Effect of Ambient Conditions and Fuel Properties on
advertisement
Effect of Ambient Conditions and Fuel Properties on Homogeneous Charge Compression Ignition Engine Operation by Morgan M. Andreae M.E.M. Engineering Management Dartmouth College, 1997 B.E. Mechanical Engineering Dartmouth College, 1997 B.A. History Haverford College, 1993 Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy at the Massachusetts Institute of Technology June 2006 MASSACHUSETTS INSTITUTE OF TECHNOLOGY C 2006 Massachusetts Institute of Technology All rights reserved JUL 14 2006 LIBRARIES Signature of Author .:........... of Signature A uthor ~~~~~..............f .... .... :......................... Department of Mechanical Engineering ARCHNES May 5 th, 2006 C ertified by ........... .................. : . ...................................................... Wai K. Cheng Professor of Mechanical Engineering Thesis Supervisor AA Approved by ...................................................... Lallit Anand Chairman, Departmental Graduate Committee MIT'Libraries Document Services Room 14-0551 77 Massachusetts Avenue Cambridge, MA 02139 Ph: 617.253.2800 Email: docs@mit.edu http://libraries.mit.edu/docs DISCLAIMER OF QUALITY Due to the condition of the original material, there are unavoidable flaws in this reproduction. We have made every effort possible to provide you with the best copy available. If you are dissatisfied with this product and find it unusable, please contact Document Services as soon as possible. Thank you. Some pages in the original document contain text that runs off the edge of the page. ( Pages 97 - 100) 2 Effect of Ambient Conditions on Homogeneous Charge Compression Ignition Engine Operation by Morgan M. Andreae Submitted to the Department of Mechanical Engineering on May 5 th 2006 in partial fulfillment of the requirements for the Degree of Doctor of Philosophy Abstract Practical application of Homogeneous Charge Compression Ignition (HCCI) combustion must demonstrate robust responses to variations in environmental conditions. This work examines the impact of ambient conditions and fuel changes on HCCI engine operation, and evaluates cam phasing as a mechanism to compensate for these changes. Experiments were carried out on a modified 2.3 L 14 production engine, and HCCI operation was achieved by the use of residual trapping by negative valve overlap. The first phase of the project examined the impact of changes in intake air temperature and humidity on HCCI operation. Exhaust cam phasing was used to control load, and intake cam phasing was use to produce a change in combustion phasing. Cam timing control was largely able to compensate for changes in combustion due to changes in air temperature and humidity. Higher intake air temperature advanced combustion phasing and resulted in a 1 bar reduction of the net indicated mean effective pressure (NIMEP) at the high load limit for lower engine speeds. Intake air temperature did have more of an impact during lean operation. Higher intake air humidity delayed combustion phasing. During stoichiometric operation, this delay allowed a small extension (a few tenths of a bar in NIMEP) in the high load limit. During lean operation, the delay in combustion timing resulted in a reduction of the high load limit. The second phase of the project examined the impact of market fuel composition variations on HCCI operation. Twelve test fuels were created to vary the composition of 5 fuel properties: Research Octane Number (RON), Reid Vapor Pressure (RVP), olefin content, aromatic content, and ethanol content. The test fuels were blends of different commercial refinery streams and contained hundreds of different hydrocarbons to be representative market gasolines. Fuel type was found to have only a small impact on the HCCI operating range, and cam phasing was largely able to compensate for changes in fuel composition. The main effect of the different fuel composition appeared to be differences in ignition delay. Thesis Supervisor: Wai Cheng, Professor of Mechanical Engineering 3 4 Acknowledgements I would like to thank my thesis committee for their invaluable guidance over the course of my thesis work. It has been a pleasure to work with all of you. I would like to thank my thesis advisor Prof. Wai Cheng for his support and advice during my time here at MIT. I would like to thank Prof. John Heywood for his input and his perspective on engine research. I would like to thank Prof. Jim Cowart for his assistance in setting up the project, and more importantly for his input and encouragement over the last two years. I would like to thank Prof. Bill Green for his advice in areas ranging from hydrocarbon chemistry to future career paths. I am particularly grateful for the assistance and support of Profs. Heywood and Cheng as I transitioned to the HCCI project. I would also like to thank Profs. Heywood and Cheng for the opportunity to work in the Sloan Automotive Laboratory. I would like to thank the project sponsors Ford Motor Company and BP for the financial support to make the project possible. In particular I'd like to thank Tom Kenney of Ford and Yi Xu of BP for the time that they have put into the project, for their input, and for their support throughout the project. I would like to thank the Sloan Automotive Lab staff. Thanks to Nancy Cook, Janet Maslow, and Karla Stryker for keeping things running smoothly. Thank you to Thane DeWitt and Raymond Phan for all of your help in the lab. I would like to thank all of my fellow students in the lab for their help in classwork, in preparing for qualifying exams, in experimental work in the lab, and in preparing for defending my thesis. Finally I'd like to thank my family for their support during my time here at MIT. 5 6 Table of Contents Abstract ............................................................................................................................... 3 Acknowledgem ents...................................................................................................... 5 List of figures and tables................................................................................................ 10 Chapter 1: Introduction.................................................................................................. 19 1.1 M otivation............................................................................................................... 19 1.2 Project Objectives................................................................................................ 20 1.3 Background.............................................................................................................20 1.4 Experim ental Strategy......................................................................................... 22 Chapter 2: Experim ental Setup...................................................................................... 26 2.1 Engine..................................................................................................................... 26 2.1.1 Engine.............................................................................................................. 26 2.1.2 Custom Cam Shafts for Residual Trapping.................................................. 26 2.1.3 Engine controller............................................................................................ 27 2.1.4 Intake M anifold........................................................................................... 27 2.1.5 Exhaust M anifold......................................................................................... 27 2.2 Fuel......................................................................................................................... 28 2.3 Engine coolant.................................................................................................... 28 2.4 Intake air conditioning......................................................................................... 28 2.5 Dynam ometer & M otor ...................................................................................... 28 2.6 M easurem ent & D ata Acquisition ...................................................................... 29 2.7 SI Baseline .............................................................................................................. 33 2.8 M odeling Tools.................................................................................................... 35 Chapter 3: Control of H CCI............................................................................................ 37 3.1 Control variable and the state of the cylinder..................................................... 37 3.2 Exhaust Cam Phasing ........................................................................................ 40 3.2.1 Exhaust Cam Phasing: Load & Combustion Phasing................. 40 3.2.2 Exhaust Cam Phasing and Efficiency......................................................... 44 3.2.3 Exhaust Cam Phasing and Em issions .............................................................. 45 3.3 Intake Cam Phasing............................................................................................. 48 7 3.3.1 Intake Cam Phasing: Load and Combustion Phasing ................. 48 3.3.2 Throttled operation........................................................................................ 58 3.3.3 Intake Cam Phasing: Efficiency....................................................................... 58 3.3.4 Intake Cam Phasing and Emissions ................................................................. 60 3.4 Fueling Rate as a Control Variable..................................................................... 3.4.1 Fueling Rate: Constant Cam Timing - Load & Combustion Phasing ...... 65 65 3.4.2 Fueling Rate: Constant Cam Timing - Efficiency & Emissions.......... 66 3.4.3 Fueling Rate: Constant Fueling - Load & Combustion Phasing.......... 68 3.4.3 Fueling Rate: Constant Fueling - Efficiency & Emissions........................... 70 3.5 Operating Limits................................................................................................... 73 3.5.1 Low Load Limit............................................................................................ 73 3.5.2 High Load Limit ........................................................................................... 75 3.5.3 Summary of Operating Limits ..................................................................... 84 3.6 Cam Phasing and the High & Low Load Limits .................................................... 85 3.6.1 High Load Limit ........................................................................................... 85 3.6.2 Low Load Limit ............................................................................................ 89 3.6.3 Repeatability ................................................................................................ 92 Chapter 4: Impact of Ambient Conditions on HCCI Operation .................................... 93 4.1 Intake Air Temperature........................................................................................ 93 4.1.1 Effect of intake air temperature at a point .................................................. 93 4.1.2 Operating Range Limits and Temperature................................................... 96 4.1.3 Operating Range and Engine Speed .............................................................. 103 4.1.4 Operating Range and Equivalence Ratio ....................................................... 110 4.2 Intake Air Humidity............................................................................................. 115 4.2.1 Effect of humidity at a point ......................................................................... 115 4.2.2 HCCI Operating Range and Humidity........................................................... 119 4.3 Fuel Effects....................................................................................................... 122 4.3.1 Fuel Composition Range............................................................................. 122 4.3.2 Effect of fuel composition on combustion.................................................... 126 4.3.3 Operating Range and Fuel Type .................................................................... 132 4.3.4 Impact of octane sensitivity ........................................................................... 152 8 4.3.5 O ther fuel com ponents................................................................................... 155 4.3.6 D ifferences in combustion ............................................................................. 158 4.3.7 Fuel com position and emissions .................................................................... 163 5.1 Operating Range Comparison............................................................................... 169 5.2 H CCI and SI Efficiency........................................................................................ 171 5.3 H CCI & SI Em issions Comparison ...................................................................... 175 Chapter 6: Conclusions................................................................................................... 178 Appendix A : Combustion Phasing ................................................................................. 181 A ppendix B: Throttled H CCI Operation ........................................................................ 183 A ppendix C: Local Knock Criterion............................................................................... 187 A ppendix D: A dditional Fuel Data................................................................................. 189 Appendix E: Volum etric Efficiency ............................................................................... 195 References....................................................................................................................... 197 9 List of figures and tables Figure 1.1 Typical SI valve timing (left) and valve timing for HCCI using residual trapping (right).......................................................................................................... 23 Figure 1.2 Cylinder pressure data for HCCI combustion with residual trapping over the course of one engine cycle..................................................................................... 24 Table 2.1 Engine geometry............................................................................................ 26 Table 2.2 Data signals................................................................................................... 29 Figure 2.1 Cylinder pressure and MAP during at intake event..................................... 30 Figure 2.2 Baseline SI net indicated efficiency versus NIMEP at MBT timing............ 34 Figure 2.3 SI baseline HC emissions versus NIMEP at MBT timing. .......................... 34 Figure 2.4 SI baseline CO emissions versus NIMEP at MBT timing. ......................... 35 Figure 3.1 Cam phasing for high load (low residual fraction) and low load (high residual ................................................................................................................................... 40 fraction)............................................................................................................................. 40 Figure 3.2 Exhaust valve closing versus NIMEP. ........................................................ 41 Figure 3.3 Exhaust valve closing versus CA50 ........................................................... 41 Figure 3.4 Chemkin calculated burned gas temperature versus residual mass fraction. .. 42 Figure 3.5 WAVE modeling: residual fraction versus cylinder temperature at 30 CAD B TC ........................................................................................................................... 43 Figure 3.6 CA50 & NIMEP versus EVC....................................................................... 44 Figure 3.7 Efficiency and PMEP versus EVC .............................................................. 45 Figure 3.8 Emissions of HC versus EVC..................................................................... 46 Figure 3.9 Emissions of CO versus EVC..................................................................... 47 Figure 3.10 Emissions of NO versus EVC. ................................................................. 47 Figure 3.11 Asymmetrical intake cam timing is shown. Symmetrical cam timing minimizes pumping losses..................................................................................... 48 Figure 3.12 IVC versus NIMEP..................................................................................... 49 Figure 3.13 CA50 versus NIMEP................................................................................ 49 Figure 3.14 Inducted mass versus intake cam timing at 1500 RPM. Dashed line shows symm etric IV C .......................................................................................................... 50 10 Figure 3.15 Inducted mass versus intake cam timing at 2500 RPM. Dashed line shows sym m etric IVC .......................................................................................................... 51 Figure 3.16 Cylinder pressure and MAP at IVC 42 ABC. .......................................... 52 Figure 3.17 Cylinder pressure and MAP at IVC 27 ABC. ............................................ 53 Figure 3.18 Mass flow rates of the two intake valve timings. BDC is at 540 CAD......... 54 Figure 3.19 Cumulative mass flow for the two intake valve timings........................... 54 Figure 3.20 NIMEP & CA50 versus IVC.................................................................... 56 Figure 3.21 NIMEP versus combustion phasing for a series of intake cam timings........ 57 Figure 3.22 Intake cam timing and combustion phasing at a constant load of 3.5 bar..... 57 Figure 3.23 Pumping loss versus intake cam timing. Dashed line shows symmetric IVC. ...................................................... ........................................................................... 59 Figure 3.24 Net indicated specific fuel consumption versus intake valve closing. Dashed line shows symmetric IVC.................................................................................... 60 Figure 3.25 HC emissions versus IVC. Dashed line shows symmetric IVC................. 61 Figure 3.26 Peak cylinder pressure versus CA50......................................................... 61 Figure 3.27 HC emissions versus peak cylinder pressure.............................................. 62 Figure 3.28 CO emissions versus IVC......................................................................... 62 Figure 3.29 CO emissions versus IVC......................................................................... 63 Figure 3.30 NO emissions versus IVC. ....................................................................... 63 Figure 3.31 NIMEP and CA50 versus equivalence ratio with constant cam timing: EVC 98 BTC, IVC 42 ABC........................................................................................... 65 Figure 3.32 Efficiency and PMEP versus phi with constant cam timing: EVC 98 BTC, IV C 42 AB C . ......................................................................................................... 66 Figure 3.33 HC emissions versus equivalence ratio.................................................... 67 Figure 3.35 CO emissions versus equivalence ratio.................................................... 67 Figure 3.35 NO emissions versus equivalence ratio..................................................... 68 Figure 3.36 NIMEP versus equivalence ratio with constant fuel pulse width............... 69 Figure 3.37 CA50 & maximum average dP/dt versus equivalence ratio with constant fuel pulse w idth................................................................................................................ 69 Figure 3.38 Efficiency & PMEP versus equivalence ratio with constant fuel pulse width. ................................................................................................................................... 70 11 Figure 3.39 HC emissions versus equivalence ratio with constant fuel pulse width........ 71 Figure 3.40 CO emissions versus equivalence ratio with constant fuel pulse width........ 71 Figure 3.41 NO emissions versus equivalence ratio with constant fuel pulse width........ 72 Figure 3.42 COV of NIMEP versus NIMEP. ............................................................... 74 Figure 3.43 NISFC versus NIMEP................................................................................ 74 Figure 3.44 Eight cycles of unprocessed audible signal data sampled at a rate of 100 kHz for four operating conditions at 2000 RPM ......................................................... 76 Figure 3.45 Cylinder pressure during the combustion event for no audible knock.......... 77 Figure 3.46 Cylinder pressure during the combustion event for loud audible knock....... 78 Table 3.1 Expected resonant frequencies in a cylinder - SAE 2003-01-3217.............. 79 Figure 3.47 FFT of the pressure signal sampled at a rate of 100kHz for four operation conditions..................................................................................................................79 Table 3.2 Survey of some HCCI knock criteria found in literature.............................. 80 Figure 3.48 Integrated FFT of audible data versus maximum dP/dt on a cycle by cycle basis........................................................................................................................... 81 Figure 3.49 WAVE modeling: cylinder temperature at IVC versus residual fraction...... 82 Figure 3.50 Combustion phasing, max dP/dt, and NIMEP over consecutive cycles at the high load lim it....................................................................................................... 83 Figure 3.51 CA50 and WAVE calculated cylinder temperature at IVO for the 4 cycles before misfire............................................................................................................ 84 Table 3.3 Summary of operating limits. ........................................................................ 84 Figure 3.52 Max dP/dt versus combustion phasing....................................................... 86 Figure 3.53 Max dP/dt versus intake cam timing......................................................... 86 Figure 3.54 High load limit: cam phasing strategy........................................................... 87 Figure 3.55 Mapping of the high load limit for one operating condition. .................... 88 Table 3.4 Comparison of two operating points at the high load limit. ......................... 88 Figure 3.56 Low load limit: cam phasing strategy. .......................................................... 89 Figure 3.57 Cam timing at the low load limit.............................................................. 90 Figure 3.58 Low load limit: NIMEP versus IVC.......................................................... 90 Figure 3.59 Air flow into the engine versus IVC at a range of EVC timings............... 91 Figure 3.60 Low load limit search procedure................................................................ 92 12 Table 3.5 Repeatability for high and low load limit search procedures. ...................... 92 Figure 4.1 NIMEP, CA50, and maximum dP/dt versus intake air temperature. .......... 94 Figure 4.2 Inducted mass versus intake air temperature................................................ 94 Figure 4.3 WAVE modeling: residual mass fraction versus intake air temp................ 95 Figure 4.4 WAVE modeling: cylinder temperature at 30 BTC versus intake air temp.... 96 Table 4.1 Operating range limits were evaluated at a range of intake air temperatures, engine speeds, and equivalence ratios. ................................................................ 96 Figure 4.5 High and low load limits for a range of intake air temperatures and equivalence ratios at 1000 rpm ............................................................................. 97 Figure 4.6 High and low load limits for a range of intake air temperatures and equivalence ratios at 1500 rpm ............................................................................. 98 Figure 4.7 High and low load limits for a range of intake air temperatures and equivalence ratios at 2000 rpm. ............................................................................ 98 Figure 4.8 High and low load limits for a range of intake air temperatures and equivalence ratios at 2500 rpm. ............................................................................ 99 Figure 4.9 High and low load limits versus intake air temperature at 1500 RPM with phi = 1. .......................................................................................................................... 100 Figure 4.10 Intake air temperature versus intake cam timing at the high load limit. ..... 100 Figure 4.11 Inducted mass at the high load limit versus intake air temperature. ........... 101 Figure 4.12 Pressure at 30 CAD BTC for the high load limits versus intake air temperature. ............................................................................................................ 102 Figure 4.13 WAVE modeling: residual fraction at the high load limit versus intake air temperature. ............................................................................................................ 103 Figure 4.14 High and low load limits for a range of intake air temperatures and engine speeds at an equivalence ratio of 1.0. ..................................................................... Figure 4.15 Air flow versus engine speed at constant cam timing................................. 104 105 Figure 4.16 Engine speed versus high load limit at phi = 1, and intake air temperature 30C .......................................................................................................................... Figure 4.17 Engine speed versus %reduction in air flow and NIMEP. ......................... 105 106 Figure 4.18 Engine speed versus residual fraction for the high load limit. .................... 107 13 Figure 4.19 WAVE modeling: Engine speed versus heat transfer over one engine cycle. ................................................................................................................................. 10 7 Figure 4.20 Cylinder temperature at 30 CAD BTC (pre-combustion) versus engine speed. ................................................................................................................................. 10 8 Figure 4.21 WAVE modeling: Residual mass % versus cylinder temperature at 30 CAD BTC at 2500 RPM. ................................................................................................. 109 Figure 4.22 WAVE modeling: cylinder temperature at IVC versus residual mass fraction for a range of intake air temperature....................................................................... 109 Figure 4.23 (same as figure 4.6). High and low load limits for a range of intake air temps and phi's at 1500 RPM. .......................................................................................... 110 Figure 4.24 Chemkin modeling: burned gas temp versus residual fraction at a range of phi. .......................................................................................................................... 1 11 Figure 4.25 WAVE modeling: cylinder temperature at IVC versus residual fraction at a range of phi. ............................................................................................................ 112 Figure 4.26 Chemkin modeling: burned gas temp versus fuel mole fraction over a range of equivalence ratios............................................................................................... 112 Figure 4.27 WAVE modeling: cylinder temperature at IVC versus inducted fuel over a range of equivalence ratios. .................................................................................... 114 Table 4.2 For each equivalence ratio, the range of air humidity was tested at 1500 RPM. ................................................................................................................................. 1 15 Figure 4.28 NIMEP vs. intake air humidity.................................................................... 116 Figure 4.29 Inducted mass versus intake air humidity. .................................................. 116 Figure 4.30 Combustion phasing versus humidity and equivalence ratio. ..................... 117 Figure 4.31 Chemkin: burned gas temperature versus ambient humidity for a range of equivalence ratios.................................................................................................... 118 Figure 4.32 WAVE modeling: cylinder temperature versus ambient humidity............. 118 Figure 4.33 High load limit versus humidity.................................................................. 119 Figure 4.34 IVC versus humidity at the high load limit for different equivalence ratios. ................................................................................................................................. 120 Figure 4.35 Low load limit versus humidity................................................................... 121 Figure 4.36 Humidity versus minimum phi.................................................................... 121 14 Table 4.3 The table shows the 12 test fuels created for testing. ..................................... 124 Figures 4.37-4.41 Variability in RVP, RON, aromatics, olefins, and octane sensitivity from B P fuel survey []. ........................................................................................... 125 Table 4.4 Test m atrix for fuel testing.............................................................................. 126 Figure 4.42 CA50 at two reference points for each test fuel. ......................................... 128 Table 4.5 Reference points. ............................................................................................ 128 Figure 4.43 RON versus CA50 at 4 reference points. .................................................... 129 Figure 4.44 RVP versus CA50 at four reference points. ................................................ 130 Figure 4.45 Aromatic content versus CA50 at four reference points. ............................ 130 Figure 4.46 Olefin content versus CA50 at four reference points.................................. 131 Figure 4.47 Ethanol content versus CA50...................................................................... 132 Figure 4.48 Operating range versus engine speed for the 12 test fuels. ......................... 133 Figure 4.49 Intake and exhaust cam timing at the limits of operation............................ 134 Figure 4.50 EVC versus IVC for the high load limits. ................................................... 135 Figure 4.51 IVC versus engine speed for the high load limits........................................ 136 Figure 4.52 IVC versus EVC for the low load limits for each test fuel.......................... 137 Figure 4.53 EVC versus engine speed at the low load limits for each fuel.................... 138 Figure 4.54 Aromatic content versus the high load limit at 1500 RPM......................... 140 Figure 4.55 Olefin content versus high load limits at 1500 RPM. ................................. 140 Figure 4.56 RON versus high load limits at 1500 RPM................................................. 142 Figure 4.57 MON versus high load limits at 1500 RPM................................................ 142 Figure 4.58 High load limits at 1000 RPM versus the high load limits at 1500 RPM... 143 Figure 4.59 High load limits at 1000 RPM versus the high load limits at 2000 RPM... 144 Figure 4.60 High load limits at 1000 RPM versus the high load limits at 2500 RPM... 144 Table 4.6 Residual fraction and load at the high load limits for High Olefin and Extremely Low Aromatic & Olefin fuels. .............................................................. 146 Figure 4.61 Aromatic content versus the low load limits at 1000 RPM......................... 147 Figure 4.62 Olefin content versus the low load limits at 1000 RPM.............................. 147 Figure 4.63 RON versus the low load limits at 1000 RPM........................................... 148 Figure 4.64 Low load limits at 1000 RPM versus the low load limits at 1500 RPM..... 149 Figure 4.65 Low load limits at 1000 RPM versus low load limits at 2000 RPM..... 150 15 Figure 4.66 Low load limits at 1000 RPM versus low load limits at 2500 RPM..... 150 Figure 4.67 Olefin content versus CA50 at 2500 RPM. EVC is 103 CAD BTC and IVC 47 CAD ATC (+/- 2 CAD ). .................................................................................... 151 Figure 4.68 Olefin fuels: MON versus CA50 for the reference points........................... 153 Figure 4.69 MON versus RON for a variety of fuels - from Shibata et. al. [14]........... 154 Figure 4.70 Olefin fuels components.............................................................................. 155 Figure 4.71 Regression: high load limits and napthenes. ............................................... 156 Figure 4.72 Napthene content versus NIMEP at the high load limit.............................. 157 Figure 4.73 Regression: high load limits and distillation temperatures......................... 158 Figure 4.74 High load limit versus combustion phasing. .............................................. 159 Figure 4.75 Maximum rate of pressure rise versus CA50 for 10 repeated low load limits. ................................................................................................................................. 160 Figure 4.76 Maximum rate of pressure rise versus CA50 at the high load limit............ 161 Figure 4.77 Maximum rate of pressure rise versus CA50 for repeated high load limits. 161 Figure 4.78 Percent burn versus CAD............................................................................ 162 Figure 4.79. Emissions of HC versus olefin content for four reference points at 1500 RPM ........................................................................................................................ 164 Figure 4.80 HC emissions versus NIMEP for 4 olefin fuels at 1500 RPM................... 165 Figure 4.81 HC emissions versus engine speed.............................................................. 166 Figure 4.82 HC emissions versus NIMEP...................................................................... 166 Figure 4.83 CO emissions versus engine speed.............................................................. 167 Figure 4.84 CO emissions versus NIMEP...................................................................... 167 Figure 4.84 NO emissions versus engine speed.............................................................. 168 Figure 5.1 HCCI & SI operating range comparison: NIMEP versus engine speed........ 169 Figure 5.2 HCCI efficiency improvement over SI efficiency......................................... 171 Figure 5.3. Vehicle speed versus time for the FTP Urban Drive Cycle. ........................ 172 Figure 5.4. Vehicle speed versus time for the FTP Highway Drive Cycle..................... 173 Figure 5.5 HCCI & SI: HC emissions versus NIMEP.................................................... 175 Figure 5.6 HCCI & SI: CO emissions versus NIMEP.................................................... 176 Figure 5.7 HCCI & SI: NO emissions versus NIMEP. .................................................. 176 Figure 5.8 HCCI & SI: NO emissions versus NIMEP. .................................................. 177 16 Figure Al. Comparison of CA50 and max dP/dt location.............................................. 182 Figure B 1. Throttle operation: NLMEP versus MAP...................................................... 183 Figure B2. Throttled operation: combustion phasing versus load.................................. 184 Table B 1. Comparison of two points with similar load: throttled and unthrottled......... 184 Figure B3. Throttled operation: load versus max rate of pressure rise........................... 185 Figure B4. Throttled operation: NIMEP versus MAP at the low load limit................... 186 Table D l. Fuel properties. .............................................................................................. 189 Table D2. Fuel properties continued............................................................................... 189 Figure Dl. HC emissions versus olefin content for ref. points at 1500 RPM. ............... 190 Figure D2. CO emissions versus olefin content for ref. points at 1500 RPM. ............... 190 Figure D3. NO emissions versus olefin content for ref. points at 1500 RPM................ 190 Figure D4. HC emissions versus aromatic content for ref. points at 1500 RPM............ 191 Figure D5. CO emissions versus aromatic content for ref. points at 1500 RPM............ 191 Figure D6. NO emissions versus aromatic content for ref. points at 1500 RPM. .......... 191 Figure D7. HC emissions versus RVP for four reference points at 1500 RPM.............. 192 Figure D8. CO emissions versus RVP for four reference points at 1500 RPM.............. 192 Figure D9. NO emissions versus RVP for four reference points at 1500 RPM. ............ 192 Figure D10. HC emissions versus RON for four reference points at 1500 RPM........... 193 Figure D11. CO emissions versus RON for four reference points at 1500 RPM........... 193 Figure D12. NO emissions versus RON for four reference points at 1500 RPM........... 193 Figure D13. HC emissions versus ethanol content for reference points at 1500 RPM. . 194 Figure D14. CO emissions versus ethanol content for reference points at 1500 RPM.. 194 Figure D15. NO emissions versus ethanol content for reference points at 1500 RPM.. 194 Figure El. Air flow into the cylinder per cycle versus engine speed. ............................ 195 Figure E2. Air flow versus engine speed with constant cam timing. ............................. 196 17 18 Chapter 1: Introduction 1.1 Motivation Internal combustion engines convert the chemical energy in fuel into mechanical work. The two most common mechanisms to release this chemical energy are spark ignition combustion and diesel combustion. A third type of combustion, Homogeneous Charge Compression Ignition (HCCI) combustion, has the potential to achieve significantly improved fuel efficiency as compared to spark ignition (SI) engine operation, without the emissions of NOx and particulates associated with diesel combustion. This work explores the impact of ambient conditions on HCCI engine operation in a gasoline internal combustion engine. HCCI combustion shares some characteristics with SI combustion and some characteristics with diesel combustion. As in the case of SI combustion, the fuel and air are evenly mixed before combustion. Similar to diesel combustion, combustion is initiated by compression heating. Ignition through compression allows the combustion of very dilute mixtures. Combustion temperatures are lower with more dilute mixtures, and low combustion temperatures avoid NOx formation. The homogeneous nature of the mixture prevents the formation of any particulates during combustion. Combustion of dilute mixtures also allows load to be controlled by dilution. Controlling load by dilution avoids the throttling losses of SI engine operation. Although the advantages of HCCI combustion appear significant, there are also significant challenges to practical application of HCCI combustion. One major challenge is lack of a simple way to control combustion timing. SI combustion is governed by flame propagation, and combustion phasing is controlled by spark timing. Diesel combustion is governed by diffusion, and combustion phasing is controlled by injection timing. HCCI combustion is governed by chemical kinetics, and there is no explicit control over combustion phasing. For HCCI combustion, the duration and location of the combustion event within the cycle are both determined by the temperature and 19 concentration of gases within the cylinder over time. In order to achieve combustion without engine knock at the desired location in the engine cycle, the conditions within the cylinder must be controlled precisely. A second challenge is the limited operating range associated with HCCI operation. High load operation can be problematic because of the tendency of the HCCI combustion to lead to engine knock. Very low load operation can be difficult because of the very low temperatures associated with high dilution rates. Many research efforts have focused on these two critical problems: the difficulty of controlling HCCI combustion timing, and the limited operating range of HCCI combustion. In order to apply HCCI combustion in a practical way, there must be some control over combustion phasing, and there must a good understanding of the HCCI operating boundaries. One area that has not received much attention is the impact of the range of real world ambient conditions on HCCI combustion timing and operating limits. Any practical engine must respond robustly to a range of intake air temperatures, intake air humidity, and fuel compositions. In the case of HCCI combustion, changing intake air temperature, or intake air humidity, or fuel composition will change the conditions within the cylinder and will impact the combustion event. In order to maintain stable operation there must be a way to compensate for changes in these ambient conditions. 1.2 Project Objectives 1. Determine how the limits of HCCI operation are affected by intake air temperature, intake air humidity, and fuel composition. 2. Evaluate cam phasing as a mechanism to control HCCI operation, and to compensate for changes in ambient conditions and fuel composition. 1.3 Background HCCI operation has been demonstrated using a range of different fuels, engine designs, and control strategies. HCCI combustion was first investigated by Onishi in 1979 [1]. The investigation was motivated by the tendency of a gasoline two stroke engine to "runon" without a spark under certain conditions. Najit and Foster first demonstrated HCCI 20 operation in a four stroke in 1984 using gasoline as a fuel [2]. In 1989 Thring proposed a hybrid engine that would operate in SI mode at high load and at very low load, and HCCI mode where possible [3]. For a dual SI/HCCI mode engine, the constraint of operating in SI mode requires a low compression ratio. The limited compression heating associated with a low compression ratio is often insufficient to achieve ignition temperatures. Thring and others have used intake air heating to raise the initial temperature of the charge. Ishibashi & Asai instead used hot residual gases to raise the temperature of the charge [4]. High residual fractions were a feature of two stroke engines, and provided additional thermal energy to the charge. But the residual fractions in four stroke engines are normally very low with conventional valve timing. Kontarakis et. al. was the first to use residual trapping to provide the thermal energy and dilution necessary for HCCI combustion [5]. More recently in 2001 Li et. al. examined cam phasing as a mechanism to provide some control over the trapped residual mass [6]. In 2004 C. Li et. al. used simulation to explore the impact of intake cam phasing on HCCI combustion [7]. Many researchers have explored HCCI operating limits, including several who have focused on the impact of intake air temperature. Aroonsrisopon considered the impact of intake air temperature using a single cylinder engine [8]. Air temperatures ranged from 300K-400K. Persson investigated the impact of temperature on the HCCI operating range using a multi-cylinder gasoline engine [9]. Temperatures ranged from 288K-323K (15-50C). Sjoberg and Dec examined the impact of intake air temperature on combustion phasing with intake temperatures ranging from 300K-420K [10]. Many researchers have also considered the impact of different fuels on HCCI combustion. The great majority of work has focused on different pure hydrocarbon fuels (PRFs, alcohol, methane, etc.). Several researchers have used real gasoline in HCCI studies, but the impact of different components of gasoline was not examined. Christensen examined HCCI combustion with mixtures of gasoline and diesel, but did not consider the impact of different components in the gasoline [11]. 21 Perhaps most applicable to this work are recent studies by Oakley, Kalghatgi, and Shibata. Oakley examined how three different blends of gasoline with very different aromatic and paraffin content performed in HCCI combustion [12]. Kalghatgi introduced an octane index (function of RON & MON) to describe the auto-ignition performance of several fuels [13]. The four fuels were two PRFs and two research gasolines. Shibata examined HCCI combustion in fuels consisting of mixtures of 11 different pure hydrocarbons [14], [15]. Although these fuels are very simple compared to regular gasoline, they do more closely approximate real gasoline than PRFs. 1.4 Experimental Strategy For this project residual trapping was selected as the strategy to achieve HCCI combustion. This strategy is sometimes referred to as Controlled Auto Ignition (CAI). All HCCI combustion strategies have some features in common: the mixture temperature must be high enough to achieve combustion at the correct point in the cycle, and the mixture must be dilute enough to avoid engine knock. In many cases varying the level of dilution allows operation without any throttling. Operation with no throttle results in a significant efficiency improvement in comparison to SI engine operation. In the case of gasoline fuel, the temperature of the mixture must reach near 1000K at the end of the compression stroke. This can be achieved in a number of ways, including increasing the compression ratio, heating the intake air, or as in the case of residual trapping, diluting the fresh charge with hot exhaust gas. The use of hot residual gases to dilute the charge offers several advantages over other strategies. Diluting with residuals allows stoichiometric engine operation, which enables the use of a 3-way catalyst to treat emissions. Dilution with residuals also heats the fresh charge to a temperature where HCCI can be achieved with compression ratios that are compatible with SI operation. These two features of dilution with residuals might allow hybrid operation of an engine in both SI mode and HCCI mode. At startup, idle, and high load the engine could operate in SI mode, and at low to mid load the engine could operate in HCCI mode. A hybrid HCCI/SI engine might allow a practical application of HCCI combustion, even with a limited HCCI operating range. 22 High residual fractions for HCCI operation are most often achieved in one of two ways: residual trapping (negative valve overlap), or residual re-induction (positive valve overlap). In the case of residual trapping the exhaust valve closes during the exhaust stroke, trapping significant quantities of hot gas in the cylinder. In the case of residual reinduction the exhaust valve remains open during the intake event allowing exhaust gases to be drawn back into the intake. Residual trapping avoids any potential problems with piston - valve interference at top center in the piston stroke, and could more easily be applied in a turbo-charged engine. Figure 1.1 shows cam phasing diagrams for a typical SI engine and for an engine operating in HCCI mode with residual trapping by negative valve overlap. The approximate locations of the valve events over the 720 degree cycle are shown mapped on to a 360 degree circle. Exhaust valve opening (EVO) and intake valve closing (IVC) are in the same position in both cases. HCCI valve timing differs in that exhaust valve closing (EVC) is relatively early, and intake valve opening (IVO) is relatively late. TC EVC ETC NO IVC EVC Inak HCIN VC tam EVO BC EVO BC Figure 1.1 Typical SI valve timing (left) and valve timing for HCCI using residual trapping (right). The early EVC in HCCI operation has several implications. Because EVO doesn't change, the early EVC means that valve event durations are very short. Early EVC traps a significant amount of hot exhaust gases in the cylinder during the exhaust stroke and these gases are compressed during the remainder of the exhaust stroke. To avoid losing the work input during the re-compression of the residuals, the IVO is delayed until the 23 pressure has decreased and the work has been extracted. There is some heat transfer during the recompression event creating a small pumping loop, but most of the work can be recovered if IVO and EVC are approximately symmetric about TC. Figure 1.2 shows the approximate timing of residual trapping valve events displayed on a plot of cylinder pressure over the course of a typical HCCI cycle. The recompression event can be seen in the peak in the cylinder pressure after EVC. As discussed, IVO is delayed to recover work input during the compression of residuals. 35. 30 25 CU -0 20 CD 0D C 15 -0 10 5 IVC U, 0 EVO 100 200 EVC 500 300 400 CAD (TC Combustion @ 180) IVO 600 - 700 Figure 1.2 Cylinder pressure data for HCCI combustion with residual trapping over the course of one engine cycle. Additional advantages of residual trapping include ease of implementation. For a port fueled injection gasoline engine, the only hardware requirement is a mechanism to control the valve timing to allow control over residual fraction. Control of residual fraction impacts the mixture dilution, and allows control over load. In comparison to other HCCI strategies, residual trapping is compatible with low compression ratio SI 24 operation, avoids additional heat exchangers to heat intake air, is compatible with boosting, allows comparatively fast control of cylinder conditions, and allows the use of off the shelf emissions after-treatment equipment. The low compression ratio ensures that peak cylinder pressures are not destructively high, which can be a problem with high compression ratio HCCI operation. The residual trapping strategy does however have drawbacks. The efficiency improvement is smaller compared to other HCCI strategies. Lean HCCI operation with intake air heating leads to a charge with lower specific heat (less C02 & H20) during compression and combustion than operation with residual trapping. Lean HCCI operation also allows operation with higher compression ratios: a mixture diluted with air has a lower starting temperature than a mixture diluted with residuals. A lean mixture with a lower temperature at IVC requires more compression heating to achieve the temperatures necessary for combustion. The most efficient HCCI strategies can achieve specific fuel consumption rates similar to diesel fuel consumption rates [16]. 25 Chapter 2: Experimental Setup 2.1 Engine 2.1.1 Engine Experiments were carried out on a modified port fuel injected 2.3L 14 16 valve production Mazda gasoline engine. Continuously variable intake cam phasing is a standard feature on this engine. The modifications made to the engine include increasing the compression ratio from 9.7 to 11.1 by modifying the piston, adding cam phasing to the exhaust side, and the use of custom cam shafts with very short duration valve events. The engine is operated as a single cylinder engine, with only one cylinder fired. The remaining three cylinders are motored. Displacement (cc) 565 Bore (mm) 87.5 Stroke (mm) 94 Connecting Rod (mm) 154.8 Wrist Pin Offset (mm) 0.8 Compression Ratio 11.1 Table 2.1 Engine geometry. 2.1.2 Custom Cam Shafts for Residual Trapping Cam event durations of 120 CAD were selected for both exhaust and intake cams based on initial modeling results [17]. The modeling suggested that cam durations shorter than 100 CAD would severely limit airflow into the engine, limiting higher load HCCI operation. Cam durations longer than 140 CAD would make trapping large residual masses difficult, limiting lower load HCCI operation. The 120 CAD cams were designed with the maximum possible lift. For these very short durations, the maximum lift was limited to 2 mm to avoid excessive cam lobe stress.The camshafts were provided by project sponsor Ford Motor Company. 26 2.1.3 Engine controller Engine control is accomplished using a desktop computer and code developed at MIT. The code allows control of fuel pulse width, spark timing, spark dwell, intake cam timing, and exhaust cam timing. Fuel pulse width, spark timing, and both cam timings can be changed in real time as desired. Cam phasing is continuously variable within a 35 degree range. If a desired cam position is outside of the 35 degree range, the default position can be reset manually to allow the desired position to be reached. The control is accurate to within +/- 1 CAD. The feedback on cam position is available once per cycle, and the resolution on the cam position measurement is 1 degree. 2.1.4 Intake Manifold A custom intake manifold was created for the one firing cylinder. The manifold allowed measurement of air pressure and air temperature 5 cm from the intake port. The intake system could be reconfigured as desired to include a manual throttle, a heater, and a chiller to condition the air flowing into the engine. The air flow to the three motored cylinders was filtered, but was not throttled or conditioned in any way. 2.1.5 Exhaust Manifold A custom exhaust manifold was created for the one firing cylinder. The manifold included ports for the measurement of exhaust temperature and pressure, and a port for an exhaust sample line. The exhaust from the three motored cylinders was separate from the firing cylinder exhaust. 27 2.2 Fuel A range of gasoline fuels were used in these experiments. For the intake air temperature and intake air humidity testing, Chevron UTG91 gasoline was used (RON 90.8, MON 82.8). The gasolines used in the fuel study are described in section 4.3.1. 2.3 Engine coolant All experiments were performed with the engine coolant at 90C (+/- 2 degrees C). Engine coolant temperature is raised using a heater and lowered using a heat exchanger. The engine thermostat was removed, and coolant continuously flows through the engine at a constant rate. For control purposes the coolant temperature is measured at the exit of the engine block. 2.4 Intake air conditioning A chiller connected to a heat exchanger and a heater allow control of the intake air temperature from -6C to IOOC+. The humidity of the intake air is controlled by using the chiller and heat exchanger to lower the temperature of the air to a desired dew point, lowering the humidity. The heater is then used to raise the temperature of the intake air to the desired level. 2.5 Dynamometer & Motor An absorbing dynamometer is installed in series with an electric motor. The motor can operate in either speed mode, or torque mode. In speed mode, the motor ensures that the speed of the system does not fall below the target value. The absorbing dynamometer prevents the speed from exceeding target value. In torque mode the motor puts out a constant torque, and the dynamometer absorbs the output from the engine and the motor keeping the system at a constant speed. All experiments were performed with engine speed held constant. 28 2.6 Measurement & Data Acquisition Labview is used to acquire data from the experimental setup. Data collected includes: Cylinder pressure (Kistler 6052b sensor) MAP (Omega PX 176 sensor) Exhaust pressure (Omega PX 176 sensor) Intake cam position (modified production sensor) Exhaust cam position (modified production intake cam sensor) Spark Fuel pulse width Fuel-Air ratio (MEXA 720NOx - combined air-fuel ratio & NOx sensor) Emissions (C02, CO, HC - Horiba MEXA 554JU portable emissions sensor unit) Emissions (MEXA 720NOx) Table 2.2 Data signals. The engine is equipped with an encoder allowing data collection in phase with the engine cycle at a rate of once per CAD. Correct alignment between encoder and engine cycle phasing was verified by examination of a motoring pressure signal at 2000 RPM. Cylinder Pressure: Cylinder pressure was collected using a Kistler 6152b pressure sensor mounted in the head approximately 3cm from the spark plug. The pressure transducer is a high temperature piezoelectric pressure sensor. Because the sensor and amplifier have an offset drift, the absolute value of the cylinder pressure has to be pegged to another sensor. In a conventional experimental engine setup, the pressure signal is usually set equal to the MAP at a point in the cycle where the MAP and cylinder pressures equalize. Usually BDC compression is selected because at that point in the cycle the piston is not moving, and late in the intake stroke enough time has passed for the pressures to equalize. However because of the very short cam events in this experimental setup, there is often not enough time for the cylinder pressure and the MAP to equalize. Figure 2.1 shows a plot of the cylinder pressure and the MAP during the intake event over the course of one 29 HCCI cycle. Unfortunately the two pressures do not equalize at BDC, so it is difficult to know exactly when the MAP equals the cylinder pressure. 1.3 - MAP 1.2- d- .9 - 0.L 0.7 - IVO 440 4b0 BDC 480 500 520 540 CAD ATC Combustion Cylinder Pressure IVC 50 58 600 Figure 2.1 Cylinder pressure and MAP during at intake event. In cases where pressure signal offset is not critical (e.g. for determining the NIMEP), the cylinder pressure can be set equal to the MAP at BDC. Although there may be some error, the error will likely be less than a few tenths of a bar. Some quick checks can be performed to minimize the error: the cylinder pressure should generally be less than or equal to MAP at BDC, and the cylinder pressure should usually be equal to or higher than atmospheric pressure during the exhaust stroke. In cases where the accuracy of the absolute value of pressure signal is important (e.g. for determining pressure at IVC), two methods have been applied. WAVE modeling has been used to simulate the cylinder pressure over the course of the cycle, and to determine cylinder pressure during the intake stroke. If the simulated net air mass flow into the cylinder, the simulated MAP, and simulated cylinder pressure all match up reasonably well with the actual experimental results, then the simulated cylinder pressure during the intake stroke is assumed to be accurate. The experimental cylinder pressure can be pegged to the modeled cylinder pressure. The second approach involves a comparison of two different calculations of the mass flow into the cylinder. The first mass flow calculation is based on the fueling rate and air- 30 fuel ratio. This calculation is believed to be quite accurate (error of < 2% based on fuel injector calibration). The second mass flow calculation is based on the pressure difference between the MAP and the cylinder pressure. The pressure difference between the MAP and cylinder pressure is assumed to be equal to the pressure difference across the intake valves. The pressure delta across the intake valves drives the flow into the cylinder. The valve open area and the pressure delta can be calculated as a function of CAD. The mass flow into the cylinder can then be calculated based on the pressure difference and the valve open area [18] over the course of the intake event. P abs -P signal +P offsel massfl rate = f (fuel _.rate, ) 1VC mass delta-P =A,,, f~av * f (A Pjnal sg + masse,, (MAP - [P,, + No offset ,, By setting the two mass calculations equal, the pressure difference between the cylinder pressure and MAP is fixed, and so the absolute value of the cylinder pressure is fixed as well. If both the WAVE modeling and the pressure delta mass flow calculation give similar values for the cylinder pressure during the intake event, then the calculated/simulated cylinder pressure can be used to fix the value of the actual cylinder pressure. MAP: Intake manifold air pressure is measured using an Omega PX-176 pressure sensor. The sensor is installed approximately 5 cm upstream of the intake port. Exhaust Pressure: Pressure in the exhaust manifold is measured with an Omega PX-176 pressure sensor. The pressure sensor temperature limit is 120C, so the pressure sensor is installed at the end of a length of tubing to isolate the sensor thermally. However the tubing also introduces some pressure dynamics into the exhaust pressure signal, and the signal is very noisy. 31 Intake & Exhaust Cam Signals: Intake and exhaust cam positions are measured using the production intake cam position sensors. The sensors are installed in the engine cover near the back of the end of the cams. These sensors are inductive and generate a voltage pulse when a metal tooth on the back of the cam shaft rotates by. Digital hardware processes the voltage pulse into a digital pulse with the leading edge at BDC compression, and the trailing edge at the cam metal tooth passing. The duration of the pulse indicates the cam timing. This duration is read by the software of the engine controller at a resolution of one sample per crank angle degree. Fuel Pulse Width: Fuel pulse width data is collected in two ways: manually and with Labview. Labview collects the fuel pulse width signal once per crank angle degree over the course of each cycle. Because a sample rate of 1/CAD limits the resolution of the fuel pulse width at low engine speeds (167 micro seconds/CAD at 1000 RPM), the commanded fuel pulse width in microseconds is also recorded manually. Air - Fuel Ratio: The air fuel ratio is measured using a Horiba MEXA-720NOx combined NO and A/F sensor and analyzer. The sensor was located 5 cm downstream from the exhaust port. The sensor was recalibrated each time a fuel was switched. Emissions: Emissions of HC, CO, and NO were measured. Emissions measurements are presented in terms of grams of emission per gram of fuel. The NO was measured using the Horiba MEXA-720NOx sensor. The analyzer was recalibrated each time a fuel was switched. The HC and CO emissions were measured using a Horiba MEXA-554J analyzer. For the MEXA-554J, a sample line was installed in the exhaust 10 cm downstream from the exhaust port. The sample was cooled in condenser to remove water vapor from the exhaust before going to the analyzer. The analyzer was recalibrated each time a fuel was switched or at the start of each day, whichever came first. 32 Engine Load Measurements: Because the electric drive motor and absorbing dynamometer are independent units, measuring the brake torque using the dynamometer is challenging. There is no way to precisely separate the engine output from the motor output. Separating the frictional load of the three motored cylinders from the one fired cylinder presents an additional challenge. Therefore load is measured by calculating indicated mean effective pressure based on the cylinder pressure signal. Fuel Consumption Measurement: Fuel consumption is measured using fuel injector pulse width. The injector was calibrated in a bench test, and gravimetric measurements were made to confirm the calibration. Fuel efficiency is calculated on an indicated basis using the NIMEP, the fuel consumption rate, and the energy in the fuel. 2.7 SI Baseline The engine was operated in spark ignition mode to provide a baseline with which to compare the HCCI results. For the baseline data, the production camshafts were used and the compression ratio was 11.1 (the same as for HCCI operation). Data was taken from 1000 to 3000 RPM from 1 bar NIMEP to wide open throttle (WOT). Figures 2.2-2.4 show baseline SI efficiency and emissions of hydrocarbons and carbon monoxide. Efficiency and emissions versus load up to approximately 7 bar NIMEP is shown (7 bar is well beyond the maximum HCCI range for this engine). 33 SI Efficiency 0.4 _ >%0.35 - 1000 A&___ RPM @1500RPM . -90.3 03 A 2000 RPM .100.25 * 2500 RPM 0.25 0.2- 03000 RPM 0.15 6 4 2 0 8 NIMEP (bar) Figure 2.2 Baseline SI net indicated efficiency versus NIMEP at MBT timing. SI HC Emissions 0.02 S1000 RPM 1 0.016 C 1500 RPM 0.012 A 2000 RPM 0 * 0.008 - E * 2500 RPM 0.004 03000 RPM 0 0 I I I 2 4 6 8 NIMEP (bar) Figure 2.3 SI baseline HC emissions versus NIMEP at MBT timing. 34 SI CO Emissions 0.1 0.0 - JCI CD 0.08 RPM 100P A*L01000 A A 2000 RPM RPM 3000 RPM 0 0.07 06-*2500 . 0.060 0.05 0.04 0 4 2 6 8 NIMEP (bar) Figure 2.4 SI baseline CO emissions versus NIMEP at MBT timing. 2.8 Modeling Tools Throughout the project the experimental work was accompanied by modeling. Modeling was used to predict HCCI behavior to better design experiments and to analyze experimental results. The modeling tools used included commercially available software, as well as software developed at MIT. Ricardo's WAVE engine simulation software [19] was used to simulate flows into and out of the cylinder and to simulate basic engine processes. A model was created based on the experimental setup, and for many purposes this model was sufficient. An example of an application of the WAVE model might be an analysis of the effect combustion phasing on NIMEP. In a WAVE model, unlike in a real engine, it is possible to hold many parameters constant and vary only those of interest. WAVE also allows the estimation of properties that are difficult to measure experimentally like cylinder temperature and heat transfer. 35 For more detailed analysis of combustion other tools were used. Chemkin was used to perform simple equilibrium calculations of burned gas temperatures. A single zone chemical kinetics combustion model developed at MIT allowed prediction of combustion timing [20]. The model was used with two different chemical mechanisms: the detailed mechanism developed at LLNL, and a reduced mechanism developed by Tanaka & Keck at MIT. When coupled to WAVE, the single zone model allowed analysis of the impact of cam timing on combustion timing. Another tool used in combustion analysis was heat release code developed at MIT [21]. This code calculated combustion characteristics based on the cylinder pressure (see description in Appendix A). 36 Chapter 3: Control of HCCI 3.1 Control variable and the state of the cylinder All practical applications of internal combustion engine technology must have some mechanisms to control the combustion processes. At a minimum, all engines must have mechanisms to allow control over load and some control over when combustion occurs. In HCCI combustion, unlike diesel or SI combustion, chemical kinetics determine when combustion starts, and how combustion proceeds. The temperature, pressure, and concentration of gases in the cylinder over time determine the load, the combustion phasing, and the combustion duration. A practical application of HCCI combustion therefore requires some control over cylinder conditions. In this work the control variables are cam phasing and fueling rate. Cam phasing allows some control over how gases enter and leave the cylinder, which determines inducted mass and trapped residual mass. The fueling rate determines the inducted fuel mass. While these variables can indirectly have some impact on cylinder pressure and temperature, the more direct impact of these variables is on mixture composition. Load is largely proportional to inducted fuel mass when there is sufficient air. If air fuel ratio is allowed to vary, controlling the fueling rate can control load. If the air-fuel ratio is fixed, then fueling rate becomes a dependent variable and load is determined by the air mass flow into the cylinder. As in an SI engine, the flow of air entering the cylinder can be controlled using a throttle. However the use of a throttle does introduce pumping losses, which have a significant negative impact on efficiency. If the flow of fuel and air into the cylinder is not restricted or boosted in any way, the inducted mass and therefore the load are largely determined by the trapped residual mass. Trapping more residual gases leaves less room in the cylinder for fresh charge during the intake stroke. The location of exhaust valve closing determines trapped residual mass and so can indirectly control load (Figure 3.1). 37 Controlling combustion offers substantial benefits including improved efficiency and a wider operating range. A combustion event can be described in terms of two characteristics: combustion duration and combustion phasing. Combustion duration impacts the tendency of the combustion event to result in engine knock. Combustion phasing impacts load, efficiency, and combustion duration. If combustion begins before top center (TC), extra work is required to finish the compression stroke, resulting in a reduction in load and efficiency. Early combustion also leads to higher peak pressures and temperatures. The higher pressures and temperatures associated with early combustion phasing contribute to shorter combustion durations and to engine knock. Combustion phasing is largely determined by the temperature of the gases in the cylinder during the compression stroke, and is therefore more challenging to control than load. The temperature of gases in the cylinder mainly depends on the temperature and amount of the fresh fuel and air entering the cylinder, those of the residual gases, and the temperature of the walls of the cylinder. Fresh air temperature is determined by ambient conditions, but residual gas temperature is determined by the previous combustion event. Burned gas temperatures are in large part a function of fuel mass and the concentration of reactants (residual fraction and air-fuel ratio). If the concentration of gases changes with load, the residual temperatures change as well. Combustion phasing also has an impact on residual temperatures. Later combustion phasing leads to higher exhaust temperatures and higher residual temperatures. There are a variety of strategies to control cylinder temperature but most have significant drawbacks: heat exchangers to heat or cool fresh charge or residuals are expensive and have a slow response time, variable compression ratio mechanisms are expensive, etc. Another strategy to control combustion phasing is to adjust the flow entering the cylinder. The flow entering the cylinder impacts not only the load, but also the concentration of gases and therefore the residual gas temperature. Additionally, if the flow into the cylinder is varied for a fixed residual mass, the pressure within the cylinder will also change. Higher cylinder pressure at IVC will lead to higher pressures and temperatures during the compression stroke, and earlier combustion phasing. In this work intake cam 38 phasing is used to impact the flow of fresh charge into the cylinder, allowing some control over combustion phasing. Control over combustion duration would allow avoidance of knock and an expansion of the HCCI operating range. However combustion duration is largely a function of the concentration of reactants within the cylinder, and so is very difficult to control independently of load. A relatively large residual fraction is always necessary to provide the thermal energy required for combustion, and to dilute the charge to slow combustion and avoid knock. This minimum requirement for residual fraction limits high load operation in unboosted operation (Many researchers have examined boosting to allow higher load operation [22], but boosting was not explored in this project.). During low load operation, very high residual fractions slow combustion significantly. At some level the high residual fraction leads to low enough combustion and residual temperatures that stable combustion cannot be maintained. 39 3.2 Exhaust Cam Phasing 3.2.1 Exhaust Cam Phasing: Load & Combustion Phasing Exhaust cam phasing has a direct impact on trapped residual mass. The timing of exhaust valve closing determines the trapped volume and trapped mass. Exhaust valve opening can impact the work extracted during the expansion stroke. However unless the opening time is significantly retarded, EVO impact on work is small in comparison to the effect of varying trapped residual mass produced by changes in EVC. Thus exhaust cam timing can be used to control residual fraction and therefore load. Early EVC traps a larger residual mass in the cylinder, allowing less fresh charge to enter the cylinder. This results in a lower load. Later EVC traps a smaller residual mass. Less residual mass leaves more room for fresh charge to enter the cylinder, and results in a higher load. TC TC EVC High Load N Low Res Mdake 1VC Low Load High Res NVO EVC IVC EVO BC EV ExBaust BC Figure 3.1 Cam phasing for high load (low residual fraction) and low load (high residual fraction). In order to better understand the impact of exhaust cam timing, a sweep of exhaust cam timing was performed with all other variables held constant under HCCI operation. Figure 3.2 shows the NIMEP versus exhaust cam timing with engine speed, equivalence ratio, intake air temperature, and intake cam timing held constant. As the exhaust cam timing is advanced, more residuals are trapped allowing less room in the cylinder for 40 fresh charge. The NIMEP drops by approximately 1.6 bar over 25 degrees of exhaust cam range. 4 Exhaust Cam Timing & Load 1500 RPM, Intake Air T 30C, Phi = 1, WOT, IVC 52 ABC 3.5 3 .0 a. uL zU2.5 2 Earlier EVC, More Residuals 1.5 75 85 105 95 EVC (CAD BTC) 115 Figure 3.2 Exhaust valve closing versus NIMEP. Ex. Cam & Combustion Phasing 1500 RPM, Intake T = 30C, Phi = 1, WOT, IVC 52 ABC 4.9V 4.7 U- 0 4.5 4.3 4.1 3.9 I 75 85 I 95 105 EVC (CAD BTC) 115 Figure 3.3 Exhaust valve closing versus CA50. Over the same range in exhaust cam phasing, there is very little impact on combustion phasing. Figure 3.3 shows a change in combustion phasing of less than 1 CAD over the range of exhaust cam timing. As the residual fraction increases, the concentration of 41 reactants declines, resulting in lower combustion temperatures and lower residual temperatures. At the same time, a greater portion of the mixture consists of hot residual gas. The two phenomena tend to cancel each other out, and the cylinder temperature changes little. Some simple equilibrium calculations can give an indication magnitude of the impact of residual fraction on burned gas temperatures. Figures 3.4 shows residual fraction versus burned gas temperature for ethylene. Ethylene was used as a substitute for gasoline because the H:C ratios are similar: 2 for ethylene and 1.85 for the gasoline. The burned gas temperatures were calculated using Chemkin. To simplify the calculation, combustion was assumed to be constant volume and stoichiometric with a starting temperature of 1000K. As residual fraction increases, there is less fuel in the cylinder, and more C02, H20, and N2 to absorb the energy of combustion, and the burned gas temperatures drop. Burned Gas Temperature Constant Volume Combustion C2H4, Phi = 1, Tinitiai = 1000K, Pinitia = 1bar 2600 2500 2400 + + D 2300 w 2200 a 2100 -a 2000 E 1900 1800 1700 0.3 0.4 0.5 0.6 0.7 0.8 Residual Mass Fraction Figure 3.4 Chemkin calculated burned gas temperature versus residual mass fraction. 42 WAVE: Residual vs. Cylinder T 1500 RPM, Phi = 1, In T 30C 885 S880CE1 @875 870 865 860 35 45 55 65 75 Residual Mass % Figure 3.5 WAVE modeling: residual fraction versus cylinder temperature at 30 CAD BTC. Modeling with WAVE allows calculation of cylinder temperatures for a range of residual fractions. Figure 3.5 shows calculated cylinder temperatures at one point in the compression stroke before combustion: 30 CAD BTC. Changing residual fractions impact cylinder temperature in a number of ways. Residual fraction impacts burned gas temperature (and therefore residual gas temperature), trapped residual mass, and the ratio of specific heats during the compression stroke. At low residual fractions, burned gas temperatures are high, but the mass of hot residual gases in the cylinder is small and the cylinder temperatures are relatively lower. As residual fraction is increased, the portion of hot gases in the cylinder increases, and the cylinder temperature increases. As residual fractions increase further, the burned gas temperatures continue to drop and the lower ratio of specific heats contributes to lower temperatures from compression. Eventually these phenomena dominate and the cylinder temperature falls. If the modeling is accurate and cylinder temperature drops with very high residual fraction, the combustion phasing should also become later at very high residual fractions. Figure 3.6 shows load and combustion phasing versus exhaust cam timing for a series of 43 points. The point with earliest exhaust cam timing should have the largest trapped residual mass, and shows the lowest NIMEP. This point also shows a later combustion phasing than the other points, which is consistent with expectations based on modeling. Exhaust Cam Phasing vs. Load & CA50 1500 RPM, Phi = 1, IVC 42 ABC, In Air T = 40C 3.5 - 3Ci z 2.5 CA50 (CAD - 2- ATC) U z2 o; 1.5- Earlier EVC, More Residuals 0.5 0 95 115 105 EVC (CAD BTC) -. m NIMEP (bar) 125 Figure 3.6 CA50 & NIMEP versus EVC. 3.2.2 Exhaust Cam Phasing and Efficiency Figure 3.7 shows the indicated efficiency for the same points shown in figure 3.6. As EVC is advanced, the efficiency drops. Over the same range, the PMEP doesn't change significantly. With exhaust cam advancement the load decreases, so the PMEP makes up a larger portion of the NIMEP. 44 Exhaust Cam Phasing vs. Load & CA50 1500 RPM, Phi = 1, IVC 42 ABC, In Air T = 40C 0.33 0.32 a ated Effic iency 2 0.31 -Indic 00.3 0.29 0U . PME P (bar) 0.28 + 0.27 0.26 95 I I 100 105 110 115 EVC (CAD BTC) Figure 3.7 Efficiency and PMEP versus EVC. 3.2.3 Exhaust Cam Phasing and Emissions Emissions of hydrocarbons, carbon monoxide, and oxides of nitrogen are sensitive to conditions within the cylinder, and are particularly sensitive to temperature. Higher temperatures during combustion generally increase rates of reaction resulting in higher levels of NO emissions and lower levels of HC and CO emissions. Lower temperatures during combustion tend to result in lower levels of NO emissions, and higher levels of HC and CO emissions. Higher loads and air-fuel ratios contribute to higher cylinder temperatures and higher emissions of NO. Lower loads and lean air fuel ratios contribute to lower cylinder temperatures and higher emissions of HC & CO. The same mechanisms that lead to emissions during SI combustion play a role in HCCI combustion, but there are some important differences. The main mechanisms responsible for unburned HC emissions in a warm engine are flame quenching at the walls, flame quenching in the crevices, adsorption of fuel vapor into the oil on the cylinder walls, and incomplete combustion [18]. The relative importance of each mechanism depends on the particular operating condition. HCCI 45 emissions of HC tend to be higher than SI emissions of HC, particularly at low loads. Recent computational and experimental studies indicate that near the low load boundary incomplete combustion (bulk gas quenching) contributes significantly to HC emissions [23]. As previously discussed, HCCI combustion is governed by chemical kinetics. The mixture temperature must be high enough for a reaction to take place. At very low loads, HCCI operation is limited by unstable combustion and misfire due to low temperatures. The low temperatures at low loads are consistent with higher HC emissions. Figure 3.8 shows EVC versus HC emissions. As EVC is advanced, more residual gases are trapped in the cylinder. Load and burned gas temperatures both decrease. As expected, HC emissions also increase. Exhaust Cam Phasing vs. Emissions 1500 RPM, Phi = 1, IVC 42 ABC, In T = 40C c 0 0.019 0.017 0.015 0.0130.011 0.009- # 0.007 95 I I I 100 105 110 115 EVC (CAD BTC) Figure 3.8 Emissions of HC versus EVC. Carbon monoxide is largely a function of air-fuel ratio. Emissions of CO during HCCI combustion tend to be similar to levels during SI combustion except at very low loads. At very low loads, HCCI emissions of CO can be significantly higher than SI emissions of CO at similar loads. Again, the low burned gas temperatures associated with very dilute HCCI combustion lead to bulk gas quenching, and high emissions of CO. Figure 3.9 shows CO emissions versus EVC. Earlier EVC traps more residuals. Higher residual fractions lead to lower loads, lower burned gas temperatures, and higher CO emissions. For the points shown here, the differences in CO emissions are small. 46 Exhaust Cam Phasing vs. Emissions 1500 RPM, Phi = 1, IVC 42 ABC, In T = 40C 0.08 -0.07 . 0.06 ,00.05 0.04 C 0.03 - 0.02 95 100 105 110 EVC (CAD BTC) 115 Figure 3.9 Emissions of CO versus EVC. The reactions that produce oxides of nitrogen (mostly NO with a small amount of N02) are very sensitive to cylinder temperatures. Higher cylinder temperatures lead to higher rates of formation of NOx. Lower residual fractions and higher loads both contribute to higher cylinder temperatures and higher levels of NOx. Figure 3.10 displays NO emissions versus EVC. EVC is advanced and load, peak cylinder temperatures, and NO emissions all decline. These emissions levels are approximately two orders of magnitude lower than traditional SI NO emissions levels. Chapter 5 compares SI and HCCI engine operation including emissions. Exhaust Cam Phasing vs. Emissions 1500 RPM, Phi = 1, IVC 42 ABC, In T = 40C 0.0008 c 0.0006 0.0004 00.0002 - 0 95 100 105 110 EVC (CAD BTC) 115 Figure 3.10 Emissions of NO versus EVC. 47 3.3 Intake Cam Phasing 3.3.1 Intake Cam Phasing: Load and Combustion Phasing The most common intake cam phasing strategy for residual trapping is to make intake valve opening symmetric with exhaust valve closing about top center (figure 3.1). Symmetrical intake cam timing minimizes the pumping loop during the gas exchange processes, but asymmetrical cam timing was examined (figure 3.11). IVO and IVC can have some affect on how much fresh charge enters the cylinder, but the impact on load is very small in comparison to the affect on load of exhaust cam phasing. Intake cam timing does however have an impact on the state of the cylinder at IVC, and so does have some affect on combustion. TC TC intake EVC .1EVC Early Intake Timing c g s 1~~VC Itk IVC BC Late Timing EVO BC EVO Figure 3.11 Asymmetrical intake cam timing is shown. Symmetrical cam timing minimizes pumping losses. Figure 3.12 shows a sweep of the intake cam timing with exhaust cam timing held. constant. As in the case of the exhaust cam sweeps, the engine speed, equivalence ratio, intake air temperature were held constant. In contrast to the exhaust cam timing sweeps shown in figures 3.2 and 3.6, intake cam timing has a small impact on NIMEP: the range in load over 35 CAD of intake cam timing is only 0.2 bar. The effect of intake cam timing on combustion phasing is more significant: over the 35 CAD range of intake cam timing, combustion phasing varies by approximately 3 degrees (figure 3.13). 48 4 Intake Cam Timing & Load 1500 RPM, Intake Air T 30C, Phi = 1 WOT, EVC 95 BTC 3.5 3 . LLU 2.5 z 2 I 1.5 20 30 50 40 IVC (CAD ABC) . 60 Figure 3.12 IVC versus NIMEP. 5 In. Timing & Combustion Phasing 1500 RPM, Intake Air T 30C, BTC WOT, Phi =1, EVC 95 4 3 2 4 0 0 20 30 I I 40 50 60 IVC (CAD ABC) Figure 3.13 CA50 versus NIMEP. The impact of exhaust cam timing on load and combustion phasing is fairly intuitive, but the impact of intake cam timing is not. As intake cam timing is advanced, IVC occurs earlier in the cycle and the trapped volume increases. One might expect trapped mass and 49 therefore load to increase as IVC is advanced. Earlier IVC also results in a larger effective compression ratio, which should lead to more compression heating, higher temperatures during compression, and earlier combustion phasing. At later intake cam timings, the load increases and combustion phasing advances as IVC is advanced, consistent with expectations. However at earlier intake cam timing positions, the load decreases and the combustion phasing is retarded as IVC is advanced. An examination of the trapped mass shows that the trapped mass is actually decreasing as intake cam timing is advanced beyond the symmetric point (symmetric cam timing is marked with the vertical dashed line). This decrease in trapped mass with advancing IVC was observed across the range of engine speeds from 1000 RPM to 2500 RPM. The two circled points in figure 3.14 are examined in detail. Inducted Mass vs. Intake Cam 1500 RPM, WOT, Phi = 1, Intake Air T 30C 240 - %. 0220 - 180 U. 160 140. 120 - ." 100 20 30 40 50 IVC (CAD ABC) 60 70 Figure 3.14 Inducted mass versus intake cam timing at 1500 RPM. Dashed line shows symmetric IVC. 50 Inducted Mass vs. Intake Cam Timing 2500 RPM, WOT, Phi =1, Intake Air T 30C 240- 220- E ~20O £U180- I I S160 140 , I-_ .S 120 100 I 20 30 40 50 IVC (CAD ABC) I 60 70 Figure 3.15 Inducted mass versus intake cam timing at 2500 RPM. Dashed line shows symmetric IVC. The inducted mass is less than expected because of the interactions of piston motion, cylinder pressure, MAP, and valve open area. Because of the very short duration cam events, the valve lift is limited to 2mm (typical SI valve lift is 8-9mm). The valve open area, even at peak lift, is very small compared to typical SI valve open areas. Flow through the intake valves is restricted, and the cylinder pressure and the MAP never equalize. Figure 3.16 examines cylinder pressure and map during the intake event for two cases: nearly symmetrical cam timing with IVC at 42 CAD ABC, and earlier intake cam timing with IVC at 27 CAD ABC (the two circled points in figure 3.14). The cylinder pressure was pegged using modeling and a comparison of mass flow calculations as described in section 2.6. The approximate location of the valve events is displayed on the figure. In the case of the symmetric cam timing, when IVO occurs MAP is slightly below cylinder pressure (heat transfer during the recompression event). Because the valve does not open to maximum lift immediately, there is very little flow into the cylinder just after valve opening. At this point in the stroke, the piston is moving relatively quickly. 51 1.2 Valve Event Cylinder Pressure IVC 27 ABC ......-MAP IVC 27 ABC 1.15 1.1 ------- Cylinder Pressure IVC 42 ABC -------- MAP IVC 42 ABC 1.05. 1 Valve Event ~ - -. 0.95 a) - ~0.9 0.85 - 0.8 - 1d 0.75I 640 I I I I 660 680 700 720 CAD (720 = BDC Compression) I I 740 760 Figure 3.16 Cylinder pressure and MAP at IVC 42 ABC. In figure 3.16, the maximum piston velocity for a given engine speed occurs at approximately 613 CAD. IVO for the nearly symmetric case is at 642 CAD. The very small valve open area and relatively fast piston motion cause the cylinder pressure to drop significantly below MAP. As the stroke progresses the valve opens further allowing more flow into the cylinder, the piston slows, and the cylinder pressure begins to recover. As the piston moves beyond bottom center and begins to move upward, cylinder pressure recovers and rises slightly above MAP before intake valve closing. In the case of the earlier intake cam timing, IVO occurs while the pressure within the cylinder is higher than MAP. Higher cylinder pressure results in a small amount of backflow into the intake just after the valve opens. By the time the cylinder pressure has fallen to the level of the MAP, the valve open area has increased. As the stroke progresses the piston moves downward, increasing the volume and reducing the pressure 52 of the cylinder. However, the larger valve open area at the point in the stroke when MAP and cylinder pressure are equal leads to less restriction of the flow into the cylinder. Less restricted flow into the cylinder leads to a slightly smaller pressure difference between MAP and cylinder pressure over the intake event. In the case with nearly symmetric timing, the very small initial valve open area leads to a reduced cylinder pressure. The lower cylinder pressure drives increased flow throughout the stroke. With earlier intake cam timing, the initial pressure drop is smaller, and the pressure difference driving the flow into the cylinder is smaller. Figure 3.17 compares the pressure differences (MAP minus cylinder pressure) in both cases over the intake valve events. 0.25 BDC 0.2 - 0.15LO CL 0.1 0.05 -- _0 U MAP - Cyl P IVC 27 ABC CL < -0.05 ------MAP - Cyl P IVC 42 ABC - -0.1 BDC 620 640 660 680 700 720 740 760 780 CAD (720 = BDC Compression) Figure 3.17 Cylinder pressure and MAP at IVC 27 ABC. The smaller difference in pressures for the early intake cam timing, and the initial high cylinder pressure, suggest that the mass flow through the intake valve will be lower in the case of the early intake valve closing. Figure 3.18 shows the calculated mass flow rates based on the pressure difference. The mass flow was calculated assuming incompressible 53 flow. The integrated calculated mass flows agree to within 5% of the actual inducted masses (actual mass flows shown in figure 3.14). Early intake cam timing results in less mass flow because of the very small valve lift combined with the dynamics of piston motion and cylinder pressure. 0.03 0.0250.02- W IVC 27 ABC 0.015 IVC 27 ABC 0 0.01 i. IVC42ABC - 2 0.005 0-0.005 440 490 540 590 CAD ATC Combustion Figure 3.18 Mass flow rates of the two intake valve timings. BDC is at 540 CAD. 250 C 200 S150- 150-- - - 100- I IVC 27 ABC - IVC 42 ABC E S500 0 50 100 Valve Event CAD (open @ 0, close @ 120) Figure 3.19 Cumulative mass flow for the two intake valve timings. 54 The reduced mass flow associated with early intake cam phasing has a significant impact on combustion phasing. The first effect is a small change in gas concentrations, and the second is a reduction in pressure at IVC. Both of these effects contribute to lower temperatures during the compression stroke, and later combustion phasing. The magnitude of the temperature decrease associated with a small change in gas concentration can be estimated with some simple calculations. If the trapped residual mass is constant, a reduction in the fresh charge mass will result in an increase in residual fraction. For the two cases under consideration the exhaust cam timing is the same, and so the trapped residual mass should be similar. WAVE modeling actually shows a very small increase (- 1%) in residual mass as IVC is advanced with constant EVC. As shown in figure 3.4, an increase in residual fraction leads to a reduction in burned gas temperature, and therefore a reduction in residual gas temperature. For the purposes of calculation, an initial residual fraction of 50% was assumed. If the residual mass is held constant and the inducted mass is reduced by 7% (the approximate decrease in trapped mass between the two points from figure 3.14), residual fraction increases to -52.5%. A 2.5% increase in residual fraction leads to a 50 degree K cooler burned gas temperature. The temperature decrease associated with a drop in pressure may be more significant. With early intake cam timing, less fresh air and fuel enters the cylinder. Early intake cam timing also results in a larger trapped volume. For the two points considered, constant EVC leads to very similar residual trapped mass. Therefore the total trapped mass at IVC will be lower. Larger trapped volume and smaller mass should contribute to lower cylinder pressure. If residual gases within the cylinder are expanded to a lower pressure, the temperature will be lower. WAVE modeling results are consistent with this argument showing lower pressures at IVC and lower temperatures during compression for earlier intake cam phasing. 55 Figure 3.16 compares the cylinder pressure of the early and nearly symmetric intake cam timing during the intake valve event (540 is BDC). The cylinder pressure of the operating point with early intake cam timing is approximately 0.05 bar lower. Assuming adiabatic and isentropic processes and a 50K lower burned gas temperature, a 0.05 bar pressure drop leads to an approximately 40K cooler cylinder temperature at intake valve closing. While a 40K change in cylinder temperature might not sound significant, experimental results show that lowering cylinder temperature by approximately 18 degrees K leads to a delay in combustion phasing of 2.5 CAD, which is consistent with the impact of IVC timing (section 4.1.1). Further intake cam timing advancement allows further delay in combustion phasing. To obtain the points in figure 3.20, the default intake cam position was advanced (the 35 CAD range was advanced). The combustion phasing and load show the same trends as in figures 3.12 and 3.13. The earlier intake cam phasing shows a combustion phasing range of 6 CAD over the symmetrical cam phasing. Intake Cam Phasing vs. Load & CASO 1500 RPM, Phi = 1, EVC 98 BTC, In T = 40C 7 6 u 5-+ CA50 (CAD 4 ATC) c; 3 - NIMEP (bar) 2 1 0 10 20 30 40 50 60 IVC (CAD ABC) Figure 3.20 NIMEP & CA50 versus IVC. One drawback of keeping exhaust cam timing constant while varying intake cam timing is that the load changes as intake cam timing is changed. To eliminate any impact from changing load, a combination of intake and exhaust cam timing was used to provide a 56 range of combustion phasing timings at the same load. Figure 3.21 shows a series of exhaust cam timing sweeps at a range of different intake cam timings. As EVC was retarded, less residual mass was trapped and the load increased. For a given load, the earliest intake cam timing results in the latest combustion phasing. Combustion Phasing vs NIMEP Sweep of Exhaust Phasings around High Load Boundary 1500 RPM, Phi=1, WOT, Intake T = 30C 9 8 7 5 C.) +IVC 62 ABC I VC 52 ABC 2vee AIVC 42 ABC m IVC 32 ABC I- 4 eV 3 2 1 0 2.5 0 9 IVC 27 ABC 0 OL 4 3.5 3 4.5 NIMEP (bar) Figure 3.21 NIIMEP versus combustion phasing for a series of intake cam timings. Combustion Phasing & Cam Timing 1500 RPM, Intake Temperature 30C, - 6 - NIMEP 3.5 bar 0 0 5 4 +EVC 97 CAD BTC *EVC 93 CAD BTC SEVC 91 CAD BTC 03 03 Q 23 Ln U 01 U 0 0 20 40 60 80 Intake Valve Closing (CAD Figure 3.22 Intake cam timing and combustion phasing at a constant load of 3.5 bar. Figure 3.22 shows intake cam timing versus combustion phasing for the points at 3.5 bar NIMEP. The plot shows that it is possible to control the combustion phasing over approximately 5 CAD while holding the load constant. Although more advanced IVC 57 restricts the flow into the cylinder, retarding EVC can compensate. Later EVC reduces trapped residual mass by a small amount, countering the flow restriction and a constant load is maintained. The inducted mass is very similar between the different operating points. In the case of constant exhaust cam timing two mechanisms contribute to later combustion phasing: an increase in dilution, and a reduction in the pressure at IVC. In the case of constant load, the inducted mass remains roughly constant. The residual mass is purposely reduced with advancing intake cam timing to counter the reduction in inducted mass seen in previous examples. With inducted mass roughly constant and residual mass decreasing, the residual fraction should decrease. Lower residual fractions will lead to higher burned gas temperatures. However the effects of lower cylinder pressure with advanced IVC are large enough to lower cylinder temperatures, resulting in later combustion phasing. 3.3.2 Throttled operation Advanced intake cam timing restricts the flow into the cylinder and results in later combustion phasing. Another way to restrict the flow into the cylinder is to throttle the flow. By using the throttle to control the flow into the cylinder it was possible to control load. Interestingly control of the throttle position and the trapped residual mass also allowed some control over combustion phasing. In general the more throttled operation showed later combustion phasing. The mechanisms responsible for later combustion phasing with throttled operation are similar to the mechanisms responsible for later combustion phasing with advanced IVC. With more throttling and lower MAP, the cylinder pressure at IVC is reduced, and the trapped residuals are expanded to a lower pressure and temperature. The results of the throttled operation are shown in Appendix B. 3.3.3 Intake Cam Phasing: Efficiency Intake cam phasing has a significant affect on efficiency. Due to the dynamics of the flow past the valve, late and early intake cam timing result in less inducted mass, and later 58 combustion phasing. If combustion occurs before TC, extra work will be required to complete the compression stroke. If the combustion occurs later in the expansion stroke, less work can be extracted from the heat of combustion. Intake cam timing can also affect efficiency by having an impact on the cylinder pressure during the intake stroke. Asymmetrical cam timing can result in the creation of a pumping loop. In the case of late intake cam timing, cylinder pressure drops below MAP before IVO, requiring additional work during the intake stroke. In the case of early intake cam timing, the intake valve opens while the pressure within the cylinder is still high. Not all of the work put into the recompression event is recovered. Figure 3.23 shows that pumping losses approach a minimum near symmetric cam timing (again symmetric cam timing is marked by the dashed vertical line). Overall engine efficiency is however a function of not only pumping loss, but also combustion phasing. The benefits of later combustion phasing can in some cases outweigh the negative impacts of higher PMEP. Figure 3.24 shows the indicated efficiency versus intake cam timing for the range points shown in figure 3.12 & 3.13. Early intake cam timing efficiencies are approximately 2% better than the efficiencies of later cam timing. Pumping Loss vs. Intake Cam 1500 RPM, Intake Air Temperature 30C, WOT, EVC 95 BTC vs. 35 CAD of IC Phasing 0.3 0.295 j 0.29 o.0.285LU I. 0.28- * 0.275 + 0.27 20 30 40 50 60 70 Intake Valve Closing (CAD ABC) Figure 3.23 Pumping loss versus intake cam timing. Dashed line shows symmetric IVC. 59 Inducted Mass vs. Intake Cam 1500 RPM, WOT, Phi = 1, Intake Air T 30C 0.334 I 0.333 I C 0.3320.331 m V -I 0.330.3290.328 0.327 0.326 20 30 40 50 60 70 IVC (CAD ABC) Figure 3.24 Net indicated specific fuel consumption versus intake valve closing. Dashed line shows symmetric IVC. 3.3.4 Intake Cam Phasing and Emissions Because intake cam timing can have a significant impact on combustion phasing, intake can timing also has an impact on emissions. Combustion closer to top center results in higher peak pressures and temperatures. Higher temperatures during combustion tend to lead to more complete oxidation of HC and CO, and higher rates of NO formation. Figures 3.25, 3.26, and 3.27 show the emissions of HC, CO, and NO versus intake cam timing with the exhaust cam timing and other variables held constant. These are the same points shown in the intake cam timing sweep of figure 3.20. 60 Intake Cam Phasing vs. Emissions 1500 RPM, Phi = 1, EVC 98 BTC, In T = 40C 0.015 * 0.0130.011 - 0.009 0.007 -I 10 I I 20 30 40 60 50 IVC (CAD ABC) Figure 3.25 HC emissions versus IVC. Dashed line shows symmetric IVC. Hydrocarbon emissions appear to reach a minimum near symmetric cam timing and increase as intake cam timing is advanced. Combustion phasing that is closer to TC will result in higher peak temperatures and more complete combustion. Figure 3.26 plots CA50 versus peak pressures for the same points. The HC emissions are plotted against peak cylinder pressures in figure 3.27. As peak cylinder pressures decrease, HC emissions increase. Over the range of these points, the load varies by 0.44 bar (figure 3.20). CA50 vs. Peak Pressure 1500 RPM, Phi =1, In T = 40C 39 *_ 37 _ .0 35 0. 33 " 31 0. 29 27 0 I I 2 4 6 8 CA50 (CAD ATC) Figure 3.26 Peak cylinder pressure versus CA50. 61 Peak Cylinder P vs. Emissions 1500 RPM, Phi= 1, In T =40C 0.0150.014 =0.013 a 0.012 JR0.011 0.01 _ z 0.009 0.008 0.00727 29 I I 31 33 I 35 37 39 Peak P (bar) Figure 3.27 HC emissions versus peak cylinder pressure. Intake Cam Phasing vs. Emissions 1500 RPM, Phi = 1, EVC 98 BTC, In T =40C 0.06 - 0.05-p0.042 0 *2 3 0.03 4 0.02 10 20 I I I 30 40 50 5 60 IVC (CAD ABC) Figure 3.28 CO emissions versus IVC. Emissions of carbon monoxide also increase with earlier intake cam timing. As discussed previously, CO emissions are very sensitive to air-fuel ratio. Figure 3.28 plots CO emissions versus intake cam timing. The CO emissions appear to correlate with intake cam timing. In contrast to the HC emissions plot, CO is lowest at the latest intake cam timing. The lower than expected CO emissions with later IVC suggest a different mechanism may be playing a role in CO emissions. Figure 3.29 plots CO emissions versus lambda. Each point is numbered to allow comparison with figure 3.28. Coincidentally the lambda is richest at the earliest cam timing, and leanest at the latest 62 cam timing. It is likely that both temperature and lambda are contributing to the range of CO emissions. Intake Cam Phasing vs. Emissions 1500 RPM, Phi = 1, EVC 98 BTC, In T = 40C 0.06 0.05 .2 03 0.04 - 4 0 0.03 0 5 0.02 0. 992 0.994 0.996 0.998 Lambda I SI I 1 1.002 1.004 Figure 3.29 CO emissions versus IVC. NO emissions generally increase with higher cylinder temperatures. As intake cam timing is advanced, combustion phasing is retarded and peak pressures and temperatures drop. As expected, NO emissions decrease as intake cam timing is advanced. However, as with CO, the emissions do not meet expectations with the latest cam timing. With IVC at 52 CAD ABC, the CA50 is approximately 1 CAD later than when IVC is at 42 CAD ABC. However the emissions at the latest intake cam timing are significantly higher, instead of similar or slightly lower. The leaner air-fuel of the latest intake cam timing point may contribute to the higher NO emissions. Intake Cam Phasing vs. Emissions 1500 RPM, Phi = 1, EVC 98 BTC, In T = 40C 0.0012 0.001 .2 0.0008 -00.0006 A A a 0.0004 z 0.0002 A A 0 I 10 20 I 30 40 IVC (CAD ABC) I 50 60 Figure 3.30 NO emissions versus IVC. 63 Another possibility is that late intake cam timing may lead to differences in-cylinder mixing. With later intake cam timing there is no backflow into the intake manifold at IVO, but there is some backflow into the manifold just before IVC. If differences in mixing lead to pockets of fresh charge that were not diluted by residuals, those pockets might have very high local temperatures and high local rates of NO formation. 64 3.4 Fueling Rate as a Control Variable 3.4.1 Fueling Rate: Constant Cam Timing - Load & Combustion Phasing In addition to intake cam phasing and exhaust cam phasing, fueling rate is another control variable. If air-fuel ratio is allowed to vary, fueling rate can be used to control load and also has an impact on combustion phasing. Lean operation is attractive because it does offer a couple advantages in terms of efficiency. Leaner mixtures increase the ratio of specific heats of the mixture allowing more efficient engine processes. For a residual trapping engine, lean operation also allows more flexibility to optimize cam timing to minimize PMEP losses and adjust combustion phasing as desired. Of course this added flexibility comes at the price of added complexity. If air fuel ratio is fixed, fueling rate is a dependant variable and only intake cam timing and exhaust cam timing must be optimized. If air fuel ratio can vary, the control strategy must optimize a third variable. Load can be controlled quite simply with fueling rate: reducing fueling reduces load. In figure 3.31 the cam timing is held constant (EVC 98 BTC, IVC 42 ABC), and the load and combustion phasing are plotted versus the equivalence ratio. Phi vs. Load & CA50 1500 RPM, Phi = 1, Intake Air T = 40C Constant Cam Timing 6- * CA50 (CAD ATC) 4 3 2- 0.6 mENIMEP (bar) 0.7 0.8 0.9 1 1.1 Phi Figure 3.31 NIMEP and CA50 versus equivalence ratio with constant cam timing: EVC 98 BTC, IVC 42 ABC. 65 As the fueling rate is reduced, the combustion phasing is retarded. Less injected fuel means there is less fuel energy to raise the temperature of the combustion products. Lower temperature burned gas results in lower temperature residuals and later combustion phasing. 3.4.2 Fueling Rate: Constant Cam Timing - Efficiency & Emissions Changing the fueling rate changes the load, the combustion phasing, and the air-fuel ratio. All three impact efficiency. Figure 3.32 shows the indicated efficiency and PMEP versus equivalence ratio for the points in figure 3.31. As the equivalence ratio decreases, the ratio of specific heat of the mixture increases contributing to better efficiency. More charge dilution leads to lower burned gas temperatures, and so less heat transfer during the recompression event. Less heat transfer reduces pumping losses. Lower residual gas temperatures also contribute to lower cylinder temperatures during compression, and later combustion phasing. Later combustion phasing improves efficiency, ensuring that no work is lost during compression. As load decreases, pumping losses usually make a larger portion of the NIMEP, reducing overall efficiency. All of these factors contribute to relatively stable efficiency across a range of loads and equivalence ratios. Phi vs. Load & CA50 1500 RPM, Phi = 1, Intake Air T = 40C Cam Timing -Constant 0.31 0.3 0.29 0.28 o 0.27 A?~ 0.26 . Indicated Efficiency ( m PMEP (bar) + j0.25 0.24 _ W 0.23 0.6 0.7 0.9 0.8 1 1.1 Phi Figure 3.32 Efficiency and PMEP versus phi with constant cam timing: EVC 98 BTC, IVC 42 ABC. 66 Just as efficiency is impacted by load, combustion phasing, and equivalence ratio, so are emissions. Figures 3.33, 3.34, 3.35 show HC, CO, and NO emissions for a range of equivalence ratios for the points in figure 3.31 (constant cam timing). The HC emissions increase at leaner equivalence ratios. During lean operation, cylinder temperatures are lower and combustion phasing is later, both of which contribute to lower peak temperatures and higher HC emissions. Phi vs. Emissions 1500 RPM, Phi =1, Intake Air T = 40C 0.025 0.02 - 0.015 0.01 z 0.005 0 0.6 0.7 I I I 0.8 0.9 1 1.1 Phi Figure 3.33 HC emissions versus equivalence ratio. CO emissions show a dramatic drop as the equivalence ratio becomes leaner. At the leanest point (near the operating limit), the emissions of CO rise again. The drop in CO is consistent with the sensitivity of CO emissions to air-fuel ratio. Phi vs. Emissions 1500 RPM, Phi = 1, Intake Air T = 40C 0.05 1 0.04 0.03 - 130.02 0 0.01 0 0.6 0.7 0.8 0.9 1 1.1 Phi Figure 3.35 CO emissions versus equivalence ratio. 67 The increase in CO at the leanest point is consistent with the increase in CO seen near the limits of operation. Near the lean limit, the low cylinder temperatures lead to incomplete combustion and higher CO emissions. Emissions of NO increase slightly at an equivalence ratio of 0.9, and then fall at leaner equivalence ratios. These results are consistent NO emissions in SI engines: NO emissions peak at an equivalence ratio of approximately 0.9. Phi vs. Emissions 1500 RPM, Phi 1, Intake Air T = 40C 0.001 0.0008 0.0006 %.00.0004 A A I I 0.9 1 A 0 _ Z 0.0002 A 0 I 0.6 0.7 0.8 1.1 Phi Figure 3.35 NO emissions versus equivalence ratio. 3.4.3 Fueling Rate: Constant Fueling - Load & Combustion Phasing In an attempt to minimize the impact of varying load on combustion phasing, efficiency, and emissions, the impact of equivalence ratio was evaluated while holding the fueling rate constant. The equivalence ratio was varied by adjusting the exhaust cam to trap more or less residual mass in the cylinder. Trapping less residual mass leaves more room in the cylinder for fresh air. If the fuel pulse width is held constant and the air flow into the cylinder is increased, the equivalence ratio will drop. Conversely trapping a larger residual mass allows less air into the cylinder, and will result in a rise in the equivalence ratio. As the equivalence ratio increases in figure 3.36, the load drops (-0.25 bar from phi = 0.8 - 1.0). Increased pumping losses account for 0.1 bar of the 0.25 drop in NIMEP (figure 3.38). More trapped residuals result in higher pressures and temperatures during recompression, and more losses from heat transfer to the cylinder walls. The remainder of the decrease in load can be attributed to changing combustion phasing. 68 Phi vs. Load & CA50 1500 RPM, Phi = 1, Intake Air T = 40C Constant Fuel Pulse Width 4 U 3.5 0 3 LO 2.5 4 2 zU 1.5 1 0.5 0 0 .75 *NIMEP (bar) . . * CA50 (CAD ATC) U 0.85 0.95 1.05 Phi Figure 3.36 NIMEP versus equivalence ratio with constant fuel pulse width. Figure 3.37 shows the combustion phasing and maximum rate of pressure rise for the same points. As the mixture becomes more and more dilute (lower equivalence ratio), combustion temperatures drop, and residual temperatures drop. At the same time, the trapped residual mass also decreases as EVC is retarded to reduce equivalence ratio. Both of these factors contribute to cooler temperatures during the compression stroke and later combustion phasing as the mixture becomes leaner. LO 0 4 3.5 3 2.5 2 1.5 1 0.5 0 0.75 Phi vs. dP/dt & CA50 1500 RPM, Phi = 1, Intake Air T = 40C Constant Fuel Pulse Width U *dP/dt (MPa/ms) R.. * CA50 (CAD ATC) U I 0.85 I 0.95 1.05 Phi Figure 3.37 CA50 & maximum average dP/dt versus equivalence ratio with constant fuel pulse width. 69 The one data point taken at a rich air-fuel ratio shows a significant delay in combustion phasing compared to stoichiometric operation. The excess fuel reduces the combustion efficiency, and increases the specific heat of the mixture. Both of these factors contribute to cooler cylinder temperatures and later combustion phasing. The maximum rate of pressure rise during the combustion event gives some indication of the combustion duration. All other factors being equal, faster combustion will lead to higher rates of pressure rise. The earliest CA50 occurs at stoichiometric operation, but interestingly the highest rate of pressure rise occurs at an equivalence ratio of 0.9. 3.4.3 Fueling Rate: Constant Fueling - Efficiency & Emissions As discussed, equivalence ratio, load, and combustion phasing all have an impact on efficiency.. Richer operation results in a lower ratio of specific heats. Lower load operation means that PMEP contributes more to overall NIMEP. If the start of combustion advances beyond TC, work is lost during the compression stroke. All of these factors contribute to the lower efficiencies at higher equivalence ratios (figure 3.38). As mentioned previously, the earlier EVC to trap more residuals (and reduce equivalence ratio) results in higher PMEP. Phi vs. Efficiency & PMEP 1500 RPM, Intake Air T = 40C 0.3 - 0.29 -- .0.29 a0 .2 8 - +Indicated Efficiency m PMEP (bar) 0.27 - -i 0.26 - W 0.25 0.24 1 0.75 0.85 0.95 1.05 Phi Figure 3.38 Efficiency & PMEP versus equivalence ratio with constant fuel pulse width. 70 Figure 3.39 shows HC emissions versus equivalence ratio for the points in figure 3.36. As the equivalence ratio is increased from lean towards stoichiometric, the HC emissions fall. As the equivalence ratio is increased towards stoichiometric, combustion advanced towards TC, and peak cylinder temperatures rise. Higher temperatures contribute to more complete combustion. HC emissions are very similar at equivalence ratios from 0.9 to 1.0. The HC emissions increase as equivalence ratio is increased above 1.0. Phi vs. Emissions 1500 RPM, Phi =1, Intake Air T = 40C 0.015 j 0.014''0.013 0.012 - z 0.011 - c 0.01- 0.75 0.85 0.95 1.05 Phi Figure 3.39 HC emissions versus equivalence ratio with constant fuel pulse width. Phi vs. Emissions 1500 RPM, Phi = 1, Intake Air T = 40C . 0.3 0.25 0.2 'M C> 0.15 0. 0 0.05 0 00.75 U U 0.85 U U 1.05 0.95 1.05 Phi Figure 3.40 CO emissions versus equivalence ratio with constant fuel pulse width. 71 As in the plots with constant cam timing, CO emissions are very sensitive to equivalence ratio. Lean air-fuel mixtures result in very low emissions of CO. Unlike the leanest point with constant cam timing, the leanest point with constant fueling is not near the operating boundary, and so the CO emissions remain low. Rich operating results in very high emissions of CO. NO emissions show similar trends to those shown with constant cam timing. NO emissions seem to peak at an equivalence ratio of 0.9. On either side of the peak, NO emissions decline. Phi vs. Emissions 1500 RPM, Phi 1, Intake Air T = 40C 0.0004 ~QQQ3A 0.0003A 0.0002 oz A A A A A 0.0001 00.75 0.85 0.95 A 1 1.05 Phi Figure 3.41 NO emissions versus equivalence ratio with constant fuel pulse width. 72 3.5 Operating Limits The HCCI operating range is determined by how the operating limits are defined. The two main criteria defining the operating limits are combustion stability and engine knock. When combustion becomes unstable or when engine knock becomes unacceptable, operation is no longer feasible. 3.5.1 Low Load Limit The first criterion defining the low load limit is misfire. Load control is accomplished by either adjusting the fueling or by controlling the trapped residual mass and maintaining a constant air-fuel ratio. Lower loads will have lower concentrations of fuel and higher. concentrations of residuals or excess air. As the dilution increases, the burned gas temperatures decrease (see figure 3.5). At some point the burned gas temperatures are so low that the residuals are not hot enough to raise the temperature of the fresh charge to the temperature necessary for ignition. If there is one misfire, the temperature of the residuals on the next cycle will be very low (no combustion), and so there will be no combustion on future cycles. The second criterion defining the low load limit is combustion stability, measured in terms of COV of NIMEP. As the residual fraction increases and the load is reduced, combustion becomes less stable. Unstable combustion will lead to misfire, but in some cases unstable combustion can be maintained before reaching misfire. A limit of 3.5% was selected as an upper limit for COV of NIMEP. If the COV increases above 3.5%, combustion was defined as unstable. This limit is consistent with industry standards. Figure 3.42 shows the COV of NIMEP at the low load limit. As load was reduced the COV remained stable until near the limit of operation. Near the limit, the COV increased quickly, and a misfire occurred before the COV reached 3.5%. The third criterion is that the specific fuel consumption of any point within the operating range must be better than SI specific fuel consumption at the same load and speed. The baseline SI data is used to provide a comparison. Near the low load limit this criterion 73 never came into effect, as the efficiency of HCCI is substantially better than throttled SI operation. Efficiency is plotted versus NIMEP near the low load limit in figure 3.43. The solid black line shows baseline SI efficiency. Low Load Limit: NIMEP vs. COV 1500 RPM, Intake T 30C, WOT, Phi =1 0.035 I. uJ 0.031 0.027 0 > 0.023 0.019 0.015 1 1.5 2 2.5 3 NIMEP (bar) Figure 3.42 COV of NIMEP versus NIMEP. Low Load Limit: Load vs Efficiency 1500 RPM, Intake 30 C, WOT, Phi =1 0.34> 0.32 .S + 0.3 - E 0.28 0 0.26 - SI Baseline 0.24 0.24C~ 0.22 -- 0.2 I 1 2 3 4 NIMEP (bar) Figure 3.43 NISFC versus NIMEP. 74 3.5.2 High Load Limit The first phenomenon defining the high load limit is engine knock. In order to increase load, the amount of fuel within the cylinder must be increased. As load increases, the concentration of fuel in the cylinder increases, and combustion temperatures increase. Higher combustion temperatures and higher concentrations of reactants lead to higher reaction rates. At some point local reaction rates become so high that pressure cannot equalize within the cylinder and pressure waves begin to resonate within the cylinder. These pressure waves can cause audible noise, and can lead to engine damage. For a small local area, the rate of pressure rise is dependent on the rate of heat release. The maximum rate of heat release without the generation of pressure waves can be calculated for the local area. The derivation is shown in Appendix C. Yelvington describes a different derivation of the same relationship in SAE 2003-01-1092 [20]. This simple relationship gives some insight into the characteristics that govern whether or not conditions will lead to engine knock. >(3;v where P q = volumetric rate of heat release (Watts/M 3 ) 7 = ratio of specific heats P = pressure within the local region a = speed of sound r = radius of local region In an SI engine, oscillations within the cylinder pressure signal indicate that there is autoignition of the end gas. In many cases this is associated with audible pinging and engine damage. In an HCCI engine the mechanism of combustion is auto-ignition. If the rates of pressure rise are slow enough, there are no pressure oscillations in the cylinder during combustion. As in SI engine operation, HCCI knocking conditions are caused by rapid 75 auto-ignition leading to engine noise and possibly engine damage. However several researchers have observed that it is possible for pressure oscillations to occur without audible knock. If pressure waves can occur without any audible knock, selecting a knock criterion becomes more challenging. High frequency data was collected from the engine to evaluate engine knock. The data included both cylinder pressure and audible data collected at a rate of 100 kHz. A microphone was placed 2 inches above the center of the firing cylinder to collect the audible data. Figure 3.44 shows high frequency audible data over the course of eight engine cycles for four different operating conditions. All four points are at 2000 RPM. The conditions vary from light load (2.3 bar NIMEP) to moderate load (3.5 bar NIMEP). At each operating condition a subjective evaluation of the knock intensity was made. The subjective knock rating labels are disvlaved near each of the four sub-plots. 1.5- 1.5 Slight Knock 1 I1I 0.5 0.5 0 0 -0.5 -0.5 -1 -1 0, 0 -1.5 0 0.1 1 5 1 0.5. 0.2 0.3 Time (s) 0.4 Knock li-A. L_.. i -1. I~ I. LL -I 0 -0.5 -1.5 0 'I I0.1 1 0.2 0.3 Time (s) I' 1 0.4 ] 1 f, I 0 0.1 1.5 0.2 0.3 Time (s) 1. 1 1 0.5 .1 1.1*1 0.4 0 -0.5 0 Heavy Knock 1.r. -1 -1.5 .1, 11 kI,- 1 0, 0 . 0.1 0.2 0.3 Time (s) 0.4 Figure 3.44 Eight cycles of unprocessed audible signal data sampled at a rate of 100 kHz for four operating conditions at 2000 RPM. 76 With the exception of the "No Knock" data, the plots show peaks that correspond to the combustion event. The relative magnitudes of the peaks of the audible data seem to correlate with the subjective knock rating: No Knock shows no audible peaks, Heavy knock has strong peaks. A sample of the pressure data for the two extreme cases is shown in figures 3.45 and 3.46. Figure 3.45 shows the cylinder pressure during a combustion event for the "No Knock" operating point. There are clearly oscillations in the pressure data, even though knock could not be heard with the ear, and the audible data showed none of the peaks that the knocking cases displayed. The magnitude of the oscillations is on the order of 1 bar. 35 30 Cylinder Pressure (bar) 25 20 15 10- -5 0 5 10 15 20 25 30 35 40 CAD (0 CAD = TC) Figure 3.45 Cylinder pressure during the combustion event for no audible knock. Figure 3.46 shows the cylinder pressure during a combustion event for the "Heavy Knock" operating point. The magnitude of the oscillations is much larger at approximately 5 bar. Additionally the oscillations do not appear to be as smooth. This could be indicative of higher frequency oscillations, or perhaps oscillations at similar frequencies that are out of phase. 77 50 45 - Cylinder Pressure (bar) 40. 3025 20 151 -5 0 5 10 15 20 25 30 35 40 CAD (0 CAD = TC) Figure 3.46 Cylinder pressure during the combustion event for loud audible knock. Some frequency analysis of the cylinder pressure signal can give further insight into knock phenomenon. The values of the expected oscillation frequencies were calculated assuming that the engine cylinder was a perfect cylinder with a fixed volume [24]. The frequencies are a function of not only the geometry of the cylinder, but also the cylinder mixture composition and temperature. The gas properties for the calculations were estimated using WAVE. The calculated frequencies range from a 1st mode of 5.75 kHz up to 16.65 kHz. The calculated frequencies and an illustration for each mode are shown in table 3.1. The range of actual frequencies seen in the pressure data is displayed in figure 3.47. The plots show the magnitude of the FFT from 4 to 20 kHz for each of the four cases shown in figure 3.44. Again each plot is labeled with a subjective knock rating. In all four cases there is a large peak in the magnitude at just over 5 kHz, as expected. The other peaks in 78 the magnitude of the FFT are close to the expected resonant frequencies. The magnitude of the peaks seems to correlate pretty well with the subjective knock ratings. Pm,n B = 87.5 mm Gamma = 1.22 R = 302 T= 2000K 2RT fm,n n 1.84 3.054 3.832 4.201 5.332 5.75 9.54 11.97 13.12 16.65 1'm,n f,n(kHZ ) Table 3.1 Expected resonant frequencies in a cylinder - SAE 2003-01-3217. 100 100 80 80 60 60 40 40 20 20 0 5 10 15 Frequency (kHz) 2) 0 100 80 80 60 60 40 40 20 20 5 10 15 Frequency (kHz) 10 15 20 Frequency (kHz) 100 0 5 20 0 5 10 15 Frequency (kHz) 20 Figure 3.47 FFT of the pressure signal sampled at a rate of 100kHz for four operation conditions. 79 The magnitudes are lowest for the "No Knock" case and highest for the "Heavy Knock" case. There does not appear to be a significant difference between the "Slight Knock" case and "Knock" case. The "Knock" case does show a small peak at approximately 12 kHz that does not appear in the "Slight Knock" case. Previous work has suggested that knock in HCCI engine operation correlates with excitation of higher mode frequencies [25]. However the correlation between the magnitude of higher frequency mode and knock intensity was not clearly better than the correlation between the magnitude of the first mode and knock intensity. Research Group SAE Paper Type of Knock Criterion Equivalent dP/dt AVL 2003-01-0754 2 bar/cad @ 3000 RPM 3.6 MPa/ms Brunel 2001-01-1030 0.5 bar oscillation on 10% of cycles Cosworth 2005-01-0133 6-8 bar/cad @ 1500 RPM 5.6-7.2 MPa/ms GM 2002-01-2859 Ringing Index - (dP/dt)A2/Pmax - 3 MPa/ms Lotus 2005-01-0157 dP/cad Lund 980787 12 bar/cad @ 1000 RPM Nipoon Oil 2005-01-0138 dP/cad 7.2 MPa/ms Table 3.2 Survey of some HCCI knock criteria found in literature. One knock criterion that has been used by a number of HCCI researchers, and one that seems to correlate well with audible knock, is the maximum rate of pressure rise within the cylinder. The local knock criterion suggests that knock will occur when rates of heat release rise above a certain threshold. Since the rate of pressure rise within the cylinder is largely proportional to the rate of heat release, it seems reasonable to set a threshold for a maximum rate of pressure rise. For this project, the knock limit was defined as an average maximum rate of pressure rise over 100 cycles of 5 MPa/ms. Figure 3.48 shows a plot of the maximum rate of pressure rise during combustion versus the integration of the magnitude of the FFT of the audible data during combustion (a measure of engine 80 noise). The maximum rate of pressure rise was calculated based on the same 1/CAD sampling rate that is used for most data collection. The FFT of the audible data was calculated using the 100 kHz sample rate. The data displays approximately 25 consecutive engine cycles for four different operating conditions. Each point shows the engine noise and maximum rate of pressure rise for a single cycle. The legend shows the subjective knock ratings for each condition. As the maximum rate of pressure rise in the cylinder increases, the audible noise also increases. The "No Knock" points all fall below the threshold of 5 MPa/ms. The "Slight Knock - Acceptable" points mostly fall below the threshold. The "Slight Knock - Unacceptable" points are mostly above the threshold, and the "Knock" points are all above the threshold. 11 1 A - 10- 8- * A E 0L 4- 3 AS, M* A_ gh - TKnock A A 2A~ A A A * A A A Slight KnockAcceptable No Knock 1 'I ' ' 7 6 5 4 3 2 1 0 Integrated Magnitude of FFT of Audible Data During Combustion X104 1 Figure 3.48 Integrated FFT of audible data versus maximum dP/dt on a cycle by cycle basis. The second criterion for the high load limit is misfire. At some operating conditions, increasing the load can result in misfire. Figure 3.49 again shows cylinder temperature versus residual fraction. As the residual fraction decreases, the burned gas temperature increases. Although the residual temperatures are higher, there are fewer residuals to heat 81 the fresh charge. This can lead to lower cylinder temperatures during compression, late combustions phasing, and in some cases misfire. WAVE: Residual vs. Cylinder T 1500 RPM, Phi= 1, In T 30C 885 I- m 880 CE, @875 870S 865 860 - 35 45 55 65 75 Residual Mass % Figure 3.49 WAVE modeling: cylinder temperature at IVC versus residual fraction. Misfire near the high load limit can also occur if combustion becomes unstable. In general HCCI combustion is stable. Late combustion on one cycle leads to higher residual gas temperatures. Higher residual gas temperatures contribute to higher cylinder temperatures during compression and earlier combustion phasing. Any variability in combustion phasing is usually self correcting and stable combustion is reached after a few cycles. However near the limits of operation there are instances when large swings in combustion phasing can lead to misfire. Figure 3.50 shows NIMEP, max dP/dt, and CA50 for 25 cycles at the high load limit at 2500 RPM. At cycle 17, there is a misfire and combustion is lost. The misfire can be seen in the NIMEP which drops from a stable value of - 2.5 bar, to -zero. From cycles 0-12, there are swings in combustion phasing demonstrating the relatively stable nature of HCCI combustion. In cycle 15 a very late combustion event leads to very hot residuals. These hot residuals lead to early combustion before top center. The maximum rate of pressure rise in this cycle is very high (as is the peak pressure) indicating a very fast combustion event. The residuals after 82 this early combustion event are cooler, and combustion cannot be maintained on the next cycle. 30 NIMEP*10 (bar) Misfire 25- I- 20- I- 0 15- CA50 (CAD ATC) C; 10 4. Max dP/dt (bar/CAD) 0- 5 10 15 20 Cycle # Figure 3.50 Combustion phasing, max dP/dt, and NIMEP over consecutive cycles at the high load limit. Figure 3.51 shows the combustion phasing for the four cycles preceding the misfire, and the simulated residual gas temperature following combustion for each of those cycles. The temperatures were calculated using WAVE. On the 15t cycle combustion is very late. This leads to very high exhaust gas temperatures at IVO (residual gas temperatures). The hot residuals mix with the fresh charge, and the higher cylinder temperature leads to early combustion on cycle 16. The early combustion of cycle 16 leads to lower residual temperatures, and misfire on cycle 17. 83 CA50 & WAVE Residual T: Cycles leading to misfire 2500 RPM, Phi = 1, Intake Air = 30C & I0 U, C.) 30 20 10 0 -10 I I 12 13 14 15 16 17 12 13 14 15 16 17 770 760 0 > 750 740 - 730 720 710 Cycle# Figure 3.51 CA50 and WAVE calculated cylinder temperature at IVO for the 4 cycles before misfire. The final criterion for the high load limit is that the specific fuel consumption for all points must be better than SI specific fuel consumption at the same speed and load. Near the high load limit, the efficiency of HCCI operation is only slightly better than SI operation, but this criterion rarely defines the high load limit. 3.5.3 Summary of Operating Limits Low Load Limit " No Misfire * COV of NIMEP < 3.5% * HCCI NISFC < SI NISFC High Load Limit " Avg. Max rate of Pressure Rise < 5 MPa/ms " No Misfire " HCCI NISFC < SI NISFC Table 3.3 Summary of operating limits. 84 3.6 Cam Phasing and the High & Low Load Limits High and low load operation is limited by how the gas exchange and combustion processes occur. Cam phasing offers some control over combustion phasing, and combustion phasing impacts both the high and low load limit. This section discusses optimal cam timing to maximize the load before reaching the high load limit, and to minimize the load before reaching the low load limit. 3.6.1 High Load Limit The high load limit is reached when the load cannot be increased further without violating the knock criteria, becoming unstable, or achieving efficiency worse than SI efficiency. For this engine, the knock criteria and stability criteria are almost always the limiting factors in high load operation. To increase the load the exhaust cam is retarded. Later exhaust valve closing traps less residual mass allowing more room for fresh charge. More fresh fuel and air allow higher load operation. As the load increases, the concentration of fuel and air increases, and the tendency of the mixture to knock unacceptably during combustion increases. As discussed in section 3.5.2, the knock criteria used in this work is an average maximum rate of pressure rise. By reducing the maximum rate of pressure rise, the high load limit can be maximized. Combustion phasing has a significant impact on the maximum rate of pressure rise. Figure 3.52 shows combustion phasing versus maximum rate of pressure rise at a constant load (same points as shown in figure 3.22). The dashed line shows the approximate location of the knock limit. Earlier combustion phasing correlates with higher rates of pressure rise. If combustion begins before top center, the motion of the piston will contribute to the rise in cylinder pressure. If combustion occurs after top center the motion of the piston will contribute to a reduction in the cylinder pressure. 85 Max dP/dt & Combustion Phasing 1500 RPM, Intake Temperature 30C, WOT NIMEP 3.5 bar 6 E 5.6 10 + 5.2 --- - 2 4.8 x (U "" .mEnockJ """" Limit *EVC 97 CAD BT( N EVC 93 CAD BTC "" *EVC 91 CAD BTC 4.4 4 0 4 2 6 CA50 (CAD ATC) Figure 3.52 Max dP/dt versus combustion phasing. Figure 3.53 shows the max rate of pressure rise versus intake cam timing for the same points. Earlier intake cam timing can retard combustion phasing, and therefore can reduce the maximum rate of pressure rise. Max dP/dt & Cam Timing 1500 RPM, Intake Temperature 30C, WOT NIMEP 3.5 bar 6 5.6 2<Knock EVC 97 CAD BTC mEVC 93 CAD BTC EVC 91 CAD BTC Limit -. 4.8 x 4.4 4 1 20 30 40 50 60 70 IVC (CAD ABC) Figure 3.53 Max dP/dt versus intake cam timing. 86 Avoiding knock through later combustion phasing will allow some extension of the high load limit. However at some point combustion will become so late that it will become unstable and a misfire will occur. The high load limit is reached when retarding combustion phasing results in unstable combustion, and advancing combustion phasing results in knock. The cam phasing diagram in figure 3.54 shows the overall cam timing strategy for maximizing the high load limit. Exhaust cam phasing is retarded to reduce the trapped residual mass allowing more fresh charge into the cylinder and higher load operation. Intake cam timing is advanced or retarded as necessary to adjust combustion phasing to avoid knock or unstable combustion. TC NO EVC C Intake BC To increase load EVC is retarded to reduce trapped residual mass, and IVC is adjusted as necessary to optimize cam timing. EVO Figure 3.54 High load limit: cam phasing strategy. Figure 3.55 shows a mapping of the high load limit for one operating condition in terms of cam timing. The red diamonds mark the knock limit, and the black circles mark the stability limit. An example of a high load search trajectory (triangles) is plotted on top of the map. The search for a high load limit begins at a stable operating point that is not knocking. In this case, the exhaust cam is retarded to increase the load. The new operating point violates the knock limit. Intake cam timing is advanced to retard combustion phasing and avoid knock, and exhaust cam timing is retarded again. 87 Exhaust & Intake Cam Timing: High Load Limit 1500 RPM, Phi = 1, WOT 24 * Acceptable Points + Knock 22 0- 20 18 0 0 * p Start U FinishAA 16 A A A Trajectory Acceptable Points A Trajectory Knock A 14 12 10 65 70 * Unstable 75 85 80 A Trajectory Unstable EVC (CAD BTC) Figure 3.55 Mapping of the high load limit for one operating condition. Point ____ NIMEP COV (bar) dP/dt (MPa/ms) AA 4.29 1.3 4.95 B . 4.32 1.5 4.82 Table 3.4 Comparison of two operating points at the high load limit. The new point is stable and is not knocking. Exhaust cam timing is retarded further to increase load. Eventually combustion starts to become unstable. This is evident by significant variability in combustion phasing and variability in maximum rate of pressure rise. Intake cam timing is retarded to advance combustion phasing. The resulting points are closer to the knock limit. As exhaust cam timing is retarded further, combustion becomes unstable and a misfire occurs. The high load limit is the stable non-knocking point with the highest load. 88 Table 3.4 shows operating characteristics for two points: the high load limit from high load search (A), and the high load limit from the mapping process (B). The NIMEP of both points is very similar, and both points are very close to the knock limit. 3.6.2 Low Load Limit At the low load limit, the load cannot be reduced any further without resulting in unstable combustion. Unstable combustion is defined by either high COV of NIMEP, or by misfire. Advancing exhaust cam timing reduces load. Earlier exhaust cam phasing traps more residuals, leaving less room in the cylinder for fresh charge. As the residual fraction increases, and the amount fuel entering the cylinder decreases, and the burned gas temperatures decrease. As burned gas temperatures fall, the cylinder temperatures fall and eventually stable combustion cannot be maintained. TC EVC IVO IVC btk BC EVO Figure 3.56 Low load limit: cam phasing strategy. Unlike the case of the high load limit, intake cam phasing has only a small impact on the low load limit. Figure 3.57 shows a map of the low load limit in terms of the intake and exhaust cam timing. At each intake cam timing, the exhaust cam timing was swept until misfire occurred or combustion became unstable. For this engine speed most advanced EVC is reached at an IVC of approximately 40 CAD ABC. 89 Cam Timing Near Low Load Limit 1500 RPM, Phi =1 104 106 - -108 -- v + I 110 ( 112 2 114> 116 m 118120 20 30 40 50 60 IVC (CAD ABC) Figure 3.57 Cam timing at the low load limit. Figure 3.58 shows the NIMEP versus intake cam timing for the same points. Over an IVC range of 25 CAD (32 CAD ABC to 57 CAD ABC), the lowest load varied from approximately 1.55 bar to 1.65 bar NIMEP. Compared to operation at the high load limit, the low load limit is not very sensitive to intake cam timing. At the high load limit, a one or two degree change of intake cam timing could lead to either misfire or knock. Low Load Limit 1500 RPM, Phi =1 2.4 2.3.. 2.2 cc 2.1 2 wlU.9 ~1.8 1.7 1.6- Y 1.5 20 30 40 50 60 IVC (CAD ABC) Figure 3.58 Low load limit: NIMEP versus IVC. 90 Figure 3.58 also shows that for this operating condition the lowest load is achieved with IVC at 42 CAD ABC. This intake cam timing also happens to be intake cam timing that maximizes the air flow into the cylinder. Figure 3.59 shows the airflow into the cylinder versus intake cam timing for a range of exhaust cam timings. At each exhaust cam timing and fuel pulse width, the intake cam timing was swept and the air fuel ratio was measured. The air flow was calculated based on the fuel flow rate and the air fuel ratio. At 1500 RPM, the airflow seems to reach a maximum at an IVC of 42 CAD ABC, independent of exhaust cam timing. Air Flow vs. Cam Timing 1500 RPM 2.7 +EVC 94 BTC, Fuel PW 4250 2.5- **EVC 2.3 o 2.121 A 99 BTC, 2.3 1.9- Fuel PW 4000 A EVC 104BTC, Fuel PW 3650 A 1.7 1.5 1 20 * EVC 109 BTC, Fuel PW 3200 30 40 50 60 IVC (CAD ABC) Figure 3.59 Air flow into the engine versus IVC at a range of EVC timings. As figure 3.58 shows, this intake cam timing coincides with the cam timing that allows the lowest load operation. At all engine speeds this trend was consistent: intake cam timing for lowest load operation was found to coincide with intake cam timing for maximum air flow into the cylinder. This result is consistent with the arguments presented to describe the effect of intake cam timing on combustion phasing. Maximizing the air flow into the cylinder also maximizes the pressure at IVC. Higher pressure at IVC should result in higher temperature at IVC. Higher temperature during the compression stroke will tend to advance combustion phasing. Earlier combustion phasing will result in higher peak pressure and higher peak temperature. Higher temperature during 91 combustion will allow stable combustion with more dilute mixtures. More dilution results in lower load. The relationship between air flow and the low load limit is used to speed the search for the low load limit. For each operating condition of interest, the intake cam timing giving the maximum air flow into the cylinder is found. Exhaust cam timing is advanced to increase trapped residual mass and reduce load. Once the lowest possible load is reached, intake cam timing is varied to ensure that the low load limit has in fact been reached. Low Load Limit 1500 RPM, Phi = 1 2.4- 2.3 2.2 2.1 2' 1.6 - 1.5 - 20 30 40 50 60 IVC (CAD ABC) Figure 3.60 Low load limit search procedure. 3.6.3 Repeatability To ensure that the methods of finding the high and low load limits were repeatable, searches for the high and low load limits were repeated 10 times each. In each case the initial cam timing was varied to ensure that the trajectories would be different for each search. The results are summarized in table 3.5. High Load Limit Low Load Limit 90% Confidence Interval 0.04 0.02 Standard Deviation (bar) 0.07 0.03 Table 3.5 Repeatability for high and low load limit search procedures. 92 Chapter 4: Impact of Ambient Conditions on HCCI Operation Because HCCI combustion timing is determined by chemical kinetics, any variable that has an impact on cylinder conditions will have an impact on combustion. Intake air temperature, intake air humidity, and gasoline fuel composition can vary daily, seasonally, and regionally. Changes in these variables do have an impact on cylinder conditions, and can affect the limits of operation. 4.1 Intake Air Temperature 4.1.1 Effect of intake air temperature at a point Production engines are generally expected to operate reliably in ambient air temperatures varying from approximately -40C to 50C, a range of 1 OC. The range of intake air temperatures for testing was therefore selected to be approximately 1 OC. The temperature of the intake air was measured approximately 5 cm from the intake port. With the selected chiller and air heat exchanger, the minimum achievable intake air temperature was -6 to -8 C (depending on operating condition). The maximum intake air temperature was selected to be 100C. Although this air temperature is higher than normal ambient operating conditions, it allows the evaluation of an approximately 110 degree C range of temperatures. First the effect of intake air temperature on one operating point was examined. Engine speed, cam phasing, and equivalence ratio (4) were held constant as the temperature of the intake air was varied. As intake air temperature increases, the density of the air decreases resulting in less inducted air mass. If the air-fuel ratio is constant, less inducted air mass results in less inducted fuel and a lower NIMEP. A reduction in the inducted fuel mass will contribute to lower combustion temperatures and lower residual temperatures. However the impact of the higher intake air temperature outweighs the impact of the lower burned gas temperature, and combustion is advanced. The NIMEP falls by 7% as the temperature of the intake air is increased by 70 degrees C, and the combustion phasing advances by 2.5 CAD (Figure 4.1). Figure 4.2 shows a reduction in inducted 93 fresh charge mass of approximately 6% over the same temperature range. The decrease in NIMEP is to a large extent due to the reduction in fueling, and to a lesser extent due to the change in combustion phasing. Effect of Intake Air Temperature 1000 RPM, WOT, Phi = 1 EVC = 108 BTC, IVC = 103 BTC 0 .) 4 0 3.5 * NIMEP (bar) 3 X 2.5 A 2 w A CA50 (CAD A ATC) A A 1.5 1 0.5 * Max dP/dt A (Mpa/ms) L 0 A 25 45 65 85 105 Intake Air Temp (C) Figure 4.1 NIMEP, CA50, and maximum dP/dt versus intake air temperature. Effect of Intake Air Temperature 1000 RPM, WOT, Phi = 1 EVC = 108 BTC, IVO =103 ATC 220 - 218 216 , 214 212 210 * 208 * 206 204 202 25 45 65 85 105 Intake Air Temp (C) Figure 4.2 Inducted mass versus intake air temperature. 94 Figures 4.3 and 4.4 show the results of modeling using WAVE engine simulation software. Valve timing, combustion phasing, and intake air temperature were input into the model. As the air temperature increases, the reduced density of the fresh air results in a small increase in residual fraction. Although a higher residual fraction would contribute to lower burned gas temperatures and lower residual temperatures, the impact of the increased intake air temperature leads to hotter cylinder temperatures. The higher cylinder temperatures seen in figure 4.4 are consistent with the earlier combustion phasing. In figure 4.3 the percentage of residual mass in the cylinder ranges from approximately 58% to 61%. WAVE modeling shows a change in cylinder temperature (30 CAD BTC combustion) of approximately 18 degrees. For an operating condition with a lower residual fraction, intake air temperature will have a larger impact on operation. If a larger fraction of the cylinder mass consists of fresh air, any changes in the temperature of that air will have a larger impact on the overall cylinder temperature. Modeling: Intake Air T and Res% 1000 RPM, WOT, Phi = 1 EVC = 108 BTC, IVO = 103 ATC 62 O61 c60 59 - * 58 57 25 i I I 45 65 85 105 Intake Air Temp (C) Figure 4.3 WAVE modeling: residual mass fraction versus intake air temp. 95 Modeling: Intake Air T and Compression T 1000 RPM, WOT, Phi = 1 EVC = 108 BTC, IVO = 103 ATC 805 800 795 790 785 25 I I I 45 65 85 105 Intake Air Temp (C) Figure 4.4 WAVE modeling: cylinder temperature at 30 BTC versus intake air temp. 4.1.2 Operating Range Limits and Temperature The effect of intake air temperature was evaluated across a range of engine speeds and equivalence ratios. The operating conditions are listed in table 4.1. Thus the experimental matrix consists of a 4x4x4 matrix for a total of 64 operating conditions. At each operating condition the high and low load limit was found. Intake Air Speed Fuel-Air Equivalence Temperature 100 1000 1.0 70 1500 0.9 30 2000 0.8 -6 2500 0.7 Table 4.1 Operating range limits were evaluated at a range of intake air temperatures, engine speeds, and equivalence ratios. 96 Figures 4.5-4.8 show the high and low load limits for a range of equivalence ratios, engine speeds, and intake air temperatures. Higher intake air temperatures lead to a reduction in the maximum achievable load at low speeds, but seem to have little impact at higher speeds (more in Section 4.1.3: Operating Range Limits and Engine Speed). Across all speeds, higher intake air temperatures allow a slight reduction in the low load limit (a few tenths of bar over 110 degrees C). Higher temperatures allow more dilute operation, and lower loads can be reached. Operation at lean equivalence ratios (0.7, 0.8) is more sensitive to intake air temperature than stoichiometric operation. During lean operation, fresh air is a larger fraction of the cylinder mass, and so the impact of changes in the temperature of that air is larger (more in Section 4.1.4: Operating Range Limits and Equivalence Ratio). 1000 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.5 5 * 4.5 U A 4 I- .0 0. 3.5 * Phi = 1 4 m Phi = 0.9 A-i A Phi = 0.8 m Phi = 0.7 3 2.5 2 1.5 1 -10 10 30 50 70 90 110 Intake Air Temperature (C) Figure 4.5 High and low load limits for a range of intake air temperatures and equivalence ratios at 1000 rpm. 1500 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.5 5 97 Figure 4.6 High and low load limits for a range of intake air temperatures and equivalence ratios at 1500 rpm. 2000 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.5 5 4.5 4 *Phi = 1 3.5 EPhi = 0.9 3 A Phi = 0.8 Z 2.5- Phi = 0.7 LU 2 1.5 1 -10 10 30 50 70 90 110 Intake Air Temperature (C) Figure 4.7 High and low load limits for a range of intake air temperatures and equivalence ratios at 2000 rpm. 2500 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.5 T 98 Figure 4.8 High and low load limits for a range of intake air temperatures and equivalence ratios at 2500 rpm. The large majority of load limits at all speeds, temperatures, and equivalence ratios are limited by combustion stability. For a given engine speed and equivalence ratio, intake cam timing can compensate to a large degree for the advanced combustion associated with higher intake air temperature. Figure 4.1 shows combustion phasing advancing by approximately 2.5 degrees for an increase in intake air temperature of 70C. Combustion phasing can be controlled over a range of at least 5 CAD (figure 3.22). Although intake cam timing can compensate for the changes in combustion phasing seen with increasing intake air temperature, higher air temperatures do have an impact on the maximum achievable load. Figure 4.9 shows the high and low load limits for stoichiometric operation at 1500 RPM. As intake air temperature is increased, the maximum attainable NIMEP decreases by 0.86 bar. Over the same temperature range the minimum attainable NIMEP is reduced by 0.31 bar. As intake air temperature increases, cylinder temperatures rise and combustion phasing advances. 1500 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 99 5.5 I I Figure 4.9 High and low load limits versus intake air temperature at 1500 RPM with phi = 1. 1500 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 40 35 30 25I*Phi =1 g 20 15 10 5 0 -20 0 20 40 60 80 100 120 Intake Air Temperature (C) Figure 4.10 Intake air temperature versus intake cam timing at the high load limit. 100 At the low load limit, higher temperatures and more advanced combustion phasing allow stable combustion with more dilute mixtures, and therefore lower minimum loads. At the high load limit, higher temperatures and more advanced combustion phasing increase the tendency of the engine to knock. Intake cam timing advancement can reduce the tendency of the engine to knock by retarding combustion phasing. The intake cam timings for the high load limits at 1500 RPM with a stoichiometric equivalence ratio are shown in figure 4.10. Advancing intake cam timing retards combustion phasing by restricting the flow into the cylinder. Restricted flow results in lower pressure at IVC, and so lower temperatures during compression (and later combustion phasing). At higher intake air temperatures, more intake cam advancement is necessary to compensate for the higher cylinder temperatures. More intake cam advancement results in lower pressures at IVC, lower density of the cylinder mixture, and less fresh charge in the cylinder. The data shows trends that are consistent with these explanations. As temperature increases and intake cam is advanced to compensate, figure 4.11 shows that the inducted mass falls. Greater intake cam advancement leads to more restriction of the flow, and the pressure at IVC falls. Inducted Mass vs. Intake Air T 1500 RPM, WOT, Phi = 1 300 290 %E 280 270 260 2 250 = 240 230220 -20 I 0 I I I 20 40 60 80 Intake Air Temp (C) 100 120 Figure 4.11 Inducted mass at the high load limit versus intake air temperature. 101 This drop in pressure with advanced intake cam timing can be seen in figure 4.12. At the higher temperatures (and more advanced intake cam positions), the cylinder pressure before at 30 CAD before TC decreases. Cylinder Pressure vs. Intake Air T 1500 RPM, Phi =1, WOT e 8.4 8.2 8 7.8 7.6 u) 7.4 - 7 .2 I -20 0 I I 20 40 I I 60 80 100 120 Intake Air Temperature (C) Figure 4.12 Pressure at 30 CAD BTC for the high load limits versus intake air temperature. Interestingly, the residual fraction remains constant across the range of intake air temperatures. The residual fractions in figure 4.13 were estimated using WAVE. This suggests that there may be a concentration limit for this operating condition. When residual fraction falls below a certain level, knock cannot be avoided no matter how much combustion is retarded. With higher intake air temperatures, more intake cam advancement is required to avoid knock. Advanced intake cam timing results in lower pressures at IVC, less dense charge, and the concentration limit is reached at a lower load. 102 Resisdual Fraction % vs. Intake Air T 1500 RPM, WOT, Phi =1 48 46 - 44 42 Lu40 s 38 7 36 34 32 -20 0 20 40 60 80 100 120 Intake Air Temp (C) Figure 4.13 WAVE modeling: residual fraction at the high load limit versus intake air temperature. 4.1.3 Operating Range and Engine Speed As seen in figure 4.14, engine speed has a much larger impact on the operating range than intake air temperature. Figure 4.14 shows the same data points as figures 4.5-4.8, but the points have been reconfigured to show the impact of engine speed. As engine speed increases, the high load limit is significantly reduced. The low load limit is also reduced as engine speed increases. The major impact of higher engine speeds is a reduction in the amount of time available for combustion, heat transfer, gas exchange, and the other processes that affect operation. As the engine speed increases, there is less time for the gas to flow into the cylinder, and less fresh charge will result in lower load. There is less time for heat transfer suggesting that the charge will be hotter. Combustion will have to occur more quickly at higher engine speeds, which will require higher cylinder temperatures. 103 Phi = 1, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.5 5 4.5 -E A 4 *100C S3.5 m 70C 3 A 30C Z 2.5- m -6C W 2 1.5- 1 500 1000 1500 2000 2500 3000 RPM Figure 4.14 High and low load limits for a range of intake air temperatures and engine speeds at an equivalence ratio of 1.0. The very small valve lift (2 mm) significantly restricts the air flow into the engine at higher speeds. Figure 4.15 shows the air inducted per cycle versus engine speed for points with constant cam timing. The air flow was calculated based on the fuel flow and the air-fuel ratio measurement. As the engine speed increases, the air flow drops. If the drop in the high load limit versus engine speed is compared with the drop in inducted air, they agree reasonably well. This suggests that most of the reduction in load at the high load limit is a function of the reduced volumetric efficiency. 104 Air Flow vs. Engine Speed EVC 98 BTC, IVC 42 ABC, Phi 260 Z240 220 E 200 -g 180 - 160 S140 * 120 100 500 1000 1500 2000 1 2500 3000 Engine Speed (RPM) Figure 4.15 Air flow versus engine speed at constant cam timing. High Load Limit: RPM & NIMEP Phi = 1, Intake Air T 30C, WOT 5.5 5Z 4.5- a. 4 23.5 3 2.51 2500 1000 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.16 Engine speed versus high load limit at phi = 1, and intake air temperature 30C. Figure 4.16 shows engine speed versus the high load limit for stoichiometric operation with the intake air temperature at 30 degrees C. As the engine speed increases, the load drops dramatically. Figure 4.17 compares the reduction in air flow (figure 4.15) with the reduction in the high load limit (figure 4.16). Both are displayed in terms of percentage of the air flow or NIMEP at 1000 RPM. The reduction in high load limits follows the reduction in volumetric efficiency fairly closely. 105 Reduction in Air Flow and High Load Limit EVC 98 BTC, IVC 42 ABC, Phi=1 110 - 100- 0 % Air Flow @1000 90+ 80 2u z 70 0%60C % NIMEP @1000 50 40 500 1000 I I I 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.17 Engine speed versus % reduction in air flow and NIMEP. However a closer look indicates that engine speed is limiting high load operation in other ways. Figure 4.18 shows the residual fraction at the high load limit. From 1000 to 2000 RPM there appears to be a very slight increase in residual fraction from 36% to 40%. The residual fraction at 2500 RPM is significantly higher at 53%. The higher residual fraction at 2500 RPM is not a result of cam phasing limits: neither the exhaust cam nor the intake cam reached the end of the possible range. The exhaust cam could have been retarded further to reduce residual fraction, but operation at a lower residual fraction was not possible without violating the high load constraints. The higher residual fraction for the high load limit at 2500 RPM suggests that combustion is limited in another way. In addition to reducing volumetric efficiency, higher engine speeds also allow less time for combustion. The ignition delay is largely a function of temperature, so as engine speed increases, the mixture temperature must also increase to maintain a constant combustion phasing. If cylinder temperature were held constant and the engine speed were increased, combustion would occur later and later in the cycle until misfire occurred. Higher engine speeds require higher cylinder temperatures to ensure that combustion occurs in the desired location. 106 High Load Limit & RPM: Res Fraction Phi =1, Intake Air T 30C 0.5$-. 0.5 w 0.45- u 0.4i 0.350.3 .i W 0.25 0.2- 500 1500 2500 Engine Speed Figure 4.18 Engine speed versus residual fraction for the high load limit. Fortunately just as higher engine speeds allow less time for combustion, higher speeds also allow less time for heat transfer. Figure 4.19 shows the heat transfer per cycle from the gases in the cylinder to the cylinder walls. The heat transfer was estimated by matching cylinder pressure data with the WAVE engine simulation cylinder pressure. The modeled points have the same operation conditions as the high load limits shown in figure 4.16. Hiah Load Limit: RPM & Heat Transfer Phi = 1, Intake Air T 30C, -% 350 300 250 200 150I 1001 500 1000 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.19 WAVE modeling: Engine speed versus heat transfer over one engine cycle. 107 As the amount of heat transfer out of the residual gases declines with higher engine speed, the temperature of the residuals increases. Hotter residual gases lead to higher cylinder temperatures during compression, and shorter ignition delays. Figure 4.20 shows the estimated cylinder temperature (WAVE) at 30 CAD BTC versus engine speed for the same operating conditions. Cylinder temperature before combustion is approximately 100 degrees higher at 2500 RPM compared to 1000 RPM. High Load Limit: RPM & Pre-Combustion T Phi = 1, Intake Air T 30C, WOT 0 910 F- 890 o 870 850+ 830 4 810 790+ *S 770 1 1 1 Q 7501 500 1000 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.20 Cylinder temperature at 30 CAD BTC (pre-combustion) versus engine speed. The cylinder temperatures shown in figure 4.20 don't by themselves explain why the residual fraction is higher at 2500 RPM. The higher residual fraction was necessary in order to achieve the temperature shown in figure 4.20. Figure 4.21 shows WAVE modeling results comparing cylinder temperature at 30 CAD BTC to residual fraction at 2500 RPM. Just as in the case of figure 3.5, combustion duration and phasing were held constant to simplify modeling. As residual fraction is increased from 40%, cylinder temperatures increase. A lower residual fraction would result in lower cylinder temperatures. Lower cylinder temperatures would increase ignition delay and slow combustion, and operation would not be possible. 108 WAVE Modeling: Cylinder T vs. Residuals 2500 RPM, Phi = 1 900 895 890 885 880 875 870 865 860 0- I 40 60 50 I' I 70 80 90 Residual Mass % Figure 4.21 WAVE modeling: Residual mass % versus cylinder temperature at 30 CAD BTC at 2500 RPM. WAVE Modeling: Cylinder T vs. Residuals 2500 RPM, Phi = 1 960 940 920 I-I900 -0@j - A Intake T = 375K m Intake T = 325K 880 * Intake T = 275K 860 840 820 40 I I I 50 60 70 80 90 Residual Mass % Figure 4.22 WAVE modeling: cylinder temperature at IVC versus residual mass fraction for a range of intake air temperature. The higher residual fractions necessary to operate at higher engine speeds also help to explain why intake air temperature has less of an effect at higher engine speeds. Figure 4.22 shows the cylinder temperature (WAVE modeling) versus the residual fraction for a 109 range of intake air temperatures. The absolute values of the temperatures may include some error, but the modeled temperatures can give an indication of the trends. As the residual fraction increases, the impact of changing intake air temperature is reduced. The difference between cylinder temperatures at high and low intake air temperatures becomes smaller. At higher engine speeds and larger residual fractions intake air has less of an impact on the high load limit. The lower load limit does not show as dramatic a reduction with engine speed as the high load limit. At the low load limit the same mechanisms play a role as at the high load limit: less time for combustion, less time heat transfer, and less time for gas exchange. At the low load limit, less time for gas exchange actually assists in low load operation. As in the case for the high load limit, shorter ignition delays require higher temperatures. The reduction in heat transfer results in significantly less heat transfer during the recompression event. The much hotter residuals allow stable HCCI operation with more dilute mixtures, and so lower load operation. 4.1.4 Operating Range and Equivalence Ratio 1500 RPM, WOT, 120 CAD Cams IVC 0- 62 ABC, EVC 65-130 BTC 5.554.5 - 4 Phi = 1 A 3.5 m Phi = 0.9 3 A Phi = 0.8 Z 2.5 m Phi = 0.7 2 ]a A 1.5 -ILA 1 -10 10 30 50 70 90 110 Intake Air Temperature (C) Figure 4.23 (same as figure 4.6). High and low load limits for a range of intake air temps and phi's at 1500 RPM. 110 Intake air temperature does have an impact on HCCI operation at lean equivalence ratios. At 1500 RPM, operation at an equivalence ratio of 0.7 is only possible at high intake air temperatures, and even at those higher intake air temperatures the high load limit is reduced. At an equivalence ratio of 0.8 the high load limit is reduced with an intake air temperature of -6C. However at the low load limit equivalence ratio seems to have little impact. Figures 4.24 & 4.25 show calculated burned gas temperature and cylinder temperature for a range of equivalence ratios versus residual fraction. As expected, the burned gas temperatures decrease with increasing residual fraction. Burned gas temperatures also decrease as the equivalence ratio is reduced. At a given pressure and residual mass fraction, a leaner equivalence ratio means less fuel in the cylinder. The cylinder temperatures for a constant equivalence ratio show the same trends as shown in figure 3.5. Cylinder temperatures are lower with low residual fractions because there is less hot trapped mass. Cylinder temperatures increase with increasing residual fraction and then decrease as falling burned gas temperatures and a lower ratio of specific heats contribute to lower temperatures. Lean operation, with more dilution, shows lower temperatures for a given equivalence ratio. Burned Gas T vs. Res Fraction C2H4, Tinitial 1000K, Pinitiai 1 bar 2700 2500 0E I. 2300 2100- & _A cc *NPhi= 1 Phi = 0.9 Phi = 0.8 G Phi = 0.7 1900 1700S15000.3 0.4 0.5 0.6 0.7 0.8 Residual Mass Fraction Figure 4.24 Chemkin modeling: burned gas temp versus residual fraction at a range of phi. 111 Cylinder Temp vs Residual Fraction 1500 RPM, Intake T = 300K E 0Dm 0 ME 810 800 790 780 . A 770 760 A 750 740 730 30 40 60 50 70 *Phi m Phi A Phi 0Phi =1 = 0.9 = 0.8 =0.7 80 Residual Mass % Figure 4.25 WAVE modeling: cylinder temperature at IVC versus residual fraction at a range of phi. While comparing calculated temperatures versus residual fraction gives some insight, the loads are not equal for the same residual fraction. For a given residual fraction, less fuel is injected at the leaner operating points. Load would be a better criterion by which to compare the temperatures. Burned Gas T vs. Fuel Mole Fraction C2H4 2700 2500 E 0 I-- 2300 CO 2100 0 0 * Phi = 1 A 2300*Ph 1900 m Phi = 0.9 A Phi =A = 0.8 0 Phi = 0.7 1700 1500 0.01 0.03 0.04 0.05 Fuel Mole Fraction 0.02 Figure 4.26 Chemkin modeling: burned gas temp versus fuel mole fraction over a range of equivalence ratios. 112 In figure 4.26 the same calculated temperatures are displayed versus fuel mole fraction within the cylinder (load correlates well with injected fuel mass). Across the range of equivalence ratios the relationship between fuel mole fraction and burned gas temperature is consistent. In figure 4.27, as the injected fuel mass increases the calculated cylinder temperature first increases slightly, and then decreases. The pattern is very similar to that seen in figure 4.25 (and figure 3.5), as there is a correlation between fuel mass and residual fraction. Even though the calculated burned gas temperatures are approximately the same for the same fuel mass, the cylinder temperatures for the lean operating points are lower. For a given fuel mass, the lean operating points have a smaller residual fraction. A smaller residual fraction allows more room fresh air. Fresh air has a lower temperature than residual gases, and so cylinder temperatures are lower. The difference between the stoichiometric cylinder temperatures and the leanest cylinder temperatures are smallest at low load and largest at high load. This helps explain why there is very little difference in the low load limit at different equivalence ratios. Conversely figure 4.27 also helps explain why the range in the high load limits for different equivalence ratios is larger than the range of low load limits. The horizontal line on figure 4.27 shows a hypothetical combustion limit (the actual combustion limit is a function of temperature over the entire compression stroke). If the cylinder temperature falls below a certain point, the cylinder gases will not reach a high enough temperature during compression to ignite. At the low end the different equivalence ratios hit the combustion limit at a similar fueling rate (load). At the higher end the points of contact with the combustion limit have a much wider spread. Some of the equivalence ratios are not limited by low temperatures, but are instead limited by knock. The vertical line shows a hypothetical knock limit. In section 4.1.2, it was argued that the knock limit is largely a function of the concentration of reactants, so a knock limit at a constant fueling rate is reasonable. In this hypothetical case, stoichiometric operation and operation at an equivalence ratio of 0.9 are limited by engine knock, and so reach the same high load limit. Leaner equivalence ratios are limited by the misfire limit, and so high load 113 operation is more limited. While the hypothetical combustion and knock limits are simplifications of the real physical limits, this explanation is consistent with the experimental results. Cylinder Temp vs Inducted Fuel Mass 1500 RPM, Intake T = 300K 810 800 0 790 - + 0- 780 S770 -A A A A A7 A 0 df @) 750 740 730 ko +Phi = 1 m Phi = 0.9 IA Phi = 0.8 e Phi = 0.7 _ 5 7 9 11 13 15 17 19 21 Fuel/Cycle (mg) Figure 4.27 WAVE modeling: cylinder temperature at IVC versus inducted fuel over a range of equivalence ratios. 114 4.2 Intake Air Humidity 4.2.1 Effect of humidity at a point Intake air humidity can vary from very low levels in very dry or very cold climates, to very high levels in hot humid climates. To test the impact of humidity on HCCI engine operation, testing was carried out over a range of humidity ranging from 3g/m3 to 30 g/m3 (10% - 100% relative humidity at 30C). The humidity was controlled by adding moisture to the intake air stream when necessary, reducing the temperature of air to a desired dew point, and then raising temperature to a constant temperature of 35C. Data was collected at several different fuel equivalence ratios at 1500 RPM. Although engines do operate at lower and higher air humidity, this range covers humidity under most conditions and is wide enough to evaluate the impact of humidity on HCCI operation. Intake Air Fuel-Air %RH 30C Equivalence 8 1.0 30 0.9 70 0.8 100 0.7 Table 4.2 For each equivalence ratio, the range of air humidity was tested at 1500 RPM. Figure 4.28 shows NIMEP versus humidity in the intake air for several equivalence ratios with cam timing held constant. As the humidity increases, some of the oxygen and nitrogen in the air is displaced by water vapor. With increasing humidity there is little impact on the load at stoichiometric operation and at an equivalence ratio of 0.9. At an equivalence ratio of 0.8, the load decreases with increasing humidity. 115 Load vs Humidity 1500 RPM; WOT; Tin = 35 'C (EVC 85 BTC; IVC 41 ABC) 3.5 . 3.3 G 3.1 i 2.9 a. . . - * * *Phi = 1 *Phi = 0.9 A Phi = 0.8 A A 2.7 A 2.5 A I 0 10 20 30 40 Ambient Humidity [g/mJ Figure 4.28 NIMEP vs. intake air humidity. The inducted mass of dry air and fuel versus ambient humidity is shown in figure 4.29. Inducted Mass vs Humidity 1500 RPM; WOT; Tin = 35 OC (EVC 85 BTC; IVC 41 ABC) 228 E) 226 224 V) 222 * Phi = 1 a Phi = 0.9 A Phi = 0.8 Ua 220 218 216 A 214 0 20 4) Ambient Humidity [g/m] Figure 4.29 Inducted mass versus intake air humidity. As the humidity increases, combustion phasing is retarded at all three equivalence ratios (figure 4.30). Over the range of humidity, the combustion phasing varies by 4-5 CAD. 116 Combustion Phasing 1500 RPM; WOT; T in = 35 *C (EVC 85 BTC; IVC 41 A 9- 7o *Phi=1 m Phi = 0.9 A 5 Phi = 0.8 gA 03 1 0 10 20 30 40 Ambient Humidity [g/nf ] Figure 4.30 Combustion phasing versus humidity and equivalence ratio. As the humidity of the intake air increases, the oxygen and nitrogen are displaced by water vapor. At constant $, the fuel per cycle also decreases. The added water vapor raises the specific heat of the mixture, and the higher specific heat has a small impact on the temperature of the mixture during the compression stroke. The impact of the change in composition and the reduction in fuel mass does however have a significant impact on the calculated burned gas temperature (figure 4.31). For this calculation, a constant residual mass fraction of 50% and an initial temperature of 1000K were assumed. As humidity increases and fuel and oxygen are displaced by water vapor, there is less fuel to burn. Lower burned gas temperatures result in lower residual temperatures, and lower cylinder temperatures. WAVE modeling results show falling cylinder temperatures as humidity increases in figure 4.32. The lower cylinder temperatures contribute to later combustion phasing. Higher concentrations of water vapor (a combustion product) may also delay the combustion reactions. The later combustion phasing affects maximum rates of pressure rise, efficiency, and in some cases load. In the case of stoichiometric operation in figure 4.28, the reduction in fueling (fuel vapor displaced by water vapor) is offset by the more favorable combustion phasing, and the NIMEP remains roughly constant. 117 Burned Gas T vs. Intake Air Humidity 50% Residuals, C2H4 2350 2300 Phi = 1 0-0 2250+ 2200 ml Phi = 0.9 a 2150 V 2 100 A Phi = 0.8 A 2050 A Phi = 0.7 2000 4 1950 1900 60 20 40 0 Ambient Humidity (gm/m3) Figure 4.31 Chemkin: burned gas temperature versus ambient humidity for a range of equivalence ratios. Humidity vs. Cylinder Temp 1500 RPM 840 830 E10 . 820 4) 810 *Phi = 1 = 0.9 UPhi A 0 800 790 A Phi = 0.8 A 780 I 0 20 1 40 * Phi = 0.7 60 Ambient Humidity (g/m3) Figure 4.32 WAVE modeling: cylinder temperature versus ambient humidity. 118 Burned Gas T vs. Intake Air Humidity 50% Residuals, C2H4 23507 0c10 C 0) I- 23001 225011 22002150-LE 21002050- *Ph= h= O Phi = 0. 9 A APi=0 APi=0 8 0 Phi = 0. 7 2000 195019001 0 20 60 40 Ambient Humidity (gm/m3) Figure 4.31 Chemkin: burned gas temperature versus ambient humidity for a range of equivalence ratios. Humidity vs. Cylinder Temp 1500 RPM 840 830 820 D +Phi = 1 Phi = 0.9 A Phi = 0.8 *Phi = 0.7 810 A 800 ) A 790 780 0 20 40 60 Ambient Humidity (g/m3) Figure 4.32 WAVE modeling: cylinder temperature versus ambient humidity. 118 in lower residual fractions than stoichiometric operation. These lower residual fractions result in lower cylinder temperatures and later combustion phasing. Because of the higher proportion of fresh air during lean operation, humidity has a larger impact. With an equivalence ratio of 0.8, the combustion phasing becomes so late that combustion becomes unstable. Cam timing cannot compensate for the late combustion phasing, and so the load is limited. Figure 4.34 shows the cam timing is constant at the high load limit for an equivalence ratio of 0.8. At each humidity level, the cam timing is adjusted to advance combustion phasing as much as possible. High Load Limit 1500 RPM; WOT; TI, = 35 cC 45 - 40 -A A A A S35 <C 30, o 25 - _A +*Phi= 1 Phi =0.9 Phi = 0.8 20 15 - 10- 0 40 30 20 10 Ambient Humidity [g/m 3] Figure 4.34 IVC versus humidity at the high load limit for different equivalence ratios. Humidity has little impact on the low load limit, except at very high humidity and in the case of very lean operation. Near the low load limit, combustion is limited largely by the amount of energy amount of energy in the fuel. 120 Low Load Limit & Humidity 1500 RPM, Intake Air Temp 35C, WOT 2.2 2 1.8 .a 0. 1.6 1. - Phi = 1 Phi = 0.9 1.2 A Phi = 0.8 1 0.8 A6 0 30 20 10 Ambient Humidity (g/mA3) 40 Figure 4.35 Low load limit versus humidity. Figure 4.36 shows the impact of humidity on the lean operating limit for a constant operation condition. At each humidity level the fueling rate was reduced until the engine misfired. The increasing humidity increases the specific heat of the mixture and reduces the burned gas temperature of the mixture. As the humidity increases, more fuel is required to ensure that the temperature within the cylinder is high enough to ensure stable combustion. Humidity vs Minimum Phi 1500 RPM; WOT; Ti, = 35 'C (EVC 85 BTC; IVC 41 ABC) 0.8 .2 c a C 0.78 0.76 0.740.72 - 0.7 ' 0.68 - * w 0.66 0.64- 0 10 20 30 40 Ambient Humidity [g/m 3J Figure 4.36 Humidity versus minimum phi. 121 4.3 Fuel Effects 4.3.1 Fuel Composition Range Fuel composition can vary from season to season and from region to region. The source of the oil as well as the details of the refining process can impact composition. Real world gasoline is not a pure substance, or even a mixture of a small number of components. Each gasoline usually consists of a mixture of hundreds of different hydrocarbons. While many studies have evaluated the impact of different pure hydrocarbons on HCCI combustion, few have evaluated how more realistic mixtures of hydrocarbons might impact HCCI combustion. A range of fuels was created to highlight the market variability in selected fuel properties. The fuel properties were selected based on two criteria: properties showing high variability in the market, and fuel properties the research literature suggested might have an impact on HCCI combustion. If a fuel property showed high variability but there was a clear consensus that this property was not important for HCCI combustion, the property was not selected. Conversely, if the property was believed to play a significant role in HCCI combustion but showed very little variability in the market, the property was not selected. Project sponsor BP provided a fuel survey to allow determination of which fuel properties showed high variability. The survey was carried out from 1999 to 2005 and contained approximately 27,000 different fuel samples from the US gasoline market. The properties selected for testing were: Reid Vapor Pressure (RVP), Research Octane Number (RON), aromatic content, olefin content, and ethanol content. Previous work suggests that RON, aromatic content, olefin content, and ethanol content all have some impact on HCCI operation. Kalghatgi's octane index is a function of RON & MON: RON - KS where S is octane sensitivity (RON-MON). Lower 01 correlates with earlier combustion phasing [13]. Shibata showed a relationship between RON and low temperature heat release (LTHR). LTHR has a strong impact on the main high temperature heat release (HTHR) [14]. Shibata also examined HCCI combustion in fuels 122 consisting of mixtures of 11 different pure hydrocarbons. Olefins and aromatics were found to retard combustion phasing, and iso-paraffins and n-paraffins advanced combustion (HTHR). Shibata does conclude that RON alone is not sufficient to characterize HCCI performance [15]. Oakley compared the combustion characteristics of 3 fuels with similar MON, but different RON and different compositions. The 3 fuels all showed similar performance. The test conditions were closer to MON conditions, perhaps explaining the lack of correlation with RON. Oakley also concluded that percent of aromatic, olefin, and paraffin had a small impact on HCCI operation. Oakley found pure ethanol allowed more dilute operation, but did not investigate blends of ethanol and gasoline [12]. Although there is no literature suggesting RVP has a significant impact, RVP was included to investigate possible impacts of fuel volatility. One additional fuel was created although it fell outside of the market variability: a fuel consisting mainly of large molecule hydrocarbons. The fuel is referred to as "Heavy". This fuel was included in the study even though the market fuel survey didn't include detailed information on fuel composition, because of a possible impact on HCCI combustion. Sulfur and MTBE content were both considered but rejected based on the assumption that neither substance will likely be a significant component in fuels in the near future. By explicitly varying the selected properties a number of other properties were also implicitly varied including octane sensitivity, MON, and paraffin content. The strategy for the creation of the test fuels was to create a base fuel, and perturb the properties to cover the range seen in the market. The fuels were created using commercial fuel streams from BP refineries, so the composition of the fuels is representative of actual market fuels. In the case of ethanol, the two test fuels were created by adding 10% ethanol by volume to the base fuel and to the high RON fuel. 123 5 Parameters Y 12 Fuels RON IRVP (psi) V .- 't Olefins Aromatics Ethanol MON 0.0% 84.1 Base fuel 92.2 7.0 11.0% 29.0% igh RVP 91.7 9.1 10.6% 27.8% 0.0% 83.3 eavy fuel Igh RON Low RON 91.5 98.2 87.7 6.8 6.6 6.4 10.8% 10.1% 0.9% 29.4% 30.0% 1.5% 0.0% 0.0% 0.0% 82.9 87.8 86.9 High Aromatic 92.5 7.2 9.2% 38.8% 0.0% 83.2 Low Aromatic ligh Olefin 91.9 92.7 Low Olefin 93.0 7.1 7.4 6.6 13.6% 18.9% 4.3% 14.3% 26.8% 32.3% 0.0% 0.0% 0.0% 84.5 82.5 84.8 Ext Low Arom & Ole 91.2 6.3 0.4% 0.8% 0.0% 89.8 91.9 8.2 9.8% 25.9% 10.0% 83.7 98.2 8.1 9.7% 26.2% 10.0% 86.7 Ethanol igh RON w/Eth Table 4.3 The table shows the 12 test fuels created for testing. Table 4.3 shows the 12 fuels created for testing. The values of each property are shown for the twelve fuels. The fuels designed to vary a particular parameter are highlighted in that parameter column. For example the High RON & Low RON fuels were designed for a comparison of different RON values with the Base fuel RON. Although MON was not explicitly varied, the MON values for the test fuels are shown as well. Figures 4.37-4.41 show how the test fuels fall within the range of market variability for some of the fuel properties. For each property the high value of the test fuels is marked with an H, the low value is marked with an L, and value of the base test fuel is marked with a B. Although octane sensitivity was not explicitly varied as a property, the range and values for the test fuels are included here because of the suggested importance in the Kalghatgi. 124 L B H 4F L H B On 7 8 9 10 11 12 13 14 15 16 RVP (psi) 88 L 0 90 92 L 10 20 30 40 50 60 B -r 70 0 10 Aromatics (v%) B L 1 3 5 4 6 7 8 RON 96 98 9 20 Olefins (v%) 30 I 10 11 100 H H I I3 I II' 94 12 13 Octane Sensitivity Figures 4.37-4.41 Variability in RVP, RON, aromatics, olefins, and octane sensitivity from BP fuel survey [26]. 125 A range of data was collected for each fuel. The testing covered a range of speeds, the high load limit, low load limit, and reference points (Table 4.4). For the high and low load points at each speed, the cam timing was optimized to maximize the range for each fuel. For the operating range data the intake air temperature was held constant at 40C, and the equivalence ratio was fixed at 1.0. For each reference point, the intake cam timing, exhaust cam timing, equivalence ratio, and intake air temperature were fixed at several different preset values for each fuel. While the high and low load limits allow comparisons of operating range between fuels, the optimized cam timing for a high load or low load point could be quite different for different fuels. This makes a comparison of combustion characteristics for different fuels difficult. The reference points allow comparison of combustion characteristics with all inputs held constant. Reference Points - 1500 RPM, IVC 42 ABC Speed Operating Range 1000 High Load Limit EVC 98 BTC Intake T 40C, Phi = 0.7-1 1500 Low Load Limit EVC 108 BTC Intake T IOOC, Phi = 1 2000 2500 Table 4.4 Test matrix for fuel testing. 4.3.2 Effect of fuel composition on combustion Fuel energy is converted to thermal energy through the combustion process. For different fuels the detailslof the combustion process will be different. Each type of hydrocarbon will undergo different reactions during the oxidation process. Unfortunately there are detailed models of the reaction mechanisms for only a few pure hydrocarbon fuels. However from the perspective of a practical application of HCCI, the only differences that really matter are differences in overall combustion characteristics. From the perspective of engine operation, combustion phasing and combustion duration describe 126 the most important features of combustion: when combustion occurs, and how quickly combustion proceeds. Just as the combustion phasing is a useful parameter to describe the combustion process, it is also a useful parameter in comparing different fuels. An ideal comparison would keep all parameters constant and observe the combustion phasing for different fuels. However in the case of residual trapping, the previous combustion event influences the combustion phasing of the next combustion event. Combustion phasing impacts residual gas temperature, and residual gas temperature impacts combustion phasing. Because of the impact of the residuals, attempting to create identical cylinder conditions for different fuels is very challenging. Another approach is to instead keep all of the inputs constant, and observe the combustion phasing for each fuel. Figure 4.42 shows the combustion phasing for two reference points for each of the test fuels. For the points with EVC at 98 CAD BTC, the load is approximately 3 bar NIMEP. For the points with EVC at 108 CAD BTC, the load is approximately 2.5 bar NIMEP. For most of the fuels, the combustion timing is quite similar: the CA50s for 10 of the 12 fuels are within approximately 2 CAD. Two of the fuels, the high olefin fuel and the extremely low aromatic and olefin fuel are significantly different. The combustion phasing for the extremely low aromatic and olefin fuel was significantly later than most of the fuels, suggesting less tendency to auto-ignite. The combustion phasing for the high olefin fuel was earlier than most fuels, indicating a greater tendency to auto-ignite at these conditions. 127 Combustion Phasing: Reference Points Phi = 1, 1500 RPM, WOT, Intake Air T 40C 5 Low RON High Olef Heavy Ex Low A&O High Arom High RON Eth Base High RON+Eth Low Arom E Low Olef 0 ULR Winter + * o A A * X * * A A 4 3 U 2 0 1 0O 0 U. +x X RA - eX -1 -2 -3 100 95 110 105 115 EVC (CAD BTC) Figure 4.42 CA50 at two reference points for each test fuel. Combustion phasing is first considered for the selected fuel properties. In figures 4.434.47, the combustion phasing of four reference points is plotted for the base fuel, the high property value, and the low property value. In the figures each reference point uses the same symbol, and the fuel name is displayed next to the four reference points. Point EVC (CAD BTC) Intake Air T (degree C) IVC (CAD ABC) Phi Speed (RPM) 1 98 40 42 1.0 1500 2 108 40 42 1.0 1500 3 98 100 42 1.0 1500 4 108 100 42 1.0 1500 Table 4.5 Reference points. 128 The first property considered is RON. RON measures the tendency of a fuel to autoignite under certain conditions: SI operation at 600 RPM, intake air temperature 65.6C (150F), and coolant temperature of IOOC. As expected, the reference points with the higher temperature intake air show more advanced combustion phasing than the points with the lower temperature intake air. The variation in intake air temperature seems to have a larger impact on combustion phasing than the variation in exhaust cam timing for these operating conditions. A good correlation was expected between RON and the ease of ignition in HCCI combustion. However figure 4.43 shows almost no difference in combustion phasing for RON values ranging from 88 to 98. RON: Ref Points & CA50 1500 RPM, Phi = 1, IVC 42 CAD ABC 100.0 *High RON & A 98.0 * EVC 98 BTC, In T 40C 96.0 z * EVC 108 BTC, In T 40C 94.0 92.0 ase A A EVC 98 BTC, In T 1OC 90.0 * EVC 108 BTC, In T 100C 88.0 86.0 Low RON I I -6 -4 -2 0 2 4 i 6 CA50 (CAD ATC) Figure 4.43 RON versus CA50 at 4 reference points. Reid Vapor Pressure (RVP) measures the volatility of a fuel. In cold weather, a fuel with a higher RVP is desirable to ensure sufficient fuel vaporization for reliable cold starts. Adding a small amount (a few percent) of a more volatile fuel like butane raises the RVP of a fuel. Because the more volatile components of the fuel are only a small percentage of the fuel mixture, RVP was not expected to have a large impact on HCCI combustion. The experimental results do not show a clear correlation between RVP and CA50. There is no consistent difference between the high RVP reference points and the base fuel reference points. 129 RVP: Ref Points & CA50 1500 RPM, Phi = 1, IVC 42 CAD ABC 10.0 ULR Winter 9.0 Q EVC 98 BTC, In T 40C A* 8.0 C. * EVC 108 BTC, In T 40C 7.0 A EVC 98 BTC, In T 100C 6.0 Ex Low Aro & Ole m EVC 108 BTC, In T 100C 5.0 4.0 -6 -4 -2 0 2 4 6 CA50 (CAD ATC) Figure 4.44 RVP versus CA50 at four reference points. Aromatics: Ref. Points & CA50 1500 RPM, Phi =1, IVC 42 CAD ABC a 45.0%E = -40.0% High Aro A o 35.0% 30.0%- BEVC Bs EVC 98 BTC, In T 40C 108 BTC, In T 40C 25.0% AEVC98BTC, InTl1OC 20.0% 1 10.0% E 5.0% L .0 1 Aro1& 01giA 41 0.0% -1Ex Low 0 -2 -4 -6 A EVC 18 BTC, In T 1OOC 2 4 6 CA50 (CAD ATC) Figure 4.45 Aromatic content versus CA50 at four reference points. 130 RON and RVP are both properties that describe the physical behavior of a fuel under certain conditions. Aromatic content and the other properties are measures of the composition of the fuel. Increasing aromatic content was expected to reduce the tendency of the fuel to auto-ignite. However figure 4.45 shows that increasing aromatic content does not seem to correlate with later ignition. If there is any correlation, increasing aromatic content seems to in fact correlate with earlier ignition. Like aromatic content, olefin content was expected to reduce the tendency of the fuel to ignite. Like aromatic content, olefin content has been shown to increase RON. Shibata's results suggest that increasing olefin content leads to later combustion. However figure 4.46 shows the opposite: increasing olefin content seems to correlate with earlier combustion phasing. Olefins: Ref. Points & CA50 1500 RPM, Phi= 1, IVC 42 CAD ABC 20.0% 208.0% A S18.0%- e * E 5 16.0% 0 14.0% 12.0% 10.0% =0 8.0%- High Ol * EVC 98 BTC, In T 40C *, A A EVC 98 BTC, In T 100C 6.0%- 40% o * EVC 108 BTC, In T 40C Base LowOl AE EVC 108 BTC, In T 100C 2.0%0.0% I -6 -4 -2 4 f , 0 2 ExLow Aro&Ol 0 4 6 CA50 (CAD ATC) Figure 4.46 Olefin content versus CA50 at four reference points. The final fuel property that was explicitly varied is ethanol content. Ethanol has increasingly been used as a fuel oxygenate in recent years. Ten percent ethanol by 131 Rf Ponts CAS Ethanl: volume was added to two fuels, the base fuel and the high RON fuel. Figure 4.47 shows that ethanol appears to have little impact on combustion phasing for the reference points. Ethanol: Ref Points &CA50 1500 RPM, Phi = 1, IVC 42 ABC 11.0% 0 E 9.0% -0 -I 7.0% 5.0% 0 C.) 0 0 w 3.0% 1.0% 00 A 0 -1.0% 1 -6 1 -4 -2 0 2 CA50 (CAD ATC) 4 6 * Base: EVC 98 BTC, In T 40C * Base: EVC 108 BTC, In T 40C A Base: EVC 98 BTC, In T 100C S Base: EVC 108 BTC, In T 100C O Hi RON: EVC 98 BTC, In T 40C * Hi RON: EVC 108 BTC, In T 40C A Hi RON: EVC 98 BTC, In T 100C o Hi RON: EVC 108 BTC, In T 100C Figure 4.47 Ethanol content versus CA50. 4.3.3 Operating Range and Fuel Type 4.3.3.1 High and Low Load Limits The high and low load operating points were found for each fuel across a range of engine speeds. The constraints in the operating range were described in section 3.5: operation had to be stable, and there could be no engine knock. As in the case of the intake air temperature and humidity tests, the cam timing was optimized to maximize the operating range. Figure 4.48 shows the operating ranges for the 12 test fuels. Each point shown is either a high load limit or a low load limit. 132 Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 6+ Low RON m High Olef 5 - 0 Heavy 5A 4 Ex Low A&O A High Arom *High RON X Eth e Base :*: High RON+Eth A 0.3 IL A 2 1 A Low Arom Low Olef 0 ULR Winter -o 0 I 500 1000 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.48 Operating range versus engine speed for the 12 test fuels. Although there is some difference in the load limits, the performance of all of the fuels is similar. The largest difference is seen at the lowest engine speed. At 1000 RPM the difference in NIMEP in high load limits is approximately 0.8 bar. The difference in NIMEP in low load limits at 1000 RPM is 0.7 bar. 4.3.3.2 Operating Range and Cam Timing For each high and low load limit the cam timing was optimized. If high load operation for a particular fuel was limited by knock, the intake cam timing was advanced to retard combustion phasing as discussed in section 3.3. If a different fuel showed fewer tendencies to knock, the intake cam timing would not need to be advanced as far. Therefore the cam timing can give some insight into the performance of the different 133 fuels. Figure 4.49 shows the intake and exhaust cam timing for each of the high and low load limits from figure 4.48. Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 60 - 50 - a I I X ) 140 A A X X EHigh Olef Heavy A A Ex Low A&O A e p+*High 30 -G (>Low RON Low Load Limits High Load Limits A High Arom RON Q X Eth U eBase I XA X High RON+Eth 20 - ALowArom U 10 AA I U 0 50 70 Low Olef 0 ULR Winter I 90 110 130 150 EVC (CAD BTC) Figure 4.49 Intake and exhaust cam timing at the limits of operation. Figure 4.50 shows the cam timing at the high load limit. For a particular fuel, each point represents a high load limit at a different engine speed. At the high load limit the tendency to knock is offset by advancing the intake cam to retard combustion phasing, and by retarding the exhaust cam to compensate for the intake cam flow restriction. Fuels with a greater tendency to knock at the high load limit should show earlier intake and later exhaust cam timing. Fuels with fewer tendencies to knock should show later intake and earlier exhaust cam timing. According to this criteria the Low RON fuel and the Extremely Low Aromatic & Olefin fuel auto-ignite less easily at all four engine speeds. The cam timing for the High Olefin fuel suggests that this fuel auto-ignites more easily. 134 Operating Range: High Load Limits Phi = 1, WOT, Intake Air T 40C 60 Less Knock High Olef A 50 40A X 0 Heavy Ex Low A&O A High Arom X A0 A * High RON 0 ~30 X Eth OBase High RON+Eth A Low Arom & Low Olef 0 ULR Winter AN 3 X > Low RON A 2K A0 10 AmA 0 * &A SMore Knock 50 60 70 80 90 100 EVC (CAD BTC) Figure 4.50 EVC versus IVC for the high load limits. The intake cam timing at the high load limit is plotted versus engine speed in figure 4.51. Generally the intake cam timing is less advanced at higher engine speeds. The reduced volumetric efficiency restricts flow into the cylinder, so it is not necessary to advance intake cam timing as much at higher engine speeds. At 2000 RPM, the intake cam timing is later than one might predict from the overall trend. The air flow at 2000 RPM is especially restricted by manifold dynamics, and the intake cam timing reflects this (see Appendix E). The High Olefin and High Aromatic fuels generally show more intake cam advancement. More advancement suggests a greater tendency to knock at the high load limit for these fuels. The Low RON and Extremely Low Aromatic & Olefin fuels show less intake cam advancement across the speed range. These fuels appear to have fewer tendencies to knock at the high load limit. At 1000 RPM, the Extremely Low Aromatic and Olefin fuel was not able to approach the knock limit when unstable combustion and 135 misfire occurred (average maximum rate of pressure rise of - 3 MPa/ms instead of 5 MPa/ms). Cam phasing could not advance combustion timing. Operating Range: High Load Limits Phi = 1, WOT, Intake Air T 40C 60 < Low RON High Olef 0 Heavy A Ex Low A&O A High Arom * High RON A* 50 4M043 30 hX o Eth e Base XK High RON+Eth 20 - A Low Arom m Low Olef ULR Winter 10 o U0 0 500 1000 1500 2000 2500 3000 Engine Speed (RPM) Figure 4.51 IVC versus engine speed for the high load limits. The cam timing at the low load limits for each fuel is shown in figure 4.52. The range of intake cam timing is much smaller at the low load limit than at the high load limit. As discussed in section 3.6.2, the low load limit is not very sensitive to intake cam timing. The range of exhaust cam timing at the low load limits is very similar to the range of exhaust cam timing at the high load limit. Unlike at the high load limit, the cam timing of one fuel does not immediately appear different from the other fuels. 136 W --- ZKZ , -* ---- ---- - Operating Range: Low Load Limits Phi = 1, WOT, Intake Air T 40C 60 <> Low RON High Olef o Heavy A Ex Low A&O A High Arom 50 - 4 X X A 40 O High RON < '(30 Eth 3X 0 Base > 20 - K High RON+Eth A Low Arom Earlier EVC, More Residual, Lower Load 10 a Low Olef o ULR Winter 0 90 110 100 120 130 140 EVC (CAD BTC) Figure 4.52 IVC versus EVC for the low load limits for each test fuel. Since there is little difference in intake cam timing at the low load limit, the exhaust cam timing versus engine speed is instead examined in figure 4.53. The range of exhaust cam timing varies from 8 CAD at 1000 RPM to 6 CAD at 2500 RPM. As engine speed increases, the falling volumetric efficiency means that less exhaust cam advancement is necessary. As noted earlier, there is a deviation from this tendency at 2000 RPM. Because of the restricted flow at 2000 RPM, not as much exhaust cam advancement is necessary at that engine speed. 137 Operating Range: Low Load Limits Phi = 1, WOT, Intake Air T 40C 3000 <> Low RON High Olef Heavy 2500 - 2 0 4) 4) C,) 4 -- A Ex Low A&O A High Arom * High RON X Eth 0 Base High RON+Eth A Low Arom o Low Olef o ULR Winter 2000 - 1500 - B)KX C1000 500 90 100 110 120 130 140 EVC (CAD BTC) Figure 4.53 EVC versus engine speed at the low load limits for each fuel. Across the range of speeds the exhaust cam timing for the High Olefin fuel tends to be more advanced than the other fuels (129 CAD BTC at 1000 RPM). This would suggest that the High Olefin fuel tends to ignite more easily than the other fuels, and allows stable combustion with a more dilute mixture (more trapped residuals). Less definitively, the Low Aromatic, Low Olefin, and Extremely Low Aromatic Olefin & Aromatic tend to show less EVC advancement. Less EVC advancement indicates fewer tendencies to ignite. 4.3.3.3 High Load Limit and Fuel Composition Examination of the cam timing at the high and low load limits showed different behavior for different fuels at the limits of the operating range. The next sections look in detail at the high and low load limits, and the impact of fuel composition on those limits. 138 Just as different fuel properties have an impact on combustion phasing at the reference points, fuel properties do have an impact on the high load limits. If a fuel is more likely to auto ignite, the fuel will likely show earlier combustion phasing and a greater tendency to knock. A greater tendency to knock can be compensated for with more advanced intake cam timing. Intake cam advancement reduces pressure at IVC, and results in a lower high load limit. The largest range in high load limits is at lower speeds, so initial attention is focused there. At 1000 RPM, the Extremely Low Aromatic and Olefin fuel was limited by unstable combustion, and was not able to approach the high load limit. If combustion could have been advanced with cam timing or higher intake air temperature, the high load limit could have been extended. Since high load operation for this fuel was limited by cam timing and not by combustion behavior, the high load limit at 1500 RPM is used instead as a basis for comparison of the fuel properties. Figure 4.54 shows the high load limits at 1500 RPM versus the aromatic content for each fuel. The fuels created to vary aromatic content are circled (High Aromatic, Base, Low Aromatic, Extremely Low Aromatic & Olefin). Most of the other fuels have a very similar aromatic content of approximately 30%. The one exception is the Low RON fuel with very low aromatics. As aromatic content is increased, the maximum load decreases. This decrease in high load operation is consistent with the combustion phasing results at the reference points. Both suggest a greater tendency for auto-ignition to occur as aromatic content is increased. The impact of olefin content at 1500 RPM also appears to be consistent with the trends seen in the combustion phasing at the reference points. Greater olefin content correlates with a greater tendency to auto-ignite, and seems to correlate with lower high load limits. Figure 4.55 shows the olefin content of all the fuels versus the high load limits at 1500 RPM. Again the fuels blended to vary olefin content are circled (High Olefin, Base, Low Olefin, Extremely Low Aromatic and Olefin). 139 High Load Limit at 1500 RPM 4.6 *Low RON n High Olef 0 Heavy Hig Ole 4.5 4.5 4.4 (L i 4.3 w ( A Ex Low A&O O eav Arom A High AEx Lw A& RON * High 4.4 4.2 X Eth 0 Base x High RON+Eth 4.3*HghRO A Low Aro o Low Olef o ULR Winter 4.1 4 3.9 0.0% 10.0% 20.0% 30.0% 40.0% 50.0% Aromatic Content Figure 4.54 Aromatic content versus the high load limit at 1500 RPM. High Load Limit at 1500 RPM 4 .6 4 .5(9 IL wj 4 4 .1 A 4 * Low RON m High Olef o Heavy A Ex Low A&O A High Arom * High RON X Eth * Base * High RON+Eth A Low Arom m Low Olef o ULR Winter 3. .9. 0.0% 5.0% 15.0% 10.0% Olefin Content 20.0% Figure 4.55 Olefin content versus high load limits at 1500 RPM. 140 As in the case of the aromatics, the trends at 1500 are counter to expectations: higher concentrations of olefins in the fuel seem to correlate with lower higher load limits. RON was expected to show some correlation with CA50, but no correlation was apparent. The high load limits do not appear to correlate with RON either. Figure 4.56 shows RON versus the high load limits at 1500 RPM. The fuels created to test the impact of RON number are circled (High RON, Low RON, and Base). Low RON achieves a higher load than either Base or High RON. The RON of the rest of the points (with the exception of High RON w/Ethanol fuel) are quite similar. Higher RON was expected to reduce the tendency of a fuel to auto-ignite, and to allow higher load operation. However the results do not support this, and show if anything, the opposite trend. Although the MON was not varied explicitly, the MON did vary as the other fuel properties were changed. Figure 4.57 shows the MON versus the high load limits at 1500 RPM. Interestingly there appears to be a much better correlation between the high load limits and MON than between the high load limits and RON. The conditions during the MON test (900 RPM and intake air temperature of 300F) are closer to the actual operation conditions than the RON test conditions. The impact of RVP and ethanol content on the high load limits was also examined, but there did not appear to be a correlation. 141 High Load Limit at 1500 RPM 4.6 <DLow RON * High Olef 0 Heavy A Ex Low A&O A High Arom * High RON X Eth * Base x High RON+Eth A Low Arom a Low Olef O ULR Winter 4.5 Q 4.4 4. A X_ 4. 2 4.2-x 4.1 3.9 - 90.0 85.0 95.0 100.0 RON Figure 4.56 RON versus high load limits at 1500 RPM. High Load Limit at 1500 RPM 4.6 < Low RON m High Olef 0 Heavy A Ex Low A&O A High Arom *High RON x Eth 4.5 4.4 L 4.3 - A 4.2 - * Base x High RON+Eth 4.1 - Low Arom AA m Low Olef 4 0 ULR Winter 3.9 82.0 84.0 86.0 88.0 90.0 MON Figure 4.57 MON versus high load limits at 1500 RPM. 142 The previous slides examined the impact of the fuel properties on the high load limit at 1500 RPM. Figures 4.58-4.60 compare the high load limits at higher speeds to the high load limit at 1000 RPM. The high load limits at 1000 and 1500 RPM show a good correlation. This indicates that the performance of the fuels at the high load limit is very similar at 1000 RPM and 1500 RPM. The Low RON and Extremely Low Aromatic and Olefin fuels achieve higher loads. As discussed, the high load operation of the Extremely Low Aromatic and Olefin fuel is limited at 1000 RPM by the inability of cam timing to advance combustion. At 1500 RPM, Extremely Low Aromatic and Olefin is limited by knock and stability, like the other fuels, and the load achieved is higher in comparison to the other fuels. The limits for High Olefin, High Aromatic, and Heavy fuels achieve among the lowest of the high load limits. Operating Range: High Load Limit Phi = 1, WOT, Intake Air T 40C Low RON High Olef Heavy Ex Low A&O A High Arom * High RON X Eth * Base K High RON+Eth A Low Arom * Low Olef o ULR Winter * n * A 4.6 4.5 4.4 4.3 0 A X 4.2 4.1 A 4 3.9 3.8 4.3 4.5 4.7 4.9 5.1 High Load Limit 1000 RPM Figure 4.58 High load limits at 1000 RPM versus the high load limits at 1500 RPM. A comparison of the high load limits at 1000 RPM with the high load limits at 2000 RPM does not show a very good correlation (figure 4.59). 143 Operating Range: High Load Limit Phi = 1, WOT, Intake Air T 40C 3.8 0 Low RON m High Olef o Heavy A Ex Low A&O A High Arom * High RON X Eth * Base * High RON+Eth A Low Arom m Low Olef o ULR Winter 3.7 0-. 3.6 2 ex 3.5 o 3.4 3.3 0 -j ''1 ' = 3.2 3.1 q 4.3 4.5 4.7 4.9 5.1 High Load Limit 1000 RPM Figure 4.59 High load limits at 1000 RPM versus the high load limits at 2000 RPM. Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 3 *Low RON m High Olef o Heavy A Ex Low A&O A High Arom * High RON X Eth * Base * High RON+Eth A Low Arom o Low Olef o ULR Winter 2.9 2.8 2.7 - p E 2.6 :3 A 2.5 X 0 A .c 2.4 :F 2.3 p 2.2 4.3 4.5 4.7 4.9 5.1 High Load Limit 1000 RPM I Figure 4.60 High load limits at 1000 RPM versus the high load limits at 2500 RPM. 144 As compared to the high load limits at 1000 RPM and 1500 RPM, the high load limit at 2000 RPM does not show the same range. The high load limit of the Low RON and Extremely Low Aromatic and Olefin fuels have dropped significantly compared to the other fuels. The high load limits for the Heavy, High Aromatic, and High Olefin fuels have increased compared to the other fuels. The dynamics of the flow complicate the picture at 2000 RPM (Appendix E). The comparison of the high load limits at 2500 RPM and 1000 RPM shows similar trends to the comparison at 2000 and 1000 RPM. The high load limits for the Extremely Low Aromatic and Low RON fuel are lower compared to the other fuels than at lower speeds. The high load limits for the High Aromatic, High Olefin, and Heavy fuel are higher compared to the other fuels than at lower speeds. The trends at 2500 RPM are closer to expectations. High RON, High Aromatic, and High Olefin fuels achieve higher loads than Low RON, Low Aromatic, and Low Olefin fuels respectively. The lack of correlation between the high load limits at low speeds and high speeds suggests that different mechanisms limit operation at different speeds. As discussed in section 4.1.3, high load operation at 2500 RPM appears to be limited by the time for combustion to occur. A shorter time for combustion requires higher temperatures to speed combustion. Higher residual fractions are necessary to achieve higher cylinder temperatures. A fuel with a greater tendency to auto-ignite should allow operation with lower cylinder temperatures and therefore lower residual fractions. Lower residual fractions result in higher loads. The High Olefin fuel demonstrated earlier combustion phasing at the reference points. The tendency of the High Olefin fuel to auto-ignite more easily is consistent with reduced high load operation at 1000 RPM. At 1000 RPM, the main mechanism limiting operation is the concentration of reactants. Because High Olefin fuel ignites more easily, it is 145 necessary to reduce the pressure at IVC, and the limiting concentration is reached with a less dense mixture. At 2500 RPM, high load operation is limited by the time for combustion to occur. In order to reduce the ignition delay, higher temperatures are necessary. Higher temperatures are achieved by higher residual fractions. High Olefin fuel seems to ignite more easily than the other fuels. High Olefin fuel can ignite with slightly lower temperature mixtures, and so can ignite with slightly lower residual fractions, and achieves a higher load. Table 4.6 shows estimated residual fractions and measured NIMEP for the High Olefin fuel and for the Extremely Low Aromatic & Olefin Fuel. Residual fractions were estimated using WAVE. Res % @ NIMEP @ Res % @ NIMEP 1000 RPM 1000 RPM 2500 RPM @ 2500 RPM High Olefin 40% 4.4 44% 2.8 Ex. Low A&O 41% 4.7 54% 2.6 Table 4.6 Residual fraction and load at the high load limits for High Olefin and Extremely Low Aromatic & Olefin fuels. 4.3.3.4 Low Load Limit and Fuel Composition A fuel with a greater tendency to auto-ignite should allow lower load operation. More easily ignited mixtures will be able to maintain stable combustion with slightly cooler, more dilute mixtures. More dilute mixtures means lower low load limits. The reference point data showed that higher olefin content seemed to correlate with earlier combustion phasing. This suggests that higher olefin content should correlate with lower minimum load operation. The largest range in the low load limits occurs at 1000 RPM, so the examination of the impact of fuel properties starts with that speed. Figure 4.61 shows the aromatic content versus the low load limits at 1000 RPM. 146 Low Load Limit at 1000 RPM 2.5 * Low RON m High Olef O Heavy A Ex Low A&O A High Arom * High RON X Eth * Base X High RON+Eth A Low Arom * Low Olef O ULR Winter 2.3 2.2 CL Lu 2.1 0 2 1.9 I I 1.8 1.7 1.6 1.5 20.0% 30.0% 40.0% 10.0% 0.0% 50.0% Aromatic Content Figure 4.61 Aromatic content versus the low load limits at 1000 RPM. Low Load Limit at 1000 RPM 2.5 2.4 2.3 2.2 (L 2.1 z 2 p A 1.9 1.8 A 1.7 1.6 * Low RON m High Olef 0 Heavy A Ex Low A&O A High Arom * High RON X Eth * Base * High RON+Eth A Low Arom o Low Olef o ULR Winter 1.5 0 .0% 5.0% 10.0% 15.0% 20.0% Olefin Content Figure 4.62 Olefin content versus the low load limits at 1000 RPM. 147 The fuels created to vary aromatic content have been circled (High Aromatic, Base, Low Aromatic, Extremely Low Aromatic and Olefin). There does not appear to be a clear correlation between the aromatic content and the lowest achievable load at 1000 RPM. Olefin content on the other hand, does appear to correlate with the low load limits. As the reference point combustion phasing results suggested, higher olefin content seems to correlate with lower minimum load operation. Figure 4.62 plots olefin content versus the low load limits at 1000 RPM. High Olefin, Base, Low Olefin, and Extremely Low Olefin and Aromatic fuels are circled. In figure 4.63, RON is compared to the low load limits at 1000 RPM. High RON, Base, and Low RON are circled. Although the three points do line up nicely, it's not clear that RON has a significant impact on the low load limit for this engine. Low Load Limit at 1000 RPM 2.5 2.4A 2.3 Low RON m High Olef 2.2 0 Heavy . W 2.1 Z 0 A Ex Low A&O X A High Arom **High RON X Eth 2 1.9 1.8 X 0 Base X High RON+Eth A Low Arom o Low Olef 0 ULR Winter A 1.7 1.6 1.5 85.0 95.0 90.0 100.0 RON Figure 4.63 RON versus the low load limits at 1000 RPM. 148 Neither ethanol content nor RVP seemed to impact the low load limit. At the high load limit, the relative performance of the fuels appeared to change with engine speed. Figures 4.64-4.66 compare the low load limits at 1000 RPM with the low load limits at higher engine speeds. The low load limits at 1000 and 1500 RPM correlate pretty well, indicating that the fuel behavior is very similar at both speeds. Operating Range: Low Load Limit Phi = 1, WOT, Intake Air T 40C 2 0Low RON 1.9 * High Olef o Heavy 0 1.8 A Ex Low A&O 1.7 A X E 1.6 0 1.5 0 -j 1.4 A High Arom * High RON X x Eth e Base * A K High RON+Eth A Low Arom a Low Olef o ULR Winter 1.3 2.5 1.2 1.5 1.7 1.9 2.1 2.3 2.3 2.5 Low Load Limit 1000 RPM Figure 4.64 Low load limits at 1000 RPM versus the low load limits at 1500 RPM. At 2000 RPM the trends remain mostly the same (figure 4.65). The High Olefin fuel achieves the lowest load, and the Extremely Low Aromatic and Olefin fuel and ULR Winter fuels have the highest low load limit. The Low Aromatic fuel does exhibit significantly different behavior, with a comparatively higher low load limit at 2000 RPM than at 1000 RPM. 149 Operating Range: Low Load Limit Phi = 1, WOT, Intake Air T 40C 2 Low RON m High Olef o Heavy A Ex Low A&O A High Arom * High RON X Eth * Base )K High RON+Eth A Low Arom o Low Olef o ULR Winter 1.9 1.8 1.7 A 0. A 0 1.6 0 0i 1.5 sK 1.4 * + A 1.3 1.2 1.9 1.7 1.5 2.3 2.1 2.5 Low Load Limit 1000 RPM Figure 4.65 Low load limits at 1000 RPM versus low load limits at 2000 RPM. Operating Range: Low Load Limit Phi = 1, WOT, Intake Air T 40C 2 * Low RON 1.9 - n High Olef 1.8 0 Heavy A Ex Low A&O A High Arom *High RON W1.7 1.7 ~*1.6 X Eth 0A . 1.5 0 Base x High RON+Eth A Low Arom o Low Olef A -1.4 0 A * 1.3 - X 0 ULR Winter __ 1.2 1.5 1.7 1.9 2.1 2.3 2.5 Low Load Limit 1000 RPM Figure 4.66 Low load limits at 1000 RPM versus low load limits at 2500 RPM. 150 The difference between the low load limits is significantly smaller at 2500 RPM than at 1000 RPM (figure 4.66). The low load limits of all twelve fuels fall within a range of approximately 0.25 bar NIMEP. Ten of the low load limits fall within a range of approximately 0.15 bar NIMEP. Even though the differences are small, the differences appear to be consistent with trends seen at lower speeds. The High Olefin fuel is among the fuels with the lowest minimum loads. Extremely Low Aromatic and Olefin fuel and ULR Winter have the highest minimum loads. At the low load limit, the behavior of the fuels seems reasonable consistent across the range of speeds. This suggests that the same mechanisms limit low load operation at 1000 RPM, and limit low load operation at 2500 RPM. Although reference points were only taken at 1500 RPM, data was taken at 2500 with very similar inputs. Figure 4.67 shows combustion phasing at 2500 RPM for the olefin fuels. Cam timing is within +/- 1 degree on the exhaust, and within +/- 2 degree on the intake for all four points. The data shows that at 2500 RPM at this particular condition, like at lower speeds, higher olefin content seems to correlate with earlier combustion phasing. Olefin Content vs. CA50 2500 RPM, Phi=1 20.0% 1 High Olef 12.0% 0 0 OBase %Low Olefin 8.0% LowA&O 80AEx 4.0% 0.0% A 1 -3 -1 1 3 CA50 (CAD ATC) Figure 4.67 Olefin content versus CA50 at 2500 RPM. EVC is 103 CAD BTC and IVC 47 CAD ATC (+/- 2 CAD). 151 This condition is quite close to the low load limit (- 1.4 bar NIMEP), and is further evidence that the behavior of fuels is consistent at the low load limits across the range of speeds. 4.3.4 Impact of octane sensitivity Contrary to expectations, higher olefin and aromatic content in the test fuels seems to correlate with earlier combustion phasing and a greater tendency for auto-ignition to occur. However the fuels tested are very complex mixtures of many hydrocarbons, like real world gasoline, so attributing an effect to one property can be challenging. When the test fuels were created, one property was varied and an attempt was made to hold the other properties constant. With the olefin fuels, olefin content was varied and aromatic content, ethanol content, RVP, and RON were held constant. However by minimizing changes in the selected properties, other fuel properties did change. Of the properties not explicitly controlled, MON seems to correlate best with combustion behavior (figure 4.57). As previously noted, operating conditions for most of the testing were more similar to MON test conditions than RON test conditions. Figure 4.68 shows MON versus CA50 of the reference points for the olefin fuels. The reference points for each fuel are labeled. The correlation between higher MON and later combustion phasing is as good as the correlation between CA50 and olefin content. In other words, olefin content is changing consistently with MON. The Extremely Low Aromatic and Olefin fuel has the highest MON and shows the latest combustion phasing. High Olefin fuel has the lowest MON and shows the earliest combustion phasing. Although at first glance the olefin fuels seem to show behavior contrary to expectations, the lower MON values associated with the higher olefin content fuels are consistent with the observed greater tendency for auto-ignition. The test fuels with higher aromatic content also have lower MON values. 152 MON of Olefins: Ref. Points & CA50 1500 RPM, Phi= 1, IVC 42 CAD ABC 91.090.0- AL Ex. Low A&O 89.0 - * EVC 98 BTC, In T40C 88.0 * EVC 108 BTC, In T 40C z 87.00 2 86.0 A EVC 98 BTC, In T 100C 85.0 A 84.0 AU" G* * Low Olefin m EVC 108 BTC, In T 100C Base 83.0 A 82.0 - -6 U r -4 G High Olefin 1 i 1 10 2 -2 r 4 6 CA50 (CAD ATC) Figure 4.68 Olefin fuels: MON versus CA50 for the reference points. The relationship between olefin and aromatic content, and MON can be understood by examining more closely how the fuels were created. As olefin content was varied, an attempt was made to keep the other properties, including RON, constant. Shibata created a plot to show RON versus MON for a range of fuels (figure 4.69). Shibata identifies the fuels by groups: olefins, aromatics, and paraffins. Olefin and aromatics tend to have larger octane sensitivity than paraffins (the difference between RON and MON is larger). RON tends to be higher for olefin fuels. Paraffin fuels show wide range of RON and MON. If olefin content is increased, Shibata's chart suggests that MON will likely decrease. For a given RON, olefins have lower MON, so the MON of the mixture will be lower. The RON of most of the olefins and all of the aromatics appear to be higher than the base RON. If RON increases with increasing olefin, a low RON component will have to be added to the fuel. The lowest RON fuels are paraffins. Figure 4.69 shows that paraffins tend to have equal RON and MON. 153 Approximal ;location of Base RON' value 120~ 2110 2: Arom atts -~ Napi finsamd 0Ohfiis W 60 I 100 120 RO N Figure 4.69 MON versus RON for a variety of fuels - from Shibata et. al. [14]. Therefore increasing the content of a paraffin with a low RON, will at the same time be increasing the content of a fuel with low MON to the mixture. As olefin content increases, low RON paraffin content will increase, and MON will decrease. As MON decreases, the tendency of a fuel to auto-ignite increases. These same arguments help explain the trends seen with the aromatic fuels. As aromatic content is increased, low RON paraffin content increases as well to keep the RON constant. This has the effect of lowering the MON, and lowering MON changes combustion behavior. The very high paraffin fuels (Extremely Low Aromatic & Olefin fuel and Low RON fuel) consist of relatively high MON paraffins, and display less tendency to auto-ignite. The low paraffin content fuels (High Olefin and High Aromatic fuels) contained lower MON paraffins, and show a greater tendency to ignite. The results are consistent with the results of Oakley, whose fuels with similar MON showed similar behavior. The results are also consistent with Kalghatgi's work, which emphasized the role of octane sensitivity. 154 4.3.5 Other fuel components The experimental results show that fuel properties do have an impact on HCCI combustion. However analyzing results for multi-component real world gasoline is complex. Because of the way the test fuels were created, many components in the test fuels were varied simultaneously. For example in order keep RON, RVP, aromatic content and ethanol content constant as olefin content was varied, other components of the fuel had to be varied. Figure 4.70 shows the composition of the different olefin fuels. Olefin Fuels Components 100.0 90.0 80.0 -W 70.0 60.0 o C 50.0 0 40.0 0_ a) 30.0 0. 20.0 10.0 0.0 0. 0% - "" 0 0 - 5.0% b - - - . g * n-Paraffin o iso-Paraffin A Napthene * Polynapthene * Aromatics 10.0% 15.0% 20.0% Olefin Content Figure 4.70 Olefin fuels components. Most of the components change very little - a couple percent across the range of olefin fuels. Iso-paraffins however change significantly and also constitute the largest single component of the fuels. Linear regression analysis was performed to examine the impact of iso-paraffins and other components on combustion. For the purposes of the regression fuel components were lumped into 7 fuel factors: aromatics, olefins, ethanol, iso-paraffins, paraffins, napthene, and polynapthene. There are 12 fuels, and so for each operating condition there are only 12 data points. With 7 factors and 12 data points, generating meaningful 155 regression results can be challenging. The regression of fuel factors and data points did not give significant results for the low load limits or for the combustion phasing at the reference points. At the high load limit, regression suggested that napthene and polynapthene were significant. A regression of the high load limits with 2 factors, napthene content and polynapthene content, showed an R2 value of 0.90. Figure 4.71 plots the high load limits at 1500 RPM versus the predicted high load limit based on the regression analysis. Regression: High Load Limit & Napthenes 1500 RPM, Phi = 1, WOT, Intake Air T 40C 4.6 * Low RON 4.5 R =0.89 n High Olef A o Heavy 4.4 A Ex Low A&O 4.3 U A High Arom * High RON 4.2 A , X Eth 4.1 C * Base K High RON+Eth 00 4 A Low Arom 3.9 o Low Olef 3.8 3.8 4 4.2 4.4 4.6 NIMEP (bar) Figure 4.71 Regression: high load limits and napthenes. Both of the coefficients were negative indicating that higher napthene content correlates with lower high load limits. Figure 4.72 plots the high load limits at 1500 RPM versus the napthene content in the fuels. Even though napthenes constitute a small portion of the fuel, they do appear to have a significant impact. 156 High Load Limit vs. Napthene Phi = 1, WOT, Intake Air T 40C 4.6 Low RON * High Olef 4.5 4.4 ccC o Heavy AA A Ex Low A&O 4.3 A High Arom * High RON A E3 uJ 4.2 2 4.1 X X Eth * Base K High RON+Eth A 4 A Low Arom 3.9 I m Low Olef I 3.8 0.00 2.00 4.00 6.00 8.00 10.00 Napthene Content % Figure 4.72 Napthene content versus NIMEP at the high load limit. Polynapthene was also examined, but the amount of polynapthene in the fuels was only a few tenths of a percent. Polynapthene content also correlates reasonably well with napthene content, so it is impossible to separate the impact of one from the other. Regression of the high load limit data points with the distillation temperatures also gave good results. Figure 4.73 plots regression for the high load limits and the temperatures at the 10%, 30%, and 90% distillation levels. Interestingly, the boiling point of napthene is 81 degrees C which is in the range of the 10% and 30% temperature levels. This result supports conclusion that napthene plays an important role in combustion at the high load limit. 157 Regression: High Load Limit & Distillatio n Temps 10%, 30%, 90% 1500 RPM, Phi = 1, WOT, Intake Air T 40C 4.0 * Low RON * High Olef o Heavy A Ex Low A&O A High Arom * High RON x Eth * Base K High RON+Eth A Low Arom 4.5 4.4 ICU 4.3 4.2 4) 0 4.1 A IL 4 3.9 a Low Olef 3.8 3.8 4 4.2 4.4 4.6 NIMEP (bar) Figure 4.73 Regression: high load limits and distillation temperatures. 4.3.6 Differences in combustion The different fuels display different combustion behavior. At the low load limit the High Olefin fuel is able to achieve lower load operation than most of the fuels. At the high load limit the Extremely Low Aromatic and Olefin fuel is able to achieve higher load operation than most of the fuels. One possibility is that the difference in performance at the limits is a function of differing combustion phasing. The different performance of some of the fuels at the limits of operation is consistent with the trends seen at reference points: the High Olefin fuel displayed earlier combustion phasing, and the Extremely Low Aromatic and Olefin fuel showed later combustion phasing. Another possibility is that there are differences in the rates of heat release between the different fuels. A difference in the rate of heat release would impact the tendency of the fuel to knock, and so would impact operation at the limits. 158 However the rate of heat release is not only a function of the fuel chemistry, but also of the combustion phasing and the operating conditions. Later combustion phasing, lower temperatures, lower pressures, and more dilution all contribute to lower reaction rates and rates of heat release. In a residual trapping engine, isolating the impact of differences in fuel composition from the impact of differences in operating conditions is challenging. If the combustion phasing or load for two fuels is different, differences in the rates of heat release may be due to differences in combustion phasing or load, and not due to differences in fuel chemistry. Figure 4.74 plots the combustion phasing versus the average maximum rate of pressure rise at the low load limit for the different fuels. The maximum rate or pressure rise is indicative of the rate of heat release during combustion. Operating Range: Low Load Limits 1000 RPM, Phi = 1, WOT, Intake Air T 40C * Low RON * High Olef 2.5- 0 Heavy 2 -__ E 1.5 1P 16 c .5 0 1 2 3 4 5 A Ex Low A&O A High Arom * High RON X Eth e Base X High RON+Eth A Low Arom m Low Olef o ULR Winter CA50 (CAD ATC) Figure 4.74 High load limit versus combustion phasing. 159 The maximum rates of pressure rise for all of the fuels are similar. The combustion phasing is clustered within a range of approximately 2 CAD. Combustion appears to be very similar for the different fuels at the low load limit. Figure 4.75 also plots combustion phasing versus the maximum rate of pressure rise, but for 10 repeated low load limits with the same fuel. Although the differences in load of the repeats were very small (standard deviation of 0.04 bar), the differences in combustion phasing and maximum rate of pressure rise for the repeats are very similar to the differences seen with the different fuels. Therefore differences in the maximum rate of pressure rise seen with different fuels are likely due to experimental variability. The main sources of experimental variability include a lack of precision in control of cam phasing (limit of +/- 1 CAD), inaccuracies in measuring cylinder pressure, and inaccuracies in analyzing combustion phasing. The rates of heat release are likely very similar for the different fuels at the low load limits. Repeated Low Load Limits 1500 RPM, Phi = 1, WOT 2.5 2 1.5 12- x 0.5 0 1 2 3 4 5 CA50 Figure 4.75 Maximum rate of pressure rise versus CA50 for 10 repeated low load limits. At the high load limit, the combustion phasing is also clustered quite closely: the different fuels show a range of - 2 CAD (figure 4.76). The maximum rates of pressure rise show more variability than at the low load limit. However this variability is also very similar to the variability seen in repeated high load limits with the same fuel (figure 160 4.77). In the repeated high load limits there does appear to be some correlation between higher rates of pressure rise and earlier combustion phasing. Operating Range: High Load Limits 1500 RPM, Phi = 1, WOT, Intake Air T 40C 5.5 - Low RON m High Olef 5 - 0 Heavy A Ex Low A&O E A High Arom 4.5 - C. * High RON X Eth * Base X High RON+Eth A Low Arom 4 - X S3.5 3 m Low Olef 9 10 11 12 CA50 (CAD ATC) 13 o ULR Winter Figure 4.76 Maximum rate of pressure rise versus CA50 at the high load limit. Repeated High Load Limits 1500 RPM, Phi = 1, WOT 5.5 Cl) .E 5 + cc -. 4 X3.5 3 9 10 11 12 13 CA50 (CAD ATC) Figure 4.77 Maximum rate of pressure rise versus CA50 for repeated high load limits. 161 Some increase in the rate of pressure rise with earlier combustion phasing is fairly intuitive. The same approximate trend appears to be present in the plot of the different fuels. Again, the differences in the rate of pressure rise for the different fuels appear to be due to experimental variability. This suggests that there is no significant difference in the main heat release event for the different fuels. If one fuel had higher or lower rates of heat release, that would have been evident in different rates of pressure rise at the high load limit (where combustion phasing was the same). A closer look at the combustion event for three fuels supports idea that the heat release events for the different fuels are very similar. In figure 4.78, the percent of fuel burned is plotted versus CAD for the High Olefin, Low RON, and Extremely Low Aromatic and Olefin fuels. High Load Limit & Combustion 1500 RPM,Phi = 1 14 12- A A 010 < O m High Olefin 86 4A_AEx. LowRON Low & Olef 24"Arom 0- 0 10 20 30 40 50 60 % Burn Figure 4.78 Percent burn versus CAD. Combustion proceeds in very similar ways in all three cases. The time (in terms of CAD) between the 2% burn and 10% burn is very similar for all fuels, and the time between 10% burn and 50% is also very similar. The main heat release event for the three 162 different fuels is therefore very similar. Any differences in the performance of the three fuels can be attributed to differences in the initial combustion chemistry of the fuels. 4.3.7 Fuel composition and emissions Comparing the emissions of different fuels in HCCI with residual trapping is particularly challenging because of the difficulty of controlling cylinder conditions precisely, and because of the lack of direct control over combustion. Load, combustion phasing, and airfuel ratio all have significant impacts on emissions. Comparing the emissions of two fuels at different loads or different combustion phasing may not be a good comparison. The impact on emissions of combustion differences may outweigh the impact of differences between the fuels. Cam timing allows some control over combustion, but impacts the mixture concentrations and temperatures, and overall efficiency. Ideally the load, combustion phasing, and mixture composition would be held constant for a comparison of different fuels, but this is not possible with the present experimental setup. Another possible way to compare the emissions of the fuels would be to measure the emissions for different fuels for a fixed engine calibration. The advantage of this approach is that it is relatively simple to implement. The disadvantage is that combustion phasing and mixture concentration are not directly controlled. A difference in emissions might result from a difference in combustion phasing, instead of a difference in fuel composition. While developing an engine calibration strategy is outside of the scope of this project, data was collected with inputs held constant (reference points). Additionally, the strategy to reach the high and low load limits was consistent for all fuels. Emissions results were examined in a number of different ways. Emissions at the reference points as a function of fuel properties were evaluated. Emissions of different fuels as a function of load were plotted. And emissions at the high and low load limits were plotted. The fuels properties as tested did not appear to show significant impacts on emissions. All of the significant differences in emissions could be attributed to differences in load and combustion phasing. 163 The HC emissions results for olefins are presented here, and emissions of the other properties at the reference points are shown in Appendix D. Figure 4.79 shows the HC emissions versus olefin content for the four reference points. There are small but not significant differences between the emissions of the High Olefin, Base, and Low Olefin fuels. The Extremely Low Aromatic and Olefin fuel does show significantly higher emissions of HC. Fuel Properties: Olefins & HC Emissions * E*HighOl A 20.0% E 7 15.0%C > EVC 98 BTC, In T 40C .0 10.0% 0 S 5.0% _e 0.0% 0.005 AG A * EVC 108 BTC, In T 40C A EVC 98 BTC, In T 1OOC EC108 BTC, In T 1 OOC N*Base e+ 4 0.01 Low 01 * * 0.015 Ex Lo Aro & Ol 0.02 HC (g/g fuel) Figure 4.79. Emissions of HC versus olefin content for four reference points at 1500 RPM. The Extremely Low Aromatic and Olefin fuel is also the fuel with the latest combustion phasing. These points are the same points shown in figure 4.46. As previously discussed, later combustion phasing is associated with higher HC emissions (lower peak pressures). If the emissions of HC are considered across the entire load range at 1500 RPM, the results are very similar. Figure 4.80 shows HC emissions versus load for the High Olefin, Base, Low Olefin, and Extremely Low Aromatic and Olefin fuels. The emissions of HC are not significantly different for the High Olefin, Base, and Low Olefin fuels. The HC emissions for the Extremely Low Aromatic and Olefin fuel are higher across the load range. These higher emissions, however, can largely be attributed to the later combustion 164 phasing, lower peak pressures, and lower peak temperatures associated with this fuel. Figure 3.26 & 3.27 shows that retarding the CA50 from 2 CAD ATC to 4 CAD ATC can result in an approximately 20% increase in HC emissions. Olefins: HC Emissions 1500 RPM, Phi= 1 0.025 0 Base 0.02 A High Olefin a) "R 0.015 A Ex Low A&O a Low Olefin 0.01 0.005 0 1 2 3 4 5 NIMEP (bar) Figure 4.80 HC emissions versus NIMEP for 4 olefin fuels at 1500 RPM. Emissions of NO and CO also showed differences between different fuels, but the differences do not appear to be significant once the effects of combustion phasing and load have been taken into account. Figures 4.81-4.86 show the HC, CO, and NO emissions for all of the fuels across the range of speeds and loads respectively. Each point is either a high load limit or a low load limit. For the HC and CO emissions, the points with the higher emissions are low load limits and the points with the lower emissions are the high load limits. For NO emissions, the high load points have the higher emissions and the low load points have the lower emissions. 165 Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.035 * Low RON 0.03 n High Olef Low Load Limits O Heavy 0.025 -) A Ex Low A&O A High Arom 0.02 * High RON x Eth x 0.015 * Base * High RON+Eth 0.01 A Low Arom m Low Olef 0.005 o ULR Winter High Load Limits 0 500 1000 2000 1500 2500 3000 Engine Speed (RPM) Figure 4.81 HC emissions versus engine speed. Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.035 * m * A 0.03 0.025 A e Low RON High Olef Heavy Ex Low A&O A High Arom * High RON X Eth 0.02 I) %0 0.015 U X * Base )K High RON+Eth A LowAron m Low Olef o ULR Winter 0A 0.01 0.005 0 0 1 2 3 4 5 6 NIMEP (bar) Figure 4.82 HC emissions versus NIMEP. 166 Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.14 Low Load Limits . 0.12 o o *Low RON m High Olef o Heavy A Ex Low A&O 0 0.1 0 I 0 U. A High Arom 0.08 * High RON x Eth * Base 0.06 I 0.04 X * High RON+Eth ALowArom a Low Olef 0.02 H d o ULR Winter High Load Limits 0 500 1000 3000 2500 2000 1500 Engine Speed (RPM) Figure 4.83 CO emissions versus engine speed. Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.14 *Low RON mHigh Olef * Heavy A Ex Low A&O 0.12 0.1 a) A High Arom 40 0.08 * High RON 0) 0.06 0) ,%A 0.04 x Eth * Base A *A x 0.02 A * High RON+Eth A Low Arom m Low Olef x o ULR Winter 0 0 1 2 3 4 5 6 NIMEP (bar) Figure 4.84 CO emissions versus NIMEP. 167 Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.014 * Low RON * High Olef 0 Heavy A Ex Low A&O A High Arom * High RON x Eth * Base K High RON+Eth A Low Arom a Low Olef o ULR Winter 0.012 High Load Limits 0.01 0.008 0.006 0) z 0 0.004 0.002 Low Load Limits A 0 1000 500 3000 2500 2000 1500 Engine Speed (RPM) Figure 4.84 NO emissions versus engine speed. Operating Range: High & Low Load Limits Phi = 1, WOT, Intake Air T 40C 0.014 * Low RON mHigh Olef o Heavy A Ex Low A&O 0.012 0.01 0)- 0 z A High Arom * High RON X Eth 0.008 0.006 %0 - * * Base * High RON+Eth A Low Arom 0.004 n Low Olef 0.002 o ULR Winter 0 0 1 2 3 4 5 6 NIMEP (bar) Figure 4.85 NO emissions versus engine speed. 168 Chapter 5: Comparison of HCCI Operation with SI Operation The experimental results show the operating limits for a particular type of engine: a low compression ratio, negative valve overlap gasoline HCCI engine with cam phasing. This project, like most gasoline HCCI research, is motivated by the potential for an improvement in fuel economy over tradition SI engine operation. A comparison of SI and HCCI operating ranges, efficiencies, and emissions can give some indication of the magnitude of the potential benefits from a practical application of this type of HCCI engine. 5.1 Operating Range Comparison One great challenge of implementing HCCI combustion is the limited operating range in which HCCI combustion is feasible. Figure 5.1 shows the HCCI and SI operating ranges for this engine. The SI operating range data was provided by project sponsor Ford Motor Company, and is consistent with SI data collected on the experimental engine. 1 SI & HCCI Operating Ranges 14 12 10 CL L. 0 - - ,. 8 6 -SIOperating Range -HCCI z Operating Range 4 2 0 0 1000 2000 3000 4000 5000 6000 Engine Speed (RPM) Figure 5.1 HCCI & SI operating range comparison: NIMEP versus engine speed. 169 The HCCI operating range is based on the data collected in this project. The HCCI operating range is clearly much smaller than the SI range. Fortunately a significant portion of engine operation during the EPA fuel economy test occurs in or near the HCCI operating range (see figures 5.3 & 5.4). However, the HCCI operating range does not extend down to idle, although it is probably within 1 bar at the low load limit. 170 5.2 HCCI and SI Efficiency Two mechanisms contribute to the efficiency improvement HCCI shows as compared to SI engine operation. The most important contributor to improved efficiency is unthrottled operation. Because load can be controlled by dilution instead of by throttling the flow, the pumping losses during the gas exchange process are much smaller than for SI operation. Very fast combustion can also contribute to improved efficiency if combustion timing is optimized. HCCI approaches the ideal constant volume combustion much more closely than SI combustion. Lean operation can also improve efficiency (see section 3.3.2), for both SI and HCCI operation. However for this comparison a constant stoichiometric air-fuel ratio was maintained. Stoichiometric operation is necessary for the use of a three-way catalyst for emissions reductions, and therefore may be necessary for any hybrid SI/HCCI applications. HCCI Efficiency Improvement over SI 1500 RPM, Phi =1 S25% E20%- o C15%>10% .5%- 1 2 3 4 5 NIMEP (bar) Figure 5.2 HCCI efficiency improvement over SI efficiency. 171 Figure 5.2 plots the efficiency improvement HCCI displays over SI operation in terms of NIMEP versus percent improvement. The efficiency improvement is much larger at lower loads than at higher loads. SI efficiency falls quickly as load is decreased because of the effects of throttling. The efficiency benefits across the range of speeds are similar, although the operating range is limited at higher speeds. To better quantify the potential benefit from HCCI operation using a residual trapping strategy, some simple modeling was carried out using Advisor vehicle simulation software. A typical mid-size vehicle was simulated over the urban and highway Federal Test Procedure (FTP) drive cycles. In each case, the engine demand in terms of BMEP was calculated at each second throughout the drive cycle. If the engine demand was greater than 1 bar BMEP and less then 4 bar BMEP, HCCI operation was considered possible. Figure 5.3 shows the portions of the cycle where HCCI operation is theoretically possible by marking those points in blue. SI operating points are marked in red. HCCI operation is theoretically possible over approximately 40% of the urban drive cycle. SI vs. HCCI Mode over the Urban FTP Fuel Economy Drive Cycle HCCI mode - 40% of cycle 30 -25 E 20 0a C. 15 - .C it Z 0 -W 0 A HCCI Mode -SI mode AIN.- INC 500 1000 1500 Time (sec) Figure 5.3. Vehicle speed versus time for the FTP Urban Drive Cycle. 172 The same simulation and calculations were performed for the FTP highway test. In this case, because of the more consistent vehicle speeds and engine demand, HCCI operation was calculated to be possible of 75% of the drive cycle (figure 5.4). FTP Highway Fuel Economy Drive Cycle HCCI Mode - 75% of Cycle 30 25 - ^A .20- A HCCIMode - 15 - SIMode o 0 10-; A > 5 0 0 200 600 400 800 1000 Time (s) Figure 5.4. Vehicle speed versus time for the FTP Highway Drive Cycle. In order to estimate the order of magnitude of the possible benefit from a hybrid SI/HCCI engine, the fuel economy drive cycle results were used to compare SI fuel consumption with SIIHCCI fuel consumption. For the SI/HCCI fuel consumption calculation, it was assumed that an immediate perfect transition from SI to HCCI operation and vice versa was possible. While this is obviously unrealistic, it does allow an upper bound calculation. It was also assumed that HCCI mode was possible in the range of 1 to 4 bar BMEP of load. The efficiency benefit was calculated based on a comparison of the efficiency of the baseline SI data and the HCCI data. This improvement was then applied to the simulated fuel consumption in the regions where HCCI operation was assumed possible. The modeling showed a fuel consumption improvement in the urban drive cycle of approximately 6%, and an improvement in the highway drive cycle of 10%. While this is a small improvement in fuel economy, the hardware costs to allow hybrid SIIHCCI operation of an engine may also be relatively small. A valvetrain with cam 173 phasing and cam lobe switching might allow transition between SI and HCCI modes, and control within HCCI mode. These valvetrains are already in production (Honda V-TEC). The fuel economy benefit of a hybrid SI/HCCI engine can be improved by expanding the HCCI operating range. Researchers have investigated boosting to expand high load operation without knocking, and considered different fuels to allow lower load operation. A second way to increase the benefit of HCCI operation is to modify the power train so that the engine is able to stay within the HCCI operating range during a larger portion of vehicle operation. A hybrid electric powertrain might be well suited for an SI/HCCI engine: the hybrid electric powertrain turns off the engine at very low load and uses electric power. The hybrid electric powertrain may also allow a smoothing of the engine load requirements enabling the engine demand to remain within the HCCI range more of the time. 174 5.3 HCCI & SI Emissions Comparison A comparison of HCCI and SI emissions shows that the results are generally consistent with expectations. HC and CO emissions are similar to SI emissions near the high end of the HCCI load range. At the low end of the HCCI range the HC and CO emissions are higher than SI emissions. The NO emissions are much lower than undiluted SI emissions across the entire HCCI load range. Figure 5.5 compares HC emissions across the HCCI load range for HCCI and SI. The SI emissions are based on HC emissions results using the baseline engine, and are shown by the black solid line. HC emissions are very similar at the high end of the load range, but HCCI emissions increase significantly as the low load boundary is approached. HC Emissions 1500 RPM, Phi =1, Intake Air T 0.02 40C - 0.015S0.01 - V 0.005 01 2 3 NIMEP (bar) 4 5 Figure 5.5 HCCI & SI: HC emissions versus NIMEP. Emissions of CO are shown in figure 5.6. CO emissions are lower than SI emissions at the high end of the load range. As the low load limit is approached, the CO emissions increase to higher levels than SI CO emissions. As discussed previously CO emissions are very sensitive to lambda, so small differences in lambda result in some increased variability. 175 CO Emissions 1500 RPM, Phi = 1, Intake Air T = 40C 0.08 - 0.06 , U 0.04 o 0.02 0 1 2 3 4 5 NIMEP (bar) Figure 5.6 HCCI & SI: CO emissions versus NIMEP. NO emissions were not collected for the baseline SI engine. Instead, NO emissions were compared with emissions from an experimental SI engine [27]. The experimental SI engine had similar geometry and compression ratio (11.7), but data was only available at 3.5 bar NIMEP. In figure 5.7, the triangles show the HCCI emissions. The circle shows the SI emissions for no EGR, and the square marks the emissions for heavy dilution with EGR. The solid black line is an estimate of SI emissions without EGR across the load range based on results in Heywood [18]. NO Emissions 1500 RPM, Phi 1, Intake Air T = 40C 0.04 0.03 0.02 0 0.01 z A 1 2 3 4 5 NIMEP (bar) Figure 5.7 HCCI & SI: NO emissions versus NIMEP. 176 Except at the high load limit, HCCI emissions of NO are close to 2 orders of magnitude lower than SI emissions. At the high load limit, NO emissions are only 4 times lower. Figure 5.8 looks more closely at the HCCI emissions. The emissions for the EGR diluted SI combustion are shown by the square. The EGR dilution level, 36%, is right at the dilution limit for SI combustion for this load. The diluted SI emissions are very similar to the HCCI emissions at this load. HCCI does have higher dilution levels, but HCCI also has shorter combustion durations, higher peak pressures. The higher dilution levels tend to reduce burned gas temperatures and NO emissions, and higher peak pressures tend to increase peak temperatures and NO emissions. NO Emissions 1500 RPM, Phi =1, Intake Air T = 40C 0.006 W 0.005- A %20.0040 0.003 %0.002 z 0.001 -_ 0 1 _AA AAA 1,AA 2 3 4 5 NIMEP (bar) Figure 5.8 HCCI & SI: NO emissions versus NIMEP. 177 Chapter 6: Conclusions In this project the impact of ambient conditions on the HCCI operating range has been evaluated over a range of engine speeds and air fuel ratios. HCCI combustion, unlike SI or diesel combustion, is governed by chemical kinetics. The state of the cylinder over time determines when and how combustion will occur. The ambient air temperature, ambient air humidity, and type of fuel can all affect the state of the cylinder during the compression stroke, and so can have an impact on combustion. Control of Combustion Experiments were carried out on a modified production 2.3L 14 gasoline engine. Dilution and heating of the charge was achieved through residual trapping (negative valve overlap). In order to control HCCI combustion, it is necessary to control the state of the cylinder: the temperature, pressure, and concentrations of gases within the cylinder. Control of HCCI operation was accomplished using cam phasing. Exhaust cam phasing controls residual fraction, which determines the amount of fresh air entering the cylinder. Intake cam phasing allows some control over combustion phasing by impacting the flow into the cylinder. Advanced intake cam timing increases the residual fraction lowering burned gas temperatures, and reduces the pressure at IVC lower the temperature of the cylinder mixture. Cam phasing was used to maximize the operating range across the range of speeds for different temperatures, equivalence ratios, and fuels. To reach the minimum load, exhaust cam phasing was advanced until misfire occurred. Intake cam timing was found to have a small impact on the lowest load operation. To reach the maximum load, exhaust cam timing was reduced until either knock or unstable combustion occurred. At the same time intake cam timing was adjusted to optimize combustion phasing. Later combustion phasing reduced the tendency of the engine to knock. Combustion occurring too far after TC resulted in misfire. 178 Intake Air Temperature Intake air temperature was varied between -6C and lOOC. Increasing temperature advanced combustion phasing, but cam phasing was largely able to compensate for the higher air temperatures. The higher intake air temperatures did result in a reduction of the high load limit at lower engine speeds. At 1000 RPM the high load limit was reduced by approximately 1 bar with higher intake air temperatures. However this reduction was significantly less than the almost 2 bar reduction in the high load limit associated with increasing engine speed. Intake air temperature had less impact on the low load limit, but higher air temperatures did allow slightly lower load operation. The higher residual fractions at low load operation minimized the impact of changes in air temperature. Intake air temperature did have more of an impact during lean operation. Humidity The humidity of the intake air was controlled from approximately 10% relative humidity at 30C, to 100% relative humidity at 30C. Increasing humidity was found to delay combustion phasing. During stoichiometric operation, this delay allowed a small extension (a few tenths of a bar in NIMEP) in the high load limit. During lean operation, the delay resulted in a reduction of the high load limit. Humidity had little impact on the low load limit, except for at very high humidity levels: the low load limit was raised by a few tenths of a bar. Fuel Composition Twelve fuels were created to vary 5 fuel properties: RON, RVP, aromatic content, olefin content, and ethanol content. The different fuels did show different combustion behavior, but cam phasing was largely able to compensate for any differences, and the operating ranges of the different fuels were quite similar. At 1000 RPM the high load limits were 179 within a range of 0.8 bar. The range in the low load limits was approximately 0.7 bar. The difference in operating ranges was small at higher speeds. At first glance the performance of the fuels seemed to contradict expectations. The high olefin and aromatic fuels seemed to auto-ignite more easily than other fuels. A closer examination showed that the paraffin content of the fuels was changing with the aromatic and olefin content. The greater octane sensitivity of the aromatic and olefin fuels, combined with the changing paraffin content, could account for the combustion behavior. Regression analysis suggested that at the high load limit napthene content seems to have a significant impact. Practical Implementation of HCCI The operating range of this negative valve overlap HCCI engine is small compared with the SI operating range, but the range does cover an important area of the SI map (low load and low speed). HCCI emissions of HC and CO are similar to SI emissions at the high end of the load range, and higher than SI at the low end. Emissions of NO are much lower than undiluted SI levels, but appear to be similar to NO emissions of SI with very high EGR. The potential fuel economy benefit of a hybrid SIIHCCI engine is estimated to be modest (6% urban FTP, 10% highway FTP), but the hardware costs may be modest as well. Ambient air temperature, humidity, and gasoline composition all impact combustion. However cam phasing is largely able to compensate for changes in combustion. An optimal control strategy would require some feedback on combustion, such as the location of CA50, or peak pressure, etc. A fixed control strategy without feedback might be possible, but it would limit the HCCI operating range. 180 Appendix A: Combustion Phasing Analysis of cylinder pressure to determine combustion phasing is an important element in combustion analysis. Combustion characteristics were analyzed using a combination of heat release analysis and the 1-D engine simulation software WAVE as described in SAE paper 2005-01-2133 [28]. The process is a multi-step iterative process: 1. Simulate the operating condition with WAVE using estimated combustion characteristics. 2. Extract from WAVE the residual fraction and use this as input into the heat release analysis code, along with actual operating characteristics. 3. Use the combustion duration and phasing results from the heat release analysis as inputs into WAVE. 4. Compare the actual cylinder pressure data with the simulated cylinder pressure from WAVE. 5. Adjust the intake and exhaust pressure in WAVE to account for any dynamics and run to get a new residual fraction. 6. Repeat steps 3-5 until the results converge (simulated cylinder pressure and actual cylinder pressure data agree, and residual temperatures and combustion characteristics reach steady values). Once the simulated and actual cylinder pressures agree, it is assumed that WAVE accurately models the cycle. This allows not only an examination of combustion characteristics and residual fraction, but also cylinder temperature, heat release, instantaneous mass flows, etc. over the course of the cycle. This process is however too time consuming to perform for each data point. Fortunately the location of the 50% mass fraction burn location seems to correlate very well with the location of the maximum rate of pressure rise within the cylinder. Operating points across a range of engine speeds and loads were analyzed using the combined WAVE + heat 181 release analysis. Figure Al shows the comparison of 50% mass fraction burn and maximum rate of pressure rise for each of the points. Correlation of 50% Burn Location and Maximum Rate of Pressure Rise Location 12- r 0 w Z W o - 0 2 4 6 8 10 12 Location of Maximum Rate of Pressure Rise (Avg. of 400 Cycles) Figure Al. Comparison of CA50 and max dP/dt location. Because the correlation is very good, the location of the maximum rate of pressure rise has been used as a first pass value for CA50. In cases where a detailed analysis was appropriate, the detailed WAVE + heat release analysis was performed. 182 Appendix B: Throttled HCCI Operation Throttle is another potential control variable for HCCI operation. Unfortunately throttling has a significant negative impact on efficiency. Throttling is however another variable that can be used to impact mixture concentrations, pressures, and temperatures. Therefore throttling can allow some control over combustion. Figure B1 shows a two sweeps of MAP. In each case all variables except MAP are held constant. Two points with the same NIMEP are highlighted: A & B. The later exhaust cam timing of A results in less trapped mass. In order to achieve the same load, A must be throttled. Throttled Operation: Load vs. MAP 1500 RPM, Phi = 1, Intake Air Temp 1OOC 1.05 1 - 0.95 .c M0 0.9 - +EVC9 BTC, 48 ABC ___IVC 0.85 - N EVC 7- 7BTC, 0.8 48 0.8 -IVO 0.75 0.7 1.5 2.5 3.5 4.5 NIMEP (bar) Figure B1. Throttle operation: NIMEP versus MAP. Figure B2 shows the combustion phasing versus load for the same points. The combustion phasing for the more throttled (lower MAP) case is significantly later than for the less throttled points. 183 Throttled Operation: Load vs. Combustion Phasing 1500 RPM, Phi = 1, Intake Temp IOOC 12 010 8 EVC 98 BTC, IVC48ABC IA 6 S4 +_ 02 EVC 77 BTC, 48 ABC _IVC B '00 -2 1 1.5 2.5 4.5 3.5 NIMEP (bar) Figure B2. Throttled operation: combustion phasing versus load. Table B 1 compares some of the important characteristics of the points A & B. Although the pumping losses are greater in A (the more throttled case), the efficiency of A is better than B. The combustion phasing of A is closer to optimal, and very little of the combustion energy is wasted. The later combustion phasing of point A also leads to lower rates of pressure rise - 4.0 MPa/ms versus 5.8 MPa/ms. Point B would not meet the selected high load knock criterion, but point A would be acceptable. A B EVC (CAD BTC) 77 98 IVC (CAD ABC) 48 48 MAP (bar) 0.85 1 NIMEP (bar) 3.21 3.22 Inducted Mass (mg) 202 206 -0.372 -0.292 ISFC (g/kw-hr) 257 262 50% Burn (CAD 6.3 -0.5 Max dP/dt (MPa/ms) 4.0 5.8 PMEP (bar) Table BI. Comparison of two points with similar load: throttled and unthrottled. 184 Figure B3 plots the two MAP sweeps showing maximum rate of pressure rise versus NIMEP. The more throttled case, with the later combustion phasing, achieves a higher load without knocking (0.3 bar higher NIMEP at high load limit). The effectiveness of throttling in retarding combustion phasing suggests that cylinder pressure at IVC does in fact play an important role in cylinder temperature and combustion phasing. Throttled Operation: Load vs. Max dP/dt 1500 RPM, Phi = 1, Intake Air T 1 OOC 7 6U E +*EVC98 BTC, RVC 48 ABC 5 S4 - 3 - __2 . + _ *EVC 77 BTC, 48 ABC 2IVC 0 1.5 2.5 3.5 4.5 NIMEP (bar) Figure B3. Throttled operation: load versus max rate of pressure rise. Throttling does offer some control over combustion phasing, and allows slightly higher load operation without exceeding the high load limit. The role of throttling at the low load limit was also examined. In the case of lean HCCI (dilution with air, not residuals), throttle has been used help achieve very low load by maintaining a desirable ratio of reactants to diluting gases. However as figure B4 shows, throttled operation did not allow lower load operation in the case of HCCI with a large residual fraction. At two different cam settings, the MAP was swept until the engine misfired. In a third case, the load was reduced by increasing residual fraction until the engine misfired. In all three cases the load is very similar. In the non-throttled case, the residual fraction increases and flame temperatures decrease until the temperature of the residuals is too low to support combustion. In the throttled case, the residual fractions are much lower for a given load, which leads to higher flame temperatures. However this effect is canceled by the cooling of the residuals when they are expanded to the lowered MAP. 185 Throttle: Load vs MAP 1500 RPM, Phi = 1, Intake Air 100C 3.3 3 2.9 +EVC 98 BTC, IVC48 ABC ~2.7 2.5 2.3 - CL W EVC 113 BTC, IVC 48 ABC ~2.1 1.9 1.7 1.5 0.75 AEVC 122 BTC, IVC 42 ABC 0.85 0.95 1.05 MAP (bar) Figure B4. Throttled operation: NIMEP versus MAP at the low load limit. When the efficiencies of the throttled and unthrottled cases are examined, it is seen that the efficiencies of the throttled case at low load are significantly worse than the unthrottled case (figure B5). Throttle: MAP vs ISFC 1500 RPM, Phi = 1, Intake Air 1OOC 280 *EVC 98 BTC, IVC48 ABC 275 270 so 265 0 ) +A 260 255 0.75 + 0.85 0.95 m EVC 113 BTC, CIVC 48 ABC A EVC 122 BTC, IVC 42 ABC 1.05 MAP (bar) Figure B5. Throttled operation: ISFC versus MAP at the low load limit. 186 Appendix C: Local Knock Criterion Professor Wai Cheng derived this criterion. Below is the derivation: This criterion seeks to determine how fast the rate of heat release must be to result in oscillating pressure waves. Consider a volumetric heat release rate q in a sphere of radius r. Assume prefect gas with constant properties. Constant pressure process in time dt: mcAT = 4Vdt Volume expansion with constant pressure for fixed mass: p(V + AV)= mR(T + AT) PAV = mR 4lvdt j _E PAV = r1 Expansion is fast enough if AV = 1 4Vdt 4Vdt 4Cr2adt , where a is the speed of sound. 187 P4fr2 adt = r I )4Vdt V =4/3)rr3 P4cr2 ad = 44 / 37rr3 dt P >3r (r-1(r/a) The criterion depends on the volumetric rate of heat release, the pressure, the composition of the gases (gamma), the speed of sound, and the size of the local area under consideration. 188 Appendix D: Additional Fuel Data T T I RON RVP (psi) Base fuel 92.2 7.0 High RVP 91.7 Heavy fuel - Y Eth LHV MJ/ka 11.0% 29.0% 0.0% 43.9 6.7 45.8 6.18 0.11 9.1 10.6% 27.8% 0.0% 43.9 10.5 44.0 5.93 0.10 91.5 6.8 10.8% 29.4% 0.0% 44.2 6.9 41.5 7.88 0.30 High RON 98.2 6.6 10.1% 30.0% 0.0% 44.3 3.0 51.8 3.61 0.05 Low RON 87.7 6.4 0.9% 1.5% 0.0% 44.9 8.8 84.1 1.58 0.00 High Aromatic 92.5 7.2 9.2% 38.8% 0.0% 43.2 10.6 34.6 5.98 0.16 Low Aromatic 91.9 7.1 13.6% 14.3% 0.0% 43.6 4.8 59.4 5.96 0.07 High Olefin 92.7 7.4 18.9% 26.8% 0.0% 43.8 5.8 39.2 7.85 0.18 Low Olefin 93.0 6.6 4.3% 32.3% 0.0% 43.7 8.1 49.8 4.63 0.05 Ext Low Arom & Ole 91.2 6.3 0.4% 0.8% 0.0% 44.8 6.3 88.1 0.66 0.00 Ethanol 91.9 8.2 9.8% 25.9% 10.0% 42.7 9.6 37.0 6.78 0.11 High RON w/Eth 98.2 8.1 9.7% 26.2% 10.0% 42.2 3.8 44.1 4.81 0.07 Ole Aro n-Par. i-Par. Na ph. Table D1. Fuel properties. MON R+M/2 sp arav R-M Base fuel 84.1 88.2 0.744 8.1 High RVP 83.3 87.5 0.735 8.5 Heavy fuel 82.9 87.2 0.756 8.6 High RON 87.8 93.0 0.746 10.4 Low RON 86.9 87.3 0.702 0.8 High Aromatic 83.2 87.9 0.761 9.3 Low Aromatic 84.5 88.2 0.720 7.4 High Olefin 82.5 87.6 0.741 10.2 Low Olefin 84.8 88.9 0.748 8.2 Ext Low Arom & Ole 89.8 90.5 0.703 1.4 Ethanol 83.7 87.8 0.747 8.2 High RON w/Eth 86.7 92.5 0.747 11.5 Table D2. Fuel properties continued. 189 poly- Olefins: Ref. Points & HC Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 20.0% 18.0% E 16.0% .0 14.0% G EVC 98 BTC, In T 40C 12.0% .a 0 0 Figure D1. HC emissions versus olefin content for ref. points at 1500 RPM. * EVC 108 BTC, In T 40C 10.0% A EVC 98 BTC, In T 100C 8.0% 6.0% A 4.0% m EVC 108 BTC, In T 100C G.* 2.0% 0.0% 0. 005 0.015 0.01 0.02 HC (g/g fuel) Olefins: Ref. Points & CO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 20.0% a) A 18.0% Figure D2. CO emissions versus olefin content for ref. points at 1500 RPM. 0K E 16.0% 0 14.0% G EVC 98 BTC, In T 40C .0 12.0% A* 10.0% 0) 0 * EVC 108 BTC, In T 40C GE A EVC 98 BTC, 8.0% 6.0% * EVC 108 BTC, InT 100C A 00 4.0% In T 100C 2.0% 0.0% 0.005 111. 0.025 GAI 0.045 0.065 0.085 CO (g/g fuel) Olefins: Ref. Points & NO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 20.0% a) * 18.0% Figure D3. NO emissions versus olefin content for ref. points at 1500 RPM. AG@ E 16.0% 0 14.0% G EVC 98 BTC, In T 40C .0 12.0% * EVC 108 BTC, In T 40C G A E* 10.0% A EVC 98 BTC, 8.0% 0 0 6.0% 4.0% 4$ In T 100C m EVC 108 BTC, In T 100C AG@ 2.0% 0.0% 0 0.0005 0.001 0.0015 NO (g/g fuel) 190 Aromatics: Ref. Points & HC Emissions 1500 RPM, Phi =1, IVC 42 CAD ABC 45.0% - 40.0%- SAD O EVC 98 BTC, In T 40C 35.0%0 30.0% - 25.0% - AG No * EVC 108 BTC, In T 40C 20.0%- A EVC 98 BTC, In T 1OOC 15.0% - E 10.0% - 0 Figure D4. HC emissions versus aromatic content for ref. points at 1500 RPM. m EVC 108 BTC, In T 1OOC 5.0%- * A 0.0% 0.005 0.02 0.015 0.01 HC (g/g fuel) Aromatics: Ref. Points & CO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 45.0% E 40.0%> 35.0% 30.0% AU S 9 * EVC 98 BTC, In T 40C - a IV U * EVC 108 BTC, In T 40C 25.0%5 20.0% Figure D5. CO emissions versus aromatic content for ref. points at 1500 RPM. - A EVC 98 BTC, In T 1OOC 15.0%- * EVC 108 BTC, InT 100C E 10.0%5.0%- 0.0% 0.005 1,* , MA , 0.045 0.065 0.085 CO (g/g fuel) 0.025 Aromatics: Ref. Points & NO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 45.0% E 40.0% 9 B Figure D6. NO emissions versus aromatic content for ref. points at 1500 RPM. * EVC 98 BTC, In T 40C 0 35.0% A 30.0% EVC 108 BTC, In T 40C 25.0% 0 20.0% A EVC 98 BTC, In T 100C 15.0% S A n EVC 108 BTC, In T 100C E 10.0% 5.0% 0.0% 0 0.0005 0.001 0.0015 NO (g/g fuel) 191 RVP: Ref Points & HC Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 10.0 9.0 * EVC 98 BTC, In T 40C AM 8.0 0. A0 0 7.0 Figure D7. HC emissions versus RVP for four reference points at 1500 RPM. * EVC 108 BTC, In T 40C * AGE A EVC 98 BTC, In T 100C 6.0 n EVC 108 BTC, In T 100C 5.0 4.0 0.005 0.01 0.015 0.02 HC (g/g fuel) RVP: Ref Points & CO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 10.0 - 9.0 - 8.0 - * EVC 98 BTC, In T 40C * EVC 108 BTC, In T 40C 0. Figure D8. CO emissions versus RVP for four reference points at 1500 RPM. 7.0 ** 6.0 A EVC 98 BTC, In T 1OOC GA m EVC 108 BTC, In T 100C 5.0 4.0 0.005 0.025 0.065 0.045 0.085 CO (g/g fuel) RVP: Ref Points & NO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 10.09.0 _ O EVC 98 BTC, In T 40C 8.0 0. * EVC 108 BTC, In T 40C 7.0 Figure D9. NO emissions versus RVP for four reference points at 1500 RPM. GA A EVC 98 BTC, In T 100C 6.0 - * EVC 108 BTC, In T 100C 5.0 4.0 - 0 0.0005 0.001 0.0015 NO (g/g fuel) 192 RON: Ref Points & HC Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 100.0 A 98.0 A * EVC 98 BTC, In T 40C 96.0 z0 cc 94.0 * EVC 108 BTC, In T 40C 92.0 A EVC 98 BTC, In T 100C 90.0 m EVC 108 BTC, In T 100C Figure D10. HC emissions versus RON for four reference points at 1500 RPM. 88.0 86.0 0. 005 0.01 0.02 0.015 HC (g/g fuel) RON: Ref Points & CO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 100.0 * A 98.0 Figure D11. CO emissions versus RON for four reference points * EVC 98 BTC, In T 40C 96.0 z at 1500 RPM. * EVC 108 BTC, In T 40C 94.0 0 A + 92.0 ae A EVC 98 BTC, In T 100C 90.0 m EVC 108 BTC, In T 100C 88.0 86.0 0. 005 0.045 0.025 0.065 CO (g/g fuel) RON: Ref Points & NO Emissions 1500 RPM, Phi = 1, IVC 42 CAD ABC 100.0 6b 98.0 A * EVC 98 BTC, In T 40C 96.0 z0 * EVC 108 BTC, In T 40C 94.0 " 92.0 E-* B A Figure D12. NO emissions versus RON for four reference points at 1500 RPM. A EVC 98 BTC, In T 100C 90.0 m EVC 108 BTC, In T 100C 88.0 Aw 86.0 0 0.0005 0.001 0.0015 NO (g/g fuel) 193 Ethanol: Ref Points & HC Emissions O Base: EVC 98 BTC, In 1500 RPM, Phi = 1, IVC 42 ABC T 40C 11.0%* Base: EVC 108 BTC, In A A* * 9) 0 T 40C E 9.0%A Base: EVC 98 BTC, In T 100C 7.0%* Base: EVC 108 BTC, In T 100C 5.0% 0 O Hi RON: EVC 98 BTC, 0 In T 40C 3.0%C * Hi RON: EVC 108 BTC, (U In T 40C 1.0% uJ AZID NCO* A Hi RON: EVC 98 BTC, In T 100C -1.0% 0.013 o Hi RON: EVC 108 BTC, 0.005 0.007 0.009 0.011 In T 100C HC (g/g fuel) Ethanol: Ref Points & CO Emissions * Base: EVC 98 BTC, In 1500 RPM, Phi = 1, IVC 42 ABC T 40C 11.0% * Base: EVC 108 BTC, In Adl O @ 0) T 40C E 9.0% A Base: EVC 98 BTC, In T 100C 7.0% * Base: EVC 108 BTC, In .0 T 100C 5.0% 2 O Hi RON: EVC 98 BTC, 0 In T 40C 3.0% * Hi RON: EVC 108 BTC, In T 40C 1.0% w A A 0 4O111 A Hi RON: EVC 98 BTC, , l 0 NO In T 100C -1.0% 0.065 o Hi RON: EVC 108 BTC, 0.025 0.045 0.005 In T 100C CO (g/g fuel) Ethanol: Ref Points & NO Emissions * Base: EVC 98 BTC, In 1500 RPM, Phi = 1, IVC 42 ABC T 40C 11.0% * Base: EVC 108 BTC, In 0 A 0D T 40C E 9.0% A Base: EVC 98 BTC, In T 100C 7.0% * Base: EVC 108 BTC, In (D T 100C 5.0% 0 O Hi RON: EVC 98 BTC, In T 40C 3.0% w * Hi RON: EVC 108 BTC, In T 40C 1.0% 4*o + BA A Hi RON: EVC 98 BTC, In T 100C -1.0% o Hi RON: EVC 108 BTC, 0.0015 0.001 0.0005 0 In T 100C NO (g/g fuel) Figure D13. HC emissions versus ethanol content for reference points at 1500 RPM. Figure D14. CO emissions versus ethanol content for reference points at 1500 RPM. Figure D15. NO emissions versus ethanol content for reference points at 1500 RPM. 194 Appendix E: Volumetric Efficiency Because of the very small valve lift (2 mm), volumetric efficiency drops more rapidly with increasing engine speed than for a traditional SI engine. Figure El shows the airflow into the cylinder per cycle as a function of engine speed for the intake system used for the fuel testing. The exhaust manifold was constant for all testing, but the intake manifold was different for the earlier intake air temperature and humidity testing. Different intake manifold configurations might not show the same response. For the data in figure El, the cam timing was held constant as the engine speed was increased. The air flow was calculated using the measured air-fuel ratio and the calculated fuel flow. As the air flow dropped the fueling rate had to be reduced. From 1500 RPM to 2800 RPM, the air flow drops by approximately 45%. Air Flow vs. Engine Speed Air Flow Calculated Using Lambda EVC 81 CAD BTC, IVC 43 CAD ABC 0" 4) 300 +4800 >250 _ 4000 E200 21m M 150 100 1200 A 3500 3200 1700 2200 2700 3200 Engine Speed (RPM) Figure El. Air flow into the cylinder per cycle versus engine speed. Figure El also shows that the reduction in volumetric efficiency is not smooth. At 2200 RPM, the flow restriction appears to deviate from the overall trend. Figure E2 plots the same air flow data points from figure El in terms of mass flow rate, instead of mass per cycle. 195 Air Flow vs. Engine Speed Air Flow Calculated Using Lambda EVC 81 CAD BTC, IVC 43 CAD ABC 3.5 - *+ *4800 U 3.1 - 0 M m4000 0 A 3500 E 2.9 o3200 I 2.72.5- 1200 1700 2200 2700 3200 Engine Speed (RPM) Figure E2. Air flow versus engine speed with constant cam timing. Both figure El and figure E2 show that the flow into the cylinder is restricted in some way at 2200 RPM. This restriction is believed to be a result of some pressure wave dynamics in either the intake manifold or the exhaust manifold. Low pressure in the intake manifold during the intake event, or high pressure in the exhaust manifold during the exhaust event could lead to a reduction in the inducted mass. The reduced flow at 2200 RPM is consistent with the cam timings seen in section 4.3.3.2. The air flow is most restricted at 2200 RPM, but is also reduced at 2000 RPM. Cam timing compensates for the flow restriction at 2000 RPM. At 2000 RPM, the intake and exhaust cam timing are later than one might predict from the overall trends. Later intake cam timing means that the air flow into the cylinder is less restricted. Later exhaust cam timing traps less residual, allowing more fresh charge into the cylinder. 196 References [1] Onishi, S., Jo, S.H., Shoda, K., Jo, P.D., Kato, S., "Active Thermo-Atmosphere Combustion (ATAC) A New Combustion Process for Internal Combustion Engines", SAE 790501, 1979 [2] Najit, P.M., Foster, D.E., "Compression-Ignited Homogeneous Charge Combustion", SAE 830264, 1983 [3] Thring, R.H., "Homogeneous Charge Compression Ignition (HCCI) Engines", SAE 892068, 1989 [4] Ishibashi, Y., Asai, M., "Improving the Exhaust Emissions of Two-Stroke Engines by Applying the Activated Radical Combustion", SAE 960742, 1996 [5] Kontarakis, G., Collings, N., Ma, T., "Demonstration of HCCI Using a Single-Cylinder, Four Stroke Engine with Modified Valve Timing", SAE 2000-01-2870, 2000 [6] Li, J., Zhao, H., Ladommatos, N., Ma, T., "Research and development of controlled auto-ignition (CAI) combustion in a 4-stroke multi-cylinder gasoline engine", SAE 2001-01-3608, 2001 [7] Li, C., Zhao, H., Jiang, X., Kalian, N., "Effects of Intake Valve Timing on Premixed Gasoline Engine with CAI Combustion", SAE 2004-01-2953, 2004 [8] Aroonsrisopon, T., Foster, D., Morikawa, T., Lida, M., "Comparison of HCCI Operating Ranges for Combinations of Intake Temperature, Engine Speed, and Fuel Composition", SAE 2002-01-1924, 2002 [9] Persson, H., Agrell, M., Olsson, J., Johansson, B., Strom, H., "The effect of intake temperature on HCCI operation using negative valve overlap", SAE 2004-01-0944, 2004 [10] Sjoberg, M., Dec, J., "An investigation of the relationship between measured intake temperature, BDC temperature , and combustion phasing for premixed and DI HCCI engines", SAE 2004-01-1900, 2004 [11] Christensen, M., Hultqvist, A., Johansson, B., "Demonstrating the multi-fuel capability of a homogeneous charge compression ignition engine with variable compression ratio", SAE 1999-01-3679, 1999 [12] Oakley, A., Zhao, H., Ladommatos, N., Ma, T., "Dilution effects on the controlled auto-ignition (CAI) combustion of hydrocarbon and alcohol fuels", SAE 2001-01-3606, 2001 [13] Kalghatgi, G., "Auto-Ignition Quality of Practical Fuels and Implications for Fuel Requirements of Future SI and HCCI Engines", SAE 2005-01-0239, 2005 [14] Shibata, G., Oyama, K., Urushihara, T., Nakano, T., "The effect of fuel properties on low and high temperature heat release and resulting performance of an HCCI engine", SAE 2004-01-0553, 2004 [15] Shibata, G., Oyama, K., Urushihara, T., Nakano, T., "Correlation of Low-Temperature Heat Release with Fuel Composition and HCCI Engine Combustion", SAE 2005-01-0138, 2005 [16] Yang, J., Culp, T., Kenney, T., "Development of a gasoline engine system using HCCI technology - The concept and the test results", SAE 2002-01-2832, 2002 [17] Chesa, J.L., Andreae, M., Green, W., Cheng, W., Cowart, J., "A Modeling Investigation Into the Optimal Intake and Exhaust Valve Event Duration and Timing for a Homogenous Charge Compression Ignition Engine", SAE 2005-01-3746, 2005 [18] Heywood, J., "Internal Combustion Engine Fundamentals", Appendix C, McGraw-Hill, 1988 197 [19] Ricardo WAVE Software [20] Yelvington, P., Green, W., "Prediction of the knock limit and viable operating range for a homogeneous-charge, compression-ignition (HCCI) engine", SAE 2003-01-1092 [21] Santoso, H., Matthews, J., Cheng, W., "Characteristics of HCCI Engine Operating in the NegativeValve- Overlap Mode", SAE 2005-01-2133, 2005 [22] Christensen, M., Johansson, B., Arn, J. Mauss, F., "Supercharged Homogeneous Charge Compression Ignition", SAE 980787, 1998 [23] Aceves, S., Flowers, D., Espinoza-Loza, F., Martinez-Frias, J., Dec, J., Sjoberg, M., Dibble, R., Hessel, R., "Spatial analysis of emissions sources for HCCI combustion at low loads using a multi-zone model", SAE 2004-01-1910, 2004 [24] Vressner, A., Lundin, A., Christensen, M., Tunestal, P., Johansson, B., "Pressure Oscillations During Rapid HCCI Combustion", SAE 2003-01-3217, 2003 [25] Eng, J.A., "Characterization of Pressure Waves in HCCI Combustion", SAE 2002-01-2859, 2002 [26] BP US Market Fuel Survey, 2005 [27] Ivanic, Z., Ayala, F., Goldwitz, J., Heywood, J., "Effects of Hydrogen Enhancement on Efficiency and NOx Emissions of Lean and EGR Diluted Mixtures in an SI Engine", SAE 2005-01-0253, 2005 [28] Santoso, H., Matthews, J., Cheng, W.K., "Characteristics of HCCI Engine Operating in the Negative Valve Overlap Mode", SAE 2005-01-2133, 2005 198
