EFFECTS OF OPERATING CONDITIONS, COMPRESSION RATIO, AND GASOLINE REFORMATE ON SI ENGINE KNOCK LIMITS by Michael D. Gerty B.A.Sc., Mechanical Engineering University of British Columbia, 2001 Submitted to the Department of Mechanical Engineering in Partial Fulfillment of the Requirements for the Degree of Master of Science in Mechanical Engineering at the MASSACHUSETTS INS OF TECHNOLOGY Massachusetts Institute of Technology JUN 16 2005 June 2005 © 2005 Massachusetts Institute of Technology All Rights Reserved LIBRARIES Signature redacted Signature of the Author ............................................................................... . Department of Mechanical Engineering May 13,2005 Signature redacted Certified by ........................................................................ J~~ 'B"li~~;~d Sun Jae Professor of Mechanical Engineering Thesis Advisor Signature redacted Accepted by ....................................................... ~.......... , .................. : ...... . Lallit Anand Professor, Department of Mechanical Engineering Chairman, Department of Graduate Commitee E. (this page intentionally left blank) EFFECTS OF OPERATING CONDITIONS, COMPRESSION RATIO, AND GASOLINE REFORMATE ON SI ENGINE KNOCK LIMITS by Michael D. Gerty Submitted to the Department of Mechanical Engineering on May 13, 2005 in Partial Fulfillment of the Requirements for the Degree of Master of Science in Mechanical Engineering ABSTRACT A set of experiments was performed to investigate the effects of air-fuel ratio, inlet boost pressure, hydrogen rich fuel reformate, and compression ratio on engine knock behavior. For each condition the effect of spark timing on torque output was measured. Knock limited spark advance was then found for a range of octane number (ON) for each of three fuel types; primary reference fuels (PRFs), toluene reference fuels (TRFs), and test gasolines. A new combustion phasing parameter based on the timing of 50% mass fraction burned, termed "combustion retard", was found to correlate well to engine performance. Increasing airfuel ratio increases the combustion retard required to just avoid knock for PRFs and has little effect for TRFs. Combustion retard also increases more with inlet pressure and decreases more with reformate addition for PRFs than for TRFs. Both fuel types responded similarly to increased compression ratio. The trends for gasoline are about halfway between PRFs and TRFs. Experiments were also performed to determine the response of mid-load indicated efficiency to air-fuel ratio, load, and compression ratio. At a compression ratio of 9.8:1, relative net efficiency improvement is about 2.5% per unit compression ratio. Efficiency peaks at about 14:1 with a maximum benefit of 6-7%. Detailed chemical kinetics were combined with a cylinder pressure based end-gas modeling methodology to successfully predicted the response of PRFs to compression ratio and air-fuel ratio, and the response of TRFs to boost. The difference between the response of PRFs and TRFs to air-fuel ratio was also captured. Constant volume chemistry modeling found that hydrogen slows alkane autoignition reactions by consuming hydroxy radicals in the end gas. Reforming 30% of the fuel entering an engine decreases the required fuel quality 10 ON or more, which would allow increased compression ratio or increased turbocharging without increasing combustion retard. A simplified analysis indicates that increasing compression ratio and downsizing the engine to maintain constant maximum torque would increase fuel efficiency by about 9%. Turbocharging and downsizing would increase fuel efficiency by about 16%. Thesis Advisor: John B. Heywood Title: Sun Jae Professor of Mechanical Engineering 3 (this page intentionally left blank) ACKNOWLEDGEMENTS I would like to thank Professor John Heywood for allowing me the invaluable opportunity to focus on a subject that I enjoy at one of the worlds greatest engineering institutions. It has been an honor to be mentored by a leader of such great character, knowledge, and integrity. His respect and high professional expectations motivated me to excel in my areas of strength, to develop new skills, and to strive for greater achievement. He was a key part of making my time at MIT a rewarding experience. This work would not have been possible without my predecessor Jennifer Topinka. Her creative and yet practical ideas pushed the plasmatron project into the new territory that I lucky to be able to explore. My valued co-workers and office mates Joshua Goldwitz and Ziga Ivanic made coming to the lab an adventure, and helped make my transition into MIT and the Sloan Automotive Laboratory much smoother and more enjoyable. Thanks also to Ferran Ayala and Bridget Revier, my teammates who will be carrying the plasmatron/knock torch when I leave. The Sloan Automotive Laboratory has been a great place to work, learn, and socialize with fellow students and dedicated MIT employees. It's impossible to name everybody, but I'd like to thank those that made life at the lab interesting and helped me with my research. Among the employees, I'd like to thank Karla Stryker for keeping Professor Heywood's chaotic schedule under control; Thane Dewitt and Raymond Phan for facilitating virtually all experimental work that goes on in the lab; and Professor Jim Keck, Professor Wai Cheng, and Dr. Jim Cowart for their useful insight into the complexities of this project. This project wouldn't exist without its sponsor Arvin Meritor with Rudy Smaling leading the way. The Plasma Science and Fusion Center also deserves credit for developing a key element of this research, the plasmatron fuel reformer. I sincerely appreciate all of the support that I received from my home country, Canada. Dr. Robert Evans at UBC, and my colleagues and supervisors at Westport Innovations afforded me the opportunity to gain experience in the engine business, which paved the way to my application and acceptance to MIT and the Sloan Lab. A scholarship from the Natural Sciences and Engineering Research Council of Canada helped to support me financially. The company of my fellow ex-pats, Jeff Jocsak, JP Urbanski, and Devon Manz, allowed me to enjoy some of the finer points of Canadian culture while away from home. Also, absence has not diminished my appreciation for Perry Wong, Kelly Cameron, Graham Knutson, and all my other friends from beautiful British Columbia. My family - Mom (Carol), Dad (Brian), Krista, and Tyler - have always been there for support and encouragement when I needed it most. Finally, my fiancee Phoi has been my most inspirational force, even from thousands of kilometers away. I'd like to express my sincere gratitude for the loving support, encouragement, and patience that she provided me every step of the way. 5 (this page intentionally left blank) 6 TABLE OF CONTENTS A bstract................................................................................. - - -. 3 ........ .. ---................-- Acknowledgements ....................................................................................... . ---------------............... L ist of Figures......................................................................... .... Chapter 1. Introduction.................................................................................. 1.1 Knock in Spark Ignition Engines ........................................................................ 1..1 A utoignitionChem istry........................................................................................ 1.1.2 FactorsAffecting Autoignition in SI Engines.................................................... 1.2 Efficiency Effects of Compression Ratio............................................................27 1.3 Evolving Engine Technologies...................................................................31 1.3.1 Fuels and Compression Ratio ........................................................................... . . ........................... ....... 1.3 .2 B o ost............................................................................. 1.3.3 Em issions Control .............................................................................................. 1.3.4 Charge and Combustion..................................................................................... 1.3.5 Engine Control............................................ 1.3.6 Modern Engine Performance.............................................................................. 5 13 .21 21 22 25 31 31 32 33 34 34 1.4 The Plasmatron Engine System........................................................................36 36 1.4.1 PlasmatronDesign and Operation..................................................................... 38 1.4.2 Benefits of Hydrogen and CarbonMonoxide Enhancement.............................. 38 1.5 Previous Work .................................................................. 1.5.1 Effects ofAir-Fuel Ratio and Fuel Reformate on Knock andAutoignition........38 1.5.2 Effects of OperatingConditionson Knock and Autoignition............................ 39 39 1.5.3 Effects of Compression Ratio on Efficiency ...................................................... ....... 41 1.6 Objectives ..............................................................-. Chapter 2. Experimental Method .................................................................. 43 2.1 Engine Setup................................................................43 2.1.1 Engine and Dynamometer Specifications...........................................................43 45 2.1.2 A ir andFuel Supply Systems.............................................................................. .... 46 2.2 Engine Control and Measurements ................................................... 46 2.2.1 Engine Control Unit ............................................................................................ 47 2.2.2 Fuel Flow Measurement................................................................................ 2.2.3 Intake PressureMeasurement and Control.........................................................49 49 2.2.4 Temperature Measurement and Control........................................................... . 49 2.2.5 A ir Flow Measurement.................................................................................. 2.2.6 A ir-FuelRatio Measurement...........................................................................-49 50 2.2.7 Cylinder PressureMeasurement....................................................................... 50 2.2.8 Knock Detection........................................... 2.3 Changing Compression Ratio...........................................................51 7 2.4 Efficiency Experim ents ........................................................................................ 51 2.4.1 Experimental Procedure.................................................................................... 51 2.4.2 Fuel ........................................................ 52 2.4.3 Operating Conditions and CompressionRatio ................................................. 52 2.5 Knock Limited Minimum Spark Retard Experiments....................52 2.5.1 ExperimentalProcedure.................................................................................... 53 2 .5 .2 Fuels ....................................................................................................................... 54 2.5.3 OperatingConditions and Compression Ratio ................................................. 56 Chapter 3. Experim ental Results .............................................................................. 59 3.1 Efficiency Results............................................................................................... 59 3.1.1 Effects ofAir-Fuel Ratio andLoad ................................................................... 59 3.1.2 Effects of CompressionRatio ............................................................................. 61 3.2 Knock Lim ited Perform ance Results.................................................................. 70 3.2.1 Ignition Timing ....................................................................................................... 3.2.2 Effects ofA ir-FuelRatio.................................................................................... 3.2.3 Effects of Boost ....................................................................................................... 3.2.4 Effects of PlasmatronReformate Addition.......................................................... 3.2.5 Effects of CompressionRatio ............................................................................. Chapter 4. Chem istry Modeling .............................................................................. 70 71 72 73 75 95 4.1 Background..............................................................................................................95 4.2 M odel Description............................................................................................... 96 4.3 Governing Equations .......................................................................................... 96 4.4 Engine Knock Sim ulations ................................................................................. 98 4.4.1 Selection of Input Parameters........................................................................... 4.4.2 Pressure-TemperatureProfiles............................................................................ 99 102 4.4.3 Model Sensitivity .................................................................................................. 106 4.4.4 Comparisonto ExperimentalResults...................................................................107 4.5 Constant Volum e Sim ulations...............................................................................112 4.5.1 Effects of Hydrogen Addition ............................................................................... 113 4.5.2 Effects of Carbon Monoxide Addition..................................................................117 Chapter 5. Engine Optim ization ................................................................................ 121 5.1 Perform ance Analysis............................................................................................121 5.1.1 CompressionRatio Optimization ......................................................................... 121 5.1.2 Fuel Octane Quality and Reformate Addition......................................................123 5.2 Efficiency Analysis ............................................................................................... 126 5.2.1 Effect of CompressionRatio on Brake Efficiency ................................................ 126 5.2.2 Effect of Boost on Brake Efficiency......................................................................128 5.3 Im plications for Engine Design ............................................................................. Chapter 6. Conclusions.............................................................................................. 8 130 131 References .............................................................................. Appendix A Appendix B ..... ...... ... 133 137 M odified Piston Dimensions .......................................................... Additional Charts for Effects of Air-Fuel Ratio on Knock ............ 141 Appendix D Additional Charts for Effects of Boost on Knock .......................... 147 Additional Charts for Effect of Fuel Reformate on Knock ............ 155 Appendix E Additional Charts for Effect of Compression Ratio on Knock ...... 163 Appendix C 9 (this page intentionally left blank) 10 LIST OF TABLES Table 1.1 - Outlet composition of an ideal plasmatron for a fuel H/C ratio of 1.9 and of a 37 typical plasm atron [8]............................................................................................ Table 1.2 - Effect of Engine Operating Conditions on Borderline Knock [16]...........40 Table 2.1 - Test Engine Specifications ....................................................................... 43 Table 2.2 - Summary of experimental engine transducers and gauges ......................... 48 55 Table 2.3 - U TG fuel properties ................................................................................... Table 4.1 -Temperatures at intake valve close estimated by WAVE engine model.......102 Table 4.2 - Sensitivity of autoignition time to model input parameters. AMign-pp is the distance in crank angle degrees from autoignition to peak-pressure.......................108 Table 5.1 - Calculated compression ratio for peak torque at 1500 rpm..........................123 Table 5.2 - Effect of downsizing with increased rc on brake efficiency.........................127 Table 5.3 - Effect of boosting and downsizing on mid-load brake efficiency................129 11 (this page intentionally left blank) 12 LIST OF FIGURES 23 Figure 1.1 - Examples of typical hydrocarbons ............................................................ Figure 1.2 - Pressure measurements from autoignition of PRFs in a rapid compression machine; fuel-air equivalence ratio: 0.4, initial temperature: 318K, initial pressure: 24 lbar, com pression ratio: 16:1 [3].......................................................................... Figure 1.3 - Alkane oxidation scheme proposed by Tanaka et al. [3]..........................24 Figure 1.4 - Impact of spark timing on cylinder pressure, 1500 rpm, WOT, X = 1.0.......26 Figure 1.5 - Indicated efficiency for the ideal gas, constant heat capacity, constant v olum e cy cle .............................................................................................................. 28 Figure 1.6 - Fuel-air results for indicated fuel conversion efficiency as a function of compression ratio. Fuel: octene; pi = 1 atm, T, = 388 K, xr = 0.05. [1]...............29 Figure 1.7 - Relative fuel conversion efficiency improvement with increasing compression ratio at wide-open throttle (data from two sources) [1]....................30 Figure 1.8 - Conceptual diagram of maximum torque curves and normal operating range of naturally aspirated and boosted downsized engines..........................................32 Figure 1.9 - Approximate maximum BMEP (normalized torque) curves for typical naturally aspirated and boosted engines .............................................................. Figure 1.10 - Schematic of the plasmatron, courtesy of A. Rabinovich [8] ................. 35 36 37 Figure 1.11 - Conceptual plasmatron engine system configuration ............................ Figure 1.12 - Lower octane fuel is supplied to the engine for audible knock when some energy is derived from H2 and CO (plasmatron reformate). Data simulates 15% and 30% of the gasoline being reformed in the plasmatron fuel reformer [14]. .......... 39 Figure 1.13 - Brake efficiency improvement with compression ratio for engines of 40 several swept volum es [17]. ................................................................................. Figure 2.1 - Base, medium, and high compression ratio pistons (left to right).............44 Figure 2.2 - Illustration of the 13.4:1 piston and cross sections of the combustion chamber 44 at top center piston position................................................................................... 45 Figure 2.3 -Intake air system ....................................................................................... Figure 2.4 - Liquid fuel system ..................................................................................... 46 Figure 2.5 - Gaseous fuel system (simulated plasmatron gas supply)...........................46 Figure 2.6 - Engine control and measurement diagram.................................................47 52 Figure 2.7 - Test matrix for efficiency experiments ..................................................... Figure 2.8 - RON and MON of toluene reference fuels as a function of volume fraction nheptane. Compiled by Shell Global Solutions [18] from ASTM vol 05.04, Table 28 56 and internal data...................................................................................................... Figure 2.9 - Test matrix for knock experiments ............................................................ 57 13 Figure 3.1 - Change of net indicated efficiency with lambda; 4.Obar NIMEP..............61 Figure 3.2 - Normalized change of net indicated efficiency with lambda; 4.Obar NIMEP. 62 ................................................................................................................................... Figure 3.3 - Change of gross efficiency with lambda; 4.Obar NIMEP..........................62 Figure 3.4 - Combustion event timing; 4.Obar NIMEP, r, = 9.8:1.................................63 Figure 3.5 - Combustion event timing; 4.Obar NIMEP, k = 1.6....................................63 Figure 3.6 - Change of net indicated efficiency with NIMEP; k = 1.0. ........................ 64 Figure 3.7 - Change of net indicated efficiency with NIMEP; k 64 = 1.3. ........................ Figure 3.8 - Normalized change of net indicated efficiency with NIMEP; k = 1.0. ......... 65 Figure 3.9 - Normalized change of net indicated efficiency with NIMEP; k = 1.3. ......... 65 Figure 3.10 - Change of gross indicated efficiency with NIMEP; k = 1.0....................66 Figure 3.11 - Change of gross indicated efficiency with NIMEP; k = 1.3....................66 Figure 3.12 - Change of net indicated efficiency with compression ratio for a range of k; 4.Obar N IME P ........................................................................................................ 67 Figure 3.13 - Normalized change of net indicated efficiency with compression ratio for a range of k; 4.Obar N IM EP. .................................................................................... 67 Figure 3.14 - Change in net indicated efficiency with compression ratio for a range of loads; k = 1.0. ........................................................................................................ . 68 Figure 3.15 - Change in net indicated efficiency with compression ratio for a range of load s; k = 1.3 . ........................................................................................................ . 68 Figure 3.16 - Normalized change in net indicated efficiency with compression ratio for a range of loads; =1 .0 ............................................................................................ 69 Figure 3.17 - Normalized change in net indicated efficiency with compression ratio for a range of loads; k= 1.3 ............................................................................................ 69 Figure 3.18 - Change of NIMEP with spark timing for a range of k; rc= 9.8:1, 1500 rpm. ................................................................................................................................... 77 Figure 3.19 - Normalized decrease of NIMEP with spark retard for a range of k; rc = 9.8:1, 1500 rpm, 10.1 bar NIMEPMBT- .------------------.-------................................-------- 77 Figure 3.20 - Normalized decrease of NIMEP with spark retard for a range of ) and rc; toluene fuel, 1500 rpm, MAP at k > 1.0 boosted to match MBT NIMEP at unboosted WO T X = 1.0. ....................................................................................... 78 Figure 3.21 - Change of combustion retard with spark retard for a range of k; rc = 9.8:1, 1500 rpm, 10.1 bar NIMEPMBT ...........------..---------.....-..-- .....--.....- 78 Figure 3.22 - Change of normalized NIMEP with spark retard for a wide range of operating conditions and compression ratios; toluene fuel except where noted, 1500 rp m ............................................................................................................................. 79 14 Figure 3.23 - Change of normalized NIMEP with combustion retard for a wide range of operating conditions and compression ratios; toluene fuel except where noted, 1500 .....-7 9 rp m ................................................................................................................------. Figure 3.24 - Averaged heat release rate profiles for several spark timings for toluene and for TRFs with octane numbers that result in near knocking conditions; rc = 9.8:1, X = 80 1.3, 1500 rpm, MAP = 1.23bar (for NIMEPMBT= 10.1bar). ................................ Figure 3.25 - Change of combustion retard with spark retard for toluene and for TRFs with octane numbers that result in near knocking conditions; rc = 9.8:1, X = 1.3, 1500 80 rpm, MAP = 1.23bar (for NIMEPMBT= 10.1 bar). ................................................ Figure 3.26 - Effects of X and PRF fuel RON on combustion retard to just avoid knock; 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted W O T ...................................................................................................... 81 Figure 3.27 - Effects of X and TRF fuel RON on combustion retard to just avoid knock; 1500 rpm, MAP at %> 1.0 boosted to match MBT NIMEP at stoichiometric 81 unboosted WO T .................................................................................................... Figure 3.28 - Effect of X on combustion retard to just avoid knock for UTG fuels; r, = 11.6:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric 82 unboosted W OT............................................................................................. Figure 3.29 - Change of combustion retard to just avoid knock with increasing X; rc = 9.8:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted W OT........................................................................................82 Figure 3.30 - Change of combustion retard to just avoid knock with increasing /; rc = 11.6:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric 83 unboosted W O T ...................................................................................................... Figure 3.31 - Change of combustion retard to just avoid knock with increasing X; rc = 13.4:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric .83 unboosted WO T ................................................................................................... Figure 3.32 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; rc = 11.6:1, X = 1.0, 1500 rpm........................................84 Figure 3.33 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; rc = 11.6:1, X = 1.0, 1500 rpm........................................84 Figure 3.34 - Lines of constant MAP calculated from Eq. (3.1) imposed on data from Figure 3.33. Horizontal lines are drawn at NIMEPMBT. Vertical distance between curves and lines is torque loss from retarding spark to avoid knock.....................85 Figure 3.35 - Effects of boosted NIMEP and PRF fuel RON on combustion retard to just 85 avoid knock; X = 1.0, 1500 rpm ............................................................................. Figure 3.36 - Effects of boosted NIMEP and TRF fuel RON on combustion retard to just 86 avoid knock; X = 1.0, 1500 rpm ............................................................................. 15 Figure 3.37 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for UTG96, PRF95 and TRF95; rc = 11.6:1, A = 1.0, 1500 rpm........86 Figure 3.38 - Increase of combustion retard with increased airflow rate from boosting at 1500 rpm, PRFs at k = 1.0 and ) = 1.3, TRFs at k = 1.0 and k = 1.3, and UTG96 at k = 1.3. Data is from all three compression ratios............................................... 87 Figure 3.39 - Increase of combustion retard with increased NIMEP from boosting at 1500 rpm, PRFs at k = 1.0 and k = 1.3, TRFs at k = 1.0 and k = 1.3, and UTG96 at k = 1.3. Data is from all three compression ratios............................................... 87 Figure 3.40 - Decrease of combustion retard to just avoid knock and associated NIMEP increase with increased reformed fuel fraction for PRFs; r, = 11.6:1, k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). ----......... 88 Figure 3.41 - Decrease of combustion retard to just avoid knock and associated NIMEP increase with increased reformed fuel fraction for TRFs; rc = 11.6:1, k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT) .----......... 88 Figure 3.42 - Increase of combustion retard with boost and decrease of combustion retard with reformate fraction for TRF95; r, = 11.6:1, k = 1.0, 1500 rpm. Curve for NIMEP vs. combustion retard at 40% boost calculated from Eq. (3.1). ............... 89 Figure 3.43 - Increase of combustion retard with rc and decrease of combustion retard with reformate fraction for TRF95; k = 1.3, 1500 rpm, MAP set to match NIMEPMBT at unboosted WOT k = 1.0. Curve for NIMEP vs. combustion retard for 89 rc = 13.4:1 calculated from Eq. (3.1)...................................................................... Figure 3.44 - Decrease of combustion retard to just avoid knock with increased reformed fuel fraction for UTG fuels with PRFs and TRFs for comparison; rc = 11.6:1, k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT).90 Figure 3.45 - Effects of reformed fuel fraction and PRF fuel RON on combustion retard to just avoid knock; k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIM EPMBT). ----...........................--------.................................................. 90 Figure 3.46 - Effects of reformed fuel fraction and TRF fuel RON on combustion retard to just avoid knock; k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIM EPMBT). ..------------------. --------------------.................................................. 91 Figure 3.47 - Decrease of combustion retard with increased reformed fuel fraction at 1500 rpm, PRFs at k = 1.0 (40% boost) and k = 1.3, TRFs at k = 1.0 (40% boost) and %=1.3,and UTGs at k = 1.3. Data is from all three compression ratios........91 Figure 3.48 - Increase of unboosted NIMEP at MBT spark timing with r. for three fuel types; k = 1.0, 1500 rpm. Closed symbols represent the first high load runs with new pistons, open symbols represent runs made after several engine hours at high lo ad . ........................................................................................................................... 92 Figure 3.49 - Normalized average increase of WOT NIMEP at MBT timing with rc; %= 1.0, 1500 rpm. Raw data is from Figure 3.48 ..................................................... 92 16 Figure 3.50 - Increase of combustion retard to just avoid knock and change of NIMEP with increased rc for PRFs; k = 1.0, 1500 rpm, unboosted WOT..........................93 Figure 3.51 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; k = 1.0, 1500 rpm, unboosted WOT. ........................ 93 Figure 3.52 - Curves for NIMEP vs. combustion retard from Eq. (3.1) imposed on data from Figure 3.51. Vertical distance between curves and horizontal lines at NIMEPMBT is torque loss from retarding spark to avoid knock. .......................... 94 Figure 3.53 - Average increase in combustion retard to avoid knock over all operating conditions considered, organized by fuel type and compression ratio interval. Bars 94 represent one standard deviation. ......................................................................... Figure 4.1 - Structure of an autoignition simulation. .................................................... 97 Figure 4.2 - A set of 90 cycles of pressure data taken under near knocking conditions with PRF90; 1500rpm, WOT, X = 1.0. The solid lines are cycles that have a location of 50% mass fraction burned that is earlier than 90% of the cycles........................100 Figure 4.3 - Residual fractions calculated by a Ricardo WAVE engine simulation; X = 1.3, 1500 rpm , M A P = 1.22 bar...............................................................................101 Figure 4.4 - Comparison of measured cylinder pressure to polytropic compression to find effective intake valve closing time; 1500 rpm, X = 1.5. .......................................... 101 Figure 4.5 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for stoichiometric and lean air-fuel ratios; iso-octane fuel, 1500 rpm, rc = 9.8:1, NIMEP = 10.1 bar, MBT spark timing. ........................................ 103 Figure 4.6 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for unboosted and boosted conditions; iso-octane fuel, 1500 rpm , rc = 9.8:1, k = 1.0, M BT spark tim ing.............................................................104 Figure 4.7 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for low and high compression ratios; iso-octane fuel, 1500 rpm, 104 M A P = 1 bar, M BT spark tim ing. ........................................................................... Figure 4.8 -Predicted non-reacting end-gas temperature profiles with the same pressure profile for PRF100 and TRF100 fuels; 1500 rpm, rc = 9.8:1, MAP = 1.0 bar, k = 1.0, M B T spark tim ing....................................................................................................105 Figure 4.9 - Predicted end-gas temperature profiles with the same pressure profile for simulations with reaction kinetics disabled and enabled for PRF95; 1500 rpm, r, = 9.8:1, M AP = lbar, X = 1.0, Osp = 60 BTC...............................................................106 Figure 4.10 - Effect of initial temperature on predicted autoignition time in increments of 5 K for PRF95; 1500 rpm rc = 9.8:1, MAP = Ibar, X = 1.0, Osp = 6' BTC. ............ 107 Figure 4.11 - Pressure traces from several spark timings and corresponding simulated end-gas temperature profiles; 1500 rpm, rc = 11.6:1, k = 1.0, MAP = 1.Obar. 4 BTC is interpreted as the predicted knock limited spark timing......................................109 17 Figure 4.12 - Comparison of model predicted combustion retard to experimental combustion retard with increasing compression ratio for PRF95; 1500 rpm, X = 1.0, 109 M AP =1 .Obar. ......................................................................................................... Figure 4.13 - Comparison of model predicted combustion retard to experimental combustion retard with increasing compression ratio for TRF95; 1500 rpm, X = 1.0, 1 10 M AP = 1.0 b ar. ........................................................................................................ Figure 4.14 - Comparison of model predicted combustion retard to experimental combustion retard with increasing air-fuel ratio for PRF95; 1500 rpm, rc = 9.8:1, N IM E PM BT 10.1 bar. ............................................................................................. 110 Figure 4.15 - Comparison of model predicted combustion retard to experimental combustion retard with increasing air-fuel ratio for TRF95; 1500 rpm, rc = 9.8:1, N IM E P M BT = 10.1 bar..............................................................................................111 Figure 4.16 - Comparison of model predicted combustion retard to experimental combustion retard with increasing boosted NIMEP for PRF95; 1500 rpm, r, = 9.8:1, 11 1 X = 1 .0 ...................................................................................................................... Figure 4.17 - Comparison of model predicted combustion retard to experimental combustion retard with increaing boosted NIMEP, PRF95; 1500 rpm, rc = 9.8:1, X = 1 12 1 .0 . ........................................................................................................................... Figure 4.18 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane and of iso-octane with 5.5% H 2 by energy; Tinit = 875 K, P init = 4 5 bar, k = 1.5................................................................................................113 Figure 4.19 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated non-reactive H 2 (nrH 2); Tinit = 875 K, Pinit = 45 bar, X = 1.5....................................114 Figure 4.20 - OH consumption by reactions with fuel molecules for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated non-reactive H 2 (nrH 2); Tinit = 875 K, P init= 45 bar, k = 1.5. .......... 115 Figure 4.21 - Rates of reaction of OH and H with iso-octane molecules for constantvolume autoignition simulations of iso-octane with 5.5% H 2 by energy, and of isooctane with simulated non-reactive H2 (nrH 2); Tinit = 875 K, Pinit= 45 bar, k = 1.5. ................................................................................................................................. 115 Figure 4.22 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated H 2 (prH2 ) that reacts only with OH; Tinit = 875 K, Pinit = 45 bar, X = 1.5.................116 Figure 4.23 - Proposed mechanism by which hydrogen impacts the autoignition of alkane hydrocarbons. Base diagram is from Tanaka et al. [3]...........................................117 Figure 4.24 - Comparison of the reaction rate constant for CO+02 => CO2+O from three sources. The rate from Scire et al. [26] was used in this study...............................118 18 Figure 4.25 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane and of iso-octane with 7% CO by energy; Tiit = 875 K, P init= 45 bar, k = 1.5................................................................................................118 Figure 4.26 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 7% CO by energy, and of iso-octane with simulated non-reactive CO (nrCO); Tinit = 875 K, Pinit = 45 bar, k = 1.5. ................................ 119 Figure 5.1 - Tradeoff between NIMEP and compression ratio. Values shown approxim ate TRF95 fuel..........................................................................................122 Figure 5.2 - Contours of constant combustion retard for varying reformate fraction and fuel RON; k = 1.0, 1500 rpm, MAP for 40% boost. Data from Figure 3.45 (PRFs, 124 left) and Figure 3.46 (TRFs, right). ......................................................................... Figure 5.3 - Contours of constant combustion retard for varying boosted NIMEP and fuel RON; k = 1.0, 1500 rpm. Data from Figure 3.35 (PRFs, left) and Figure 3.36 12 5 (TRF s, right). ........................................................................................................... Figure 5.4 - Contours of constant combustion retard for varying compression ratio and fuel RON; k = 1.0, 1500 rpm, MAP = 1.0 bar.........................................................125 Figure 5.5 - Estimated increase of mid-load brake efficiency with re, with and without downsizing the engine to maintain constant maximum torque output; 1500 rpm, k = 1.0, 2.6 bar baseline B M EP.....................................................................................128 Figure 5.6 - Estimated increase of mid-load net and brake efficiencies with boosting and downsizing; 1500 rpm, k = 1.0, 2.6 bar baseline BMEP.........................................129 19 (this page intentionally left blank) 20 CHAPTER 1. INTRODUCTION Knock in spark ignited internal combustion engines is the external sound produced by pressure oscillations in the cylinder caused by spontaneous autoignition of the air-fuel mixture before it is consumed by the propagating flame front [1]. The knocking noise is unacceptable in today's refined vehicles, and repeated exposure to the extreme local pressures and temperatures caused by very rapid autoignition are damaging to engine components. Since the inception of the spark ignited internal combustion engine, knock has been one of the fundamental limiting factors. Historically, its main effect has been to limit the maximum compression ratio, and thus the efficiency, of automotive engines. As fuel quality and engine technology have improved, engine compression ratios have increased to the point that further increases result in smaller efficiency improvements. Currently, the main effect of knock is reduced low speed torque - due to spark retard on naturally aspirated engines, and limited boost on turbocharged or supercharged engines. The motivation of this work is to investigate the tradeoffs between fuel type, compression ratio, boost level, air-fuel ratio, and spark timing at the knock limit such that informed decisions can be made during the selection of new engine configurations. 1.1 KNOCK IN SPARK IGNITION ENGINES The term "knock" describes the sound resulting from the spontaneous ignition of the unburned air-fuel mixture ahead of the advancing turbulent flame front. The unburned mixture is compressed by the expansion of the burned gases produced in the turbulent flame propagating from the spark plug outwards. The increases in unburned mixture temperature and density associated with this pressure rise initiate the chemical reactions that lead to autoignition. If the flame does not consume the unburned fuel, or "end-gas", before the chemical reactions proceed to the point of rapid bulk heat release, the resulting autoignition causes local temperatures and pressures to increase sharply. Simulations performed at Leeds University [2] indicate that the local pressure may rise to upwards of 300 atmospheres in as little as one microsecond. A pressure wave then propagates outwards from the location of autoignition at the local speed of sound and resonates within the engine cylinder. The pressure oscillations cause vibration of the head and block, resulting in a clanging or pinging sound. While light autoignition can be 21 functionally acceptable, audible knock is disagreeable, and the high local temperatures and pressures associated with heavy knock cause engine damage. Since knock can damage the engine and produces a disagreeable sound, engine design and operating parameters are adjusted for its avoidance. Knock is mainly a problem during high load, which creates high peak pressures, and low engine speed, which provides a long residence time for the end-gas. To avoid knock under these conditions engine control systems typically retard spark timing to lowers peak pressure, but high spark retard results in a significant decrease in engine torque and combustion stability and an increase in exhaust temperatures. To avoid the need for excessive spark retard, compression ratio is limited, and for turbocharged or supercharged engines boosted inlet pressure is limited. 1.1.1 Autoignition Chemistry Gasoline consists of approximately 60% alkane hydrocarbons and 30% aromatic (cyclic) hydrocarbons, with the remainder composed of alkenes, alcohols, and additives. Figure 1.1 shows some examples of typical hydrocarbons. The low temperature (less than 1000K) chemistry of alkanes larger than C5 are most relevant to gasoline autoignition, so they are the topic of most work on the subject. Figure 1.2 shows the evolution of pressure in the combustion chamber of a rapid compression machine for several mixtures of primary reference fuels (iso-octane and n-heptane). Two-stage autoignition, an important characteristic of alkane autoignition, is clearly visible. The increasing duration of the second stage with octane number results in an overall increase in autoignition time. Three sequences are responsible for the alkane autoignition process.[3] A highly exothermic chain-branching reaction sequence is responsible for the first stage: RH+OH+202 =>P +20H In this sequence of reactions, a hydroxy radical combines with oxygen and a fuel molecule to produce two hydroxy radicals, which carry on the sequence. As the temperature rises, a weakly exothermic competing reaction sequence slows down the temperature rise, causing a transition to the second stage: 2RH+ 20H + Q2 = P +H202 22 In this sequence of reactions hydroxy radicals are tied up into hydrogen peroxide. Finally, as temperature and radical concentrations rise, decomposition of the hydrogen peroxide initiates rapid exothermic combustion (autoignition): H2 0 2 ->2OH Alkanes: CH3 C I I CH3 -C-CH 2 C-CH 3 C CH3 CH3-CH 2-CH2 -CH 2-CH 2-CH 2 -CH 3 Isooctane (2,2,4-Trimethylpentane) ON=100 Aromatics: n-Heptane ON=0 /CH 2 /CH 3 /CH-C\ H CH CH-CH Toluene RON-120 /CH-C CH H CH 3 CH-CH Ethylbenzene Figure 1.1 - Examples of typical hydrocarbons Figure 1.3 shows the three sequences in an overall autoignition scheme. The lower loop is the chain branching sequence. The fuel molecule is attacked by a hydroxy radical and then combines with oxygen and undergoes internal isomerization before releasing two hydroxy radicals. The upper loop is the high temperature mechanism. At higher temperatures, after being attacked by the hydroxy radical, the alkyl radical reacts with oxygen to produce olefin and an HO 2 radical, which combines to produce hydrogen peroxide. As temperatures rise, the built up hydrogen peroxide molecules dissociate into hydroxy radicals and autoignition occurs. Although the typical trend is for decreasing time to autoignition with increased temperature, variations in the rates of intermediate reactions result in complex variations of autoignition time. For alkanes such as n-heptane and iso-octane there is a temperature interval (roughly 750 K to 850 K), called the Negative Temperature Coefficient (NTC) region, for which autoignition time increases with temperature. Mixtures of fuels show more complicated variations of autoignition time with temperature. Although 23 experimental ignition times give an indication of a fuel's autoignition properties, due to transient pressures and temperatures in SI engine end-gas, they do not relate directly to knock propensity. Prm---e-r---F-ldTac B (K) 7 1393 6 1194 0~ 4 31 5~.995 75 90100 7so-Ocane) ON=O0 (n-Heptane 795 597 2 398 1 19 0 .5 5 10 time (msec) Is 2 Figure 1.2 - Pressure measurements from autoignition of PRFs in a rapid compression machine; fuel-air equivalence ratio: 0.4, initial temperature: 318K, initial pressure: lbar, compression ratio: 16:1 [3]. High Temperature -H02 H202 Olefin 02 ----- - - R-----RH+0 2R V- M RH - H O H20 02 IROO hr erazakat Low Temperature -ROOH OH 02 -OOROOH OROOH *ORO Figure 1.3 - Alkane oxidation scheme proposed by Tanaka et al. [3] 24 The system most commonly used to rate the knock propensity of SI engine fuels is the octane rating method. Fuels are compared to Primary Reference Fuels (PRFs) in a standardized single cylinder engine under two operating conditions. American Society for Testing and Materials standard ASTM D-2699 specifies the research method, which measures the RON (Research Octane Number). ASTM D-2700 specifies the motoring method, which gives the MON (Motoring Octane Number). For each test the compression ratio is increased until the engine just knocks with the fuel under test. The Octane Number (ON) of the fuel is the ON of the PRF that just knocks under the same conditions. The motoring method is performed with higher inlet temperature, higher speed, and more spark advance than the research method. For most fuels, knock propensity increases more with severity of operating conditions than it does for PRFs, thus the MON is typically lower than the RON. 1.1.2 Factors Affecting Autoignition in SI Engines Under most conditions heat release from the end-gas prior to autoignition is small, so compression and normal combustion dictate the evolution of cylinder pressure. Cylinder pressure, however, has a significant effect on end-gas thermodynamic state and reaction rates. An estimate of how pressure affects end-gas temperature can be gained by looking at the P-T relationship for adiabatic compression: T= Tinitial ( ( Pinitial / Higher pressures result in higher temperatures and increased reaction rates. Factors that affect cylinder pressure are: Inlet pressure - Controls the amount of air-fuel mixture entering the cylinder. Higher inlet pressure increases engine torque output and increases cylinder pressures. A ir-fuel ratio- Affects combustion rates and the energy released during combustion. Dilution from excess air and changes in thermodynamic properties of the mixture with air-fuel ratio affect how the heat release impacts pressure evolution. Spark timing - Affects combustion phasing. Late combustion phasing results in lower cylinder pressures because the bulk of the combustion process occurs in a larger, faster expanding volume. Figure 1.4 shows how cylinder pressures change with spark timing. 25 60 Spark Timing 50- 50% mass... 40 - burned 30 - -16deg -8deg BTC -4deg ATC BTC (MBT) - 20 10 - Opark A -45 0 Crank Angle 45 90 (0ATC) Figure 1.4 - Impact of spark timing on cylinder pressure, 1500 rpm, WOT, X = 1.0 Compression ratio- The ratio of the maximum cylinder volume to minimum cylinder volume. As compression ratio is increased the cylinder volume before and during combustion becomes smaller, resulting in higher cylinder pressures. Engine speed - The duration of the engine cycle scales inversely with engine speed. As speed increases, the duration of time for which unburned end-gas is subjected to high pressures and temperatures decreases. Chargepreparation- Turbulence and mixture homogeneity affect speed of combustion. As combustion duration decreases the end-gas is consumed faster, resulting in less time for autoignition reactions to occur. Combustion chamber geometry -Longer flame travel distances result in longer burn duration and more time for autoignition to occur. Factors that affect the evolution of temperature and the chemical composition of the endgas in response to the evolution of pressure are: 26 Mixture composition - The ratio of specific heats (y) is an important parameter in Eq. ( 1.1). Higher y results in higher compression temperatures. Air has a higher ratio of specific heats (y = 1.4) than gasoline (y~1.0) and burned gas (y~1.3), so increased air-fuel ratio increases y and increased residual fraction reduces y. Reactant concentrations, fuel types, and fuel additives also have important effects on autoignition reaction rates. Initial temperature- End-gas compression temperature scales closely with initial temperature, as indicated by Eq. ( 1.1). Higher initial temperatures result in higher unburned gas temperatures throughout the compression and combustion processes, increasing the rates of autoignition reactions. Initial temperature is affected by inlet air temperature, heat transfer, and the temperature and quantity of residual gas. Heat transfer during compression and combustion affects both the evolution of pressure and the temperature of the end-gas. During the first part of compression heat is transferred to the unburned gas. During the second part of compression and during combustion heat transfer from the burned gas reduces cylinder pressure and heat transfer from the end-gas reduces its temperature and slows autoignition reactions 1.2 EFFICIENCY EFFECTS OF COMPRESSION RATIO To avoid knock, the maximum compression ratio of an engine must be limited. The improvement of engine efficiency with compression ratio is a key motivation behind work to improve knock performance. This section gives an overview of the effect of compression ratio on engine efficiency. An illustrative model of an SI engine cycle is the ideal gas, constant heat capacity, constant volume heat addition cycle. In this cycle the working fluid is compressed isentropically, heat is added at constant volume, and then work is extracted as the working fluid is isentropically expanded. A constant volume heat rejection (fluid exchange) process completes the cycle. For this cycle the efficiency is: qf/ig = 1- 1 (1.2) rrl The results of this equation are shown in Figure 1.5. Efficiency increases with 7 and increases with rc with decreasing returns at high rc. This cycle is illustrative, but it is missing many real-engine effects. 27 0.75 y = 1.4 0.7 - (air) 0.65 y = 1.3 0.6 0.55 ' 0.5 y=1.2 0.45 0.4 0.35 0.3 0.25 0 5 10 15 20 25 30 rc Figure 1.5 - Indicated efficiency for the ideal gas, constant heat capacity, constant volume cycle For a more accurate estimate of real engine performance, the effects of fluid properties and combustion chemistry must be taken into account. The working fluid does not have constant specific heats and its composition changes during the cycle. Combustion does not go to completion - the burned gases are approximately in equilibrium at high temperatures and have a frozen composition at low temperatures. The fuel-air cycle takes these effects into account [1]. The results of the air-fuel cycle can be calculated using fits to thermodynamic data and equilibrium computer code. Figure 1.6 shows indicated efficiency as a function of compression ratio for a series of fuel/air equivalence ratios ( = (F/A)/(F/A)stoich). Lines of constant j are slightly lower than those predicted by Eq. ( 1.2) from the ideal cycle. Efficiency increases with decreasing 4 due to an increased value of burned gas y (from decreased temperature and an increased proportion of diatomic gases). The higher y causes the burned gases to expand through a greater temperature ratio for a given expansion ratio, increasing work output [1]. Efficiency decreases with increasing * on the rich side because the fuel is not fully oxidized due to insufficient air. 28 0.65 -- 0.4 0.5 1.0 at0.6 -. - =388K p, = 0.60 81.0 x, = 0.05 - - -- 0. 55 - 0.50 1.2 - 7)f i 0.45 ---- 40 - 0. 35 - 0.25 1.4 // 12'1 0 5 10 I 20 15 Compression ratio r 25 30 Figure 1.6 - Fuel-air results for indicated fuel conversion efficiency as a function of compression ratio. Fuel: octene; p1 = 1 atm, T1 = 388 K, xr = 0.05. [1] In a real spark ignited engine several other factors impact the effect of compression ratio on efficiency. Figure 1.7 shows a comparison of real engine performance (circa 1960, there is a lack of data for a wide range of rc in modem engines) and the fuel-air cycle results. At low rc the experiments match the air-fuel cycle well. At high rc, heat transfer, crevice effects, and friction cause efficiency to decrease. Under normal operation, heat transfer from the combustion chamber corresponds to roughly 30% of the fuel energy, similar to the fuel conversion efficiency of the engine. Thus, small changes in heat transfer can have a great impact on engine performance. Increasing the compression ratio decreases temperatures late in the expansion stroke because more energy is removed as work. However, increasing rc increases temperatures during combustion and early expansion, and increases the heat transfer coefficient due to increases in density and mixture motion. Changes in combustion chamber surface area also affect heat transfer, but are dependent on engine geometry. The combined effect is that heat transfer decreases up to a rc of about 10:1 and increases above that [1]. 29 1.3 Fuel-air cycle 1.2 Git 93 ;(CN) 1.1- Wide-open throttle 1.0I. 8 I 10 I 12 I 14 I 16 I 18 I 20 I 22 24 Compression ratio rc Figure 1.7 - Relative fuel conversion efficiency improvement with increasing compression ratio at wide-open throttle (data from two sources) [1]. Crevices between the piston and the bore, above and behind the top compression ring, and around the valves and spark plug comprise approximately 2-3% of the combustion chamber clearance volume. The fraction of the charge that gets trapped in the crevices during combustion is even greater because the cold unburned gas in the crevices is much more dense than the hot burned gas in the cylinder. The flame does not propagate into the crevices and thus detracts from the energy extracted from the charge during the main combustion event. As the combustion chamber volume is decreased to increase compression ratio the crevice volume becomes a greater fraction of the total clearance volume, and higher cylinder pressures increase the density of the mixture in the crevices. The higher mass trapped in the crevices leads to reductions in efficiency. Increasing the compression ratio increases pressures in the cylinder during compression, combustion, and expansion. This increases the force between the piston and the bore, which increases sliding friction, and between the piston and the connecting rod, which increases friction at the wrist pin and crankshaft bearings. 30 1.3 EVOLVING ENGINE TECHNOLOGIES The spark ignition engine is being steadily improved upon. From 1984 to 2000, maximum torque normalized by engine displacement has increased by about 1.5% per year [4]. Engine performance has also been impacted by government emissions regulations and, more recently, by vehicle fuel efficiency standards. The technologies employed to increase torque output and meet government regulations has important implications on engine configuration, operating parameters, and knock control. 1.3.1 Fuels and Compression Ratio From the 1920s to the 1960s steady improvements in fuel octane quality allowed SI engine compression ratios to rise steadily [5]. The improvements have been made through better hydrocarbon selection and processing, and through addition of octane enhancers such as TEL (tetraethyl lead) and, more recently, oxygenates like MTBE (methyl tert-butyl ether). Since the 1960s fuel octane quality has been relatively constant and variations in r, have been in response to changes in regulations and engine technology. Compression ratios of modem naturally aspirated SI engines are steadily increasing at about 0.1 units every three years, and are currently between 9.5:1 and 11:1. 1.3.2 Boost Historically inlet pressure boosting has been used to increase the power output of high performance engines. Increasingly, most notably in Western Europe, smaller boosted engines are being used as replacements for larger naturally aspirated engines. The most common method of boosting inlet pressure is through turbocharging, where energy in the exhaust gas is extracted by a turbine and used to power an inlet air compressor. Most modem turbochargers have variable geometry nozzles or turbines for a greater dynamic range and faster throttle response. An intercooler is usually used with boosted engines to reduce the temperature of the compressed inlet air before it enters the engine cylinder. Downsizing an engine requires it to operate at higher specific loads under normal driving conditions. Figure 1.8 shows the increased BMEP range for a downsized engine (BMEP is Brake Mean Effective Pressure, a measure of torque output normalized by engine displacement). Engine downsizing improves efficiency by reducing throttling losses, friction, and heat transfer. To maintain acceptable maximum torque in a downsized engine, inlet boosting is required. To avoid knock, however, boosted engines typically 31 require a reduced compression ratio. In practice, boosted downsized engines have moderate efficiency benefits. If an engine could be boosted without reducing part-load compression ratio, through methods such as compression variation or fuel reforming, the efficiency benefits of boosting and downsizing could be improved significantly. 16 CU ~12 E Normal operating range for boosted and downsized engine. 8a) Normal operating range for 0 -Baseline .4-apiated engine. - NA Boosted 0 0 3500 7000 Speed (RPM) Figure 1.8 - Conceptual diagram of maximum torque curves and normal operating range of naturally aspirated and boosted downsized engines. 1.3.3 Emissions Control Since the 1960s emissions regulations have imposed important restrictions on SI engine operation. To meet current emission standards engine designers employ a combination of emissions reduction techniques. Improvements in air-fuel ratio control and reductions of crevice volume reduce emissions of hydrocarbons and carbon monoxide. Exhaust Gas Recirculation (EGR) reduces the temperature of the burned gas and slows NOx production. Additionally, virtually every SI engine powered vehicle in developed countries comes with a 99% efficient Three-Way Catalyst (TWC) to oxidize unburned hydrocarbons (HC) and carbon monoxide (CO) and reduce oxides of nitrogen (NOx). Efficient operation of a TWC requires restrictions to be placed on engine operating parameters, the most important of which is a stoichiometric air-fuel ratio. Significant 32 efficiency benefits can be realized by running an engine lean of stoichiometric, but under oxygen-rich conditions the TWC is unable to reduce NOx. If the engine is run sufficiently lean, NOx emissions drop to the point that further reductions are not necessary. With current technology, however, the high air-fuel ratio required is in excess of the "lean limit" - the limit beyond which combustion stability becomes unacceptable. Research is underway at MIT [6] and elsewhere to extend the lean limit using charge motion and fuel reforming. Additionally, lean NOx reduction technology currently under development for diesel engines may eventually make its way to SI engine applications. TWCs also place limits on maximum exhaust temperature during full load operation, limiting maximum spark retard. Enriching the air-fuel ratio at high loads is commonly used to increase the heat capacity of the burned gas and reduce the exhaust temperature. TWCs also place restrictions on fuel composition. Impurities such as sulfur and additives such as octane-enhancing TEL poison catalytic converters and reduce their efficiency over time. 1.3.4 Charge and Combustion As described in the previous section, most SI engines run a stoichiometric air-fuel ratio over much of the engine map. Emissions requirements prevent lean operation, but enrichment is sometimes used at high loads to decrease knock tendencies and reduce exhaust temperature. EGR is commonly used for emissions and efficiency benefits. There are several ways to introduce fuel into the inlet air. The most common is sequential port injection. In this configuration fuel injectors at the inlet ports are directed at the back of the intake valves. Increasingly common is Gasoline Direct Injection (GDI). With GDI, the fuel is injected directly into the cylinder during the intake or compression strokes. In the USA GDI strategies are limited to stoichiometric, usually early injection. In places where NOx regulations are not as stringent, lean stratified charge late injection strategies may be used. Under high loads, early injection while the intake valve is still open cools the charge. This increases its density, allowing more airfuel mixture into the cylinder, and slows autoignition reactions, allowing a higher compression ratio or increased boost. Most modern SI engines employ some kind of charge motion control to improve mixing and combustion and reduce cycle-to-cycle variability under low load conditions. Charge motion has also been used to improve knock performance by speeding up combustion, 33 thus reducing end-gas residence time [7], and can be used to extend the lean limit [6]. Methods for introducing charge motion include adjustable vanes in the inlet manifold close to the intake port, valve masking, and variable valve actuation. Intake tuning adds significantly to modem SI engine performance. Most engines have intake runners and plenum volumes tuned to take advantage of pressure waves to increase the pressure at the valve at the end of the intake stroke. Many engines have variable intake geometry such that they can be tuned to several engine speeds. Although not as effective, exhaust tuning also helps to improve volumetric efficiency. Modem combustion chambers have been optimized for fast combustion. Central spark plug location and compact combustion chamber geometry reduces the distance that the flame has to travel and increases flame area. Reduced combustion time results in less cycle-to-cycle variability under low loads and reduced time for autoignition reactions to occur at high loads. 1.3.5 Engine Control The ECU (Engine Control Unit) of a modem engine receives inputs from many sensors and controls many operating parameters and peripheral systems. The control parameters most important to engine operation are fuel injection quantity and spark timing. Under normal part-load operation the ECU executes closed-loop control of the fuel quantity using feedback from the exhaust 02 sensor. Spark timing is retarded about 5' CA from MBT to reduce NOx emissions with a very small efficiency penalty. At high loads spark retard is used to lower cylinder pressures for knock avoidance. The amount of spark retard is usually based on operating conditions. In some vehicles a knock detection sensor is employed. 1.3.6 Modern Engine Performance For a typical spark ignition engine, maximum efficiency occurs at about mid speed, three-quarters load. The reasons for decreasing efficiency around this point are: " Decreased speed - at lower speeds there is more time for heat transfer in each cycle. " Increased speed - at high speeds rubbing and fluid friction become more important. " Decreased torque - throttling increases pumping work, and the relative importance of friction increases 34 * Increased torque - spark retard is required to avoid knock and enrichment is used to protect the catalytic converter. Figure 1.9 shows approximations of BMEP as a function of engine speed for typical naturally aspirated and boosted engines. Maximum torque reaches a peak at mid speed and drops off in either direction. For naturally aspirated engines the peak is governed by the amount of air that flows into the engine, which depends on tuning and flow restrictions. For boosted engines the peak is determined by the size of the turbocharger and the maximum cylinder pressures that the engine can structurally endure. At lower speeds the maximum torque drops off due to the spark retard required to avoid knock, and due to increased heat transfer and decreased intake manifold tuning effects. At higher speeds the maximum torque is reduced by increased friction and flow restriction. 20 maximum cylinder pressure (mechanical integrity) and/or turbocharger performance IL E 16 - airflow restriction and/or turbocharger performance spark retard and reduced boost (to avoid knock) 12 limited airflow through valves and 0 8 - spark retard (to avoid knock) and heat transfer intake air flow restriction through ports and valves 4- Boosted -Baseline NA 0 0 2000 4000 6000 8000 Speed (RPM) Figure 1.9 - Approximate maximum BMEP (normalized torque) curves for typical naturally aspirated and boosted engines 35 1.4 THE PLASMATRON ENGINE SYSTEM "Plasmatron" is the name used to describe a partial oxidation thermal fuel reformer under development by the MIT Plasma Science and Fusion Center in cooperation with Arvin Meritor. The hydrogen rich reformate produced by the plasmatron is beneficial to several aspects of engine operation. 1.4.1 Plasmatron Design and Operation The plasmatron, shown schematically in Figure 1.10, uses a plasma arc to partially oxidize a stream of rich air-fuel mixture. Ideally, all of the carbon in the fuel would be converted to CO and all of the hydrogen would be converted to H2 : CmHn + M(O2 + 3.773N 2 )=> mCO+L! H 2 + 3.773-mN 2 2 2 2 (1.3) In practical applications the partial oxidation is not ideal, resulting in a mixture of several species in the outlet gases. For a typical plasmatron, about 20% of the fuel heating value is released as heat and the rest is converted to the chemical energy in CO and H2 . Table 1.1 shows ideal and typical plasmatron outlet compositions. Fuel Air I 1-Plas matron 2- 1 Stage Reactor 3-Nozzle Section 4- 2 nd Stage Reactor Air 2 Air 3 Fuel Figure 1.10 - Schematic of the plasmatron, courtesy of A. Rabinovich [8] 36 Table 1.1 - Outlet composition of an ideal plasmatron for a fuel H/C ratio of 1.9 and of a typical plasmatron [81 H2 CO N2 Ideal Plasmatron 25% 26% 49% Typical Plasmatron 20% 22% 51% CO 2 and H2 0 smaller hydrocarbons 0% 0% 6% ~1% Species The conceptual configuration of the plasmatron engine system is shown in Figure 1.11. Gasoline reformate would be used to enhance the air-fuel mixture entering the engine with H2 and CO. The proportion of fuel delivered to the plasmatron would be modulated to achieve the required combustion stability or knock avoidance while minimizing the efficiency penalty. Since the size of the plasmatron is limited by cost and space constraints, the maximum fuel reformed fraction would likely be limited to 20-30%. ,,,1 "I 3 r V air V H2 ,CO, gasoline N 2 (eg.85%) exhaust Figure 1.11 - Conceptual plasmatron engine system configuration 37 1.4.2 Benefits of Hydrogen and Carbon Monoxide Enhancement Hydrogen has two important properties that are beneficial to SI engine performance when used in conjunction with gasoline: 1) The laminar flame speed of hydrogen is about three times that of gasoline (at 1000 C, 1 atm, stoichiometric SL,H2= 170 cm/s [9] and SL,gasoine= 45.3 cm/s [1]). This stabilizes combustion and extends the dilution limit, either with air or with recirculated exhaust gas [10]. Faster combustion also consumes the unburned end-gas faster and helps to prevent knock at high loads. 2) Hydrogen has a high octane number. Since it is more knock-resistant than the highest ON PRF, it cannot be rated on the octane scale as specified by the ASTM standards. A previous literature review estimated the relative octane number of hydrogen to be between 130 and 140 [11][12][13]. A rating of 140 showed good correlation to experimental results when used in a bond-weighted mixing octane number estimation method [14]. This is partially due to its high flame speed, but mainly due to its resistance to autoignition. Recent work by Topinka et al. [14] has shown that replacing 10% of the fuel with H2 (by energy) lowers the required octane number of the primary fuel by 10 points for the same indicated torque. Carbon monoxide has a mild effect on increasing laminar flame speed. It is also resistant to knock and recent tests have indicated an ON of about 106. Experiments by Topinka et al. [14] have shown that CO improves knock resistance of the primary fuel about half as much as hydrogen for the same fuel fraction (by energy). 1.5 PREVIOUS WORK This section summarizes the results of a preceding MIT investigation of the effects of airfuel ratio and fuel reformate on knock. Relevant studies of the effects of operating conditions on knock, and the efficiency effects of compression ratio are also described. 1.5.1 Effects of Air-Fuel Ratio and Fuel Reformate on Knock and Autoignition In a preceding study by Topinka et al. [14] [8], the effects of air-fuel ratio and fuel reformate on the octane number of the PRF required to just avoid knock were investigated. It was found, contrary to previous studies [15], that at constant indicated torque the octane requirement of the engine increased slightly with air-fuel ratio (about 2 38 ON for relative air-fuel ratio from 1.1 to 1.7). Through extensive experiments with ideal fuel reformate addition (25% H2 , 26% CO, balance N2 ), it was found that the octane number of the PRF main fuel could be reduced as fuel reformate was added. The magnitude of this reduction is shown in Figure 1.12. It was also found that by adding just the H2 from the fuel reformate about 50% of the reformate benefit could be achieved, and by adding just the CO from the fuel reformate about 30% of the reformate benefit could be achieved. This study also proposed a method for modeling autoignition in SI engines using a zero-dimensional end-gas simulation with chemical kinetics. 30 -4-high load, lambda=1.1 -*-- high load, lambda=1.3 25 U- 20 -A- high load, lambda=1.5 -- high load, lambda=1.7 -8-mid load, lambda=1.1 15 --E-mid load, lambda=1.3 -A-mid load, lambda=1.5 0 0 (D 0hCU M 10 0) 5 0 0 0.05 0.1 0.15 0.2 0.25 0.3 Fuel Energy From H2+CO [%/100] Figure 1.12 - Lower octane fuel is supplied to the engine for audible knock when some energy is derived from H2 and CO (plasmatron reformate). Data simulates 15% and 30% of the gasoline being reformed in the plasmatron fuel reformer [141. 1.5.2 Effects of Operating Conditions on Knock and Autoignition Experiments by Russ [16] on a single cylinder engine have been used to estimate the magnitude of the effects that engine operating conditions and compression ratio have on the fuel ON required to avoid knock. The results are summarized in Table 1.2. 1.5.3 Effects of Compression Ratio on Efficiency Although there are several studies on the effects of small compression ratio changes, there are few on the effects of wide rc sweeps on a modem engine. Muranaka et al. [17] 39 compiled results of r, sweeps from earlier references and added their own data (Figure 1.13). They found that efficiency improvement with rc depended on operating conditions and swept volume. For a 500 cc cylinder at part load, efficiency initially increases at about 3% per unit r, at 9:1 and peaks between 13:1 and 15:1. Their analysis indicates that the major factors for the efficiency limit are heat transfer and unburned fuel. Table 1.2 - Effect of Engine Operating Conditions on Borderline Knock [16] Effect on ON requirement Operating condition Spark advance 1 ON/iV spark advance Intake air temperature 1 ON/7 K Air/fuel ratio Intake pressure peaks 5% rich of stoichiometric, 2 ON/air-fuel ratio around peak 3-4 ON/10 kPa Compression ratio 5 ON/compression ratio Coolant temperature 1 ON/10 K 2 00- LU 15 WOT I 4RL Ref1) [66.J 0- LU c-_ Ref (2) (497] [43] 4 5 Z -A12A [309] [SWEPT VOLUME) A1 [246] c 1 05 6 8 10 12 1 14 16 18 20 COMPRESSION RATIO Figure 1.13 - Brake efficiency improvement with compression ratio for engines of several swept volumes [171. 40 1.6 OBJECTIVES The objective of this work is to contribute to the knowledge required to optimize the configuration of an SI engine to properly balance tradeoff between mid-load efficiency and low speed maximum torque. The main factor affecting this tradeoff has traditionally been compression ratio. More recently, intake pressure boosting has also become a significant factor, with reduced compression ratios being necessary for satisfactory boosted engine performance at the knock limit. The development by MIT of an on-board fuel reformer that has been shown to lower the octane number requirement of the primary fuel [14] has introduced an additional factor. Five main tasks were carried out: 1. Investigate the efficiency effects of increasing the compression ratio at several load and air-fuel ratio conditions. 2. Investigate the relationship between spark retard and torque reduction. 3. Investigate the trends in knock limited spark timing with changes in compression ratio, intake boost, reformate fraction, air-fuel ratio, and fuel type. 4. Refine a previously developed knock modeling methodology and investigate its accuracy and sensitivity to initial conditions. 5. Apply experimental data to the optimization of engine performance and efficiency. 41 (this page intentionally left blank) 42 CHAPTER 2. EXPERIMENTAL METHOD To explore the effects of compression ratio, operating conditions, and fuel reformate on engine efficiency and knock limited performance, a series of low to medium load efficiency experiments and a series of high load knock experiments were performed. The set of experiments was repeated for each of three compression ratios. The experimental methods and test matrices were designed to define as many trends as possible without a prohibitively large number of test runs. 2.1 ENGINE SETUP The test cell contains a single cylinder research engine, a dynamometer, fuel and air supply systems, and related support equipment. 2.1.1 Engine and Dynamometer Specifications The test engine is a Ricardo MK III single cylinder research engine retrofitted with a Volvo B5254 pent-roof, 4-valve, central spark plug cylinder head. A tumble/swirl inducing plate between the inlet manifold and the head is used at low and medium loads to increase charge turbulence. The engine specifications are shown in Table 2.1. Table 2.1 - Test Engine Specifications Bore (mm) Stroke (mm) Displacement Volume (cm 2 ) Connecting Rod Length (mm) Piston 1 Clearance Vol. (cm 2)/Comp. Ratio Piston 2 Clearance Vol. (cm 2)/Comp. Ratio 83 90 487 158 55 / 9.8:1 46 / 11.6:1 Piston 3 Clearance Vol. (cm 2)/Comp. Ratio 39 /13.4:1 Valve Ia Timing Tii IVO 00 BTC, IVC 600 ABC EVO 480 BBC, EVC 120 ATC Three custom-made forged pistons were used for the experiments, one for each compression ratio. The baseline piston is similar to the stock Volvo piston, with the same crown shape and first land height. The two high rc pistons have partially raised piston crown "pop-ups" to decrease the combustion chamber clearance volume. Figure 2.1 43 shows a photograph of the three pistons and Appendix A shows the dimensions of the high r. piston crowns. Figure 2.2 is an illustration of cross sections of the combustion chamber at the top center position for the 13.4:1 piston. Figure 2.1 - Base, medium, and high compression ratio pistons (left to right) Figure 2.2 - Illustration of the 13.4:1 piston and cross sections of the combustion chamber at top center piston position An EATON 6000 series 50HP electric dynamometer is connected to the engine by a drive shaft. This dynamometer is able to motor the engine or absorb power from it and automatically adjusts its torque output to maintain a constant speed at a manual set-point. 44 2.1.2 Air and Fuel Supply Systems The engine intake system, pictured in Figure 2.3, can be switched between two configurations - ambient or boosted. In the ambient configuration the air is drawn through a filter from the test cell. In the boosted configuration air is supplied by a compressor in a separate test cell, then is filtered and regulated to the desired engine inlet pressure. In both cases the air then goes through a laminar flow element and a pulsationdamping tank before being throttled into the engine intake. regulator to amiet ompresso a filter 0-brggvalve 4bar gage laminar flow Figure 2.3 -Intake air system The engine liquid fuel system, outlined in Figure 2.4, can be configured to draw out of either of two fuel tanks, and can be purged with nitrogen gas. During normal operation the pump draws the fuel out of the main fuel tank. From the pump, the fuel flows through the filter to the engine. At the engine, a regulator keeps the pressure in the fuel line at a constant differential (approximately 3 bar) from the average intake manifold pressure. Excess fuel that is not consumed by the injector is returned to the fuel tank by the regulator. For high accuracy fuel flow measurements the fuel supply and return can be switched to the second fuel tank, which is equipped with a balance. The nitrogen purge system allows the fuel system to be cleared of fuel, except for the small volumes of the accumulator, the regulator, and the injector supply line. Figure 2.5 shows the gaseous fuel system used to supply simulated plasmatron gas to the engine. A regulator is used to control the pressure upstream of the critical flow orifice. 45 o leslo t p regulator to 1bar gage valve3 to fuel injector slop tank 1I N2 selector fuel tank1I from intake manifold fle hose pump ------------valvel valve5 differential backpressure regulator (~3bar) valve2 fuel valve4 accumulator tank 2 balance ............ return Figure 2.4 - Liquid fuel system adjustable regulator solenoid valve critical flow orifice flame arrestor to intake manifold H2, co, N2 Figure 2.5 - Gaseous fuel system (simulated plasmatron gas supply) 2.2 ENGINE CONTROL AND MEASUREMENTS A diagram of transducers and gauges is shown in Figure 2.6. Descriptions are listed in Table 2.2. The rest of this section describes the control and measurement system in further detail. 2.2.1 Engine Control Unit A MoTeC M4 engine controller is used to control the injector and ignition system. The injection timing is set to 3850 BTC and the dwell is set between 4 ms and 8 ms, 46 depending on the compression ratio of the engine (high dwell for high re). The injector pulse width and spark timing can be adjusted while the engine is running. selector fuel tank 1 hose ........................... funel tank 2 TT2 PT1 throttle P damping from air supply laminar flow element inj. from gaseous fuel supply .............. P2critical flow orifice P P P dynamometerEN heated oil supply heatedlcooled coolant supply exhaust to trench Figure 2.6 - Engine control and measurement diagram 2.2.2 Fuel Flow Measurement There are two methods of liquid fuel flow measurement. For fast measurements with limited accuracy (+/- 2%) the fuel injector pulse width is used to determine the amount of fuel injected for each cycle. The fuel pressure regulator keeps a constant differential between the average intake manifold pressure and the fuel injector supply pressure. Thus the flow through the injector orifice is constant when the injector is open. Using an experimental calibration, the mass of fuel injected can be calculated from the injector pulse width. The accuracy of this method is limited because of fuel temperature-density effects, and because high-speed pressure fluctuations in the intake manifold vary the pressure across the injector at the time of injection as operating conditions vary. For more accurate measurements (+/- 0.5%) a "pail and scale" method is used. For this 47 method the fuel supply and return are switched to a container on a balance. The mass of fuel consumed by the engine over a set time period is calculated by subtracting the final mass of the container from the initial mass. The fuel mass flow rate is the mass of fuel consumed divided by the time period. Table 2.2 - Summary of experimental engine transducers and gauges Name CP1 CP2 CP3 CP4 DPT ENC Measurement crankshaft position crankshaft position camshaft position camshaft position differential pressure transducer crankshaft position encoder 02T P1 P2 PTl PT2 PT3 oxygen sensor oil pressure gauge gaseous fuel pressure gauge air supply absolute pressure transducer manifold absolute pressure transducer cylinder pressure transducer TT1 TT2 TT3 TT4 TT5 WB engine coolant inlet thermocouple air inlet thermocouple exhaust thermocouple engine coolant inlet thermocouple engine oil inlet thermocouple balance Details for engine speed display for engine control unit for pressure acquisition system for engine control unit for air volume flow rate display 360 pulses per revolution, for pressure acquisition system for lambda meter/display for air supply pressure display for MAP display Kistler pressure transducer, converted to voltage signal by charge amplifier, measured by computer DAQ card for coolant temperature display for inlet air temperature display for exhaust temperature display for engine coolant heater thermostat for engine oil heater thermostat/display displays mass of fuel tank 2 A critical flow orifice is used to control the gaseous fuel flow rate. So long as the absolute gas pressure upstream of the orifice is kept higher than double the manifold pressure, flow through the orifice will be choked. For choked flow, the mass flow rate depends only upon the orifice dimensions, the upstream pressure and temperature, and the specific heat and molecular weight of the gas. The orifice calibration (performed with air) is modified to take into account the reformate gas properties. The temperature upstream of the orifice is relatively constant at room temperature due to the long flowpath through a copper tube. Thus, adjusting the upstream gas pressure with a regulator 48 sets the orifice flow rate to a known value. Since the range of each orifice is limited, several orifices are required to achieve the required range. 2.2.3 Intake Pressure Measurement and Control Absolute pressure transducers measure the intake pressure in two locations. One is after the air flow meter but before the throttle plate (air tank pressure), and the second is between the throttle and the engine (Manifold Absolute Pressure - MAP). Digital readouts display the results. The manifold pressure is controlled by manually adjusting the angle of a throttle valve that is driven by a stepper motor. 2.2.4 Temperature Measurement and Control There are four temperature measurement points on the engine: " Engine coolant inlet " Intake air, in the tank between the flow meter and the throttle " Exhaust, about 2 cm from the exhaust port outlet " Engine oil inlet The engine coolant temperature is controlled to 900 C +/- 20 C by an electric heater connected to an electronic thermostat, and a cold-water heat exchanger connected to a mechanical thermostat. An electric heater connected to an electronic thermostat controls the oil temperature to approximately 700 C. 2.2.5 Air Flow Measurement Air volume flow rate is obtained by measuring the differential pressure across a laminar flow element. The volume flow rate is converted to a mass flow rate by using the ideal gas law with the air tank pressure and temperature measurements. The air mass flow rate is corrected for water content using a humidity measurement. The accuracy of the air mass flow measurement is approximately +/- 2%. 2.2.6 Air-Fuel Ratio Measurement A Universal Exhaust Gas Oxygen (UEGO) sensor measures the oxygen content of the exhaust gas. A Horiba Mexa-11 O analyzer interprets the signal and displays the air-fuel equivalence ratio (2). 49 2.2.7 Cylinder Pressure Measurement A Kistler 6125A piezoelectric pressure transducer equipped with a flame arrestor measures the cylinder pressure. A charge amplifier converts the current signal from the transducer to a voltage signal. The voltage signal is measured once every crank degree by a National Instruments 6023E data acquisition card triggered by a BEI crankshaft encoder. A program written in National Instruments LabVIEW records 300 cycles of cylinder pressure data when triggered by the user. Burn rate analysis software is used to calculate several quantities from the acquired cylinder pressure data. They include: 0 NIMEP (Net Indicated Mean Effective Pressure) 0 COV (Coefficient of Variation) of NIMEP 0 GIMEP (Gross Indicated Mean Effective Pressure) 0 COV of GIMEP 0 Peak pressure 0 Crank angle of peak pressure * 0-10% mass fraction burned time * 0-50% mass fraction burned time 0 10-90% mass fraction burned time A Visual Basic macro automatically runs the burn rate analysis software with inputs exported from the data collection spreadsheet and writes the results to the spreadsheet. 2.2.8 Knock Detection An audible knock method was used to detect knock for this work. An equalizer set to remove all frequencies except those near 6 kHz and 12 kHz filtered the signal from a microphone placed approximately 1 cm above the valve cover. Headphones were used to listen to the resulting audio signal and identify knock. This method was selected because it was shown in previous work to be consistent [8], because it closely matches the method used during real engine calibration, and because it avoided the complexity of a sensorbased knock detection system. 50 2.3 CHANGING COMPRESSION RATIO Three pistons were used during the experiments. When the first piston was installed in the engine, a new cylinder liner, new rings, and new connecting rod bearings were also installed. The engine was then broken in for 50 hours at varying speeds and loads. Motoring and firing cylinder pressure data was recorded and analyzed to ensure proper sealing and proper operation of the data acquisition system. Efficiency and emissions data was recorded and compared to the stock Volvo piston to ensure consistency. To minimize variation in ring pack performance and eliminate the need for cylinder liner honing, the first and second compression rings were re-used for the second and third piston. After each piston was installed, the engine was broken in for 20 hours and cylinder pressure data was recorded under motoring and firing conditions to ensure proper sealing and proper operation of the data acquisition system. 2.4 EFFICIENCY EXPERIMENTS A series of tests were performed to determine trends in indicated efficiency and other engine parameters with changes in operating conditions and compression ratio. 2.4.1 Experimental Procedure Before each set of data is taken, the engine is fully warmed up. The procedure for each data point is as follows: 1. Set engine speed in the dynamometer controller. 2. Make an initial estimate of MBT (Maximum Brake Torque) timing from previous trends. 3. Adjust fueling and air flow to achieve the desired air-fuel ratio and NIMEP (from a real-time readout). 4. Perform a timing sweep while recording and processing cylinder pressure data to identify MBT timing to within 10 CA. Re-check air-fuel ratio and NIMEP. 5. At MBT timing, record operating conditions and cylinder pressure data. 6. Keeping control parameters constant, record the change in fuel tank mass (from the balance) over a 600 sec (1500 rpm) or 420 sec (2500 rpm) time period. 51 The experimental results are recorded in a spreadsheet and used to calculate indicated efficiencies, mass fraction burned angles, and other useful parameters. The k measurement from the UEGO sensor and the ), calculated from the measured air and fuel flow rates are compared to ensure consistency. 2.4.2 Fuel The fuel used for the efficiency experiments, termed "toliso", is a mixture of 70% isooctane with 6.0 mL of TEL per gallon (PRF RON of 120) and 30% toluene. A high-ON fuel was required to avoid knock at high loads and high compression ratios. This mixture was selected because it has a similar alkane/aromatic ratio, H/C ratio, energy content, and specific gravity to gasoline. The high vaporization temperature was not considered to be important since all experiments were performed under fully warmed conditions. 2.4.3 Operating Conditions and Compression Ratio The test matrix for the efficiency experiments is shown in Figure 2.7. At each rc and each engine speed an air-fuel ratio sweep was performed and load sweeps at two air-fuel ratios were performed for a total of 42 data points. Rc 9.8:1 11.6:1 1=1.3 X=1.0 2bar NIMEP 2bar NIMEP 2500rpm X=1.0 X=1.3 X=1.6 1500rpm 4bar NIMEP 4bar NIMEP 4bar NIMEP X=1.0 8bar NIMEP X=1.3 Speed 1500rpm 2500rpm 13.4:1 13.4:1 1500rpm 160p 2500rpm 8bar NIMEP Figure 2.7 - Test matrix for efficiency experiments 2.5 KNOCK LIMITED MINIMUM SPARK RETARD EXPERIMENTS A series of tests were performed to determine the relationship between spark timing and engine output and to identify trends in knock limited spark timing with changes in operating conditions and compression ratio. 52 2.5.1 Experimental Procedure Before each set of data is taken, the engine is fully warmed up. Each data point consists of a spark-timing sweep with a high-octane fuel - PRF 120 for a PRF data point and toluene for a TRF data point (see Section 2.5.2 ) - and a Knock Limited Spark Advance (KLSA) fuel octane sweep. The procedure for each data point is as follows: Spark timing sweep 1. Set the engine speed with the dynamometer controller. 2. Position the throttle plate in the wide-open position. 3. Make an initial estimate of MBT (Maximum Brake Torque) timing from previous trends. 4. For naturally aspirated points, adjust the fuel injector pulse width to get a stoichiometric air-fuel ratio. For boosted points, adjust the boost pressure regulator and injector pulse width to achieve the desired NIMEP and air-fuel ratio. For points with fuel reformate, the gaseous fuel flow rate must also be adjusted to achieve the desired reformate fraction. 5. Set spark timing to approximately 40 CA earlier than estimated MBT timing. If the spark timing cannot be advanced because the point is knock limited (high rc, high boost points), set the spark timing as early as possible while avoiding the onset of pressure oscillations. If the spark timing cannot be advanced because the point is cylinder pressure limited (high re, high boost points), set the spark timing as early as possible without exceeding the maximum cylinder pressure of 11 Obar. 6. Record operating conditions and cylinder pressure data. 7. Retard the spark timing in steps of 2' CA, repeating step 6 for every spark timing until combustion becomes unstable or exhaust temperature approaches 7500 C. At some timings spark discharge noise will interfere with the data acquisition system, so step sizes must be adjusted by 0.5' CA. At retarded spark timings the throttle must be closed slightly to maintain the correct exhaust air-fuel ratio. Fuel octane sweep 1. Keeping the engine motoring, flush the fuel system with nitrogen. 2. Load the fuel system with the highest ON fuel that will knock with spark timing set to MBT. 53 3. Let engine stabilize at a spark timing that avoids knock for 2-3 minutes. 4. Keeping fuel flow and air-fuel ratio constant, advance spark timing by increments of 10 CA until knocking cycles are clearly heard through the audio system. 5. Retard the spark timing by 1 CA and record operating conditions and cylinder pressure data. 6. Keeping the engine motoring, flush the fuel system with nitrogen and load the next lower ON fuel. 7. Repeat steps 3 - 6 until spark is retarded to the point that combustion is unstable, exhaust temperature approaches 750 C, or runaway knock cannot be avoided. For UTG fuels (see Section 2.5.2 ) the spark timing cannot be advanced to MBT during the spark-timing sweep, so the procedure is modified as follows: 1. Set the spark timing 4-6* CA later than the expected knock limit. 2. Since timing cannot be advanced to MBT due to knock, the required boost pressure must be approximated by matching the NIMEP of the toluene sparktiming sweep under similar operating conditions. Adjust the boost pressure regulator and injector pulse width to achieve the desired X and match the NIMEP of the corresponding toluene data point. For points with reformate, the gaseous fuel flow rate must also be adjusted to achieve the desired reformate fraction. 3. Retard spark timing until combustion becomes unstable or exhaust temperature approaches 7500 C. 4. Keeping fuel flow and X constant, advance spark timing by increments of 1 or 20 CA until knocking cycles are clearly heard through the audio system. Record operating conditions and cylinder pressure data at each spark timing. The experimental results are recorded in a spreadsheet and used to calculate bum angles and other relevant parameters. The X measurement from the UEGO sensor and the X calculated from the measured air and fuel flow rates are compared to ensure consistency. 2.5.2 Fuels Since gasoline is available only in a limited set of formulations, three types of fuels were used in the experiments to explore the entire range of operating conditions and fuel qualities of interest: 54 UTG Unleaded test gasolines, commonly termed "indolene", are standardized fuels typically used for emissions certification and laboratory work. For this work UTG91 and UTG96 were obtained from Chevron Phillips Chemical Company LLC to represent the performance of gasoline. The relevant specifications of these fuels are in Table 2.3. Table 2.3 - UTG fuel properties Specification RON MON Sensitivity Alkanes,%vol Aromatics,%vol Olefins,%vol H/C ratio Typical UTG91 90.8 83 7.8 70 24 6 1.89 Values UTG96 96.1 87 9.1 67 28 5 1.86 PRF Primary reference fuels are the reference for the ASTM octane rating methods. By definition, the RON and MON of PRFs are equal. For ON of 0 to 100, PRFs are mixtures of iso-octane and n-heptane, the octane numbers of which are the volume fraction of isooctane under standard conditions. For ON of 100 to 120, PRFs are mixtures of iso-octane and TEL (tetraethyl lead), the proportions of which are specified in the ASTM 2699 and ASTM 2700 standards by: ON =100 + 28.28T 2 1+ 0.736T + 1+ 1.472T - 0.035216T (2.1) Where T is the amount of TEL in mL per gallon of iso-octane. For this work, PRFs were mixed in increments of 5 ON. To avoid directly handling concentrated TEL, which is toxic, fuels above PRF 100 were mixed by diluting PRF 120 with iso-octane to obtain the required TEL concentration. 55 TRF To better represent the sensitivity and aromatic content of gasolines, the knock experiments were also performed with toluene reference fuels. TRFs are mixtures of toluene and n-heptane. TRF RON and MON are plotted as a function of n-heptane volume fraction in Figure 2.8. For this work TRFs were mixed in increments of 5 ON. 110 RON = -0.0097x 2 - 0.4419x + 111.29 + RON M MON 100 - E 90 z I. 0 0 80 70 60 MON 50 = -O.0031x 2 - 0.7498x + 103.24 i 10 20 30 40 50 60 %v n- heptane (bulk toluene) Figure 2.8 - RON and MON of toluene reference fuels as a function of volume fraction n-heptane. Compiled by Shell Global Solutions [18] from ASTM vol 05.04, Table 28 and internal data. 2.5.3 Operating Conditions and Compression Ratio The test matrix for the knock experiments is shown in Figure 2.9. For each r" and each fuel type an air-fuel ratio sweep was performed and boost sweeps at two air-fuel ratios were performed. Reformate sweeps were performed from a boosted stoichiometric point and a base lean point. Not all of the points were taken for the UTG fuels. In total there are 83 data points. 56 Rc Fuel Type X=1.0* WOT PRF 9.8:19.1 - - - - - TRF X=1.3* I PRF 11.6:1 TRF UTG91* UTG96*** X=1.0 200/a boost (1.2X WOT NIMEP) X=1.3** 15% boost (1.15X X=1.0 WOT NIMEP) TRF UTG96* 'match X= 1.0 WOT NIMEP 15% reformate fraction I__ PRF 13.4:1 =1.6* bo osted to mat ch X=1.0 WO T NIMEP X=1.3* boosted to match X=1.0 OT NIMEP 400/ boost (1.4X WOT NIMEP) )L=1.3** 300a boost (1.3X X=1.0 WOT NIMEP) X=1.3* match 2=1.0 WOT NIMEP 3 00 A reformate fraction 400h boost 15% reformate fraction k=1 .0 400a boost 3 00 / reformate fraction Figure 2.9 - Test matrix for knock experiments 57 (this page intentionally left blank) 58 CHAPTER 3. EXPERIMENTAL RESULTS This chapter presents the data collected using the apparatus and procedures described in Chapter 2. The experimental results are used to evaluate the effects of engine operating parameters on efficiency and knock-limited performance. An extended data set is available in the appendices. 3.1 EFFICIENCY RESULTS The data presented in this section was collected as described in Section 2.4 . The effects of air-fuel ratio and load on indicated efficiency are discussed first, followed by a description of the efficiency effects of compression ratio. 3.1.1 Effects of Air-Fuel Ratio and Load Figure 3.1 shows the experimental net indicated efficiency (rjin) for air-fuel equivalence ratio (k) sweeps at the mid-load point of 4.0 bar NIMEP (Net Indicated Mean Effective Pressure). Data is displayed for all three compression ratios at 1500 rpm and 2500 rpm. The normalized improvement from the stoichiometric efficiency is shown in Figure 3.2. The curves are drawn to illustrate the expected trends between data points, and match previous work that found peak fin to be at k = 1.5 for a similar load, 1500 rpm, and 10:1 rc [6]. Figure 3.3 shows the effect of k on gross indicated efficiency (rtig) for the same data points. The maximum increase in rl'ig is about two-thirds that of the maximum increase in tifn, indicating that about two-thirds of the improvement from increasing air fuel ratio is from increased y and decreased heat transfer. The remaining improvement is from reduced pumping work. At low k, the normalized increase of agin with k appears stronger for low rc than for high rc. Figure 1.5 shows that the magnitude of the efficiency improvement from increasing y (i.e. from increasing k) remains approximately constant as rc increases. Since efficiency is increasing with compression ratio, at high rc the improvement from increasing y is smaller as a fraction of the baseline efficiency. At low rc, i1fig peaks and starts to decay at a lower air-fuel ratio than for high rc. The extension of X for peak efficiency at high rc is likely due to reduced combustion durations 59 and cycle-to-cycle variability. The same trend is visible for rIjn, but decreasing pumping work extends the peaks to higher air-fuel ratios. Figure 3.4 shows combustion event timing at the 9.8:1 r, as a function of X. The crank angles of spark and 10%, 50%, and 90% mass fraction burned are shown. As the mixture becomes leaner the laminar flame speed decreases, leading to longer combustion durations. Figure 3.5 shows combustion event timing under lean conditions as a function of rc. As r, increases, the pressures and temperatures in the combustion chamber increase, resulting in a higher laminar flame speed, which is evident in the decreased 0-10% mass fraction burned times. The decrease in 10-90% mass fraction burned time is probably due to a combination of the increased laminar flame speed, and an increase in turbulence from the fluid being "squished" towards the center of the chamber by the piston "pop-up". As engine speed increases, there is less time for heat loss from the burned gases for each cycle, so in all cases qfig at 2500 rpm is higher than at 1500 rpm. Net efficiency is also higher at the higher speed, but by a lesser amount because the increased efficiency from lower heat loss must be counteracted with increased throttling to maintain constant load, resulting in increased pumping losses. Figure 3.6 and Figure 3.7 show the increase in Tlin with increasing load for X of 1.0 and 1.3. The efficiency improvement is normalized around the mid-load 4 bar NIMEP point in Figure 3.8 and Figure 3.9. At mid-load, net efficiency improves by about 7% per bar NIMEP relative to the efficiency at 4 bar NIMEP. Although well aligned, the curves for high r, are slightly steeper than for low rc. This is likely because the higher efficiency at high r, reduces the intake manifold pressure necessary to maintain the same load. At the reduced manifold pressures, the reduction in throttling losses from an incremental load increase improves efficiency more than at higher manifold pressures. The improvement in rjin with load for the stoichiometric air-fuel ratio is slightly better than for the lean airfuel ratio. Again, this is because manifold pressures are higher under lean conditions so there is not as much benefit from reduced throttling. The gross indicated efficiencies, shown in Figure 3.10 and Figure 3.11, increase with load at approximately one quarter the rate of the net efficiencies. Since the difference between rfin and t mig is due to pumping work, reduction in throttling is responsible for about 75% of the improvement in agin as NIMEP increases. The rest of the improvement is mostly due to heat loss becoming a smaller portion of the total charge energy. 60 3.1.2 Effects of Compression Ratio Figure 3.12 shows the change of net indicated efficiency with compression ratio for a range of X at mid-load. The efficiency increase is normalized in Figure 3.13. At the base rc of 9.8:1, 11in improves by about 2.5% per unit rc. Efficiency appears to peak at a rc of about 14:1 with an improvement of 6-7%. This agrees well with existing data [17]. The improvement is better at 2500 rpm than at 1500 rpm, likely because the increased heat transfer at high r0 does not have as much of an effect when the amount of time for heat transfer is lower. There does not appear to be a clear trend with X. Figure 3.14 and Figure 3.15 show the increase of Tgi. with increasing r, for a range of loads at X of 1.0 and 1.3. The efficiency improvement from the base compression ratio is normalized in Figure 3.16 and Figure 3.17. The improvement of jfin is better at higher loads, where the increased heat transfer from increasing rc is a smaller fraction of the overall charge energy. The experimental results also show that load has a greater effect on the relationship between re and ifi under lean conditions than it does at , = 1.0. 0.36 -+- Rc = 9.8:1, 1500 rpm -- Rc = 9.8:1, 2500 rpm -- Rc = 11.6:1, 1500 rpm 0.35 ' -z-Rc = 11.6:1, 2500 rpm -- Rc = 13.4:1, 1500 rpm -e-Rc = 13.4:1, 2500 rpm 0.34- 4 0.33 0.32- z 0.31 0.30.290.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Lambda Figure 3.1 - Change of net indicated efficiency with lambda; 4.Obar NIMEP. 61 12 -+- Rc = Rc = -Rc = -A- Rc = -4-Rc = S10- 9.8:1, 1500 rpm 9.8:1, 2500 rpm 11.6:1, 1500 rpm 11.6:1, 2500 rpm 13.4:1, 1500 rpm -e-- Rc = 13.4:1, 2500 rpm C S- 2- Z 40)- 0.9 1.1 1 1.3 1.2 1.4 1.5 1.6 1.7 Lambda Figure 3.2 - Normalized change of net indicated efficiency with lambda; 4.Obar NIMEP. 4) 0.4--+ Rc = 9.8:1, 1500 rpm -- Rc = 9.8:1, 2500 rpm 0.39 -- Rc = Rc = -- Rc = -e-Rc = >-% S0.38 - .9 11.8:1, 11.6:1, 13.4:1, 13.4:1, u'0.37 0.36- 0.35 0.34 0.330.9 1 1.1 1.2 1.3 1.4 Lambda 1.5 1.6 1.7 Figure 3.3 - Change of gross efficiency with lambda; 4.bbar NIMEP. 62 1500 2500 1500 2500 rpm rpm rpm rpm 30 -+- .......... Spark, 1500 rpm -m-10% mfb, 1500 rpm 20- mfb, 1500 rpm -0-90% mfb, 1500 rpm -&-50% 10- -*- Spark, 2500 rpm -e- 10% mfb, 2500 rpm -A- 50% mfb, 2500 rpm 0-10- -e- 90% mfb, 2500 rpm (-20-30-40-50 - -60 1.1 0.9 1.7 1.5 1.3 Lambda Figure 3.4 - Combustion event timing; 4.Obar NIMEP, r, = 9.8:1. 30 20 - 10 - -4-$Spark, 1500rpm -U-10% mfb, 1500 rpm -+-50% mfb, 1500 rpm +90% mfb, 1500 rpm -+- Spark, 2500 rpm -E- 10% mfb, 2500 rpm 10 0 0- -A- 50% mfb, 2500 rpm - -e- 90% mfb, 2500 rpm (-20-30-40-50-60 I 9 10 I 11 I I 12 13 14 RF Figure 3.5 - Combustion event timing; 4.Obar NIMEP, X = 1.6. 63 0.38 Rc = 98:1 1500 rpm -e- Rc = 9.8:1, 2500 rpm -- 0.36 - -- Rc = -r-Rc = -- Rc = -e- Rc = 0.34- 11.6:1, 11.6:1, 13.4:1, 13.4:1, 1500 2500 1500 2500 rpm rpm rpm rpm lw 0.32 o 0.3 - z 0.280.26 - 0.24 - i i i i 1 2 3 4 5 6 7 8 9 NIMEP (bar) Figure 3.6 - Change of net indicated efficiency with NIMEP; X = 1.0. 0.4 -+-Rc = 9.8:1, 1500 rpm 0.38 - -+-Rc = 11.6:1, 1500 rpm --A- Rc = 11.6:1, 2500 rpm 0.36 - -4- Rc = 13.4:1, 1500 rpm -e-Rc = 13.4:1, 2500 rpm --- Rc = 9.8:1, 2500 rpm w 0.34V .- * U 0.32 - 0.30.28- 0.26 1 2 3 4 5 6 7 8 9 NIMEP (bar) Figure 3.7 - Change of net indicated efficiency with NIMEP; X 64 = 1.3. 20 8 - -*- Rc = 9.8:1, 1500 rpm ?15 - ~+-Rc = 9.8:1, 2500 rpm -Rc = 11.6:1, 1500 rpm 10- -- Rc = 11.6:1, 2500 rpm -4--Rc = 13.4:1, 1500 rpm -e-Rc = 13.4:1, 2500 rpm -U Z -10- & -15 0 -20- -25 1 2 I 3 I 5 I 4 I 6 7 I 8 9 NIMEP (bar) Figure 3.8 - Normalized change of net indicated efficiency with NIMEP; X 20 -- -- = 1.0. -+--Rc = 9.8:1, 1500 rpm 615 + Rc = 9.8:1, 2500 rpm -+- Rc = 11.6:1, 1500 rpm S10Z --A- Rc = 11.6:1, 2500 rpm -0-Rc = 13.4:1, 1500 rpm -e-Rc = 13.4:1, 2500 rpm - z -10C CM -15S-20-25- I 1 2 3 4 5 6 7 8 9 NIMEP (bar) Figure 3.9 - Normalized change of net indicated efficiency with NIMEP; X = 1.3. 65 0.4 -+-Rc =9.8:1, 1500 rpm +R %= .a8 : , 2500 rpml - 0.38 - Rc = 11.6:1, 1500 rpm ,A- Rc = 11.6:1, 2500 rpm -4-Rc = 13.4:1, 1500 rpm -e-Rc = 13.4:1, 2500 rpm 0.36- 0.34- 0.32- 0.3 1 2 3 I 4 I 5 I 6 I 7 8 9 NIMEP (bar) Figure 3.10 - Change of gross indicated efficiency with NIMEP; X = 1.0. 0.42 -+- Rc = 9.8:1, 1500 rpm -e- Rc = 9.8:1, 2500 rpm -0.- Rc = 11.6:1, 1500 rpm --A- Rc = 11.6:1, 2500 rpm -4-Rc = 13.4:1, 1500 rpm -e--Rc = 13.4:1, 2500 rpm W 0.38 -) 0.36 0 0.34- 0.32 1 2 3 4 5 I 6 I 7 I 8 9 NIMEP (bar) Figure 3.11 - Change of gross indicated efficiency with NIMEP; X 66 = 1.3. n0 -+-Lambda -s-Lambda Lambda -- Lambda 0.35 > 0.34 - = 1.0, 1500 rpm = 1.0, 2500 rpm = 1.3, 1500 rpm = 1.3, 2500 rpm -*--Lambda = 1.6,1500 rpm -e- Lambda = 1.6, 2500 rpm 0.320.31 Z 0.3 - 0.29 0.28 - ' i 9 10 11 12 13 14 Figure 3.12 - Change of net indicated efficiency with compression ratio for a range of X; 4.Obar NIMEP. 8 -- 7 --4-e- Lambda = 1.0, 1500 rpm -+- Lambda = 1.0, 2500 rpm 6 W. 5 Z Lambda = Lambda = Lambda = Lambda = 1.3, 1.3, 1.6, 1.6, 1500 2500 1500 2500 rpm rpm rpm rpm 3 2 0 T 9 - 10 11 12 13 14 Figure 3.13 - Normalized change of net indicated efficiency with compression ratio for a range of X; 4.Obar NIMEP. 67 0.38 - -+-NIMEP = 2.0 bar, 1500 rpm 0.36 - -+- NIMEP = 2.0 bar, 2500 rpm - NIMEP = 4.0 bar, 1500 rpm 0.34 - -- -*-NIMEP = 8.0 bar, 1500 rpm e- NIMEP = 8.0 bar, 2500 rpm 0.32- . 0.3- W ' z NIMEP = 4.0 bar, 2500 rpm 0.28 - 0.26 - 0.24 - 0.22 - : 0.2 9 10 11 12 13 14 Figure 3.14 - Change in net indicated efficiency with compression ratio for a range of loads; X=1.0. 0.4- -- 0.38 - 0.36 - -,- NIMEP = 4.0 bar, 1500 rpm -A- 0.34- . W 0.32 NIMEP = 2.0 bar, 1500 rpm -+- NIMEP = 2.0 bar, 2500 rpm NIMEP = 4.0 bar, 2500 rpm -0- NIMEP = 8.0 bar, 1500 rpm -e- NIMEP = 8.0 bar, 2500 rpm - 0.3 0.28- 0.26 - 0.24 - 0.22 i 9 10 11 12 13 14 Figure 3.15 - Change in net indicated efficiency with compression ratio for a range of loads; X =1.3. 68 8 NIMEP = 2.0 bar, 1500 -+- NIMEP = 2.0 bar, 2500 -A-NIMEP = 4.0 bar, 1500 ,h- NIMEP = 4.0 bar, 2500 -- S6 0-l 0 7 -0--NIMEP = 8.0 bar, 1500 rpm -e- NIMEP = 8.0 bar, 2500 rpm W. 5 cc z rpm rpm rpm rpm (4 S3 4)2 I1 a) cc 0 9 - 1 9 10 11 12 13 14 Figure 3.16 - Normalized change in net indicated efficiency with compression ratio for a range of loads; X=1.0. 8 -+-NIMEP = 2.0 bar, 1500 rpm -- NIMEP = 2.0 bar, 2500 rpm -h-NIMEP = 4.0 bar, 1500 rpm -A- NIMEP = 4.0 bar, 2500 rpm 06 -*- NIMEP = 8.0 bar, 1500 rpm -e- NIMEP = 8.0 bar, 2500 rpm S W5 3 z C S2 C 0 I 9 10 1 12 11 12 13 14 Figure 3.17 - Normalized change in net indicated efficiency with compression ratio for a range of loads; X=1.3. 69 3.2 KNOCK LIMITED PERFORMANCE RESULTS This section begins with a description of the relationship between spark timing and engine torque output. The effects of varying air-fuel ratio, boost pressure, fuel reformate fraction, and compression ratio on knock limited engine performance are subsequently presented. Data was collected according to the procedure described in Section 2.5 3.2.1 Ignition Timing The change of NIMEP with spark timing for a range of X with PRF 120 and toluene fuel is shown in Figure 3.18. Spark timing for maximum NIMEP (MBT timing) advances with increased X due to a slower flame speed. MBT timing is also more advanced for PRF 120 than for toluene. These results agree with the findings of [19] which show that TRF has similar combustion duration to gasoline while PRF is slightly slower. NIMEP drops off as spark timing is retarded from the optimum because later combustion decreases the volume ratio, and consequently temperature ratio, through which the burned gases are expanded, resulting in less work extraction. Figure 3.19 shows the same data as Figure 3.18, except with NIMEP normalized to maximum NIMEP, and with spark timing normalized to MBT timing. The normalized spark timing, termed "spark retard", is the spark timing, in 'ATC, minus the MBT spark timing. It represents the number of crank degrees that the spark timing has been shifted from that for maximum torque. It is evident from this chart that NIMEP of the test points with slower combustion (i.e. with more advanced MBT timing) does not decrease as quickly with spark retard as it does for test points with faster combustion. This trend can also be seen in Figure 3.20, where the NIMEP of the low r, test points, which are slower burning, drops more slowly with spark retard than for the high r, test points. Another timing parameter, termed "combustion retard", has been developed to better describe the change in combustion phasing from the optimal. Combustion retard is the location of 50% mass fraction burned, in 'ATC, minus the location of 50% mass fraction burned for MBT spark timing. It represents the number of crank degrees that the center of the combustion event has been shifted from the timing for maximum torque. Figure 3.21 shows the relationship between spark retard and combustion retard for a range of X with PRF120 and toluene fuel. It can be seen that the timing of 50% mass fraction burned reacts differently to spark timing as operating conditions change. Figure 3.22 70 shows the change of normalized NIMEP with spark retard for a range of compression ratios, fuels, air-fuel ratios, and intake pressures. There is significant spread, but when combustion retard is substituted for spark retard, as shown in Figure 3.23, all of the points fit well to a single diagonally asymptotic curve. The equation of the curve fit is: NIMEP NIMEP =I - 0.168 - + 4.443 \ 10-3(50%mjb - 2\0.s431~~ 050%mfb,MBT 1] (3.1) NIMEPMBT where 0 50%m.t is the crank angle of 50% mass fraction burned in "ATC and 0 50%mfb,MBT the crank angle of 50% mass fraction burned at MBT spark timing. The quantity 0 50%mfb,MBT 0 is 50%mf- is the combustion retard. Under near-knocking conditions, the relationship between spark timing and location of 50% mass fraction burned is affected by early chemical heat release. Figure 3.24 shows the heat release profiles averaged over 300 cycles for several spark timings. The solid lines are from data taken using toluene as a fuel. The dashed lines are from data taken using TRFs with octane numbers that yield close-to-knocking conditions (i.e. advancing the spark timing by one degree would produce audible knock). As the spark timing becomes more retarded and fuel octane number is decreased to maintain near-knocking conditions, early (non-flame) heat release becomes more significant and flame speed appears to increase. Under severely retarded conditions there are two distinct periods of heat release - one from spontaneous reactions in the unburned gas, which peaks near top center crank position, and a later one from the propagation of the turbulent flame. These reactions, although they do not proceed to full autoignition and knock does not occur, result in a shortening of the burn durations. Figure 3.25 shows that the reduced burn duration causes 50% mass fraction burned to occur up to 4 degrees earlier for the same spark timing. Since the time of 50% mass fraction burned gives a better indication of when combustion is occurring, and since it correlates better to torque loss, combustion retard is used in the knock experiments as the indicator of combustion phasing. 3.2.2 Effects of Air-Fuel Ratio Figure 3.26 and Figure 3.27 display the combustion retard required to just avoid knock as a function of fuel RON and air-fuel ratio for PRFs and TRFs respectively. The stoichiometric data points were taken at WOT (Wide-Open Throttle) with no boost. Air71 fuel ratio was increased from stoichiometric by boosting inlet pressure to match NIMEP at MBT timing to the stoichiometric WOT case. High-octane fuel (PRF 120 for PRF experiments, toluene for TRF experiments) was used when adjusting air-fuel ratio to ensure that timing was not knock-limited. The data is also available as 2D charts for each compression ratio and fuel type in Appendix B. Figure 3.28 shows the change in combustion retard for UTG91 and UTG96 with k at a compression ratio of 11.6:1, with reference fuel data plotted for comparison. There are distinct differences between the results for PRFs, for which combustion retard increases with k, and the results for TRFs, for which combustion retard decreases with k. The trends for the UTG fuels are roughly half way between the PRFs and the TRFs, which results in little net change with k. Figure 3.29, Figure 3.30, and Figure 3.31 show the normalized change in the combustion retard required to just avoid knock with changing k for the three compression ratios and a range of fuels. These charts show more clearly that PRFs require increasing combustion retard with increasing k. The increase is mild at the 9.8:1 re, which agrees with previous findings [8], and becomes more severe at high rc. The increase is also more severe for the change from k of 1.3 to 1.6 than from 1.0 to 1.3. At all compression ratios the trend for TRFs is a mild decrease in the amount of combustion retard required with increasing k. Again, it is apparent that the trend for UTG fuels is roughly in between the two types of reference fuels. To increase X in these experiments the inlet pressure was boosted, decreasing the ratio between the inlet pressure and the peak pressure. A possible reason why PRFs require more spark retard than TRFs is that the lower specific heat of TRFs causes peak end-gas temperatures to decrease more with the decreased pressure ratio. This effect is discussed in more detail in Section 4.4.2 . 3.2.3 Effects of Boost The boost sweeps were performed by setting MAP for the required amount of boost, then successively lowering the octane number of the fuel and measuring the amount of combustion retard required to avoid knock. Figure 3.32 and Figure 3.33 show a sample of the results for PRFs and TRFs respectively at k = 1.0 and r, = 11.6:1. Lines are drawn through points with the same fuel mixture. Figure 3.34 shows the same TRF data, except with the axes swapped and with lines of constant MAP drawn using the combustion retard-NIMEP relationship from Eq. ( 3.1). The vertical distance between the point on 72 the curve and the horizontal line at NIMEPMBT represents the torque lost from retarding spark to avoid knock. As MAP is increased, NIMEPMBT increases, but torque loss also increases. The complete sets of PRF and TRF combustion retard data for stoichiometric boosting are plotted in Figure 3.35 and Figure 3.36 respectively. The remaining 2D plots of stoichiometric boosting data and results for boosting at k of 1.3 are available in Appendix C. For the range of boost levels and fuel ON investigated, NIMEP always increases with MAP, indicating that the torque gained from increasing the manifold pressure is greater than the torque lost from retarding combustion. PRFs consistently require more combustion retard than TRFs as NIMEP is increased. One explanation is that for these boosting experiments, the peak pressure rises approximately proportionally to MAP. Since the pressure ratio through which the endgas is compressed does not change significantly, there is no significant change in compressed end-gas temperature. Simulated end-gas temperature profiles for boosted and non-boosted cases are discussed in Section 4.4.2 . Autoignition reaction rates for PRFs are more sensitive to pressure than for TRFs [20], so PRFs require more combustion retard to avoid knock. Figure 3.37 shows a comparison of UTG96 to PRF95 and TRF95 for boosting at a k = 1.3 and rc = 11.6:1. As was seen with the reaction of UTG fuels to increases in air-fuel ratio, UTG96 behaves approximately half way between PRF and TRF. This trend is also visible in Figure 3.38 and Figure 3.39, which show the increase of combustion retard with boosted airflow and boosted NIMEP respectively for almost all of the test points considered. (Points with very high combustion retard - greater than 300 - or with spark timing earlier than MBT were omitted.) Although there is some spread to the data, it is clear that PRFs, which require about 5' CA of combustion retard per bar NIMEP, need about three times as much combustion retard as TRFs when boosted to achieve the same NIMEP. 3.2.4 Effects of Plasmatron Reformate Addition Fuel reformate was added under two different operating conditions for each compression ratio. The first condition was stoichiometric with 40% boost, for which inlet pressure was boosted so that NIMEP at MBT spark timing was 1.4 times NIMEPMBT at unboosted WOT. The second condition was at =1 .3 with MAP boosted to match NIMEPMBT at 73 unboosted stoichiometric WOT. In all cases the boost pressure was set while operating with high ON fuel (PRF 120 for PRF experiments, toluene for TRF experiments) to ensure that timing was not knock-limited. The reformed fraction is defined as the mass of the fuel in the gaseous reformate, which is the mass of the carbon and hydrogen constituents, divided by the total mass of the liquid and reformed fuel. Figure 3.40 and Figure 3.41 show knock limited combustion retard and NIMEP for reformate addition to PRFs and TRFs respectively. Data for boosted stoichiometric operation at rc of 11.6:1 are shown. Data for other operating conditions and compression ratios can be found in Appendix D. As reformate is added, autoignition reactions are slowed and combustion retard can be decreased, resulting in increased NIMEP. Figure 3.42 shows how reformate addition can be used to recover torque lost from retarding timing to avoid knock with increased boost. As boost is applied NIMEP increases, but this increase is moderated because spark timing must be moved away from MBT timing to avoid knock. Then, as reformate is added, spark timing can be advanced and NIMEP increases further according to the relationship from Eq. ( 3.1). Reformate can also be used to recover torque lost from spark retard as compression ratio is increased, as shown in Figure 3.43. Figure 3.44 shows a comparison of reformate addition to UTG fuels, which represent gasoline, as compared to PRFs and TRFs at X = 1.3 and r, = 11.6:1. The results indicate that the decrease of combustion retard for gasolines with increased reformed fraction is roughly halfway between that of TRFs and PRFs with similar ON. The complete set of data for adding reformate at the stoichiometric boosted condition to PRFs and TRFs are shown as 3D surfaces in Figure 3.45 and Figure 3.46 respectively. The decrease of combustion retard with reformate addition for each fuel is normalized in Figure 3.47. Reformate addition is most effective when applied to PRFs, which are composed entirely of alkane hydrocarbons. The combustion retard required to just avoid knock decreases by about 2' CA per 3% reformed fraction. When applied to TRF fuels with octane numbers of 95 or lower, which are composed of more than 20% n-heptane, reformate addition is only slightly less effective than it is for PRF fuels. As the alkane content decreases and aromatic content increases, reformate addition appears to become less effective for TRFs to the point that there is no benefit for reformate addition to pure toluene. 74 Autoignition chemistry modeling, described in Section 4.5 , indicates that the hydrogen in the reformate slows down autoignition by converting hydroxy radicals to hydrogen radicals and water. Hydrogen radicals are not as effective at initiating the chain branching reaction sequence as hydroxy radicals are. In the toluene oxidation mechanism proposed by Emdee et al. [21], which is a reference for the LLNL detailed toluene mechanism [22], most of the initiation and propagation reactions involving the hydroxy radical have an analogous reaction with the hydrogen radical with similar rate constants. This indicates that the conversion of hydroxy radicals into hydrogen radicals by adding hydrogen rich reformate will not significantly affect the rate of the autoignition reactions for toluene. Since the hydrogen in the fuel reformate works to slow down the autoignition reactions in alkanes, reformate becomes more effective for TRFs as nheptane is added to decrease RON. 3.2.5 Effects of Compression Ratio All eleven operating conditions were repeated for each of three compression ratios; 9.8:1, 11.6:1, and 13.4:1. An important effect of increasing compression ratio is an improvement in the thermodynamic efficiency of the engine, resulting in higher torque output for equivalent manifold pressures. Figure 3.48 shows how NIMEP at MBT spark timing and stoichiometric, WOT conditions changes with rc. There is some scatter in the points due to changing environmental factors such as atmospheric pressure, temperature, and humidity. The data for each compression ratio is averaged and plotted as the percentage increase of NIMEP with rc in Figure 3.49. Extrapolating the curve, the maximum NIMEP increase is about 9% and occurs at an rc of about 14:1. The equation for the quadratic curve fit is: NIMEP NIMEPMBT,9.8:1 0.126+ O.MBT 0.137RC - 0.00487RC 2 (3.2) Figure 3.50 and Figure 3.51 show how NIMEP and the combustion retard required to just avoid knock change with compression ratio for PRFs and TRFs, respectively. The data shown is for stoichiometric, unboosted, WOT conditions. Plots for the other operating conditions are available in Appendix E. As the compression ratio is increased, peak cylinder pressures and temperatures increase, so the spark must be retarded to avoid knock, resulting in increased combustion retard. Two effects influence NIMEP when rc is increased, as illustrated in Figure 3.52. The first is an increase in torque due to increased engine efficiency. The second is an increase in torque loss due to increased 75 combustion retard. At low compression ratios and mild amounts of spark retard, NIMEP increases with rc. At higher compression ratios and higher amounts of spark retard, NIMEP decreases with rc. For each fuel type and operating condition there is an optimum r, where engine torque output is at a maximum. For the conditions shown in Figure 3.50 and Figure 3.51 the optimum rc is about 11:1 for 95 RON fuels, which is similar to compression ratios used in modem automotive engines. The tradeoff between torque increase from higher thermal efficiency and torque decrease from retarding spark to avoid knock is discussed further in Section 5.1.1 . The average increase with compression ratio of combustion retard required to avoid knock is shown in Figure 3.53, separated by fuel type and rc interval. Points with very high combustion retard - greater than 300 - or with spark timing earlier than MBT were omitted. The bars, which reflect one standard deviation in each direction, indicate that there is significant scatter in the data. There do not appear to be any clear trends to the scatter save that for X, which is discussed in Section 3.2.2 . The average increase of combustion retard with compression ratio, about 3' CA per unit r", does not seem to be significantly affected by fuel type or rc range. 76 11 A A A A1 A A 0 - 10.5 - 10 - 9.5 * Lambda = 1.0, Toluene A Lambda = 1.3, Toluene * Lambda = 1.6, Toluene o Lambda = 1.0, PRF120 A Lambda = 1.3, PRF120 -9 c -8.5 IL * Lambda = 1.6, PRF120 0 0 A AD1 1 A A 30 A -8 13 A -7.5 -7 - 6.5 -6 50 60 -10 0 10 20 30 40 Spark Timing (*BTC) Figure 3.18 - Change of NIMEP with spark timing for a range of X; r, = 9.8:1, 1500 rpm. 1.05 mLambda = 1.0, Toluene A Lambda = 1.3, Toluene I * Lambda = 1.6, Toluene *A t- 0.95 - A DA I I a. A I 0.9 * Lambda = 1.0, PRF120 A Lambda = 1.3, PRF120 * Lambda = 1.6, PRF120 96 A IW A I z 0.85 0.8 - A I 96 I 0.75 0.7 A - -1 ) 0 Spark Retard 10 (esp-OSP,MBT, 20 30 *ATC) Figure 3.19 - Normalized decrease of NIMEP with spark retard for a range of X; r, = 9.8:1, 1500 rpm, 10.1 bar NIMEPMBT. 77 1.05 mRc = 9.8:1, Lambda = 1.0 A Rc = 9.8:1, Lambda = 1.3 I * * A * I iE 0.95 - 0.9 - 0.85 - 0.8 - 0.75 - ++ 9.8:1, Lambda = 1.6 11.6:1, Lambda = 1.0 11.6:1, Lambda = 1.3 11.6:1, Lambda = 1.6 o Rc = 13.4:1, Lambda = 1.0 A Rc = 13.4:1, Lambda = 1.3 o Rc = 13.4:1, Lambda = 1.6 0W w Rc = Rc = Rc = Rc = ~ A" 0Z A* A* 0.7 0 -10 20 10 Spark Retard (Osp-OP,MBT. 30 0 ATC) Figure 3.20 - Normalized decrease of NIMEP with spark retard for a range of X and re; toluene fuel, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at unboosted WOT X = 1.0. 50 * Lambda = 1.0, Toluene A Lambda = 1.3, Toluene * Lambda = 1.6, Toluene 4 1 40 I- A 0 A+' W 30 - 20 - c0 o Lambda = 1.0, PRF120 * Lambda = 1.3, PRF120 * Lambda = 1.6, PRF120 IIt 0 100 0- .0 -10 -10 0 10 Spark Retard 20 30 40 50 0 (OP- S,MBT, *ATC) Figure 3.21 - Change of combustion retard with spark retard for a range of X; r, = 9.8:1, 1500 rpm, 10.1 bar NIMEPMBT. 78 1.05 *rc9.8, I , 0.95 - + (- LU 0.9 - i a. 0.85 LU Z 0.8 K >%. 0.75 - + 4 7 . 0.7 - _ -1 0 Spark Retard 30 20 10 0 (0sp-Osp,MBTO 0ATC) 11.0, n1O.1 A rc9.8, 11.3, n10.1 * rc9.8, 11.6, nl0.1 Drc9.8, 11.0, n1O.1, prl120 A rc9.8, 11.3, nI0.1, prf20 * rc9.8, 11.6, n10.1, pr1120 X rc9.8, 11.3, n10.1, 15%ref * rc9.8, 11.3, n10.1, 30%ref * rc9.8,11.3, nl 1.6 + rc9.8, 11.3, n13.1 -rc9.8, 11.0, n12.1 -rc9.8, 11.0, n14.1 rc9.8, 11.0, n14.1, 15%ref 3 rcg.8, 11.0, n14.1, 30%ref A rc1.6, 11.0, n10.7 X rc1.6, 11.3, n10.7 * rc1 1.6, 11.6, n 10.7 * rcl 1.6, 11.3, n1 0.7, 15%ref + rcl1.6, 11.3, n10.7, 30%ref -rcl1.6, 11.3, n12.2 - rcl1.6, 11.3, n13.8 *rc1.6, 11.0, n12.7 * rcl 1.6, 11.0, n14.7 A rcl1.6, 11.0, n14.7, 15%ref X rcl1.6, 11.0, n14.7, 30%ref * rcl3.4, 11.0, nl1.0 + rcl3.4,11.3, n11.0 + rc1 3.4, 11.6, n 11.0 11.3, n11.0, 15%ref -1rc13.4, -rc13.4, 11.3, n11.0, 30%ref +rcl3.4, 11.3, n12.5 * rcl 3.4, 11.3, n14.0 * rcl 3.4, 11.0, n 13.3 Xrcl 3.4, 11.0, n1 5.4 * rcl3.4, 11.0, n15.4, 15%ref *rcl3.4, 11.0, n15.4, 30%ref Figure 3.22 - Change of normalized NIMEP with spark retard for a wide range of operating conditions and compression ratios; toluene fuel except where noted, 1500 rpm. 1.05- -Correlation Y = 1 - 0.168((1 + 4.443E-3XA2)A0.5 RMS Error = 0.005 1- - LU 0.95 - 0.9 - 0.85 - 0.8 - 0.75 - 0.7-10 - * 1) 20 10 Combustion Retard 0 0 ( 50%- 60%,MBT, 30 0 ATC) rc9.8, 11.3, n10.1 rc9.8, 11.6, n10.1 rc9.8, 11.0, n0.1, prfl120 A rc9.8, 11.3, n0.1, prf120 * rc9.8, 11.6, n10.1, p1f120 X rc9.8, 11.3, n1O.1, 15%ref K rcg.8, 11.3, n10.1, 30%ref * rc9.8,11.3, nl1.6 + rc9.8, 11.3, n13.1 - rc9.8, 11.0, n12.1 rc9.8, 11.0, n14.1 rc9.8, 11.0, n14.1, 15%ref o rc9.8, 11.0, n14.1, 30%ref A rc1.6, 11.0, n10.7 X rc1.6, 11.3, n10.7 )K rc1 1.6,11.6, n 10.7 * rc11.6,11.3, n10.7, 15%ref + rcl1.6, 11.3, nl0.7, 30%ref - rc1.6,11.3, n12.2 rc1.6,11.3, n13.8 Srcl1.6, 11.0, n12.7 * rc1.6, 11.0, n14.7 A rc11.6, 11.0, n14.7, 15%ref X rc11.6, 11.0, n14.7, 30%ref )K rc1 3.4, 11.0, n1 1.0 * rc13.4,11.3, n11.0 + rcl3.4,11.6, n11.0 - rc13.4, 11.3, nl1.0, 15%ref rc3.4, 11.3, n11.0, 30%ref + rc13.4, 11.3, n12.5 * rcl3.4, 11.3, n14.0 A rc13.4, 11.0, n13.3 x rc13.4, 11.0, n15.4 K rc13.4, 11.0, n15.4, 15%ref o rc13.4, 1.0, n154 30%ref * o ++ 0 rc9.8, 11.0, n1O.1 A 40 Figure 3.23 - Change of normalized NIMEP with combustion retard for a wide range of operating conditions and compression ratios; toluene fuel except where noted, 1500 rpm. 79 0.05 toluene O~p 28--28*BTC Se,= 0,,=16*BTC 90.04- .-...... TRF near knock 12*BTC -08,= 0 =8*BTC -- 0.03 -o 4.1P 0,,=4*BTC 0.02- 0.01 - -30 -15 -15 0 0 Cran 60 45 30 is 75 Crank Angle (*ATC) Figure 3.24 - Averaged heat release rate profiles for several spark timings for toluene and for TRFs with octane numbers that result in near knocking conditions; r,= 9.8:1, X = 1.3, 1500 rpm, MAP = 1.23bar (for NIMEPMBT 10.1bar). 40 -0-Toluene -+- TRF, near knock * 30 2 C *0 ,- - 0- 10- 0- E 0 -10 ' -10 0 I I 10 20 I 30 40 0 Spark Retard (Op.-0sp,MBT9 ATC) Figure 3.25 - Change of combustion retard with spark retard for toluene and for TRFs with octane numbers that result in near knocking conditions; rc = 9.8:1, x = 1.3, 1500 rpm, MAP = 1.23bar (for NIMEPMBT = 10.1bar). 80 - rc =13.4:1 rc = 11.6:1 = 9.8:1 -- - .rc '02 5- 20 15 .... 10 5 E 20 -5 ..... 80 1 .90 Lambda 1.4 100 ::. ::. 1610Fuel RON Figure 3.26 - Effects of X and PRF fuel RON on combustion retard to just avoid knock; 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. - -- .. rc = 13.4:1 rc = 11.6:1 rc = 9.8:1 - 30 25 - -...- 20 10 -.. .. 5 -- -5 100 1.4 Lambda 80 . -- 1 1.6 110 Fuel RON Figure 3.27 - Effects of X and TRF fuel RON on combustion retard to just avoid knock; 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 81 35 0 Torque Loss 30 - 2UO 25 - - +- -PRF95 . -- Ii-.- 20 - 15 - 10 - -. E 0 TRF95 --- UTG96 . PRF90 - .: 10% - -0- 0. (L) - -- 0'& ' ,-.---- --- TRF90 UTG91 3% 5- 1% 0- -5 i 0.9 1.1 1.3 Lambda 1.7 1.5 Figure 3.28 - Effect of A on combustion retard to just avoid knock for UTG fuels; r, = 11.6:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 12 PRF95 --- -e- PRF90 8- -e--PRF85 - PRF80 4- -*- 0 E) .0 -11 0 1 - - -- 0 -0- TRF90 TRF85 -0- U TRF95 TRF80 -4- .8 - -12 I 1.2 1.4 1.6 Lambda Figure 3.29 - Change of combustion retard to just avoid knock with increasing X; r, = 9.8:1, 1500 rpm, MAP at X> 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 82 12 - -M- PRF105 -h- PRFIOO 8- - PRF95 PRF90 -U- 4- .0 -A- TRF100 01: - . - %- A. ':--: E -0 .-- -4- -0- TRF95 -I- TRF90 - UTG96 - - - -a - -11 -0 UTG91 -8 -12 1 1.2 1.6 1.4 Lambda Figure 3.30 - Change of combustion retard to just avoid knock with increasing X; r, = 11.6:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. - 12 8 -)K-- PRF110 -X- PRF105 -+- PRFI 00 ... -+-PRF95 -a 4 0 - - -- - -12 TRF105 -A- TRF100 -0- TRF95 - -o- -UTG96 -4- -8 -X- - ' I 1.2 1.4 1.6 Lambda Figure 3.31 - Change of combustion retard to just avoid knock with increasing X; r, = 13.4:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 83 35 I- Torque Loss -PRF115 -W-PRF110 30 - -X-PRF105 25 l -e-PRFI 00 CD 9 20 - 10% -+PRF95 U PRF90 150 7E 10 - 3% 5 .0 E 0 1% ----- 0 ------- 0. -5 i I I 1 9 10 11 12 ---------1 13 1 15 1 14 15 16 NIMEP (bar) Figure 3.32 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; r,= 11.6:1, X = 1.0, 1500 rpm. 35 - Torque Loss -- 30 - 20% oI TRF105 -*--TRF100 0 IP 25 9 -- TRF95 - -U--TRF90 20 - 10 0 15 - 10 - 3% 5 1% 0 0 e-0000 U 0 - - ------------- -5 9 10 11 12 13 114 15 16 NIMEP (bar) Figure 3.33 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; r,= 11.6:1, X = 1.0, 1500 rpm. 84 16 X TRF105 A TRF100 15 - IL .0 14 - 13 - 12 * TRF95 * TRF90 -WOT torque loss - 20% boost .. ....... .................................. -40% boost - 11 10a 5 0 -5 15 10 Combustion Retard 25 20 (O05%-00%MBT, 30 35 *ATC) Figure 3.34 - Lines of constant MAP calculated from Eq. (3.1) imposed on data from Figure 3.33. Horizontal lines are drawn at NIMEPMBT- Vertical distance between curves and lines is torque loss from retarding spark to avoid knock. .. 35 . -......... --........ Rc=13.4: Rc=1 1.6:1 Rc=9.8:1 . 35 25. Boosted .... 8 20 (b- 0 - .....: 1 5 ...... *0 L) 10 80. ... -... ...--.. ~ ............. ... 90 100 RON .... ::.-.. 16 .1n12............. 110 120 8 10 Boosted NIMEP (bar) Figure 3.35 - Effects of boosted NIMEP and PRF fuel RON on combustion retard to just avoid knock; X = 1.0, 1500 rpm. 85 -Rc=1 3.4:1 Rc=1 1.6:1 35 . ...........3 0 Rc=9.8:1 --...---....- -- -........ -- .0 20 g15 . -.......... 10 E 0 ........-.... 2.......... 1 1.... R8 N 120 t...... .N ME ( r 0 (bar)fe ONo omutonrtr Figure 3.36 - Effects of b~~~oosted NIMEP 00 to just avoid knock; X = 1.0, 1500 rpm. 12028 %IME,(bar sPRF95 35 oos1e 030 - 20%. .TR9 -+-UTG96 25 - Torqu-Los..-4-P-F9 .a 15-- C 1 30 03% -- 0 - -------- -- -- 5 - 1% E .5 -I 9 10 1 1 11 12 13 NIMEP (bar) Figure 3.37 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for UTG96, PRF95 and TRF95; re = 11.6:1, X = 1.0, 1500 rpm. 86 15 0 a) 10 - 0 0 ** D E 0 * PRFs o TRFs S S Se 0 UTG96 0 0 I- * - Linear (PRFs) - Linear (TRFs) - Linear (UTG96) 0 S00 5 0 -OOC b 0 0 1.4 1.3 1.2 1.1 1 massfiowalrmassflowalr,unboosted Figure 3.38 - Increase of combustion retard with increased airflow rate from boosting at 1500 rpm, PRFs at X = 1.0 and X = 1.3, TRFs at X = 1.0 and X = 1.3, and UTG96 at X = 1.3. Data is from all three compression ratios. 15 S + UTG96 S 0* 7E * e PRFs o TRFs 0 0 10 - 0 - Linear (PRFs) ** Linear (TRFs) - 5 * 000-0 E 0 Se0* S. 5- 59 * 0 * o00 Q-6 Linear (UTG96) 0 e-< 5 P0 000 0 0 1 2 3 4 NIMEP-NIMEPunboosted (bar) Figure 3.39 - Increase of combustion retard with increased NIMEP from boosting at 1500 rpm, PRFs at , = 1.0 and X = 1.3, TRFs at X = 1.0 and k = 1.3, and UTG96 at X = 1.3. Data is from all three compression ratios. 87 16 35 --- PRFI10 C.R. U 30 - 25 - 20 - 15 ... ;.X . .- - --- 14 I- 0 E 15 I10 13 I uJ - -- -- - - --- - - - - -- - -X- PRF105 C.R. -- PRF100 C.R. -+-- PRF95 C.R. --- PRF90 C.R. 'PRF110 - -- - -x- -PRF105 NIMEP 12 5- NIMEP .0 E 04 11 0- e- - -PRFIOO NIMEP - PRF95 - - NIMEP 10 -5 0 5 15 10 20 30 25 - - 35 PRF90 NIMEP Reformate Fraction (%) Figure 3.40 - Decrease of combustion retard to just avoid knock and associated NIMEP increase with increased reformed fuel fraction for PRFs; r, = 11.6:1, x = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 35 - - 16 -- 4-TRF105 C.R. 30 -- 15 .~~~~~~~. 25 . . . . . . -. - -. -. . . -- TRF100 C.R. - x - -- U TRF95 C.R. -14 20 - 15 - 0 E 10 - -U-TRF90 -- 13 a. ------------ -- - 0 -x- TRFIO5 NIMEP --- - - -- 12 (.. E C.R. - - -A- TRF100 NIMEP 5: - TRF95 - .a. TRF90 11 0-5 - 0 5 10 15 20 25 30 10 NIMEP NIMEP 35 Reformate Fraction (%) Figure 3.41 - Decrease of combustion retard to just avoid knock and associated NIMEP increase with increased reformed fuel fraction for TRFs; re = 11.6:1, x = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 88 16 * TRF95 C.R. 15 -40% boost ......... 1,,,.................................................................. - i 14 c- 30% ref. 15% ref. - torque loss 40% boost 13 - U 0. z 12 20% boost 11 - 10 - WOT, no boost a -5 0 5 30 25 20 15 10 35 0 0 Combustion Retard (6 o%- 60%MBT, *ATC) Figure 3.42 - Increase of combustion retard with boost and decrease of combustion retard with reformate fraction for TRF95; r, = 11.6:1, X = 1.0, 1500 rpm. Curve for NIMEP vs. combustion retard at 40% boost calculated from Eq. ( 3.1). 12 * TRF90 C.R. - 11 - .. .................... ... ............................ torque loss 15% re . IL (U Rc=13.4:1 10 - 11.6:1 z 9.8:1 9 13.4:1 8 -5 0 5 10 15 20 25 30 35 0 0 Combustion Retard (Oso%- 5O%MBT9 ATC) Figure 3.43 - Increase of combustion retard with r, and decrease of combustion retard with reformate fraction for TRF95; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at unboosted WOT X = 1.0. Curve for NIMEP vs. combustion retard for rc= 13.4:1 calculated from Eq. ( 3.1). 89 Torque Loss ) 30 - 20% -- UTG96 C.R. -E- UTG91 C.R. 25 - +- PRF95 - TRF95 C.R. PRF90 C.R. ' -C.R. 9 20 10% 215 W: - 10 - 3% ----.. '-. -C.R. C .0 E U 04 -,-- -- ---- TRF90 - 5 - 1%' 0 -- - - - - - - - - - - - - - - - - - - -'----50 10 5 15 20 35 30 25 Reformate Fraction (%) Figure 3.44 - Decrease of combustion retard to just avoid knock with increased reformed fuel fraction for UTG fuels with PRFs and TRFs for comparison; r, = 11.6:1, X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). ----. .... Rc= 13.4:1 Rc=11.6:1 Rc=9.8:1 40 030 G30 - .......... -... ... ... 25 20 15 .10 E 0 -5 -. .......... .... ... 20 Reformate Fraction (%) 100 -... 010 30 110 RON Figure 3.45 - Effects of reformed fuel fraction and PRF fuel RON on combustion retard to just avoid knock; k = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 90 .............. ........ R c= 1 3 .4:1 Rc=1 1.6:1 ...... Rc=9.8:1 . 3-~ 20'R 5 .. 8 215-.8 10-- 0 .5-... 0~2 ........R... Reformats Fraction (%) 3 2.... -...... 5 Figure 3.46 - Effects of reformed fuel fraction and TRF fuel RON on combustion retard to just avoid knock; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 TR.0 '.. ... 80 times unboosted NIMEPMBT). 30 * PRFs 00 10 ....... -..... a TRFBO-TRF95 -20 5 0 - .2 15TRF1O5 -x E10 -5 0 5 25 Reformate Fraction (%) 10 15 20 30 35 Figure 3.47 - Decrease of combustion retard with increased reformed fuel fraction at 1500 rpm, PRFs at X = 1.0 (40% boost) and X = 1.3, TRFs at X =1.0 (40% boost) and X=1.3, and UTGs at A = 1.3. Data is from all three compression ratios. 91 11.4 A 1st Toluene A 2nd Toluene 11.2 - * Ist PRF120 * 2nd PRF120 I.. .0 * 1st Toliso * Average 11 10.8 - 10.6 - 10.4 - - Poly. (Average) AA 2 10.2 10 -9 10 12 11 14 13 Figure 3.48 - Increase of unboosted NIMEP at MBT spark timing with r, for three fuel types; X = 1.0, 1500 rpm. Closed symbols represent the first high load runs with new pistons, open symbols represent runs made after several engine hours at high load. 9 y = -4.87E-01x 2 + 1.37E+Olx - 8.74E+01 8 0 j7 6K 0 -5 w 4 3 2 0 9 10 12 11 13 14 Rc Figure 3.49 - Normalized average increase of WOT NIMEP at MBT timing with r,; ) = 1.0, 1500 rpm. Raw data is from Figure 3.48. 92 35 11 ~ 0to- 30 e( b •• 0 .. - ~PRF105 ........ --~ C.R. • • • • • • • b • • • • • • • • • • • • "JO. ~--.- .: 25 III ~ ""'-PRF100 C.R. 10 --'-PRF95 C.R. ~ 0 • • • • • • • • • • e on cp 20 ~ 9 ~ 0 on s '0 15 ---PRF90 C.R. :E • -x· ' PRF105 NIMEP 8 Z (1) .. ~ ecu Il. W ~ cu 0:: - 10 t: • -co- 'PRF100 0 U) :::l NIMEP 5 .c 7 E 0 0 - -¢- - -0. ' PRF95 NIMEP ------------------ 0 'PRF90 NIMEP 6 -5 9 11 10 12 13 14 Rc Figure 3.52 - Increase of combustion retard to just avoid knock and change of NIMEP with increased rc for PRFs; A = 1.0, 1500 rpm, unboosted WOT. 40 0 to- 11 35 . e( 0 b ••• .:III 30 _ .. - .. _ .. ---'_ ... _-- ~ on cp . -- ~ 25 9 ~ 0 on S .. cu w 15 - "JO.. - -<>- :E 8 Z (1) t: ~ cu Il. 0:: 0 - ---TRF90 C.R. e 20 '0 ~ --+-TRF95 C.R. 10 ~ 0 ---TRF100 C.R. 10 - o[J. TRF100 NIMEP TRF95 NIMEP TRF90 NIMEP U) :::l .c 5 7 E 0 0 0 -5 +-----~------~------.-----~------+6 9 10 12 11 13 14 Rc Figure 3.53 - Increase of combustion retard to just avoid knock and change of NIMEP with increased rc for TRFs; A = 1.0, 1500 rpm, unboosted WOT. 93 A TRF100 C.R. 11 - ............ torque loss .... ........... . 10 - * TRF95 C.R. l t .................. ........ * TRF90 C.R. -9.8:1 (L -11.6:1 (U 9 -13.4:1 8 7 0 -5 5 10 15 Combustion Retard 20 0 35 30 25 ( 60%-8o%MBT, 40 0 ATC) Figure 3.52 - Curves for NIMEP vs. combustion retard from Eq. (3.1) imposed on data from Figure 3.51. Vertical distance between curves and horizontal lines at NIMEPMBT is torque loss from retarding spark to avoid knock. 0-%6 4 3 II I I 0 (2 E 0 0 9.8:1-11.6:1 PRF 11.6:1-13.4:1 PRF 9.8:1-11.6:1 TRF 11.6:1-13.4:1 TRF Figure 3.53 - Average increase in combustion retard to avoid knock over all operating conditions considered, organized by fuel type and compression ratio interval. Bars represent one standard deviation. 94 CHAPTER 4. CHEMISTRY MODELING The results of a knock model under development at MIT were compared to experimental data to ascertain the sensitivity of the model to input parameters and define the range of the models validity. It was found that high accuracy in the determination of initial conditions and thermodynamic environment is required for consistent results. Calibration of baseline initial conditions was required to match experimental data. The model successfully predicted the response of PRFs to compression ratio and air-fuel ratio and the response of TRFs to boost. The difference between the response of PRFs and TRFs to air-fuel ratio was also captured. Constant volume simulations were used to help identify the mechanism by which hydrogen and carbon monoxide affect autoignition. 4.1 BACKGROUND Mechanisms for predicting the response of autoignition time to thermodynamic conditions are available in varying levels of complexity and accuracy. Most assume that the rates of reaction of the end-gas have an exponential temperature dependence combined with a pressure dependent term. One of the simplest forms is to invert a single empirical reaction rate equation to define an induction time and then integrate the induction time over the evolution of the pressure and temperature profile of the end-gas. Although useful for first-order approximations, this method is unable to reproduce the complicated temperature and pressure dependence of hydrocarbon autoignition reaction mechanisms. To better replicate the chemical kinetics of hydrocarbon oxidation, mechanisms involving multiple reactions have been developed. These range from simple or "reduced" mechanisms with several species and reactions to "detailed" mechanisms with thousands of species and reactions. One can still consider detailed mechanism to be reduced from millions of possible species and reactions. This study utilized detailed mechanisms designed by Lawrence Livermore National Laboratories. The LLNL PRF mechanism is a combination of the LLNL isooctane mechanism [23] and n-heptane mechanism [24]. It was selected because it has been shown to be valid over a wide range of thermodynamic conditions and because it is well supported by LLNL researchers. The TRF mechanism was created for this study by LLNL researcher William Pitz. It is a combination of the n-heptane mechanism and a toluene mechanism [22]. 95 4.2 MODEL DESCRIPTION The model is designed to simulate an adiabatic homogenous element of end-gas subjected to the cylinder pressure generated by compression and combustion. It was assumed that, prior to autoignition, pressure is uniform throughout the cylinder. The simulation starts at the time of intake valve close and proceeds until autoignition or until a specified point late in the expansion stroke if autoignition does not occur. The time of autoignition is assumed to be the point at which the temperature rises faster than a critical rate. A rate of 2 x 106 K/s, or 222 K/0 CA at 1500 rpm, was found to give good accuracy while maintaining numerical stability. 4.3 GOVERNING EQUATIONS The structure of the simulation is shown in Figure 4.1. The rate constants and species thermodynamics are taken from the PRF or TRF detailed reaction mechanisms. The initial conditions consist of the molar concentrations of the reactants and the initial temperature. The thermodynamic constraint for the engine knock simulation is the cylinder pressure. For the constant volume simulation it is the volume, which is calculated from the initial pressure using the ideal gas law. The molar production rates for each species in the mechanism are calculated according to the equation: dC, dt k U ' j ... (4.1) rx Where C, is the molar concentration of species i, Jis the set of reactions for which i is a product (both forward and reverse), Cs are the concentrations of the reactants in reactionj, and ovj is the stoichiometric coefficient for species i in reactionj. The rate coefficient, k is generally calculated according to the Arrhenius equation: ki = A T"' exp ~ I RT) (4.2) Where T is the temperature, R is the ideal gas constant (molar), and A1 , nj, and Ej are, respectively, the pre-exponential factor, the temperature exponent, and the activation 96 energy for reactionj given by the reaction mechanism. The rate constant is multiplied by a pressure dependent term for some reactions. The internal energy and specific heat for each species is calculated from a polynomial fit to temperature given by the thermodynamics section of the reaction mechanism. ) -'N Equations molar production rates from reactions conservation of mass equation of state (Ideal Gas Law) conservation of energy Inputs species & species thermodynamics reactions & reaction rate constants (A,E,n) initial conditions const raints (ie. P(t) or V(t)) Numerical Integrator Outputs species concentrations vs. time thermodynamic state (ie. temperature) vs. time autoignition time (derived) Figure 4.1 - Structure of an autoignition simulation. For the engine knock simulation using cylinder pressure as a constraint the ideal gas law is represented by: p = MaveP RT (4.3) Where p is the density, Mave is the average molecular weight for the mixture, and P is the total pressure. The equation for conservation of mass is written for each species, and takes the form: dc dt dC. dt irX ~ 1 dP (P dt 1 dT 1 T dt (4.4) 97 Where the first term is the change of concentration due molar production from reactions and the second term is the change of concentration due to changing mixture density. The base equation for conservation of energy is: dU = SQ - SW (4.5) Since heat transfer is neglected, 6Q is zero. Internal energy, dU changes due to the change of thermodynamic state and due to changes in mixture composition: dU = cVdT + I dmiu, (4.6) Where c, is the specific heat at constant volume, dmi is the differential change of mass of species i, and ui is the internal energy for species i. Using the definition of work and the ideal gas law: SW = Pdv = RdT - vdP (4.7) Where v is the specific volume of the mixture. Equating the two and collecting terms: cdT +Z dmju, =vdP (4.8) Where c, is the specific heat at constant pressure. For use in the model molar quantities were used and the equation was rearranged and differentiated in time: ,Pp dT Mave dt ~ dC 'idt dP dt For the constant volume simulations the calculation of reaction rates and species thermodynamics is the same as described above. The ideal gas equation of state and conservation equations are similar to those shown above, but rearranged to use volume as the constraint instead of pressure. 4.4 ENGINE KNOCK SIMULATIONS The Jacobian IDE solver by Numerica Technology LLC was used to integrate the engine knock simulation. Numerica Technology's Open Chem Pro routines were used to calculate molar production rates and species thermodynamics. The rest of the equations 98 were written directly into the IDE. Each simulation from intake valve closing time to autoignition takes approximately 30 minutes to run on a modem PC. 4.4.1 Selection of Input Parameters Besides the chemistry mechanism, the other key inputs to the knock simulation are the pressure trace, the initial species concentrations, the initial pressure, and the initial temperature. For this application these inputs were derived from the detailed experimental data corresponding to the results presented in Section 3.2 . Analysis of pressure data by Topinka et al. indicates that at borderline knocking conditions, cycles with peak pressure locations that are more than one standard deviation earlier than the mean are most likely to knock [14]. Thus, for previous work, data from a cycle with a peak pressure that is one standard deviation earlier than the mean was selected as an input to the knock model. The method used in this work is similar, but needed to be modified for the conditions under which data was collected. Instead of using peak pressure as an indicator of combustion phasing, the location of 50% mass fraction burned was used. This was necessary because when spark is significantly retarded, as it is for many of the data points in this study, there can be two pressure maxima, one near TDC from compression, and a later one from combustion. Also, since the data taken in this study was at a spark timing one degree later than that for borderline knock, not as many cycles are likely to autoignite. Observations of pressure data, such as that shown in Figure 4.2, indicate that under near-knock conditions approximately the earliest 10% of cycles show mild autoignition. Thus the pressure trace selected from a set of data for input to the knock model was one with a location of 50% mass fraction burned that is earlier than 90% of the cycles in the set (i.e. 1 0 th percentile early). There are three components to the intake charge that must be taken into account when determining the in-cylinder mixture composition. They are intake air, fuel, and residual gas. The air flow and fuel flow were taken directly from experimental measurements. The humidity of the air was measured and taken into account. Residual mass fractions were approximated using linear fits to results from a Ricardo WAVE 1-dimensional fluid flow and engine simulation. An example of the simulation results is shown in Figure 4.3. Since the valve overlap on the experimental engine is small, most of the residual is from the cylinder clearance volume, which is at approximately atmospheric pressure at the end of the exhaust stroke. Increased residual temperature decreases density and thus decreases residual mass, which explains the downward trend of residual fraction with 99 increased spark retard. It also decreases with increased charge mass due to increased dilution. For stoichiometric boosted operation the model residual output scales inversely with charge mass to within 2% of the absolute value. 45 - 40 Earlier than 90% of cycles Later than 90% of cycles - INi 30 U) M. -"D~C"- 4,, - *44 - . 25 - 20 15 - i 0 10 20 30 40 50 Crank Angle (*ATC) Figure 4.2 - A set of 90 cycles of pressure data taken under near knocking conditions with PRF90; 1500rpm, WOT, X = 1.0. The solid lines are cycles that have a location of 50% mass fraction burned that is earlier than 90% of the cycles. Initial temperature was estimated for unboosted data points using the ideal gas law with the volume at intake valve close, the measured manifold pressure, and the number of moles of charge in the cylinder calculated from air flow, fuel flow, and residual fraction: Tec = PintVeIVC (4.10) nchR Where Pin, is the intake manifold pressure and nch is the number of moles of charge in the cylinder. Since compression begins before the actual intake valve closing time, an 100 effective intake valve closing time, VeIvc, was estimated by comparing cylinder pressure data to a polytropic compression curve. Figure 4.4 shows that the two curves become aligned at about 400 ABC. 5 - - 4 0 - UU 0L 32- 1-~ -- Rc=9.8:1 y(9.8) = -0.0354x + 4.1435 -- Rc=1 1.6:1 y(11.6) = -0.0381x + 3.9022 +Rc=1 3.4:1 y(13.4) = -0.0414x + 3.6633 0 10 0 20 30 e50%mfb ( 0ATC) 40 50 Figure 4.3 - Residual fractions calculated by a Ricardo WAVE engine simulation; X = 1.3, 1500 rpm, MAP = 1.22 bar. 2C1.8 - - Measured pressure --- Polytropic compression (k=1.29) Effective, IVC,' - CL 1.4 1.2 - 1 180 20 0 220 240 260 Crank Angle (*ATC) Figure 4.4 - Comparison of measured cylinder pressure to polytropic compression to find effective intake valve closing time; 1500 rpm, X = 1. 5. 101 Under boosted conditions the throttle had to be closed part way to control air flow. This induced pressure fluctuations in the intake manifold and made experimental manifold pressure measurements unreliable. For these cases, the WAVE engine model was used. For the unboosted baseline case, WAVE predicted an initial temperature 11 K lower than the experimental value calculated from the ideal gas law. The error is most likely due to errors in modeled heat transfer coefficients and surface temperatures. Instead of using the WAVE predicted values directly, the variation of the predicted initial temperature was added to the experimentally estimated initial temperature from the baseline case. The results are shown in Table 4.1. The intake manifold pressure was then estimated using the ideal gas law: P= (4.11) Ivc VeIVC nch Table 4.1 -Temperatures at intake valve close estimated by WAVE engine model. TeIVC (K) X=1.0, unboosted X = 1.3, X = 1.6, 20% 40% boosted to boosted to boosted boosted NIMEPMBT, maintain maintain NIMEPMBT, NIMEPMBT NIMEPMBT X = 1.0 = 1.0 WAVE predicted: 339 339 339 336 334 Adjusted Adjistd 351 (from ideal 351 351 348 346 experimental: gas law) I 4.4.2 Pressure-Temperature Profiles In order to gain an understanding of the temperatures that SI engine end-gas is subjected to in response to cylinder pressure, simulations were run with the chemical kinetics disabled. The results help to explain the trends in knock behavior with changes in operating conditions. Figure 4.5 shows the pressure and temperature profiles for two air fuel ratios, one with a X of 1.0 and one with a X of 1.6. Spark timing is set for MBT and NIMEP is the same for both cases. Even though cylinder pressure is higher for the lean case, the peak end-gas temperature is slightly lower. For adiabatic compression, the increase in temperature is 102 dependant on the pressure ratio through which the mixture is compressed. The increased initial pressure from boosting to maintain constant NIMEP decreases the pressure ratio and thus decreases the temperature ratio. The decreased pressure ratio from boosting more than compensates for the increase in temperature from the increased y for lean mixtures. 1000 100 1.0 Pressure -Lambda 90 - ----- Lambda 1.6 Pressure 900 -Lambda 1.0 Temperature - - - Lambda 1.6 Temperature 80 800 70 cc 60 700 50 cc 600 40 - 500 30 20 - - -400 10 - 0 1 -90 1 -75 1 -60 300 -45 -30 -15 0 15 30 45 Crank Angle (*ATC) Figure 4.5 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for stoichiometric and lean air-fuel ratios; iso-octane fuel, 1500 rpm, r, = 9.8:1, NIMEP = 10.1 bar, MBT spark timing. Pressure and temperature profiles for boosted operation are compared to unboosted operation in Figure 4.6. As with lean operation, although the pressure for the boosted case is increased, the pressure ratio remains relatively constant, so peak temperatures do not change significantly. Increasing compression ratio increases both pressure and temperature, as shown in Figure 4.7. Inlet pressure remains constant, but peak pressure increases. This increases the pressure ratio and so increases the temperature ratio. The initial temperature is slightly lower due to decreased residual fraction, but its effect is insignificant compared to the increase from higher compression pressures. 103 100 90 - 1Obar NIMEP Pressure .. l.. 14bar NIMEP Pressure -1Obar 80 I 1000 - - - 14bar NIMEP Temperature - 800 70 IC 40 ~0 .0 I- 900 - NIMEP Temperature 60 - 700 2 - 600 CL - 500 - 400 50 40 30 - E 20 10 .- 300 0 -90 -75 -60 -45 -30 -15 0 15 30 45 Crank Angle (*ATC) Figure 4.6 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for unboosted and boosted conditions; iso-octane fuel, 1500 rpm, r, = 9.8:1, X = 1.0, MBT spark timing. 100 1000 a 9 8:1 Drafssre I ------ 13.4:1 Pressure 9.8:1 Temperature 90 80 I- -900 - - 13.4:1 Temperature 70 cc .01 60 - U) 50 - 40 - 30 - - 800 - 700 2 -e I- -600 E0. 20 10 - 400 - 300 - -90 -75 -60 -45 -30 -15 0 15 30 45 Crank Angle (*ATC) Figure 4.7 - Experimental pressure profiles and predicted non-reacting end-gas temperature profiles for low and high compression ratios; iso-octane fuel, 1500 rpm, MAP = 1 bar, MBT spark timing. 104 Figure 4.8 shows the differences in temperature profiles between mixtures of air and PRF 100 (iso-octane) and mixtures of air and TRF 100 (82% toluene). Toluene has lower specific heats than iso-octane does, which causes it to have a higher y. At 600 K the stoichiometric PRF mixture has a y of 1.311 while the stoichiometric TRF mixture has a y of 1.323. For adiabatic compression a higher y increases the change in temperature in response to a given change in pressure. The result is an increase in peak temperature. I 1000 100 90 - Pressure PRF100 Temperature - - 900 - - - TRFIOO Temperature 80 70 VU IL -800 I- - 60 I- - 700 2 - 600 E /- 50 40 30 - 500 - 400 20 101 300 -90 -75 -60 -45 -30 -15 0 15 30 45 Crank Angle (*ATC) Figure 4.8 -Predicted non-reacting end-gas temperature profiles with the same pressure profile for PRF100 and TRF100 fuels; 1500 rpm, re = 9.8:1, MAP = 1.0 bar, X = 1.0, MBT spark timing. Since TRF fuels have a higher y, they are more sensitive to changes in pressure ratio than PRF fuels are. The differences in y between TRFs and PRFs help to explain the differences between the trends of knock limited spark timing with air-fuel ratio presented in Section 3.2.2 . In the experiments the air-fuel ratio was adjusted by increasing the inlet pressure. This reduced the ratio between inlet pressure and peak pressure. Since TRF fuels have a higher y, the peak temperatures for TRF fuels decrease more with decreased pressure ratio than they do for PRF fuels. The decreased peak temperature for TRF fuels compensates for the increased reaction rates from increased absolute pressure. For PRF 105 fuels the decrease in peak temperature is not enough to compensate for the increased reaction rates at elevated pressures, so more spark retard is required. Figure 4.9 shows the end-gas temperature profiles for two PRF95 simulations, one with the chemical kinetics enabled and one with the chemical kinetics disabled. For the simulation with kinetics enabled the rapid temperature rise at 270 ATC represents autoignition. The temperature from the reacting simulation does not depart from the nonreacting one until about 100 CA before autoignition at a temperature of about 950 K. 60 50 - I 1200 Pressure Inert Temperature - - - Reacting Temperature - 1100 - 1000 40 - - I.- L 30 - 900 - 800 L 0 - 700 20 - E - 600 10 - 0 -75 -60 -45 -30 -15 0 15 30 - 600 ' 400 45 Crank Angle (*ATC) Figure 4.9 - Predicted end-gas temperature profiles with the same pressure profile for simulations with reaction kinetics disabled and enabled for PRF95; 1500 rpm, re = 9.8:1, MAP = lbar, X = 1.0, O,p = 6'BTC. 4.4.3 Model Sensitivity Qualitative observations of pressure data such as that shown in Figure 4.2 indicate that under near knock conditions, autoignition occurs near where peak pressure would be in a non-knocking cycle. To assess the accuracy required of the model input parameters, the variation of the autoignition time with respect to the time of peak pressure for variations of each parameter was noted. Figure 4.10 shows the effect of changing initial temperature on predicted autoignition time for a simulation with PRF95. For autoignition 106 to occur near peak pressure, an initial temperature of about 423 K is required. This temperature is about 70 K higher that that estimated using the method described in Section 4.4.1 . Since the actual value of the initial temperature of the portion of the endgas that autoignites is not well known, initial temperature was used as a variable for calibrating the model. Variation of autoignition time with initial temperature for autoignition times near peak pressure is about 20 CA per 5 K. The rest of the results of the sensitivity study are listed in Table 4.2. 50 -1200 45 - Pressure - Temperature ",j=4iK / Tjng=41OK 1100 TjT=405K 40/ 1000 * 35 - E Tinj=400K 1000 30 900 25 800 20 0 30 15 Crank Angle (0ATC) 45 Figure 4.10 - Effect of initial temperature on predicted autoignition time in increments of 5 K for PRF95; 1500 rpm, r, = 9.8:1, MAP = lbar, X = 1.0, , = 6' BTC. 4.4.4 Comparison to Experimental Results In order to assess the model's accuracy, it was compared to experimental results for lambda sweeps, boost sweeps, and compression ratio sweeps for PRF95 and TRF95. For each fuel, the initial temperature was calibrated at the baseline condition (rc = 9.8:1, X = 1.0, MAP = 1.Obar, spark timing for near knock) so that autoignition occurred at peak pressure. As operating conditions were changed from the baseline conditions, the initial 107 temperature in the model was adjusted by the same amount that the estimated experimental initial temperature changed. Table 4.2 - Sensitivity of autoignition time to model input parameters. AOign.pp is the distance in crank angle degrees from autoignition to peak-pressure. Parameter: Effect on AOign-pp (0CA): Initial temperature Fuel octane 2' advance per 5 K increase 20 retard per 5 ON increase number Manifold pressure (for pegging P,,,) Pressure trace location of 50%mfb 20 retard per 20 retard per 30 retardof i.ncr Imicrease ____m_ For each operating condition, the model was run with pressure traces extracted from spark timing sweep data under the same condition. The pressure traces were selected from each set of data according to the criteria described in Section 4.4.1 . The spark timing for which autoignition occurred closest to peak pressure was interpreted as the knock limited spark timing, as shown in Figure 4.11. Figure 4.12 to Figure 4.17 show comparisons of the model output to experimental data along with the estimated experimental initial temperatures and those used for the simulation. The combustion retard shown is that corresponding to the predicted knock limited spark timing. The PRF simulations match the experimental results very well for changes in compression ratio and changes in air-fuel ratio. The model under predicts the combustion retard required to avoid knock for boosted operation. This is probably mostly due to errors in the estimated initial conditions. Combined errors in initial temperature and manifold pressure can add up to several degrees of error in the simulation results. The chemistry mechanism, which has not been completely verified under the conditions being explored, may also contribute to the difference. The TRF simulations overestimate the effects of compression ratio and air-fuel ratio, but the direction of the trends are correct. The errors may be partially due to errors in the initial conditions, but this combined reaction mechanism has not yet been well tested and is likely responsible for much of the difference. The boosted experiments appear well matched. 108 10 1200 55 55 0 6 0BTC 4 BTC 20 TC 0 ep=8 BTC 50 - 1100 45 - 1000 IL. 35 - E a' O*BTC 30 - 900 25- 800 20 -15 0 45 30 15 Crank Angle (*ATC) Figure 4.11 - Pressure traces from several spark timings and corresponding simulated end-gas temperature profiles; 1500 rpm, r, = 11.6:1, X = 1.0, MAP = 1.Obar. 40 BTC is interpreted as the predicted knock limited spark timing. 35 - I- 30 - 25 - 20 - 0 + PRF95 * Model PRF95 Exp. Teivc,.st=343K TKvc,.e=342K T.jvc,.t=351 K Ti. 15 - =I4 Tingt=423K 10 - 50 E 0 0-5 9 10 11 12 13 14 Rc Figure 4.12 - Comparison of model predicted combustion retard to experimental combustion retard with increasing compression ratio for PRF95; 1500 rpm, X = 1.0, MAP = 1.0bar. 109 35 - -- - od --lTjlf=30 TRF95 Model --- - 0 Ti,,I=350K 30 - -- +-TRF95 Exp. 0 25 - Tiit=348K CP 20 - 15 - , T.ves,.t=344K 9 T..vc,.=342K Tifft=357K ,10 T.jvc,..t=351K .0 E 0 C. 5 0 - ' 1 9 10 -5 ,I 12 14 13 12 11 Rc Figure 4.13 - Comparison of model predicted combustion retard to experimental combustion retard with increasing compression ratio for TRF95; 1500 rpm, X = 1.0, MAP = 1.0 bar. - 35 I- 30 - - * PRF95 Model --- PRF95 Exp. 25 o 0 C 20 - 15 - 10 - T.vc,..t=351 K Ti"t=423K T.vc,..t=351 K T.vc,..t=3 5 1K T101t=423K T10ft=423K 0 5E 0 L) 01---- - - - -------------------------5 0.9 1.1 1.3 1.5 1.7 Lambda Figure 4.14 - Comparison of model predicted combustion retard to experimental combustion retard with increasing air-fuel ratio for PRF95; 1500 rpm, r, = 9.8:1, NIMEPMBT= 10.1 bar. 110 35 -TRF95 Model - I- 30 -- +--TRF95 Exp. 0 0 So 0 (U VSo 25 20 15 T.ivce..t=35 1 K c- a) T.vc,..=351 K 10 Tinit=357K ' 0 .0 E 0 0. , T.s 6.-t=351 K 5 Ti =357K fT 0 , -- T=357K Tj, -5 1.5 1.3 1.1 0.9 1.7 Lambda Figure 4.15 - Comparison of model predicted combustion retard to experimental combustion retard with increasing air-fuel ratio for TRF95; 1500 rpm, r, = 9.8:1, NIMEPMBT = 10.1 bar. 35 30 - - * PRF95 Model -- * PRF95 Exp. TervC,..t=346K TeivC,..j=348K 25 - Teive,...=334 C 20 - T~jvcA=:.K Tikt=418K 15 - ,0- TifTn=420K C "0 0 .0 E .) T,,t=423K 10 5 - 0-5 8 9 10 11 12 13 14 NIMEP (bar) Figure 4.16 - Comparison of model predicted combustion retard to experimental combustion retard with increasing boosted NIMEP for PRF95; 1500 rpm, r, = 9.8:1, x = 1.0. 111 35 0 30 +TRF95 Model -- +-TRF95 Exp. 25 T. 1vC,.m=346K 2 CP 20Tev,..t=348K 0 15- T.vC,..=351 K 10 - C 05 - , Tjflt=352K '"t=354K +T Ti.K=357K 5- E 0 * ---------------------- 0-------5 - 8 9 I I I I 10 11 12 13 14 NIMEP (bar) Figure 4.17 - Comparison of model predicted combustion retard to experimental combustion retard with increaing boosted NIMEP, PRF95; 1500 rpm, r, = 9.8:1, X = 1.0. 4.5 CONSTANT VOLUME SIMULATIONS Constant volume autoignition simulations were used to gain insight into the mechanisms by which hydrogen and carbon monoxide affect autoignition of alkane fuels. Chemkin v3.7 Aurora by Reaction Design was used as the solver. A constant volume perfectly mixed adiabatic reactor simulator is included in the Aurora package. The user must supply the reaction mechanism, species thermodynamics, and initial conditions. To simulate conditions in an engine cylinder an initial pressure of 45 atm was selected. An initial temperature of 875 K was chosen so that both the low temperature and high temperature oxidation mechanisms were active and could be observed. For this study the species mole fractions were chosen to represent a mixture of air and iso-octane with X 1.5. Hydrogen and carbon monoxide were added as 5.5% and 7% of the fuel energy, respectively, similar to a mixture with 15% fuel reformed fraction. 112 = 4.5.1 Effects of Hydrogen Addition Figure 4.18 shows a comparison between the temperature and OH radical concentration profiles of autoignition simulations for iso-octane with air, and for iso-octane plus 5.5% H2 by energy with air. The two-stage nature of autoignition is evident in the evolution of the OH concentration. The first stage, consisting of rapid initiation and moderate heat release, lasts for about 2 ms. The second stage lasts until just before the time of autoignition. It is slower and less exothermic than the first stage. As expected, the simulation predicts that the autoignition time for the case with hydrogen addition is longer than that for just iso-octane. 1.E-02 3000 2500 - Temperature Mole Fraction -OH - CeH 18 C8H18 -- 1.E-04 + H2 0 I.E-06 2000 - U. L 0 E 1500 - 1.E-08 - 0 1000 0UU ---- - . 1 0.000 0.003 0.006 00 0.009 I.E-10 1 1.E-12 0.012 Time (s) Figure 4.18 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane and of iso-octane with 5.5% H2 by energy; Tiit= 875 K, Pi1it= 45 bar, X = 1.5. To evaluate the effect of thermal dilution from the added hydrogen, a new species was added to the simulation. The new species, nrH2, is thermodynamically identical to hydrogen, but it is non-reactive. An autoignition simulation for which H2 was replaced with nrH2 is compared to the original H2 addition case in Figure 4.19. The increased thermal dilution from adding hydrogen accounts for about half of the overall effect. The rest of the effect must be accounted for with changes in the chemical kinetics. 113 3000 1.E-02 -Temperature Mole Fraction -OH 1.E-04 2500 C8H18 + nrH2 C8H,8 +H 2 0 I 2000 - 1.E-06 I EU E 1500 - 0.0 .1.E-08 1.E-10 1000 - Soo 0.000 I.E-12 0.003 0.006 0.009 0.012 Time (s) Figure 4.19 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated non-reactive H2 (nrH2); Ti it= 875 K, Piit= 45 bar, X = 1.5. A limiting factor to the progression of the reactions leading to autoignition is the availability of OH radicals to react with the alkane to produce an alkyl radical. Figure 4.20 shows the rates of reaction for OH radicals reacting with iso-octane and hydrogen molecules, as compared to total OH consumption for the H2 addition and non-reacting nrH2 addition simulations. For the reacting H2 case there is a slightly wider gap between total OH consumption and OH consumption from the reaction with iso-octane. Consumption of OH by H2 is negligible for the nrH2 case, but is approaching the same order of magnitude as the reaction with iso-octane in the reacting H2 case. This indicates that due to reactions with H2, less OH is available to initiate the chain branching reaction sequence with iso-octane. A product of the reaction between H2 and OH are H radicals. Although H radicals also react with fuel molecules to initiate the chain branching sequence, Figure 4.21 shows that in both cases, the rate of this reaction is not significant compared to the reaction with OH. Closer inspection of the reaction mechanism shows that the rates for reactions of H with iso-octane are one to two orders of magnitude less than those for OH. 114 1.E+02 + nrH2 Sn 1.E+00 E Total OH consumption C811 + H2 - -C8H8+OH H2+OH => H+H20 - I- => C8H17+H20 0 Ii 0. E I .E-040 c. 1.E-08 0.009 0.006 0.003 0 0.012 Time (s) Figure 4.20 - OH consumption by reactions with fuel molecules for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated non-reactive H 2 (nrH2); Tinit= 875 K, Pinit= 45 bar, X = 1.5. 1.E+02 -C8H8+OH - cv> 1.E+00E CsH1 8 + nrH2 -0 0 S1.E-02 => C8H17+H20 C8H8+H => C8H17+H2 CBH 18 + H2 - 0. E . .. - , ... 1.E-04 - 1.E-06 - 0 1.E-08 'I 0 0.003 0.006 0.009 0.012 Time (s) Figure 4.21 - Rates of reaction of OH and H with iso-octane molecules for constantvolume autoignition simulations of iso-octane with 5.5% H 2 by energy, and of isooctane with simulated non-reactive H2 (nrH2); Tinit = 875 K, Piit= 45 bar, X = 1.5. 115 To test the theory that consumption of OH radicals by H2 is the reason for the nondilution portion of the increased autoignition delay, a different new species was added to the simulation. The second new species, prH2 , is similar to the non-reacting species nrH 2 , except that it is allowed to react with OH only through the reaction "prH 2 + OH => H + H2 0". The results of a simulation with prH2 are compared to a simulation with fully reacting H2 in Figure 4.22. The autoignition times of the two cases are almost identical. ---- . 1.E-02 - 1.E-04 2000 - - 1.E-06 t E 1500 - - 1.E-08 3000 - - Temperature OH Mole Fraction 2500- CBH 1 s C8H1 8 + H2 +prH 2 0 ME 0 1000 - 500 0.0 00 - . . . . 0.003 0.006 0.009 0.012 I.E-10 1.E-1 2 Time (s) Figure 4.22 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 5.5% H2 by energy, and of iso-octane with simulated H2 (prH2) that reacts only with OH; Tiit= 875 K, Pinit= 45 bar, ) = 1.5. The proposed mechanism by which H2 slows down the reactions leading to autoignition is shown in Figure 4.23. According the simulations run with the LLNL PRF mechanism, H2 intercepts the OH radicals produced by the chain branching reactions and turns them into H radicals. The H reaction with the alkane fuel is much slower than that for OH, so the rate at which the chain branching reaction sequence is initiated is reduced. This mechanism is similar to that proposed in [25], except that the participation of the H radical in the reaction sequence has been neglected. An effect that is not included in the reaction mechanism employed, and so was not investigated, is the reaction of H2 with radicals that are intermediate to the chain branching reaction sequence. These types of 116 reactions, however, would have the same effect as the reaction of H2 with OH, interrupting the chain branching sequence and slowing autoignition. High Temperature H202 -HO2 RH Olefin -4 RH+O2 2 -H20 R.- 1-+--* H 02.~ Low Temperature 4 fonarkate ROOHI OH IOr- ~OOROOH -~*OROOH 0R0 Figure 4.23 - Proposed mechanism by which hydrogen impacts the autoignition of alkane hydrocarbons. Base diagram is from Tanaka et al. [3]. 4.5.2 Effects of Carbon Monoxide Addition Before simulations were performed to investigate the effect of carbon monoxide addition, the reaction rate for "CO + 02 => CO 2 + 0" was changed from that in the original LLNL mechanism to that suggested by Scire et al. [26]. The original rate worked well for simulations with normal concentrations of CO, but William Pitz at Lawrence Livermore National Laboratories suggested the new rate for this application. A comparison of the rates is shown in Figure 4.24. Figure 4.25 shows a comparison between the temperature and OH radical concentration profiles of autoignition simulations for iso-octane with air, and for iso-octane plus 7% CO by energy with air. The simulation predicts that the autoignition time for the case with carbon monoxide addition is longer by about half as much as it is for hydrogen addition. This agrees qualitatively with the results of engine experiments that show that addition of CO in quantities corresponding to that in plasmatron reformate has about onehalf to two-thirds the effect of H2 on knock [14]. 117 8 - - Original Mechanism --- Tsang & Hampson Scire et al. 640S CDi % \ \ '1% %~ %% 2- .2 0-2-4-6 40.0006 . 0.0010 0.0018 0.0014 1/T Figure 4.24 - Comparison of the reaction rate constant for CO+02 => C02+O from three sources. The rate from Scire et al. [26] was used in this study. 3000 -OH 2500 1.E-02 -Temperature Mole Fraction - C8H18 H1 +8C O ~ 1.E-04 + CO 0 2000 - - 1.E-06 LA. - 1.E-08 E 1500 - 0 1000 - 500F0.000 - 1.E-10 1 0.003 1 1 0.006 0.009 I 1.E-12 0.012 Time (s) Figure 4.25 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane and of iso-octane with 7% CO by energy; Tiit = 875 K, Pinit = 45 bar, X= 1.5. 118 As was done with hydrogen addition, a non-reactive species, nrCO, was created to investigate the effect of thermal dilution. An autoignition simulation for which CO was replaced with nrCO is compared to the original CO addition case in Figure 4.26. The result shows that the increased thermal dilution from adding non-reacting carbon monoxide extends the autoignition time just past that for reacting CO. This implies that the main effect of CO is thermal dilution of the charge with a less reactive species, and that the chemical kinetics that involve CO may actually work to slightly reduce autoignition time. Although several reactions involving CO were found to influence autoignition time, the changes made by adjusting those reactions were small compared to the accuracy of the simulation, so no further conclusions were made. 3000 - 1.E-02 - -- Temperature OH Mole Fraction 1.E-04 2500C8H16 +CO COH 18 +nrCO - 1.E-06 '5 2000 CU- 1.E-08 - E 1600 1.E-1 0 1000- 500 0.000 0.003 0.006 0.009 1.E-12 0.012 Time (s) Figure 4.26 - Temperature and OH concentration for constant-volume autoignition simulations of iso-octane with 7% CO by energy, and of iso-octane with simulated non-reactive CO (nrCO); Tinit = 875 K, Piit = 45 bar, X = 1.5. 119 (this page intentionally left blank) 120 CHAPTER 5. ENGINE OPTIMIZATION This chapter investigates some of the performance and efficiency optimization issues relevant to modern internal combustion engines. Suggestions for maximizing low speed torque and the best utilization of improved fuel octane quality for improved efficiency are made and supported by processing the experimental data and applying a simplified analysis. 5.1 PERFORMANCE ANALYSIS This section investigates optimization of low speed torque by varying compression ratio, and the effects of operating conditions on the fuel octane number requirement. The results of the fuel octane number requirement analysis are used in the last section of this chapter to analyze the benefits of increased fuel knock resistance. 5.1.1 Compression Ratio Optimization A parameter that may be useful for the design of naturally aspirated engines is the compression ratio for maximum low speed torque. The intake manifold pressure controls the amount of air that is drawn into a spark ignition engine, and consequently the amount of fuel energy that can be released. At low speeds, where intake manifold tuning is not effective, the manifold pressure of naturally aspirated engines is limited to approximately atmospheric pressure. Since the maximum amount of air-fuel mixture that can enter the engine is limited, maximum torque is impacted by how efficiently the engine can turn the air-fuel mixture into mechanical work. Compression ratio has two effects; increasing rc increases maximum efficiency, but the spark retard required to avoid knock decreases efficiency. The result is a tradeoff for which there is an optimum compression ratio. Figure 5.1 illustrates this tradeoff for 1500 rpm. The top line is calculated from Eq. ( 3.2) and represents the increase in maximum torque at MBT spark timing with increased compression ratio. This torque benefit is only available with a very high ON fuel. The combustion retard required to avoid knock, shown with the dashed line, was increased at a rate of 3' CA per unit rc, which is consistent with the mean shown in Figure 3.53. Combustion retard at the baseline 9.8:1 compression ratio was chosen to be 10" CA, which matches that for TRF95. The bottom line is the decrease in torque from the retarded combustion timing, calculated from Eq. ( 3.1). The middle solid line represents 121 the engine output NIMEP. It is the product of the increase from increased maximum efficiency and the decrease from increased combustion retard: NIMEP (Eq.3.2). NIMEPMBT NIMEPIMEPMBT,9.8:1 NIMEP (Eq.3.1) (5.1) NIMEPMBT -25 1.1 increased NIMEPMBT from increased R- 1.05 20 CL increased combustion retard to avoid knock LLI 0 15 Ile, Q ME0.95 -- o 101% 4 product of increase from Rc aed 0.9 E and decrease from C.R. NIMEP/NIMEPMBT 5 from increased C.R. 0.85 9 10 1 11 1 12 . 0 13 14 15 Rc Figure 5.1 - Tradeoff between NIMEP and compression ratio. Values shown approximate TRF95 fuel. Under these conditions torque output peaks at a compression ratio of approximately 11.2:1. Table 5.1 shows how the compression ratio for peak NIMEP changes with fuel type (which changes the combustion retard at the baseline 9.8:1 r,), and for values of 2, 3, and 4' CA of combustion retard per unit rc, chosen to represent the spread indicated by Figure 3.53. The trend with increased fuel octane for 3" CA per unit rc is similar to the experimental values indicated by Figure 3.51. At typical fuel RON values of 90 and 95, the compression ratio for peak NIMEP is relatively insensitive to the rate of combustion 122 retard increase with r. The average compression ratio for peak NIMEP at 1500 rpm is approximately 11:1 for this engine. As engine speed increases, and the engine becomes less knock limited, the compression ratio for maximum NIMEP would approach 14:1. Table 5.1 - Calculated compression ratio for peak torque at 1500 rpm. Fuel type TRF85 TRF90 TRF95 TRF100 Corresponding combustion retard at 9.8:1 r, ('CA) 22 18 10 -5 Increase of combustion retard with compression ratio ("CA per unit r,) 2 3 4 9.8:1 10.6:1 11.6:1 10.0:1 10.8:1 11.8:1 10.6:1 11.4:1 12.2:1 12.4:1 13.0:1 13.6:1 5.1.2 Fuel Octane Quality and Reformate Addition For typical engine applications the shape of the low speed portion of the torque curve is determined by the amount of combustion retard required to avoid knock. Maximum combustion retard must be limited to avoid unsatisfactory performance, and to avoid excessive exhaust temperatures and combustion instability. Figure 5.2 shows contours of constant combustion retard on a chart of fuel RON vs. reformate fraction at 1500 rpm, 11.6:1 rc, and 40% boost. They represent the decrease of the octane number requirement of the primary fuel to maintain a given combustion retard with increased reformate fraction. Previous work with PRFs indicates that reforming 30% of the fuel results in a decrease of about 20 ON in the PRF RON required to avoid knock at MBT spark timing [14]. The PRF data in Figure 5.2 agrees with the previous work and extends it to operating conditions with retarded spark. As discussed in Section 3.2.4 , hydrogen is not effective at slowing autoignition reactions for toluene. Thus the decrease of combustion retard with reformate addition for high RON TRFs, which are mostly toluene, is not as pronounced as it is for TRFs with moderate RONs, which have a significant concentration of n-heptane. For TRF RONs between 90 and 100, adding 30% reformate reduces ON requirement by about 10 ON. The increase of fuel octane number requirement with boosted NIMEP for several values of combustion retard at 1500 rpm are shown on Figure 5.3. As indicated by the 123 experimental data in Section 3.2.3 , PRFs are more sensitive to boost than TRFs. For a real boosted application the inlet temperature would increase by 10-50 K, depending on the effectiveness of the intercooler, so the increase of fuel RON requirement with boosted NIMEP would be somewhat steeper. Experiments by Russ give the increase of octane number requirement with inlet temperature to be 1 ON per 7 K [16]. I Iur 110 N umbers are Numbers are combustion retard (CAD) combustion retard 105 (~ C I 05 z0 z 0 W100 00 U- 95 90- 95 -------- 5 20 15 10 Reformate Fraction (%) 25 30 o 5 25 20 15 10 Reformate Fraction (%) 30 Figure 5.2 - Contours of constant combustion retard for varying reformate fraction and fuel RON; X = 1.0, 1500 rpm, MAP for 40% boost. Data from Figure 3.45 (PRFs, left) and Figure 3.46 (TRFs, right). Figure 5.4 shows the increase of fuel octane requirement with compression ratio at 1500 rpm, wide-open throttle. PRFs and TRFs behave similarly. The slope for most values of combustion retard is 2-3 ON per unit compression ratio. This is somewhat less than that predicted by Russ [16]. The differences may be partially due to the experiments in this work being performed under retarded-spark conditions while the experiments in the reference were performed only at MBT spark timing. They may also be partially due to differences in engine configuration and inlet conditions. 124 10 U1 95- 100- z0 z0 951- 90 Ix a. 85 90[- Numbers are combustion retard (CAD) Numbers are combustion re tard (CAD) 9 11 12 10 Boosted NIMEP (bar) 13 14 9 08 "'V 10 11 14 13 12 Boosted NIMEP (bar) Figure 5.3 - Contours of constant combustion retard for varying boosted NIMEP and fuel RON; X = 1.0, 1500 rpm. Data from Figure 3.35 (PRFs, left) and Figure 3.36 (TRFs, right). 1uu 00 105 95 100 z 0z w 0 L- 0. 90- 95 N Numbers are combustion retard (CAD) %A[I 10 10.5 11 11.5 Rc 12 12.5 13 85- combustion retard (CAD) -0 10.5 11 11.5 Rc 12 12.5 13 Figure 5.4 - Contours of constant combustion retard for varying compression ratio and fuel RON; X = 1.0, 1500 rpm, MAP = 1.0 bar. 125 5.2 EFFICIENCY ANALYSIS As discussed further in the next section, improvement in fuel knock-resistance would allow design parameters that are typically limited by knock to be extended. The simple analysis presented in this section looks at the possible efficiency benefits of increasing compression ratio, and of inlet boosting and engine downsizing. 5.2.1 Effect of Compression Ratio on Brake Efficiency The effect of compression ratio on net efficiency is shown in the experimental results of Section 3.1 . For vehicular applications, a more important parameter is brake efficiency, which dictates fuel consumption. If the engine size is not changed, and engine friction does not change significantly, then brake efficiency will stay proportional to the net efficiency as compression ratio increases. The improvement in brake efficiency will then be the same as the improvement in net efficiency. Also, since the engine size doesn't change, the maximum torque output will increase as compression ratio increases. If, however, the size of the engine is decreased so that maximum torque output remains constant, a reduction in pumping work and friction causes brake efficiency to improve more than net efficiency. An illustration of this effect is shown in Table 5.2. For the analysis it was assumed that mechanical friction normalized by engine size, FMEP, stays constant at 0.8 bar [28]. This assumption is reasonable because friction torque increases only slightly with load compared to the total torque, and friction torque scales approximately with engine size. Although maximum NIMEP would occur at a higher speed, the data for 1500 rpm presented in Figure 3.48 was used because it was readily available and should scale similarly to higher speeds. Maximum BMEP was estimated by subtracting the assumed FMEP from the experimental maximum NIMEP: BMEP = NIMEP - FMEP (5.2) As compression ratio increases BMEPax increases and the engine displacement is decreased to keep maximum torque constant: Torque(Nm) = 8 - BMEP(bar) -Displacement(L) 126 (5.3) Displacementdownsized BMEPmax,baseline Displacementbaseline BMEPmax,downsized (54) Since the engine displacement is reduced, mid-load BMEP must increase to keep midload torque constant: BMEPmidoaddownsized BMEPmid-load baseline _ Displacementbaseline Displacementdownsized BMEPmax,downsized BMEPmax,baseline (55) A mid-load BMEP of 2.6 bar at 1500 rpm was selected for the baseline 9.8:1 compression ratio because it is representative of the mid-load cruising condition and is influential on the overall vehicle efficiency [27]. Mid-load NIMEP is calculated by adding FMEP to the downsized BMEP. Net efficiency is interpolated from Figure 3.6. Brake efficiency is then estimated by multiplying the net efficiency by the ratio of BMEP to NIMEP. (5.6) = ne - Bbrake 77brak NIME Figure 5.5 compares the improvement in brake efficiency with downsizing, calculated as described above, to the improvement in brake efficiency without downsizing, calculated by assuming constant mechanical efficiency. Downsizing the engine to maintain a constant maximum torque output with increased rc increases the efficiency benefit by about 60% compared to the case without downsizing. Table 5.2 - Effect of downsizing with increased r, on brake efficiency. R 9.83 11.59 13.40 Max Max NIMEP Max Max BMEP Midload BMEP Midload NIMEP (bar) (bar) (bar) (bar) 10.16 10.74 11.01 9.36 9.94 10.21 2.60 2.76 2.84 3.40 3.56 3.64 Net efficiency Brake efficiency Increase in brake efficiency (%) 0.289 0.302 0.309 0.221 0.234 0.241 0 6.1 9.3 127 12 ,a- Without Downsizing I -~10 VVILII LJUW"ZI1,LiII 8 W 1 _ 4- 02 9 10 12 11 13 14 Rc Figure 5.5 - Estimated increase of mid-load brake efficiency with r,, with and without downsizing the engine to maintain constant maximum torque output; 1500 rpm, X = 1.0, 2.6 bar baseline BMEP. 5.2.2 Effect of Boost on Brake Efficiency Turbocharging an engine increases its maximum BMEP. If engine size is held constant, this can be used to increase the torque output of the engine without significantly affecting efficiency (if compression ratio is not changed). However, if the engine is downsized to keep torque output constant, as described for increased compression ratio in the previous section, decreased friction, pumping work, and heat transfer lead to an increase in brake efficiency. The effect of boosting and downsizing is illustrated in Table 5.3. As in the previous case, FMEP was assumed to be constant at 0.8 bar, baseline mid-load BMEP is assumed to be 2.6 bar, and mid-load BMEP is scaled with maximum BMEP. A baseline maximum BMEP of 10.4 bar was chosen to match the performance of modem engines [4]. The increase in net efficiency is interpolated from Figure 3.6. As in the previous section, brake efficiency is estimated by multiplying the net efficiency by the ratio of BMEP to NIMEP. The improvements in net and brake efficiencies are plotted in Figure 5.6. The 128 improvement in net efficiency is due to decreased pumping work and heat transfer, the additional improvement in brake efficiency is due to reduced friction. Table 5.3 - Effect of boosting and downsizing on mid-load brake efficiency. NIMEP boost bevsl level (%) 0 20 40 Midload BMEP Max BMEP baE Max NIMEP NIaEP (bar) (bar) (bar) Net eff. Brake eff. (bar) Increase in brake efficiency (%) 0.289 0.302 0.312 3.40 3.96 4.52 2.60 3.16 3.72 10.40 12.64 14.88 11.20 13.44 15.68 Midload NIMEP 0.221 0.241 0.256 0 9.0 16.2 20 Efficiency -Net --- Brake Efficiency 16 C 12 0 8 C) 4 0 0 10 20 30 40 50 NIMEP Boost Level (%) Figure 5.6 - Estimated increase of mid-load net and brake efficiencies with boosting and downsizing; 1500 rpm, X = 1.0, 2.6 bar baseline BMEP. 129 5.3 IMPLICATIONS FOR ENGINE DESIGN Modem engines are designed to maximize performance and fuel efficiency within the constraints of available fuel quality. The experimental results presented in this work indicate that an improvement of 10 ON or more can be realized by converting a portion of the fuel entering an engine to a hydrogen rich reformate. Consequently, it is an interesting exercise to analyze what the efficiency benefits would be for a 10 ON improvement in fuel knock resistance, whether it be through on-board reforming, or through improved fuel refining. Assuming that gasoline performs in between PRFs and TRFs, Figure 5.3 implies that boosting NIMEP by 40% while maintaining constant combustion retard would require an improvement of fuel quality of about 6 ON. If inlet temperature increases by about 30 K, it would bring the increased octane requirement to about 10 ON [16]. Alternatively, Figure 5.4 implies that increasing fuel quality by 10 ON would allow compression ratio to be increased from 9.8:1 to 13.4:1 without increasing the amount of combustion retard required to avoid knock. The estimations made in this chapter indicate that with the improved fuel quality, increasing compression ratio could potentially improve brake fuel conversion efficiency by about 9%. Adding a turbocharger, which would have a somewhat higher cost, could improve brake fuel conversion efficiency by up to 16%. To maximize efficiency benefit given an improvement in fuel octane quality over today's levels, this analysis implies that inlet boosting and engine downsizing is a better strategy than increasing compression ratio. Since the calculations presented required many assumptions and simplifications, a more accurate engine performance analysis may be necessary to confirm the magnitude of the efficiency benefits. 130 CHAPTER 6. CONCLUSIONS * Mid-load net efficiency improvement from increased air-fuel ratio peaks at about 10% relative to stoichiometric values. About 2/3 of the improvement is from the increased ratio of specific heats for lean mixtures and reduced heat transfer; the rest is from reduced pumping losses. The air-fuel ratio for peak efficiency increases with increasing compression ratio. Relative net efficiency improvement from increasing load is about 7% per bar NIMEP at mid-load. About 75% of the improvement is from reduced pumping losses and 25% is from heat loss becoming a smaller portion of the overall charge energy. At a compression ratio of 9.8:1, relative net efficiency improvement is about 2.5% per unit compression ratio. Efficiency peaks at about 14:1 with a maximum benefit of 6-7%. Efficiency improves more with compression ratio at high speeds and loads due to the reduced importance of heat loss. * A new combustion phasing parameter, termed "combustion retard", has been developed. It is the location of 50% mass fraction burned minus the location of 50% mass fraction burned for MBT spark timing. It represents the crank angle that the center of the combustion event has been shifted from the timing for maximum torque. Loss of indicated torque correlates well to a diagonally asymptotic relationship with combustion retard for the wide range of operating conditions considered. * The knock behavior of three fuel types was investigated. Primary reference fuels are mixtures of iso-octane and n-heptane or tetraethyl lead that are used in the ASTM octane rating procedures. Toluene reference fuels are mixtures of toluene and nheptane that are used to better represent the aromatic content and sensitivity to operating conditions of gasoline. The octane number rating of both of these fuel types can be adjusted by adjusting the mixture composition. Two formulations of unleaded test gasolines were used to represent regular automotive fuel. The knock behavior of PRFs and TRFs is similar for changes in compression ratio, but PRFs are more sensitive to boosted inlet pressure, air-fuel ratio, and fuel reformate than TRFs. Test gasolines behave about halfway between PRFs and TRFs. * The combustion retard required to just avoid knock for PRFs increases with air-fuel ratio, with the increase becoming more severe at higher compression ratios. For TRFs, required combustion retard decreases slightly with air-fuel ratio for all 131 compression ratios. PRFs, which require about 5' CA of combustion retard per bar NIMEP, need about three times as much combustion retard as TRFs when boosted to achieve the same NIMEP. Both fuel types require an average of about 30 CA of combustion retard per unit of increased compression ratio. The standard deviation of the data is about 50% of the total effects. The wide spread is mainly due to the nonlinear nature of the knock phenomenon, but is also affected by the accuracy of audible knock detection and changing atmospheric conditions. " The combustion retard required to just avoid knock decreases by about 20 CA per 3% reformed fraction for PRFs. For TRFs with low alkane content reformate addition is less effective. Constant volume iso-octane autoignition modeling shows that hydrogen converts hydroxy radicals to hydrogen radicals, which are not as effective at initiating the chain branching reaction sequence. In the toluene oxidation mechanism investigated, most of the initiation and propagation reactions involving the hydroxy radical have an analogous reaction with the hydrogen radical with similar rate constants, so the overall reaction rate is not affected. * Detailed chemical kinetics mechanisms were combined with a cylinder pressure based end-gas modeling methodology. It was found that high accuracy in the determination of initial conditions is required for consistent results. Calibration of baseline initial temperature is required to match experimental data. The model successfully predicted the response of PRFs to compression ratio and air-fuel ratio and the response of TRFs to boost. The difference between the response of PRFs and TRFs to air-fuel ratio was also captured. Simulations with reaction kinetics disabled show that due to a higher ratio of specific heats, TRFs are more sensitive to the pressure ratio through which the end-gas is compressed than PRFs. This effect helps to explain the different responses of PRFs and TRFs to boosted lean operation. * Reforming 30% of the fuel entering an engine decreases the required fuel octane number by 10 ON or more depending on fuel composition. This improvement would allow the compression ratio of an engine to be raised from 9.8:1 to approximately 13.4:1 without increasing combustion retard, or it would allow the engine to be boosted to a 40% higher NIMEP. A simplified analysis indicates that increasing compression ratio and downsizing the engine to maintain constant maximum torque would increase mid-load fuel efficiency by about 9%. Boosting and downsizing would increase mid-load fuel efficiency by about 16%. 132 REFERENCES [1] Heywood, J.B., Internal Combustion Engine Fundamentals,McGraw Hill Inc., New York, 1988 [2] Pan, J., et al., "End-Gas Inhomogeneity, Autoignition and Knock," SAE 982616 [3] Tanaka et al., "Two-stage ignition in HCCI combustion and HCCI control by fuels and additives," Combustion and Flame, Volume 143, pp219-239, 2003. [4] Chon, D.M., and Heywood, J.B., "Performance Scaling of Spark Ignition Engines: Correlation and Historical Analysis of Production Engine Data," SAE 2000-01-0565 [5] Taylor, C., The Internal Combustion Engine in Theory and Practice, M.I.T. Press, Cambridge, MA 1985 [6] Goldwitz, J.A., and Heywood, J.B., "Combustion Optimization in a HydrogenEnhanced Lean-Bum SI Engine," SAE 2005-01-0251 [7] Hisato Hirooka, Sachio Mori and Rio Shimizu, "Effects of High Turbulence Flow on Knock Characteristics," SAE 2004-01-0977 [8] Topinka, J., Knock Behavior of a Lean-Burn, H2 and CO Enhanced, SI Gasoline Engine Concept ,M.S. Thesis, MIT, May 2002 [9] Glassman, I., Combustion, Academic Press Inc., California, 1996 [10] Ivanic, Z. et al., "Effects of Hydrogen Enhancement on Efficiency and NOx Emissions of Lean and EGR Diluted Mixtures in a SI Engine," SAE 2005-010253 [11] Natkin, R., Ford Motor Company, Private Communication, June 2003. [12] Green, R., and Pearce, S., "Alternative Transport Fuel," Energy World Journal, pp. 8-11. October 1994. [13] Tang, X., Heffel, J., et al., "Ford P2000 Hydrogen Engine Dynamometer Development", SAE 2002-01-0242 133 [14] Topinka, J.A. et al., "Knock Behavior of a Lean-Burn, H2 and CO Enhanced, SI Gasoline Engine Concept," SAE 2004-01-0975 [15] Gruden, D., and Hahn, R., "Performance, Exhaust Emissions and Fuel Consumption of a IC Engine with Lean Mixtures," I Mech E Publication (C111/79), 1979. [16] Russ, S., "A Review of the Effect of Engine Operating Conditions on Borderline Knock," SAE 960497 [17] Muranaka, S., Takagi, Y., and Ishida, T., "Factors Limiting the Improvement in Thermal Efficiency of S.I. Engine at Higher Compression Ratio," SAE 870548 [18] Kalghatgi, G., Shell Global Solutions, Private Communication, April 2004. [19] Burluka, A., et al., "The Influence of Simulated Residual and NO Concentrations on Knock Onset for PRFs and Gasolines," SAE 2004-01-2998 [20] Kalghatgi, G., "Auto-ignition quality of practical fuels and implications for fuel requirements of future SI and HCCI engines," SAE 2005-01-0239 [21] Emdee, J., Brezinsky, K., and Glassman, I., "A Kinetic Model for the Oxidation of Toluene Near 1200K," J. Phys. Chem. 1992, 96, 2151-2161 [22] Pitz, W., et al., "Chemical Kinetic Study of Toluene Oxidation Under Premixed and Nonpremixed Conditions," LLNL report UCRL-CONF-201575, 2003 [23] Curran, H. J., et al., "A Comprehensive Modeling Study of iso-Octane Oxidation," Combustion and Flame, Volume 129, pp253-280, 2002. [24] Curran, H. J., et al., "A Comprehensive Modeling Study of nHeptane Oxidation," Combustion and Flame, Volume 114, pp149-177, 1998. [25] Tomohiro Shinigawa et al., "Effects of Hydrogen Addition to SI Engine on Knock Behavior," SAE 2004-01-1851 [26] Scire, J. J., et al., "Flow Reactor Studies of Methyl Radical Oxidation Reactions in Methane-Perturbed Moist Carbon Monoxide Oxidation at High Pressure With Model Sensitivity Analysis," International Journal of Chemical Kinetics, Volume 33, pp75-100, 2001 134 [27] Leone, T., Ford Motor Company, Private Communication, April 2005. [28] Sandoval, D., and Heywood, J.B., "An Improved Friction Model for SparkIgnition Engines," SAE 2003-01-0725 135 (this page intentionally left blank) 136 APPENDIX A MODIFIED PISTON DIMENSIONS 137 ref 6.64 71.23 ref 1.36 36 0 33100 137 ref 16,08 56.12 ------ - - --5 - - -- -- -- -- - 35.9 30 87~ -- ---- 12:1 CR Piston Mike Gerty Oct 22, 20 3 138 ref 6.64 71,23 ref 1.36 36 0 33.90 13.7 ref 16 08 56 12 -- 077.51 ------ ---- ----- 35.9 5,4 87 14:1 RC Piston Mike Gerty Oct 22, 2003 139 (this page intentionally left blank) 140 APPENDIX B ADDITIONAL CHARTS FOR EFFECTS OF AIR-FUEL RATIO ON KNOCK 141 35 C- Torque Loss 30 1 20% 25 - -g-PRF100 -*- PRF95 UPRF90 -6-PRF85 20 d 10% -+-PRF80 C 15 0 0 .0 - 10 - 3% 5 1% 0 -5 1.1 0.9 1.3 1.5 1.7 Lambda Figure B. 1 - Effect of X on combustion retard to just avoid knock for PRFs; r. = 9.8:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. . 35 a I- Torque Loss -*- TRF100 30- 20% -4-TRF95 25- -U-TRF90 0 m U C so 0 -- 20- 10% C so -+-TRF80 0 I- TRF85 15- (U 4-. (U 10 - 3% 0 (I) 5 1% .0 E 0 C-) 0A -5 0.9 'PO 1.1 1.3 1.5 1.7 Lambda Figure B. 2 - Effect of X on combustion retard to just avoid knock for TRFs; r, = 9.8:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 142 35 Torque Loss 0 30 - 20% C.) 25 - -#- PRFIIO -X-PRFIO5 *PRF100 20 - 10% -+- PRF95 - PRF90 15- 10- 3% 0 X. 5 1% E 0 - -- 0 - -- - -- -- - --- -- -- - -5 1.1 0.9 1.7 1.5 1.3 Lambda Figure B. 3 - Effect of A on combustion retard to just avoid knock for PRFs; re = 11.6:1, 1500 rpm, MAP at X> 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 35 Torque Loss -*- TRF100 30 - 20% -+-TRF95 25 - -6-TRF90 -4-TRF85 20- 10% 15 0 U 10- 3% 5 1% 0 ------------------------5 -t 0.9 1.1 1.3 1.5 1.7 Lambda Figure B. 4 - Effect of X on combustion retard to just avoid knock for TRFs; re = 11.6:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 143 35 .oqueLs Torq ue Loss I 30 --- PRF115 20% -*- PRF1 10 25 - -X- PRF105 -PRF00 20 1 10% -+PRF95 15 - 10 - 3% - - - - - - - - -- - - - - - - - 0 1% 5 0 -- ' -5 1.1 0.9 1.3 1.5 1.7 Lambda Figure B. 5 - Effect of X on combustion retard to just avoid knock for PRFs; r, = 13.4:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 35 Torque Los -0--TRF90 30 - 20% 0 -4-TRF95 25 - -i- 20 - 10% -X-TRF105 TRF100 15 10- 3% 5 1% 0 0 -- --- -5 0.9 1.1 1.3 Lambda 1.5 1.7 Figure B. 6 - Effect of X on combustion retard to just avoid knock for TRFs; r, = 13.4:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 144 35 T Torque Loss - 30 - 20% - +- PRF95 -TRF95 -- +-UTG96 25 E C 20 - 10% --.. , C 15 0 10 - 3% . 0 E .0 0 5 1% 0 -5 0.9 1.1 1.3 1.5 1.7 Lambda Figure B. 7 - Effect of X on combustion retard to just avoid knock for UTG96 fuel; r,= 13.4:1, 1500 rpm, MAP at X > 1.0 boosted to match MBT NIMEP at stoichiometric unboosted WOT. 145 (this page intentionally left blank) 146 APPENDIX C ADDITIONAL CHARTS FOR EFFECTS OF BOOST ON KNOCK 147 35 I- Torque Loss -)*-PRF105 30 - 20% -4-PRF100 0 25 - 20 d 10% 0 -- zl"MOoe PRF95 -0-PRF90 101 3% 0 5E 0 1% ---- 0- --- ----------- -5 8 9 10 11 12 13 14 15 NIMEP (bar) Figure C. 1 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; r,= 9.8:1, X = 1.0, 1500 rpm. , 35 I- Torque Loss -*-TRF100 30- 20% -+-TRF95 25 - -U-TRF90 -0-TRF85 20- 10% 0 15- 10 - 3% 0 1% 5- E 0 0- ---- -- - --- - --- --- -- -5 8 9 10 11 12 13 14 15 NIMEP (bar) Figure C. 2 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; re = 9.8:1, X = 1.0, 1500 rpm. 148 35 I Torque Loss -- PRF100 30 - 20% -- PRFI5 25 - -N- PRF110 -PRF115 20 - 10% / 0 15 E ~0 0 10 - -- PRF120 3% 5- 1% 0- -5 i 9 1 1 10 11 1 1 16 15 14 13 12 NIMEP (bar) Figure C. 3 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; r,= 11.6:1, X = 1.0, 1500 rpm. 35 I- - Torque Loss -*-Toluene -+- 30- 20% TRF105 -*-TRF100 25- -4-TRF95 0 20 - 10% 0 15- E 0 10 - 3% 0 1% 5 0 -5 9 10 11 12 13 14 15 16 NIMEP (bar) Figure C. 4 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; r,= 11.6:1, X = 1.0, 1500 rpm. 149 ............... 35. .. in25 Rc=13.4:1 Rc= 11.61 -... Rc=9.8:1 -. --.... 0 'a C- 20..... 1 5 .... - .... .. .. ........ 1 0 .. E 0 5.. . 0- 90 -.... 100- 10 110 RON 9 1 0 12 13 14 Boosted NIMEP (bar) Figure C. 5 - Effects of boosted NIMEP and PRF fuel RON on combustion retard to just avoid knock; X = 1.3, 1500 rpm. .............-..... 35 ................ 3 0 ............. ..... ........ 2.. -. Rc= 1 3 .4 :1 Rc=11.6:1 Rc=9.8:1 0 0) -o 20 . ..... --... 0 L) E .. -5 ....-.. -.. ..... 0 ... 105 12014 120 9 10Boosted NIMEP (bar) Figure C. 6 - Effects of boosted NIMEP and TRF fuel RON on combustion retard to just avoid knock; X = 1.3, 1500 rpm. 150 35 I- I Torque Loss -X-PRF105 -h-PRF100 30 -120% +- PRF95 25 - // -U- 20 -i 10% PRF90 -S- PRF85 a 15 - // 10 1I3% U- 5 E 0 1% 0 -------------------5 9 8 11 10 NIMEP (bar) 12 13 Figure C. 7 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; r,= 9.8:1, k = 1.3, 1500 rpm. 35 I- 30- Torque Loss -e- TRF100 20% -+-TRF95 -U-TRF90 0 *E 25- -+-TRF85 C 2015 - 10- 3% .0 I- 0 5 1% 0 E ---- 0 -5 - ----- 9 10 11 12 13 14 NIMEP (bar) Figure C. 8 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; r,= 9.8:1, X = 1.3, 1500 rpm. 151 . Torque Loss 35 C- - PRFI15 30 - 20% -*- PRF110 25 - -<-PRF105 -*-PRF100 20- 1U0/ -4-PRF95 0 15 - E 10- 3% 1% 5 0 .0 0 -------- --------------- -5 9 10 11 12 13 14 NIMEP (bar) Figure C. 9 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for PRFs; re = 11.6:1, X = 1.3, 1500 rpm. 35 30 - 25 - Torque Loss -M-TRF90 20% -+-TRF95 -*-TRF100 0 -)-TRF105 20 - 10% 0 E 0 15 - 10 - 3% 5 1% A- U - - - - - - -- 0 -- - - - - - -5 9 10 11 12 13 14 NIMEP (bar) Figure C. 10 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; r,= 11.6:1, X = 1.3, 1500 rpm. 152 . It Torque Loss -PRFI15 30 - 20% I0 25 -*-PRFIIO 7/M - -N-PRF105 C: V -*-PRF100 20 - 10% 7 15 - 10 - 3% 5 1% 0 0 -5 ' 11 10 9 14 13 12 NIMEP (bar) Figure C. 11 - Increase of combustion retard to just avoid kn )ck with increased NIMEP from boosting for PRFs; r,= 13.4:1, X = 1.3, 1500 rpi 1. 35 L- 0 Torque Loss -- TRF95 30- 20% -6-TRF100 25- -)-TRF105 20- 10% I- 0 10d 3% 0 (U 5 1% An -I-J------------------------5 9 10 11 12 13 14 NIMEP (bar) Figure C. 12 - Increase of combustion retard to just avoid knock with increased NIMEP from boosting for TRFs; re = 13.4:1, X = 1.3, 1500 rpm. 153 (this page intentionally left blank) 154 APPENDIX D ADDITIONAL CHARTS FOR EFFECT OF FUEL REFORMATE ON KNOCK 155 35 16 -+-PRF100 C.R. 30 15 C- 25 -U--PRF90 - C.R. 14 w 201 0 PRF95 C.R. - - 4- PRFIOO - -- - a- PRF90 NIMEP -, 130.i 1 IL .D 10 12 - 0 -- , ' . . - - - -- - -. , PRF95 NIMEP NIMEP - 5- E 0 0-5 10 0 5 10 15 20 30 25 35 Reformate Fraction (%) Figure D. 1 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for PRFs; r, = 9.8:1, X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 35 30 C0 w 16 - TRF100 --- C.R. - -- 15 -+-TRF95 C.R. 25 -U-TRF90 -- 14 20 C.R. -- C 0 TRF85 C.R. 15 - 10 - ---------- -- - -- - - -12 - 4- TRF100 NIMEP z - TRF95 -- NIMEP 50 E 0 - - 11 0- 110 -5 - 0 5 10 15 20 25 30 10 - - TRF90 NIMEP - .0- TRF85 NIMEP 35 Reformate Fraction (%) Figure D. 2 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for TRFs; r, = 9.8:1, X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 156 . 16 35 I- 30 - -~--PRF110 C.R -+4-PRF105 - 15 0 - - S..- - 10 25 -* 0 C C 0 - 10 - PRF100 - 20U- C.R. PRF95 C.R PRF115 + - 13 z - -- NIMEP - 12 0 5 E 0 C.R -14 15 PRF115 C.R --- -- - - - - - - - - - 0 r:: - *- -PRFIIO - -x- NIMEP PRFIO5 NIMEP -PRFIOO - -- -5 i 0 I5 10 15 25 20 30 NIMEP 1IV PRF95 35 NIMEP Reformate Fraction (%) Figure D. 3 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for PRFs; r, = 13.4:1, X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 16 35 ---- Toluene C.R. 30 .. - - - - - . x- - - - - -- +-TRF105 15 C.R. 250- - -+--TRF100 C.R. 14 20 U --- 13 Iw 15 E 0 TRF95 C.R. - -- z 10 12 0 oluene NIMEP - X- TRF105 NIMEP 5 11 - TRFIOO - NIMEP 0 --- ' -5 0 5 10 15 20 25 30 10 TRF95 NIMEP 35 Reformate Fraction (%) Figure D. 4 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for TRFs; r, = 13.4:1, X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT)157 - .. ................... -- .. 4 Rc=13.4:1 R c= 1 1 .6 :1 Rc=9.8:1 0 ...--.... R30 eo 3 3525- ..... 20 15 0 -..-....-... 10 20 Reformate Fraction (%) ..... 30 ..... .....100 90 RON Figure D. 5 - Effects of reformed fuel fraction and PRF fuel RON on combustion retard to just avoid knock; X=1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 0 8 Rc=13.4:1 ......Rc=1 1.6:1 - Rc=9.8:1 ( 35 -......... 30 .-... 25 0 ..... 101E 0~2 R.N ...... Reformate Fr....n.(%).. Figue ffecsD.6 ofrefomed uelfracion nd TF retard t........ ~0 s... i k..k ,.....,.A.s.to .. nb..sted ............. i st h5mt 5 8....8 .... ful.R.......b.ti. m tc.....IM *v .=. P BT a 35 30 - -- *--PRF95 C.R. -- U-PRF90 11 C.R. 25: -*-PRF85 20 1 C.R. ---- PRF80 A .. 10 0 15 - 10 - w a. ~- a - - - - - -- -- - - -o- C.R. PRF75 C.R. PRF95 NIMEP - 9 5 E 0 --- -PRF90 - NIMEP - -- - -PRF80 0 -5 NIMEP I 0 PRF85 I 5 i I ' I 25 20 15 10 Reformate Fraction (%) 30 8 NIMEP 35 - -x- PRF75 NIMEP Figure D. 7 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for PRFs; r, = 9.8:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 35 -- +-TRF95 C.R. 30 - 25 - 0 20 - a 15 - I- - 11 -U- TRF90 -e- IL U~ 10 w - - - -- - ----- - - -- -+-TRF80 C.R. TRF75 --C.R. - -o- TRF95 - -- - - NIMEP -9 - -- 5- E 0 C.R. TRF85 C.R. TRF90 NIMEP - <- TRF85 0- NIMEP - +-'- TRF80 I-8 -5 0 5 10 15 20 25 Reformate Fraction (%) 30 35 NIMEP - -- -TRF75 NIMEP Figure D. 8 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for TRFs; r, = 9.8:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 159 35 PRF100 C.R. PRF95 C.R. PRF90 C.R. -+I- 30 - 11 -- 0 .- 0 25 1 0 - 20 - -10 0~ 1 is - -0 0 -+-PRF100 NIMEP -PRF95 - NIMEP --p 0 9 '0' 5 11 0 C.R. PRF80 C.R. - 10- (0 E PRF85 E -- -- ---------- -- -PRF90 - NIMEP - + 8 -5 0 5 10 15 20 25 NIMEP 35 30 PRF85 - PRF80 -- NIMEP Reformate Fraction (%) Figure D. 9 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for PRFs; r, = 11.6:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 35 -+-TRF100 C.R. 30 ~ 25 - . -. . . - - -.-. . . 11 -- 'TRF95 C.R. -4-TRF90 3 0 0 15,a 10 w -+--TRF80 C.R. z - - - -9 5 - +~ TRF95 - a- -TRF90 NIMEP 0 - NIMEP - .0- -5 TRF100 NIMEP - --- - - -- - - E 0 C.R. TRF85 C.R. --- 20 ,8 10 5 10 15 20 25 Reformate Fraction (%) 30 35 TRF85 NIMEP - +*- TRF80 NIMEP Figure D. 10 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for TRFs; r, = 11.6:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 160 - 35 -- +- PRF105 C.R. I- 30 - 0 25 - 11 U -- PRFI0 C.R. - PRF95 00 C.R. 0 0,~ 20 - PRF90 C.R. S OPRF85 - 10 a. w C.R. - 10 - -x- 'PRF105 NIMEP 0 9 5- PRF100 - -A- NIMEP -PRF95 - -- E 0 - - - - - - - - - - - - 0 -- - - - - - - - - NIMEP - -PRF90 NIMEP 8 -5 0 10 5 35 30 25 20 15 -PRF85 - NIMEP Reformate Fraction (%) Figure D. 11 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for PRFs; r, = 13.4:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 35 I- 30 --. .- C -- ) -TRF105 11 . - . --. + - -- -- 25 - ... . ... . -. -. - - - A - 0 20 10 E 15 a. w - 0 .- - - - - - - -- - . ,- z 10 C.R. -4-TRF95 C.R. -U-TRF90 C.R. -+-TRF85 C.R. - -x- TRF105 NIMEP 0 @1 TRF100 C.R. 9 - - 5- TRFIOO NIMEP - -- - a- TRF95 NIMEP 0-5 0 5 10 15 20 25 Reformate Fraction (%) 30 35 TRF90 NIMEP 8 -0 . TRF85 NIMEP Figure D. 12 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for TRFs; r, = 13.4:1, X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 161 35 +- UTG96 0 30 -1C.R. - -- UTG91 - 2C.R. 25- - ,- -10 -.- UTG96 NIMEP UTG91 o -NIMEP 15 ~- ~X20 - 2W , w10 - -- 9 5 .0 E 0 0 --- - --- -- - -- 5 10 15 -- -- -- -- -- - -5 8 0 20 25 30 35 Reformate Fraction (%) Figure D. 13 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for UTGs; re = 11.6:1, x = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 35 Q --- UTG96 C.R. -1 - UTG96 30 - o -NIMEP 25 20 - -- 10 00 15 - 00 U) 5z 0--9 05- E 0 --------------------- 0---------- -5 -8 0 5 10 15 20 25 30 35 Reformate Fraction (%) Figure D. 14 - Decrease of combustion retard to just avoid knock and associated increase in NIMEP for increased reformed fuel fraction for UTG96; r, = 13.4:1, x = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 162 APPENDIX E ADDITIONAL CHARTS FOR EFFECT OF COMPRESSION RATIO ON KNOCK 163 - 35 I 30 11 ---- PRF105 C.R. - -- *-PRF100 C.R. - 10 5. 25- -2015 -9 10 - 0 E 0 i -- e--PRF90 C.R. 0. - - -- - -- - - - -- - -- - -x- PRF105 - -- NIMEP -8 2 - PRF95 C.R. -PRFIOO NIMEP 5-7 - -- NIMEP 0- -PRF90 NIMEP -6 -5 10 9 11 12 13 PRF95 14 Rc Figure E. 1 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. - 35 11 C.R. 30 10 a 25 - 20 - TRF90 C.R. -,+-TRF85 C.R. 1510 -+-TRF95 C.R. --- I0 M -- *-TRF100 ------ - -- - - ------- - - 4- TRF100 NIMEP 8 E - - 0 4- TRF95 NIMEP 50 7 - a- TRF90 NIMEP - .- 06 -5 9 10 11 12 13 TRF85 NIMEP 14 Rc Figure E. 2 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 164 ' . 35 )K. 30 - -*-PRF110 C.R. -- - , k PRF105 C.R. . . .. . . . . 25 - 0 11 - PRFI C.R. - 20 0 - -- *-PRF95 C.R. 0 E- 15 - (0 E 0 PRF110 - -X- NIMEP - - -PRF105 5- NIMEP 7 E -- -PRF100 - 0- C.) z 8 10 00 -5 - - - - - - - - - - - - - - - -- -- - I 9 10 I I - I - - -PRF95 6 I NIMEP 14 13 12 11 NIMEP Rc Figure E. 3 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.6, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. ' 11 35 -+-TRF100 C.R. 30 - C 25 - - -. . - - -. - 10 - -'&--TRF90 C.R. M at 20 -- +-TRF95 C.R. 94 - 0W 8 00 15- -+-TRF85 C.R. - 4- TRF100 NIMEP 0 10 - - - TRF95 NIMEP 0 E 0 5- 7 0-5 4 9 6 10 12 11 13 - -- - .o- TRF85 TRF90 NIMEP NIMEP 14 Rc Figure E. 4 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; X = 1.6, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT. 165 - 35 30 14 PRF115 C.R. PRF110 C.R. - - ---13 25 - CP 20 -PRFI -- I - 12 ~ -- -- 0 C.R. CL 15 - - I- 10 4' 11 - PRF95 C.R. PRF115 NIMEP E 0 05 C.R. PRFIO9 - 5- -- PRF110 NIMEP 10 0- - -X- PRF105 - -- -PRF100 NIMEP ' -5 9 9 10 11 12 13 NIMEP 14 - -o- PRF95 NIMEP Rc Figure E. 5 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for PRFs; X = 1.0, 1500 rpm, MAP for 20% boost (NIMEPMBT 1.2 times unboosted NIMEPMBT). 35 30 - 14 -- TRF105 C.R. - 25 - 20 - 13 I.----. --- TRF100 C.R. --- TRF95 C.R. . . . . . . .. - 12'g 0 --U-TRF90 .0 C.R. 15- - -TRF105 A- -TRF100 - E 0 NIMEP -112 10- - 0 NIMEP 5 - 0 10 4 -- - - - - - - - - - - -- - - - - - - - - -9 -5 9 10 11 12 13 - -- -TRF95 NIMEP - -- -TRF90 NIMEP 14 Rc Figure E. 6 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; X = 1.0, 1500 rpm, MAP for 20% boost (NIMEPMBT 1.2 times unboosted NIMEPMBT). 166 is 35 - - I- 30 - - - - .. .. .. . + . --- - - - 14 ~- 25 - -- *-PRF110 CP 20 - -13 15 - 0 10 - - - - - - - - - - - - - PRF1 20 C.R. PRF115 C.R. - - - - -~ -12 - j -- C0 0. -- z - C.R. -PRF105 C.R. PRFI 00 C.R. - - PRF120 NIMEP PRFII5 - -- NIMEP PRF115 - -- - -- - -- 50 -11 -PRFI10 0- NIMEP -5 i 9 10 12 11 1 13 410 PRF105 NIMEP 14 --- PRFIOO NIMEP Rc Figure E. 7 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 16 35 -M--TRF105 C.R. I- 30 - 15 IrOO000000000000-000a O 25 - 20 - 14 'C' CD 10 C.R. -- +-TRF95 C.R. - -000-0- 15 -- 4-TRF100 . - -- w-TR90 C.R. - -TRF105 )- NIMEP 13 z - -TRFIOO - -- 0 c NIMEP 5- E '0 12 0-5 - -~- NIMEP - ' 9 1 10 I1 11 1 13 12 13 11 TRF95 - TRF90 NIMEP 14 Rc Figure E. 8 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for TRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT). 167 35 30 13 - - --12 0a 25 - ~---PRF1 05 20 - C.R. -- h- PRFI00 C.R. -- +-PRF95 C.R. - -- PRF115 11 CL I .0 E 0 C. PRF115 C.R. PRF110 C.R. 15 - 10 - -- - -- - -- - -- - -- - -- - 102 NIMEP 5 - NIMEP 9 0 ---5 i 9 'PRF110 - -- - -x- PRF105 - NIMEP I I 1 1 . 10 11 12 13 14 - -PRFlOO - - 8 NIMEP PRF95 NIMEP Rc Figure E. 9 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.3, 1500 rpm, MAP for 15% boost (NIMEPMBT 1.15 times NIMEPMBT at stoichiometric unboosted WOT). 35 30 T3 -13 - - - -.- - - 25 - 20 - 15 - - 1- E 0 . - - - - - - - - - - - - - 12 -- -- Ar--TRF100 C.R. -- +-TRF95 C.R. -112 -.- TRF90 C.R. - 10 TRF105 C.R. - - NIMEP - )I(- 0 -TRF105 -- 10 i A- NIMEP 5- 9 - *- - - 8 -5 10 12 11 13 TRF95 NIMEP 01+-- 9 -TRF1OO -TRF90 NIMEP 14 Rc Figure E. 10 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for TRFs; X = 1.3, 1500 rpm, MAP for 15% boost (NIMEPMBT 1.15 times NIMEPMBT at stoichiometric unboosted WOT). 168 35 15 - ---- 30 14 25 - 20 - PRF1 20 C.R. PRF15 C.R. -- *-PRF110 C.R. -- X- PRF1 05 13j C.R. 0 15 - . . . . - - : - - + E 0 10 12 - --- PRFIOO - -+-- C.R. PRF120 NIMEP '--x- - -- 5 - NIMEP 11 C) PRF115 - <- -PRF110 - -x- PRF105 - e- 0- NIMEP i 10 -5 i 9 10 12 11 13 NIMEP 14 -PRFIOO NIMEP Rc Figure E. 11 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.3, 1500 rpm, MAP for 30% boost (NIMEPMBT 1.3 times NIMEPMBT at stoichiometric unboosted WOT). . 35 30 - 15 - - -- '-TRF100 - 14 C.R. 25 I- 20 ---+-TRF95 C.R. - - 13 ' 0 U .. 15 - 10 - -U-TRF90 C.R. - E C I-TRF105 C.R. <- -TRF105 NIMEP -12 2 - -- -TRF1OO NIMEP 5- 11 0-5 ' 10 i 9 10 12 11 13 - -- - - TRF95 NIMEP -TRF90 NIMEP 14 Rc Figure E. 12 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; ) = 1.3, 1500 rpm, MAP for 30% boost (NIMEPMBT 1.3 times NIMEPMBT at stoichiometric unboosted WOT). 169 35 0 30 41 2 16 --- PRF110 C.R. -- - - 25 - - RF0 C.R. - 15 - PRF100 C.R. E 2 4 4 , ----------- ~15 C - --- W - 1 - --- -x- -PRF105 NIMEP - 12 0 -- - - - - - - 0 4-PRFIOO - - - - - - - - - - - - - - -NME - -- 511 9 PRF110 NIMEP , E PRF95 C.R. PRF90 C.R. -- --------- -PRF95 NIMEP 10 11 12 13 14 -.a--PRF90 NIMEP Rc Figure E. 13 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT), 15% fuel reformate fraction. 16 35 0 30 -R 30- -416TRF105 C.R. -- *--TRF100 9C . . . . -. . 25 - 15 C.R. -- S20 - - TRF95 C.R. 14 V -- m-TRF90 U 13 z - C.R. 10- - -- 12 E S0--- - - 11 9 10 12 11 13 - - -- TRFI00 NIMEP TRF95 NIMEP --------------- -5 -TRF105 NIMEP TRF90 NIMEP 14 Rc Figure E. 14 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for TRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT), 15% fuel reformate fraction. 170 16 35 PRF100 C.R. +- PRF95 C.R. -+- 30 15 P - 25- -- 0 -- 20 .a ' - - -- 15 - 10 - - - -4"- ..- ""~ -- -- m-PRF90 o -U--- 14 - cc ~~~~ 0 13 z NIMEP 512 LI) E 0 C.R. PRF85 -+C.R. - PRF80 C.R. - -- -PRFIOO - +-. PRF95 - NIMEP -- -PRF90 NIMEP - -PRF85 011 -5 9 12 11 10 13 NIMEP 14 -- -PRF80 NIMEP Rc Figure E. 15 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT), 30% fuel reformate fraction. -16 35 L- 30 - 15 25 - -4(-TRF105 C.R. -- *-TRF100 C.R. -- +-TRF95 C.R. ---- 20 - 14 V. a. . - - -- 0 15 - TRF90 C.R. -4--TRF85 C.R. -+-TRF80 C.R. -13 z 10 - - -x- TRF105 NIMEP -TRF100 - -(U E 0 NIMEP 5- 0-5 -- - - - - - - - - - - - - - - - - - - - - ' 11 ' 9 12 10 12 11 Rc 13 14 - - TRF95 NIMEP - .0- TRF90 NIMEP +-o TRF85 NIMEP --- TRF80 NIMEP Figure E. 16 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; X = 1.0, 1500 rpm, MAP for 40% boost (NIMEPMBT 1.4 times unboosted NIMEPMBT), 30% fuel reformate fraction. 171 , 35 I- '- a) 11 -r-kPRFI0O C.R. 30 25 - 20 - 15 - 10 - - - - - - -- - . .- - - - - - , 0 I , -PRF95 . - 10 C.R. -U-PRF90 C.R. -+-PRF85 -9 C.R. PRF80 C.R. - CL - .'i .PRFI0O NIMEP 0 - 5E 0 PRF95 +- NIMEP -7 -PRF90 - 0-4--- NIMEP - -5 t 1 9 10 a 11 12 14 13 - -+- PRF85 NIMEP PRF80 NIMEP Rc Figure E. 17 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for PRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT, 15% fuel reformate fraction. 11 35 - ----- 30 -- *-TRF100 - - - -- C.R. -+-TRF95 C.R. - 10 I- 25 - -- TRF90 C.R. 0 20 -4-TRF85 'U 15 10 8 2 - -- 5 7 0 -- - - TRF95 NIMEP - e- - -0- TRF90 NIMEP 6 -5 9 C.R. 'TRF1OO - -- NIMEP *0 E 0 C.R. -+--TRF80 10 12 11 Rc 13 14 TRF85 NIMEP - -- -TRF80 NIMEP Figure E. 18 - Increase of combustion retard to just avoid knock and change of NIMEP with increased re for TRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT, 15% fuel reformate fraction. 172 35 11 .. - I- 30 -.-. -E-PRF90 - - - --.. C.R. - - -- - - -.. - - Ca -- 10 25 20 - 15 - ---- -- - - - - - - -.)-- -- - -+- - - - 8 z 10 - U 0 -- C.R. PRF75 C.R. PRF70 C.R. PRF90 NIMEP - -o- PRF85 5- -- 7 -- E 0 - PRF85 C.R. PRF80 NIMEP - -+- 0 - NIMEP - 6 -5 9 PRF80 10 NIMEP 14 13 12 11 -K- -PRF75 - -- PRF70 NIMEP Rc Figure E. 19 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for PRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT, 30% fuel reformate fraction. 11 35.- 4---- 00a 0 30 - . -. --...- - --- -U-TRF90 . ,------ -- --- e 10 25 - 20 - 15 - -- +-TRF80 9': . - - - - - --- -- - - --- - 10- 8 z C.R. ---- TRF75 C.R. TRF70 C.R. - -TRF90 - -- 57 0 E 0 C.R. TRF85 C.R. NIMEP TRF85 NIMEP - -~- -TRF8O 0- NIMEP - -K- -. 9 10 12 11 Rc 13 13 6 14 TRF75 NIMEP - -- TRF70 NIMEP Figure E. 20 - Increase of combustion retard to just avoid knock and change of NIMEP with increased r, for TRFs; X = 1.3, 1500 rpm, MAP set to match NIMEPMBT at stoichiometric unboosted WOT, 30% fuel reformate fraction. 173