration at retarded timings may be due to different injection rate shapes, where for retarded timings more gas is injected later and at higher injection rates. 1100,  ,  .  ,  .  ,  ,  ,—  <> t  1000  0.3  900  •  800  0.4 0.5  700 "co  £  600  D_  LU  500 400 300 200 100 0  Figure 5-1  0  5 10 15 20 Gas Start of Injection (circATDC)  25  Comparison of I M E P for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  48  160  i  140 T n  |  y% !  \  i  120  i  /  80  o. o  60  J  _  _  ;  :  i  I  (GSOI (°CA)  1  r\  y  100 w w  .J  !—-V\  1  \\  1  8> 1  40 20  * '• "  0 -60  1  -40  -20  0  i  i  i  i  '  20  40  60  80  100  Crank Angle [°CA] Figure 5-3  In-cylinder pressure for various injection timings at 1200 rpm, 4 g/ cycle air, 0=0.4.  49  5  tjl  Figure 5-4  1  -5  1  0  1  1  .  1  5 10 15 20 50% Heat Release ("ATDC)  1  25  1  Burn duration for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  The variation of the start of combustion is shown i n Figure 5-5, where the start of combustion (SOC) is defined as 10% heat release. There is some scatter i n the data, and no consistent trend is present for <() of 0.3. The prominent scatter for <> | of 0.3 is probably due to the pilot comprising almost 10% of the total energy. For each equivalence ratios of 0.4 and 0.5, the variability appears to generally increase with retarded timings. The increased variation with retarded timing is likely due to combustion occurring later in the stroke and therefore at lower incylinder temperatures. The variation in engine output by coefficient of variation ( C O V ) of I M E P is shown in Figure 5-6. There is no discernible effect on variation in engine output for § of 0.4 and <)) of 0.5. The large scatter and slightly higher variability for § of 0.3 is likely due to the short gas injection duration for low fuelling rates. The C O V decreased with retarded injection, which may be explained by the increasing gas injection duration with retarded injection, which allows more stable injector operation.  50  o 0.25  o x  0.2  C c o  x V  |o.15  o o o  V  co  I  o  0.1  V  V V  M.  -cr  X  O V  0.3 0.4 0.5  0.05  -10  Figure 5-5  -5  0  5  Gas Start of Injection (°ATDC)  10  15  Standard deviation of 50% heat release for various injection timings at 1200 rpm, 144 kg/hr air and 3 equivalence ratios.  3.5  r  X  O V  0.3 0.4 0.5  3-2.5  S  2  CO  >  o c 1.5  'o  "S3 o O  O O V  V  0.5  0i  Figure 5-6  1  1  1  1  1  1  -5  0  5  10  15  20  50% Heat Release (°ATDC)  1  1  25  Coefficient of variation of I M E P for various injection timings at 1200 rpm, 4 g/cycle air and 3 equivalence ratios.  51  A n examination of the H R R shown in Figure 5-7 for different speeds gives information about combustion intensity and burn duration. For each case, the engine had the same amount of air and fuel injected per cycle. With respect to intensity, the maximum rate of heat release appears to be similar for all speeds. The 800 rpm case has a burn duration of 19° (3.96 ms) which is significantly lower than the 1200 and 1600 rpm cases which have B D ' s of 25° (3.47ms) and 2 9 ° (3.02 ms) respectively. The 1600 rpm case appears to burn the fastest on an absolute time basis. A s the fuel is injected at a constant rate regardless of speed, the faster burn implies that the mixing rates are higher at increased engine speeds. A s the burn duration is slower in terms of crank angle, the mixing rate is not directly proportional to speed. This is not surprising as the injection momentum of the natural gas injection contributes significantly to in-cylinder mixing.  2.5  800 rpm 1200 rom  2.0 ro k_ 0) (0  ra Q> <u k_ +->  ra a>  JZ  a>  -1600 rpm  1.5 •  1.0  •  0.5  j |  \ . :  y  / /  ; /  ^ }j\ >  / 1  0.0  •  •0.5 -20  -10  0  10  20  30  40  50  crank a n g l e [°]  Figure 5-7  Apparent heat release rate for various injection timings and 3 speeds at <|)=0.4, with 50% Heat Release- + 1 0 ° A T D C . G S O I : 800 rpm = +5°; 1200 rpm = 0 ° ; 1600 rpm = -5°.  It can also be inferred from the different injection timings depicted i n Figure 5-7 that the centroids of injection and heat release w i l l shift with respect to G S O I as speed is changed. This is because the injection rate is constant on an absolute time basis, but the injection crank duration gets longer as speed is increased. There may also be a hydraulic delay between commanded G S O I and true G S O I , which w i l l exacerbate the shift. This means that the engine performance w i l l change for a fixed G S O I across different speeds, as the timing of the combustion event influences engine performance. The timing of the cumulative 50% heat release (HR50) approximates the center of combustion and, as such, the H R 5 0 was chosen as the independent  52  variable for efficiency comparison. The effects of injection timing on I S F C at different speeds are shown in Figure 5-8. The I S F C is plotted against both G S O I and H R 5 0 and the trends for each speed collapse onto one another when plotted against H R 5 0 , but not for G S O I . This illustrates a relationship between the H R 5 0 timing and efficiency. The indicated fuel consumption w i l l be lower than the brake fuel consumption as indicated power neglects friction and auxiliary engine loads. B y neglecting these parasitic losses however, indicated efficiency can provide fundamental comparisons between different engine conditions. This provides knowledge for better understanding of the pilot ignited H P D I natural gas combustion process.  220 200  m  180 160 140 120 100 80 60 40 20 0  -10  -5  0  5  -*-  RPM 800  •  1200  V  1600  10  15  G a s Start of Injection (°ATDC)  220 200 180 160 140  i  CO  120 100 RPM  80 60 40 20 0  -5  0  5  10  15  -*-  800  •  1200  V  1600  20  25  5 0 % Heat R e l e a s e (°ATDC)  Figure 5-8  Indicated specific fuel consumption for various injection timings at <|)=0.4, 4 g/cycle air and 3 speeds, plotted against a) G S O I and b) 50% Heat Release Crank Angle.  53  The most efficient performance appears to be obtained when the H R 5 0 occurs at approximately 5+° A T D C , which appears to be the local minimum for specific fuel consumption. While running at this optimum timing, 1600 rpm appears slightly more efficient than 800 rpm, but this difference is eliminated as injection is retarded. The efficiency difference at optimum timing is likely due to different heat transfer at different speeds. The peak heat transfer to the piston and cylinder walls occurs at the high peak temperatures and pressures associated with combustion hear piston top dead center ( T D C ) . The transition through this point is faster at higher speeds and there is less heat lost because less time is available for heat transfer. Increased speeds however, w i l l introduce more mechanical friction and reduce brake efficiency. It appears that changing the speed for injection timings retarded past optimum did not affect the efficiency significantly. A s equivalence ratio is increased, the centroids of injection and combustion shift later in the cycle. For this reason, H R 5 0 is used to compare efficiency different equivalence ratio. A s shown in Figure 5-9, the indicated specific fuel consumption is statistically within error for each equivalence ratio tested at 1200 rpm. In terms of gross efficiency performance, there are no significant differences between different fuelling rates.  200  _j_ _§  180 160 140 ^ 120 — 100  o  LL  52 80 60  *  <> t  •  0.4  V  0.5  40 20 0 Figure 5-9  -5  0  5 10 15 20 50% Heat Release (°ATDC)  0.3  25  Comparison of I S F C for various injection timings and 3 equivalence ratios, 1200 rpm, 4 g/cycle air.  54  5.2  Nitrogen Oxides This section examines the effects of injection timing on N O emissions with particular x  consideration given to the timing of combustion. It is widely accepted that peak cylinder temperature is the most critical factor governing N O production in C I engines [14]. For x  combustion occurring after T D C , the H R 5 0 may correspond to the timing of maximum cylinder temperature as both pressure and temperature drop after combustion as the cylinder volume expands. The timing of 50% heat release w i l l approximate the centroid of combustion for retarded timings, whereas G S O I is merely an indication of where injection begins. Plotting N O emissions as functions of H R 5 0 and G S O I are compared in Figure 5-10. The specific N O  x  x  emissions appear as independent trends for each (j) plotted against G S O I , but when plotted against H R 5 0 they collapse within error into a single trend. This indicates that FfR50 is indeed a meaningful independent variable for comparing N O emissions for an H P D I engine. For x  consistency, all further emissions comparisons are made using H R 5 0 instead of G S O I as the independent variable. For all loads tested, there was a decrease of N O emissions with retarded timing as shown x  in Figure 5-10. A s the thermal mechanism of N O production dominates i n most engine x  applications, the lower peak temperatures associated with retarded timing likely influence the reduction of N O emissions. However, as the 50% heat release is increased beyond 15° A T D C , x  there are no further significant reductions in specific N O . This agrees with the data found in x  Dumitrescu [8], where brake specific N O emissions at high load showed a lower limit for x  retarded injection. It seems there is a limit to the extent which N O emissions can be mitigated by x  retarding gas injection. . A s specific N O emissions appear to be the same regardless of equivalence ratio, this x  implies that changing the <t) (at least between 0.3-0.5) does not affect N O production. The x  amount of N O emitted however, is roughly proportional to the amount of fuel burned for a fixed x  H R 5 0 . This is true as both specifc N O emissions and specific fuel consumption do not x  significantly change with equivalence ratios at a constant H R 5 0 for the conditions tested . A confirming figure is found in Appendix H . This proportionality could be due to the largely mixing-limited combustion in H P D I engines, which proportion is consistent for all timings. The relationship between N O production and H R 5 0 could be examined with diesel fuel to see i f the x  55  relationship holds true for diesel. The apparent collapse of N O production at different x  equivalence ratios onto a single trend is unique to gross indicated specific power, which references against zero power. The trend w i l l not collapse with brake specific power as there is a constant amount of mechanical friction and auxiliary engine loads at any given speed, although the H R 5 0 timing beyond which no further reductions in N O w i l l be the same. x  20  r  *  18  X  16  •  0.4  14  V  0.5  0.3  a  J= 12 i  X  O 8  -10  -5 0 , 5 10 Gas Start of Injection (°ATDC)  15  20 18 --16  1  T  ,- -T  j  i.  14  *  j= 12 |  —  1°  ...  !  1: i  X  O 8  \f~T~~  ~i  6  !_ i  *  i  w  i_  j  r  n  4 i  2 0 Figure 5-10  -5  0  r :]:  ft  f  i  I f  5 10 15 20 50% Heat Release (°ATDC)  if  25  Nitrogen oxides emissions for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air (a) plotted against Gas Start of Injection (b) plotted against H R 5 0  56  The N 0 emissions were also observed to be dependent on speed changes. A s shown in X  Figure 5-11, the N O production for (J) of 0.4 is significantly higher at 800 rpm than at 1200 or x  1600 rpm. This is an expected result as the kinetics of N O formation are limited by less x  available time at high temperatures at faster speeds, even though in-cylinder surface temperatures are higher. A l s o , the mixing rates are lower at lower speeds, and as such, the products of combustion take longer to mix with cooler excess air. It also appears that N O is reduced as x  H R 5 0 is retarded to 2 0 ° A T D C at 1600 rpm. A s no N O reduction is gained at 1200 rpm beyond x  15° A T D C , the H R 5 0 limit for N O reduction via retarded injection appears to increase with x  speed. The N O emissions at 1600 rpm were biased compared to lower speeds as the intake x  manifold temperature rose to 31°C for the 1600 rpm case, as compared to 25°C for 800 and 1200 rpm. Increasing the intake charge air temperature w i l l result in more N O emitted[14]. The x  aftercooler air outlet temperature was set to 30°C, but at lower speeds (and airflow rates), the inlet air was cooled throughout the piping. The difference between speeds may be shown to be more proportional with tests at a different aftercooler setting. A s the piping is ambient temperature, it would be appropriate to reduce the aftercooler setting to approximately 2 5 ° C .  RPM 800 •  1200  V  1600  ^ 20  ^5 Figure 5-11  0  5 10 T5 20 50% Heat Release (°ATDC)  Nitrogen oxides production for various timings and 3 speeds, <j)=0.4, 4 g/cycle air.  57  5.3  Total H y d r o c a r b o n s The total hydrocarbon emissions are an indication of relatively how much fuel is escaping  combustion. For comparison of T H C emissions at different injection timings, both the raw T H C emissions and normalized emissions are plotted in Figure 5-12. The first observation that can be made is that raw levels of T H C are similar across equivalence ratios for early injection timings and decline modestly as injection is slightly delayed. The T H C emissions begin to rise significantly as injection is further delayed, and this effect is more pronounced for low § than for high <(). This compares well with Dumitrescu [8] whose results indicate that retarded timings at low loads had increased C H 4 and T H C emissions. Hydrocarbon oxidation rates slow as temperature in the cylinder drops [14]. A s the in-cylinder temperatures are higher for higher the correlation of oxidation with temperature may explain why unburned hydrocarbons are lower at retarded timings for higher (j). The variability of combustion noted in section 5.1 may also be linked to the increased T H C emissions. The precise source of the apparent equivalent raw T H C emissions at early timings is unknown, with several possibilities: a similar amount some of the fuel may escape combustion despite the length of injection, or some lubricating oil may be expelled. The possibility also exists of diesel leakage from the injector, which would be very small and likely be consistent regardless of quantity of gas injected.  58  35 30 25  +  0.3  •  0.4  V  0.5  O 20 O 15  a  10  -5  0  5 10 15 20 50% Heat Release (°ATDC)  25  2 1.8 1.6 1.4  -+•  0.3  •  0.4  V  0.5  0 1.2  1  1  O 0.8 x 0.6 * --  0.4 0.2 0  Figure 5-12  0  5 10 15 20 50% Heat Release (°ATDC)  25  Total Hydrocarbon production for various timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air (a) non-normalized, (b) normalized  The effects of speed on T H C emissions are shown i n Figure 5-13. The amount of T H C emitted is the same within experimental error for 1200 and 1600 rpm, however the T H C emissions at 800 rpm are slightly lower. A s shown i n Figure 5-7, the combustion intensity is similar for between each speed. Therefore, the lower T H C emissions at 800 rpm is probably due 59  to the increased time available for the residuals to oxidize. The effects of retarding injection on T H C emissions at different speeds appears to be consistent as the increase in T H C emissions with retarded injection is similar. Total hydrocarbon emissions do not collapse onto a single trend with H R 5 0 . A s discussed earlier in section 5.1, plotting against H R 5 0 actually shifts later with increasing equivalence ratio and speed. A s T H C emissions at lower equivalence ratios are more affected than at higher equivalence ratios, using H R 5 0 as the independent variable instead of G S O I causes the trends between equivalence ratios to diverge. Therefore, H R 5 0 is not a good independent variable for comparing total hydrocarbon emissions at different equivalence ratios. This also means that the T H C emissions are not related to the timing of peak cylinder temperature in the way that N O emissions relate. x  1.8  RPM 800  1.6  1.4  •  1200  V  1600  _1.2  o h  1  30.8 O X  ~ 0.6 0.4 0.2 0 Figure 5-13  5.4  0  5 10 15 20 50% Heat Release (°ATDC)  25  Total hydrocarbon emissions for various timings and 3 speeds at 0=0.4, 4 g/cycle of air.  Carbon Monoxide The C O production at different equivalence ratios is presented in the form of raw emission  rates in Figure 5-14. A s can be seen in Figure 5-14(a.), the absolute C O production for equivalence ratios of 0.3 and 0.4 is the same for early injection timings within experimental error. A t a very late injection timing, C O production increases for 0 of 0.3 and 0 of 0.4. This differs slightly from the findings of Dumitrescu [8], where no increase in C O was found. This may be  60  due to different timings tested or a difference i n engine or injector performance. The C O could be produced in local rich conditions; as a result of the variability of combustion mentioned in section 5.1; quenching of the C O oxidation process as residuals are mixed with cooler bulk air; or a combination of the three. The C O emissions at § of 0.5 did not exhibit consistent behavior, increasing with retarding injection to approximately 12°c.a., then decreasing dramatically to 20°c.a. before rising again at extremely late injection. The injection timings corresponding to H R 5 0 of 5 and 10° c a . were repeated to confirm the C O levels. The repeated points confirmed the results within experimental error. The relatively high C O emissions at (j) of 0.5 around 12°c.a. are indicate poor air utilization of the fuel. Poor air utilization is likely a result of interaction of the fuel jet with the piston bowl. The effect of injection timing on C O emissions at different speeds is shown in Figure 515. Carbon monoxide production is observed to be relatively consistent for different speeds and H R with the exception of 1600 rpm, H R 5 0 of 15°c.a.. If the observed increase in C O production is simply due to limited availability of oxidation time, it would be expected to have lower C O levels at low speed, which has more time available for oxidation. The C O emissions measured at 1200 and 1600 rpm for late timing appears to be equivalent. Therefore, C O emissions appear to be more affected by equivalence ratio and timing than by speed. A s with the T H C emissions, C O emissions at late timings and lower equivalence ratios are more affected by retarded timing than at higher equivalence ratios. A l s o similar is that H R 5 0 as an independent variable causes C O emission trends to diverge at different equivalence ratios. Therefore, plotting against H R 5 0 exacerbates C O and t H C emissions caused by retarded timing and is not the best independent variable for comparison.  61  80 70 60  +  < > l  0.3  0  0.4  •  0.5  50  I  40  D  30 20 10 0  +  2.5  i  0  5 10 15 20 50% Heat Release (°ATDC)  25  0  5 10 15 20 50% Heat Release (°ATDC)  25  < > l  0.3  •  0.4  V  0.5  1.5  O  o 0.5  Figure 5-14  Carbon monoxide emissions for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air. (a) non-normalized (b) normalized  62  3  2.5  #  .1.5 O  o  -tr 0.5  0  Figure 5-15  i  1  -5  1  0  1  ,  1  ,  5 10 15 20 50% Heat Release (°ATDC)  ,  1  25  Carbon monoxide emissions for various timings and 3 speeds at (|)=0.4, 4 g/cycle of air.  5.4.1 Particulate M a t t e r The effect of changing injection timing on P M emissions at different equivalence ratios is shown in Figure 5-16. N o trends are apparent within experimental error. For most timings, it appears that P M production is the same within error for all equivalence ratios. However, there is a maximum P M emitted for <> | of 0.5 at H R 5 0 of +10°ATDC. Interestingly, this coincides with the same conditions that generate the maximum C O production. Testing at increased equivalence ratios and/or extending the sampling period may be more provide more informative regarding injection timing and particulate matter emissions. The effect of speed on P M production is shown in Figure 5-17. It appears that more P M is generated at 1600 rpm than lower speeds. There is no statistical difference in P M production between 800 and 1200 rpm for all timings tested. The effect of timing on P M production at 1600rpm is inconsistent, increasing with retarded timing to 15°c.a., then decreasing at 20°c.a. and then increasing again. When comparing the C O emissions to the P M emissions at 1600 rpm versus timing, the trends appear to somewhat correlate. This requires further investigation. The 1600 rpm, <b=0.4 case is the only condition, where this engine shows a N O - P M trade-off with x  timing prevalent in diesel engines [14,25].  63  20 18  +  16 14  <t>  0.3  —  •  0.4  --  V  0.5  ?12 | 1a>io e,  I  8  6 4 2 0  Figure 5-16  0  5 10 15 20 50% Heat Release (°ATDC)  25  Particulate matter emissions for various timings and 3 equivalence ratios at 1200 rpm, 4g/cycle air.  35 RPM 800  30 25  •  1200  V  1600  : 20 E  '15 10  0 Figure 5-17  5 10 15 20 50% Heat Release (°ATDC)  25  Particulate matter emissions for various injection timings and 3 speeds at 0=0.4, 4g/cycle air.  64  5.5  Summary • B y using 50% heat release (HR50) instead of G S O I injection timing, the N O production x  was found to collapse to a single curve for all equivalence ratios at a constant speed and air flow rate. N O x emissions are proportional to fuel burned for a fixed H R 5 0 . • Using 50% heat release was found to correlate efficiency well at different speeds. • Retarding absolute injection timing: a) improves efficiency until H R 5 0 reaches +5°c.a. and then decreases efficiency; b) decreases specific N O emissions until 50% heat release reaches approxix  mately +15°c.a., after which specific N O emissions are constant. x  This  result is consistent with the findings of Dumitrescu [8]; c) modestly decreases T H C emissions for slight retard, and then increases with further retard.  Higher speeds and lower equivalence ratios more  affected. This is consistent with data found in Dumitrescu [8]; d) increases C O emissions for <)) of 0.3 and 0.4 at all speeds tested; the raw C O emissions are the same at early timings; e) increases C O emissions for § of 0.5 at 1200 rpm as 50% heat release reaches +12°c.a., then decreases C O to +22°c.a., then increases C O beyond +22°c.a.. This may be explained by natural gas impinging on the piston bowl; f) causes no significant effect on P M emissions for cj) of 0.3, 1200 rpm; nor for <|) of 0.4 at 800 or 1200 rpm; g) increases and decreases P M emissions in a manner that seems to correlate with C O emissions for <j) of 0.5 at 1200 rpm; h) generally increases P M emissions for <)) of 0.4 at 1600 rpm. • Increasing speed: a) reduces absolute time of burn duration, implying increased mixing rates; b) improves efficiency for optimum timing, with no effect on efficiency at later timings; c) reduces specific N O emissions and delays the apparent 50% heat release x  limit for N O emission improvements; x  65  d) does not significantly affect C O production or P M emissions between 800 and 1200 rpm; e) increases T H C emissions; f) increases P M emissions at 1600 rpm as compared to 1200 rpm;  66  6. RELATIVE INJECTION TIMING This chapter examines the effects on performance and emissions of changing the relative injection timing (RIT) between the diesel pilot and the natural gas. The relative timing between the injections of natural gas and diesel w i l l dictate the amount of mixing of natural gas and air before ignition. This amount of mixing influences the proportions of pre-mixed and mixinglimited combustion, which may affect the emissions and engine efficiency. A s diesel substitution with natural gas in direct injection diesel engines has proven to reduce emissions while maintaining efficiency[4,5,8], diesel quantity should be as low as possible for best emissions. Wakenel et al. [18] found there was a minimum amount of diesel pilot required to promote good ignition of the natural gas in a pilot-ignited, natural gas fueled diesel engine [18]. Optimization of the relative injection timing between the natural gas and the pilot w i l l facilitate the minimum amount of diesel required for good ignition. Results are compared with injection delay simulations from Ouellette [23] and experimental results from Dumitrescu [8]. For this study, an R I T of 1.8 ms is considered normal, the value at which all other studies were conducted on this engine [9,30]. Relative timings below 1.8 ms are considered 'short' and timings above 1.8ms are considered 'extended'. A l l tests were conducted with 4g fresh air per cycle at an overall equivalence ratio (())) of 0.4. The effect of R I T on heat release is considered for separate cases of short and extended RIT. The effect of R I T on efficiency and emissions is compared for a set of tests where both gas and pilot injections were varied to maintain a constant 50% heat release (HR50). The influence of speed and absolute timing on the effects of R I T are also considered.  6.1 Performance Short Relative Injection Timing For this set of tests, the pilot start of injection (PSOI) was held constant at -5° A T D C and the natural gas injection timing varied. The purpose of this test is to clearly examine ignition and heat release at short RIT. B y maintaining a constant pilot injection, the conditions during pilot autoignition are consistent and therefore effects of changing relative timing are independent of changes in diesel ignition. Nielsen et al. [36] investigated the effect of methane on diesel autoignition by inducting methane into the intake manifold of a diesel engine. The results indicated that methane inhibited the diesel autoignition. It was unclear however, whether the 67  effect was due to chemical effects or lower in-cylinder temperatures due to reduced specific heat of the charge air mass. M t u i et al.[37], in study of an early sequential H P D I injected engine found there was no increase in autoignition delay of diesel when natural gas was injected shortly after the diesel. The diesel represented 30-50% of total energy. M u c h smaller amounts of pilot were used for the current study, and injection of natural gas before the diesel pilot is possible with this injector. The heat release data for pilot ignition at relative timings where natural gas was injected before and after the diesel pilot for a constant P S O I is shown in Figure 6-1. It appears that the initial change in heat release rate (-2° A T D C ) , which is assumed to be the diesel ignition event, is unchanged for the range of R I T tested. This agrees well with the experimental results of Mtui et al. [37], even with negative relative timings. It appears that natural gas does not affect diesel autoignition for this injector configuration.  10 20 Crank angle [°CA]  Figure 6-1  30  40  Comparison of apparent heat release rate for various short relative injection timings at 0=0.4, P S O I = - 5 ° A T D C , 1200 rpm.  As gas start of injection (GSOI) is advanced from + 8 ° to -5°, the combustion appears to become more premixed as the maximum heat release rate increases and the burn duration decreases. A s the gas is injected before the diesel ignites for G S O I of -5° and -12°, the natural gas likely mixes with more air before ignition. O f interest is the negative R I T (GSOI -12°) results in a lower apparent heat release rate than the 0 R I T case. However, as the natural gas is injected further before diesel ignition, it likely creates a mixture with leaner components, and the additional time for mixing creates a more voluminous combustible mixture. This is consistent with simulations presented by Ouellette [23], which indicated that no injection delay resulted in over-mixed methane. A l s o , the flame speed becomes slower with leaner mixtures[15]. Premixed flame speed is the fastest when the mixture fraction gradient is the sharpest. The gradient w i l l be  68  the most sharp with some premixing and but the gradient w i l l even out with further mixing (i.e. leaner mixture). The slower flame speed in a greater volume of combustible mixture is the most likely explanation for lower heat release rate and longer burn duration at negative relative timings. A l s o of interest is the shape of the heat release rate for negative R I T where the change i n heat release appears rough (0-4° A T D C ) before continuing into the next portion of combustion with smooth changes in heat release rate. While both fuels are present and reacting in a heterogeneous fuel mixture they are likely competing for oxygen; also, combustion kinetics may be different from a single fuel environment.  Variable Pilot Injection Timing For these tests, the gas injection timing was held constant at -5° A T D C and the pilot injection timing varied i n small increments between -32 and +7 °c.a. relative to gas injection. Positive crank angle indicates that the pilot was injected after the natural gas. The heat release for a range of extended R I T (PSOI -22 to -37°c.a.) is shown in Figure 6-2. The ignition of the natural gas appears to retard as R I T is extremely long. Combustion of the diesel may not be focused enough to ignite the new jet area. This is because the long R I T allows the diesel pilot to completely burn and mix with cool bulk air, which reduces ignition capability and quality. The gas jet may also travel further to reach the burning pilot, which would allow more natural gas to mix with air before igniting, resulting in more premixed combustion. The shape of the heat release rate at long relative timing also seems to change, where the 'peak' of combustion occurs at a later crank angle. This may be due to weak ignition of the natural gas.  -0.5 I -30 Figure 6-2  :  !  J  !  !  -20  -10  0  10  20  Crank Angle (°)  1 30  Comparison of apparent heat release rate for various extended relative injection timings at <b=0.4, 1200 rpm, G S O I = - 5 ° . 69  A n examination of the heat release data of extended R I T shown i n Figure 6-2 yields an apparent minimum threshold before the diesel pilot w i l l ignite. When the diesel pilot is injected at -27° A T D C and at -37° A T D C , the heat release rate is negligible until approximately -20° c a . . The minimum auto-ignition temperature of diesel is approximately 650 K [14], and according to polytropic compression relations, the in-cylinder temperatures reaches autoignition threshold at approximately -30° A T D C . After the diesel reaches 650 K and is mixed with oxygen, there is a delay while the diesel acquires enough energy to activate combustion reaction[14]. A t 1200 rpm, this appears to take approximately 10 degrees. The effects of R I T on the combustion events: start of combustion (SOC), end of combustion (EOC) and 50% heat release (HR50) are shown in Figure 6-3. The S O C is defined as the point where 10% total heat release is attained, and E O C is defined at 90% of total heat release. For these heat release comparisons only one standard deviation is used for experimental error. This error estimate is employed as engine conditions were not changed except for relative injection timing, and thus the experimental variation between relative injection tests is much smaller than the repeatability tests, where engine conditions were dramatically adjusted between samples.  The S O C appears to be the most affected by RIT, where it increases as R I T is changed  from 12°. The increased S O C changes more strongly for short RIT.  The H R 5 0 appears less  affected by R I T than S O C , as H R 5 0 is only shifted for R I T of 0 and 32°c.a.. The E O C does not appear to move consistently with R I T as S O C shifts. It is interesting that the E O C does not change much for very short or long relative timings even though the S O C does. This implies that the combustion rate must be higher for those timings than for a 'normal' relative injection timing.  70  25  + soc  20 nr  O 15 Q  m  o  HR50  •  EOC  •= 10 C <  c  (  IT  O 5  -at  at  5Z-  ' TIT  10 15 20 Relative Injection (°CA) Figure 6-3  2H "  25  30  Comparison of combustion events at (])=0.4, G S O I = - 5 ° A T D C , 1200 rpm at various RIT.  The burn duration, which is the difference between the S O C and E O C is another measure for examining combustion. The effect of relative injection timing on burn duration is shown in Figure 6-4. The maximum burn duration occurs with a relative injection timing between 8 and 22°c.a.. This range likely corresponds to the maximum proportion of mixing-limited combustion as compared to other relative timings. A s R I T is increased from this point, the burn duration declines slightly. A s R I T is reduced from 8°c.a. the burn duration decreases to relative injection of -4°c.a., where pilot is injected after natural gas. This minimum burn duration is where the natural gas has mixed sufficiently to burn the fastest. When the relative injection of the pilot is delayed to -7°c.a., the burn duration appears to increase. This indicates that the combustible natural gas mixture is likely more lean for pilot injections delayed beyond approximately -4°c.a.. Simulations of 3 injection delays presented by Ouellette [23] indicated that injection delay of 0.25 ms resulted in the shortest burn duration, and 0.5 ms delay was faster than no delay. The results from the current study indicate that slightly negative injection delay results in the fastest combustion. The difference could be due to the dissimilar configurations between simulation and this injector, or possibly a difference between commanded and actual start of injection times.  71  20[  O 15  r  o  o  10  3  oo  0  i  -10  Figure 6-4  .  -5  1  0  ,  1  ,  5 10 15 Relative Injection (°CA)  ,  20  1  25  30  ,—i  Burn duration for various relative injection timings at 0=0.4, G S O I = - 5 ° A T D C , 1200 rpm.  The effect of changing relative injection timing on variability of engine output is shown in Figure 6-5. The scatter in the data is noticeable, and only negative relative injection timings show moderately higher variability. It appears that R I T does not strongly affect engine output variability, therefore the C O V of indicated mean effective pressure w i l l not be used for further comparisons.  The effect of changing relative injection timing on the variability of combustion events is shown in Figure 6-6. The two parameters chosen for the variability of combustion are the standard deviations of start of combustion and of 50% heat release. Both parameters increase with negative relative timing, and increase dramatically for relative timing above 20°. Because the correlation coefficient is 0.98, only the standard deviation 50% heat release w i l l be used for performance comparisons at different operating conditions. The high correlation implies that the entire combustion event is varying with R I T and not simply the start of combustion. The increased variability is expected coincide with increased C O and T H C emissions as the variation is an indication of combustion stability.  72  7  o •S4 CO >  •  c,  •  D  cr  o  CD O  1  O  -10  Figure 6-5  5  10  15  Relative Injection (°CA)  20  25  30  Coefficient of variation of I M E P for various relative injection timings at <j)=0.4, G S O I = - 5 ° A T D C , 1200 rpm.  2  10% Heat Release 50% Heat Release  * o  1.8  • O-  1.6 <1.4 §1.2  o  Q  i0.8  o  "D  |  0.6  *  0.4  *  0.2  *I  °  -5  5  o  o ---£----< *•  0  Figure 6-6  o  10  * 15  Relative Injection (°CA)  20  25  30  Standard deviation of combustion events for various relative injection timings at ^=0.4, GSOI=-5° A T D C , 1200 rpm.  73  6.2 Variable Pilot and Natural Gas Injection The tests presented in this section were accomplished by varying the injection of both fuels to maintain a constant point of 50% heat release. This was done as H R 5 0 was previously shown in section 5.2 to be a marker for efficiency and N O production. This provides a consistent x  comparison for emissions and efficiency. The influence of timing on R I T effects was examined at 1200 rpm, for of H R 5 0 at +5 and +15°ATDC held within 0.5°. The increments for the relative timings were chosen with baseline points of -0.5, 0, 1.8 (standard), 3 ms plus two variable timings, one short and one extended. These points are illustrated in Figure 6-7 with ' A ' and ' B ' denoting the short and extended timings respectively. The G S O I was held constant and the ITR50 was determined for R I T of 1.8 ms from real-time heat release analysis on the D A Q computer. The two relative timings where H R 5 0 shifted +0.5° relative to that timing were arbitrarily chosen as the variable timings. These variable points were dependent upon engine conditions.  4.0 -I 0.0  ^ 0.5  1  1  ,  ,  ,  ,  1.0  1.5  2.0  2.5  3.0  3.5  1  4.0  1 4.5  Relative Injection Timing (ms)  Figure 6-7  Illustration of the determination of the variable relative injection timing settings.  A t 1200 rpm, the variable timings were 2° and 29°c.a. at H R 5 0 of +5°, and 6° and 29°c.a. at FfR50 of +15°. B y using the same strategy at 800 rpm, H R 5 0 of +15°, the timings were 3° and 25 °c.a.. The extended variable R I T was utilized as an arbitrary limit of how long the relative timing may be before substantial deterioration of natural gas ignition. The different limits at different operating conditions suggest that operating conditions influence the effects of relative injection timing. The influence of speed on R I T effects was examined by comparing 800 rpm  74  and 1200 rpm at a constant H R 5 0 of +15°ATDC. A s negative relative timings caused audible knock and degradation of emissions, the minimum R I T at 800 rpm was set to 0°. 6.2.1 Performance To compare performance, relative injection timing was varied to determine impact on efficiency, burn duration, and variability of 50% heat release. The indicated specific fuel consumption is constant within experimental error for all relative timings for each absolute timing at 1200 rpm as shown in Figure 6-8. A s noted earlier, a slightly negative relative timing resulted in the shortest burn duration. A s burn duration becomes shorter, the engine performance becomes closer to the ideal otto engine cycle. A s such, it was expected that efficiency would improve with the most rapid burning. However, it appears that the change i n burn duration is not significant enough to alter engine efficiency for this fuelling rate. This may be due to incomplete combustion or the change in burn duration is not large enough to affect efficiency. Despite the lack of change in efficiency for this fuelling rate, shorter burn duration has the potential to improve efficiency and should be tested at other loads and higher speeds. The results of this study differ from the findings of Dumitrescu [8], which found decreased thermal efficiency for short relative timing. It was noted by Dumitrescu [8] that the decreased thermal efficiency may have been due to injector limitations at short RIT, rather than due to the combustion effects of a short RIT.  10 15 20 Relative Injection (°CA)  Figure 6-8  25  30  Efficiency for various relative injection timings at <J>=0.4,1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  75  A s shown in Figure 6-9, the longest burn duration at 1200 rpm occurs at a relative timing of 13°c.a. (1.8 ms) for both absolute timings. The minimum burn duration occurs at a negative relative timing of -4° for both absolute timings. The minimum burn duration is lower at F£R50 of +5° than +15, presumably due to the higher cylinder pressures and temperatures which promote combustion. The results indicate that there is little influence of changing absolute injection timing on burn duration for zero RIT. It is apparent that R I T does not affect the short burn duration for retarded timings that was noted in chapter 5. The effect of R I T on burn duration is shown for two speeds in Figure 6-10. The results are shown in absolute time format and also in crank angle format. It is apparent that burn duration does not collapse onto a single trend versus R I T for either crank angle or absolute time. The relative timing effects at short R I T are dependent on the diesel pilot autoignition delay and the distance the natural gas jet travels to reach the burning pilot. Diesel autoignition delay is strongly dependent on pressure and temperature, which are weakly affected by speed. The gas jet speed is the same, regardless of engine speed. Both of these factors are mostly time dependent phenomenon and this is apparent in Figure 6-10-a for short RIT, where the proportional decrease in burn duration appears to match on an absolute time basis. The effects at long R I T probably have more to do with mixing and cooling of the diesel pilot after combustion, where mixing is roughly proportional to speed. This is apparent in Figure 6-10b, where after the maximum burn duration for each speed, the slope of the burn duration vs. R I T seem to match. Therefore, it appears the characteristics of burn duration at short R I T are time dependent, while the burn duration characteristics at long R I T are speed dependent.  76  -  20  < '-  c)  t)  - -f  -  <>  c>  (  5  HR50 + +5° o  Ol  -5  Figure 6-9  1  0  1  5  !  1  1  10 15 20 Relative Injection (°CA)  25  +15° '  30  ^  35  Comparison of burn duration for various relative injection timings at 0=0.4, 1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  77  22 20 18 16  6  14  | 12 ro  «1 c  5  m  1 0  8 6 4 2  1  0  5  Figure 6-10  2  3  4  Relative Injection Timing (ms)  10 15 20 Relative Injection (°CA)  25  30  Burn duration for various relative injection timings at two speeds, <f)=0.4, FfR50=+15° plotted against: (a) absolute timing, (b) crank angle  The effect of R I T on variability appears similar for both absolute timings tested at 1200 rpm as shown in Figure 6-11. Only at a negative R I T does the variation appear to be lower an earlier absolute timing. This could be due to higher in-cylinder temperatures which should  78  promote more consistent ignition of the natural gas. The influence of speed on combustion variability due to R I T is shown in Figure 6-12. It appears that with short RIT, the variability at 800 rpm is slightly lower than that of 1600 rpm. There is not much difference in variation at extended RIT. 1  +5 ° ATDC +15 ° ATDC  0.9 0.8 <0.7 •2-'  re 5 0.5 Q 10.4 "O  |  0.3 0.2 0.1 °5  Figure 6-11  10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two absolute timings, (|)=0.4, 1200 rpm.  1 0.9  * O  800 rpm 1200 rpm  o  0.8  o  <0.7 §0.6 Q  I 0.4 o | 0.3 0.2 0.1 0  Figure 6-12  o  _Q_  o  _o_ 10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C .  79  6.2.2 Emissions Nitrogen Oxides The effects of relative injection timing on N O emissions at 1200 rpm is shown in Figure x  6-13. The minimum N O emissions for both absolute timings occur at relative injection timing of x  13° (1.8 ms). A s R I T is extended from 13°, there is a slight increase in N O emissions. A s x  relative injection timing is shortened from 13°, there is a dramatic increase N O emissions. The x  effects of R I T on N O emissions are stronger for the H R 5 0 of +5 as compared to the H R 5 0 of x  +15. This likely has to do with the lower in-cylinder temperature at later absolute timing. It is interesting that at negative R I T the apparent heat release rate is lower for than for 0 RIT, but the N O emissions are higher. A n explanation for this is that the natural gas mixture is more x  voluminous before ignition at very negative relative timings. Once burned, the products of the mixture may take longer to m i x with the cooler bulk air, which allows more time for N O  x  chemical kinetics to proceed. The effects of R I T on N O emissions at different speeds is shown in Figure 6-14. The x  minimum N O production at 800 rpm is level between 9° and 14° relative timings. A t 800 rpm, x  the N O increases from a minimum of 6 g/kW-hr to 12 g/kW-hr at 0 RIT. A t 1200 rpm, the N O x  x  increases from a minimum of 4.4g/kW-hr to 8.6 g/kW-hr. It appears that the effects of relative timing on N O emissions are greater at lower speeds. This is probably due to increased mixing x  rates at higher speeds. Dumitrescu [8] found that short R I T increased N O emissions while holding P S O I x  constant. Chapter 5 illustrated that shifting combustion earlier in the cycle increases N O  x  emissions. It is not surprising that the 3° shorter R I T in [8], which shifted the gas combustion earlier i n the cycle, resulted in increased N O emissions. The results from the current study x  indicate that the increased N O found by [8] would be partly due to premixing, and partly due to x  shifting the combustion event.  80  5 Figure 6-13  10 15 20 Relative Injection (°CA)  25  30  Comparison of N O production for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm. x  14  — Speed + 800 rpm  12  o  1200 rpm  10  O  °  5 Figure 6-14  10 15 20 Relative Injection (°CA)  25  30  N O emissions for various relative injection timings at two speeds, x  0=0.4, FfR50=+15°ATDC.  81  Total Hydrocarbons and Carbon Monoxide The results of the effects of relative injection timing on both total hydrocarbons and carbon monoxide are shown in Figure 6-15 for two absolute timings. A s with N O emissions, the x  minimum C O and T H C emissions occur at an R I T of 13°. A t H R 5 0 of +5°, changing R I T increases both T H C and C O emissions for both extended and negative relative timings. Relative timing appears to affect C O and t H C emissions more strongly at HR50+15. The in-cylinder pressures and temperatures are lower for the later absolute timing, making it more difficult to ignite the natural gas mixture. In addition, there is less time for T H C s and C O to oxidize at late absolute timings. The presence of more incompletely burned products at short R I T for late absolute timing indicates that combustion is less efficient and as a result, the corresponding shorter burn durations are less likely to improve engine efficiency (at late absolute timings). The more moderate levels of T H C and C O emissions for short R I T at early timings provide more opportunity for the short burn duration to improve efficiency (at early absolute timings). The simulations of Ouellette [23] found high levels of C O formation with no ignition delay. This was mostly due to impingement of the gas jet on a wall as the flame could not reach the end of the jet before the jet reached the wall. The 1200 rpm experimental results of Dumitrescu [8] indicated that higher levels of C O and T H C emissions occur with a 3° RIT, as compared to a 6° RIT. The trends of C O and T H C emissions in this study agree with trends at short R I T found by Dumitrescu [8] and C O results of Ouellette. The effects of R I T on T H C and C O emissions at different speeds is shown in Figure 6-16. The minimum T H C and C O emissions at 800 rpm occur at the same relative timing as with N O emissions between 9° and 14°. The effects of changing R I T on T H C and C O emissions appears similar for both speeds.  82  x  ffi  I I  Mol / HR50 THC / +5  O  THC / +15  •  CO / +5  V  CO /+15  I  c o  .i  LU  i I I  1 1  _L  3E  5  Figure 6-15  +  10 15 20 Relative Injection (°CA)  25  30  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm.  + THC / 800 rpm  ffi  _£ I  t  o  THC/'1200 rpm  •  CO / 800 rpm  v  CO / 1200 rpm  4  c  •to3 2  en  0  E LU  2 -_t 1 5  I  5 Figure 6-16  1-  IE 1 , 5  10 15 20 Relative Injection (°CA)  25  30  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C  83  Particulate Matter There was no statistical change in P M emissions for the relative timing sweeps at each operating condition. The experimental error masked any effect of relative timing, partly because the P M emissions at this equivalence ratio are very low. To determine the effect of R I T on P M , this study should be repeated for an engine operating condition where the engine is known to produce higher levels of P M .  Overall Emission Discussion To examine a perceived correlation between emissions and burn duration, the correlation coefficient between burn duration and emission species was calculated using:  Cov(X, Y) ? ,y= X  0  x  ,  _  .  1X  (El- - ) 6  0  y  1  Where o~ is the standard deviation for each variable and the Cov is given by:  Cov(X, Y) = -  (*. - ii )(y. x  I  - up  (Eq. 6.2)  =1  The correlation coefficient for each emission species, averaged over each mode tested is listed in Table 6.1. It is apparent that the maximum burn duration corresponds to minimum emissions when changing relative timing at all speeds and absolute timings tested. This likely has to do with increasing the proportion of pre-mixed combustion, which burns faster, but causing more emissions. Burn duration could be used for determining optimal RIT, without requiring lengthy emission data samples and analysis, although this should be verified. The flat optimal emission/ burn duration response to changing relative timing at 800 rpm suggest that more pilot is injected than required for good ignition. This can be inferred as an optimal minimum amount of pilot w i l l likely have an distinct maximum burn duration versus RIT. The accuracy of the burn duration measurements can be improved by averaging more cycles of in-cylinder data. This correlation w i l l not hold for extreme negative relative timings where the burn duration increases or for extremely long R I T where the ignition of the natural gas is very weak. The exact limits for good correlation are unknown. This correlation is only valid for a fixed operating condition.  84  Table 6.1: Burn Duration Correlations Emission Species  Correlation Coefficient with Burn Duration  Carbon monoxide  -0.97  Nitrogen oxides  -0.98  Total hydrocarbon  -0.96  Hypothetically, smaller quantities of pilot in the current configuration would result in a smaller window of good ignition. This is because a smaller quantity of diesel pilot w i l l cool faster and thereby reduce the ignition capability. Higher speed or high swirl may require more amounts of pilot, as the pilot w i l l disperse and cool faster than at low speeds. Conversely, the amount of diesel may be reduced at low rpm as it does not disperse as quickly. A l s o , R I T may become significant at high E G R , as E G R w i l l increase the ignition delay [14], which w i l l affect ignition strength due to mixing. The minimum amount of pilot is a complicated problem involving engine conditions plus injection duration, injection rate and shape, injector configuration, and nozzle shape. The sheer number of variables involved suggest that numerical simulation studies be conducted to optimize the diesel charge before experiments. Noting that the diesel accounts for as little as 3% of the total fuel (at high load), further reduction of pilot would result in only nominal emission reductions due to fuel substitution. However, optimized pilot charge may improve natural gas ignition, where minimal pre-mixing of the natural gas should result i n the lowest possible N O  x  formation. The emission results indicate that optimal timing between gas and pilot injection occurs at approximately 13° and 8-14° for 1200 and 800 rpm respectively. The optimum timing for both speeds included the 'normal' timing of 1.8 ms. The emission response was more flat in terms of absolute time at 800 rpm than at 1200 rpm. This is likely influenced by enhanced mixing at higher speeds. The fact that optimal emissions resulted at delayed sequential injection concurs with diesel/CH4 simulations by [23], which determined that an injection delay of gas after diesel reduces chance of fuel over-mixing and impingement. In those simulations, increasing the injection delay from zero decreased C O formation and reduced unburned C H 4 .  The results of  these experiments indicate that a minimum of premixed combustion, with good ignition of the natural gas, brings about the best possible emissions.  85  6.3 Summary •  There is an optimum relative injection timing (or range of timings) for a minimum of all emissions at fixed operating condition, the optimum relative timing appears to be independent of speed or absolute injection timing. The optimum relative timing included the 'normal' 1.8 ms (9° and 13° for 800 and 1200 rpm respectively). The emission response was flatter at 800 rpm than at 1200 rpm on an absolute time basis.  •  The efficiency is not significantly affected by relative injection timing. However, the ability to reduce burn duration provides potential for improved efficiency at early absolute timings.  •  When 50% heat release is held constant, burn duration correlates well with emissions. The longest burn duration is produced by combustion most favorable for emissions.  •  For short relative injection timings, an optimal delay avoids premixed combustion-i.e. it is desirable for the diesel pilot to be ignited before the N G is injected. It seems that combustion appearing more pre-mixed generates worse emissions. This event is more dependent on absolute time than by crank angle.  •  Agreement with the emission trends at short relative timings with Dumitrescu [8], however the efficiency findings between studies differ. The difference in thermal efficiency found in that study was likely due to injector design.  •  At extended relative injection timings, apparent heat-release data indicate that natural gas ignition by pilot is weakened by mixing of the pilot, therefore higher speeds and high swirl engines may require more diesel for good ignition of the natural gas.  •  Future study should include simulation of the diesel pilot with a focus of improving ignition of the natural gas.  •  The effect of relative injection timing on particulate matter emissions should be studied at conditions generating more particulate matter (higher equivalence ratios). 86  A n examination of the heat release data of extended R I T shown i n Figure 6-2 yields an apparent minimum threshold before the diesel pilot w i l l ignite. When the diesel pilot is injected at -27° A T D C and at -37° A T D C , the heat release rate is negligible until approximately -20° c a . . The minimum auto-ignition temperature of diesel is approximately 650 K [14], and according to polytropic compression relations, the in-cylinder temperatures reaches autoignition threshold at approximately -30° A T D C . After the diesel reaches 650 K and is mixed with oxygen, there is a delay while the diesel acquires enough energy to activate combustion reaction[14]. A t 1200 rpm, this appears to take approximately 10 degrees. The effects of R I T on the combustion events: start of combustion (SOC), end of combustion ( E O C ) and 50% heat release (HR50) are shown in Figure 6-3. The S O C is defined as the point where 10% total heat release is attained, and E O C is defined at 90% of total heat release. For these heat release comparisons only one standard deviation is used for experimental error. This error estimate is employed as engine conditions were not changed except for relative injection timing, and thus the experimental variation between relative injection tests is much smaller than the repeatability tests, where engine conditions were dramatically adjusted between samples.  The S O C appears to be the most affected by RIT, where it increases as R I T is changed  from 12°. The increased S O C changes more strongly for short RIT.  The H R 5 0 appears less  affected by R I T than S O C , as H R 5 0 is only shifted for R I T of 0 and 3 2 ° c a . . The E O C does not appear to move consistently with R I T as S O C shifts. It is interesting that the E O C does not change much for very short or long relative timings even though the S O C does. This implies that the combustion rate must be higher for those timings than for a 'normal' relative injection timing.  70  25  20  O Q  +  SOC  o  HR50  •  EOC  15  £10 c <  c O  I at  -5  Figure 6-3  jg;.  °  10 15 20 Relative Injection (°CA)  25  30  Comparison of combustion events at 0=0.4, GSOI=-5° ATDC, 1200 rpm at various RIT.  The burn duration, which is the difference between the SOC and EOC is another measure for examining combustion. The effect of relative injection timing on burn duration is shown in Figure 6-4. The maximum burn duration occurs with a relative injection timing between 8 and 22°c.a.. This range likely corresponds to the maximum proportion of mixing-limited combustion as compared to other relative timings. As RIT is increased from this point, the burn duration declines slightly. As RIT is reduced from 8°c.a. the burn duration decreases to relative injection of -4°c.a., where pilot is injected after natural gas. This minimum burn duration is where the natural gas has mixed sufficiently to burn the fastest. When the relative injection of the pilot is delayed to -7°c.a., the burn duration appears to increase. This indicates that the combustible natural gas mixture is likely more lean for pilot injections delayed beyond approximately -4°c.a.. Simulations of 3 injection delays presented by Ouellette [23] indicated that injection delay of 0.25 ms resulted in the shortest burn duration, and 0.5 ms delay was faster than no delay. The results from the current study indicate that slightly negative injection delay results in the fastest combustion. The difference could be due to the dissimilar configurations between simulation and this injector, or possibly a difference between commanded and actual start of injection times.  71  0  l  -10  Figure 6-4  '  -5  '  0  '  5  '  10  '  15  Relative Injection (°CA)  '  20  '  25  '—I  30  Burn duration for various relative injection timings at cf»=0.4, GSOI=-5° ATDC, 1200 rpm.  The effect of changing relative injection timing on variability of engine output is shown in Figure 6-5. The scatter in the data is noticeable, and only negative relative injection timings show moderately higher variability. It appears that RIT does not strongly affect engine output variability, therefore the C O V of indicated mean effective pressure will not be used for further comparisons.  The effect of changing relative injection timing on the variability of combustion events is shown in Figure 6-6. The two parameters chosen for the variability of combustion are the standard deviations of start of combustion and of 50% heat release. Both parameters increase with negative relative timing, and increase dramatically for relative timing above 20°. Because the correlation coefficient is 0.98, only the standard deviation 50% heat release will be used for performance comparisons at different operating conditions. The high correlation implies that the entire combustion event is varying with RIT and not simply the start of combustion. The increased variability is expected coincide with increased CO and THC emissions as the variation is an indication of combustion stability.  72  7  c  o |4[>  o  =3  •  'o o 9^  Oi  <  -10  Figure 6-5  -5  1  0  1  1  1  D  a  r  1  5 10 15 20 Relative Injection (°CA)  •  1  1  25  30  1  Coefficient of variation of IMEP for various relative injection timings at 0=0.4, GSOI=-5° ATDC, 1200 rpm.  2  * o  1.8  10% Heat Release 50% Heat Release  -O-  1.6 <1.4 §1.2 to Q  l  o  *0.8 •a | 0.6  *  0.4  o  *  0.2 0  Figure 6-6  o  -5  o  \I*  °*  o * •  **  *  5 10 15 20 Relative Injection (°CA)  25  30  Standard deviation of combustion events for various relative injection timings at 0=0.4, GSOI=-5° ATDC, 1200 rpm.  73  6.2 Variable Pilot and Natural Gas Injection The tests presented i n this section were accomplished by varying the injection of both fuels to maintain a constant point of 50% heat release. This was done as H R 5 0 was previously shown in section 5.2 to be a marker for efficiency and N O production. This provides a consistent x  comparison for emissions and efficiency. The influence of timing on R I T effects was examined at 1200 rpm, for of H R 5 0 at +5 and +15°ATDC held within 0.5°. The increments for the relative timings were chosen with baseline points of -0.5, 0, 1.8 (standard), 3 ms plus two variable timings, one short and one extended. These points are illustrated in Figure 6-7 with ' A ' and ' B ' denoting the short and extended timings respectively. The G S O I was held constant and the H R 5 0 was determined for R I T of 1.8 ms from real-time heat release analysis on the D A Q computer. The two relative timings where H R 5 0 shifted +0.5° relative to that timing were arbitrarily chosen as the variable timings. These variable points were dependent upon engine conditions.  4.0-1 0.0  h  0.5  >  '  '  1.0  1.5  2.0  2.5  .  •  3.0  3.5  1  4.0  1 4.5  Relative Injection Timing (ms)  Figure 6-7  Illustration of the determination of the variable relative injection timing settings.  A t 1200 rpm, the variable timings were 2° and 29°c.a. at H R 5 0 of +5°, and 6° and 29°c.a. at H R 5 0 o f + 1 5 ° . B y using the same strategy at 800 rpm, H R 5 0 of +15°, the timings were 3° and 25 °c.a.. The extended variable R I T was utilized as an arbitrary limit of how long the relative timing may be before substantial deterioration of natural gas ignition. The different limits at different operating conditions suggest that operating conditions influence the effects of relative injection timing. The influence of speed on R I T effects was examined by comparing 800 rpm  74  and 1200 rpm at a constant H R 5 0 of + 1 5 ° A T D C . A s negative relative timings caused audible knock and degradation of emissions, the minimum R I T at 800 rpm was set to 0 ° . 6.2.1 Performance To compare performance, relative injection timing was varied to determine impact on efficiency, burn duration, and variability of 50% heat release. The indicated specific fuel consumption is constant within experimental error for all relative timings for each absolute timing at 1200 rpm as shown in Figure 6-8. A s noted earlier, a slightly negative relative timing resulted in the shortest burn duration. A s burn duration becomes shorter, the engine performance becomes closer to the ideal otto engine cycle. A s such, it was expected that efficiency would improve with the most rapid burning. However, it appears that the change in burn duration is not significant enough to alter engine efficiency for this fuelling rate. This may be due to incomplete combustion or the change in burn duration is not large enough to affect efficiency. Despite the lack of change in efficiency for this fuelling rate, shorter burn duration has the potential to improve efficiency and should be tested at other loads and higher speeds. The results of this study differ from the findings of Dumitrescu [8], which found decreased thermal efficiency for short relative timing. It was noted by Dumitrescu [8] that the decreased thermal efficiency may have been due to injector limitations at short RIT, rather than due to the combustion effects of a short RIT. 200 180  -f-T"  160 140 120 o)100  L? 80 CO  60 40 20 5 Figure 6-8  10 15 20 Relative Injection (°CA)  30  Efficiency for various relative injection timings at <b=0.4,1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  75  A s shown in Figure 6-9, the longest burn duration at 1200 rpm occurs at a relative timing of 13°c.a. (1.8 ms) for both absolute timings. The minimum burn duration occurs at a negative relative timing of -4° for both absolute timings. The minimum burn duration is lower at F1R50 of +5° than +15, presumably due to the higher cylinder pressures and temperatures which promote combustion. The results indicate that there is little influence of changing absolute injection timing on burn duration for zero RIT. It is apparent that R I T does not affect the short burn duration for retarded timings that was noted in chapter 5. The effect of R I T on burn duration is shown for two speeds in Figure 6-10. The results are shown in absolute time format and also i n crank angle format. It is apparent that burn duration does not collapse onto a single trend versus R I T for either crank angle or absolute time. The relative timing effects at short R I T are dependent on the diesel pilot autoignition delay and the distance the natural gas jet travels to reach the burning pilot. Diesel autoignition delay is strongly dependent on pressure and temperature, which are weakly affected by speed. The gas jet speed is the same, regardless of engine speed. Both of these factors are mostly time dependent phenomenon and this is apparent in Figure 6-10-a for short RIT, where the proportional decrease in burn duration appears to match on an absolute time basis. The effects at long R I T probably have more to do with mixing and cooling of the diesel pilot after combustion, where mixing is roughly proportional to speed. This is apparent in Figure 6-10b, where after the maximum burn duration for each speed, the slope of the burn duration vs. R I T seem to match. Therefore, it appears the characteristics of burn duration at short R I T are time dependent, while the burn duration characteristics at long R I T are speed dependent.  76  20  <  O 15  c o «  10  °  m  10 15 20 Relative Injection (°CA) Figure 6-9  25  30  35  Comparison of burn duration for various relative injection timings at <b=0.4, 1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  77  o  1 4  §12  8  W  3 CQ  1  2  3  Relative Injection Timing (ms)  4  22 r 2018CA)  1614-  -i  }•  o 12 « 5 10-- -o- Q ^_  m  86speed 800 rpm  420-5  Figure 6-10  1200 rpm 5  10 15 20 Relative Injection (°CA)  25  30  Burn duration for various relative injection timings at two speeds, (j)=0.4, H R 5 0 = + 1 5 ° plotted against: (a) absolute timing, (b) crank angle  The effect of R I T on variability appears similar for both absolute timings tested at 1200 rpm as shown in Figure 6-11. Only at a negative R I T does the variation appear to be lower an earlier absolute timing. This could be due to higher in-cylinder temperatures which should  78  promote more consistent ignition of the natural gas. The influence of speed on combustion variability due to R I T is shown in Figure 6-12. It appears that with short RIT, the variability at 800 rpm is slightly lower than that of 1600 rpm. There is not much difference i n variation at extended RIT. 1  +5 ° ATDC +15° ATDC  0.9 0.8 <0.7 gO.6 Q  I  0-4  | 0.3 0.2 0.1 0 Figure 6-11  10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm.  1 0.9  *  O  800 rpm 1200 rpm  o  0.8  o  <0.7  g 0.5 o Q I 0.4 "O  | 0.3  o _Q.  o  -0_  0.2 0.1 0  Figure 6-12  5  10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C .  79  6.2.2 Emissions Nitrogen Oxides The effects of relative injection timing on N O emissions at 1200 rpm is shown in Figure x  6-13. The minimum N O emissions for both absolute timings occur at relative injection timing of x  13° (1.8 ms). A s R I T is extended from 13°, there is a slight increase in N O emissions. A s x  relative injection timing is shortened from 13°, there is a dramatic increase N O emissions. The x  effects of R I T on N O emissions are stronger for the H R 5 0 of +5 as compared to the H R 5 0 of x  +15. This likely has to do with the lower in-cylinder temperature at later absolute timing. It is interesting that at negative R I T the apparent heat release rate is lower for than for 0 RIT, but the N O emissions are higher. A n explanation for this is that the natural gas mixture is more x  voluminous before ignition at very negative relative timings. Once burned, the products of the mixture may take longer to m i x with the cooler bulk air, which allows more time for N O  x  chemical kinetics to proceed. The effects of R I T on N O emissions at different speeds is shown in Figure 6-14. The x  minimum N O production at 800 rpm is level between 9° and 14° relative timings. A t 800 rpm, x  the N O increases from a minimum of 6 g/kW-hr to 12 g/kW-hr at 0 RIT. A t 1200 rpm, the N O x  x  increases from a minimum of 4.4g/kW-hr to 8.6 g/kW-hr. It appears that the effects of relative timing on N O emissions are greater at lower speeds. This is probably due to increased mixing x  rates at higher speeds. Dumitrescu [8] found that short R I T increased N O emissions while holding P S O I x  constant. Chapter 5 illustrated that shifting combustion earlier in the cycle increases N O  x  emissions. It is not surprising that the 3° shorter R I T in [8], which shifted the gas combustion earlier in the cycle, resulted in increased N O emissions. The results from the current study x  indicate that the increased N O found by [8] would be partly due to premixing, and partly due to x  shifting the combustion event.  80  25  50% Heat Release + +5  o  20  5 Figure 6-13  +15  10 15 20 Relative Injection (°CA)  25  30  Comparison of N O production for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm. x  14  — Speed + 800 rpm  12  o 1200 rpm  10  "5>  5 Figure 6-14  10 15 20 Relative Injection (°CA)  N O emissions for various relative injection timings at two speeds, x  0=0.4, H R 5 0 = + 1 5 ° A T D C .  81  Total Hydrocarbons and Carbon Monoxide The results of the effects of relative injection timing on both total hydrocarbons and carbon monoxide are shown in Figure 6-15 for two absolute timings. A s with N O emissions, the x  minimum C O and T H C emissions occur at an R I T of 13°. A t H R 5 0 of +5°, changing R I T increases both T H C and C O emissions for both extended and negative relative timings. Relative timing appears to affect C O and t H C emissions more strongly at HR50+15. The in-cylinder pressures and temperatures are lower for the later absolute timing, making it more difficult to ignite the natural gas mixture. In addition, there is less time for T H C s and C O to oxidize at late absolute timings. The presence of more incompletely burned products at short R I T for late absolute timing indicates that combustion is less efficient and as a result, the corresponding shorter burn durations are less likely to improve engine efficiency (at late absolute timings). The more moderate levels of T H C and C O emissions for short R I T at early timings provide more opportunity for the short burn duration to improve efficiency (at early absolute timings). The simulations of Ouellette [23] found high levels of C O formation with no ignition delay. This was mostly due to impingement of the gas jet on a wall as the flame could not reach the end of the jet before the jet reached the wall. The 1200 rpm experimental results of Dumitrescu [8] indicated that higher levels of C O and T H C emissions occur with a 3° RIT, as compared to a 6° RIT. The trends of C O and T H C emissions in this study agree with trends at short R I T found by Dumitrescu [8] and C O results of Ouellette. The effects of R I T on T H C and C O emissions at different speeds is shown i n Figure 6-16. The minimum T H C and C O emissions at 800 rpm occur at the same relative timing as with N O emissions between 9° and 14°. The effects of changing R I T on T H C and C O emissions appears similar for both speeds.  82  x  MOI/HR50 HI  + THC / +5  I  o  THC/+15  •  CO / +5  v  CO /+15  as  i  . i  mI  11  I  01 -5  Figure 6-15  1  1  0  5  .  .  ,  1  10 15 20 Relative Injection (°CA)  L.  25  30  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm.  + THC / 800 rpm  m  _i 1  I  4  o  THC/ 1200 rpm  •  CO / 800 rpm  v  CO / 1200 rpm  I  in tz  •in 3 Q  to  0  E  LU  10 15 20 Relative Injection (°CA) Figure 6-16  25  30  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C  83  Particulate Matter There was no statistical change in P M emissions for the relative timing sweeps at each operating condition. The experimental error masked any effect of relative timing, partly because the P M emissions at this equivalence ratio are very low. To determine the effect of R I T on P M , this study should be repeated for an engine operating condition where the engine is known to produce higher levels of P M .  Overall Emission Discussion To examine a perceived correlation between emissions and burn duration, the correlation coefficient between burn duration and emission species was calculated using: Px,y  '  Cov(X,Y) 0  x  .  0  ,„  ,  (Eq-6.1)  ^y  Where o is the standard deviation for each variable and the Cov is given by: Cov(X, Y) =  - ^ n  (x.-^Xy.1  up  (Eq. 6.2)  = 1  The correlation coefficient for each emission species, averaged over each mode tested is listed in Table 6.1. It is apparent that the maximum burn duration corresponds to minimum emissions when changing relative timing at all speeds and absolute timings tested. This likely has to do with increasing the proportion of pre-mixed combustion, which burns faster, but causing more emissions. Burn duration could be used for determining optimal RIT, without requiring lengthy emission data samples and analysis, although this should be verified. The flat optimal emission/ burn duration response to changing relative timing at 800 rpm suggest that more pilot is injected than required for good ignition. This can be inferred as an optimal minimum amount of pilot w i l l likely have an distinct maximum burn duration versus RIT. The accuracy of the burn duration measurements can be improved by averaging more cycles of in-cylinder data. This correlation w i l l not hold for extreme negative relative timings where the burn duration increases or for extremely long R I T where the ignition of the natural gas is very weak. The exact limits for good correlation are unknown. This correlation is only valid for a fixed operating condition.  84  Table 6.1: Burn Duration Correlations Emission Species  Correlation Coefficient with Burn Duration  Carbon monoxide  -0.97  Nitrogen oxides  -0.98  Total hydrocarbon  -0.96  Hypothetically, smaller quantities of pilot in the current configuration would result in a smaller window of good ignition. This is because a smaller quantity of diesel pilot w i l l cool faster and thereby reduce the ignition capability. Higher speed or high swirl may require more amounts of pilot, as the pilot w i l l disperse and cool faster than at low speeds. Conversely, the amount of diesel may be reduced at low rpm as it does not disperse as quickly. A l s o , R I T may become significant at high E G R , as E G R w i l l increase the ignition delay [14], which w i l l affect ignition strength due to mixing. The minimum amount of pilot is a complicated problem involving engine conditions plus injection duration, injection rate and shape, injector configuration, and nozzle shape. The sheer number of variables involved suggest that numerical simulation studies be conducted to optimize the diesel charge before experiments. Noting that the diesel accounts for as little as 3% of the total fuel (at high load), further reduction of pilot would result in only nominal emission reductions due to fuel substitution. However, optimized pilot charge may improve natural gas ignition, where minimal pre-mixing of the natural gas should result in the lowest possible N O  x  formation. The emission results indicate that optimal timing between gas and pilot injection occurs at approximately 13° and 8-14° for 1200 and 800 rpm respectively. The optimum timing for both speeds included the 'normal' timing of 1.8 ms. The emission response was more flat in terms of absolute time at 800 rpm than at 1200 rpm. This is likely influenced by enhanced mixing at higher speeds. The fact that optimal emissions resulted at delayed sequential injection concurs with diesel/CH4 simulations by [23], which determined that an injection delay of gas after diesel reduces chance of fuel over-mixing and impingement. In those simulations, increasing the injection delay from zero decreased C O formation and reduced unburned C H 4 .  The results of  these experiments indicate that a minimum of premixed combustion, with good ignition of the natural gas, brings about the best possible emissions.  85  A n examination of the heat release data of extended R I T shown in Figure 6-2 yields an apparent minimum threshold before the diesel pilot w i l l ignite. When the diesel pilot is injected at -27° A T D C and at -37° A T D C , the heat release rate is negligible until approximately -20° c a . . The minimum auto-ignition temperature of diesel is approximately 650 K [14], and according to polytropic compression relations, the in-cylinder temperatures reaches autoignition threshold at approximately -30° A T D C . After the diesel reaches 650 K and is mixed with oxygen, there is a delay while the diesel acquires enough energy to activate combustion reaction [14]. A t 1200 rpm, this appears to take approximately 10 degrees. The effects of R I T on the combustion events: start of combustion (SOC), end of combustion (EOC) and 50% heat release (HR50) are shown in Figure 6-3. The S O C is defined as the point where 10% total heat release is attained, and E O C is defined at 90% of total heat release. For these heat release comparisons only one standard deviation is used for experimental error. This error estimate is employed as engine conditions were not changed except for relative injection timing, and thus the experimental variation between relative injection tests is much smaller than the repeatability tests, where engine conditions were dramatically adjusted between samples.  The S O C appears to be the most affected by RIT, where it increases as R I T is changed  from 12°. The increased S O C changes more strongly for short RIT.  The H R 5 0 appears less  affected by R I T than S O C , as H R 5 0 is only shifted for R I T of 0 and 3 2 ° c a . . The E O C does not appear to move consistently with R I T as S O C shifts. It is interesting that the E O C does not change much for very short or long relative timings even though the S O C does. This implies that the combustion rate must be higher for those timings than for a 'normal' relative injection timing.  70  25 + SOC 20 i n nr  0 15?  o  HR50  •  EOC  c <  J  5i  -^J:  at  jg;  SJ:-  10 15 20 Relative Injection (°CA) Figure 6-3  25  30  Comparison of combustion events at 0=0.4, G S O I = - 5 ° A T D C , 1200 rpm at various RIT.  The burn duration, which is the difference between the S O C and E O C is another measure for examining combustion. The effect of relative injection timing on burn duration is shown in Figure 6-4. The maximum burn duration occurs with a relative injection timing between 8 and 22°c.a.. This range likely corresponds to the maximum proportion of mixing-limited combustion as compared to other relative timings. A s R I T is increased from this point, the burn duration declines slightly. A s R I T is reduced from 8°c.a. the burn duration decreases to relative injection of -4°c.a., where pilot is injected after natural gas. This minimum burn duration is where the natural gas has mixed sufficiently to burn the fastest. When the relative injection of the pilot is delayed to -7°c.a., the burn duration appears to increase. This indicates that the combustible natural gas mixture is likely more lean for pilot injections delayed beyond approximately -4°c.a.. Simulations of 3 injection delays presented by Ouellette [23] indicated that injection delay of 0.25 ms resulted in the shortest burn duration, and 0.5 ms delay was faster than no delay. The results from the current study indicate that slightly negative injection delay results in the fastest combustion. The difference could be due to the dissimilar configurations between simulation and this injector, or possibly a difference between commanded and actual start of injection times.  71  20  o  < O 15 C  o 3  Q 10 3  CQ  5  rji -10  Figure 6-4  , -5  , 0  , , , 5 10 15 Relative Injection (°CA)  , 20  , 25  ,—i 30  Burn duration for various relative injection timings at <b=0.4, G S O I = - 5 ° A T D C , 1200 rpm.  The effect of changing relative injection timing on variability of engine output is shown in Figure 6-5. The scatter in the data is noticeable, and only negative relative injection timings show moderately higher variability. It appears that R I T does not strongly affect engine output variability, therefore the C O V of indicated mean effective pressure w i l l not be used for further comparisons.  The effect of changing relative injection timing on the variability of combustion events is shown in Figure 6-6. The two parameters chosen for the variability of combustion are the standard deviations of start of combustion and of 50% heat release. Both parameters increase with negative relative timing, and increase dramatically for relative timing above 2 0 ° . Because the correlation coefficient is 0.98, only the standard deviation 50% heat release w i l l be used for performance comparisons at different operating conditions. The high correlation implies that the entire combustion event is varying with R I T and not simply the start of combustion. The increased variability is expected coincide with increased C O and T H C emissions as the variation is an indication of combustion stability.  72  7  c o CO  >  •  f3-  CD  D  •  O  o  0  i  1  -10  Figure 6-5  -5  1  1  0  5  1  .  10  15  Relative Injection (°CA)  >  1 20  ,  25  1  30  Coefficient of variation of I M E P for various relative injection timings at (t)=0.4, G S O I = - 5 ° A T D C , 1200 rpm.  2r  10% Heat Release 50% Heat Release  * o  1.8 1.6-  <1.4  g  1 -2  O  CD  Q  co  h o  0.8  Jo.6  r  0.4-  O o  0.2-  o  o  o  •  -  1  -10  Figure 6-6  -5  5  10  15  Relative Injection (°CA)  20  25  30  Standard deviation of combustion events for various relative injection timings at 0=0.4, GSOI=-5° A T D C , 1200 rpm.  73  6.2 Variable Pilot and Natural Gas Injection The tests presented in this section were accomplished by varying the injection of both fuels to maintain a constant point of 50% heat release. This was done as HR50 was previously shown in section 5.2 to be a marker for efficiency and N O production. This provides a consistent x  comparison for emissions and efficiency. The influence of timing on RIT effects was examined at 1200 rpm, for of HR50 at +5 and +15°ATDC held within 0.5°. The increments for the relative timings were chosen with baseline points of -0.5, 0, 1.8 (standard), 3 ms plus two variable timings, one short and one extended. These points are illustrated in Figure 6-7 with ' A ' and ' B ' denoting the short and extended timings respectively. The GSOI was held constant and the FIR50 was determined for RIT of 1.8 ms from real-time heat release analysis on the D A Q computer. The two relative timings where HR50 shifted +0.5° relative to that timing were arbitrarily chosen as the variable timings. These variable points were dependent upon engine conditions.  4.0  J  |n  ,  ,  ,  ,  ,  ,  0.0  0.5  1.0  1.5  2.0  2.5  3.0  3.5  1  4.0  1 4.5  Relative Injection Timing (ms)  Figure 6-7  Illustration of the determination of the variable relative injection timing settings.  At 1200 rpm, the variable timings were 2° and 29°c.a. at HR50 of +5°, and 6° and 29°c.a. at HR50 of +15°. By using the same strategy at 800 rpm, HR50 of +15°, the timings were 3° and 25 °c.a.. The extended variable RIT was utilized as an arbitrary limit of how long the relative timing may be before substantial deterioration of natural gas ignition. The different limits at different operating conditions suggest that operating conditions influence the effects of relative injection timing. The influence of speed on RIT effects was examined by comparing 800 rpm  74  and 1200 rpm at a constant H R 5 0 of +15°ATDC. A s negative relative timings caused audible knock and degradation of emissions, the minimum R I T at 800 rpm was set to 0 ° . 6.2.1 Performance To compare performance, relative injection timing was varied to determine impact on efficiency, burn duration, and variability of 50% heat release. The indicated specific fuel consumption is constant within experimental error for all relative timings for each absolute timing at 1200 rpm as shown in Figure 6-8. A s noted earlier, a slightly negative relative timing resulted in the shortest burn duration. A s burn duration becomes shorter, the engine performance becomes closer to the ideal otto engine cycle. A s such, it was expected that efficiency would improve with the most rapid burning. However, it appears that the change in burn duration is not significant enough to alter engine efficiency for this fuelling rate. This may be due to incomplete combustion or the change in burn duration is not large enough to affect efficiency. Despite the lack of change in efficiency for this fuelling rate, shorter burn duration has the potential to improve efficiency and should be tested at other loads and higher speeds. The results of this study differ from the findings of Dumitrescu [8], which found decreased thermal efficiency for short relative timing. It was noted by Dumitrescu [8] that the decreased thermal efficiency may have been due to injector limitations at short RIT, rather than due to the combustion effects of a short RIT. 200 r 180 160 140 $ 120 | o)100  L? 80 co  10 15 20 Relative Injection (°CA) Figure 6-8  25  30  Efficiency for various relative injection timings at 0=0.4,1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  75  A s shown in Figure 6-9, the longest burn duration at 1200 rpm occurs at a relative timing of 13°c.a. (1.8 ms) for both absolute timings. The minimum burn duration occurs at a negative relative timing of -4° for both absolute timings. The minimum burn duration is lower at H R 5 0 of +5° than +15, presumably due to the higher cylinder pressures and temperatures which promote combustion. The results indicate that there is little influence of changing absolute injection timing on burn duration for zero RIT. It is apparent that R I T does not affect the short burn duration for retarded timings that was noted in chapter 5. The effect of R I T on burn duration is shown for two speeds in Figure 6-10. The results are shown in absolute time format and also in crank angle format. It is apparent that burn duration does not collapse onto a single trend versus R I T for either crank angle or absolute time. The relative timing effects at short R I T are dependent on the diesel pilot autoignition delay and the distance the natural gas jet travels to reach the burning pilot. Diesel autoignition delay is strongly dependent on pressure and temperature, which are weakly affected by speed. The gas jet speed is the same, regardless of engine speed. Both of these factors are mostly time dependent phenomenon and this is apparent in Figure 6-10-a for short RIT, where the proportional decrease in burn duration appears to match on an absolute time basis. The effects at long R I T probably have more to do with mixing and cooling of the diesel pilot after combustion, where mixing is roughly proportional to speed. This is apparent in Figure 6-10b, where after the maximum burn duration for each speed, the slope of the burn duration vs. R I T seem to match. Therefore, it appears the characteristics of burn duration at short R I T are time dependent, while the burn duration characteristics at long R I T are speed dependent.  76  -  20  <»  c)  c  >  (  <>  t> - -<  5  HR50 + +5°  o +15° Oi  -5  Figure 6-9  1 1  0  I 1  5  1 1  1 1  1 1  10 15 20 Relative Injection (°CA)  1  1  25  —  i  30  =^  35  Comparison of burn duration for various relative injection timings at <b=0.4, 1200 rpm, with H R 5 0 timings of +5 and + 1 5 ° A T D C .  77  22 20-  18 16  § 12h  g1t>r  I  Si  CD  642 0 -  1  2  3  4  Relative Injection Timing (ms)  22 20 18 C)  16 2514  §12  o  10  c 3 8 m  5  Figure 6-10  10 15 20 Relative Injection (°CA)  25  30  Burn duration for various relative injection timings at two speeds, ct)=0.4, H R 5 0 = + 1 5 ° plotted against: (a) absolute timing, (b) crank angle  The effect of R I T on variability appears similar for both absolute timings tested at 1200 rpm as shown i n Figure 6-11. Only at a negative R I T does the variation appear to be lower an earlier absolute timing. This could be due to higher in-cylinder temperatures which should  78  promote more consistent ignition of the natural gas. The influence of speed on combustion variability due to R I T is shown in Figure 6-12. It appears that with short RIT, the variability at 800 rpm is slightly lower than that of 1600 rpm. There is not much difference in variation at extended RIT.  +5 ° ATDC +15 ° ATDC  0.9 0.8 <0.7  Q  10.4 "D  | 0.3 0.2 0.1 0 Figure 6-11  10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm.  1  * O  0.9  800 rpm 1200 rpm  _o_  0.8 o  <0.7  CD u - u  Q  I  o  0.4  "D  | 0.3 0.2  -Q.  o -O-  0.1 0  Figure 6-12  10 15 20 Relative Injection (°CA)  25  30  Standard deviation of 50% heat release for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C .  79  6.2.2 Emissions Nitrogen Oxides The effects of relative injection timing on N O emissions at 1200 rpm is shown in Figure x  6-13. The minimum N O emissions for both absolute timings occur at relative injection timing of x  13° (1.8 ms). A s R I T is extended from 13°, there is a slight increase in N O emissions. A s x  relative injection timing is shortened from 13°, there is a dramatic increase N O emissions. The x  effects of R I T on N O emissions are stronger for the H R 5 0 of +5 as compared to the H R 5 0 of x  +15. This likely has to do with the lower in-cylinder temperature at later absolute timing. It is interesting that at negative R I T the apparent heat release rate is lower for than for 0 RIT, but the N O emissions are higher. A n explanation for this is that the natural gas mixture is more x  voluminous before ignition at very negative relative timings. Once burned, the products of the mixture may take longer to mix with the cooler bulk air, which allows more time for N O  x  chemical kinetics to proceed. The effects of R I T on N O emissions at different speeds is shown in Figure 6-14. The x  minimum N O production at 800 rpm is level between 9° and 14° relative timings. A t 800 rpm, x  the N O increases from a minimum of 6 g/kW-hr to 12 g/kW-hr at 0 RIT. A t 1200 rpm, the N O x  x  increases from a minimum of 4.4g/kW-hr to 8.6 g/kW-hr. It appears that the effects of relative timing on N O emissions are greater at lower speeds. This is probably due to increased mixing x  rates at higher speeds. Dumitrescu [8] found that short R I T increased N O emissions while holding P S O I x  constant. Chapter 5 illustrated that shifting combustion earlier in the cycle increases N O  x  emissions. It is not surprising that the 3° shorter R I T in [8], which shifted the gas combustion earlier in the cycle, resulted in increased N O emissions. The results from the current study x  indicate that the increased N O found by [8] would be partly due to premixing, and partly due to x  shifting the combustion event.  80  5  Figure 6-13  10 15 20 Relative Injection (°CA)  Comparison of N O production for various relative injection timings at two absolute timings, <b=0.4, 1200 rpm. x  14  12  +  Speed 800 rpm  o  1200 rpm  10  I ° O  b  5  Figure 6-14  10 15 20 Relative Injection (°CA)  25  30  N O emissions for various relative injection timings at two speeds, x  0=0.4, H R 5 0 = + 1 5 ° A T D C .  81  Total Hydrocarbons and Carbon Monoxide The results of the effects of relative injection timing on both total hydrocarbons and carbon monoxide are shown in Figure 6-15 for two absolute timings. A s with N O emissions, the x  minimum C O and T H C emissions occur at an R I T of 13°. A t H R 5 0 of +5°, changing R I T increases both T H C and C O emissions for both extended and negative relative timings. Relative timing appears to affect C O and t H C emissions more strongly at HR50+15. The in-cylinder pressures and temperatures are lower for the later absolute timing, making it more difficult to ignite the natural gas mixture. In addition, there is less time for T H C s and C O to oxidize at late absolute timings. The presence of more incompletely burned products at short R I T for late absolute timing indicates that combustion is less efficient and as a result, the corresponding shorter burn durations are less likely to improve engine efficiency (at late absolute timings). The more moderate levels of T H C and C O emissions for short R I T at early timings provide more opportunity for the short burn duration to improve efficiency (at early absolute timings). The simulations of Ouellette [23] found high levels of C O formation with no ignition delay. This was mostly due to impingement of the gas jet on a wall as the flame could not reach the end of the jet before the jet reached the wall. The 1200 rpm experimental results of Dumitrescu [8] indicated that higher levels of C O and T H C emissions occur with a 3° RIT, as compared to a 6° RIT. The trends of C O and T H C emissions in this study agree with trends at short R I T found by Dumitrescu [8] and C O results of Ouellette. The effects of R I T on T H C and C O emissions at different speeds is shown i n Figure 6-16. The minimum T H C and C O emissions at 800 rpm occur at the same relative timing as with N O emissions between 9° and 14°. The effects of changing R I T on T H C and C O emissions appears similar for both speeds.  82  x  7  Mol/HR50 m.  + THC / +5  I  _  1  I  o  THC/+15  •  CO /+5  v  CO /+15  A  i i i  1  1 1  1 0  Figure 6-15  !  >  —5  1  0  5  1  m  1  ^  .  I  10 15 20 Relative Injection (°CA)  —  25  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two absolute timings, 0=0.4, 1200 rpm.  + THC / 800 rpm  nx  I*  _ i i  o  THC/ 1200 rpm  •  CO / 800 rpm  v  CO / 1200 rpm  tn c  • 3 Q  tn  E  LU  . j Figure 6-16  0  ll  i 5  10 15 20 Relative Injection (°CA)  I i 25  I «  m  30  Carbon monoxide and total hydrocarbon emissions for various relative injection timings at two speeds, 0=0.4, H R 5 0 = + 1 5 ° A T D C  83  Particulate Matter There was no statistical change i n P M emissions for the relative timing sweeps at each operating condition. The experimental error masked any effect o f relative timing, partly because the P M emissions at this equivalence ratio are very low. To determine the effect o f RIT on P M , this study should be repeated for an engine operating condition where the engine is known to produce higher levels o f P M .  Overall Emission Discussion To examine a perceived correlation between emissions and burn duration, the correlation coefficient between burn duration and emission species was calculated using: CovjX, Pr  v  Y) (Eq. 6.1)  =  **• y Where o is the standard deviation for each variable and the Cov is given by: Cov(X,  Y)=  -^(x i=\ t  uJO,.- ( V  (Eq. 6.2)  n  The correlation coefficient for each emission species, averaged over each mode tested is listed i n Table 6.1. It is apparent that the maximum burn duration corresponds to minimum emissions when changing relative timing at all speeds and absolute timings tested. This likely has to do with increasing the proportion o f pre-mixed combustion, which burns faster, but causing more emissions. Burn duration could be used for determining optimal R I T , without requiring lengthy emission data samples and analysis, although this should be verified. The flat optimal emission/ burn duration response to changing relative timing at 800 rpm suggest that more pilot is injected than required for good ignition. This can be inferred as an optimal minimum amount o f pilot w i l l likely have an distinct maximum burn duration versus R I T . The accuracy o f the burn duration measurements can be improved by averaging more cycles o f in-cylinder data. This correlation w i l l not hold for extreme negative relative timings where the burn duration increases or for extremely long R I T where the ignition o f the natural gas is very weak. The exact limits for good correlation are unknown. This correlation is only valid for a fixed operating condition.  84  Table 6.1: Burn Duration Correlations Emission Species  Correlation Coefficient with Burn Duration  Carbon monoxide  -0.97  Nitrogen oxides  -0.98  Total hydrocarbon  -0.96  Hypothetically, smaller quantities of pilot in the current configuration would result in a smaller window of good ignition. This is because a smaller quantity of diesel pilot w i l l cool faster and thereby reduce the ignition capability. Higher speed or high swirl may require more amounts of pilot, as the pilot w i l l disperse and cool faster than at low speeds. Conversely, the amount of diesel may be reduced at low rpm as it does not disperse as quickly. A l s o , R I T may become significant at high E G R , as E G R w i l l increase the ignition delay [14], which w i l l affect ignition strength due to mixing. The minimum amount of pilot is a complicated problem involving engine conditions plus injection duration, injection rate and shape, injector configuration, and nozzle shape. The sheer number of variables involved suggest that numerical simulation studies be conducted to optimize the diesel charge before experiments. Noting that the diesel accounts for as little as 3% of the total fuel (at high load), further reduction of pilot would result in only nominal emission reductions due to fuel substitution. However, optimized pilot charge may improve natural gas ignition, where minimal pre-mixing of the natural gas should result in the lowest possible N O  x  formation. The emission results indicate that optimal timing between gas and pilot injection occurs at approximately 13° and 8-14° for 1200 and 800 rpm respectively. The optimum timing for both speeds included the 'normal' timing of 1.8 ms. The emission response was more flat in terms of absolute time at 800 rpm than at 1200 rpm. This is likely influenced by enhanced mixing at higher speeds. The fact that optimal emissions resulted at delayed sequential injection concurs with diesel/CH4 simulations by [23], which determined that an injection delay of gas after diesel reduces chance of fuel over-mixing and impingement. In those simulations, increasing the injection delay from zero decreased C O formation and reduced unburned C H 4 .  The results of  these experiments indicate that a minimum of premixed combustion, with good ignition of the natural gas, brings about the best possible emissions.  85  6.3  Summary •  There is an optimum relative injection timing (or range of timings) for a minimum of all emissions at fixed operating condition, the optimum relative timing appears to be independent of speed or absolute injection timing.  •  The optimum relative timing included the 'normal' 1.8 ms (9° and 13° for 800 and 1200 rpm respectively). The emission response was flatter at 800 rpm than at 1200 rpm on an absolute time basis.  •  The efficiency is not significantly affected by relative injection timing. However, the ability to reduce burn duration provides potential for improved efficiency at early absolute timings.  •  When 50% heat release is held constant, burn duration correlates well with emissions.. The longest burn duration is produced by combustion most favorable for emissions.  •  For short relative injection timings, an optimal delay avoids premixed combustion-i.e. it is desirable for the diesel pilot to be ignited before the N G is injected. It seems that combustion appearing more pre-mixed generates worse emissions. This event is more dependent on absolute time than by crank angle.  •  Agreement with the emission trends at short relative timings with Dumitrescu [8], however the efficiency findings between studies differ. The difference i n thermal efficiency found i n that study was likely due to injector design.  •  A t extended relative injection timings, apparent heat-release data indicate that natural gas ignition by pilot is weakened by mixing of the pilot, therefore higher speeds and high swirl engines may require more diesel for good ignition of the natural gas.  •  Future study should include simulation of the diesel pilot with a focus of improving ignition of the natural gas.  •  The effect of relative injection timing on particulate matter emissions should be studied at conditions generating more particulate matter (higher equivalence ratios). 86  7. INJECTION PRESSURE This chapter presents the effects of changing injection pressure on engine performance and emissions. The injection pressure of the fuel influences the fuel penetration, injection rate, and mixing rate. These in turn influence the fuel combustion, performance and emissions of the engine. The experiments in this section were conducted at two injection pressures and various absolute timings. The variation of injection timings are required to understand the relationships between emissions, injection pressures, and shifts in combustion events. Three sets of tests were conducted: constant air/cycle for 3 fueling rates at one speed; constant equivalence ratio for 3 air flow rates at one speed; one equivalence ratio (ty) and constant air/cycle for 3 speeds. The 50% heat release crank angle is used as the independent variable for comparison. 7.0.1 G a s Injection M e c h a n i c s Natural gas is injected into the cylinder through a small nozzle at sonic or near-sonic conditions. The Reynolds number of the jet at sonic velocities is approximately 5 x l 0 , which is 6  fully turbulent. A s the jet penetrates into the cylinder it quickly mixes with the chamber air. The goal of injection is to achieve proper air utilization while avoiding over-penetration, which is fuel impingement onto the piston bowl or the cylinder walls [14]. It is also desirable to avoid underpenetration as it results in poor air utilization. Ouellette [23] noted that over-mixing of gas jets is also a potential issue for pilot ignition, defined as too much lean mixture prior to ignition. That study showed increasing injection pressures increases injection rate, mixing, and also gas jet penetration. The critical pressure ratio for sonic injection of methane at 300K is 1.84, based on the perfect gas law, injection-to-cylinder pressure ratios above this w i l l result in an underexpanded jet. The experiments by Ouellette [23] also found that the jets observe selfsimilarity with no discontinuity at sonic injection. This means that the gas jet w i l l exhibit consistent characteristics above and below the critical pressure ratio. In the case of operation at high load, 1200 rpm, the in-cylinder pressure approaches that of the injection pressure. The peak cylinder pressure reaches 18.5 M P a at 5° A T D C while the injection duration is from -10° to +7° A T D C . It is evident that the injection at this point is no longer choked flow, as the pressure ratio is near unity. The choked state of the gas jet is dependent on the speed and load of the engine as well as timing and pressure of the gas injection.  87  However, from a practical perspective, the injection pressure should be kept as low as possible to minimize energy required for compressing the natural gas. 7.0.2 C o m b u s t i o n Effects Increasing the gas pressure boosts the rate of fuel injection, which should enhance mixing rates, and thereby increase the heat release rate. A n increase of injection pressure causes an increase in momentum injection rate, which thereby mixes the gas faster. A l s o the fuel injection rate increases, which injects more gas in the cylinder for a given crank angle (within the injection duration). The effects of changing injection pressure on heat release rate is shown in Figure 7-1 for <|) of 0.5 at a fixed start of gas injection (GSOI). It appears that the pilot combustion and the ignition of the natural gas is the same for both injection pressures. However, at the higher injection pressure, the heat release rate appears to increase faster (steeper slope) than at the low injection pressure. This is probably an indication of increased mixing rates. A l s o , the maximum apparent heat release rate is higher for increased injection pressure. The heat release rate for 19 M P a injection pressure shows a distinctive 'double hump', which indicates a progression from pre-mixed to mixing-limited combustion. However, for 23 M P a , the 'double hump' occurs more rapidly and at a higher rate. This is probably due to a combination of the increased injection and mixing rates. It also appears that the burn duration is shorter for higher injection pressures. This is confirmed by the results shown i n Figure 7-2, which shows a general reduction i n burn duration with increasing injection pressure of approximately 7°. A s the G S O I was the same for both injection pressures, it is apparent that the 50% heat release shifts earlier in the cycle for a higher injection pressure.  Figure 7-1  Apparent heat release rate for two injection pressures at, 1200 rpm, 0=0.5, G S O I 0° and 4 g/cycle of air. 88  5 10 15 50% Heat Release (°ATDC)  Figure 7-2  Burn duration for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  7.0.3 V a r y i n g C h a r g e A i r M a s s The effects of injection pressure are influenced by the ratio of the injection pressure to i n cylinder pressure. A s such, the in-cylinder pressure, or charge air mass, was varied. B y varying the charge air mass or variable supercharging, emissions and performance can be affected, without changing injection pressure. Supercharging influences the in-cylinder mixing rates, where the mixing rate is proportional to P  c y  i  1 / 4  for a constant injection momentum[23]. To  examine the effects of variable supercharging, equivalence ratio was held constant at ty of 0.4 while charge air mass was varied between 3 and 6.5 g/cycle. Emission results at 6.5 g/cycle of air are not reported due to inaccurate air flow rate measurements. Studies of supercharging with diesel found that increasing supercharging with rapid injection rates allowed for N O reductions x  with few penalties from other emissions [38]. Supercharging is effective for diesel N O reductions as the increased mixing and shorter diesel ignition delay reduces premixed combustion.  89  x  7.1  Performance The effects of injection pressure on the performance of the engine is demonstrated by  comparison of efficiency using Indicated Specific Fuel Consumption (ISFC) across a range of injection timings. The independent timing variable used for I S F C comparison is the 50% heat release crank angle (HR50). Three different parameters were varied with injection pressure; equivalence ratio, speed, and charge air mass. The equivalence ratio tests were conducted at 1200 rpm fortyof 0.3 and 0.5, with various injection timings, and injection pressures of 19 and 23 M P a . The second set of tests were conducted attyof 0.4 for speeds between 800 and 1600 rpm. Equivalence ratio was held constant as it was previously found to affect emissions. The last set of tests were conducted at 1200 rpm and a constant ty of 0.4, while varying the charge air mass between 3 and 6.5 g/cycle. The relative injection timing (RIT) was held constant at 1.8 ms, while the absolute injection timing was varied in 5° increments. The effects on performance of increasing injection pressure from 19 to 23 M P a at different equivalence ratios is shown i n Figure 7-3. The I S F C trends appear statistically the same with changes in injection pressure for the range of fueling tested at 4 g of charge air. The results of changing injection pressures at different speeds are shown in Figure 7-4. It is reaffirmed that the gross efficiency near optimum timing is greater at 1600 rpm than 800 rpm. It is apparent that when considering I S F C in terms of H R 5 0 , injection pressure does not affect efficiency.  200 180 160 _140 x: § 120 |ioo  —  LL  T  « 80  inj  60  * °  0.3/19 MPa 0.3 / 23 MPa  40  v  0.5/19 MPa  20  a  0.5 / 23 MPa  °5 Figure 7-3  d> / p..  5 10 15 50% Heat Release ("ATDC)  20  25  Efficiency for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  90  200  160  I 140  120  S100 o  *  co 80 60 40 20  Figure 7-4  RPM / P. . J 800 / 19 MPa ln  LL  -5  0  D  800 / 23 MPa  v  1600/ 19 MPa  o  1600/23 MPa |  "~5 10 15 50% Heat Release (°ATDC)  20  25  Efficiency for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  The effect of changing the injection pressure at different charge air masses with 0 of 0.4 is shown in Figure 7-5. The air supply for 6.5 g/cycle was a reciprocating compressor and the screw compressor was used for 3g/cycle. The efficiency at 3g/cycle was unaffected by injection pressure. There was no effect of injection pressure on efficiency trends at late timings. However, for an early injection timing at 19MPa, efficiency was lower than comparable timings at 23 M P a injection pressure. The poor efficiency at early timing, 19 M P a injection, was due to the high incylinder pressure, which reached 18.4 M P a at 5° A T D C , while the injection duration was from lOo to 6.3oc.a.. The low pressure ratio between the injector and cylinder that was reached likely results in poor mixing and a long burn duration, which is less efficient. A s a comparison, the burn duration at 19 M P a is 32°, whereas for 23 M P a injection pressure, the burn duration is only 23°. This indicates that injection pressure has an effect at high loads and early injection timings when the in-cylinder pressure approaches injection pressure.  A comparison between the efficiency  curves at different charge air masses indicates that the efficiency is greater at higher charge air mass. This was likely due to the dramatic load difference between 3 and 6.5 g/cycle which was 6.5 bar and 13.5 bar I M E P respectively. A t high loads there was likely a smaller percentage of energy lost to heat transfer. The I S F C was improved at high pressure despite the increased burn duration at high load, with the exception of operation at 6.5 g/cycle of air, 19 M P a , G S O I - 1 0 ° .  91  200 180 ! _ l  .g_^____^_§.l  11  £  160 140  | 120  3100 o  Air / Pinj  W 80  +  3g/19MPa  60  •  3g/23MPa  40  v  6.5g/19 MPa  20 -5 Figure 7-5  0  o 6.5g / 23 MPa [ 5 10 15 20 50% Heat Release (°ATDC)  25  Efficiency for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  The earliest timing for 19 M P a injection pressure, 6.5g/cycle supercharging where efficiency appears relatively unaffected, is G S O I = 0 ° . The in-cylinder pressure remained almost constant at 14 M P a throughout the injection duration of 10.5°c.a.. Assuming there is only a small amount of losses through injection, the pressure ratio between injector and cylinder is approximately constant at 1.37 for this engine condition. A pressure trace for this condition is included in Appendix J. The pressure ratio went from 2.55 down to 1.02 for the condition of GSOI=-10°, <)) of 0.4, 6.5 g/cycle of air and 19 M P a injection. Therefore, efficiency appears unaffected when minimum ratio between the injection and in-cylinder pressure remains greater than 1.37. However, the true limit is not defined by these experiments, but lies somewhere between pressure ratios of 1.02 and 1.37. A s the pressure ratio is less than the critical 1.84, apparently sonic injection is not necessary from an efficiency perspective for this engine, but the lower pressure ratio may affect emissions due to penetration and mixing effects. The studies of injection pressure by Dumitrescu [8] found the best thermal efficiency averaged over all loads, was at intermediate pressure. The injector used in that study did not allow for independent control of the critical timing of gas injection. A s such, the pilot start of combustion and R I T were held constant, which allowed the centroid of gas injection to vary with injection pressure. Douville [28] found the best thermal efficiency at high load was also an  92  intermediate pressure with simultaneous injection. Results from the current study suggest that such variation of the gas injection would have caused the centroid of combustion to vary with injection pressure, possibly on either side of optimum timing. This may explain the optimum intermediate injection pressure found by Dumitrescu[8] and Douville [28]. B y comparing I S F C with 50% heat release crank angle, it is shown that the injection pressure does not generally affect the combustion efficiency at the conditions tested, except for high load, early timings where the in-cylinder pressure approaches injection pressure. Another consideration that was not encountered in these experiments is the desire to avoid overly late combustion due to slow injection at high speeds as identified by [7], which notes a sufficient injection pressure is required to avoid this scenario. 7.2 Emissions The effects of injection pressure on engine emissions are presented for N O , C O , T H C , x  and P M emissions. For consistency, a l l emissions results in this chapter are compared using timing of 50% cumulative heat release (HR50). The first set of tests that are presented were conducted at 1200 rpm for equivalence ratios of 0.3 and 0.5 with various injection timings for injection pressures of 19 and 23 M P a . The second set of tests at 1200 rpm included varying the charge air mass between 3 and 4 g/cycle at a constant (j) of 0.4. The equivalence ratio was held constant as emissions were previously found to change with equivalence ratio. The last set of tests presented were conducted at 0 of 0.4 for speeds between 800 and 1600 rpm. 7.2.1 Nitrogen Oxides The results of changing injection pressure on specific N O emissions at different x  equivalence ratios is shown in Figure 7-6. The N O emissions for both injection pressures and x  both equivalence ratios follow the same trend when plotted against 50% heat release. This implies that the increased combustion rate due to increased injection pressure does not adversely affect N O production. This differs from the studies by Dumitrescu [8] and Douville [28], which x  indicated increased N O with increasing injection pressure. Neither study however, accounted for x  changes in injection rate. It seems that the N O formation does not depend on the combustion x  duration or intensity, but primarily depends on when the combustion occurs within the engine cycle. This also indicates that the amount of premixed combustion does not change significantly  93  with the increase in backpressure. This can be inferred by remembering from chapter 6 that a short R I T resulted i n premixed combustion and correspondingly more N O x emissions.  25  r——~  i i  i —  * T  5  I i__  8  / P. .  inj  *  0.3/19 MPa  °  0.3/23 MPa  v  0.5/19 MPa  •  0.5 / 23 MPa  i 1  5  —  $  5 o,  -5  Figure 7-6  0  5 10 15 50% Heat Release (°ATDC)  20  25  Oxides of nitrogen emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  The results of increasing injection pressure on N O emissions at different charge air x  masses for 0 0.4 is shown in Figure 7-7. The N O emissions follow the same trend within x  experimental error for both charge air masses and injection pressures. Thus, supercharging does not appear to affect H P D I natural-gas N O emissions significantly. L i k e l y this is because much x  of the combustion is already mixing-limited and the portion of premixed combustion is relatively unaffected by slight changes in the mixing rate. Further investigation is required to confirm this hypothesis for a wider range of charge air masses. A s it is demonstrated that injection pressure has no significant effect on N O production using H R 5 0 as an independent variable, no further x  discussion is included for different speeds. A confirming figure is found in Appendix J . While the minimum N O emissions at very retarded timings (HR50 > +15°c.a.) are within experimental x  error, the N O levels appear slightly higher for 23 than 19 M P a injection for all cases. Perhaps a x  study using finer instrumentation focussed at very late injection, a small amount of N O reduction x  may be obtainable by decreasing injection pressure. These changes would be small compared to retarding the injection timing from piston T D C . This is different from injection pressure results  94  of Stumpp et al.[25] in a diesel engine, where increasing injection pressure significantly increases the minimum possible N O x emissions.  25  .. _  r  I  Air / P. . I  f  h  +  3g/19MPa  D  3g / 23 MPa  v  4g/19MPa  o  4g / 23 MPa  -I  15  I  .  ;  ^ 3c-  I  i  t  5  1  f  If  50% Heat Release (°ATDC)  Figure 7-7  Nitrogen Oxides emissions for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  7.2.2 Total H y d r o c a r b o n s The effect of injection pressure on total hydrocarbon ( T H C ) emissions at different timings and equivalence ratios is shown in Figure 7-8. A n examination of T H C emissions at 0 of 0.5 reveals that there is little difference between the two injection pressures, but at 0 of 0.3, the high T H C emission rate at late timings is exacerbated by increased injection pressure. The increased T H C emissions at retarded timings is assumed to be almost all methane. Justification for this assumption is found in Appendix K . The increased mixing rates may over-lean some of the fuel and/or quench some of the flame at lower temperatures. A t higher temperatures (higher 0), the effect is less significant due to higher average temperatures and less charge air for the flame to mix with. Data in Dumitrescu [8] reveals that increasing injection pressure caused slight increases of C H 4 across all loads at the timing chosen. This supports the hypothesis that increasing injection pressure may over-mix some of the fuel i n H P D I of natural gas. The effects of injection pressure on total hydrocarbon emissions at different speeds is shown in Figure 7-9. The results appear consistent with the mixing hypothesis, as at 800 rpm the  95  T H C emissions are lower than at 1600 rpm, where more mixing occurs due to higher speed. A t slow speeds, adequate time is available for oxidation of the fuel and T H C emissions appear irrespective of injection pressure. A t higher speeds and retarded timings however, the increased mixing effects of higher pressures appear to quench the reaction and there are more unburned hydrocarbons. The equivalence ratio presented in Figure 7-9 is 0.4, as compared totyof 0.3 affected by injection pressure at 1200 rpm. The results for changing the amount of supercharging at a constant equivalence ratio is shown in Figure 7-10. The amount of T H C emissions is unaffected by injection pressure except for 4g/cycle between 12° and 2 0 ° A T D C . There is no statistical difference in T H C emissions between 3 and 4g/cycle, and therefore T H C appear unaffected by changing the amount of supercharging at a constant equivalence ratio. More mixing is expected with 4g/cycle due to a combination of higher in-cylinder pressures and more injection momentum due to more fuel mass injected. The increase in T H C production due to injection pressure is slightly stronger at late injection for 4g/cycle as compared to 3g/cycle. Further study at a wider range of charge air mass is needed to determine i f this is a consistent effect. The fueling rate fortyof 0.4 at 3g charge air is equivalent toty0.3, 4g charge air. It appears there is a greater effect on T H C emissions by changing ty than by changing the charge air mass at a constant ty.  2.5 —  1  <b / p.. Y  1  I  in]  +  0.3/19 MPa  °  0.3/23 MPa  v  0.5/19 MPa  •  0.5 / 23 MPa  I  . . J_  o 1  I  h I 0, -5  Figure 7-8  0  1  ! *i  5 10 15 50% Heat Release (°ATDC)  i 20  25  Total hydrocarbon emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings.  96  2  i  RPM / P. .  1.8 1.6  i  800 /23 MPa  • i i i  v  5.1.2  1  r  1600/ 19 MPa  i  IT  1 i  1  i i  1 1*  o 1600/23 MPa  1  t  I I  O 0.8 x  I I  T-  0.6  f  0.4 0.2 0 Figure 7-9  ;  + 800 / 19 MPa D  1.4  i  J-  I  -r  i * {- U > U  ;  i  -L 1  [ 1  i i  1  "i  1  -5  10 15 50% Heat Release (°ATDC)  20  25  Total hydrocarbons emissions for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  1.8 1.6  +  1.4  •  o 4g / 23 MPa  1  D)0.8  3g / 23 MPa  V 4g/19 MPa  .1.2  o  Air / P.mj. 3g/19 MPa  -4  o X  - 0.6 0.4 0.2 0,  Figure 7-10  5 10 15 50% Heat Release (°ATDC)  20  25  Total hydrocarbons at 0=0.4, for two injection pressures, air flowrates, 1200 rpm.  97  7.2.3 C a r b o n M o n o x i d e The effects of injection pressure on C O emissions as a function of timing for three equivalence ratios at 1200 rpm is shown in Figure 7-11. To clarify the results, the figure includes approximate trend-lines for each test condition and arrows showing the increased injection pressure. A t retarded timings for a 0 of 0.3, C O emissions are unaffected by injection pressure at early timings. A s injection is retarded however, C O emissions are higher for injection at 23 M P a as compared to 19 M P a . This suggests that the C O oxidation process is being quenched due to either increased wall impingement or ultra-lean conditions generated by the enhanced mixing. For a (> | of 0.5, C O emissions are reduced by increasing injection pressure. The higher levels of C O emission indicates poor air utilization, and lower C O emissions indicate improved mixing. It seems that the increased mixing induced by increasing injection pressure improves fuel airutilization at higher equivalence ratios, but can also quench C O oxidation at retarded timing at lower equivalence ratios. A s the C O emissions for retarded timings at 0 of 0.5 are actually reduced with increasing injection pressure, it suggests wall impingement is not a factor at retarded timings. Instead, it is likely that the increases in C O emissions found at <t) of 0.3 are due to quenched oxidation caused by increased mixing. It is interesting that the conditions tested appear to reach the extremes of both air under-utilization and over-mixing. Dumitrescu [8] also concluded that C O was unaffected by injection pressure, however an examination of the data yields a discernible reduction of C O emissions at high load with increasing injection pressure. Therefore, it appears that injection pressure can be used to improve air utilization for engine conditions where C O emissions require mitigation. The absolute value of the C O emissions for all of these cases is very low.  98  0.5  0  i  -5 Figure 7-11  1 5 10 15 20 25 50% Heat Release (°ATDC) C O emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings. Arrows indicate increasing injection pressure. .  1  1  1  1  0  The effects of injection pressure on C O emissions at ty of 0.4 across different speeds is shown in Figure 7-12. Injection pressure does not significantly affect the relationship between C O emissions and injection timing for the speeds tested. The effect of changing injection pressure on C O emissions for different charge air masses at a constant equivalence ratio is shown in Figure 7-13. The C O production is statistically the same for both injection pressures for both charge air masses. A s a result C O emissions seem unaffected by moderate changes in supercharging. A s with T H C emissions, C O emissions trends appear the same at a constant equivalence ratio for different supercharging rates. This suggests that equivalence ratio has more effect on C O and T H C emissions than charge air mass (at constant ty). Considering that the mixing rate changed by less than 10% due to a 25% change in-cylinder density, emission studies should be examined for larger differences in charge air mass. Increasing the supercharging rates is not expected to drastically affect the amount of premixed natural gas combustion, which appeared to generate more pollutant emissions in chapter 6.  99  3 RPM / P. . in] 800 / 19 MPa  +  2.5 2  D  800 / 23 MPa  v  1600/ 19 MPa  o  1600/23 MPa  3 1.5  4 $  O  o •El.  0.5  5 10 15 50% Heat Release (°ATDC) Figure 7-12  20  25  Carbon Monoxide emissions for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  Air / P. .  -  +  3g/19MPa  D  3g / 23 MPa  v  4g/19MPa  () —  . . .  —  o 4g/23MPa i  !  — 1 c ]  -5 Figure 7-13  :i  -  0  )  T  ii <T 1  > L| 1 ii  5 10 15 50% Heat Release (°ATDC)  I-! 20  —  25  Carbon monoxide emissions for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  7.2.4 Particulate Matter The effects of injection pressure on P M emissions at 1200 rpm is shown i n Figure 7-14. Injection pressure does not significantly affect P M emissions at 0 of 0.3. However, the P M 100  emissions are lower around H R 5 0 o f 10°c.a. fortyof 0.5 at 23 M p a injection as compared to 19 M P a injection. This differs from the results of Baribeau [17], which showed improvement at low load, but no change in P M at high load. This may be due to the differences in injection pressures and loads tested between studies. The effects of injection pressure on P M emissions across different speeds are shown in Figure 7-15. A t 800 rpm, there is no apparent effect of injection pressure on P M emissions. A t 1600 rpm both injection pressures exhibit low P M emissions near H R 5 0 of 20°c.a. and then P M increases with late timing. Therefore, it appears that injection pressure has little effect on P M emissions for these conditions. The effects of changing injection pressure and changing the charged air mass on P M emissions attyof 0.4 is shown in Figure 7-16. There is no statistical difference in P M emissions between 3 and 4 g/cycle at 23 M P a injection. A s well, there is no statistical difference between injection pressures at 4 g cycle of air. The P M data for 19 M P a at 1200 rpm, ty of 0.4, 3g/cycle was faulty and not reported. A s the results indicate there is statistically no effect of injection pressure on P M emissions for ty of 0.4 at any other condition, there is likely no effect of pressure on P M for 3g/cycle attyof 0.4. 20 18  < > t / p.. T  inj  4- 0.3/19 MPa  16  o 0.3 / 23 MPa  14  V 0.5/19 MPa  ?12  I  0.5 / 23 MPa  •  "5.10 E, 2 8 Q.  6 4 2 5 10 15 20 25 50% Heat Release (°ATDC) Particulate matter emissions for two injection pressures at two equivalence ratios, 1200 rpm, 4g/cycle of air and various timings. 0.  Figure 7-14  0  101  35 RPM / P 800 / 19 MPa  +  30  •  25  800 / 23 MPa  V 1600/ 19 MPa  | 20  1600/23 MPa  o  E  r 15  10  5 10 15 50% Heat Release (°ATDC) Figure 7-15  20  25  Particulate matter for two injection pressures at two speeds, 0=0.4, 1200 rpm, 4g/cycle of air and various timings.  Air / P. . mj  [ ]  c]  D  3g/23MPa  *  4g/19 MPa  o  4g / 23 MPa  ti  -  '  7  []  - c)  t c)  [  c)  ]  1  c  >  Ic  c)  0i  -5  Figure 7-16  1  >  1  >  0  1  u t  1_  5 10 15 20 25 50% Heat Release (°ATDC) Particulate matter emissions for two injection pressures at two air flow-rates, 0=0.4, 1200 rpm, and various timings.  102  7.3  Summary • Increasing injection pressure from 19 to 23 M P a : a) increases combustion rate and decreased burn duration; b) shifts the 50% heat release earlier in the cycle for a fixed start of gas injection; c) causes no change in efficiency trends for 4g/cycle of charge air, or 6.5 g/ cycle charge air at late injection timing, when comparing against 50% heat release; d) improves of engine efficiency only for early injection with 6.5 g/cycle charge air; e) causes no change in N O x emission trend when compared at 50% heat release, for all conditions; f) generally decreases C O emissions, but causes no effect on T H C emissions, for <> | of 0.5, at 1200 rpm; g) increases C O and T H C emissions for late injection timings, for ty of 0.3 at 1200 rpm; h) no change i n C O emissions, but increases T H C emissions at 1600 rpm, for 0 of 0.4; i) causes no change in T H C emissions for ty of 0.4, at 800 or 1200 rpm; j) causes no significant change in P M emissions for ty of 0.3 or 0.4; k) decreases peak P M emissions for ty of 0.5, 1200 rpm, which coincides with C O reduction; • The changes in T H C and C O emissions consistent with, and are most likely due to increased mixing rates at higher injection pressure. • Changing equivalence ratio causes more considerable change in T H C and C O emissions than moderately changing the load at a constant equivalence ratio (i.e. supercharging). • The effects on emissions of high rates of supercharging should be tested with an accurate air flow-rate measurement.  103  8. CONCLUSIONS 8.1  Introduction The general focus of this study has been to provide knowledge of how changing the  injection parameters affect the emissions and efficiency of a modified Cummins I S X engine fueled with pilot-ignited, high pressure direct injection (HPDI) of natural gas. This knowledge is needed to develop optimized injection strategies for similar engines. The injection parameters studied include the absolute injection timing, the relative injection timing between the diesel pilot and natural gas, and the injection pressure.  The performance measures studied included the  efficiency, and pollutant emissions of nitrogen oxides ( N O ) , carbon monoxide ( C O ) , total x  hydrocarbons (THC)  and particulate matter (PM).  The engine was a supercharged, single-cylinder engine and several considerations are important for operation, particularly concerning the exhaust back pressure. The effects of back pressure on emissions and efficiency were studied to determine a consistent testing procedure to simulate turbo-charged conditions. Increasing back pressure can decrease N O emissions, though x  these effects are reduced as speed is increased, or injection timing is retarded. Increasing back pressure can significantly affect T H C emissions, where the effects are delayed as equivalence ratio is increased.  Increasing back pressure reduces C O emissions i f they are  greater than  approximately 1 g/kW-hr, where otherwise there is no significant effect. Increasing back pressure affects P M emissions at some conditions, where P M reductions coincide with operating points where back pressure induces reductions in C O emissions. Bearing in mind the emissions, the back pressure was set to 150 k P a (absolute) which, depending on speed, was 15 to 40 k P a lower than intake manifold pressure at time of intake valve opening. The rate of supercharging was also briefly considered. 8.2 Conclusions Based on measurements which cover a range of equivalence ratios between 0.3 and 0.5, corresponding to loads mainly between 6 and 10 bar I M E P , and speeds between 800 and 1600 rpm, the following conclusions have been drawn:  104  8.2.1 Relative Injection Timing The timing between the start of injections of the pilot and natural gas was varied at 800 and 1200 rpm for an equivalence ratio of 0.4. This relative timing was found to significantly affect engine performance and emissions in the following ways: • When the 50% cumulative heat release timing is held constant and relative timing is varied, the burn duration correlates well with N O , T H C , and C O emissions. The maxix  mum burn duration corresponds to the lowest emissions. • The diesel pilot should be ignited before the natural gas is injected to obtain the minimum emissions. The emission trends exhibited at short relative timings agree with data found in Dumitrescu [8]. • There is a relative injection timing that minimizes all emissions, which appears to be independent of speed or absolute injection timing.  The optimum relative timing  included 1.8 ms for all speeds. • For long relative injection timings, apparent heat-release data indicates the natural gas ignition by the pilot is weakened by excessive mixing of the burned products of the pilot combustion. This being so, the minimum pilot required for good ignition of the natural gas is probably greater at higher speeds and high swirl engines. Relative injection timing does not significantly affect efficiency or particulate matter for the conditions tested.  This efficiency result differs from the reduced efficiency found by  Dumitrescu [8] with short relative injection.  8.2.2 Absolute Injection Timing Absolute injection timings sweeps were conducted with a constant relative injection between pilot and natural gas. The timing of 50% cumulative heat release (HR50) was found to be a good independent variable for comparison of N O emissions and efficiency. x  • Indicated fuel consumption as a function of H R 5 0 is almost independent of speed. The optimum timing for efficiency occurs when the 50% heat release is approximately + 5 ° A T D C . A s timing is retarded beyond this, efficiency declines. Equivalence ratio does not affect efficiency significantly. • Specific N O emissions are a strong function on H R 5 0 and are independent of equivax  lence ratio for constant air injection. This indicates that the timing of the combustion  105  event is a critical factor in N O production for the H P D I engine. There is a limit to x  which N O can be reduced by retarding timing, which occurs when the H R 5 0 reaches a x  threshold of approximately 15° A T D C at 1200 rpm. A lower limit for retarded timing N O x reduction is compatible with data found in Dumitrescu [8]. The engine should be operated such that absolute injection timing is not retarded beyond the apparent limit defined by the H R 5 0 at each speed. Beyond this point no further reductions in N O are x  gained and, efficiency is reduced and other emissions deteriorate. • The N O x emitted is proportional to the fuel burned for fixed H R 5 0 , speed, and air flowrate for equivalence ratios between 0.3 and 0.5. • Carbon monoxide emissions increase with retarding gas injection for late timings for (j) of 0.3 and 0.4 at all speeds tested. For (j) of 0.5 at 1200 rpm however, retarding injection causes C O emissions to increase, then decrease and then increase again. • Retarding injection increases T H C emissions for all conditions tested. This is expected as combustion temperatures are lower and less time is available for oxidation as timing is retarded.  The effect is more pronounced at lower equivalence ratios and higher  speeds. • Particulate matter emissions are unaffected by injection timing for equivalence ratios of 0.3 and 0.4 at 800 and 1200 rpm; retarding timing generally increases P M emissions for 0 of 0.4 at 1600 rpm; and retarding timing causes P M to vary i n a similar manner as C O for <j) of 0.5 at 1200 rpm. • Raising speed for various timings: increases specific T H C emissions; does not significantly affect C O emissions; generally increases P M emissions; decreases indicated specific N O production; and delays the timing of the N O reduction limit. x  x  8.2.3 Injection Pressure a n d Supercharging Increasing injection pressure shifts the combustion event earlier in the cycle for a constant start of gas injection, and the H R 5 0 timing was employed for showing the effects of injection pressure on emissions and efficiency. • A s a function of H R 5 0 , efficiency is generally independent of injection pressure. Injection pressure only significantly affects efficiency trends when the in-cylinder pressure approaches injection pressure (at high load, early timing), where increasing injection  106  pressure improves efficiency. The findings differ from the findings of Dumitrescu [8] and Douville [28], who found the best efficiency at an intermediate pressure based on a constant start of injection and did not examine H R 5 0 . • N O emissions trends with H R 5 0 are not significantly affected by injection pressure. x  • Increasing injection pressure generally adversely affects t H C at late timings. • Increasing injection pressure can either increase or decrease C O emissions depending on equivalence ratio and absolute timing. Increasing the injection pressure at an equivalence ratio of 0.4 does not significantly affect C O emissions. The complex dependendence of C O emissions on injection pressure for equivalence ratios of 0.3 and 0.5 is shown i n Figure 7-11. • Increasing injection pressure can decrease P M for an equivalence ratio of 0.5 at 1200 rpm. Particulate matter emissions are not significantly affected at other engine conditions. • Changing the load by 25% while maintaining a constant equivalence ratio of 0.4 with supercharging does not appear to affect any emissions. Changing the load by 25% with a fixed charge air mass did affect t H C and C O emissions, implying that equivalence ratio is an important factor for these emissions. 8.3  Recommendations for F u t u r e Study This study represents a partial understanding of the pilot-ignited H P D I process.  The  operating conditions tested were somewhat limited. Several avenues of investigation may expand understanding and application of pilot-ignited H P D I . • Experiments attempting minimize the diesel pilot should be conducted, using the burn duration as guidance for optimization of the relative timing. This method w i l l be faster than taking emission measurements and requires less data processing. Heat release data may also be useful in determining the minimum amount of diesel for good ignition. Pilot quantities may likely be reduced at slow speeds as compared to high speeds • Injection parameters warrant further investigation for a wider range of equivalence ratios, and particular attention should be paid to the N O versus 50% heat release x  trends. Emissions should be studied for higher supercharging rates with an accurate air-  107  flow measurement. These tests should be conducted with an injector that provides better air utilization at high equivalence ratios (i.e. new injector geometry). Experiments should be conducted as to whether the S A E recommended N O correction x  factor for diesel engines, which was originally determined for diesel fuel, is valid for natural gas. Future investigation should include variation of diesel pilot quantity and alteration of rate shape of pilot injection. Effects of the gas injection rate shape should be studied. The effect of relative injection timing on particulate matter emissions should be studied at conditions generating more particulate matter (higher equivalence ratios).  To  improve accuracy of P M measurements, the test period should be extended. The loose correlation of C O emissions trends with P M emission trends warrants fruther investigation.  108  REFERENCES [I]  U.S. Environmental Protection Agency (EPA). "2000 A i r Quality Trends Report." E P A 454/K-01-001,2000.  [2]  U.S. Environmental Protection Agency (EPA). "Health assessment document for diesel engine exhaust." EPA/600/8-90/057F, 2002.  [3]  www.dieselnet.com/standards.html  [4]  Douville,B., Ouellette,P.,Touchette,A.,Ursu,B., "Performance and Emissions of a TwoStroke Engine Fueled Using High-Pressure Direct Injection of Natural Gas", SAE Technical Paper Series, Paper N o . 981160, 1998.  [5]  Douville,B., Ouellette,P.,Hill,P.,Ursu,B., " N O x Reduction in a Directly Injected Natural Gas Engine", Proceedings of the 1998 Fall Technical Conference of the ASME Internal Combustion Engine Division, I C E Vol. 31-3, 1998.  [6]  Hodgins, K . , H i l l , P., Ouellette, P. and Hung, P., "Directly Injected Natural Gas Fueling of Diesel Engines", SAE Technical Paper Series, Paper N o . 961671, 1996.  [7]  Harrington, J., Munshi, S., Nedulcu, C , Ouellette, P., Thompson, J., and Whitfield, S., "Direct Injection of Natural Gas in a Heavy-Duty Diesel Engine", SAE Fuels and Lubricants Conference May 2002, Paper N o . 2002-01-1630, 2002.  [8]  Dumitrescu, Silviu., "Pilot Ignited H i g h Pressure Direct Injection of Natural Gas Fueling of Diesel Engines," M . A . S c . Thesis, University of British Columbia, September 1999.  [9]  McTaggart-Cowan, Gordon., "The Application of Exhaust Gas Recirculation to a Single Cylinder Compression Ignition Engine Fueled with Natural Gas", M . A . S c . Thesis, University of British Columbia, December 2001.  [10]  C o x , Jim., Personal Communication,  [II]  Borman, G., Ragland, K . , Combustion Engineering.  [12]  Munshi, Sandeep., Personal Communication,  [13]  Huang, J., H i l l , P., Bushe, K . , Munshi, S., "Experimental Shock Tube Study of Methane Ignition under Engine-relevant Conditions", Unpublished-currently under review for Combustion and Flame Journal, 2002.  [14]  Hewood, John., Internal Combustion Engine Fundamentals. 1988.  B . C . Gas, October 2002.  109  M c G r a w - H i l l , N e w York, 1998.  Westport, October 2000.  M c G r a w - H i l l , N e w York,  [15]  Warnatz, J, Maas, U . , Dibble, R.W., Combustion-Physical and Chemical Fundamentals, Modeling and Simulation, Experiments, Pollutant Formation. 2nd Ed., Springer, Berlin 1999.  [16]  Easley, W . L . , and Mellor, A . M . , " N O Decomposition in Diesel Engines", SAE Technical Paper Series, 1999-01-3546.  [17]  Baribeau, A . M . (2000) Master's Thesis: P M Emissions from a Cummins I S X 400 Engine Operating on High-Pressure Direct Injection of Natural Gas with Diesel Pilot. University of British Columbia, Vancouver, B C .  [18]  Wakenell, O ' N e a l , and Baker., "High-Pressure Late Cycle Direct Injection of Natural Gas in a R a i l Medium Speed Diesel Engine", SAE Technical Paper Series, Paper N o . 872041, 1987.  [19]  Lorn, E . , Ly, K . , " H i g h Pressure Injection of Natural Gas in a Two Stroke Diesel Engine", SAE Technical Paper Series, Paper N o . 902230, 1990.  [20]  Miyake, M . , B i w a , T., Endoh, Y., Shimotso, M . , Mukararu, S. and Kumoda, T. "Development of H i g h Efficient Gas Burning Diesel Engines", C I M A C Paper, Conference Proceedings, Paris (1983).  [21]  Meyers, D.P., Bourn, G.D., Hedrick, J . C , Kubesh, J.T., "Evaluation of Six Natural Gas Combustion Systems for L N G Locomotive Applications", SAE Technical Paper Series, Paper N o . 972967.  [22]  Hodgins, K . , Gunawan, H , H i l l , P , "Intensifier-Injector for Natural Gas Fueling of Diesel Engines", SAE Technical Paper Series, Paper N o . 921553, 1992.  [23]  Ouellette, Patric, "Direct Injection of Natural Gas for Diesel Engine Fueling," P h . D . Thesis, University of British Columbia, A p r i l 1996.  [24]  L i , G , Ouellette, P., Dumitrescu, S., H i l l , P., "Optimization Study of Pilot-Ignited Natural Gas Direct-Injection in Diesel Engines", SAE Technical Paper Series, Paper N o . 1999-013556, 1999.  [25]  Stumpp, G , Polach, W., Muller, N . , Warga, J., "Fuel Injection Equipment for Heavy Duty Diesel Engines for U . S . 1991/1994 Emission Limits", SAE Technical Paper Series, Paper N o . 890851, 1989.  [26]  Uchida, N . , Daisho, Y. and Saito, T., "The Control of Diesel Emissions by Supercharging and Varying Fuel-Injection Parameters", SAE Technical Paper Series, Paper N o . 920117, 1992.  110  [27]  Zhang, L . , Takatsuki, T., Yokota., " A n Observation and Analysis of the Combustion Under Supercharging on a D I Diesel Engine", SAE Technical Paper Series, Paper N o . 940844, 1994.  [28]  Douville, Brad, "Performance, Emissions and Combustion Characteristics of Natural Gas Fueling of Diesel Engines", M . A . S c . Thesis, University of British Columbia, A p r i l 1994.  [29]  Taylor, Jay, Personal Communication,  [30]  Brakel, Thomas., "The Effect O f Exhaust Gas Recirculation O n Particulate Matter Emissions F r o m A Compression-ignition, Natural Gas Fuelled Engine", M . A . S c . Thesis, University of British Columbia, A p r i l 2002.  [31]  Park, Sung-Hoon, Personal Communication,  [32]  S A E Recommended Practice, 1995, "Measurement of Carbon Dioxide, Carbon Monoxide, and Oxides of Nitrogen in Diesel Exhaust", S A E J177 J U N 9 5 .  [33]  Randolph, Andrew. "Cylinder-Pressure-Based Combustion Analysis in Race Engines", SAE Technical Paper Series, Paper N o . 942487, 1994.  [34]  Ladommatos, N . , Abdelhalim, S., Zhao, H . , "The Effects of Exhaust Gas Recirculation of Diesel Combustion and Emissions", International Journal of Engine Research, V o l . 1, N o . 1,2000.  [35]  Munshi, Sandeep., Personal Communication,  [36]  Nielson, O., Qvale, B . , Sorenson, S., "Ignition Delay in the Dual Fuel Engine", SAE Technical Paper Series, Paper N o . 870589, 1987.  [37]  Mtui, P., H i l l , P., "Ignition Delay and Combustion Duration with Natural Gas Fueling of Diesel Engines", SAE Technical Paper Series, Paper N o . 961933, 1996.  [38]  Uchida, N . , Daisho, Y , Saito, T. and Sugano, H . , "Combined Effects of E G R and Supercharging on Diesel Combustion and Emissions", SAE Technical Paper Series, Paper No. 930601, 1993.  [39]  Perry,S., Chemical Engineering Handbook, 6th Ed., M c G r a w - H i l l Book Company, 1984, pp. 2-65 and 2-66.  [40]  Grimm, B . , and Johnson, R., "Review of Simple Heat Release Computations", SAE Technical Paper Series, Paper N o . 90044, 1990.  U . B . C . , M a y 2002.  Ill  U . B . C . , November 2002.  Westport, August 2002.  APPENDIX A. CALIBRATION Emission Bench Procedure Alternative Fuels Laboratory, UBC Mechanical Engineering NOTES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.  Parti 1. 2. 3. 4. 5. a. b. c. d.  When calibrating - ensure the flow rate is correct for each step. Do not change the pressure on the bottle regulators (exception-NOx). Order gas when cylinder pressure falls below 500 psi to ensure enough lead time. Do NOT turn off analyzers in cabinet #2 (breakers 2 & 6A) unless a week without testing is anticipated. Turn off Ratfisch FID (heater Oven, Pump, then Power) when not using to minimize bottled fuel and air consumption. If flow rates are all too low - check heated filters in cabinet #1 and replace as necessary. Sample pump must be shut off and the filters cold. If there are still problems with flow rate, check the sample pump (not a trivial procedure). Balance analyzer flow rates once sampling from hot exhaust has commenced (engine conditions may affect flow rates) Wait at least 60s after turning on calibration gas before activating the calibration function. Ensure that the heated enclosure has reached appropriate temperature before sampling. The NOx analyser (API) is challenging to calibrate - be careful! NOx component of Siemens Ultramat 22P is currently disabled (see maintenance records).  -Start-Up Open the compressed gas cylinders fully (beside the fume hood). Turn on breakers in the back of cabinet #1, except NOx converter (breaker 8A) and only select one of breakers 1A, 1B, & 3A (appropriate engine's heated sample line). Select a sample stream from the appropriate engine by turning its valve to "ON" cabinet #1, ensure all other streams are selected to "OFF' Check heated enclosure temperature is set on Ogden dial to 190°C - for all engines. Select the analog output path in Cabinet #2 by sliding all switches to left (Ricardo) or right (SCRE) Turn on the Ratfisch FID: Press "Power" Press "Heater-oven" button Wait 10 minute for the FID oven temperature to reach at least 150 C before attemptng ignition Hold " H Over" button and adjust FUEL to 0.5 bar and AIR to 0.8 2  112  e. f.  Continue holding " H Over" and press "Ignition" button - hold both buttons until the ignition light (button) goes out If ignition does not occur quickly (<20 sees), try switching to "cal gas" and repeat. If the bench has not been used for several days, it may be necessary to purge the H fuel line (remove/replace the fuel hose located on the back of Ratfisch inside cabinet #1) and repeat (e). Plug in NOx vacuum pump (in Fume Hood) 2  2  6.  Part 2 - Calibration 1. 2.  Ensure cabinet #2 has warmed up for at least 1 hour. Ensure the Ratfisch is ignited for at least 40 minutes prior to calibration (other analyzers can be calibrated in the meantime).  ZERO CH4 (Ultramat 22P) and NOx (API) cabinet #2 (NOTE - NOx currently disabled on Ultramat 22P) o Turn NO or CH4 switches to ZERO (linked) o Wait 10 minutes (or for reading to stabilize) for API NOx analyzer: o Press 'TEST' until sample flow rate shows at top of screen o This should read 289-295 cc/min o If flow-rate requires adjustment -use black regulator in back of cabinet #2. CW to increase flow (very sensitive) and CCW to decrease flow, o Memorize flow rate o press CAL o press ZERO o press ENTER for CH4 analyzer: o  Press ">0<" button to zero the gas and again when zeroing is complete. (Note: during zeroing, flow should read ~2.0 L/min. If it doesn't, correct it using the adjustment knob inside the cabinet).  CALIBRATE CH4 Analyzer (Siemens Ultramat 22P) cabinet #2 Range: 1 - 5V for 0 - 5000 ppm o o o  Turn NO switch to RUN and turn CH4 switch to SPAN, Check flow rate is 2.0 L/min. Adjust the potentiometer so that the display reads 3947ppm.  CALIBRATE C 0 2 analyzer (Beckman 880) cabinet #2 Range: 0 - 5V for 0 - 20% o o  Ensure that valve on top of cab. 2 is set to "Ricardo" Turn the C02 switch to zero (top Of cab #2).  113  o o o o o o  Check flow rate is 1.0 LVmin (2 SCFH), (using "Flow to analyzers" knob). Press "Zero" then "Enter". Adjust with arrows to read 0% on left of display, and press "Enter" again, Turn C02 switch to SPAN, Check flow rate is 1.0 L/min (2 SCFH). Press "Span", "Enter". Adjust with arrows to read ~16% on left of display, then press "Enter". Turn C 0 2 switch to RUN. Adjust flow rate to 1.0 L/min (2 SCFH) again  Calibrate (if desired) Low Range C02 (California Analytical) cabinet #2 Range: 0-10 V for 0-2% (1) or 0-10% (2) Note: Valves in back of cabinet should be pointed parallel right for low-range operation (intakedilution C02) and left for high range operation. o o o o o o o o o o o o o  Select range 1 or 2 using knob on analzyer Turn valves on LHS at back of cabinet # 2 to 'exhaust line' (both handles should point left) Turn valve on top of cabinet 2 to "SCRE" Turn C 0 2 switch to ZERO Adjust flow rate to 1 L/min, wait 2 minutes Adjust 'zero' potentiometer so that display reads 0. (can be done while calibrating the Beckman) Turn C02 switch to SPAN Adjust flow rate to 1 L/min Adjust 'span' potentiometer so that display reads 9.08 / 1.80 (depending on whetherrange '1' or '2' is selected) Turn C 0 2 switch to RUN To measure intake/dilution C02, switch valves at back of cabinet 2 to 'intake'. Plug in separate sample pump, Set sample flow rate in back of cabinet #2 to 1 L/min Set the drier air flow rate to 1.5 L/min  Calibrate CO analyzer (Siemens Ultramat 21P)  cabinet #2  Range: 0 - 5V for 0 - 10,000 ppm o Turn the CO switch to ZERO. o Press ">0<" button to zero the gas and again when zeroing is complete. (Note: during zeroing, flow should read 2 L/min. If it doesn't, correct it using the adjustment knob inside the cabinet). o Turn CO switch to SPAN. o Check flow rate is 2 L/min, (adjust using "Flow to analyzers" knob if necessary), o Adjust the CO pot. So that display reads 2077ppm, (i.e. read 1.039 +/- 0.003V on Chessel display - for this ensure that the SCRE DAQ chassis is on). Note about 5s delay between changing pot and response o Turn CO switch to RUN. Adjust flow rate to 2 L/min again if necessary.  114  Calibrate 0 2 analyzer (Oxymat 5E) cabinet #2 Range: 0 - 5V for 0 - 2 1 % Note: very sensitive to flow rate-ensure proper flow rates for every step o o o o o o o o o o o  Turn the 02 switch to ZERO. Check flow rate is 0.7 L/min. Enter ".111" to make Code 1 light go out. Set analyzer to "Calibration" mode by pressing "Meas/Cal" button. Press "5", then press "Enter" to zero the analyzer. Wait until "not ready" light is off. Turn the 02 switch to SPAN. (Check flow rate is 0.7 L/min). Press "8", then press "Enter" to span, Wait until "not ready" light is off. Press "Meas/Cal" button to set analyzer to Measure mode. Turn 0 2 switch to RUN. Adjust flow rate to 0.7 L/min again if necessary.  Calibrate THC analyzer (Ratfisch RS-55) cabinet #1 Range: 0 - 1 0 V for: either Range 4: 0 - 10 000 ppm (Range 3): (0 - 1000 ppm) o o o o o o o o o  o o  (default) (use values in brackets for calibrating this range)  Wait 40 minutes after ignition before calibration, Select appropriate range. Select appropriate calibration gas using the valve at the bottom of cabinet #1. Check that ignition light is OFF. If not, follow steps e and f again (under start-up), Turn large black knob to "ZeroGas" position. Set sample backpressure at 200 mbar, maintain at every step, (sensitive) Adjust "Zero" on Gossen display using potentiometer, Turn large back knob to "CalGas" Set sample backpressure at 200 mbar and turn fuel knob back until Gossen display starts decreasing, then return to the maximum value. (Note: if this value is not -3.5 on the fuel gauge, instrument needs to warm up more before calibrating). Adjust "Gain" pot to obtain a reading of 3947 (253.0 for range3 - note the decimal is burnt out) on the Gossen display. Turn large back knob to "Sample" and reset backpressure at 200 mbar.  Check chiller is cold (<3°C) and heated enclosure is 190°C. Commence sampling from engine. Balance (set) all flow rates - this is an iterative procedure. Last: Calibrate NOx analyzer (API) - next page  115  Last: Calibrate NOx analyzer  (API)  Range: 0 - 5 V for 0 - 3000 ppm  o o o o o o o o o o  Check Range of instrument - should be 3000ppm for Cummins SCRE and 4500ppm for Ricardo. (Press "set", "range", check the range and adjust if necessary, then "enter"). Turn NO switch on front of cabinet #1 to SPAN Adjust regulator ON THE NOx CYLINDER so that the sample flow rate is exactly what was used to zero the instrument, Wait 2-3 minutes NO MORE/NO LESS Press 'SPAN' Display should read 1957 ppm NOx Press 'ENTER' Press 'EXIT' Calibration (yellow) light should go out and Sample (green) light should go on Reset the flow rate to memorized value using black regulator in back of cabinet #2.  FINAL NOTE: C h e c k flow rates before r e c o r d i n g e m i s s i o n data  Ensure that NOx flow-rate does not change more than +/-1 cc/min (each 1 cc/min affects results -0.5%))  116  APPENDIX B. TEST PROCEDURE 1. Start engine 2. Wait for o i l temperature and diesel mass to stabilize 3. R u n engine up at 12 bar load, 1200 rpm for 15 minutes 4. Record ambient conditions 5. Set desired speed with dynamometer 6. Set desired air flow rate 7. Set appropriate timing 8. Set appropriate equivalence ratio with fuel 9. Set appropriate back pressure 10. Check air flow (will be affected by backpressure) 11 .Wait for exhaust temperature to stabilize 12. Take high speed data and data process immediately, check for a reasonable I M E P 13. Record time of day, commanded parameters, o i l temperature, and I M E P 14. Take 300 samples (5 minutes) of data 15. Take high speed data, process immediately and check for identical I M E P Repeat from 4.  117  APPENDIX C. LIST OF ACQUIRED PARAMETERS  Hand Recorded  Data Acquisition System  Date of the test (dd/mm/yy)  IMEP (kPa)  Time  Post Aftercooler Air Temperature (°C)  Mode Number  Supercharger Intake Temperature (°C)  Barometric Pressure [kPa]  Supercharger Exhaust Temperature (°C)  Relative humidity (%)  Pre-aftercooler air temperature (°C)  Ambient air temperature (°C)  Exhaust Manifold Temperature (°C)  PSOI [msBTDC]  Intake Manifold Temperature (°C)  PW_Diesel [microseconds]  Exhaust Backpressure (kPag)  Gas Delay [milliseconds]  Pre-aftercooler Air Pressure (kPag)  P W _ C N G [microseconds]  Post-aftercooler Air pressure (kPag)  Engine Oil Temperature (°C)  Engine Speed (RPM) Dynamometer Torque (N*m) Vector Motor Torque (N*m) Diesel fuel temperature (°C) Natural Gas Fuel Temperature (°C) Diesel Fuel Pressure (Mpa) C N G Fuel Pressure (Mpa) C N G Fuel Flow (kg/hr) Intake Airflow (kg/hr) C0 0  2  - high range (%) (%)  2  CO (ppm) NO  x  (ppm)  THC (ppm) C0  118  2  - low range (%)  APPENDIX D. NATURAL GAS PROPERTIES Table D . l Properties of the Components of B . C . Natural Gas  Compound  Molecular Fraction (%)  Molecular Mass (kg/kmol)  Lower Heating Value (kJ/kg)  Methane  95.945  16.043  50030  Ethane  1.9549  30.070  47511  Propane  0.5547  44.097  46333  i-Butane  0.0689  58.123  45560  n-Butane  0.1116  58.123  45719  i-Pentane  0.0252  72.150  45249  n-Pentane  0.0201  72.150  45345  Hexane  0.0248  86.177  45103  Carbon Dioxide  0.4248  44.010  0  Nitrogen  0.870  28.013  0  The lower heating value is for gaseous components at STP.  The molecular mass and lower heating value of the natural gas are based on the weighted average of the components.  119  Stoichiometric Air-to-Fuel Ratio The stoichiometric air-to-fuel ratio is calculated on a mass-basis from the complete combustion of one mole of natural gas with dry air as follows:  A _ (moles of air) • (molecular mass of air) F (moles of fuel) • (molecular mass of fuel)  (2.038)-(137.36) (1). (16.63)  Natural Gas Summary: Molecular Weight  16.63 kg/kmo  Hydrogen/Carbon Ratio: 3.924 Lower Heating Value:  49110 kJ/kg  Air-to-Fuel Ratio:  16.83  120  =  1 6  -  8 3  APPENDIX E. CHAUVENET'S CRITERION Chauvenet's criterion is a statistical method for determining 'outlier' measurements, which are values determined to be anomalous and discarded. This method assumes that the sample distribution is gaussian (or 'normal'). A reading may be rejected if the probability of obtaining a particular deviation is less than l/2*number of samples, this is expressed by: P{ -x)<±. Xi  where n is the number of samples, x is an individual sample and x is the sample mean. t  For example if 20 samples are taken, samples would be rejected if: P{x -x)< t  0.025  When comparing the z value, the standard normal variable, as defined by:  where a is the standard deviation of the sample. From normal distribution tables and a probability of 0.025, values may be rejected when corresponding z values are greater than 2.24.  121  APPENDIX F. SMOOTHED AND UNSMOOTHED HEAT RELEASE An unsmoothed heat release trace is found in Figure F-l and the smoothed heat release is found in Figure F-2. The difference is apparent for the pilot combustion event occuring at approximately -16°. The pilot event becomes more disctinct with smoothing, albeit more 'squat' shaped. The start of natural gas combustion, occuring shortly after -10° is also much more clear with smoothing. The oscillations present in the interval between the pilot and natural gas ignition are likely due to signal noise and not part of combustion. The smoothing algorithm seems to improve clarity of combustion events without affecting the net integrated heat release. 2.0  -0.5 A -30  1 -20  1  1  1  -10  0  10  i 20  1  1  1  1  30  40  50  60  crank angle [deg]  Figure F-l 2.0 -,  -30  Unsmoothed heat release trace for mode 7 ;  ,  ,  ,  ,  ,  :  :  -20  -10  0  10  20  30  40  50  crank angle [deg]  Figure F-2  Smoothed heat release trace for mode 7  122  60  APPENDIX G. BACKPRESSURE  Table G.1 Instantaneous and Average Intake Manifold Pressures Manifold IVO (kPa-a)  Manifold IVC (kPa-a)  Manifold p  800 rpm  164  171  164  1200 rpm  181  165  179  1600 rpm  184  175  200  X  1  ]  Figure G-l  L  !  J  i  L  GSOI  i  ( 1  t  ! *  i i  -L  1  i  ;  *  ave  A  (kPa-a)  J  _  J _  i  ' J  -1  y.  i. i  •  i_  J  I _  i_  u  5  I  I  0  20  40  I  I  I  -  s | ±  u  i i •  11  °  L  i |  ]  J  -  +10°  I  60 80 100 120 Backpressure (kPa-g)  I  I  140  160  I 180  NOx emissions vs. backpressure at 800 rpm, <j) 0.4 and 2 timings  123  CO vs. Backpressure at 800 rpm, E.Q. 0.6 30.0 25.0 _l LU ZJ  20.0 15.0  CO  O O  10.0 5.0 0.0 0.0  50.0  100.0  150.0  200.0  Backpressure (kPag)  Figure G-2  C O emissions vs. backpressure at 800 rpm, ty 0.6, G S O I -5  ---  150i  800 rpm  1 5 10 15 Time after TDC Intake (ms) Intake manifold pressure versus time for three speeds. 1  1  0  Figure G-3  The period of oscillation is similar for the three speeds as shown in Figure G - 3 , which indicates that the mechanical dynamics of the intake manifold cause the pressure oscillations and not purely gas dynamics. If the oscillation were purely gas dynamic, it should damp out after the intake stroke.  124  APPENDIX H. FUEL SPECIFIC N O EMISSIONS Y  140 120  100  CT>  Oi  Figure H - l  1  -5  1  0  1  .  1  1  5 10 15 20 50% Heat Release (°ATDC)  1  25  1  Nitrogen oxides emissions normalized with fuel for various injection timings and 3 equivalence ratios at 1200 rpm, 4 g/cycle air.  125  APPENDIX I. RELATIVE INJECTION TIMING 1.1 Variable Pilot Injection Timing For these preliminary tests, a broad R I T sweep was conducted with small increments by changing the injection timing of the pilot while maintaining a constant gas injection timing. The test conditions were 1200 rpm, at an equivalence ratio of 0.4 and G S O I of -5° A T D C . A rough look at how changing R I T affects emissions is shown in Figure 1-1 where there is a local minimum between 1.2 and 3.0 ms. O f note is the decrease in N O x as R I T is shifted negative. The reason for this is the pilot has now been injected well after the natural gas-effecting a change in timing of the combustion.  15r  i - i - i -  10  I f  CD E  5  1  Figure 1-1  I  x  NOx  •  CO  :)  tHC  2 RIT (ms)  Comparison of emissions production at <b=0.4,1200 at various RIT, G S O I held constant at -5° A T D C .  126  0.8  *  O  800 1200  0.7  0.6  O  ; 0.5  I  0.4  | 0.3  *  0.2  Q  A  _ .  o  0.1  RIT (ms) 0.8  r  *  O  0.7  800 rpm 1200 rpm o  0.6^ o  § 0.5^  .Is i l l 0.4 [ "CJ  k_ CO  O  o  CO  00  0.2  *  o -©  —  *  •  -  -©-  0.1 10 15 20 Relative Injection (°CA) Figure 1-2  25  Standard Deviation of S O C , 0= 0.4, two speeds  127  30  .  ,  ,  !  -p.  i  •  i  50% Heat Release -+ +5 o  +15  c>  <  2 Q_  4  0  Figure 1-3  (>  i  i  -5  )—  1  0  i  5  1  .  .  >  10  15  20  25  Relative Injection (°CA)  30  Particulate matter emissions at various relative timings at 0=0.4, for 1200 rpm  128  APPENDIX J. INJECTION PRESSURE  35 RPM / P. . +  30  •  25 !20  O  inj  800 / 19 MPa 800 / 23 MPa  V  1600/ 19 MPa  o  1600/23 MPa  1 0  10 — _  1_.  5 10 15 50% Heat Release (°ATDC) Figure J - l  20  25  N O x emissions for two injection pressures, two speeds and various timings.  129  160 140 120  ro £, 100 0) 5  80  in  a> 5. 60 o  40 20 >0 -£ 0 -410 -30 -20 -1 0 0 -e  Figure J-2  ()  10  20  30  40  50  60  70  80  cIrank Angle [°CA ]  In-cylinder pressure trace for GSCT=0, 6.5 g/cycle air, phi=0.4,  130  90  APPENDIX K. METHANE AND NON-METHANE EMISSIONS A methane analyzer was available, however there was an error with the zero and data was not presented for the study. However, the sensitivity of the instrument is reliable. Methane and non-methane emissions for a timing sweep are shown in Figure K - l and Figure K - 2 respectively. Methane emissions increased for the low equivalence ratio. Non-methane hydrocarbons were negative, but showed no change with injection timing for either equivalence ratio. This indicates that the increase in total hydrocarbons found in at late timings is due to unburned methane fuel.  — 14  <>t  IP..  °  0.3/23 MPa  •  0.5/23 MPa  12 10  X  o  5  10  15  50% Heat Release (°ATDC)  Figure K - l  20  25  Methane emissions for various timings, two injection pressures, two equivalence ratios at 1200 rpm.  131  0 <b / P.. mj T  5 10 15 50% Heat Release (°ATDC)  +  0.3/23 MPa  •  0.5/23 MPa  20  25  Non-methane emissions for various timings, two injection pressures, two equivalence ratios at 1200 rpm.  132  APPENDIX L. CRANKANGLE OFFSET EFFECTS A s there is a known offset between the shaft encoder and the crankshaft, the effects on I M E P and heat release should be understood and quantified. To quantify these effects, apparent heat release was calculated with various offsets at an intermediate equivlanence ratio, 1200 rpm, and 4 g/cycle of air.  It is shown in Figure L - l that the offset affects the apparent heat transfer  from the cylinder (before combustion, near T D C ) , where an offset of 1.2 °c.a. (encoder signal is too early and thus the offset retards crank angle values) shows no heat transfer. There must be some heat transfer as the in-cylinder temperature is roughly 900 K and the cylinder wall termperatures is approximately 380 K . A s such, the offset used of 1.7 °c.a. has merit, although the exact magnitude of the heat transfer is unknown and may actually be greater than what is calculated with an 1.7  0  offset.  12  1  "  -  !  - 1.2 1.7 2.2  10  1  f  1  t j  8 T  "O i  1  " I  j  r  vv  CO I  -3  +  <D 2 tn re <D "53 0 a> -2  I  / j  V  L  Ti  1  1 1  >-  | i  !  330  Figure L - l  335  340  345 350 355 360 Crank Angle (360 = TDC)  365  370  Apparent heat release near T D C for various offsets at 1200 rpm, P S O I -2°, G S O I +10°, <>| = 0.4, air = 4g/cycle, y = 1.32  How specific heat ratio affects heat release was only examined near T D C to determine i f it strongly affects the apparent heat transfer from the cylinder. The sensitivity of apparent heat  133  release near T D C to different specific heat ratios (7) is shown in Figure L - 2 . The shape of heat release are similar and the value of heat release at T D C is identical for each offset. According to equation 3.5, changes in the specific heat ratio is w i l l obviously affect the value of apparent heat release. A specific heat ratio of 1.30 was used for heat release calculations for combustion events and any error due to an incorrect specific heat ratio w i l l be consistent for all plots in this study.  12  1  1  [ 1  1.2 1.7  10  2.2  T  1  co  -E 4 cu 2 co CD <D  1  1  1  j  ;  r  r  \  i 1  Figure L - 2  335  340  •>  ]' •  -  1  /  if  1:  '<  330  ''  1  i  "53 0 W - \ \ tr "5 „ CD -2  1  1 4 Ik  i  8 1  1  1  A  A  1  ^  „  v  345 350 355 Crank Angle (360 = TDC)  360  365  370  Apparent heat release for various offsets at 1200 rpm, P S O I -2°, G S O I +10°, <(> = 0.4, air = 4g/cycle, y = 1.35  134  The effect of the offset on the apparent heat release throughout combustion is shown in figure L - 3 . The shape of the heat release for the main combustion event relatively unaffected, but the magnitude of the peak heat release decresases with increasing offset.  330 Figure L - 3  340  350  360 370 380 390 Crank Angle (360 = TDC)  400  410  420  Apparent heat release for various offsets at 1200 rpm, P S O I - 2 ° , G S O I +10°, ty - 0.4, air = 4g/cycle, 7 = 1.32  The error associated with changing the offset -0.5 and +0.5 from 1.7° is displayed i n table L . l for operation at ty = 0.4, 1200 rpm, G S O I = +10°. Obviously changes in the I M E P w i l l affect the I S F C . This is why the measurement error is so high as compared to the repeatability unceratinty as noted in Table 3.7. A n y error in the offset is systematic and w i l l not affect any comparisons between operating conditions on this engine.  Table L . l Measurement Offset Sensitivity for 1200 rpm, +10° G S O I . Parameter  IMEP Max  H R P v  Offset -0.5  +0.5  +5%  -5%  +2%  -2%  135  APPENDIX M. IMPINGEMENT SIMULATION M.l  Introduction To investigate the interaction of the natural gas jet with the piston bowl at different  timings, 3 dimensional simulations were conducted using a modified K I V A 3 V numerical code. It has been found that a burning jet of methane has a similar penetration to that of an unburned jet . A s such, the simulations were conducted using methane with no chemistry. The K I V A 3 V 1  code solves 3 dimensional turbulent fluid flows using a k-e reynolds-averaged navier-stokes model and is also capable of simplified chemistry. The mesh is block-structured and the geometry, which matches the S C R E engine and injector geometry, was generated by Guowei L i of Westport Innovations. Modifications to allow dual-fuel and constant pressure injections were also made by Guowei L i . The initial conditions were validated. N o sensitivity studies were conducted.  M.2  Initial Condiditon Validation The progress of pressure was used to validate the initial conditions selected for this study.  The simulation begins at -150° A T D C where the initial pressure was matched to the in-cylinder pressure data of the S C R E engine. The simulated air was set to the humidity of air during the experiment. The cylinder wall and cylinder head temperatures were set to the coolant temperature of 80°C.  Using an initial temperature of intake manifold plus 25 K , a power regression fit of the  simulated pressure-volume data from -150° to 10° A T D C is within 0.001 of experimental data (-1.371 versus -1.3705). This is considered good agreement for validation of initial conditions.  M.3  Settings The amount of fuel injected was set to match the equivalence ratio i n experiments. The  experiments in question are equivalence ratio 0.4, 0.5. The conditions simulated were G S O I of 0° T D C for 0 of 0.4 and G S O I of 0° and + 1 0 ° A T D C for <J) of 0.5. The injection pressure was set  1. Ouellette, P a t r i c , "Direct Injection of Natural Gas for Diesel Engine Fueling," P h . D . Thesis, University of British Columbia, A p r i l 1996.  136  to 19 M P a to match experimental conditions. The injection mode was set to constant pressure injection. Chemistry was not enabled. The standard k-e constants were employed. M.4  Discussion o f Results The simulations results of slices extracted through the injector centerline are shown in  Figure M - l for 0=0.5. The slice was chosen at 15° after G S O I as that is late i n the combustion event, but still during combustion where production of C O is likely to occur. The highest concentration in the legend, which indicates a rich mixture (by definition), was not obtained. This may be due to the fact that K i v a 3 V simulations are known to underpredict methane jet penetration and over-mix radially. However, it is very apparent in Figure M - l (a) that there is 1  much more of the richest mixture than in Figure M - l ( b ) . The C O emissions are much higher for a G S O I of 0° than + 1 0 ° for 0=0.5 (as found i n Figure 5.13). Upon visual inspection, the area within the richest contour in Figure M - l ( b ) is similar to the richest contour found in Figure M - 2 . When comparing the emissions (in Figure 5.13) for these two cases (0=0.5, G S O I = +10°, 0=0.4, G S O I = 0°), the C O emissions are within error. This combination of simulation data and experimental C O production concurs with the known correlation between rich fuel mixture and the amount of C O produced. There are essentially two opposing processes, C O production and oxidation. There is likely more interaction of the burning jet with the piston at early timings, however the earlier that the C O is produced, the more opportunity it has to mix and oxidize. This may be why injection at -10° A T D C exhibits lower values of C O than at G S O I of 0°, even though there is more jet interaction with the piston. Increasing injection pressure w i l l also increase mixing, thereby reducing C O emissions, agreeing well with experimental results.  The simulations indicate that there is considerable interaction of the burning natural gas jet with the piston bowl for rich mixtures and early timings. The richness of the fuel mixture that the jet creates at the piston bowl is a likely source of carbon moxide emissions. The simulation trends coincide with the experimental results.  1. Personal Communication, Colin Blair, UBC, March 2002. And Ouellette, Patric, "Direct Injection of Natural Gas for Diesel Engine Fueling," Ph.D. Thesis, University of British Columbia, A p r i l 1996.  137  Figure M - 2  Contour plots of methane concentration, 15°c.a.after injection at G S O I of 0° for 0=0.4.  138  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0080985/manifest

Comment

Related Items