A RAPID COMPRESSION MACHINE – DESIGN, CHARACTERIZATION, AND AUTOIGNITION INVESTIGATIONS
by GAURAV MITTAL
Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy
Dissertation Adviser: Prof. Chih-Jen Sung
Department of Mechanical and Aerospace Engineering CASE WESTERN RESERVE UNIVERSITY
January, 2006
CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
Gaurav Mittal ______________________________________________________ candidate for the Ph.D. degree *.
C. J. Sung
(signed)_______________________________________________ (chair of the committee)
D. L. Feke
________________________________________________
J. M. Prahl
________________________________________________
Y. Kamotani
________________________________________________
________________________________________________
________________________________________________
8/29/05
(date) _______________________
*We also certify that written approval has been obtained for any proprietary material contained therein.
iii Table of Contents LIST OF TABLES
vi
LIST OF FIGURES
vii
ACKNOWLEDGEMENTS
xvii
ABSTRACT
xviii
Chapter 1 Introduction
1
1.1 Chemical Kinetics
1
1.2 Experimental Facilities for Studying Homogeneous Combustion
3
1.2.1 Static Reactors
3
1.2.2 Flow Reactors
4
1.2.3 Well-Stirred Reactor
5
1.2.4 Shock Tubes
7
1.2.5 Rapid Compression Machines
8
1.3 Advantages and Challenges with Rapid Compression Machines
9
1.4 Specific Objectives
13
1.5 Structure of Dissertation
14
Chapter 2 Experimental Facility
19
2.1 Introduction
19
2.2 Review of Existing RCMs
21
2.3 Design and Operation of Rapid Compression Machine
25
2.3.1 Rapid Compression Machine Configuration
25
2.3.2 Optimization of Reactor Piston Crevice
28
2.3.3 Experimental Procedure
32
iv Chapter 3 Aerodynamics Inside The Rapid Compression Machine
38
3.1 Introduction
38
3.2 Experimental
44
3.3 Experimental Results
46
3.3.1 Ignition Delay
46
3.3.2 Acetone Fluorescence
47
3.3.3 Flat Piston
49
3.3.3.1 Low Compressed Gas Pressure
49
3.3.3.2 High Compressed Gas Pressure
50
3.3.4 Creviced Piston
51
3.3.4.1 Low Compressed Gas Pressure
51
3.3.4.2 High Compressed Gas Pressure
52
3.4 Numerical Analysis
52
3.5 Computational Results
54
3.6 Assessment of Adiabatic Core Hypothesis
57
3.7 Effect of Piston Head Configuration on Ignition Delay
62
3.8 Summary
63
Chapter 4 Performance Characterization of Rapid Compression Machine
85
4.1 Introduction
85
4.2 Numerical Model
86
4.3 Results and Discussion
90
4.3.1 Characterization Experiments: Inert Mixtures
90
4.3.2 Characterization Experiments: Reactive Mixtures
91
v 4.3.3 Rapid Sampling Apparatus
94
4.3.4 Modeling Results
95
4.4 Summary
97
Chapter 5 Autoignition Investigations of H2/CO System
110
5.1 Importance of Hydrogen-Carbon Monoxide Oxidation
110
5.2 Prior Work on Hydrogen-Carbon Monoxide Oxidation
112
5.2.1 H2/O2 System
113
5.2.2 H2/CO/O2 System
115
5.3 Experimental Specifications
119
5.4 Kinetic Models
120
5.5 Results and Discussion
121
5.5.1 Comparison with Previous Studies
122
5.5.2 Autoignition of Hydrogen
123
5.5.3 Autoignition of Hydrogen/Carbon Monoxide Mixtures
124
5.5.4 Refinement of Rate Constants
130
5.6 Summary
135
Chapter 6 Summary and Future Work
150
6.1 Summary
150
6.2 Future Work
153
Appendix A PLIF of Acetone
156
Appendix B Piston Velocity Calculation
165
References
168
vi List of Tables
Table 2.1 Summary of experimental conditions and features of various RCMs reported in the literature.
24
Table 4.1 – Coefficient for vp(t) for curve fits shown in Fig. 4.11.
107
Table 5.1 – Compositions of gas mixtures investigated.
120
Table 5.2 – Relevant portions of the Mechanism of Li et al. (2004b)
127
vii List of Figures
Fig. 1.1 - Typical pressure trace for compression of inert gas in an RCM. Test gas – nitrogen; Initial Conditions: P0 = 1000 torr; T0 = 297 K; Temperature at top dead center, Tc = 800 K.
16
Fig. 1.2 – Typical Pressure trace during ignition of a reactive mixture in an RCM. Molar composition: H2/O2/N2/Ar = 2/1/2.9/10.1. Initial Conditions: P0 = 705.8 torr; T0 = 298.1 K. Temperature at top dead center, Tc = 977 K.
17
Fig. 1.3 - Creation of roll-up vortex due to piston motion during the compression stroke.
18
Fig. 2.1 – Schematic of the present RCM. Enlarged views of the end reaction chamber and creviced piston are also shown.
34
Fig. 2.2 – Schematic of the rapid sampling apparatus
35
Fig. 2.3 - Computational grid distribution for a creviced piston at the end of compression. All length dimensions are in mm.
36
Fig. 2.4 - STAR-CD simulation of temperature (left panel) and velocity (right panel) fields at the end of compression for nitrogen gas. Conditions: stroke – 254 mm; compression ratio – 19.14 (without considering crevice volume); wall temperature – 297 K. (a) Flat piston without crevice. (b) Creviced piston with dimensions (in mm): A=0.50, B=4.0, C=0.15, D=8.0, and E=1.50. (c) Creviced piston with dimensions (in mm): A=0.50, B=4.0, C=0.15, D=20.0, and E=1.50. Successive isotherms are 21.4 K apart for (a), and 14.3 K apart for (b) and (c). Maximum gas speed: (a) 12.7 m/s, (b) 9.5 m/s, and (c) 18.3 m/s
37
viii
Fig. 3.1 –Creviced piston head configuration. Dimensions (in mm): A=0.50, B=4.0, C=0.15, D=20.0, and E=1.50.
65
Fig. 3.2 – PLIF Schematic. M1, M2 – UV mirrors for lowering the height of laser beam to RCM quartz window height. Ls – Spherical lens, Lc – Cylindrical lens.
66
Fig. 3.3 – Ignition delay versus adiabatic core temperature for stoichiometric isooctane/oxygen/inert mixtures. Composition: iC8H18/O2/inert=1/12.5/47. Initial conditions: P0=331 torr and T0=297 K. Adiabatic core temperature is varied by changing the composition of inert gases.
67
Fig. 3.4 – A representative single-shot PLIF image and the associated photon counts. Chamber conditions: pressure=44 bar and temperature=297 K. Laser sheet enters from right side of the image.
68
Fig. 3.5 – Deduction of temperature from fluorescence signal. Chamber conditions: pressure=44 bar and temperature=297 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity by using the BeerLambert law. Solid line: deduced temperature distribution.
69
Fig. 3.6 – PLIF intensities and the deduced temperature distributions at varying times after compression for a simulated flat piston head. Gas composition: 2% acetone
in
nitrogen.
Conditions
at
TDC:
pressure=12.5
bar
and
temperature=780 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
70
ix Fig. 3.7 – PLIF intensities and the deduced temperature distributions at varying times after compression for a simulated flat piston head. Gas composition: 1% acetone
in
nitrogen.
Conditions
at
TDC:
pressure=39.5
bar
and
temperature=815 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
71
Fig. 3.8 – PLIF intensities and the deduced temperature distributions at varying times after compression for a creviced piston head. Gas composition: 1% acetone
in
nitrogen.
Conditions
at
TDC:
pressure=11.95
bar
and
temperature=760 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
72
Fig. 3.9 – PLIF intensities and the deduced temperature distributions at varying times after compression for a creviced piston head. Gas composition: 1% acetone
in
nitrogen.
Conditions
at
TDC:
pressure=39.5
bar
and
temperature=770 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
73
Fig. 3.10 – STAR-CD simulation of temperature and velocity fields at varying times after compression for flat piston at a pressure of 12.8 bar at TDC under laminar condition.
74
x Fig. 3.11 – STAR-CD simulation of temperature and velocity fields at varying times after compression for flat piston at a pressure of 13.14 bar at TDC under turbulent condition.
75
Fig. 3.12 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 11.9 bar at TDC under laminar condition.
76
Fig. 3.13 – STAR-CD simulation of temperature field at varying times after compression for flat piston at a pressure of 40.7 bar at TDC under laminar condition.
77
Fig. 3.14 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 39.9 bar at TDC under laminar condition.
78
Fig. 3.15 – STAR-CD simulation of temperature field at varying times after compression for flat piston at a pressure of 20.5 bar at TDC under laminar condition.
79
Fig. 3.16 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 19.7 bar at TDC under laminar condition.
80
Fig. 3.17 – Assessment of adiabatic core hypothesis based on computational results. Pressure at TDC is indicated in the figure. Dashed line, solid line, and line with symbols respectively represent pressure, calculated temperature using adiabatic core hypothesis, and actual maximum temperature in the chamber.
81
xi Fig. 3.18 – Extent of core region, calculated from computational results, as a function of post-compression time. (a) Effect of pressure and piston head configuration. (b) Effect of turbulence. Pressure at TDC is indicated in the figure. Solid line: laminar calculation. Dashed line: turbulent calculation.
82
Fig. 3.19 – Time variation of maximum velocity in the cylinder from the end of compression for laminar calculation. Pressure at TDC is indicated in the figure.
83
Fig. 3.20 – Computational results on the effect of piston head configuration under the same conditions of temperature and pressure at TDC. Cases of flat and creviced piston head configurations with nitrogen as the test gas are shown.
83
Fig. 3.21 – Comparison of measured pressure profiles for iso-octane auto-ignition experiments using flat and creviced piston head configurations. Composition: iC8H18/O2/inert=1/12.5/47. Adiabatic core temperature at TDC is indicated in the figure. Solid line: flat piston. Dashed line: creviced piston.
84
Fig. 4.1 - Typical pressure traces for inert gas tests, demonstrating experimental repeatability. Subscripts: 0 - initial condition; c - condition at TDC; ac – condition at TDC for a truly adiabatic compression.
99
Fig. 4.2 - Demonstration of experimental repeatability for iso-octane auto-ignition. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/28.2/18.8. Conditions: P0 = 331 torr; T0 = 297 K; Tc = 745 K.
100
Fig. 4.3 - Demonstration of experimental repeatability for iso-octane auto-ignition under conditions of two-stage ignition. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/37.6/9.4. Conditions: P0 = 324.7 torr; T0 = 297 K; Tc = 713 K.
100
xii Fig. 4.4 - Pressure profiles at different compressed gas temperatures for iso-octane auto-ignition. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 379.7 torr; T0 = 297 K. Compressed gas temperature is varied by changing the composition of inert gases in the mixture.
101
Fig. 4.5 - Pressure profiles at different compressed gas temperatures for iso-octane auto-ignition. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Compressed gas temperature is varied by changing the composition of inert gases in the mixture.
102
Fig. 4.6 - Effect of pressure on auto-ignition of iso-octane mixtures. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/37.6/9.4. Initial temperature is 297 K. Pressure at TDC is varied by changing the initial pressure.
103
Fig. 4.7 - Ignition delay of iso-octane mixture verses adiabatic core temperature. Molar composition: iC8H18/O2/inert = 1/12.5/47. Adiabatic core temperature is varied by changing the composition of inert gases in the mixture.
103
Fig. 4.8 - Pressure traces during quenching experiments using a rapid sampling apparatus. Molar composition: CH4/O2/Ar = 1/2/24.2. Initial conditions: P0 = 320 torr; T0 = 296 K.
104
Fig. 4.9 - HID chromatogram of methane auto-ignition at two different quenching times using Hayesep DB column. Test conditions are the same as those in Fig. 4.8. Fig. 4.10 - Comparison of experimental pressure traces for pure Ar and pure N2 with model predictions. Initial conditions: argon – P0 = 600 torr; T0 = 297 K,
105
xiii nitrogen – P0 = 1000 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction.
106
Fig. 4.11 – Volume expansion trace, vexp(t), (solid line) and corresponding curvefitted polynomial function, vp(t), (dotted line) for N2 and Ar determined by using the pressure traces of Fig. 4.10.
107
Fig. 4.12 - Comparison of experimental pressure profile for stoichiometric isooctane mixture with numerical prediction. Molar composition: iC8H18/O2/N2 = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction. Comparison for the corresponding non-reactive mixture is also shown.
108
Fig. 4.13 - Comparison of experimental pressure profile for stoichiometric isooctane mixture with numerical prediction. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction.
108
Fig. 4.14 - Comparison of experimental and computed ignition delays with varying adiabatic core temperatures. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K.
109
Fig. 4.15 - Comparison of experimental pressure profile for stoichiometric hydrogen mixture with numerical prediction. Molar composition: H2/O2/N2/Ar = 2/1/2.9/10.1. Pressure traces for the corresponding non-reactive cases are also shown.
109
Fig. 5.1 - Explosion characteristics of homogeneous hydrogen/oxygen mixtures.
137
Fig. 5.2 - Explosion characteristics of homogeneous wet CO mixtures.
137
xiv Fig. 5.3 – Typical pressure trace and experimental repeatability for ignition of H2/CO
mixture.
Molar
composition:
H2/CO/O2/N2/Ar
=
9.375/3.125/6.25/18.125/63.125. Initial conditions – T0 = 298.7 K, P0 = 661 torr. Conditions at TDC – TC= 999 K, PC = 30 bar.
138
Fig. 5.4 – Comparison of measured ignition delay times on the present RCM with previous measurements.
139
Fig. 5.5 – Measured (filled symbol) and calculated (open symbol) ignition delays for
H2
autoignition.
Molar
composition:
H2/O2/N2/Ar
=
12.5/6.25/18.125/63.125. Pressure at TDC, PC = 30 bar.
140
Fig. 5.6 – Measured (filled symbol) and calculated (open symbol) ignition delays for H2 autoignition at PC of 15 and 50 bar. Molar composition: H2/O2/N2/Ar = 12.5/6.25/18.125/63.125.
140
Fig. 5.7 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 50 bar.
141
Fig. 5.8 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC= 30 bar.
141
Fig. 5.9 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 15 bar.
142
Fig. 5.10 – Comparison of experimental effect of CO addition with model calculations.
Molar
composition:
(H2+CO)/O2/N2/Ar
12.5/6.25/18.125/63.125. Conditions at TDC: PC= 30 bar and TC = 1010.5 K.
= 142
xv Fig. 5.11 – Comparison of experimental effect of CO addition with model calculations.
Molar
composition:
(H2+CO)/O2/N2/Ar
=
12.5/6.25/18.125/63.125. Condition at TDC: PC= 15 bar and TC = 1028.5 K.
143
Fig. 5.12 – Comparison of inhibition effect of CO addition with model calculations. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Condition at TDC, PC = 50 bar, TC = 1044 K.
143
Fig. 5.13 – Effect of equivalence ratio on ignition delay. Composition: Mixtures #6 to #10 in Table 5.1. Pressure at TDC, PC = 50 bar.
144
Fig. 5.14 – Effect of Equivalence ratio on ignition delay. Composition: Mixtures #6 to #10 in Table 5.1. Conditions at TDC, PC = 50 bar and TC = 990 K (top group), 1009 K (bottom group).
144
Fig. 5.15 – Sensitivity analysis of ignition delay at different CO additions. Conditions at TDC – PC = 30 bar and TC = 1010.5 K.
145
Fig. 5.16 – Specie concentration for adiabatic, constant volume calculations at different pressures. Molar composition: H2/O2/N2/Ar = 12.5/6.25/18.125/63.125. Initial temperature = 1010.5 K. Initial pressure: (a) 1 bar, (b) 5 bar, (c) 30 bar, (d) 50 bar
146
Fig. 5.17 – Sensitivity analysis of ignition delay at different pressures for RCO = 0.50. Conditions at TDC – TC= 1010 K.
147
Fig. 5.18 – Sensitivity analysis of ignition delay at different equivalence ratios. Conditions at TDC – PC = 50 bar and TC = 1009 K.
147
xvi Fig. 5.19 – Comparison of rate constants for k28. k28 for the measurements Baldwin et al. (1965, 1970, 1972) is calculated by using (k18+k19) from Hippler et al. (1990).
148
Fig. 5.20 – Comparison of measured and calculated ignition delays for different values of k28 while using the mechanism of Li et al. (2004b). Calculations are conducted by multiplying k28 from Li et al. (2004b) by a constant factor.
149
Fig. A.1 – Simplified model of acetone’s photophysical behavior with multistep decay of the excited singlet. Source: Thurber et al. (1997, 1998).
159
Fig. A.2 - Fluorescence per unit laser energy per unit mole fraction at atmospheric pressure, normalized to the room-temperature value. At constant pressure and seeding fraction, temperature can be inferred from a fluorescence measurement using this plot. For clarity, only five excitation wavelengths are displayed. Source: Thurber et al. (1997, 1998). Fig. A.3 - Absorption coefficient as a function of temperature.
160 161
Fig. A.4 – Fluorescence per unit laser energy per unit mole fraction normalized to the room temperature.
161
Fig. A5- Temperature as a function of fluorescence per unit laser energy per unit mole fraction normalized to the room temperature.
162
Fig. B.1 – Experimental pressure trace for compression of N2. Intial conditions – P0 = 1000 torr and T0 = 297 K.
166
Fig. B.2- Calculated combustion chamber volume as a function of time (solid line) and the fitted profile (dashed line).
167
Fig. B.3- Deduced piston velocity history.
167
xvii ACKNOWLEDGEMENTS I would like to express my gratitude to Dr. Sung for giving me a very valuable opportunity and learning environment, while working on my thesis under his guidance. Throughout my stay at Case Western, I felt very motivated and peaceful in working under him. He provided a very healthy environment of helping individuals in the Combustion Diagnostics Lab. His always insisting on good quality of work and thought provoking discussions helped me gain better insights in my research. I am very grateful to Dr. P.V. Krishnan at IIT Delhi for encouraging me to pursue Ph.D. at Case Western. His example as a responsible Professor will always remain an inspiring factor in my life. I am very thankful to David Conger in helping me a lot in the design of my experimental set up. Sam Cordaro, Wayne Schmidt and John Weber fabricated the machine. I also want to thank all my labmates, Dr. A. Ibarreta, Dr. Y. Huang, Dr. B. Han, J. Nanduri, K. Kumar, S. Volchko, and C. Mento, for their encouraging support. I also thank Prof. Prahl, Prof. Kamotani, and Prof. Feke for serving as my examination committee members and giving their valuable time. I would like to thank O. Reed, C. Wilson, A. Szakacs, J. A. Stiggers, and M. Marietta for their help whenever required. I am especially thankful to my friends, P. Raju, A. Kumar, A. Garg, and R. Varadarajan for their emotional support and providing me a homely atmosphere. Finally, I am grateful for the support of my parents and family members, whose well wishing must have certainly made a lot of difference.
xviii A Rapid Compression Machine – Design, Characterization, and Autoignition Investigations
Abstract By
GAURAV MITTAL A rapid compression machine (RCM) has been designed for the purpose of chemical kinetics studies at elevated pressures and temperatures. The present RCM is designed as a versatile tool and includes the features of a well-defined core region, fast compression, ability to vary stroke and clearance, optical accessibility, and capability for specie measurement. The machine is pneumatically driven and hydraulically actuated and stopped. Characterization experiments establish the suitability of the machine for chemical kinetic studies and show that highly repeatable experimental conditions up to 50 bar and greater than 1000 K can be obtained. A
numerical
model
accounting
for
compression and heat loss is also developed to simulate the RCM experiments. Experimental and computational investigations of aerodynamics inside the machine by using planar laser induced fluorescence of acetone and Star-CD CFD package substantiate the importance of piston head design for achieving a homogeneous core region inside a rapid compression machine. Results show that the flat piston head design leads to significant mixing of cold vortex with hot core region, eventually leading to the failure of adiabatic core assumption. Whereas, creviced piston head configuration is demonstrated to result in drastic reduction of the effect of vortex, and adiabatic core assumption is found to be valid for long time after compression.
xix Using this facility, autoignition investigations are conducted for iso-octane, H2, and H2/CO system. Iso-octane autoignition is investigated at pressures up to 22 bar, and temperature from 680 K to 880 K. Whereas H2, and H2/CO systems are investigated at pressures from 15 to 50 bar, and temperatures from 950 K to 1100 K. Comparisons of experimental results with numerical predictions of detailed mechanisms show that existing mechanisms fail to predict the behavior of these systems. Particularly, for isooctane ignition significant differences in simulated and experimental results are noted for experimental conditions in the NTC regime. For H2/CO system, the existing mechanisms fail to describe the inhibition effect of CO addition on H2. Kinetic analysis is shown to further identify the controlling reaction steps, which require modification of rate constant. Further investigation of these systems over a wide range of physical conditions is warranted.
1
CHAPTER 1 INTRODUCTION 1.1 Chemical Kinetics Chemical reactions converting fuel to combustion products are complex and depend in subtle ways on the conditions under which combustion takes place. Detailed study of rates and mechanisms of combustion reactions has been in the mainstream of combustion research after the recognition that tackling of practical combustion problems requires detailed understanding of combustion kinetics. For instance, investigation of formation and destruction of combustion generated pollutants (e.g. SOx, NOx, soot etc.), relation of autoignition chemistry to knock, development of fuels and fuel additives for engines, combustion safety etc., require understanding of detailed combustion chemistry. Further, with the availability of large amount of elementary kinetic data and increased computational capability, combustion modeling considering the effects of finite rate chemistry has become an important tool in the analysis of combustion systems. Such modeling leads to better understanding and improvement of combustion systems. At the most fundamental level, combustion modeling involves reaction mechanism consisting of large number of elementary reactions, generally known as comprehensive or detailed kinetic mechanism. A comprehensive kinetic mechanism is a reaction mechanism which has been tested and validated with experimental data over a wide range of physical conditions. The purpose of a comprehensive mechanism is to have a generalized reaction mechanism which can reproduce all of the available experimental data and furthermore predict with some confidence the kinetics of systems for which
2
experimental data is not available. This latter condition is important when kinetics to be predicted falls outside the boundaries for which mechanism has been validated. The development, verification, and further refinement of a reaction mechanism is largely dependent on the continued collection and evaluation of experimental data. Reaction mechanisms need to be validated with experimental results of high fidelity over a wide range of temperature and pressure, before they can be used for predictions. Chemical kinetic data which extend beyond the limits of the model validation are especially valuable and interesting for kinetic modeling. Successful matching of such experimental data with modeling results is a good indication that the chemistry of a reaction mechanism is fundamentally understood. Whereas any mismatch inspires further refinement of mechanism by increasing the constraints defining the acceptability of predictive comparisons. Determination of kinetic data over the range of pressure and temperature relevant to combustion systems requires the use of several experimental approaches. Consequently, there is considerable interest in having experimental facilities which can simulate the combustion environment relevant to the operating conditions of practical combustion devices and provide insights into the development of detailed chemical kinetics. Experimental data itself is important from both practical and mechanistic point of view. Although it is relevant to study combustion processes in practical combustion devices, it is extremely difficult to obtain data of sufficient quality to draw conclusions regarding the kinetic processes of importance. The difficulty is primarily caused by complicated fluid mechanics effects in practical devices. Complex interaction of chemistry with fluid mechanics makes it very difficult to interpret the data obtained from
3
practical combustion devices. A common practice is the use of experimental devices which allow study of homogeneous chemistry and where complicated fluid mechanical effects are negligible. The obtained experimental data provide important information for the development of detailed chemical kinetic mechanisms. These types of simpler experimental configurations include static reactors, flow reactors, well-stirred reactors, shock tubes, and rapid compression machines. Each type of configuration has comparative merits and drawbacks associated with accessible experimental ranges and characteristic times, complexity in data interpretation and modeling. In their entirety, the data obtained using these various experimental facilities can span large ranges of temperature, pressure, and measured observables. Before a detailed description of the rapid compression machine designed and used in the present study, brief discussion of various experimental facilities used for homogeneous combustion studies is provided in order to put its capabilities into a broader perspective.
1.2 Experimental Facilities for Studying Homogeneous Combustion 1.2.1 Static Reactors Static reactors are conceptually simple, but interpretation of the data acquired is very difficult. They consist typically of a spherical vessel that is filled with the reactive mixture. The gas phase reactants are, at least initially, maintained at the desired temperature by an oven or a thermostated bath. The progress of reaction is observed by measuring the change in pressure (reaction takes place at constant volume) or by detecting concentrations as a function of time for one or several specie from withdrawn,
4
quenched gas samples. A typical assumption associated with static reactors is that the reacting mixture is homogeneous, i.e. no temperature and/or concentration gradients are present. So both convective and diffusive effects are negligible. To strictly meet this assumption, the characteristic reaction times in static reactor experiments should be longer than characteristic diffusion times. This restriction imposes significant limitations on the operating range of static reactors (typically lower than 750 K) and results in comparatively long experimental timescales (on the order of seconds to minutes). Therefore static reactors are most useful for comparatively slow reactions. Static reactors have been extensively used to delineate the pressure-temperature boundaries for many fuel-oxidizer mixtures which separate the regions of slow and fast reaction. Such pressure-temperature boundaries are called explosion limits for the given mixture. For instance, explosion limits of H2-O2 system are generally determined using a spherical static reactor. However, due to high sensitivity to surface effects under conditions of long reaction times, static reactor experiments are generally not as useful for quantitative analysis as alternate techniques.
1.2.2 Flow Reactors A flow reactor is a continuous flow device, which brings reactant mixture to the desired temperature and pressure and then allows it to convect down a tube. The reactor design may involve heating of premixed gases or separate heating of reactants. In flow reactors, designed for chemical kinetic interpretation, the objective is to achieve a plug flow situation, where composition and temperature are uniform over the cross section of reactor and axial diffusion is negligible. In order to achieve this, the mixture is highly
5
diluted with inert gas and the total flow rate is large. This stretches the reaction zone over a significant physical distance, and the axial concentration gradients and thermal gradients are small. Under these conditions diffusive effects can be neglected, which are negligible in comparison to the convection. The large physical distance of reaction zone also allows gas sampling of products as well as gas temperature measurement. Samples are extracted at different axial locations in the flow direction, with different residence times. With measurement or assumption of the gas velocity in a flow reactor, the sampling distance is correlated with experimental time and thus transforms the onedimensional, quasi-steady spatial problem to a zero-dimensional temporal problem. In this way, a flow reactor provides an experimental opportunity for direct and detailed measurement of chemical kinetics. Data from flow rector may be influenced by experimental uncertainties. Potential problems with this technique include imperfect mixing of reactants, radial gradients of temperature and concentration and problems associated with gas sampling technique. In essence, a flow reactor offers ease of specie and temperature measurement, but is suitable for relatively slow reaction conditions under highly diluted mixture. The variablepressure flow reactor at Princeton University can operate in the pressure range of 0.2 to 20 bar and temperature up to 1200 K (Vermeersch et al., 1991; Kim et al., 1994), which is a typical upper limit of operating conditions for flow reactor.
1.2.3 Well-Stirred Reactors The well-stirred, or perfectly stirred, reactor is that in which perfect mixing is achieved inside the reactor. It is a fixed volume reactor with inlet and outlet mixtures
6
entering and exiting, respectively. These systems usually operate at constant pressure and at steady state. Well-stirred reactors are usually designed based upon employing high velocity inlet-jet approaches. Because of high-intensity turbulent mixing, temperature and concentrations can ideally be assumed to be homogeneously distributed. The rapid mixing thus results in sample conditions that are purely kinetically controlled. Because of mixing, the mixture coming out from exhaust is at same conditions as the reactor interior mixture. The composition of exiting mixture can be measured as a function of residence time. Mean residence time in these reactors is usually defined as the ratio of mass inside the reactor to mass flow rate of reactants, which can be varied by varying mass flow rate. Reactants are followed from low conversion to high conversion by varying the residence time. Most important practical problem with stirred reactors is the achievement of sufficiently rapid mixing. An implicit assumption is that the time scale for mixing is much shorter than chemical reaction time of the fluid in the reactor. In experimental situations, departure from this behavior may exist. Particularly at high temperatures most combustion reactions are very fast and chemical time constants may be comparable to mixing time. Under these conditions experimental results are mixing influenced. As a rule of thumb, well-stirred reactor data obtained at temperatures higher than 1300 K to 1400 K should be used with caution. However, since there is no reaction zone to be spatially resolved (as with flow reactors), well-stirred reactors can operate with lower dilution and shorter residence times, and therefore permit study of reactions at higher temperatures than flow reactors. But a high degree of dilution is still used in order to reduce temperature gradients and heat release. Thus operation at steady state is possible.
7
Since these reactors operate under steady state condition, information regarding time evolution of chemical reaction is not readily available. As an example, investigation of n-heptane oxidation has been studied in a jet stirred reactor in the temperature range of 550 K to 1150 K and up to 40 bar of pressure, but mixture used was highly dilute (0.1% fuel) in order to reduce temperature gradients and heat release inside the reactor and maintain a steady state (Dagaut et al., 1995).
1.2.4 Shock Tubes Shock tubes are widely used to study autoignition characteristics of gas mixtures at higher temperatures and pressures than those achieved by flow reactors and well stirred reactors. In a shock tube, an inert driver gas at high pressure is separated by a thin diaphragm from a several meter long section containing the potential reactants. Reactants are usually diluted in inert gas and prepared at significantly lower pressure. Puncturing the diaphragm creates a shock wave which compresses the mixture of reactants to a desired temperature and pressure. Usually ignition delay time is measured behind the reflected shock waves. Ignition delay time is defined as the time interval between shock arrival, which is determined from a pressure trace, and the onset of combustion, which is usually inferred from either a pressure trace or the emission/absorption spectra of an intermediate combustion species (e.g., CO, OH). In shock tubes, pressure and temperature up to 550 bar and 2800 K have been obtained (e. g. Sivaramakrishnan et al., 2004). The high pressure shock tube at Stanford University is capable of attaining reflected-shock pressures as high as 1000 atm (Petersen et al., 1996) However, due to interference from boundary layer effects and/or reflected waves, observation times are
8
typically limited to less than 5 ms. Therefore, experimental conditions are constrained to pressure and temperature regimes where chemical induction times are very short.
1.2.5 Rapid Compression Machines Rapid Compression Machine (RCM) simulates a single compression event of an engine. It provides a simple method of simulating the process of adiabatic compression and ignition. Enclosed gases in the reaction chamber are compressed to high pressure and temperature in few milliseconds (typically 20 to 40 ms) and reaction is allowed to proceed in a constant volume, constant mass chamber. Typically, post compression pressure greater than 50 bar and temperature greater than 1000 K can be obtained. In order to avoid significant heat loss and also reactions from happening before reaching its end position, the piston is required to travel very fast. To achieve these high velocities, the piston is usually driven pneumatically. The compression ratio, initial pressure, temperature and composition of the mixture can be varied to control the pressure and temperature histories after compression. In a rapid compression machine, primary experimental data consists of the pressure trace of reacting mixtures. A typical pressure trace using N2 as a test gas, Fig. 1.1, shows rapid rise in pressure during the compression stroke followed by gradual decrease in pressure due to heat loss from a constant volume chamber at the end of compression. Further, for a reactive mixture, ignition is observed after certain ignition delay, as shown in Fig. 1.2, when suitable conditions of pressure and temperature exist. Ignition delay is usually defined as the time from end of compression stroke, when a peak
9
in pressure is observed, to the time of rapid rise in pressure due to ignition. Thus, a rapid compression machine gives direct measure of ignition delay. In the following section, advantages and challenges associated with rapid compression machines are discussed in detail.
1.3 Advantages and Challenges with Rapid Compression Machines A rapid compression machine is an excellent tool to study high pressure autoignition of combustible mixtures as it gives direct measure of ignition delay. Further, autoignition chemistry can be investigated by taking out samples from reacting mixture during the induction time and analyzing the specie concentration. Such measurements, although more difficult than the specie measurement in a flow reactor, are still possible. Advantage of RCM over shock tubes lies in its capability to provide longer time duration for experiment. Experimental duration in shock tube is limited to less than 5 ms due to interference from reflected waves. In an RCM, experimental duration of an order of magnitude larger than shock tube can be obtained, as there is no interference problem and experimental duration is limited only by heat loss to the cold wall. As a result, RCM gives accessibility to study combustion at elevated pressure under conditions for which reaction is too slow for shock tubes, which typically spans intermediate range of temperature from 600 K to 1100 K. Phenomenon of autoignition in the intermediate temperature region is of practical importance in a large range of applications. For example, it is the desired way of achieving combustion in Homogeneous Charge Compression Ignition (HCCI) engines, while it can produce undesirable ‘knock’ in the Spark Ignition (SI) engines (Hu and
10
Keck.,1987). Chemical studies in the intermediate temperature range, where phenomenon of two-stage ignition and negative temperature coefficient (NTC) occurs (e.g. Griffiths et al., 1993b), can provide a better understanding of hydrocarbon oxidation, with the ultimate aim of controlling the autoignition phenomenon. Control of autoignition by using additives is particularly relevant to HCCI engines, where combustion phasing and control of rate of heat release over a range of engine load and speed are critical challenges (Tanaka et al., 2003b). From safety point of view, situations can arise where unwanted auto-ignition may occur. For example, premixed fuel and oxidant may be present in the inlet manifolds of combustion chambers and, if the auto-ignition delays are shorter than the residence time, ignition can occur in the manifold rather than as desired in the combustion chamber. In particular, in lean premixed combustion systems operating at high pressure and temperature ranges, the phenomena of autoignition and flashback in the premixing ducts is proven to be a serious issue which is leading to questions regarding reliability of these advanced systems. Development of HCCI engines and LPP (Lean Prevaporized Premixed) gas turbines critically depend on the ability to understand, control, and accurately predict autoignition. From fundamental perspective, high pressure autoignition is important for the development of kinetic mechanisms for modeling combustion. Like other experimental facilities, RCM also has challenges associated with it. It has been observed that data obtained in different RCMs even under similar conditions can be quantitatively different (Minetti et al., 1996a; Griffiths et al., 1997; Silke et al., 2005; Wurmel and Simmie, 2005). This difference is attributed to complicated aerodynamic and heat loss effects in various RCMs. Although in principle RCM
11
simulates a single compression event and avoids the complicated aerodynamics that exists in an engine, complex aerodynamic features affecting the state of the reacting mixture can exist in the reaction chamber. Various studies (Daneshyar et al., 1973; Griffiths et al., 1993a; Lee and Hochgreb, 1998a; Clarkson et al., 2001) have shown that the motion of the piston during the compression stroke creates a roll-up vortex, which results in mixing of pockets of cold gas from the boundary layer with the hot gases in the core region. Figure 1.3 depicts the creation of the vortex due to piston motion. Such mixing leads to difficulty in accurate characterization of the state of reacting mixture. As a result of roll-up vortex, there is temperature inhomogenity in the reaction chamber and reaction will proceed to different extents at different locations. Further, since direct measurement of temperature in an RCM is not possible due to rapid nature of compression and ignition process (typically less than 100 ms), temperature is determined indirectly using experimental pressure trace. In deducing the temperature, a hypothesis called ‘adiabatic core hypothesis’ is used. It is assumed that the effect of heat loss to the wall during and after the compression stroke is limited to a thin boundary layer along the wall of reaction chamber and the majority of reaction chamber away from the wall remains unaffected by heat loss. The unaffected region of the chamber is called core region, in which uniformity of temperature field is assumed to exist. It means that although compression is not truly adiabatic, but the core region of chamber away from wall can be assumed to be compressed adiabatically. If the compression process is truly adiabatic, temperature and pressure at top dead center (TDC) would be given by the following relations
12 Tac
= ln (CR ) ∫ T0 γ − 1 T
Tac
1 dT
γ
dT
(1.1)
P
ac ∫ γ − 1 T = ln( P ) , T0 0
(1.2)
where Tac and Pac are temperature and pressure at TDC for truly adiabatic compression, CR is compression ratio, T0 and P0 are initial temperature and pressure before compression begins, and γ is the ratio of specific heat at constant pressure to specific heat at constant volume. But due to heat transfer from hot gases to the wall, compression is not truly adiabatic and realized pressure and temperature are lower than those achieved for adiabatic compression, i.e. Pac and Tac. Based on the adiabatic core hypothesis, temperature at TDC, Tc, is determined by using actual pressure at TDC, Pc, which is measured by pressure transducer. Therefore, according to the hypothesis, calculated temperature at TDC, Tc, is Tc
γ
dT
∫ γ −1 T
T0
= ln(
Pc ) P0
.
(1.3)
In the above relation, everything except Tc is a measured parameter; therefore, with known temperature dependence of γ, Tc can be determined by numerical integration. After the end of compression, if the assumption of adiabatic core is assumed to hold, temperature evolution can be determined from the measured pressure as well. The assumption of adiabatic core will hold if the effect of heat loss is indeed limited to the boundary layer. However, if the aerodynamic effects due to vortex roll-up and mixing become substantial, adiabatic core hypothesis will break down and there will be ambiguity in accurate determination of temperature of reacting mixture. From kinetic study point of view, accurate determination of temperature is of prime importance due to
13
exponential nature of dependence of reaction rate on temperature. Therefore, it is important that any designed RCM minimize the effect of roll-up vortex and establish the validity of adiabatic core hypothesis before experimental data can be obtained. Thus, existence of a thermodynamically well-defined homogeneous reactant mixture is the key in obtaining meaningful data from RCM.
1.4 Specific Objectives In view of the suitability of RCM for combustion kinetics studies at elevated pressures, we design and fabricate a RCM. The present RCM is designed as a versatile tool which integrates the key features of various existing RCMs and allows wide range of experimental conditions. Particular attention is given to elimination of the effect of rollup vortex and ensuring the homogeneity of the reaction chamber. Further, the RCM is designed with ability to vary stroke and clearance to get a wide range of experimental conditions and ability for carrying out various diagnostics by providing for features of optical accessibility and sampling apparatus for specie measurements. After the fabrication of RCM, characterization experiments are conducted to demonstrate the suitability of the designed machine for combustion related research. First, these experiments characterize the aerodynamics and temperature field inside the RCM and demonstrate the suitability of the designed reaction chamber configuration in unambiguously determining the thermodynamic state of reacting mixtures. Thereafter, experiments are conducted with inert gases and reactive mixtures to ascertain that the designed machine meets various design objectives and is capable of capturing essential
14
features of combustion. A numerical model, capable of simulating the performance of RCM is also developed. In the next phase of the research, we focus on detailed experimental investigations and modeling of oxidation of H2/CO system by using the designed RCM. Recognizing the gap in experimental investigations for H2/CO oxidation at high pressures, which will be discussed in Chapter 5, the present study aims at investigating the kinetics of H2/CO oxidation at high pressure relevant to practical combustion devices. The lack of knowledge of ignition at high pressures and moderate temperatures prevents a better understanding of physical and chemical processes in practical combustion devices. Therefore, further experimental investigations and verifications under conditions of elevated pressures are conducted. Experiments are conducted for a wide range of pressure, temperature, mixture composition and equivalence ratio. Experimental results are compared with model predictions using various mechanisms available in the literature. Comparison of the experimental data with modeling calculations will test the mechanism under the conditions that were not validated during its development and can further identify the important reaction steps for which accurate rate parameters are needed.
1.5 Structure of Dissertation Chapter 2 delineates the design and key features of the present RCM. Configuration of the designed RCM, computational investigations on the design of piston head and experimental procedure are discussed.
15
In Chapter 3, investigations on aerodynamics and resulting temperature field inside the designed RCM for various operating conditions and configurations of piston head is presented. Both, experimental results using planar laser induced fluorescence (PLIF) of acetone and numerical results of computational fluid dynamics (CFD) simulation are discussed. In Chapter 4, characterization experiments using inert gases and reactive mixtures are presented. These experiments demonstrate experimental reproducibility and the range of physical conditions attainable in the designed RCM. Utility of rapid sampling apparatus for quantitative analysis is also demonstrated. A numerical model, which accounts for the effect of heat loss during the compression and reaction period as well as accurately characterizes the thermodynamic state of reacting mixture, is also presented for simulating the experimental data with detailed chemistry. The numerical model is also shown to appropriately predict the performance of the machine. In Chapter 5, results of detailed investigations of H2/CO autoignition are presented. Experimental measurements are compared with several kinetic mechanisms. These comparisons are shown to highlight the deficiencies in the present understanding of H2/CO oxidation at high pressure and intermediate temperature. Refinements to existing kinetic mechanisms are also suggested. In Chapter 6, results are summarized and recommendations for further studies are discussed.
16
Pressure (bar)
Compression Stroke 50
Pressure drop due to heat loss
40 End of Compression
30
Nitrogen T0 = 297 K
20
P0= 1000 torr Tc = 800 K
10
Pc = 47.5 bar
0
-20
0
20 40 Time (ms)
60
80
100
Fig. 1.1 - Typical pressure trace for compression of inert gas in an RCM. Test gas – nitrogen; Initial Conditions: P0 = 1000 torr; T0 = 297 K; Temperature at top dead center, Tc = 800 K.
17
50
Pressure (bar)
40 30
P0= 705.8 torr T0 = 298.1 K Tc= 977 K
Ignition Delay
20 10 0 -20
0 Time (ms)
20
Fig. 1.2 – Typical Pressure trace during ignition of a reactive mixture in an RCM. Molar composition: H2/O2/N2/Ar = 2/1/2.9/10.1. Initial Conditions: P0 = 705.8 torr; T0 = 298.1 K. Temperature at top dead center, Tc = 977 K.
18
Piston Motion
Piston Head Piston Motion
Cylinder Wall
Piston Head
Shear Area
Entrained Vortex Area Area
Fig. 1.3 - Creation of roll-up vortex due to piston motion during the compression stroke.
19
CHAPTER 2 EXPERIMENTAL FACILITY 2.1 Introduction As previously discussed, in principle, RCM simulates a single compression event. Enclosed gases in the reaction chamber are compressed to high pressure and temperature in few milliseconds (typically 20 to 40 ms) and reaction is allowed to proceed in a constant volume, constant mass chamber at the end of compression stroke. Also, at the end of compression stroke, fast moving piston is required to stop almost instantaneously and get locked in the final position of top dead center (TDC). Both, overshoot and rebound at the time of stopping of piston, have to be avoided in order to get a constant volume chamber. These requirements are very severe because the stopping of the piston can generate large inertia forces. In turn, such forces can induce vibrations in the machine and deteriorate the quality of acquired data. Effect of vibration or change in volume of reaction chamber on acquired data is maximum at TDC because of small clearance at the end of compression stroke. Also, uncontrolled inertia forces may lead to mechanical damage. Considering the stringent requirements, RCMs have developed through many phases since the time of the earliest RCM. A review of previous RCMs is presented by Affleck and Thomas (1968). Earliest of these machines were driven by falling weights or flywheel-driven crank. But these machines were limited in terms of achievable compression ratio and speed of compression. With further developments, machines were designed in which piston was driven by compressed air. These machines were able to achieve high compression ratios and compression time of the order of 20 ms. Separation
20
of driving and compression piston allowed the operation of machine with low driving gas pressure because large thrust could still be obtained by using a driving piston of large diameter in comparison to the diameter of driven piston. In these RCMs, various mechanisms were used for stopping of piston at the end of compression stroke. These consisted of using powerful springs, air-cushion system, absorbing piston energy by plastic deformation of metal or transferring piston momentum to an auxiliary floating mass. These methods either required considerable amount of mechanical manipulation between runs of the machine or gave inferior performance. A hydraulic stopping mechanism, first used by Rogowski (1963), could overcome these advantages. In this mechanism, piston is decelerated and stopped by controlled venting of hydraulic fluid. After this development, various other RCMs (Beeley et al., 1980a; Park and Keck, 1990) have been successfully developed using this concept, which are still in operation. An alternative design is developed in the Lille laboratory (Minetti et al., 1994). This RCM uses the shape of cam for stopping of piston. A free piston based compression facility in University of Michigan uses the interference fit of a polymeric free piston in the bore of reaction chamber as the means of stopping the piston (Donovan et al., 2004). In any case, ability for fast compression and vibration-free stopping are the prime requirements of an RCM. In addition, existence of a thermodynamically well-define core region of compressed gas is essential. Features of optical accessibility, ability to vary stroke and clearance and ability to rapidly extract reacting mixture for specie analysis are important to carry out various diagnostic studies under a wide range of experimental conditions.
21
2.2 Review of Existing RCMs A summary of features of various RCMs and contributions of various research groups, who have actively used RCM for intermediate temperature and high pressure investigations, was recently reported (Donovan et al., 2004). Here, key features of various RCMs in operation are briefly discussed and the capabilities of the designed RCM are contrasted with existing machines. The rapid compression machine at National University of Ireland, Galaway, is perhaps the oldest machine which is currently being used for combustion studies. It uses a unique dual-opposed-piston configuration (Affleck and Thomas, 1968). A dual piston set up was chosen to reduce the compression time and provide a high degree of mechanical balance. Using this RCM, autoignition of methane (Brett et al., 2001), propane (Gallagher et al., 2003), and isomers of heptane (Silke et al., 2005) has been investigated. Compression time of machine is less than 20 ms and compression pressure and temperature of 40 bar and 1060 K have been reported. Machine is capable of withstanding post-ignition pressure up to 135 bar (Gallagher et al., 2003). Recently, Wurmel and Simmie (2005) used a CFD model to investigate the temperature field and optimize the piston head crevice design for this machine. Rapid compression machine at University of Leeds is pneumatically driven and hydraulically stopped (Beeley et al., 1980a). Typically, experiments using this machine have reported compression ratio of approximately 11, with compression time around 22 ms. Compressed gas temperature and pressure up to 1000 K and 20 bar are realized. Numerous autoignition studies have been conducted on this machine (Beeley et al., 1980a; Griffiths et al., 1993b; Griffiths et al., 1997; Griffiths et al., 2002a). RCM is
22
optically accessible which allowed study of temperature field using PLIF of acetone (Clarkson et al., 2001). Limited specie measurement studies have also been conducted using a rapid sampling apparatus (Beeley et al., 1980b). Rapid compression machine at Lille laboratory (University of Science and Technology) uses a right angle configuration with driving and compression piston connected by a cam (Minetti et al., 1994). Two pistons are stopped independently – the driving piston by a conventional shock absorber and the compression piston by the shape of cam at the end of stroke. Compression ratio of machine is 9.8 and compression time can be varied between 20 and 80 ms by changing the pressure in the driving cylinder. Compressed gas pressure and temperature up to 16.8 bar and 900 K are reported in literature. This machine has the facility of optical accessibility (Desgroux et al., 1995) and specie measurement. Primarily, oxidation and autoignition chemistry of several hydrocarbon fuels has been studied using rapid sampling apparatus (Minetti et al., 1994; Minetti et al., 1995; Minetti et al., 1996a; Minetti et al., 1996b; Minetti et al., 1999; Ribaucour et al., 2000). Rapid compression machine at MIT uses a pneumatically driven and hydraulically stopped mechanism (Park and Keck, 1990). Machine achieves compression ratio up to 19 when using a flat piston head and is designed with easily adjustable stroke and clearance. Compression time is 10 to 30 ms. Maximum post compression pressure and temperature of 70 bar and 1200 K can be attained. This machine incorporates a creviced piston optimized to suppress the effect of roll-up vortex (Lee and Hochgreb, 1998a). This machine has been used for autoignition studies of Primary Reference Fuels (Park and Keck, 1990) and hydrogen (Lee and Hochgreb, 1998a). Recently this machine was used
23
to study the effect of fuel structure and additives on combustion of hydrocarbon fuels (Tanaka et al., 2003a, 2003b). Recently, another rapid compression machine has been designed at MIT (Kitsopandis, 2004). For this RCM, compression stroke ranges from 8 to 10 inches, resulting in a compression ratio between 12.5 and 16.5. The machine is optically accessible and has been used to study the soot formation process under diesel-engine like conditions (Kitsopandis, 2004). At University of Michigan, a free-piston rapid compression machine has been developed. In this facility, a novel sabot design prevents the entrance of roll-up vortex in the main combustion chamber (Donovan et al., 2004). Fast moving sabot (free piston) is stopped at the end of compression stroke by interference fit with inside diameter of combustion chamber. As a result, nose portion of the piston gets deformed after few runs and needs to be replaced. Machine achieves a range of compression ratio from 16 to 37 and compression time is approximately 100 ms. Compressed gas pressure of more than 20 bar, and temperature of 1000 K for nitrogen and 2000 K for argon can be obtained. This facility is instrumented with optical access for laser diagnostics and sampling apparatus for specie measurements. Recently, this facility was used to study ignition and reaction kinetics of silane combustion (Donovan et al., 2005). Table 2.1 summarizes the range of experimental conditions attainable on various RCMs and their features. It can be seen in Table 2.1 that the present RCM is designed as a versatile tool which integrates the key features of various existing RCMs and allows a wide range of experimental conditions. Specifically, operational and diagnostic capabilities of present RCM include a well-defined core region, fast compression, ability
24
Affiliation
Driving, Stopping Mechanism
Compression Pmax (bar), Ratio, Tmax (K) Compression Time (ms) >50 bar, 21 (flat piston) >1100 K (post 15.1 (creviced compression); piston), 350 bar (post 25-40 ms ignition)
Key Features
Case Western Reserve University
Pneumatically driven, hydraulically stopped
University of Ireland, Galaway
Dual-opposed piston configuration. Pneumatically driven, hydraulically stopped Pneumatically driven, hydraulically stopped
40 bar, 1060 K (post compression); 135 bar (post ignition)
13.4, < 22 ms
20 bar, 1000 K (post compression)
<14.6, 22 ms
Optically accessible, Specie measurement, Adjustable stroke
Pneumatically driven, stopped by cam Pneumatically driven, hydraulically stopped
17 bar, 900 K (post compression)
9.8, 20-80 ms
Optically accessible, Specie measurement
70 bar, 1200 K (post compression)
19 (flat piston) 10-30 ms
Creviced piston, Adjustable stroke and clearance
MIT (2004)
Pneumatically driven, hydraulically stopped
40 bar, 900 K (post compression)
12.5-16.5, 15 ms
University of Michigan
Pneumatically driven, stopped by interference fit of sabot
20 bar, 1000 K (N2), 2000 K (Ar) (post compression)
16-37, 100 ms
Optically accessible, Adjustable stroke Detachable combustion chamber Sabot prevents vortex roll-up, Optically accessible, Specie measurement, Adjustable clearance
University of Leeds
University of Science and Technology MIT (1990)
Creviced piston, Optically accessible, Specie measurement, Adjustable stroke and clearance Creviced piston, Adjustable Stroke
Table 2.1 Summary of experimental conditions and features of various RCMs reported in the literature.
25
to vary stroke and clearance, optical accessibility, and capability for specie measurement using fast sampling apparatus. In the following section, design, key features, and operation of present RCM are described in detail.
2.3 Design and Operation of Rapid Compression Machine 2.3.1 Rapid Compression Machine Configuration The design of the present RCM is based on several other compression machines reported in the literature, such as those of Affleck and Thomas (1968), Beeley et al. (1980a), Beeley et al. (1980b), Park and Keck (1990), and Minetti et al., (1994). As shown in Fig. 2.1, the present RCM system consists of a driver piston, reactor piston, hydraulic motion control chamber, and a driving air tank. Driver cylinder and the reactor cylinder have a bore of 5 inch and 2 inch, respectively. This allows the pressure of driving air to be a factor of 6.25 smaller than the final desired pressure achieved by the reactor piston. Stroke can be varied between 7 and 10 inches, by adjusting the spacers on the hydraulic cylinder. Clearance is also adjustable and can be varied by using split shims between hydraulic cylinder head and reactor cylinder. Compression ratio of up to 21 (15.1) can be achieved when using a flat (creviced) piston head, which can be further increased by adding a metal disc on the top of reactor piston or an end plug. The reaction chamber is designed to withstand the high pressure generated during combustion and is equipped with the sensing devices for measuring pressure and temperature, gas inlet/outlet port for preparing reactive mixture, and quartz windows for optical access. Cross-sectional view of the end reaction chamber is also shown in Fig.
26
2.1. Special care has been taken in the design of gas inlet/outlet port to minimize dead volume. In order to achieve this, a Swagelok rising plug valve is mounted on the reactor cylinder. A metal plug with a hole diameter of 0.040 inch is welded inside the valve on the reactor side and feeding is accomplished through a channel of 0.040 inch diameter in the reactor cylinder. This setup of gas inlet/outlet port yields very low dead volume, which is approximately 0.25% of the compressed gas volume and less than 0.02% of initial gas volume. The end of the reaction chamber can be either a steel plate or a quartz window, and the latter allows optical accessibility to the entire reaction chamber. A rapid sampling apparatus (to be discussed in due course) for analyzing specie concentration can also be mounted when needed. The reaction cylinder is wrapped with a heating tape, which allows variation of initial charge temperature. Hydraulic chamber performs three distinct functions. First, it provides a way of holding the piston in position before machine is actuated. In this position, the rear portion of hydraulic piston seals against hydraulic seal and front of hydraulic piston is pressurized with oil. Second function of hydraulic piston is to control the speed of piston assembly during compression. During motion, oil ahead of hydraulic piston is released from the clearance between the piston and bore of hydraulic chamber. Venting through this clearance area controls the speed of piston. Third function of hydraulic system is to stop the piston at the end of compression. Stopping is achieved by a hydraulic ring and groove mechanism. As stopping ring on hydraulic piston enters the stopping groove, high pressure is generated by the oil trapped in the groove ahead of stopping ring. Precisely machined clearance between ring and the groove allows progressive venting of compressed oil. Stopping ring has 4 steps on it, which reduces the clearance for oil
27
venting as piston slows down. This arrangement gives almost uniform deceleration during the stopping distance, which is 0.630 inch. Since the moving piston assembly weighs around 7.5 kg, the machine is mounted on a heavy-duty table to minimize the vibrations during experiments. Driving piston, reactor piston, and its connecting rod with hydraulic piston are made from 7075-T6 aluminum, whereas the hydraulic piston and its connecting rod with driving piston are made from steel. For corrosion resistance, the interior and exterior of reactor cylinder and hydraulic chamber are chrome plated and black oxide finished, respectively. The entire machine is enclosed inside a polycarbonate cage for enhanced safety of operation. Dynamic pressure during compression is measured using a Kistler 6125B transducer with 5010B charge amplifier. Pressure is recorded using a National Instruments data acquisition card (PCI 6030E) with 16-bit resolution and 100,000 samples/second. Data acquisition card is controlled by Labview software from National Instruments using a driver written for our data acquisition system. The present design of rapid sampling apparatus for specie measurement is modified from the earlier designs of Roblee (1961), Beeley et al. (1980b), and Minetti et al. (1994). This apparatus is employed to quench the reaction at predetermined time and collect the reaction products for analysis. Quenching is achieved by puncturing an aluminum diaphragm by a needle held above the diaphragm. Once the diaphragm is ruptured, the quenched gaseous sample is dumped to an expansion tank of approximately 1.8 liter volume, resulting in a volume ratio of around 50 between the reaction chamber and expansion tank. The expansion tank is cylindrical in shape with a dished bottom, as shown in Fig. 2.2. Concaved configuration of the bottom ensures that shock waves
28
generated on bursting the diaphragm and reflected from the bottom of the expansion tank will be diverged and hence weakened (Roblee, 1961). The needle for puncturing the diaphragm is held in position by an electromagnet and driven by a compression spring. Initially the holding force of electromagnet is greater than the opposing force of spring. When the electromagnet is deactivated, the needle is driven forward by the action of spring and diaphragm is punctured. Instant of the electromagnet deactivation or the diaphragm puncturing is varied by using a time delay circuit. Triggering signal for deactivating electromagnet can be taken from either the releasing of solenoid valve for firing the machine or the output of pressure transducer. After the rupture of diaphragm, the extracted reaction intermediaries are analyzed using a SRI 8610C Gas Chromatography (GC). This GC is equipped with two automatic gas sampling valves, a flame ionization detector (FID), and a helium ionization detector (HID).
2.3.2 Optimization of Reactor Piston Crevice The key feature of an RCM is the ability to have a high pressure, high temperature, constant volume, homogeneous core of reaction mixture. As previously discussed, various studies (Daneshyar et al., 1973; Griffiths et al., 1993a; Lee and Hochgreb, 1998a; Clarkson et al., 2001) have shown the existence of a roll-up vortex due to piston motion, which leads to the mixing of pockets of cold gases from wall boundary layer with the hot gases in the core region. This resulting non-uniformity of the reacting core can cause serious discrepancies between data taken from different RCMs. Lee and Hochgreb (1998a) have theoretically demonstrated that this problem of vortex roll-up can be counteracted by deliberately machining a crevice on the piston. The location of the
29
crevice is on the cylindrical surface of the piston and the piston face is kept flat. As a result of the crevice along the cylindrical periphery of the piston, cold gases can flow into the crevice and their roll-up, as depicted in Fig. 1.3, may be prevented. In order to effectively suppress the vortex for the designed RCM, appropriate geometry of the crevice is identified with the aid of numerical simulation. In order to systematically examine the effect of crevice geometry on vortex suppression, STAR-CD CFD package is employed to facilitate the design of reactor piston head. Because of cylindrically symmetric geometry of the combustion chamber, numerical simulation is carried out on an axisymmetric grid distribution. In actual physical situation in the RCM, flow may deviate from axisymmetric behavior and exhibit asymmetrical patterns.
However, axisymmetric configuration is nevertheless chosen
because it considerably reduces computational time while capturing the essentials of the underlying fluid mechanics. Simulation is carried out for laminar conditions with a constant wall temperature and nitrogen as the test gas. Computations are performed from the beginning of compression stroke with a compression time of 30 ms. Initially, the test gas is at rest and specified with uniform temperature and pressure. In the simulation, piston starts from rest and its motion is given in a similar manner as the piston motion in an engine, by specifying dimensions of crank radius and connecting rod length. At the top dead center (TDC), piston comes to rest. A time step of 55.55 µs is taken as further reduction of time step by a factor of two results in identical solution. During the compression stroke, vertex filling methodology of StarCD (Star-CD Version 3.20 User Guide) is chosen for the motion of grid distribution. In
30
this method, while the same number of grids is maintained between cylinder head and piston head, the grids become compressed as piston moves. Computational grid distribution for the creviced piston configuration at the end of compression is shown in Fig. 2.3. The main reaction chamber consists of 100 grids in the radial direction and 140 grids in the axial direction. In the radial direction, finer grid is used near the wall boundary. In the axial direction, accordion distribution is used, with finer grids near the piston head and cylinder head and coarser grids in between. In order to ascertain the quality of the grid distribution employed, selective cases are calculated on a finer grid distribution of 150 × 200 in the radial and axial direction, respectively. The use of finer grids results in identical flow pattern and less than 0.25% change in predicted temperature. As a result, all calculations are performed on a 100 × 140 grid distribution. From Star-CD simulation, it is observed that a slowly converging crevice design as proposed by Lee and Hochgreb (1998a) is capable of suppressing the roll-up vortex. An enlarged view of a typical creviced piston head configuration is also shown in Fig 2.1. In Star-CD simulation, dimensions A to E are varied systematically to investigate their individual and combined effects on vortex suppression. The findings are summarized as follows. (1) The total volume of crevice should be large enough to completely contain the boundary layer; otherwise suppression of vortex is only partial. (2) Crevice entrance, dimension A, should be large enough to allow the boundary layer to flow into crevice. Dimension A of 0.50 mm is found sufficient and increasing it to 1.0 mm results in flow circulation within the crevice inlet.
31
(3) Increasing the length of the crevice inlet, dimension B, does not affect fluid mechanics in any manner, except that the gas in crevice is cooler due to increased heat loss from crevice entrance. (4) When dimension C is increased beyond a critical value, the undesirable back flow from crevice to the main cylinder volume begins to occur. In the simulation, no back flow is observed when dimension C is 0.25 mm, but it occurs at C=0.50 mm. Even a very small value of C (0.07 mm) is found sufficient to allow boundary layer to flow in without any back flow for the conditions tested. Figure 2.4 further demonstrates the computed results of temperature and velocity fields at the end of compression for different piston head configurations, including a flat piston (Fig. 2.4(a)) and a creviced piston (Figs. 2.4(b) and 2.4(c)). The term flat piston indicates a piston that does not have a crevice along the cylindrical periphery of the piston, and is flat on face and sides. The pressure at the end of compression is around 40 bar for all three cases. It is seen that at the end of compression, vortex fills a significant portion of core region for flat piston, whereas vortex is drastically suppressed by using the crevice design of appropriate configuration. Some temperature non-homogeneity on the face of the piston head, however, could not be eliminated completely even with the creviced design. In Fig. 2.4(b), a smaller vortex than that of Fig. 2.4(a) exists because the crevice volume is not sufficiently large. Finally, the crevice piston with the dimensions as given for Fig. 2.4(c), A=0.5 mm, B=4.0 mm, C=0.15 mm, D=20.0 mm, and E=1.5 mm, is shown to successfully contain the vortex even for a stroke of 10 inch, which is the largest stroke of the present RCM. As such, the configuration of Fig. 2.4(c) is selected for the final piston head design.
32
It is further noted that in Fig. 2.4(a) with flat piston, the maximum velocity in the reaction chamber is approximately 12.7 m/s. For Fig. 2.4(c), although there is relatively large motion within the crevice volume where the maximum gas velocity is 18.3 m/s, the gas velocity within the main cylinder volume is less than 0.4 m/s, except on the top of piston, where the gas velocity is slightly higher. Furthermore, the peak temperature for creviced piston is slightly lower simply because of the smaller effective compression ratio. These calculations of Fig. 2.4 suggest that an appropriate crevice configuration exists, which can capture the cold boundary layer and suppress vortex roll-up.
2.3.3 Experimental Procedure Test charge is prepared manometrically inside a 5.5-liter vacuum tank and is allowed to be homogenized. A large tank is employed so that many runs can be conducted with the same charge. A Baratron 626B (0-1000 torr) capacitance diaphragm manometer is used for measuring partial pressures. Feeding lines equipped with Swagelok VCR connections are found to be leak tight. Combined leak rate of all the fittings and reaction chamber is found to be less than 0.020 torr per minute. The piston assembly is retracted by pressurizing the front side of the driver piston with air. This retraction also seals the hydraulic piston against hydraulic seal. Then, the hydraulic chamber is pressurized with oil up to 1000 psi. Reaction chamber is evacuated and fed with premixture from the mixing tank. Subsequently, the rear side of the driver piston is pressurized by filling the 20-gallon air tank, followed by releasing air on the front side of driver piston.
33
After pressurizing the air tank, the maximum pressure of driving air is such that the thrust of air is less than the opposing thrust of oil on hydraulic piston, and hence the piston assembly stays in its position. When the hydraulic pressure is released by venting oil from the hydraulic chamber through a solenoid valve, the thrust at driving piston is no longer balanced and the whole piston assembly accelerates. Piston speed can be controlled by changing the driving gas pressure. Before releasing the solenoid valve, data acquisition system is actuated to record pressure traces. Furthermore, pressure and temperature of compressed mixture can be varied by changing initial pressure (P0), initial temperature (T0), and compression ratio or by altering the specific heat of reacting mixture. Specific heat is altered by varying the composition of inert gases (Ar, N2, and CO2) in the reacting mixture.
34
Rapid sampling apparatus
Reactor Cylinder end region Cylinder with quartz window, pressure transducer, thermocouple & gas line
Hydraulic piston seal Hydraulic Chamber
Piston stopping groove
Spacers for adjusting stroke
Driver Piston
Air line from tank Hydraulic line for filling, draining, and solenoid release Hydraulic Piston
Port for gas inlet/outlet valve
Ports for quartz windows
Pressure transducer
Thermocouple
End Reaction Chamber
Creviced Piston
Fig. 2.1 – Schematic of the present RCM. Enlarged views of the end reaction chamber and creviced piston are also shown.
35
Diaphragm
Electromagnet Compression spring
Expansion Tank
Fig. 2.2 – Schematic of the rapid sampling apparatus
Needle
Reaction Cylinder
36
Volume for piston seal Cylinder wall
Centerline
44
Crevice
Piston head
14 25
Fig. 2.3 - Computational grid distribution for a creviced piston at the end of compression. All length dimensions are in mm.
37
12
Pc=40.7 bar
Wall
Centerline
Tc=904.4 K
8
4
0
5
10 15 Radius (mm)
20
25
(a)
Piston Head
Wall
Centerline
Tc=842.5 K Pc=40 bar
(b)
Piston Head
Wall
Centerline
Tc=806.9 K Pc=39.9 bar
(c) Fig. 2.4 - STAR-CD simulation of temperature (left panel) and velocity (right panel) fields at the end of compression for nitrogen gas. Conditions: stroke – 254 mm; compression ratio – 19.14 (without considering crevice volume); wall temperature – 297 K. (a) Flat piston without crevice. (b) Creviced piston with dimensions (in mm): A=0.50, B=4.0, C=0.15, D=8.0, and E=1.50. (c) Creviced piston with dimensions (in mm): A=0.50, B=4.0, C=0.15, D=20.0, and E=1.50. Successive isotherms are 21.4 K apart for (a), and 14.3 K apart for (b) and (c). Maximum gas speed: (a) 12.7 m/s, (b) 9.5 m/s, and (c) 18.3 m/s
38
CHAPTER 3 AERODYNAMICS INSIDE THE RAPID COMPRESSION MACHINE 3.1
Introduction In an RCM, vigorous motion created by the piston along the wall typically creates
a vortex, which promotes mixing of colder mixture from walls with the hotter core region. When modeling the RCM experiments, for simplicity it is often assumed that the aerodynamic effects do not play any significant role at the short time scales encountered in the RCM. As such, temperature evolution is determined from the measured pressure profile by assuming the adiabatic core hypothesis. However, substantial discrepancies have been observed between data taken from different rapid compression machines even under similar conditions of temperature and pressure. These discrepancies are attributed partly to different heat loss characteristics after the end of compression stroke and partly to the difference in aerodynamics between various machines. Effect of aerodynamics is particularly more complicated because it does not show up in the pressure trace and it may lead to significant temperature gradients and ultimately to the failure of adiabatic core hypothesis. If aerodynamic effect becomes significant and adiabatic core hypothesis fails, there is no easy way to determine the temperature inside the reaction chamber. As a result, unambiguous determination of the state of the reacting mixture and systematic characterization of the resulting aerodynamic field inside RCM are important for obtaining reliable kinetic data from RCMs and bridging the gap between data taken from different machines. Several studies have contributed to the understanding of aerodynamics and temperature field inside an RCM, by computational fluid dynamics (CFD) calculations
39
and experimental measurements. Griffith et al. (1993a) numerically showed that the hot core region generated at the end of compression is virtually adiabatic and spans approximately 70% of the volume of combustion chamber at the end of compression. Griffith et al. (1993a) also observed differences between computational results using spatially uniform conditions and CFD simulation.
Differences between two sets of
computations were attributed to the effect of temperature gradient that was accounted for in CFD analysis (Griffith et al., 1993a). In the recent study of Clarkson et al. (2001), temperature field was imaged by Rayleigh scattering and laser induced fluorescence (LIF) of acetone. Acetone-LIF was found to nicely characterize temperature variations in the RCM whereas Rayleigh scattering was relatively less sensitive (Clarkson et al., 2001). It was experimentally observed that the roll-up vortex had penetrated the center of the combustion chamber at the end of compression (Clarkson et al., 2001).
The
temperature difference between the hot gases and roll-up vortex was estimated to be 50 K.
Furthermore, LIF of acetone generated by decomposition of di-t-butyl peroxide
unambiguously showed the temperature stratification at the center of the reaction chamber (Clarkson et al., 2001). Griffiths et al. (2001) investigated temperature and concentration fields in a rapid compression machine by using a number of experimental techniques, including schlieren imaging, PLIF of acetone, PLIF of formaldehyde, and chemiluminescence. In order to illustrate the interaction of chemistry with the temperature field in the RCM, Griffiths et al. (2001) contrasted the behavior of a di-t-butyl peroxide decomposition, which exhibits an overall positive temperature dependence of reaction rate, with n-pentane, which exhibits an overall negative temperature dependence. Imaging was limited to 10 ms after
40
the end of compression. Results showed that for di-t-butyl peroxide decomposition, reaction proceeded faster in the zone of peak temperature. A somewhat similar behavior was observed for n-pentane combustion when temperature was at the lower end of negative temperature dependence range. By contrast, at temperatures close to the upper end of the negative temperature dependence region, reaction in the cooler zone developed faster and the temperature inhomogeneity inside the reaction chamber rapidly smoothed out. However, the authors discussed that spatial inhomogeneity of concentrations intermediates can be there, which would affect the eventual evolution of ignition (Griffiths et al., 2001). Griffiths et al. (2002b) used chemiluminescence imaging along with filtered Rayligh scattering to characterize the transition from non-knocking to knocking reaction and the evolution of spatial development of the reaction. Imaging was done during the final event of ignition after the ignition delay period. Results from filtered Rayleigh scattering gave evidence of cooler core region at temperature below the negative temperature dependence region. In contrast, Rayleigh scattering did not show a cooler core at temperature within the region of negative temperature dependence. This behavior was attributed to the smoothing effect on temperature in the region of negative temperature dependence. Lee and Hochgreb (1998a) theoretically demonstrated that roll-up vortex due to piston motion can be counteracted by deliberately machining a crevice on the piston head. Creviced piston allowed more accurate predictions of the reacting temperature from pressure-time records. Predictions of a simple heat transfer model in conjunction
41
with a creviced piston were found to agree well with experimental pressure history (1998). Desgroux et al. (1995, 1996) made direct measurements of temperature in an RCM using thermocouple and single point Rayleigh scattering.
These measurements
confirmed the existence of adiabatic core gas at the end of compression. Rayleigh scattering measurements were made with temperature accuracy of 3-4%, but after the end of compression the local temperature exhibited a standard deviation of 10-15% (Desgroux et al., 1995). Desgroux et al. (1996) conducted fine wire thermocouple measurements for unreactive and reactive mixtures at different radial locations. They observed uniform temperature field for few milliseconds after the end of compression and subsequent development of temperature inhomogenity due to heat loss to the wall and gas recirculation.
For unreactive mixture temperature inhomogenity persisted after
compression, whereas for reactive iso-octane mixtures a temperature leveling effect, ascribed to negative temperature coefficient of reaction rate, was observed (Desgroux et al., 1996). Donovan et al. (2004) took direct thermocouple measurement in a free-piston rapid compression facility for the assessment of adiabatic core hypothesis. Measurements were corrected for slower response of thermocouple. Donovan et al. (2004) asserted that experimental measurements at the end of compression stroke justify the core gas assumption. However, thermocouple measurements failed to give true representation of temperature after the end of compression because of the failure of the model used for correcting time response of the thermocouple (Donovan et al., 2004).
42
Recently, Wurmel and Simmie (2005) conducted CFD studies of a twin-piston rapid compression machine using Star-CD. From CFD simulations, it was found that the setup was not sensitive to a nonsynchronized or nonuniform piston release of the twinpiston machine, provided the compression ratio was kept constant. Piston head crevices were optimized for the twin-piston RCM, in terms of volume, location, and the dimension and geometry of the channel connecting crevice and chamber. The channel connecting the crevice and chamber was optimized and the ideal geometry was found to be rectangular. An angled channel design was seen to help somewhat in the cooling of the trapped gas in the piston head crevice. Authors also observed strong dependence of the crevice performance on the test gas used. While they were able to define an optimal crevice for nitrogen, oxygen, and argon, this was not possible for helium, due to high heat loss associated with helium. Simulated profiles of pressure were compared to experimental data for a creviced piston head, for a variety of pure unreactive gases. Simulated data of pressure using creviced piston was shown to agree well with experimentally obtained profiles for nitrogen and oxygen. However, substantial difference was observed for argon and helium. Instead of helium, use of xenon as a bath gas for RCM experiments was recommended (Wurmel and Simmie, 2005). The present investigation aims to extend previous efforts by experimentally and numerically studying the effect of the piston head geometry on aerodynamics and temperature field inside an RCM. It has implications for kinetic modeling in the sense that how well the prediction of adiabatic core hypothesis matches with actual temperature in the reaction chamber for a long time duration after the end of compression. Requirement of experimental data, particularly under the conditions of homogeneous
43
charge compression ignition (HCCI) operation, requires the ability to sustain unambiguous reaction conditions for longer times, around 100 ms. This is so because the relevant HCCI conditions are extremely lean or highly diluted, which significantly increases the ignition delay times. Therefore, it is particularly important to characterize the state of aerodynamics and the resultant temperature field for long time duration after the end of compression stroke. In the present work, temperature field inside an RCM is studied using planar laser induced fluorescence (PLIF) of acetone. Laser diagnostic technique in an RCM offers many advantages in comparison to thermocouple measurements.
Thermocouple
measurements suffers from the drawback that one cannot readily obtain information about instantaneous spatial variation of temperature.
Moreover, there will be
aerodynamic disturbances if the thermocouple probe is not sufficiently fine (Desgroux et al., 1996).
Additionally, correction of the measured temperature due to slow time
response of the thermocouple requires the knowledge of velocity field inside the chamber. Validity of the model used for determining velocity field and correcting for time response may be under question. By resorting to longer compression time, as conducted by Desgroux et al. (1996), while the effect of thermal inertia of the thermocouple can be minimized, aerodynamics at long compression time may not be true representation of what happens at short compression times. In contrast, temperature mapping using laser techniques is practically non-intrusive and is capable of giving instantaneous spatially-resolved temperature field. Both a flat head piston and a creviced head piston are employed and compared in the present study.
These experiments will demonstrate whether the uniformity of
44
temperature field is indeed improved by using a creviced head piston, as suggested by the earlier theoretical study by Lee and Hochgreb (1998a). The present investigation is particularly useful because experiments are conducted on the same RCM. Mapping of temperature field is carried out for two different piston head configurations under varying operating conditions for long time duration after compression. In order to investigate the effect of piston geometry on ignition delay, experiments are also conducted for autoignition of iso-octane using both piston head configurations. In addition, numerical simulations are carried out using Star-CD CFD package.
Computed results further
provide insights into the nature of aerodynamics inside a RCM.
Based on these
experimental and computational studies, conditions under which adiabatic core hypothesis can be satisfactorily applied are delineated. In the following sections the experimental setups for ignition delay measurements and PLIF of acetone are first described, followed by experimental and numerical results on characterization of aerodynamics inside the RCM. 3.2
Experimental In order to study the effect of piston head design on aerodynamics in RCM,
experiments are conducted for two different piston head configurations: a creviced piston head and a simulated flat piston head. An enlarged view of the configuration of the creviced piston head is shown in Fig. 3.1. Flat piston head is simulated by filling the crevice volume with a pack of o-rings.
This filling results in approximately 90%
reduction in the crevice volume and gives an almost flat piston. Two sets of experiments are conducted and compared using both piston head configurations.
First, ignition delays for stoichiometric iso-octane/oxygen/inert gas
45
mixtures are measured in the temperature range of 684 K to 878 K, from which the effect of piston head configuration on auto-ignition is demonstrated. Second, PLIF imaging of acetone for temperature field mapping is carried out under conditions of relatively low pressure (approximately 12 bar) and high pressure (approximately 39.5 bar). These experiments also provide insights into the effect of pressure on the resulting temperature field inside the rapid compression machine. Schematic of PLIF set up is shown in Fig. 3.2. For fluorescence measurements, mixture consists of 1 to 2% (by concentration) acetone and remaining nitrogen. An approximately 5 mm wide laser sheet with a waist of 50 µm is made to traverse the central plane of combustion chamber. Combustion chamber is equipped with two 0.66inch diameter optically flat quartz windows for the traversal of laser sheet. End of the chamber is fitted with a 2-inch diameter and 1.7-inch thick quartz window, which provides full view of the combustion chamber for fluorescence imaging.
WG305
(Schott) filter is used to filter scattered light from fluorescence signal. A frequency doubled continuum Nd:YAG system is used in conjunction with a dye laser filled with Rhodamine 590 soultion. The resulting laser energy at wavelength of 279 nm is around 8 mJ/pulse. Synchronization of the laser firing with machine is achieved in the following manner. Laser is continuously fired at a repetition rate of 10 Hz. A triggering signal from the laser control unit is used to trigger the camera control module, which synchronizes image acquisition by ICCD camera with the laser pulse. Simultaneously, the camera control module also gives a triggering signal, which is used to actuate a relay after a specified time delay.
Time delay is specified through DG-535 timing generator.
46
Actuation of relay fires the RCM. A subsequent laser pulse, which occurs after the end of compression, provides the pumping source for the induced fluorescence. Timings of laser pulses and pressure trace of RCM are simultaneously recorded using a data acquisition system, which allows accurate determination of the timing of laser pulse relative to the end of compression. After every experiment, laser optics is realigned to correct for any minor movement of machine as a result of firing. Since the repetition rate of laser system is 10 Hz, PLIF images are single shot measurements. PLIF images are acquired at different post compression times simply by changing the time delay in DG535.
3.3
Experimental Results
3.3.1 Ignition Delay Experiments are generally conducted for a stroke of 10 inch and clearance of 0.525 inch. Time for compression stroke is approximately 30 ms. Figure 3.3 shows a plot of ignition
delay
versus
adiabatic
core
temperature
for
stoichiometric
iso-
octane/oxygen/inert gas mixtures, measured in the present RCM.
Experiments are
conducted using a flat head piston and a creviced head piston.
Adiabatic core
temperature is varied by changing the composition of inert gases (argon and nitrogen), while keeping a fixed compression ratio. Results of the present work are also compared with the data from Minetti et al. (1996b) under similar conditions of pressure and composition. For the creviced piston, initial pressure (P0) and initial temperature (T0) are kept fixed at 331 torr and 297 K, respectively. For the flat head piston, clearance is increased to compensate for the absence of crevice so that the conditions at top dead
47
center (TDC) are identical to that for the creviced piston when same initial conditions of pressure, temperature, and composition are used.
This results in adiabatic core
temperature ranging from 684 K to 878 K and pressure at TDC varying from 13.3 bar to 16.25 bar. In spite of similar conditions, Fig. 3.3 demonstrates substantial discrepancies among different sets of experimental data. This comparison shows that not only data from different RCMs under similar test conditions may be different, but also for the same RCM there may be significant differences depending upon the configuration of piston head and final reacting volume. Furthermore, it is seen from Fig. 3.3 that when a flat piston head is used, ignition delays are significantly reduced as compared to those obtained using creviced piston head. It is therefore expected that changing from creviced piston head to flat head significantly affects the heat loss and the resulting aerodynamics inside the RCM, and the effect reflects in the form of considerable change in ignition delay. This example is presented here to highlight the importance of aerodynamic effects on ignition delay.
3.3.2 Acetone Fluorescence Acetone fluorescence signal can be highly sensitive to temperature. At 279 nm excitation, fluorescence signal reduces by 42% as temperature increases from 600 K to 800 K. Figure 3.4 shows a representative single-shot PLIF signal of a homogeneous acetone/nitrogen mixture at room temperature of 297 K, along with the associated photon count integrated along the width of the laser sheet. Laser sheet enters from right side of the image. Location of cylinder wall is also shown in the figure. Decay in the
48
fluorescence signal from right to left is because of strong absorption. According to the Beer-Lambert law, in a homogeneous field with uniform temperature and concentration, signal decays exponentially due to absorption of the laser intensity. By correcting for absorption and incorporating for temperature sensitivity of fluorescence signal, it is possible to determine the resultant temperature field from fluorescence signal. Parameters for temperature sensitivity of acetone fluorescence are taken from Thurber et al. (1997, 1998). Using the fluorescence intensity shown in Fig. 3.4, Fig. 3.5 demonstrates the procedure for deducing the temperature distribution. Detailed procedure for deducing temperature from fluorescence signal along with curve fittings used to obtain parameters for laser absorption and temperature sensitivity of fluorescence are documented in Appendix A. In Fig. 3.5, radial direction equal to 0 corresponds to central axis of the cylindrical chamber, while 2.54 cm and -2.54 cm represent cylinder wall on either side. Moreover, dotted line is the actual fluorescence signal, while dashed line represents the exponential decay in fluorescence intensity, as calculated by using the Beer-Lambert law and the known temperature of 297 K. In addition, the deduced temperature distribution is denoted by solid line. It is seen from Fig. 3.5 that in a uniform temperature field, signal follows the exponential decay feature and the deduced temperature nicely predicts the temperature distribution. In Fig. 3.5, the deduced temperature is noted to be within ± 5 K, which is uncertainty in the deduced temperature due to noise in the fluorescence signal. When there exists temperature non-uniformity in the chamber, knowledge of temperature at one anchor point is required to deduce temperature from fluorescence signal. Relative to this anchor point, temperature along the radial direction can be
49
subsequently deduced by knowing the temperature dependence of fluorescence intensity. However, such an anchor temperature is difficult to be measured during the actual RCM experiment. In the present work, the maximum temperature in the chamber at any instant is taken as the adiabatic core temperature, calculated from the pressure trace. It has to be pointed out that this assumption does not affect the pattern of the deduced temperature in any way. If the actual maximum temperature is somewhat lower than the adiabatic core temperature, the entire temperature profile would shift downward, without affecting the pattern of temperature distribution. 3.3.3 Flat Piston 3.3.3.1 Low Compressed Gas Pressure Figure 3.6 show the experimental results conducted at initial pressure and temperature of 236 torr and 297 K using the simulated flat piston. Mixture consists of 2% acetone in nitrogen. At the end of compression, the gas pressure and temperature are 12.5 bar and 780 K, respectively. Again, the raw fluorescence signal is shown as dotted line and the deduced temperature is denoted as solid line.
Exponential decay of
fluorescence intensity, as may be observed in a uniform temperature field, is shown as dashed line. Time “0” is taken as the end of compression stroke. Fluorescence signal at 1 ms after compression shows increased intensity in the central portion of the chamber. Any deviation of fluorescence signal from exponential decay is indicative of temperature inhomogenity. In addition, any hump corresponds to reduced temperature, while any depression indicates an increase in temperature.
At 1 ms, increased fluorescence
intensity in the central portion is due to the effect of roll-up vortex, which brings in cold gases from boundary layer to the center of the chamber. In the temperature field, this
50
increase in fluorescence intensity corresponds to approximately 100 K reduction in temperature. At 6 ms post-compression, as we go from wall to centerline, there are alternating cold and hot regions. Apart from the temperature reduction of approximately 50 K in the central regime, there is another zone of low temperature near the wall. Reason for this type of temperature non-uniformity will be discussed later on along with the discussion of computational results. At time steps of 8 and 18 ms, the width of central depression zone increases due to thermal transport, whereas temperature difference associated with this depression decreases. In addition, the effect of the side depression becomes more pronounced. Similar features are observed in temperature field at subsequent time steps. Even at long time scale of 129 ms, sharp temperature gradients exist in the chamber. Since only one PLIF measurement can be imaged during every run of the RCM, aerodynamic features captured at the specified time instance may vary slightly in different runs. For instance, at 19 ms of Fig. 3.6, instead of depression in temperature at the center, a peak in the center and two adjacent zones of low temperature are also observed. This would suggest that the cold gases flowing across the piston head did not reach the center of chamber from either direction. Although slightly different pattern may be observed in different runs even at the same reference time after compression, it does not alter the conclusion that the use of flat head piston leads to substantial temperature non-uniformity due to the effect of roll-up vortex. 3.3.3.2 High Compressed Gas Pressure Figure 3.7 illustrates another set of experimental results using simulated flat piston, with initial pressure and temperature of 700 torr and 297 K. The molar concentration of
51
acetone is 1% in nitrogen. This results in compressed gas pressure of 39.5 bar and compressed temperature of 815 K at TDC. Features of temperature field are similar to those obtained at low compressed pressure.
Specifically, there exists temperature
depression in the center at the end of compression and side depressions appear at the subsequent time steps. In contrast to the measurement at low pressure, at high pressure the central depression in temperature is smaller, being around 65 K. At the subsequent time steps, central temperature depression at higher pressure of 39.5 bar is generally less than that at low pressure of 12.5 bar.
3.3.4 Creviced Piston 3.3.4.1 Low Compressed Gas Pressure Using mixture consisting of 1% acetone in nitrogen, a series of experiments with creviced head piston are conducted. Initial conditions of 275 torr and 297 K yield compressed gas conditions of 11.95 bar and 760 K. PLIF results and the deduced temperature distribution at varying time steps are shown in Fig. 3.8. At 4.1 ms post compression, effect of the vortex has not reached the center of the chamber and the depression in temperature is observed to be approximately 35 K. At 6.2 ms, the central depression corresponds to approximately 30 K, as shown in Fig. 3.8. However, at 12 ms the effect of vortex has grown and central depression in temperature is approximately 80 K. In comparison with the case of flat piston at low pressure, the effect of vortex is reduced, although it is not eliminated. Figure 3.8 also demonstrates that the development of temperature inhomogenity is relatively gradual for the case of creviced piston.
52
3.3.4.2 High Compressed Gas Pressure Figure 3.9 shows the experimental results for the mixture consisting of 1% acetone in nitrogen, with the initial conditions of 870 torr and 297 K. At the end of compression, the compressed gas conditions are 39.5 bar and 770 K. It is seen from Fig. 3.9 that there is no evidence of any temperature inhomogenity at 2 ms post compression. Temperature in the chamber is quite uniform and fluorescence signal shows an exponential decay. At subsequent time steps, even up to 114 ms, there is no effect due to vortex. Eventually, the effect of the vortex becomes noticeable. At 200 ms, depression due to vortex is approximately 40 K, while at 390 ms central depression corresponds to approximately 100 K. These results clearly show the effects of pressure and piston head design on aerodynamics within a RCM. In comparison to the results using creviced piston at low pressure (Fig. 3.8), at high pressure the effect of vortex roll-up is significantly reduced. While at low pressure the effect of vortex appears from the end of compression, at high pressure such an effect becomes noticeable only at time greater than 114 ms postcompression. Furthermore, in comparison to the results using flat piston at high pressure, creviced piston at high pressure yields a higher degree of homogeneity inside test chamber for long time duration.
3.4
Numerical Analysis In Chapter 2, some results on numerical analysis using Star-CD CFD package were
presented in the context of optimization of piston head crevice. Here, additional simulations are performed to simulate the experimental conditions of PLIF measurements
53
and obtain insights into the detailed evolution of aerodynamic field inside a RCM. As with experimental PLIF measurements, simulation is conducted for conditions of low, intermediate, and high compressed gas pressures, and for flat as well as creviced piston head configurations. For CFD simulation, similar computational grid and time step size, as described in Chapter 2, are used. Computations are performed from the beginning of compression stroke with a compression time of 30 ms. Initially, the test gas, nitrogen, at rest is specified with uniform temperature and pressure.
At the wall boundary, constant
temperature and no slip condition are specified. In the simulation, piston starts from rest and travels to TDC, where it remains at rest for the subsequent time steps. During the compression stroke, the vertex filling methodology of Star-CD is chosen for motion of the grid, but after the end of compression stroke computational grid remains the same for the post compression period. In addition, calculations are performed and compared for both laminar and turbulent flow conditions. Various turbulent models are considered. A standard version of k-ε model is found to yield extremely high turbulent mixing. On the other hand, calculations using RNG model yield relatively less turbulent mixing and are found to be closer to experimental observation in comparison to the results obtained using the k-ε model. Furthermore, computational results show that the features of aerodynamics are better predicted by laminar calculations, while the actual aerodynamic pattern is expected to be somewhere between laminar and turbulent calculations.
54
3.5
Computational Results Using flat piston, Fig. 3.10 shows computed fields of velocity and temperature for
various time steps after the end of compression under laminar conditions. This is the case with relatively low pressure of 12.8 bar at TDC. At the end of compression, i.e. at 0 ms, maximum velocity in the chamber is 12.33 m/s. There is a big vortex spanning almost entire chamber, which shears cold gases from the boundary and brings them inside. As such, the regime near the centerline is greatly affected by the flow of cold gases from the boundary. Along with the main vortex, the computed results show the formation of corner vortices near the cylinder wall, piston head, and the centerline. At 20 ms after compression, maximum velocity has reduced to 3.27 m/s, but the velocity pattern is markedly different from that at the end of compression. Corner vortex on the piston head also becomes significantly large. At the subsequent time steps, the pattern of flow field remains almost identical and velocity in the chamber continues to decay. At 60 ms, the maximum velocity is reduced to 1.19 m/s. As a comparison, Fig. 3.11 shows the computed results using the RNG turbulent model under similar conditions of Fig. 3.10. The compressed pressure at TDC is now 13.14 bar. For turbulent calculations, the maximum velocities at the end of compression and at 10 ms post compression are respectively 13.38 and 5.48 m/s, which are not much different from those of laminar calculations. However, turbulent calculations do not exhibit any corner vortex, as seen in the laminar model. This is expected as turbulence suppresses the onset of flow separation. Figure 3.11 also demonstrates that the nature of velocity field remains identical for all time steps calculated. In the temperature field, enhanced mixing and shallow temperature gradient are shown in Fig. 3.11.
55
When comparing PLIF data with computational results, it can be clearly seen that simulation using laminar model agrees well with the features of aerodynamics observed in experiments, whereas turbulent calculations fail to do so.
Existence of sharp
temperature gradients in the experimentally deduced temperature profile is also shown in the laminar flow calculations. It should be mentioned that although turbulent calculations may not completely capture the qualitative trend observed in PLIF experiments, the comparison of Figs. 3.10 and 3.11 provide insights into the effect of turbulence on the aerodynamics inside a RCM. As with the experimental results, the effect of vortex on the temperature distribution near the central portion of the chamber at the end of compression is also noted in numerical simulation.
However, depression in temperature predicted by
simulation is greater (200 K) than that obtained by experimental data (100 K). This quantitative difference can be attributed to a number of reasons as follows. The piston used in the experiment is a simulated flat piston with approximately 10% crevice volume and not a truly flat piston as employed in the simulation. Therefore, relatively smaller effect of vortex in the experiment than what would occur for a truly flat piston for which simulation is calculated is expected.
In addition, the simulation is conducted for
conditions of constant wall temperature. But in the actual experiment, it is expected that the wall temperature increases by few Kelvin, resulting in smaller temperature nonuniformity. Furthermore, there exist alternating hot and cold regions in the experimental results using flat piston (cf. Figs. 3.6 and 3.7). Moving from the wall boundary towards the center, there is a zone of high temperature. After this zone, there is low temperature
56
regime, followed by another zone of high temperature. Finally, the center of chamber is at low temperature. Such a temperature field can be produced only by multiple vortices, rotating in different directions.
Similar features can be observed in the computed
temperature and velocity fields at 20 ms for laminar calculations as shown in Fig. 3.10(b). There are alternating hot and cold regions and velocity field exhibits primarily two vortices. Based on laminar calculations, Fig. 3.12 shows the computed temperature fields for creviced piston at a relatively low pressure of 11.9 bar at TDC.
For ease of
representation, only the main chamber volume, apart from the crevice zone, is shown in Fig. 3.12. At the end of compression (0 ms), there is small temperature non-uniformity near the piston head. The maximum predicted gas velocity at this time is 2.2 m/s, which is much smaller than the gas velocity for flat piston. Effect of the vortex is seen to gradually grow at subsequent time steps. Even at 100 ms post compression, unlike flat piston, significant part of the chamber is unaffected by the vortex, as shown in Fig. 3.12(f). Figure 3.13 shows the simulated temperature fields for flat piston at a higher pressure of 40.7 bar at TDC, using the laminar model.
Patterns of velocity and
temperature fields are identical to those calculated at low pressure. However, at higher pressure, the effect of vortex penetrates at much slower rate. For instance, temperature distribution at 40 ms under higher compressed pressure (Fig. 3.13(c)) is very similar to that at 20 ms under lower compressed pressure (Fig. 3.10(b)). Comparing with Fig. 3.13, Fig. 3.14 shows the laminar simulation for creviced piston at a high pressure of 39.9 bar at TDC. At the end of compression, apart from
57
minor non-uniformity at the center, the effect of vortex is practically non-existent. Only at time of 100 ms, the vortex effect reaches the central plane of the chamber. Similar features are observed in experimental results. However, laminar calculations slightly over-predict the onset of temperature inhomogenity, as experimental results do not observe any vortex until 114 ms post compression. Figures 3.15 and 3.16 shows the temperature field for flat and creviced piston respectively at an intermediate pressure of approximately 20 bar. For both piston head configurations, effect of vortex at the intermediate pressure penetrates at a rate which is slower than rate of penetration at low pressure and faster than that at high pressure. In addition, the effect of pressure on vortex formation for creviced piston is consistent in both experimental and computational results. Increase in pressure reduces the effect of vortex. For a flat piston, although such a pressure effect cannot be clearly observed in experimental data, it is demonstrated in the computational results. Effect of pressure on the vortex roll-up and the induced temperature inhomogenity can be attributed to the change in thermal diffusivity. At higher compressed gas pressure, the mixture thermal diffusivity is lower, thereby leading to thinner thermal boundary layer. Since temperature inhomogenity is induced by shearing cold gases from boundary to the interior of cylindrical chamber, a thinner boundary layer under high compressed gas pressure will slow down the development of temperature inhomogenity.
3.6
Assessment of Adiabatic Core Hypothesis Adiabatic core hypothesis is based on the assumption that the effect of heat loss due
to wall does not affect the temperature of the core region at the short time scales
58
encountered in RCM. From the present experimental results using flat and creviced piston heads, the validity of adiabatic core hypothesis can be clearly identified. For the case of flat piston at low pressure, at the end of compression (Fig. 3.6(a)) there is large temperature gradient at the center of the chamber, while the rest of the chamber is not much affected. However, at post compression time of 6 ms (Fig. 3.6(b)), temperature gradient is present across the entire domain of the chamber and the effect of rolled-up cold gases has spread everywhere. Therefore, at 0 ms there is a possibility that adiabatic core assumption may hold, but the actual temperature quickly begins to deviate from adiabatic core temperature. From 6 ms onwards, the failure of adiabatic core hypothesis is expected. For flat piston at high compressed pressure, temperature gradient across the entire chamber is observed at 15.8 ms post compression (Fig. 3.7(c)). For the case of creviced piston at high pressure, even at 114 ms post compression (Fig 3.9(f)), the effect of vortex is not observed. This suggests the validity of adiabatic core assumption up to this time when using creviced piston at high pressure. However, at low pressure, the validity of adiabatic core hypothesis is not as good. Similar features regarding the validity of adiabatic core hypothesis are also observed in the computational results. Figure 3.17 plots the time evolution of pressure and maximum temperature, obtained from the simulations employing different conditions and piston head configurations. In addition, the temperature profile calculated by using adiabatic core hypothesis is also shown and compared in Fig. 3.17. At the end of compression, temperature predicted by the hypothesis exactly matches with the maximum temperature of CFD analysis. For flat piston, however, computed maximum temperature quickly begins to significantly deviate from the temperature of adiabatic core
59
as time proceeds. Laminar simulation for flat head at low pressure (Fig. 3.17(a)) shows that at 50 ms post compression, adiabatic core assumption over-estimates the peak temperature by 42 K, while at 100 ms the over-estimate is 67 K. Therefore, the use of adiabatic core assumption for the flat piston runs becomes questionable, especially considering the exponential dependence of reaction rate on temperature. For flat piston at higher pressure, it is seen from Fig. 3.17(b) that adiabatic core hypothesis is valid up to 20 ms post compression, and deviations are 16 K and 46 K at 50 ms and 100 ms, respectively. When turbulent simulation is used, larger over-prediction resulting from adiabatic core assumption is observed, as shown in Fig. 3.17(c). For creviced head piston under laminar conditions, Figs. 3.17(d) and 3.17(e) show that adiabatic core hypothesis accurately predicts maximum temperature for time greater than 100 ms. Effect of turbulence can be seen by comparing Figs. 3.17(e) and 3.17(f). Although prediction of maximum temperature becomes slightly worse when considering turbulent mixing, for creviced piston at high pressure, adiabatic core hypothesis is still quite adequate for a long period of time. Figures 3.18(a) and 3.18(b) show the extent of core region from the end of compression for laminar and turbulent calculations, respectively. Here the core region is defined as the ratio of the volume of the chamber that is within 5% of the maximum instantaneous temperature to the total volume of the chamber at the end of compression. For the case of creviced piston, only the main cylinder volume, apart from crevice, is considered for calculation of core region. Note that the core region may not be the geometric center of the chamber, but the region of maximum temperature.
60
For flat piston at low pressure of 12.8 bar at TDC, while at the end of compression (time = 0) core region encompasses approximately 70% volume of the chamber, it rapidly reduces to 35% at 20 ms and continues to reduce as time proceeds. In a previous CFD work (Griffiths et al., 1993a), similar results were reported regarding the state of core region at the end of compression. It was also reported (Griffiths et al., 1993a) that although gases in the chamber were moving at the velocity of the piston at the end of compression, predicted mixture velocity reduced rapidly within very short time interval. The present simulation demonstrates that although relatively large gas volume is at adiabatic core temperature at the end of compression, heat transport rates from this core region are not low. The failure of adiabatic core assumption after the end of compression results from high velocity field in a small confined chamber. This can be understood from the time evolution of maximum velocity from the end of compression, as shown in Fig. 3.19.
Significantly higher velocities with a flat head piston in
comparison to creviced piston are seen in Fig. 3.19. High velocities at the end of compression quickly bring the effect of wall into the core region. For flat piston at high pressure of 40.7 bar, Fig. 3.18(a) shows that the extent of core region is relatively larger than that at low pressure. In contrast, with a creviced piston, the extent of core is significantly increased. At low pressure of 11.9 bar, the core region is 89% of the volume at the end of compression and 53% at 60 ms postcompression. For creviced piston at higher pressure of 39.9 bar, the extent of core is even greater. It is 93% at the end of compression and 70% at a long post-compression time of 120 ms.
61
Figure 3.18(b) further demonstrates the effect of turbulence on the extent of core region. In general, the effect of turbulent mixing lowers the temperature of the region of maximum temperature while increases its spatial extent. It should be noted that under both laminar and turbulent conditions the primary reason for the failure of adiabatic core hypothesis is due to the mixing of cold gases sheared from the wall boundary with the core region, rather than the heat loss from cold wall. Figure 3.19 also illustrates that the use of creviced piston head substantially reduces gas velocity at the end of compression. As a result, temperature predicted by adiabatic core hypothesis matches closely with the actual maximum temperature for long time interval, as shown in Fig. 3.17. Additionally, the extent of core region is greatly improved, as seen in Fig. 3.18. Since the determination of temperature in an RCM is typically indirect, the present numerical experiments therefore demonstrate that it is extremely important to have proper operating conditions and geometric configuration of the test chamber so that the adiabatic core hypothesis is valid. Although various RCMs have different operating conditions of clearance, stroke, etc., similar features of aerodynamics as observed in this study are expected. Depending upon the extent of vortex effect, various RCMs would vary in terms of the validity of adiabatic core hypothesis after the end of compression, although such a hypothesis is expected to be accurate at TDC. This would explain the discrepancies in the reported experimental data from different RCMs, even under similar conditions.
62
3.7 Effect of Piston Head Configuration on Ignition Delay It has been shown in Fig. 3.3 that under similar conditions the use of different piston head configurations can lead to different ignition delay. In order to address the reasons, additional simulations and experiments are carried out based on the same compressed gas temperature and pressure for a flat and a creviced piston head configurations. To achieve the same condition at the end of compression, clearance is increased to account for the absence of crevice volume in the case of flat head piston. Figure 3.20 shows the numerical comparison using nitrogen. Although pressure and temperature at TDC are the same for both piston head configurations, the rate of pressure drop for creviced piston is higher than that for the flat piston. This result is attributed to the larger surface area to volume ratio for the creviced piston case. The crevice is a zone of small volume but large surface area, which accelerates the rate of heat loss to the wall. Due to higher post-compression pressure in the case of flat piston, temperature calculated by adiabatic core hypothesis is also higher. However, it is seen from Fig. 3.20 that the computed maximum temperature for the flat head piston case can become lower than that for the creviced head piston case.
This is because of the
deviation from the adiabatic core hypothesis when using a flat head piston. Consistent with the computational results shown in Figs. 3.17-3.19, for the flat piston case the adiabatic core assumption is not valid, and the significant effect of roll-up vortex decreases the core temperature, despite of higher pressure. The resulting lower core temperature and higher pressure for the use of flat piston will in turn affect the ignition delay.
63
Figure 3.21 compares the experimental pressure traces using stoichiometric isooctane/O2/inert gas mixtures for both piston head configurations under similar composition, compressed gas temperature, and TDC pressure. Results for three different adiabatic core temperatures, namely 684 K, 745 K, and 878 K at TDC, are compared in Fig. 3.21.
Both single-stage (878 K) and two-stage (684 K and 745 K) ignition
phenomena are clearly shown. At TDC, exact agreement between pressure traces for both pistons is observed. It is also seen in Fig. 3.21 that post-compression pressure for flat head piston is higher, as demonstrated by the computational results. The combined effect of lower core temperature and higher pressure for the flat piston case alters the ignition delay as depicted in Fig. 3.21.
3.8
Summary The purpose of this work is to investigate the effects of aerodynamics on the
experimental data in a rapid compression machine. Such an understanding is important in order to unambiguously characterize the state of reacting mixture and to explain the reasons for the mismatch of ignition delay data obtained from various RCMs. Experimental and computational studies are conducted for a flat and a creviced piston head configurations.
Experimental results using PLIF of acetone demonstrate the
suitability of a creviced piston in providing a homogeneous reacting core. Computational results under laminar conditions closely reproduce the features of aerodynamics as observed in experiment. For a flat piston, although adiabatic core assumption accurately predicts the temperature at the end of compression, it significantly over-predicts postcompression temperature as the effect of fast-moving cold gases quickly spreads all over
64
the chamber. A creviced piston head is found to significantly increase the extent of core region. Therefore, when using a properly designed creviced piston, the post-compression temperature can be accurately predicted based on adiabatic core hypothesis. It is also observed that the effect of roll-up vortex is reduced as the compressed pressure is increased. The present results clearly demonstrate that using a creviced piston design, the actual temperature in a RCM can be deduced from the pressure trace. From the kinetic modeling point of view, unambiguous determination of temperature is the prime requirement, as heat loss associated with a RCM can be easily accounted for by using an empirical relation. This work suggests that when a properly designed creviced piston is used, a zero-dimensional model should satisfactorily model the experimental results using a RCM. We shall detail how the comparison of experimental and computational results is performed in the following two Chapters. However, when a flat head piston is employed, the use of a zero-dimensional model to simulate experimental data is deemed inadequate.
65
Fig. 3.1 –Creviced piston head configuration. Dimensions (in mm): A=0.50, B=4.0, C=0.15, D=20.0, and E=1.50.
66
Controller with Pulse Generator
Nd:YAG
DYE Doubler
Intensified CCD Camera
Iris M1
WG305 Filter
Ls M2
Lc
DG 535
PC
Laser sheet at 279 nm
PC For actuating relay to fire the RCM
Beam Dump
Fig. 3.2 – PLIF Schematic. M1, M2 – UV mirrors for lowering the height of laser beam to RCM quartz window height. Ls – Spherical lens, Lc – Cylindrical lens.
67
120
Present work, Creviced Piston Present Work, Flat Piston
Ignition Delay (ms)
Minetti et al. (1996b), Flat Piston
100 80 60 40 20 650
700 750 800 850 Adiabatic Core Temperature (K)
900
Fig. 3.3 – Ignition delay versus adiabatic core temperature for stoichiometric isooctane/oxygen/inert mixtures. Composition: iC8H18/O2/inert=1/12.5/47. Initial conditions: P0=331 torr and T0=297 K. Adiabatic core temperature is varied by changing the composition of inert gases.
68
2
x 106 Cylinder Wall
Intensity (A.U.)
1.5
1
0.5
0
0
200
400
600
800
1000
Pixel
Fig. 3.4 – A representative single-shot PLIF image and the associated photon counts. Chamber conditions: pressure=44 bar and temperature=297 K. Laser sheet enters from right side of the image.
69
6
350
325 1 300
0
-2
-1 0 1 Radial Direction (cm)
2
Temperature (K)
Intensity (A.U.)
2 x 10
275
Fig. 3.5 – Deduction of temperature from fluorescence signal. Chamber conditions: pressure=44 bar and temperature=297 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity by using the Beer-Lambert law. Solid line: deduced temperature distribution.
70
6 2 x 10
6 2 x 10
800
(a) 1 ms
800
(b) 6 ms
T
-2
-1 0 1 Radial Direction (cm)
2
6 1.5 x 10
650
2
6 1.5 x 10
700
2
6 2 x 10
750
T
1
0.5
700
-2
-1 0 1 Radial Direction (cm)
2
650
700
-2
-1 0 1 Radial Direction (cm)
2
750
1
700
-2
-1 0 1 Radial Direction (cm)
2
6 2 x 10
650
800
(h) 129 ms
1.5
750
T
1
0.5
650
800
T
1.5
0.5
650
Intensity (A.U.)
Intensity (A.U.)
1.5
1
(f) 22 ms
800
(g) 57.5 ms
750
6 2 x 10
Intensity (A.U.)
1
-1 0 1 Radial Direction (cm)
800
1.5
0.5
Temperature, T (K)
Intensity (A.U.)
750
-2
Intensity (A.U.)
650
T
650
T
800
(e) 19 ms
0.5
Temperature, T (K)
700
Temperature, T (K)
Intensity (A.U.)
1
-1 0 1 Radial Direction (cm)
2
(d) 18 ms
750
-2
-1 0 1 Radial Direction (cm)
6 2 x 10
T
0.5
-2
-2
Temperature, T (K)
700
0.5
800
(c) 8 ms
1
Temperature, T (K)
0.5
750
Temperature, T (K)
700
1.5
-1 0 1 Radial Direction (cm)
700
2
Temperature, T (K)
1
Intensity (A.U.)
750
Intensity (A.U.)
1.5
Temperature, T (K)
T
650
Fig. 3.6 – PLIF intensities and the deduced temperature distributions at varying times after compression for a simulated flat piston head. Gas composition: 2% acetone in nitrogen. Conditions at TDC: pressure=12.5 bar and temperature=780 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
71
6 1.5 x 10
6 2 x 10
(b) 2.1 ms
(a) 1.2 ms
800
1
775 750
0.5
Intensity (A.U.)
T
Temperature, T (K)
800
1
775 0.5
750 725
725 0 -1 0 1 Radial Direction (cm)
2
6
2
6 2 x 10
2 x 10
(c) 15.8 ms
(d) 25 ms
775 1
750
1.5 Intensity (A.U.)
800
1.5
Temperature, T (K)
825
T Intensity (A.U.)
-1 0 1 Radial Direction (cm)
825 T
800
1
775
0.5
750
725 0.5
-2
-1 0 1 Radial Direction (cm)
725 0
2
6 2 x 10
-1 0 1 Radial Direction (cm)
2
(f) 47.8 ms
800 T 775
1
750
Intensity (A.U.)
825 Temperature, T (K)
Intensity (A.U.)
-2
6 2 x 10
(e) 39.6 ms
1.5
825 800
1.5
T 775
1
750
725 0.5
-2
-1 0 1 Radial Direction (cm)
725 0.5
2
6
-1 0 1 Radial Direction (cm)
2
1.5 x 10
(g) 60.6 ms
(h) 130 ms
825
T 775 1
750
Intensity (A.U.)
800
1.5
825 Temperature, T (K)
Intensity (A.U.)
-2 6
2 x 10
800
1
775 T
0.5
750
725 0.5
-2
Temperature, T (K)
-2
-2
Temperature, T (K)
0
-1 0 1 Radial Direction (cm)
2
Temperature, T (K)
Intensity (A.U.)
1.5
T
Temperature, T (K)
825
825
725 0
-2
-1 0 1 Radial Direction (cm)
2
Fig. 3.7 – PLIF intensities and the deduced temperature distributions at varying times after compression for a simulated flat piston head. Gas composition: 1% acetone in nitrogen. Conditions at TDC: pressure=39.5 bar and temperature=815 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
72
x 105
5
x 10
(a) 4.1 ms
(b) 6.2 ms
-2
-1 0 1 Radial Direction (cm)
2
750 T
Intensity (A.U.)
650
3
Temperature, T (K)
Intensity (A.U.)
700
4
700
3
650
600
-2
5
5
x 10
-1 0 1 Radial Direction (cm)
2
(d) 20 ms
-1 0 1 Radial Direction (cm)
2
700
650
2
600
-2
-1 0 1 Radial Direction (cm)
2
5
x 105
x 10
(f) 90 ms
(e) 30.8 ms
3
650
-1 0 1 Radial Direction (cm)
2
Intensity (A.U.)
700
3 T
700
650
2
600
-2
x 105
-1 0 1 Radial Direction (cm)
2
600
5 4 x 10
(g) 128 ms
(h) 336 ms
4
750
T 3 650
600 -1
0
1
Radial Direction (cm)
2
Intensity (A.U.)
700
Temperature, T (K)
750
-2
Temperature, T (K)
750
750 T
Temperature, T (K)
4
-2
600
3 700 T 2
1
650
-2
-1 0 1 Radial Direction (cm)
2
Temperature, T (K)
-2
T
Temperature, T (K)
650
3
3 Intensity (A.U.)
Intensity (A.U.)
700
750 Temperature, T (K)
750 T
4
Intensity (A.U.)
600
x 10
(c) 12 ms
Intensity (A.U.)
Temperature, T (K)
750 T
4
600
Fig. 3.8 – PLIF intensities and the deduced temperature distributions at varying times after compression for a creviced piston head. Gas composition: 1% acetone in nitrogen. Conditions at TDC: pressure=11.95 bar and temperature=760 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
73
5 15 x 10
(a) 2 ms
700 5
650
Intensity (A.U.)
750
10
Temperature, T (K)
T
800 T
10
750 700
5
650
600 -1 0 1 Radial Direction (cm)
2
5
15 x 10
2
(d) 38 ms
750
T
700 5
650
Intensity (A.U.)
800 Temperature, T (K)
Intensity (A.U.)
-1 0 1 Radial Direction (cm)
15 x 10
(c) 25 ms
10
-2 5
10
800 750
T
700 5
650
600 0
-2
-1 0 1 Radial Direction (cm)
600 0
2
5
-1 0 1 Radial Direction (cm)
2
15 x 10
(e) 70 ms
(f) 114 ms
750 T 700
5
650
Intensity (A.U.)
800 Temperature, T (K)
Intensity (A.U.)
-2 5
15 x 10
10
800 750
10 T
700 5
650
600 0
-2
-1 0 1 Radial Direction (cm)
600 0
2
5 15 x 10
-1 0 1 Radial Direction (cm)
2
(h) 390 ms
750
10 T
700
5
650
Intensity (A.U.)
800 Temperature, T (K)
Intensity (A.U.)
-2
5 15 x 10
(g) 200 ms
800 750
10
700
T 5
650
600 0
-2
Temperature, T (K)
-2
600 0
Temperature, T (K)
0
-1 0 1 Radial Direction (cm)
2
Temperature, T (K)
Intensity (A.U.)
(b) 16 ms
800
Temperature, T (K)
5 15 x 10
600 0
-2
-1 0 1 Radial Direction (cm)
2
Fig. 3.9 – PLIF intensities and the deduced temperature distributions at varying times after compression for a creviced piston head. Gas composition: 1% acetone in nitrogen. Conditions at TDC: pressure=39.5 bar and temperature=770 K. Dotted line: raw fluorescence intensity. Dashed line: calculated fluorescence intensity in a uniform temperature field by using the Beer-Lambert law. Solid line: deduced temperature distribution.
74
(a) 0 ms
Piston Head
12
8
4
0
5
(b) 20 ms
10 15 Radius (mm)
Max. Velocity: 12.33 m/s
Temperature (K)
20
25
892.8 871.4 849.9 828.5 807.1 785.7 764.2 742.8 721.4 699.9 678.5 657.1 635.7 614.2 592.8
Temperature (K)
12
8
4
0
5
10 15 Radius (mm)
20
25
Max. Velocity: 3.27 m/s
824.9 803.5 782.0 760.6 739.2 717.8 696.3 674.9 653.5 632.0 610.6 589.2 567.8 546.3 524.9
(c) 40 ms
Temperature (K)
Max. Velocity: 1.818 m/s
794.2 772.8 751.3 729.9 708.5 687.1 665.6 644.2 622.8 601.3 579.9 558.5 537.1 515.6 494.2
(d) 60 ms
Temperature (K)
Max. Velocity: 1.192 m/s
771.9 750.5 729.0 707.6 686.2 664.7 643.3 621.8 600.4 579.0 557.5 536.1 514.7 493.2 471.8
(e) 80 ms
Temperature (K)
Max. Velocity: 0.911 m/s
749.4 728.0 706.5 685.1 663.7 642.3 620.8 599.4 578.0 556.5 535.1 513.7 492.3 470.8 449.4
Fig. 3.10 – STAR-CD simulation of temperature and velocity fields at varying times after compression for flat piston at a pressure of 12.8 bar at TDC under laminar condition.
75
(a) 0ms
Piston Head
12
8
4
0
5
(b) 40 ms
10 15 Radius (mm)
Max. Velocity: 13.38 m/s
Temperature (K)
20
25
898.6 877.2 855.7 834.3 812.9 791.5 770.0 748.6 727.2 705.7 684.3 662.9 641.5 620.0 598.6
Temperature (K)
12
8
4
0
5
10 15 Radius (mm)
20
25
Max. Velocity: 1.547 m/s
738.8 717.4 695.9 674.5 653.1 631.7 610.2 588.8 567.4 545.9 524.5 503.1 481.7 460.2 438.8
(c) 80 ms
Temperature (K)
Max. Velocity: 0.6566 m/s
672.5 651.1 629.6 608.2 586.8 565.4 543.9 522.5 501.1 479.6 458.2 436.8 415.4 393.9 372.5
(d) 100 ms
Temperature (K)
Max. Velocity: 0.482 m/s
648.5 627.1 605.6 584.2 562.8 541.4 519.9 498.5 477.1 455.6 434.2 412.8 391.4 369.9 348.5
Fig. 3.11 – STAR-CD simulation of temperature and velocity fields at varying times after compression for flat piston at a pressure of 13.14 bar at TDC under turbulent condition.
76
(a) 0 ms
Piston Head
Temperature (K)
12
8
4
0
(b) 20 ms
5
(c) 40 ms
10 15 Radius (mm)
20
25
Temperature (K)
(d) 60 ms
Temperature (K)
Temperature (K) 730.6 715.6 700.6 685.6 670.6 655.6 640.6 625.6 610.6 595.6 580.6 565.6 550.6 535.6 520.6
741.3 726.3 711.3 696.3 681.3 666.3 651.3 636.3 621.3 606.3 591.3 576.3 561.3 546.3 531.3
(e) 80 ms
Temperature (K) 762.6 747.6 732.6 717.6 702.6 687.6 672.6 657.6 642.6 627.6 612.6 597.6 582.6 567.6 552.6
789.6 774.6 759.6 744.6 729.6 714.6 699.6 684.6 669.6 654.6 639.6 624.6 609.6 594.6 579.6
(f) 100 ms
722.8 707.8 692.8 677.8 662.8 647.8 632.8 617.8 602.8 587.8 572.8 557.8 542.8 527.8 512.8
Temperature (K) 716.6 701.6 686.6 671.6 656.6 641.6 626.6 611.6 596.6 581.6 566.6 551.6 536.6 521.6 506.6
Fig. 3.12 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 11.9 bar at TDC under laminar condition.
77
(a) 0 ms
Piston Head
Temperature (K)
(b) 20 ms
12
8
4
0
5
(c) 40 ms
10 15 Radius (mm)
20
25
Temperature (K)
(d) 60 ms
Temperature (K)
Temperature (K) 833.3 811.9 790.4 769.0 747.6 726.2 704.7 683.3 661.9 640.4 619.0 597.6 576.2 554.7 533.3
858.8 837.4 815.9 794.5 773.1 751.7 730.2 708.8 687.4 665.9 644.5 623.1 601.7 580.2 558.8
(e) 80 ms
Temperature (K) 883.0 861.6 840.1 818.7 797.3 775.9 754.4 733.0 711.6 690.1 668.7 647.3 625.9 604.4 583.0
904.4 883.0 861.5 840.1 818.7 797.3 775.8 754.4 733.0 711.5 690.1 668.7 647.3 625.8 604.4
(f) 100 ms
805.0 783.6 762.1 740.7 719.3 697.9 676.4 655.0 633.6 612.1 590.7 569.3 547.9 526.4 505.0
Temperature (K) 794.2 772.8 751.3 729.9 708.5 687.1 665.6 644.2 622.8 601.3 579.9 558.5 537.1 515.6 494.2
Fig. 3.13 – STAR-CD simulation of temperature field at varying times after compression for flat piston at a pressure of 40.7 bar at TDC under laminar condition.
78
(a) 0 ms
Piston Head
Temperature (K)
(b) 20 ms
806.9 792.6 778.3 764.0 749.8 735.5 721.2 706.9 692.6 678.3 664.0 649.8 635.5 621.2 606.9
12
8
4
0
5
(c) 40 ms
10 15 Radius (mm)
20
788.4 774.1 759.8 745.5 731.3 717.0 702.7 688.4 674.1 659.8 645.5 631.3 617.0 602.7 588.4
25
Temperature (K)
(d) 60 ms
Temperature (K)
Temperature (K) 767.1 752.8 738.5 724.2 710.0 695.7 681.4 667.1 652.8 638.5 624.2 610.0 595.7 581.4 567.1
776.4 762.1 747.8 733.5 719.3 705.0 690.7 676.4 662.1 647.8 633.5 619.3 605.0 590.7 576.4
(e) 80 ms
Temperature (K)
(f) 100 ms
759.7 745.4 731.1 716.8 702.6 688.3 674.0 659.7 645.4 631.1 616.8 602.6 588.3 574.0 559.7
Temperature (K) 753.6 739.3 725.0 710.7 696.5 682.2 667.9 653.6 639.3 625.0 610.7 596.5 582.2 567.9 553.6
Fig. 3.14 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 39.9 bar at TDC under laminar condition.
79
(a) 0 ms
Piston Head
Temperature (K)
(b) 20 ms
898.2 876.8 855.3 833.9 812.5 791.1 769.6 748.2 726.8 705.3 683.9 662.5 641.1 619.6 598.2
12
8
4
0
5
(c) 40 ms
10 15 Radius (mm)
20
853.6 832.2 810.7 789.3 767.9 746.5 725.0 703.6 682.2 660.7 639.3 617.9 596.5 575.0 553.6
25
Temperature (K)
(d) 60 ms
Temperature (K)
Temperature (K) 793.9 772.5 751.0 729.6 708.2 686.8 665.3 643.9 622.5 601.0 579.6 558.2 536.8 515.3 493.9
811.4 790.0 768.5 747.1 725.7 704.3 682.8 661.4 640.0 618.5 597.1 575.7 554.3 532.8 511.4
(e) 80 ms
Temperature (K)
(f) 100 ms
778.5 757.1 735.6 714.2 692.8 671.4 649.9 628.5 607.1 585.6 564.2 542.8 521.4 499.9 478.5
Temperature (K) 763.3 741.9 720.4 699.0 677.6 656.2 634.7 613.3 591.9 570.4 549.0 527.6 506.2 484.7 463.3
Fig. 3.15 – STAR-CD simulation of temperature field at varying times after compression for flat piston at a pressure of 20.5 bar at TDC under laminar condition.
80
(a) 0 ms
Piston Head
Temperature (K)
(b) 20 ms
797.4 783.1 768.8 754.5 740.3 726.0 711.7 697.4 683.1 668.8 654.5 640.3 626.0 611.7 597.4
12
8
4
0
5
(c) 40 ms
10 15 Radius (mm)
20
771.3 757.0 742.7 728.4 714.2 699.9 685.6 671.3 657.0 642.7 628.4 614.2 599.9 585.6 571.3
25
Temperature (K)
(d) 60 ms
Temperature (K)
Temperature (K) 746.0 731.7 717.4 703.1 688.9 674.6 660.3 646.0 631.7 617.4 603.1 588.9 574.6 560.3 546.0
756.3 742.0 727.7 713.4 699.2 684.9 670.6 656.3 642.0 627.7 613.4 599.2 584.9 570.6 556.3
(e) 80 ms
Temperature (K)
(f) 100 ms
738.5 724.2 709.9 695.6 681.4 667.1 652.8 638.5 624.2 609.9 595.6 581.4 567.1 552.8 538.5
Temperature (K) 732.6 718.3 704.0 689.7 675.5 661.2 646.9 632.6 618.3 604.0 589.7 575.5 561.2 546.9 532.6
Fig. 3.16 – STAR-CD simulation of temperature field at varying times after compression for creviced piston at a pressure of 19.7 bar at TDC under laminar condition.
81
Flat Piston 900
8
700
Pressure
650
Deduced Temperature using Adiabatic Core Hypothesis
6
Temperature (K)
750
Pressure (bar)
10
800
10
750
8
700
600
550
2
550
500
0 100
500
Computed Maximum Temperature
-20
0
20 40 Time (ms)
60
80
6
650
4
600
12
850
12
800
4 2 -20
(b) 40.7 bar, Laminar Conditions
900
80
0 100
40
20
650
800
30
750 700
20
650 600
10
550
Pressure (bar)
700
Temperature (K)
30
750
Pressure (bar)
10
550 -20
0
20 40 Time (ms)
60
80
500
0 100
-20
(c) 41.5 bar, Turbulent Conditions
900
0
20
40 60 Time (ms)
80
100
120
0
(f) 38.9 bar, Turbulent Conditions
900
40
850
40
30
750 700
20
650 600
Temperature (K)
800
Pressure (bar)
850
10
550
800
30
750 700
20
650 600
Pressure (bar)
Temperature (K)
60
850
600
Temperature (K)
20 40 Time (ms)
(e) 39.9 bar, Laminar Conditions
40
800
500
0
900
850
500
14
900
14
850 Temperature (K)
Creviced Piston (d) 11.9 bar, Laminar Conditions
16
Pressure (bar)
(a) 12.8 bar, Laminar Conditions
10
550 -20
0
20 40 Time (ms)
60
80
0 100
500
-20
0
20 40 Time (ms)
60
80
0 100
Fig. 3.17 – Assessment of adiabatic core hypothesis based on computational results. Pressure at TDC is indicated in the figure. Dashed line, solid line, and line with symbols respectively represent pressure, calculated temperature using adiabatic core hypothesis, and actual maximum temperature in the chamber.
82
100 Extent of Core Region (%)
(a) Creviced, 39.9 bar
80 60
Creviced, 11.9 bar 40
Flat, 40.7 bar
20 0
Flat, 12.8 bar
0
20
40
60 Time (ms)
80
100
120
100 Extent of Core Region (%)
Creviced, 38.9 bar 80
(b) Creviced, 39.9 bar
60
Flat, 41.5 bar
40 Flat, 40.7 bar
20 0
0
20
40
60 Time (ms)
80
100
120
Fig. 3.18 – Extent of core region, calculated from computational results, as a function of post-compression time. (a) Effect of pressure and piston head configuration. (b) Effect of turbulence. Pressure at TDC is indicated in the figure. Solid line: laminar calculation. Dashed line: turbulent calculation.
83
Maximum Velocity (m/s)
12 Creviced Piston, 11.9 bar
10
Creviced Piston, 39.9 bar Flat Piston, 12.8 bar
8
Flat Piston, 40.7 bar
6 4 2 0
0
20
40 60 Time (ms)
80
100
Fig. 3.19 – Time variation of maximum velocity in the cylinder from the end of compression for laminar calculation. Pressure at TDC is indicated in the figure.
850
12
Pressure, Flat
Temperature (K)
750
8
700 650
Adiabatic Core Temperature, Flat
4
Pressure (bar)
Pressure, Creviced
800
Computed Maximum Temperature, Flat Adiabatic Core Tempertaure, Creviced
600
Computed Maximum Temperature, Creviced
550
-20
0
20
40 Time (ms)
60
80
0 100
Fig. 3.20 – Computational results on the effect of piston head configuration under the same conditions of temperature and pressure at TDC. Cases of flat and creviced piston head configurations with nitrogen as the test gas are shown.
84
878 K
684 K
745 K
Pressure (bar)
15
10
5
0
-20
0
20
40 Time (ms)
60
80
100
Fig. 3.21 – Comparison of measured pressure profiles for iso-octane auto-ignition experiments using flat and creviced piston head configurations. Composition: iC8H18/O2/inert=1/12.5/47. Adiabatic core temperature at TDC is indicated in the figure. Solid line: flat piston. Dashed line: creviced piston.
85
CHAPTER 4 PERFORMANCE CHARACTERIZATION OF RAPID COMPRESSION MACHINE 4.1 Introduction Before using the designed RCM for combustion related studies, the performance of machine must be characterized. In the previous chapter, we have shown that the RCM fulfils the important requirement of unambiguous determination of the state of reacting mixtures. Further, it is necessary to demonstrate that the present RCM is capable of generating a wide range of test conditions for studying various combustion phenomenon of interest. These conditions should be repeatable in order to provide experimental data of high fidelity. From mechanical design point of view, requirements of fast compression and vibration-free stopping of moving piston need to be fulfilled. Also, performance of rapid sampling apparatus in terms of fast quenching of reaction products and quantitative analysis using GC needs to be demonstrated. In order to assess the predictability of a detailed kinetic mechanism, a numerical model, which accounts for the effect of heat loss during the compression and reaction period as well as accurately characterizes the thermodynamic state of reacting mixture, is also required for simulating the experimental data with detailed chemistry. It will be shown in due course that the present numerical model appropriately predicts the performance of the machine for non-reacting mixtures. Therefore, the comparison of the experimental results using reacting mixtures and the simulated results with detailed
86
chemistry can be used to assess the “comprehensiveness” of the detailed kinetic mechanism employed. In this chapter, numerical model developed for simulating RCM experiments is first described, followed by a discussion of characterization experiments performed to demonstrate the performance of RCM and establish its suitability for combustion related studies. A creviced piston head is used for all the data shown in this chapter.
4.2 Numerical Model As discussed in Chapter 1, in the present RCM, measurements of gas pressure and initial temperature (T0) are direct, whereas the determination of compressed gas temperature is indirect. If the compression process is truly adiabatic, the adiabatic compression temperature at TDC, Tac, is calculated by Tac
∫
T0
1 dT = ln ( CR ) , γ −1 T
(4.1)
where γ is the specific heat ratio and CR represents the compression ratio. However, due to heat transfer from hot gases to the wall, compression is not truly adiabatic. Although the actual pressure and temperature are lower than those achieved for adiabatic compression, various studies (Griffiths et al., 1993a; Desgroux et al., 1995) have shown that it is reasonable to assume an isentropically-compressed core region of the gas mixture, with the effective compression ratio being modified by heat transfer to wall. In Chapter 3, it was demonstrated that while this assumption may fail for a flat piston head due to the effect of roll-up vortex, it remains valid for a properly designed creviced piston for long time duration after the end of compression stroke. Compressed gas temperature at TDC, Tc, based on the assumption of an adiabatic core can be calculated by
87
Tc
γ
dT
⎛P ⎞
c ∫ γ − 1 T = ln⎜⎜ P ⎟⎟ T0 ⎝ o⎠
(4.2)
,
where Pc is the measured pressure at the end of compression. Difference between the calculated Tc and Tac indicates the degree of non-adiabaticity of compression. Based on the adiabatic core hypothesis, computed pressure during and after compression is matched to experimental pressure trace by using a zero-dimensional model with empirically determined parameters. Two different approaches have been proposed to account for the effect of heat loss. One is to include a heat loss term in energy equation (Ribaucour et al., 2000; Westbrook et al., 2002), and the other is to specify a volume expansion term (Tanaka et al., 2003b). For the present RCM, the approach of specifying an “effective volume” term to simulate the effect of heat loss is adopted. In this approach, conditions prevailing in the core region are reproduced through the “effective volume”, Veff, which is a function of time. During the compression period, it is assumed that the piston initially accelerates at a constant rate and subsequently moves at constant speed. At the end of compression it is brought to stop by constant deceleration. An empirically-determined parameter, Vadd, is added to the actual time-dependent geometric volume of the combustion chamber, Vg, in order to match the simulated pressure at TDC with the experimental value. Here, Vadd simulates heat loss during compression. After the end of compression, t > 0, volume expansion is specified in terms of a polynomial fit, vp(t). Effective volume after compression is then taken as the product of effective volume at TDC (t = 0) and the fitted volume expansion term. Namely, Veff(t) = Vg(t) + Vadd
t≤0
Veff(t) = Veff(0) vp(t)
t > 0.
(4.3)
88
Therefore, Vadd and the polynomial function vp(t) are the key parameters of the present heat transfer model. For simulating the RCM experiment of a given reactive mixture, a non-reactive experiment is first carried out with a mixture of the same values of heat capacity, initial pressure, and initial temperature as for the reactive mixture. As such, the parameters required for the present heat transfer model, including Vadd and vp(t), can be determined from the pressure trace, P(t), of the non-reactive experiment. The empirically-determined parameters are then used for simulating the corresponding reactive case. Moreover, vp(t) is curve-fitted to a volume expansion trace, vexp(t): 1/ γ
⎛ P(0) ⎞ vexp (t ) = ⎜ ⎟ ⎝ P(t ) ⎠
t ≥ 0.
(4.4)
Numerical calculations are performed using the Sandia SENKIN code (Lutz et al., 1988) in conjunction with CHEMKIN (Kee et al., 1989). For modeling RCM, conditions specified in SENKIN calculations are those of a closed adiabatic system with volume specified as a function of time. Initial conditions of pressure, temperature, composition and the time-dependent effective volume, Veff(t), are specified in each SENKIN calculation. Since the effect of heat loss is taken into account by Veff(t), the computed time evolutions of temperature and species composition with detailed chemical kinetics can be compared with the experimental data. Governing equations for mass and energy conservation as used by SENKIN are as follows. The reacting mixture is treated as a closed system with no mass crossing the boundary, so the total mass of the mixture m = ∑k =1 mk is constant, and dm / dt = 0 . Here K
89
t is time, mk is the mass of the kth species and K is the total number of species in the mixture. The individual species are produced or destroyed according to dmk = Vω k W k dt
k= 1,2,…………..,K
(4.5)
where t is time, ωk is the molar production rate of the kth species, Wk is the molecular weight of the kth species, and V is the volume of the system, which may vary in time. Since the total mass is constant, this can be written in terms of the mass fractions as dYk = vω k Wk dt
k= 1,2,…………..,K
(4.6)
where Yk = mk/m is the mass fraction of the kth species and v = V/m is the specific volume. The First law of thermodynamics for a pure substance in an adiabatic, closed system states that de + pdv = 0 ,
(4.7)
where e is the internal energy per mass and p is the pressure. This relation holds for an ideal mixture of gases, with the internal energy of the mixture given by e = ∑k =1 ek Yk , K
where ek is the internal energy of the kth specie. Differentiating the internal energy of the mixture leads to the expression de = ∑k =1 ek dYk + ∑k =1 Yk dek K
K
.
(4.8)
Assuming thermally perfect gases, we write dek = cv,kdT , where T is the temperature of the mixture, and cv,k is the specific heat of the kth species evaluated at constant volume. Defining the mean specific heat of the mixture, c v = ∑k =1 cv ,k Yk and differentiating with K
respect to time, the energy equation (4.7) becomes
90
cv
dY dT dv K + ∑k =1 ek k + p = 0. dt dt dt
(4.9)
Substitution of the species production rate from equation (4.6) gives
cv
dT dv K + v ∑k =1 ek ω k Wk + p = 0. dt dt
(4.10)
The ideal gas equation of state is used to compute the pressure, p=
ρRT W
,
(4.11)
where R is the universal gas constant, W is the mean molecular weight of the mixture, and ρ is the mass density. When volume is provided as a function of time, the specific volume and its rate of change are v(t ) =
V (t ) dv 1 dV = and m dt m dt
.
(4.12)
The system of equations for SENKIN calculations for modeling RCM consists of equation (4.10) for the energy, and the K equations (4.6) for the species mass fractions.
4.3 Results and Discussion 4.3.1 Characterization Experiments: Inert Mixtures
Characterization experiments with inert gases are conducted using nitrogen and argon as the test gases to demonstrate the capability of the designed RCM. Figure 4.1 shows typical pressure profiles for both gases. An overlap of four experiments using nitrogen as the test gas and four experiments using argon are shown in Fig. 4.1. Pressure traces depict raw experimental data without any data processing. Time zero is taken as the end of the compression stroke, when pressure peaks. Operating conditions for the
91
experiments are also indicated in Fig. 4.1. Driving air pressure for these experiments is 150 psi. These inert gas tests identify several important features of the present RCM performance. (1) In Fig. 4.1, overlapping pressure traces follow each other exactly, indicating that highly repeatable experimental conditions are obtained on the designed RCM. (2) Pressure trace is smooth and free from any significant disturbances. This shows that the hydraulic stopping mechanism minimizes the vibrations caused by stopping of piston, and hence this RCM yields reliable pressure data. (3) Peak compressed pressures of up to 50 bar can be obtained routinely. (4) Compression process is very fast. It takes approximately 27 ms to travel 10 inch of the stroke. Average piston speed during compression is 9.3 m/s, with a peak speed of 12.7 m/s. (Procedure for deducing the piston velocity is outlined in Appendix B.) (5) Approximately 46% pressure rise occurs in the last 2 ms of compression stroke. Such a rapid pressure rise is desirable in order to minimize the possibility of chemical reaction during compression when using a reactive mixture. (6) Compression is much closer to the truly adiabatic compression for nitrogen gas in that the pressure at TDC (Pc) is closer to the adiabatic compression pressure (Pac). This is because heat loss is higher when using argon due to very high compressed gas temperature and its high thermal diffusivity. 4.3.2 Characterization Experiments: Reactive Mixtures
Reactive tests are conducted using stoichiometric iso-octane/oxygen/inert gas mixtures. The purpose of tests on iso-octane mixtures is to demonstrate the capability of
92
the designed machine in conducting experiments under reactive conditions and capturing the key features of auto-ignition phenomena. The molar composition of reactants is represented as 1.0 iC8H18+12.5 O2+47.0 (Ar+N2). Again, primary data from RCM are pressure-time traces. In order to ensure the reproducibility of experiments under reactive conditions, many experiments are conducted and compared under the same conditions. Figure 4.2 demonstrates such a comparison by overlapping pressure traces of seven reactive experiments with molar ratio of N2/(N2+Ar) being 0.6. It is seen from Fig. 4.2 that all the pressure traces closely follow each other. Figure 4.3 further illustrates the repeatability under the condition of two-stage ignition, in which N2/(N2+Ar)=0.8. Both first- and second-stage ignition delays match very well for all the experiments, as shown in Fig. 4.3. Steep rise in pressure at the end of the second-stage ignition indirectly indicates the uniformity of reacting mixture and homogenous ignition. Figure 4.4 compares pressure profiles for iso-octane auto-ignition at different compressed gas temperatures (Tc), while fixing P0=379.7 torr and T0=297 K for all the cases. Compressed gas temperature is varied by changing the composition of inert gases in the reacting mixture. Several interesting features of iso-octane auto-ignition can be observed in Fig. 4.4. At low compressed gas temperatures, ignition exhibits a two-stage phenomenon. As Tc is increased, the ignition delay associated with the first-stage ignition becomes smaller and eventually ignition becomes single stage. Furthermore, it is seen from Fig. 4.4 that energy release related to the first-stage ignition, represented by pressure rise, decreases with increasing Tc. This observation is in line with the existence of a cut-off temperature for the low temperature cycle of the two-stage ignition (Tanaka
93
et al., 2003a; Silke et al., 2005). When compressed gas temperature is low, the exothermic reactions associated with the first-stage ignition proceeds till this cut-off temperature is reached. With increasing compressed gas temperature, the heat release required to raise the mixture temperature to the cut-off temperature is less. As a result, heat release of the first-stage ignition decreases with increasing Tc. Figure 4.5 further compares the experimental results with slightly lower initial pressure of 331 torr, and shows the similar trend as Fig. 4.4 in auto-ignition with varying compressed gas temperatures. Figure 4.6 demonstrates the effect of pressure on ignition delay. For all the cases, initial temperature and composition are fixed, while initial pressure is varied. Although higher initial pressure results in higher compressed gas pressure, the change in the temperature at TDC is minor when keeping the clearance and stroke constant. Specifically, Tc varies slightly from 713 K to 718 K as initial pressure is increased from 324.7 torr to 503.4 torr. At higher pressures, thermal diffusivity of the gas decreases, resulting in less heat loss and slight increase in the temperature. In spite of small variation in temperature, the results shown in Fig. 4.6 primarily reflect the effect of pressure on ignition delay. In Fig. 4.6, while ignition delays decrease monotonically with increasing pressure, the phenomenon remains two-stage ignition. Figure 4.7 further plots ignition delay as a function of compressed gas temperature for three different charge densities corresponding to initial pressures of 324.7, 331, and 379.7 torr at T0=297 K. It is seen that at low and high Tc, ignition delay decreases with increasing Tc, whereas in the intermediate temperature range a zone of negative
94
temperature coefficient (NTC) exists such that ignition delay increases with increasing Tc. The above experimental results using reactive mixtures demonstrate the capability of the designed RCM in obtaining highly repeatable test data and capturing the essential features of combustion phenomena. During various reactive experiments, post-ignition pressure up to 350 bar is safely achieved in the present RCM.
4.3.3 Rapid Sampling Apparatus
As discussed previously, the rapid sampling apparatus enables the extraction of product mixture at specific post-compression time for species measurement. Such a determination of species concentrations allows the comparison with the detailed reaction mechanism to the next level of rigor, in addition to the validation of global response like ignition delay. In order to freeze the reaction at the desired instance, fast quenching of reacting mixture in the expansion tank is required. Experiments are conducted using this apparatus to characterize its performance in terms of ability for fast expansion. Further, the extracted sample is analyzed using a GC to quantitatively determine species concentrations. Representative pressure traces during quenching experiments are shown in Fig. 4.8. The solid line depicts the pressure trace if the diaphragm is not punctured, while the dashed line shows fast decay in pressure when diaphragm is punctured. These experiments are conducted for a CH4/O2/Ar mixture in the molar ratio of 1/2/24.2. Initial pressure and initial temperature are 320 torr and 296 K, respectively. Two instances of expansion at different post-compression times of 64.5 ms and 89 ms are demonstrated in
95
Fig. 4.8, showing very fast decay in pressure. Such a fast cooling rate is expected to freeze the composition of the gaseous products. The mixture composition collected in the expansion tank is then analyzed by using a GC. Extracted products are separated using a 30 m Hayesep DB column along with a 13X molecular sieve. Quantification is carried out using a flame ionization detector (FID) for organic compounds and a helium ionization detector (HID) for organic compounds as well as permanent gases. Figure 4.9 shows HID chromatogram for samples extracted at post-compression times of 64.5 and 89 ms. At 64.5 ms, there is no change in CH4 concentration and hence major product species such as CO2 are not detected. At 89 ms, part of CH4 has been oxidized and CO2 is present in the extracted mixture. Since any H2O present in the extracted mixture is adsorbed in the GC column, it is not identified in the analysis. In this way, by quenching the experiment at different post-compression times, time variation of specie concentrations during the experiment can be obtained.
4.3.4 Modeling Results
Figure 4.10 shows a comparison of experimental pressure traces (solid line) and model predictions (dashed line) for the RCM runs using pure N2 and Ar as a test gas. Computed profile is seen to closely follow the experimental pressure trace. Figure 4.11 shows corresponding volume expansion trace vexp(t) for N2 and Ar determined by using Equation 4.4 and the pressure traces of Fig. 4.10. The corresponding polynomial function, vp(t), which is curve fitted to vexp(t), is also shown in Fig. 4.11. The associated coefficients for vp(t) are tabulated in Table 4.1.
96
For the case of reactive mixture, a non-reactive experiment with the same mixture heat capacity is first carried out to determine the parameters of heat transfer model. Figure 4.12 compares experimental pressure traces (solid line) and model prediction (dashed line) for a reactive iso-octane mixture. Experimental conditions are indicated in Fig. 4.12. Detailed kinetic mechanism of iso-octane oxidation is taken from Curran et al. (2002). Experimental and computed pressure profiles for the corresponding non-reactive mixture are also shown in Fig. 4.12. For the non-reactive case, experimental and simulated pressure traces are seen to match exactly. However, there exists large discrepancy for the reactive case. Especially, the experimental ignition delay is much shorter than the computed result, as shown in Fig. 4.12. This discrepancy is attributed to the deficiency of the reaction mechanism under the conditions of relatively low compressed temperature of 684 K. Figure 4.13 further compares experimental and predicted pressure traces for the isooctane mixture under different conditions. Experimental pressure profile is shown with solid line and numerical prediction with dashed line. Unlike Fig. 4.12, it is seen from Fig. 4.13 that the computed ignition delay is now smaller than the experimental value at higher compressed temperatures. It is also noted that at Tc=782 K, experimental pressure profile exhibits single-stage ignition, while computed result shows a two-stage ignition phenomenon. Since the two-stage ignition phenomenon ceases to exist beyond the NTC region, the existence of two-stage ignition for the computed response at 782 K suggests that the iso-octane mechanism of Curran et al. (2002) predicts the NTC region at a higher temperature than what is observed in the experiment. At Tc=878 K, both experimental and computed pressure traces show a single-stage ignition phenomenon, indicating that
97
prediction using the mechanism of Curran et al. (2002) at this compressed temperature has crossed the NTC region. Figure 4.14 further compares the variation of ignition delay with compressed temperature for both experimental and computational results. Such a comparison clearly demonstrates the shifting of the predicted NTC region and the higher cut-off temperature for the first-stage ignition obtained from the mechanism of Curran et al. (2002). In addition, within the NTC region, the computed ignition delays are noted to be significantly smaller than the experimental results. Since the oxidation of hydrogen has been extensively studied and is considered to be well-established, additional experiments are conducted using a H2/O2/N2/Ar mixture in molar ratio of 2/1/2.9/10.1. Figure 4.15 compares experimental and computed pressure traces under varying conditions. Three recent hydrogen oxidation mechanisms of Li et al. (2004), O’Conaire et al. (2004), and Davis et al. (2005) are employed in the simulation. In Fig. 4.15, experimental pressure traces for the corresponding non-reactive mixture are also plotted as a reference. Differences between predictive capabilities of different mechanisms under the tested conditions can be noticed. However, in contrast to the isooctane cases, agreement of experiments with simulated results is seen to be much better. Particularly, the mechanism of O’Conaire et al. (2004) is seen to adequately predict the experimental data for the conditions tested. Further investigation over a wide range of pressures and temperatures is warranted, which is an emphasis of Chapter 5. 4.4 Summary
Characterization experiments are conducted to demonstrate the suitability of the designed machine for combustion related research.
98
Experiments using inert gases and reactive mixtures show that the present RCM is capable of providing highly repeatable data. Using this facility, ignition delay measurements have been conducted for various fuel mixtures, which capture the essential combustion characteristics. The robust mechanical design allows the compressed gas pressure at TDC in excess of 50 bar. Utility of a rapid sampling apparatus in rapidly expanding the reacting gases and thereby freezing the reaction composition for the subsequent quantitative analysis is also demonstrated. All these features make the designed RCM an excellent tool for studying combustion phenomena at elevated pressure/temperature conditions. A numerical model accounting for heat loss is developed to simulate experimental RCM data with detailed chemistry in order to assess the “comprehensiveness” of the detailed kinetic mechanism employed. Since the present numerical model is shown to appropriately predict the performance of the machine for non-reacting mixtures, the comparison of the experimental results using reacting mixtures and the simulated results with detailed chemistry can illustrate the predictive capability of the existing mechanisms.
99
50
Pressure (bar)
40 30
Argon T0 = 297 K P0 = 600 torr Tc = 1509 K Pc = 46.6 bar Tac = 1806 K Pac = 73 bar
20 10 0
-20
0
20 40 Time (ms)
Nitrogen T0 = 297 K P0 = 1000 torr Tc = 800 K Pc = 47.5 bar Tac = 838 K Pac = 56.5 bar
60
80
100
Fig. 4.1 - Typical pressure traces for inert gas tests, demonstrating experimental repeatability. Subscripts: 0 - initial condition; c - condition at TDC; ac – condition at TDC for a truly adiabatic compression.
100
40 35 P0 = 331 torr T0 = 297 K Tc = 745 K N2 /(Ar+N2) = 0.6
Pressure (bar)
30 25 20 15 10 5 0
-20
0
20
40 60 Time (ms)
80
100
Fig. 4.2 - Demonstration of experimental repeatability for iso-octane auto-ignition. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/28.2/18.8. Conditions: P0 = 331 torr; T0 = 297 K; Tc = 745 K.
20 First-stage ignition
Pressure (bar)
15
10 P0 = 324.7 torr T0 = 297 K Tc = 713 K N2 /(Ar+N2) = 0.8
5
0
-20
0
20 40 Time (ms)
Second-stage ignition
60
80
Fig. 4.3 - Demonstration of experimental repeatability for iso-octane auto-ignition under conditions of two-stage ignition. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/37.6/9.4. Conditions: P0 = 324.7 torr; T0 = 297 K; Tc = 713 K.
101
25 746.6
714.6
20
Tc=880 K
Pressure (bar)
685.5
783.7
828
15 10 P0 = 379.7 torr T0 = 297 K
5 0
-20
0
20 Time (ms)
40
60
Fig. 4.4 - Pressure profiles at different compressed gas temperatures for iso-octane autoignition. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 379.7 torr; T0 = 297 K. Compressed gas temperature is varied by changing the composition of inert gases in the mixture.
102
20
Pressure (bar)
15
826
713
Tc=878 K 684
745
782
10 P0 = 331 torr T0 = 297 K
5
0
-20
0
20
40 60 Time (ms)
80
100
120
Fig. 4.5 - Pressure profiles at different compressed gas temperatures for iso-octane autoignition. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Compressed gas temperature is varied by changing the composition of inert gases in the mixture.
103
P0=503.4 torr
25
460.8 379.7
Pressure (bar)
331
20 324.7
15 10 T0 = 297 K N2/(Ar+N2) = 0.8
5 0
-20
0
20 40 Time (ms)
60
80
Fig. 4.6 - Effect of pressure on auto-ignition of iso-octane mixtures. Molar composition: iC8H18/O2/N2/Ar = 1/12.5/37.6/9.4. Initial temperature is 297 K. Pressure at TDC is varied by changing the initial pressure.
324.7
Ignition Delay (ms)
120 331
100 80 60
P0=379.7 torr
40 20
700
750
800
850
900
Adiabatic Core Temperature (K) Fig. 4.7 - Ignition delay of iso-octane mixture verses adiabatic core temperature. Molar composition: iC8H18/O2/inert = 1/12.5/47. Adiabatic core temperature is varied by changing the composition of inert gases in the mixture.
104
25 P0 = 320 torr T0 = 296 K
Pressure (bar)
20 15
64.5 ms
89 ms
80
100
10 5 0
-20
0
20
40 60 Time (ms)
Fig. 4.8 - Pressure traces during quenching experiments using a rapid sampling apparatus. Molar composition: CH4/O2/Ar = 1/2/24.2. Initial conditions: P0 = 320 torr; T0 = 296 K.
105
4000 Time: 64.5 ms Ar O2 Signal (A.U.)
CH4
0
4000
Ar Time: 89 ms
Signal (A.U.)
O2
CH4 CO2
0 0
2
4
6
8
10
Retention Time (min)
Fig. 4.9 - HID chromatogram of methane auto-ignition at two different quenching times using Hayesep DB column. Test conditions are the same as those in Fig. 4.8
106
50
Pressure (bar)
40 30 Argon T0 = 297 K P0= 600 torr
20
Nitrogen T0 = 297 K P0= 1000 torr
10 0
-20
0
20
40 60 Time (ms)
80
100
120
140
Fig. 4.10 - Comparison of experimental pressure traces for pure Ar and pure N2 with model predictions. Initial conditions: argon – P0 = 600 torr; T0 = 297 K, nitrogen – P0 = 1000 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction.
107
1.5 1.4 Argon
vexp(t)
1.3
Nitrogen
1.2 1.1 1
0
20
40
60 80 Time (ms)
100
120
140
Fig. 4.11 – Volume expansion trace, vexp(t), (solid line) and corresponding curve-fitted polynomial function, vp(t), (dotted line) for N2 and Ar determined by using the pressure traces of Fig. 4.10 Coefficients
Argon
Nitrogen
a0
1.009
1.011
a1
1.380x10-2
5.649 x10-3
a2
- 3.895 x10-4
-1.215 x10-4
a3
7.197 x10-6
1.864 x10-6
a4
- 7.419 x10-8
- 1.690 x10-8
a5
3.908 x10-10
8.099 x10-11
a6
-8.184 x10-13
-1.574 x10-13
Table 4.1 – Coefficient for vp(t) for curve fits shown in Fig. 4.11. vp(t)= a0 + a1 t+ a2 t2 + a3 t3 + a4 t4 + a5 t5+ a6 t6
108
20 Experimental
Reactive
Pressure (bar)
15 Model
10 P0 = 331 torr T0 = 297 K Tc = 684 K N2/(Ar+N2) = 1.0
5
0
-20
0
20
Non-Reactive Experimental and Model
40 60 80 Time (ms)
100
120
140
Fig. 4.12 - Comparison of experimental pressure profile for stoichiometric iso-octane mixture with numerical prediction. Molar composition: iC8H18/O2/N2 = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction. Comparison for the corresponding non-reactive mixture is also shown.
20 Tc = 878 K Tc = 782 K
Pressure (bar)
15
10 P0 = 331 torr T0 = 297 K
5
0
-20
0
20
40 60 Time (ms)
80
100
120
Fig. 4.13 - Comparison of experimental pressure profile for stoichiometric iso-octane mixture with numerical prediction. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K. Solid line – experimental pressure. Dashed line – model prediction.
109
140
Ignition Delay (ms)
120
Experimental
100 80 60
Computed
40 20 0
700
750
800
850
900
Adiabatic Core Temperature (K) Fig. 4.14 - Comparison of experimental and computed ignition delays with varying adiabatic core temperatures. Molar composition: iC8H18/O2/inert = 1/12.5/47. Initial conditions: P0 = 331 torr; T0 = 297 K.
50
Pressure (bar)
40
P0= 705.8 torr Tc= 977K
P0 = 640 torr Tc = 1010.5 K
30 T0 = 298 K
20
Experimental O'Conaire et al. (2004)
10
Davis et al. (2005) Li et al. (2004)
0 -15
-10
-5
0 5 Time (ms)
10
15
20
Fig. 4.15 - Comparison of experimental pressure profile for stoichiometric hydrogen mixture with numerical prediction. Molar composition: H2/O2/N2/Ar = 2/1/2.9/10.1. Pressure traces for the corresponding non-reactive cases are also shown.
110
CHAPTER 5 AUTOIGNITION INVESTIGATIONS OF H2/CO SYSTEM 5.1 Importance of Hydrogen-Carbon Monoxide Oxidation Over the history of combustion studies, considerable efforts have been expended in studying the combustion characteristics of the hydrogen-carbon monoxide-oxygen system, motivated primarily by its essential role in the hierarchical structure of oxidation models of hydrocarbon fuels. In the hierarchical development of mechanisms, it generally begins with the simplest reacting system and systematically incorporates more species and reactions in increasing order of complexity. At each level, the developed mechanism is tested against experimental data base. In the hierarchical structure of comprehensive mechanisms of complex hydrocarbons, H2/CO/O2 system is the simplest fundamental sub-mechanism. Therefore, characterization of H2/O2 and H2/CO/O2 systems is taken as of primary importance in combustion kinetics. Recently, the prospect of a hydrogen-based economy has prompted increased interest in the use of hydrogen as a fuel given its high chemical energy per unit mass and cleanliness (O’Conaire et al., 2004). It appears that most of the technological problems in using hydrogen in spark-ignited internal combustion engines, including NOx emissions (Heffel et al., 2003), have now been solved; vehicular on-board storage is probably the one remaining difficulty (Hirscher et al., 2003). As a result, there is continued interest in developing a better understanding of the hydrogen oxidation at conditions of high pressure relevant to engine combustion.
111
Furthermore, investigation of H2/CO oxidation is relevant in its own right. In recent years, motivated by a push towards more efficient and environment friendly coalbased power generation, significant research efforts have been directed towards development of Integrated Gasification Combined Cycle (IGCC) power generation systems. IGCC is a promising process, which uses coal and other solid or liquid fossil fuels of inferior quality in a clean, economical, and environmentally efficient manner to produce power. IGCC combines two technologies – gasification and combined cycle – to offer the potential to achieve the environmental benefits of gas-fired generation with the thermal performance of a combined-cycle plant, yet with the low fuel cost associated with coal. Gasification is the process by which coal is partially combusted with oxygen and steam to form what is commonly called syngas (primarily a blend of H2 and CO). The syngas then is cleaned to remove particulates and other polluting elements for the use in the combustor. The environmental benefits of IGCC stem from the capability to cleanse as much as 99 percent of the pollutant-forming impurities from coal-derived gases, which are much difficult to remove when coal is directly used for combustion in coal-fired combustors. Efficiency benefits of IGCC come from the use a ‘combined cycle’ power plant. In a typical coal combustion plant, heat from burning coal is used to boil water, making steam that drives a steam turbine-generator. Only a third of the energy value of coal is actually converted into electricity by most combustion plants, the rest is lost as waste heat. An IGCC power plant, however, typically gets dual duty from the gases it produces. First, the syngas, cleaned of their impurities, are fired in a gas turbine - much like natural gas - to generate one source of electricity. The hot exhaust of the gas turbine
112
is then used to generate steam for a more conventional steam turbine-generator. This dual source of electric power, called a combined cycle, converts much more of coal's inherent energy value into useable electricity. The fuel efficiency of an IGCC power plant can be boosted to 50 percent or more. Composition of syngas derived from various sources and varieties of fossil fuels can vary a lot and there is practical interest in developing a better understanding of the kinetics of syngas for optimization and safety of combustors. Also, H2/CO mixture obtained from reforming of hydrocarbon fuels, known as reformer gas, has been proposed as a means of reducing cold start emission of an SI engine (Sung et al., 2001; Huang et al., 2002).
5.2 Prior Work on Hydrogen-Carbon Monoxide Oxidation Since the focus of present work is on studying homogeneous chemistry, work done with non-homogeneous systems, such as flames, is not discussed here. Earlier work on the H2/CO/O2 system has been reviewed in detail by Dixon-Lewis and Williams (1977) up through 1975 literature, by Gardiner and Olson (1980) through 1979, and subsequently by Westbrook and Dryer (1984) and Warnatz (1984). Oxidation mechanism of H2 is very sensitive to the physical conditions and can be divided into three regimes, depending upon the nature of radical pool and predominant chemistry. At high temperature and low pressure, radical pool is dominated by O, H, and OH radicals and the controlling reactions are of chain branching. Whereas at low temperature and high pressure, radical pool is dominated by HO2 and H2O2, and chemistry involving HO2 and
113
H2O2 becomes important. In the intermediate regime, concentrations of O, H, OH, HO2, and H2O2 are of the same order of magnitude.
5.2.1 H2/O2 System Autoignition and explosion characteristics of hydrogen-oxygen mixtures have been explored extensively. Figure 5.1 shows the well-known inverted “S” shaped curve, which divides the explosive and nonexplosive regions (Glassman, 1996). The second explosion limit is defined by the competition of the chain-branching and chainrecombination pathways, i.e. H + O2 → OH + O (Reaction 1) and H + O2 + M → HO2+M (Reaction 9) as the total concentration [M] = 2k1/k9, where M is the third body participating in Reaction 9, [M] represents the total concentration, and k1 and k9 are rate constants for Reaction 1 and Reaction 9 respectively. At higher pressures, branching involving the formation and decomposition of H2O2 characterizes the third explosion limit. Within the explosive region, Saytzev and Soloukin (1962) and Strehlow and Cohen (1962), first noted that ignition regime can be separated in weak and strong ignition. At a constant pressure, there was a sharp increase in the ignition delay, with a decreased temperature at the extension of the second limit (cf. Fig. 5.1). Ignition at lower temperature was termed as ‘weak’ ignition and that at higher temperature as ‘strong’ ignition. Voevodsky and Soloukhin (1965) explained the change controlling chemistry from change-branching to chain-propagation at the extended second limit as the reason for change in character of ignition form strong to weak. Meyer and Oppenheim (1970) presented a high-speed schlieren film, which showed different nature of ignition in the
114
‘weak’ and strong’ regimes. Kordylewski and Scott (1984) endeavored to account for the influence of heat release at pressure above the second limit and suggested that the energy release during the induction time becomes significant. It was numerically shown (Kordylewski and Scott, 1984) that whereas at second explosion limit the extent of selfheating is negligible and ignition is controlled by relative rate of chain-branching and chain-termination, as pressure is increased to third limit, the interaction between the thermal feedback and chain-branching determines the condition of ignition. Yetter et al. (1991c) showed by doing eigenvalue analysis that the transition from a weak to strong ignition corresponds to a change in the controlling reactions from reactions involving HO2 and H2O2 to the H+O2 branching reaction. Previous experimental studies using shock tubes mostly determined autoignition times in the low-pressure, high-temperature, strong explosion region. Schott and Kinsey (1958) measured ignition delay times of two H2/O2/Ar fuel mixtures behind incident shock waves over the temperature range 1085-2700 K at 1 atm. Skinner and Ringrose (1965) extended the range of temperatures to 965–1076 K at reflected shock pressure of 5 bar, determining a change in activation energy as the state of the mixture crosses the extended second limit. Craig (1966) measured ignition delay of H2/O2/N2 mixtures near the extended second explosion limit, in the pressure range of 1-2 atm. Bhaskaran (1973) measured ignition delays at pressure of 2.5 atm in the temperature range 800 to 1400 K. Slack (1977) studied stoichiometric hydrogen–air mixtures in a shock tube and measured induction times near the second explosion limit at a reflected shock pressure of 2 atm in the temperature range of 980–1176 K. Cheng and Oppenheim (1984) obtained ignition delay times in the range of temperature from 800 to 2400 K and pressures from 1 to 3
115
bar. Petersen et al. (2003) measured ignition delay times in three highly dilute H2/O2/Ar mixtures at temperatures of 1010–1750 K, around atmospheric pressure. Amongst few investigations at high pressure, Petersen et al. (1996) investigated the ignition delay time of dilute H2-O2 in a shock tube at pressures from 33 to 87 atm and temperatures between 1189–1876 K. Lee and Hochgreb (1997) investigated hydrogen autoignition above the extended second explosion limit using a rapid compression machine, in the pressure and temperature range of 6 to 40 bar and 950-1050 K. Subsequently, Mueller et al. (1999a) investigated H2/O2 reaction under dilute conditions using a variable pressure flow reactor over pressure and temperature ranges of 0.3-15.7 atm and 850-1040 K and developed a kinetic mechanism. Overall data set consisted primarily of measured profiles of H2, O2, and H2O. Their data spanned the explosion limit behavior and placed significant emphasis on HO2 and H2O2 kinetics above the extended second limit. Results showed that the extended second limit is an important boundary in H2/O2 kinetics. Near this limit, small increase in pressure can result in more than a two orders of magnitude reduction in reaction rate.
5.2.2 H2/CO/O2 System Early experimental work on the oxidation of CO was confused by the presence of any hydrogen-containing impurity. The rate of CO oxidation in the presence of species such as water is substantially faster than the “bone dry” condition. Even very small quantities of hydrogen, of the order of 20 ppm, will substantially increase the rate of CO oxidation (Brokaw, 1967). As a result, it is now well-established that the hydrogen oxidation mechanism is invariably imposed on the CO oxidation mechanism.
116
Consequently, most of CO oxidation studies have been done by carefully adding a controlled amount of H2 or H2O and examining the effects of varying its concentration. CO oxidation with addition of hydrogen or water is termed as ‘wet’ oxidation. In addition, explosion limits of dry CO are neither well defined nor reproducible, principally because of the difference in dryness in the various experiments determining the limits (Glassman, 1996). Experimental results at atmospheric pressure show that CO oxidation increases with increasing H2O concentration. However a dilution effect is observed in which increasing amounts of H2O addition has a smaller effect on the acceleration of the reaction (Yetter et al., 1991b). With increased H2O addition, second explosion limit for wet CO oxidation is predicted to coincide with the second explosion limit of H2/O2 mixtures (Kim et al., 1994). Thus, H2/O2 submechanism controls the explosion under such conditions. However a third limit for CO oxidation has not been defined. Figure 5.2 shows typical explosion limits for wet CO oxidation which separates explosive and nonexplosive regions. Early contributions to the kinetics of H2/CO oxidation were based on shock tube experiments (e.g. Brabbs et al., 1970; Gardiner et al., 1971, 1972; Dean et al., 1978). In these experiments, attention was usually directed to the exponential growth zone of the explosion. In particular, Brabbs et al. (1970) analytically studied H2/CO/O2 system using the exponential growth constant of blue CO flame band emission behind incident shock. Measurements were in the temperature and pressure range of 1160-1900 K and 1-2 atm. Gardiner et al. (1971, 1972) conducted shock tube experiment in the temperature range of 1400 to 2500 K and pressure less than 0.3 atm. Observations were made for OH radical profiles by absorption spectroscopy, CO2 profiles by thermal emission intensity, and
117
[CO][O] profiles by CO flame spectrum emission, and elementary reaction rates were deduced. Experiments of Dean et al. (1978) consisted of measuring emissions at 450 nm ([CO][O]) and 4.27 µm ([CO2]) when a variety of mixture containing H2, CO, O2 or N2O, and Ar were heated behind reflected shock waves to temperatures of 2000 – 2850 K and total concentrations near 5x1018 molecules/cc. Lyon and Hardy (1980, 1985) studied oxidation kinetics of wet CO in trace concentrations in a flow reactor at temperatures from 1123 K to 1298 K and pressure of 1.22 atm to 2.44 atm. Moist CO oxidation has been extensively studied in an atmospheric pressure flow reactor (Yetter et al., 1991a, 1991b, 1992; Roesler et al., 1994) and a variable presure flow reactor (Kim et al., 1994) at Princeton University at temperature 850 to 1200 K and pressure up to 9.6 bar. Yetter et al. (1991b) obtained kinetic data for CO/H2O/O2 oxidation in the temperature range of 852-1138 K and at 1 atmosphere pressure. In particular, these results defined explosion limit temperatures for near stoichiometric mixtures of CO/H2O/O2 and H2/O2 systems at atmospheric pressure and demonstrated the complex relationship between water vapor concentration and the carbon monoxide oxidation rate (Yetter et al., 1991b). A comprehensive reaction mechanism for the oxidation of carbon monoxide in the presence of hydrogen for pressures between 0.3 and 2.2 atm and temperature range of 823-2870 K was also described (Yetter et al., 1991a). Kim et al. (1994) investigated the kinetics of moist CO oxidation for pressures from 1 to 9.6 atm, temperatures from 960 to 1200K, and equivalence ratios from 0.33 to 2.1 in a variable pressure flow reactor. This set of data was used in combination with previous experimental data to develop a revised comprehensive reaction mechanism for the CO/H2O/O2 system (Kim et al., 1994). Roesler et al. (1994) showed complex
118
dependence of rate of moist CO oxidation on O2 concentration at atmospheric pressure. It was experimentally observed that at temperature of 1000 K and atmospheric pressure, counter-intuitively, moist CO oxidation is inhibited by oxygen addition. Recently, Sivaramakrishna et al. (2005) studied the oxidation of H2/CO mixture in a high pressure shock tube facility at pressures ranging from 25 to 600 bar. Experiments were conducted under highly diluted conditions with total fuel concentration being less than 650 ppm. Their experimental data showed that CO oxidation is inhibited as pressure is increased. Numerical simulations (Sivaramakrishna et al., 2005) showed that GRIMech 3.0 (Smith et al., 1999) failed to predict the behavior of the combustion system, whereas the trend was well described by the mechanism of Davis et al. (2005).However, the performance of the kinetic model of Davis et al. deteriorated at higher pressure. From literature survey for the homogeneous kinetics investigation on H2/CO system, it is evident that numerous studies have been conducted under conditions of low pressure and high temperature. Hence, the mechanism of chain-branching reactions, which occurs under these conditions, is well understood. On the other hand, investigations at high pressure, which are more relevant to practical combustors and internal combustion engines, are meager. It is further noted that under the conditions of high pressure and low temperature, the reactions involving formation and consumption of HO2 and H2O2 are expected to become significantly important. Experimental data in the relevant pressure and temperature ranges are therefore needed to scrutinize the existing rate constants of the reactions related to HO2 and H2O2. Recognizing the gap in experimental investigations for H2/CO oxidation at high pressures and low temperatures, the present study aims at investigating the associated
119
kinetics using a rapid compression machine. Experiments are conducted for homogeneous H2/CO mixtures over a wide range of pressure, equivalence ratio, and mixture compositions. Experimental results are subsequently compared with calculations using various detailed reaction mechanisms. Kinetic analysis is also carried out to delineate the controlling chemistry and identify the key elementary reactions that require further study and refinement in order to match the experimental data.
5.3 Experimental Specifications Autoignition investigations for H2/CO/O2/N2/Ar mixtures are conducted in temperature range of 950-1100 K, pressure range of 15-50 bar, and equivalence ratio ranging from 0.36 to 1.6. All the gases used in experiments are of ultra high purity - H2: 99.999%, CO: 99.99%, O2: 99.993%, N2: 99.999%, Ar: 99.995%. Further, mixture composition is varied by changing the relative proportion of H2 and CO in the reacting mixture. For a given mixture composition, compressed gas temperature at the end of compression (top dead center, TDC) is varied by altering the compression ratio, whereas the desired pressure at TDC is obtained by varying initial pressure of the reacting mixture. The majority of the experiments are conducted with the compositions of gas mixtures as tabulated in Table 5.1. RCO is defined as the mole percentage of CO in the combined H2/CO fuel mixture, and is expressed as RCO=FCO/(FH2+FCO), where FCO and FH2 are respectively the mole fractions of CO and H2 in the fuel mixture. Mixtures #1 to #5 have the global equivalence ratio (φ) of unity and keep the concentrations of diluents constant, while RCO varies from 0 to 0.80. For Mixtures #6 to #10, the mole fractions of
120
H2 and CO are kept constant such that RCO is fixed at 0.25, while the global equivalence ratio is varied from 0.36 to 1.6 by changing the amount of oxygen in the total mixture. In addition, for Mixtures #6 to #10 the mole fractions of N2 and Ar are adjusted accordingly to maintain the same level of mixture heat capacity in order to yield the same temperature and pressure at TDC when the same initial pressure and temperature are used.
Mixture #
φ
RCO
H2
CO
O2
N2
Ar
1
1.0
0
12.5%
0%
6.25%
18.125%
63.125%
2
1.0
0.25
9.375%
3.125%
6.25%
18.125%
63.125%
3
1.0
0.50
6.25%
6.25%
6.25%
18.125%
63.125%
4
1.0
0.65
4.375%
8.125%
6.25%
18.125%
63.125%
5
1.0
0.80
2.5%
10%
6.25%
18.125%
63.125%
6
0.36
0.25
6.667%
2.222%
12.345%
14.418%
64.348%
7
0.72
0.25
6.667%
2.222%
6.173%
21.586%
63.352%
8
1.0
0.25
6.667%
2.222%
4.444%
23.600%
63.067%
9
1.3
0.25
6.667%
2.222%
3.419%
24.782%
62.910%
10
1.6
0.25
6.667%
2.222%
2.777%
25.511%
62.823%
Table 5.1 – Compositions of gas mixtures investigated.
5.4 Kinetic Models Experimental results are compared with simulations using various detailed mechanisms available in the literature. Specifically, three recent hydrogen and
121
hydrogen/carbon monoxide oxidation mechanisms of Li et al. (2004b), O’Conaire et al. (2004), and Davis et al. (2005) are employed in the simulation. In addition, calculations are also performed using GRI Mech 3.0 (Smith et al., 1999). It is further noted that the mechanism of Li et al. (2004b) is developed for C1 species combustion in a hierarchical manner. The H2 oxidation subset of this mechanism is validated for a wide range of experimental conditions: temperature ranging from 298 K to 3000 K, pressure from 0.3 atm to 87 atm, and equivalence ratio from 0.25 to 5.0 (Li et al., 2004a) whereas the H2/CO subset is validated up to a maximum pressure of 9.6 bar (Li et al., 2004b). The mechanism of O’Conaire et al. (2004) is for H2 oxidation and consists of 19 reactions, which is validated in the temperature range from 298 K to 2700 K, pressures from 0.05 atm to 87 atm, and equivalence ratios from 0.2 to 6.0 (O’Conaire et al., 2004). The mechanism of Davis et al. (2005) is an optimized version for H2/CO combustion and is validated against similar experimental dataset as the mechanism of Li et al. (2004b). GRI-Mech 3.0 is the latest version of GRI mechanisms and is optimized for natural gas combustion in the range of 1000-2500 K, 10 Torr to 10 atm, and equivalence ratios from 0.1 to 5.0.
5.5 Results and Discussion Figure 5.3 shows an example of a typical pressure trace for ignition of an H2/CO/O2/N2/Ar mixture. Mixture #2 is used here with initial temperature (T0) of 298.7 K and initial pressure (P0) of 661 torr. The pressure and temperature at the end of compression are TC=999 K and PC=30 bar, respectively. Overlapping pressure traces of seven experiments conducted under similar conditions are also compared in Fig. 5.3,
122
demonstrating that all the pressure traces closely follow each other and experimental data is highly repeatable It is noted from Fig. 5.3 that the compression stroke is of the order of 30 ms, and the ignition delay (τ) is defined as the time from the end of the compression stroke, where the pressure peaks, to the rapid rise in pressure due to ignition.
5.5.1
Comparison with Previous Studies In order to illustrate the type and quality of data attainable from the RCM,
experiments are first conducted and compared with the published data in the literature. Since extensive experimental data are available on H2 autoignition, these experiments are carried out for the case of pure H2, without any CO addition. Figure 5.4 is a plot of ignition delay times from various previous studies for H2 ignition, scaled to first order with oxygen concentration, versus the reciprocal of temperature. Three sets of the present data in the pressure range of 29.6 to 33.6 atm are also included. For plotting RCM data, temperature and pressure conditions at TDC, i.e. TC and PC, are taken as the reference conditions. It is seen from Fig. 5.4 that data basically follow two trends depending upon the conditions of pressure and temperature. At high temperature and low pressures, the effective activation energy is lower, represented by the lower gradient of the dataset in Fig. 5.4. At relatively low temperatures and high pressures, however, dataset follows a steeper variation, which is indicative of higher effective activation energy. This transition from shallower slope to steeper slope is pressure dependent and is related to change in the controlling chemistry from chain-branching regime at higher temperatures and low pressures to a chain-propagation regime dominated by HO2 chemistry at higher pressures and low temperatures.
123
We also note that the present measurements shown in Fig. 5.4 lie in the regime above the extended second explosion limit, and are consistent with the data of Lee and Hochgreb (1998b). Further comparing the data obtained on the present RCM with the shock tube measurements of Petersen et al. (1996) at 33 atm, it is seen that the present RCM results appear to lie on an extension of data obtained by Petersen et al. (1996) and follow a similar slope. From this comparison, we demonstrate that measurements from the present RCM are well characterized for chemical kinetics studies. In addition, the RCM data are shown to complement the data obtained from shock tubes, as RCM provides a means of experimental investigation in the intermediate temperature range, under which the associated ignition delay could be too long to be captured in shock tubes.
5.5.2
Autoignition of Hydrogen Figure 5.5 presents the measured ignition delay times for H2 at a pressure of 30 bar
at TDC as a function of inverse of temperature. Calculated ignition delays using various kinetic mechanisms are also shown and compared. All the mechanisms considered follow the same trend with temperature as the experimental observation. Particularly, the mechanism of O’Conaire et al. (2004) shows the best agreement with the present experimental data. The Mechanism by Li et al. (2004b) predicts slightly higher ignition delays. Whereas, the predicted ignition delays by using the mechanism of Davis et al. (2005) and GRI-Mech 3.0 are even higher. Further, Fig. 5.6 presents similar comparisons of experimental and calculated ignition delays at PC of 15 and 50 bar. At 50 bar, comparison of calculated and measured values shows a similar trend as at 30 bar. However, predictions using the mechanism of
124
O’Conaire et al. (2004) become worse towards lower temperatures, where predicted values are higher than the experimental data. For PC=15 bar, except for GRI-Mech 3.0, all mechanisms predict a lower ignition delay than experimental measurements at high temperatures and a higher ignition delay at low temperatures. Thus, at 15 bar calculated results show a steeper slope than the slope of the measured values. In general, the mechanism of O’Conaire et al. (2004) is still seen to perform better than other mechanisms. These comparisons suggest that some further refinements can still be made to the existing kinetic mechanism for H2 oxidation in order to improve its predictions under the conditions of intermediate temperature and high pressure, where HO2 and H2O2 chemistry becomes important.
5.5.3
Autoignition of Hydrogen/Carbon Monoxide Mixtures Figure 5.7 presents the measured ignition delay times for Mixtures #1 to #5 at a
pressure of 50 bar at TDC. These tests demonstrate the effect of CO addition by increasing RCO, while the global equivalence ratio and total fuel mole fraction in the mixture are kept constant. Clearly, even at RCO of 0.25, CO addition is shown to inhibit the ignition in comparison to pure H2 and result in longer ignition delays. A further increase in CO mole fraction results in a monotonous increase in ignition delay. Figures 5.8 and 5.9 plot similar experimental results, but at lower pressures of 30 bar and 15 bar, respectively. At all the pressures investigated, the inhibition effect of CO addition is observed. Comparison of Figs. 5.7-5.9 further illustrates that the inhibition effect is much more pronounced as pressure is increased. In Fig. 5.9, at a relatively lower pressure of 15 bar, ignition delay lines for increasing amount of CO addition are closer to
125
each other. Whereas, at 30 bar (Fig. 5.8), ignition delay lines are relatively far apart. At a higher pressure of 50 bar (Fig. 5.7), ignition delay lines are farther apart than at 30 bar. However, the incremental change in the enhancement of inhibition effect in varying from 30 bar to 50 bar is much less than the change in varying from 15 bar to 30 bar. Increase in inhibition effect of CO addition at higher pressure indicates that CO oxidation is inhibited at higher pressure. This finding is consistent with the recent observations of Sivaramakrishna et al. (2005). Figure 5.10 plots the experimental results as a function of RCO at 30 bar and 1010.5 K, showing the inhibition effect of CO addition in the H2/CO mixture, together with the simulated results using various mechanisms. Large discrepancies between experimental data and calculated values can be seen. Predictions using the mechanisms of Li et al. (2004b) and Davis et al. (2005) are quite close to each other. However, from either of these two mechanisms, significant inhibition effect of CO is not observed until RCO reaches 0.80. Predictions using GRI-Mech 3.0 even show the opposite trend of reduction in ignition delay with CO addition and subsequent increase in ignition delay for large CO percentage. Figures 5.11 and 5.12 show similar comparisons at PC/TC of 15 bar/1028.5 K and 50 bar/1044 K, respectively. At 15 bar (Fig. 5.11), the mechanisms of Li et al. (2004b) and Davis et al. (2005) predict slight reduction in ignition delay with CO addition and subsequent increase for RCO greater than 0.65. Similar discrepancy between experimental and calculated values is observed under all the conditions investigated in the present study. Figure 5.13 presents the effect of change in equivalence ratio on ignition delay at PC of 50 bar. Experiments are conducted for Mixtures #6 to #10. For these mixtures, fuel
126
loading is kept constant by keeping the same mole fractions for both H2 and CO, and equivalence ratio is changed by varying O2 mole fraction through the mole fraction variation of inerts. In Fig. 5.13, calculated ignition delays for equivalence ratio of 1 by using the mechanism of Li et al. (2004b) are also shown as a comparison. Since the chemistry of HO2 and H2O2 is significantly important under the experimental conditions of this study, addition of oxygen is expected to promote HO2/H2O2 reactions and reduction of oxygen can diminish HO2/H2O2 chemistry. It is seen from Fig. 5.13 that for a given TC, ignition delay decreases with increasing O2 addition or decreasing equivalence ratio, although the effect is observed to be rather weak. Figure 5.14 further shows the effect of equivalence ratio on ignition delay along with simulated results by using various mechanisms. For all mechanisms considered, calculated values are seen to follow the same trend as experimental data, despite quantitative differences are noted. A sensitivity analysis is subsequently conducted to identify the dominant elementary reactions for the ignition delays of Mixtures #1 to #5. For this, the rate constant of each reaction, ki, is individually doubled. The resulting percentage change in the ignition delay, namely 100×[τ(2ki)-τ(ki)]/τ(ki), is then taken as the percent sensitivity of that particular reaction. Figure 5.15 shows such a sensitivity analysis for pure H2 (RCO=0) and H2/CO mixtures of RCO=0.50 and 0.80 at TC=1010.5 K and PC=30 bar. Here, the mechanism of Li et al. (2004b), as listed in Table 5.2, is employed for illustration. The use of the mechanism of Davis et al. (2005) yields the similar results.
127
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Reaction H+O2=O+OH O+H2=H+OH H2+OH=H2O+H O+H2O=OH+OH H2+M=H+H+Ma H2+AR=H+H+AR H2+HE=H+H+HE O+O+M=O2+Ma O+O+AR=O2+AR O+O+HE=O2+HE O+H+M=OH+Mb H+OH+M=H2O+Mc H+O2(+M)=HO2(+M)d HO2+H=H2+O2 HO2+H=OH+OH HO2+O=O2+OH HO2+OH=H2O+O2 HO2+HO2=H2O2+O2 Duplicate HO2+HO2=H2O2+O2 Duplicate H2O2(+M)=OH+OH(+M)e
26
H2O2+H=H2O+OH H2O2+H=HO2+H2 H2O2+O=OH+HO2 H2O2+OH=HO2+H2O Duplicate H2O2+OH=HO2+H2O Duplicate CO+O(+M)=CO2(+M)f
27 28 29 30 31 32 33 34 35 36 37 38
CO+O2=CO2+O CO+HO2=CO2+OH CO+OH=CO2+H HCO+M=H+CO+Mg HCO+O2=CO+HO2 HCO+H=CO+H2 HCO+O=CO+OH HCO+OH=CO+H2O HCO+O=CO2+H HCO+HO2=CO2+OH+H HCO+HCO=H2+CO+CO HCO+HCO=CH2O+CO
21 22 23 24 25
k0 k∞
k0 k∞
k0 k∞
A 3.55×1015 5.08×104 2.16×108 2.97×106 4.58×1019 5.84×1018 5.84×1018 6.16×1015 1.89×1013 1.89×1013 4.71×1018 3.80×1022 9.04×1019 1.48×1012 1.66×1013 7.08×1013 3.25×1013 2.89×1013 4.20×1014
n -0.4 2.7 1.5 2 -1.4 -1.1 -1.1 -0.5 0 0 -1 -2 -1.50 0.6 0 0 0 0 0
Ea 16599 6290 3430 13400 104380 104380 104380 0 -1788 -1788 0 0 492 0 823 295 0 -497 11982
1.30×1011
0
-1629.3
1.20×1017 2.95×1014 2.41×1013 4.82×1013 9.55×106 1.00×1012
0 0 0 0 2 0
45500 48430 3970 7950 3970 0
5.80×1014
0
9557
1.55×1024 1.80×1010 2.53×1012 3.01×1013 2.23×105 4.75×1011 7.58×1012 7.23×1013 3.02×1013 3.02×1013 3.00×1013 3.00×1013 3.00×1012 3.00×1013
-2.79 0 0 0 1.9 0.7 0 0 0 0 0 0 0 0
4190 2384 47700 23000 -1158.7 14874 410 0 0 0 0 0 0 0
k=ATnexp(-Ea/RT); A Units are in mole-cm3-sec-K, Ea units are in cal/mole. a
Efficiency factors are: ε H 2 = 2.5, ε H 2O = 12, ε CO = 1.9, ε CO2 = 3.8, ε Ar = 0, ε He = 0.
128
b
Efficiency factors are: ε H 2 = 2.5, ε H 2O = 12, ε CO = 1.9, ε CO2 = 3.8, ε Ar = 0.75, ε He = 0.75.
c
Efficiency factors are: ε H 2 = 2.5, ε H 2O = 12, ε CO = 1.9, ε CO2 = 3.8, ε Ar = 0.38, ε He = 0.38.
d
When the main bath gas is Ar. Efficiency factors are: ε H 2 = 3, ε H 2O = 16, ε CO = 2.7,
ε CO = 5.4, ε O = 1.1, ε He = 1.2. Troe parameter, Fc = 0.5. 2
e
f
g
2
Efficiency factors are: ε H 2 = 2.5, ε H 2O = 12, ε CO = 1.9, ε CO2 = 3.8, ε Ar = 0.64, ε He = 0.64. Troe parameter, Fc = 0.5. Efficiency factors are: ε H 2 = 2.5, ε H 2O = 12, ε CO = 1.9, ε CO2 = 3.8, ε Ar = 0.87. Troe parameter, Fc = 1.0. Efficiency factors are: ε H 2 = 2.5, ε H 2O = 6, ε CO = 1.9, ε CO2 = 3.8.
Table 5.2 – Relevant portions of the Mechanism of Li et al. (2004b)
Under the experimental conditions investigated, ignition delay for pure H2 is dominated by following reactions H+O2=O+OH
R1
H+O2(+M)=HO2(+M)
R13
HO2+HO2=H2O2+O2
R18, R19
H2O2(+M)=OH+OH(+M)
R20
H2O2+H=HO2+H2
R22
Computed results also show that during the induction period prior to ignition, radical concentrations are in the ratio of O/OH/H/HO2/H2O2 = 1/2.3/8.5/5000/26000. In order to illustrate the effect of pressure on the radical pool, Fig. 5.16 shows calculated radical concentrations at various pressures for adiabatic, constant volume calculations. It can be noticed that with the increase in pressure, concentrations of HO2 and H2O2 radicals become more pronounced. Clearly, at 30 bar concentrations of HO2 and H2O2 are
129
several orders of magnitude higher than those of H, O, and OH. Consequently, in Fig. 5.15, autoignition is more sensitive to reactions involving formation and decomposition of H2O2. Flux analysis further illustrates that R18/R19 and the reverse of R22 are main reactions leading to the formation of H2O2, which subsequently decomposes through R20 and produces OH radical. It is noted that during the induction period the net rate of R22 is in the reverse direction, which promotes the formation of H2O2 and H radical. Since the reactions of HO2+HO2→H2O2+O2 (R18, R19) and HO2+H2→H2O2+H (-R22) are respectively “chain-termination” and “chain-branching” in nature, increasing the reaction rates of the former and the latter respectively leads to the increase and reduction in ignition delay, as shown in Fig. 5.15. With increasing addition of CO, sensitivity to the chain-branching cycle via R1-R3 and chain-termination via R13 increases, whereas sensitivity to R22 decreases. The reduced sensitivity of R22 is related to the decrease in H2 concentration. Sensitivity to R20 also reduces with increasing CO addition as the mechanism predicts reduction in H2O2 formation. Additionally, flux analysis shows that R28, CO+HO2=CO2+OH, is the primary reaction for the consumption of CO during the induction period. However, when ignition takes place and temperature increases, R29, CO+OH=CO2+H, consumes almost the entire remaining CO. For RCO=0.50 and TC=1010 K, Fig. 5.17 compares the percent sensitivities at compressed pressures of 15, 30, and 50 bar. At a lower pressure of 15 bar, it is seen that autoignition is highly sensitive to the chain-branching reaction R1 and chain-terminating reaction R13. As PC is increased, sensitivity to R1 and R13 decreases significantly,
130
whereas sensitivity to the formation and decomposition of H2O2 through reactions R18/R19, R22, and R20 increases. It is also noted that sensitivity to R28 reduces with increasing compressed pressure. Sensitivity analysis for different equivalence ratios, corresponding to Mixtures #6#10, at RCO=0.25, PC=50 bar, and TC=1009 K is shown in Fig. 5.18. It can be observed that there is no noticeable change in the importance of HO2 and H2O2 chemistry as equivalence ratio varies from 0.36 to 1.6. As such, the increase in oxygen mole fraction does not significantly alter the radical pool size. For lean equivalence ratios, calculations demonstrate that there is a slight increase in the concentrations of O, OH, HO2, and H2O2 and relatively larger decrease in H concentration. However, the radical pool remains dominated by HO2 and H2O2, whose concentrations are several orders of magnitude larger than those of O, OH, and H. Furthermore, HO2 is primarily consumed through reaction R18/R19, followed by the decomposition of H2O2 by R20. Addition of oxygen only slightly accelerates the ignition by increasing the concentrations of HO2 and H2O2.
5.5.4
Refinement of Rate Constants
Comparison of experimental results and model predictions, as shown in Figs. 5.105.12, demonstrates that the existing kinetic models significantly under-predict the inhibition effect of CO addition. By comparing the sensitivity analysis of pure H2 and H2/CO mixtures shown in Fig. 5.15, due to the increasing importance of R28 with increasing RCO, this reaction is expected to be responsible for the mismatch between experimental and calculated results. Figure 5.15 also shows that the percent sensitivities of R13 and R22 differ significantly as RCO is varied. Specifically, an increase in the
131
reaction rate of R22, k22, will reduce ignition delay of pure H2 to a larger extent than the reduction in ignition delay of H2/CO mixtures, leading to greater inhibition effect of CO addition. Similarly, an increase in k13 will also lead to greater inhibition effect of CO addition as it will increase the ignition delay of H2/CO mixtures to a larger extent than the increase in ignition delay of pure H2. Since any changes in k13 and k22 will involve a reexamination of H2 system and Figs. 5.5 and 5.6 demonstrate relatively acceptable predictability of the existing mechanisms, the reaction rate of R28 appears to be the primary candidate requiring refinement, which is discussed in the following. Recognizing that the reaction rate of R28 appears to be the primary candidate requiring refinement, a literature review of its rate constant is first conducted. Early investigations of the rate constant for R28 were summarized and reported by Lloyd (1974). Amongst these investigations, Baldwin et al. (1965) obtained a value for k28/(k18+k19)1/2 at 773 K from a study of slow reaction in the H2/CO/O2 system in a boricacid coated vessel, quoting an accuracy of 10-15%. Utilizing another technique, Baldwin et al. (1970) studied carbon monoxide sensitized decomposition of hydrogen peroxide in
a flow system using nitrogen as the carrier gas and determined a value for k28/(k18+k19)1/2 at 713K. It was observed that R28 had to be included into mechanism to achieve agreement between experimental and calculated decomposition rates at high CO concentrations (1970). Lloyd (1974) suggested that since it was assumed that all the participating reactions had been considered, other reactions aiding the decomposition or under-estimation of surface decomposition effects could affect k28. Baldwin et al. (1972) quoted an unpublished work for a value of k28/(k18+k19)1/2 which was obtained at 773 K and was a factor of 24 less than the previous measurement of Baldwin et al. (1965) at the
132
same temperature. Lloyd (1974) reported the values of k28 obtained by using the recommended expression for (k18+k19) along with the measurements of Baldwin et al. (1965, 1970, 1972). However, when the recently recommended value of Hippler et al. (1990) for (k18+k19) is used, values of k28 are a factor of 3 less than those calculated by Lloyd (1974). Vardanyan et al. (1975) investigated the oxidation of formaldehyde in the presence of CO to determine rate constant of R28 in the temperature range of 878-952 K. Without considering R17 – HO2+OH=H2O+O2, CO was assumed to be formed only by R28 and R29 – CO+OH=CO2+H. Vardanyan et al. (1975) asserted that under the conditions investigated, only 3-4% of the amount of CO2 determined experimentally was produced by R29 and therefore the main source of CO2 production was taken as R28. Based on this, k28 was determined therein. It is noted that k28 values of Vardanyan et al. (1975) are faster than those adopted by Li et al. (2004b). In fact, a calculation by using the mechanism of Li et al. (2004b) under the experimental conditions similar to those of Vardanyan et al. (1975) shows that R29 can account for 33% of CO2 production. Azatyan (1971) investigated the influence of CO addition on the first and second explosion limits of H2/O2 system. Azatyan (1971) concluded that in comparison with other reactions, rate of R28 is very much slower and he estimated the upper limiting values of k28 at 743 K and 945 K. Atri et al. (1977) noted a marked increase in the reaction rate by replacing N2 by CO in a H2/O2/N2 mixture undergoing slow reaction at 500 K and determined k28 from the experimentally determined ratio of k28/(k18+k19)1/2. However, Mueller et al. (1999b) pointed out that a re-interpretation of their rate data using recent value of (k18+k19) from Hippler et al. (1990) supports a 50% reduction in k28.
133
Colket et al. (1977) estimated k28 by using the rate of CO2 formation in the oxidation of acetaldehyde between 1030 K and 1115 K in a turbulent flow reactor. Colket et al. (1977) observed a steady increase in CO2 mole fraction throughout their
experiments and the estimated value of k28 at 1100 K was a factor of 3.5 higher than the extrapolated value of Atri et al. (1977). However, Colket et al. (1977) suggested that their value of k28 was accurate only within a factor of 4. In a more recent repetition of the experiments of Colket et al. (1977) by Zarubiak (1997) in a similar atmospheric pressure flow reactor and a variable pressure flow reactor, he obtained good agreement with the results of Colket et al. (1977) with the exception that he did not observe production of CO2 downstream of the injector in both reactors. Zarubiak (1997) further noted that the required disruption in radical concentrations which would accommodate the rate of CO2 production in the experiments of Colket et al. (1977) would substantially alter the rate of reaction and distribution of products, which was not observed in his experiment. Zarubiak (1997) then suggested that the production of CO2 in the experiments of Colket et al. (1977) could have been influenced by the impurities present in acetaldehyde, whereas his fuel was purified. Therefore, the work of Zarubiak (1997) would indicate a substantially lower value of k28. Recognizing this, in their mechanism Mueller et al. (1999b) justified a reduction in value of k28 and used a value which is factor of 2 smaller than those reported by Atri et al. (1977). Mueller et al. (1999b) also recommended further investigation of this reaction. In addition, Vandooren et al. (1986) obtained k28 by investigating the structure of a lean formaldehyde-oxygen flame burning at 22.5 torr. Figure 5.19 shows an Arrhenius plot of
134
suggested expressions for k28 along with the experimental measurements available in the literature. Since literature review of k28, particularly the recent finding of Zarubiak (1997), suggested that significant reduction in k28 may be due, further calculations are conducted to examine whether agreement between the measured and computed ignition delays for the present study can be improved by altering k28. We first note that GRI Mech 3.0 uses the value of k28 recommended by Baulch et al. (1976), which is a factor of 5 greater than the value used by the mechanisms of Li et al. (2004b) and Davis et al. (2005) that is taken from Mueller et al. (1999b). From simulations, it is observed that by reducing the value of k28 in GRI-Mech 3.0, contradictory trend predicted in the reduction of ignition delay with RCO (cf. Figs. 5.10-5.12) can be corrected. A comparison of experimental results and predicted ignition delay times by using different values of k28 based on the mechanism of Li et al. (2004b) is shown in Fig. 5.20. In Fig. 5.20, calculations are carried out by multiplying k28 from Li et al. (2004b) by a constant factor. Comparison shows that significant improvements in the prediction of existing mechanisms can be made by reducing k28 by a factor of 4. Additional simulations also confirm that the reduction in k28 alone does not affect much the predictions for low pressure literature data such as those of Kim et al. (1994). Finally, it has to be pointed out that while calculations with the reduction of k28 are shown to agree better with the inhibition effect of CO addition observed in RCM experiments, the modification of k28 alone still cannot capture all the experimental phenomena, especially the pressure dependence of ignition delay. Further kinetic studies are certainly warranted.
135
5.6 Summary
Autoignition of H2 and H2/CO fuel mixtures with O2/N2/Ar oxidizing mixtures is investigated in a rapid compression machine at pressures from 15 to 50 bar, temperatures from 950 K to 1100 K. The present experimental measurements lie above the extended second explosion limit, where CO oxidation at elevated pressures has not been studied thoroughly. Under these experimental conditions of intermediate temperature and high pressure, reactions involving formation and consumption of HO2 and H2O2 become significantly important. Data obtained herein is of direct importance in the development of a comprehensive mechanism for H2/CO oxidation, which can be used for modeling of practical combustors. Experiments demonstrate that replacement of even small amounts of H2 with CO leads to inhibition of autoignition. At all the pressures investigated, the inhibition effect of CO addition is observed. The inhibition effect is seen to be much more pronounced at higher pressures than at lower pressures. Comparison of experimental measurements with modeling results shows that for H2 autoignition, the mechanism of O’Conaire et al. (2004) generally shows good agreement, although some improvements are still required. However, for H2/CO mixtures, discrepancies are noted between experimental data and calculated results. Existing mechanisms of Li et al. (2004b), Davis et al. (2005), and GRI-Mech 3.0 (Smith et al., 1999) fail to describe the inhibition effect of CO addition. In contrast to experimental results, modeling results obtained from the mechanisms of Li et al. (2004b) and Davis et al. (2005) indicate that the inhibition effect of CO is not observed until CO constitutes
80% of the total fuel mole fraction. Whereas, predictions using GRI-Mech 3.0 show an
136
opposite trend of reduction in ignition delay with CO addition and subsequent increase in ignition delay for large CO percentage. The effect of O2 addition shows that at a pressure of 50 bar, ignition delays increase slightly with reduced amount of oxygen, although the effect is observed to be rather weak. Existing mechanisms are seen to properly describe the trend with oxygen addition. Kinetic analysis shows that under the present experimental conditions, the most important reactions related to autoignition are the chain-branching reaction H+O2=O+OH and the reactions involving H2O2 and HO2, including H+O2(+M)=HO2(+M), HO2+HO2=H2O2+O2,
H2O2(+M)=OH+OH(+M),
H2O2+H=HO2+H2,
and
CO+HO2=CO2+OH. Reaction of CO+HO2=CO2+OH is the dominant reaction for direct oxidation of CO during the induction period, and is identified as the primary reaction responsible for the mismatch of experimental and calculated ignition delays. Significant improvements in the prediction of existing mechanisms are made possible by reducing k28 by a factor of 4.
137
Weak chain branching H+O2+M→ HO2+M (k9) (k17) HO2+H2 → H2O2+H H2O2+M → OH+OH+M (k15)
3rd limit
Weak ignition
Extended 2nd limit
Chain termination H+O2+M→ HO2+M (k9)
P
2nd limit
Strong chain branching H+O2 → O+OH (k1) O+H2 → H+OH (k2) OH+H2 → H2O+H (k3)
1st limit
Strong ignition
T Fig. 5.1 - Explosion characteristics of homogeneous hydrogen/oxygen mixtures
Steady reaction
2nd limit
P Explosion
1st limit
T Fig. 5.2 - Explosion characteristics of homogeneous wet CO mixtures
138
60
Pressure (bar)
50 40
Ignition Delay, τ
Pc = 30 bar Tc = 999 K
30 20 10 0 -30
-20
-10 Time (ms)
0
10
Fig. 5.3 – Typical pressure trace and experimental repeatability for ignition of H2/CO mixture. Molar composition: H2/CO/O2/N2/Ar = 9.375/3.125/6.25/18.125/63.125. Initial conditions – T0 = 298.7 K, P0 = 661 torr. Conditions at TDC – TC = 999 K, PC = 30 bar.
139
1
10
Craig_1 (1966) Craig_2 (1966) Skinner/Ringrose (1965) Bhaskaran et al. (1973) Cheng/Oppenheim (1984) Schott/Kinsey (1958) Petersen et al._1 (1996) Petersen et al._2 (1996) Petersen et al._3 (1996) Lee/Hochgreb (1998b) Present_1 Present_2 Present_3
0
3
τ [O2] (µs mol/cm )
10
-1
10
10-2 -3
10
10-4 10-5 10-6
0.4
0.6 0.8 1000/T (1/K)
1
Fig. 5.4 – Comparison of measured ignition delay times on the present RCM with previous measurements. Craig_1 (1966)
H2/O2/N2 - 2/1/3.76
1.02 atm
Craig_2 (1966)
H2/O2/N2 - 2/1/3.76
2.04 atm
Skinner/Ringrose (1965)
H2/O2/Ar - 8/2/90
5 atm
Bhaskaran et al. (1973)
H2/O2/N2 - 2/1/3.76
2.5 atm
Cheng/Oppenheim (1984)
H2/O2/Ar - 6.67/3.33/90
1-3 atm
H2/O2/Ar - 5/5/90
1-3 atm
Schott/Kinsey (1958)
H2/O2/Ar mixtures
0.15 - 9.5 atm
Petersen et al._1 (1996)
H2/O2/Ar - 0.5/0.25/99.25
33 atm
H2/O2/Ar - 2/1/1997
33 atm
H2/O2/Ar -0.33/0.167/99.5
64 atm
H2/O2/Ar -0.1 /.05/99.85
64 atm
Petersen et al._3 (1996)
H2/O2/Ar - .5/.25/99.25
57 atm
Lee/Hochgreb (1998b)
H2/O2/Ar - 2/1/5
6-40 atm
Present_1
H2/O2/N2/Ar - 2/1/7/45
33.6 atm
Present_2
H2/O2/N2/Ar - 2/1/15/82
33.55 atm
Present_3
H2/O2/N2/Ar - 2/1/2.9/10.1
29.6 atm
Petersen et al._2 (1996)
140
Ignition Delay, τ (ms)
100 Experimental Li et al. Davis et al. GRI-Mech 3.0 O'Conaire et al.
10
1 0.94
0.96
0.98 1 1.02 1000/T (1/K)
1.04
1.06
Fig. 5.5 – Measured (filled symbol) and calculated (open symbol) ignition delays for H2 autoignition. Molar composition: H2/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 30 bar.
100
}
Ignition Delay, τ (ms)
PC=15 bar
}
PC=50 bar
10 Experimental Li et al. Davis et al. GRI-Mech 3.0 O'Conaire et al.
1
0.94
0.96
0.98
1 1.02 1000/T (1/K)
1.04
1.06
Fig. 5.6 – Measured (filled symbol) and calculated (open symbol) ignition delays for H2 autoignition at PC of 15 and 50 bar. Molar composition: H2/O2/N2/Ar = 12.5/6.25/18.125/63.125.
141
100 Ignition Delay, τ (ms)
PC = 50 bar
0.50
0.80 0.65
0.25 RCO=0
10
1 0.96
0.98 1 1000/T (1/K)
1.02
1.04
Fig. 5.7 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 50 bar.
100 Ignition Delay, τ (ms)
PC = 30 bar
0.50 0.80 0.65
RCO=0 0.25
10
1 0.96
0.98
1 1.02 1000/T (1/K)
1.04
1.06
Fig. 5.8 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 30 bar.
142
Ignition Delay, τ (ms)
100
RCO 0 0.25 0.50 0.65
10
0.80
PC = 15 bar
1 0.94
0.96 0.98 1000/T (1/K)
1
1.02
Fig. 5.9 – Measured ignition delays for H2/CO autoignition. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Pressure at TDC, PC = 15 bar.
30 Ignition Delay, τ (ms)
Li et al.
25
PC = 30 bar TC = 1010.5 K
GRI-Mech 3.0 Davis et al.
20
Experimental
15 10 5 0
0
0.2
0.4
0.6
0.8
1
RCO Fig. 5.10 – Comparison of experimental and computed results, showing the effect of CO addition on ignition delay. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Conditions at TDC: PC = 30 bar and TC = 1010.5 K.
143
Ignition Delay, τ (ms)
20
PC = 15 bar TC = 1028.5 K
Li et al. GRI-Mech 3.0
15
Davis et al. Experimental
10
5
0
0
0.2
0.4
0.6
0.8
1
RCO Fig. 5.11 – Comparison of experimental and computed results, showing the effect of CO addition on ignition delay.Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Condition at TDC: PC= 15 bar and TC = 1028.5 K.
4 Li et al.
Ignition Delay, τ (ms)
3.5
GRI-Mech 3.0
3
PC = 50 bar TC = 1044 K
Davis et al.
2.5
Experimental
2 1.5 1 0.5 0
0
0.2
0.4
0.6
0.8
1
RCO Fig. 5.12 – Comparison of experimental and computed results, showing the effect of CO addition on ignition delay. Molar composition: (H2+CO)/O2/N2/Ar = 12.5/6.25/18.125/63.125. Condition at TDC, PC = 50 bar, TC = 1044 K.
Ignition Delay, τ (ms)
144
Computational
Experimental φ 0.36 0.72 1.0 1.3 1.6
10
φ=1.0, Li et al.
PC = 50 bar
1
0.96
0.98 1 1000/T (1/K)
1.02
1.04
Fig. 5.13 – Effect of equivalence ratio on ignition delay. Composition: Mixtures #6 to #10 in Table 5.1. Pressure at TDC, PC = 50 bar.
Li et al. GRI-Mech 3.0 Davis et al. Experimental
Ignition Delay, τ (ms)
20 18 16
TC = 990K
14 12 10 TC = 1009 K
8 6 4
0.4
0.6
0.8 1 1.2 Equivalence Ratio
1.4
1.6
Fig. 5.14 – Effect of Equivalence ratio on ignition delay. Composition: Mixtures #6 to #10 in Table 5.1. Conditions at TDC, PC = 50 bar and TC = 990 K (top group), 1009 K (bottom group).
Reaction Number
145
29 28 27 26 25 22 21 20 19 18 17 16 15 14 13 3 2 1
TC = 1010.5 K PC = 30 bar
RCO 0 0.50 0.80
-60
-40
-20
0
20
40
60
Percent Sensitivity
Fig. 5.15 – Sensitivity analysis of ignition delay at different CO additions. Conditions at TDC – PC = 30 bar and TC= 1010.5 K.
146
10
-2
10
-4
10
-6
100
H2 O2
H
HO2
Concentration (mol fraction)
Concentration (mol fraction)
100
H2O
10-8 -10
OH
10
O
H2O2
10-12 10-14
(a) 1 bar 0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
10
-2
10
-4
10
-6
HO2 H 2O H2O2
10-8
H
-10
10
(b) 5 bar
10-12 0
0.5
1
1.5
Time (ms) 0
10
-2
10
-4
10
-6
10
-8
10
H2 O2 H 2O H2O2
HO2
H
-10
10
O
OH
-12
10
(c) 30 bar
-14
10
2
2.5
3
Time (ms)
Concentration (mol fraction)
Concentration (mol fraction)
10
OH
O
10-14
0.4
H2 O2
10
0
H2
-2
O2 H 2O
10-4 10
-6
10
-8
H2O2
HO2
H
-10
10
O
OH
-12
10
(d) 50 bar
-14
0
1
2
3
4
5
10
0
Time (ms)
0.5
1
1.5
2
2.5
3
3.5
4
Time (ms)
Fig. 5.16 – Specie concentration for adiabatic, constant volume calculations at different pressures.
Molar
composition:
H2/O2/N2/Ar
=
12.5/6.25/18.125/63.125.
temperature = 1010.5 K. Initial pressure: (a) 1 bar, (b) 5 bar, (c) 30 bar, (d) 50 bar
Initial
147
29
TC = 1010 K RCO = 0.50
28 27
Reaction Number
26 25 22 21 20
PC
19
15 bar 30 bar 50 bar
18 17 15 14 13 1
-80
-40
0
40
80
Percent Sensitivity Fig. 5.17 – Sensitivity analysis of ignition delay at different pressures for RCO = 0.50. Conditions
Reaction Number
at TDC – TC = 1010 K. 29 28 27 26 25 22 21 20 19 18 17 16 15 14 13 3 2 1
RCO=0.25 TC = 1009 K PC = 50 bar
φ 1.6 1.0 0.36
-40
-30
-20
-10
0
10
20
Percent Sensitivity Fig. 5.18 – Sensitivity analysis of ignition delay at different equivalence ratios. Conditions at TDC – PC= 50 bar and TC = 1009 K.
148
Mueller et al. (1999b) Lloyd (1974) Atri et al. (1977) Vardanyan et al. (1975) Vandooren et al. (1986) Baulch et al. (1976) Mueller et al. (1999b) / 4.0
1013
k28 (cm3/mol/s)
1011
Baldwin et al. (1965) Baldwin et al. (1972) Baldwin et al. (1970) Azatyan (1971) Atri et al. (1977) Vardanyan et al. (1975)
109
107
105
103 0.5
1
1.5
2
2.5
3
1000/T (1/K)
Fig. 5.19 – Comparison of rate constants for k28. k28 for the measurements Baldwin et al. (1965, 1970, 1972) is calculated by using (k18+k19) from Hippler et al. (1990).
149
Ignition Delay, τ (ms)
20
15
k28x0.1 k28x0.25 k28x0.5 k28 Experimental
PC = 15 bar TC = 1028.5 K
10
5
0
Ignition Delay, τ (ms)
25
k28x0.1
PC = 30 bar TC = 1010.5 K
k28x0.25
20
k28x0.5 k28
15
Experimental
10 5 0
Ignition Delay, τ (ms)
6
4
k28x0.1 k28x0.25 k28x0.5 k28 Experimental
PC = 50 bar TC = 1044 K
2
0 0
0.2
0.4
0.6
0.8
1
RCO
Fig. 5.20 – Comparison of measured and calculated ignition delays for different values of k28 while using the mechanism of Li et al. (2004b). Calculations are conducted by multiplying k28 from Li et al. (2004b) by a constant factor.
150
CHAPTER 6 SUMMARY AND FUTURE WORK 6.1 Summary A rapid compression machine is an excellent tool to study high pressure autoignition of combustible mixtures. Experimental duration in shock tube is limited to less than 5 ms due to interference from reflected waves. In an RCM, experimental duration of an order of magnitude larger than shock tube can be obtained. As a result, RCM gives accessibility to study combustion at elevated pressure under conditions for which reaction is too slow for shock tubes, which typically spans intermediate range of temperature from 600 to 1100 K. In view of the suitability of RCM for combustion kinetics studies, we have designed and fabricated a RCM. The present RCM is designed as a versatile tool which integrates the key features of various existing RCMs and allows a wide range of experimental conditions. Key design objectives include a well-defined core region, fast compression, ability to vary stroke and clearance, optical accessibility, capability for specie measurement using fast sampling apparatus, and minimum mechanical vibrations. The present RCM is pneumatically driven and hydraulically actuated and stopped. Stroke of the machine varies from 7 to 10 inches and clearance is also adjustable. Compression ratio of up to 15.1 can be obtained while using a creviced piston. A deliberately machined crevice on the piston head has been optimized, using STAR-CD CFD package, in order to suppress the formation of the roll-up vortex and provide a homogeneous core of reacting mixture.
151
Aerodynamics inside the rapid compression machine after the end of compression is investigated experimentally using PLIF of acetone. In order to study the effect of reaction chamber configuration on the resulting aerodynamics and temperature field, experiments are conducted and compared using a creviced piston head and a flat piston head under varying conditions. Results show that the flat piston head design leads to significant mixing of cold vortex with hot core region, which causes alternate hot and cold regions inside the combustion chamber. At higher pressures, the effect of vortex is reduced.
Creviced piston head configuration is demonstrated to result in drastic
reduction of the effect of vortex. Experimental conditions are also simulated using Star-CD CFD package. Numerical results indicate that with a flat piston head design, gas velocity after compression is very high and the core region shrinks quickly due to rapid entrainment of cold gases. Whereas, for a creviced piston head design, gas velocity after compression is significantly lower and the core region remains unaffected for a long duration. As a consequence, for the flat piston head, adiabatic core assumption can significantly overpredict the maximum temperature after the end of compression. For the creviced piston, adiabatic core assumption is found to be valid even up to 100 ms after compression. These investigations suggest that when a properly designed creviced piston is used, a zero-dimensional model should satisfactorily model the experimental results using a RCM. However, when a flat head piston is employed, the use of a zero-dimensional model to simulate experimental data is deemed inadequate. Characterization experiments are conducted to demonstrate the suitability of the designed machine for combustion related research. Experiments with either inert gases or
152
reactive mixtures demonstrate the reproducibility of pressure traces. Compression process is also shown to be very rapid and free from any significant mechanical vibrations. Measurements show that highly repeatable compressed conditions of up to 50 bar and greater than 1000 K can be obtained. In addition, a rapid sampling apparatus is incorporated in the reaction chamber for determining specie concentration at specific post-compression time. A numerical model accounting for heat loss is also developed to simulate the RCM data. All these features make the designed RCM an excellent tool for studying combustion phenomena at elevated pressure/temperature conditions. Using this facility, autoignition investigations are conducted for iso-octane and H2/CO system. Experiments using iso-octane reveal the interesting features of two-stage ignition and negative temperature dependence of reaction. In addition, significant discrepancies are noted between experimental results and predictions of mechanism. Investigations for H2 and H2/CO system are conducted at pressures from 15 to 50 bar, temperatures from 950 K to 1100 K, and equivalence ratios from 0.36 to 1.6. In addition, the effect of change in relative concentrations of H2 and CO is investigated by replacing H2 with CO, while keeping the total fuel mole fraction of combined H2/CO fuel constant. Under these experimental conditions of intermediate temperature and high pressure, reactions involving formation and consumption of HO2 and H2O2 become significantly important. Results show that for H2 autoignition, the mechanism of O’Conaire et al. (2004) agrees well with the experimental data, although some improvements can still be made. At all the pressures investigated, replacement of some amount of H2 by CO in a reacting mixture leads to the increase in ignition delay, even with small amount of CO addition.
153
The inhibition effect of CO addition is also observed to be much more pronounced with increasing pressure. For H2/CO mixtures, existing mechanisms of Li et al. (2004b), Davis et al. (2005), and GRI-Mech 3.0 fail to describe the experimental trend of ignition delay. At a pressure of 50 bar, ignition delays are seen to increase monotonically with O2 addition, although such an effect is rather weak. Kinetic analysis further demonstrates that under the present experimental conditions, CO+HO2=CO2+OH is the primary reaction responsible for the mismatch of experimental and calculated ignition delays. Significant improvements in the predictions of existing mechanisms can be made by reducing this reaction rate.
6.2 Future Work Research conducted in this work can be further extended in a number of ways, which are discussed below. In the present investigations, CFD simulations are shown to be extremely helpful in describing the aerodynamics inside the machine. These computational studies can be further extended to reactive mixtures as well. Simulations with reactive mixtures can help in understanding the effect of spatial inhomogeneity on autoignition. Further, the development of temperature field for reactive mixtures can also be studied experimentally. A number of experimental techniques, for example PLIF of OH, PLIF of formaldehyde, chemiluminescence imaging, can be used for this purpose. Experiments using iso-octane demonstrate that the predictions of the existing detailed mechanism significantly differ from experimental results. Although the mechanism of Curran et al. (2002) is generally known for giving reasonable predictions
154
under conditions of high temperature, it fails under conditions of intermediate temperature which are accessible on RCM. Further experiments can be conducted for isooctane ignition over a range of physical conditions with the aim of developing deeper insights in the low temperature chemistry of this system, which is not properly understood. Investigations of H2 and H2/CO system demonstrate that the existing kinetic mechanisms fail to describe the experimental behavior. Further investigations of the chemical kinetics of these systems can be undertaken on the facilities available in the Combustion Diagnostics Laboratory, with the ultimate aim of development of comprehensive kinetic mechanism for this system. Specifically, investigation of CO oxidation in the presence of water can provide additional information about kinetics of CO oxidation. In addition, a detailed kinetic analysis can be undertaken to understand the uncertainties and modifications required in kinetic parameters, which would enable the prediction of experimentally observed trend of CO inhibition and pressure effect. Experiments can also be conducted on low-stretch diffusion flame burner and counterflow burner, which already exist in the Combustion Diagnostics Lab. In the present work, utility of rapid sampling apparatus along with Gas Chromatography (GC) in quantitative analysis of reacting gases is demonstrated for methane combustion. However, the identification of reaction products becomes increasingly difficult when the sample to be analyzed is a complex mixture, which is the case during combustion of higher hydrocarbons. A Mass Spectroscopy (MS) unit combined with GC is capable of unambiguous identification and quantification of complex mixtures, and can also be incorporated in the present RCM. Addition of
155
combined GC/MS unit to the present RCM will enhance the diagnostic capabilities of the RCM and enable measurement of intermediate specie concentrations for chemical kinetic investigations of various hydrocarbon fuels. Finally, since the designed experimental system is demonstrated to be suitable for autoignition investigations, such investigations can be conducted for a variety of fuels.
156
Appendix – A PLIF of Acetone PLIF stands for planar laser-induced fluorescence. Molecules have certain quantized energy states. When a molecule is excited to a higher energy level, it seeks to return to the lowest energy state, the ground state. Two methods of relaxing back to the ground state are important - radiationless relaxing, often through collisions with other molecules (quenching), and relaxing by emitting light. There are two main processes that relax the molecule through emission; fluorescence and phosphorescence. Energy states of a molecule are described as singlet or triplet based on the spins of the electrons. Fluorescence occurs when light is emitted without a change in spin (singlet to singlet or triplet to triplet). Phosphorescence occurs when a spin change must also happen during the emission (triplet to singlet or singlet to triplet). This phosphorescence process is has a much longer lifetime than fluorescence. Laser-induced fluorescence simply means that a laser beam at an appropriate wavelength is used to excite the molecules. Laser energy needs to match the energy needed to excite the molecule to a state from which it can fluoresce. Planar refers to the shape of the laser beam. A series of lenses is used to make narrow laser sheet from the original laser beam. The strength of PLIF-based diagnostics is their ability to non-intrusively provide instantaneous information in an entire plane of the flow field, without the line-of-sight averaging. Signals are typically stronger than for Rayleigh or Raman scattering. Particularly for confined geometries, Rayleigh scattering results in significant noise from reflection from surfaces, whereas PLIF is easily applied because incident laser is at a different wavelength than fluorescence signal and therefore can be filtered.
157
For fluorescence measurement, the chosen fluorescent specie can be that exists in the flow of interest (e.g. OH in flames) or can be seeded into the flow as a tracer. Acetone has found wide use as a tracer molecule in PLIF imaging. Its appeal includes low toxicity, low cost, and a high vapor pressure that permits seeding in high concentrations. A broad absorption feature (225 to 320 nm) is accessible with high-energy, pulsed UV lasers. Fluorescence occurs between 350 and 550 nm. Particularly, for the purpose of temperature imaging, which is the objective of this work, the use of acetone as a tracer offers the advantage of high temperature sensitivity of fluorescence intensity. A simplified model of acetone’s photophysical behavior taken from Thurber et al. (1997, 1998) is shown in Fig. A.1. Typical fluorescence decay of an excited molecule is shown. This molecule initially sits in thermal equilibrium at some energy ∆Ethermal(T) above the ν’’=0 level of the ground singlet state S0, where ν’’ is the vibrational quantum number. It is excited by a laser, which pumps it with energy ∆Elaser to the excited singlet state S1. The combination of the initial thermal energy and the laser excitation energy determines the molecule’s initial energy E above the excited singlet origin. The molecule decays in E because of discrete collisional events. From each vibrationally excited level that it occupies in the S1 manifold, the molecule can nonradiatively decay through intersystem crossing, with a rate kNR(E); fluoresce back to S0, with a rate kf; or decay to a lower vibrational level in S1, with a rate equal to a collision rate kcoll(T, P). Rapid intersystem crossing from S1 to the first excited triplet T1 limits the fluorescence yield, and allows phosphorescence to occur from T1 on a long time scale. Detailed low-pressure studies of the intersystem crossing process in acetone have revealed it to be intermediate in character between large and small-molecule behavior.
158
The excited singlet evolves rapidly (in less than 5 ns) into a state of mixed singlet-triplet character in an intramolecular dephasing process. At pressures greater than 10 torr, collisional quenching of the mixed singlet-triplet state into the triplet is very rapid; therefore, the conceptual model of a direct nonradiative, irreversible transition from singlet to triplet is considered adequate under most conditions. Eventually the molecule reaches the thermalized level ∆Ethermal(T), above the origin of S1, from which it can either cross to the triplet or fluoresce back to the ground singlet. In this model, although the nonradiative electronic relaxation of the excited singlet is assumed to be due to intersystem crossing, it could equally well be due to any combination of intramolecular processes, including internal conversion or predissociation, Fluorescence from a molecule with broadband absorption and fluorescence features such as acetone can be modeled as shown in Equation (A.1). For weak excitation, the fluorescence signal, in number of photons collected, is given by Sf =
χ P E ηopt dVc acetone σ (λ , T )φ (λ , T , P, ∑ χ i ). hc / λ kT i
(A.1)
Here, E is the laser fluence [J/cm2], (hc/ λ) is the energy [J] of a photon at the excitation wavelength λ, ηopt is the overall efficiency of the collection optics, and dVc is the collection volume [cm3]. The bracketed term is the acetone number density [cm-3], given as the product of mole fraction χacetone and total pressure P divided by the Boltzmann constant k times temperature T. The final two quantities are σ, the molecular absorption cross-section of the tracer [cm2], and φ, the fluorescence quantum yield.
159
Fig. A.1 – Simplified model of acetone’s photophysical behavior with multistep decay of the excited singlet. Source: Thurber et al. (1997, 1998). At conditions of uniform seeding and total pressure, a single-wavelength excitation strategy is possible for determination of temperature. Denoting Sf+ as the fluorescence signal per unit acetone mole fraction at constant pressure, we see that:
S +f ∝
1 σ (λ , T )φ (λ , T , P) . T
(A.2)
If the temperature dependence of the right-hand side of Equation (A.2) has been determined for excitation wavelength λ and total pressure P, measurement of fluorescence in conjunction with a single calibration point in the flow provides an indication of temperature.
160
Fig. A.2 - Fluorescence per unit laser energy per unit mole fraction at atmospheric pressure, normalized to the room-temperature value. At constant pressure and seeding fraction, temperature can be inferred from a fluorescence measurement using this plot. For clarity, only five excitation wavelengths are displayed. Source: Thurber et al. (1997, 1998). Figure A.2 shows temperature dependence of normalized fluorescence per unit laser energy per unit mole fraction, S +f = (Sf+(T) / Sf+(295 K)), for different excitation wavelengths. Clearly, any wavelength between 248 and 289 nm is suitable for temperature imaging, due to high sensitivity of fluorescence to temperature. In this work, excitation wavelength of 279 nm is used. Additionally, data for absorption coefficient is required to correct for the attenuation of laser intensity as laser traverses the zone of interest. Data for absorption coefficient (σ) and normalized fluorescence signal per unit acetone mole fraction ( S +f ) for excitation wavelength of 276 nm and 282 nm were reported by Thurber et al. (1997, 1998). For 279 nm excitation, reported data is
161
interpolated. Interpolated curves and corresponding equations are shown in Figs. A.3, A.4, and A.5. x10
-20
2
Absorption coefficient, σ (cm )
7.5 7 6.5 6
282 nm 276 nm 279 nm (curve fit)
σ
5.5 5 4.5 300
400
500 600 Temperature, T (K)
700
800
900
Fig. A.3 - Absorption coefficient as a function of temperature. Absorption coefficient at 279 nm : σ = (-6.17183E-09*T3 + 7.85630E-06*T2 + 1.61383E-03*T + 3.71465) 10-20. (A.3) 1
0.8 282nm 276nm 279 nm (curve fit)
~
Sf+
0.6
0.4
0.2
0 300
400
500 600 Temperature, T (K)
700
800
900
Fig. A.4 – Fluorescence per unit laser energy per unit mole fraction normalized to the room temperature. S +f at 279 nm: S +f = 1.16016E-11*T4 - 3.14317E-08*T 3 +3.177761E-05*T 2 -1.518611E-02*T + 3.42820
(A.4)
162
1000
Temperature, T (K)
900 800
282 nm 276 nm 279 nm (curve fit)
700 600 500 400 300 200 0
0.2
0.4
~ S+ f
0.6
0.8
1
Fig. A5- Temperature as a function of fluorescence per unit laser energy per unit mole fraction normalized to the room temperature. T = 870.361 S +f 4 - 2078.670 S +f 3 +2429.144 S +f 2 –2181.93 S +f +1259.214.
(A.5)
Deduction of temperature
Fluorescence intensity is integrated along the width of the laser sheet, which gives fluorescence intensity versus pixel number as seen in Fig. 3.4. The following procedure is used for deduction of temperature from this integrated fluorescence intensity. An anchor point in the fluorescence signal is arbitrarily chosen at some pixel inside the chamber and given a value of temperature equal to adiabatic core temperature in the cylinder. Subsequently, temperature at other radial locations is determined by marching radially in both directions from this anchor point. After each iteration, temperature of anchor point is adjusted for the next iteration such that eventually the maximum deduced temperature in the chamber equals the adiabatic core temperature.
163
At the anchor point (location 1), normalized fluorescence per unit laser energy per unit mole fraction, S +f ,1 , is calculated from Equation (A.4) by using the assumed temperature, T1. Apparent incident laser intensity at anchor point is given by I1 = Sexp,1 / S +f ,1
,
(A.6)
where Sexp,1 is given by Sexp,1 = Sexp,1 (T1 ) / Sexp (295K ) ,
(A.7)
where Sexp,1(T1) is the experimentally measured fluorescence intensity at the anchor point. As laser traverses a length of size dx at the anchor point to location 2, attenuation of incident laser intensity according to Beer’s law is expressed as I2/I1 = exp(-σ1 n1 dx) ,
(A.8)
where σ1 is calculated from Equation (A.3) and n1 represents acetone number density given by n1= PN/RT1 ,
(A.10)
P is the pressure, R is the universal gas constant, and N is the Avogadro number. Now, for subsequent marching, experimentally determined normalized fluorescence intensity per unit incident laser intensity per unit acetone mole fraction is expressed as S +f ,exp,2 = Sexp,2 / I 2 ,
(A.10)
where Sexp,2 is expressed in a manner similar to Equation (A.7). By using Equations (A.6)-(A.9), Equation (A.10) can be written as S +f ,exp,2 =
Sexp,2 S +f ,1 Sexp,1 exp(−σ 1n1dx)
(A.11)
164
Now, the normalized fluorescence intensity can be calculated from Equation (A.11), and temperature is deduced by using Equation (A.5). Further attenuation in incident laser intensity at this point can be calculated as temperature is known and hence σ and acetone number density, n, are also known. In this way temperature is deduced by marching from the anchor point.
165
Appendix – B Piston Velocity Calculation An approximation for the velocity of the piston during the compression process can be made from the experimental pressure trace. Here, the procedure for obtaining velocity of the piston is described. Compression process can be approximated as polytropic in nature. Index of polytropic compression, n, which is taken as a constant, can be expressed by the relation
Pc = (CR) n , P0
(B.1)
where P0 and Pc are the initial pressure and pressure at TDC respectively, and CR is the geometric compression ratio. From known P0, Pc, and CR, the constant index of polytropic compression, n, can be calculated from Equation (B.1). Based on polytropic compression, volume of the combustion chamber during compression, V(t), can be expressed as 1/ n
⎡ P ⎤ V (t ) = V0 ⎢ 0 ⎥ ⎣ P (t ) ⎦
,
(B.2)
where V0 is the initial volume before compression begins and P(t) is the experimental pressure trace. Subsequently, velocity of the piston during compression, Vel(t), can be obtained by
Vel (t ) =
dV / dt π D2 / 4 ,
(B.3)
where D is the diameter of the reaction chamber and dV/dt is the time change of volume of the chamber. In order to obtain dV/dt, a piecewise continuous polynomial (including
166
continuous first derivatives) is fitted to the time-history of the calculated volume of the chamber. The polynomial fit for V(t) is then used to determine the derivative dV/dt. A typical pressure trace for compression of N2 from 1000 torr to 47.5 bar is shown in Fig. B.1. For this experiment, the geometric compression ratio (including the effect of piston crevice and other dead volumes), CR, is 15.1, and the initial combustion chamber volume, V0, is 533.8 cm3. Figure B.2 shows the calculated combustion chamber volume (solid line) using Equation (B.2) and the associated piecewise polynomial fit (dashed line). Derived piston velocity based on Equation (B.3) is shown in Fig. B.3. It is seen that after the initial rapid acceleration, the piston accelerates at a slower rate and reaches the peak velocity. Finally, there is rapid but uniform deceleration and the piston comes to stop as it reaches the TDC.
50
Pressure (bar)
40 30 20 10 0 -30
-20
-10
0
10 20 Time (ms)
30
40
50
Fig. B.1 – Experimental pressure trace for compression of N2. Intial conditions – P0 = 1000 torr and T0 = 297 K.
167
500
V(t) (cm3)
400
300
200
100
0 -30
-25
-20
-15 Time (ms)
-10
-5
0
Fig. B.2- Calculated combustion chamber volume as a function of time (solid line) and the fitted profile (dashed line).
Piston velocity, Vel(t) (m/s)
15
10
5
0 -30
-25
-20
-15 Time (ms)
Fig. B.3- Deduced piston velocity history.
-10
-5
0
168
References Affleck, W. S., and Thomas, A. (1968) An Opposed Piston Rapid Compression Machine for Preflame Reaction Studies. Proc. Inst. Mech. Eng., 183, 365-387. Atri, G. M., Baldwin, R. R., Jackson, D., and Walker, R. W. (1977) The Reaction of OH Radicals and HO2 Radicals with Carbon Monoxide. Combust. Flame, 30, 1-12. Azatyan, V. V. (1971) Dokl. Akad. Nauk., SSSR 196, 617. Baldwin, R. R., Jackson, D., Walker, R. W., and Webster, S. J. (1965) The Use of the Hydrogen-Oxygen Reaction in Evaluating Velocity Constants. Proc. Combust. Inst., 10, 423-433. Baldwin, R. R., Walker, R. W., and Webster, S. J. (1970) Combust. Flame, 15, 167. Baldwin, R. R., Jackson, D., Melvin, A., and Rossiter, B. N. (1972) The Second Limit of Hydrogen + Carbon Monoxide + Oxygen Mixtures. Int J. Chem. Kin., 4, 277-292. Baulch, D. L., Drysdale, D. D., Duxbury, J., and Grant, S. J. (1976) Evaluated Kinetic Data for High Temperature Reactions , Vol. 3, Butterworths, London. Beeley, P, Griffiths, J. F., and Gray, P. (1980a) Rapid Compression Studies on Spontaneous Ignition of Isopropyl Nitrate, Part I: Nonexplosive Decomposition, Explosive Oxidation and Conditions for Safe Handling. Combust. Flame, 39, 255268. Beeley, P., Griffiths, J. F., and Gray, P. (1980b) Rapid Compression Studies on Spontaneous Ignition of Isopropyl Nitrate, Part II: Rapid Sampling, Intermediate Stages and Reaction Mechanisms. Combust. Flame, 39, 269-281. Bhaskaran, K.A., Gupta, M. C., and Just, Th. (1973) Combust. Flame, 21, 45-48.
169
Brabbs, T. A., Belles, F. E., and Brokaw, R. S. (1970) Shock-Tube Measurements of Specific Reaction Rates in Branched-Chain H2-CO-O2 Systems. Proc. Combust. Inst., 13, 129-136. Brett, L., Macnamara, J., Musch, P., and Simmie, J. M. (2001) Simulation of Methane Autoignition in a Rapid Compression Machine with Creviced Pistons. Combust. Flame, 124, 326-329. Brokaw, R. S. (1967) Ignition Kinetic of the Carbon Monoxide-Oxygen Reaction. Proc. Combust. Inst., 11, 1063-1073. Cheng, R. K., Oppenheim, A. K. (1984) Autoignition in Methane-Hydrogen Mixtures. Combust. Flame, 58, 125–139. Clarkson, J., Griffiths, J. F., Macnamara, J. P., and Whitaker, B. J. (2001) Temperature Fields During the Development of Combustion in a Rapid Compression Machine. Combust. Flame, 125, 1162-1175. Colket, M. B., Naegeli, D. W., and Glassman, I. (1977) High Temperature Oxidation of Acetaldehyde. Proc. Combust. Inst., 16, 1023–1039. Craig, R. R. (1966) A Shock Tube Study of the Ignition Delay of Hydrogen-Air Mixtures Near the Second Explosion Limit. AFAPL-TR-66-74, AF Aeropropulsion Lab, Wright- Patterson. Curran, H. J., Gaffuri, P., Pitz, W. J., and Westbrook, C. K. (2002) A Comprehensive Modeling Study of iso-Octane Oxidation. Combust. Flame, 129, 253-280. Dagaut, P., Reuillon, M., Cathonnet, M. (1995) Experimental Study of the Oxidation of n-Heptane in a Jet Stirred Reactor from Low to High Temperature and Pressures up to 40 Atm. Combust. Flame 101, 132-140.
170
Daneshyar, H., Fuller, D. E., and Deckker, B. E. L. (1973) Vortex Motion Induced by the Piston of an Internal Combustion Engine. Int. J. Mech. Sci., 15, 381-390. Davis, S. G., Joshi, A. V., Wang, H., and Egolfopoulos, F. (2005) An Optimized Kinetic Model of H2/CO Combustion. Proc. Combust. Inst., 30, 1283-1292. Dean, A. M., Steiner, D. C., and Wang, E. E. (1978) A Shock Tube Study of the H2/O2/CO/Ar and H2/N2O/CO/Ar Systems: Measurement of the Rate Constant for H+N2O=N2+OH*. Combust. Flame, 32, 73-83. Desgroux, P., Gasnot, L., and Sochet, L. R. (1995) Instantaneous Temperature Measurement in a Rapid-Compression Machine Using Laser Rayleigh Scattering. Applied Physics B, 61, 69-72. Desgroux, P., Minetti, R., and Sochet, L. R. (1996) Temperature Distribution Induced by Pre-Ignition Reactions in a Rapid Compression Machine. Combust. Sci. Tech., 113114, 193-203. Dixon-Lewis, G., and Williams, D. J. (1977) Comprehensive Chemical Kinetics. Vol. 17, Elsevier, Amsterdam, p. 1-248 Donovan, M. T., He, X., Palmer, T. R., Zigler, B. T., Wooldridge, M. S., and Atreya, A. (2004) Demonstration of a Free-piston Rapid Compression Facility for the Study of High Temperature Combustion Phenomenon. Combust. Flame, 137, 351-365. Donovan, M.T., He, X., Zigler, B., Palmer, T.R., Walton, S.M., and Wooldridge, M.S. (2005) Experimental Investigation of Silane Combustion and Particle Nucleation Using a Rapid-compression Facility. Combust. Flame, 141, 360-370. Gallagher, S. M., Curran, H. J., Wurmel, J., and Simmie, J. M. (2003) The Oxidation of Propane at Elevated Pressures in a Rapid Compression Machine: An Experimental
171
and Modeling Study. Proc. of the Third Joint Meeting of the US Sections of the Combust. Inst.. Gardiner, W. C., McFarland, M., Morinaga, K., Takeyama, T., and Walker, B. F. (1971) Initiation Rate for Shock-Heated Hydrogen-Oxygen-Carbon Monoxide-Argon Mixtures as Determined by OH Induction Time Measurements. J. Phys. Chem., 75, 10, 1504-1509. Gardiner, W. C., Mallard, W. G., McFarland, M., Morinaga, K., Owen, J. H., Rawlins, W. T., Takeyama, T., and Walker, B. F. (1972) Elementary Reaction Rates from Post-Induction-Period Profiles in Shock-Initiated Combustion. Proc. Combust. Inst., 14, 61-75. Gardiner, Jr., W. C., and Olson, D. B. (1980) Chemical Kinetics of High Temperature Combustion. Ann. Rev. Phys. Chem., 31, 377-399 Glassman, I. (1996) Combustion. 3rd Edition, Academic Press. Griffiths, J. F., Jiao, Q., Schreiber, A., Meyer, J., Knoche, K. F., and Kardylewski, W. (1993a) Experimental and Numerical Studies of Ditertiary Butyl Peroxide Combustion at High Pressure in a Rapid Compression Machine. Combust. Flame, 93, 303-315. Griffiths, J. F., Halford-Maw, P. A., and Rose, D. J., (1993b) Fundamental Features of Hydrocarbon Autoignition in a Rapid Compression Machine. Combust. Flame, 95, 291-306. Griffiths, J. F., Halford–Maw, P. A., and Mohamed, C. (1997) Spontaneous Ignition Delays as a Diagnostic of the Propensity of Alkanes to Cause Engine Knock. Combust. Flame, 111, 327-337.
172
Griffiths, J. F., MacNamara, J. P., Mohamed, C., Whitaker, B. J., Pan, J., and Sheppard, C. G. W. (2001) Temperature Fields During the Development of Autoignition in a Rapid Compression Machine. Faraday Discuss., 119, 287-303. Griffiths, J. F., and Whitaker, B. J. (2002a) Thermokinetic Interactions Leading to Knock During Homogeneous Charge Compression Ignition. Combust. Flame, 131, 386399. Griffiths, J. F., MacNamara, J. P., Shepperd, C. G. W., Turton, D. A., and Whitaker, B. J. (2002b) The Relationship of Knock During Controlled Autoignition to Temperature Inhomogeneities and Fuel Reactivity. Fuel, 81, 2219-2225. Hardy, J. E., and Lyon, R. K. (1980) Isothermal Quenching of the Oxidation of Wet CO. Combust. Flame, 39, 317-320. Heffel, J. W. (2003) NOx Emission and Performance Data for a Hydrogen Fueled Internal Combustion Engine at 3000 rpm Using Exhaust Gas Recirculation. Int. J. Hydrogen Energy, 28, 1285–1292. Hippler, H., Troe, J., and Willner, J. J. (1990) Shock Wave Study of the Reaction HO2 + HO2 → H2O2+O2: Confirmation of a Rate Constant Minimum Near 700 K. J. Chem. Phys., 93, 1755-1760. Hirscher, M., and Becher, M. (2003) Hydrogen Storage in Carbon Nanotubes. J. Nanosci. Nanotech, 3, 3–17. Hu, H., and Keck, J. (1987) Autoignition of Adiabatically Compressed Combustible Gas Mixture. SAE 872110
173
Huang, Y., Sung, C. J., and Eng, J. A. (2002) Effects of n-Butane Addition on Reformer Gas Combustion: Implications for the Potential of Using Reformer Gas For an Engine Cold Start. Proc. Combust. Inst., 29, 759-765. Kee, R. J., Rupley, F. M., and Miller, J. A. (1989) Chemkin-II: A Fortran Chemical Kinetics Package for the Analysis of Gas Phase Chemical Kinetics. Sandia National Laboratories, Report No. SAND89-8009B. Kim, T. J., Yetter, R. A., and Dryer, F. L. (1994) New Results on Moist CO Oxidation: High Pressure, High Temperature Experiments and Comprehensive Kinetic Modeling. Proc. Combust. Inst., 25, 759-766. Kitsopanidis, I. (2004) Experimental and Computational Study of Soot Formation Under Diesel Engine Conditions. Ph.D. Thesis, Department of Mechanical Engineering, MIT. Kordylewski, W., Scott, S. K. (1984) The Influence of Self-heating on the Second and Third Explosion Limits in the O2+H2 Reaction. Combust. Flame, 57, 127-139. Lee, D., and Hochgreb, S. (1998a) Rapid Compression Machine: Heat Transfer and Suppression of Corner Vortex. Combust. Flame, 114, 531-545. Lee, D., and Hochgreb, S. (1998b) Hydrogen Autoignition at Pressures above the Second Explosion Limit (0.6–4.0 MPa). Int. J. Chem. Kin., 30, 385-406. Li, J., Zhao, Z., Kazakov, A., and Dryer, F. L. (2004a) An Updated Comprehensive Kinetic Model of Hydrogen Combustion. Int. J. Chem. Kin., 36, 10, 566–575. Li, J., Zhao, Z., Kazakov, A., and Dryer, F. L. (2004b) A Comprehensive Kinetic Mechanism for C1 Species Combustion. Proc. Combust. Inst., 30, 1F1-04 WIP Poster
174
Lloyd, A. C., (1974) Evaluated and Estimated Kinetic Data for Phase Reactions of the Hydroperoxyl Radical. Int. J. Chem. Kin., 6,169-228 Lutz, A. E., Kee, R. J., and Miller, J. A. (1988) Senkin: A FORTRAN Program for Predicting Homogeneous Gas Phase Chemical Kinetics with Sensitivity Analysis. Sandia National Laboratories, Report No. SAND 87-8248. Lyon, R. K., Hardy, J. E., and William, V. H. (1985) Oxidation Kinetics of Wet CO in Trace Concentrations. Combust. Flame, 61, 79-86. Meyer, J. W., and Oppenheim, A. K. (1970) On the Shock Induced Ignition of Explosive Gases. Proc. Combust. Inst., 13, 1153-1164. Minetti, R., Ribaucour, M., Carlier, M., Fiischen, C., and Sochet, L. R. (1994) Experimental and Modeling Study of Oxidation and Autoignition of Butane at High Pressure. Combust. Flame, 96, 201-211. Minetti, R., Ribaucour, M., Carlier, M., Therssen, E., and Sochet, L. R. (1995) A Rapid Compression Machine Investigation of Oxidation and Auto-ignition of n-Heptane: Measurements and Modeling. Combust. Flame, 102, 298-309. Minetti, R., Carlier, M., Ribaucour, M., Therssen, E., and Sochet, L. R. (1996a) Comparison of Oxidation and Autoignition of the Two Primary Reference Fuels by Rapid Compression Proc. Combust. Inst., 26, 747-753. Minetti, R., Ribaucour, M., Carlier, M., and Sochet, L. R. (1996b) Autoignition Delays of a Series of Linear and Branched Chain Alkanes in the Intermediate Range of Temperature. Combust. Sci. Tech., 113-114, 179-191.
175
Minetti, R., Roubaud, A., Therssen, E., Ribaucour, M., and Sochet, L. R. (1999) The Chemistry of Pre-ignition of n-Pentane and 1-Pentene. Combust. Flame, 118, 213220. Mueller, M. A., Kim, T. J., Yetter, R. A., and Dryer, F. L. (1999a) Flow reactor studies and Kinetic Modeling of the H2/O2 Reaction. Int. J. Chem. Kin., 31, 2, 113 - 125. Mueller, M. A., Yetter, R. A., and Dryer, F. L. (1999b) Flow Reactor Studies and Kinetic Modeling of the H2/O2/NOx and CO/H2O/O2/NOx Reactions. Int. J. Chem. Kin., 31, 705-724. O’Conaire, M., Curran, H. J., Simmie, J. M., Pitz, W. J., and Westbrook, C. K. (2004) A Comprehensive Modeling Study of Hydrogen Oxidation. Int. J. Chem. Kin., 36, 11, 603-622. Park P., and Keck J. C. (1990) Rapid Compression Machine Measurements of Ignition Delays for Primary Reference Fuels. SAE Paper 900027. Petersen, E., Davidson, D., Rohrig, M. and Hanson, R. (1996) High-pressure Shock-tube Measurements of Ignition Times in Stoichiometric H2/O2/Ar Mixtures. 20th International Symposium on Shock Waves, 941-946. Petersen, E. L., Kalitan, D. M., Rickard, M. J. A. (2003) Chemical Kinetics of OH* Chemiluminescence in High Temperature Reacting Flows. Proc. of the Third Joint Meeting of the U. S. Sections of the Combustion Institute, Chicago, IL USA, Paper PL05. Ribaucour, M., Minetti, R., Sochet, L. R., Curran, H. J., Pitz, W. J., and Westbrook, C. K. (2000) Ignition of Isomers of Pentane: An Experimental and Kinetic Modeling Study. Proc. Combust. Inst., 28, 1671-1678.
176
Roblee, L. H. S. (1961) Technique for Sampling Reaction Intermediates in a Rapid Compression Machine. Combust. Flame, 5, 229-234. Roesler, J. F., Yetter, R. A., and Dryer, F. L. (1994) On the Dependence of the Rate of Moist CO Oxidation on O2 Concentration at Atmospheric Pressure. Combust. Sci. Tech., 95, 161-171. Rogowski, A. R. (1963) A New Machine for Studying Combustion of Fuel Sprays With Contolled Air Motion. SAE Preprint 436F. Saytzev, S. G., and Soloukin, R. I. (1962) Study of Combustion of an AdiabaticallyHeated Gas Mixture. Proc. Combust. Inst., 8, 344-347. Schott, G. L., and Kinsey, J. L. (1958) Kinetic Studies of Hydroxyl Radicals in Shock Waves. II. Induction Times in the Hydrogen-Oxygen Reaction. J Chem. Phys. 29, 1177–1182. Silke, E. J., Curran, H. J., and Simmie, J. M. (2005) The Influence of Fuel Structure on Combustion as Demonstrated by the Isomers of Heptane: a Rapid Compression Machine Study. Proc. Combust. Inst., 30, 2639-2647. Sivaramakrishnan, R., Tranter, R.S., and Brezinsky, K. (2004) High-pressure, Hightemperature Oxidation of Toluene. Combust. Flame, 139, 340-350. Sivaramakrishnan, R., Comandini, A., Tranter, R. S., and Brezinsky, K. (2005) High Pressure Experiments on the Combustion of H2/CO Mixtures. 6th Int. Conf. on Chem. Kin., NIST, Poster L14. Skinner, G. B., and Ringrose, G. H. (1965) Ignition Delays of a Hydrogen—Oxygen— Argon Mixture at Relatively Low Temperatures. J Chem. Phys, 42, 2190–2192.
177
Slack, M. W. (1977) Rate Coefficient for H + O2 + M = HO2 + M Evaluated from Shock Tube Measurements of Induction Times. Combust. Flame, 28, 241–249. Smith, G. P., Golden, D. M., Frenklach, M., Moriarty, N. W., Eiteneer, B., Goldenberg, M., Bowman, C. T., Hanson, R. K., Song, S., Gardiner, W. C., Lissianski, V. V., and Qin, Z. (1999) GRI-Mech 3.0, http://www.me.berkeley.edu/gri_mech. Strehlow, R. A., and Cohen, A. (1962) Initiation of Detonation. Physics of Fluids, 5, 97101. Sung, C. J., Huang, Y., and Eng, J. A. (2001) Effects of Reformer Gas Addition on the Laminar Flame Speeds and Flammability Limits of n-Butane and iso-Butane Flames. Combust. Flame, 126, 1699-1713. Tanaka, S., Ayala, F., Keck, J. C., and Heywood, J. B. (2003a) Two-Stage Ignition in HCCI Combustion and HCCI Control by Fuels and Additives. Combust. Flame, 132, 219-239. Tanaka, S., Ayala, F., and Keck, J. C. (2003b) A Reduced Chemical Kinetic Model for HCCI Combustion of Primary Reference Fuels in a Rapid Compression Machine. Combust. Flame, 133, 467-481. Thurber, M. C., Grisch, F., and Hanson, R. K. (1997) Temperature Imaging With Singleand Dual-wavelength Acetone Planar Laser-Induced Fluorescence. Opt. Lett., 22, 251–253. Thurber, M. C., Grisch, F., Kirby, B. J., Votmeier, M., and Hanson, R. K. (1998) Measurements and Modeling of Acetone Laser-Induced Fluorescence with Implications for Temperature-Imaging Diagnostics. Appl. Opt., 37, 4963–4978.
178
Vandooren, J., Oldenhove de Guertechin, L., and Van Tiggelen, P. J. (1986) Kinetics in Lean Formaldehyde Flame. Combust. Flame, 64, 127-139. Vardanyan, I. A., Sachyan, G. A., and Nalbandyan, A. B. (1975) The Rate Constant of the Reaction HO2+CO = CO2+OH. Int. J. Chem. Kin., 7, 23-31. Vermeersch, M. L., Held, T. J., Stein, Y., and Dryer, F. L. (1991) Auto-Ignition Chemistry Studies of n-Butane in a Variable Pressure Flow Reactor. SAE Paper 912316. Voevodsky, V. V., and Soloukhin, R. I. (1965) On the Mechanism and Explosion Limits of Hydrogen-Oxygen Chain Self-Ignition in Shock Waves. Proc. Combust. Inst., 10, 279-283. Warnatz, J. (1984) Combustion Chemistry, Editor (Gardiner, Jr., W. C.). 197 -346 Westbrook, C. K., and Dryer, F. L. (1984) Chemical Kinetic Modeling of Hydrocarbon Combustion. Prog. Energy Combust. Sci., 10, 1-57 Westbrook, C. K., Pitz, W. J., Boercker, J. E., Curran, H. J., Griffiths, J. F., Mohamed, C., and Ribaucour, M. (2002) Detailed Chemical Kinetic Reaction Mechanisms for Autoignition of Isomers of Heptane under Rapid Compression. Proc. Combust. Inst., 29, 1311-1318. Wurmel, J., and Simmie, J. M. (2005) CFD Studies of a Twin-piston Rapid Compression Machine. Combust. Flame, 141, 417-430. Yetter, R. A., Dryer, F. L., and Rabitz, H. (1991a) A comprehensive Reaction Mechanism for Carbon Monoxide/ Hydrogen/ Oxygen Kinetics. Combust. Sci. and Tech., 79, 97-128.
179
Yetter, R. A., Dryer, F. L., and Rabitz, H. (1991b) Flow Reactor Studies of Carbon Monoxide/ Hydrogen/ Oxygen Kinetics. Combust. Sci. Tech., 79, 129-140. Yetter, R. A., Rabitz, H., and Hedges, R. M. (1991c) A Combined Stability-Sensitivity Analysis of Weak and Strong Reactions of Hydrogen/Oxygen Mixtures. Int. J. Chem. Kin., 23, 251-278. Yetter, R. A., and Dryer, F. L. (1992) Inhibition of Moist Carbon Monoxide Oxidation at Trace Amounts of Hydrocarbons. Proc. Combust. Inst., 24, 757-767. Zarubiak, D. C. Z. (1997) Study of Acetaldehyde Combustion Kinetics in a Variable Pressure Flow Reactor. M.S. Thesis, Department of Mechanical and Aerospace Engineering, Princeton University, Princeton, NJ.
