METHYLCYCLOHEXANE IGNITION DELAY TIMES UNDER A WIDE RANGE OF CONDITIONS. Thesis. Submitted to. The School of Engineering of the UNIVERSITY OF DAYTON

Transcription

1 METHYLCYCLOHEXANE IGNITION DELAY TIMES UNDER A WIDE RANGE OF CONDITIONS Thesis Submitted to The School of Engineering of the UNIVERSITY OF DAYTON In Partial Fulfillment of the Requirements for The Degree of Master of Science in Mechanical Engineering By Aditya Nagulapalli Dayton, Ohio May 2015

2 METHYLCYCLOHEXANE IGNITION DELAY TIMES UNDER A WIDE RANGE OF CONDITIONS Name: Nagulapalli, Aditya APPROVED BY: Sukh S. Sidhu, Ph.D. Advisory Committee Chairman Division Head Distinguished Research Engineer Energy Technologies & Materials Division University of Dayton Research Institute Philip. H. Taylor, Ph.D. Committee Member Group Leader Distinguished Research Scientist Environmental Engineering Group University of Dayton Research Institute Moshan Kahandawala, Ph.D. Committee Member Group Leader Senior Research Engineer Bioenergy & Carbon Mitigation Group University of Dayton Research Institute John G. Weber, Ph.D. Associate Dean School of Engineering Eddy M. Rojas, Ph.D., M.A, P.E Dean, School of Engineering ii

3 Copyright by Aditya Nagulapalli All rights reserved 2015 iii

4 ABSTRACT METHYLCYCLOHEXANE IGNITION DELAY TIMES UNDER A WIDE Name: Nagulapalli, Aditya University of Dayton RANGE OF CONDITIONS Advisor: Dr. Sukh. S. Sidhu During the last century, our dependence on oil has increased rapidly and is projected to increase for several decades. There is a critical need to improve the design of the combustion chamber for different kinds of engines to reduce fuel consumption. Chemical kinetics of the fuel plays an important role in reducing emissions and improving engine efficiency. Studying single components of a conventional fuel allows a fuller understanding of the physical and chemical behavior of the real fuel. Many studies have been conducted on all classes of hydrocarbons, with the exception of cycloalkanes. Only a few studies exist on cycloalkanes, which is an important class of hydrocarbons. Methylcyclohexane (MCH), which is widely used as a surrogate to represent the cycloalkane portion of a fuel, was chosen as the subject of this study. The shock tube is an established tool used for measuring the ignition delay, and was used as the experimental apparatus. Ignition delay was measured using the end-plate pressure rise, the OH * and CH * chemiluminescence and white light emission. iv

5 In addition, experimental results were compared with kinetic modeling data using detailed MCH mechanisms developed by Pitz et al. and Orme et al. Different modeling approaches, such as constant volume and internal energy (with and without experimental pressure profiles) and constant pressure, were used to validate the models by comparing against experimental ignition delay data. It was observed that the equivalence ratio affects the ignition delay time. For the lower argon concentration (Ar = 93%) and higher pressure (P ~ 16 atm), ignition delay times were longest for rich conditions. Additionally, they were shorter at lower temperatures (T 1250 K) for stoichiometric conditions in comparison to lean values, but the opposite trend was observed at the higher temperatures (T > 1250 K). Ignition delay times of stoichiometric mixtures were longer than lean mixtures across the studied temperature range for low pressure (P = 2 atm) and argon concentration (Ar = 93%), as well as high pressure (P ~ 16 atm) and argon concentration (Ar = 98%). The Orme et al. model using the approach of constant U,V assumption with experimental pressure profile showed a better agreement with experimental results at low temperatures than the approach without experimental pressure profile. Both models and approaches underestimate the experimental ignition delay times at high temperatures. v

6 DEDICATION Dedicated to Family and Friends vi

7 ACKNOWLEDGMENTS I would like to express my sincere gratitude to my research advisor, Dr. Sukhjinder S. Sidhu, for his substantial support and motivation throughout my master s work done at the Shock Tube Research Laboratory. Besides my advisor, I want to extend my earnest thanks to Dr. Moshan Kahandawala for his guidance, inspiration and efforts throughout my research work. I am indebted to the University of Dayton Research Institute for providing a stimulating environment, as well as the Air Force Research Laboratory for their generous support. Specifically, I would like to express my appreciation for the contract monitor of this project, Mr. Edwin Corporan. I consider it an honor to work on this project as a member of the Sustainable Environmental Technologies Group, and thank them for their continued support. The help of Dr. Saumitra Saxena and Mr. Giacomo Flora in their in-depth guidance and constant availability were of great value during this work. Thesis editing by Dr. Jeremy Cain is also greatly appreciated. Finally, this effort would have been incomplete without the encouragement and the abundant love of my parents, brother and friends. I thank them for their support and understanding during the long years of my education. vii

8 TABLE OF CONTENTS ABSTRACT... iv DEDICATION... vi ACKNOWLEDGMENTS... vii TABLE OF CONTENTS... viii LIST OF FIGURES... xii LIST OF TABLES... xvii LIST OF SYMBOLS... xviii LIST OF ABBREVIATIONS... xix CHAPTER INTRODUCTION Background Ignition Delay Time Shock Tube Methylcyclohexane Current Study CHAPTER EXPERIMENTAL SETUP AND PROCEDURE viii

9 2.1 Shock Tube Components Operation of Shock Tube CHAPTER CALCULATIONS End Plate Velocity Calculation Post-Reflected Temperature and Pressure Calculations Uncertainties of Post-Reflected Pressures and Temperatures Calculation of the Uncertainties Using the Perturbation Method Example of Uncertainty Propagation Fuel Mixture Calculation Fuel Mixture Preparation CHAPTER RESULTS AND DISCUSSION Overview of Results Ignition Delay Correlation Uncertainty Analysis Modeling Modeling Approach Discussion ix

10 4.2.1 Ignition Delay Trends at Different Conditions Impact of Pressures, Equivalence Ratios and Diluent Ratios on Ignition Delay Times Comparison of MCH with Other HC s CHAPTER CONCLUSION CHAPTER FUTURE STUDIES BIBLIOGRAPHY APPENDIX A IGNITION DELAY TIMES MEASURED FROM END PLATE USING OH *, CH *, WL AND PRESSURE APPENDIX B UNCERTAINTIES ON POST-REFLECTED PRESSURES AND TEMPERATURES 72 APPENDIX C UNCERTAINTIES ON IGNITION DELAY TIMES APPENDIX D CONDITION FOR UNCERTAINTY ANALYSIS ON P5 AND T APPENDIX E CHEMKIN INPUT FILE x

11 APPENDIX F SAMPLE FILE FOR FUEL MIXTURE PREPARATION APPENDIX G EXAMPLE OF END PLATE VELOCITY CALCULATIONS APPENDIX H IGNITION DELAY DATA FOR 2-METHYLHEPTANE, N-DODECANE, M-XYLENE AND M-XYLENE/N-DODECANE (23% / 77 %) BLEND xi

12 LIST OF FIGURES Figure 1 Schematic of the shock tube Figure 2 Detailed view of the shock tube test section Figure 3 Schematic of sample preparation unit Figure 4 Shock tube test section with time and distance measurements from the end plate Figure 5 Experimental pressure profiles from oscilloscope showing end plate and sidewall measurements and corresponding incident times Figure 6 Sample oscilloscope trace showing end plate and sidewall pressure profiles from the combustion of MCH Figure 7 Sample oscilloscope trace showing end plate pressure profile, OH * and CH * chemiluminescence and WL emissions from the combustion of MCH Figure 8 Ignition delay measurement using the maximum slope extrapolation method Figure 9 End plate pressure profiles from MCH combustion at 93% argon dilution and 16 atm. Profiles are shown for equivalence ratios of (a) 0.5, (b) 1.0 and (c) xii

13 Figure 10 End plate pressure profiles for = 0.5 and 1 at pressures of 2 and 16 atm with 93% and 98% argon dilution, respectively Figure 11 Normalized ignition delays times for equivalence ratios of (a) = 3.0, (b) = 1.0 and (c) = Figure 12 Pressures scaled to 20 atm for low argon concentration at (a) stoichiometric and (b) fuel lean mixtures Figure 13 Ignition delay for = 0.5, Ar = 93% and P = 20 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 14 Ignition delay for = 1.0, Ar = 93% and P = 20 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 15 Ignition delay for = 3.0, Ar = 93% and P = 20 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend. Scaled values assume n = Figure 16 Ignition delay for = 0.5, Ar = 93% and P = 2 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et xiii

14 al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 17 Ignition delay for = 1.0, Ar = 93% and P = 2 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 18 Ignition delay for = 0.5, Ar = 98% and P = 20 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 19 Ignition delay for = 1.0, Ar = 98% and P = 20 atm. Experimental (OH *, CH *, WL and pressure measured at the end plate) and modeling (Pitz et al. 10 and Orme et al. 21 ) results are shown. A curve fit for modeling with a pressure profile shows its trend Figure 20 Comparison of equivalence ratios of 0.5, 1 and 3 (symbols) with Orme et al. 21 modeling (lines) for 93% argon dilution at 20 atm Figure 21 Rate of production analysis for = 0.5, P = 20 atm, Ar = 93%, T = 1000 K Figure 22 Rate of production analysis for = 1, P = 20 atm, Ar = 93%, T = 1000 K. 51 xiv

15 Figure 23 Rate of production analysis for = 0.5, P = 20 atm, Ar = 93%, T = 1500 K Figure 24 Rate of production analysis for = 1, P = 20 atm, Ar = 93%, T = 1500K.. 52 Figure 25 Comparison of equivalence ratios of 0.5 and 1 (symbols) with Orme et al. 21 model (lines) for 93% argon dilution at 2 atm Figure 26 Comparison of equivalence ratios of 0.5 and 1 (symbols) with Orme et al. 21 model (lines) for 98% argon dilution at 20 atm Figure 27 Comparison of argon concentrations of 93% and 98% (symbols) with Orme et al. 21 model (lines) for equivalence ratio of 0.5 at 20 atm Figure 28 Comparison of argon concentrations of 93% and 98% (symbols) with Orme et al. 21 model (lines) for equivalence ratio of 1 at 20 atm Figure 29 Comparison of MCH ignition delays at pressures of 2 and 20 atm (symbols) with Orme et al. 21 model (lines) for equivalence ratio of 1 at 93% argon dilution Figure 30 Comparison of MCH ignition delays at pressures of 2 and 20 atm (symbols) with Orme et al. 21 model (lines) for equivalence ratio of 0.5 at 93% argon dilution. 58 Figure 31 Comparison of MCH with other hydrocarbons at lean conditions: 2- methylheptane, 47, 48 n-dodecane, 7 m-xylene, 8 and n-dodecane/m-xylene blend Figure 32 Comparison of MCH with other hydrocarbons at stoichiometric conditions: 2- methylheptane, 47, 48 n-dodecane, 7 m-xylene, 8 and n-dodecane/m-xylene blend xv

16 Figure 33 Comparison of MCH with other hydrocarbons at rich conditions: 2- methylheptane, 47 n-dodecane, 7 m-xylene, 8 and n-dodecane/m-xylene blend Figure 34 Variation of shock wave velocity with axial distance in shock tube xvi

17 LIST OF TABLES Table 1 Properties of methylcyclohexane...7 xvii

18 LIST OF SYMBOLS τign ignition delay time (ms) tincident incident shock arrival time (μs) tdwell dwell time (ms) U1 incident shock velocity (m s -1 ) R universal gas constant (cal mol -1 K -1 ) n number of moles Vi volume fraction (m 3 ) MW molecular weight (g mol -1 ) specific gravity (g ml -1 ) Xi mole fraction of species Pi partial pressure of species (Torr) (O/F)act actual oxygen-to-fuel ratio (O/F)stoic stoichiometric oxygen-to-fuel ratio equivalence ratio xviii

19 LIST OF ABBREVIATIONS AEO HCs ID JP MCH MW PMT PT RCM ROP RP SPU UDRI U,V WL annual energy outlook hydrocarbons inner diameter jet propulsion methylcyclohexane molecular weight photo multiplier tube pressure transducer rapid compression machine rate of production rocket propulsion sample preparation unit University of Dayton Research Institute internal energy and volume white light xix

20 CHAPTER 1 INTRODUCTION 1.1 Background During the last century, the dependence on oil has increased rapidly and is projected to increase for the next several decades. As a result, pollutants (e.g., NOx, SOx, PM and HC s) are emitted into the atmosphere that cause adverse environmental and health effects, among which is an increase in global temperatures from greenhouse gases. 1 According to the Annual Energy Outlook (AEO), the transportation sector is a major consumer of fossil fuels. 2 It consumed 71% of the total liquid fuels in 2009, and usage by transportation is expected to increase to 73% by The sensitive relationship between the increasing domestic crude oil production and rising fuel prices shows the strong demand and potential for an alarming global fossil fuel crisis. Although the AEO also reported that the overall pollution levels will be slightly lower than current values by 2035, the projected values of greenhouse gas emissions (particularly CO2) will still be higher than current acceptable limits. Efforts to ensure efficient, clean combustion of fuels will positively contribute to the aforementioned issues through reduced fuel consumption and pollutant emissions. Thus, there is a critical need to improve the combustion chamber design in engines. Knowledge of the kinetic mechanisms governing the combustion of fuels is crucial to 1

21 achieving this. Chemical kinetics plays an important role in identifying the key reactions responsible for fuel oxidation and pollutant formation. The reactions describing the global conversion of reactants to products via many intermediate reactions makeup a chemical kinetics mechanism. Computer software is used to process the mechanism in simulating the fuel s combustion and, thus, obtain temporal evolution of species concentrations, temperature and pressure. However, before utilizing the mechanism to solve applied problems, it must be validated against relevant experimental data. Practical fuels such as gasoline, diesel and jet fuels consist of hundreds to thousands of different hydrocarbons, including paraffins (normal, branched and cyclo), olefins and aromatics. 3 Combustion simulation of these fuels involve complex mechanisms (large number of reactions and species), and their comprehensive modeling is currently not feasible due to the high computational cost and time. 3 One common approach used to reduce the computational efforts is to replace the practical fuel with a surrogate fuel that consists of a mixture of selected representative species. The surrogate formulation is chosen to reproduce the physical and/or chemical properties of the real fuel. For this purpose, three types of surrogate fuels exist: physical surrogates (e.g., boiling point, flash point and molecular weight), chemical surrogates (e.g., chemical composition and sooting tendency) and comprehensive surrogates (both physical and chemical). The importance of fuel surrogate components, their relevance to practical systems, and the degree of understanding of their properties and mechanisms has been previously studied. 3 It was explained there that the mechanisms for straight and branched alkanes are well understood, 2

22 but those for cycloalkanes and single- and multi-ring compounds are least understood, in addition to the properties of cycloalkanes and aromatics. Various hydrocarbon classes such as normal and branched alkanes 4-7 and aromatics 8 have been studied extensively, and literature reviews are available. 9 Thus, it is necessary to focus on other significant and lesser understood species such as cycloparaffins. 10 Cycloalkanes are an important class of hydrocarbons 11 that constitute ~ 60% by volume in RP-1, 12 ~ 20% by volume in Jet-A/Jet-A1/JP-8, 13 and 30 65% by weight in diesel fuels. 14 Methylcyclohexane is widely used to represent the cycloalkane portion in diesel and jet fuels. 3, 10, 15 The goal of the present study is to provide experimental ignition delay data to the cycloparaffin database by characterizing the ignition behavior of MCH. It also seeks to utilize the data to validate existing chemical kinetic mechanisms. 1.2 Ignition Delay Time Ignition delay time represents a global parameter for the validation of chemical kinetic mechanisms. It also provides useful information, such as the chemical kinetic time scale at any temperature for the design of combustion chambers. 16 In general, ignition delay is defined as the time taken for a fuel-oxidizer mixture at an initial temperature and pressure to initiate the ignition process. 16 In many chemically reacting systems (e.g., combustion), reaction processes are governed by chain reactions. The main elements in chain reactions are chain carriers because they are directly responsible for the global reaction progress. In combustion, radicals that serve as chain carriers have been identified as transient intermediate species. 3

23 Ignition takes place once the amount of radicals (radical pool) becomes large enough to consume the fuel. Ignition is experimentally observed from radical species such as OH and CH when they are in an excited state (OH * and CH * ). 16 Due to the large amount of heat released during combustion, both temperature and pressure rise during ignition. Large pressure rise are typical of ignition conditions involving high fuel concentration, 17 and its delayed response is typical of low fuel concentration. The ignition delay time can be defined either by the maximum rate of change or the time when a specie or pressure or visible light emission reaches its peak value; it can also be derived by extrapolating the maximum slope to zero signal level. 17 The ignition delay time depends on various parameters such as pressure, temperature, equivalence ratio and fuel composition. Correlations developed from ignition delay data at different parameter settings help understand the dependency on these parameters, as shown in previous investigations. 6, 15 The correlations are useful for modelers as an initial step in understanding the kinetic processes. 17 Moreover, they allow comparison of data sets at different conditions. Plotting the natural logarithm of ignition delay time against the inverse of pre-ignition temperature gives the global activation energy for the fuel. Ignition delay can be characterized into physical and chemical ignition delay. The former depends on parameters such as atomization, vaporization and mixing of fuel and air, while the latter depends on the fuel s chemical composition. Physical effects can be separated from chemical effects by pre-vaporizing and premixing fuel with air and making a homogeneous gas-phase fuel-air mixture. Previous studies on ignition delay times of methylcyclohexane were performed using rapid compression machines 18, 19 and shock tubes. 15, Comparing the data from different reactors with different reaction 4

24 conditions is difficult. A better approach to cover a wide range of conditions is to use the same system. The shock tube can help understand the effects on MCH ignition delay by studying the chemical effects without reactor specific physical effects. Data from different reactors helps validate the kinetic mechanism and computational approaches taken to simulate the various reactor environments (e.g., fluid dynamic and thermal processes) with a previously validated model. 11 The current study uses a shock tube to obtain the ignition delay time data due to its advantage over other systems, as explained below. 1.3 Shock Tube The shock tube is a well-established tool for chemical kinetic studies. It allows a gas mixture at an initial pressure and temperature to almost instantaneously reach high temperature and pressure by a reflected shock wave. Furthermore, shock tube experiments have good reproducibility and allow smaller fuel volumes (microliters) to be tested compared to experimental engines (gallons). The design and principles of shock tube operation have been explained elsewhere. 24, 25 The shock tube consists of a high pressure (driver) section and a low pressure (driven) section separated by a diaphragm. A sudden burst in the diaphragm produces a planar shock wave as the high pressure gas flows into the driven section. The shock wave compresses the driven gas as it travels into that section. An expansion fan (rarefaction wave) is also produced by the diaphragm burst, and it travels into the driver section and reflects back from the driver end plate. A contact surface is formed at the interface between the driver and driven gases; it travels rapidly behind the shock front. The incident shock wave advances quickly into the driven section, impinges upon the driven end plate and is reflected back into the driven section, thus, elevating the 5

25 temperature and pressure to a maximum in the shock tube driven section. The reflected expansion wave quenches the reactions in the driven section upon coming into contact with the post-reflected shocked gas. The interactions between the reflected shock wave and contact surface are explained by Gaydon and Hurle. 25 The region behind the reflected shock wave creates a uniform temperature and pressure region until it is quenched by the rarefaction wave. Thus, the shock tube provides a great environment for chemical kinetic studies. However, non-ideal effects such as incident shock attenuation and boundary layer development behind the reflected shock can lead to uncertainties in pressures and temperatures. Petersen 26 explained the non-ideal effects behind a reflected shock wave and developed a reflected shock gas dynamic model to measure post-reflected pressure and temperature uncertainties, which were in good agreement with his experimental data. Studies on boundary layer instabilities, 27 turbulent boundary layer effects on shock tube test times 28 and flow instabilities due to shock wave boundary layer contact surface interactions 29 have been conducted. It has been shown that boundary layer effects can by minimized by tailoring the shock wave. 25 Hong et al. 30 developed a theoretical model based on their contact surface tailoring studies to estimate the contact surface tailoring conditions in a convergent shock tube. The ignition delay measured in the shock tube is defined as the time interval between the arrival of the incident shock at the end of the driven section (i.e., test section end plate) and detection of the onset of sustained ignition. Shock tubes have been used to measure ignition delay of different fuels. Comparisons are made with ignition delay from other fuel surrogates at similar conditions, 7, 8 as well as other 15, 20, 22, 23 studies on MCH. 6

26 1.4 Methylcyclohexane Methylcyclohexane (C7H14) is a colorless cycloalkane with a faint benzene-like smell. It is a component of jet fuel, and it is widely used in surrogate fuels to represent the cycloalkane portion of jet fuel. Properties of MCH are given in Table 1. Table 1 Properties of methylcyclohexane. 19 Property Value Molar mass g mol -1 Density 0.77 g cm -3 Melting point -195 F Boiling point 214 F Flash point 25 F Auto ignition temperature 482 F Several ignition delay studies have previously been performed on MCH. Pitz et al. 10 investigated its ignition behavior over a temperature range of K at pressures of atm with diluents of argon, nitrogen and 50/50 (weight %) nitrogen/argon at stoichiometric conditions using a rapid compression machine. A negative temperature coefficient (NTC) region was observed in the experimental data. Pitz et al. also developed a new low temperature oxidation mechanism for MCH and combined it with an existing high temperature mechanism from Orme et al. 21 Their computational work indicated that using n- and iso-alkane based estimates of RO2 isomerization rate constants does not show the NTC region and estimates much longer ignition delays at lower temperatures. Upon replacing the RO2 isomerization rate constants with experimental values from Gulati and 7

27 Walker, 31 the computed ignition delay times showed a better agreement at low temperatures, as well as the NTC region with experimental ignition delay times from Pitz et al. 10 Tanaka et al. 19 studied some pure fuels using a rapid compression machine (RCM) and showed that combustion characteristics are strongly dependent on the fuel structure: either single- or two-stage ignition occurred. A two-stage ignition phenomenon was observed for all saturated compounds (including MCH) and olefins tested, while unsaturated cyclic compounds showed a single-stage ignition. The authors explained that the absence of two-stage ignition is most likely associated with the low heat release during the first stage of combustion. Faster burning rates were noticed in MCH combustion than iso-octane and unbranched cyclic compounds. The phenomenon of two-stage ignition was explained differently by Mittal et al. 18 They used an RCM to investigate the auto-ignition of MCH at = , pressures of 15.1 and 25.5 bar and a temperature range of K. Two-stage ignition and a strong NTC behavior were observed. They explained that the ignition behavior was strongly dependent on the ignition temperature: two-stage ignition occurred at low temperatures and single-stage ignition was observed at high temperatures. A shift in the NTC region to higher temperatures at high pressure and oxygen-rich conditions was also observed. Hawthorn and Nixon 20 measured ignition delay times of MCH behind incident shock waves. Very low concentrations of fuel (5% and 1% fuel-oxygen mixtures) were used with argon as a bath gas. A few experiments were run replacing argon with nitrogen, but no significant effect on ignition delay was observed. The ignition delay of MCH/O2/Ar was recorded for a temperature range of K, at pressures of 0.61, 1.02 and 1.7 8

28 atm and = They observed that the influence of equivalence ratio is qualitatively larger in stoichiometric and rich regions. Although the effect of pressure, reactant concentrations and equivalence ratios were studied, no correlations with these variables were developed. Orme et al. 21 studied the ignition delay times of MCH/O2/Ar behind the reflected shock wave over a temperature range of K for reflected shock pressures of 1, 2 and 4 atm and equivalence ratios of 0.5, 1.0 and 2.0. The ignition delay times measured ranged from less than 50 s to greater than 1 ms. However, since the measurement uncertainty was very high, the small ignition delays were ignored. Orme et al. replicated the experimental conditions previously studied by Hawthorn and Nixon; 20 good agreement between both data sets were observed. A detailed chemical kinetic model was also created by Orme et al. Their model showed a good agreement with their experimental measurements of ignition delay times. Vasu et al. 15 measured the ignition delay times of MCH/O2/Ar and MCH/air mixtures behind the reflected shock wave in two different shock tubes. The experiments were conducted at low and high pressure conditions. At low pressure conditions, the ignition delay was measured over post-reflected temperatures of K, pressures of atm and equivalence ratios of with argon as the diluent. An ignition delay correlation developed for the low pressure data showed a strong dependency on equivalence ratio and oxygen concentration, and the ignition delay times showed a good agreement with several mechanisms. The high pressure ignition delay time data were for a post-reflected shock temperature range of K, pressures of atm and = 1 in air. Vasu et al. also studied the effects of pressure on MCH ignition delay by 9

29 complementing earlier measurements of MCH made in an RCM by Pitz et al. 10 Strong ignition was seen in the high pressure data, with lower pressures having longer ignition delays. The NTC behavior was also observed at 45 atm for T < 880 K. Another study on MCH using a shock tube was conducted by Vanderover et al., 22 who measured the ignition delay of MCH/air at reflected shock temperatures and pressures of K and atm, respectively, at equivalence ratios of 0.25, 0.5 and 1.0. The ignition delay was measured using the end plate OH * and sidewall pressure profile. The effect of equivalence ratios, pressures and temperatures on ignition delay was characterized. Inverse dependencies on pressure and equivalence ratios were observed for stoichiometric and lean conditions with a wide range of pressure scaling parameters. Their experimental data showed the NTC behavior for T < 1000 K, which agrees well with prior work done by Orme et al., 21 Pitz et al. 10 and Vasu et al. 15 Hong et al. 23 measured the ignition delay times for MCH behind reflected shock waves, and compared the ignition delay times with cyclohexane and butylcyclohexane (BCH) at pressures of 1.5 and 3 atm for = 0.5 and 1 over a temperature range of K. They observed that the ignition delay times measured were longer for MCH than BCH. 1.5 Current Study Several studies were conducted on the ignition delay of MCH prior to this work. The present study focused on the ignition delay time measurements from reflected shock wave combustion of MCH/O2/Ar mixtures with argon dilutions of 93% and 98%, pressures of 2 ± 0.06 and 16 ± 1.5 atm, equivalence ratios of 0.5, 1 and 3, and pre-ignition 10

30 temperatures of K. The experiments were conducted in a shock tube to study the effects of pressure, diluent concentration, equivalence ratio and temperature on the ignition delay times of MCH. The experiments included a wide range of conditions and were intended to cover fuel-rich ( = 3) and high pressure (16 atm) conditions using argon as a bath gas, which has not been studied before. Thus, the goal of this investigation was to extend the MCH ignition delay database. All experiments were conducted using tailored conditions. Experiments were performed for a wide range of conditions over constant dwell times (7.6 ± 0.2 ms) and scaled to common pressures (2 and 20 atm). The uniqueness of this study is to develop a correlation for ignition delay that depends on all the parameters (i.e., pressure, argon concentration, equivalence ratio and temperature) for the wide range of data set, where other studies have looked at these effects individually. Ignition delay was also modeled using the mechanisms of Orme et al. 21 and Pitz et al. 10 under different modeling approaches to validate the models with measurements of this study. The ignition delay of MCH was plotted against other hydrocarbons to study the behavior of cycloalkanes relative to other classes. The uncertainties on P5, T5, fuel concentration and incident shock velocity were calculated, and the experimental uncertainties of ignition delay measurements were also calculated. 11

31 CHAPTER 2 EXPERIMENTAL SETUP AND PROCEDURE The shock tube used in this study is a single pulse, double diaphragm, non-heated shock tube, and is made from SS 304 with ½ thick walls. It is used to study the high pressure, high temperature combustion of various hydrocarbon fuels behind reflected shock waves. The shock tube mainly consists of four sections: a driver section that provides the energy to produce the supersonic shock wave, a driven section that allows a planar shock front to develop, a breaker section connecting the driver and the driven sections that also helps initiate the shock and a test section that contains the test mixture. The experimental device is located in the Shock Tube Laboratory at the University of Dayton Research Institute. The shock tube was designed to study post-reflected temperatures of K, post-reflected pressures of atmospheres and dwell times up to 12 ms. It is used for the combustion of hydrocarbon fuels in the gas phase (i.e., fuels with relatively high vapor pressure). 2.1 Shock Tube Components The shock tube is comprised of multiple sections, as shown in Figure 1. The driver section of the shock tube is 274 cm long and has an inner diameter (ID) of 7.62 cm. Its inner surface is finely polished. Although the driver section is a single section, it 12

32 is divided into two parts (driver and breaker sections) for operational purposes. The driver portion is always maintained at a higher pressure than the other sections so that the pressure differential between the chambers will create a shock wave. The gases to be filled in the driver section are chosen with great care so that they do not react with the test gas. It is also important to ensure that the molecular weight is low because the speed of the sound increases with a decrease in molecular weight. Helium is chosen for the driver section, and a predetermined percentage of argon is added to the helium to obtain the desired velocities. Argon addition also helps match the acoustic impedance of the gases to obtain uniform post-reflected shock pressures of a perfectly tailored shock wave, which is achieved by the tailored interface technique. 32 The breaker section attached to the driver section plays an important role in initiating the shock wave. This section is cm long and has an ID of 5.08 cm. It is located between the driver and the driven sections, and has Mylar diaphragms on either side. The diaphragm thickness varies with the differential pressure; a thicker diaphragm is used when higher pressures are maintained in the driver section. A pneumatic valve connects the breaker section to the breaker vacuum chamber that is initially maintained under vacuum. The driven section is 274 cm long and has an ID of 5.08 cm. Its inner surface is finely polished. The driven section contains ports for filling this portion with test gases or to obtain shock wave pressure-time histories. A dump tank (ID = cm, length = cm) is also connected to the driven section; this assures only a single reflected pulse and prevents repetitive heating of the test gas by the shock wave. 13

33 Dump Tank Isolation Valve Vacuum Pump Test Section with Access Ports Isolation Valve Driver Section Breaker Section Breaker Vacuum Section Driven Section Optical Diagnostic Setup Figure 1 Schematic of the shock tube. 14

34 The test section is an important component of the shock tube because the ignition and combustion of flammable reactant mixtures occur there. The test section is cm long and 5.08 cm in ID. It also has a finely polished inner surface similar to the driven section. Figure 2 shows the schematic of the test section. It has six access ports: two are used for purging with helium, one is used for introducing the test gases into the test section from the sample preparation unit (SPU) and three are used for measuring pressure. There are four piezoelectric pressure transducers (PCB Piezotronics, model 113A24) total: one at the end plate and three on the sidewall of the test section. All pressure and optical events are monitored using two four-channel digital oscilloscopes (Tektronix, models TDS3014B and TDS3014). A pressure gauge and capacitance manometer are also connected to monitor the pressure inside the test section. Figure 2 Detailed view of the shock tube test section. The SPU is attached to the test section of the shock tube, as shown in Figure 3. This setup is used to vaporize the liquid fuel (e.g., methylcyclohexane), dilute it with bath gas and introduce it into the test section. The SPU is attached to the test section through 15

35 the injection port. The SPU consists of a glass bulb for introducing the fuel into a 14.5 L chemically inert air sampling canister (Restek, SilcoCan ) that holds the fuel mixture, a capacitance manometer for pressure measurement, a stainless steel manifold for gas delivery/transport and a vacuum pump. Figure 3 Schematic of sample preparation unit. The experiments were conducted with the diagnostic setup positioned at the end plate of the shock tube, as shown in Figure 2. A quartz rod is mounted flush at the end plate and acts as a window for optical diagnostic setup to measure OH * and CH * chemiluminescence and white light (WL). All measurements were relative to the pressure signal generated by the pressure transducer located at the end plate. A photomultiplier tube (PMT), fitted with a 20 nm band pass interference filter centered at = 430 nm, was used to detect CH A-X chemiluminescence. A second PMT, fitted with a 20 nm band pass interference filter centered at = 307 nm, was used to detect OH A-X chemiluminescence. Both PMTs (Hamamatsu Corporation, model H ) included the high voltage power supply. Light traveling through the quartz rod is passed through a beam splitter to divert 16

36 it towards these PMTs. Unfiltered visible light passes through the end of the quartz rod to a PMT for white light emission. The rise time of PMT signals was approximately 2 s. The shock tube contains two pneumatic valves, one in the breaker section and another in the test section. The former separates the pressurized breaker section and the adjoining vacuum section. Upon opening this valve, a negative pressure will be exerted on the diaphragms, causing them to rupture and create a shock wave. The pneumatic ball valve that separates the test section from the driven section instantaneously opens for 0.5 s and subsequently closes by programming command. The system allows the shock wave to enter the test section, ignite the mixture and trap the products for further study. A control panel is utilized to control the pneumatic valves, turbo pump, monitor the pressure and control the flows of different gases into the driver, breaker and driven sections. 2.2 Operation of Shock Tube Shock tube operation consists of several processes. The shock tube should be cleaned between tests. The cleaning procedure for the test section involves removing diaphragm debris and soot particles (incomplete combustion products). The diaphragm debris is cleaned up using Kimwipes; soot and other products are thoroughly cleaned by wiping the tube inside with hexane-treated Kimwipes until no soot appears on the wipe. An oxygen shock is carried out (after every three combustion experiments) to prevent buildup of any residual combustion products on the walls after cleaning with the hexanetreated Kimwipes. An oxygen shock is performed by filling the empty test section with oxygen and creating a shock wave to combust any impurities (soot, dust, etc.) present. 17

37 After cleaning, the pneumatic ball valve is closed, the pressure gauges are isolated, and the test section is purged at a high pressure with helium for approximately two minutes. Two Mylar diaphragms of different thickness are placed in the breaker section, one at each end. The diaphragm thickness depends upon the post-reflected pressure (P5) to be generated. The driver, driven, breaker assembly and test sections are pumped down to pressures of 10-3 Torr using vane pumps. The test and driven sections are pumped down further to 10-6 Torr using a turbopump. The test section is monitored for leaks before each experiment. The leak rates measured in the test section should be less than 0.01 Torr per minute. Once the tube is under vacuum, both pneumatic valves are closed using electronic switches on the control panel. Both diaphragms respond to filling helium in the driver and breaker sections. The driver pressure is maintained at a higher value than the breaker pressure; thus, the diaphragms deflect towards the low pressure side. The test section is filled with the fuel (MCH) and argon from the canister and oxygen and additional argon (makeup argon for required diluent gas partial pressure) directly from the cylinders. The fuel mixture preparation and calculation are explained in Chapter 3. After filling the test section with the test gas mixture, it is kept undisturbed for 30 minutes for homogenous mixing. Meanwhile, the driver, driven, and breaker sections are filled with appropriate gases. Once all the sections are filled with the desired gases, the two pneumatic valves are manually put into auto position. They are electronically opened in sequence to initiate the shock. Rupture of the diaphragms causes a shock wave to travel into the test gas, 18

38 compressing it and elevating its temperature and pressure to high values for the combustion to take place. The control system is set up to trap the products in the test section by immediately closing the pneumatic ball valve within 0.5 s. The shock wave arrival is detected by the piezoelectric pressure transducers located on the sidewalls and the end plate of the test section and can be monitored using the four channel digital oscilloscope. The diagnostic setup is attached to the shock tube end plate and is capable of capturing simultaneously the WL and chemiluminescence emissions of CH * and OH * on to another four channel digital oscilloscope. The ignition delay can be measured from the time of the shock front arrival at the end plate to the onset of the WL and emissions of CH * and OH *. A typical trace of both oscilloscopes is explained in Chapter 4. 19

39 CHAPTER 3 CALCULATIONS 3.1 End Plate Velocity Calculation The velocity of the incident shock wave at the end plate of the test section was computed using a linear extrapolation of the mean velocities between the pressure transducers located along the side wall of the test section in relation to the location of the end plate. Figure 4 shows the test section of the shock tube, which contains three pressure transducers on the sidewall and one at the end plate. From the oscilloscope trace shown in Figure 5, the three incident times t1, t2 and t3 are measured. The distances d1, d2 and d3 corresponding to the distances between each pressure transducer and the end plate are known from the shock tube s physical dimensions. Velocities V1, V2 and V3 can be calculated by dividing the corresponding distance by time. Thus, velocities are calculated as follows: V i = d i t i, i = 1 6. (3.1) The mean velocities V4, V5 and V6 are calculated using distances and times that are given by the following expressions: t 4 = t 1 t 3, (3.2a) 20

40 t 5 = t 2 t 1, t 6 = t 2 t 3, d 4 = d 1 d 3, d 5 = d 2 d 1, d 6 = d 2 d 3. (3.2b) (3.2c) (3.3a) (3.3b) (3.3c) The distances from PT1 to the midpoint of two sidewall pressure transducers (PT2 and PT3, PT3 and PT4 and PT2 and PT4) were calculated. They are denoted by 4, 5 and 6, and are given by the following: 4 = d 5 2, (3.4) 5 = d 5 + d 6 2, (3.5) 6 = d 6 2. (3.6) The calculated distances 4, 5 and 6 were plotted against the velocities V4, V5 and V6 and extrapolated to the end plate to obtain the velocity there. An example of this calculation is found in Appendix G. Figure 4 Shock tube test section with time and distance measurements from the end plate. 21

41 Figure 5 Experimental pressure profiles from oscilloscope showing end plate and sidewall measurements and corresponding incident times. 3.2 Post-Reflected Temperature and Pressure Calculations The post-reflected temperatures and pressures are calculated using CHEMKIN- PRO, 33 which utilizes iterative methods to solve the 1 D shock equations. CHEMKIN- PRO takes initial parameters such as temperatures, pressures, velocities and mole fractions as inputs to generate the post-reflected conditions (Appendix E). The calculation of these post-reflected pressures and temperatures are explained in detail in the CHEMKIN manual. 3.3 Uncertainties of Post-Reflected Pressures and Temperatures Perturbation methods have been used to calculate the uncertainties of post-reflected pressures and temperatures. The upper and lower perturbed values of the post-reflected conditions are calculated from the uncertainties in the initial temperatures, pressures, velocities and test section mixture. The root sum square formula was applied on the averaged values of the upper and lower perturbations to obtain the uncertainties of postreflected pressure and temperature. An example of the results obtained from a particular 22

42 condition is shown in Appendix D. The effects from the above parameters on post-incident and post-reflected pressures and temperatures are listed as absolute values and percentages. It is important to note that the uncertainty of the incident shock velocity contributes more than other uncertainty values in the uncertainty of post-reflected pressure and temperature. It is also seen that the uncertainty of the initial pressure contributed less than the uncertainty of the initial temperature, and that the uncertainty of post-reflected temperature and pressure were approximately two orders of magnitude greater than those of the initial temperature and pressure. 3.4 Calculation of the Uncertainties Using the Perturbation Method given by: Let x be the error of quantity x. The upper and lower perturbed values of y are y + = R (x + x), (3.7) y = R (x x), (3.8) where R is the function of y in terms of x. Changes to y values are given by: d y + = y y +, (3.9) d y = y y. (3.10) The error y can be calculated from the above values through the expression: y = (d + y + d y ) 2. (3.11) Therefore, the percentage error results are given by: 23

43 % Error = 100 ( y y). (3.12) Example of Uncertainty Propagation The uncertainty of argon mole fraction in the reactant mixture was calculated from the uncertainty in the partial pressures. Let P1 be the sum of partial pressures of fuel, oxygen and argon in the test section: P 1 = P Ar + P Ox + P f. (3.13) The argon mole fraction is expressed as Differentiating 3.14 with respect to PAr gives X Ar P Ar = 1 P Ar + P Ox + P f X Ar = P Ar P = Ar. (3.14) P 1 P Ar + P Ox + P f P Ar (P Ar + P Ox + P f ) 2 = P Ox + P f (P Ar + P Ox + P f ) 2 = P 1 P Ar P 1 2. (3.15) The uncertainty of the argon partial pressure may be determined by using the root sum square formulae: up Ar = (( X Ar P Ar ) P) 2 = P P 1 P Ar 2. (3.16) P Fuel Mixture Calculation The MCH/O2/Ar mixture was studied at different equivalence ratios and argon dilutions. Calculations for one of the conditions using 93% argon and = 3 is presented in this section. The fuel mixture was prepared in the SPU before being introduced into the test section of the shock tube. Preparation of the fuel mixture is explained in the next section. The calculations for the amounts of fuel and argon to be put into the SPU are explained in this section. 24

44 The complete combustion of fuel (MCH: C7H14) into carbon dioxide and water vapor is given below in the balanced, stoichiometric global reaction: C 7 H O 2 7 CO H 2 O. (3.17) The stoichiometric oxygen-to-fuel ratio is readily calculated: (O F) stoic = = (3.18) The relation between the actual and stoichiometric oxygen-to-fuel (O/F) ratios can be given as: Assuming Ф = 3, the oxygen-to-fuel ratio is (O F) actual = (O F) stoic. (3.19) (O F) actual = = 3.5. (3.20) For an argon dilution of 93%, the remaining 7% is the fuel and oxygen mixture: X O2 + X f = (3.21) The mole fractions of individual components can be calculated using 3.22 and 3.23: X O2 = 3.5 X f, (3.22) X f X f = (3.23) Solving 3.22 and 3.23 gives Xf = , XO2 = and XAr = This allows the partial pressures of all species to be obtained: P f = X f P, P O2 = X O2 P, P Ar = X Ar P. (3.24a) (3.24b) (3.24c) By substituting the density relationship into the ideal gas equation for the canister, the following expression may be generated for the fuel injection volume: 25

45 V inj = P f V canister ρ f T (R u ). (3.25) MW f 3.6 Fuel Mixture Preparation The initial step in the fuel mixture preparation is the injection of the fuel into the glass bulb. Equation 3.25 gives the amount of fuel to be injected into the glass bulb. The SPU is evacuated to ~ 10-5 Torr before injecting the fuel into the glass bulb. At the time of injection, the glass bulb is isolated from the canister. Upon the injection of a known volume of liquid fuel, some of the MCH vaporizes and reaches the saturated pressure, which is read from the digital pressure measurement. After keeping the fuel in the glass bulb for two minutes to allow the pressures to stabilize, the canister is opened and the MCH vapors start to fill the canister. The canister pressure is monitored at regular intervals. Once the canister pressure stabilizes, a predetermined amount of argon is put into the canister. This argonfuel mixture is utilized in every experiment, with the oxygen being added separately into the test section for each test. An example case of the fuel mixture preparation is given in Appendix F for reference. 26

46 CHAPTER 4 RESULTS AND DISCUSSION 4.1 Overview of Results Methylcyclohexane ignition delay times were measured behind the reflected shock wave in a single pulse reflected shock tube. Pressures of 2 and 16 atm, equivalence ratios of 0.5, 1 and 3, argon dilutions of 93% and 98% and a temperature range of K were tested. The experimental ignition delay data are presented in Appendix A. Ignition delay times were measured from the pressure profiles recorded at the end plate of the test section, as well as the chemiluminescence emissions of CH * and OH * and white light (WL) broadband emissions from the same location. Example oscilloscope traces of these signals from the combustion of MCH are given in Figures 6 and 7. In this study, the ignition delay time is defined as the time difference between the arrival of the incident shock wave at the test section end plate and the maximum rate of change of the optical emissions (OH *, CH * and WL) extrapolated to the pre-ignition baseline, as shown in Figure 8. The dwell time, which is the time available for combustion, is defined as the time interval between the arrival of the incident shock wave at the test section end plate and the arrival of the reflected rarefaction wave at the same location. The measurement of the dwell time is also shown in Figure 7. 27

47 Figure 6 Sample oscilloscope trace showing end plate and sidewall pressure profiles from the combustion of MCH. Figure 7 Sample oscilloscope trace showing end plate pressure profile, OH * and CH * chemiluminescence and WL emissions from the combustion of MCH. The arrival and reflection of the incident shock wave at the end plate is observed from the first sudden rise in the pressure profile recorded there. The incident times are measured using the three sidewall and end plate pressure profiles, i.e., measuring the time 28

48 between the pressure rise due to the arrival of the incident shock wave at the sidewall and end plate. These times are subsequently used to calculate the incident shock velocity at the test section end plate. The procedure used for this computation is given in section 3.1. Figure 8 Ignition delay measurement using the maximum slope extrapolation method. The experimental pressure profiles recorded at the test section end plate are plotted in Figures 9 and 10. They show the post-reflected pressures, dwell times and the rise in pressures due to ignition (in some instances). Strong oscillatory peaks in the pressure profile are observed for almost all conditions having a 93% bath gas (argon) concentration. This indicates a strong ignition with a progression to a detonation-like phenomenon Saxena et al. 37 observed a similar effect in the combustion of ethylene for the same diluent gas and concentrations. When a high fuel concentration mixture (93% bath gas dilution) containing a large volume of unreacted gas homogeneously ignites, strong ignition with transition to supersonic combustion phenomena takes place. 36 Supersonic combustion can lead to shock waves that cause strong pressure peaks and steep pressure gradients. 29

49 However, the detonation process has no effect on the ignition delay since ignition occurs before detonation commences. 38 Although strong peaks are seen at the low pre-ignition pressure (~ 2 atm), a sudden rise in pressure does not appear due to the lower concentration of fuel present in the shock tube test section. A mild ignition with no strong rise in the pressure profile due to ignition was noticed at the higher argon dilution (Ar = 98%). Although it is likely not the case here, it is noted that mild ignition can also occur due to local hot spots (non-homogeneities in temperature) that can lead to one or several flame kernels. 36 Figure 9 End plate pressure profiles from MCH combustion at 93% argon dilution and 16 atm. Profiles are shown for equivalence ratios of (a) 0.5, (b) 1.0 and (c)

50 Figure 10 End plate pressure profiles for = 0.5 and 1 at pressures of 2 and 16 atm with 93% and 98% argon dilution, respectively. A gradual rise in the pre-ignition pressure is observed in the end plate pressure profiles, as shown in Figures 9 and 10. This phenomenon is possibly due to a boundary layer that builds up behind the incident shock wave. The interaction of the reflected shock 17, 26 wave with the boundary layer gives a gradual rise in pressure. Previous studies observed that the boundary layer is formed in shock tubes with small internal diameters. A previous investigation using a ½ diameter shock tube observed an attenuation of the incident shock velocity, indicating the presence of a boundary layer with considerable thickness. 25 A similar phenomenon is observed for the low fuel concentration at high postreflected pressure in this study (see Figure 10). Oehlschlaeger et al. 39 measured the ignition delay times for the combustion of iso-octane in a shock tube with an internal diameter of 31