PILOT-IGNITED N A T U R A L GAS COMBUSTION IN DIESEL ENGINES By Peter Mtui B. Sc., University of Dar es Salaam, Tanzania, 1986 M . Sc., University of Strathclyde, U.K., 1989 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Mechanical Engineering We accept this thesis as conforming . to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October, 1996 ©Peter Mtui, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of fA&eJL.. £ k g x^oAAxV^ The University of British Columbia Vancouver, Canada Date H>tf?6 DE-6 (2/88) ABSTRACT The purpose of the work was to determine the conditions under which pilot liquid diesel flame (as a source of heat) can ignite natural gas directly injected into the cylinder for diesel engines. The work included experimental and numerical simulation of a conventional diesel and natural gas fueling. The engine was operated at medium speed and load conditions of 1250 R P M and 3 bar (brake mean effective pressure). Measurements and computations were performed by varying the fuel injection timing for baseline diesel (100% diesel) and for diesel-gas (30% pilot diesel and 70% natural gas). High speed flame photography based on the endoscope technique was used to obtain combustion flame pictures of the firing engine. Measured cylinder pressure data were analyzed by a multi-zone combustion model for mass burning rate. Three-dimensional numerical simulation based on the KIVA code was used to predict thermal and flow field in the engine chamber. The KIVA code was modified for diesel-gas combustion. A single-step chemical reaction coupled with a mixing-controlled combustion model were implemented in the modified KIVA code to predict ignition and combustion. Successful ignition of natural gas by a pilot diesel flame depends strongly on the relative injection timing of the pilot diesel and natural gas. With the beginning of natural gas injection 3 degrees after the injection of the pilot diesel, a successful ignition is possible over a wide range of injection timings. Successful ignition with late injection timing is associated with low emissions of oxides of nitrogen (NOx). With pilot injection, the natural gas is ignited by contact with the pilot diesel combustion products rather than the combustion enhanced compression of the unburned fuel and air. Ignition of the natural gas depends strongly on the combustion duration of the pilot diesel. ii Under the conditions studied, the diesel-gas combustion has about the same ignition delay and combustion duration as baseline diesel. Combustion with the diesel-gas case is apparently smoother (i.e. lower cycle-to-cycle variations) than with baseline diesel. iii TABLE OF CONTENTS A B S T R A C T i i TABLE OF CONTENTS iv LIST OF SYMBOLS x LIST OF TABLES xiv LIST OF FIGURES xv A C K N O W L E D G M E N T xxiii CHAPTER ONE: INTRODUCTION 1.1 Natural Gas as Fuel 1 1.2 Direct Injection Diesel Engines 1 1.3 Pilot Fuel Injection Engines 3 1.4 Diesel-Gas Engines 7 1.5 Internal Combustion Engine Modeling 8 1.6 Fuel Combustion and NOx Formation 9 1.7 Summary of Prior Knowledge 10 1.7.1 Some Knowns 11 1.7.2 Some Important Questions .', 11 1.8 Objectives 12 1.9 Methodology 13 CHAPTER TWO: COMBUSTION IN DIRECT INJECTION ENGINES 2.1 Introduction 15 2.2 Fuel Injection in Direct Injection Engines 15 2.2.1 Injection Velocity 16 2.2.2 Injection Pressure 19 2.2.3 Injection Hole Diameter 20 2.2.4 Injection Geometric Angle 22 2.2.5 Injection Rate and Duration 24 2.2.6 Summary 25 iv 2.3 Ignition in Direct Injection Diesel Engines 26 2.3.1 Effect of Temperature 26 2.3.2 Ignition of Liquid Diesel 27 2.3.3 Ignition of Gaseous Fuel (Natural Gas) 28 2.3.4 Diesel-Natural Gas Ignition 29 2.3.5 Summary 30 2.4 Combustion in Direct Injection Diesel Engines 30 2.4.1 Pilot Diesel Natural Gas Combustion 31 2.4.2 Summary 33 2.5 NOx Emissions in Diesel Engines 34 2.5.1 NOx Reduction in Diesel Engines 35 2.5.2 Summary. 36 CHAPTER THREE: ENGINE TESTING APPARATUS 3.1 Introduction 38 3.2 The Test Engine 38 3.3 Test Conditions 39 3.3.1 Diesel Fuel Flow Rate 42 3.3.2 Natural Gas Flow Rate 42 3.3.3 Intake Air Flow Rate 42 3.3.4 Manifold Temperature and Pressure 43 3.4 The Dual Fuel Injection 43 3.5 Engine Instrumentation and Data Acquisition 48 3.5.1 Torque Measurement 49 3.5.2 Engine Speed and Crank Position 49 3.5.3 Cylinder Pressure 50 3.5.4 Exhaust Emissions 52 3.5.5 Combustion Flame Visualization 54 3.6 Evaluation of Performance Parameters 58 3.6.1 Brake Mean Effective Pressure (BMEP) 58 3.6.2 Thermal Efficiency (n) • • 60 v 3.6.3 Brake Specific Emissions (BSE) 61 3.8 Summary 62 CHAPTER FOUR: DISCUSSION OF EXPERIMENTAL RESULTS 4.1 Introduction 64 4.2 The Combustion Flame Photographs 67 4.3 Cylinder Pressure 69 4.3.1 Cycle-to-Cycle Variations 70 4.3.2 Premixed and Diffusive Combustion 71 4.4 Ignition Delay and Combustion Duration 72 4.4.1 The Ignition Delay 72 4.4.2 The Combustion Duration : 73 4.5 Burned Gas Temperature (Flame Temperature) 73 4.6 Unburned Gas Temperature 74 4.7 Formation of NOx 74 4.8 Wall Heat 74 4.9 Summary 75 4.9.1 Ignition Delay 76 4.9.2 Combustion Duration 76 4.9.3 Chamber Temperature and NOx Emissions 77 CHAPTER FIVE: TURBULENT COMBUSTION 5.1 Introduction 90 5.2 ignition 91 5.3 Ignition of Methane 93 5.3.1 Ignition Delay 95 5.4 Laminar Flame Speed 97 5.5 Flame Propagation in Turbulent Combustion 99 5.6 Chemical Reaction Time Scale 100 5.7 Mixing-Controlled Combustion 100 5.7.1 The Magnussen Combustion Model 101 vi 5.8 Fuel-Air Mixing Dual-Fuel 104 5.9 Chemical Interaction in Pilot-Ignited Natural Gas Engines 108 CHAPTER SIX: TURBULENT COMBUSTION MODELING 6.1 Introduction 109 6.2 The KIVA-II Code 110 6.2.1 The Governing Equations I l l 6.2.2 The Chemical Reactions 112 6.3 Initial and Boundary Conditions 113 6.3.1 Initial Conditions 113 6.3.2 Boundary Conditions 113 6.4 Modifications Done on the KIVA-II Code... 115 6.4.1 Diesel Fuel Library Data 116 6.4.2 Natural Gas Injection 116 6.4.3. Chemical Reactions 118 6.5 Numerical Computation 119 6.5.1 Grid Sensitivity , 120 CHAPTER SEVEN: DISCUSSION OF COMPUTED RESULTS 7.1 Introduction 123 7.2 The Cylinder Pressure 124 7.3 The Mass Burned Fraction 125 7.4 Ignition Delay 126 7.5 Combustion Duration 127 7.6 Burned Gas Temperature (Flame Temperature) 127 7.7 Unburned Gas Temperature 128 7.8 The Emissions of NOx 129 7.9 Summary 130 7.9.1 The Cylinder Pressure 130 7.9.2 Ignition Delay and Combustion Duration 130 7.9.3 Combustion Chamber Temperature and NOx Emissions 131 vii CHAPTER EIGHT: DISCUSSIONS OF SIMULATION RESULTS 8.1 Introduction 145 8.2 Fuel Injection and Combustion 145 8.2.1 Jet Induced Turbulence by Dual Fuel Injection 146 8.2.2 The Mixing Time Scale 147 8.2.3 The Equivalence Ratio 148 8.2.4 The Diesel Vapour and Natural Gas in the Chamber 148 8.2.5 Ignition Delay and Combustion Duration 150 8.2.6 The Cylinder Temperature 151 8.2.7 The NOx Formation 152 8.4 Summary 152 8.4.1 Jet Induced Mixing and Mixing Time 152 8.4.2 The Equivalence Ratio 153 8.4.3 The Diesel Vapour and Natural Gas in the Chamber 153 8.4.4 Ignition Delay and Combustion Duration 153 8.4.5 Cylinder Temperature and NOx Emissions 154 CHAPTER NINE: PARAMETRIC SIMULATION RESULTS 9.1 Introduction 180 9.2 Injection Angle 181 9.2.1 Jet Induced Turbulence by Dual Fuel Injection 181 9.2.2 The Mixing Time Scale 182 9.2.3 The Equivalence Ratio 182 9.2.4 The Diesel Vapour and Natural Gas in the Chamber 183 9.2.5 Ignition Delay and Combustion Duration 183 9.2.6 The Cylinder Temperature 185 9.2.7 The NOx Formation 185 9.3 The Injection Timing 186 9.3.1 Simultaneous Injection of Pilot Diesel and Natural Gas 186 9.3.2 Late Natural Gas Injection 187 9.4 Ignition Delay and Combustion Duration 187 viii 9.5 Summary 187 9.5.1 Injection Angle 188 9.5.2 The Injection Timing 188 CHAPTER TEN: CONCLUSIONS AND RECOMMENDATIONS 10.1 Introduction 205 10.2 Ignition 206 10.3 Combustion , 207 10.4 Multi-Zine Combustion Model (XPNOX code) 209 10.5 3-D Numerical Simulation (KIVA-II code) 209 10.6 Recommendations for Future Work 210 10.6.1 Experiments 210 10.6.2 Numerical Simulations 210 REFERENCES ...211 Appendix A: Engines Technical Specifications 219 Appendix B: Pressure Transducer Testing 220 Appendix C: Calculation of Mass-Burned Fraction (XPNOX Code) 223 Appendix D: Diesel Fuel Data and Reaction Scheme Constants 226 Appendix E: Constants for NOx Chemical Reaction Schemes 228 Appendix F: Modeling of the KIVA-II Code 222 Appendix G: Error Analysis 233 ix LIST OF SYMBOLS a empirical oxygen concentration exponent [- 1 b empirical fuel concentration exponent [ "] A pre-exponential constant [- 1 Am/ Bm constants in Magnussen model [- 1 C F fuel concentration [mol/cm 3] Co2 oxygen concentration [mol/cm 3] CpR products concentration [mol/cm 3] D diffusion constant [- 1 E a activation energy [J/kg.K] enthalpy of specie m [kJ/mol] I Internal energy [J] k turbulence kinetic energy [m2/s 2] k f reaction rate constant [ " ] 1 turbulence length scale [cm] P pressure [bar] Pr Prandtl number [ "] Q heat release per unit volume [J/m3] R gas constant [kJ/kg.K] Re Reynolds number [ " ] r Pressure exponent [ " ] S F Stoichiometric ratio [ - 1 S C Schmidt number [ " ] T b bulk temperature [ K ] T c core temperature [ K ] T u unburned gas temperature [ K ] S F stoichiometric ratio [ "] SL laminar flame speed [m/s] SLO laminar flame speed constant [m/s] ST turbulent flame speed [m/s] turbulence intensity molecular weight [m/s] [kg/kmol] xi G R E E K S Y M B O L S a fuel stoichiometric coefficient [-] P oxygen stoichiometric coefficient [-] 5 products stoichiometric coefficient [-] e injection angle (relative to the cylinder head shown in Fig. 6.1) [-] s dissipation rate of kinetic energy [J/s] * equivalence ratio (also the injection angle shown in Fig. 6.1) ["] temperature exponent [-] V kinematic viscosity [mVs] p density [ k g / m 3 ] CO reaction rate [kmol/m3.s] T time [s] 0. relative injection angle between pilot diesel and natural gas [ ° ] SUPERSCRIPTS A N D SUBSCRIPTS b burned ch chamber ign ignition m specie mix of mixing o stagnation (pressure) PR products r reaction r xii ABBREVIATIONS BOI beginning of injection B T D C before top dead center C A crank angle CI compression ignition C N G compressed natural gas D D C Detroit Diesel Corporation D D E C Detroit Diesel Electronic Control DSL diesel dt time increment EOI end of injection G A S natural gas KTVA -II three-dimensional numerical simulation code L H V lower heating value NOx oxides of nitrogen PR pressure ratio PSI pressure (pounds per square inch) PW injection duration in crank angle R P M revolutions per minute T D C top dead center T O T total (diesel plus natural gas) X P N O X multi-zone model for mass burn analysis xiii LIST OF TABLES Table 3.1 Diesel and Natural Gas Injection Parameters 40 Table 3.2 Test Conditions 41 Table 4.1 Test Conditions 66 Table 7.1 Experimental and Simulation Test Conditios Conditions 124 Table 8.1 KiVA-II Simulation Conditions 145 Table 9.1 Parametric Test Conditions 180 Table 9:2 Ignition Delay and Combustion Duration for Pilot Diesel 204 Table A . l Technical Data for D D C 6V92 and D D C 1-71 Engines 219 Table D . l Diesel Fuel Library Data (Varnavas [1990]) 226 Table D.2 Kinetic Rate Constants for Diesel and Natural Gas 227 Table D.3 Mixing Limited Rate Constants for Diesel and Natural Gas.. 227 Table 1.3 Reaction Rate Constants for NOx Formation (Heywood [1986]) 228 Table F . l Constants for the k-e Turbulence Model (Amsden (1989) 230 Table G . l Errors in Exhaust Gas Measurements 232 xiv LIST OF FIGURES Fig. 1.1 Heat Release Rate in Diesel Engines 3 Fig. 1.2 Ignition Delay for Natural Gas Combustion (Naber et al. [1994]) 6 Fig. 1.3 Pilot Diesel Natural Gas Injection in CI Engines 6 Fig. 2.1 Effects of Liquid Diesel Injection Velocity (Beck [1988]) 17 Fig. 2.2 Effects of Pressure Ratio on Jet Penetration (Chepakovich [1993]) 20 Fig. 2.3 Fuel Mass Fraction in the Chamber at (J) < 2 (Jennings and Jeske [1994 II]) 22 Fig. 2.4 Contours of 170 180 190 200 210 Crank angle, deg Fig. 1.1: Typical heat release rate in diesel engines (Heywood [1988]) 1.3 Pilot Fuel Injection Engines Ignition delay determines the ignition and combustion behavior in the later stage of the combustion. For practical application of any fuel in diesel engines, ignition delay should be less than 10 or at most 15 deg. crank angle. For an engine rotating at 1250 rpm this means an ignition delay of 1.5 to 2 ms and combustion duration of about 5 ms. 3 Fraser et al. [1991] and Naber et al. [1994] demonstrated that the auto-ignition of natural gas in compression ignition engines would require a cylinder temperature of 1200 - 1300 K to reduce the ignition delay of natural gas to less than 2 ms. They used a constant volume combustion vessel to approximate the engine combustion chamber close to TDC (during which the rage of change of volume is minimal). Figure 1.2 shows that the auto-ignition delay of natural gas is strongly dependent on temperature and decreases continuously with increasing temperature. The filled symbols (Fig. 1.2) are experimental data by Naber et al. [1994] obtained in a pressurized constant-volume chamber preheated to the ignition temperature of natural gas. A naturally aspirated diesel engine with bottom dead center (BDC) temperature of 325 K would require a compression ratio of 46:1 to reach a top dead center (TDC) temperature of about 1250 K, assuming a polytropic compression (assume polytropic index of 1.35). This compression ratio is unrealistically high. Conventional diesel engines which have compression ratios of about 17:1 (corresponding to about 850 K) to be adapted for natural gas, a source of heat is required to raise the temperature to the level required so that the ignition delay of natural gas is less than 2 ms. A pilot fuel as a source of heat for the ignition of natural gas is the subject of this thesis. In a pilot-diesel-injection engine, a charge of air is compressed in the cylinder to the pressure level of conventional diesel engines. A small amount of pilot diesel injected near the TDC starts to burn after an ignition delay (see Fig. 1.3). After a few degrees crank angle, the gaseous fuel (natural gas) is injected at high pressure (about 140 bar) and is ignited by the pilot combustion. Figure 1.3 illustrates the working principle of the pilot-diesel natural-gas combustion. The injection angle (for pilot diesel and natural gas) relative 4 to the cylinder head is denoted by 0. The pilot diesel spray and natural gas jet are separated in the circumferential direction by angle (J). With such configuration (Fig. 1.3), the gas/air mixture burns as the result of the action of multiple flame front propagation initiated by the burning pilot spray. After the ignition of the natural gas, the mode of combustion can be dominated either by premixed or diffusive combustion as shown in Fig 1.2. Which combustion mode dominates depends significantly on the injection timing of the pilot fuel relative to that of the natural gas, i.e., the natural gas injected significantly before the injection of pilot fuel could lead to a premixed-dominated combustion and vice versa. A question may be raised whether natural gas is ignited by the burning pilot or by the high chamber temperature raised by the burning pilot fuel or by the pre-existing residuals. This will be considered by determining the extent to which compression (partly due to pilot combustion) raises the unburned gas temperature. Numerical modeling discussed in Chapter 6 has been a valuable tool in the detailed exploration of the m-cylinder process by mvestigaring the effect of injection timing of natural gas relative to the pilot diesel on the ignition and combustion. 5 Tc [K] „ 1600 1400 1200 1000 30r-T 1 1 1 —1 — 1 l 1000/Tc[1/K] Fig. 1.2: Ignition delay for natural gas combustion (Naber et al [1994]) Natural gas Pilot diesel Fig. 1.3: Pilot-diesel-natural gas injection in CI engines (Courtesy, P. Ouellette) 6 1.4 Diesel-Gas Engines In this context, diesel-gas engines are defined as engines which operate with two fuels, the pilot and the main fuel. Pilot fuel (diesel) provides heat for the ignition of the main fuel (i.e. natural gas). The diesel-gas engines can be grouped in three categories: (i) those which use fumigation method in which natural gas is carburetted with air during the intake process and ignited by the burning pilot fuel which is injected near TDC; (ii) those having timed port injection in which the natural gas is timely injected in the intake manifold immediately before the intake ports close; the timed injection provides stratified charge which is easier to ignite than category (i), and (iii) those with high pressure natural gas injection in which natural gas injection takes place near TDC almost simultaneously with the pilot fuel. As in categories (i) and (ii), the burning of pilot fuel initiates the ignition of the natural gas. The fumigated and timed port injection engines demonstrate that a dual fuel engine operating under categories (i) and (ii) respectively suffers a drawback of possible engine knocking due to premature self-ignition of the natural gas during the compression stroke especially at high loads. The third category of diesel-gas engine, the main theme of this thesis, does not suffer from knock because, as in a conventional diesel, only air is compressed during the compression process. As noted in Chapter 2, the combustion of high pressure natural gas with pilot diesel in conventional direct injection (Dl) diesel engines has been studied in the past decade. Some pertinent research work is reported to have been systematically performed in which natural gas was injected into the combustion chamber at a pressure of about 200 bar and 7 ignited by pilot diesel of. the order of 5% (on energy basis). The conversion of diesel engines to run on natural gas was found to produce the same power as the conventional diesel engines while reducing NOx and particulate matter (e.g., see Miyake [1983]). 1.5 Internal Combustion Engine Modeling Recent advances in multi-dimensional computer modeling have been valuable for mterpreting and supplementing experimental results. In a Dl engine the m-cyhnder processes include transient three-dimensional turbulent flow of evaporating or non-evaporating fuel spray mteracting with multi-component gases undergoing mixing, ignition, chemical reaction and heat transfer. In studying the m-cylinder processes experimentally, there are limitations in efforts to develop a more complete understanding of a firing engine because the processes are complex and occur in environments which pose difficulty for measuring instruments. Certain features such as local temperature and NOx concentration transients are particularly difficult to measure. In recent years, however, computer technology and numerical simulation techniques have advanced to the point that multi-dimensional numerical simulations of the reactive flow in an internal combustion engine cylinder can be performed. Such methods aid in the interpretation of data and in studying the possible consequences of design changes. The recently increased use of multi-dimensional numerical simulation has provided more evidence that the turbulent mixing process plays a significant role in the combustion process. During the intake process the flow produces shear layers at the intake valve/port 8 Such layers are unstable and merge to form larger-scale vortices. Chandrsuda et al. [1978] reported that the intake jet flow induces a recirculation region which is sensitive to minor variations in the flow which is believed to contribute to cycle-to-cycle variations of the engine combustion. In DI engines, the flow is complicated by jet-induced turbulence as a result of fuel injected into the combustion chamber near TDC. Depending on the injection location and direction, general circulation may be enhanced or reduced, and small-scale turbulence will be generated by the injection process. The turbulence generated by the fuel jet has been reported to be responsible for the formation of the major portion of fuel/air mixture. Ignition and combustion have been shown by Tabaczynski et al. [1977] to depend significantly on the structure of the small scale flow which influences flame propagation. The sudden rise of temperature due to combustion increases the fluid viscosity, thereby increasing the rate of decay of turbulence so that molecular mixing is enhanced. 1.6 Fuel Combustion and NOx Formation Combustion can be simulated by two categories: (i) the simplified chemical kinetic with few manageable chemical reaction steps. The chemical kinetics of the Arrhenius type are strongly temperature dependent The modeling of fuel ignition and NOx formation in the high temperature zone are based on the chemical kinetics, and (ii) the mixmg-limited chemical reaction which is controlled by the rate of mixing of the reactants and the oxidant 9 Categories (i) and (ii) lack universality so that empirical constants are introduced and adjusted to match experimental results. In IC engines the NOx formation is of particular interest The mode of NOx formation is well understood and developed; it is based on extended Zeldovich mechanism. NOx is generated mainly in the localized high temperature zones in the engine combustion chamber. Any possibility of operating the engine with short combustion duration and/or lower cylinder temperatures tends to reduce the amount of NOx emissions. The reaction rates governing the production of NOx in diesel engines are much slower than those for combustion and have time scales of the same order of magnitude as the turbulence fluctuations (Brown and Heywood [1986]). The NOx formation has been modeled by chemical kinetics using extended Zeldovich mechanism (discussed in Chapter 2). As in the case of fuel combustion, the NOx formation model involves empirical constants to fit with experimental results. 1.7 Summary of Prior Knowledge Prior knowledge of some knowns and some important questions about the pilot-ignited natural gas combustion engines is summarized below. 10 1.7.1 Some Knowns (i) Previous experimental work (discussed in Chapter 2) has shown that natural gas can be used in diesel engines, say, when a source of heat provides ignition for the natural gas. The pilot-ignited natural gas combustion engine can provide power and efficiency comparable to that of the diesel engine with the advantage of reduced NOx emissions; (ii) The auto-ignition time delay of natural gas is strongly temperature dependent and decreases continuously with increasing temperature. The auto-ignition delay (< 2 ms) for natural gas which consists of over 90% methane, corresponds to temperatures of 1200-1300 K, so that ignition assistance is essential for conventional diesel engine fueled by natural gas; (iii) The mechanisms of the formation of NOx are well understood and developed. The modeling of NOx formation is based on the extended Zeldovich mechanism which involves empirical constants to agree with the experiment The reaction rate for these mechanisms are strongly temperature dependent 1.7.2 Some Important Questions (i) Whether the ignition of the natural gas by the pilot diesel is due to direct contact with the burned products (of pilot diesel) or by the temperature rise due to the increased compression ratio resulting from burning of the pilot diesel; 11 (ii) How the high pressure natural gas injection influences the flame structure and penetration of the pilot-ignited natural gas; (iii) How the natural gas ignition delay and combustion duration is influenced by pilot liquid diesel injection timing, spacing and amount; (iv) How the natural gas injection angle and pressure influence the ignition delay, combustion duration and NOx emissions; (v) Whether the temperature field on which the NOx formation is strongly dependent can be controlled in such a way as to minimize NOx emissions; (vi) How the ignition delay and combustion duration of dual fuel combustion is influenced by engine load, swirl, speed, natural gas injection pressure and angle. 1.8 Objectives The objectives of this thesis are to shed light on some of the unanswered questions addressed above pertaining to the combustion of diesel-gas fueled engines as follows. (1) Find out whether the injection of high pressure natural gas jet results in a radical change of the flame penetration and structure compared to the baseline diesel. (2) Establish whether the natural gas is ignited by thermal contact with the burned product of the pilot diesel or by self-ignition due to the compression temperature created by the expanding combustion products of the pilot-diesel. 12 (3) Find out the level of cycle-to-cycle variations in the diesel-gas combustion and whether it is dominated by pre-mixed or diffusive combustion. (4) Determine the influence of injection parameters (i.e. injection timing and geometric angle) on: (i) ignition and combustion, and (ii) NOx emissions in the combustion of the pilot-diesel-natural gas. (5) To evaluate the validity of the numerical modeling (KIVA-II code) based on the experimental results so that possible design changes of the engine operating conditions can be assessed by KIVA-II predictions. 1.9 Methodology Experimental and numerical methods have been employed to accomplish the objectives stated. The combustion of pilot-ignited natural gas engine has been studied experimentally by employing a prototype dual-fuel injector with the Detroit Diesel engine DDC 6V-92TA. The test engine was instrumented with electronic control and a computerized data acquisition system which provides on-line engine performance and exhaust gas emissions. The prototype injector was designed and manufactured in the Department of Mechanical Engineering at University of British Columbia (UBC). The ignition and combustion processes in the combustion chamber were studied visually by combustion flame imaging. The DDC 6V-92TA engine, used for flame visualization, has the advantage of mounting the optical instrument Modifications were made to mount 13 the endoscope and the high-speed camera to capture the ignition and combustion events of the firing engine. The endoscope, an optical probe inserted into a small hole in the cylinder head of the engine, provides a view about 78° solid angle to the combustion chamber. Measured cylinder pressures were analyzed by the XPNOX code (Hill and Douville [1996]) which is a multi-zone model for the determination of mass burning rate and NOx emissions. Detailed modeling of the injection, combustion and NOx emissions were simulated by the multi-dimensional KTVA-II code. The KTVA-II computer code solves the three-dimensional unsteady equations of motion for a chemically reactive mixture of ideal gases. For this thesis, the code was modified for pilot-diesel-natural-gas injection and combustion. A mixing-controlled combustion model was incorporated to the original combustion model in the KIVA-II code. The numerical approach allows interpretation and supplement of experimental results. The numerical model will be particularly a valuable tool for studying (in this work) the m-cylinder processes, such as ignition, combustion and formation of NOx for pilot-ignited natural gas engines. 14 CHAPTER TWO - COMBUSTION IN DIRECT INJECTION ENGINES 2.1 Introduction Fuel combustion in direct-injection (DI) engines is essentially controlled by the rate of turbulent mixing of fuel and air to form a combustible mixture. Previous research has shown that combustion of the DI diesel engine is significantly influenced by the m-cylinder fluid flow and the fuel injection characteristics. This chapter reviews selected studies on: (i) fuel injection and mixing; (ii) ignition and combustion, and (iii) formation of NOx. Both conventional diesel engines and pilot-ignited natural gas combustion engines are considered. 2.2 Fuel Injection in Direct Injection Engines The fuel-air mixing process of liquid fuel spray or gaseous fuel jet is influenced by the turbulent mixing as the result of the interaction of fuel jet with the cylinder air motion. The fuel injection characteristics and the formation of combustible mixture are important in the combustion and pollutant formation of DI diesel engines. Fuel injection parameters of interest which influence air-fuel mixing are: (i) velocity; (ii) pressure; (iii) geometrical angle and hole diameter, and (iv) rate and duration of injection. 15 2.2.1 Injection Velocity Kuo and Bracco [1983] performed numerical simulations to study the effects of liquid fuel injection velocity and chamber temperature on the diesel fuel evaporation and penetration. They reported that, a 13% increase in velocity results in a 30% reduction in drop size diameter. Lower injection velocity results in larger drops (due to poor atomization) which lead to longer evaporation time. For liquid fuels, however, at exceedingly high injection velocity the breakup of fuel droplets increases so that penetration is retarded. The retardation of penetration of the diesel droplets is primarily due to drag resistance and entrainment by the surrounding air. Therefore, optimal injection velocity leading to desired penetration and fuel-air mixing is essential because, over-mixing leads to lean mixtures which is difficult to ignite or does not burn to completion. Computed results (Gosman and Johns [1980]) and shadowgraph pictures ([Shundow [1991]) suggest that jet fuel-air mixing is mainly dependent on its own turbulence kinetic energy created by high jet velocity. Gosman and Johns noted that the fuel spray velocities induce turbulence levels comparable to those produced by other mechanisms such as swirl and squish. Squish motion is the radial flow in the cylinder towards the end of compression stroke. Swirl is defined as the rotational flow witMn the cylinder about its axis. Further, they reported that, the resulting zone of high turbulence near the injector contributes to rapid mixing of air and fuel. This is consistent with the results discussed in this thesis, in which the high pressure natural gas jet (at 460 m/s) results in turbulence kinetic energy significantly higher than the diesel spray injected at 140 m/s. 16 Figure 2.1 shows that, when diesel injection velocity exceeds 200 m/s, penetration of the jet and the core (inner part of jet) decreases with increasing velocity. The parameters Uj, E>h and s represent injection velocity, hole diameter and jet penetration, respectively. Similarly, higher chamber temperature which increases the evaporation rate, results in lower jet penetration. Such results shown in Fig. 2.1 suggests that, if high penetration is desirable during diesel injection, the injection velocity should not exceed 200 m/s. Fig. 2.1: Effect of liquid diesel injection velocity (Beck [1988]) Endoscopic pictures by Werlbeger and Cartellieri [1987] suggest that for an engine operating above 50% load, large droplets of liquid diesel fuel strike the piston surface and form a liquid layer rundering mixing. The observed large droplets could be associated with high fuel injection rate at high loads. Abraham et al. [1994] performed calculations to investigate the mixing effectiveness between liquid diesel and natural gas. From the results, Abraham concluded that, under the same injection momentum, liquid diesel spray mixes more quickly than NG jet Further 17 he noted that, within the first 10 deg. CA of injection, the flammable air-fuel fraction of the spray, say diesel, is more than double that of the NG jet It was suggested that higher diesel viscosity enhances air entrainment and faster mixing than NG. This is contrary to schlieren photographs (Miyake et al. [1983]) of injection velocities of natural gas and diesel (409 m/s and 380 m/s, respectively) in a constant volume chamber. With similar injection momentum, Miyake et al. reported similar penetration for diesel and natural gas. The results reported in this work (discussed in Chapters 8 and 9) indicate that, though natural gas penetration and turbulent kinetic energy k is higher than diesel (due to higher injection velocity), the rates of fuel-air mixing (i.e. mixing times) is similar for diesel and natural gas. The explanation is that, although the natural gas k is higher than for diesel, the rate of dissipation of k, (defined here as e) is also high, such that the mixing time defined k/e is about the same for diesel and natural gas. The fact that diesel and natural gas penetration and mixing processes are about the same, is of great importance in the use of gaseous fuels in diesel engines, because some of the knowledge of liquid fuel sprays can fairly be applied to gaseous fuels. When natural gas (about 90% methane) is used as fuel with lower heating value of about 50 MJ/kg, and is injected at high velocity (close to critical velocity), it is possible to form a jet having the same energy value as the conventional liquid spray by simply adjusting the nozzle diameter. The numerical simulation of Jennings and Jeske [1994, (I)] investigated the behavior of natural gas jet injected in DI (quiescent chamber) diesel engine. The turbulence generated 18 by the jet was reported by Jerrnings and Jeske to be responsible for the jet mixing within the first 2 ms after the jet injection. Further, they reported that, fuel-air mixing increases exponentially within the first 1.0 ms in which the jet has penetrated only 56% of the chamber radius, i.e., most of the NG fuel remains confined at the central part of the chamber, resulting in higher penetration for liquid diesel injection. 2.2.2 Injection Pressure Kuo and Yu [1984] and Shundow [1991] observed that higher injection pressure enhanced mixing by improving fuel atomization and penetration. Penetration of the NG jet was found experimentally by Chepakovich [1993] to depend on the pressure ratio rather than on the absolute pressure. Pressure ratio (PR) is the ratio of injection pressure to chamber pressure (Pch). Figure 2.2 shows the effect of PR on NG jet penetration in which BOI, in this context, is defined as the beginning of NG injection. At PR of 2.0, for various chamber pressure values, the penetration is within experimental error limits. Chepakovich's results indicate that increasing the pressure ratio from 1.5 to 3.0 almost doubles the jet penetration at 6 ms after the start of injection. At a given time t, the natural gas jet penetration Z t been shown by Hill and Ouellette [1996] to be related as Zt = 3 * [M / p C J 1 / 4 f1/2 where M and pa, are the momentum rate and chamber density, respectively. For liquid sprays and gaseous jets, the penetration Zt is essentially proportional to W1. High pressure injection of natural gas for pilot diesel combustion engines has been studied by Miyake et al. [1983], Einang et al. [1983], Tao [1993] and Douville [1994]. Miyake 19 reported an increase of thermal efficiency by 3.5% (Max. thermal efficiency was 43% at 100% engine load) by increasing the natural gas injection pressure from 205 bar to 250 bar. Results discussed in this work (under conditions which were investigated) indicate that, the combustion of natural gas takes place near the squish region resulting in high wall heat losses. Therefore, care ought to be exercised with high injection velocity of natural gas since this can lead to over-penetration. A A • • A A PR=3 Pch=3550 ft P R = 2 Pch=450 Q jjv P R = 2 Pch=1820 <3> PR=2 Pch=3550 • • PR=1.5 Pch=355() (Pch is in kPa) 1 3 4 5 6 Time after BOI , ms 7 8 9 Fig. 2.2: Effect of pressure ratio on jet penetration (Chepakovich [1993]) 2.2.3 Injection Hole Diameter Computed results reported by Kuo and Yu [1984] indicate that the injector hole size (i.e. number of holes) influences fuel-air mixing. Because of spray-to-spray interaction, poor mixing of fuel and air is encountered for an injector with more than 12 holes; otherwise having hole number below 4 tends to promote faster penetration and thereby hinder 20 mixing, i.e., the fewer the number of holes the larger the amount of fuel required per hole so that higher jet momentum and penetration result Jennings and Jeske [1994 (I)] performed muti-dimensional numerical computation to simulate auto-ignition of NG in CI engines. The simulated chamber temperature and pressure at TDC were 970 K and 98 bar when the natural gas was injected at a pressure of 200 bar. Their computed results shown in Fig. 2.3 indicate the influence of the injector hole diameter on the air-fuel mixing for equivalence ratio less than 2.0. Equivalence ratio cj> is the ratio of actual to theoretical fuel-air ratio. For 16 holes (Fig. 2.3) the plumes merge together and reduce plume surface, or in the worst case result in the jet attaching to the cylinder head wall (Coanda effect), which adversely affects mixing. They suggested that an optimal number of holes is the one that keeps the jets separate so that fuel-air mixing can be enhanced. The optimal number of holes can be approximated from the spreading angle of natural gas injection (e.g. schlieren pictures of Chepakovich [1993]). The schlieren pictures indicate natural gas spreading angle is less than 30°. Such a typical spreading angle results in about 12 (i.e. 360°/30) nozzles, which is consistent with Jennings and Jeske's findings. 21 2.Z4 Injection Geometric Angle For a poppet injector (approximation of injector with large number of holes), the computed results by Jennings and Jeske [1994 JTJ reveal that, NG jet penetration is influenced by the jet injection angle relative to the cylinder head surface. At a small injection angle, the fuel jet attaches itself on to the cylinder head (Coanda effect) retarding jet penetration and mixing. The Coanda effect reported by Jennings and Jeske is consistent with jet penetration and wall attachment photographs observed by Ouellette and Hill [1992] for a poppet injector. The influence of the Coanda effect on the NG jet penetration is shown is shown in Fig. 2.4. The fuel injector is located at point "P" in Fig. 2.4 with the jet positions of NG at 2 ms after BOI. 22 Fig. 2.4: Contour of NO + N (2.2a) N + O2 NO + O (2.2b) N + OH <-+ NO + H (2.2c) Since NOx is mainly formed within the localized high temperature zones in the combustion chamber, control of flame temperature distribution appears to be an effective approach for NOx reduction. 2.5.1 NOx Reduction in Engines For a given fuel, various techniques have been used to reduce the amount of NOx formation in IC engines. Injection timing retard has been performed by Wade [1987], Kato [1989] and Aoyama [1990] to limit the level of NOx emissions. A lower peak cylinder temperature is attained for retarded injection and thus results in low NOx emissions. The experimental results of Aoyama [1990] include NOx emission of 275, 230 and 185 ppm (parts per million) for the injection timings of 10, 7 and 4 deg. BTDC, respectively. Pilot injection (multiple injections of diesel fuel) has been demonstrated by Schulte [1988] and Aoyama [1990] to be an effective technique of reducing NOx emissions. Pilot injection, by promoting ignition and combustion of the main injection facilitates injection retard (resulting in low local spatial temperature) resulting to low NOx emissions. The experiment by Aoyama [1990] on pilot injection to match the cylinder pressure for normal 35 injection showed NOx emission for pilot and normal injection to be 385 and 300 ppm, respectively. High diesel injection pressure has been reported by Shundow [1991] to reduce NOx emission. For the range of injection timing of 15 to 5 deg. BTDQ Shundow reported reduction of NOx by 25% when the injection pressure was increased from 100 MPa to 200 MPa. Previous experimental work on a pilot diesel NG combustion has shown a substantial reduction of NOx due to lower peak cylinder temperature. Tao, Hodgins and Hill [1995] and Douville [1994] reported a reduction of NOx by 25% (compared to baseline diesel) while using 30% diesel and 70% NG (on energy basis). Einang et al. [1983] reported a reduction of 24% NOx level with 73% NG at full engine load. 2.5.2 Summary & The mechanisms of NOx formation are well understood and developed. NOx formation is strongly temperature dependent and increases exponentially with temperature. Previous work has not addressed the influence, if any, of turbulence on the formation of NOx; & Various methods to limit the peak cylinder temperature and to control NOx formation include retard and split injection. The use of alternative fuels (natural gas) is a possibility of reducing NOx emissions; 36 The injection timing has been shown to influence NOx emissions. The influence of injection timing and angle for pilot diesel and NG on combustion duration and NOx emissions has not been addressed adequately. CHAPTER THREE - ENGINE TESTING APPARATUS 3.1 Introduction The purpose of this chapter is to discuss the experimental setup and procedure used to achieve the objectives discussed in Chapter 1. The experimental part is divided into two categories: (i) performance and emissions, and (ii) combustion flame photography. A brief discussion of the existing experimental setup is explained. The existing instrumentation and the modifications carried out to provide optical access to the engine combustion chamber are discussed. 3.2 The Test Engine The DDC 6V-92TA engine was used for the main part of the experiment The DDC 6V-92TA engine is turbocharged and the air charge is after-cooled by an air-water heat exchanger placed after the blower. This engine provides the needed space for optical access through the existing glow-plug hole. Thus, the DDC 6V-92TA engine was modified for the high speed flame photographic setup whose results are discussed in Chapter 4. One of the cylinders was converted to a pilot-diesel-natural gas injection because only one prototype dual fuel injector was available during the course of this experiment However, a naturally aspirated single cylinder engine (DDC 1-71) was used for performance and emissions experiments because it avoids the interference (which might be caused by the adjacent 38 cylinders) of performance and emission data. Error analysis of the measured quantities are documented in Appendix G. 3.3 Test Conditions The medium speed of 1250 rpm and mid-range load of 3 bar (brake mean effective pressure, BMEP) were chosen as a typical operating condition with 140 bar natural gas injection pressure. Three test cases were studied (shown in Table 3.1) and are defined as: (i) case A where 100% diesel was used and is referred to as baseline-diesel; (ii) case B in which 30% diesel and 70% natural gas (on energy basis) was used, herein referred to as the gas-diesel, and (iii) case A* where 30% pilot diesel and 0% natural gas (was injected using the injector of case B) and is referred to as the pilot-diesel-only. The purpose of case A* was to provide visual observation of the flame during the ignition of the pilot diesel in the absence of natural gas. Throughout this thesis, the results pertaining to pilot-ignited natural gas combustion are based on test conditions shown in Tables 3.1 and 3.2. In all cases, the total amount of fuel injected was about 30 mg, except for case A* in which only pilot diesel (9 mg) was injected. Since the lower heating value for diesel (43 MJ/kg) and natural gas (50 MJ/kg) are about the same, the percentage energy ratio of the fuel can be related to the mass with an error of about 5%. Thus, throughout in this thesis, the 30% and 70% pilot diesel and natural gas will be referred to as 9 mg and 21 mg, respectively, totaling 30 mg. 39 Table 3.1: Geometry of the dual fuel injector and the injection parameters PGAS TGAS VDSL VGAS V p E T 0 DDSL DGAS (bar) (K) (m/s) (m/s) (m/s) (deg) (mm) (mm) 140 350 140 460 80 10 0.05 0.5 Table 3.1 shows the injection parameters for diesel and natural gas. The natural gas injection pressure and temperature are PGAS and TGAS, respectively, are similar to the high pressure natural gas bottles shown in Fig. 3.1. The diesel injection velocity VDSL was chosen as typical injection velocity for diesel engines (Varnavas [1991]). The natural gas injection velocity VGAS was calculated from stagnation conditions (supply temperature and pressure TGAS and TGAS) by applying one-dimensional compressible flow equations. The piston speed VpBT/ determined from the engine speed, is used to establish the boundary conditions for the natural gas injection (discussed in Chapter 6). The injection angle 0 is measured relative to the cylinder head surface as shown in Figs. 1.3 and 6.1. The hole diameters for the pilot diesel and natural gas are denoted as DDSL and DGAS; they were chosen such that the amount of fuel injected per cycle is the same for the dual fuel injector and the standard injector (originally supplied by the engine with six holes) for 100% diesel (discussed in Section 3.4). The dual fuel injector shown in Fig. 3.1 consists of 6 holes for pilot diesel and 6 holes for natural gas; they are equally spaced by (j>=30o (e.g. see Table 3.2 case B). 4 0 Table 3.2: Test conditions C A S E SPEED BOI a PW L O A D D S L : G A S 4> (RPM) (DEG) (DEG) (DEG) (BMEP) RATIO (DEG) C A S E A 1250 8* 3 10 3 100% : 0% -C A S E A * 1250 0* 3 7 1 30% : 0% -C A S E B 1250 6* 3 14 3 30% : 70% 30 0* the injection timing setting at 4 deg. BTDC for flame photographs 0* the injection timing setting at 4,8,12 and 16 deg. BTDC for cylinder pressure measurements Table 3.2 shows the operating conditions for the cases defined above. Throughout this thesis (unless stated otherwise), BOI is defined as the beginning of injection of diesel (or pilot diesel) referenced to TDC, and PW is defined as the pulse width, i.e. the injection duration. The symbol Q, indicates the difference in injection timing of natural gas relative to the pilot diesel. For case B, the commencement of natural gas was determined by optical inspection with schlieren photography of the injection pattern and the hydraulic pressure rise during fuel injection (discussed in Section 3.4), to be 3 deg. crank angle after pilot-diesel injection. The DSL:GAS is the energy ratio (in %) of the pilot diesel to the natural gas. The injection angle cj) is the circumferential angle of natural gas injection relative to that of the pilot diesel (shown in Figs. 1.3 and 6.1). The test conditions (load, speed and BOI) for the DDC 1-71 engine were set similar to those of the DDC 6V92 engine (cases A and B) in order to evaluate the thermal efficiency and NOx emissions. Test results from DDC 1-71 engine (from Douville [1994]) are discussed later in this chapter; they were useful to establish the test conditions for DDC 6V92 engine. 41 3.3.1 Diesel Fuel Flow Rate Diesel mass flow rate is measured based on the principle of the hydraulic "Wheatstone Bridge". Excess diesel from the engine flows to the recirculating tank so that the measuring device measures the net fuel consumption. The Wheatstone Bridge allows the diesel mass flow rate to be determined by relating differential pressure. The differential pressure is converted into an electrical signal and sent to the data acquisition system. Douville [1994] provides details of the measuring principle for the diesel flow rate for the DDC 6V 92-TA. 3.3.2 Natural Gas Flow Rate Natural gas mass flow rate to the engine was measured by a mass flow instrument (Micro Motion, Model DH012) which was installed in the natural gas line between the gas bottles (140 bar) and the engine intake as shown in Fig 3.1. Natural gas flows through a vibrating U-tube whose frequency is proportional to the product of gas velocity and the density. The frequency of the signal is converted to electric signal which is transmitted to the data acquisition system. The error specified for this instrument is ±0.4% at full scale reading. 3.3.3 Intake Air Flow Rate Intake air to the engine was measured by a turbine flow meter. Within the turbine flowmeter, the flowing diesel engages the vaned rotor, causing it to rotate at an angular 42 velocity proportional to the flow rate. The rotating turbine induces an AC voltage to a magnetic pickup coil mounted externally to the fluid process. The generated voltage is sent to the data acquisition system which is initially calibrated for flow measurements. 3.3.4 Manifold Temperature and Pressure The air intake and exhaust pressure were measured by pressure transducers mounted on the inlet manifolds. Similarly, temperature sensors were mounted on the manifolds. The manifold signals were transmitted to the data acquisition system. The ambient conditions (temperature, pressure and humidity) were read directly from analog meters located in the building where the experiment was performed. These values were manually entered into the computer to correct (standardize) the measured engine torque and power as recommended by SAE Handbook (Vol. 3 [1992]). 3.4 The Dual Fuel Injection Diesel fuel injection for the baseline case was injected by a commercially available injector (originally supplied with the DDC 6V92 engine). However, diesel was injected to the chamber by the prototype injector (Fig. 3.1) for pilot-ignited natural gas combustion. For the test conditions in which experiments and computations were performed (see Table 3.2 and 7.1), the amount of fuel injected per cycle was about 30 mg. The injection duration, PW, were adjusted to be about the same (Table 3.2) so that for 100% and diesel-gas injection 43 durations were 10 and 14 deg. crank angle. The observed late combustion of natural gas (Section 4.3.2) could be partly due to longer injection duration for diesel-gas (about 40% higher) higher than 100% diesel. A prototype dual fuel injector was designed and patented by P. G. Hill and K. B. Hodgins at the Department of Mechanical Engmeering, University of British Columbia (UBC). The injector delivers a fixed amount of pilot diesel, about 9 mg per cycle. The quantity of natural gas injection increases with increasing engine load so that the ratio of pilot diesel and natural gas decrease with increasing load. Figure 3.1 shows that the holes of the injector are designed such that the gas jets are directed so as to avoid collision with the pilot diesel. With this jet orientation the pilot diesel seeks the zones of available air in the combustion chamber for auto-ignition. This solenoid operated injector is of the same size as the conventional diesel engine injectors. The working principle of this injector is that, when the DDEC (Detroit Diesel Electronic Control) control unit sends an electric signal to command the beginning of injection (BOI) the solenoid operates the spool valve. The rotating cam pushes down the plunger thus building up high pressure for the injection of pilot diesel. As the pressure builds up, the gas needle lifts up against the spring to allow the injection of the gas (at 140 bar supplied by the intensifier) and simultaneously blocking further pilot diesel injection. The injection duration and the relative injection timing of the pilot diesel and natural gas (i.e. Q.) are largely determined by the strength of the spring above the gas needle. 44 VALVE Fig. 3.1: Pilot diesel natural gas injector (Courtesy, K. B. Hodgins) The injector was tested to establish the actual beginning of injection (BOI) for pilot diesel and natural gas in order to compensate for the effect of mechanical/hydraulic delay. This injection delay is referred to as the time interval between the solenoid injection command signal (i.e. "electrical BOI") and the actual opening of the needle valve (i.e. "physical BOI"). The hydraulic pressure for opening of the injector was provided by a cam (similar to the engine) rotating at 715 rpm. The schlieren technique shown in Fig. 3.2 was used where the dual fuel injection took place in a pressurized constant volume chamber (transparent) to simulate the engine conditions near TDC. The lenses with large focal length provide parallel beam of light through the combustion quiescent chamber. The high speed jet is visible (schieren pictures) because of the density gradient of the high pressure jet in the chamber. A single-shot high speed video camera was used to capture the jet position as it penetrates in the combustion chamber. The jet position was used to establish the "injection 45 delay" relative to the solenoid command of BOI. Figure 3.3 illustrates the schlieren picture for the jet position after 6.5° crank angle injection delay (corresponding to actual BOI) after the solenoid command. \ F L O W V I S U A L I Z A T I O N - S C H L I E R E N CW MERCURY ARC LAMP CAMERA SPEC. |TRIGGER| - CCD B&W VIDEO - 1/10000 t h OF A SECOND ELECTRONIC SHUTTER - OUTPUT TO FRAME GRABBER BOARD IN PC FOR ANALYSIS 1 Fig. 3.2: Schlieren flow visualization (Courtesy, P. Ouellette) Because it was difficult to distinguish the BOI for the pilot-diesel and the start of natural gas injection from the schlieren pictures, a hydraulic pressure rise in the injector was used to establish the start and end of injection of the pilot-diesel and natural gas, respectively. Figure 3.4 shows the injection characteristics for the pilot-diesel and natural gas. Point Pi corresponds to the start of pilot diesel injection which occurs at about 6.5° crank angle after the electronic command signal was initiated by the control unit Relative to the BOI of pilot-diesel, i.e. Pi, natural gas injection starts about 3° later as shown in Fig. 3.4. 4 6 Fig. 3.3: Jet position after 6.5 degrees crank angle Crank Angle Position (Degrees) Fig 3.4: Injector hydraulic pressure rise 47 3.5 Engine Instrumentation and Data Acquisition The fuel injection is controlled electronically by the use of an electronic control module (ECM) which contains a microprocessor. The E C M picks the current from individual sensors mounted on the engine (e.g. intake temperature, fuel flow rate, etc.) and processes the data necessary for feedback information to determine the beginning of injection (BOI) and pulse width (PW) for the next cycle. The output signals from all the engine sensors (except the E C M readings, cylinder pressure data and video signal) are sent to an intermediate board for conditioning. The conditioned signals are transmitted to a 16 channel analog/digital board (PCL 818) installed in a P C / A T computer. The BOI and PW are displayed directly on the E C M . The A / D board (PCL 818) conversion accuracy is specified by the manufacturer as ±0.02% full scale reading. The signal from the piezoelectic pressure transducer (model PCB 112A10) is conditioned by the charge amplifier before being sent to the ISAAC, a hardware that samples the cylinder pressures (at high speed) up to 100 consecutive cycles. The BDC signal triggers: (i) E C M to initiate injection; (ii) C C D camera for flame recording, and (hi) ISAAC to acquire cylinder pressure data. The crank angle signal acts as an external clock for ISAAC to determine the sampling crank angle interval (was 1 ° CA) for the cylinder pressure recording. The digitized pressure data is temporarily stored in ISAAC and later transferred to the data acquisition by a general purpose interface board (GPIB). Simultaneous screen display and 48 floppy disk data storage of the recorded pressure cycles takes place as ISAAC transfers data to the computer. 3.5.1 Torque Measurement A n eddy-current dynamometer was used to measure the torque developed by the engine. The dynamometer is an assembly of stator and rotor. The rotor is free to rotate. The rotation of the rotor is limited by a torque arm which applies force on the load cell (strain gauge). The calibrated load cell transmits the electric signal (developed by the load cell) to the data acquisition system. The error specified in this instrument is ±0.1% at full scale reading. 3.5.2 Engine Speed and Crank Position The engine speed and crank position was measured by an optical encoder (SoftPort Model S1-360IB) assembly which was attached to the crankshaft The sensor consists of a magnetic pickup with a 60-tooth gear which develops a frequency signal that is proportional to the engine speed. The engine speed is estimated by measuring the time elapsed between the occurrence at two successive signals. The digitized information is transmitted to the data acquisition system. The resolution of the speed sensor specified by the manufacturer is ±0.25°. 49 3.5.3 Cylinder Pressure Combustion cylinder pressure history data are useful for combustion analysis (discussed in Chapter 4). Among the engine parameters which can be deduced from cylinder pressure are: (i) rate of combustion, i.e. mass-burn-rate; (ii) engine indicated work; (iii) frictional losses, and (iv) cyclic variations in the combustion cycle. Foster [1985] reported that the rehability of cylinder pressure data is very important because a 1% error in pressure measurements can result in an approximately 50% error in heat release rate. Poor transducer linearity and sensitivity can adversely affect the measured data. Preliminary tests were performed to determine the most reliable pressure transducer among the commonly used and which were available within the UBC engine laboratory. They were tested for thermal effects by considering the high temperature combustion chamber environment Four transducers were tested by Hodgins and Mtui [1994]. Three were non-water-cooled piezoelectric transducers (model PCB112A10, PCB112M297A, PCB112M298) and one was a water-cooled (model AVL-8QP500C). In every test case the transducer was coupled to a charge amplifier (Kistler 5001). The test results (Appendix B) of the transducer performance show that the transducer 112A10 performs best in terms of sensitivity and linearity for the range of load tested. This is the type of transducer (with time response of 2 us) used throughout for cylinder pressure measurements reported in this thesis. The maximum error limit provided by the manufacturer is less than 1% for the range of pressure of 0-3000 psi (with a sensitivity setting at 1.10 pC/psi). 50 During the test, twenty consecutive cycles of pressure data were acquired and averaged (shown in Figs. 4.2a and 4.2b). The actual number of cycles required depend on the cycle-to-cycle variation of the engine. It has been determined from prior studies by Douville [1994] that 20 cycles is a statistically significant sample for pressure averaging (a minimum sample in which the standard deviation of the IMEP remains unchanged) for the DDC 6V-92TA and 1-71 engines. The variation of cylinder pressure from cycle-to-cycle was evaluated by the coefficient of variability (COV) in the indicated mean effective pressure. A typical COV value for twenty consecutive cycles for the experiments reported in this thesis was 0.52%. The COV is defined as the ratio of standard deviation in the indicated mean effective pressure (IMEP) to the average IMEP per work cycle. Determination of IMEP is discussed in Appendix B. Figure 3.5 illustrates a typical cyclic variation (COV =0.78%) of cylinder pressure for a sample of five firing consecutive cycles. The magnitude of the spikes in each cycle is due to cycle-to-cycle variations. Possible sources of cycle-to-cycle variations are discussed in Section 4.3.1. The mass-burned fractions which are reported in this thesis were obtained from averaged (20 consecutive cycles) measured pressures, so that the effect of cycle-to-cycle variations are minimized. COVMEP = ^-^*100% (3.1) IMEP Heywood [1988], stated that the vehicle drivability becomes difficult when COVIMEP exceeds 10%. 51 CA (DEG) Fig. 3.5: Typical cyclic variations in cylinder pressure 3.5.4 Exhaust Emissions The exhaust gas emission console consists of six sampling probes for the determination of exhaust concentrations of CO2, CO, NOx, (oxides of nitrogen) O2, C H 4 . and THC (total hydrocarbons). The emission console and exhaust gas measurements are briefly explained while detailed information on the apparatus and mstrumentation is provided by Tao [1993]. The exhaust gas sample is tapped approximately 6 m downstream of the exhaust manifold to allow for a uniform mixture of sample from the cylinder(s) as recommended by SAE Handbook (Vol. 3 [1992]). 52 Infrared analyzers were used to determine the concentrations of carbon dioxide (Beckman, Model 880), carbon monoxide (Siemens, Model 21P), oxides of nitrogen and methane (Siemens, Model 22P). When infrared radiation passes through a sample and a reference cell the intensity of the infrared light passing though the sample is attenuated. The attenuation is proportional to the difference of the concentration of the gas in the two cells. This attenuation is converted to concentration from the previously calibrated digital display. Total hydrocarbons were measured by a flame ionization analyzer (Model Ratfisch RS55). Burning of hydrocarbon in a hydrogen-air flame produces positive ions and electrons. An ion detector counts the number of ions which indicate the number of carbon atoms introduced into the flame. The number of carbon atoms is proportional to the concentration of the total hydrocarbon in the exhaust gas. The oxygen concentration was measured by Siemens gas analyzer (Model 5E) which utilizes the paramagnetic property of oxygen. Molecules of oxygen are attracted to the stronger part of the magnetic field when they are passed through a non-uniform magnetic field. A pressure difference is created when oxygen exhaust and reference sample are passed in a non-uniform magnetic field. This pressure difference is proportional to the oxygen concentration in the sample. The exhaust measuring console is calibrated prior to commencement of the experiment The uncertainties involved in these measurements are discussed in Appendix G. 53 3.5.5 Combustion Flame Visualization An endoscope coupled with single-shot CCD video camera (black and white Pulnix TM-745) was used for combustion flame visualization. The electronic shutter speed of the CCD camera is 0.0001 s. The endoscope is an optical probe (5 mm dia. by 550 mm length) which consists of an object lens to collect the image from the source and transmit it to the high speed camera for picture recording. Usually, it is positioned in the combustion chamber nearly in the same way as a pressure transducer. Figure 3.6 shows the endoscope installed in the combustion chamber with an angle (solid angle) of view of 78 deg. The endoscope probe is protected by a quartz window that withstands severe conditions (high temperature and pressure) in the combustion chamber and is sufficiently transparent for high speed photography. Cooling air (supplied at 5 bar) is circulated in the endoscope assembly to control the temperature to within the allowable limits. High gas velocity in the combustion chamber provides some degree of self cleaning of the window surface so that it remains clean from soot deposition over a reasonable period of operation. However, if the window surface is hit directly by the diesel fuel spray during the injection process, a significant darkening of the window is likely to occur. 54 Fig. 3.6: Positioning of the endoscope in the combustion chamber The endoscope assembly was adapted to the engine cylinder head by taking the advantage of the already existing glow-plug holes. The modification was done such that the hole is interchangeable with the pressure transducer for cylinder pressure measurements. The events occurring in the combustion chamber (fuel ignition, flame development and propagation) were captured instantaneously by a CCD video camera at a prescribed crank angle interval referenced to BDC. The control unit triggers the video camera after getting a command signal from the BDC marker. The video images are captured by a PC-based 55 frame grabber board and digitized. Image display and analysis is done by a PC-based image processing system. Figure 3.7 illustrates the layout of the experimental mstrumentation. 56 C l ) P C C O M P U T E R C 2 ) I S A A C U N I T C 3 ) C O N T R O L U N I X C 4 ) M O N I T D R C 5 ) O S C I L L O S C O P E C 6 ) E N G I N E ( 7 ) V I D E D C A M E R A ( 8 ) I N J E C T O R ( 9 ) E C M U N I T C I O ) S M O K E M E T E R (11) EMISSIONS CONSOLE (12) SIGNAL CONDITIONER (13) R P M / T D C E N C O D E R ( 1 4 ) E N D O S C O P E 3.6 Evaluation of Performance Parameters The parameters discussed below are useful in estabhshing the engine performance. 3.6.1 Brake Mean Effective Pressure (BMEP) This can be thought of as a constant pressure which if acting on the piston for the complete power cycle would produce the same amount of work as the actual varying pressure. This parameter scales (power level of) different sizes of engines. It is defined as: where BMEP is brake mean effective pressure in kPa and PB is the measured (brake) power in kW. VD is the total displaced volume in dm3 and N is the engine speed in rev/ s . The factor n R = 1 and 2 for a two and four stroke cycle engine, respectively. The brake power is given by: where TB is the measured torque (in N.m) and Ns is the engine speed in m/s. The indicated mean effective pressure (IMEP) is the theoretical value which is the sum (before losses take toll) of BMEP and the engine mechanical losses and energy consumed by accessories mounted on the engine. The IMEP is given by: BMEP = PB *103*nR VD*N (3.2) PB = 2*n*Ns *TB (3.3) 58 JMEP = W'™*103 ( 3 4 ) where WIND (in kj) and VCYL (in dm3) are the indicated work per stroke and cylinder volume, respectively. The LMEP is evaluated from the pressure-volume (p-v) diagram obtained from experimental measurements. A typical (experimental) p-v diagram is shown in Fig 3.8. The WIND is given by eqn. (3.5). W IND = o z LLJ o 20 LL uu 15 rr w 10 .DIESEL.100% .DSL30% + GAS70% 0 L J _ U _l_ _ L -16 -14 -12 -10 INJECTION TIMING (DEG CA) -8 Fig. 3.9: Thermal efficiency (exp. at DDC 1-71 engine) . Q I . i I i . . . I . . i i I , . i . I i , . , I -16 -14 -12 -10 -8 INJECTION TIMING (DEG CA) Fig. 3.10: Emissions of NOx (exp. at DDC 1-71 engine) 63 CHAPTER FOUR - DISCUSSION OF EXPERIMENTAL RESULTS 4.1 Introduction This chapter discusses the experimental results with the purpose to show: (i) directly observed features for pilot-ignited natural gas combustion, (ii) differences/similarities between the 100% diesel and pilot-ignited natural gas combustion, and (iii) the features which may be inferred by the XPNOX code from the measured data. XPNOX is a multi-zone thermodynamic model (Hill and Douville [1996]) to determine combustion rate and NOx formation from a measured cylinder pressure data and performance of the diesel engine. Conservation of energy at each crank angle interval is satisfied and combustion of fuel-air mixture is assumed near stoichiometric. Wall heat transfer during combustion is determined by non-linear fit to a Gaussian distribution, rather than by the use of the well known Woschni [1967] method which may be uncertain by 50-100% when applied to a particular diesel engine. The XPNOX model (described in Appendix C) solves iteratively for the wall heat transfer, so that it satisfies the overall energy balance, i.e. the calculated heat transfer dependent on engine operating conditions obtained from the measured cylinder pressure and inlet and exhaust conditions. The skewed Gaussian distribution from the wall heat loss is consistent with measured results reported in literature (e.g. see Dent and Sulaiman [1977]). The method utilizes discretized burning zones (about 100 zones), separated by burned zones, which remain unmixed for a typical mixing time of about 0.5 ms (initially determined from the KIVA-II 64 results discussed in Chapter 6). Mixing of residual gas with burned, and unburned gas takes place at the end of combustion time step (typically 1 deg. crank angle) so that NOx formation ceases after mixing. The effects of the pilot diesel and natural gas injection conditions (see Table 3.1), were accounted for by determining the enthalpies (of the diesel and natural gas) based on the injection temperature and pressure. On the other hand, the effects of the residual gas were considered in the XPNOX code. With the KIVA-II code (discussed in Chapter 6), the boundary conditions during injection were specified at the injector location. However, KTVA-II (in the current state of development) does not account for the residual gas in the combustion chamber. Experiments were done at BOI of 4, 8, 12 and 16, but emphasis on the results is put on the BOI of 4 deg. BTDC (e.g. the flame photographs discussed in section 4.2) because the preliminary results discussed in Sections 3.6.2 and 3.6.3 show that late injection timing (say, at BOI of 4) provides advantages of high thermal efficiency with low NOx emissions. The test conditions for the experiment are shown in Table 4.1 for cases A, A* and B. The choice of the test conditions (speed and load) is based on a typical engine operating at medium conditions. The injection duration (PW) was chosen to match the amount of fuel needed for such load condition. The geometric (circumferential) injection angle Patm) because these radicals participate in the chain branching and propagation which leads to self-sustained combustion. Zhou and Karim [1994 I) suggested that the reaction of methyl radicals CH3 is one of the main reaction steps governing the oxidation of methane. During the ignition 91 delay period, C H 3 radicals determine which pathway the reaction proceeds, e.g. branching, recombination, etc. Typical oxidation path of methane was reported by Zhou and Karim as CH4—»CH3-> C H 2 O -»HCO -> CO -» CO2. Figure 5.1 idealizes the chemical reaction of hydrocarbon fuels depicting the intermediate active species which are consumed as the chemical reaction progresses. The rate of heat release q' increases rapidly with temperature T as the intermediate products are converted to the final products. P r e h e a t zone Reac t ion zone 1 Reactants time-*--Fig. 5.1: Intermediate species during combustion (Glassman [1987]) Usually, chemical reaction occurs in a number of fundamental steps. Zhou and Karim [1994 I] pointed out that, even the simplest hydrocarbon fuel consists of over a hundred reaction steps. The computer time and memory storage required to solve for such copious reactions is overwhelming with even today's super-computers. Heywood [1988] stated that 92 practical fuels, e.g., diesel and gasoline are composed of a mixture of numerous species, and that, detailed modeling and vahdation of the reaction mechanisms are difficult and uncertain. Due to the difficulties in computing detailed reaction steps for practical use, smplifications have been made by Westbrook and Dryer [1981], Zellat et al. [1990], Mulholland [1992] and Sloane et al. [1992] so that the complex multi-step reaction process can be approximated by only the few reaction steps which were thought to be dominant Limitations of simplified chemical for methane reaction kinetics are discussed below. 5.3 Ignition of Methane Westbrook and Dryer [1981] considered a simplified reaction mechanism (single- and two-step) for the oxidation of hydrocarbon fuels. They found that a single overall reaction could reproduce the flame speeds of hydrocarbons over a wide range of stoichiometric mixtures. Westbrook, however, did not discuss how well a single step reaction rate reproduces the thermal structure (temperature and heat release profiles). Sloane and Roney [1992] stated that care must be taken in applying single-step chemical kinetics because the simplified reaction rate fails to describe both ignition and flame propagation and leads to large discrepancy between computed and experimental results. The main discrepancy is associated with the conditions pertaining to the developing of the flame kernels which were ignited by a localized energy source (in pre-mixed flames) on which the models were calibrated. 93 Further, Sloane and Roney reported that, the activation energy for the single-step overall reaction is much higher than the thermal energy (denoted by E/Ru*Tad) of the gas at the adiabatic flame temperature, so that the total flame is concentrated on a small region. They pointed out that, in such reactions, the correct heat release profile in the reaction zone of the developed flame front is where the fuel and oxygen concentrations are low relative to the initial values. Further, they noted that, the single-step reaction over-predicts flame temperature (above 1700 K) leading to higher reaction rates. Thus, the use of single-step reaction rates (particularly in non-premixed conditions) in combustion modeling leads into significant level of uncertainties, e.g. determming the ignition delay, combustion duration and rate of heat release. The single step reaction for methane (CH4) is expressed as a CH4 + P 02 > Y C02 +5 H20 (5.1) The constants a through 5 are stoichiometric coefficients. The reaction rate co (mol/ cm3.s) for methane is given by co = kflCmriOif (5 .2 ) where [CH4] and [O2] are the concentrations (mols/cm3) for methane and oxygen, respectively. A generalized form of eqn. (5.2) is shown re-written as eqn. (6.3) in Chapter 6 which allows for the forward and backward chemical reactions. The reaction rate constant kf is expressed as 94 kf = A*7/ c * exp -V (5.3) RU*T J where A and X, are empirical constants and T is the local temperature. EA and Ru are the activation energy and universal gas constant, respectively. The EA is reported by Zhou and Karim [1994 I) that it is not necessarily constant and can vary locally with temperature, pressure and equivalence ratio. Thus, care must be taken in applying such correlations in different circumstances. The form of eqns. (5.1) to (5.3) is applicable for combustion of other hydrocarbon fuel with the appropriate constants. The constants for diesel and methane are shown in Appendix D. The pre-exponential constants shown in Appendix D are adjusted to match the experiment results. 5.3.1 Ignition Delay Ignition delay of natural gas in an engine-like condition has been reported by Naber et al. [1994]. The experimental vessel of Naber et al. was a disk-type constant volume chamber (diameter 114 mm x width 28.6 mm) with heated wall to about 450 K, similar to engines wall temperature close to TDC. A two-step combustion process was performed as follows: (i) generation of engme-like condition (high temperature and pressure) by spark-ignition of a lean mixture (with some additives) whose combustion products are similar to air (i.e. 21% O2 and 79% N2), and (ii) high pressure natural gas (about 207 bar) was injected in the hot chamber so that it auto-ignites. A rotating fan was incorporated in the chamber to generated turbulence. Such test conditions represent quantitatively some important features of IC engine near TDC. 95 For the ignition of the injected natural gas, Naber et al. reported that below 1200 K the natural gas ignition delay is of Arrhenius type. Above 1300 K, the ignition delays approached a limiting value depending on the injection system, say, time response of the injector. The ignition delay T i g was correlated by Naber et al . (eqn. 5.4) and fitted in experimental results shown in Chapter one (Fig. 1.3). The ignition temperature for natural gas (~ 1250 K) reported by Naber et al. has demonstrated the need of ignition assistance (pilot diesel) for diesel engines fueled with natural gas. Modeling and implementation (in the KfVA-II code) of ignition and combustion of hydrocarbon fuels are discussed in Chapter 6. C2 + A: exp Ku J c J 1/2 (5.4) where C, p and p 0 are constant representing the physical delays, chamber pressure and reference pressure. Rw and EA are the universal gas constant and activation energy. The core temperature and pressure exponent are denoted by T c and r, respectively. The physical delay constant (C « 0.41 ms) is mainly due to finite fuel injection rate, pressure sensor delay and fuel/air mixing. The pressure exponent for methane is r * -0.94. The overall average core temperature T c is related (eqn 5.5) by the vessel wall temperature T w and the mass averaged bulk temperature Tb determined from the pressure, density and molecular weight using the perfect gas law. 96 T ^-=l + a ( \ (Tb } 1- w \ + b (5.5) Using the constants a and b of 0.005 and 0.108, respectively. The core region of the combustion chamber is defined as the inner 90% of the vessel volume. For the wall temperature of 450 K and density of 20.4 kg/m3, Neber et al. concluded that the ratio Tc/Tb ranges from 1.08 at 1500K to 1.04 at 700K. 5.4 Laminar Flame Speed The laminar burning flame speed SL, is the velocity at which unburned gases move perpendicular to the combustion wave surface. Flame propagation in practical combustion systems plays a key role for combustion. Figure 5.2 illustrates flame propagation by thermal diffusion where burning is taking place at zone I. The temperature at zone TJ is raised to ignition due to the heat diffusion from zone I to zone TJ. The unburned gas temperature, ignition temperature, and flame temperature are TQ, Ti and Tf, respectively. The reaction zone (flame thickness) is denoted by 5, (=CC/SL) where a is the thermal diffusivity and SL is the laminar flame speed. The reaction surface is assumed by Chomiak [1990] as the reacting region at which 50% of the reactants are changed to products. 97 X Fig. 5.2: Laminar flame reaction zones (Glassman [1987]) Westbrook and Dryer [1981] pointed out that, typical flame speed of paraffin fuels at atmospheric pressure is about 40 cm/s (at cb=l), while for methane is 38 cm/s. The dependence of reaction rate on the flame speed (discussed below) shows that, for given same conditions, NG burns more slowly than most paraffin fuels. For most hydrocarbon fuels correlated (Adamczk and Lavoie [1978]) for simplified global kinetics (eqn. 5.6), the flame speed decreases with increasing pressure. SL « PR (5.6) where the exponent r = (a+b-2)/2. The constants a and b are empirically determined so that the flame speed agrees with the experiment Westbrook and Drayer [1981] suggest P 0 3 for methane (i.e a=-0.3 and b=1.3) and P 0 1 3 for gasoline. As hydrocarbon fuels, methane and gasoline are denoted by CH* and CsHis, respectively. 98 Metghalchi and Keck [1982] have provided flame speed correlation (eqn. 5.6) which is applicable for engine conditions for pre-mixed combustion: where T 0 and P0 are the reference temperature (298 K) and pressure (1 atm.), respectively. SL C V OC and P and are constants for a given fuel and equivalence ratio. T u is the unburned gas temperature and P is the local pressure. 5.5 Flame Propagation in Turbulent Combustion The diffusive flame is characterized by the burning rate (or fuel consumption) which is determined by the rate at which the fuel and oxidizer are brought together in proper proportions to form a combustible mixture. Such a mixing-controlled combustion is valid when the Damkdhler number (Damkohler number defined in Section 5.7) is high and ignition has occurred. The integral length scale is a characteristic dimension for larger scale structure in turbulent combustion. Combustion is controlled by large scale mixing and diffusion occurs at small scale. The integral length scales have been computed by KIVA code and were about 0.5 mm for diesel-gas injection under test conditions shown in Tables 3.1 and 3.2. Such small turbulence length scales are consistent with the jet induced turbulence which is about 10 times higher for diesel-gas injection compared to 100% diesel (5.7) (5.7) 99 (discussed in Chapters 8 and 9). Such turbulence increases the flame surface area so that combustion rate is increased compared to laminar case. 5.6 Chemical Reaction Time Scale Based on the laminar flame speed, the chemical reaction time scale is defined as T C =X/SL where A is the Taylor microscale and 5L (aX) is the laminar flame thickness. The chemical reaction time is defined in this manner because laminar diffusion occurs over the Taylor microscale. Combustion times are usually much smaller than ignition delay times. In the review paper by Daneshyar and Hil l [1987], they suggest that the typical values for (spark engines) for flame thickness and speed are, respectively, 6L~0.01 mm and SL^O.S m/s resulting in xc ~ 20 us (compared to t i g ~ 2 ms). 5.7 Mixing-Controlled Combustion Combustion in compression ignition (CI) engines is comprised of two phases: (i) the premixed, and (ii) the mixmg-limited (mixing-controlled) combustion. Heywood [1988] states that the majority of fuel (more than 75%) in CI engine burns under mixmg-limited combustion phase during which the rate of burning and heat release is considerably lower than those of the premixed phase. This is consistent with the findings of Plee and Ahmad [1983] noted that the majority of the diesel combustion process is mixmg-limited because TC « Tnv where T c and x m are the combustion and mixing times, respectively. The mixing time is defined as T m =L/u ' where u1 is the root-square-mean (rms) velocity fluctuation and L is 100 the turbulence integral length scale. Fraser et al. [1986] suggested that L is about one-fifth of the clearance height at TDC. The fluctuation velocity varies between 10 m/s (close to the nozzle) to 1 m/s (far away from the nozzle). Comparison of chemical reaction time scale T c to mixing time-scale xm provides a non-dimensional parameter for the relative rate of reaction. The non-dimensional Damkohler number is defined as: Da = — = - * — (5.9) where L / u ' defines the mixing time and the term SL/5L accounts for the presence of the flame. As discussed in Chapter 8 and 9, the characteristic mixing time xm~0.5 ms (>TC), therefore combustion is controlled by the rate of mixing. Damkohler number has been used by Wong W. et al. [1979] and Plee S. et al. [1983] to distinguish between the kinetically control ("slow" reaction) and mixing-controlled ("fast" reaction) rates. For D a of the order of unity, both time scales determine the reaction rate. The chemical reaction, however, is considered to be kinetically controlled for D a « l (in which the mixing rate is small) and mixmg-limited if D a » l (for intense mixing). 5.7.1 The Magnussen Combustion Model Magnussen and Hjertager [1976] proposed a mixing-controlled combustion model, on the assumption that the fuel and oxidant exist in different eddies and the chemical reaction is 101 controUed by the rate of mixing of these eddies. Product concentration also affect reaction rate because heat is transported from the reacting zones of fuel and oxidant Fig. 5.3 illustrates the mixing-limited combustion/ showing the diffusion of fuel and oxidant mixture to the reaction zone. Mixing Qxidizer reaaani. Fuel vapor crji droplets (contact surface) Diffusion combustion Oxidizer hurnt fases fuel vapor g^d droplets Fig. 5.3 Description of mixing-controlled-combustion Oxidizer and hurnt fases (diffusion flame) For fuel deficient zones, the reaction rate COF (mols/cm3.s) is controlled by the fuel concentration (eqn. 5.10a) and for the oxygen deficient zones, the oxygen concentration controls the reaction rate 0002 (eqn. 5.10b): to F = —— *[FUEL] Tm - Am m [OXYGEN] (5,10a) (5.10b) where the concentrations (mols/cm3) of fuel and oxygen are [FUEL] and [ O X Y G E N ] , respectively. SF is the oxygen-fuel stoichiometric ratio for diesel and for natural gas. A M is model empirical constant Tm (defined as k/ e) is the characteristic turbulence mixing time 102 where k is the turbulent kinetic energy and e is the rate of dissipation of k. The quantities k and e are determined from the turbulence modeling described in Appendix F. In the premixed flame, the fuel and oxygen are in mixed structures separated by hot combustible products. In this circumstance, the reaction rate (eqn. 5.10c) is determined by the dissipation of the structures contaming the hot combustible products which provide the premixed structures with the activation energy needed for combustion. The concentration of products (mols/cm3), is denoted by [PRODUCTS] while Bm is an empirical constant - r _ _ K .[PRODUCTS, ( 5 1 0 C ) T m SF + 1 Gosman et al [1982] simulated combustion in a homogeneous charge engine by combining Magnussen's equations (5.10a) to (5.10c) to the form shown in equation (5.11). In the application of the model (eqn. 5.11) the smaller term in the bracket controls the reaction rate. The constants for the mixing-controlled reaction i.e. Am, Bm, and SF for diesel and natural gas are shown in Appendix D. T m SF SF + 1 The Magnussen combustion model has been the choice for this thesis because, though simple, it represents the mixing effect on combustion; a phenomenon common in diesel engine combustion. Although this model lacks universality because it requires the "tuning" of the empirical constants, it has been shown by Pinchon [1989], Kuo and Ritz 103 [1989], and Varnavas et al. [1991] capable of reasonably reproducing the features of both premixed and diffusion flames. For pre-mixed and dtffusion flames, Magnussen suggested Am=4 and Bm=0.5 for the model shown in eqn. (5.11). Adjustments of constants have been reported by Pinchon (Am=16 and Bm=2) and Varnavas(Am=0.5 and Bm=0.5). For the results reported in this thesis, both for 100% diesel and diesel-gas combustion, the constant were modified to Am=l and Bm=l (see Appendix D) to match the experiment Tuning of constants seems to be necessary, partly, because of the limitations of the model to account for engines operating conditions. In such modeling, the transitional point between ignition (kinetically controlled) and combustion (mixing-controlled) has been defined by Varnavas as 2% of the mass burned fraction. The transitional point for the results reported in this thesis is discussed in Section 6.4.3. 5.8 Fuel Air Mixing in Dual-Fuel Injection This section discusses the possibilities of injection timing of pilot diesel relative to the natural gas and subsequent formation of combustible mixture resulting from the pilot diesel and natural gas. Three possible mjecting timings of pilot diesel gas can be categorized as: (i) diesel injection precedes that of natural gas; (ii) natural gas injection precedes that of diesel, and (iii) diesel and gas injected simultaneously. 104 Figures (5.4) through (5.6) idealize the ignition delays relative to the positions of fuel injection. Results of the influence of the relative injection timing are discussed in Chapter 9. The ignition delay has been defined (in this thesis) as 0-5% mass burn fraction. The ignition delay for pilot-ignited natural gas engine, say, for injection pattern shown in Fig. 5.4 (case study in this thesis), a complex interaction of pilot diesel and natural gas is inevitably during the pre-ignition process. The effects of such interactions are discussed in the next section. i -o LU CO Q i i l l l l l l l l l DSL IGNITION DELAY o UJ CO < CD GAS IGNITION DELAY •>CA (DEG) Fig. 5.4: Pilot diesel injection prior to natural gas o I -o UJ 2 o co < co Q ->CA (DEG) DSL IGNITION DELAY GAS IGNITION DELAY Fig. 5.5: Natural gas injection prior to pilot diesel 105 o DSL IGNITION DELAY GAS IGNITION DELAY Fig 5.6: Simultaneous injection of pilot diesel and natural gas For the injection configurations shown in Figs. 5.4 to 5.6, the pilot diesel and natural gas necessarily will share the air in the combustion chamber. Thus, the derivation of the equivalence ratio accounts for the presence of pilot diesel and natural gas in the vicinity of available air. The total air-fuel ratio and the resultant equivalence ratio derived in the following section were implemented in the KTVA-II code discussed in Chapter 6. The overall equivalence ratio is defined in the conventional manner as the ratio of the actual fuel-air ratio divided by the stoichiometric fuel-air ratio. The equivalence ratio for the pilot diesel and natural gas are given by eqns. (5.12) and (5.13). _ TTlDSL /THAR>DSL / r i n \ cJ)Dsi = (5.12) FARDSL and 106 O G A S = ( 0 . 1 3 ) FARGAS Where FARDSL and FARGAS is the stoichiometric fuel-air ratio for diesel and natural gas, respectively. The rn.AiR.DSL and m A i R G A S are given by eqns. (5.16) and (5.17). The resultant equivalence ratio which accounts for the presence of pilot diesel and natural gas is given by: , (niDSL+ ntGAs) / niAm / C I A \ Oror = (5.14) FARTOT where T?AT> - M D S L + M < 3 A S 1 c\ FARTOT = (5.15) m Am, DSL + niAm, GAS The mass of air needed for combustion of diesel (m AIR DSL) and natural gas ( m A i R G A s ) are respectively given by equations (5.16) and (5.17). The assumption is that the air required for the combustion of the individual fuel is proportional to the fraction of the stoichiometric air mass. The total mass of air, m A r c = m A i R r e L + m A i R G A S . r niDSL / FARDSL , * / r - i ^ \ m Am, DSL = I ; ; J mAm [p. lb) niDSL' FARDSL + ntGAsf FARGAS r niGAs / FARGAS 7 . _ , _. triAm,GAS = / : ; J niAm (5 .17) ntDSLI FARDSL + niGAs' FARGAS 107 5.9 Chemical Interaction in Pilot-Ignited Natural Gas Engines The introduction of natural gas in the diesel engine (say, the test conditions in Table 3.2 in which natural gas injection takes place before pilot diesel ignition, similar to Fig. 5.4) can affect significantly both the physical and chemical processes during the ignition delay period, modifying the ignition delay of diesel in the absence of natural gas. Changes in physical properties due to natural gas injection, may be due to temperature, pressure and properties of natural gas (e.g. specific heat and molecular weight). In this work, no attempt has been made to establish whether the ignition delay is mainly controlled by the chemical (resulting from chemical interaction of diesel and natural gas) or physical factors. However, due to strong temperature dependency of ignition delay, it may be assumed that, the ignition delay is mainly controlled by chemical factors (active species formed during the pre-ignition period) so that the type and quantity of gaseous fuel in the combustion chamber play a role during the pre-ignition period. Results on the ignition and combustion duration for pilot-ignited natural gas combustion (including injection timing leading to long pre-ignition period) are discussed in Chapters 4 and 7. 108 CHAPTER SIX - COMBUSTION MODELING IN IC ENGINES 6.1 Introduction The experimental results obtained are limited because they provide average parameters of the measured quantities, e.g. thermal efficiency and NOx emissions. The average parameters provide a limited possibility of detailed study of the interaction of fluid mechanics and the combustion process. In order to obtain a detailed spatial and temporal processes occurring in an internal combustion engine (say, ignition sites, temperature distribution, etc.), a parametric study of the engine's combustion process is necessary to supplement and provide some more information of the m-cylinder processes. It is therefore the purpose of this chapter to discuss the use of multi-dimensional computer modeling using the KTVA-II code (for this thesis) because experimental parametric studies are very costly and time consuming. As will be discussed in Chapters 7 through 10, the KTVA-II code provides valuable information pertaming to the thermal and flow field occurring in the combustion chamber. Currently, internal combustion modeling is playing a significant role for engine detailed studies and has recently increased steadily with increasing computer storage capacity and speed. In the long term, mathematical modeling is cost-effective to study combustion mechanisms as well as to guide development and design of engines. Computer simulation of the mixture formation and combustion process in direct injection (Dl) engines has made it possible to reasonably model the m-cylinder phenomena and predict the engine 109 performance and the pollutant formation. There are several phenomenological models (Chiu et al. [1976] and Mansouri [1982]) which describe the spray behavior and turbulent mixing as dominant factors of diesel engine combustion. Recently, a detailed multi-dimension model for diesel engine combustion has been developed by introducing numerical fluid dynamics into the description of the m-cyhnder processes. Such multi-dimension models performed by Gosman et al. [1982], Su and Traci [1987] and Amsden et al. [1989] have included gas motion, spray behavior, ignition, combustion and pollutants formation. The KTVA-II code is briefly described in this chapter. 6.2 The KIVA-II Code The KTVA-II computer program developed by Amsden et al. [1989] has been modified (for this thesis) to perform numerical simulation for pilot-ignited natural gas combustion engines. The general features of the KTVA-II code is discussed briefly in this section while the detailed information is documented by Amsden et al. [1989]. KTVA-II is a multi-dimensional code which is specialized for IC engines. It embodies engine submodels including fuel spray, atomization and fuel drops breakup/coalescence. A user input file permits changes of the input variables, e.g. chamber geometry, engine speed, injection velocity, etc. 110 6.2.1 The Governing Equations The governing equations are typical of the Navier-Stokes transport equations for the conservation of mass (for arbitrary number of species), momentum and energy for the gas phase in the chamber. Turbulence is modeled by the widely used k-e model. The KTVA-TJ code solves for the finite difference approximations of the governing equations (shown in Appendix F) discretized both in space and time. The spatial discretization of the computational domain is constructed using a finite volume method which preserves the local conservation properties of the differential equations. Two categories of chemical reactions are considered in the KIVA-II code: (i) the slow reactions which proceed kinetically (Arrhenius form), and (ii) the fast reactions which are assumed to be in equilibrium. The gas mixture is assumed to be an ideal gas with uniform transport coefficients as shown in eqns. 6.1a, 6.1b and 6.1c for mass, momentum and energy. D'-T- (6-la) pSc u = ua l r+ c M — (6.1b) K = (6.1c) where D, u and K are the mass diffusion, viscosity, and heat conduction, respectively. I l l 6.2.2 The Chemical Reactions The rate of consumption of fuel and the formation of NOx are assumed to proceed by chemical kinetics . For reaction r and specie ra, the chemical reaction proceeds as: (6.2) where Xm represents the mole of specie ra. amr and bmr are the stoichiometric coefficients of reaction r. The kinetics reaction rate (A is given by the expression co m \ "m ) m (6.3) where Dm and W m are the density and molecular weight of specie ra. The constants a'mr and b'mr are the forward and backward empirically determined reaction orders for specie ra in reaction r. The constants a'^ and b'mr are not necessarily equal to ami and bmr, respectively because they are adjusted to match the experiment The forward and backward reaction rates constants kf and kb are evaluated from Arrhenius expression of general form as: kr = ArT^exp E 1 -V Ku 1 J (6.4) where A r is the pre-exponential constant, k is the temperature exponent and E r is the activation energy for reaction r. Ru is the universal gas constant and T is the local temperature. 112 The dissociation of reaction rate for species is assumed to proceed infinitely fast such that equmbrium reactions are invoked. The equihbrium reaction proceeds as shown in eqn. (6.5) n Pm ^ yWmJ KC(T) (6.5) where amr and tw are the respective forward and backward constants of species m in reaction r. Kc(T) is a temperature dependent equihbrium constant for reaction r. 6.3 Initial and Boundary Conditions Initial and boundary conditions are established prior to the commencement of the computation. 6.3.1 Initial Conditions Prior to computation, initial conditions are established to match the experiment The initial conditions pertains to velocity, temperature, pressure, turbulent kinetic energy and its rate of dissipation. 6.3.2 Boundary Conditions The law of the wall functions which were initially suggested by Launder and Spalding [1974] are assumed. Wall functions are used to specify the fluxes of the momentum and 113 energy at the wall. This is accomplished by matehing the computed fluid velocities and temperature at the grid points adjacent to the wall. The wall shear stress T w is calculated from the frictional velocity u* (i.e. Tw=pu*2) given by equation (6.6) and (6.7) for laminar and turbulent flow, respectively. In the laminar sub layer, a laminar velocity profile is assumed which results in eqn. (6.6). U ~ l ( y U ' (6.6) v ' In the fully turbulent region, the logarithmic law velocity is used (eqn. 6.7). u 1 , — = — In u K ' yu x V v J + B (6.7) where K is the von Karman constant (« 0.4327) and y is the distance from the cell face the adjacent wall, v is the kinematic viscosity. The constant B is of the order of 5.5. The wall condition is specified by a constant temperature. The heat flux at the wall, J w is determined by the formula: Jw 1 Pu'cP(T-Tw) where T is the gas temperature, PRL is the laminar Prandtl number and C p is the specific heat at constant pressure. 114 The turbulent kinetic energy, k and the rate of its dissipation, e at the wall are defined respectively by equations below. k = C„ e = C M . I P (6.9) (6.10) where k and e are evaluated at a distance y from the wall and the constant C u e is given by CM£ = K K - c , ) l 1/2 (6.11) The values of constants Q i and C£2 are shown in Appendix F. 6.4 Modifications Done to the KIVA-II Code The simulated results discussed in Chapters 7 through 9 were performed after the following modifications were done to the KTVA-II code: (i) update diesel fuel library data; (ii) incorporate mixing-controlled combustion to co-exist with the ignition model, and (iii) provide for injection and combustion of pilot-diesel and natural gas. 115 6.4.1 Diesel Fuel Library Data The diesel fuel library was replaced with updated data (Varnavas [1990]) which is more representative for IC engine simulation. The proposed fuel chemical formula (Appendix D) is of the form C13H23 which is within the range of light fuels of the general form CxHi.8X suggested by Heywood [1988] and heavy fuels proposed by Ferguson [1985] as Q4.4H24.9. The diesel fuel injection was injected with a velocity of 140 m/s with assumed droplet of Sauter mean diameter (SMD) of 20 um. The injection angle relative to the cylinder head was 10° as shown in Table. 3.1. 6.4.2 Natural Gas Injection NG injection requires inflow boundary condition. The injection velocity and reference density of the NG are calculated from one dimensional compressible fluid flow of ideal gas relationships. The NG injection pressure and temperature shown in Table 6.1 are specified to match the experiment The choked condition was assumed because the NG injection pressure was 140 bar while the cylinder pressure was about 45 bar (at TDC). With natural gas injection conditions shown in Table 3.1, the jet is fully turbulent The nozzle exit density, mass flow rate, temperature and velocity are computed from the upstream specified conditions shown in Table 3.1. 116 Applying the ideal gas relationships for injection conditions shown in Table 3.1, the resulting natural gas density at nozzle exit is 77 kg/ m 3 compared to 750 kg/ m 3 for diesel fuel. With the injection conditions shown in Tables 3.1 and 3.2 (engine speed of 1250 rpm) the injection rates are 22.5 mg/ms and 14.3 mg/ms for cases A and B, respectively. The PW for 21 mg of natural gas is 11° (i.e. 14-n=ll°, where H=3 was determined as discussed in Section 3.4). The total mass of fuel injected per cycle is 30 mg both for cases A and B. The 70% of the total fuel (i.e. 21 mg of natural gas ) for case B was used to calculate the momentum rate, for comparison with case A. With the above injection rate coupled with the injection velocity (Table 3.1), the rates of momentum injection for cases A and B are 3150 and 6586 kg.m/s2, respectively. This indicates that, the momentum injection rate for case B is about 2 times that of case A, resulting in higher penetration rate for case B (discussed in the next chapters). For three-dimensional computation, the velocity of the pilot diesel spray and natural gas jet (shown in Fig. 6.1) is decomposed in the x-y-z directions with respect to angles 0 and 4», respectively. The face cell velocities are prescribed where the injector is located in the computation cells. In addition to the injection velocity, inflow boundaries for the inlet turbulence kinetic energy TKEJN , and turbulent length scale XIN were specified. The TKEEM was evaluated as 10% of the square of the mean piston velocity (Grasso et al. [1987]), i.e. TKEIN " (VPBT)2 • The value of mean piston speed is given Table 3.1. The length scale XIN was comparable to the grid size (~ 1 mm) at the location of the injector hole. 117 6.4.3 Chemical Reactions Ignition for pilot-diesel and NG was modeled by a single step, kinetic chemical reactions (see Sections 5.3 and 6.2.2). The mixing limited combustion model has been incorporated in the code and allow for a user input ignition transition temperature. Ignition transition temperature, in this context, is defined as the transitional temperature form kinetically- to mixing-controlled combustion. In the results reported in this thesis, the transition temperature corresponds to the diesel ignition temperature (* 825 K) so that ignition delay is less than 2 ms. Ignition and combustion co-exist so that the overall chemical reaction is governed by eqn. (6.12) which is derived from eqn. (6.3) and (5.11). The chemical reaction is controlled by the slowest rate. The kinetic and mixing limited reaction rates are respectively ick and com. In the quenching zones as well as excessively lean or rich mixtures zones the conversion rate is controlled by chemical kinetics. The kinetics and mixing-controlled reaction rate constants are shown in Appendix D. w (6.12) (i/co* + i / w j The formation of NOx was a modeled by the Zeldovich mechanism given by eqns. 2.2a-c discussed in Chapter 2. The reaction rate in the formation of NOx is assumed to be of Arrhenius type. The reaction rate constants for the formation of NOx are given in Appendix D. 118 6.5 Numerical Computation Computation was commenced at the intake port closure (i.e. 125 degrees CA BTDC) of the combustion chamber. The grid arrangement for the computational domain is shown in Fig. 6.2 with grid size: 32(radial), 20(axial) and 6(circumferential). The grids in the bowl are body-fitted while those in the squish region undergo compression or expansion as the piston moves towards or away from the BTDC. It is noted that the grid density is higher close to the injector location (point P in Fig. 6.2) so that higher resolution of the fuel injection and ignition is achieved. The injector tip contains twelve equal-spaced holes around the tip in two planes arranged alternately as shown schematically in Fig. 6.1. The upper and lower planes consist of six holes for natural gas and pilot diesel, respectively. These holes are mclined at 10 degrees from the surface of the cylinder-head fire-deck. This arrangement separates the pilot diesel holes from those of natural gas by angles 0 and <$> of 10 and 30 degrees, respectively. The computational grid arrangement in Fig. 6.2 is a 60-degree pie-shaped sector so that it contains two holes; one for pilot-diesel and the other for natural gas. Calibration of the KIVA-II code was performed in which the wall heat loss was adjusted (by increasing the heat loss by about 2.5) such that the computed results in the absence of combustion matches with experiment (motored engine) as shown in Fig. 6.3, i.e. to isolate the effect of combustion. Computational results discussed in Chapters 7 through 9 were performed with the calibrated KIVA-II code. 119 6.5.1 Grid Sensitivity Errors in computational fluid flow arise from the modeling of the physical process and the approximations in the numerical discretization of the equations. Thus, grid dependency test is necessary on which numerical computation is performed. However, a compromise should be reached to ensure reasonable computing time and memory. For the case of the grid size shown in Fig. 6.2, (i.e. 4851 total grid points) the computation time was about 8.5 hours on a HP 715-75 computer. In view of possible grid dependency, grid sensitivity on the simulated results was performed by increasing the grid density by 25% at BOI of 4 deg. BTDC. The ignition delay was set as a criteria to determine the effect of grid size effect for 100% diesel and for diesel-gas combustion . Figure 6.4 shows the computation results are grid-dependent because the ignition delay increases by 60% when the grid density is increased by 25%. The KIVA-II results reported in the thesis are therefore highly grid-dependent This indicates that with the present grid size (limited by a time and computer memory), KTVA-LI can not be considered to have an absolute predictive power. However, at the given engine operating conditions and grid size, some parameters (e.g. reaction rate constants) can be adjusted to match the experimental results. 120 Fig. 6.1: Dual fuel injection geometry Fig. 6.2: Computational grid arrangement 121 -100 -50 0 CA(DEG) 50 100 Fig. 6.3: Calibration of the KTVA-II code 5 -DIESEL100% DSL30% + GAS70% _L i i i i _1_ 5.0000E3 5.2000E3 5.4000E3 5.6000E3 5.8000E3 TOTAL NUMBER OF GRID POINTS Fig. 6.4: Dependence of ignition delay on grid size 122 CHAPTER SEVEN: DATA COMPARISON 7.1 Introduction The main objective of this chapter is to evaluate the KIVA-II code results based on the experimental data and provide ground for detailed KIVA-II results discussed in Chapters 8 and 9. The results discussed in this chapter are: (i) the cylinder pressure; (ii) mass-burned fraction; (iii) ignition delay; (iv) combustion duration, and (v) NOx emissions. The data taken for comparison are the test cases A and B shown in Table 7.1 in which case A pertains to 100% diesel and case B is for 30% pilot diesel and 70% natural gas. The results predicted by the KIVA-II code should not be treated as absolute considering the limitations of the simplified chemical kinetics (Section 5.7) used in the combustion modeling. On the other hand, the unknown effect of chemical and physical interactions of the pilot diesel and natural gas (particularly during the pre-ignition period) contribute to the uncertainties associated with combustion modeling. Therefore, the predicted results reported in this thesis should be considered as approximate for the reasons stated above. In this chapter, three types of results are discussed: (i) the experimental results which were obtained directly from measurement; (ii) the computed results in which the measured cylinder pressure data were analyzed using the multi-zone (XPNOX model described in Appendix C), and (iii) the simulated results which were predicted by KIVA-II numerical simulation. 123 Table 7.1: Experimental and simulation test conditions CASE SPEED BOI a PW LOAD DSL: GAS 4> (RPM) (DEG) (DEG) (DEG) (BMEP) RATIO (DEG) CASE A 1250 0* 3 10 3 100% : 0% -CASE B 1250 0* 3 14 3 30% : 70% 30 6* the injection tuning setting at 4,8,12 and 16 deg. BTDC 7.2 The Cylinder Pressure The measured and KIVA-II simulated cylinder pressure results for cases A and B are shown in Figs. 7.1a-7.1d for BOI of 4, 8,12 and 16 deg. BTDC. A reasonable agreement of the experimental and KIVA-II results is observed, particularly for the BOI of 4 BTDC in which the KIVA-II peak cylinder pressure is about 2% higher than for the corresponding experiment. Similar comparison of results for case B are shown in Figs. 7.2a-7.2b. At BOI of 4 deg. BTDC, KTVA-II predicts the cylinder pressure is about 6% higher than the corresponding experiment. Generally, at a given injection timing, the measured and KIVA-II simulated cylinder pressure (cases A and B) are in a reasonable agreement in terms of peak pressures and the "shape" of the pressure profile. The predicted cylinder pressure profile for case B (Figs. 7.4a-7.4d) are higher than the corresponding measurements because of the limitations of the single-step chemical reaction modeling (used in the KIVA-II code) which tends to over-predict the rate of 124 heat release. This is consistent with Sloane [1992] who pointed out that, the correlations for methane combustion suggested by Westbrook [1981] over-predicts the rate of heat release for T>1700 K. The results reported in this thesis are based on the Westbrook's correlations (Chapter 5 and 6) in which the ignition of natural gas (by pilot diesel) is in the high temperature range. 7.3 The Mass Burned Fraction The results of mass-burned fraction for case A calculated by XPNOX and KIVA-II are shown in Figs. 7.3a-7.3d. The XPNOX and KIVA-II results agree closely at BOI of 4 deg. BTDC with wider deviation at BOI of 16 deg. BTDC. Similar comparison of results for case B are shown in Figs. 7.4a-7.4d. For the BOI of 4, 8 and 12, the KIVA-II results predict higher burning rate than the corresponding XPNOX results. The high rate of diesel-gas combustion (Figs. 7.5a-7.5b) is consistent with the cylinder pressure results discussed above. The limitation of combustion model results in high burning rate. However, at the BOI of 16, the diesel-gas combustion rate is much slow such that it results in incomplete combustion due to significantly long ignition delay for advanced injection timing (discussed below). The ignition delay and combustion duration results for cases A and B discussed in the next section were respectively derived from the cumulative mass-burned shown in Figs. 7.3a-7.3b and 7.4a-7.4b. 125 7.4 Ignition Delay The ignition delay (0-5% mass burned) results for cases A and B calculated by X P N O X and KIVA-II code are shown in Figs. 7.5a and 7.5b. The KIVA-II results predict slightly longer ignition delay for both cases. Say, at BOI of 4 deg. BTDC, the ignition delays calculated by X P N O X for cases A and B are about 20% lower than the corresponding KIVA-II predictions. It appears that the ignition delay for case A is not sensitive to the injection timing while that of case B increases with advancing injection timing. This could be explained by the fact that the average cylinder temperature (from KIVA-II) does not change appreciably during compression from BOI of 16 to 4 deg. BTDC which were 800 K and 850 K, respectively. This temperature range is within the ignition temperature of diesel fuel, so that the ignition delay of 100% diesel becomes independent of injection timing (within this range). However, with diesel-gas fueling (case B), the ignition delay is dependent on injection timing because: (i) the injection of cold ( about 350 K throughout the range of injection timing ) natural gas jet in the combustion chamber lowers the temperature resulting in longer ignition delay, and (ii) the pre-ignition process may produces active species (from diesel and natural gas) which influence the ignition delay of the pilot diesel. The large discrepancy between XPNOX and KIVA-II could be explained by: (i) uncertainties involved in approximations of XPNOX and KIVA-II simulation, and (ii) incomplete combustion for BOI of 16 deg. BTDC (both for XPNOX and KIVA-II in Fig. 7.4a) contributing to uncertainties in the results. 126 7.5 Combustion Duration Comparison of the combustion duration results for cases A and B as calculated by XPNOX and KIVA-II are shown Figs. 7.6a and 7.6b. For cases A and B at BOI of 4 deg. BTDC, the combustion durations calculated by KTVA-II are respectively 10% and 20% shorter than the corresponding XPNOX calculations. The higher burning rate for diesel-fuel compared to 100% diesel is consistent with the results reported by Sloane [1992] who demonstrated that the correlations for methane combustion suggested by Wesstbrook [1981] over-predict the burning rate for T > 1700 K. 7.6 Burned Gas Temperature (Flame Temperature) The flame temperature for cases A and B calculated by the XPNOX and KTVA-II are compared as shown in Figs. 7.7a-7.7d and Figs. 7.8a-7.8d. The flame temperature predicted by KTVA-II was evaluated as the highest temperature in the local computational cell as a function of crank angle. This may imply that there are only few zones with high temperature compared with the XPNOX results in which combustion is assumed to occur within a stoichiometric mixture. This will be shown to be the case (in Section 7.8) by determining the NOx level both from the XPNOX and KTVA-II with similar empirical constants for NOx formation using Zeldovich mechanism. Here, the assumption is that only temperature is responsible for NOx formation. 127 Generally, the results indicate that the flame temperatures for case A are higher or similar to that of case B. Say, at BOI of 4 deg. BTDC, the peak flame temperatures simulated by KIVA-II are about 5% (case A) and 3% (case B) higher than corresponding XPNOX calculations. At a given BOI, the XPNOX results (Figs. 7.7a-7.7d and 7.8a-7.8d) show high temperature values in the absence of combustion which is not realistic for reasons explained in the next paragraph. However, during the main combustion phase the XPNOX and KIVA-II results are similar. The discrepancies in flame temperatures (Figs. 7.7a-7.7d and 7.8a-7.8b) during the ignition delay as well as in the later part of combustion duration (i.e. beyond 40 deg. ATDC) could be explained as follows: (i) the XPNOX assumes stoichiometric combustion, leading to a significantly high flame temperature even if an infinitesimal mass of fuel is burned during the ignition delay (perhaps lean mixture) or in the later part of combustion period; (ii) combustion in the KIVA-II code takes place within the local flammable mixture which does not necessarily result in high flame temperature, and (iii) the mixing effect (k-s model in KIVA-II) of the burned and unburned gases which results in significantly lower temperature during ignition and late in the combustion cycle. 7.7 Unburned Gas Temperature The unburned gas temperature for cases A and B computed by XPNOX and KIVA-II are shown in Figs. 7.9a-7.9d and Figs. 7.10a-7.10d. At TDC, the unburned gas 128 temperature in all cases is about 850 K. The results agree closely for case A than case B. Say, at BOI of 4 deg. BTDC, the unburned gas temperature predicted by KIVA-II for cases A and B are respectively 4% and 2% less than the corresponding XPNOX calculations. For diesel-gas combustion, the unburned gas temperature computed by XPNOX and by KIVA-II (shown in Figs. 7.10a-7.10d) do not agree closely in the later stage of combustion. For 100% diesel the temperatures are in good agreement as shown in Figs. 7.9a-7.9d. The explanation is that, unlike XPNOX, the KIVA-II code is capable of simulating mixing process in the combustion chamber in which the hot combustion gases raise the temperature of the cold gases through mixing. 7.8 The NOx Emissions Figures 7.11a-7.11d show the NOx formation calculated by XPNOX and KIVA-II for cases A and B. It is observed that both XPNOX and KIVA-II under-predict the NOx emissions compared to the measured results for DDC 1-71 engine (shown in Fig. 3.10), but the measured NOx results for the 6V92 and 1-71 engines are not necessarily the same even with similar operating conditions. It should be pointed out that all the XPNOX and KIVA-II results reported in this thesis are based on 6V92 engine. The NOx results shown in Figs. 7.11a-7.11d indicate that KIVA-II predictions are consistently lower than those of XPNOX. This is consistent with lower flame temperature from 129 KIVA-II predictions compared to XPNOX (discussed in Section 7.6). The discrepancy between KIVA-II and XPNOX calls for research work. 7.9 Summary This section gives a brief summary of the results discussed in this chapter. 7.9.1 Cylinder Pressure The measured and the simulated cylinder pressure show: (i) reasonable agreement within 6% error with highest deviation for case B at BOI of 8 and 16 deg. BTDC; (ii) the measured and simulated cylinder pressure increase with advancing injection timing. 7.9.2 Ignition Delay and Combustion Duration The ignition delays which were derived from the mass burned fraction indicated that: (i) the ignition delays computed by XPNOX and KIVA-II are of the same order of magnitude. The ignition delay for case B was found to increase with advancing injection timing; 130 (ii) the ignition delays for cases A and B computed by XPNOX were about 20% lower than the KIVA-II predictions. (iii) the combustion durations computed by XPNOX and KIVA-II are generally insensitive to injection timing; (iv) the combustion durations predicted by KIVA-II for cases A and B were about 10% and 20%, respectively, shorter than the XPNOX calculations. 7.9.3 Combustion Chamber Temperature and NOx Emissions Some features depicted by the cylinder temperature and NOx emissions are: (i) the flame temperatures for case A were generally higher than the corresponding case B. For cases A and B, KIVA-II results were about 5% and 3%, respectively, higher than XPNOX results; (ii) the unburned gas temperatures were of the order of 850 K which is much lower than the ignition temperature of natural gas (1200-1300 K); (iii) the NOx emissions calculated by XPNOX and KIVA-II are lower that those measured in DDC 1-71 engine; (iv) the predicted NOx emissions by KIVA-II are generally lower than the corresponding XPNOX results. For case A, the XPNOX and KIVA-II under-predict by about 1% and 20%, respectively. Similarly, for case B, the XPNOX and KIVA-II under-predict by 30% and 60%, respectively. 131 From the comparison of the results in this chapter, it appears that the KIVA-II code is capable of predicting the features and trend of the experimental results and those inferred from experiment using the XPNOX code. Thus, KIVA-II can be used as a tool for detailed analysis for design changes to investigate the possible consequence of the engine operating conditions. 132 Fig. 7.1c: Cylinder pressure (case A at BOI=12 deg) Fig. 7.1d: Cylinder pressure (case A at BOI=16 deg) 40 £ J *\ J "\ J '\ S 30 // s \ • V \\ UJ •'/ Q_ rr 20 UJ z _ J o MEASURED (DSL30% + GAS70%) 10 KIVA (DSL30% + GAS70%) (BOM DEG BTDC) . . i . . . . i . . . . i . . . . i . . . . i . . . . i . . -20 -10 0 10 20 30 40 50 60 CA(DEG) Fig. 7.2a: Cylinder pressure (case A at BOI=4 deg) LO I- • . . . i . . . . i . . . . i • . . . i • . . . i • . . . i . . . . i . . . . i -20 -10 0 10 20 30 40 SO 60 CA(DEG) i . . . . i . . . . i i . . . . i . . . . i . . . . i . . . . i -20 -10 0 10 20 30 40 50 60 CA(DEG) Fig. 7.2b: Cylinder pressure (case A at BOI=8 deg) • * • • • • - • - - • • • • • - - ' - • - • ' - - • - • - - - - ' - • • - « • - - - ' -20 -10 0 10 20 30 40 50 60 CA (DEG) Fig. 7.2c: Cylinder pressure (case A at BOI=12 deg) Fi8- 7 2 d : Cylinder pressure (case A at BOI=16 deg) SET Fig. 7.4a: Mass-burned fraction (case B at BOI=4 deg) Fig. 7.4b: Mass-burned fraction (case B at BOI=8 deg) u> (Ti CA(DEQ) Fig. 7.4c: Mass-burned fraction (case B at BOI=12 deg) Fig. 7.4d: Mass-burned fraction (case B at BOI=16 deg) 100% DIESEL 14 r -12 -10 -8 INJECTION TIMING (DEG CA) -6 Fig. 7.5a: Ignition delay for cases A 14 O 4 DSL30% + GAS70% _ XPNOX -+—KIVA 0 • ' ' ' -16 I L _ l _ -14 -12 -10 -8 INJECTION TIMING (DEG CA) Fig. 7.5b: Ignition delay for cases B 137 DIESEL100% 50 r -16 -14 -12 -10 -8 -6 -4 INJECTION TIMING (DEG CA) Fig. 7.6a: Combustion duration for cases A DSL30% + GAS70% 138 g 2000 ui cc z> H ui 1500 a. s 1000 DIESEL100% 10 . XPNOX . KIVA (801-4 DEG BTDC) 30 40 CA (DEG) 50 Fig. 7.7a: Flame temperature (case A at BOI=4) DIESEL 100% 2500 -— 2000 I 2 3 1000 20 XPNOX KIVA (BOI-8 DEG BTDC) 30 40 CA (DEG) 50 Fig. 7.7b: Flame temperature (case A at BOI=8) DIESEL 100% ui cc 3 2000 1500 1000 500 -10 -XPNOX - KIVA (BOM 2 DEG BTDC) 10 20 30 40 CA (DEG) 50 60 2500 i . 2000 cc Z3 a. 2 ui s 3 DIESEL 100% . XPNOX -KIVA (BOI-16 DEG BTDC) -10 10 20 30 40 CA (DEG) 60 Fig. 7.7c: Flame temperature (case A at BOI=12) Fig. 7.7d: Flame temperature (case A at BOI=16) DSL30% + GAS70% 2500 2 2000 UJ 1500 2 1000 500 - XPNOX - KWA (BOM DEG BTDC) 10 20 30 40 CA (DEG) 50 60 Fig. 7.8a: Flame temperature (case B at BOI=4) DSL30% + GAS70% Z> ' 1500 3 iooo 500 - XPNOX . KiVA (BOI-8 DEG BTDC) 20 30 40 CA(DEG) 60 Fig. 7.8b: Flame temperature (case B at BOI=8) o DSL30% + GAS70% 2500 * 2000 1500 5 iooo XPNOX KIVA (BOM 2 DEG BTDC) •10 30 40 CA (DEG) DSL30% + GAS70% £ 2000 -2 3 1000 - XPNOX - KlVA (BOI-16 DEG BTDC) 20 30 40 CA(DEG) 60 Fig. 7.8c: Flame temperature (case B at BOI=12) Fig. 7.8d: Flame temperature (case B at BOI=16) DIESEL100% Q- 700 600 500 -XPNOX . KIVA ( B O M DEG BTDC) 10 30 40 CA (DEG) 60 DIESEL100% 400 (BOI-8 DEG BTDC) 10 20 30 40 CA (DEG) 60 Fig. 7.9a: Unburned gas temp, (case A at BOI=4 deg) Fig. 7.9b: Unburned gas temp, (case A at BOI=8 deg) i-1 DIESEL100% 0_ 700 s U l O 600 O U l z cc Z> 400 h . XPNOX KIVA (BOI-12 DEG BTDC) • i I i i i i I i i i 10 30 40 CA(DEG) DIESEL100% 900 -800 JNBURNED GAS TEMP JNBURNED GAS TEMP 500 XPNOX KIVA (BOI-16 DEG BTDC) 400 . . 1 . . . . 1 -10 0 10 20 30 40 CA (DEG) 50 60 Fig. 7.9c: Unburned gas temp, (case A at BOI=12 deg) Fig. 7.9d: Unburned gas temp, (case A at BOI=16 deg) DSL30% + GAS70% 400 . XPNOX . KIVA (BOM DEG BTDC) 20 30 40 CA (DEG) 60 DSL30%+GAS70% - XPNOX . KIVA (BOI-8 DEG BTDC) i 20 30 40 CA (DEG) Fig. 7.10a: Unburned gas temp, (case B at BOI=4 deg) Fig. 7.10b: Unburned gas temp, (case B at BOI=8 deg) |4> CO OSL30% + GAS70% 800 & - 7 0 0 z UNBURNED GAS TE 1 i XPNOX KIVA 400 (BOI-12 DEG BTDC) i . . . . i . . . . i . . . . 1 -10 0 10 20 30 40 CA (DEG) 50 60 DSL30% + GAS70% 800 z CO < o a Ul z CC 13 500 - XPNOX - KIVA -10 10 20 30 40 CA (DEG) 50 60 Fig. 7.10c: Unburned gas temp, (case B at BOI=12 deg) Fig. 7.10d: Unburned gas temp, (case B at BOI=16 deg) 100% DIESEL-Fig. 7.11a: Emissions of NOx (case A at BOI=4 deg) lO U) . : •__ 100% DIESEL 250 = 200 (L '- 1 in EMISSIOf EMISSIOf / 6 100 z XPNOX (DIESEL 100%) _ _ KIVA (DIESEL 100%) SO ' 1 1 (BOI-12 DEG BTDC) 0 J 1 i . . i i 0 2 0 CA (DEG) 4 0 60 100% DIESEL 250 -200 -S a m 1 5 0 ; EMISSION o / O z 50 / ' XPNOX (DIESEL 100%) KIVA (DIESEL 100%) (BOI-8 DEG BTDC) J •'. i . . . . i . . . . i 0 2 0 CA (DEG) 4 0 6 0 Fig. 7.11b: Emissions of NOx (case A at BOI=8 deg) 100% DIESEL 300 Ox EMISSIONS (PPM) 1 z 100 / XPNOX (DIESEL 100%) KIVA (DIESEL 100%) (BOI-16 DEG BTDC) o i . . . . i . . . i 0 2 0 CA (DEG) 4 0 60 Fig. 7.11c: Emissions of NOx (case A at BOI=12 deg) Fig. 7.11d: Emissions of NOx (case A at BOI=16 deg) DSL30% + GAS70% 100 NOx EMISSIONS (PPM) 1 / XPNOX (DSL30% + GAS70%) IS - - - KlVA(DSL30% + GAS70%) / / (BOI-4 DEG BTDC) n . i — < ^ ' . . i . . . . i . . . . i . . . . i . . . . i V 0 10 20 30 40 50 60 CA (DEG) Fig. 7.12a: Emissions of NOx (case B at BOI=4 deg) DSL30% + GAS70% 200 NOx EMISSIONS (PPM) •8 § 8 J •' XPNOX (DSL30% + GAS70%) / KIVA (DSL30% + GAS70%) / (BOH12 DEG BTDC) o / i i . . . . . i . . . . i . . . . i . . . . i . . . . i . . . . i •10 0 10 20 30 40 50 60 CA (DEG) Fig. 7.12c: Emissions of NOx (case B at BOI=12 deg) DSL30% + GAS70% Fig. 7.12b: Emissions of NOx (case B at BOI=8 deg) OSL30% + GAS70% 300 -(PPM) CO z y \ NOxEMISSIC 1 XPNOX (DSL30% + GAS70%) KIVA (DSL30% + GAS70%) (BOI-16 DEG BTDC) 0 i . i . f . . i 0 2 0 CA (DEG) 4 0 6 0 Fig. 7.12d: Emissions of NOx (case B at BOI=16 deg) CHAPTER EIGHT - DISCUSSIONS OF SIMULATION RESULTS 8.1 Introduction This purpose of this chapter is to examine the structure of the flow and thermal field, as represented by IQVA-II corresponding to the experiments for cases A and B. Test conditions for the results discussed in this chapter are shown in Table 8.1. Criteria for the choice of the test conditions are discussed in Section 4.1. Table 8.1: KTVA-II simulation conditions C A S E SPEED (RPM) BOI (DEG) (DEG) PW (DEG) L O A D (BMEP) D S L : G A S RATIO 4> (DEG) C A S E A 1250 0* 3 10 3 100% : 0% -C A S E B 1250 8* 3 14 3 30% : 70% 30 0* the injection timing setting at 4,8,12 and 16 deg. BTDC 8.2 Fuel Injection and Combustion Description of the combustion chamber processes during injection and combustion are illustrated by viewing half the cross section area of the combustion chamber as shown in Fig. 8.1 for the computational grid arrangement Three views are defined as bottom, middle and top. The views were taken at different combustion chamber depths to help provide details of the contour plot along the depth of the piston bowl. These views are the cross 145 sections where the contour plots are viewed as discussed later in this Chapter. The contour plots correspond to the crank angle positions of 0, 5,10 and 20 degrees ATDC as shown in Fig. 8.1. A typical velocity field at 10 deg. ATDC (viewed at the middle) in the combustion chamber with the engine operating at a swirl ratio of 1.0 is shown in Fig. 8.2. The length of the vector lines indicate higher swirl velocity near the cylinder wall. The spatial and temporal contours lines discussed in this chapter are those of: (i) temperature (K); (ii) NOx (ppm), (iii) turbulence kinetic energy (joules/gm), and (iv) mixing time (ms). The contour lines for: (i) baseline diesel; (ii) pilot-diesel; (iii) natural gas, and (iv) carbon dioxide are expressed as mass fraction (percentage of the total mass) while the equivalence ratio resulting from the pilot diesel and natural gas is given as the ratio of the actual to the theoretical stoichiometric air fuel ratio. The contour lines are defined as the highest (H) and lowest (L) with an increment A. 8.2.1 Jet-Induced Turbulence by Dual Fuel Injection The injection geometry shown in Chapter 6 (Fig. 6.1) indicates that the pilot-diesel spray and natural gas jet are separated by angles (j> and 6. The differential injection velocity of pilot diesel («140m/s) and natural gas («460m/s) suggests a creation of a shear layer resulting into high turbulence level. A question arises as to what extent the interference of pilot diesel and natural gas (during injection and combustion) influences turbulence intensity and mixing rate. This question is addressed by examming the turbulence level and the characteristic mixing time during injection and combustion. 146 Figure 8.3 shows the average turbulence kinetic energy (TKE) for cases A and B. In general it is observed that higher turbulence level is closer to the jet center-line and decreases towards the periphery of the jet During the injection period, the TKE for cases B is about 10 times that of case A. The combined effect of pilot diesel and natural gas (cases B) together with high natural gas injection velocity relative to that of pilot diesel is largely responsible for higher TKE. The spatial and temporal distribution of TKE is shown in Figs. 8.4a and 8.4b for cases A and B. The zones of the high TKE are observed to correspond to the locations of the jet, suggesting a shear layer created by the jet as it penetrates the combustion chamber. 8.2.2 The Mixing Time Scale The air-fuel mixing rate and the subsequent formation of the flammable mixture was evaluated based on the characteristic mixing time. The mixing time Tm, was evaluated as the ratio of the turbulence kinetic energy k to the rate of dissipation of kinetic energy e. The quantities k and e are determined from the turbulence modeling discussed in Appendix F. Figure 8.5 illustrates the mixing time for cases A and B, in which a typical mixing time during injection is about 0.25 ms. The contour lines of constant mixing time are shown in Figs. 8.6a and 8.6b for cases A and B, respectively. The zones of smaller mixing time suggest higher mixing rates; they corresponds with the high turbulence level zones shown in Figs. 8.4a and 8.4b. Although high mixing rate is desirable for better fuel-air mixing, if is in excess, it can lead to excessively lean mixtures as well as enhancing heat loss. 147 8.2.3 The Equivalence Ratio Figure 8.7 illustrates the instantaneous formation of the lean, mean (flammable) and rich mixture in the combustion for case B. The lean, flammable and rich mixtures were defined in their respective range of equivalence ratios, i.e. for lean 4F=0.25 - 0.5, flammable cb=0.5 -2.25 and rich (J>=2.25 - 5.0. The equivalence ratios were computed by KIVA-II with the resultant equivalence ratio defined by eqn. 5.14 (in Chapter 5). Such equivalence ratios were normalized (for the sake of plotting) by 0.5, 2.25 and 5.0, respectively. The mean (flammable) mixture is minimum at about 8.5 deg. ATDC which corresponds to the ignition of the natural gas resulting in the consumption of the flammable mixture. The distributions of combustible mixture in Figs. 8.8a and 8.8b show the contour lines of constant equivalence ratio of a combustible mixture (i.e. cj) = 0.25 - 2.25) in the combustion chamber for cases A and B. The spatial and temporal distribution of equivalence ratio for case B is more uniformly distributed than for case A. 8.2.4 The Diesel Vapour and Natural Gas in the Chamber The total fuel injected and its rate of consumption is shown in Figs. 8.9a and 8.9b for cases A and B. The solid lines in Fig. 8.9a show injected diesel increases linearly to a maximum of 30 mg at 10 deg. after BOI. The dashed line illustrates the instantaneous diesel vapour in the chamber in which a sudden decrease of diesel vapour concentration at about 10 deg. corresponds with the premixed combustion phase. The increased amount of diesel vapour 148 between 10 and 15 deg. ATDC is due to vapourization of the liquid diesel by the hot gases (resulted premixed burning phase) and a slower rate of burning during the diffusive-combustion phase. Figure 8.9b shows: (i) the total injected pilot diesel liquid (9 mg) and natural gas (21 mg), and (ii) the instantaneous pilot diesel vapour and the natural gas. The pilot diesel is shown to have all been consumed by 10 deg. ATDC. The main combustion phase of the natural gas seems to commence 10 deg. ATDC in which the natural gas concentration falls exponentially with crank angle as the combustion proceeds. The unburned diesel vapour (case A) in the combustion chamber is shown by the contour plots in Fig. 8.10a as a percentage of mass fraction. The diesel vapour increases rapidly from maximum of 0.3% at TDC to 19.5% at 10 deg. ATDC, which is consistent with Fig. 8.9a. The effect of swirl is noticeable at 10 deg. ATDC in which the diesel vapour is "turned" in the clockwise direction (see the direction of swirl velocity in Fig. 8.2). The pilot diesel fuel vapour for case B is shown by the contour lines in Fig. 8.10b. Beyond 5 deg. ATDC, the spatial spreading and penetration of the pilot diesel is higher for case B than for case A because the pilot diesel is entrained in the high velocity natural gas jet The effect of swirl is higher because of the lower momentum of pilot diesel in case B compared to case A. Although the momentum rate for diesel 100% is lower than for natural gas jet (discussed in Section 6.4.2) it does not necessarily mean that increasing the diesel injection velocity (for higher momentum rate) will result in higher penetration for the reasons explained in Section 2.2.1, i.e. diesel injection velocity beyond 200 m/s retards penetration. 149 The unburned natural gas (case B) is shown by the contour lines in Figs. 8.11. At a given crank angle position, natural gas spreading and penetration higher for case B compared to 100% diesel (case A) shown in Fig. 8.10a. The natural gas concentration is negligible at 0 deg. ATDC, but by 5 and 20 deg. ATDC, jet penetration in the chamber radius is about 50% and 100%, respectively. 8.2.5 Ignition Delay and Combustion Duration Figure 8.12a shows the C O 2 from the combustion of the diesel for case A. The normalizing factor is the total quantity of the C O 2 (i.e. 98.5 mg) produced by burning 30 mg of diesel fuel. Distinction was made between the C O 2 produced by the pilot diesel and the natural gas as shown in Fig. 8.12b. The normalizing factor (81.0 mg) is the total C O 2 which is the sum of C O 2 produced of pilot diesel and natural gas combustion. The total C O 2 in Figs. 8.12a and 8.12b represent the corresponding fuel mass-burned fraction. The change in slope of total C O 2 in Fig. 8.12b at about 9 deg. ATDC indicates the transitional point between the end of combustion of the pilot diesel and the ignition of the natural gas. It is observed that the combustion of the natural gas starts just when the pilot diesel combustion finishes at about 10 deg. ATDC. The spatial and temporal location and quantity of the combustion product (i.e. CO2) for case A is shown in Fig. 8.13. The C O 2 produced from the combustion of the pilot diesel and natural gas (case B) are shown in Figs. 8.14a and 8.14b, respectively. For case A at 20 deg. crank angle (with already 50% mass burned), a large quantity of C O 2 is concentrated about 150 half way of the chamber radius, suggesting that significant quantity of diesel burns in the piston bowl. In contrast, large quantities of C O 2 for case B are found in the squish region where combustion is mainly taking place. The higher flame penetration for case B than A is partly due to higher momentum rate for case B (discussed in Section 6.4.2). The spatial location of the zones of high concentration of CO2 from burning pilot diesel (Fig. 8.14a) corresponds to those from burning natural gas (Fig 8.14b), mdicating a single flame resulting from a pilot diesel combustion. Thus, natural gas is ignited by heat diffusion from the burning pilot diesel rather than due to the compression temperature resulting from expanding gases due to the pilot diesel combustion. 8.Z6 The Cylinder Temperature The maximum local cylinder temperature for cases A and B are shown in Fig. 8.15 in which case A results in the highest temperature. This was expected because the diesel flame temperature is higher than for the natural gas. After ignition delay for cases A and B, i.e. about 8.5 deg. after BOI, the peak cylinder temperature rises sharply to above 2500 K. The same phenomena can be observed by temperature distribution in the combustion chamber for cases A and B as shown in Figs. 8.16a and 8.16b for the deg. CA positions of 0 to 20 deg. ATDC. General features of the contour lines of constant temperature are: (i) the overall temperature is highest for case A than case B; (ii) at a given crank angle position the flame penetration is about 15% higher for cases B than A, and (ii) the temperature gradients for cases B are about 3% lower than case A. 151 8.2.7 The NOx Formation The rate of NOx formation for cases A and B are shown in Fig. 8.17. It is seen that the maximimi NOx level from simulation results is lower than the corresponding XPNOX and experimental for DDC 1-71 shown Fig. 3.10 and Figs. 4.13a-4.13d, respectively. However, both experiment and simulation results show lower NOx level for diesel-gas combustion compared to the baseline diesel; this calls for more research. The zones of formation of NOx in the combustion chamber are shown in Figs. 8.18a and 8.18b for cases A and B. In all cases (A and B ), the NOx formation starts effectively at 10 deg. ATDC and corresponds to the high temperature zones shown in Figs. 8.16a and 8.16b. 8.4 Summary The following summary pertains to the discussions of the KIVA-II simulation results discussed in this chapter. 8.4.1 Jet-Induced Turbulence and Mixing Time (i) The jet-induced turbulence kinetic energy (TKE) is highest within the jet location; (ii) The TKE induced by natural gas jet is about 10 times that of the corresponding 100% diesel injection; (iii) The mixing time is lowest during the injection period, i.e. corresponding to the highest TKE; 152 (iv) During the injection period, the turbulence mixing time for pilot diesel natural gas is about 5 times that of the 100% diesel; (v) The contour plots for case B show low mixing time (i.e. high mixing rate) in the squish than for case B which is mainly takes place in the piston bowl. 8.4.2 The Equivalence Ratio (i) The spatial and temporal "penetration" of combustible mixture (i.e. the contours of 4>= 0.5 - 2.25) is higher for case B than case A; (ii) The flammable mixture for case B is more imiformly distributed with smaller gradients of equivalence ratio compared to case A. 8.4.3 The Diesel Vapour and Natural Gas in the Chamber (i) The spatial and temporal penetration of unburned fuel is higher for case B than case A; (ii) Unburned diesel vapour and natural gas (case B) result to smaller gradients of the than for case A. 8.4.4 Ignition Delay and Combustion Duration (i) At a given crank angle, say 10 deg. ATDC, the concentrations of CO2 is located in the squish region, and in the piston bowl for case B and A, respectively. Such a 153 phenomena indicates that combustion for case A takes place mainly in the piston bowl while for case B it takes place in the squish region; (ii) The spatial location of zones of concentration of CO2 from pilot diesel and natural gas (case B) are the same mdicating a single flame which propagates from the pilot diesel ignition to the natural gas; (iii) The rate of formation of C O 2 is higher for case A than case B as a result of higher burning rate for case A. 8.4.5 Cylinder Temperature and NOx Emissions (i) The maximum chamber temperature is higher for case A than case B; (ii) The chamber temperature for case B is uniformly distributed with smaller gradients of about 3% compared to case A; (iii) At a given crank angle position, the flame penetration is about 15% higher for case B than case A. (iv) For a BOI of 4 deg. BTDC, the formation of NOx starts effectively at 10 deg. ATDC; (v) The contour plots depict that the zones of NOx corresponds to those of high temperature. 154 10 Fig. 8.1: Combustion chamber cross section area at 0, 5,10 and 20 degrees ATDC 155 156 4 O 5? UJ ^ 2 O UJ ,DIESEL100% . DSL30% + GAS70% ( B O M DEG BTDC) 0 F 1 1 1 1 i 1 1 ' 1 1 1 • ' • 1 -10 10 20 30 CA (DEG) 40 50 60 Fig. 8.3: Turbulence kinetic energy in the combustion chamber 157 BOTTOM L=0 H=0.02 L=0 H=0.04 L=0 H=0.06 L=0 H=0.02 MIDDLE L=0.02 H=0.2 L=0.02 H=0.2 L=0 H=0.08 L=0 H=0.02 TOP L=0.05 H=0.4 L=0.04 H=0.3 L=0 H=0.07 CA (ATDC) 10 L=0 H=0.02 20 Fig. 8.4a: Contours of constant turbulent kinetic energy in joules/gm for case A (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 158 BOTTOM MIDDLE TOP CA (ATDC) H=0.02 L=0 H=0.04 L=0.01 H=0.08 L=0 H=0.03 L=0.02 H=0.16 L=0.03 H-0.3 L=0.05 H=0.47 L-0.01 H=0.06 L=0.25 H=2.2 L=0.42 H=3.8 L=0.02 'H-1.8 L=0.01 H=0.07 Fig. 8.4b: Contours of constant turbulence kinetic energy in joules/gm for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 159 Fig. 8.5: Characteristic turbulent mixing time in the combustion chamber 1 6 0 BOTTOM MIDDLE Fig. 8.6a: Contours of constant turbulence mixing time in ms for case A (H-Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 161 BOTTOM MIDDLE TOP CA (ATDC) L=1.1 H=4.2 L=1.0 H=4.5 L=0.7 H=3.1 L=1.1 H=2.7 L=0.6 H=4.1 L=0.6 H=4.1 L=0.5 H=2.3 L=0.8 H=2.4 L=0.4 H=3.1 L=0.4 L=0.4 H=3.2 H=2.9 L=0.6 H=2.5 10 20 Fig. 8.6b: Contours of constant turbulence mixing time in ms for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 162 (DSL30% + GAS70%) Fig. 8.7: Resultant equivalence ratio in the combustion chamber 163 BOTTOM MIDDLE TOP CA (ATDC) Fig. 8.8a: Contours of constant overall equivalence ratio for case A (H=2.25 and L=0.25) in all cases. The cross sectional views are shown in Fig. 8.1 164 CA BOTTOM ' MIDDLE TOP (ATDC) 0 5 10 20 Fig. 8.8b: Contours of constant overall equivalence ratio for case B (H=2.25 and L=0.25) in all cases. The cross sectional views are shown in Fig. 8.1 165 (DIESEL.100%) s -J UJ I \ DIESEL (Injeted) DIESEL (Instantaneous) (BOM DEG BTDC) -i L 1 0 CA (DEG) 2 0 30 40 Fig. 8.9a: Total injected fuel and vapour in the combustion chamber for case A (DSL30% + GAS70%) 20 -15 O 2 UJ E '10 .DSL (Injected) GAS (Injected) DSL (Instantaneous) GAS (Instantaneous) (BOM DEG BTDC) • 1 0 CA (DEG) 2 0 30 40 Fig. 8.9b: Total injected fuel and vapour in the combustion chamber for case B 166 BOTTOM • MIDDLE TOP CA (ATDC) L=0.0 H=0.2 L=1.6 H=14.4 L=1.9 H=16.9 L=0.0 H=0.3 L=2.2 H=19.5 L=2.1 •H=19.1 L=0.0 H=0.3 U2.2 H=19.5 L=2.2 H=19.5 Fig 8.10a: Contours of constant local mass fraction of diesel as % of the total mass for case A (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 167 CA BOTTOM MIDDLE TOP (ATDC) 0 L=0.0 H=0.2 L=0.0 . H=0.3 L=0.0 H=0.3 5 L=0.0 H=0.4 L=0.2 H=1.4 L=0.2 H=1.9 10 L=0.4 H=3.4 L=1.1 H=9.6 L=1.8 H=16.5 20 L=0.4 H=2.7 L=0.7 H=5.1 L=0.7 Fig. 8.10b: Contours of constant local mass fraction of pilot diesel as % of the total mass for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 168 BOTTOM Fig. 8.11: Contours of constant local mass fraction of natural gas as % of the total mass for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 169 1.0 r 0.8 0.6 O O —i O >-00 Q UJ N < 0.4 CC o z C M 8 0.2 0.0 (DSL30% + GAS70%) / / / / / / / /' / _ -! / /" / / , r — ( i C0 2 (From pilot DSL) / C0 2 (From GAS) / C02(Total-81mg) (BOI-4 DEG BTDC) ' —f *r . . i i i 1_ 1 1 1 1 1 0 2 0 CA (DEG) 4 0 6 0 Fig. 8.12b: Burning rate (CO2 quantity) for case B 170 BOTTOM Fig. 8.13: Contours of constant local mass fraction of C O 2 (from diesel combustion ) as a % of the total mass for case A (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 171 BOTTOM L=0.0 H=0.2 L=0.0 H=0.4 L=0.4 H=3.4 L=0.4 H=2.7 MIDDLE L=0.0 H=0.3 L=0.2 H=1.4 L=1.1 H=9.6 L=0.7 H=5.1 TOP L=0.0 H=0.3 L=0.2 H=1.9 L=1.8 H=16.5 L=0.7 CA (ATDC) 10 20 Fig. 8.14a: Contours of constant local mass fraction of CO2 (from pilot diesel combustion ) as a % of the total mass for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 172 BOTTOM L=0.0 H<0.01 H<0.01 L=0.0 H=0.5 L=0.5 H=4.2 MIDDLE L=0.0 H<0.01 L=0.0 H<0.01 L=0.3 H=2.8 L=1.1 H=9.1 TOP L=0.0 H=0.01 L=0.0 H<0.01 1=0.3 H=2.9 CA (ATDC) L=1.0 H=9.1 Fig. 8.14b: Contours of constant local mass fraction of CO2 (natural gas combustion ) as a % of the total mass for case B (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 173 2500 Sj 2000 LU t -cc UJ Q • 1500 > o —I < ^ 1000 3 500 DIESEL100% DSL30% + GAS70% (BOI-4 DEG BTDC) -20 -10 10 20 30 CA (DEG) 40 50 60 Fig. 8.15: Maximum local cylinder temperature for cases A and B 174 BOTTOM MIDDLE TOP CA (ATDC) L=871 H=906 U705 H=918 L=528 H=870 L=872 H=1046 L=691 H=1153 U597 HU1272 L=1020 H=2263 L=994 H=2595 L=1000 H=2586 L=953 H=2117 L=996 H=2344 L=940 H=2417 Fig. 8.16a: Contours of constant temperature in kelvin for case A (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 175 BOTTOM MIDDLE TOP CA (ATDC) L=872 H=907 L=858 H=916 L=922 H=1303 L=937 H=1674 L=727 H=894 L=808 H=1005 L=1004 H=1993 L=1032 H=2345 L=510 H=865 L=750 H=1050 L=973 H=2382 L=944 H=2374 Fig. 8.16b: Contours of constant temperature in kelvin for case B (H-Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 1 7 6 100 ! . • 80 (PPM) t EMISSIONS i ~ I s I * o z 20 I ' I / I s Is _ • 7 1 1 f 1 1 I niFRFi inno/„ DSL30% + GAS70% (BOM DEG BTDC) 0 / 1 . , i , . , . i . , . . i . , , , i 0 10 20 30 40 50 CA (DEG) 60 Fig. 8.17: Formation of NOx for cases A and B 177 BOTTOM MIDDLE TOP CA (ATDC) L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H<0.1 L=0.0 H<0.1 L=0.0 H<0.1 1=23 H=212 L=98 H=889 L=110 H=980 10 L=38 H=337 L=94 H=850 L=139 H=1258 20 Fig. 8.18a: Contours of Constant NOx Emission in ppm for case A (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 178 BOTTOM MIDDLE TOP CA (ATDC) L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=0.0 L=0.0 H=3.0 L=19.0 H=60.0 L=5.5 H=50.0 L=57.0 H=576.0 L=23.0 H=200.0 L=96.0 H=874.0 10 20 Fig. 8.18b: Contours of Constant NOx Emission in ppm for case B (H=Highest and L=Lowesf). The cross sectional views are shown in Fig. 8.1 179 CHAPTER NINE - PARAMETRIC SIMULATION RESULTS 9.1 Introduction The purpose of this chapter is to examine the theoretical consequences of two specific variable design changes, i.e. the geometrical injection angle and the injection timing of natural gas relative to the pilot diesel. The influence of these variables on the ignition delay, combustion duration and NOx emissions will be investigated. All the test conditions discussed in this work are shown in Table 9.1 for clarity. The test cases C, D and E are defined as follows: (i) case C refers to the pilot diesel spray and natural gas jet overlapping longitudinally (i.e. angle cb=0 in Fig. 6.1); (ii) the simultaneous injection of pilot diesel and natural gas is case D in which the relative injection timing angle Q. of natural gas and pilot diesel is zero, and (iii) the relative injection timing angle Q of 8° is referred to as case E such that the pilot diesel is injected 8 deg. earlier than natural gas. Table 9.1: Parametric test conditions CASE SPEED (RPM) BOI (DEG) n (DEG) PW (DEG) LOAD (BMEP) DSL: GAS (RATIO) (DEG) 8 (DEG) CASE A 1250 4 - 10 3 100% : 0% - 10 CASE A* 1250 4 - 7 1 30% : 0% - 10 CASE B 1250 4 3 14 3 30% : 70% 30 10 CASEC 1250 4 3 14 3 30% : 70% 0 10 CASED 1250 4 0 14 3 30% : 70% 30 10 CASEE 1250 12 8 14 3 30% : 70% 30 10 180 9.2 The Injection Angle The effect of interaction of pilot-diesel spray and NG jet on the formation of combustible fuel-air mixture was studied by varying angle 4> of natural gas injection relative to the pilot-diesel. The simulated results showing the contour plots in the combustion chamber are discussed with similar defixurions as described in Section 8.2. The influence of the injection angle on the following parameters are discussed: (i) jet-induced turbulence by dual fuel injection; (ii) mixing time scale; (iii) equivalence ratio; (iv) pilot diesel vapour and natural gas concentration in the combustion chamber; (v) ignition delay and combustion duration; (vi) cylinder temperature, and (vii) formation of NOx. 9.2.1 Jet-Induced Turbulence by Dual Fuel Injection The injection geometry shown in Fig. 6.1 shows that the pilot-diesel spray and natural gas jet are separated by angles <}> and 6 of 30 and 10 degrees, respectively. A question arises as to what extent the pilot diesel and natural gas interact during injection and combustion influences turbulence intensity and the mixing rate. To address this question, the level of turbulence kinetic energy (TKE) and the characteristic mixing time in the combustion chamber are discussed. Figure 9.1a shows the average TKE for cases A and C. The combined effect of pilot diesel and natural gas (case C) together with high natural gas injection velocity relative to that of 181 pilot diesel is largely responsible for higher TKE. The spatial and temporal distribution of TKE is shown in Fig. 9.1b for case C. The zones of the highest TKE are observed to correspond to the location of the jet as a result of a shear layer created by the jet as it penetrates the combustion chamber. 9.2.2 The Mixing Time Scale The air-fuel mixing rate and the subsequent formation of the flammable mixture were evaluated based on the characteristic mixing time. The mixing time Tm was evaluated as the ratio of the turbulence kinetic energy k to the rate of dissipation of kinetic energy s. The quantities k and z are determined from the turbulence modeling discussed in Appendix F. Figure 9.2a illustrates the mixing time for cases A and C; a typical mixing time during injection period is about 0.25 ms. The contour lines of constant mixing time are shown in Fig. 9.2b for case C. The zones of smaller mixing time (i.e. higher mixing rates) corresponds to zones of high turbulence as shown in Fig. 9.1b. 9.2.3 The Equivalence Ratio The distribution of combustible mixture for case C (shown in Fig. 9.3) depicts the lines of constant equivalence in the combustion chamber for cb=0.25 - 2.25. It is observed that the spatial penetration of the flammable mixture for case C (Fig. 9.3) is higher and more uniformly distributed than case A shown in Fig. 8.8a. 182 9.2.4 The Diesel Vapour and Natural Gas in the Chamber The pilot diesel fuel vapour for case C is shown by the contour hues in Fig. 9.4. Beyond 5 deg. ATDC, the spatial spreading and penetration of the pilot diesel are higher for case C than case A shown in Fig. 8.10a. The unburned natural gas for case C is shown by the contour lines in Fig. 9.5. At a given crank angle position, spatial spreading and penetration is higher than the corresponding case A shown in Fig. 8.10a. The natural gas concentration is negligible at 0 deg. ATDC, and it penetrates rapidly to about 50% and 100% of the chamber radius at 5 and 20 deg. ATDC, respectively. 9.2.5 Ignition Delay and Combustion Duration Figure 9.6a shows the normalized CO2 from the combustion of the pilot diesel and the natural gas for cases C. The normalizing factor of 81.0 mg was the total CO2 produced from the combustion of the pilot diesel and the natural gas for case C. The rate of formation of the C O 2 represents the chemical reaction rate (see Fig. 9.6b). It is observed that the ignition of the natural gas starts at about 10 deg. ATDC when the pilot diesel combustion is about to finish so that the chamber temperature is sufficiently high for natural gas ignition. Figures 9.6a and 9.6b show that the ignition delay of pilot diesel is not influenced by angle (b of natural gas relative to the pilot diesel. At about 20 deg. ATDC (Fig. 9.6b), the combustion rate for cases A, B and C are about the same. However, beyond 20 deg. ATDC, 183 the combustion rate for case C is lower than cases A and B. Comparing results for diesel-gas combustion in Figs. 9.7 and 8.14, it is seen that the quantity of C O 2 (representing combustion rate) is lower for case C. A possible explanation is that higher interaction of pilot diesel and natural gas (case C) result in competition of available air for the fuels to burn leading to rich mixture burriing at a slower rate. The spatial and temporal location and quantity of the combustion product (i.e. CO2) from pilot diesel is shown in Fig. 9.7 for case C. For 100% diesel (Fig. 8.10a) a large quantity of C O 2 is concentrated about half way of the chamber radius at 20 deg. CA with already 50% mass burned. In contrast, large quantities of CO2 for case C are close to the squish region. This indicates that the 100% diesel (under the test conditions investigated) combustion takes place in the piston bowl while diesel-gas, burns near the wall regions. Figure 9.8 shows the contours of C O 2 (which indicates the zones which combustion is most complete) due to natural gas combustion for case C. The zones of high concentration of C O 2 from burning natural gas shown in Fig. 9.8 corresponds to those pilot diesel (Fig. 9.7). This suggests a single flame from the pilot diesel is responsible for the natural gas ignition, in other words, natural gas is ignited by thermal contact (and possibly chemical) with the burned products of the pilot diesel. 184 9.2.6 The Cylinder Temperature The maximum local cylinder temperature for cases A, B and C are shown in Fig. 9.9. Case C results in lower peak temperatures than cases A and B. The higher peak temperature for case A was expected because the diesel flame temperature is higher than for the natural gas. The peak cylinder temperature for case C is lower than B. The reason is due to simultaneous competition of available air by the pilot diesel and natural gas (due to close interaction of pilot diesel and natural gas, at cb=0) resulting in burning a richer mixture (than case B) with lower peak temperature. The spatial and temporal distribution of temperature in the combustion chamber for case C is shown in Fig. 9.10 for CA positions of 0 to 20 deg. ATDC. Comparing results of Figs. 9.10 with those in Figs. 8.16a and 8.16b, the following features are observed: (i) the overall temperature is highest for case A than case C; (ii) smaller temperature gradients for cases C compared to A, and (iii) flame penetration is higher in case C than A. 9.2.7 The NOx Formation The rates of NOx formation for cases A, B and C are shown in Fig. 9.11. The NOx emissions level for case C is lower than for cases A and B. This is consistent with the peak cylinder temperature discussed above. The spatial and temporal formation of NOx in the combustion chamber are shown in Fig. 9.12 for case C. It is observed that the zones of high concentration of NOx corresponds to the zones of high temperature shown in Fig. 9.10. 185 However, the NOx formation commences effectively at 10 deg. ATDC for cases A and B, while for case C it starts later at about 15 deg. ATDC. 9.3 The Injection Timing Experimental results of cylinder temperatures and contour plots from KIVA-II have shown that natural gas is ignited by heat diffusion (and possible chemical action) from the pilot-diesel burning zones. A question arises whether is the ignition delay or the combustion duration of the pilot diesel is critical for ignition of the natural gas. To answer this question, computed results for different injection timings of natural gas relative to the pilot diesel were investigated. These cases were: (i) the pilot diesel and natural gas injected simultaneously (Cl=0), and (ii) natural gas injected significantly later after the pilot diesel injection (0=8),. Computed results for case B in which experimental results are available are presented for comparison. 9.3.1 Simultaneous Injection of Pilot Diesel and Natural Gas Computed results of mass-burned fraction of the pilot diesel and natural gas for a case D are shown in Figs. 9.13. For comparison, similar results for case B are shown in Fig. 9.15. The results in Fig. 9.13 reveal that, neither the pilot diesel nor natural gas burns to completion, implying poor combustion efficiency for simultaneous injection of pilot diesel and natural gas. A test case for late natural gas injection was performed for further evaluation as described below. 186 9.3.2 Late Natural Gas Injection Figure 9.14 shows the mass-burned fraction for the pilot diesel and natural gas for case E where the pilot diesel and NG injection take place at 12 and 4 deg. BTDC, respectively. It is noted that the pilot diesel and natural gas burn to completion (i.e. 5-95% mass-burned). Also the ignition of natural gas takes place just after the completion of pilot diesel ignition 9.4 Ignition Delay and Combustion Duration Ignition delay and combustion duration of the pilot diesel for case B and D (extracted from Figs. 9.13, 9.14 and 9.15) are shown in Table 9.2 for comparison. Comparing the ignition delays, the pilot diesel ignition delay (about 9 deg. CA) for cases D and E are similar to case B in which the pilot diesel precedes the natural gas by 3 deg. CA. On the other hand, the combustion duration of the pilot diesel for case D is very long, i.e. incomplete combustion. This shows that (i) the ignition delay is similar for case B and E, and (ii) the combustion duration for case E is about 20% higher than for case B. This shows that the pilot diesel combustion duration is a critical parameter for stable ignition of natural gas leading to complete combustion. 9.5 Summary This section gives the summary of parametric results discussed in this Chapters. 187 9.5.1 Injection Angle (i) The jet-induced turbulence and the characteristic mixing time for cases B and C were similar; (ii) the ignition delay for cases B and C are similar; (iii) The peak cylinder temperature was lower for case C compared to cases A and B; (iv) The NOx emissions level for case C were about 2/3 that of case B at a BOI of 4 deg. BTDC; (v) The flame penetration was similar for case B and C so that combustion takes place mainly in the squish region. 9.5.2 The Injection Tinting (i) The combustion efficiency (i.e. complete combustion) is critically dependent on the injection timing of the natural gas relative to the pilot diesel injection. Optimal injection timing of natural gas relative to the pilot diesel results in complete combustion of the pilot diesel and natural gas. 188 4 O CO UJ LU -DIESEL100% -DSL30% + GAS70% (BOI-4 DEG BTDC) 1 h 1 1 1 • • • -10 10 20 30 CA (DEG) 40 50 60 Fig. 9.1a: Turbulence kinetic energy in the combustion chamber 189 CA BOTTOM MIDDLE TOP (ATDC) L=0.09 H=0.4 L=0.1 H=0.5 L=0.08 H=0.7 Fig. 9.1b: Contours of constant turbulence kinetic energy in joules/gm for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 190 Fig. 9.2a: Characteristic turbulent mixing time in the combustion chamber 191 BOTTOM MIDDLE TOP CA (ATDC) L-0 H=0.02 L=0.02. H-0.15 L=0.12 H-1.1 5 L=0 H=0.03 L=0.03 H=0.27 L=0.04 H=3.7 L=0.01 H=0.08 L=0.03 H=0.39 L=0.02 H=1.9 20 Fig. 9.2b: Contours of constant turbulent mixing time in ms for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 1 9 2 20 Fig. 9.3: Contours of constant overall equivalence ratio for case C (H-2.25 and L=0.25) in all cases. The cross sectional views are shown in Fig. 8.1 19 3 BOTTOM MIDDLE TOP CA (ATDC) L=0.5 H=4.8 L=0.9 H=8.6 U1.1 H=9.8 \ w- -i S L is \ ' n \ v I L i—i ^ L=0.6 H=2.4 L=1.0 H=4.4 L=0.6 H=5.3 Fig. 9.4: Contours of constant local mass fraction of pilot diesel as % of the total mas; for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 194 BOTTOM MIDDLE TOP CA (ATDC) 0 5 L=0.0 H=0.4 L=0.4 H=2.4 L=0.9 H=8.5 10 L=0.02 H=0.2 L=0.04 H=3.5 L=0.8 H=7.7 20 U0.01 H=0.5 L<0.01 H=0.7 L=0.2 H=1.8 Fig. 9.5: Contours of constant local mass fraction of natural gas as % of the total mass for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 195 (DSL30% + GAS70%, PHI-O) 1.0 8 0 8 < & £ 0.6 m o UJ < 0.4 •X o 2 8 0.2 0.0 . C0 2 (From pilot DSL) • CO s (From GAS) •COj(Total-81mg) (BOI-4 DEG BTDC) 2 0 CA (DEG) 4 0 6 0 Fig. 9.6a: Burning rate (CO2 quantity) for case C 1.0 r CA (DEG) Fig. 9.6b: Mass burned fraction for cases A, B and C 196 BOTTOM L=0.0 H=0.2 MIDDLE L=0.0 H=0.3 TOP UO.O H=0.3 CA (ATDC) L=0.5 H=4.8 L=0.9 H=8.6 L=1.1 H=9.8 1 0 L=0.6 H=2.4 L=1.0 H=44 L=0.6 H=5.3 2 0 Fig. 9.7: Contours of constant local mass fraction of C O 2 (from pilot diesel combustion ) as a % of the total mass for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 197 BOTTOM MIDDLE TOP CA (ATDC) 0 5 10 L=0.2 H=1.4 L=0.4 H=3.5 L=0.5 H=4.3 L=0.7 H=5.8 L=09 H=8.0 Fig. 9.8: Contours of constant local mass fraction of C O 2 (natural gas combustion) as a % of the total mass for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 198 2500 g | / ^ 2000 UJ (-CC i / i s X ' \ :ALCYLINDE -ii MAXLOC ii Li DIESEL100% DSL30% + G A S 7 0 % (PHI-30) . DSL30% + GAS70%(PHI-0) (BOI-4 D E G BTDC) 500 . . i . . , , { . , . , i . ; . . i , . . , i . t . . . . i . . . . i -20 -10 0 10 20 30 40 50 60 C A (DEG) Fig. 9.9: Maximum local cylinder temperature for cases A, B and C 199 BOTTOM MIDDLE TOP CA (ATDC) L=872 H=908 1=716 H=896 L=546 H=865 L=862 H=952 L=789 H=998 L=772 H=1175 10 20 L=951 H=1493 L=1063 H=2185 L=930 H=2329 Fig. 9.10: Contours of constant temperature in Kelvin for case C (H=Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 200 CA(DEG) Fig. 9.11: Formation of NOx for cases A and C 201 BOTTOM MIDDLE TOP CA (ATDC) 0 5 10 L=1,0 H=10.0 L=10.0 H=88.0 L=18.0 H=160.0 20 L=2.0 H=16.0 U21.0 H=193.0 L=61.0 H=550.0 Fig. 9.12: Contours of Constant NOx Emission in ppm for case C (H-Highest and L=Lowest). The cross sectional views are shown in Fig. 8.1 202 (BOldsl-4, BOIgas-4) 20 O 15 z cc to 10 CO CO < s _ l UJ z> PILOT DIESEL (30%) NATURAL GAS (70%) 10 20 30 40 CA (DEG) • ' • 50 60 Fig. 9.13: Mass burning rate for the pilot diesel and natural gas (case D) (BOIdsl-12, BOIgas-4) 20 PILOT DIESEL (30%) NATURAL GAS (70%) 20 30 40 CA(DEG) 50 60 Fig. 9.14: Mass burning rate for the pilot diesel and natural gas (case E) 203 (BOIdsl-4, BOIgas-1) i i i Fig. 9.15: Mass burning rate for the pilot diesel and natural gas (case B) Table 9.2: Ignition delay and combustion duration for pilot diesel P A R A M E T E R CaseB CaseD Case E Ignition delay (deg) 9.0 9.5 9.5 Combustion duration (deg) 13.5 16.5 20 O 1 5 o UJ z cc m 10 to 204 CHAPTER TEN - CONCLUSIONS AND RECOMMENDATIONS 10.1 Introduction This chapter summarizes the principal findings from experimental and numerical simulations of diesel and natural gas fueling of diesel engines. Throughout this chapter, the term baseline diesel refers to 100% diesel while diesel-gas represents the 30% pilot diesel and 70% natural gas fuel ratio. The purpose of the pilot diesel was to ignite natural gas injected directly into the combustion chamber. The emissions of oxides of nitrogen are collectively referred to as NOx. The most important questions addressed in this work concern the ignition and combustion of the pilot diesel and the natural gas and the validity of three-dimensional numerical simulations in comparing specific ignition and combustion reaction rates. The test conditions for the experimental and numerical simulations were based on medium range of speed and load, i.e. 1250 RPM and 3 bar BMEP (brake mean effective pressure) which is representative of engine operating conditions. The effect of injection timing was investigated for cases of beginning of injection (BOI) timed at 4, 8,12 and 16 degrees before top dead center (BTDC), but special attention was devoted to the injection timing of 4 deg. case because this is associated with lowest NOx emissions. Strictly speaking, the following conclusions pertain directly to the stated test conditions. However, as it will be seen, a number of them have wider implications. 205 10.2. Ignition 1) With pilot diesel preceding natural gas by 3°, it is possible for pilot diesel to successfully ignite the natural gas over a large range of injection timing, including late injection which is favourable for low NOx emissions. Experimental results show that, changing the injection timing from 4° to 16° BTDC results in an increase of ignition delay by 50%. 2) Injection timing of natural gas relative to pilot diesel is critically important if ignition is to lead to complete combustion. Injection of pilot diesel preceding natural gas has demonstrated successful ignition of natural gas. Simulation results indicate that, changing the injection interval, say, from 3° (i.e. pilot diesel preceding natural gas) to 0° crank angle leads to the ignition of pilot diesel, but the combustion period of pilot diesel becomes much longer and natural gas fails to burn completely. 3) Ignition delays deduced from mass-burned fraction (from measured data) and from numerical simulations are consistent with appearance of flame growth observed in flame photographs. At late injection timing, the ignition delay for diesel-gas is similar to that of the corresponding baseline diesel case. However, ignition delay for diesel-gas increases with advance injection. 4) The natural gas is ignited by direct contact with the products of pilot combustion, 206 rather than by auto-ignition in the unburned region. The maximum unburned gas temperature with pilot diesel combustion (determined from the analysis of measured cylinder pressure data and from numerical simulations) was about 850 K. This shows that, though the pilot diesel combustion raises the unburned gas temperature, it does not rise high enough for spontaneous ignition of the natural gas. The relative thermal and chemical contribution from the burned products of the pilot combustion to the natural gas ignition have not been determined 5) Ignition delay for diesel-gas appears to be tittle dependent on the ckcumferential geometric injection angle of the pilot diesel relative to natural gas. Simulation results have shown that decreasing the circumferential spacing of pilot diesel and natural gas from 30° to 0° does not significantly influence the ignition delay of the pilot diesel. 10.3 Combustion 1) The combustion duration for diesel-gas is the same (within experimental uncertainty) as for baseline diesel, and essentially independent of injection timing. Simulations indicate that the high gas-jet-induced turbulence has tittle effect on the main combustion period. Despite the high turbulence level (calculated from simulations during the injection period), the combustion photographs show that the large scale flame structure is not radically wrinkled by the high velocity natural gas injection. However, small scale turbulence (comparable to the flame thickness) may exist although not obvious from the flame photographs. 207 2) Flame photos indicate a transitional phase between the completion of pilot diesel combustion and the beginning of natural gas combustion. This is consistent with the change in the slope of the mass-burned fraction (from measurements and simulations) at the transitional phase. Simulations indicate that, for a given relative injection timing, the length of the transition period is independent on the absolute (beginning of pilot) injection timing. 4) For the experimental conditions under study, the diesel-gas combustion (from measured data) results in about half of the baseline diesel cycle-to-cycle variations. 5) The diesel-gas combustion (under the test conditions) does not indicate a pre-mixed-dominated combustion despite high injection velocity of natural gas, which would tend to promote mixing of fuel and air before ignition. 6) With high pressure natural gas injection (about 140 bar), simulations show that diesel-gas combustion takes place near the wall regions resulting in high heat losses, while baseline diesel burns mainly in the piston bowl. 7) Simulation results indicate that, zero relative injection angle (geometric angle in the circumferential direction) results in about 100% lower NOx emissions compared with 30° spacing of the pilot diesel sprays and natural gas jets; this indicate the injection geometry between the pilot diesel and the natural gas deserve further study. 208 10.4 Multi-Zone Combustion Model (XPNOX code) The use of the XPNOX code has been a valuable tool for detennining parameters of interest from the measured data which otherwise would have been difficult do obtain. Such parameters are: ignition delay, combustion duration and wall heat loss. 10.5 3-D Numerical Simulation (KIVA-II code) Predictions of KIVA (in its present state of development) are generally outside the bound of experimental errors, but the KIVA simulation represents the experimental data within the following discrepancies: peak cylinder pressure (3%), peak burned gas temperature (6%), unburned gas temperature (2%), ignition delay (20%), and combustion duration (20%). KIVA-II (in the current form) is not an absolute predictive tool, but with modest adjustments in some parameters (e.g. reaction rate constants and wall heat transfer). For given a given experimental conditions, the KIVA simulations can be valuable because they can offer: 1) possible explanations of experimental trends; 2) possible design changes to improve efficiency and reduce NOx, e.g. by providing spatial and temporal details of the zones of fuel-air mixing, ignition, combustion, NOx formation, etc. in the combustion chamber 209 10.6 Recommendations for Future Work 10.6.1 Experimental i) Perform measurements and flame visualization a over wider range of engine operating conditions, i.e. load, speed, and injection timing ii) Operate the engine at lower ratio of pilot diesel to natural gas 10.6.2 Numerical Simulation i) Perform numerical simulations corresponding to the experiments, i.e. loads, speeds and injection timings ii) Perform numerical simulations for the engine operate at lower ratio of pilot diesel to natural gas iii) Determine whether the relative injection timing of pilot diesel and natural gas is dependent on engine speed, load and the ratio of diesel to natural gas iv) Establish whether multiple injections of natural gas result in lower NOx emissions without compromising fuel consumption and engine performance v) Improve the ignition and combustion models in the KIVA code 210 REFERENCES Abraham, J. Bracco, F. V. and R. D. Ritz A Comparison of Computed and Measured Premixed Charge Engine Combustion Combustion and Flame, Vol 60, pp 309-322 (1985) Adamczyk, A. A. and G. A. Lavoie Laminar Head-on Flame Quenching - A Theoretical Study SAE Transactions, Vol. 87, SAE 780969 (1978) Aoyama ,T. Mizuta, J. and Y. Oshima NOx Reduction by Injection Control SAE Paper No 900637 (1990) Amsden, A .A. O'Rouke, P. J. and T. D. Buttler Kiva Computer Program for Chemically reactive Flows with Sprays Los Alamos National Laboratory, LA 11560 - MS (1989) Beck, N. J. Uyehara, O. A. and W. P. Johnson Effects of Fuel Injection on Diesel Combustion SAE Paper No 880299 (1988) Befrui, A. Brandsttter, W. and H. Kratochwill Calculations of Inhomogenous-Charge Combustion in Swirl Assisted Lean Burn Engine SAE Paper 910266 (1991) Brown A. J. and J. B. heywood A Fundamentally Based Stoichiostic Mixing Model Method for Predicting NO and Soot Emissions from Direct Injection Diesel Engines Calculations of Reactive Flows -AMD Vol. 81 (1986) Chandrsuda, Mehta, R. P. Wier, A. D. and P. Bradshaw Effects of Free Stream Turbulence on Large Scale Structures in Turbulence Mixing Layers /. Fluid Mech. Vol 85 pp 693-704 (1978) Chepakovich A. Visualization of Transient Single and Two Phase Jets Created by Diesel Engine Injectors M. A. Sc. Thesis, University of British Columbia (1993) Chiu, W. S. Shahed, S. M. and W. T. Lyn A Transient Spray Mixing Model for Diesel Combustion SAE Paper 760128(1976) Chomiak, J. Combustion: A Study in Theory, Fact and Application Abacus Press/Gordon and Breach Science Publishers (1990) 211 Daneshyar, H. and P. G. Hill The Structure of Small-Scale Turbulence and Its Effect on Combustion in Spark Ignition Engines Prog. Energy Combust. Sci. 1987, Vol 13, pp 47-73 Dent, J. C. and S. J. Sulaiman Convective and Radiative Heat Transfer in a High Swirl Direct Injection Diesel Engine SAE Paper No 77047 (1977) Dig, X and P. G. Hill Emissions and Fuel Economy of Prechamber Diesel Engines with Natural Gas Dual Fueling SAE Paper 860069 (1986) Douville, B. Performance, Emissions and Combustion Characteristics of Natural Gas Fueling of Diesel Engines M. A. Sc. Thesis, University of British Columbia (1994) Einang, M. P. Koren, S. Kvamsdal, R. Hansen, T. and A. Sasten High Pressure Digitally Controlled Injection of Gaseous Fuel in Diesel Engine, With Special Reference to Boil-Off from LNG Tankers C Z A 4 A C Conference Proceedings, Paris (1983), pp 1285-1308 Einang, M. P. Engja H. and R. Vestergen Medium Speed 4-Stroke Diesel Engine Using High Pressure Gas Injection Technology CIMAC Conference Proceedings, Paris, (1984), pp 916-932 Ferguson, C. R. Internal Combustion Engines: Applied Thermosciences John Willey and Sons (1985) Ferguson, C. R. Green, R. M. and R. P. Lucht Unburned Gas Temperature in an Internal Combustion Engine BL Heat Realease Computations Combustion Sci. and Tech., Vol. 55, pp. 63-81, (1987) Foster, C. R. An Overview of Zero Dimensional Thermodynamic Models for Internal Combustion Engine Data Analysis SAE Paper 852070(1985) Fraser ,R. Siebers, L. D. and C. F. Edwards Auto-Ignition of Methane and Natural Gas in a Simulated Diesel Environment SAE Paper 910227 (1991) 212 Fraser, R. A. Felton, P. G. Bracco, F. V. and D. A. Santavicca Preliminary Turbulence Length Scale in Motored Engine SAE Paper 860021 (1986) Fukuda, T. Komoda, T. Furusima, K. Yanagihara, M. and Y. Ito Development of a Highly Efficient Gas Diesel Engine with Glow Plug Ignition Assist for Degeneration Systems JSME-ASME International Conference on Power Engineering (1993) ICOPE -1993, Vol. 1 Glassman I Combustion Academic Press, Second Edition, 19987 Gosman, A. D. and R. J. Johns Computer Analysis of Fuel Air Mixing and Combustion in Direct Injection Diesel Engines SAE Paper 800091(1980) Gosman, A. D. and P. S. Harvey Computer Analysis of Fuel Air Mixing and Combustion in Axisymetric Direct Injection Diesel Engines SAE Paper 820036 (1982) Grasso, F. Wey, M. J. Bracco, F. V. and J. Abraham Three Dimensional Computations in Stratified-Charge Rotary Engine SAE Paper 870409 (1987) Heywood, J. B. Internal Combustion Engine Fundamentals McGraw Hill Book Company (1988) Heywood, J. B. Fluid Motion Within the Cylinder of Internal Combustion Engines -A Rewiew Journal of Fluid Engineering, Vol 109, March 1987 Hill, P. G. and B. Douville NOx Formation in Two Stroke Engines (to be published) Hill, P. G. and P. Ouellette Self Similar Transient Jets (to be published, 1996) 213 Hires, S. D. Ekchian, A. Heywood, J. B. Tabaczynski, R. J. and J. C. Wall Performance and NOx Emissions Modelling of Jet Ignition Prechamber Stratified Charge Engine SAE Paper 760161 (1976) liiroyasu, H. and T. Kadota Models of Combustion and Formation of Nitric oxide and Soot in Direct Injection Diesel Engines SAE Paper 760129 (1976) Hodgins, K. B. and P. L. Mtui Evaluation of PCB Pressure Transducers Department of Mechanical Engineering, UBC Internal Report, June 1994 Iida, N. and G. T. Sato Temperature and Mixing Effects on NOx and Particulate SAE Paper 880424 (1988) Jennings, M. J. and F. R. Jeske Analysis of Injection Process in Direct Injected Natural Gas Engines: Part I - Study of Unfined and m< y^linder Plume Behaviour Journal of Engineering far Gas Turbines and Power, Vol. 116 (1994) Jennings, M. J. and F. R. Jeske Analysis of Injection Process in Direct Injected Natural Gas Engines: Part II - Effects of Injector and Combustion Chamber Design Journal cf Engineering far Gas Turbines and Power, Vol. 116 (1994) Karim, G. A. An Examination of Some Measures for Improving the Performance of Gas Fueled Diesel Engines at Light Load SAE Paper 912366 (1991) Karim, G. A. Ito, K. Abraham, M. and L. Jensen Examination of the Role of Formaldehyde in the Ignition Processes of a Dual Fuel Engine SAE Paper 912367 (1991) Kato, T. Tsujimura, K. Sruntani, M. Miami, T. and I. Yamaguchi Spray Characteristics and Combustion Improvement of Dl Diesel Engine with High Pressure Fuel Injection SAE Paper 890265 (1989) Kono, K. and N. Akmito Prediction of m-Cyhrider Flow Spray Formation Effects on Combustion in Dl Diesel Engines SAE Paper 850108 (1985) 214 Kuo, T. W. and F. V. Bracco Computation of Drop Sizes in Pulsating Sprays and Liquid-Core Length in Vapourizing Sprays SAE Paper 820133 (1983) Kuo, T. W. and R. C. Yu Modeling of Transient Evaporating Spray Mixing Processes - Effects of Injection Characteristics SAE Paper 840226 (1984) Kuo, T. W. and R. D. Ritz Computation of Premixed Charge Combustion in a Pancake and Pent Roof Engines SAE Paper 890670 (1989) Launder, B. E. and D. B. Spalding The Numerical Computation of Turbulence Flows Computer Methods in Applied Mechanics and Engineering 3 (1974) 269-289 Lorn, E. J. and H. K. Ly High Pressure Injection of Natural Gas in a Two Stroke Diesel Engines SAE Paper 902230 (1990) Magnussen, B. F. and B. H. Hjertager On the Mathematical ModeUing of Turbulent Combustion with Special Emphasis on Soot Formation and Combustion 16th Symposium on Combustion, Combustion Institute (1976), pp 719-729 Mansouri S, H. Heywood, J. B. and K. Randrishan Divided Chamber Diesel Engine, Part I: A Cycle Simulation which Predicts performance and Emissions SAE Paper 820273 Metghalchi, M. and J. Keck Buirung Velocities of Mxtures of Air with Methanol, Iso-octane and Indolene at High Pressure and Temperature Combustion and Flame, Vol 48, pp 191-210 (1982) Minami, Toshitaka, Takeuchi, Kouichi, Simakazi and Naoki Reduction of Diesel Engine NOx Using Pilot Injection SAE Paper No 950611 Miyake, M. Biwa, T. Endoh, Y. Shimotso, M. Mukarani, S. and T. Kumoda Development of High Efficient Gas Burning Diesel Engines CIMAC Paper, Conference Proceedings, Paris (1983) 215 Mulholland, J. A. Sarofin,A. F. and J. N. Beer On the Derivation of Global Ignition Kinetics from a Detailed Mechanisms for Simple Hydrocarbons Combustion Science and Technology, 1992, Vol 87, pp 139-156 Naber, J. D. Siebers, D. L. Di Julio and C. K. Westbrook Effects of Natural Gas Composition on Ignition Delay Under Diesel Conditions Combustion and Flame 99 2 Nov 1994. pp 192 - 200 Plee, S. L. and T. Ahmad Relative Roles of Premixed and Diffusion Burning in Diesel Combustion SAE Paper 831733 (1983) Pinchon, P. Three Dimensional Modelling of Combustion in a Pre-chamber Diesel Engine SAE Paper 89066 (1989) Schenkck H. The Theories of Engineering Experimentation Hemisphere Pubhshing Corportion Third Edition, 1979 Schulte, H. Scheid, E. Pischinger, F. and U. Reuter Pre-Injection: A Measure to bifluence Exhaust Quality and Noise of Diesel Engine ASME, ICE-Vol 4(1988), 97 Shundoh, H. Kakegawa, T. Tsujimura K. and S Kobayashi The Effect of Injection Parameters and Swirl on Diesel Combustion with High Pressure Fuel Injection SAE Paper 910489 (1991) Sloane, T. M and P. D. Ronney A Comparison of Ignition Phenomena Modelled with Detailed and S i m p l i f i e d Kinetics Combustion Science and Technology, Vol 88, pp 1 -13 (1992) Spalding, D. B. Mixing and Chemical Reaction in Steady Confined Turbulent Flames 13th Symposium (Int) on Combustion (1971) Combustion Institute pp 649 - 657 Su, Y. S. and R. M. Traci The Application of MAGIC Internal Combustion Engine Model: Turbulent Combustion Results SAE Paper 872101 (1987) 216 Tabacznski, R. J. Ferguson, C. R. and Radhakrishnan A Turbulent Entrapment Model for Spark Ignition Engine Combustion SAE Trans. Vol 86 Paper 770647, pp 2414-2433 Tao, Y. Performance and Emission Characteristics of Gas Diesel Engines M. A. Sc. Thesis, University of British Columbia (1993) Tao Y., Hodgins B. and P. G. Hill NOx Emissions From a Diesel Engine Fueled With Natural Gas ASME Transactions, pp 290-296, Vol. 117, Dec. 1995 Ulrich, S. and R. Krebs Optical Fibre Technique as a Tool to Improve Combustion Efficiency SAE Paper 902138 (1990) Varnavas, C. A. Studies of Diesel Engine Combustion Using a Multi-Dimesional Computer Simulation M. Sc. Thesis, University of Illinois, (1990) Varnavas C. A. and D. A. Assanis The Effect of Spray, Mixing and Combustion Model Parameters on KTVA-II Predictions SAE Paper 911785 (1991) Wade, W. R. Hunter, C. E. Trinker, F. H. and H. A. Cikanek The Reduction of NOx and Particulate Emissions in Diesel Combustion Process ASME, 87-ICE-37 (1987) Wakenell, F. J. O'neal G. B. and Q. A. Baker High Pressure Late Cycle Direct Injection of Natural Gas in Rail Medium Speed Diesel Engine SAE Paper 872041 (1987) Walker, R. W. Reaction of H O 2 Radicals in Combustion Chamistry 22nd Symposium (International) on Combustion The Combustion Institute, (1988), pp 883-892 Tao Y., Hodgins B. and P. G. Hill NOx Emissions From a Diesel Engine Fueled With Natural Gas ASME Transactions, Vol. 117, Dec. 1995 Wong, W. Y. Midkiff, K. C. and S. R. Bell Performance and Emissions of Natural Gas Dual-Fuelled Indirect Injection Engine SAE Paper 911766(1991) 217 Woschni, G. A Univesally Applicable Equation for the Instantaneous Heat Transfer Coefficient Internal Combustion Engine SAE Paper 670931 (1967) Zellat,M. RollanctTh. and F. Poplow Three Dimensional Modelling of Soot Formation in Indirect Injection Diesel Engines SAE Paper 900254 (1990) Zou, G. and G. A. Karim A Comprehensive Kinetic Model for the Oxidation of Methane Emerging Technology, ASME Vol. 57,1994 (I) Zhou, G. and G. A. Karim The Concentration of Radicals and Intermediate Reactive Species in Methane Oxidation Emerging Technology, ASME Vol. 57,1994 (II) 218 A P P E N D I X A : T E C H N I C A L S P E C I F I C A T I O N S F O R T H E E N G I N E S Table A . l : Technical data for DDC 6V-92TA and DDC 1-71 engines ENGINE PARAMETER DDC 6V-92TA DDC 1-71 Number of cycles 2 - Cycle 2 - Cycle Number of cylinders 6 (V-type) 1 Cylinder bore 123 mm 108 mm Stroke length 127 mm 127 mm Conrod length 257.5 mm 254.0 mm Displacement 9.05 Liters 1.162 Liters Compression ratio 17:1 16:1 Rated power 225KW@1200RPM 11.2KW@1200RPM Rated BMEP 9.2BAR@1200RPM 4.8BAR@1200RPM Rated torque 1322NM@1200RPM 76NM@1200RPM Inlet port closure 55 Deg ABTD 60 DegABDC Outlet port open 85 Deg ATDC 85 Deg ATDC BBDC refers to Before Bottom Dead Center ABDC refers to After Bottom Dead Center 219 APPENDIX B: PRESSURE TRANSDUCER TESTING After each transducer was installed, the engine was warmed to the operating temperature. A speed set point (1250 or 1800 rpm) was set to a dynamometer controller. Engine throttle position sensor (TPS) was adjusted to give desired load of 1, 3, and 6 bar brake mean effective pressure (BMEP). Twenty consecutive cycles of pressure data were acquired for the six test points. Prior studies have shown that twenty cycles is a statistically significant sample for cylinder pressure averaging. Total Frictional Mean Effective Pressure (TFMEP) was evaluated and used to establish the performance of the transducers. TFMEP is defined as the difference between the indicated mean effective pressure (IMEP) and break mean effective pressure (BMEP). BMEP is given by equation B. l BMEP = P°*1()3*nR (B.l) VD*N where PB is the measured power (kW), V D is the total engine displaced volume (dm3) and N is the engine speed (rev/ s). The factor nR equals 1 and 2 for two and four stroke engine respectively. The measured power is given by eqn. B.2. PB=2*n*N*TB (B.2) where TB is the measured torque (N.m) and N is the engine speed (rev/s). IMEP= W>ND*103 (B-3) Where WIND is the work per revolution as calculated from j) pdv. 220 Theoretical TFMEP was approximated from experimental correlations provided by Heywood [1988] by eqn B.4. TFMEP(kPa) = Ci + C2-^— + C 3 S 2 (B.4) 1000 where N is the engine speed (RPM) and S is the piston mean speed (m/s). Coefficients Ci , C2 and C3 are determined by engine configuration and operating conditions. Figure B. l and B.2 shows the TFMEP and the coefficient of variability C O V (defined in Section 3.5.3) at engine speed of 1250 RPM, respectively for the transducers tested. It is noted that the TFMEP and C O V decrease with increasing load. The TFMEP obtained by transducer Model 112A10 agrees closely with theory (eqn. B^l) and that it is relatively insensitive to thermal effects since it varies linearly horizontal with load. Similarly, at high loads, say, 5 bar BMEP, the C O V is lowest for transducer model 112A10. 221 0.0% 0 1 2 3 4 5 bar Engine Load (BMEP) Fig. B.2: Coefficient of variability @1250RPM for 100% diesel 222 APPENDIX C: CALCULATION OF MASS BURNED FRACTION (XPNOX CODE) The mass burned fraction is calculated (from the measured cylinder pressure and flow rates of intake air and fuel) with a procedure proposed by riill and Douville [1996] which is formulated for two-stroke diesel engines. Conditions at inlet port closure are determined from a scavenge air flow estimate coupled with measurements of air-box temperature and exhaust temperature, which together allow residual gas temperature and composition to be determined. Low temperature products of combustion are assumed to consist of Oz, N2, CO2, and H2O/ except that N O produced at high temperatures and in quantities too small to affect the engine heat balance) is assumed not to decay on cooling to exhaust temperature. 1. Combustion Analysis Combustion process is performed in a series of 1 deg. crank angle increment k*. In a time interval d6, a quantity of fuel dmfk is assumed to burn in stoichiometric fuel-air ratio. The products of combustion at temperature Tbk (determined from mass and energy conservation) are assumed to be in equilibriirm. The burned charge mixes with air-residual mixture to form burned-and mixed mass. For the results reported in this thesis, the characteristic mixing time is about 0.5 ms. The burned-and-mixed mass then undergo isentropic compression or expansion. The burned-and-mixed mass will have different temperature; formation of NOx ceases after mixing. During the time interval dd, a quantity of fuel dmjwj having a specific enthalpy hjmj is injected. Work done on the piston is dW=pdVafi results from combustion the injected fuel. The wall heat loss dQ is evaluated as discussed below 223 2. Mass Conservation Considering the end of combustion process, the mass conservation is expressed in terms of volume as shown in eqn. (Cl) . where Vty and V u are the instantaneous cylinder volume and total air residual volume. Vb is the total volume of the previously burned (but not yet mixed) and Vbm is the burned and previously mixed gas volume. The total volume of fuel injected but not yet burned is denoted by V^. At k"1 interval a fuel quantity dmfk (of specific volume V b k ) is injected. 3. Energy Conservation Corresponding to eqn. (Cl ) the total energy in the cylinder at the end of combustion interval is in which each of the E terms has the same definition as eqn. (Cl ) except that the symbol e stands for the specific internal energy. The change in cylinder energy during combustion interval is Vcy=Vu+ V b + Vbm+ V M + d r a ^ k (Cl ) E