UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Emissions characterization of paired gaseous jets in a pilot-ignited natural-gas compression-ignition… Mabson, Christopher William John 2015

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

Item Metadata

Download

Media
24-ubc_2015_may_mabson_christopher.pdf [ 3.09MB ]
Metadata
JSON: 24-1.0167675.json
JSON-LD: 24-1.0167675-ld.json
RDF/XML (Pretty): 24-1.0167675-rdf.xml
RDF/JSON: 24-1.0167675-rdf.json
Turtle: 24-1.0167675-turtle.txt
N-Triples: 24-1.0167675-rdf-ntriples.txt
Original Record: 24-1.0167675-source.json
Full Text
24-1.0167675-fulltext.txt
Citation
24-1.0167675.ris

Full Text

EMISSIONS CHARACTERIZATION OF PAIRED GASEOUS JETS IN A PILOT-IGNITED NATURAL-GAS COMPRESSION-IGNITION ENGINE  by CHRISTOPHER WILLIAM JOHN MABSON B.S. Washington State University, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Mechanical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  February 2015 Β© Christopher William John Mabson, 2015ii Abstract Heavy-duty engines must meet strict emission standards and retain high fuel efficiency. This thesis examines a new type of fuel injector nozzle for a pilot-ignited direct-injection natural gas engine. The nozzle uses paired jets that increase mixing with air during combustion, which aims to reduce the amount of particulate matter (PM) formed. Tests were performed for different speeds and loads and over engine parameter sweeps (including timing, EGR, EQR, and diesel pilot mass) to compare the effects on the emissions to a single-hole nozzle. Low-PM strategies and morphology of the soot were compared as well. Contrary to expectations, the tests showed large increases in CO and PM from all the paired nozzles at all modes compared to the single-holed injector. Changing speed and load did not affect the relative emissions so further tests were only done with the paired nozzle that had the least emissions. The engine parameter sweeps at mid speed, high load (B75) showed similar emission patterns for the paired-hole nozzle and the single-hole nozzle. This suggests that the reasons for the high emissions lie in the characteristics of the jets, which are not changed much under normal HPDI operation. Injecting the natural gas before the pilot injection reduced PM. Late post-injection of some of the gas reduced PM by 50% without increasing other emissions for both injector types. Apparently, these strategies could work for other HPDI injectors. Compared to the reference injector, the paired-hole nozzle produced larger soot aggregates and larger numbers of particles but soot primary particle size showed different trends at different conditions. Soot fractal dimensions were the same and consistent with conventional diesel soot. CFD simulations showed that fuel packets moved through a richer high PM and CO forming zone during combustion for the paired nozzle. This high sooting zone had to be avoided either by further mixing or less mixing to avoid the high emissions produced. The results presented here were developed on a single-cylinder engine. While trends are expected to be similar to those from an equivalent multi-cylinder engine, emission levels and fuel consumption are not directly comparable to production multi-cylinder engines.   iii Preface The original design of the paired nozzles was done by Ehsan Faghani. I planned and conducted the entire engine test matrix and performed the post-processing of the engine data. For the TEM sampling, Ramin Dastanpour provided assistance with the thermophoretic sampler and took the images of the grids using the microscope. I processed the images with a semi-automatic code Ramin had written. I selected the tests to run with CFD. The CFD cases and post-processing of the data was run by Pooyan Kheirkhah.   iv Table of Contents Abstract ......................................................................................................................................................... ii Preface ......................................................................................................................................................... iii Table of Contents ......................................................................................................................................... iv List of Tables .............................................................................................................................................. viii List of Figures ............................................................................................................................................... ix List of Acronyms ........................................................................................................................................... xi Acknowledgments ....................................................................................................................................... xii 1. Introduction .............................................................................................................................................. 1 1.1 Emissions from CI Engines and PM Health Effects .............................................................................. 1 1.2 Formation of Emissions in a CI Engine ................................................................................................ 1 1.3 Emission Regulations and After-Treatment Strategies ....................................................................... 4 1.4 Benefits of Natural Gas as a Fuel ........................................................................................................ 6 1.5 High Pressure Direct Injection of Natural Gas (HPDIβ„’) ...................................................................... 7 1.6 Previous Literature Review for Paired Jets ......................................................................................... 8 1.7 Objectives and Scope ........................................................................................................................ 12 2. Experimental Setup ................................................................................................................................. 14 2.1 Injector Configurations ..................................................................................................................... 14 2.2 Single Cylinder Research Engine (SCRE) ............................................................................................ 15 2.3 In-Cylinder Pressure Measurement .................................................................................................. 17 2.4 Heat Release Rate ............................................................................................................................. 17 2.5 Gaseous Emissions Measurement .................................................................................................... 18 2.6 PM Emissions Measurement ............................................................................................................ 19 2.6.1 Sampling System ........................................................................................................................ 19 v 2.6.2 DustTrak DRX Aerosol Monitor .................................................................................................. 21 2.6.3 Tapered Element Oscillating Microbalance (TEOM) .................................................................. 21 2.6.4 Scanning Mobility Particle Spectrometer (SMPS) ...................................................................... 22 2.6.5 Transmission Electron Microscopy (TEM) Imaging .................................................................... 24 2.7 GOLD CFD Model .............................................................................................................................. 24 2.8 Data Processing ................................................................................................................................. 25 2.9 Repeatability of Experiments ............................................................................................................ 27 3. Results ..................................................................................................................................................... 30 3.1 Multi-Mode Tests .............................................................................................................................. 32 3.1.1 Emissions .................................................................................................................................... 32 3.1.2 Effect of Emissions Formation on Fuel Consumption ................................................................ 36 3.1.3 Heat Release Rates ..................................................................................................................... 37 3.1.4 Discussion of Multi-Mode Results ............................................................................................. 40 3.2 Emissions Characterization with Parameter Sweeps ........................................................................ 40 3.2.1 Timing Sweep ............................................................................................................................. 40 3.2.2 Exhaust Gas Recirculation (EGR) Sweep .................................................................................... 42 3.2.3 Equivalence Ratio (EQR) Sweep ................................................................................................. 44 3.2.4 Diesel Mass Sweep ..................................................................................................................... 46 3.2.5 Parameter Sweep Discussion ..................................................................................................... 47 3.3 Low PM Strategies ............................................................................................................................ 48 3.3.1 Late Post Injection ...................................................................................................................... 48 3.3.2 Negative PSEP ............................................................................................................................ 51 3.3.3 Slightly Premixed Combustion ................................................................................................... 53 3.3.4 Low PM Strategies Discussion .................................................................................................... 55 3.4 Response Surfaces of Experiments ................................................................................................... 55 vi 3.5 PM Characteristics Comparison for LHSA and Reference Injector ................................................... 58 3.5.1 Sizing of Primary Particles and Aggregates ................................................................................ 59 3.5.2 Fractal Analysis of Aggregates ................................................................................................... 61 3.5.3 Differences between HPDI and Diesel PM ................................................................................. 64 3.5.4 Effects of PM Morphology on a DPF .......................................................................................... 64 3.5.5 Discussion of PM Characteristics Comparison for LHSA and Reference Injector ...................... 65 3.6 Interpretation of Measurements ...................................................................................................... 65 3.6.1 Computational Fluid Dynamics (CFD) Studies ............................................................................ 65 3.6.2 Combustion Maps ...................................................................................................................... 72 4. Conclusions ............................................................................................................................................. 77 5. Recommendations for Further Work ...................................................................................................... 78 Bibliography ................................................................................................................................................ 79 Appendices .................................................................................................................................................. 85 Appendix A: Raw Data Tables ................................................................................................................. 85 Multi-Mode Data ................................................................................................................................ 85 Timing Sweep ...................................................................................................................................... 87 EGR Sweep .......................................................................................................................................... 89 EQR Sweep .......................................................................................................................................... 91 Diesel Mass Sweep .............................................................................................................................. 93 Negative PSEP Sweep .......................................................................................................................... 95 Optimal Points – Negative PSEP, SPC .................................................................................................. 97 LPI ........................................................................................................................................................ 99 PM Characteristics ............................................................................................................................ 100 Full Factorial Matrix .......................................................................................................................... 102 Appendix B: SCRE Operating Procedures .............................................................................................. 103 vii Appendix C: Airflow Calibrations .......................................................................................................... 110 Appendix D: Piezoelectric Transducer Calibration ................................................................................ 115 Appendix E: High Range CO Calibration ................................................................................................ 124    viii List of Tables Table 1 - European Heavy Duty Emissions Limits Steady-State (adapted from www.dieselnet.com) ......... 4 Table 2 - European Heavy Duty Emissions Limits Transient (adapted from www.dieselnet.com) .............. 4 Table 3 - Injector Nozzle Specifications ...................................................................................................... 15 Table 4 - Custom DMA Column Dimensions and SMPS Flows .................................................................... 23 Table 5 - Allowable Operator Variability for B75 ........................................................................................ 27 Table 6 - Coefficient of Variance B75 Points over Time .............................................................................. 29 Table 7 - Summary of Experiments ............................................................................................................. 31 Table 8 - Loss in Energy due to Formation of Emissions ............................................................................. 37 Table 9 - Coefficients of Variation LPI ......................................................................................................... 50 Table 10 - Coefficients of Variation SPC ...................................................................................................... 54 Table 11 - R2 values for predicting different output emissions .................................................................. 57 Table 12 - DRX and CO Sensitivities at B75 relative to 2 parameters changing.......................................... 58 Table 13 - Morphology of PM at B75 .......................................................................................................... 60 Table 14 - 95% Confidence Intervals for π‘˜π‘“, 𝐷𝑓 ......................................................................................... 63 Table 15 - Inputs for GOLD Simulation ....................................................................................................... 66 Table 16 - SCRE vs. GOLD Emissions ........................................................................................................... 71    ix List of Figures Figure 1 - Soot Formation Process (adapted from Tree and Svensson, 2007) ............................................. 1 Figure 2 - TEM image of Soot Aggregates and Primary Particles .................................................................. 2 Figure 3 - HPDI Injector (www.westport.com) ............................................................................................. 7 Figure 4 - Typical HPDI Injection Strategy ..................................................................................................... 8 Figure 5 - Mixture Fraction of Single and Paired Jets (adapted from Faghani and Rogak)......................... 10 Figure 6 - Group-Hole Nozzle Design (adapted from Zhang et al.) ............................................................. 10 Figure 7 - Reference Injector and Paired Nozzle Configuration.................................................................. 14 Figure 8 - Single Cylinder Research Engine (SCRE): ..................................................................................... 16 Figure 9 - PM Dilution System ..................................................................................................................... 20 Figure 10 - TEOM Measurements ............................................................................................................... 22 Figure 11 - SMPS System ............................................................................................................................. 23 Figure 12 - Image Processing of an Aggregate ............................................................................................ 24 Figure 13 - Point over Time Variability: B75 Baseline ................................................................................. 28 Figure 14 - Multi-Mode PM Emissions ........................................................................................................ 33 Figure 15 - CO Normalized Emissions ......................................................................................................... 34 Figure 16 - Normalized NOx Emissions ........................................................................................................ 35 Figure 17 - Normalized Methane Emissions ............................................................................................... 35 Figure 18 - Normalized Fuel Economy ........................................................................................................ 36 Figure 19 - Paired Nozzle Heat Release Rates at Mode B75 ....................................................................... 38 Figure 20 - Paired Nozzle Heat Release Rates at Mode A75 ....................................................................... 39 Figure 21 - 50% IHR Sweep Emissions (7-15Β°aTDC) .................................................................................... 41 Figure 22 - EGR Sweep Emissions (0-24%) .................................................................................................. 43 Figure 23 - EQR Sweep Emissions (0.50-0.7) ............................................................................................... 45 x Figure 24 - Diesel Mass Sweep Emissions (6-22.5 mg/inj) .......................................................................... 47 Figure 25 - Late Post Injection Strategy ...................................................................................................... 49 Figure 26 - LPI Emissions normalized by B75 single injection strategy ....................................................... 50 Figure 27 - Negative PSEP Injection Strategy .............................................................................................. 51 Figure 28 - Negative PSEP PM Emissions (-2.5ms β†’ +0.3ms) ..................................................................... 52 Figure 29 - SPC Emissions normalized by B75 single injection strategy ..................................................... 54 Figure 30 - Response Fit and Residual Plot for LHSA DRX ........................................................................... 57 Figure 31 - TEM vs. SMPS Aggregate Sizing ................................................................................................ 59 Figure 32 - Fits of Fractal Dimension .......................................................................................................... 63 Figure 33 - Gas/Diesel Angle ....................................................................................................................... 67 Figure 34 - Interlace Angle HRR GOLD Simulation ...................................................................................... 67 Figure 35 - In-Cylinder Combustion for Different Interlace Angles ............................................................ 68 Figure 36 - GOLD HRR vs SCRE HRR, Reference Injector and LHSA (8 Degree)........................................... 69 Figure 37 - Local Ο†-T Difference Plot 25% IHR B75 ..................................................................................... 70 Figure 38 - Penetration of Jets (Left) and Entrainment of Air (Right) ......................................................... 70 Figure 39 - PM vs. NOx over entire B75 dataset for LHSA and Ref Inj......................................................... 72 Figure 40 - PM vs. NOx Normalized over entire B75 dataset for LHSA and Ref Inj ..................................... 73 Figure 41 - CO vs. PM over entire B75 dataset for LHSA and Ref Inj .......................................................... 74 Figure 42 - CO vs. PM Low PM Strategies ................................................................................................... 75 Figure 43 - CO vs. PM all PH injectors at B75 .............................................................................................. 76    xi List of Acronyms aTDC - after Top Dead Centre   38 CFD - Computational Fluid Dynamics   24 CI - Compression Ignition   1 CNG - Compressed Natural Gas   6 CPC - Condensate Particle Counter   22 DMA - Differential Mobility Analyzer   22 DOC - Diesel Oxidation Catalyst   5 DPF - Diesel Particulate Filter   5 DRX - DustTrak DRX   20 GIMEP - Gross Indicated Mean Effective Pressure   17 GRP - Gas Rail Pressure   16 GSEP - Gas Separation   18 HPDIβ„’ - High Pressure Direct Injection   7 HRR - Heat Release Rate   17 IHR - Integrated Heat Release   18 LPI - Late Post Injection   51 NOx - Nitric Oxides   1 PM - Particulate Matter   1 PSOI - Pilot Start of Injection   18 SCR - Selective Catalytic Reduction   5 SMPS - Scanning Mobility Particle Spectrometer   22 SPC - Slightly Premixed Combustion   53 TEM - Transmission Electron Microscope   2 TEOM - Tapered Element Oscillating Microbalance   20   xii Acknowledgments I would like to thank everyone who has helped to get me this far in my education. First, I would like to thank my supervisor Dr. Steve Rogak for providing me guidance throughout my work. I am thankful to have had him during my time at UBC. I would also like to thank Dr. Pat Kirchen, Dr. Ning Wu, and Dr. Phil Hill for the weekly check-ins at our SCRE meetings to help me improve my communication and presentation skills.  Being able to work alongside an industry partner has been an exciting experience. I would like to thank all the Westport support engineers including Dr. Gord McTaggart-Cowan, Dr. Ning Wu, and Bronson Patychuk for their support and feedback. I would like to thank Bob Parry, our engine technician for all the help he has provided fixing the SCRE especially with diagnosing problems with the emissions bench. Thanks to Glenn Jolly and Sean Buxton in the instrument shop for LabVIEW assistance and reminding me how circuits work. I would like to thank my fellow researchers on the APC project, Ehsan Faghani and Pooyan Kheirkhah for the constant critique and support of our work. I would like to thank Ramin Dastanpour for all his help with the TEM image processing. I would also like to thank the guys on the Proteus group, especially Jeremy Rochussen and Jeff Yeo for being able to take a break once in a while. Finally I would like to thank my family, especially my parents for helping me get this far. Last I have to thank Brie for the constant support through the past 2 years.    GG. 1 1. Introduction 1.1 Emissions from CI Engines and PM Health Effects Compression-ignition (CI) engines are used in the transportation of goods by large and medium duty trucks, buses, and large marine ships. CI engines have higher fuel efficiency, power, and reliability than spark-ignition (SI) engines. However, the non-premixed combustion that makes the CI engine more efficient results in high emissions of particulate matter (PM) and nitrogen oxides (NOx). Engine exhaust PM is actually a mixture of several types of aerosols including black carbon (soot), semi-volatile organic carbon, and small amounts of ash, metal particles, and sulfates (Khalek, Bougher, Merritt, & Zielinska, 2011) caused from fuel or engine wear. Non-volatile particles are deposited in the lungs and can cause inflammation and oxidative stress (Ristovski et al., 2012). The surface area of these particles is very important as the amount that is absorbed into the lungs depends on the surface area in contact with the lining of the lungs. Smaller particles are able to penetrate deeper into the lungs and have more surface area. The importance of sizing of the PM on health has been discussed in several studies (B. Giechaskiel, AlfΓΆldy, & Drossinos, 2009; Silverman et al., 2012). Therefore qualities of the PM including size and particle number need to also be considered. 1.2 Formation of Emissions in a CI Engine The formation of soot occurs in fuel rich regions of the combustion chamber at elevated temperatures where there is a local lack of oxygen for complete combustion. A hydrocarbon goes through several stages to become soot aggregates which are described in Figure 1.  Figure 1 - Soot Formation Process (adapted from Tree and Svensson, 2007) First is pyrolysis of a hydrocarbon, when a hydrocarbon alters its structure into some form of polycyclic aromatic hydrocarbon with the aid of increased temperatures (Tree & Svensson, 2007). C2H2 (acetylene) is typically modeled as a precursor for soot formation sites due to its strong carbon-carbon bond. These PAHs then combine together to form ring structures such as a benzene ring. 2 During nucleation, the rings then combine with more acetylene and with other ring structures to increase the number of reactive sites and the size of the molecules into small particles between 1 and 2nm in diameter. These are known as nucleation particles and serve as the inception point for larger particles to form. The next stage is known as surface growth which is the process of increasing the mass of a nucleated soot particle (Tree & Svensson, 2007). This is accelerated with higher temperatures and a lack of oxygen and many reactive sites on the outside of the particles. This can be described as HACA or H-abstraction-C2H2-addition (Frenklach, 2002). Surface growth is strongly dependant on local temperature and oxygen concentration and the number of reactive sites available. The next stages are known as coalescence and agglomeration and they all take place at the same time as surface growth. Coalescence occurs as two smaller particles combine into roughly spherical primary particles. When fewer active sites remain this process slows. Agglomeration is the process of primary particles combining with other primary particles or larger chains (aggregates) to form larger structures. These structures are known as soot aggregates which can be chains or clumps of other primary particles and smaller aggregates. Figure 2 shows a sample a Transmission Electron Microscope (TEM) image of soot aggregates.  Figure 2 - TEM image of Soot Aggregates and Primary Particles As the particles move through the combustion chamber, the local oxygen concentration increases, which promotes oxidation of the particles. Higher temperatures also promote increased oxidation of the particles. The total engine-out PM is a balance of the formation rates and the oxidation rates. For further details about soot formation and PM composition see (Tree & Svensson, 2007), (Khalek et al., 2011), and (O’Connor & Musculus, 2013). 3 Additionally, gaseous vapors from fuel or lubricating oils can condense on the agglomerates that are formed. This is known as semi-volatile organic carbon, and can be removed by heating the particles to oxidize the volatile components. Previous measurements (Patychuk, 2013) of elemental carbon (EC) and semi-volatile organic carbon (OC) for this HPDI engine showed that the EC/OC ratio was mainly a function of load. CO is formed from rich areas in the cylinder. In addition to being hazardous to human health, CO also represents energy that is not fully utilized (Pulkrabek, 2004) as a triple bond is formed. CO forms near the outside layer of a burning jet but still within the jet where local temperatures are high and oxygen concentration is low. High temperatures are needed to initially form CO, but low oxygen prevents full oxidation from occurring. CO can also be formed when CO molecules are formed and mix with air quickly, lowering the temperature to prevent oxidation to CO2 despite a high oxygen concentration. NOx is formed on the outside of a burning jet where local temperature is very high and there is plenty of oxygen. The NOx formation process is described by the extended Zeldovich mechanism (Heywood, 1988) and is mainly a function of combustion temperature and local equivalence ratio. In-cylinder reduction of NOx can be done with Exhaust Gas Recirculation (EGR) to reduce oxygen concentration and temperature in the cylinder. In HPDI combustion, unburned hydrocarbon emissions are either from the diesel pilot or the methane main injection. The emissions bench measures both the methane concentration and the total hydrocarbon emissions. The difference is typically the quantity of unburned fuel that arises from incomplete combustion of the diesel pilot. From experimental testing in this work and previous the majority of hydrocarbon emissions are due to incomplete combustion of the methane injection as the total hydrocarbon concentration is typically close to the methane concentration. Methane emissions are formed from areas of the cylinder that are unable to burn. The two main processes that cause hydrocarbon emissions are over-leaning and under-mixing. In over-leaning, areas of the cylinder mix with the air to form a mixture that is too lean to support a flame. In HPDI this can be caused by an extended duration for natural gas to premix before ignition due to the inability of the fuel to auto-ignite. In under-mixing, rich areas in the core of the jet are unable to mix with the air and are too rich to burn. Areas that impinge on a cooled piston surface or the walls of the cylinder may be unable to burn due to lack of oxygen and low temperature (McTaggart-Cowan, 2006). They are only able to burn later in combustion after further mixing with air. 4 1.3 Emission Regulations and After-Treatment Strategies Regulations by the US EPA for heavy-duty CI engines have been in effect since 1974 (US EPA, 2013). PM mass (per unit of engine work) has dropped by a factor of 60, CO by a factor of 3 and NOx by a factor of 8. The European Commission has similar regulations in Europe which is currently being phased into Euro VI which adds a limit on number of particles smaller than 100nm. This is to address the growing concern of the health effects caused by these small particles. The current emission limits for steady state testing and transient testing are included in Table 1 and Table 2 below. For diesel engines the engine must comply with tests for the current stage. The World Harmonized Stationary Cycle is used for steady-state testing, and the World Harmonized Transient Cycle is used for transient tests. Further information can be found in (European Commission, 2014).  Table 1 - European Heavy Duty Emissions Limits Steady-State (adapted from www.dieselnet.com)  Table 2 - European Heavy Duty Emissions Limits Transient (adapted from www.dieselnet.com) To comply with regulations, there are three main ways of reducing emissions: ο‚· In cylinder strategies ο‚· Changes to fuel or lubrication systems ο‚· Adding exhaust after-treatment First, the production of emissions can be reduced in-cylinder through combustion strategies that minimize emissions at the exhaust valve. Technology including variable valve timing, electronic direct injection, higher injection pressures, and cooled EGR all can greatly reduce engine out emissions. A motivation for this thesis is based on this strategy. S age Date Test CO g/kW-hr NOx g/kW-hr PM g/kW-hr PN #/kW-hrEuro I 1992 ECE R-49 4.5 8.0 0.36 -Euro II 1998 ECE R-49 4.0 7.0 0.15 -Euro III 2000 ESC & ELR 2.1 5.0 0.10 -Euro IV 2005 ESC & ELR 1.5 3.5 0.02 -Euro V 2008 ESC & ELR 1.5 2.0 0.02 -Euro VI 2013 WHSC 1.5 0.40 0.01 8.0E+110.13HC g/kW-hr1.11.10.660.460.46Stage Date Test CO g/kW-hr NMHC g/kW-hr CH4 g/kW-hr NOx g/kW-hr PM g/kW-hr PN #/kW-hrEuro III 2000 ETC 5.5 0.78 1.6 5.0 0.16Euro IV 2005 ETC 4.0 0.55 1.1 3.5 0.03Euro V 2008 ETC 4.0 0.55 1.1 2.0 0.03Euro VI 2013 WHTC 4.0 0.16 0.5 0.46 0.01 6.0E+115 Second, fuels can be changed, particularly with the introduction of Ultra Low Sulfur Diesel in 2006. This has led to a reduction in both SOx emissions (which can lead to acid rain) and has reduced the amount of nucleation particles which reduces soot formation. Finally, emissions can be treated after combustion before the tailpipe. For CO and unburnt hydrocarbons a diesel oxidation catalyst can be used (DOC). A diesel oxidation catalyst serves 4 purposes (Reif, 2014): 1. Oxidize CO and unburned hydrocarbons by over 90% once the catalyst is at sufficient temperature (light-off temperature, which is approx. 170-200Β°C). 2. Oxidize unburned hydrocarbons coating PM to reduce PM mass (15-30%). 3. Oxidize NO to NO2 for downstream SCR treatment. 4. Provide increased exhaust temperatures to a Diesel Particulate Filter for regeneration (oxidation of CO and unburned hydrocarbons is an exothermic process). Diesel Particulate Filters (DPF) can be used to collect engine-out PM. DPFs use a porous ceramic honeycomb to deposit particles on the walls of the channels. Over time, soot builds up creating a pressure drop across the filter. Eventually the filter becomes full and requires regeneration. Regeneration is a process where the exhaust temperature is raised to about 600Β°C to oxidize the soot particles and clean the filter. This is done using the excess oxygen that is normally present in diesel exhaust. This usually occurs every 500km and takes 10-15 minutes but this time is very dependent on the amount of PM emitted at the operating condition. Fuel injection strategies are altered and efficiency is lowered to provide these higher exhaust temperatures. There are other strategies that can be used with DPFs, including using a catalyzed diesel particulate filter, using fuel additives (often cerium or iron) or using a continuously regenerating trap (Reif, 2014). For control of NOx, a NOx storage catalyst can be used. NO is oxidized to NO2 and then stored with a barium catalyst. When the catalyst is full the nitrogen oxides must be removed, which is done with CO and hydrocarbons. This requires engine operation at rich conditions to limit the oxygen in the exhaust for a brief period of time. The storage phase takes between 30-300s and the regeneration phase 2-10s. During this time, engine efficiency is reduced. NOx can also be removed using Selective Catalytic Reduction (SCR). A reducing agent (usually urea) reacts to form ammonia. The ammonia then reacts with NO and NO2 to form H2O and N2 via 3 possible reactions. At low temperatures the NO and NO2 must be in concentrations about 1:1. An upstream 6 oxidation catalyst or the DOC can be used to achieve these concentrations. This system can reduce NOx emissions by 90% (Reif, 2014) but requires several sensors and an injection system and control unit for operation. Overall, aftertreatment provides an added cost for an engine both in performance (efficiency) and in capital and operating cost, so elimination of exhaust emissions before the exhaust valve is preferred. 1.4 Benefits of Natural Gas as a Fuel Increasingly many countries have attempted to shift away from crude oil for both political and economic reasons. Natural gas is a promising fuel as it tends to have a less volatile price than oil (and is usually lower) and has large reserves in North America. Increased techniques in unconventional extraction (fracking) have made the fuel increasingly viable. Natural gas is a fossil fuel that is normally about 92% methane (CH4) in British Columbia but also varies geographically depending on where it is mined (Korakianitis, Namasivayam, & Crookes, 2011). The fuel can be used in either SI or CI engines. It has several unique properties that give it certain benefits. Methane has the lowest carbon-to-hydrogen ratio, which means carbon dioxide emissions are lowest per molecule. Carbon dioxide is known to be a leading contributor to global warming so this is a benefit. However, methane is a much more powerful greenhouse gas so it is critical that natural gas engines have high combustion efficiency and low CH4 emissions. Natural gas has a high octane number near 120 (Heywood, 1988) which means higher thermodynamic efficiencies can be achieved due to higher compression ratios of cylinders. This is beneficial to spark ignition engines, as the compression ratio is limited based on the fuel’s octane number to prevent knocking from occurring. For operation in a CI engine an alternative source of ignition is needed which can be a spark plug, glow plug, or pilot fuel, as methane does not auto-ignite at normal end-of-compression conditions. In HPDI, a diesel pilot is used to begin combustion and this is discussed further in section 1.5. The primary drawback of natural gas as a fuel is that the storage of the fuel limits the range of the vehicle. Natural gas can be stored in compressed tanks (CNG) or in a cryogenic liquid state (LNG). CNG has approximately one quarter of the range compared to a gasoline tank before requiring refueling for the same volume. Liquefied natural gas (LNG) increases the density by liquefying the gas, so an equivalent volume tank provides approximately three quarters the equivalent range of a gasoline tank. 7 1.5 High Pressure Direct Injection of Natural Gas (HPDIβ„’) Westport Innovations’ HPDIβ„’ system combines the benefits of a high thermal efficiency diesel engine with a clean burning fuel. The key to this system is having a mechanism to cause complete combustion of the natural gas and reliable ignition. Westport’s J36 injector achieves this by using a diesel pilot which ignites quickly but provides most of the energy from a natural gas injection. Dual concentric needles are used with two electronic actuators that can be individually electronically controlled (Figure 3).  Figure 3 - HPDI Injector (www.westport.com) Several parameters control the injector during the compression and power stroke which is depicted in Figure 4. Typically, diesel fuel is injected first and the time of injection is specified by the Pilot Start of Injection (PSOI) signal sent from the control computer and determined by the crank encoder. The amount of diesel fuel is based on the length of the pulse (PPW – Pilot Pulse Width). There is then a delay between the diesel injection and the gas injection which is specified as Pulse Separation (PSEP). Alternatively the timing of the gas injection can be independently specified using crank-angle degrees (Gas Start of Injection: GSOI) based on the encoder signal. The length of the gas injection is based on the Gas Pulse Width sent to the injector (GPW). 8  Figure 4 - Typical HPDI Injection Strategy Because of the electronic actuation of the injector and separate injections, all the parameters can be changed to affect combustion individually. Typical (β€œconventional”) HPDI combustion uses PSEP>0ms. 1.6 Previous Literature Review for Paired Jets The mixing process of fuel has a strong effect on the output emissions from combustion in CI engines. In HPDI, a gaseous fuel is injected in the combustion chamber and forms transient turbulent gaseous jets as it mixes with the air. Two parameters can be used to describe the properties of the jet that can affect the mixing phenomenon: penetration of a jet, and entrainment of air in a jet.  Previous work has shown that the penetration of a gaseous jet increases with higher momentum, more time, and lower in-cylinder density (Hill & Ouellette, 1999). The equation describing this phenomenon for a single jet is described below: π‘₯𝑑𝑖𝑝 = Ξ“(?Μ‡?πœŒπ‘Ž)1/4𝑑1/2 Where π‘₯𝑑𝑖𝑝 is the penetration length of the jet, Ξ“ is a constant, ?Μ‡? is the momentum of the jet, πœŒπ‘Ž is the ambient density (cylinder density) and 𝑑 is time. The factor most easily changed in HPDI is the momentum of the jet, which can be changed by changing gas rail pressure. Entrainment of air can be described from the work of (Ricou & Spalding, 1961). The steady state entrainment for single jets can be described by: 9 π‘šπ‘Žπ‘–π‘Ÿ+𝑓𝑒𝑒𝑙 = π‘Žπ‘₯𝑑(πœŒπ‘Žπ‘–π‘ŸπœŒπ‘“π‘’π‘’π‘™)1/2π‘šπ‘“π‘’π‘’π‘™ Where π‘šπ‘Žπ‘–π‘Ÿ+𝑓𝑒𝑒𝑙 is the total mass of the jet, π‘Ž is a constant, 𝑑 is nozzle diameter, and π‘šπ‘“π‘’π‘’π‘™ is the total mass injected from a single hole nozzle. Primarily, entrainment can be increased using a smaller nozzle diameter. The idea of using a smaller pair of jets is explored as way to maintain penetration (by maintaining momentum) and increase entrainment (by using smaller holes and/or changing the angle between the holes). Two separate jets that then merge together are intended to entrain more air initially, and then merge to increase total penetration. Faghani & Rogak (2011) modelled twin non-reacting jets to explore the penetration characteristics of twin jets and predict mixing qualities (Figure 5). They found that initially, the twin jet would penetrate slowly due to the jets being separate. However, after they merged together the penetration would increase. This showed that the penetration and mixing was also dependant on whether the jet had reached this β€˜combined point’. To compare with a single-hole baseline configuration several designs were evaluated. Either the total flow rate for a nozzle was kept constant (twin jets would have smaller diameters), or the total penetration was kept constant (total flow rate was higher for the twin jet). Several angles between the twin jets were evaluated to explore the change in penetration and entrainment as the jets merged sooner or later during an injection. Four nozzle designs formed the basis for this work (two different angles between the jets, and two different hole sizes matching either total penetration or flow rate of a single jet). The use of twin jets showed that the peak mass fraction in a jet was lower for the paired nozzle, which implied that it might be possible to lower soot emissions as they are strongly affected by the mixture fraction in a jet. 10  Figure 5 - Mixture Fraction of Single and Paired Jets (adapted from Faghani and Rogak) While the idea of twin gaseous jets will be experimentally pursued on an HPDI engine, it is of interest to compare this work with diesel engine literature. It is important to note that the fundamental process during an injection event is different when using a liquid fuel. Primarily the atomization process does not occur with gaseous fuel injections. However, some diesel studies have used a type of twin jet design with varying success. The group-hole design, which is discussed in a few works was first examined in (Zhang, Nishida, Nomura, & Ito, 2003). The group-hole design uses two jets that are closely spaced injecting vertically.  Figure 6 - Group-Hole Nozzle Design (adapted from Zhang et al.) 11 Laser absorption and scattering were used to determine vapor concentration and droplet density for four different configurations (changing flow area and angle between the jet pairs). They found that the mass of entrained air, spray volume and excess air ratio were greater for the group nozzle than the single hole nozzle. They found that the area between the two jets had improved penetration but could not conclude whether this aided fuel atomization. The group-hole design was examined in (Gao, Matsumoto, Namba, & Nishida, 2007) changing the same parameters and in a free spray environment or in an optical diesel engine. Six pairs of group-holes were used with a 10 degree angle between jets. Results showed that the group-hole nozzles could maintain penetration and increase evaporation of the fuel. Other work (Moon, Gao, Nishida, Matsumoto, & Zhang, 2008) used a similar chamber and noticed that larger divergent angles between the jets would increase the vapor mass and thus entrained more air. They suggested that there is probably a critical angle as the angle is increased. The effect on penetration was also noticed with larger divergent angles shortening the maximum penetration. Larger angles between the jets were also evaluated in (Gao, Moon, Zhang, Nishida, & Matsumoto, 2009) in a constant volume chamber that simulated diesel engine conditions. Higher included angles gave higher OH intensity, which implied better fuel/air mixing and improved combustion. They stressed that the hole configuration and alignment with the piston bowl is important for correct air utilization. From the experimental results a trade-off seems to exist in that the group-hole or paired nozzle would increase the amount of entrainment in both diesel and gaseous jets while maintaining similar penetration characteristics.  While there are limited studies using the group-hole nozzle in an actual engine, one study has shown that under specific conditions the group-hole nozzle could improve fuel economy up to 3% under throttled conditions for stoichiometric diesel conditions (intended for using a 3 way catalyst for aftertreatment), but no benefits were seen under high boost and EGR conditions (Kim, Park, Andrie, Reitz, & Sung, 2009). Another factor that needs to be considered is the effect of jet-to-jet interaction when an injector has many holes. One study noticed increased interaction as the number of holes was increased from 6 to 18 holes (Benajes & Molina, 2006). They suggested that this interaction reduces oxygen concentration and thus soot oxidation, producing higher soot levels. As the paired nozzles have 14 holes this will also be briefly discussed. 12 Numerical studies generally agreed with the experimental findings. In (Nishida, Tian, Sumoto, & Long, 2009), CFD results showed larger angles between jets would reduce the amount of evaporation of the diesel jets. They found the 10 degree angle has the longest spray tip penetration which decreased as the angle between the jets increased. Work at the University of Wisconsin-Madison (Park & Reitz, 2008) found that the group-hole nozzle had the same spray tip penetration as the conventional nozzle as each group had the same momentum. The spray showed that the group nozzle had similar spatial fuel-air mixture distribution but lower inhomogeneity, which gave better fuel consumption and lower CO and soot emissions. However the impact on emissions is still unclear as other studies they performed (Park & Reitz, 2009) showed that in some cases emissions could be improved due to spray targeting of the fuel and piston alignment, but in other configurations plume-to-plume interaction of the jets deteriorates fuel consumption and increases CO emissions. The deteriorated emissions are most consistent with the work in this thesis. For diesel literature, the group-hole nozzle has been seen to slightly reduce penetration while mainly improving mixing with the air through improved atomization depending on the angle between the jets. This has the potential to improve CO and PM emissions under certain conditions. However the different configuration of the nozzles, number of nozzles, and piston bowl orientation, can play a significant part in the mixing process and end emissions. 1.7 Objectives and Scope While there is some literature with paired-jet diesel injections, there are no studies on the application of paired gaseous injections for HPDI. Based on the work from (Faghani & Rogak, 2011) it was expected that the group nozzle may provide some benefits in increasing fuel/air mixing and reducing rich zones that promote soot and CO emissions. However, the exact results were unclear, as the CFD model did not involve real cylinder geometry (effects of impingement etc.) or simulate combustion behaviour and emission formation. This thesis aims to fill these gaps through experimental tests on a research HPDI engine. Four designs were tested and compared to a baseline single-hole nozzle. The nozzle holes were specified by either maintaining penetration or flow rate. There were also two different angles between the holes for a total of four configurations. Further details on these designs are included in section 2.1. This study will examine how engine parameters can be changed to improve the performance and emissions of the paired nozzles in an attempt to find the most beneficial conditions to use this strategy in HPDI. 13 The overall objective was to assess the emissions of the pair-hole nozzle in a single cylinder research engine using HPDI. The four specific objectives were as follows: Objective 1: Evaluate the four paired nozzle designs at five engine operating modes based on the ESC cycle to evaluate combustion performance and emissions in an HPDI engine. Objective 2: Evaluate the effects on PM and gaseous emissions by changing engine parameters to gauge sensitivity. Compare the effect of using previously developed HPDI low PM strategies by partial premixing of the fuel and late post injections with the paired nozzle. Objective 3: Compare engine out soot characteristics and morphology between the paired nozzle and the reference injector for clues into soot formation and oxidation process differences. Compare the differences between HPDI and diesel PM for these characteristics.  Objective 4: Interpret results using CFD as a tool to explain reasons for the PM and CO emissions. Show the strategies used in this work on a PM-NOx trade-off to assess the effectiveness of each strategy. Compare the emission patterns on a PM-CO graph to see if a constant ratio exists across a variety of tests on the SCRE with the paired nozzle and reference injector.  14 2. Experimental Setup 2.1 Injector Configurations The β€œreference injector” was based on a Westport J36 injector. There are some differences between the reference injector and the J36. The reference injector nozzle has 7 gas holes and 7 diesel holes while a J36 has 9 gas holes and 7 diesel holes and is not keyed allowing the two nozzles to freely rotate. In the reference injector, the gas needle was keyed to keep the diesel and gas holes aligned (interaction between the two fuels is constant for each cycle). Some previous work was done comparing emissions from a baseline J36 injector to the reference injector. The emissions were shown to be very similar. For this study the reference injector is used to compare the effect of paired jets versus a normal single jet (Figure 7).  Figure 7 - Reference Injector and Paired Nozzle Configuration. Paired Nozzle designs are referred to by hole size and angle (SHSA denotes Small-Hole-Small-Angle Nozzle) The paired nozzles are keyed so the diesel hole is centred between the two gas holes. Table 3 shows the dimensions for the paired nozzle designs and the reference injector. 15  Table 3 - Injector Nozzle Specifications Four paired nozzle designs were built changing the diameter of the holes and the angle between the gas jets based on the work of (Faghani & Rogak, 2011). The small-hole paired injectors have the same flow area as the reference injector, while the large-hole injectors have 20% extra flow area. The extra flow area is to maintain the same penetration of the jet as the reference injector. The gas-hole dimension is nominal and set by flow testing the nozzle until the specified flow is reached. The angle between the holes in the gas pair was set to be 10 or 18 degrees. The 10 degree angle should have equivalent penetration rate while increasing the width of the jet. The 18 degree angle is expected to have less penetration and a wider jet and increased air entrainment. For all the results presented, the nozzles are defined on the size of the gas holes and angle between the gas holes (i.e. small-hole-large-angle nozzle or SHSA). As will be discussed in section 3.6.1.1, the gas and diesel holes were slightly misaligned for some of the prototypes. 2.2 Single Cylinder Research Engine (SCRE) All tests performed in this work were on the SCRE (Figure 8) located in the Clean Energy Research Centre. The SCRE is a 6-cylinder four-stroke ISX 400 heavy duty diesel engine that has been modified to run on one cylinder and in an HPDI configuration. The engine uses an internal common rail and has dummy injectors installed in the 5 non-fired cylinders. The valves are bolted shut with the rocker arms removed. Additionally, the deactivated pistons are installed with only 1 piston ring.  Nozzle Reference Injector SHSA SHLA LHSA LHLANumber of Gas holes 7 14 14 14 14Nominal gas hole D (mm) 0.73 0.52 0.52 0.57 0.57Angle between gas jets (Β°) - 10 18 10 18Gas flow (LPM) 61 61 61 73 7316  Figure 8 - Single Cylinder Research Engine (SCRE): Displacement 2.5L Compression Ratio 17:1 Bore 137mm Stroke 169mm A 24V Cummins starter is used to start the engine. While the engine is running, a Baldor 40hp electric motor is controlled with a vector drive to provide additional torque to the engine. This compensates for the increased friction on the engine caused by the pistons in the deactivated cylinders. Power is absorbed with a water-cooled eddy current dynamometer with an absorption capacity of 200hp. An Ingersoll-Rand rotary screw compressor supplies dried air at a pressure of 100psi. Pulsations are damped out with a pressurized storage tank. Airflow rate is measured with a venturi and confirmed with a hot film anemometer. Airflow measurements are used in the carbon balance and measurements of EGR and EQR. Appendix C describes the methods used for airflow calibrations. A computer-controlled pneumatic regulator lowers the air pressure to a maximum of 3bar. Air then passes through an intake surge tank to dampen pressure pulses from valves and allow EGR to mix with the incoming charge air. Natural gas is supplied with a multistage screw type compressor. A coriolis mass flow meter measures the flow rate to the engine. Diesel fuel is measured on a scale and the flow rate is determined by the change in mass over time. Diesel is pressurized in two stages to obtain rail pressure. The diesel and gas rail pressure are set with a dome loaded regulator. Gas rail pressure is maintained slightly below diesel pressure using a needle valve to set a bias pressure. Gas rail pressure (GRP) is set between 20 and 30 MPa. Diesel pressure is set 0.5-2 MPa higher than gas to prevent gas from getting into the diesel lines. 17 Cylinder pressure is collected at half crank angle increments determined by a crank encoder. This is used for calculation of heat release rates and GIMEP and normally done over a 45 cycle average. Data for all other measurements are collected once per second normally for 3 minutes total. Information on post processing of data is in section 2.8. The engine and test cell setup is described elsewhere (Brown, 2008; Jones et al., 2011; Patychuk, 2013). The operating procedure is described in Appendix B. The latest version of the SCRE user’s manual is provided in the electronic appendix of this thesis. 2.3 In-Cylinder Pressure Measurement The SCRE uses a flush-mounted Kistler 6067C water-cooled piezoelectric pressure transducer. The transducer produces a charge proportional to pressure change. A charge amplifier is used to convert the charge to a voltage signal. As a small amount of charge will leak over time, an accurate measurement of starting pressure is needed to β€œpeg” the pressure. This is done using a piezo-resistive pressure transducer mounted in the intake manifold; pegging occurs when the intake valve closes (the intake manifold pressure is equal to the in cylinder pressure) at every cycle. During this work the transducer’s calibration was checked and confirmed and details on the procedure are included in Appendix D.  Because the SCRE’s 5 non-firing pistons produce a significant amount of additional friction, we characterize engine load using gross indicated mean effective pressure (GIMEP), which is defined below: 𝐺𝐼𝑀𝐸𝑃 =  ∫ 𝑃𝑑𝑉𝐸𝑉𝑂𝐼𝑉𝐢𝑉𝑠𝑀𝑒𝑝𝑑 Where 𝐸𝑉𝑂, 𝐼𝑉𝐢, and 𝑉𝑠𝑀𝑒𝑝𝑑 are 130 degrees bBDC, 144 degrees aBDC, and 2.491 liters respectively. For the tests herein GIMEP varied between 5.5 bar to 16.7 bar. See Table 7 for details. 2.4 Heat Release Rate An apparent heat release rate (HRR) can be calculated based on the first law of thermodynamics, the ideal gas law and an in-cylinder pressure trace and is included below. The apparent net heat release 𝑄𝑛 is the rate at which work is done on the piston plus the rate of change of internal energy of the cylinder contents (Heywood, 1988). 18 π‘‘π‘„π‘›π‘‘πœƒ=𝛾𝛾 βˆ’ 1π‘π‘‘π‘‰π‘‘πœƒ+1𝛾 βˆ’ 1π‘‰π‘‘π‘π‘‘πœƒ 𝑝 is the in-cylinder pressure, 𝑉 is the cylinder volume at crank angle πœƒ, and 𝛾 is the specific heat ratio (taken as 1.3 here). The HRR can be affected by having an incorrect pressure trace or 𝛾 being unsuitable. If 𝛾 is too low, the HRR will be overestimated. There is more uncertainty in the exact pressure at each crank angle due to the sensitivity constants of the piezoelectric transducer. This was calibrated before this work commenced and is discussed in Appendix D.  Setting the engine combustion timing involves matching the 50% integrated heat release (IHR) point, which is calculated by integrating the heat release rate over the cycle and dividing by the total heat released. Typically, the pilot start of injection (PSOI) is advanced or retarded to achieve this while PSEP and/or GSEP (separation between 2 consecutive gas pulses) are constant. 2.5 Gaseous Emissions Measurement The gaseous emissions are measured with an AVL CEB II Emissions Bench. A small amount of raw exhaust passes through a heated line and filter. At the bench the flow is separated. One side goes to measure hydrocarbon emissions and NOx emissions which are measured wet (water not removed), while the other side measures CO, CO2, and O2 after running the samples through a dryer. The hydrocarbon emissions are measured using a Flame Ionization Detector. A hydrogen flame ionizes sample hydrocarbons to produce a current that is proportional to the amount of carbon present. There are 2 FIDs in the emissions bench. A portion of the sample passes through a converter that removes all non-methane hydrocarbons. The second FID measures the concentration of methane. The total hydrocarbon (tHC) emissions are converted to a methane carbon equivalent by dividing the concentration by 3 (propane to methane carbon number ratio). This gives an idea of how much hydrocarbon emissions are coming from the methane alone, which is typically most of the hydrocarbon emissions. The NOx emissions are measured with a Chemiluminescent detector. Oxygen is ionized to ozone via low pressure and fed to a reactor chamber which converts NO to NO2. During this reaction light is released which is captured with a photomultiplier. As the engine gives off mostly NO but also some NO2 it is also possible to measure NO2. To measure the amounts of NOx (NO + NOx), it first is converted to NO, and the total concentration of NOx is measured. The difference in readings is the concentration of NO2. 19 CO and CO2 emissions are measured with a nondispersive infrared gas analyzer. CO and CO2 absorb different wavelengths of light and this can be measured based on motion of detector cells. Oxygen is measured based on its paramagnetic properties. A dumbbell-shaped body is repelled by a magnetic field. As oxygen enters the surroundings it is attracted to one side of the field, causing a force on the dumbbell. The amount of torque needed to keep the body stationary is proportional to the concentration of oxygen. During each test day, all analyzers are calibrated using zero and span gases at the beginning and end of the day and any drift was recorded. If the measured value deviated by 5% from the actual value, the bench needs to be recalibrated. Doing this at the beginning of a test day should find any issues before data was taken. For the data presented in this thesis, there were no issues due to drifting of the measurements. For further information see (Brown, 2008; Ladommatos & Zhao, 2001). After a majority of the tests were taken, it was found that the mass flow controller on the high range CO analyzer (2000ppm and higher) was faulty, providing a variable flow rate through the analyzer when taking data. This only affected data points operating in the COH range (over 3000ppm). For the data this affected, a correction scale was applied based on the low range CO analyzer (0-2000ppm) readings using a linear comparison between the two sensors when within the same scales to provide a better reading. The linear scale was based on 410 data points taken between September 25, 2013 and March 21, 2014. Only a small number of points were in this high CO region and needed the linear fit. After applying the correction the values were very close to the lower range readings. Further details are provided in Appendix E. 2.6 PM Emissions Measurement 2.6.1 Sampling System PM emissions sampling on the SCRE is separate from the gaseous emissions. First, the sampled exhaust needs to be diluted. This is to both simulate the dilution of exhaust exiting a tail pipe and due to the raw concentrations being too high for most instruments. Dilution air is set by a rotameter and heated to the exhaust temperature at this location (70Β°C). This is done to prevent condensation or nucleation of semi-volatile material in the exhaust (Patychuk, 2013) and enters an ejector dilutor which pulls in raw exhaust and dilutes it in a mixing chamber. More dilution air is added that cools the sample to 55Β°C and stops 20 further PM growth. The sampling line then splits to several instruments while kept at a constant temperature (Figure 9).  Figure 9 - PM Dilution System The dilution ratio is found by comparing the CO2 concentration of the diluted sample to the AVL bench CO2 concentration using a California Analytical Instruments Infrared Analyzer (Model 100). One line goes to the TEOM with a differential pressure gage measuring line pressure vs. ambient. Pressure should be slightly above ambient pressure for the tests or there are flow problems with the instruments. The last line goes through a UBC-developed thermodenuder (Patychuk, 2013) and to the DustTrak DRX (DRX). The thermodenuder heats the sampled exhaust to 200Β°C to remove the volatile component of the PM. For SMPS and TEM measurements, the denuded line was switched to go to the other instruments after data was taken with the DRX. 21 2.6.2 DustTrak DRX Aerosol Monitor The TSI DustTrak DRX (Model 8533) measures light scattering to measure PM concentrations. The calibration for the DustTrak DRX is based on Arizona Test Dust (ISO 12103-1 A1, TSI Incorporated, 2012a) which includes a large range of particle sizes. If particles have a different size and morphology, the DRX cannot be expected to read mass concentrations correctly. Previous studies (Patychuk & Rogak, 2012) have shown that if the sample has large semi-volatile fractions, the mass concentration is overestimated. This is due to the volatile component having different refractive properties. The strength of the instrument lies in that is able to take fast measurements, and is sensitive at lower concentrations than the TEOM (Maricq, 2013; TSI Incorporated, 2012b). 2.6.3 Tapered Element Oscillating Microbalance (TEOM) The Tapered Element Oscillating Microbalance (TEOM) measures engine-out PM mass. The advantage of the TEOM is a faster direct mass measurement (compared to gravimetric filters) which does not rely on empirical corrections for measurement (DustTrak DRX, SMPS). The TEOM is a hollow tube that is clamped on one end and oscillates. A filter is placed on the end and the diluted exhaust is drawn through the filter and into the tapered element. The flow is maintained at a constant rate via a mass flow controller. The element vibrates at its natural frequency and an electric circuit measures the oscillation. As PM collects on the filter, the mass increases which causes the frequency of the oscillations to decrease similar to a spring mass system. The equation below correlates the change in mass with the change in frequency: π‘‘π‘š = 𝐾0 (1π‘“π‘“π‘–π‘›π‘Žπ‘™2 βˆ’1π‘“π‘–π‘›π‘–π‘‘π‘–π‘Žπ‘™2 ) where 𝐾0 is a spring constant. The TEOM is slower than the DRX and requires higher loading. Figure 10 below shows how it is much harder to determine the mass loading for the low load condition. 22  Figure 10 - TEOM Measurements For most points data is taken over 180 seconds. For PM concentrations below 0.01 g/kW-hr typically the DustTrak DRX measurement is used for PM mass and the trend is confirmed with the TEOM. For more information see (Brown, 2008) and (Patashnick & Rupprecht, 1991). 2.6.4 Scanning Mobility Particle Spectrometer (SMPS) A TSI scanning mobility particle spectrometer (SMPS, Wang & Flagan (2007)) was used to obtain size distributions. The SMPS consists of two instruments (Figure 11), a differential mobility analyzer (DMA) and a condensate particle counter (CPC). The DMA applies a charge to the incoming aerosol particles which then enter a classifying column. An electric field is created from a voltage difference between the centre column and the wall. This only permits particles of a specific mobility to exit the column depending on the voltage. After the particles exit the DMA they are counted by a condensation particle counter (CPC). The CPC condenses butanol onto the particles in a supersaturated environment and grows them into droplets that are larger than 1ΞΌm. Then they pass through a viewing window where incident laser light scatters onto a photo detector that records the signal. The combination of the DMA with the CPC is known as an SMPS system. There are two ways of obtaining size distributions. In a stepping mode, the voltage is set to a constant value and a distribution is discretely formed. This method can take from 15-50 minutes to perform a full scan. The second method is to constantly change voltages over the range during a scan. This is possible if the voltage is changed exponentially during a scan with time. During a scan the DMA is able to obtain a size distribution in 23 minutes. In this work the second method was used and the time needed per scan was 5 minutes or less depending on the concentration of the aerosol. In this work a DMA column that was built at UBC was used. The dimensions and input details for the SMPS system are shown below in Table 4. Length 41.29 cm Inner radius 0.945 cm Outer radius 1.927 cm Sheath flow 3 liter per minute Aerosol flow 0.3 liter per minute Particle density used 1.2 g/cc Table 4 - Custom DMA Column Dimensions and SMPS Flows   Figure 11 - SMPS System For the work here, a best-fit lognormal curve was used to fit the data for a robust estimation of the distribution mode. TSI provided software (AIM) used a multiple-charge correction algorithm to account for particles that received more than 1 charge. Because the SMPS measures the equivalent mobility diameter, using this will overestimate the mass of particles due to the non-spherical morphology of the particles. 24 2.6.5 Transmission Electron Microscopy (TEM) Imaging For TEM imaging, a UBC developed thermophoretic sampler was used to collect denuded PM onto copper grids. The samples were then examined and photographed with a transmission electron microscope (Hitachi H7600). At least 30 sample images were chosen from specific areas of the grid to minimize human bias of particles. The images were taken by Ramin Dastanpour who also provided a post processing code to measure aggregate and primary particle sizes. The user selects an aggregate from an image and adjusts a filter to remove background noise (Figure 12). The program then creates a binary image and measures the size of the aggregate. For primary particle sizing, the user has to manually measure the individual particles. For this work 370-500 primary particles and 50-80 aggregates were measured per operating condition.  Figure 12 - Image Processing of an Aggregate An automatic code was also used which measured the primary particle sizes based on the pair correlation function (Dastanpour, Boone, & Rogak, 2015). 2.7 GOLD CFD Model Modelling of the paired-jet combustion was done with a Westport-developed Computational Fluid Dynamics (CFD) package called β€œGOLD”. GOLD is based on OpenFoam and uses Westport-developed chemistry and heat transfer models. The computational mesh only simulates 1/7 of the combustion chamber and one set of injector holes (out of 7). Boundaries of the mesh were set so mass leaving one side of the slice would enter at the other side. The chemical kinetic mechanisms were implemented with the trajectory-generated low-dimensional manifold method. The manifolds were generated using detailed chemical kinetic mechanisms for natural gas and heptane. A 71 species, 379 reaction modified GRI mechanism was used for natural gas. For heptane a 170 species, 1500 reaction Lawrence Livermore National Lab mechanism was used. Large-eddy-simulation (LES) was the turbulence closure model used for these simulations as it gives better 25 flow realization than other mechanisms. GOLD has been used extensively in Westport research and can predict in-cylinder pressure, heat release rates, and gaseous and PM emissions. Computation fluid dynamics (CFD) simulations reported here were done by Pooyan Kheirkhah (Kheirkhah, 2015). Additional information about the model can be found in (McTaggart-Cowan, Mann, Huang, & Wu, 2012; Munshi, McTaggart-Cowan, Huang, & Hill, 2011). 2.8 Data Processing Data are taken through LabVIEW via the β€œfast” collection and β€œslow” collection radio buttons. Data are saved in .csv files that are processed in Excel with several Visual Basic macros explained here. Filenames for slow and fast data always should be the same for the processing routines to work best. Fast data (in-cylinder pressure, intake manifold pressure, and crank encoder) are taken at half degree increments for the specified number of cycles. Once this is taken it is manually processed in a file called β€œNew Data Processing Routine2” which is saved in the processing routines folder on the test cell computer. Open the routine, click β€œProcess Pressure Data” and select the .csv file in the Fast folder on the test cell computer. The data are processed and saved in a file with the same name as the name given in LabVIEW in the Fast folder. The routine averages the in-cylinder pressure over the number of cycles and calculates GIMEP and heat release rates. For setting the engine point, the values from this spreadsheet are used for measuring 50% IHR and GIMEP. Slow data (all other values) are collected at 1 Hz for the specified duration. Normally this is done over 3 minutes to allow time for a steady reading for the TEOM. Data are saved in a .csv file in the Slow folder on the test-cell computer. During normal engine operation the operator sets all parameters based on the readings in LabVIEW except GIMEP and 50% IHR (use the New Data Processing Routine2 to check these values). It is good practice to only keep the fast data files when the values are in the correct range so there is only one file with that name. If a new file is processed with the same name the processing routine throws an error. When satisfied, the operator then collects slow data. At the end of the day, both fast and slow data need to be combined into one spreadsheet. Cut and Paste all files from the Fast folder and Slow folder into a new folder with the current date that should also contain the emissions bench calibrations for the day. Fast and Slow data will be combined using the files named β€œSlow Collection” and β€œSCRE Emissions Spreadsheet”. Copy these files to this folder and then 26 open the Slow Collection file. Be sure to always copy a new one to keep the ones in the routine folder blank. There are several macros in Slow Collection written in Visual Basic.  In the Slow Collection file press Ctrl-s to execute the main macro. Go to the folder and select all files that begin with the name β€œslow”. The slow collection file takes the data and averages values over the sample time and saves them in the tab averaged data. Different rows denote different test points. Then the values are moved to the Analysis tab. Excel will then prompt if you want to load fast data. Unless fast data is unavailable the operator should select β€œyes”. A routine runs that automatically searches for fast files that begin with the same name as the slow file and loads the data. If Excel cannot find it, Excel will prompt you to locate the file. Once all data is loaded, the macro asks if you want to move the data to the SCRE Emissions Spreadsheet file. Click yes to do so. The Emissions Spreadsheet will be populated based on data from the slow collection file. Each column represents a different test point. Once this has been done save both the emissions spreadsheet and the slow collection file. For post-processing of this data via plotting, selecting data within bounds, and creating fits MATLAB is used for the work in the thesis. The codes used were initially written by Ehsan Faghani although I made several modifications to his code. Background knowledge of MATLAB is very useful to understand and modify the codes. Initially data is loaded from all the SCRE Emissions Spreadsheets and saved into a large matrix via a code having the name β€œDataLoader*” in the title. This code will also pull information from the pressure traces from the unprocessed fast files and save images of that data for the user. Once the entire data is loaded, several matrices are saved: ο‚· DataBaseData contains all values from the SCRE spreadsheets. ο‚· DataBaseString contains string values from the spreadsheet including the data label name. ο‚· CA contains information for plotting the heat release rate crank angle values. ο‚· HRR contains the heat release rate data. ο‚· RowsTitle contains the row names and is useful when plotting different rows. Next a new matrix is created that only consists of points the user is interested in. The .m file called β€œPointFinder*” is used to create this filter. The user can select which values from the SCRE Emissions 27 Spreadsheet to keep, their names, and the range of acceptable data. A matrix called ACCdatabase is generated. In addition, an .xls file is generated to give a table of the acceptable points. This can all be easily adjusted by the user to use the appropriate parameters. For plotting, a new .m file can load the ACCdatabase and the user can select what to plot. Sample files are included in the electronic appendix. For this work a separate folder was used for each type of test. 2.9 Repeatability of Experiments With engine testing, setting a specific operating point by keeping over 5 variables constant will always have some error from the exact point. Error can be described as point-to-point (the same point set several times in the same day), and point-over-time (which can be associated with instrument drift or changing hardware on the engine). Response fits can be used to correct or predict emissions based on unintentional (but measured) variations in engine parameters. They can also provide insight in understanding the sensitivity of emissions to a specific parameter. Point-to-point variation and point-over-time variation are discussed here. Response fits of parameters are discussed in section 3.4. Point-to-point variation can be reduced by having operators follow a specific procedure setting engine conditions. If a parameter is out of the permitted range, the data is not used. The point variation is defined to be as accurate as an operator can set the condition and listed in the table below. Furthermore, two data points are repeated to quantify output emissions and performance parameters due to the point-to-point variability. Appendix A lists the points as an average of the points taken for the experiment. Normally 2 points per condition were taken each day.  Table 5 - Allowable Operator Variability for B75 The instruments and operating conditions can change over time. This can be due to instrument drift or changing engine calibrations. Drift is often small and sometimes not noticed until months after data has been taken and processed. Therefore, every day a low load zero EGR repeatability point (β€œRepeat Point”) is taken. In addition, a B75 EGR baseline condition is also taken. Over time, instrument drift can be P ra et r B75 Set Point Allowable Operator VariabilityEngine S eed (rpm) 1493 15GIMEP (bar) 16.6 0.3EGR (%) 18 1.50EQR 0.61 0.015GRP (MPa) 25.4 0.3IHR 11 1.5Diesel Mass (mg/inj) 10.5 2.528 realized if measurements are continually taken within the permitted engine parameter variation. Additionally the emissions bench is calibrated at the beginning and end of each test day and checked with zero and span gases. Any deviation above 4% from the span gas requires recalibration of the bench. Points were taken over the entire dataset usually with at least 1 acceptable B75 point per day. The paired nozzle and LHSA injector are both plotted alongside the standard J36 injector that is usually used for other engine experimentation. It is important to note the scales on each plot.  Figure 13 - Point over Time Variability: B75 Baseline J36 data is provided by Ehsan Faghani and will be published in future work 2013-09-25 2014-01-04 2014-04-15 2014-07-2600.050.10.150.20.25DateDustTrak (g/kW-hr)   RefInjLHSAJ362013-09-25 2014-01-04 2014-04-15 2014-07-26-0.100.10.20.3DateTEOM (g/kW-hr) 2013-09-25 2014-01-04 2014-04-15 2014-07-260102030DateCO (g/kW-hr)    2013-09-25 2014-01-04 2014-04-15 2014-07-261.11.21.31.41.51.6DateNOx (g/kW-hr) 2013-09-25 2014-01-04 2014-04-15 2014-07-260.40.60.81DateCH4 (g/kW-hr) 2013-09-25 2014-01-04 2014-04-15 2014-07-26170175180185190DateGISFC 29 The gaps in the data are periods without testing. It can be seen that the J36 and reference injector have similar emissions over the same time frame. However emissions do change over time for all the injectors and this is important when planning to compare experiments from different dates.  For most of this work the experiments were taken with one injector right after another with less than a week before the tests were completed. Other sources of error can be attributed to operator inexperience, however for the data above the same operator ran all the reference injector and LHSA tests for this dataset. A second operator ran the J36 tests. For this dataset the average values for the B75 baseline condition is included below with a table of coefficients of variance for each injector.  Table 6 - Coefficient of Variance B75 Points over Time As seen in Table 6, NOx and CH4 tend to have more consistent measurements than CO or PM across all injectors. The reference injector and the J36 have comparable emissions with the reference injector having slightly lower CO and CH4 overall. The reference injector and LHSA have emissions that are much greater than the bounds of the coefficient of variance. The coefficients of variance are similar for all injectors which indicate the repeatability of each injector is comparable. For the rest of this work it will be seen that the coefficients of variance are much better for individual experiments as the time aspect is minimized.   Average COV Average COV Average COVDustTrak (g/kW-hr) 0.015 44% 0.108 34% 0.018 19%TEOM (g/kW-hr) 0.034 38% 0.146 28% 0.045 32%CO (g/kW-hr)   4.64 24% 13.98 33% 5.39 16%NOx (g/kW-hr) 1.37 5% 1.21 7% 1.36 5%CH4 (g/kW-hr) 0.46 5% 0.65 17% 0.53 9%GISFC - diesel equivalent (g/kW-hr) 177 2% 184 2% 176 1%RefInj LHSA J36Point over Time Average and Coefficient of VarianceEmission30 3. Results Results are organized according to the thesis objectives. First, the paired-hole (PH) injectors are compared to the reference injector over several operating modes. Next, the best PH injector is compared to the reference injector for several engine parameter sweeps. Response surfaces are fit to PM to compare its sensitivity to each parameter at B75. Low PM strategies are then examined to evaluate the effectiveness of each strategy compared to a single injection strategy. PM characteristics including mobility diameter and morphology of the aggregates are studied for further information about the differences in soot formation between the two injectors. Last, CFD studies are run at baseline conditions to investigate the causes of differences in emissions. All points in this work are then compared on PM/NOx and PM/CO plots noting the region of operation for the paired nozzle and the reference injector. The experiments in this work are summarized in Table 7. Parentheses note the baseline B75 parameters. All points were measured twice except for the full factorial matrix which was only run once due to the increased time needed for all the data points. The baseline test conditions were defined specifically for the SCRE. They were selected to be similar to multi-cylinder engine conditions at equivalent modes but are not specifically matched to any specific multi-cylinder engine operating mode.    31  Table 7 - Summary of Experiments   Test RPMGIMEP (bar)EQR EGR (%)GRP (MPa)PPW (ms)50% IHR (Β°aTDC)PSEP (ms) InjectorsB75 1493 16.6 0.61 18 25.4 0.62 11 0.3 Ref Inj, 4 PH NozzlesA75 1222 16.3 0.61 15 24.8 0.66 16 0.4 Ref Inj, 4 PH NozzlesC75 1763 14.1 0.59 19 25.4 0.59 9 0.4 Ref Inj, 4 PH NozzlesB50 1493 11.0 0.55 19 24.5 0.63 11 0.7 Ref Inj, 4 PH NozzlesB25 1493 5.5 0.48 20 19.3 0.64 13 1.0 Ref Inj, 4 PH NozzlesTestRPMGIMEP (bar)EQR EGR (%)GRP (MPa)DIESEL (mg/inj)50% IHR (Β°aTDC)PSEP (ms) Injectors50% IHR 1493 16.6 0.61 18 25.4 10.57, 9, (11), 13, 150.3 Ref Inj, PH LHSAEGR 1493 16.6 0.610, 6, 12, (18), 2425.4 10.5 11 0.3 Ref Inj, PH LHSAEQR 1493 16.6.50, .54, .58, (.61), .7018 25.4 10.5 11 0.3 Ref Inj, PH LHSADiesel Mass 1493 16.6 0.61 18 25.46, (10.5), 15, 19.5, 22.511 0.3 Ref Inj, PH LHSATest RPMGIMEP (bar)EQR EGR (%)GRP (MPa)PPW (ms)50% IHR (Β°aTDC)PSEP (ms) Injectors1493 16.6 0.61 0 22.1 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSA1493 16.6 0.61 0 25.4 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSA1493 16.6 0.61 0 28.7 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSA1493 16.6 0.61 18 22.1 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSA1493 16.6 0.61 18 25.4 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSA1493 16.6 0.61 18 28.7 0.62 8, 11, 14 -0.6, 0.3, 1.0 Ref Inj, PH LHSATestFI (%)GSEP (ms)EQR EGR (%)GRP (MPa)PPW (ms)50% IHR (Β°aTDC)PSEP (ms) Injectors85 2 0.61 18 25.4 0.62 11 0.3 Ref Inj, PH LHSA80 1.5 0.61 18 25.4 0.62 11 0.3 Ref Inj, PH LHSAB75 PSEP Sweep - - 0.61 18 25.4 0.62 110.3, -0.4, -1.1, -1.8, -2.5Ref Inj, PH LHSAB75 SPC - - 0.7 25 25.4 0.62 11 -1.1 Ref Inj, PH LHSATestFI (%)GSEP (ms)EQR EGR (%)GRP (MPa)DIESEL (mg/inj)50% IHR (Β°aTDC)PSEP (ms) InjectorsB75 Baseline - - 0.61 18 25.4 10.5 11 0.3 Ref Inj, PH LHSAB75 0 EGR - - 0.61 0 25.4 10.5 11 0.3 Ref Inj, PH LHSAB75 SPC Optimal - - 0.70 25 25.4 10.5 11 -1.1 Ref Inj, PH LHSAB75 LPI Optimal 85 2.0 0.61 18 25.4 10.5 11 0.3 Ref Inj, PH LHSAMulti ModeLow PM StrategiesPM ComparisonParameter SweepsFull Factorial MatrixFull Factorial Matrix of EGR, GRP, IHR, PSEPB75 LPI32 3.1 Multi-Mode Tests All injectors were tested on 5 modes approximating modes of the European Stationary Cycle. Because the SCRE does not exactly match a full multi-cylinder engine, we do not use the ESC notation but instead label the modes by the speed (low to high A,B,C) and the load (approx. % of full load). For these experiments the diesel injection was set according to previous J36 diesel PPW settings. The J36, paired nozzles, and reference injector were later found to require different pulse widths for the same injection of diesel. In the parameter sweep section later in this thesis, a diesel PPW sweep was run to see the effect of higher diesel mass on emissions and results showed that a small change in pilot injected was unlikely to account for the differences. The emissions and performance raw data for the complete test matrix are included in Appendix A. A longer version is provided in the electronic appendix. 3.1.1 Emissions For these tests, the TEOM measured the total engine out PM from the exhaust and the DustTrak DRX measured the PM with the semi-volatiles removed. The multi-mode paired nozzle results are below in Figure 14. The bar is represents the mean of two points with the error bars the maximum and minimum. 33  Figure 14 - Multi-Mode PM Emissions There is an increase in PM by using all of the paired nozzles at all modes compared to the reference injector. The DRX and TEOM readings match qualitatively, except at B25 where a large fraction of the PM is removed by the thermodenuder and is difficult to measure with the DRX. Each paired nozzle seems to give the same PM output for different modes relative to others (the SHSA and LHLA are particularly bad, and the SHLA and LHSA tend to be better). Given that each operating condition in the multi-mode matrix are drastically different from each other (different GRP, PSEP, 50%IHR, speed, and load), this seems to indicate that high PM at one mode indicates high PM at all modes for all the paired nozzles. The gaseous emissions and GISFC are normalized relative to the reference injector. This was not done for the PM measurements as some values from the TEOM were negative at low PM points and would not give useful information. B75 A75 C75 B50 B2500.050.10.150.20.25TEOM (g/kW-hr)MM Emissions  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2500.050.10.15DustTrak (g/kW-hr)MM Emissions  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B250.60.70.80.911.1NOxMM Emissions  RefInjSHSASHLALHSALHLA34 Figure 15 shows the CO emissions normalized by the reference injector for each mode. For all gaseous emissions and GISFC the error bars represent the maximum and minimum value normalized by the reference injector baseline value.  Figure 15 - CO Normalized Emissions The results from the multi-mode tests show that all the paired nozzles have significantly higher CO emissions than the reference injector, especially for the SHSA injector. This could indicate that the formation of CO is affected less by targeting and impingement of the piston bowl and more by jet dynamics as several variables are changed across each mode. The penetration of the jets and air entrainment will be discussed further in section 3.6.1. The PM and CO graphs show very similar trends in emissions across modes for the paired nozzles. Consistently the SHSA nozzle has the highest emissions for both CO and PM at all modes. As CO and PM both are formed in oxygen deficient, hot zones, it is possible that the paired nozzles have increased these areas, which promotes both to form. B75 A75 C75 B50 B2505101520253035COMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2500.511.522.53CH4MM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B250.9511.051.11.15GISFCMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLA35  Figure 16 - Normalized NOx Emissions NOx emissions are also normalized by the reference injector (Figure 16). The paired nozzles exhibit lower NOx emissions at all modes by up to 20% compared to the reference injector. This reduction could be due to lower maximum in-cylinder temperature or slower combustion for the paired nozzle. These results indicate that the paired nozzles might require less EGR for optimal emissions. Operating with lower EGR should reduce PM and CO somewhat and is further tested in the EGR sweep. As with the PM and CO emissions, the relative trends for each injector is seen for NOx that each nozzle seems to perform the same at each mode relative to the others.  Figure 17 - Normalized Methane Emissions Figure 17 shows methane emissions normalized by the baseline reference injector emissions. The paired nozzles have higher methane emissions at high loads. These methane emissions are likely because of under-mixing as the high methane emissions are also seen with high PM and CO which can indicate problems mixing with the air. For the low load condition, smaller pulse widths and more time for combustion seem to have a benefit in methane emissions. However in terms of PM reduction, the B25 B75 A75 C75 B50 B2500.050.10.150.20.25TEOM (g/kW-hr)MM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2500.050.10.15DustTrak (g/kW-hr)MM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B250.60.70.80.911.1NOxMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2505101520253035COMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2500.511.522.53CH4MM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B250.9511.051.11.15GISFCMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLA36 point is less important than the high load conditions due to a higher volatile fraction being easier to eliminate in a DOC. The gross specific fuel consumption (relative to the reference injector) is shown in Figure 18.  Paired hole nozzles show higher fuel consumption in a pattern that follows the CO, PM, and methane graphs. The effect of formation of CO, NOx, and CH4 on fuel consumption is discussed in the next section.  Figure 18 - Normalized Fuel Economy 3.1.2 Effect of Emissions Formation on Fuel Consumption An analysis was done to compare the effect of the increased emissions on fuel consumption for the SHSA paired injector and the reference injector, considering the total enthalpy of the reaction using the same amount of fuel and air in the reaction. The chemical analysis assumes a constant volume combustion process with initial and final temperature of 298Β°C. The equation for this therefore becomes: π‘„π‘œπ‘’π‘‘ =βˆ‘ π‘π‘Ÿ(β„ŽΜ…π‘“Β° + β„ŽΜ… βˆ’ β„ŽΜ…Β° βˆ’ 𝑅𝑒𝑇)π‘Ÿπ‘’π‘Žπ‘π‘‘π‘Žπ‘›π‘‘π‘ βˆ’βˆ‘ 𝑁𝑝(β„ŽΜ…π‘“Β° + β„ŽΜ… βˆ’ β„ŽΜ…Β° βˆ’ 𝑅𝑒𝑇)π‘π‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘  As the process assumes a constant start and end temperature the equation is simplified to: π‘„π‘œπ‘’π‘‘ =βˆ‘ π‘π‘Ÿ(β„ŽΜ…π‘“Β° βˆ’ 𝑅𝑒𝑇298)π‘Ÿπ‘’π‘Žπ‘π‘‘π‘Žπ‘›π‘‘π‘ βˆ’βˆ‘ 𝑁𝑝(β„ŽΜ…π‘“Β° βˆ’ 𝑅𝑒𝑇298)π‘π‘Ÿπ‘œπ‘‘π‘’π‘π‘‘π‘  B75 A75 C75 B50 B2505101520253035COMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B2500.511.522.53CH4MM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLAB75 A75 C75 B50 B250.9511.051.11.15GISFCMM Emissions Normalized by Reference Injector  RefInjSHSASHLALHSALHLA37 The general equation for the reaction is: π‘Ž(𝑂2 + 3.76𝑁2) + 𝑏(𝐢𝐻4) + 𝑐(𝐢10𝐻20) + 𝑑(𝑁𝑂 + 𝐢𝑂 + 𝐢𝑂2 + 𝑂2 +𝐻2𝑂 +𝑁2) β†’ 𝑒(𝑁𝑂 + 𝐢𝑂 + 𝐢𝐻4 + 𝐢𝑂2 + 𝑂2 +𝐻2𝑂 + 𝑁2) where the coefficients π‘Ž- 𝑒 represent the number of moles of air, methane, diesel, EGR, and exhaust. For these two tests, the total GIMEP was kept constant, so the total heat released using the reference injector is compared to the SHSA paired nozzle by keeping exhaust composition constant. When the total available energy is negative that means the reaction is exothermic. The results for this model at B75 are given in Table 8. From this simplified analysis, it appears that about half of the fuel consumption penalty appears to be due to high emissions (mainly CO).   Table 8 - Loss in Energy due to Formation of Emissions Chemical Formation Penalty and Actual Penalty 3.1.3 Heat Release Rates Heat release rates (HRR) for the paired nozzle and reference injector are presented (average of 45 cycles) for mode B75 in Figure 19. The coefficient of variation of Pmax for the 45 cycles was less than 1%, which is typical of B75 for the reference injector. Recall that the SHSA injector had the highest emissions, and the LHSA injector had the lowest emissions of the paired-hole nozzles. The HRR peaks are offset by less than 0.5o. Parameter Ref Inj SHSA Paired Nozzle SHSA Fuel PenaltyQout (kJ/hr) -4.13E+05 -4.02E+05 2.6%GISFC (g/kW-hr) 182.02 193.07 5.738  Figure 19 - Paired Nozzle Heat Release Rates at Mode B75 HPDI combustion has four stages. First diesel fuel is injected and evaporates before igniting. The heat release rate goes negative as the fuel evaporates, and a small initial peak of diesel combustion is seen at approximately -18 degrees aTDC. The peak is small as the energy from the diesel is only a small fraction of the total energy (β‰ˆ5%). Premixed combustion of the natural gas is seen at about -2 to 5 crank angles. This is where the outside of the jet becomes mixed with air to a combustible stoichiometry and burns quickly. The reference injector has a higher peak than the SHSA injector which could indicate more fuel is premixed initially. The other paired injectors have approximately the same height as the reference injector. This stage may be shorter for the paired injectors as the HRR suddenly drops. It is very difficult to define the end of this phase so strong conclusions cannot be drawn from a single point. The third stage of combustion is mixing controlled, where the rate of combustion is controlled by how fast air and fuel can mix. The reference injector moves from premixed to mixing controlled gradually so -20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  SHSARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  SHLARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  LHSARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  LHLARefInj39 the exact CA where premixed combustion ends is difficult to quantify. However the paired nozzles look different. The SHSA injector drops suddenly and then exhibits a second peak which could indicate that fuel is not mixing until much later or there is a split form of combustion perhaps due to impingement with the piston. This could hint at mixing problems with this nozzle. The LHSA nozzle on the other hand drops initially and has a plateau from 5 to 10 degrees aTDC. This could also indicate problems mixing for this injector but to a lesser extent than the SHSA nozzle. The other two nozzles drop faster than the reference injector. At approximately 15 degrees the HRR for all 3 injectors becomes the same as combustion is entirely mixing controlled.  When comparing to another high load point (A75), the same shapes are also noticed for the paired nozzles in Figure 20. Cyclic variability of Pmax was less than 1% for these 45 cycles.  Figure 20 - Paired Nozzle Heat Release Rates at Mode A75 It is difficult to connect differences in HRR between the injectors to emissions without having optical access to the combustion chamber. However it appears there is a fundamental difference in HRR for the paired nozzles that could be related to the higher emissions seen from these nozzles. It also appears that -20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  SHSARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  SHLARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  LHSARefInj-20 0 20 40050100150200CA [deg]HRR [kJ/m3/deg]  LHLARefInj40 changing modes does not affect the relative shapes of the HRR. In section 3.6.1 this will be further examined in the CFD simulations based on these same operating points. 3.1.4 Discussion of Multi-Mode Results At all modes tested for all pair-hole nozzles, PM and CO emissions were much higher than the reference injector. Fuel consumption is also higher, partly due to these emissions. Because methane emissions are also higher this could suggest that increased fuel-rich regions are being formed. NOx is slightly lower, but not enough that changes in timing or EGR could bring PM emissions to the PM levels of the reference injector.  The heat release rates for the paired nozzles are considerably different from the reference injector. The small-hole small-angle nozzle exhibits a double peak while the large-hole small-angle nozzle has a plateau between the premixed and fully mixing controlled areas of combustion. The other two nozzles have the same peaks but drop off faster, potentially indicating less mixing controlled combustion. Due to the high PM and CO emissions from the SHSA nozzle, the double peak we see in the HRR may indicate rich unmixed areas of the combustion chamber are not able to burn completely. 3.2 Emissions Characterization with Parameter Sweeps Parameter sweeps changing 50%IHR, EGR %, EQR, and diesel pilot mass were studied for the reference and LHSA injectors at mode B75. The results from this study are used with the full factorial matrix in section 3.4 to understand how the sensitivities of various parameters compare with the two injectors. Raw data for all these sweeps are included in Appendix A. 3.2.1 Timing Sweep  The results for the timing sweep are included in Figure 21. These are raw values and not normalized. 41  Figure 21 - 50% IHR Sweep Emissions (7-15Β°aTDC) 1493 RPM 16.6bar GIMEP 0.61EQR 18%EGR 25.4MPa GRP 10.5mg/inj. Diesel 0.3ms PSEP Over the timing sweep, the same trends are observed as have been with traditional HPDI. As combustion is advanced, NOx increases due to increased peak temperature in the cylinder. CO and PM decrease due to increased time available for oxidation and higher combustion temperatures. The parabolic trend in PM has been frequently observed with this engine (McTaggart-Cowan & Rogak, 2004; Patychuk, 2013). The reference injector shows a slight parabola with the DustTrak reading; however this is not seen on the TEOM. As combustion is retarded, NOx decreases due to a drop in peak combustion temperature. PM increases at mid-timing for the paired nozzle. PM emissions decrease with later timings due to decreased formation rate caused by lower in-cylinder temperatures. NOx emissions also decrease due to lower 6 8 10 12 14 1600.050.10.150.2Timing Sweep Emissions50% IHRDustTrak (g/kW-hr)  RefInjLHSA6 8 10 12 14 1600.050.10.150.2Timing Sweep Emissions50% IHRTEOM (g/kW-hr)6 8 10 12 14 16051015Timing Sweep Emissions50% IHRCO (g/kW-hr)   6 8 10 12 14 160.511.522.5Timing Sweep Emissions50% IHRNOx (g/kW-hr)6 8 10 12 14 160.40.50.60.7Timing Sweep Emissions50% IHRCH4 (g/kW-hr)6 8 10 12 14 16175180185190Timing Sweep Emissions50% IHRGISFC (g/kW-hr)42 peak cylinder temperature. Earlier combustion provides greater fuel efficiency; later timing reduces efficiency due to less efficient peak pressure timing. The paired nozzle and reference injector show similar trends for the timing sweep. At all cases there is an increased shift for the PM, CO, methane emissions and fuel consumption. There is no β€˜sweet spot’ where PM or CO emissions between the paired nozzle and reference injector become comparable or beneficial. NOx remains almost identical for both injectors over the entire timing sweep. This shows that the local temperatures are approximately the same for both injectors. 3.2.2 Exhaust Gas Recirculation (EGR) Sweep The results from the EGR sweep are below in Figure 22. 43  Figure 22 - EGR Sweep Emissions (0-24%) 1493 RPM 16.6bar GIMEP 0.61EQR 25.4MPa GRP 10.5mg/inj. Diesel 0.3ms PSEP 11Β° 50%IHR EGR increases PM and CO as oxygen is displaced with an inert gas. This is due to slowing soot oxidation mechanisms with a lower concentration of oxygen and lower temperatures which reduces the reaction rate for soot oxidation (Tree & Svensson, 2007). For the paired nozzle, PM at the baseline condition (18% EGR) can be reduced significantly when EGR is eliminated however this does not reduce the gap between the two injectors at the same conditions. CO emissions can also be reduced with the elimination of EGR but not to near the reference injector levels. 0 5 10 15 20 2500.050.10.150.2EGR %DustTrak (g/kW-hr)EGR Sweep  RefInjLHSA0 5 10 15 20 2500.050.10.150.2EGR %TEOM (g/kW-hr)EGR Sweep0 5 10 15 20 2505101520EGR %CO (g/kW-hr)   EGR Sweep0 5 10 15 20 25012345EGR %NOx (g/kW-hr)EGR Sweep0 5 10 15 20 250.20.40.60.81EGR %CH4 (g/kW-hr)EGR Sweep0 5 10 15 20 25170175180185190EGR %GISFC (g/kW-hr)EGR Sweep44 EGR shows that NOx emissions respond similarly for both injectors. As EGR is added, combustion temperature drops which reduces NOx emissions significantly. At higher levels of EGR more methane emissions are noticed for the paired nozzle, which could indicate increased over-leaning zones in the cylinder. This has implications in the entrainment aspect of the jets and could show that the paired nozzle has improved fuel and air mixing characteristics, but we are unable to burn this fuel as it is too lean. At low EGR fraction methane emissions are comparable. This is all similar to diesel and HPDI literature where EGR moves emissions along a PM/NOx trade-off curve (Lee, Zhu, & Song, 2008). While EGR can improve emissions for the paired nozzle, it cannot reduce them below reference injector levels. 3.2.3 Equivalence Ratio (EQR) Sweep The results from the EQR sweep are below in Figure 23. Due to significant quantities of EGR in the intake, the equivalence ratio is based on the amount of oxygen present in the charge as there is oxygen present in EGR for lean stoichiometry. For the EQR sweep the baseline point was 0.61. As EQR increases the intake oxygen mass fraction will decrease with constant EGR. 45  Figure 23 - EQR Sweep Emissions (0.50-0.7) 1493 RPM 16.6bar GIMEP 18% EGR 25.4MPa GRP 10.5mg/inj. Diesel 11Β° 50%IHR 0.3ms PSEP Lowering EQR has a large benefit for both injectors. At the low EQR points CO and PM emissions are reduced by 50% for both injectors which show that the same benefits are possible at these conditions compared to the baseline EQR for both injectors. For gaseous emissions, NOx drops as the global equivalence ratio becomes richer. This is due to a drop in oxygen concentration that reduces NOx formation. Methane emissions appear to drop slightly for the reference injector, while rising for the paired nozzle. It is expected that higher EQR would reduce the methane emissions. This is due to higher cylinder temperatures and fewer regions in the cylinder that are able to become too lean to burn due to excess oxygen. 0.5 0.55 0.6 0.65 0.7 0.7500.050.10.150.2Oxygen equivalence RatioDustTrak (g/kW-hr)EQR Sweep  RefInjLHSA0.5 0.55 0.6 0.65 0.7 0.7500.10.20.30.4Oxygen equivalence RatioTEOM (g/kW-hr)EQR Sweep0.5 0.55 0.6 0.65 0.7 0.750102030Oxygen equivalence RatioCO (g/kW-hr)   EQR Sweep0.5 0.55 0.6 0.65 0.7 0.750.511.52Oxygen equivalence RatioNOx (g/kW-hr)EQR Sweep0.5 0.55 0.6 0.65 0.7 0.750.20.40.60.81Oxygen equivalence RatioCH4 (g/kW-hr)EQR Sweep0.5 0.55 0.6 0.65 0.7 0.75160170180190Oxygen equivalence RatioGISFC (g/kW-hr)EQR Sweep46 With the paired nozzle the spread in data is large and further tests would be needed to confirm this result for the paired nozzle due to the variability in the methane emissions. The paired-hole nozzle shows similar sensitivity to changing global EQR for PM and CO emissions. Despite these improvements however, the total emissions are still higher and the ability to provide higher EQR depends on the ability to provide higher boost pressure. For an actual vehicle, this would likely add a parasitic loss in fuel efficiency to drive the compressor, which is not shown in a plot of GISFC. 3.2.4 Diesel Mass Sweep The results for the diesel sweep are below in Figure 24. This test was done to see the sensitivity of the paired nozzle on emissions with different diesel quantities. It was also a follow-up to the multi-mode tests where different injectors had slightly different quantities of diesel injected per cycle to show that this was not the key cause for the high emissions seen from the paired nozzle injectors.  47  Figure 24 - Diesel Mass Sweep Emissions (6-22.5 mg/inj) 1493 RPM 16.6bar GIMEP 0.61 EQR 18% EGR  25.4MPa GRP 11Β° 50%IHR 0.3ms PSEP There is an increase in PM for both injectors as the diesel pilot is increased. However when comparing the sensitivity of the change versus the baseline conditions, the increase is the same for both injectors. This shows that a small increase in diesel is not the reason for the high emissions for the multi-mode tests. Other emissions (NOx CH4 and GISFC) remained relatively constant over this test. 3.2.5 Parameter Sweep Discussion For all the parameter sweeps, the emissions for both injectors change similarly to each other. Higher EQR is able to reduce emissions significantly for the paired nozzle, but at the same conditions it still has 5 10 15 20 2500.10.20.30.4Diesel injection mass (mg/inj)DustTrak (g/kW-hr)Diesel Sweep  RefInjLHSA5 10 15 20 2500.10.20.30.4Diesel injection mass (mg/inj)TEOM (g/kW-hr)Diesel Sweep5 10 15 20 250102030Diesel injection mass (mg/inj)CO (g/kW-hr)   Diesel Sweep5 10 15 20 2511.21.41.61.8Diesel injection mass (mg/inj)NOx (g/kW-hr)Diesel Sweep5 10 15 20 250.40.60.81Diesel injection mass (mg/inj)CH4 (g/kW-hr)Diesel Sweep5 10 15 20 25170175180185190Diesel injection mass (mg/inj)GISFC (g/kW-hr)Diesel Sweep48 higher emissions than the reference nozzle. The other sweeps (IHR, EGR, and diesel) showed the same trends for both injectors. These results indicate that individual engine parameter sweeps are unlikely to find a β€˜sweet spot’ that will optimize the paired nozzle to have lower emissions than the reference nozzle. 3.3 Low PM Strategies Previous research at UBC and Westport has found several strategies that are able to reduce PM significantly, while minimizing any influence on other emissions. To attempt to reduce PM from the paired-hole nozzle, three of these strategies are studied: Late Post Injection (LPI), negative PSEP, and Slightly Premixed Combustion (SPC), a variant of negative PSEP. 3.3.1 Late Post Injection Splitting the gas injection into two pulses has several effects. The diesel literature has shown that post injections have the potential to reduce PM emissions but carry a fuel consumption penalty (β‰ˆ7%) (O’Connor & Musculus, 2013). In recent HPDI work, the application of post injections has been studied (Faghani, Patychuk, McTaggart-Cowan, & Rogak, 2013; Faghani, 2015) and post injections with a sufficient delay between the two injections are able to reduce PM with only a slight fuel penalty (β‰ˆ1%) based on experiments using the SCRE. This strategy is known as Late Post Injection (LPI), where the delay between 2 gas injections is at least 1ms. The LPI conditions tested here are optimized cases based on the previous study for the J36 injector where 10-15% of the fuel is present in the second pulse which is 1.5 to 2ms after the first pulse. The LPI point is set based on total fuel needed for an operating condition. First, a single injection reference point is set to obtain the total amount of gas required. Two additional parameters are needed for a split injection (Figure 25). GSEP is gas pulse separation, which is the time between the end of the first gas pulse and start of the second pulse. The SCRE has the capability to also set injections on crank angle location but for this study GSEP and PSEP and 50%IHR set this timing. The other parameter is FI, which is the percentage of total natural gas fuel in the first injection. For the two tests this was either 80% or 85%. The second injection is trimmed to match the engine load. The total quantity of fuel in the second injection will be close to 15% when the load is reached. 49  Figure 25 - Late Post Injection Strategy The emissions in Figure 26 are presented as bars that have been normalized by the values for single-injection operation for each injector. This shows the relative effectiveness of the LPI strategy. The coefficients of variance for these points are then included in Table 9. 50  Figure 26 - LPI Emissions normalized by B75 single injection strategy FI85 GSEP 2.0ms, FI80 GSEP1.5ms 1493 RPM 16.6bar GIMEP 0.61 EQR 18% EGR 25.4MPa GRP 10.5mg/inj. Diesel 11Β° 50%IHR 0.3ms PSEP   Table 9 - Coefficients of Variation LPI FI85GSEP2.0 FI80GSEP1.500.20.40.6PM DRXLPI Emissions Normalized  Ref InjLHSAFI85GSEP2.0 FI80GSEP1.500.20.40.60.8COLPI Emissions Normalized  Ref InjLHSAFI85GSEP2.0 FI80GSEP1.50.60.811.2CH4LPI Emissions Normalized  Ref InjLHSAFI85GSEP2.0 FI80GSEP1.500.20.40.60.8PM TEOMLPI Emissions Normalized  Ref InjLHSAFI85GSEP2.0 FI80GSEP1.50.60.811.2NOxLPI Emissions Normalized  Ref InjLHSAFI85GSEP2.0 FI80GSEP1.50.60.811.2GISFCLPI Emissions Normalized  Ref InjLHSACo ffic e of VariationInjector Ref Inj LHSA Ref Inj LHSADustTrak (g/kW-hr) 31% 27% 38% 5%TEOM (g/kW-hr) 30% 17% 34% 14%CO (g/kW-hr) 21% 12% 34% 2%CH4 (g/kW-hr) 6% 9% 1% 5%NOx (g/kW-hr) 6% 8% 3% 3%GISFC - diesel equivalent (g/kW-hr) 0.7% 0.7% 0.2% 0.3%FI 85 GSEP 2.0 FI 80 GSEP 1.551 The results of this experiment show that the LPI strategy provides the same benefit (i.e., % reduction) for both injectors. The FI85 GSEP2.0 point has better emissions of the LPI points. For both injectors, PM and CO is reduced by over 50%, CH4 is reduced slightly, and NOx is approximately constant. From our measurements GISFC improved slightly for the paired nozzle. The emissions reduction for late post injection is identical for the paired nozzle as the reference injector, despite having much higher PM and CO emissions to start with. The coefficients of variation for the tests are consistent with the B75 baseline variation, to show that this emissions benefit is consistent with normal operating conditions. Both the CO and PM (DRX and TEOM) emissions have very high COV, indicating that the variations are not likely caused by a faulty instrument. 3.3.2 Negative PSEP Previous work has been done adjusting the relative timing of the diesel and natural gas injections in HPDI. This was first done in (McTaggart-Cowan, Bushe, & Rogak, 2003). By injecting natural gas before the diesel (Figure 27), the gas is given more time to mix with the air before being ignited by the diesel. Under certain circumstances this can greatly reduce PM emissions as the local equivalence ratios are reduced when the fuel is premixed. However there is a trade-off where more premixing increases NOx and methane emissions due to over-leaning, oxygen availability and high local temperatures.  Figure 27 - Negative PSEP Injection Strategy On the SCRE, the diesel pilot is typically injected 0.3-1ms before the start of the gas injection (see Figure 4). PSEP is defined as the start of gas injection minus the end of the diesel injection. When there is overlap or if the gas injection is before the diesel the result is negative (negative PSEP). The paired-hole nozzles were designed to increase the amount of air that is entrained into the jet and by adjusting this delay might be more effective at reducing PM for the paired nozzle. For this experiment the PSEP was varied from 0.3ms (baseline) to -2.5ms (very premixed). As the PM levels are very low for some of these tests, the DustTrak measurements are more meaningful than the TEOM measurements. 52  Figure 28 - Negative PSEP PM Emissions (-2.5ms β†’ +0.3ms) 1493 RPM 16.6bar GIMEP 0.61 EQR 18% EGR 25.4MPa GRP 10.5mg/inj. Diesel 11Β° 50%IHR PM is greatly reduced as the amount of time for premixing to occur is increased. This is most evident between -0.4 and -1.1ms. PM levels for the paired nozzle become comparable to the reference injector once PSEP is -1.1ms. After this point PM emissions reach a minimum value where increased ignition delay gives no additional benefit. Due to the majority of PM being eliminated by -1.1ms, this shows that this early part of combustion greatly influences the amount of PM formed. This optimal timing is the same for both injectors. -3 -2 -1 0 100.050.10.15PSEP (ms)DustTrak (g/kW-hr)PSEP Sweep (Negative PSEP)  RefInjLHSA-3 -2 -1 0 100.050.10.150.2PSEP (ms)TEOM (g/kW-hr)PSEP Sweep (Negative PSEP)-3 -2 -1 0 105101520PSEP (ms)CO (g/kW-hr)   PSEP Sweep (Negative PSEP)-3 -2 -1 0 112345PSEP (ms)NOx (g/kW-hr)PSEP Sweep (Negative PSEP)-3 -2 -1 0 10123PSEP (ms)CH4 (g/kW-hr)PSEP Sweep (Negative PSEP)-3 -2 -1 0 1160170180190PSEP (ms)GISFC (g/kW-hr)PSEP Sweep (Negative PSEP)53 As PSEP becomes more negative, NOx emissions increase due to a larger diffusion flame area and high local temperatures and oxygen availability but remain comparable with the paired nozzle. Methane increases due to increased over-leaned areas of the cylinder that are unable to ignite. For the paired nozzle this region seems to be greater, so more mixing may be occurring. This might be because of increased air entrainment of the jets. Slightly more mixing could also be occurring due to a smaller GPW for the paired nozzle which would give more time for the fuel to mix before diesel is added. This is due to the LHSA injector having a larger flow area than the reference injector so a shorter pulse is needed to inject the same quantity of fuel. Fuel consumption decreases due to more efficient combustion and more heat released earlier in the cycle. There are several other drawbacks with negative PSEP. Due to the increased amount of fuel and air available to burn when combustion begins, the pressure rise rate is much higher when operating with negative PSEP. This has the effect of increased wear on the engine and the potential of engine knock occurring. 3.3.3 Slightly Premixed Combustion Faghani (2015) developed an operating method named Slightly Premixed Combustion (SPC). The optimal point for the J36 is the basis of this experiment. Details on SPC, including application to a multi-cylinder engine, is also reported by (Patychuk, 2013). SPC uses the negative PSEP concept of premixing the fuel while changing the in-cylinder composition to optimal values (25% EGR .70EQR). Due to low PM emissions, higher levels of EGR can be used to reduce NOx. Higher methane emissions are then reduced by an increase in EQR. SPC results are presented in Figure 29. In this test a PSEP sweep was done at high EGR and EQR, however only the best point is presented here which is the same timing as the optimum negative PSEP point (-1.1ms). For this case the best SPC and negative PSEP point (18%EGR, 0.61EQR) were compared by normalizing emissions against the baseline single injection strategy for each injector. The coefficient of variation for the emissions is included below in Table 10. 54  Figure 29 - SPC Emissions normalized by B75 single injection strategy B75NegPSEP=0.61 EQR 18% EGR, SPCNegPSEP=0.70 EQR 25% EGR 1493 RPM 16.6bar GIMEP 25.4MPa GRP 10.5mg/inj. Diesel  11Β° 50%IHR -1.1ms PSEP   Table 10 - Coefficients of Variation SPC B75NegPSEP SPCNegPSEP00.10.20.30.4PM DRXSPC Emissions Normalized  RefInjLHSAB75NegPSEP SPCNegPSEP0.40.60.811.2COSPC Emissions Normalized  RefInjLHSAB75NegPSEP SPCNegPSEP1.21.41.61.822.2CH4SPC Emissions Normalized  RefInjLHSAB75NegPSEP S CNegPSEP00.10.20.30.4PM TEOMSPC Emissions Normalized  RefInjLHSAB75NegPSEP S CNegPSEP0.511.522.5NOxSPC Emissions Normalized  RefInjLHSAB75NegPSEP S CNegPSEP0.60.811.2GISFCSPC Emissions Normalized  RefInjLHSACo ffic e t of VariationInjector Ref Inj LHSA Ref Inj LHSADustTrak (g/kW-hr) 7% 28% 5% 25%TEOM (g/kW-hr) 147% 25% 587% 78%CO (g/kW-hr) 13% 4% 19% 12%CH4 (g/kW-hr) 9% 9% 18% 4%NOx (g/kW-hr) 5% 1% 10% 5%GISFC - diesel equivalent (g/kW-hr) 0.2% 1.0% 0.9% 2.1%B75 Neg PSEP SPC Neg PSEP55 Overall, the SPC strategy can reduce PM up to 90%, and keep CO and NOx levels the same as the baseline single injection strategy. GISFC is unchanged. Methane emissions increase by 60% for both injectors. For the J36 these EQR and EGR set points would keep methane close to the single injection strategy. For these injectors, methane emissions are higher than the baseline case, possibly because EQR and EGR values are not optimal. However both injectors have the same response to SPC. The repeatability of these points is similar to B75 mode, which shows that the emissions reduction is a consistent strategy. As the PM levels are very low, the TEOM is not an accurate instrument at these conditions. 3.3.4 Low PM Strategies Discussion Based on the engine parameter sweeps and low PM strategies studied, we cannot improve PM emissions on the paired nozzle to beat the reference injector. The LPI strategy can be used to greatly improve emissions from the paired nozzle as with the reference injector. Negative PSEP can reduce PM emissions to almost reference injector levels. SPC can keep NOx at baseline levels with an increase in methane emissions. These strategies indicate that these strategies work for both injectors and could potentially reduce emissions for the other paired nozzles that were not tested in this experiment as the % reduction is the same. 3.4 Response Surfaces of Experiments Response surfaces were generated based on data from the parameter sweeps, the negative PSEP sweep, and from the full factorial matrix to obtain a large dataset based around the B75 baseline condition. The goal was to show the sensitivity of emissions to the various engine parameters used to set the point (diesel pilot mass, gas rail pressure, EQR, EGR, PSEP and IHR timing). The response surface uses the same concept as (Laforet, 2009; Patychuk, 2013) based on a first order system (Myers, Montgomery, & Anderson-Cook, 2009). The general equation is below: 𝑦 = 𝛽0 + 𝛽1π‘₯1 + 𝛽2π‘₯2 + 𝛽3π‘₯3 + 𝛽4π‘₯4 +β‹―+ 𝛽𝑖π‘₯𝑖 + πœ€ Where 𝑦 is the dependant variable we are trying to fit, 𝛽𝑖 are the regression coefficients, π‘₯𝑖 are the input parameters (diesel, GRP etc.), and πœ€ is the error from the fit. The dependant variables were the emissions, primarily the PM emissions. All parameters were assumed to have linear responses with the emissions except for PSEP and 50% IHR which assumed second order dependence. This is due to the parabolic response seen on these parameter sweeps for PM emissions. The model is evaluated based on 56 the R2 value and its response over the dataset. Evaluating this model involves using coded variables instead of the natural input variables (such as EQR). The coded variables have mean of 0 and range that scales with the range of the parameter being fit (maximum EQR has coded variable of +1 and minimum has coded variable of -1). This is widely used in fitting linear regression models (Myers et al., 2009). To calculate the coefficients matrix we use: 𝐡 = (𝑋′𝑋)βˆ’1𝑋′𝑦 Where 𝑋 and 𝑦 are the input and output matrices that have been scaled. 𝐡 is the coefficients matrix for the scaled parameters. Finally, the model is described in: ?Μ‚? = 𝑋𝐡 Where ?Μ‚? are the output values of the fit. To obtain the values for 𝛽 this result needs to be unscaled to have an equation that is more easily manipulated. The equation for CO for the LHSA nozzle is provided here for reference and the other equations are included in the electronic appendix: 𝐢𝑂 = 7.65 βˆ’ 2.65 βˆ— 𝐷𝑖𝑒𝑠𝑒𝑙 + 1.16 βˆ— 𝐺𝑅𝑃 βˆ’ 20.26 βˆ— 𝐸𝑄𝑅 βˆ’ 0.0023 βˆ— 𝐸𝐺𝑅 βˆ’ 3.20 βˆ— 𝑃𝑆𝐸𝑃 βˆ’ 0.74βˆ— 𝐼𝐻𝑅 βˆ’ 0.079 βˆ— 𝐼𝐻𝑅2 βˆ’ 0.15 βˆ— 𝑃𝑆𝐸𝑃2 + 0.033 βˆ— 𝐷𝑖𝑒𝑠𝑒𝑙 βˆ— 𝑃𝑆𝐸𝑃 βˆ’ 0.065 βˆ— πΊπ‘…π‘ƒβˆ— 𝑃𝑆𝐸𝑃 + 10.33 βˆ— 𝐸𝑄𝑅 βˆ— 𝑃𝑆𝐸𝑃 + 0.21 βˆ— 𝐸𝐺𝑅 βˆ— 𝑃𝑆𝐸𝑃 βˆ’ 0.195 βˆ— 𝑃𝑆𝐸𝑃 βˆ— 𝐼𝐻𝑅 + 0.26βˆ— 𝐷𝑖𝑒𝑠𝑒𝑙 βˆ— 𝐼𝐻𝑅 βˆ’ 0.146 βˆ— 𝐺𝑅𝑃 βˆ— 𝐼𝐻𝑅 + 7.09 βˆ— 𝐸𝑄𝑅 βˆ— 𝐼𝐻𝑅 + 0.019 βˆ— 𝐸𝐺𝑅 βˆ— 𝐼𝐻𝑅 The units for CO are in g/kW-hr, and the input units are from setting the point (i.e. for diesel it is mg/inj., GRP is MPa, see Table 7). One model was created for each injector to show differences in sensitivity for the PM and gaseous emissions. The R2 values for each fit are below in Table 11. The fit for the LHSA paired nozzle and studentized residual plot for the DRX are below in Figure 30. As the majority of the points are within 2 standard deviations this shows the fit is valid over the range of experimental data and the fit consistently predicts the value for the entire range of the experimental data. 57   Figure 30 - Response Fit and Residual Plot for LHSA DRX   Table 11 - R2 values for predicting different output emissions Overall, the surface predicts NOx and CH4 the best, while PM and CO emissions have lower R2 values. The response surfaces are able to do a good job predicting the emissions over the range tested. To analyze the sensitivity of the 2 injectors at B75 baseline, the 6 parameters were adjusted to show the dependency of each individual parameter and when changing 2 parameters together. Table 12 shows the percent emission change when the input variables are increased by 10% of the baseline B75 value (Table 5). The response surface accounts for changes due to interaction of both variables. 0 0.1 0.2 0.3 0.4-0.100.10.20.3Experimental DataResponse FitLHSA DRX Fit r2=0.746  Response Fit1:10.05 0.1 .15 2-2-1012Experimental DataStudentized ResidualStudentized Residual Plot DRX LHSARefInj LHSADRX 0.646 0.746TEOM 0.71 0.81CO 0.823 0.72NOX0.97 0.967CH4 0.951 0.965InjectorFitted Variable:R2 Values58  Table 12 - DRX and CO Sensitivities at B75 relative to 2 parameters changing For the sensitivity table GRP consistently has benefits when it is increased regardless of other parameters (except when combined with EQR). This was only changed in the full factorial experiments. EQR seems to have the strongest influence on PM as well. The reading for the reference injector is more sensitive due to the B75 baseline reading being lower and small changes affect the value more. Generally the CO sensitivity follows the PM sensitivity. Response surfaces are most accurate when a large quantity of data exists which is why the B75 speed and load combination was chosen. Using these surfaces to extrapolate to other modes or to other injectors would be difficult to have confidence in the results. Additionally, a robust fit is based on the operator’s knowledge of the changing variables. This fit was simply based on 6 parameters and was not investigated whether a simpler fit would work. This could be investigated in future work.  Overall, the response surfaces are able to predict emissions but normally require significant time to take sufficient experimental data. Further work could attempt to combine data sets and apply them to the SCRE directly to map output emissions with these parameters. 3.5 PM Characteristics Comparison for LHSA and Reference Injector PM characteristics resulting from the paired-hole nozzle and reference injector were compared at four engine operating conditions. The characteristics measured included mobility diameter, aggregate diameter, primary particle diameter, and fractal dimension. The points represented the baseline operating condition, the 0 EGR condition, and the two optimized low PM strategies explored earlier (LPI and SPC) all based at B75. Diesel GRP EQR EGR PSEP IHR Diesel GRP EQR EGR PSEP IHRDiesel 0% 9%GRP -42% -43% -16% -25%EQR 76% 33% 76% 26% -9% 17%EGR 8% -35% 83% 8% 17% -18% 24% 7%PSEP 4% -40% 80% 11% 3% 11% -23% 18% 9% 2%IHR -4% -49% 46% 1% -3% -7% 12% -30% 40% 8% 2% 0%Diesel GRP EQR EGR PSEP IHR Diesel GRP EQR EGR PSEP IHRDiesel -1% 2%GRP -24% -23% -7% -10%EQR 61% 40% 62% 32% 21% 30%EGR 4% -18% 67% 5% 6% -5% 34% 4%PSEP 0% -22% 64% 6% 1% 3% -9% 31% 5% 1%IHR -6% -29% 72% -1% -5% -6% -1% -18% 28% -1% -5% -5%ParameterReference Injector LHSAEffect on CO with 10% Change in input ParameterReference Injector LHSAEffect on DRX reading with 10% increase in input ParameterParameter59 Samples for TEM were collected using the thermophoretic sampler. Particles are selected from specific areas of the TEM grid to minimize human bias in selection of particles. For each case at least 30 images, comprising at least 40 aggregates and 350 primary particles are measured. For the SMPS scans, 3 sequential scans were taken and averaged. The results of the analysis are included in Table 13. SMPS results are included in Appendix A. 3.5.1 Sizing of Primary Particles and Aggregates The aggregates form lognormal size distributions which can be seen in the SMPS scans. Because of this type of distribution, mean diameter will be larger than the median. Therefore a geometric mean of the TEM-measured projected-area diameter is used to evaluate the sizes of the particles, and is compared with SMPS mobility measurements (Figure 31). The error bars are the geometric standard deviation of the points.  Figure 31 - TEM vs. SMPS Aggregate Sizing Error bars are Geometric Standard Deviation of Measurements The SMPS scans tend to predict slightly smaller size distributions. This has been seen before (Patychuk, 2013). The processing code for TEM aggregate sizing is semi-automatic and can be adjusted by the operator based on the focus of the image, so some error can be attributed due to blurry images or to 0 50 100 150050100150SMPS Geometric Mean (nm)TEM Geometric mean (nm)TEM vs. SMPS Aggregate Sizing  B75 Baseline RefInjB75 Baseline LHSAB75 0EGR RefInjB75 0EGR LHSALPI RefInjLPI LHSASPC RefInjSPC LHSA1:160 the angle of the aggregate on the grid (possibly aggregates align with the long axis along the grid). Table 13 below shows an overview of the SMPS and TEM data.  Table 13 - Morphology of PM at B75 The standard error for 𝑑𝑝 was less than 0.5nm and the standard error for dA was less than 9nm for all TEM measurements above. The SMPS concentrations were corrected for dilution which was normally 10:1. For these conditions the primary particle diameter was measured manually by the operator. During this study an automatic code (Dastanpour, Boone, & Rogak, 2015) was also used to size primary particles. The automatic results were then compared to the manual results. The automatic code operates based on the pair correlation function. Some preliminary validation work with this code has shown for branched aggregates, the primary particle sizing matches closely with the manual sizing. Overall, the manual measurements are slightly higher than the automatic results. Some of this can be attributed to operator inexperience selecting the correct boundaries of the primary particles. Previous work has shown that manual measurement can introduce errors if the operator is inexperienced with TEM primary particle sizing (Kondo, Aizawa, Kook, & Pickett, 2013). However, the same operator measured all cases, which means trends between operating conditions should still be valid. The B75 baseline point shows larger aggregates for the LHSA injector. The LHSA also has larger primary particles and particle number is also larger. PM formation depends on surface growth, coagulation, and agglomeration - processes which all happen simultaneously. The increased primary particle and aggregate sizes show these processes all happen to a greater extent than for the reference injector. This could indicate more areas to grow PM in cylinder, or that fuel packets spend more time in these zones. Geometric Mean dA (nm)Radius of Gyration Rg (nm)Manual: meandp (nm)Automatic: meandp (nm)Fractal Prefactor kgFractal Prefactor kfFractal Dimension DfGeometricMean (nm)Total conc(#/cm3)Baseline Ref Inj 84.3 50.7 21.8 18.6 1.43 4.75 1.73 85.2 2.05E+06Baseline LHSA 131.4 87.0 25.7 18.2 1.48 4.72 1.67 121.9 3.78E+060EGR Ref Inj 72.0 42.5 22.7 18.1 2.26 6.12 1.44 59.7 1.11E+060EGR LHSA 98.6 59.3 25.1 20.1 1.70 5.51 1.69 92.1 2.87E+06LPI Ref Inj 84.5 46.8 25.0 22.8 2.02 5.92 1.55 62.7 1.39E+06LPI LHSA 119.3 68.2 30.6 25.1 2.53 6.87 1.44 99.4 2.78E+06SPC Ref Inj 62.2 35.4 24.3 21.7 2.11 6.16 1.55 45.3 5.82E+05SPC LHSA 87.9 49.1 21.8 20.1 2.45 7.11 1.54 66.1 1.05E+06Operating ConditionTEM SMPS61 The increase in particle number indicates that there are more nucleation particles that eventually grow into primary particles and aggregate chains. The 0 EGR condition has the same primary particle sizes as the baseline 18% EGR condition but aggregate size and number concentration decreases for both injectors. Investigation into HPDI and diesel literature for these conditions shows that aggregate sizes tend to increase as EGR is added. The effect on primary particles is less clear. Several diesel studies and previous SCRE work have seen that primary particle size decreases when EGR is added (Graves, Olfert, Patychuk, Dastanpour, & Rogak, 2014; Lee et al., 2008; Mathis, Mohr, Kaegi, Bertola, & Boulouchos, 2005), however this work and some other studies (Lapuerta, Martos, & Herreros, 2007) do not see this same trend. This test was only done for one snapshot of an engine condition and the TEM grid could be an unrepresentative sample of the final results, however the fact that no change in primary particle size is seen for both injectors shows that the effect of EGR on PM is more sensitive to final aggregate size. For the optimal LPI point, particle number decreases by 25% for both injectors. The aggregate size stays the same, while the primary particle size increases for both injectors. Previous work on soot morphology tends to agree that smaller aggregates tend to have smaller primaries as well (Arai, Amagai, Nakaji, & Hayashi, 2005; Dastanpour & Rogak, 2014; Mathis et al., 2005). As the processes of coagulation, surface growth, and agglomeration are simultaneous processes, it is difficult to speculate why primary particles grow only. As this is noticed for both injectors, it could be something inherent to the soot formation process in LPI, particularly with soot formation in the first injection. The optimal SPC point was also tested. For the reference injector, the premixed point reduced aggregate size and particle number while increasing primary particle size. The paired nozzle has the same effect except the primary particle size decreases. Previous HPDI work saw that the SPC condition would reduce primary particles from 30nm to 24nm (Patychuk, 2013). Automatic sizing shows larger primary particles also which contrasts to that work. The different results for this point could be experimental error and further work is under way to confirm the trend. 3.5.2 Fractal Analysis of Aggregates Fractal dimension is a method to explain the compactness of an aggregate. It describes whether an aggregate tends to form a long chain or if the aggregates are clumps of primary particles. Previous work (KΓΆylΓΌ, Faeth, Farias, & Carvalho, 1995) has discussed that the fractal properties tend to be independent of fuel and flame condition. There are several equations that can be used to calculate fractal dimension, 62 but they all relate to a ratio between the aggregate area and primary particle area. For this work the equation described in (Brasil, Farias, & Carvalho, 1999) is used and included below. 𝑁 = π‘˜π‘” (2𝑅𝑔𝑑𝑝)𝐷𝑓 Where π‘˜π‘” is a fractal pre-factor, 𝑅𝑔 is the radius of gyration for an aggregate, 𝑑𝑝is the mean primary particle diameter and 𝐷𝑓is the fractal dimension. 𝑁 is the number of primary particles and is calculated by dividing the aggregate projected area over the sauter projected area. Each operating condition for each injector was separated and fit to calculate π‘˜π‘”and 𝐷𝑓 individually using aggregate sizes and the primary particle sizes with the automatic code for that test point. The work of (Brasil et al., 1999) found values of π‘˜π‘” fall close to the range of 1.5 and 3.1. The fractal dimension (𝐷𝑓) was near 1.78. For this work, values of π‘˜π‘” and 𝐷𝑓 are lower (π‘˜π‘” of 1.43-2.53; 𝐷𝑓 1.44-1.73) but relatively close to these values. A second method uses the pre-factor π‘˜π‘“instead of π‘˜π‘”. π‘˜π‘“ is related to π‘˜π‘” in the equation below. π‘˜π‘“ =π‘˜π‘”2𝐷𝑓 To compare to (KΓΆylΓΌ et al., 1995) the pre-factor π‘˜π‘“ was used. Values from that study showed that soot aggregates from various fuels and flame conditions fell near 8.5 for π‘˜π‘“, and 1.82 for the fractal dimension. For this study 𝐷𝑓 and π‘˜π‘“ are both lower than the previous study, however both injectors are in the same range for each constant which is the same as what was seen with the comparison to Brasil. The fractal dimensions for both injectors are similar for all the conditions and for both injectors. This means that all operating conditions seem to have the same compactness regardless of the amount of PM formed. Figure 32 shows the regressions generated from the aggregates measured using Brasil’s method. This plotted range is where 95% of the aggregates lie on. 63  Figure 32 - Fits of Fractal Dimension For the range of data used, the fractal dimension (slope of log-log curve) is essentially constant for the conditions measured. The pre-factor changes as the aggregate shapes change for the different conditions. This is consistent with the literature.  Table 14 - 95% Confidence Intervals for π’Œπ’‡, 𝑫𝒇 There are some limitations with this method for measuring fractal dimension. First, the number of points for each fit is less than 100 and a more robust fit would be achieved if more aggregates were analyzed. Second, there is error in both the primary particle sizing and aggregate sizing so this has an effect on the fit and must be noted when comparing to numerical simulations. 1001011001011022*Rg/ Mean dpAggregate Area / Primary Sauter Diameter AreaFits for Range of 2*Rg/ Mean dp  Ref Inj BLLHSA BLRef Inj EGRLHSA EGRRef Inj LPILHSA LPIRef Inj SPCLHSA SPC95% Confidence IntervalsBaseline Ref Inj 4.11 5.40 1.67 1.80Baseline LHSA 3.90 5.55 1.59 1.750EGR Ref Inj 5.69 6.55 1.39 1.480EGR LHSA 5.14 5.87 1.66 1.73LPI Ref Inj 5.68 6.17 1.53 1.58LPI LHSA 6.29 7.45 1.39 1.50SPC Ref Inj 5.93 6.40 1.52 1.58SPC LHSA 6.43 7.80 1.48 1.59Dfkf64 3.5.3 Differences between HPDI and Diesel PM Several studies have analyzed the size of primary particles in diesel exhaust. The sizes of primary particles from a medium duty diesel engine was measured in (Neer & Koylu, 2006) and compared to other diesel literature. The mean sizing of primary particles was found to be between 20nm and 35nm over a range of operating conditions changing speeds and loads. This was also found for other work (Zhu, Lee, Yozgatligil, & Choi, 2005). HPDI engines have the same size range for primary particles seen also in (Graves et al., 2014; Patychuk, 2013). For aggregate sizing, the mean radius of gyration tends to be between 75nm and 180nm in the diesel literature (Neer & Koylu, 2006; Zhu et al., 2005). This is related to the aerodynamic diameter measured from the SMPS. However it is very dependent on the operating condition of the engine. In this work the radius of gyration for all the data was between 35nm and 87nm which is considerably smaller than the diesel studies. Fractal dimension was also compared. (Neer & Koylu, 2006) found that the fractal dimension for diesel soot is in the range of 1.7-1.9. For other diesel literature the fractal dimension ranged from 1.5-1.9. Previous work with the SCRE have found 𝐷𝑓 to be 1.68-1.78 (Soewono, 2008). In this work 𝐷𝑓 was smaller (1.44-1.73), however more tests are needed to confirm that they are any different from the previous work as the engine was the same and fractal dimension tends to be constant for an engine. Overall, the output PM from the SCRE in HPDI has similar characteristics as do other diesel studies in the literature. Soot aggregates and mass are smaller. 3.5.4 Effects of PM Morphology on a DPF A DPF requires regeneration when it is full, noted by a large pressure drop. The amount of specific surface area in the aggregates is the main factor in the amount of time the regeneration process takes (Lapuerta, Oliva, Agudelo, & Boehman, 2012). This is related to the sizes of the primary particles. The pressure drop increases as a function of mass, so large aggregates with large primary particles require more frequent regeneration cycles with longer times needed to complete the process. Further information can be found in (Cordiner, Mecocci, Mulone, & Rocco, 2011), (Beatrice, Iorio, Guido, & Napolitano, 2012), and (Barouch Giechaskiel, Munoz-Bueno, & Rubino, 2007). 65 3.5.5 Discussion of PM Characteristics Comparison for LHSA and Reference Injector PM characteristics were analyzed at 4 operating conditions for the reference injector and the LHSA paired-hole nozzle. For all conditions the paired-hole nozzle produced larger aggregates and higher particle number. This indicates that there are more areas in the cylinder for PM formation or more time available for fuel packets to form soot. When compared to the diesel literature, HPDI PM is similar in fractal dimension and primary particle sizes, though aggregate sizing is smaller. Because of the larger aggregates than the reference injector, the paired-hole nozzle would impact a larger fuel penalty if it was used on a production engine. 3.6 Interpretation of Measurements This section first presents the CFD studies done for B75 for the reference injector and LHSA. Then a discussion of all the experiments from this work is presented with PM/CO and PM/NOx emissions maps, including the CFD emission predictions. 3.6.1 Computational Fluid Dynamics (CFD) Studies Westport’s in-house CFD software (β€œGOLD”) was used to determine the reasons for the high CO and PM emissions. There were 4 main parts in this study. ο‚· Gas Diesel Alignment: matching correct geometry of nozzles ο‚· Simulated and measured HRR and emissions comparison for the LHSA and reference injector ο‚· Use of a equivalence ratio-temperature map to compare emissions formation for the LHSA and reference injector ο‚· Evaluate penetration and entrainment differences of the jets Table 15 includes the information from the experimental runs that was used for the simulation. 66  Table 15 - Inputs for GOLD Simulation Pooyan Kheirkhah ran the GOLD experiments based on experimental data from the SCRE. After the first simulation the model needed tuning to match the 50%IHR. To do this the PSOI and GSOI were delayed slightly to match the actual opening of the injector (PSEP remained constant) to set the ignition timing. This was due to the finite amount of time it takes for the injector to open and close. The gas and pilot pulse widths also needed to be increased slightly to account for the closing time of the injector. Further detail on tuning the model is included in (Kheirkhah, 2015). 3.6.1.1 Gas/Diesel Alignment The angle between the diesel and gas jets (interlace angle) on the LHSA did not match the manufacturing drawing. The diesel hole and centre of the gas jets should be aligned but they were offset by 5-10 degrees. The effects of changing this angle on the LHSA nozzle was studied further in GOLD as this nozzle was used the most in experimental testing. GOLD Inputs Reference Injector LHSAEngine Speed (rpm) 1494 1497Diesel Rail Pressure (MPa) 27.0 27.2Diesel Injected Mass (mg) 10.9 13.4PPW (ms) 0.62 0.62PSOI (deg aTDC) -22.5 -22.5GSOI (deg aTDC) -14.62 -14.57Gas Rail Pressure (MPa) 25.4 25.6Gas Injected Mass (mg) 173.7 178.867  Figure 33 - Gas/Diesel Angle A baseline B75 study for the LHSA was run with input information from the engine from the baseline B75 tests and using the two alignments to see differences in HRR and emissions. GOLD used a 0 and 8 degree angle. A diagram of these angles is included in Figure 35. The HRR for the changing interlace angle experiments are compared below in Figure 34.  Figure 34 - Interlace Angle HRR GOLD Simulation The HRR for the model shows that as the angle between the gas and diesel jets is changed the amount of energy released from the premixed combustion is less. In-cylinder temperature plots at the start of -20 -10 0 10 20 30 40-50050100150200250CA [ο‚°]HRR [kJ/(m3.deg)]  LHSA, interlace: 0 degLHSA, interlace: 8 deg68 combustion (10%IHR) for the two cases were generated to identify the source of this difference. The surface is an iso-stoichiometric surface where Ο†=1.  Figure 35 - In-Cylinder Combustion for Different Interlace Angles Upon further examination, changing the interlace angle affects where the gas jets ignite. For the 0-degree angle the diesel jet rotates away from the gas jet, increasing the amount of time for air to mix before the gas ignites. When the diesel hole is rotated 8 degrees ignition of the gas jet occurs sooner as the diesel jet rotates toward the gas jets due to swirl. The increased fuel available to burn for the 0 degree angle is probably due to this delay and the additional time required to fully ignite the jets increases time to create a premixed charge with the air and is shown on the HRR graphs. For the 8 degree case, the jets ignite immediately and no additional time is available for further mixing. 3.6.1.2 HRR Comparison to Experimental Data The B75 baseline condition was run for both injectors in GOLD. Figure 36 shows the GOLD predictions and SCRE measurements. 69  Figure 36 - GOLD HRR vs SCRE HRR, Reference Injector and LHSA (8 Degree) The start of combustion and the initial premixed peak are well predicted for the LHSA injector (right panel), however the second peak in the GOLD HRR curve is not present in the SCRE measurements. For the reference injector, GOLD predicts that there are two peaks, but only one is observed experimentally. 3.6.1.3 Equivalence Ratio – Temperature Map An equivalence ratio-temperature map (or Ο†-T map) can be used to show differences in combustion for the paired nozzle and the reference injector for the same B75 condition. Kheirkhah and Faghani (Faghani, 2015; Kheirkhah, 2015) generated maps based on Cantera using the GRI3 kinetic mechanism for natural gas. For each packet of fuel, emissions of CO, NOx and C2H2 (precursor for soot) are formed depending on the location on the map and a fixed reaction time. The following plot compares the mass of fuel in the cylinder for the two injectors. For these plots, the fuel mass for the reference injector in each zone (area d x dT on the map) is subtracted from the paired nozzle. Regions with more mass for the paired nozzle are in red, and more mass for the reference injector are in blue. The axes on this plot are local equivalence ratio and temperature. The Ο†-T plot shows that the trajectory from high EQR low temperature to high temperature lean regions is largely the same for the paired nozzle and reference injector. The main difference is where the fuel is located at this time-step. At 25% IHR, there is much more fuel for the LHSA in high CO and PM forming zones. The intention of the paired nozzle was to reduce the local equivalence ratios and avoid much of the soot generation process by moving the curve below these areas. However this map shows that fuel has passed the cold, rich limit where no soot can form and is now able to soot. Further leaning or less leaning would be necessary to prevent this. -20 -10 0 10 20 30 40-50050100150200250CA [ο‚°]HRR [kJ/(m3.deg)]  SCREReference Inj-20 -10 0 10 2 30 40-50050100150200250CA [ο‚°]HRR [kJ/(m3.deg)]  SCRELHSA, interlace: 8 deg70  Figure 37 - Local Ο†-T Difference Plot 25% IHR B75 (adapted from Pooyan Kheirkhah’s thesis) As combustion progresses to 50% IHR these increased sooting zones still exist, though they move toward the same high NOx zone before combustion ends. Another method to evaluate the mixing of the fuel is by comparing the penetration and air entrained into the jets to understand how this is occurring. 3.6.1.4 Penetration and Entrainment Comparison  The penetration of the jets and the amount of air entrained by the jets was computed by GOLD (Kheirkhah, 2015).  Figure 38 - Penetration of Jets (Left) and Entrainment of Air (Right) measured from GOLD B75 Baseline Case Ref Inj, LHSA 0 0.5 1 1.5 2 2.5 3010203040506070Time after start of injection [ms]Jet penetration [mm]Penetration of Jets Reference Injector, LHSA  Ref InjLHSA0 10 20 30 40 50012345Penetration [mm]Normalized entrained airEntrainment of Air as Function of Jet Penetration  Ref InjLHSA71 The LHSA was designed (Faghani & Rogak, 2011) to have penetration similar to that of the reference injector. In the previous CFD this penetration was slower due to time needed to allow the jets to combine. In Kheirkhah’s simulations we also see that the LHSA jet penetrates slower than the reference injector. While the jets travel slower, the jets are able to pull more air into the jet, and based on the Ο†-T map this is able to lean the jets. However, despite entraining more air, sooting is increased as the very rich mixture is pulled into a zone that allows soot to form, instead of leaning enough to prevent soot from forming entirely. 3.6.1.5 Evaluation of GOLD Emissions at B75 The engine-out emissions for the SCRE measurements and GOLD predictions are compared in Table 16. As the B75 condition was repeated many times, an averaged value is used for the comparison. The uncertainty is expressed as a standard deviation of the repeats.  Table 16 - SCRE vs. GOLD Emissions Predictions of NOx, CH4 and fuel consumption are quite good, but GOLD has difficulty estimating the CO and PM emissions, especially for the changing interlace angle cases. A possible reason is that GOLD estimates emissions at the exhaust valve while the actual emissions are measured further downstream. 3.6.1.6 CFD Discussion CFD simulations using GOLD were run with the LHSA paired nozzle and the reference injector. Combustion of the paired nozzle showed that fuel packets travelled through a richer high PM forming region during combustion promoting PM and CO emissions. This contrasts to the work by (Faghani & Rogak, 2011) which predicted that the paired jet would entrain more air to reduce the sooting zones. It seems that the zones have increased, by shifting the high equivalence packets to the sooting zone. Gold EmissionsGold 0 Degree InterlaceGOLD 8 Degree interlacePM (DRX) (g/kW-hr) 0.053 0.015 Β±0.01 0.062 0.060 0.101 Β±0.02CO (g/kW-hr) 10.58 4.52 Β±1.1 3.36 10.22 13.70 Β±3.6NOx (g/kW-hr) 1.24 1.37 Β±0.07 1.09 1.14 1.22 Β±0.06CH4 (g/kW-hr) 0.41 0.46 Β±0.03 0.39 0.39 0.63 Β±0.07GISFC (g/kW-hr) 170.47 176.80 Β±3.5 168.81 167.93 182.45 Β±2.7SCRE vs. GOLD E is i nsSCRE Β± uncertaintySCRE Β± uncertaintyReference Inj LHSA72 3.6.2 Combustion Maps Emissions from CFD and the entire dataset at B75 were compared together on PM vs. NOx and PM vs. CO plots. Because the paired-hole nozzle has much higher PM and CO emissions the trade-off is expected to look different. Additionally, these maps illustrated the locations of the low PM strategies discussed earlier. The full factorial dataset and parameter sweeps were combined with the low PM strategies dataset to compare the emissions trade-off curves for both injectors. This was to show the normal PM/NOx trade-off for single non-premixed conditions and show where the low PM injection strategies deviated from the normal curve. 3.6.2.1 PM vs. NOx The PM vs. NOx plot (Figure 39) includes fitted curves for positive PSEP (normal HPDI) points.  Figure 39 - PM vs. NOx over entire B75 dataset for LHSA and Ref Inj The paired-hole nozzle produces more PM than the reference injector, but similar NOx, so the curve is simply shifted rightward. There is considerable variability to the B75 baseline point as this data was taken over a large period of time, and exact engine emissions are known to vary over time and the error 0 0.05 0.1 0.150123456DustTrak (g/kW-hr)NOx (g/kW-hr)PM vs NOx  RefInj-fitLHSA-fitLPI-RefInjLPI-LHSAB75Baseline-RefInjB75Baseline-LHSASPC-RefInjSPC-LHSAPSEP=-1.1 RefInjPSEP=-1.1 LHSA73 bars are noted for the baseline condition. However the low PM strategies are clearly seen to deviate from the β€œnormal” (fit) curve. As PSEP decreases PM decreases and NOx increases. The region of the graph is the same for both injectors as the combustion becomes more premixed. For the SPC point, NOx emissions drop and are lower for the paired nozzle. For LPI, PM is reduced due to reduced PM formation with only a small increase in NOx. By normalizing emissions using the values of the baseline single injection operation, Figure 40 is obtained.  Figure 40 - PM vs. NOx Normalized over entire B75 dataset for LHSA and Ref Inj As discussed previously, the relative effect of the strategies are very similar, and this is shown in Figure 40 by the blue and red points/curves almost overlapping. For negative PSEP, the PM reduction effect on the paired-hole nozzle is slightly better than the reference injector. For LPI the effects are similar. For SPC the PM reduction for the paired nozzle is also slightly better. This underlines the robustness of each strategy. Based on these results, we would predict similar benefits for other HPDI injectors operating under these conditions. 3.6.2.2 CO vs. PM The CO vs. PM plot (Figure 41) shows a cloud of data with the paired-hole nozzle data usually further from the origin. A fitted curve based on normal operation (PSEP>0) was fit to the plot for the reference 0 0.5 1 1.500.511.522.53DustTrak (g/kW-hr)/BLavgNOx (g/kW-hr)/BLavgPM vs NOx, Normalized by Baseline  RefInj-fitLHSA-fitLPI-RefInjLPI-LHSAB75Baseline-RefInjB75Baseline-LHSASPC-RefInjSPC-LHSAPSEP=-1.1 RefInjPSEP=-1.1 LHSA74 injector. Remarkably, this fit also describes the LHSA data. When premixing is increased (PSEP<0) for both injectors the CO/PM ratio increases (blue points).  Figure 41 - CO vs. PM over entire B75 dataset for LHSA and Ref Inj Several studies (N. Clark, Gautam, Lyons, & Bata, 1997; N. N. Clark, Jarrett, & Atkinson, 1999; McKain, Wayne, & Clark, 2012; Taylor & Clark, 2004) have characterized the CO/PM ratio of many diesel engines, but have found that individual engines follow different and highly scattered CO/PM trends, in contrast to our findings for HPDI in Figure 41.  Gold CFD points were also included on the graph, and show that for an optimized run, GOLD can match the CO/PM ratio measured in the SCRE. However, we cannot explain the deviation for the 0 degree interlace angle. Figure 42 plots the CO/PM ratios for the low PM strategies. The fit from Figure 41 is transferred to this graph to illustrate the β€œnormal” trend. 10-310-210-110010-1100101102DustTrak (g/kW-hr)CO (g/kW-hr)   CO vs. PM  PSEP-2.5-2-1.5-1-0.500.51RefInjFitRef InjLHSAGOLD - RefInjGOLD - LHSA 8degGOLD - LHSA 0deg75  Figure 42 - CO vs. PM Low PM Strategies For the low PM strategies, the emission effects are the same for each strategy. The LHSA paired nozzle has higher overall emissions at all cases. Lastly, CO/PM ratios for all paired-hole injector data (positive PSEP, B75) is compared with the reference injector trend in Figure 43. Remarkably, the paired-hole measurements (except for the SHSA measurements) fall very close to the reference injector trend. 10-310-210-110010-1100101102DustTrak (g/kW-hr)CO (g/kW-hr)   CO vs. PM Low PM Strategies  RefInjFitRefInj BaselineLHSA BaselineRefInj LPILHSA LPIRefInj SPCLHSA SPCRefInj Neg PSEPLHSA Neg PSEP76  Figure 43 - CO vs. PM all PH injectors at B75 The SHSA produced so much CO that the data would have been corrected by the high-range correction mentioned earlier in Chapter 2, and there might have been additional errors in this correction. 3.6.2.3 Combustion Map Discussion and Interpretation When comparing PM/NOx trade-offs the entire curve for the paired nozzle is shifted rightward, but the strategies are equally as effective at reducing emissions. The reference and paired HPDI injectors fall into the same areas of the curve. The PM/CO ratio showed that both injectors shared the same trend along all operating conditions. For non-premixed conditions, a constant ratio is noticed for the reference injector and all paired injectors. For more premixed (PSEP<0) conditions, the CO/PM ratio increases and the LHSA and reference injector stay on the same trend-line. This shows that a fairly constant relationship for these emissions exists for the SCRE regardless of injector used. The close association between PM and CO is interesting because the two pollutants are formed by different kinetic mechanisms and over regions of the phi-T map that are different.  10-310-210-110010-1100101102DustTrak (g/kW-hr)CO (g/kW-hr)   CO vs. PM at B75 all PH injectors  RefInjFitRefInjLHSALHLASHSASHLA77 4. Conclusions There are six key conclusions from this work: 1. Previous work concluded that increasing mixing with air and fuel should always provide a PM benefit as the sooting regions are smaller or there is less time available for sooting to occur. However, the paired-hole nozzle does not work in this way. It seems to provide enough mixing to start to produce soot instead of providing enough mixing to get past this sooting zone. NOx emissions from the paired-hole nozzle are comparable to those of the reference injector over the entire dataset, which shows that local temperature is very similar using both injectors. 2. The work has shown that regardless of starting emissions, LPI, SPC, and negative PSEP can reduce emissions by similar percentages, even for nozzles that have very high baseline emissions. For the paired nozzle methane emissions increase faster when PSEP<-1.1ms, which indicates more mixing is occurring. These strategies were able to significantly reduce emissions with the LHSA paired nozzle and it is possible that these strategies work for other HPDI injectors. 3. The response surfaces can yield information about the combustion processes especially in small ranges (such as between -0.4ms PSEP and -1.1ms) when PM emissions change significantly if there is enough data. Furthermore they can provide direct information about emission responses to each parameter. 4. Further information about the sizing of primary particles is needed to explain the reasoning behind the sizes of the LPI primary particles and why they are larger when number and aggregate size are both smaller. SPC results also did not match previous work and these need to both be further investigated. This could provide more information about why the strategy works, or help to decouple coagulation and surface growth processes from agglomeration. 5. CFD can provide in-cylinder clues we cannot see without optical techniques, and the equivalence ratio-temperature map was able to explain why the emissions from the paired-hole nozzle were so high. Based on the results from the -T difference maps, CFD simulations could have predicted high emissions before the nozzles were built. CFD also provides the flexibility in changing injector geometry but exact emissions prediction is still a challenge so experimental data are still required. 6. Based on this work it seems that there is a constant trajectory seen on the CO/PM plot for all non-premixed conditions regardless of injector used and independent of engine parameters. In addition, the low PM strategies have the same effect on a PM/NOx and CO/PM plot.  78 5. Recommendations for Further Work There are several recommendations that come out of this work. 1. Optical techniques with the jets may give more information why there is not enough mixing, or confirm the CFD results that led to the design of these nozzles. This should be done with Schlieren imaging. 2. The low PM strategies should be further tested with other HPDI injectors to see if the same benefits are seen. 3. The CFD results provided information that could not be explained from emissions results alone. While experimental tests will always be needed to confirm the results (especially CO and PM), CFD provides more flexibility with altering nozzle design. For future nozzles, it is recommended that β€œdifference -T” maps be produced. 4. It is recommended that response surface fits be produced for all multi-parameter engine studies. Further investigation using these surfaces and combining with other datasets could be useful to understand operating areas. 5. It is recommended that this CO/PM ratio be mapped for the large dataset available for the J36 injector and for future HPDI experimental studies.   79 Bibliography Arai, M., Amagai, K., Nakaji, T., & Hayashi, S. (2005). Primary and Aggregate Size Distributions of PM in Tail Pipe Emissions form Diesel Engines. JSME International Journal Series B, 48(4), 639–647. doi:10.1299/jsmeb.48.639 Beatrice, C., Iorio, S. Di, Guido, C., & Napolitano, P. (2012). Detailed characterization of particulate emissions of an automotive catalyzed DPF using actual regeneration strategies. Experimental Thermal and Fluid Science, 39, 45–53. doi:10.1016/j.expthermflusci.2012.01.005 Benajes, J., & Molina, S. (2006). The use of micro-orifice nozzles and swirl in a small HSDI engine operating at a late split-injection LTC regime. In Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering (Vol. 220, pp. 1807–1817). doi:10.1243/09544070JAUTO383 Brasil, A. M., Farias, T. L., & Carvalho, M. G. (1999). a Recipe for Image Characterization of Fractal-Like Aggregates. Journal of Aerosol Science, 30(10), 1379–1389. doi:10.1016/S0021-8502(99)00026-9 Brown, S. (2008). High-pressure Direct-Injection of Natural Gas with Entrained Diesel into a Compression-Ignition Engine. University of British Columbia. Retrieved from https://circle.ubc.ca/bitstream/id/13917/ubc_2008_fall_brown_benjamin_scott.pdf Clark, N., Gautam, M., Lyons, D., & Bata, R. (1997). Natural Gas and Diesel Transit Bus Emissions: Review and Recent Data. SAE International, (412). Retrieved from http://papers.sae.org/973203/ Clark, N. N., Jarrett, R. P., & Atkinson, C. M. (1999). Field Measurements of Particulate Matter Emissions, Carbon Monoxide, and Exhaust Opacity from Heavy-Duty Diesel Vehicles. Journal of the Air & Waste Management Association, 49(9), 76–84. doi:10.1080/10473289.1999.10463880 Cordiner, S., Mecocci, F., Mulone, V., & Rocco, V. (2011). Particle Number Emissions: An Analysis by Varying Engine/Exhaust-System Design and Operating Parameters. SAE International. doi:10.4271/2011-24-0170 Dastanpour, R., Boone, J., & Rogak, S. (2015). Automated Determination of Primary Particle Diameter of Nanoparticle Aggregates by TEM Image Analysis. Journal of Aerosol Science. Dastanpour, R., & Rogak, S. N. (2014). Observations of a Correlation between Primary Particle and Aggregate Size for Soot Particles. Aerosol Science and Technology, (October), 1043–1049. doi:10.1080/02786826.2014.955565 European Commission. (2014). Transport & Environment - Environment - European Commission. Retrieved October 27, 2014, from http://ec.europa.eu/environment/air/transport/road.htm Faghani, E. (2015). Effect of Injection Strategies on Particulate Matter Emissions in HPDI Natural-Gas Engines. University of British Columbia. 80 Faghani, E., Patychuk, B., McTaggart-Cowan, G., & Rogak, S. (2013). Soot Emission Reduction from Post Injection Strategies in a High Pressure Direct Injection Natural Gas Engine. SAE International, 13. doi:10.4271/2013-24-0114 Faghani, E., & Rogak, S. N. (2011). Penetration and Flow Field Characteristics of Dual-Hole Transient Gas Jets. In Proceedings of Combustion Institute Canadian Section Spring Technical Meeting (CICS 2011). Winnipeg, Canada. Frenklach, M. (2002). Reaction mechanism of soot formation in flames. Physical Chemistry Chemical Physics, 4(11), 2028–2037. doi:10.1039/b110045a Gao, J., Matsumoto, Y., Namba, M., & Nishida, K. (2007). Group-hole nozzle effects on mixture formation and in-cylinder combustion processes in direct-injection diesel engines. SAE International, (724), 776–790. Retrieved from http://papers.sae.org/2007-01-4050/ Gao, J., Moon, S., Zhang, Y., Nishida, K., & Matsumoto, Y. (2009). Flame structure of wall-impinging diesel fuel sprays injected by group-hole nozzles. Combustion and Flame, 156(6), 1263–1277. doi:10.1016/j.combustflame.2009.01.014 Giechaskiel, B., AlfΓΆldy, B., & Drossinos, Y. (2009). A metric for health effects studies of diesel exhaust particles. Journal of Aerosol Science, 40(8), 639–651. doi:10.1016/j.jaerosci.2009.04.008 Giechaskiel, B., Munoz-Bueno, R., & Rubino, L. (2007). Particle Measurement Programme (PMP): Particle Size and Number Emissions Before, During and After Regeneration Events of a Euro 4 DPF Equipped Light-Duty, 1540–1553. Retrieved from http://papers.sae.org/2007-01-1944/ Graves, B., Olfert, J., Patychuk, B., Dastanpour, R., & Rogak, S. N. (2014). Morphology and Volatility of Particulate Matter Emitted from a Natural Gas Direct-Injection Compression-Ignition Engine. In Proceedings of the Combustion Institute - Canadian Section. Heywood, J. B. (1988). Internal Combustion Engine Fundamentals. McGrawHill series in mechanical engineering (Vol. 21, pp. 741–742). Hill, P., & Ouellette, P. (1999). Transient turbulent gaseous fuel jets for diesel engines. Journal of Fluids Engineering, 121(March 1999), 93–101. Retrieved from http://fluidsengineering.asmedigitalcollection.asme.org/article.aspx?articleid=1428857 Jones, H., McTaggart-Cowan, G., Brown, B. S., Laforet, C., Wu, N., & Patychuk, B. (2011). Single Cylinder Research Engine (SCRE) Users Manual. Khalek, I. a., Bougher, T. L., Merritt, P. M., & Zielinska, B. (2011). Regulated and Unregulated Emissions from Highway Heavy-Duty Diesel Engines Complying with U.S. Environmental Protection Agency 2007 Emissions Standards. Journal of the Air & Waste Management Association, 61(4), 427–442. doi:10.3155/1047-3289.61.4.427 Kheirkhah, P. (2015). CFD Modelling of Non-Conventional Injection Strategies in a High-Pressure Direct-Injection (HPDI) Natural Gas Engine. The University of British Columbia. 81 Kim, J., Park, S., Andrie, M., Reitz, R., & Sung, K. (2009). Experimental investigation of intake condition and group-hole nozzle effects on fuel economy and combustion noise for stoichiometric diesel combustion in an HSDI. SAE International, 2(1). Retrieved from http://papers.sae.org/2009-01-1123/ Kondo, K., Aizawa, T., Kook, S., & Pickett, L. (2013). Uncertainty in Sampling and TEM Analysis of Soot Particles in Diesel Spray Flame. SAE International, (1). doi:10.4271/2013-01-0908 Korakianitis, T., Namasivayam, a. M., & Crookes, R. J. (2011). Natural-gas fueled spark-ignition (SI) and compression-ignition (CI) engine performance and emissions. Progress in Energy and Combustion Science, 37(1), 89–112. doi:10.1016/j.pecs.2010.04.002 KΓΆylΓΌ, Ü. Γ–., Faeth, G. M., Farias, T. L., & Carvalho, M. G. (1995). Fractal and projected structure properties of soot aggregates. Combustion and Flame, 100(4), 621–633. doi:10.1016/0010-2180(94)00147-K Ladommatos, N., & Zhao, H. (2001). Engine Combustion Instrumentation and Diagnostics (p. 842). Warrendale, PA: Society of Automotive Engineers, Inc. Laforet, C. (2009). Combustion of Natural Gas with Entrained Diesel in a Heavy-Duty Compression-Ignition Engine. Lapuerta, M., Martos, F. J., & Herreros, J. M. (2007). Effect of engine operating conditions on the size of primary particles composing diesel soot agglomerates. Journal of Aerosol Science, 38(4), 455–466. doi:10.1016/j.jaerosci.2007.02.001 Lapuerta, M., Oliva, F., Agudelo, J. R., & Boehman, A. L. (2012). Effect of fuel on the soot nanostructure and consequences on loading and regeneration of diesel particulate filters. Combustion and Flame, 159(2), 844–853. doi:10.1016/j.combustflame.2011.09.003 Lee, K. O., Zhu, J., & Song, J. (2008). Effects of exhaust gas recirculation on diesel particulate matter morphology and NOx emissions. International Journal of Engine Research, 9(2), 165–175. doi:10.1243/14680874JER02307 Maricq, M. M. (2013). Monitoring Motor Vehicle PM Emissions: An Evaluation of Three Portable Low-Cost Aerosol Instruments. Aerosol Science and Technology, 47(5), 564–573. doi:10.1080/02786826.2013.773394 Mathis, U., Mohr, M., Kaegi, R., Bertola, A., & Boulouchos, K. (2005). Influence of diesel engine combustion parameters on primary soot particle diameter. Environmental Science & Technology, 39(6), 1887–92. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/15819252 McKain, D. L., Wayne, S., & Clark, N. (2012). Relationship between Carbon Monoxide and Particulate Matter Levels across a Range of Engine Technologies. SAE International. doi:10.4271/2012-01-1346 82 McTaggart-Cowan, Bushe, & Rogak. (2003). Injection parameter effects on a direct injected, pilot ignited, heavy duty natural gas engine with EGR. SAE Transactions, (724). Retrieved from http://cat.inist.fr/?aModele=afficheN&cpsidt=16124690 McTaggart-Cowan, G. (2006). Pollutant Formation in a Gaseous-Fuelled, Direct Injection Engine. The University of British Columbia. McTaggart-Cowan, Mann, Huang, & Wu. (2012). Particulate Matter Reduction From a Pilot-Ignited, Direct Injection of Natural Gas Engine. In Proceedings of the ASME 2012 Internal Combustion Engine Division Fall Technical Conference (pp. 1–11). Retrieved from http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid=1721010 McTaggart-Cowan, & Rogak, S. (2004). Effect of operating condition on particulate matter and nitrogen oxides emissions from a heavy-duty direct injection natural gas engine using cooled exhaust gas. International Journal of Engine Research, 5(6), 499–511. doi:10.1177/146808740400500602 Moon, S., Gao, J., Nishida, K., Matsumoto, Y., & Zhang, Y. (2008). Ignition and combustion characteristics of wall-impinging sprays injected by group-hole nozzles for direct-injection diesel engines. SAE Int J Engines, 1(1), 1205–1219. Retrieved from http://papers.sae.org/2008-01-2469/ Munshi, S. R., McTaggart-Cowan, G., Huang, J., & Hill, P. G. (2011). Development of a Partially-Premixed Combustion Strategy for a Low-Emission, Direct Injection High Efficiency Natural Gas Engine. In Proceedings of the ASME 2011 Internal Combustion Engine Division Fall Technical Conference (pp. 1–14). Retrieved from http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid=1627317 Myers, R. H., Montgomery, D. C., & Anderson-Cook, C. (2009). Response Surface Methodology: Process and Product Optimization Using Designed Experiments . Wiley series in probability and statistics (p. 704). Retrieved from citeulike-article-id:3931224\nhttp://www.amazon.ca/exec/obidos/redirect?tag=citeulike09-20&amp;path=ASIN/0470174463 Neer, a, & Koylu, U. (2006). Effect of operating conditions on the size, morphology, and concentration of submicrometer particulates emitted from a diesel engine. Combustion and Flame, 146(1-2), 142–154. doi:10.1016/j.combustflame.2006.04.003 Nishida, K., Tian, J., Sumoto, Y., & Long, W. (2009). An experimental and numerical study on sprays injected from two-hole nozzles for DISI engines. Fuel, 88(9), 1634–1642. doi:10.1016/j.fuel.2009.01.003 O’Connor, J., & Musculus, M. (2013). Post Injections for Soot Reduction in Diesel Engines: A Review of Current Understanding. SAE International. doi:10.4271/2013-01-0917 Park, S., & Reitz, R. (2008). Modeling the effect of injector nozzle-hole layout on diesel engine fuel consumption and emissions. Journal of Engineering for Gas Turbines and Power, 130(3), 032805. doi:10.1115/1.2835352 83 Park, S., & Reitz, R. (2009). Optimization of fuel/air mixture formation for stoichiometric diesel combustion using a 2-spray-angle group-hole nozzle. Fuel, 88(5), 843–852. doi:10.1016/j.fuel.2008.10.028 Patashnick, H., & Rupprecht, E. G. (1991). Continuous PM-10 Measurements Using the Tapered Element Oscillating Microbalance. Journal of the Air & Waste Management Association, 41(8), 1079–1083. doi:10.1080/10473289.1991.10466903 Patychuk, B. (2013). Particulate matter emission characterization from a natural-gas high-pressure direct-injection engine. The University of British Columbia. Retrieved from https://circle.ubc.ca/handle/2429/44341 Patychuk, B., & Rogak, S. N. (2012). Particulate Matter Emissions from a Natural Gas Fueled High Pressure Direct Injection Engine tested over the SCRE 9 Mode Test Cycle Title. In Proceedings of the Combustion Institute Canadian Section Spring Technical Meeting. Pulkrabek, W. W. (2004). Engineering Fundamentals of the Internal Combustion Engine. Journal of engineering for gas turbines and power (Vol. 126, p. 478). Pearson Prentice Hall. doi:10.1115/1.1669459 Reif, K. (2014). Diesel Engine Management. Springer. Retrieved from http://link.springer.com/content/pdf/10.1007/978-3-658-03981-3.pdf Ricou, F. P., & Spalding, D. B. (1961). Measurements of entrainment by axisymmetrical turbulent jets. Journal of Fluid Mechanics, 11, 21. doi:10.1017/S0022112061000834 Ristovski, Z. D., Miljevic, B., Surawski, N. C., Morawska, L., Fong, K. M., Goh, F., & Yang, I. a. (2012). Respiratory health effects of diesel particulate matter. Respirology (Carlton, Vic.), 17(2), 201–12. doi:10.1111/j.1440-1843.2011.02109.x Silverman, D. T., Samanic, C. M., Lubin, J. H., Blair, A. E., Stewart, P. a, Vermeulen, R., … Attfield, M. D. (2012). The Diesel Exhaust in Miners study: a nested case-control study of lung cancer and diesel exhaust. Journal of the National Cancer Institute, 104(11), 855–68. doi:10.1093/jnci/djs034 Soewono, A. (2008). Morphology and Microstructure of Diesel Particulates. The University of British Columbia. Taylor, S., & Clark, N. (2004). Diesel emissions prediction from dissimilar cycle scaling. In Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering (Vol. 218, pp. 341–352). Retrieved from http://pid.sagepub.com/content/218/3/341.short Tree, D., & Svensson, K. (2007). Soot processes in compression ignition engines. Progress in Energy and Combustion Science, 33(3), 272–309. doi:10.1016/j.pecs.2006.03.002 TSI Incorporated. (2012a). DustTrak DRX Aerosol Monitor Theory of Operation (pp. 1–4). 84 TSI Incorporated. (2012b). Mass Concentration Comparison Between the DustTrak DRX Aerosol Monitor and TEOM. US EPA. (2013). Heavy-Duty Highway Compression-Ignition Engines And Urban Buses -- Exhaust Emission Standards | Emission Standards Reference Guide | US EPA. Retrieved October 27, 2014, from http://www.epa.gov/otaq/standards/heavy-duty/hdci-exhaust.htm Wang, S. C., & Flagan, R. C. (2007). Spectrometer Scanning Electrical Mobility Spectrometer, (July 2014), 37–41. doi:10.1080/02786829008959441 Zhang, Y., Nishida, K., Nomura, S., & Ito, T. (2003). Spray characteristics of group-hole nozzle for DI diesel engine. SAE International, (724). Retrieved from http://papers.sae.org/2003-01-3115/ Zhu, J., Lee, K. O., Yozgatligil, A., & Choi, M. Y. (2005). Effects of engine operating conditions on morphology, microstructure, and fractal geometry of light-duty diesel engine particulates. Proceedings of the Combustion Institute, 30(2), 2781–2789. doi:10.1016/j.proci.2004.08.232   85 Appendices Appendix A: Raw Data Tables Data Tables for experiments are included generally in the order presented in the thesis. Typically each column represents an average of 2 points taken on the same day.  Multi-Mode Data A summary of all data is presented first. A longer version is included in the electronic appendix.   86   Parameter Ref Inj LHSA SHLA LHLA SHSACO (g/kW-hr)   4.15 11.59 10.47 14.23 26.07NOx (g/kW-hr) 1.43 1.32 1.25 1.21 1.16TEOM (g/kW-hr) 0.03 0.14 0.12 0.16 0.25DustTrak (g/kW-hr) 0.01 0.09 0.09 0.1 0.12CH4 (g/kW-hr) 0.47 0.56 0.58 0.75 1.04GISFC - diesel equivalent (g/kW-hr) 182.02 186.17 187.28 190.79 193.07Peak cylinder pressure (bar) 142.36 144.6 143.56 147.19 146.59Parameter Ref Inj LHSA SHLA LHLA SHSACO (g/kW-hr)   0.64 2.87 2.67 8.9 20.64NOx (g/kW-hr) 1.51 1.3 1.29 1.15 1.14TEOM (g/kW-hr) 0 0.02 0.03 0.13 0.2DustTrak (g/kW-hr) 0 0.01 0.01 0.1 0.13CH4 (g/kW-hr) 0.35 0.3 0.34 0.38 0.59GISFC - diesel equivalent (g/kW-hr) 184.28 189.86 188.72 195.17 201.96Peak cylinder pressure (bar) 116.28 116.61 117.45 119.96 118.39Parameter Ref Inj LHSA SHLA LHLA SHSACO (g/kW-hr)   3.52 8.1 7.88 10.36 20.73NOx (g/kW-hr) 1.46 1.39 1.35 1.41 1.24TEOM (g/kW-hr) 0.04 0.11 0.11 0.15 0.22DustTrak (g/kW-hr) 0.01 0.06 0.05 0.08 0.11CH4 (g/kW-hr) 0.62 0.68 0.66 0.77 0.96GISFC - diesel equivalent (g/kW-hr) 182.41 186.29 186.5 187.52 194.05Peak cylinder pressure (bar) 134.13 134.22 134.32 136.96 136Parameter Ref Inj LHSA SHLA LHLA SHSACO (g/kW-hr)   0.94 3.49 2.48 7.98 14.82NOx (g/kW-hr) 1.45 1.32 1.32 1.29 1.24TEOM (g/kW-hr) 0 0.03 0.04 0.09 0.13DustTrak (g/kW-hr) 0 0.01 0.01 0.06 0.06CH4 (g/kW-hr) 0.63 0.53 0.58 0.65 0.65GISFC - diesel equivalent (g/kW-hr) 177.67 184.45 182.21 186.63 190.54Peak cylinder pressure (bar) 97.34 98.86 100.68 100.04 102.24Parameter Ref Inj LHSA SHLA LHLA SHSACO (g/kW-hr)   2.02 2.17 1.51 2.97 2.53NOx (g/kW-hr) 1.49 1.2 1.14 1.28 1.21TEOM (g/kW-hr) -0.01 -0.01 -0.01 0.02 0.03DustTrak (g/kW-hr) 0 0 0 0 0CH4 (g/kW-hr) 1.42 0.9 0.92 0.83 0.94GISFC - diesel equivalent (g/kW-hr) 183.35 184.28 182.34 186.49 185.27Peak cylinder pressure (bar) 51.85 50.69 51.75 52.5 50.67B75A75C75B50B2587 Timing Sweep Reference Injector Parameter IHR7 IHR9 IHR11 IHR13 IHR15 CO (g/kW-hr)    2.76 3.84 4.15 3.23 2.52 NOx (g/kW-hr) 2.15 1.83 1.43 1.12 1.07 TEOM (g/kW-hr) 0.02 0.03 0.03 0.02 0.02 DustTrak (g/kW-hr) 0.01 0.01 0.01 0.01 0.01 CH4 (g/kW-hr) 0.48 0.47 0.47 0.48 0.47 GISFC - diesel equivalent (g/kW-hr) 178.47 180.72 182.02 183.95 185 Peak cylinder pressure (bar) 159.49 152.44 142.36 132.75 126.95 CO2 (kg/kW-hr) 0.42 0.42 0.43 0.43 0.43 O2 (kg/kW-hr) 0.35 0.35 0.35 0.35 0.36 Injector 1 1 1 1 1 Engine speed (rpm) 1489.39 1494.04 1494.34 1492.66 1490.81 Gross IMEP (bar) 16.71 16.63 16.6 16.57 16.67 CNG downstream pressure (MPa) 25.24 25.35 25.38 25.22 25.19 Oxygen equivalence Ratio 0.62 0.62 0.62 0.62 0.61 EGR Flow Rate (%) 17.72 17.37 17.87 18.5 17.75 50% IHR 7.22 8.65 10.88 13.4 14.98 Diesel flow rate (kg/hr) 0.5 0.48 0.49 0.49 0.5 CNG flow rate (kg/hr) 7.64 7.75 7.79 7.84 7.93 Carbon balance ratio 0.95 0.96 0.96 0.95 0.95 Oxygen balance ratio 1.01 1.01 1.01 1.01 1.01 PSOI set [deg] -26 -24.5 -22.5 -20.5 -19 PPW [ms] 0.62 0.62 0.62 0.62 0.62 PSEP (ms) 0.26 0.26 0.26 0.26 0.26 GPW [ms] 1.62 1.62 1.62 1.65 1.66 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 207.36 210.07 210.28 210.81 215.53 Thermal Efficiency (%) 0.47 0.47 0.46 0.46 0.45 Mass deposition rate [ug/s] 0.02 0.03 0.03 0.02 0.02   88  LHSA Parameter IHR7 IHR9 IHR11 IHR13 IHR15 CO (g/kW-hr)    8.63 10.73 11.59 10.93 9.54 NOx (g/kW-hr) 2.17 1.7 1.32 1.11 0.92 TEOM (g/kW-hr) 0.07 0.12 0.14 0.14 0.12 DustTrak (g/kW-hr) 0.04 0.06 0.09 0.1 0.08 CH4 (g/kW-hr) 0.55 0.56 0.56 0.55 0.56 GISFC - diesel equivalent (g/kW-hr) 180.71 184.11 186.17 187.5 188.73 Peak cylinder pressure (bar) 163.75 154.54 144.6 136.79 128.28 CO2 (kg/kW-hr) 0.42 0.42 0.43 0.43 0.44 O2 (kg/kW-hr) 0.36 0.36 0.36 0.36 0.36 Injector 2 2 2 2 2 Engine speed (rpm) 1485.69 1486.75 1488.35 1485.6 1484.77 Gross IMEP (bar) 16.65 16.62 16.59 16.6 16.63 CNG downstream pressure (MPa) 25.47 25.5 25.52 25.46 25.45 Oxygen equivalence Ratio 0.61 0.61 0.62 0.61 0.61 EGR Flow Rate (%) 17.65 17.66 18.21 17.84 18.54 50% IHR 6.71 8.71 10.94 13.06 15.27 Diesel flow rate (kg/hr) 0.6 0.58 0.57 0.56 0.55 CNG flow rate (kg/hr) 7.61 7.76 7.86 7.92 7.99 Carbon balance ratio 0.96 0.96 0.96 0.96 0.96 Oxygen balance ratio 1.01 1.01 1.01 1.01 1.01 PSOI set [deg] -26 -24 -22 -20 -18 PPW [ms] 0.62 0.62 0.62 0.62 0.62 PSEP (ms) 0.26 0.26 0.26 0.26 0.26 GPW [ms] 1.25 1.29 1.32 1.35 1.37 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 211.14 213.61 214.29 216.6 217.87 Thermal Efficiency (%) 0.47 0.46 0.45 0.45 0.45 Mass deposition rate [ug/s] 0.08 0.14 0.15 0.13 0.11   89 EGR Sweep Reference Injector Parameter 24%EGR 18%EGR 12%EGR 6%EGR 0%EGR CO (g/kW-hr)    4.86 3.94 3.45 2.74 2.33 NOx (g/kW-hr) 0.78 1.21 1.79 2.61 3.81 TEOM (g/kW-hr) 0.04 0.02 0.03 0.02 0.01 DustTrak (g/kW-hr) 0.02 0.01 0.01 0.01 0 CH4 (g/kW-hr) 0.67 0.51 0.41 0.34 0.28 GISFC - diesel equivalent (g/kW-hr) 175 175.7 176.64 175.63 177.48 Peak cylinder pressure (bar) 138.55 132.27 131.96 129.58 126.47 CO2 (kg/kW-hr) 0.41 0.42 0.42 0.42 0.43 O2 (kg/kW-hr) 0.31 0.34 0.35 0.37 0.4 Injector 1 1 1 1 1 Engine speed (rpm) 1490.46 1493.12 1477.13 1485.99 1487.18 Gross IMEP (bar) 16.62 16.71 16.8 16.77 16.61 CNG downstream pressure (MPa) 25.65 25.37 25.14 25.2 25.67 Oxygen equivalence Ratio 0.61 0.61 0.62 0.62 0.62 EGR Flow Rate (%) 24.59 18.25 13.02 6.75 0.23 50% IHR 11.16 11.84 11.38 11.27 11.19 Diesel flow rate (kg/hr) 0.47 0.44 0.44 0.39 0.46 CNG flow rate (kg/hr) 7.48 7.58 7.59 7.62 7.58 Carbon balance ratio 0.97 0.97 0.97 0.97 0.98 Oxygen balance ratio 1.01 1.01 1.01 1.01 1.01 PSOI set [deg] -22 -21.5 -21.5 -21.5 -21 PPW [ms] 0.65 0.65 0.65 0.63 0.62 PSEP (ms) 0.3 0.3 0.3 0.3 0.3 GPW [ms] 1.41 1.44 1.46 1.46 1.48 Air flow (kg/hr) 198.62 205.78 208.26 213.91 220.09 Thermal Efficiency (%) 0.48 0.48 0.48 0.48 0.47 Mass deposition rate [ug/s] 0.04 0.02 0.02 0.01 0.01   90  LHSA Parameter 24%EGR 18%EGR 12%EGR 6%EGR 0%EGR CO (g/kW-hr)    10.9 16.61 8.8 7.75 6.77 NOx (g/kW-hr) 0.75 1.13 1.8 2.52 3.65 TEOM (g/kW-hr) 0.14 0.19 0.09 0.08 0.05 DustTrak (g/kW-hr) 0.14 0.1 0.06 0.04 0.03 CH4 (g/kW-hr) 0.84 0.75 0.46 0.36 0.28 GISFC - diesel equivalent (g/kW-hr) 180.49 187.83 181.78 180.53 183.58 Peak cylinder pressure (bar) 142.3 139.21 136.21 133.37 133.17 CO2 (kg/kW-hr) 0.41 0.44 0.43 0.43 0.44 O2 (kg/kW-hr) 0.33 0.35 0.38 0.4 0.42 Injector 2 2 2 2 2 Engine speed (rpm) 1491.02 1494.81 1495.94 1499.85 1501.68 Gross IMEP (bar) 16.68 16.63 16.61 16.66 16.61 CNG downstream pressure (MPa) 25.38 25.51 25.38 25.44 25.43 Oxygen equivalence Ratio 0.61 0.62 0.61 0.61 0.61 EGR Flow Rate (%) 24.18 17.73 11.89 6.5 0.2 50% IHR 10.99 10.94 10.86 10.89 10.71 Diesel flow rate (kg/hr) 0.55 0.58 0.53 0.55 0.57 CNG flow rate (kg/hr) 7.68 7.99 7.75 7.72 7.84 Carbon balance ratio 0.96 0.99 0.97 0.98 0.98 Oxygen balance ratio 1.01 1 1.01 1.01 1.01 PSOI set [deg] -22.5 -22 -22 -22 -22 PPW [ms] 0.62 0.62 0.62 0.62 0.62 PSEP (ms) 0.3 0.3 0.3 0.3 0.3 GPW [ms] 1.23 1.21 1.22 1.2 1.21 Air flow (kg/hr) 204.87 218.77 218.45 224.46 232.44 Thermal Efficiency (%) 0.47 0.45 0.46 0.47 0.46 Mass deposition rate [ug/s] 0.1 0.19 0.07 0.06 0.04   91 EQR Sweep Reference Injector Parameter 0.70 0.61 0.58 0.54 0.50 CO (g/kW-hr)    10.09 4.34 3.26 2.06 1.67 NOx (g/kW-hr) 1 1.37 1.4 1.71 1.85 TEOM (g/kW-hr) 0.08 0.03 0.02 0.03 0.03 DustTrak (g/kW-hr) 0.03 0.01 0.01 0.01 0.01 CH4 (g/kW-hr) 0.35 0.43 0.46 0.47 0.49 GISFC - diesel equivalent (g/kW-hr) 180.91 173.97 171.53 168.55 166.32 Peak cylinder pressure (bar) 130.36 136.19 139.22 144.96 147.03 CO2 (kg/kW-hr) 0.43 0.42 0.41 0.41 0.41 O2 (kg/kW-hr) 0.22 0.32 0.35 0.42 0.47 Injector 1 1 1 1 1 Engine speed (rpm) 1494.89 1491.77 1492.81 1491.01 1490.24 Gross IMEP (bar) 16.62 16.63 16.79 16.76 16.65 CNG downstream pressure (MPa) 25.28 25.28 25.25 25.29 25.33 Oxygen equivalence Ratio 0.7 0.61 0.59 0.54 0.5 EGR Flow Rate (%) 18.71 18.6 18.93 18.17 18.11 50% IHR 11.24 11.03 11.17 11.09 10.85 Diesel flow rate (kg/hr) 0.5 0.5 0.53 0.51 0.46 CNG flow rate (kg/hr) 7.74 7.42 7.36 7.21 7.1 Carbon balance ratio 1 1 0.99 1 1.01 Oxygen balance ratio 1 1 1 1 1 PSOI set [deg] -23.5 -23.5 -23.5 -23.5 -23.5 PPW [ms] 0.7 0.7 0.72 0.73 0.73 PSEP (ms) 0.3 0.3 0.3 0.3 0.3 GPW [ms] 1.56 1.5 1.51 1.46 1.44 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 187.98 203.44 209.77 223.84 233.61 Thermal Efficiency (%) 0.47 0.48 0.49 0.5 0.51 Mass deposition rate [ug/s] 0.1 0.04 0.02 0.03 0.03    92  LHSA Parameter 0.70 0.61 0.58 0.54 0.5 CO (g/kW-hr)    24.64 15.97 14.28 9.52 6.34 NOx (g/kW-hr) 1.04 1.21 1.26 1.5 1.77 TEOM (g/kW-hr) 0.25 0.15 0.15 0.1 0.06 DustTrak (g/kW-hr) 0.17 0.1 0.1 0.06 0.04 CH4 (g/kW-hr) 0.73 0.61 0.65 0.59 0.57 GISFC - diesel equivalent (g/kW-hr) 189.29 181.8 178.42 175.12 171.55 Peak cylinder pressure (bar) 133.99 138.76 142.28 145.01 149.41 CO2 (kg/kW-hr) 0.42 0.42 0.42 0.42 0.41 O2 (kg/kW-hr) 0.26 0.34 0.37 0.43 0.49 Injector 2 2 2 2 2 Engine speed (rpm) 1490.48 1489.75 1483.96 1492.38 1491.63 Gross IMEP (bar) 16.7 16.65 16.67 16.59 16.73 CNG downstream pressure (MPa) 25.28 25.29 25.27 25.22 25.19 Oxygen equivalence Ratio 0.7 0.61 0.58 0.54 0.5 EGR Flow Rate (%) 17.64 18.09 19.09 18.76 18.05 50% IHR 11.16 11.16 10.88 10.92 10.99 Diesel flow rate (kg/hr) 0.51 0.49 0.49 0.5 0.5 CNG flow rate (kg/hr) 8.13 7.77 7.6 7.45 7.35 Carbon balance ratio 0.98 0.99 0.99 0.99 1 Oxygen balance ratio 1 1 1 1 1 PSOI set [deg] -22.5 -21.8 -22.5 -22.5 -22.5 PPW [ms] 0.57 0.55 0.57 0.57 0.57 PSEP (ms) 0.34 0.32 0.34 0.34 0.34 GPW [ms] 1.4 1.3 1.25 1.2 1.18 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 198.41 212.21 216.35 226.63 241.07 Thermal Efficiency (%) 0.44 0.46 0.47 0.48 0.49 Mass deposition rate [ug/s] 0.31 0.17 0.19 0.12 0.07    93 Diesel Mass Sweep Reference Injector Parameter 6mg 10.5mg 15mg 19mg 22.5mg CO (g/kW-hr)    5.09 5 5.21 5.57 5.75 NOx (g/kW-hr) 1.32 1.33 1.31 1.48 1.48 TEOM (g/kW-hr) 0.04 0.03 0.04 0.05 0.06 DustTrak (g/kW-hr) 0.01 0.01 0.02 0.02 0.03 CH4 (g/kW-hr) 0.45 0.45 0.46 0.46 0.45 GISFC - diesel equivalent (g/kW-hr) 173.77 174.37 175.56 175.61 176.23 Peak cylinder pressure (bar) 136.98 137.23 135.15 138.83 137.59 CO2 (kg/kW-hr) 0.42 0.42 0.42 0.42 0.42 O2 (kg/kW-hr) 0.31 0.32 0.32 0.32 0.32 Injector 1 1 1 1 1 Engine speed (rpm) 1492.8 1488.23 1488.91 1489.84 1488.37 Gross IMEP (bar) 16.73 16.66 16.58 16.69 16.66 CNG downstream pressure (MPa) 25.45 25.38 25.3 25.2 25.18 Oxygen equivalence Ratio 0.62 0.61 0.62 0.62 0.62 EGR Flow Rate (%) 18.81 18.67 18.59 18.3 18.33 50% IHR 10.75 10.83 11.18 10.64 10.95 Diesel flow rate (kg/hr) 0.27 0.46 0.69 0.87 1.01 CNG flow rate (kg/hr) 7.65 7.46 7.28 7.18 7.06 Carbon balance ratio 0.99 0.98 0.98 0.98 0.97 Oxygen balance ratio 1 1 1 1 1 PSOI set [deg] -22.5 -23.5 -25.5 -28.5 -30 PPW [ms] 0.54 0.69 0.98 1.24 1.48 PSEP (ms) 0.3 0.3 0.3 0.3 0.3 GPW [ms] 1.53 1.49 1.48 1.46 1.45 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 203.06 203.33 203.06 204 203.81 Thermal Efficiency (%) 0.48 0.48 0.48 0.48 0.48 Mass deposition rate [ug/s] 0.04 0.03 0.05 0.05 0.06    94  LHSA Parameter 6mg 10.5mg 15mg 19mg 22.5mg CO (g/kW-hr)    18.68 14.46 22.12 19.91 20.73 NOx (g/kW-hr) 1.22 1.2 1.19 1.31 1.31 TEOM (g/kW-hr) 0.13 0.12 0.22 0.23 0.26 DustTrak (g/kW-hr) 0.08 0.09 0.22 0.2 0.24 CH4 (g/kW-hr) 0.65 0.57 0.82 0.73 0.78 GISFC - diesel equivalent (g/kW-hr) 182.33 181.03 186.54 186.28 186.19 Peak cylinder pressure (bar) 140.35 136.39 140.3 140.7 141 CO2 (kg/kW-hr) 0.42 0.42 0.42 0.43 0.43 O2 (kg/kW-hr) 0.35 0.34 0.35 0.37 0.36 Injector 2 2 2 2 2 Engine speed (rpm) 1488.66 1488.61 1491.66 1491.1 1490.83 Gross IMEP (bar) 16.72 16.6 16.61 16.56 16.65 CNG downstream pressure (MPa) 25.48 25.36 25.33 25.28 25.23 Oxygen equivalence Ratio 0.61 0.62 0.62 0.61 0.61 EGR Flow Rate (%) 17.74 18.23 18.16 17.74 17.96 50% IHR 10.91 11.39 11.12 11.18 11.26 Diesel flow rate (kg/hr) 0.29 0.5 0.66 0.87 1 CNG flow rate (kg/hr) 8.01 7.7 7.83 7.6 7.53 Carbon balance ratio 1 1 0.99 0.99 0.98 Oxygen balance ratio 1 1 1.01 1.01 1.01 PSOI set [deg] -19.5 -22 -23 -26 -27.5 PPW [ms] 0.37 0.57 0.74 1.07 1.25 PSEP (ms) 0.3 0.34 0.3 0.3 0.3 GPW [ms] 1.23 1.35 1.21 1.19 1.19 2GPW [ms] 0 0 0 0 0 Air flow (kg/hr) 214.67 210.22 216.6 218.43 218.45 Thermal Efficiency (%) 0.46 0.46 0.45 0.45 0.45 Mass deposition rate [ug/s] 0.13 0.15 0.21 0.23 0.27   95 Negative PSEP Sweep Reference Injector Parameter Baseline PSEP=-0.4 PSEP=-1.1 PSEP=-1.8 PSEP=-2.5 CO (g/kW-hr)    3.94 3.63 2.91 2.44 2.61 NOx (g/kW-hr) 1.21 1.4 2.44 4.08 4.59 TEOM (g/kW-hr) 0.02 0.01 0.01 0 -0.01 DustTrak (g/kW-hr) 0.01 0.01 0 0 0 CH4 (g/kW-hr) 0.51 0.54 0.87 1.35 1.86 GISFC - diesel equivalent (g/kW-hr) 175.7 173.87 169.45 166.53 164.52 Peak cylinder pressure (bar) 132.27 134.49 139.35 142.13 135.47 CO2 (kg/kW-hr) 0.42 0.42 0.41 0.41 0.4 O2 (kg/kW-hr) 0.34 0.32 0.32 0.31 0.31 Injector 1 1 1 1 1 Engine speed (rpm) 1493.12 1494.56 1494.78 1497.53 1499.26 Gross IMEP (bar) 16.71 16.74 16.63 16.75 16.56 CNG downstream pressure (MPa) 25.37 25.18 25.16 25.45 25.31 Oxygen equivalence Ratio 0.61 0.61 0.61 0.61 0.61 EGR Flow Rate (%) 18.25 18.1 18.63 18.23 17.66 50% IHR 11.84 11.21 11.48 10.87 10.88 Diesel flow rate (kg/hr) 0.44 0.46 0.46 0.44 0.46 CNG flow rate (kg/hr) 7.58 7.51 7.26 7.21 7.03 Carbon balance ratio 0.97 1 1 1 1.01 Oxygen balance ratio 1.01 1.01 1.01 1.01 1.01 PSOI set [deg] -21.5 -16 -14.5 -13.5 -13 PPW [ms] 0.65 0.65 0.65 0.65 0.66 PSEP (ms) 0.3 -0.36 -1.06 -1.76 -2.46 GPW [ms] 1.44 1.47 1.46 1.45 1.5 Air flow (kg/hr) 205.78 204.11 198.31 196.84 193.5 Thermal Efficiency (%) 0.48 0.48 0.5 0.51 0.51 Mass deposition rate [ug/s] 0.02 0.01 0.01 0 -0.01    96  LHSA Parameter Baseline PSEP=-0.4 PSEP=-1.1 PSEP=-1.8 PSEP=-2.5 CO (g/kW-hr)    16.61 10.71 5.04 3.62 3.39 NOx (g/kW-hr) 1.13 1.42 2.6 4.03 4.32 TEOM (g/kW-hr) 0.19 0.1 0.03 0 0.01 DustTrak (g/kW-hr) 0.1 0.08 0.01 0 0 CH4 (g/kW-hr) 0.75 0.65 0.88 1.85 2.85 GISFC - diesel equivalent (g/kW-hr) 187.83 179.18 173.41 168.62 168.47 Peak cylinder pressure (bar) 139.21 139.94 140.89 144.34 135.67 CO2 (kg/kW-hr) 0.44 0.42 0.41 0.4 0.39 O2 (kg/kW-hr) 0.35 0.35 0.34 0.33 0.33 Injector 2 2 2 2 2 Engine speed (rpm) 1494.81 1498.84 1496.44 1500.7 1498.06 Gross IMEP (bar) 16.63 16.66 16.7 16.67 16.62 CNG downstream pressure (MPa) 25.51 25.26 25.13 25.13 25.31 Oxygen equivalence Ratio 0.62 0.61 0.61 0.6 0.61 EGR Flow Rate (%) 17.73 17.47 17.91 18.03 18.6 50% IHR 10.94 10.83 11.25 10.5 11.21 Diesel flow rate (kg/hr) 0.58 0.58 0.58 0.55 0.53 CNG flow rate (kg/hr) 7.99 7.63 7.38 7.19 7.16 Carbon balance ratio 0.99 0.97 0.96 0.98 0.95 Oxygen balance ratio 1 1.01 1.01 1.01 1.01 PSOI set [deg] -22 -16 -14 -13 -13 PPW [ms] 0.62 0.62 0.62 0.62 0.62 PSEP (ms) 0.3 -0.4 -1.1 -1.8 -2.5 GPW [ms] 1.21 1.13 1.1 1.14 1.19 Air flow (kg/hr) 218.77 210.56 204.02 200.17 195.54 Thermal Efficiency (%) 0.45 0.47 0.49 0.5 0.5 Mass deposition rate [ug/s] 0.19 0.1 0.02 0 0.01    97 Optimal Points – Negative PSEP, SPC Reference Injector Parameter Baseline NegPSEPOptimal SPCBaseline SPCOptimal CO (g/kW-hr)    3.94 2.91 13.8 4.58 NOx (g/kW-hr) 1.21 2.44 0.59 1.39 TEOM (g/kW-hr) 0.02 0.01 0.09 0 DustTrak (g/kW-hr) 0.01 0 0.09 0 CH4 (g/kW-hr) 0.51 0.87 0.81 1.08 GISFC - diesel equivalent (g/kW-hr) 175.7 169.45 182.31 170.25 Peak cylinder pressure (bar) 132.27 139.35 136.5 136.51 CO2 (kg/kW-hr) 0.42 0.41 0.42 0.41 O2 (kg/kW-hr) 0.34 0.32 0.23 0.21 Injector 1 1 1 1 Engine speed (rpm) 1493.12 1494.78 1493.02 1493.99 Gross IMEP (bar) 16.71 16.63 16.65 16.7 CNG downstream pressure (MPa) 25.37 25.16 25.2 25.31 Oxygen equivalence Ratio 0.61 0.61 0.7 0.69 EGR Flow Rate (%) 18.25 18.63 25.34 25.32 50% IHR 11.84 11.48 10.58 11.26 Diesel flow rate (kg/hr) 0.44 0.46 0.47 0.47 CNG flow rate (kg/hr) 7.58 7.26 7.83 7.32 Carbon balance ratio 0.97 1 0.98 0.99 Oxygen balance ratio 1.01 1.01 1.01 1.01 PSOI set [deg] -21.5 -14.5 -24 -19 PPW [ms] 0.65 0.65 0.66 0.66 PSEP (ms) 0.3 -1.06 0.34 -1.06 GPW [ms] 1.44 1.46 1.58 1.51 Air flow (kg/hr) 205.78 198.31 184.7 175.3 Thermal Efficiency (%) 0.48 0.5 0.46 0.49 Mass deposition rate [ug/s] 0.02 0.01 0.1 0   98  LHSA Parameter Baseline NegPSEPOptimal SPCBaseline SPCOptimal CO (g/kW-hr)    16.61 5.04 18.52 11.36 NOx (g/kW-hr) 1.13 2.6 0.61 1.25 TEOM (g/kW-hr) 0.19 0.03 0.23 0.01 DustTrak (g/kW-hr) 0.1 0.01 0.11 0.01 CH4 (g/kW-hr) 0.75 0.88 1.14 1.14 GISFC - diesel equivalent (g/kW-hr) 187.83 173.41 187.63 178.34 Peak cylinder pressure (bar) 139.21 140.89 135.81 140.84 CO2 (kg/kW-hr) 0.44 0.41 0.39 0.39 O2 (kg/kW-hr) 0.35 0.34 0.25 0.23 Injector 2 2 2 2 Engine speed (rpm) 1494.81 1496.44 1494.4 1495.96 Gross IMEP (bar) 16.63 16.7 16.58 16.66 CNG downstream pressure (MPa) 25.51 25.13 25.33 25.41 Oxygen equivalence Ratio 0.62 0.61 0.71 0.7 EGR Flow Rate (%) 17.73 17.91 24.9 25.01 50% IHR 10.94 11.25 11.02 10.88 Diesel flow rate (kg/hr) 0.58 0.58 0.57 0.57 CNG flow rate (kg/hr) 7.99 7.38 7.96 7.59 Carbon balance ratio 0.99 0.96 0.91 0.93 Oxygen balance ratio 1 1.01 1.02 1.01 PSOI set [deg] -22 -14 -22.5 -17 PPW [ms] 0.62 0.62 0.62 0.62 PSEP (ms) 0.3 -1.1 0.3 -1.1 GPW [ms] 1.21 1.1 1.26 1.12 Air flow (kg/hr) 218.77 204.02 186.19 179.01 Thermal Efficiency (%) 0.45 0.49 0.45 0.47 Mass deposition rate [ug/s] 0.19 0.02 0.22 0.01   99 LPI LPI Tests SingleInj SingleInj LPIFI85 LPIFI85 LPIFI80 LPIFI80 Injector Ref Inj LHSA Ref Inj LHSA Ref Inj LHSA CO (g/kW-hr)    6.8 11.59 1.42 5.91 1.41 6.41 NOx (g/kW-hr) 0 1.32 1.3 1.13 1.38 1.25 TEOM (g/kW-hr) 0.05 0.14 0.02 0.07 0.02 0.07 DustTrak (g/kW-hr) 0.02 0.09 0 0.05 0.01 0.06 CH4 (g/kW-hr) 0.46 0.56 0.43 0.49 0.5 0.61 GISFC - diesel equivalent (g/kW-hr) 181.56 186.17 175.61 181.48 175.19 180.62 Peak cylinder pressure (bar) 145.35 144.6 137.08 141.85 136.64 141.55 CO2 (kg/kW-hr) 0.42 0.43 0.42 0.45 0.42 0.42 O2 (kg/kW-hr) 0.35 0.36 0.33 0.32 0.34 0.35 Injector 1 2 1 2 1 2 Engine speed (rpm) 1494.76 1488.35 1492.29 1494.64 1492.72 1492.34 Gross IMEP (bar) 16.72 16.59 16.8 16.79 16.63 16.69 CNG downstream pressure (MPa) 25.49 25.52 25.54 25.12 25.68 25.32 Oxygen equivalence Ratio 0.61 0.62 0.61 0.62 0.6 0.61 EGR Flow Rate (%) 18.23 18.21 18.22 18.2 18.15 18.02 50% IHR 10.91 10.94 11.25 11.11 11.14 10.93 Diesel flow rate (kg/hr) 0.48 0.57 0.42 0.59 0.48 0.55 CNG flow rate (kg/hr) 7.83 7.86 7.64 7.76 7.49 7.7 Carbon balance ratio 0.96 0.96 0.98 1.02 0.97 0.97 Oxygen balance ratio 1.01 1.01 1.01 1 1.01 1.01 PSOI set [deg] -23 -22 -22.5 -23.5 -22.5 -23 PPW [ms] 0.62 0.62 0.65 0.62 0.65 0.62 PSEP (ms) 0.26 0.26 0.3 0.26 0.3 0.3 GPW [ms] 1.7 1.32 1.16 1.01 1.09 0.91 FI 100 100 85 85 80 80 GSEP (ms) 0 0 2 2 1.5 1.5 2GPW [ms] 0 0 0.52 0.55 0.53 0.39 Air flow (kg/hr) 212.05 214.29 206.61 213.07 206.34 211.46 Thermal Efficiency (%) 0.46 0.45 0.48 0.46 0.48 0.47 Mass deposition rate [ug/s] 0.05 0.15 0.02 0.05 0.02 0.05   100 PM Characteristics SMPS scans are included for the 4 points discussed.      102012345678x 107Diameter dp (nm)dN/dlogdp (1/cm3)B75 Baseline  RefInjLHSA102012346x 107Diameter dp (nm)dN/dlogdp (1/cm3)B75 0 EGR  RefInjLHSA101       10201234567x 107Diameter dp (nm)dN/dlogdp (1/cm3)B75 LPI FI 85 GSEP2.0  RefInjLHSA10200.511.522.53x 107Diameter dp (nm)dN/dlogdp (1/cm3)B75 SPC (EQR=0.70 EGR=25% PSEP=-1.1ms)  RefInjLHSA102 Full Factorial Matrix A full appendix of this large quantity of data points is included in the electronic copy of this thesis.   103 Appendix B: SCRE Operating Procedures Current SCRE Operating Procedures 1. Pre Start Procedures: 1.1 Start PM Collection Systems (1.1.1) Turn on Power Bar behind TEOM. (1.1.2) Start TEOM computer: Login: Scott, PW: ISX400. (1.1.3) Start program RP1105A located on the desktop, hit β€˜Ok’ on dialog box ensuring the SCRE mode is selected. (1.1.4) Go to Instrument -> Start. (1.1.5) Turn on TEOM, ensure pressure is below 15in Hg, otherwise filter needs to be replaced. (1.1.6) Turn on DustTrak, hit start on DustTrak. (1.1.7) When 3 lights are green, TEOM is warmed up and ready to take data (~30min). Refer to TEOM Operating Procedure in Appendix for Calibration and operation of TEOM.  1.2 Start Ingersoll Rand Compressor (1.2.1) Compressor is located in Room 2.79 the Mechanical Room at the back of the upstairs office Room 278. (1.2.2) Turn on Dryer Unit, this is the low box with the fan on top behind the Ingersoll Rand Compressor, by pushing green button. (1.2.3) Turn on Compressor Contactor on wall in front of IR Compressor. (1.2.4) Turn on IR Compressor and wait until the pressure starts to rise. 1.3 Start Natural Gas Compressor (1.3.1) Turn the natural gas compressor controller (located in the Northwest corner of the Control Room next to the PLC) from β€˜off’ to β€˜auto’ (located beside the PLC). (1.3.2) Check that the supply pressure from the natural gas compressor is above 4200 psi (the lower limit set point) and the upper limit is set to 4800 psi. To do this, press β€˜Escape’ to get to the main menu, select β€˜User Menu’, and scroll down to see the set pressures. 1.4 Start Cell Ventilation Fan   (1.4.1) At Veco touch screen, push Configure and enter PW (911). (1.4.2) Open Fan 102 Config. 104 (1.4.3) Push Manual and Restart Fan if Required. (1.4.4) Set Output to 50% and push Return. (1.4.5) In summertime set fan to auto.  2. Prepare for Engine Start: 2.1 Confirm that control panel main power switch is OFF. Grab keys from desk drawer. 2.2 Turn on three auxiliary power bars in the test cell.  Confirm that AVL pressure sensor cooling pump is operating.  IMPORTANT that these are turned on BEFORE starting control computer.  2.3 Ensure Vent Valve on Fuel Conditioning Panel is closed and NG Shut Off Valve is open, then open Natural Gas Ball Valve at rear wall of engine lab.  2.4 Turn on the diesel scale. 2.5 Unlock and turn on Vector Drive (engine drive) and Intake Heater Inverter Drive. 2.6 Slowly open air hand valve completely. 2.7 Open water hand valve about 1/3 open. 2.8 Check coolant level in main tank.  The coolant supply tank should be about Β½ full. 2.9 Turn on control panel’s main power switch. 2.10 On the control computer, start Windows XP using the User β€˜CERC’ (password: ISX400). 2.11 Start the β€˜UBC Engine Monitor’ shortcut located on the desktop. 2.12 Go to the β€˜ESD Shutdown’ tab and ensure that only the β€˜Engine Shutdown’ and β€˜Engine Shutdown Input’ indicators are lit.  Hit the β€˜Reset RESD button’ to reset the indicators and controls. The 12V system will not start if there are any other errors. Ie: a frozen diesel mass input. 2.13 Ensure the β€˜Intake Air Selector’ is set to β€˜Naturally Aspirated’. Set the β€˜Intake Air Pressure’ set point to 20 kPA.  3. Start Engine Procedure: 3.1 On the control panel, press and hold the Green 12V reset button (at bottom of panel). If the green light will not turn on, wiggle the 12 and 24V relays (at EH-215 and EH-212) and ensure the connections are secure. 3.2 Leave the logfile dialog box blank if you are starting the engine (fill in the reason for a shutdown if you are restarting after an RESD). Check that none of the warning LEDs are lit. 3.3 Inside the test cell ensure that the ignition switch on the β€˜engine control box’ is switched β€˜on’.  The β€˜warning’ and β€˜maintenance’ lights should be on). 3.4 On control panel, check that: ο‚· Intake heater switch is disabled. 105 ο‚· Intake heater control potentiometer is set to zero. ο‚· β€œTorque Mode Enable” switch set to disabled. ο‚· Torque control potentiometer set to zero. ο‚· Emergency stop button is out. ο‚· All white push buttons on the Digalog dyno controller are set out. ο‚· β€˜Diesel refill’ switch is off. ο‚· EGR valve is closed (0% open, pot fully ccw). ο‚· Backpressure valve is open (100% open, pot fully cw). ο‚· β€˜Diesel Systems’ switch is off. ο‚· β€˜HPDI Systems’ switch is off. ο‚· β€˜HPDI Engine Operation’ switch is off. 3.5 Enter the Venturi calibration offset value, so that the carbon balance is 1. 3.6 In the Injection Control Menu set the injection parameters as follows:    J36 Co-Injector B Single Needle Co-Injectors PSOI (deg) -8.00 -30 0 PPW (ms) 0.80 4.2 0 GSOI (deg) -4.50 -14 -14 GPW (ms) 0.80 0.65 0.65  3.7 Turn on oscilloscope. 3.8 In the FPGA page, switch from β€œNaturally Aspirated” to β€œCompressed Air”, and set intake pressure to 120 kPa. 3.9 Turn on Digilog dynamometer controller. 3.10 Set RSP (rpm set point) to 1250 rpm (Left digital display). 3.11 Turn on β€œDiesel Systems”, β€œHPDI Systems”, and β€œHPDI Operation” switches. 3.12 Turn diesel pressure dome loaded regulator, to the right of the control monitors) until the pressure starts to rise.  Check on Page 2 on LabView that both diesel and gas pressures have risen and are about the same.  Set initial pressure to 20 Mpa on Page 1. 3.13 Ensure oil temperature and coolant temperature are at least ambient temperature. 3.14 Turn on β€˜Enable Injection’. 3.15 On Control Panel, confirm β€œTorque Mode Enable” is set to β€œEnable”. 106 3.16 Press and hold start button and turn the dial up as the engine starts to turn. Wait for oil pressure to rise before increasing torque from dyno. 3.17 Increase the dial setting until engine is turning at more than 400 RPM. Release the starter button as the engine begins to fire. The injection should start as soon as the engine is turning over 400. 3.18 Advance timing to about -19 degrees and GPW to 1.0. 3.19 After engine is running smoothly set vector control to about 40% (~3100-3400 mV indicated). 3.20 Enable PSEP, then adjust value to 1.0. 3.21 Turn back pressure control knob to 7%, approximately 7/8 turns back from fully closed. 3.22 Set intake pressure to 120, back pressure to 80. 3.23 Monitor that oil pressure, engine temperature and other operating conditions are within specs. 3.24 If needed, turn on Charge Air Heater and set control to about 2600 for warm up. 3.25 Increase engine speed with Dyno Control to 1200 rpm and warm up engine to 70 + degrees water temperature, 80 + degrees oil temperature and intake manifold temperature to greater than 70 degrees. 4. Emission Bench Calibration  4.1 Open all bottles to the left of emissions bench, ensure pressures are set to 50 psi. 4.2 Check the 2 filters in the emission bench cabinet. If they are contaminated, change them together with the main filter located in the test cell. 4.3 AVL bench should be on, go to sample selection page and confirm that selecting Ln 1 selects sample point 3. Ensure that selecting Ln 3 selects sample 2. 4.4 Click Standby for all sensors that will be used. Do not forget the EGR sensor on the last page. Ensure the FID flames turn on for the THC and CH4 boxes. It may require clicking standby several times. 4.5 Calibrate the gas sensors by clicking β€œAutoCal”. 4.6 Flow zero and span gas and record values in spreadsheet on the desktop. This will be saved in the day file with the testing data. Zero and span will be done at the beginning and end of the day. Ensure both ranges are spanned. THC and CH4 cannot be done at the same time. If there are hashtags when calibrating, make sure to recalibrate at least 2 times since the bench averages the previous calibrations. 4.7 Select auto range for all analyzers with multiple ranges. 4.8 Click flow sample for all analyzers to be ready to take data. 5. CO2 Analyzer Calibration (analyzer needs 30 mins to warm up) 5.1 Turn on the power at the back. This normally will be done in step 1.1.1 5.2 Turn on the pump on the front panel. 107 5.3 Turn the valve from run to Cal. 5.4 Turn the valve to Zero and open the N2 valve. 5.5 Adjusting the zero knob, set the value to 0. 5.6 Close N2 bottle and turn valve from 0 to span and open CO2 valve. 5.7 Adjusting the span knob, set the value to the CO2 certified concentration. 5.8 Shut off CO2 valve, turn valve from Cal to Run. 5.9 Ensure both compressed gas bottles are firmly closed.  6. Charge Air Heater Operation: 6.1 Ensure that air flow is 120 Kg/hr or greater. 6.2 Turn the β€œIntake Heater Enable” switch on the Control Panel to β€œEnable”. 6.3 Increase the voltage to the heater with the potentiometer to about 2600. 6.4 Monitor intake temperature and heater wall temperature.  Never exceed 450 degrees. 6.5 On shut down turn off heater and run air through system to reduce temperature before shutting down engine.  108 7. Normal Shutdown Procedure 7.1 If running with the intake heater: ο‚· Ensure the airflow rate is at least 120 kg/hr. ο‚· Dial the temperature potentiometer back to zero (fully CCW) and set the β€˜intake heater enable’ switch to β€˜disable’. ο‚· Let air flow through the heater for at least 2 minutes in order to cool the heating coils. 7.2 If running at high cylinder pressure, reduce intake pressure and gas fuelling and/or retard timing until the cylinder pressure is below 100 bar. 7.3 If running with EGR, close the EGR valve.  Reduce the intake air pressure in order to maintain or reduce the intake air pressure as the EGR valve closes. 7.4 Decrease the back pressure by setting the exhaust back pressure to 0 kPa.  Open the motorized back pressure valve by turning the BP potentiometer on the CP fully CW. 7.5 Reduce fuel flow so that engine is running at 600-800 rpm, with 20 Nm of supplemental torque, no load on dynamometer.  7.6 Reduce manifold pressure to approx. 15 kPa, and then switch the combustion air line selector to β€˜naturally aspirated’.  7.7 Shut down engine:  ο‚· On the Injection Control Menu, disable all injections ο‚· Dial torque control potentiometer to 0 (fully ccw).  Wait until the engine comes to a complete stop.  Switch the β€˜vector drive enable’ switch to β€˜disabled’. ο‚· Turn down the fuel pressure to 0. 7.8 Turn off β€˜HPDI Operation’, β€˜HPDI Systems’ and β€˜Diesel Systems’ switches.  Turn off both yellow power buttons on Digalog.  Left button first, then right button. 7.9 Turn off control panel 7.10 Turn off the auxiliary power bars located in the test cell. 7.11 Shut high pressure natural gas hand valve (HV-NG-500). 7.12 Open β€˜natural gas vent’ valve on the FCM.  7.13 Close β€˜vent’ valve on the FCM when methane stops flowing. 7.14 Turn off the scale. 7.15 If EGR was used open and close HV-IST-100.  7.16 Shut down and lock-out vector and inverter drives. 7.17 Shut off water supply. 7.18 Close compressed air supply valve (HV-INT-100).  7.19 Turn the NG Compressor Controller from β€˜auto’ to β€˜off’. 7.20 On the PLC, stop the fan for test cell 102.  This is done from the β€œFan Control” page. 109 7.21 Quit the Labview control program.  Back up any data collected onto the SSH server (see U1-PROC-002). 7.22 Check calibration on the emissions bench if desired.  This is done by flowing a zero and then a span through the gas sensors and recording on the same spreadsheet from the beginning of the day. 7.23 On TEOM: (7.23.1) Go to β€œInstrument”, click β€œStop”. (7.23.2) Turn off the TEOM. (7.23.3) Close the program β€œRP1105A” and shut down computer. (7.23.4) Turn off CO2 analyzer and pump. (7.23.5) Turn off power bar after computer has shut down. 7.24 On AVL bench: De-select all gas sensors including the intake CO2 sensor. Set the Emissions bench to β€˜standby’. Set bench to β€˜pause’. 7.25 Shut down the control computer. 7.26 Ensure the valves on all of the bottled gases are firmly shut. 7.27 In the mechanical room, stop the air dryer and air compressor. Turn off the compressor and disconnect the compressor from the power grid. 7.28 Check cell for any signs of problems: fuel or oil leaks, loose fittings, etc.    110 Appendix C: Airflow Calibrations The SCRE uses two sensors to determine intake air flow to the firing cylinder. The Venturi and hot film anemometer are used to check our airflow. The hot film anemometer was originally installed to check whether the Venturi was drifting throughout a day of engine testing. Every 6 months the two meters should be compared to our Laminar Flow Element to verify our airflow measurement. This is a simple procedure that can be done in a few hours.  1. Obtain Fluke pressure calibrator from instrument shop. 0-10 inches of water range. 2. Check to make sure air is shut off to SCRE. 3. Place a notice that air is disconnected at control panel such that the engine is not inadvertently started during operation. The engine can run naturally aspirated in the new configuration, but we don’t want to do this. 4. The laminar flow element has been epoxied to a metal flange that is bolted to the lid. The epoxy is strong but care must be taken to bend the lid as little as possible when it is opened to reduce stress. Take room barometer measurement off the lid and plug it in near the drum to measure room pressure. Carefully remove leverlock on drum lid and open drum to obtain 1 in flange and hose section. Carefully close the lid, making sure the rubber seal is seated correctly and secure the lever. 5. Disconnect these two flanges. The nuts are 7/8 inches. Make sure to not lose the metal gasket when it falls out. 111  6. Bolt on new 1 inch flange (should be stored in blue drum that contains Laminar Flow Element). Make sure the metal gasket is installed.  7. Attach pipe elbow at the union and tighten with a pipe wrench. Make sure the pipe bend ensures a vertical hose connection. 8. Attach hose to other end of pipe elbow and tighten with a pipe wrench. 112  9. Move drum to approximately this location. With some adjusting of the pipe elbow, the hose should be the exact correct length. Do not connect hose to drum just yet. 10. Connect pressure calibrator to NPT hose connections on the drum lid. Make sure to put Teflon tape on the threads and tighten securely. Make sure high pressure from the laminar flow element is connected to high pressure on the pressure calibrator. Connect lead to digital readout and make sure it is reading units of inches of water at 4 degrees C. Zero the readout. 11. Connect thermocouple to a spot on the DAQ. Previously channel 10 (disconnected sample line temp) was used. This will read out on the PM page. 12. Turn on air compressor and dryer upstairs in CERC. Turn on aux power in cell, air valve, and SCRE control computer. 113 13. Set pressure to 0 kPa and switch to compressed air. Make sure valve turns. Double check hose is not connected to drum lid. 14. Set pressure to 30 kPa. Go into cell and check that air is flowing through the hose and that there aren’t any leaks. The hose will not fully inflate but should be stable. Turn off the air and connect hose to drum lid. Finger-tighten the union joint, and make sure it is secure with pipe wrench but don’t reef on it. The connection to the drum lid is through plastic so we want to be careful when we tighten the other end. Take lid off of laminar flow element to allow air to flow out of the drum. 15. Open spreadsheet file. When testing we will want to record Venturi Delta P, venturi line temp, venturi line pressure, hot film mass flow, drum temperature, room pressure, and delta P on the fluke calibrator. These are all marked in red on the spreadsheet. 16. Record venturi offset and take 0 readings. Check humidity of the air from the dryer. It should be near 5%. 17. Flow air through the system, setting the intake pressure from 30 – 140 kPa in increments of 5 kPa. Record all values. Do this twice, making sure that you check the venturi zero halfway done. Observe the error % values for the range. High error at zero flow can be ignored. If necessary, adjust hot film calibration curve or cv for venturi. If cv needs to change, make sure to remeasure the areas of the venturi first. 18. When done turn air down and switch to naturally aspirated to reduce pressure. Shut off air and disconnect system. Be sure to securely reconnect the 1 in flange at the intake heater. Cover laminar flow element and place hose and 1 in flange inside.   114 Examining Results The venturi mass flow equation is below: ?Μ‡? = πΆπ‘œπ‘›π‘ π‘‘βˆš            𝑃1𝜌 (𝑃2𝑃1)2π‘˜(1 βˆ’ (𝑃2𝑃1)π‘˜βˆ’1π‘˜)(1 βˆ’ (𝑃2𝑃1)2π‘˜)(𝐴2𝐴1)2 πΆπ‘œπ‘›π‘ π‘‘ = 𝐢𝑣𝐴2√2π‘˜π‘˜ βˆ’ 1 The spreadsheet uses P1 from the line pressure, rho is calculated using the ideal gas law P1/RT and k is 1.4. A1 and A2 have been measured as of August 2013 and should remain constant. Cv should also remain constant. A1 is 1.5 inches (bore) and A2 is .8975 inches (throat). The laminar flow element measures actual volumetric flow rate that is then converted to standard volumetric flow rate (SCFM) and then to mass flow. The humidity, temperature of the air and pressure of the air directly upstream of the element are used to provide correction factors to change units from flow at room conditions to standard volumetric flow. If humidity is below 20% there is no correction factor that needs to be added for humidity. Air Temperatures ranging outside of 72-79 F in the drum may need slight alterations to that correction factor. See Laminar Flow Elements Installation and Operation Instructions that can be obtained in CERC or online for details. The hot film anemometer has a fourth order equation to match air mass flow to a voltage. While mostly linear, a fourth order equation was used to provide a near 0 reading at no airflow conditions. The equation can be obtained from MAX on the control computer. After data is inputted note the average error for the venturi and HF sensor. At calibration of the HF, we obtained error ~.8% and ~1.5% on the venturi. Large deviations in error, or deviations on the parity plots will mean a new calibration for the HF or venturi will be needed. For the venturi this is done by adjusting Cv.   115 Appendix D: Piezoelectric Transducer Calibration Introduction Piezoelectric transducers operate on a dynamic basis. They are best suited for situations where pressure changes rapidly and where a starting pressure is known. Because they operate with a charge signal, some charge is inherently lost over time, but this can be accounted for by knowing the amount of time that has passed in a test (i.e., compression stroke time). The following report details the calibration of these types of sensors.  Equipment / Procedure For an accurate calibration of the sensors, it is best to use a similar setup to the actual system the sensor will be used on. For our calibration, we used the same charge amplifier, cable, and simulated the capacitance of an attached low capacitance BNC cable with a capacitor (not shown) that is used on the SCRE. To calibrate to a known pressure, a static pressure transducer is needed. For this case we used a 0-5000 psi transducer from the IVC and calibrated it before the testing began. It is located upstream of the ball valve on the test chamber so the charge pressure is known before the valve is opened.   116  The entire rig must be pressure tested first before charging with a pressurized gas. This should be done hydraulically only. Never pressure test with a gas. If there is a leak when pressurized with a gas, components can become projectiles that can injure or kill. Safety is more important than saving a few hours’ time! This has happened previously on the IVC and was entirely preventable if procedures were followed. For ANY questions about pressure testing, talk with Marcus Fengler in the Machine Shop, the designated safety officer, Dr. Rogak, or Dr. Kirchen. It is important this is always done with an appropriate factor of safety. If unsure of a FOS use 1.5 times the maximum test pressure.  The calibration procedure is as follows:  Install sensor in test chamber. Torque the sensor to suggested torque level (15Nm) with special Β½ inch split socket. It has been found torque on the sensor is not a huge factor when calibrating sensitivity (explained later), but it is best to install the sensor exactly the same as it would be installed on the SCRE.  Check all pipe connections to be sure they are tightened. Start DaqX and set data collection to 20kHz -5 to 5 V.  Open nitrogen bottle and set pressure to desired level (in this report 500psi and 1000psi were used). Keep ball valves closed.  Open charge ball valve slowly and nitrogen should flow to the test chamber. Keep vent valve closed, check for leaks from the test chamber, if there is a hissing sound Teflon tape may be needed to seal the pressure transducer. Also check for leaks in any other part of the setup. This can be done by monitoring the static pressure and closing the nitrogen bottle. If there are leaks, close the nitrogen bottle and vent before attempting to remove the sensor or tighten any connections. 117  If no leaks are detected the system is ready to go.  For a test, a set pressure is needed, and the set calibration constants on the charge amplifier must be known. Make sure these are written down before starting tests.  With both valves closed, set the upstream pressure to a known value. Wait a few seconds and begin taking data. Open the valve and wait a few seconds (length of time needed for data depends on discharge time constant of charge amplifier). Usually about 20 seconds is plenty to get a good curve.  Repeat the experiment opening the ball valve at different speeds.  Repeat the experiment at several pressure levels to get an idea of the behaviour of the sensor.  118 Results  Sample Voltage Curve (fast opening time). The start time and maximum voltage are marked on the graph and the time difference between is the rise time for the sensor.    2 2.5 3 3.5x 104-0.500.511.522.533.544.55time (20kHz)voltageUDA212.mat1004psi26.64pC/bar15Nmtorque1/2insocketcapKistler26119 Sample Voltage Curve (slow opening time)  To obtain the pC/bar sensitivity for the system, you then plot max P/Voltage vs charge time/Pmax. You can linearly interpolate the sensitivity based on the constant used on the charge amplifier. To do this you note the difference in pressure between start time and end time, and that pressure (based off the charge amp pressure/voltage ratio, in our case 20) is compared to the maximum pressure based off the static sensor. You multiply this by the pC/bar sensitivity on the charge amp to give the corrected pC/bar constant based on this charge time. 4 4.5 5 5.5 6x 104-0.500.511.522.533.544.55time (20kHz)voltageUDA218.mat1007psi26.64pC/bar15Nmtorque1/2insocketcapKistler26120 500psi Kistler26  1000psi Kistler26  The corrected sensitivity is based on charge time = 0. In our case it would be 26.26 pC/bar for 500psi (34.5 bar) pressure or 26.63 pC/bar for 1000psi (69 bar) pressure. The sensitivity for our sensor was previously found to be approximately 26.6 pC/bar for 0-200bar, and 26.4 for 0-50 bar so we can say this method is an effective way of checking the sensitivity of our sensor.  y = -230.42x + 26.296 RΒ² = 0.9777 23.623.82424.224.424.624.82525.225.425.60 0.002 0.004 0.006 0.008 0.01 0.012pC/bar Sensitivity charget/Pmax Series1Linear (Series1)y = -229.16x + 26.627 RΒ² = 0.9847 24.424.624.82525.225.425.625.82626.226.40 0.002 0.004 0.006 0.008 0.01pC/bar Sensitivity charget/Pmax Series1Linear (Series1)121 Effect of torque on sensor and sensor repeatability For a separate sensor (AVL p206), the mounting torque and installation repeatability was adjusted to see its effect on sensitivity. For this experiment the sensor was installed at 15Nm and tested at 500psi and 1000 psi. It was then removed and installed again at the same torque setting and the experiment was repeated. It was then removed and installed at 10Nm and run at 1000psi. Finally it was removed and installed at 20Nm and run at 1000psi. For each pressure, a total of 6 points were taken (2 fast, 2 medium, and 2 slow charge time speeds). The results of those tests are below:  500psi tests AVL p206   It seems that the error in sensitivity is more affected by repeatability than mounting torque.  25.1525.225.2525.325.3525.40 5 10 15 20 25Sensitivity (pC/bar) Mounting Torque (Nm) 15Nm500-115Nm500-210Nm500-120Nm500-1122 1000psi tests AVL p206   While this looks like an increased torque load increases the sensitivity, the difference from 10-15Nm is very slight. Also compared to the figure below (all tests). The 20Nm 1000psi point appears to be an outlier as all other points are centred around 25.3 pC/bar.  All tests AVLp206  25.125.225.325.425.525.625.725.825.9260 5 10 15 20 25Sensitivity (pC/bar) Mounting Torque (Nm) 15Nm1000-115Nm1000-210Nm1000-120Nm1000-125.125.225.325.425.525.625.725.825.9260 5 10 15 20 25Sensitivity (pC/bar) Mounting Torque (Nm) 15Nm500-115Nm1000-115Nm500-215Nm1000-210Nm500-110Nm1000-120Nm500-120Nm1000-1123  The sensitivity for this sensor after factory calibration was between 25.2 and 25.4 (pC/bar). This test confirms the charge method for calibration.  Conclusion  Based on the above method, we can calibrate piezoelectric transducers accurately. It has been found that mounting torque does not affect the sensitivity of the sensor as much as repeatability. Previously on all engine tests, the transducer was torqued to an unknown level. A special socket has been fabricated to ensure that installation of the sensor is always at the correct torque level. In addition, we are working on a second method to monitor brake torque from the SCRE as a second check on these sensors.  A MATLAB processing code is available to post process data taken for pressure calibration. This is included in the electronic appendix.   124 Appendix E: High Range CO Calibration In June 2014 the high range CO analyzer was found to have a faulty mass flow meter. To address the solution with the data taken before this problem was realized, a different routine was used that would fit a line in the region where COM and COH were both valid (2000-3000ppm). COH was scaled down to the COM value to increase accuracy above 3000ppm.  The linear fit was based on a large amount of data COH and COM data taken for the paired nozzle tests between Sept 2013 and June 2014. The fit was written in the macros in the slow processing routine .xls file.  In Fall 2014 a new mass flow controller was installed, and the routine was put back to normal on November 18, 2014. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0167675/manifest

Comment

Related Items