Fuel Effects on Homogeneous Charge Compression Ignition Combustion

Transcription

1 Fuel Effects on Homogeneous Charge Compression Ignition Combustion by Jacob Richard Zuehl A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science (Mechanical Engineering) at the University of Wisconsin Madison 2009

2 ii Approved: Date: Professor Jaal B. Ghandhi Department of Mechanical Engineering University of Wisconsin Madison

3 iii Abstract Homogeneous charge compression ignition (HCCI) is one combustion strategy that has shown the possibility of both lower emissions and lower fuel consumption than spark ignition combustion. However, HCCI combustion can be sensitive to changes in fuel composition. The effects of fuel composition on HCCI combustion were experimentally investigated in a single cylinder 4-stroke SI engine adapted to run HCCI combustion with the Negative Valve Overlap (NVO) valve timing strategy. This strategy traps large amounts of hot residual gases in the cylinder to raise the mixture temperature and promote auto-ignition. Direct fuel injection was used, with one injection occurring during the NVO period to increase the residual temperature and for fuel reforming, and the other injection occurring just after intake valve opening. The timing and duration of the NVO injection were used to control the combustion phasing; an advance in timing or increase in delivered fuel mass caused an advance in combustion. The testing was performed at three operating conditions: 2100RPM 3bar IMEP, 2100RPM 2bar IMEP, and 3500RPM 2bar IMEP. Three gasoline-like fuels and an 87 octane pump gasoline were tested to examine the effects on fuel consumption and HC, CO, and NO x emissions. The trends for the fuels were consistent over all conditions and correlated well with combustion phasing. Most differences were small with a few exceptions. One of the test fuels, which was the only oxygenated fuel tested, showed significantly higher fuel consumption, which was mainly due to its lower energy density. This fuel also had lower HC emissions than the other test fuels; however, this may have been caused by the HC analyzer s lower response to oxygenated hydrocarbons. The pump

4 iv gasoline had significantly higher NO x emissions, but the reason for this difference is less clear and may have been caused by chemical effects. The combustion phasing of the four fuels was also compared. Heptane and Iso-Octane were also tested as they provided limits of low and high octane rating respectively. The phasing of the three test fuels, which ranged in RON from 66 to 88 and MON from 69 to 98, were all advanced compared to the pump gasoline by 2º at 2100RPM and 1º at 3500RPM. The phasing of Heptane was advanced by 6º at 2100RPM and 3º at 3500RPM, while the Iso-Octane was retarded 6º at 2100RPM and 3º at 3500RPM. Correlations of RON and MON with the change in phasing were poor, but Octane Index correlations were good.

5 v Acknowledgements First and foremost I need to thank my wonderful wife, Miyoshi, for working through this process with me. I know you ve made this experience infinitely easier and more enjoyable. Thank you Mom and Dad for all your support, guidance, and instilling in me the drive to do my best and always challenge myself. The financial support was quite helpful too. Thank you to the rest of my family for always caring about me and helping me whenever I needed it. I want to thank Professor Jaal Ghandhi for giving me this opportunity and teaching me more in a year and a half than I thought possible. I also wanted to thank the rest of the people at ERC for working so hard to provide one of the best preparations for engine research in the world. Thanks to Keitaro Okuno who rotated engine duties with me as we collected too much data. And thanks to Ralph Braun who provided a wealth of knowledge about the labs and one much needed ride to the ER. Thank you to Chevron for funding this research and Chris Hagen and Bill Cannella for sitting through my teleconferences. Finally, I would like to dedicate this to my Grandfathers. To my grandpa Roger who taught me to take pride in my work, whatever it might be. And to my grandpa Rashleger who I never met, but I know would be proud of my accomplishments in a field he loved.

6 vi Table of Contents Abstract... iii Acknowledgements...v List of Figures...x List of Tables... xiv Nomenclature...xv Chapter 1 Introduction HCCI Combustion Motivation Objectives...4 Chapter 2 Literature Review Introduction HCCI History Chemical Kinetics Low Temperature High Temperature HCCI Control NVO Valve Timing NVO Injection Second Injection In-Cylinder Deposits Fuel Effects on HCCI Low Temperature Heat Release Range Expansion Octane Number HCCI Fuel Index Fuel Consumption and Emissions...21 Chapter 3 Experimental Setup...23

7 vii 3.1 Laboratory Updates Engine Specifications Intake System Exhaust System Fuel System Engine Control Unit Oil System Cooling System dynamometer Specifications Emissions Bench Data Acquisition High-Speed (In-Cylinder Pressure) Low-Speed Heat Release Calculations...33 Chapter 4 Background Experiments Operating Range Repeatability Combustion Phasing Drift Testing Method Repeatability Results Optimization Investigation Introduction Combustion Phasing Emissions and Fuel Consumption...45 Chapter 5 Results and Discussion Fuel Specifications Combustion Phasing Comparison Methodology for Comparing Combustion Phasing Combustion Phasing Results 2100RPM 3bar IMEP Combustion Phasing Results 2100RPM 2bar IMEP Combustion Phasing Results 3500RPM 2bar IMEP...53

8 viii Octane Index Analysis Low Temperature Heat Release Performance Comparison Methodology for Comparing Engine Performance Engine Performance Results 2100RPM 3bar IMEP Engine Performance Results 2100RPM 2bar IMEP Engine Performance Results 3500RPM 2bar IMEP Confirmation of Fuel Comparison Trends Comparison of Emissions Measurements from Horiba Bench FTIR Analysis of Engine Performance Differences ISFC Lower Heating Value Pumping Work Combustion Efficiency Heat Transfer Specific Heat Ratio Effects AFR ISNO x Load Limit Comparison...85 Chapter 6 Conclusions and Recommendations Overview Background Experiments Combustion Phasing Performance Comparison Recommendations...92 References...94 Appendix...98 Appendix A High Pressure Fuel Pump Noise...98 Appendix B Injector Deposits...99 Appendix C Additional Explanation for Fuel Comparison Methodology Appendix D Comparison of CA50 Trends with Varied NVO% and SOI NVO...103

9 ix Appendix E Ratio of Specific Heat Calculation Appendix F Thermodynamic Engine Model...106

10 x List of Figures Figure 1.1.1: Comparison of SI and HCCI operating ranges...2 Figure 2.4.1: NVO valve timing and injection strategy...10 Figure 2.5.1: Progression of heat release rate plots with time starting with a clean engine...12 Figure 2.6.1: HCCI combustion with LTHR for various fuels...13 Figure 2.6.2: Reduction of LTHR with increasing intake temperature...14 Figure 2.6.3: In-cylinder temperature and pressure for HCCI with and without LTHR...14 Figure 2.6.4: HCCI with Standard Cam (low residual) and Re-breathing Cam (high residual) 15 Figure 2.6.5: HCCI operating ranges of different fuels...16 Figure 2.6.6: HCCI operating range of gasoline (Reference) and a diesel-like fuel (DLF 3)...17 Figure 2.6.7: Trends of changing OI with changing engine condition for different fuels...20 Figure 2.6.8: Correlations of combustion phasing and RON, MON, and HCCI Index...21 Figure 3.1.1: Pictures of laboratory setup...23 Figure 3.2.1: Pictures of cylinder head...24 Figure 3.4.1: Schematic of intake and exhaust systems...26 Figure 3.5.1: Schematic of original fuel system...28 Figure 3.5.2: Schematic of new fuel system...29 Figure 3.7.1: Schematic of oil system...30 Figure 3.7.1: Schematic of cooling system...31 Figure 4.1.1: Operating range of engine at 40%NVO and 300afTDC SOI_NVO...35 Figure 4.2.1: CA50 drift at 2100RPM 3bar IMEP 300afTDC SOI_NVO...37 Figure 4.2.2: Spark plugs before and after experiment...38 Figure 4.2.3: Daily running procedure...39 Figure 4.3.1: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI_NVO at 2100RPM 3bar IMEP on July 2, Figure 4.3.2: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI_NVO at 2100RPM 3bar IMEP on July 4,

11 xi Figure 4.3.3: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI_NVO at 2100RPM 3bar IMEP on July 16, Figure 4.3.4: Heat release rate plots for varied NVO% at 330afTDC SOI_NVO...43 Figure 4.3.5: Response of T IVC to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO...44 Figure 4.3.6: Response of Φ NVO to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO...44 Figure 4.3.7: Heat release rate plots for varied SOI NVO at 20%NVO...45 Figure 4.3.8: ISFC response to varied combustion phasing for all three optimization matrices...46 Figure 4.3.9: ISCO response to varied combustion phasing for all three optimization matrices...47 Figure : ISHC response to varied combustion phasing for all three optimization matrices...47 Figure : ISNO x response to varied combustion phasing for all three optimization matrices...48 Figure 5.2.1: CA50 of all Fuels compared to Base 2100RPM 3bar IMEP...51 Figure 5.2.2: CA50 of all Fuels compared to Base 2100RPM 2bar IMEP...52 Figure 5.2.3: CA50 of all Fuels compared to Base 3500RPM 2bar IMEP...53 Figure 5.2.4: OI RPM 3bar IMEP...54 Figure 5.2.5: OI RPM 2bar IMEP...54 Figure 5.2.6: OI RPM 2bar IMEP...55 Figure 5.2.7: NVO injection timing sweeps 2100RPM 3bar IMEP...56 Figure 5.2.8: NVO injection timing sweeps 2100RPM 2bar IMEP...57 Figure 5.2.9: Heat Release Rate for Heptane - Varied SOI NVO at 2100RPM 3bar IMEP...58 Figure : Heat Release Rate for Heptane - Varied SOI NVO at 2100RPM 3bar IMEP...58 Figure : NVO injection timing sweeps 2100RPM 3bar IMEP...59 Figure : Heat Release Rate for Test 1 - Varied SOI NVO at 2100RPM 3bar IMEP...60 Figure : Heat Release Rate for Test 3 - Varied SOI NVO at 2100RPM 3bar IMEP...60 Figure : LTHR Heat release rate v. Temperature for Heptane and Test 1 and Test Figure : Response of T IVC with SOI NVO for LTHR comparison...61

12 xii Figure 5.3.1: Varied NVO% and SOI NVO data for all fuels at 2100RPM 3bar IMEP...63 Figure 5.3.2: Varied NVO% and SOI NVO data for all fuels at 2100RPM 2bar IMEP...64 Figure 5.3.3: Varied NVO% and SOI NVO data for all fuels at 3500RPM 2bar IMEP...65 Figure 5.4.1: Emissions comparison between Horiba bench and MKS FTIR...67 Figure : Comparison at 3500RPM 2bar IMEP of Fuel Consumption and Fuel Efficiency...69 Figure : Comparison at 2100RPM 2bar IMEP of Fuel Consumption and Fuel Efficiency...70 Figure : Comparison at 2100RPM 3bar IMEP of Fuel Consumption and Fuel Efficiency...70 Figure : Response of η th with CA Figure : Response of η th with PMEP...72 Figure : Response of PMEP with HR NVO...73 Figure : η comb trends with CA Figure : Response of η th with η comb...74 Figure : Response of HT MAIN with CA Figure : Response of η th with HT MAIN...76 Figure : Response of AFR with CA50; Response of HT MAIN with AFR...77 Figure : Experimental and predicted response of η th with AFR...79 Figure : Test AFR v. Base AFR at fixed combination of NVO% and SOI NVO...80 Figure : Test HR NVO v. Base HR NVO at fixed combination of NVO% and SOI NVO...81 Figure : Response of T IVO with HR NVO...81 Figure : Response of Mass AIR with T IVO...81 Figure : Response of AFR with HR NVO...81 Figure : Experimental and predicted response of Mass AIR with HR NVO...82 Figure : Test v. Base AFR and CA50 at fixed combination of NVO% and SOI NVO...83 Figure : AFR v. CA50 with Base shifted by difference in CA Figure : Response of ISNO x with Φ...84

13 xiii Figure : Response of ISNO x with T MAX...85 Figure : Response of MRPR with IMEP for Base at 2100RPM 40%NVO 300afTDC SOI NVO...86 Figure : Response of MRPR with CA50 for all fuels...86 Figure : Response of CoV IMEP with IMEP for Base at 2100RPM 40%NVO 300afTDC SOI NVO...87 Figure : Response of CoV IMEP with CA50 for all fuels...88 Figure A.1: Continuous low speed data acquisition measurements with the high pressure pump on and off...98 Figure A.2: High speed data acquisition signals with high pressure pump on and off...99 Figure B.1: Response of total injection duration over time at the repeatability point Figure C.1: ISHC differences at a fixed NVO% and SOI NVO and at a fixed CA Figure C.2: ISNO x differences at a fixed NVO% and SOI NVO and at a fixed CA Figure C.3: η th differences at a fixed NVO% and SOI NVO and at a fixed CA Figure D.1: Response of CA50 with NVO% and SOI NVO for all fuels at 2100RPM 3bar IMEP Figure F.1: Diagram of intake portion of thermodynamic engine model...106

14 xiv List of Tables Table 2.6.1: Operating conditions for RON and MON tests...18 Table 3.2.1: Engine specifications...25 Table : Low speed data acquisition inputs...33 Figure 4.1.1: Operating conditions...36 Table 4.2.1: Repeatability point parameters...39 Table 4.2.2: Repeatability point variability...40 Table 5.1.1: Fuel Specifications...49 Table 5.2.1: CA50 and performance of all Fuels compared to Base 2100RPM 3bar IMEP...51 Table 5.2.2: CA50 and performance of all Fuels compared to Base 2100RPM 2bar IMEP...52 Table 5.2.3: CA50 of all Fuels compared to Base 3500RPM 2bar IMEP...54 Table 5.3.1: Approximate performance change from Base fuel at 2100RPM 3bar IMEP...63 Table 5.3.2: Approximate performance change from Base fuel at 2100RPM 2bar IMEP...64 Table 5.3.3: Approximate performance change from Base fuel at 3500RPM 2bar IMEP...65 Table 5.3.4: Approximate performance change from Base fuel at 2100RPM 3bar IMEP...66

15 xv Nomenclature AFR aftdc BDC CA50 CCD CI CO CO2 CoV IMEP DI EVC EVO HC HCCI HRNVO HRR HTHR HTMAIN IMEP ISCO ISFC ISHC ISNOx IVC IVO LHV LTHR MassAIR MassRES MON MRPR MRPR NOx NVO NVO% O2 OI ON PI PMEP PRF RON RPM air fuel ratio after firing top dead center bottom dead center crank angle at 50% burn combustion chamber deposits compression ignition (engine) carbon monoxide (emissions) carbon dioxide (emissions) coefficient of variation of IMEP direct injection exhaust valve close exhaust valve open hydrocarbon (emissions) homogeneous charge compression ignition NVO heat release heat release rate high temperature heat release heat transfer from IVC to EVO indicated mean effective pressure indicated specific CO emissions indicated specific fuel consumption indicated specific HC emissions indicated specific NOx emissions intake valve close intake valve open lower heating value low temperature heat release mass of air inducted mass of residual gas motor octane number maximum rate of pressure rise maximum rate of pressure rise nitrogen oxides (emissions) negative valve overlap percent of fuel injected during NVO oxygen octane index octane number port injection pumping mean effective pressure primary reference fuel research octane number revolutions per second

16 xvi RVP SI SOINVO TDC TIVC TIVO TMAX ηcomb ηth Φ γ Reid vapor pressure spark ignition (engine) start of NVO injection timing top dead center temperature at IVC temperature at IVC maximum in-cylinder temperature combustion efficiency thermal efficiency equivalence ratio specific heat ratio

17 1 Chapter 1 - Introduction 1.1 HCCI Combustion The two main combustion modes for internal combustion engines have been spark ignition (SI) and compression ignition (CI). Spark ignition engines use a homogeneous air-fuel mixture that is compressed and subsequently ignited by an electric spark. The timing of an SI engine is controlled by when the spark is fired. Compression ignition engines compress, to a higher level than SI engines, the air charge and then the fuel is injected into the air, which is hot enough after compression to ignite the fuel. This results in a highly inhomogeneous mixture. The timing of a CI engine is controlled by when the injection occurs. It has been shown that an engine can be run using a combination of the SI and CI strategies, by utilizing a homogeneous mixture, but relying on the compression to ignite the mixture. This approach is called homogeneous charge compression ignition (HCCI). Many different strategies have spread from this basic idea and they have different names; controlled auto-ignition (CAI), low temperature combustion (LTC), premixed-charge compression ignition (PCCI), etc. Though the type of combustion used in this study may be more accurately described by another term, it will be referred to as HCCI combustion. The relative advantages of HCCI depend on what conventional combustion mode to which it is being compared. The comparison to only SI combustion will be made as that is the relevant combustion mode in this study. Homogeneous charge compression ignition operates much leaner and has no flame front, which results in low in-cylinder temperatures and thus low NO x formation. Load in an HCCI engine is controlled by the amount of fuel, allowing unthrottled operation. This reduces pumping losses and decreases fuel consumption.

18 2 The challenge of producing a real HCCI engine has proven very difficult to overcome. HCCI combustion has no direct event to initiate combustion; the mixture is allowed to react and combust on its own. This is a problem because the combustion timing in HCCI engines is significantly affected by many parameters including intake temperature and pressure, coolant and oil temperature, load, engine speed, cylinder deposits, and exhaust gas recirculation rate. The parameters can be very difficult to control in real world applications, but some form of combustion phasing control is necessary. Since the mixture is homogeneous, the pressure and temperature, for the most part, are homogeneous and when ignition does occur, it occurs almost simultaneously throughout the mixture. This rapid release of energy causes high rates of pressure rise that can cause engine damage, and this limits the maximum achievable load. The minimum load is limited by having enough energy in the system to initiate and complete combustion. Homogeneous charge compression ignition is also limited to relatively low speed because the time available for reactions to take place gets too short. Figure shows a comparison of an engine s SI and HCCI operating ranges. Figure 1.1.1: Comparison of SI and HCCI operating ranges [4]

19 3 This small operating range has made the possibility of a fully HCCI engine impractical. However, dual mode SI-HCCI engine prototypes have been successfully tested and are the HCCI combustion strategy closest to production. These engines are able to transition from SI combustion at idle, high load, and high speed to HCCI combustion at low loads and speeds. This system is the closest to production because the necessary hardware is becoming more and more common in current production engines: variable valve timing (capable of trapping large amounts of internal residuals), direct injection, and relatively high compression ratio. 1.2 Motivation The lower fuel consumption and lower engine-out emissions of HCCI combustion helps abet many wide-ranging problems. Engine and vehicle manufacturers constantly spend large amounts of money and resources to meet new and increasingly more stringent government emissions and fuel economy regulations. The price of fuel can undergo sudden, large fluctuations [5]. Hydrocarbon combustion produces CO 2, which is a greenhouse gas and may contribute to global climate change. Other emissions like NO x, CO, and HC can cause local and regional human health problems, property damage, and many environmental problems. The extraction and production of fuels has been the root of many global and local conflicts. These problems are not likely to be solved in the near future by simply ridding the world of internal combustion engines as they are used in nearly all of the different types of transportation, and nothing has yet been able to come close to their combination of low price, weight, small size, and proven reliability. Many studies have been performed that show fully HCCI engines to be significantly affected by fuel composition [6]. However, very few studies have been done on an engine

20 4 capable of performing an SI-HCCI mode switch. Because this type of engine is the closest to production, it is important to know what effect fuel composition has on the combustion, emissions, and fuel consumption of this HCCI strategy. This is also important because fuels can vary significantly in composition from country to country, region to region, and even station to station. It may also be found that a certain fuel blend or additive works better than a conventional fuel. 1.3 Objectives The first objective was to perform several background experiments in order to better understand the operating range, repeatability, and phasing control of the engine. These initial tests were performed with only one fuel. From the results of these tests, a methodology for comparing the fuels was developed that included which operating conditions to use, how to reduce repeatability error, and how to use the phasing control. The main objective of this thesis was to study the fuel effects on the HCCI mode of an SI-HCCI dual-mode engine. Four fuels were tested at three operating conditions and differences in combustion, emissions, and fuel consumption were analyzed. Only a few fuels were tested because a more in-depth study was desired which would examine the differences of the fuels at several operating conditions and many combustion timings.

21 5 Chapter 2 - Literature Review 2.1 Introduction HCCI combustion has been researched for three decades and is nearing possible implementation in production engines. This is due in large part to an increased ability to control the combustion phasing. It is therefore important to understand how fuel composition affects the combustion process and the resulting emissions and fuel consumption. This section presents a brief background on HCCI research, with the main focus on previous HCCI fuel studies. 2.2 HCCI History The first study of HCCI combustion was performed on a 2-stroke engine by Onishi et al. [7] and originally called Active Thermo-Atmospheric Combustion (ATAC). Using large amounts of trapped residuals to increase the mixture temperature, this and subsequent studies [8] showed this new combustion s cycle-to-cycle stability was much better than that of normal 2- stroke operation and also had much lower HC emissions and fuel consumption. They also showed, through high speed photography, that ATAC has no single ignition point and flame propagation, but rather autoignition occurred at many points simultaneously. An operating range limited to low speeds and low to medium load, was noted. Building from these studies, the first study of HCCI combustion in a four-stroke engine was performed by Najt and Foster [9]. Their focus was not on engine performance, but on the chemical kinetics governing the combustion. With his study on the operating range of a 4- stroke HCCI engine, Thring [10] was the first to suggest that an engine use HCCI combustion for low speed and light load and switch to conventional SI combustion for high speed and load.

22 6 2.3 Chemical Kinetics To better understand HCCI combustion, a basic understanding of hydrocarbon autoignition mechanisms is needed. The chemical kinetics of hydrocarbon autoignition are extremely complex, however, simple models such as the one presented by Hu and Keck [11] provide useful insight into the autoignition process and fairly good agreement with experiments. The reactions used can be classified into four types: chain initiation, chain propagation, chain branching, and chain termination. The reactions can also be divided into those that dominate at low temperatures (<1000K) and those that dominate at high temperature (>1000K) [9] Low Temperature The seven reactions used by Hu and Keck [11] that dominate at low temperatures are presented below. The chain initiating reaction is hydrogen abstraction from a saturated hydrocarbon producing an alkyl radical and hydroperoxy radical (R1). (R1) The alkyl radical reacts with oxygen to form an alkylperoxy radical (R2), which isomerizes to form a hydroperoxyalkyl radical (R3). (R2) (R3) This hydroperoxyalkyl radical oxidizes and forms a radical (R4). This decomposes into a hydroxyl radical and hydroperoxide (R5). (R4)

23 7 (R5) The hydroxyl radical reacts with a fuel molecule to complete the chain cycle (R6). (R6) The hydroperoxide decomposition provides the chain branching reaction (R7). (R7) As the temperature rises above 800K, these oxidation reactions (R2-R7) become less important than the cycle below (R8-R10), which has the decomposition of hydrogen peroxide as the branching reaction. (R8) (R9) (R10) At temperatures below 1000K, production is much faster than decomposition and its concentration increases. Once above 1000K, decomposition accelerates and a pool of OH radicals is formed, which leads into the high temperature regime High Temperature Although thousands of reactions can take place above 1000K to complete combustion, a few basic reactions have been shown to be sufficient [12]. The three main initiation reactions are below (R11-R13). (R11)

24 8 (R12) (R13) The alkyl radicals then decompose to alkenes (R14, R15). (R14) (R15) The alkenes then react with oxygen to form water and carbon monoxide (R16). The carbon monoxide is then oxidized to carbon dioxide (R17), which brings combustion to completion. (R16) (R17) The high temperature reactions are extremely important because they are the main contributors to the heat release. As discussed above, the autoignition process is extremely complex, but a few general kinetic rules can be very helpful in qualitatively understanding HCCI combustion. Higher mixture temperatures accelerate autoignition. This can be achieved through higher initial temperature, lower heat transfer, and thermal stratification. Higher fuel and air concentrations also accelerate autoignition. This can be achieved through higher initial pressure or lower diluent concentration.

25 9 2.4 HCCI Control The lack of direct control of the ignition event in HCCI combustion can be overcome by controlling several parameters. The most common are intake temperature, intake pressure, compression ratio, valve timing, and direct injection (if used) timing. In a production engine, the control method would need to respond to changes quickly, which would be extremely difficult for the intake temperature and pressure strategies. Many designs for a variable compression ratio engine have been, and are being, attempted but none have reached production yet and is therefore an unlikely control strategy for the near future. The most likely control strategy would use variable valve timing and/or varied direct injection timing and these are the strategies discussed below NVO Valve Timing Controlling the valve timing allows control of the amount of trapped residuals [13]. The valve timing strategy discussed here is called negative valve overlap (NVO) where the exhaust valve closing (EVC) is relatively early and the intake valve opening (IVO) is relatively late, which results in recompression of high amounts of trapped residuals. The residual mass influences combustion in two opposing ways. The first is that the hot residuals increase the mixture temperature and can advance combustion, which allows the use of ambient intake temperatures. The second is that the residuals act as a diluent to retard combustion by lowering the concentration of reactants and increase combustion duration by absorbing thermal energy. Kaahaaina [14] varied EVC timing and found that as it was delayed, residual mass decreased and phasing advanced. Nitz [3] however, saw the opposite for the engine used in this study, as the residual mass increased, phasing advanced because the temperature increased. This discrepancy

26 10 is likely caused by the engines being operated in different regimes where the different effects of the residual mass dominate. Despite this, studies have successfully used valve timing to control combustion phasing [15] NVO Injection The timing of direct injection events has been found to effectively control HCCI combustion. Direct injection timing can affect phasing by varying the homogeneity of the mixture [16]. A dual injection strategy called NVO injection, Figure 2.4.1, can provide further phasing control and is the injection strategy used in this study. The first injection takes place during the NVO period and the second injection occurs sometime during the intake or compression stroke. Figure 2.4.1: NVO valve timing and injection strategy Many studies have investigated this strategy and provide similar results [17, 18, 19, 20, 21]. Combustion phasing advances as the percent of total fuel injected during the NVO period

27 11 increases. It also advances as the NVO injection timing advances. The NVO injection advances combustion by two effects [20]. The fuel molecules are fragmented during the NVO period and thus are more reactive during compression; this is called the chemical effect. The fuel vaporization during the NVO period will act to lower temperatures, but chemical reactions can release heat and raise temperature, this is called the thermal effect. Depending on the operating condition, the chemical effect can dominate the thermal effect or vice versa. The influence of varying the percent of total fuel injected during the NVO and the NVO injection timing on emissions and fuel consumption were less clear. All the study showed an increase in NO x with increasing percent or advancing timing, but the results for HC, CO, and fuel consumption were contradictory Second Injection Previous work on this engine [3] observed two important trends concerning the second injection. First, whether the second injection was from the DI or PI had little effect on combustion and performance. Second, timing of the second injection had some effect on combustion phasing, but it was less consistent than varying the timing of the first injection. It was suggested that for control purposes, the second injection be fixed and only the timing of the second injection be varied. The best performance was found when the timing was set near the intake valve open timing. 2.5 In-Cylinder Deposits Engines build up a thin layer of cylinder deposits on the combustion chamber surfaces over time and can affect knocking in an SI engine [22]. In addition, deposits formed in the injector can affect injector spray, which can affect engine performance [23]. Güralp et al. [24]

28 12 showed that the buildup of deposits can advance HCCI combustion phasing, Figure Cao et al. [18] found that NVO HCCI engines could produce significant amounts of soot, which could further increase in-cylinder deposits. This could cause significant variability in the NVO HCCI experiments, especially those experiments performed to test fuels as fuel composition affects deposit formation. Figure 2.5.1: Progression of heat release rate plots with time starting with a clean engine [24] 2.6 Fuel Effects on HCCI Many studies on the effects of fuel composition on HCCI combustion have been performed and a brief summary of the general findings are presented. The studies discussed in this section were performed with fully premixed charges and conventional valve timing.

29 Low Temperature Heat Release Several papers by Shibata et al. [25, 26, 27,6] describe their work on a phenomena called low temperature heat release (LTHR), which has been observed in many HCCI studies. Low temperature heat release is the heat released by the reactions taking place at temperatures below ~1000K (R1-R10). The LTHR is separate from the high temperature heat release (HTHR) as shown in Figures Figure 2.6.1: HCCI combustion with LTHR for various fuels [25] It can also be seen from Figure that the amount and timing of LTHR can vary with fuel composition. It was found that a good correlation existed between increasing LTHR amount or advancing LTHR timing and advanced HTHR timing for most cases. The LTHR increases both temperature and the concentration of radical species, which shorten the ignition delay of the HTHR. Through the use of 15 different fuels composed of varied amounts of 11 pure hydrocarbons, Shibata et al. ordered the hydrocarbon families according to their tendency to advance HTHR: n-paraffins>iso-paraffins>olefins>naphthenes=aromatics=oxygenates. The amount of LTHR decreased with increasing intake temperature, which caused the differences in HTHR of different fuels to decrease as well. Figure shows that at low intake temperature

30 14 the variation in LTHR and HTHR is large, but at high intake temperature there is no LTHR and there is almost no variation in HTHR. Figure 2.6.2: Reduction of LTHR with increasing intake temperature [27] It was also found that as pressure decreased, the LTHR amount decreased. Figure shows the in-cylinder temperature and pressure conditions at which LTHR occurs. Figure 2.6.3: In-cylinder temperature and pressure for HCCI with and without LTHR [6]

31 15 This is relevant to this study as engines with large amounts of residuals, like HCCI engines with NVO, have high temperatures and low pressures, which would fall in the region of no LTHR and therefore changes in phasing between fuels have been found to be small Figure [28,29]. Figure 2.6.4: HCCI with Standard Cam (low residual) and Re-breathing Cam (high residual) [28] Range Expansion In general, the operating range of HCCI combustion is limited to low speed and low to moderate load, but the actual size of the operating range can be very fuel specific, Figure [25]. The low-load limit is reached when combustion phasing becomes too retarded and misfire occurs, conversely the high load limit is reached when combustion phasing becomes too advanced and the maximum rate of pressure rise (MRPR) is too high.

32 16 Figure 2.6.5: HCCI operating ranges of different fuels [25] Ogawa et al. [30] were able to extend the high load limit compared to diesel using alcohols, which inhibit LTHR and retard HTHR, lowering MRPR. But when compared to gasoline, methanol was observed to advance combustion and extend the low load limit [31, 32]. Jeuland et al. [33] compared 15 fuel s operating ranges in an engine using NVO and found, in general, that diesel-like fuels extend the low-load limit, oxygenated fuels extend the high-load limit, and gasoline-like fuels fall in between. Zhong et al. [34] also found that adding diesel to gasoline extended the low load-limit. Some fuels, such as the diesel-like fuel (DLF3), were found to extend both limits as compared to gasoline (Reference), shown in Figure [33].

33 17 Figure 2.6.6: HCCI operating range of gasoline (Reference) and a diesel-like fuel (DLF 3) [33] In contrast to Jeuland et al. [33], Xie et al. [35] showed that methanol and ethanol advanced combustion as compared to gasoline and extended the low load limit. These opposing results show the need for more testing to achieve a better understanding of fuel effects on NVO HCCI. Dec and Sjöberg [36] found that some fuel s combustion phasing was more sensitive to changes in load, which suggests that less sensitive fuels have a larger operating range. This is discussed further in the HCCI Fuel Index section below Octane Number The octane rating was developed to quantify a fuel s ability to resist knocking in an SI engine. An octane number (ON) is given based on a fuel s knocking tendency as compared to a mixture of iso-octane (ON=100) and n-heptane (ON=0). Two octane ratings are given to a fuel based on different engine test conditions, research octane number (RON) and motor octane number (MON). The differences between the tests are given in Table and for the purposes

34 18 of this paper the main difference is that the MON test has lower in-cylinder pressure and higher temperature than the RON test. Table 2.6.1: Operating conditions for RON and MON tests RON MON Engine Speed 600 RPM 900 RPM Intake Temperature 52 C 149 C Ignition Timing 13 bftdc bftdc Knocking is the autoignition of the unburned fuel-air mixture ahead of the flame front. It would then be reasonable to think that if RON and MON are measures of autoignition of a fuel in an SI engine, they might describe an autoignition process such as HCCI combustion. Many studies have been performed that attempted to correlate RON and MON to HCCI combustion phasing. Some, such as Shibata et al. [25, 6] and Kalghatgi et al. [37] found some operating conditions where a good correlation existed, but other conditions had no correlation. Many other studies found no correlation [9, 38, 39, 40,41]. The conditions that had a correlation were found to have similar in-cylinder temperature and pressure history as those of the end gases in the RON and MON tests [37]. The conditions that deviated had in-cylinder temperatures that were too low or too high for a given pressure, the latter being the case of NVO HCCI HCCI Fuel Index As a result of the octane number s inability to describe HCCI combustion, several attempts have been made to create an HCCI index that would predict the combustion phasing of a fuel based on its composition. Kalghatgi et al. [37] showed that Octane Index (OI) could be used to correlate HCCI combustion phasing and fuel octane rating over a fairly large range of

35 19 fuels and operating conditions. Octane Index comes from the multiple linear regression between CA50 and RON and MON. (2.1) (2.2) (2.3) (2.4) (2.5) (2.6) The variable S is called the sensitivity of the fuel. Because the reference fuels for the RON and MON tests are iso-octane and n-heptane, the sensitivity of any mixture of these two fuels will be zero. The constant K is fuel independent and depends on the temperature and pressure history (engine and operating condition dependent) and K=0 for MON test conditions and K=1 for RON test conditions. Kalghatgi et al. [39] used OI required (OI 0 ) to describe how the fuel requirements of an engine can change with operating conditions. The OI 0 is defined as the OI that results in CA50=0º. Therefore, if the OI of a fuel at a given operating condition is less than the OI 0, then combustion phasing will be before TDC and if the OI is greater than the OI 0, then combustion phasing will be after TDC. Figure provides an example of this and how OI 0 can differ from the OI provided by the fuel and in this example, the 90RON, 79MON fuel (high sensitivity) would see less combustion variation than the 90PRF fuel (zero sensitivity) between the two operating conditions.

36 20 Figure 2.6.7: Trends of changing OI with changing engine condition for different fuels [39] The fact that the OI 0 of an engine changes and the OI of a fuel changes explains how fuels can have different operating ranges. If an engine s OI 0 varies a lot, which is the case for an HCCI engine, then a fuel with a large variation in OI (high sensitivity) will have more consistent combustion phasing and a larger operating range than a fuel with little variation in OI (low sensitivity) [39]. Kalghatgi suggests this would also be true of a SI/HCCI dual mode engine that would be switching between two very different operating modes [42]. The OI was found in one study to correlate for NVO engines up to 40% residual mass, but above 40% the correlation becomes worse [29]. With these high amounts of residuals, the temperatures are very high and phasing differences are small and necessarily become more difficult to correlate. Another HCCI Index was formulated by Shibata et al. [6]. This index was created from a large data set and uses the fuel s MON, percent volume of each hydrocarbon component, and percent volume of oxygenate to predict combustion phasing. The index works very well, Figure

37 , but is only calibrated for T comp15 between 670K and 820K, which is too low for NVO HCCI engines. Figure 2.6.8: Correlations of combustion phasing and RON, MON, and HCCI Index [6] Fuel Consumption and Emissions Even though HCCI combustion can provide emissions and fuel consumption benefits compared to SI combustion, it is still important to investigate the effect of fuel composition on emissions and fuel consumption. Little has been reported in this area using NVO HCCI engines, with even less discussion of how the fuel actually affects performance. Part of the difficulty has been that the emissions changes have been relatively small and the comparisons have been with a variety of parameters held constant. Oakley et al. [31] found that methanol and ethanol had slightly lower fuel consumption, 4%, and NO x emissions compared to gasoline and PRF 95. In the study by Zhong et al. [34], adding diesel to gasoline decreased fuel consumption, but only by about 1% for a 20% diesel mixture. The diesel mixture also produced lower CO emissions with slightly higher NO x, only 5ppm higher. Shen et al. [15] tested 10 gasoline-like fuels and attempted to correlate fuel consumption and emissions to RON, S, T10, T90, % Aromatics, % Olefins, % n-paraffins, % n-butane, and % Iso-Paraffins. Only a few reasonable correlations were found, but they were not consistent across all engine speeds. Xie et al. [35] found that

38 22 ethanol and methanol had lower NO x and HC emissions and comparable CO emission, but provided little explanation for the differences.

39 23 Chapter 3 - Experimental Setup 3.1 Laboratory Updates The engine test setup was originally built nine years ago. It has been used by several students for several different projects [1, 2, 3]. During this time, many aspects of the setup changed and resulted in many redundancies that caused difficulty in troubleshooting problems and making changes to the setup. Consequently, the intake, exhaust, fuel, oil, cooling, data acquisition, and electrical systems were all either extensively modified or replaced. Figure 3.1.1: Pictures of laboratory setup 3.2 Engine Specifications The engine is a Yamaha OU17 single cylinder, four stroke research motorcycle engine. As motorcycle engine blocks are integrated with the transmission, the transmission housing is used, but the transmission gears were replaced with a single rigid shaft to remove fluctuations at low speeds. The engine was also fitted with a prototype cylinder head, Figure

40 24 Figure 3.2.1: Pictures of cylinder head The cylinder head is a dual over head cam design. Five intake and six exhaust cams were available, but only one set was used in this testing. The cylinder head has three intake and two exhaust valves, two spark plug holes, a side-mount pressure transducer hole, and a centrally located direct injector hole. The side-mount transducer was removed, the hole was plugged; and a larger, liquid cooled Kistler 6061B pressure transducer was mounted in one of the spark plug holes. This was implemented to provide more consistent in-cylinder pressure measurements [3]. Four different pistons were available providing compressions ratios of 11.59, 12, 13, and 14.1, but only the piston was used in these experiments. The engine specifications as tested are listed below in Table

41 25 Table 3.2.1: Engine specifications Bore 73 mm Stroke 56.9 mm Displacement 250 cc Geometric Compression Ratio # of Intake Valves 3 # of Exhaust Valves 2 Intake Valve Lift 1.2 mm Exhaust Valve Lift 1.5 mm Intake Valve Open (IVO) 420 aftdc Intake Valve Close (IVC) -130 aftdc Exhaust Valve Open (EVO) 135 aftdc Exhaust Valve Close (EVC) 302 aftdc Fuel Injection Type Port and Direct 3.3 Intake System The intake system, Figure 3.4.1, could use either ambient room air or the building s pressurized air system to supply the engine. The engine was started with room air each day, but all tests were performed using the pressurized system. This allowed the intake pressure to be held at 100kPa for all tests. The pressurized system consisted of a pressure regulator upstream of three critical flow orifices. The airflow was measured based on which orifices were open and the upstream pressure. After the orifices, the pressurized system connected with the room air circuit. From this junction, the air passed through a Meriam Instruments 50MC2-2 laminar flow element, which provided a check to the airflow calculated from the critical flow orifices; the difference never varied by more than 1.5%. The air then passed through a surge tank and three heaters, which were used to raise the temperature of the intake slightly above ambient and maintain it at 35 C for all tests. The final components of the intake system were an intake box, where the 35 C temperature and 100kPa pressure were measured, and a throttle body, which was set at wide open throttle for all tests.

42 Exhaust System The exhaust system, Figure 3.4.1, was connected to the building s exhaust and also to the Horiba emission bench. The exhaust gas temperature was measured just after the exhaust port and then it passed through a mixing tube inside a surge tank where its pressure was measured. After this, the gas either went to the building s exhaust through a backpressure control valve that maintained the exhaust pressure at 100.5kPa, or the emission bench. The emission bench sample line consisted of a Unique Products FLT-1584 heated filter with a 0.03 micron filter, condenser, water trap, and heated sample line. The filter was necessary because of the significant amounts of soot produced. The heated filter and line were kept at 190 C. Figure 3.4.1: Schematic of intake and exhaust systems 3.5 Fuel System The engine could use either a port injector or direct injector, but only the direct injector was used for this study. The port injector was a Mitsubishi OR A73019 injector with a

43 27 60 degree cone angle and the direct injector was a Mitsubishi OR A73014 injector also with a 60 degree cone angle. The fuel pressure was measured just before the split between the port and direct injectors and regulated to 50bar. The fuel temperature was measured just before each injector. The fuel flow rate was measured with an Endress+Hauser Promass 63 Coriolis mass flow meter. Two fuel delivery systems were used for these experiments. The first was essentially the same system used in previous studies [1, 2, 3] done on this engine, Figure 3.5.1, the only significant change was replacing the single fuel tank with two fuel tanks. The system consisted of the two fuel tanks (base fuel and test fuel), a lift pump, filter, high pressure fuel pump, relief valves, and several two and three way valves. Depending on the orientation of the valves, the system could be run on one tank, switched to the other tank, the lines flushed, and then the engine run with the other fuel. This system prevented cross contamination of the fuels and worked well, however, the fuel switching procedure was complicated and required a relatively large flushing volume of 650ml. To simplify the switching procedure and reduce the flushing volume, a completely new system was built.

44 28 Figure 3.5.1: Schematic of original fuel system The second system, Figure 3.5.2, consists of two separate, but identical systems that join at a 3-way valve just before the fuel flow meter. Each system consists of a fuel tank, Tobul 4.5A20 piston-type accumulator, lift pump, filter, 2-way valve, and a 3-way valve. A 3-way valve determines if the accumulators are connected to either a high pressure nitrogen tank or the building s exhaust, which either pressurizes or depressurizes the system. Switching fuels could be done while the engine was running in about 30 seconds using only 100ml of fuel as opposed to about 5 minutes and 650ml with the engine turned off for the first system. It also removed electrical noise that was being caused by the high pressure pump, see Appendix A.

45 29 Building Exhaust 7 <1psi N 2 Base Tank 4 gal bar N 2 Test Tank 4 gal Filter Base 1 gal Test 1 gal Filter Low Pressure Pump T,P Low Pressure Pump Fuel Flow Meter 1 Flush Volume 100ml PI Injector DI Injector Figure 3.5.2: Schematic of new fuel system 3.6 Engine Control Unit The engine control unit was from Yamaha Motor Co. and uses proprietary software. It was capable of firing the spark plug up to twice per cycle and firing the port injector once and the direct injector up to twice per cycle (total of three possible injection events per cycle). Both the duration and timing of any of the spark and injections events could be changed. The port injector was not used and the spark was only used to initiate combustion, it was turned off while data were being taken. The system also outputted the values of various parameters to the lowspeed data acquisition system.

46 Oil System The oil system, Figure 3.7.1, consisted of a reservoir with a heater, oil pump, pressure relief valve, and filter. The oil drain plug was removed and used as a drain to the oil reservoir. From the reservoir, the oil pump moved the oil back to the engine with the relief valve maintaining the pressure at 3.5bar. The heater maintained the oil temperature at 90 C. Figure 3.7.1: Schematic of oil system 3.8 Cooling System The cooling system, Figure 3.8.1, used a pump to flow coolant through a heater, the engine, and then a heat exchanger. As the system is a closed system, a coolant cap maintained the pressure below 15psi. To remove heat from the system and keep the coolant at 90 C, a PID controlled solenoid valve allowed building water to pass through the heat exchanger.

47 31 Figure 3.7.1: Schematic of cooling system 3.9 Dynamometer Specifications A Sprague electric dynamometer and Reliance Electric MaxPak Plus VS Drive control system controlled the engine speed. An Omega LCDA-150 load cell measured the dynamometer torque. A flywheel coupling the dynamometer to the engine was used to reduce low speed vibrations, but consistent low speed operation was still limited to an engine speed of 1500RPM. The maximum engine speed was 5000RPM Emissions Bench A Horiba Instruments Inc. emissions bench measured the levels of CO 2, CO, O 2, HC, and NO x from the exhaust. Horiba AIA-23 nondispersive infrared analyzers measured CO and CO 2, a Horiba MPA-21A magneto pneumatic analyzer measured O 2, a Horiba FIA-23A flame ionization detector (FID) measured HC, and a Horiba CLA-22A chemiluminescence detector measured NO x. The bench was spanned at the beginning of each day and checked several times during the day.

48 Data Acquisition High-Speed (In-Cylinder Pressure) The in-cylinder pressure was measured by a Kistler 6061B water-cooled pressure transducer and its charge was converted to a voltage by a Kistler 5010 charge amplifier. To peg the in-cylinder pressure, a Kulite XTM-190 pressure transducer was located just before the intake port and its signal was amplified by an Entran PS-A amplifier. The crankshaft angular position was measured by a Hengstler RI58-O shaft encoder. It provided two sets of outputs, one pulse every 0.5 CA and one pulse once every revolution. The data from the transducers and encoder were recorded by a DSP Technology Inc. Porta 416 data acquisition system. The data acquisition s software, ACAP, averaged the pegged in-cylinder pressure over 240 cycles and this average pressure data was used to perform the in-cylinder calculations. ACAP s calculations of IMEP, Coefficient of Variation (CoV) of IMEP, and MRPR were also used Low-Speed Controlled by LabVIEW 8.2, a National Instruments CompactDAQ with four NI-9211 thermocouple modules, one NI-9205 differential voltage module, and one NI-9201 voltage module recorded the low speed test data. The data were sampled at 10Hz and averaged over 75 seconds for each test. A table with the inputs to the data acquisition system is given below.

49 33 Table : Low speed data acquisition inputs Type Source Type Source Thermocouple Upstream of Orifice Abs. Pressure Upstream of Orifice Thermocouple LFE Abs. Pressure LFE Thermocouple Intake Box Abs. Pressure Intake Box Thermocouple Exhaust Abs. Pressure Exhaust Thermocouple Oil Abs. Pressure Oil Thermocouple Coolant In Abs. Pressure Coolant Thermocouple Coolant Out Abs. Pressure Fuel Thermocouple DI Fuel Diff. Pressure LFE Thermocouple PI Fuel Torque Dynamometer Engine RPM ECU Voltage Flow Fuel Flow Meter Spark Timing ECU Voltage CO2% Emissions Bench SOI_NVO ECU Voltage CO% Emissions Bench SOI_INT ECU Voltage O2% Emissions Bench SOI_PI ECU Voltage HC ppmc3 Emissions Bench DUR_NVO ECU Voltage NOx ppm Emissions Bench DUR_INT ECU Voltage DUR_PI ECU Voltage Heat Release Calculations The in-cylinder pressure, engine speed, mass of fuel, mass of air, combustion efficiency (η c ), NVO%, and exhaust O 2 volume percent data were used in the heat release program. A heat release program originally developed in Matlab was implemented in IGOR Pro 6 and some calculations were modified and added to include the NVO period. The program provided the heat release rate, cumulative mass fraction burned, and heat transfer rate. The equivalence ratio during the NVO period was also calculated using Equation 3.1. (3.1)

50 34 Where H/C is the hydrogen to carbon ratio of the fuel, m residual is the residual mass, m fuel is the mass of fuel, O 2 % is the exhaust O 2 volume percent.

51 IMEP [bar] 35 Chapter 4 - Background Experiments 4.1 Operating Range The operating range of the engine was determined at a fixed mass fraction of fuel delivered during the NVO period (NVO%) of 40% and start of injection during NVO (SOI NVO ) of 300afTDC. The remaining 60% of the fuel was injected at -260afTDC. The upper limit was bounded by excessive noise, MRPR>5 bar/deg, and the lower by CoV of IMEP>5%. The results of the experiment are shown in Figure Data Point Fuel Study Operating Condition CoV of IMEP>5% MRPR>5bar/deg Engine Speed [RPM] Figure 4.1.1: Operating range of engine at 40%NVO and 300afTDC SOI NVO The upper and lower load limits were smooth above 2000 RPM and the general trend was consistent with previous studies performed on this engine [2, 3]. Combustion was possible at speeds as low as 1500 RPM, but it was very inconsistent and combustion was eventually lost most of time. For this reason and to provide a large range in operating conditions, the engine

52 36 speeds and loads in Table were chosen. Only three conditions were used to reduce the number of experiments as at least 25 data points would be taken at each condition for each fuel. The need for so many data points will be discussed in section 4.3 below. Figure 4.1.1: Operating conditions Engine Speed Load (IMEP) 2100 RPM 3 bar 2100 RPM 2 bar 3500 RPM 2 bar 4.2 Repeatability Combustion Phasing Drift During the initial experiments, it was observed that for a fixed engine speed, load, NVO%, and SOI NVO the combustion phasing could drift significantly. Figure shows the results from one day when the engine was run at fixed engine speed and load and three different NVO% and their effect on combustion phasing over time, Figure The engine was first run at the repeatability point, 40%NVO, (diamonds) and then run at the lowest NVO% possible (squares) for approximately 30 minutes. The engine was then run at a high NVO% (triangles) and the CA50 advanced from 9º to 5º over one hour and then stabilized at 5º for the next hour. The repeatability point was run again and found to be advanced 1.5º from the beginning of the day. Next, the low NVO% was run again for 2 hours and CA50 retarded 1º. Finally, the repeatability point CA50 was recorded again and had retarded slightly.

53 CA50 [aftdc] % 40% 80% 11:00 AM 12:00 PM 1:00 PM 2:00 PM 3:00 PM 4:00 PM Time Figure 4.2.1: CA50 drift at 2100RPM 3bar IMEP 300afTDC SOI NVO It was consistently observed that running at high NVO% advanced combustion and running at low NVO% retarded combustion at the repeatability point. This would be consistent with combustion chamber deposits (CCD) either increasing or decreasing, which would advance or retard combustion respectively [24]. Soot formation can affect CCD formation and it has been shown that NVO injection can form significant amounts of soot because the NVO conditions are favorable for soot production, i.e. rich and high temperature [18]. Indeed, the calculated equivalence ratio (Φ) during the NVO period for these experiments ranged from Φ=1 for 20%NVO to Φ=5 for 100%NVO. It is believed that higher NVO% leads to increased soot formation, which increases CCD formation and advances combustion. Although no soot measurements have been made during these experiments, it was clear that soot formation occurred as the emission sample filter becomes clogged with soot within several hours and the spark plug becomes coated within minutes, see Figure

54 38 Figure 4.2.2: Spark plugs before and after experiment Testing Method In order to reduce the engine variability discussed above, the following testing method was implemented for all experiments. A repeatability point was run as the first and second to last data point, and motoring data were taken as the last point each day. The repeatability point was also run every three to five experiments to ensure drifting had not occurred. If the CA50 at the repeatability point had drifted, the engine was run until the proper CA50 was achieved, and then experiments could resume. The repeatability point parameters are given in Table and a flowchart of the testing schedule is given in Figure This point was chosen because it was very stable, in the middle of the load range, and similar to the repeatability point used in previous studies.

55 39 Figure 4.2.3: Daily running procedure Table 4.2.1: Repeatability point parameters Engine Speed Load (IMEP) SOI_NVO NVO% CA RPM 3 bar 300 aftdc 40 5±1 aftdc Repeatability Results The variations of several parameters in the data taken at the repeatability point are given in Table The variation in these data, which were taken over 5 months, is similar to that of previous studies by Waldman and Nitz [2, 3] done over 3 months and 1 month respectively. One possible cause for repeatability issues is presented in Appendix B for possible future study.

56 40 Table 4.2.2: Repeatability point variability Average Std. Dev. Max Min Range IMEP % CA ISFC [g/kwhr] % ISCO [g/kwhr] % ISHC [g/kwhr] % ISNOx [g/kwhr] % Optimization Investigation Introduction As discussed in the Literature Review and Experimental Setup chapters, varied SOI NVO and NVO% were used to control combustion phasing. It was understood that increasing the NVO% or advancing SOI NVO advanced combustion phasing [19], but the effects on emissions and fuel consumption were less clear. This is important when comparing fuels because if all parameters are held constant when a new fuel is tested combustion changes can be observed, however, they may be caused by combustion phasing variations, in particular sub-optimal phasing for one fuel and optimal for another. It was therefore deemed necessary to perform an investigation to better understand how phasing (through variations in NVO% and SOI NVO ) affects emissions and fuel consumption and if any optimum exists. Data were taken at 2100RPM and 3bar IMEP. A matrix of five SOI NVO timings ( aftdc) and five NVO% (20-100%) was investigated. Zero NVO% was not used as stable combustion was not possible at any SOI NVO. Additionally, three matrices were compiled over the period of several days to verify the repeatability of the trends.

57 Combustion Phasing Combustion phasing was analyzed by comparing CA50. Figures 4.3.1, 4.3.2, and show the results from three different data sets. The data are in agreement with previous studies that increasing NVO% or advancing SOI NVO advanced combustion. Although all three data sets show very similar trends, the values are shifted slightly; this is believed to be caused by the variability discussed earlier. Figure 4.3.1: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO at 2100RPM 3bar IMEP on July 2, 2008

58 42 Figure 4.3.2: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO at 2100RPM 3bar IMEP on July 4, 2008 Figure 4.3.3: Response of CA50 to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO at 2100RPM 3bar IMEP on July 16, 2008

59 Heat Release Rate [J/deg] 43 Aroonsrisopon et al. [20] attributed the phasing change to the NVO injection affecting the chemical and thermal properties of the residual mass. For a fixed SOI NVO and increasing NVO%, Figure shows that the NVO heat release decreased slightly and the temperature at the start of compression (T IVC ) changed only slightly, Figure The lower NVO% resulted in higher NVO heat release because the equivalence ratio during the NVO period was more favorable to combustion, Figure The NVO heat release increased and yet combustion retarded, suggesting that only the chemical effect (production of intermediates) advanced combustion when NVO% increased Increased NVO% 20% 40% 60% 80% 100% Crank Angle [aftdc] Figure 4.3.4: Heat release rate plots for varied NVO% at 330afTDC SOI NVO

60 44 Figure 4.3.5: Response of T IVC to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO Figure 4.3.6: Response of Φ NVO to variation in NVO% (Total Fuel = 5.3mg) and SOI NVO

61 Heat Release Rate [J/deg] 45 However, for a fixed NVO% and advancing SOI NVO, Figure shows that NVO heat release increased and combustion advanced. The chemical effect likely contributed to advanced combustion as the fuel had more time to produce intermediates, but the thermal effect also contributed as the IVC temperatures increased, Figure Advanced SOI NVO Crank Angle [aftdc] Figure 4.3.7: Heat release rate plots for varied SOI NVO at 20%NVO Emissions and Fuel Consumption The trends of ISFC, ISCO, ISHC, and ISNO x were also consistent for all three data sets, though the values did vary. It was found that the emissions and fuel consumption correlated with CA50. The spread in values for a given CA50 was not found to correlate with any other parameters and no optimal combination of NVO% and SOI NVO was found to produce best performance. Fuel consumption decreased slightly as combustion retarded, Figure NO x emissions also decreased as combustion retarded and in-cylinder temperatures decreased, Figure This decrease in temperature caused HC emissions to increase, Figure The CO

62 ISFC [g/kwhr] 46 emissions trend is less clear as for most of the data there appears to be a minimum around 8º or 10º aftdc, Figure Carbon monoxide may be high as CA50 approaches TDC because more fuel was forced into the ring pack and when it comes out as the pressure decreases, the temperatures may be too low for complete oxidation CA50 [aftdc] July 2 July 14 July 16 Figure 4.3.8: ISFC response to varied combustion phasing for all three optimization matrices

63 ISHC [g/kwhr] ISCO [g/kwhr] CA50 [aftdc] July 2 July 14 July 16 Figure 4.3.9: ISCO response to varied combustion phasing for all three optimization matrices CA50 [aftdc] July 2 July 14 July 16 Figure : ISHC response to varied combustion phasing for all three optimization matrices

64 ISNO x [g/kwhr] CA50 [aftdc] July 2 July 14 July 16 Figure : ISNO x response to varied combustion phasing for all three optimization matrices From these trends, it was decided that the engine performance was mainly dependent on CA50 and not the specific combination of NVO% and SOI NVO required to achieve that CA50. This could be important in reducing the number of tests required to compare different fuels. For this study however, to ensure no possible optimum for a different fuel is missed, the full matrices were run for each test fuel.

65 49 Chapter 5 - Results and Discussion 5.1 Fuel Specifications Six fuels were tested during this study. The specifications are given in Table The Base fuel was a generic 87 octane pump gasoline provided by Haltermann Products. The specifications, combustion phasing results, and performance results will all be referenced to the Base fuel. Three Test fuels were provided by Chevron Corporation. Test 1 had lower RON, MON, octane sensitivity, and Reid vapor pressure (RVP); and higher energy density. Test 2 had lower RON, MON, and octane sensitivity; similar RVP; and higher energy density. Test 3 had higher RON, MON, and octane sensitivity; similar RVP; and lower energy density. Test 3 was also the only oxygenated fuel. The Heptane and Iso-Octane fuels were obtained from Sigma- Aldrich Co. Heptane had lower RON, MON, octane sensitivity, and RVP; and higher energy density. Iso-Octane had higher RON, MON, and energy density; and lower octane sensitivity and RVP. Limited data for Heptane and Iso-Octane were taken as they were only used to provide a larger range of RON and MON for the combustion phasing comparison. Table 5.1.1: Fuel Specifications Fuel FR46118 Base FR44868 Test 1 FR47085 Test 2 FR47090 Test 3 Heptane Iso-Octane Source Haltermann Chevron Chevron Chevron Sigma-Aldrich Sigma-Aldrich Calc.RON Calc.MON S=RON-MON T10 ( F) T90 ( F) BP BP C/H (wt) H/C/O (wt) 13.8/86.2/0 14.5/85.5/0 14.1/85.3/0 12.0/81.0/ /84.1/0 16.1/83.9/0 RVP(psi) ~ Spec.grav LHV (kj/kg)

66 Combustion Phasing Comparison Methodology for Comparing Combustion Phasing In order to compare the combustion phasing of each fuel at all three operating conditions, data were taken with the Base fuel; the fuel system was then switched to the Test fuel at the same NVO% and SOI NVO with the overall fueling rate adjusted to maintain the same load. This was performed several times for each fuel at each operating condition to ensure long-term repeatability Combustion Phasing Results 2100RPM 3bar IMEP For the 2100RPM 3bar IMEP condition with 40%NVO and 300afTDC SOI NVO, the combustion phasing differences of the fuels as compared to Base are shown in Figure All three Test fuels were advanced by approximately 2 compared to the Base fuel. Heptane was advanced approximately 6.5. Stable combustion with Iso-Octane was only possible at 100%NVO and 290afTDC SOI NVO (the most advanced timing possible) and therefore was compared to Base at that same condition. Iso-Octane was retarded approximately 6.5. The actual average change in CA50, fuel consumption, and emissions for each fuel is given in Table Of the test fuels provided by Chevron, Test 2 was most advanced followed by Test 3 and then Test 1, although the differences are very small and probably not statistically significant.

67 CA50 Test [aftdc] Advanced to 1 Test 1 Test 2 Test 3 Heptane Iso-Octane CA50 Base [aftdc] 8 10 Figure 5.2.1: CA50 of all Fuels compared to Base 2100RPM 3bar IMEP Table 5.2.1: CA50 and performance of all Fuels compared to Base 2100RPM 3bar IMEP Δ [Test-Base] ΔCA50 ΔISFC [g/kwhr] ΔISCO [g/kwhr] ΔISHC [g/kwhr] ΔISNOx [g/kwhr] Test Test Test Heptane Iso-Octane Combustion Phasing Results 2100RPM 2bar IMEP For the 2100RPM 2bar IMEP condition with 60%NVO and 300afTDC SOI NVO, the combustion phasing differences of the fuels as compared to Base are shown in Figure Like the 3bar IMEP condition, all three Test fuels were advanced compared to the Base fuel by

68 CA50 Test [aftdc] 52 approximately 2. Heptane was advanced approximately 5. Stable combustion was not possible with Iso-Octane. The actual average change in CA50, fuel consumption, and emissions for each fuel is given in Table Of the test fuels provided by Chevron, Test 1 was most advanced followed by Test 2 and then Test 3, although the differences are again very small to 1 Advanced 2 Test 1 Test 2 Test 3 Heptane CA50 Base [aftdc] 6 Figure 5.2.2: CA50 of all Fuels compared to Base 2100RPM 2bar IMEP Table 5.2.2: CA50 and performance of all Fuels compared to Base 2100RPM 2bar IMEP Δ [Test-Base] ΔCA50 ΔISFC [g/kwhr] ΔISCO [g/kwhr] ΔISHC [g/kwhr] ΔISNOx [g/kwhr] Test Test Test Heptane Iso-Octane NA NA NA NA NA

69 CA50 Test [aftdc] Combustion Phasing Results 3500RPM 2bar IMEP For the 3500RPM 2bar IMEP condition with 40%NVO and 300afTDC SOI NVO, the combustion phasing differences of the fuels as compared to Base are shown in Figure All three Test fuels were advanced compared to the Base fuel by approximately 1. Heptane was run at 330afTDC SOI NVO to ensure safe MRPR and was advanced approximately 3.5. Stable combustion with Iso-Octane was possible at the same NVO% and SOI NVO as the other fuels and was retarded approximately 3. The actual average change in CA50, fuel consumption, and emissions for each fuel is given in Table Of the test fuels provided by Chevron, Test 1 was most advanced followed by Test 3 and then Test 2, although the differences are very small. 8 Advanced to 1 Test1 Test2 Test3 Heptane Iso-Octane CA50 Base [aftdc] 8 Figure 5.2.3: CA50 of all Fuels compared to Base 3500RPM 2bar IMEP

70 RON RON MON MON Octane Index K=5.5 Octane Index K= Table 5.2.3: CA50 of all Fuels compared to Base 3500RPM 2bar IMEP Δ [Test-Base] ΔCA50 ΔISFC [g/kwhr] ΔISCO [g/kwhr] ΔISHC [g/kwhr] ΔISNOx [g/kwhr] Test Test Test Heptane Iso-Octane Octane Index Analysis The change in CA50 of each Test fuel from Base was compared to the fuel s RON, MON, and octane index. The K value in the octane index equation was calculated using linear regression, see equations 2.1 to 2.6. These results are presented in Figures 5.2.4, 5.2.5, and for the three conditions R 2 =0.90 R 2 = Base 20 Test 0 1 Test Test 80 3 Heptane 60 Iso-Octane R 2 =0.79 R 2 = R 2 = R 2 = CA CA50 Figure 5.2.4: OI RPM 3bar IMEP Figure 5.2.5: OI RPM 2bar IMEP

71 RON RON MON MON Octane Index K=5.3 Octane Index K= R 2 =0.88 R 2 =0.66 R 2 = CA R 2 =0.90 Figure 5.2.6: OI RPM 2bar IMEP Base Test 1 Test 2 Test 3 Heptane Iso-Octane CA50 The correlations between the change in CA50 and RON and MON for all three conditions were poor, which was expected as the in-cylinder temperatures are much higher and pressures much lower than the RON and MON tests. If considering RON and MON, the amount the phasing was retarded for Iso-Octane was much larger than anticipated, as Test 1, Test 3, and Heptane had a much larger change in RON and MON from Base than Iso-Octane but Iso-Octane had a larger change in CA50. Octane Index correlated better for 2100RPM 3bar and3500rpm 2bar, but was similar for 2100RPM 2bar as Iso-Octane was not run there and a larger spread in CA50 was not possible. The small change in CA50 for the three test fuels and Base suggests that any normal differences in gasoline from station to station would not cause significant changes in combustion phasing. If a very different fuel were used, OI might be able to predict the CA50 change, though more data would be needed to verify this.

72 CA50 [aftdc] Low Temperature Heat Release During the initial attempt to test Heptane, it appeared that as SOI NVO was retarded, CA50 advanced. A sweep of SOI NVO was performed with Heptane at 2100RPM 3bar and 2bar IMEP to confirm this observation and understand why this opposite trend was occurring. The results of the sweeps are presented and compared with the other fuels in Figures and Base Test 1 Test 2 Test 3 Heptane SOI NVO [aftdc] Figure 5.2.7: NVO injection timing sweeps 2100RPM 3bar IMEP

73 CA50 [aftdc] SOINVO [aftdc] 360 Figure 5.2.8: NVO injection timing sweeps 2100RPM 2bar IMEP All fuels followed the same trend and had similar slope of retarded CA50 with retarded SOI NVO. However, at 330afTDC for 3bar IMEP and 350afTDC for 2bar IMEP, the trend reversed for Heptane and CA50 advanced rapidly as SOI NVO was retarded. The reason for this behavior is that as SOI NVO was retarded, the NVO heat release decreased and the mixture temperature decreased. The compression temperature was then low enough to result in some LTHR, which in turn advanced main combustion. This appearance of LTHR and the increase of LTHR as SOI NVO was retarded is shown in Figure and for 3bar and 2bar IMEP, respectively.

74 Heat Release Rate [J/deg] Heat Release Rate [J/deg] Retarded SOI NVO Retarded SOI NVO Crank Angle [aftdc] Figure 5.2.9: Heat Release Rate for Heptane - Varied SOI NVO at 2100RPM 3bar IMEP Retarded SOI NVO Retarded SOI NVO Crank Angle [aftdc] Figure : Heat Release Rate for Heptane - Varied SOI NVO at 2100RPM 3bar IMEP This was the first time LTHR was observed in this engine and it was of interest to see if LTHR would occur with any of the other fuels. Sweeps of SOI NVO were performed at 2100RPM 3bar IMEP with Test 1and Test 3 to try to achieve LTHR. The SOI NVO was retarded until misfire occurred. The tests were performed at 100%NVO as this provided the most advanced

75 CA50 [aftdc] 59 CA50 and the SOI NVO could be retarded the most. As shown in Figure , neither fuel experienced an inversion of the SOI NVO v. CA50 trend, although the trend for Test 1 did flatten out, but the CoV of IMEP for these points was greater than 5%. The heat release rate plots for the Test 1 and Test 3 sweeps are given in Figures and respectively. There are small heat release spikes for Test 1 at SOI NVO of 370 and 380 and for Test 3 at SOI NVO of 360. At first glance, these appeared to be LTHR, however, they occurred at temperatures above 1000K, whereas the LTHR for Heptane occurred at 800K each time, and so these were not believed to be LTHR, Figure They were also not likely LTHR as the T IVC for Test 1 or Test 3 were not as low as for the Heptane cases with LTHR, Figure Test 1 Test SOI NVO [aftdc] 380 Figure : NVO injection timing sweeps 2100RPM 3bar IMEP

76 Heat Release Rate [J/deg] Heat Release Rate [J/deg] Retarded SOI NVO Retarded SOI NVO Crank Angle [aftdc] Figure : Heat Release Rate for Test 1 - Varied SOI NVO at 2100RPM 3bar IMEP Retarded SOI NVO Crank Angle [aftdc] Figure : Heat Release Rate for Test 3 - Varied SOI NVO at 2100RPM 3bar IMEP

77 T IVC [K] Heat Release Rate [J/deg] Heat Release Rate [J/deg] Heptane 3bar 330 3bar 340 2bar 350 2bar 360 3bar 290 (No LTHR) Test 1 SOI NVO -370 Test 1 SOI NVO -380 Test Temperature [K] Temperature [K] 1100 Figure : LTHR Heat release rate v. Temperature for Heptane and Test 1 and Test LTHR SOI NVO [aftdc] 380 Test 1 Test 3 Heptane(3bar) Heptane(2bar) Figure : Response of T IVC with SOI NVO for LTHR comparison

78 Performance Comparison Methodology for Comparing Engine Performance To compare the engine performance of the different fuels, a full test matrix of five SOI NVO and five NVO% for each test fuel was performed at all three conditions. From these data, the fuel consumption and emissions were compared for a given CA50 for the reasons discussed in the previous chapter. The data for all the test fuels showed very similar CA50 trends to varied NVO% and SOI NVO as those shown above for Base. These data are presented in Appendix D. The emissions and fuel consumption also correlated with CA50 and were independent of the combination of NVO% and SOI NVO required to achieve that CA50. Additional discussion and plots describing the need for this comparison in addition to the combustion phasing comparison is provided in Appendix C Engine Performance Results 2100RPM 3bar IMEP The results for the 2100RPM 3bar IMEP condition are presented below in Figure The ISNO x for all three test fuels are similar and about 50% lower than the Base. Base, Test 1, and Test 2 showed similar ISHC with Test 3 about 10% lower. The ISCO for Test 1 and Test 2 were similar, Base was slightly higher, and Test 3 was the highest. Test 1 and Test 2 had similar ISFC, Base had higher, and Test 3 had the highest. A summary of the test fuel performance as compared to Base is given in Table

79 ISHC [g/kwhr] ISFC [g/kwhr] ISNOx [g/kwhr] ISCO [g/kwhr] Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] 12 Figure 5.3.1: Varied NVO% and SOI NVO data for all fuels at 2100RPM 3bar IMEP Table 5.3.1: Approximate performance change from Base fuel at 2100RPM 3bar IMEP Test 1 Test 2 Test 3 ISFC -3% -3% 4% ISCO -12% -6% ISNOx -50% -50% -50% ISHC -12% Engine Performance Results 2100RPM 2bar IMEP The results for the 2100RPM 2bar IMEP condition are presented below in Figure The ISNO x for Test 1, Test 2, and Test 3 are similar and approximately 30% lower than Base. The ISHC for Base and Test 2 are similar and about 5% higher than Test 1 and Test 3. Test 3 had the highest ISCO and Base, Test 1, and Test 2 had similar levels, with Test 1 perhaps slightly lower. Test 1 and Test 2 had slightly lower ISFC than Base while Test 3 had about 7% higher.

80 ISHC [g/kwhr] ISFC [g/kwhr] ISNOx [g/kwhr] ISCO [g/kwhr] 64 90x Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] 6 8 Figure 5.3.2: Varied NVO% and SOI NVO data for all fuels at 2100RPM 2bar IMEP Table 5.3.2: Approximate performance change from Base fuel at 2100RPM 2bar IMEP Test 1 Test 2 Test 3 ISFC -2% -1% 7% ISCO 10% ISNOx -30% -30% -30% ISHC -5% -5% Engine Performance Results 3500RPM 2bar IMEP The results for the 3500RPM 2bar IMEP condition are presented below in Figure All fuels had similar ISNO x with Base having perhaps the highest. Base and Test 2 again had similar ISHC and Test 1 and Test 3 had similar and lower ISHC. Base, Test 1, and Test 2 had similar ISCO and ISFC and Test 3 had 10% and 8% higher ISCO and ISFC respectively.

81 ISHC [g/kwhr] ISFC [g/kwhr] ISNOx [g/kwhr] ISCO [g/kwhr] Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] 8 Figure 5.3.3: Varied NVO% and SOI NVO data for all fuels at 3500RPM 2bar IMEP Table 5.3.3: Approximate performance change from Base fuel at 3500RPM 2bar IMEP Test 1 Test 2 Test 3 ISFC -2% 8% ISCO 10% ISNOx ISHC -10% -5% -10% Confirmation of Fuel Comparison Trends It was deemed necessary to confirm the trends from the data reported in section as they were taken over a period of 4 months. Two comparisons were performed. The first was a comparison of Base data taken at the repeatability point each day and test fuel data from the same day that had the same CA50, referred to as Repeatability in Table The second comparison was performed after the matrix data were taken. In a single day, at 2100RPM 3bar IMEP, a test fuel was run at five CA50 timings (4º, 6º, 8º, 10º, 12º aftdc) and Base was run at

82 66 these same timings for a more direct comparison. Each test fuel was retested in this manner over a total period of three days. The results from these comparisons and the matrix data are provided in Table as the column labeled fixed CA50. Table 5.3.4: Approximate performance change from Base fuel at 2100RPM 3bar IMEP Matrices Fixed CA50 Repeatability 2100RPM 3bar 2100RPM 3bar 2100RPM 3bar Test 1 Test 2 Test 3 Test 1 Test 2 Test 3 Test 1 Test 2 Test 3 ISFC -3% -3% 4% -2% -1% 7% -1% -1% 6% ISCO -12% -6% 5% -6% -4% 4% -7% -3% 9% ISNO -50% -50% -50% -50% -30% -33% -32% -31% -28% ISHC -12% -7% -5% -3% -14% The comparisons from the full matrix data (from all three conditions), repeatability data, and single day data all showed consistent trends, but the relative differences varied slightly. The trends can be summarized in the following manner: ISNO x : Base > Test 1 = Test 2 = Test 3 ISHC: Base = Test 1 = Test 2 > Test 3 ISCO: Test 3 > Base > Test 1 = Test 2 ISFC: Test 3 > Base Test 1 = Test Comparison of Emissions Measurements from Horiba Bench to FTIR All the fuels had similar ISHC results except for Test 3, which was the only oxygenated fuel. The HC emissions were measured with a FID, which is known to have a low response to oxygenated hydrocarbons. In order to confirm the lower HC emissions for Test 3, a simple comparison was performed with an MKS 2030 FTIR, which can speciate the hydrocarbons and

83 THC [ppmc 3 ] NO x [ppm] HC [ppm C 3 ] CO 2 % CO% 67 more accurately measure oxygenated hydrocarbons. A CA50 sweep at 2100RPM 3bar IMEP was first performed with Base and then with Test 3. Measurements were recorded at the same time from both the Horiba emissions bench and the MKS FTIR. The results, Figure 5.4.1, show that the bench and the FTIR had good agreement in the trends and relative difference between Base and Test 3 for the CO, CO 2, and NO x measurements. The HC measurements showed good agreement in the trends; however, the bench measured lower emissions for Test 3 than Base while the FTIR measured higher emissions for Test 3 than Base CA50 [aftdc] 5:30 PM 6:00 PM Time CA50 [aftdc] 8 9 Base Bench Base FTIR Test 3 Bench Test 3 FTIR Figure 5.4.1: Emissions comparison between Horiba bench and MKS FTIR This comparison was not intended to provide a quantitative correction for the FID HC measurements, but rather some insight into the possible inaccuracy of the FID in measuring HC

84 68 emissions from oxygenated fuels. With that in mind, these limited results do suggest that the HC response of the bench was low for oxygenated fuels and the differences seen in HC emissions between Test 3 and the other fuels was likely incorrect. In order to better quantify the difference between the measurements, more experiments would be necessary. 5.5 Analysis of Engine Performance Differences ISFC The differences in fuel consumption between the fuels, Figures 5.3.1, 5.3.2, and 5.3.3, were significant and warranted further analysis. To understand why the fuel consumption differed, the role of five parameters, which were known to affect fuel consumption, was investigated Lower Heating Value The fuels had different lower heating values (LHV) as shown in Table above. Test 1 and Test 2 had LHVs about 1% higher than Base and Test 3 had a LHV about 9% lower than Base. Since ISFC is the mass of fuel per unit of work produced, a fuel with lower energy per mass (lower LHV) will likely have a higher ISFC if a given amount of work is desired. Because of this, some of the differences in ISFC between the fuels were likely due to their different LHV. The ISFC of each fuel was multiplied by its LHV and the differences in the actual efficiency of the fuel could be seen, Figures , , The differences in fuel consumption for the 3500RPM 2bar IMEP condition as compared to Base for Test 1, Test 2, and Test 3 were -1%, -1%, and 8%, respectively. These were approximately the same difference in LHV of the fuels and, as seen in Figure , the fuel consumption changes seem to all be caused by the differences in the LHV as no significant variation in fuel efficiency was observed. However, for

85 ISFC [g/kwhr] ISFC*LHV [MJ/kWhr] 69 the 2100RPM 2bar IMEP condition, Figure , there were some differences in fuel efficiency, 1% between Base and Test 1 and Test 2 and 2% between Base and Test 3. For 2100RPM 3bar IMEP, Figure , Test 1 and Test 2 had 2% higher fuel efficiency than Base. The efficiency of Test 3 was 4% higher than Base. These results suggest that the differences in LHV were likely a main contributor to differences in fuel consumption. In order to account for the effect of the variation in LHV, thermal efficiency (η th ) will be used instead of ISFC to compare the fuels in the following sections, as shown in Figure Also, as the 2100RPM 3bar case had the largest difference in efficiency, that case will be the focus of the following discussion Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] 8 Figure : Comparison at 3500RPM 2bar IMEP of Fuel Consumption and Fuel Efficiency

86 ISFC [g/kwhr] ISFC*LHV [MJ/kWhr] ISFC [g/kwhr] ISFC*LHV [MJ/kWhr] Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] 8 Figure : Comparison at 2100RPM 2bar IMEP of Fuel Consumption and Fuel Efficiency Base Test 1 Test 2 Test CA50 [aftdc] CA50 [aftdc] Figure : Comparison at 2100RPM 3bar IMEP of Fuel Consumption and Fuel Efficiency

87 ISNO x [g/kwhr] th CA50 T MAX [aftdc] [K] Base Test 1 Test 2 Test 3 Figure : Response of η th with CA Pumping Work Although the pumping work (PMEP) for each fuel varied, the differences were small and no correlation between pumping work and fuel efficiency existed, Figure

88 ISNO x [g/kwhr] th PMEP 2000[bar] T MAX [K] Base Test 1 Test 2 Test 3 Figure : Response of η th with PMEP In the interest of better understanding how the engine operates, it should be noted here that PMEP did correlate with the total amount of heat released during the NVO period (HR NVO ), Figure This relationship and the gas exchange of the engine are further explained in section below.

89 ISNO PMEP x [g/kwhr] [bar] HR 2000 NVO 2050 [J] 2100 T MAX [K] Base Test 1 Test 2 Test 3 Figure : Response of PMEP with HR NVO Combustion Efficiency Test 3 had very slightly higher combustion efficiency (η comb ) than the other fuels, but it was only by about 0.5%, Figure This may be caused by the FID inaccuracy discussed previously; the low HC measurement from the FID resulted in higher η comb. Figure shows that for a given η comb, the η th was different for different fuels and therefore, the variations in η comb are not the cause for the differences in η th.

90 ISNO x [g/kwhr] th ISNO x [g/kwhr] comb CA50 T MAX [aftdc] [K] Base Test 1 Test 2 Test 3 Figure : η comb trends with CA T MAX [K] comb Base Test 1 Test 2 Test 3 Figure : Response of η th with η comb

91 ISNO HT MAIN x [g/kwhr] [J] Heat Transfer The amount of heat transfer during the closed portion of the cycle from IVC to EVO (HT MAIN ) for the Test fuels was lower than Base, Figure HT MAIN was found to correlate well with η th, Figure The correlation seemed to be relatively independent of the fuel type for Base, Test 1, and Test 2. The η th for Test 3 is still higher, but the difference is less than for a given CA50. This suggests that the lower heat transfer for the Test fuels was a contributor to lower fuel consumption CA50 T MAX [aftdc] [K] Base Test 1 Test 2 Test 3 Figure : Response of HT MAIN with CA50

92 ISNO x [g/kwhr] th HT T MAX [K] MAIN [J] Base Test 1 Test 2 Test 3 Figure : Response of η th with HT MAIN Test 1 and Test 2 probably had lower heat transfer than Base for a given CA50 because they had higher AFR for a given CA50, Figure (left). At the right in Figure it can be seen that the response of HT MAIN with AFR was very much fuel independent for Base, Test 1, and Test 2. Higher AFRs had higher air mass (dilution), which lowered in-cylinder temperatures and thus heat transfer. The reason for Test 3 having lower heat transfer than Base was unclear as it had much lower AFR than Base. One possibility is that the oxygenated fuel used in Test 3 had a much higher heat of vaporization which affected the in-cylinder temperatures.

93 AFR AFR HT MAIN [J] CA50 [aftdc] AFR Base Test 1 Test 2 Test 3CA50 [aftdc] Base Test 1 Test 2 Test Figure : Response of AFR with CA50; Response of HT MAIN with AFR Specific Heat Ratio Effects As AFR increases, the ratio of specific heats (γ) increases (air has a higher γ than fuel) and η th increases. Figure shows that Test 1 and Test 2 have the highest AFR for a given CA50 followed by Base and then Test 3. With respect only to γ, it would be expected that Test 1 and Test 2 would have the highest η th followed by Base and then Test 3, however, as shown in Figure , Test 3 had the highest η th. Figure can help to explain this apparent contradiction. It shows the experimental data and two calculated values for η th using the Otto efficiency, Equation 5.1. The data labeled Otto Air/Fuel Only was calculated assuming only air and Iso-Octane as the working fluid and the data Otto w/ Residuals included the residual gases with air and Iso-Octane as the working fluid. The calculations are presented in Appendix E. (5.1)

94 78 First, Test 1 and Test 2 will be compared to Base. All three fuels fall on the same line of η th v. AFR. The slope of the line was similar to both predictions. This suggests that a second mechanism for changes in η th with AFR was the effect on γ. The magnitude of the experimental data was lower than both predictions, but this was expected as the predictions did not take into account any irreversibilities. Second, Test 3 will be compared. Test 3 had much lower AFR, but higher η th and the experimental data clearly did not fall in line with the other fuels, though the slope was similar. It was thought that perhaps the γ of Test 3 is sufficiently higher than the other fuels as to increase the overall γ and increase η th. The predictions (green) were made assuming the fuel to be 90% iso-octane (γ = 1.04) and 10% ethanol (γ = 1.14), a representative oxygenate used for these calculations because the actual oxygenate was not known. The predicted difference in η th between Test 3 and the other fuels was similar to the experimental difference. These predictions suggested that even a small addition of an oxygenate could significantly increase γ and could be the reason for the higher η th of Test 3. However, it should be emphasized that this was a very simple approach with several significant assumptions and would require further investigation and knowledge of the fuel composition to confirm this hypothesis.

95 th Otto w/ Residual (Slope = 0.47) Otto Air/Fuel Only (Slope = 0.63) Experiment (Slope = 0.71) AFR Base Test 1 Test 2 Test 3 Test 3 (Ethanol) Figure : Experimental and predicted response of η th with AFR AFR The difference in AFR between Base and Test 1 and Test 2 seems to be the main contributor to the difference in η th, so it is then important to understand why the AFR differs. The AFR of Base was expected to be slightly lower because it needed more fuel to achieve the same load because of its lower LHV. It differed by about 0.35 AFR at a fixed combination of

96 80 NVO% and SOI NVO, Figure , which was more than the amount expected from just the difference in LHV. AFR Test AFR AFR Base Figure : Test AFR v. Base AFR at fixed combination of NVO% and SOI NVO This difference was attributed mainly to a difference in the total heat released during the NVO period (HR NVO ). Test 1 and Test 2 had lower HR NVO than Base by about 2.5J, Figure As HR NVO decreased, the temperature at IVO (T IVO ) decreased, Figure The lower T IVO resulted in an increase of the mass of air inducted (Mass AIR ), see Figure , which in turn increased the AFR, see Figure A trend line of the AFR v. HR NVO data suggested that a 2.5J decrease in HR NVO, the same as the difference between Base and Test 1 and 2, would result in an increase in AFR of 0.2, which was close to the difference observed between the fuels. It was not exactly equal to the actual AFR difference, but provides one likely cause.

97 Mass AIR [mg/cycle] T IVO [K] AFR HR NVO,Test [J] T IVO [K] T IVO [K] HR 2.5[J] HR NVO,Base [J] HR NVO [J] HR NVO [J] Base Test 1 Test Base 2 Test 1 Test 2 Figure : Test HR NVO v. Base HR NVO at fixed combination of NVO% Figure : Response of T IVO with HR NVO and SOI NVO [J] T IVO [K] HR NVO [J] HR NVO [J] 35 Base Test 1 Test 2 Figure : Figure : Response of Mass AIR with T IVO Response of AFR with HR NVO The change in HR NVO changes the AFR because the intake was maintained at a constant pressure and if the residual mass temperature, T IVO, decreased the residual mass required less volume in the cylinder, which allowed more volume for fresh air. A simple thermodynamic model of the engine cycle was created to help confirm this hypothesis. The model uses only the

98 Mass AIR [mg/cycle] 82 measured T IVO and the residual mass amount (Mass RES ). Figure shows the comparison of the experimental and predicted correlations between Mass AIR and HR NVO. They both showed trends of increased Mass AIR with decreased HR NVO, though the slope and absolute values differed, which was expected as the model was quite simple. Details of the model are given in Appendix F Experiment Predicted Slope=-1.16 Slope= Indicated HR NVO [J] Figure : Experimental and predicted response of Mass AIR with HR NVO The AFR differed by only 0.35 for a given combination of NVO% and SOI NVO, but it differed by about 1 for a given CA50, Figure This increased difference for a given CA50 occurred because the AFR of Test 1 and Test 2 was higher than Base for a given combination of NVO% and SOI NVO, and the CA50 was also advanced for a given combination, Figure The test fuels having advanced CA50 increased the differences in AFR for a given CA50, Figure

99 T IVO [K] AFR CA50 Test [aftdc] AFR Test AFR Advanced AFR Base CA50 Base [aftdc] Figure : Test v. Base AFR and CA50 at fixed combination of NVO% and SOI NVO CA50 HR [aftdc] NVO [J] Base Test 1 Test 2 Figure : AFR v. CA50 with Base shifted by difference in CA ISNO x The ISNO x for Base was consistently higher than all the test fuels and an attempt to explain this observation was made. The best fuel independent correlation for ISNO x was found to be Φ, although there was still a significant spread. This relationship seemed reasonable; as Φ

100 ISNO x [g/kwhr] 84 decreased and the dilution increased, reducing in-cylinder temperatures, giving rise to less NO x. It was therefore expected that ISNO x would be the same for a given maximum in-cylinder temperature (T MAX ) independent of fuel type. This was not the case however as the three test fuels had similar ISNO x response to T MAX, but Base ISNO x was higher for a given T MAX, Figure This does not completely disprove the line of reasoning as the calculation of T MAX has many assumptions and a potentially large error range, but it does cause significant doubt as to dilution being the only reason for the differences in ISNO x. The true reason for the difference could just as readily have been a chemical cause that would have required further knowledge of the fuels and perhaps a modeling investigation to confirm Base Test 1 Test 2 Test 3 Figure : Response of ISNO x with Φ

101 ISNO x [g/kwhr] T MAX [K] Base Test 1 Test 2 Test 3 Figure : Response of ISNO x with T MAX Load Limit Comparison Although a direct comparison of the operating ranges of the fuels was not performed, the data obtained could provide some insight as to the different load limits of the fuels. In order to confirm the inferences made below, however, a direct comparison would need to be performed. The high-load limit is limited by either high NO x emissions or a high MRPR. If NO x emissions were the limiting factor, Base would be expected to have a lower high-load limit as its NO x emissions were significantly higher than the test fuels as discussed above. Figure shows that there was a very linear relationship between IMEP and MRPR for Base and it was expected that similar trends would be observed for the other fuels, such that if the MRPR for one fuel were lower than Base at a given IMEP and CA50, the MRPR would probably be lower at other IMEPs. The differences in MRPR were very small, Figure , and even though Test 3

102 MRPR [bar/deg] MRPR [bar/deg] 86 seemed to have lower MRPR, the amount would likely have provided an insignificant increase in the high-load limit IMEP [bar] 3.2 Figure : Response of MRPR with IMEP for Base at 2100RPM 40%NVO 300afTDC SOI NVO CA50 [aftdc] Base Test 1 Test 2 Test 3 Figure : Response of MRPR with CA50 for all fuels

103 CoV IMEP [%] 87 The low-load limit is determined by combustion stability, CoV IMEP. Like MRPR, a very linear relationship between IMEP and CoV IMEP for Base was observed toward the low loads, Figure , and it was expected that similar trends would be observed for the other fuels, such that if the CoV IMEP for one fuel were lower than Base at a given IMEP and CA50, the CoV IMEP would probably be lower at other IMEPs. Figure shows that there was no consistent difference in CoV IMEP for the fuels at a given CA50, especially at the more retarded CA50 where the engine would likely be run as lowest ISFC occurred there. Based on CoV IMEP, no significant difference in the low-load limit would be expected for the fuels. However, as the most advanced CA50 for a given fuel is limited by 100%NVO and 290afTDC SOI NVO, the test fuels could possibly have had lower load-limits than Base because for a given combination of NVO% and SOI NVO their combustion phasing was advanced. Even so, the difference would be expected to be minimal as the difference in CA50 between Base and the test fuels was only 2º IMEP [bar] 3.2 Figure : Response of CoV IMEP with IMEP for Base at 2100RPM 40%NVO 300afTDC SOI NVO

104 CoV IMEP [%] CA50 [aftdc] Base Test 1 Test 2 Test 3 Figure : Response of CoV IMEP with CA50 for all fuels

105 89 Chapter 6 Conclusions and Recommendations 6.1 Overview Homogeneous charge compression ignition combustion has shown promise as a combustion mode that offers improved efficiency and lower emissions for internal combustion engines. Before it can be implemented in production engines, several challenges must be overcome. One of these is that HCCI combustion can be significantly affected by fuel composition. This project investigated fuel effects on the HCCI combustion mode of an SI- HCCI dual-mode engine. Several background experiments investigating the operating range, repeatability, and phasing control of the engine were performed with the Base fuel to aid in developing a methodology for carrying out fuel effects experiments. A comparison methodology was created and used to test three fuels against the Base fuel. These fuels were tested at three operating conditions and the combustion phasing, emissions, and fuel consumption were compared. 6.2 Background Experiments A set of experiments were run to determine the operating range of the engine and define which operating conditions would be used in the fuel study. The operating range was found to be relatively small, which was consistent with other HCCI studies. Three conditions were chosen that provided a spread in engine speed and load: 2100RPM 3bar IMEP, 2100RPM 2bar IMEP, and 3500RPM 3bar IMEP. The repeatability of the engine was analyzed and found to vary significantly depending on the operating condition. It was found that in general running at high NVO% advanced combustion and running at low NVO% retarded combustion, and that these changes could occur

106 90 fairly quickly. As the fuel experiments would be run at varied levels of NVO%, a running procedure was created that relied on running at a repeatability point several times a day in between experiments to ensure combustion phasing had not drifted. If phasing had drifted, the engine was run at the repeatability point until normal phasing was achieved. An experimental matrix of varied NVO% and varied SOI NVO was run to investigate the phasing control s effect on CA50, ISFC, ISCO, ISHC, and ISNO x. The matrix was run three times and all three showed similar trends of advanced CA50 with advanced SOI NVO and increased NVO%. It was also observed that ISFC, ISCO, ISHC, and ISNO x correlated well with CA50 independent of the combination of NVO% and SOI NVO required to achieve that CA50. In general, ISFC and ISNO x increased with advanced CA50, while ISHC decreased and ISCO decreased slightly then increased with a minimum around 8º aftdc. It was decided that this matrix would be performed for each fuel at each operation condition to ensure any differences in behavior would be seen. 6.3 Combustion Phasing In addition to the four test fuels, the combustion phasing of Heptane and Iso-Octane were also tested to provide a larger spread of octane number. The CA50 of the three test fuels were advanced 2º compared to Base, while Heptane was advanced 6º and Iso-Octane was retarded 6.5º for both 2100RPM conditions. The three test fuels were advanced 1º, Heptane was advanced 3º, and Iso-Octane was retarded 3º at 3500RPM. RON and MON did a poor job of predicting CA50 changes with changes in octane number. Octane index correlated well with the change in CA50 and could be useful in predicting combustion phasing changes of other fuels, but more testing with other fuels would be necessary to confirm this. The small differences in CA50 between the

107 91 test fuels and Base suggested that other gasoline-like, full boiling point fuels would not produce significant changes in CA50 in this engine. Heptane showed a trend of first retarded and then advanced CA50 with retarded SOI NVO. As SOI NVO was retarded, HR NVO decreased and gave rise to lower in-cylinder temperatures. The compression temperature became low enough for a significant amount of low temperature reactions to occur and LTHR caused CA50 to advance. The LTHR was not observed with any other fuel. 6.4 Performance Comparison The comparison of the matrices for the four fuels at the three operating conditions showed the following trends: ISNO x : Base > Test 1 = Test 2 = Test 3 ISHC: Base = Test 1 = Test 2 > Test 3 ISCO: Test 3 > Base > Test 1 = Test 2 ISFC: Test 3 > Base Test 1 = Test 2 η th : Test 3 > Test 1 = Test 2 > Base These trends were confirmed by two other comparisons. The first was a comparison of the Base repeatability point data taken each day and the test fuel data point with similar CA50 from that day. The second comparison was a sweep of five CA50 timings taken for a test fuel and the Base in a single day, which was performed for each fuel for a total of three days.

108 92 The lower HC emissions of Test 3 were possibly an artifact of the FID s low response to oxygenated hydrocarbons. The HC emissions for Test 3 were found to increase relative to Base when an FTIR was used. The higher ISFC of Test 3 was caused by its lower LHV. Test 1 and Test 2 had lower ISFC than Base in part because of their higher LHV. Differences in PMEP and η comb were not found to correlate with differences in η th. One possible cause for Test 3 having had the highest η th is that the oxygenated portion of the fuel could have resulted in a higher γ. Test 1 and Test 2 had higher η th than Base probably because of their higher AFR, which increased γ and also lowered heat transfer by lowering in-cylinder temperatures. The cause for Base having had higher ISNO x was not clear. Base did have lower dilution, which should have resulted in higher temperatures, but the trends of ISNO x with T MAX did not support this. The inaccuracy in the T MAX estimation may have played a role, but it was possible that a chemical effect was the cause, which would have been impossible to conclude with the data available. All four fuels had similar MRPR and CoV for a given CA50. This suggested that neither the high-load nor the low-load limits would differ much between the fuels. 6.5 Recommendations Perhaps the most important recommendation for future fuel studies would be to reduce variability as much as possible. This would be crucial if differences in engine performance were the focus of a study as the changes seen in this study were fairly small and close to the repeatability level. Because so many data points, ~600, were need for this study, more variability than would be desired was allowed in the interest of time. It was seen that NO x

109 93 emissions varied greatly and it is believed that much of the variation could be reduced by reducing the variation of all the constant parameters, but especially IMEP. Narrowing the range that the CA50 of the repeatability point is allowed to fluctuate would likely reduce variability as well. These measures would increase the time needed to take data significantly, however, it has been sufficiently shown in this study that a full matrix is not needed for each fuel and only a few data points taken at a fixed CA50 with several combinations of NVO% and SOI NVO should be sufficient to compare engine performance. Further experiments would need to be performed in order to confirm the predictive ability of the Octane Index. PRFs could be easily made and run to provide more data points between Heptane and Iso-Octane. Fuels such as Ethanol or Methanol could be run to provide a limit of high sensitivity. It would also be interesting to run a conventional diesel fuel and E85 as they are readily available and the small difference in CA50 of the fuels already tested suggested that this combustion mode may be capable of running a very large range of fuels with little adjustment. From the emissions filter, spark plug, and other studies, it was clear that significant amounts of soot were being formed at some operating conditions. It would be of interest to determine the amount and type of soot being formed and their response to changes in CA50, fuel type, operating condition, etc. This would be important as the levels of soot seen in some other studies would be too high for a production engine.

110 94 References 1. Aroonsrisopon, T., An Experimental Investigation of Homogeneous Charge Compression Ignition Operating Range and Engine Performance with Different Fuels, M.S. Thesis, Mechanical Engineering Department, University of Wisconsin-Madison, Waldman, J., O., Investigating the Effects of Direct Fuel Injection during the Negative Valve Overlap Period in a Gasoline Fueled HCCI Engine, M.S. Thesis, Mechanical Engineering Department, University of Wisconsin-Madison, Nitz, D. G., Investigation into Expanding the Operating Range in a Gasoline Fueled Negative Valve Overlap HCCI Engine, M.S. Thesis, Mechanical Engineering Department, University of Wisconsin-Madison, Koopmans, L., et al., Demonstrating a SI-HCCI-SI Mode Change on a Volvo 5-Cylinder Electronic Valve Control Engine SAE Paper , Collantes, G. O., Electrifying Our Way to Fuel Economy: Regulatory Perspectives on Hybrid Vehicles, SAE Paper , Shibata, G. and Urushihara, T., Auto-Ignition Characteristics of Hydrocarbons and Development of HCCI Fuel Index, SAE Paper , Onishi, S., et al., Active Thermo-Atmosphere Combustion (ATAC) A New Process for Internal Combustion Engines, SAE Paper , Noguchi, M., et al., A Study on Gasoline Engine Combustion by Observation of Intermediate Reactive Products during Combustion, SAE Paper , Najt, P. and Foster, D., Compression-Ignited Homogeneous Charge Combustion, SAE Paper , Thring, R., H., Homogeneous-Charge Compression Ignition (HCCI) Engines, SAE Paper , Hu, H., Keck, J., Autoignition of Adiabatically Compressed Combustible Gas Mixtures, SAE Paper , Najt, P.M., Compression-Ignited Homogeneous Charge Combustion M.S. Thesis, Mechanical Engineering Department, University of Wisconsin-Madison, Milovanovic, N., Chen, R., Turner, J., Influence of the Variable Valve Timing Strategy on the Control of a HCCI Engine, SAE Paper , 2004.

111 Kaahaaina, N. B., Simon, A. J., Caton, P. A., Edwards, C. F., Use of Dynamic Valving to Achieve Residual-Affected Combustion SAE Paper , Shen, Y., et al., Fuel Chemistry Impacts on Gasoline HCCI Combustion With Negative Valve Overlap and Direct Injection SAE Paper , Marriot C. et al., Experimental Investigation of Direct Injection Gasoline for Premixed Charge Compression Ignited Combustion Phasing Control, SAE Paper , Urushihara, T., et al., Expansion of HCCI Operating Region by the Combination of Direct Fuel Injection, Negative Valve overlap and Internal fuel Reformation, SAE Paper , Cao, L., et al., Investigation Into Controlled Auto-Ignition Combustion in a GDI Engine With Single and Split Fuel Injections, SAE Paper , Waldman, J., et al., Experimental Investigation Into the Effects of Direct Fuel Injection During the Negative Valve Overlap Period in an Gasoline-Fueled HCCI Engine, SAE Paper , Aroonsrisopon, T., et al., A Computational Analysis of Direct Fuel Injection During the Negative Valve Overlap Period in an Iso-Octane-Fueled HCCI Engine, SAE Paper , Berntsson, A. W., et al., A LIF-Study of OH in the Negative Valve Overlap of a Spark- Assisted HCCI Combustion Engine, SAE Paper , Kalghatgi, G. T. Deposits in gasoline engines~a literature review, SAE Paper , Lindgren, R., et al., The influence of injector deposits on mixture formation in a DISC SI engine, SAE Paper , Guralp, O., et al., Characterizing the Effect of Combustion Chamber Deposits on a Gasoline HCCI Engine, SAE Paper , Oyama, K., et a;., The effect of fuel properties on low and high temperature heat release and resulting performance of an HCCI engine, SAE Paper Shibata, G., et al., Correlation of Low-Temperature Heat Release with Fuel Composition and HCCI Engine Combustion, SAE Paper , Shibata, G., et al., The Interaction Between Fuel Chemicals and HCCI Combustion Characteristics Under Heated Intake Air Conditions, SAE Paper , 2006.

112 Eng, J. A., et al., The effect of POx on the autoignition chemistry of n- heptane and isooctane in an HCCI engine, SAE Paper , Risberg, P., et al., The Influence of EGR on Autoignition Quality of Gasoline-Like Fuels in HCCI Engines, SAE Paper , Ogawa, H., et al., Combustion control and operating range expansion in an HCCI engine with selective use of fuels with different low temperature oxidation characteristics, SAE Paper , Oakley, A., et al., Dilution effects on the controlled auto-ignition (CAI) combustion of hydrocarbon and alcohol fuels, SAE Paper , Pucher, G. R., et al., Alternative combustion systems for piston engines involving homogeneous charge compression ignition concepts~a review of studies using methanol, gasoline and diesel fuel, SAE Paper Jeuland, N., et al., Engine and fuel-related issues of gasoline CAI (controlled auto ignition) combustion, SAE Paper , Zhong, S., et al., Promotive Effect of Diesel Fuel on Gasoline HCCI Engine Operated With Negative Valve Overlap (NVO), SAE Paper , Xie, H., et al., Comparison of HCCI Combustion Respectively Fueled With Gasoline, Ethanol and Methanol Through the Trapped Residual Gas Strategy, SAE Paper , Dec, J. E. and Sjoberg, M., Isolating the effects of fuel chemistry on combustion phasing in an HCCI engine and the potential of fuel stratification for ignition control, SAE Paper , Kalghatgi, G. T., et al., A method of defining ignition quality of fuels in HCCI engines, SAE Paper , Sjoberg, M. and Dec, J. E., Combined effects of fuel-type and engine speed on intake temperature requirements and completeness of bulk-gas reactions for HCCI combustion, SAE Paper , Risberg, P., et al., Auto-ignition quality of gasoline-like fuels in HCCI engines, SAE Paper , Koopmans, L., et al., The influence of PRF and commercial fuels with high octane number on the auto-ignition timing of an engine operated in HCCI combustion mode with negative valve overlap, SAE Paper , 2004.

113 Kalghatgi, G. T. and Head, R. A., The available and required autoignition quality of gasoline-like fuels in HCCI engines at high temperatures, SAE Paper , Kalghatgi, G. T., Auto-Ignition Quality of Practical Fuels and Implications for Fuel Requirements of Future SI and HCCI Engines, SAE Paper , 2005.

114 Fuel [bar] Intake [kpa] O2 % 98 Appendix Appendix A High Pressure Fuel Pump Noise Figure A.1 shows the continuous measurements of five of the low speed data acquisition channels taken as the electric high pressure pump was turned on and then turned off. Figure A.2 shows the high speed data acquisition measurements with the pump on and off Pump Off Pump On Pump Off RPM Mass Fuel [mg] Time [sec] Figure A.1: Continuous low speed data acquisition measurements with the high pressure pump on and off

115 Pegging Pressure [kpa] In-cylinder Pressure [kpa] Pump On Pump On Pump Off Pump Off Crank Angle [aftdc] Crank Angle [aftdc] Figure A.2: High speed data acquisition signals with high pressure pump on and off Appendix B Injector Deposits Figure B.1 shows that the total injector duration command necessary to inject the same amount of fuel varied during the course of the study. It appears that Test 2 accelerated nozzle deposits and Test 3 effectively cleaned nozzle deposits.

116 Duration [usec] Mass FUEL [mg] IMEP [bar] Emissions Calc Flow Meter /1/ /1/2008 1/1/2009 3/1/2009 Figure B.1: Response of total injection duration over time at the repeatability point Appendix C Additional Explanation for Fuel Comparison Methodology The engine performance was not compared at a fixed NVO% and SOI NVO because the CA50 varied at a fixed NVO% and SOI NVO for the different fuels. The parameters of interest in this study all correlated with CA50 and therefore comparisons of engine performance at a varied CA50 would have been misleading. Also, production engines are calibrated to run at a fixed CA50, so comparing performance at a fixed CA50 was more representative of changes that would be experienced by a production engine. Figure C.1 shows that at a fixed NVO% and SOI NVO Base had higher ISHC by 10%, but at a fixed CA50 there was no significant difference.

117 ISNO x Test [g/kwhr] T IVO [K] ISNO x [g/kwhr] ISHC Test [g/kwhr] T IVO [K] ISHC [g/kwhr] 101 Figures C.2 shows that at a fixed NVO% and SOI NVO Base had similar ISNO x, but at a fixed CA50 Base ISNO x is 50% higher. Figure C.3 shows that at a fixed NVO% and SOI NVO Base had lower η th by 1%, but at a fixed CA50 Base had lower η th by 3% ISHC Base [g/kwhr] HR NVO [J] CA50 [aftdc] 10 Base Test 1 Test 2 Figure C.1: ISHC differences at a fixed NVO% and SOI NVO and at a fixed CA ISNO x Base [g/kwhr] HR NVO [J] CA50 [aftdc] 10 Base Test 1 Test 2 Figure C.2: ISNO x differences at a fixed NVO% and SOI NVO and at a fixed CA50

118 T IVO [K] th,test th th th,base HR NVO [J] 0 35 th CA50 [aftdc] Base Test 1 Test 2 Figure C.3: η th differences at a fixed NVO% and SOI NVO and at a fixed CA50 Appendix D Comparison of CA50 Trends with Varied NVO% and SOI NVO

119 103 Figure D.1: Response of CA50 with NVO% and SOI NVO for all fuels at 2100RPM 3bar IMEP Appendix E Ratio of Specific Heat Calculation Air/Fuel Only: w/ Residuals:

120 104

121 105 Appendix F Thermodynamic Engine Model Figure F.1: Diagram of intake portion of thermodynamic engine model Global Equation IVO