Experimental and Kinetic Modeling Study of Ethyl Levulinate Oxidation in a Jet-Stirred Reactor
Thesis by Jui-Yang Wang
In Partial Fulfillment of the Requirements For the Degree of Master of Science in Chemical and Biological Engineering
King Abdullah University of Science and Technology Thuwal, Kindom of Saudi Arabia
May, 2017
2
EXAMINATION COMMITTEE PAGE
The thesis of Jui-Yang Wang is approved by the examination committee.
Committee Chairperson: Professor Subram Maniam Sarathy Committee Members: Professor Aamir Farooq, Professor Robert W. Dibble, Professor Klaus-Viktor Peinemann
3
COPYRIGHT PAGE
© May, 2017 Jui-Yang Wang All Rights Reserved 4
ABSTRACT
A jet-stirred reactor was designed and constructed in the Clean Combustion
Research Center (CCRC) at King Abdullah University of Science and Technology
(KAUST); was validated with n-heptane, iso-octane oxidation and cyclohexene pyrolysis.
Different configurations of the setup have been tested to achieve good agreement with results from the literature. Test results of the reactor indicated that installation of a pumping system at the downstream side in the experimental apparatus was necessary to avoid the reoccurrence of reactions in the sampling probe.
Experiments in ethyl levulinate oxidation were conducted in the reactor under several equivalence ratios, from 600 to 1000 K, 1 bar and 2 s residence time. Oxygenated species detected included methyl vinyl ketone, levulinic acid and ethyl acrylate. Ethylene, methane, carbon monoxide, hydrogen, oxygen and carbon dioxide were further quantified with a gas chromatography, coupled with a flame ionization detector and a thermal conductivity detector.
The ethyl levulinate chemical kinetic model was first developed by Dr. Stephen
Dooley, Trinity College Dublin, and simulated under the same conditions, using the
Perfect-Stirred Reactor code in Chemkin software. In comparing the simulation results with experimental data, some discrepancies were noted; predictions of ethylene production were not well matched. The kinetic model was improved by updating several classes of reactions: unimolecular decomposition, H-abstraction, C-C and C-O beta-scissions of fuel radicals. The updated model was then compared again with experimental results and good agreement was achieved, proving that the concerted eliminated reaction is crucial for the kinetic mechanism formulation of ethyl levulinate. In addition, primary reaction pathways 5 and sensitivity analysis were performed to describe the role of molecular structure in combustion (800 and 1000 K for ethyl levulinate oxidation in the jet-stirred reactor).
6
ACKNOWLEDGEMENTS
I would like to thank my advisor, Prof. S. Mani Sarathy, who provided me plenty of opportunities and ideas for this work and inspired me in different ways.
I thank my thesis committee members, including Profs. Aamir Farooq, Robert W.
Dibble, and Klaus-Viktor Peinemann, for correcting my thesis and providing valuable comments. I would especially like to thank Dr. Zhandong for helping out setting up the jest-stirred reactor with unlimited support, assistance, and advices.
It is a privilege to finish the thesis alongside a group of talented and passionate scientists in Clean Combustion Research Center. A very big thank you to Nour Atef and
Samah Mohamed for their help on optimizing the kinetic model and simulation. To several patient academic- and life-mentors: Eshan Singh, Ehson Nasir, and Bingjie Chen who helped me a lot on experimental setup and validation, I am deeply indebted. Special thanks to Tony Bissoonauth for improving the fluency of my thesis.
To some senior scientists Zeyad Talla, Misjudeen Raji, and Najeh Kharbatia from
Analytical Core Laboratory, I express my appreciation for their kindly technical assistance on coupling my setup with their apparatus for species analysis.
Particularly, many thanks go to some best friends inside or outside KAUST, Joel,
Aniela, Jam, Mafer, Moe, Joao, Daniel, Zhibek, Omar, Dora, Ken, Charlene, and Robert for being there with me for all the ups and lows and sharing their memories and experience with me.
Finally I would like to thank my parents and my sister for their patience, encouragement and support through my education. 7
TABLE OF CONTENTS
EXAMINATION COMMITTEE PAGE ...... 2 COPYRIGHT PAGE ...... 3 ABSTRACT ...... 4 ACKNOWLEDGEMENTS ...... 6 LIST OF FIGURES ...... 9 LIST OF TABLES ...... 12 Chapter 1 INTRODUCTION ...... 13 Chapter 2 BACKGROUND RESEARCH ...... 15 2.1 Introduction to Bio-derived Fuels ...... 15 2.2 Biofuel Use in Engines ...... 15 2.3 Bio-derived Fuel Combustion Chemistry ...... 17 2.3.1 Alcohols Chemistry ...... 17 2.3.2 Biodiesel Chemistry ...... 18 2.4 From Lignocellulosic Biomass to Ethyl Levulinate ...... 21 Chapter 3 JET-STIRRED REACTOR VALIDATION ...... 25 3.1 Literature Review on Biofuels Combustion in JSR ...... 25 3.2 Experimental Set-Up at KAUST ...... 30 3.2.1 Fuel Injection System...... 33 3.3 Verification of Criteria for KAUST JSR ...... 36 3.4 Temperature Homogeneity Measurement ...... 38 3.5 Sampling Methods ...... 39 3.5.1 Downstream Sampling Design ...... 42 3.6 Analytical Techniques ...... 45 3.6.1 Gas Chromatography ...... 46 3.6.2 Column Selection ...... 48 3.6.3 Principles of GC as a Refinery Gas Analyzer (RGA) ...... 49 3.6.4 Principles of GC/MS/FID/TCD ...... 52 Chapter 4 JSR VALIDATION EXPERIMENT ...... 58 4.1 N-heptane Oxidation ...... 58 4.1.1 Oxidation Experiment of n-heptane ...... 61 4.2 Cyclohexene Pyrolysis ...... 65 4.3 Iso-octane Oxidation ...... 67 Chapter 5 ETHYL LEVULINATE MODELING AND EXPERIMENT ...... 69 5.1 Ethyl levulinate Oxidation Mechanism ...... 69 8
5.1.1 Kinetic Model ...... 69 5.1.2 Ethyl Levulinate Thermochemistry ...... 74 5.2 Ethyl Levulinate Oxidation Experiment ...... 75 5.3 Kinetic Model Adjustment ...... 79 5.3.1 H-abstraction from The Fuel ...... 79 5.3.2 Fuel Radical Derived C-C and C-O Scissions...... 83 5.3.3 Fuel Unimolecular Decomposition ...... 85 5.3.4 EL Concerted Elimination Reaction ...... 87 5.4 Simulation Results from Updated Kinetic Mechanism ...... 88 Chapter 6 CONCLUSION AND FUTURE WORK ...... 94 REFERENCES ...... 97 SUPPLEMENTAL MATERIAL ...... 109 Low-temperature Oxidation Chemistry ...... 109 Introduction to Kinetics and Thermodynamics ...... 112
9
LIST OF FIGURES
Figure 2.1: Ethanol elimination reaction ...... 18 Figure 2.2: (a) Transesterification from triglyceride and methanol to esters and glycerin. R1, R2, R3 represent hydrocarbon chains of fatty acid [24]. (b) Esterification of free fatty acid in oil with a catalyst sulfuric acid, R1 as a hydrocarbon chain...... 20 Figure 2.3: (a) methyl butanoate (b) ethyl propanoate unimolecular decomposition. Oxygens from 1 to 2 in green, carbons from 1 to 4 in red. Ea for each transition state was calculated by El-Nahas et al. with several techniques in Gaussian software [29]...... 21 Figure 2.4: Pathways from biomass to levulinic acid via acid-catalyzed hydrolysis process [33]...... 22 Figure 3.1: JSR with crossed nozzles configuration scheme...... 25 Figure 3.2: Scheme of the experimental set-up at KAUST...... 32 Figure 3.3: Scheme of homemade vaporizer (not to scale)...... 33 Figure 3.4: Scheme of microprobe for fuel introduced to JSR pre-mixed zone (not to scale)...... 35 Figure 3.5: Scheme of two heating zones prior to JSR, orange region indicates preheating zone (373-493 K); red region represents furnace heating zone (> 500 K)...... 36 Figure 3.6: Jet-Stirred reactor scheme and temperature measurement at different locations. Temperature points separated by10 mm. SP represents set point where reactor is monitored and controlled...... 38 Figure 3.7: Dashed lined are temperature homogeneity oven testing of jet-stirred reactor at 650, 800, 950 and 1100K. Fig. 3.6, including position 0 as set-point spot, is portrait positioned...... 39 Figure 3.8: Schematic of jet-stirred reactor with molecular beam sampling system [93]. I and II are differential pumping and ionization chambers with mass spectrometer, respectively. 1, sVUV beamline; 2, to turbo molecular pump; 3, molecular beam from sampling gas mixture; 4, reflectron time-of-flight mass spectrometer; 5, ion trajectory; 6, MCP; 7, quartz cone-like nozzle; 8, jet-stirred reactor...... 42 Figure 3.9: Syringe extraction scheme. The red sampling port is a septum used for the GC injector...... 44 Figure 3.10: GC instrument scheme, including carrier gas as helium, gas injector, stationary-phase column, 3-way splitters to different detectors...... 47 Figure 3.11: GC injector scheme [23]...... 47
Figure 3.12: Plumbing diagram in GC-RGA. V1 and V5 are ten-port valves with backflush function. V2, V3 are six-port valves with functions of column isolation to TCD, column sequence reversal with backflush of the pre-column. V4 is the 10
second six-port valve with a loop for sampling upstream between V1 and V5 [97]...... 49 Figure 3.13: Flowsheet of light hydrocarbon gases in valve one and two prior to backflush process...... 50 Figure 3.14: Flowsheet of valve two as isolation valve is switched on, valve one is off with backflushing...... 51 Figure 3.15 Flowsheet of hydrocarbons in valves three and four, before backflush ...... 51 Figure 3.16: Flowsheet of backflush and reverse flow as valve three is switched on...... 52 Figure 3.17: Simplified scheme of Mass Spectrometer Detector with electrical ionization method as the ion source...... 53 Figure 3.18: Mass spectrum of ethyl levulinate from NIST library...... 54 Figure 3.19: A basic design of flame ionization detector [98]...... 55 Figure 3.20: Bridge circuit used in two-cell detector. R3 is used for sample flow detection from column. R4 measures a reference flow from nitrogen or helium [96]...... 57 Figure 4.1: Fuel injection probe I.D. 1.6 mm with 0.7 mm exit. From top to bottom: without sampling probe; with sampling probe; with sampling probe and pumping device...... 61 Figure 4.2: Experimental results for n-heptane oxidation with 0.5% fuel concentration, 1 atm, 2 s residence time and equivalence ratio 1. Figure 4.1-a without sampling probe, 4.1-b with sampling probe, respectively. Configurations for blue and red dots in portrait position. From left to right, from top to bottom: n-C7H16, O2, CO2, and CO...... 64 Figure 4.3: Comparison of results of n-heptane, O2, CO, and C2H4 (top to bottom, left to right) with different experimental configurations...... 65 Figure 4.4: Experimental results of cyclohexene pyrolysis in JSR. Only ethylene and 1,3- butadiene quantified under 1 atm, 0.1% fuel and 1s residence time from 750 to 1025 K...... 66 Figure 4.5: Experimental results of cyclohexene pyrolysis with sampling probe and pumping device configuration with identical conditions and modeling as Fig. 4.4...... 66 Figure 4.6: JSR species concentration profiles (points) and simulations, using updated mechanism from Atef et al. [102] for 0.5% fuel and 2 s residence time under atmospheric pressure...... 68 Figure 5.1: (a) Bond dissociation energy of ethyl propanoate (EP) (kJ/mol). Black numbers are C-C or C-O BDE, red numbers are C-H BDE [29]. α-carbon is designated to the carbon next to carbonyl group (C(=O)). (b) Radical stabilization by the carbonyl group...... 70 Figure 5.2: Chemical structure and bond dissociation energy (BDE) of (a) ethyl levulinate calculated by Dooley et al. [123], (b) ethyl propanoate (EP) [122] and (c) methyl 11
ethyl ketone (MEK) [121]. Red numbers represent C-H BDE, black numbers are C-C or C-O BDE. (Unit: kcal/mol) ...... 71 Figure 5.3: GC-MS spectrum of EL oxidation under 850K in JSR unknown species identified with MS detector. From A to G: A: nitrogen; B: ethanol; C: methyl vinyl ketone; D: ethyl acrylate, E: angelica lactone, F: ethyl levulinate and G: levulinic acid...... 76 Figure 5.4: Temperature profiles of ethyl levulinate (EL) oxidation, with 0.5% fuel, 1 bar, 2 s residence time and equivalence ratio from top to bottom: φ= 0.5, 1, 2...... 78 Figure 5.5: Radicals or fuels nomenclature of Ethyl levulinate, ethyl propanoate, and methyl levulinate...... 81 Figure 5.6: Rate coefficient comparison of H-abstraction from fuel. Solid line: original rates adopted from other molecules; dashed line: rates from the methyl levulinate theoretical kinetic study [44]...... 82 Figure 5.7: Rate coefficient comparison of EL3J decomposing to smaller species via beta-scission at different C-C bond positions...... 85 Figure 5.8: Rate coefficient comparison of EL unimolecular decomposition, including C- C and C-O bond scission...... 86 Figure 5.9: Rate coefficient comparison in two different studies. Black line is EP oxidation mechanism [27], used in this study. Red line is shock tube experimental results [45] ...... 88 Figure 5.10: Updated mechanism of EL oxidation with experimental data identical to Fig 5.4. Solid lines represent updated model. Dashed lines are from previous simulation...... 91 Figure 5.11: Reaction path analysis for EL/O2 mixture at 1atm, 800 K (red) and 1000 K (black), φ = 1. Percentages represent decomposition percent from parent to daughter species...... 92 Figure 5.12: Sensitivity coefficients for 0.5 % ethyl levulinate in N2 under (a) 800 and (b) 1000 K at 1 bar, φ = 1...... 93
12
LIST OF TABLES
Table 2.1: Heating values of ethyl levulinate and other typical fuels...... 23 Table 3.1: Comparison of experimental conditions for jet-stirred reactor facilities in CNRS and Nancy...... 26 Table 3.2: Literature review of experimental data of biofuel oxidation in JSR...... 26 Table 3.3: Physical properties of the fuel tested and measured...... 34 Table 3.4: Runtime event for all valves in RGA...... 52 Table 3.5: Contributions to the effective carbon number [99]...... 55 Table 4.1: JSR experimental parameter comparison between Nancy and KAUST ...... 59 Table 5.1: Ethyl levulinate and ethyl propanoate mechanism rate coefficients for H abstraction by OH radical; cm3 mol-1 s-1 cal-1 units...... 81 Table 5.2: Rate coefficients for fuel radical derived C-C and C-O scissions, only EL1J, EL3J and EL4J were altered...... 83 Table 5.3: Rate coefficients for EL unimolecular decomposition reactions...... 85
13
Chapter 1 INTRODUCTION
Today’s average global temperature is the warmest in recorded history; studies cite the cause as cumulative carbon dioxide (CO2) emissions [1]. Because petroleum-derived fuels are the main source of CO2 emissions, alternative renewable energy options are gaining increasing attention. As a transition from non-renewable (petroleum derived) fuels to carbon-neutral energy, biofuels are the only feasible option for replacing conventional transportation liquid fuels used in internal combustion engines, due to their liquid nature portability, renewability, high combustion efficiency, and reduced net greenhouse gas emission [2]. However, to avoid the ethical issues and social effects of first generation biofuels, the application of second or third generation biofuels has become a more viable option. Cellulosic biomass, a second generation biofuel, is the most abundant renewable resource available [3]. With several pretreatments and hydrolysis conversion, ethyl levulinate is one of the main end-products derived from lignocellulosic biomass. When tested in diesel engines, blends of ethyl levulinate with diesel fuels showed promising results [4]; however, the detailed combustion mechanism of this promising biofuel has not been fully understood or developed.
The objective of this study is to develop a detailed chemical kinetic mechanism for ethyl levulinate, one of the most promising biofuels. Dr. Stephen Dooley has completed a preliminary model development of ethyl levulinate. Several reaction classes used analogies to two mechanisms--ethyl propanoate and methyl ethyl ketone--since their mechanisms were validated with other ideal reactors. To achieve validation of the developed mechanism, an ideal reactor (the jet-stirred reactor (JSR), similar to a continuous stirred-tank reactor
(CSTR) but used for gas-phase kinetic study [5]) was implemented and constructed at the 14
Combustion and Pyrolysis Chemistry Laboratory. Before measuring the end-gas product profiles of the target biofuel, the JSR model should be validated by conducting oxidation and pyrolysis experiments with several reference fuels, e.g. n-heptane, iso-octane and cyclohexene. Once the setup validation has been achieved, the oxidation of ethyl levulinate can be studied. Hydrocarbons were detected and quantified using a gas chromatography equipped with a flame ionization detector and a thermal conductivity detector
(GC/FID/TCD).
Further detailed adjustments have been made to the existing EL mechanism based on comparison of experimental and simulation results. Eventually, reaction flux and sensitivity analyses of ethyl levulinate were performed to describe the influence of several fuel-related reactions. 15
Chapter 2 BACKGROUND RESEARCH
2.1 Introduction to Bio-derived Fuels
In December 2015 the United Nations Framework Convention on Climate Change
(UNFCCC) proposed a new international climate agreement to prevent natural catastrophes caused by greenhouse gas (GHG) emissions; a conference of the parties (COP21) was held in Paris, France. Their aim was worldwide reduction of GHG emissions and petroleum dependence. An agreement was made to avoid the rise of global average temperatures above 2 oC by the end of the 20th century. The four main ways to meet this goal included carbon capture and storage, energy efficiency, nuclear power and renewable energy [6].
For renewable energy, biofuels, are the most viable option, produced from biological products like biomass, liquid and gaseous fuel [7, 8]. Their main advantages are life-cycle energy and reduced greenhouse gas emissions. Hossain et al. [9] concluded that the total input energy required to produce one MJ of soybean oil is 0.31 MJ, while the energy needed for one MJ of fossil fuel is 1.2 MJ. In addition, life-cycle CO2 emission is evaluated as zero, since the plants, as the sources of biofuels, absorb CO2 during photosynthesis. One of the most valuable applications for biofuel is in combustion engines, where liquid phase bio-oils can be used after blending with conventional hydrocarbons.
2.2 Biofuel Use in Engines
Biofuels can be used in both spark-ignition and compression-ignition transport engines [10]. The requirements for alternative transport fuels are that they be in liquid phase and miscible with conventional hydrocarbons. The molecular structure of biofuels is similar to petroleum-derived fuels, and engines require little or no alteration to accommodate them. Molecular structure is decisive in biofuel application; they can be 16 classed either as biodiesel or bio-gasoline for use in compression-ignition and spark- ignition engines, respectively.
In compression-ignition (CI) engines, fuel quality evaluation is based on auto- ignition property. The cetane number (CN) can be used as an index of fuel auto-ignition property, and the value for different fuels is derived from two primary reference fuels: n- hexadecane (CN = 100) and 2,2,4,4,6,8,8-heptamethylnonane (iso-octane, CN = 15). The higher the CN the higher the reactivity, as shown by short ignition delay time. Recently,
Hossain et al. [9] and Ramadhas et al. [11] tested a range of raw plants and vegetable oils in CI engines, measuring and reviewing several physical properties of bio-oils, like cloud point, viscosity etc., and evaluating their pollutant emissions. Karanja oil (CN = 47) was one example tested [12]. Comprised of several fatty acids such as oleic acid
(CH3(CH2)7CH=CH(CH2)7COOH), it has a long aliphatic chain with one double bond in the middle of its molecular structure. By blending Karanja methyl ester oil with diesel fuel, exhaust gases were analyzed and results indicated that Karanja fuel blends reduced CO, hydrocarbon and NOX exhaust emissions [13].
Spark-ignition (SI) engines, on the contrary, require fuels that are less reactive in order to avoid engine knock. Pre-ignition is a phenomenon in which unburned fuel-air mixtures spontaneously ignite before the spark timing. This unfavorable effect is followed by engine knocking that results in reduced efficiency and even engine damage [14]. Some biofuels, such as bio-derived ethanol, are proven to be better knock-resistant fuels than typical gasoline fuels; and fuel additives with better knock resistance lead to higher compression ratios and increasing the engine efficiency [14]. Therefore, mixing ethanol, or other bio-derivatives as a typical octane booster with gasoline, appears to have an 17 important beneficial effect. The research octane number (RON) represents the tendency of fuels to resist auto-ignition; in ethanol, the RON is defined as 109 [15]. The primary reference fuels used for characterizing the RON scale are n-heptane (RON = 0) and 2,2,4- trimethyl-pentane (iso-octane, RON = 100). Wei-Dong et al. [16] investigated the engine performance and emissions of different mixtures of ethanol and gasoline (0%-30% ethanol); their experimental results showed that a higher blending percentage significantly decreased the emission of CO and unburned hydrocarbon, but increased CO2 emission quantity; this is similar to emissions from biodiesel-diesel fuel blends. However, in terms of heat release rate, calorific value and pressure in the combustion chamber, the engine performance of gasoline-ethanol blends does not achieve the same level as fossil fuels [17].
2.3 Bio-derived Fuel Combustion Chemistry
Combustion is a series of species production and consumption that includes thousands of reactions, eventually releasing heat in the form of power for engine use. These reaction classes include decompositions and re-combinations of both pyrolysis and oxidation. Because they contain oxygen atoms that promote engine oxidation and achieve nearly complete combustion, biofuels are treated as partially oxidized hydrocarbons, even to their molecular structure. However, disparities in functionality produce unpredictable effects on basic combustion chemistry, directly impacting fuel blending and engine selection. In the following section, another classification of biofuels is discussed which includes alcohol and carboxyl in its functional group.
2.3.1 Alcohols Chemistry
Among all the alcohols extracted from biomass, ethanol has been the most widely used. Ethanol is primarily derived from corn and sugar through a fermentation process; in 18
2015 global ethanol production surpassed 20 billion gallons [18]. The United States and
Brazil are the two major ethanol fuel producing countries, accounting for 85% of ethanol worldwide. Yet social and ethical issues, such as stable food pricing in developing countries, dictate that edible crops be grown to produce second or third generation biomass for the energy section [19] rather than being converted to ethanol.
Regarding its molecular structure, some unimolecular decompositions are critical to high-temperature combustion. With hydroxyl functionality, the oxygen atom in alcohol attracts a hydrogen atom bonded to a carbon (Fig. 2.1), forming hydrogen bonding that causes direct molecular elimination reactions (because hydrogen bonding is strong enough to break the C-H and C-O bonds), and producing alkene and water. However, because of the electronegativity that promotes hydrogen abstraction from other radicals during combustion, the oxygen atom weakens neighboring C-H bonds.
+
ethanol ethylene + water
Figure 2.1: Ethanol elimination reaction 2.3.2 Biodiesel Chemistry
Biodiesels are mostly extracted from vegetable oils and animal fats. Typically, mono-alkyl esters are produced after transesterification from long-chain fatty acid and alcohols; the carbon number ranges from 12 to 18. However, because of their high viscosity and combustion-relative problems [20], neat vegetable oils are not appropriate for direct feeding into diesel engines. Farwood et al. evaluated short- and long-term effects, such as 19 plugging and gumming of filters, lines and injectors, and coking of a piston injector due to high viscosity [21]. One way to improve the fuel properties of vegetable oils and avoid fuel polymerization is to produce monoesters from triglyceride and methanol using a base as a catalyst (Fig. 1.2 (a)), also known as transesterification. These monoesters usually contain long aliphatic chains of 12 to 18 carbons with 10 to 11% oxygen weight percentage [22], and they exhibit high CN for diesel engine use.
Free fatty acids (FFA) are also derived from biomass such as oleic acid in Karanja oil. FFAs easily react with the trans-esterifying catalysts by forming soap and water, complicating the conversion process and reducing the yield of esters. Pretreatment is required if the FFA in oil is higher than 5%. To solve the problem, FFAs can undergo esterification with methanol and a strong acid (such as sulfuric acid), so that the hydroxyl group is replaced by an alkyl group and converted to methyl ester and water, described in
Fig. 1.2 (b) [23].
(a) O
O
= = CH3 – O – C –R1 CH2 – O – C –R1 CH2 – OH
O O
= = CH – O – C –R + CH – OH 2 + 3 CH3OH CH3 – O – C –R2 O
= O
CH – O – C –R = CH – OH 2 3 CH – O – C –R 2 3 3
(b) O H2SO4 O
= = + CH OH + H O HO – C –R 3 CH - O – C –R 2 1 3 1 20
Figure 2.2: (a) Transesterification from triglyceride and methanol to esters and glycerin.
R1, R2, R3 represent hydrocarbon chains of fatty acid [24]. (b) Esterification of free fatty
acid in oil with a catalyst sulfuric acid, R1 as a hydrocarbon chain.
Some short carbon-chain esters can be blended with gasoline as octane boosters [25].
At the molecular level, a carboxyl group is the significant difference between some biofuels and conventional gasoline or diesel fuel. Under high temperature combustion conditions
(> 700 K), unimolecular decomposition is critical and its rate coefficient must be precisely determined. In general, high temperatures help alkyl esters to decompose to carboxylic acids and olefins through four or six-member transition state rings [26]. Taking methyl butanoate (MB) and ethyl propanoate (EP) for example, 50% of MB fuel is consumed at
10 bar, 900 K, through a six-centered concerted elimination reaction, shown in Fig. 2.3 (a).
The most favorable hydrogen site to be attacked by the carbonyl group is on C4, due to the six-membered transition state ring. When the C4-H bond breaks, a beta-scission reaction occurs via C2-C3 bond cleavage which produces ethylene and methyl ethanoate. EP has a similar reaction (Fig. 2.3 (b)). The O1 connected to C1 tends to bond with the hydrogen on
C3, forming another six-centered transition state. Then the C3-H and O2-C2 bonds break followed by O1-H single and C2=O2 double bonds formation. Comparing the two transition states, the activation energy of the six-centered ring for EP is 75.53 kJ/mole lower than MB. This decrease in activation energy results in higher reactivity and faster ignition of the EP molecule [27-29]. 21
(a) 1 4 1
2 3 Ea = 209 kJ 3 1 2 4 2 1 2 1 4 3 1
2 2
(b) 1 3
1 2 2 1 1 3 2 3 2 Ea = 284.5 kJ 1 1 2 2
Figure 2.3: (a) methyl butanoate (b) ethyl propanoate unimolecular decomposition.
Oxygens from 1 to 2 in green, carbons from 1 to 4 in red. Ea for each transition state was
calculated by El-Nahas et al. with several techniques in Gaussian software [29].
2.4 From Lignocellulosic Biomass to Ethyl Levulinate
Aside from chemistry moieties, biofuels can be divided into several categories, based on their feedstock. The Renewable Fuel Standard (RFS) program [30], assigned to replace the current petroleum fuel from U.S. national policy, categorizes all biofuels into four sections: Biomass-based diesel, cellulosic biofuel, advanced biofuel and total renewable fuel. Cellulosic biofuels, such as wood or straw, achieve lifecycle GHG emission reductions as low as 60%, the highest percentage among the other three sources; it is 22 considered the most promising alternative natural resource for the sustainable supply of fuels and chemicals in the future. The amount of cellulose production is estimated at approximately 2 × 109 tons per year [31]; and these products can be utilized in food, fiber, and pharmaceutical applications. With acid-catalyzed dehydration, cellulose is decomposed into smaller molecules: glucose (or hexose), by breaking the C-O bonds connected to each cellulose unit. Furthermore, the hydrolysis of glucose produces 5- hydromethylfurfural (5-HMF), [32]. Finally, heat and an acid catalyst convert 5-HMF to levulinic acid (LA) as the main product and formic acid as a byproduct. The pathways from feedstock to the desired product--levulinic acid--are described in Fig. 2.4.
Hemicellulose Glucose Biomass Cellulose
+ Lignin -3H2O +H
+2H2O
+H+, -HCOOH
Levulinic acid 5-hydroxymethylfurfural (HMF)
Figure 2.4: Pathways from biomass to levulinic acid via acid-catalyzed hydrolysis
process [33].
In 2000, Bozell [3] proposed that LA has many potential applications for the food industry, electronics, drug delivery and fuel, and in plasticizers [34]. In the energy sector, the most promising diesel and gasoline additives from cellulosic feedstock are the esters of
LA [35] or 2-methyltetrahydrofuran (2-MTHF) [36], whose flame combustion mechanism is well understood [37]. Using an acid catalyst, levulinic esters can be synthesized via the 23 esterification of LA with alcohols. Ethyl levulinate (EL), is one of these esters, and it is produced predominantly over methyl and propyl esters, since ethanol is widely produced and commercialized.
EL is considered a potential fuel surrogate after blending with diesel fuels for several reasons: First, the physical properties of EL are similar to that of biodiesel fatty acid methyl esters (FAME). Joshi et al. [38] mentioned that blends of EL and cottonseed oil methyl esters, or poultry fat methyl esters, could improve cold flow properties by decreasing cloud and pour point, beneficial for transportation in high latitudes. Second, Christensen and coworkers [4] reported that EL may be used as a fuel lubricant to expand the lifetime of engine components. Klemm et al. [39] also suggested that a diesel engine could continue to operate, without modification to the engine components, when the percentage of EL in the blend was 20%. Moreover, the lower heating value (LHV) of EL was investigated and compared with other fuels (see Table 2.1), where LHV is defined as the heat released from the fuel combustion (initially at 25 °C), and the temperature of the combustion products is returned to 150 °C. Note that the latent heat of vaporization of water in the reaction products is not considered [40]. It is worth noting that blending 3% of EL with diesel fuel caused 0.5 MJ/kg heating value loss, but reduced CO and particulate matter emission significantly. Similarly, blends of butyl levulinate with diesel fuel can decrease soot emissions by as much as 95% [41]. However, Christensen questioned the feasibility of EL, because of its low CN and poor solubility in diesel blends at low temperatures.
Table 2.1: Heating values of ethyl levulinate and other typical fuels.
Fuel Lower Heating value (MJ/kg) Ref.
Ethyl levulinate 24.8 [42] 24
Butyl levulinate 27.4 [4]
Methanol 20.1 [43]
Ethanol 26.9 [43]
Butanol 34.4 [43]
Diesel fuel 43.2 [42]
Biodiesel 40.0 [42]
Conventional gasoline 43.4 [43]
Natural gas 47.1 [43]
3% EL in Diesel 42.7 [42]
The only theoretical and experimental studies concerning levulinic esters have been conducted by Thion et al. [44] and AlAbbad et al. [45]. Thion carried out the calculation of methyl levulinate (ML) oxidation by OH and CH3 radicals, forming ML radicals and exploring the possible pathways from ML to smaller species such as methyl vinyl ketone and methyl acrylate. AlAbbad conducted the measurement of EL unimolecular decomposition reaction to LA and ethylene under different pressure conditions at high temperature (900 to 1500 K). (Some rate coefficients in this study were adopted from their results and will be further mentioned in chapter 5.)
25
Chapter 3 JET-STIRRED REACTOR VALIDATION
Experimental work focused exclusively on constructing a new jet-stirred reactor
(JSR) facility at King Abdullah University of Science and Technology (KAUST), Clean
Combustion Research Center (CCRC).
Figure 3.1: JSR with crossed nozzles configuration scheme.
3.1 Literature Review on Biofuels Combustion in JSR
The JSR was first introduced by Villermaux [5]; it is suitable for gas-phase kinetic studies because it is taken as a zero-dimension reactor, since fuel transporting properties need not be considered, unlike co-flow or counter-flow flame. Currently, two groups are still validating gas-phase kinetic mechanisms with JSR facilities in France. In Dagaut’s model (CNRS France), the volume of the reactor is 35 cm3, equipped with four nozzles, 2 mm in diameter, to achieve perfect stirring. Liquid fuel is injected at constant flow rate, using a HPLC pump connected to a vaporizer. To reach pressures of 40 bar, the inlet and outlet pipelines are inside a stainless steel jacket, and to prevent temperature gradients, the setup is insulated with heating tapes. Nitrogen gas is the carrier gas. Species are analyzed with several gas chromatography machines installed with different columns, flame 26 ionization detectors and thermal conductivity detectors [46-48]. Another JSR setup, established in Nancy, uses a reactor volume of 95 cm3. The number of preheating zones in this setup is divided into two for progressive heating of the mixture gas [49]. Table 3.1, presents experimental setup comparisons of JSR facilities in different research institutes.
Table 3.1: Comparison of experimental conditions for jet-stirred reactor facilities in
CNRS and Nancy.
JSR in CNRS# JSR in Nancy* Pressure (bar) 10-40 1.067 Residence time (s) 0.01-3 2 Equivalence ratio φ 0.15-4.0 0.5, 1.0, 2.0, 4.0 Temperature range (K) 900-1200 500-1000 Fuel concentration (%) 0.15-0.5 0.5 Reactor volume (cm3) 35 95 Number of preheating zones 1 2
Carrier gas N2 He Sampling method online and offline online and offline # parameters were referred to [46]. * parameters in this table were adopted from n-heptane oxidation measurement [50].
A summary of available studies of fundamental reaction kinetics regarding all biofuels, including alcohols, ethers, ketone, and methyl/ethyl esters is provided in Table 3.2. The number of ethanol oxidation studies far exceeds that of other biofuels.
Table 3.2: Literature review of experimental data of biofuel oxidation in JSR.
Temperature Pressure Fuel Equivalence ratio Reference range (K) (bar) 27
Singh et al. 1979 1870 0.93 0.7-1.4 [51] Methanol Burke et al. 2016 800-1200 10 0.2-2 [52]
Togbe et al. 2009 Methanol, M85 surrogate 770-1140 10 0.35-2 [53]
Dagaut et al. Ethanol ~1000 1 0.2-2 1992 [54]
Dagaut and Ethanol, E85 surrogate 800-1100 10 0.6-2 Togbe, 2008 [55]
Galmiche et al. n-propanol 770-1190 10 0.35-2 2011 [56]
Dagaut et al. 750-1100 10 0.5-2 2009 [57] n-butanol Sarathy et al. 850-1250 1 0.25-2 2009 [58]
n-butanol/ n-heptane Dagaut and 530-1070 10 0.5 and 1 mixture Togbe, 2009 [59]
Togbe et al. 2010 2- and iso-butanol 770-1250 10 0.275-4 [60]
Serinyel et al. 2-methyl-1-butanol 700-1200 10 0.5-4 2014 [61] 28
423, 600- Togbe et al. 2010 n-hexanol 10 0.5-3.5 1100 [62]
Dayma et al. iso-pentanol 530-1220 10 0.35-4 2011 [63]
423, 770- Togbe et al. 2011 n-pentanol 10 0.7-1.4 1220 [64]
Mze-Ahmed et al. n-hexanol/ jet A-1 mixture 560-1030 10 0.5-2 2012 [65]
Cai et al. 2015 n-octanol 500-1200 10 0.5-2 [66]
Gaïl et al. 2007 800-1350 1 1.13 [67]
Sarathy et al. 850-1350 1 1 2007 [68] Methyl butanoate Gaïl et al. 2008 850-1400 1 0.375, 0.75 [69]
Hakka et al. 2010 800-850 1 0.5-1 [70]
Gaïl et al. 2008 Methyl crotonate 850-1400 1 0.375-1 [69]
Dayma et al. Methyl hexanoate 500-1100 10 0.5-1.5 2008 [71] 29
Dayma et al. Methyl heptanoate 550-1150 10 1-2 2009 [72]
Dayma et al. Methyl octanoate 800-1350 1 0.5-2 2011 [73]
Glaude et al. Methyl decanoate 500-1100 1.06 1 2010 [74]
Hakka et al. 2009 Methyl palmitate 500-1000 1 1 [75]
Bax et al., 2010 Methyl oleate 550-1100 1 1 [76]
Rapeseed methyl esters Dagaut et al. 800-1400 1-10 0.25-1.5 (RME) 2007 [77]
Dagaut et al. RME/kerosene mixture 740-1200 1 0.5-1.5 2007 [78]
Metcalfe et al. Ethyl propanoate 750-1100 1 0.3-1 2009 [27]
Dayma et al. Ethyl pentanoate 560-1160 10 0.6-2 2012 [79]
Commercial B30 fuel and Ramirez et al. 560-1030 6 and 10 0.25-1.5 B30 surrogate fuel 2011 [80]
Serinyel et al. n-methyl butanal 500-1200 10 0.35-4 2017 [81] 30
Rodriguez et al. n-hexanal 475-1100 1 0.25-2 2017 [82]
Thion et al. 2016 Cyclopentanone 730-1280 1, 10 0.5-2 [83]
Thion et al. 2017 Methyl ethyl ketone 700-1150 10 0.5-2 [84]
Shahla et al. 2017 Diethyl carbonate 680-1220 1 0.5-2 [85]
Curran et al. 1998 Dimethyl ether 800-1300 1, 10 0.2-2.5 [86]
Daly et al. 2001 Dimethoxymethane 800-1200 5.07 0.444-1.778 [87]
Methyl/ethyl tertiary butyl Glaude et al. 750-1150 10 0.5-2 ether 2000 [88]
3.2 Experimental Set-Up at KAUST
The JSR-type selection follows rules established by Ayass et al. [89], who
recommended that as the experiment with larger residence times (from 0.5 to 5 seconds),
the JSR should be selected with a larger diameter and crossed nozzles configuration for
ensuring better mixing; for this reason, a reactor volume of 76 cm3 is included in this model
(see Fig. 3.1). The nozzle diameter is 0.33 mm.
The setup was validated using several primary reference fuels such as n-heptane, which
has been studied using a setup and experimental conditions similar to those used in Nancy. 31
The iso-octane was also tested as representative of a high-octane-number species.
Afterward, the setup was used for conducting EL oxidation and compared with simulation results.
The JSR in this study was made from quartz glass. The reactor was connected to both upstream and downstream quartz tubes of diameter 25.4 mm. The values of residence time for all experiments in this study were set to two seconds and operated under atmospheric pressure. The temperature of the electrical furnace can reach 1500 K maximum. Figure 3.2 shows a schematic diagram of the complete JSR setup. The apparatus used includes a syringe pump injecting liquid fuel, a multi-gas controller from MKS, coupled with three mass flow controllers, four K-type thermocouples, a homemade vaporizer, a microprobe, an electrical heating furnace, three electrically insulated resistors, four thermometers, and a sampling tube with a glass finger. The mixture was analyzed using a gas chromatography (GC), coupled with a mass spectrometry detector (MSD), flame ionization detector (FID) and a thermal conductivity detector (TCD), or a GC designed as a refinery gas analyzer (RGA), one FID and two TCDs. Different columns, such as DB-624, DB-1, GAS-Pro, Molecular Sieve 5A were implemented for species separation and detection. 32
Multi-gas controller
Pressure gauge MFC O2 supply
N2 supply JSR Sampling tube
Pressure valve GC
Vaporizer Thermocouple
Fuel Temperature controller
Thermometer Syringe pump Electrical heating furnace
Figure 3.2: Scheme of the experimental set-up at KAUST. 33
3.2.1 Fuel Injection System
The fuel was stored in a 25-milliliter syringe provided by SGE Analytical Science and injected using a syringe pump from New Era Pump Systems Inc.; this syringe pump can be operated with a wide volume flow rate, ranging from μl per hour to ml per minute, with syringes of different volumes. Fuel in liquid phase was injected into a homemade vaporizer, which was heated to a specific temperature point 30 K higher than the boiling point of the fuel in use (Table 3.3). Fig. 3.3 shows the gap between the fuel injection pipe and the chamber, filled with very small stainless-steel chips for better heat transfer. The temperature was measured using a stainless-steel K-type thermocouple inserted between the chamber jacket and the heating tapes.
vaporized fuel diluted in N2
heating tape
stainless N2 as steel chips carrier gas
liquid fuel in I.D: 1.6 mm pipeline
Figure 3.3: Scheme of homemade vaporizer (not to scale). 34
Table 3.3: Physical properties of the fuel tested and measured.
Vaporizer setpoint Vapor density Fuel tested Boiling point (°C) (°C) (kg/m3) Ethyl levulinate 206-207 240 5.948 iso-octane 99.3 130 3.930 n-heptane 98.4 130 3.070
Nitrogen (N2) was used as a carrier gas, transporting the fuel to the microprobe (see
Fig. 3.4 for microprobe dimensions). At the microprobe, the fuel was mixed with O2 10 mm prior to the JSR nozzle entrance. After the desired residence time in the JSR, the unreacted fuel/end-products flowed to a detection device. All transfer lines were heated by insulated resistors coupled with thermocouples and monitored using temperature controllers to prevent condensation from the outlet of vaporizer to the microprobe
(expressed as preheating zone in Fig. 3.5). The temperature was set 20 K below the vaporizer temperature. As the mixture traveled into the furnace heating zone, it flowed through a 70 mm-long microprobe, to avoid temperature gradients, and then entered the
JSR. The residence time of the fuel in the microprobe was calculated; the speed of the gas mixture could reach high velocity over 400 cm/s at 700 K because the capillary had a 0.8 mm inner diameter which resulted in 0.0175 s residence time of the fuel traveling through the tube. Since this is a small value it can be assumed that it did not significantly affect the results. These parameters were calculated based on Eq. 3.1 and 3.2. The volume of microprobe is expressed as 푉; r is the radius; 푙 is the length at the same temperature as reactor; 푄 is the total volume flow rate of fuel/N2 mixture and 휏 is the residence time. 35
푉 = 휋푟2푙 (Eq. 3-1)
휏 = 푉/푄 (Eq. 3-2)
Oxygen was introduced outside the microprobe and diluted with N2. The two flows met 10 mm away from the nozzle entrance, defined as the gas pre-mixing zone. The time that the fuel/oxygen/nitrogen mixture remained in the pre-mixing zone was estimated by Eq. 3.1 and 3.2 and the calculation was near 0.05 s at 700 K. In total, the residence time of the fuel in the furnace zone (microprobe + pre-mixing zone) prior to the JSR was approximately
3.5 % of the residence time in the reactor. The conclusion can be made that the residence time of the fuel in both microprobe and the premixing zone with oxidizer was relatively insignificant (3.5%) in comparison with the residence time 2 s of the reactor.
Fuel and O2 were premixed at the inlet of the fused silica reactor (pre-mixing zone), which achieved good fuel/O2 premixing. Although this resulted in the mixture starting to react before entering the JSR, it was preferable to poor fuel/O2 mixing. Several high- temperature-resistance O-rings, with sizes ranging from a quarter to one inch were used for the adaptors between the glass and metal tubes.
O.D: 2 mm O.D: 6.35 mm I.D: 0.8 mm I.D: 2 mm
320 mm
Figure 3.4: Scheme of microprobe for fuel introduced to JSR pre-mixed zone (not to
scale). 36
70 mm 10 mm
Fuel/N2 O2/N2
Preheating zone Furnace heating zone
Figure 3.5: Scheme of two heating zones prior to JSR, orange region indicates preheating
zone (373-493 K); red region represents furnace heating zone (> 500 K).
3.3 Verification of Criteria for KAUST JSR
Lignola [90] designed another jet-stirred reactor in Italy and established some criteria to ensure that the gas mixture in the jet-stirred reactor reached homogeneity; the design was thoroughly re-examined by Herbinet and Dayma [49] from Nancy. Along with the guidelines, three criteria are listed below:
1. Jets from the four nozzles must be turbulent.
2. The four jets must provide good mixing of the entire gas phase in the reactor; the jets
must provide an intense internal recycle system.
3. The velocity of the jets at the outlet of the nozzles must not exceed the speed of sound,
which could interfere with the mixing process in case of shock wave formation.
For verification, several parameters were limited to ensure that the JSR achieved the criteria. Some of these equations are referred from Matras 1975 [91]. To meet the first criterion, eq. 3.3 should be satisfied. 휏 represents residence time; η is dynamic viscosity; 휌 is the specific weight of the gas; 푅 is the radius of the reactor; 푑 is the diameter of the 37 nozzle and 퐴 is a dimensional constant as a function of temperature, 0.3 in the case of air at 293.15 K and atmospheric pressure. In this work, studying nitrogen gas at temperatures of 700 and 1000 K: 휌 is 0.483 and 0.337 kg/m3; 푅 is 2.915 x 10-2 m; η is around 0.032 and
0.0415 mPa/s; 푑 is 0.3 mm and 퐴 is π/4. Ultimately the right side of eq. 3.3 is equal to 4.26 and 2.3 s, larger than the assigned residence time 2 s in this study.
휌퐴푅3 휏 ≤ (Eq. 3-3) 230η푑
To meet the second criterion, the required ratio between radius of the reactor and the internal diameter of the nozzles must satisfy equation 3.4. The calculated value is equal to 97.2, allowing for perfect mixing.
푅 ≥ 64 (Eq. 3-4) 푑
Of equal importance, the velocity u0 can be expressed as a function of the radius of the reactor 푅, the nozzle diameter 푑 and the residence time 휏, leading to the requirement, as shown in eq. 3-6. Gas flowing through the reactor is assumed to be nitrogen and 푐푠표푢푛푑 , which can be calculated from the equation 3-5, with 훾 as heat capacity ratio, R as ideal gas constant, 푇 as temperature and 푀 as molecular weight. Considering 푇 to be 700 and 1000
K, 푐푠표푢푛푑 is calculated as 539.4, 644.7 m/s. Hence the minimum residence time is 0.68 and
0.57 s.
훾푹푇 푐 = √ (Eq. 3-5) 푠표푢푛푑 푀
4 푅3 휏 ≥ 2 (Eq. 3-6) 3 푑 푐푠표푢푛푑(푇,푃)
In conclusion, experimental conditions in this case satisfied the stated criteria. The residence time (2 s in this study), is within 0.68 and 4.26 s at 700 K and between 0.57 and 38
2.3 s at 1000 K. Also, turbulent jets, the desired recycling rate, and a flow less than that of the sonic limit were all achieved.
3.4 Temperature Homogeneity Measurement
Temperature homogeneity is considered to be one of the most critical factors in reaction gas-phase kinetic study, since rate coefficients are very sensitive to temperature.
In the JSR set in this study, a total length of 200 mm, starting from the upstream microprobe to the downstream sampling probe, was placed and partially heated in the furnace.
Temperature distribution in the reactor and the tubes were measured to prove thermal homogeneity. Different locations were measured, as shown in Fig. 3.6. Set point (SP) was defined as the position of the ideal reactor temperature and N2 was set at a constant flow rate during the measurements.
-40 -30 -20 -10 SP 10 20 30
Jet-stirred reactor
Electric furnace
Figure 3.6: Jet-Stirred reactor scheme and temperature measurement at different locations. Temperature points separated by10 mm. SP represents set point where reactor
is monitored and controlled. 39
The temperature profile is presented in Fig. 3.7. Each location in the figure was measured with 1% error and repeated more than five times. The largest deviation is located at 650 K, which is 10 K lower than the SP. Downstream locations showed fewer temperature gradients than upstream. Overall, the furnace maintained good temperature homogeneity, especially under high-temperature set points.
1200
1050
900
Temperature (K) Temperature 750
600
-30 -20 -10 0 10 20 30 Position (mm)
Figure 3.7: Dashed lined are temperature homogeneity oven testing of jet-stirred reactor
at 650, 800, 950 and 1100 K. Fig. 3.6, including position 0 as set-point spot, is portrait
positioned.
3.5 Sampling Methods
Sampling is critical to the accuracy of gas-phase composition analysis. In Dagaut’s group [49], online and offline analysis were applied for different end-product collections, based on their physical properties. Another sampling technology--molecular beam sampling--is effective for unstable species, especially oxygenated compounds with relatively short lifetimes. 40
1. Online sampling: The online analysis is the most effective and useful method for
keeping the composition nearly identical to the outlet of the reactor. As species are
transported in the downstream tube connected to the JSR, a sampling sonic probe
is installed, fed into the reactor and connected to the transfer line. All tubes are
heated to prevent condensation. Through a 6- or 10-way injection valve system on
a gas chromatograph, a constant volume of sampling gas is introduced from the
loop to a column for species separation and analysis.
2. Offline sampling: Offline sampling was used to collect low vapor pressure species
in a glass vessel under very low temperature or low pressure conditions. Liquid
nitrogen is usually used as a bath matrix for cooling to the necessary low
temperatures. Condensed samples are stored in solid phase and then re-heated with
the addition of a solvent like acetone. However, the vessel can also be connected to
a differential pump. When low pressure conditions are reached in the vessel, species
with low vapor pressure can be stored in gas phase. This method was implemented
for n-dodecane pyrolysis [92].
3. Molecular beam mass spectrometry: The sampling methods mentioned above are
used mainly for stable species analysis. It has been proven that the unstable species,
such as hydroperoxide, decomposes, or reacts to other species instantly, prior to
sampling and analysis. Molecular beam sampling was first introduced by Battin-
Leclerc and Herbinet [93, 94] at the National Synchrotron Radiation Laboratory
(NSRL) in Hefei, China. (Experimental model described in Fig. 3.8.) This setup
was designed for unstable species which were analyzed using tunable synchrotron
vacuum ultraviolet (sVUV) photoionization mass spectrometry. With a quartz cone 41 and an orifice attached to a JSR, all species form a molecular beam as the pressure drops from 105 to 0.067 Pa and are then introduced at high velocity into a vacuum chamber under a pressure of 8.3x10-4 Pa. Horizontally a sVUV light source with
7.8 to 24 e.V tunable photon energy was introduced to ionize the sampled gases.
Eventually, in the photoionization chamber, species were accelerated into the mass spectrometer, due to the electric field, and separated by a mass reflectron technology. Ions were amplified with a microchannel plate (MCP) and signals were detected with a photo-multiplier tube (PMT). A recent contribution of this apparatus was the discovery of several third-O2-addition products under low- temperature condition of alkane oxidation [95]. 42
Figure 3.8: Schematic of jet-stirred reactor with molecular beam sampling system [93]. I
and II are differential pumping and ionization chambers with mass spectrometer,
respectively. 1, sVUV beamline; 2, to turbo molecular pump; 3, molecular beam from sampling gas mixture; 4, reflectron time-of-flight mass spectrometer; 5, ion trajectory; 6,
MCP; 7, quartz cone-like nozzle; 8, jet-stirred reactor.
3.5.1 Downstream Sampling Design
Although the JSR is considered to be a perfect stirred reactor for gas-phase mixtures, concentration and velocity distributions still exist in the reactor [89]. This discrepancy can be neglected once the species concentrations are averaged after sampling quantitatively and qualitatively. According to Herbinet’s research [49], online analysis with a sampling 43 probe installation can maintain the composition of stable species in the gas phase--even away from the reaction zone--with the incorporation of an additional heating zone. Since sampling is critical to the accuracy of product analysis, two methods were tested in this study--with and without sampling tubes. The species are then transported to the detection apparatus via either online analysis or syringe injection. All combinations are listed and explained below:
1. With or without sampling probe: The sampling probe was composed of relatively
fragile fused silica (Fig. 3.8); it was designed for temperature measurement and
sample collection. As noted in Chapter 3.3, the thermocouple is inserted into a fused
silica glass finger next to the sampling probe at the exit of the JSR. The sampling
tube was designed with 2 mm I.D., the opposite end was designed to fit the
connection to the other stainless steel tube, similar to the sample injection
microprobe configuration. The method without a sampling probe was tested for
comparison when the sampling probe was not available for use.
2. Pumping system:
A vacuum diaphragm pump from KAMOER® was installed at the downstream port
out of the GC system for shorter retention time of the products. The volume flow
rate through the sampling probe was improved from 10 standard cubic centimeter
per minute (sccm) to over 60 sccm, causing a sextuple decrease of residence time
in the sampling probe.
3. Online analysis and gas-tight syringe injection:
Online sampling connects the downstream probe to the analytical detectors through
heated pipelines, preventing sample condensation. However, when the online 44
analysis from reactor to detectors was unavailable, the gas-tight syringe method
was implemented, depending on the configuration of the gas chromatography
system. During species transportation in the pipeline, a GC-suitable septum was
installed and locked to maintain downstream pressure and prevent leakage. The
instrumental setup is shown in Fig. 3.9. By extracting a constant volume of the gas-
phase sample (500 μl from the downstream side, with a gas-tight syringe), the
sample was injected immediately and manually through the septum at the injector
part of the GC system. Hexane and acetone were used to keep the syringe clean and
uncontaminated.
Thermocouple
Sampling port with a septum
End-gas exhaustion
Figure 3.9: Syringe extraction scheme. The red sampling port is a septum used for the GC
injector. 45
3.6 Analytical Techniques
Two analytical apparatus were used for downstream species qualification and quantification:
1. Gas chromatography (GC), coupled with a mass spectrometry detector (MSD), flame ionization detector (FID) and thermal conductivity detector (TCD).
2. Gas chromatography (GC), with refinery gas analysis (RGA) valve system, a
FID, and two TCDs.
The samples were analyzed with gas chromatography using an Agilent 7890 (Agilent
Technologies, Santa Clara, CA.), equipped with a split/splitless injector. Injector operating conditions were as follows: The injection volume was 500 μl, inlet temperature was maintained at 280 °C to prevent condensation. Helium gas flow was used as the carrier gas. The programmable oven was set at 50 °C for 2 min, then 14 °C/min to 195 °C for
0.6 min, then 10 °C/min to 250 °C for 5.5 min. The total run time was 23.96 min. The analytical column used was a DB-624 capillary column (6% Cyanopropylphenyl
94% dimethylpolysiloxane; 60-m × 250-μm i.d. and 1.4-μm film thickness). The system worked at a constant flow. The exit end of the analytical column was connected to one of the four ports on the three-way splitter. This device divides the effluent from the analytical column between three different detectors: (1) to the interfaced mass selective detector
(Agilent 5975 MSD), (2) to the thermal conductivity detector (TCD), and (3) to the flame ionization detector (FID). Three deactivated fused silica capillary restrictors, located from the three-way splitter to each of the three detectors, were installed with different lengths and diameters: MSD restrictor (1.44-m × 0.18-mm i.d.), TCD restrictor (1.06-m × 0.18-mm i.d.), and FID restrictor (1.06 m × 0.18 mm). The split ratio applied was 2:1:1 (MSD: TCD: 46
FID). The splitter uses an auxiliary pneumatic control module (PCM) working to keep the pressure constant at 0.262 bar, thus maintaining the split ratio fixed in each run.
The MSD worked in PEI mode, operating in full-scan mode; monitoring was from
4 to 200 m/z. All the gases used in this work had a high purity of 99.9995% or better. The instrument control, data acquisition, and processing were performed using Agilent MSD
ChemStation software (E.02.00.3 version).
For product identification purposes, including hydrocarbons, permanent gases (e.g. carbon dioxide, carbon monoxide, oxygen, hydrogen, etc.), MSD was used. TCD was used for non-organic species analysis, while the FID detector performed quantitative analyses of hydrocarbons.
3.6.1 Gas Chromatography
The gas chromatography (GC) (Fig. 3.10) is considered to be one of the most useful instruments for separating and quantifying volatile organic compounds in mixtures by means of unique retention times, based on their physical properties. However, it cannot identify molecular structures unless the retention time for the species has already been detected using a standard solution/cylinder under the same temperature program. After injection (Fig. 3.11), the mixture is diluted and carried by mobile phase (helium) and introduced to a packed, or a capillary, column. Each compound acts as a sorbate; its flowing speed and separation process are determined by the interaction of its physical or chemical properties with the solid packing coated in the column. Complete separation can be achieved when the column is long enough, or has the appropriate diameter. A longer column is necessary if the peak is too broad, while a smaller diameter column provides 47 better high resolution and retention. The run time is lengthy, but this allows the peaks to be easily separated and analyzed.
3-way Splitters
Injector Detectors
GC/MS/FID/TCD system Carrier gas Column (Helium) Oven
Figure 3.10: GC instrument scheme, including carrier gas as helium, gas injector,
stationary-phase column, 3-way splitters to different detectors.
Figure 3.11: GC injector scheme [23]. 48
3.6.2 Column Selection
Generally, there are two types of column for the GC system, the packed and the capillary columns, with different functionalities. The packed column is composed of three parts: tube, packing retainers and packing materials, serving as stationary phases [96].
Packing materials are designed with enough surface space for compounds to adsorb, and
GC efficiency is highly relative to the polarity and selectivity of the mixture. For the columns employed in this study, DB-624 was used to separate esters, or some oxygenated hydrocarbons classified as having medium polarity; DB-5-MS, for separating hydrocarbons like n-heptane, has low polarity. However, capillary columns (also referred to as open tubular columns), separate very light and highly volatile compounds such as methane, ethane, carbon monoxide etc., in a relatively shorter time of analysis. The stationary phase is usually constructed using either metal or fused silica and the most common column type is the porous-layer open-tubular (PLOT), which uses metal oxide as the stationary phase (e.g. Al2O3-PLOT, KCl-PLOT). GS-GASPro is another useful capillary column used in this study to separate C1-C3 hydrocarbons. A molecular sieve can be used to separate and quantify CO, CH4, O2 and N2.
The inner diameter and length of the column are critical factors for better separation and quantification, but optimizing one of these factors requires a sacrifice from the other, depending on what the experiment requires. For example, a smaller I.D. is more efficient for separation, but it has lower sampling capacity, resulting in signal saturation. Decreasing the thickness of the packed column also decreases film thickness, improving resolution and signal-to-noise ratio (S/N ratio); the disadvantage is that it also decreases sampling capacity. 49
3.6.3 Principles of GC as a Refinery Gas Analyzer (RGA)
The GC system for quantifying O2, CO2, CO, CH4, C2H4, etc. is a specific gas chromatograph configured with three parallel channels which can analyze different species in several columns simultaneously; it is used mainly in refinery gas analysis. After the sampling injection, analytes are divided into three channels, one FID and two TCDs. Light hydrocarbons from C1 to C6 are detected on the FID channel while permanent gases are determined by a TCD. The second TCD is specifically designed for H2 detection, with only
N2 as the reference gas. In the valve system, two ten-port valves are used for looping prior to sampling, back-flushing heavier components in lines [97]. Three six-port valves are assigned different tasks before injection to each detector. PCM is the pressure control module, which provides two channels of gas control. EPC is used to control the pressure at the inlet section. The runtime event for all valves is described in Table 3.4.
V5
V4 V1
V3 V2
Figure 3.12: Plumbing diagram in GC-RGA. V1 and V5 are ten-port valves with backflush function. V2, V3 are six-port valves with functions of column isolation to TCD, 50
column sequence reversal with backflush of the pre-column. V4 is the second six-port
valve with a loop for sampling upstream between V1 and V5 [97].
3.6.3.1 Analysis Sequence (Permanent and Light Gases Channel)
In diagram 3.12, a sampling gas mixture is imported from PCM B Channel one and two, flowing through valve one, column one, column two, valve two, column three and, eventually introduced to TCD. A simplified graph is shown in Fig. 3.4. The heaviest species are trapped in column one and finally flushed away by the backflush function. O2 and H2 are light enough to be separate from the rest. The remaining light gases are trapped, either in 4Ft Unibeads 1S or Molecular Sieve 5A, depending on their physical properties.
After approximately three minutes, valve one switches off and C3+ compounds are backflushed away from column one, as shown in Fig. 3.5. In the meantime, valve two (also called the isolation valve), is switched on to introduce ethane and carbon dioxide directly to TCD. The purpose of isolation is to protect the Molecular Sieve column from carbon dioxide, since it would be trapped and deactivate the column. Carbon monoxide and methane are left in Molecular Sieve 5A. Finally, approximately five minutes after, valve two is switched off so that carbon monoxide and methane elute from the molecular sieve column to the TCD.
Valve 1 Valve 2 TCD
C2+ CO2, C2H6 CO, CH4, N2 O2, H2
2Ft Unibeads 1S 4Ft Unibeads 1S Mol Sieve 5A (Col. 3) (Col. 1) (Col. 2)
Figure 3.13: Flowsheet of light hydrocarbon gases in valve one and two prior to
backflush process. 51
CO2 CO, CH C2 4 C2H6 +
2Ft Unibeads 1S 4Ft Unibeads 1S Mol Sieve 5A (Col. 3) TCD (Col. 1) (Col. 2)
Figure 3.14: Flowsheet of valve two as isolation valve is switched on, valve one is off
with backflushing.
3.6.3.2 Analysis Sequence (Hydrocarbon Channel)
The hydrocarbon mixture separation process is simpler than the process using light and permanent gases, As valve four switches on, samples in the loop from the ten-port valve are introduced to DB-1 and HP-Al/S columns (Fig. 3.15). After a time period, C5 elutes to the HP-Al/S column where valve three is located. The purpose of valve three is to make C6+ exit early as one peak on the chromatograph so that backflush and reverse flow of DB-1 are implemented (Fig. 3.16).
FID
C1-C5 C6 +
25 m HP-Al/S 2m DB-1 (Col. 7) (Col. 6)
Figure 3.15 Flowsheet of hydrocarbons in valves three and four, before backflush 52
FID
C1-C5 C6 +
25 m HP-Al/S 2m DB-1 (Col. 7) (Col. 6)
Figure 3.16: Flowsheet of backflush and reverse flow as valve three is switched on.
Table 3.4: Runtime event for all valves in RGA.
Time Set point (min) Position On/Off 0.1 Valve 1 On 0.1 Valve 5 On 0.1 Valve 4 On 0.5 Valve 3 On 0.5 Valve 5 Off 1.0 Valve 4 Off 3.0 Valve 1 Off 4.20 Valve 2 On 5.0 Valve 2 Off 5.9 Valve 3 Off
3.6.4 Principles of GC/MS/FID/TCD
The other system used in this study was GC/MS/FID/TCD, for measuring levulinic acid, EL, ethyl vinyl ketone and other oxygenated species. With this device, sampling in lines were separate, with splitters dividing into three flows to MS, FID, and TCD from the mainstream.
3.6.4.1 Mass Spectrometry Detector
The Mass Spectrometry Detector (MSD), shown in Fig. 3.17, coupled with a GC system for species identification, is the most widely used detector. After separation from 53 the GC columns--and depending on their polarities and molecular weights--mixtures are delivered into MSD and ionized, first with either electrical ionization (EI) or chemical ionization (CI) under ultra-low pressure (1.33x10-5 Pa). However, EI is considered to be a hard ionization method that forms fragments, and the signal on the mass spectrometer of the parent compound is relatively small. All ions are introduced into the quadrupole ion- trap mass analyzer for further separation. Eventually, species are magnified with a multiplier and several mass-to-charge (m/z) ratios are presented as peaks on a spectrum.
With an already existing NIST library, each spectrum from GC/MS analysis is compared with the database to determine the exact structure, name and molecular weight. Figure 3.18 shows the EL mass spectrum from the GC/MS system; smaller numbers are fragments occurring after the ionization process.
Focal lenses
MSD
- e
- e Signal - e
GC column Ion trap analyzer
Electrical ionization
Figure 3.17: Simplified scheme of Mass Spectrometer Detector with electrical ionization
method as the ion source. 54
Figure 3.18: Mass spectrum of ethyl levulinate from NIST library.
3.6.4.2 Flame Ionization Detector and Effective Carbon Number
The flame ionization detector is designed mainly for organic compound detection.
With organic compound combustion in hydrogen flame, ions are generated and flame intensity is proportional to the concentration of organic species. Figure 3.19 illustrates a sample flow representing organic compounds passing through a hydrogen diffusion flame, diluted in the air in which the unknown species are combusted, ionized and eventually collected by an electrode [98]. The ion formation in the flames derives mostly from a chemi-ionization of CHO that produces an electron, described below:
CH + O → CHO+ + e-
Based on this theory, different chemicals with a specific number of carbons and hydrogens have different electrons that are generated after ionization; this unique response can be interpreted as the effective carbon number (ECN). This method can be applied when standard internal and external quantification methods are unavailable. However, a heteroatom (neither H nor C), or the carbon nearby in an organic compound, often offers a less than normal contribution to the response. For example, the carbon in a carbonyl group 55 makes zero contribution to the ECN and an oxygen atom in an ether has a negative contribution. A reference method, clarifying the contributions of functional groups, is listed in Table 3.5.
Charge collector
- +
Hydrogen Air Sample flow
Figure 3.19: A basic design of flame ionization detector [98].
Table 3.5: Contributions to the effective carbon number [99].
Atom Functional group ECN contribution
C Aliphatic 1
C Aromatic 1
C Olefinic 0.95
C Carbonyl 0
C Carboxyl 0
O Primary alcohol -0.5
O Ether -1.0
56
3.6.4.3 Thermal Conductivity Detector
The thermal conductivity values of non-organic species’ detection, such as O2, N2,
CO, etc. are all different, so this material property can be implemented to construct a detector for analyzing gas mixtures. When a species is introduced to a regime with temperature gradients, conduction of heat takes place from the higher side to the lower temperature points. In Eq. 3-7, with a known heat flow 푄, surface area 퐴, temperature points (푇1, 푇2) and the distance 푥, thermal conductivity 휆 can be calculated and identified.
퐴(푇 −푇 )휆 푄 = 1 2 (Eq. 3-7) 푥
In the design of TCD, several coiled filaments are mounted in the chamber to gain maximum resistance. In the design of the Wheatstone Bridge, for example [96], the column gas flow was measured, transformed to an electrical signal and compared with the reference flow. The only disadvantage of TCD is that it is less sensitive and quantitative than FID for good resolution of organic species analysis; for this reason alone N2, O2, CO, and CO2 were analyzed by TCD in this study. 57
Figure 3.20: Bridge circuit used in two-cell detector. R3 is used for sample flow detection from column. R4 measures a reference flow from nitrogen or helium [96].
58
Chapter 4 JSR VALIDATION EXPERIMENT
Because the system was not validated in the CCRC, before the oxidation study of ethyl levulinate (EL) began, the jet-stirred reactor model was tested.
Fuels with relatively well-developed mechanisms such as n-heptane, and iso-octane
(2,2,4-trimethylpentane) have been studied by Herbinet (2012 [50]), Zhang (2016 [100]),
Tsang in (2015 [101]) and Atef (2017 [102]), respectively; and they can be used to provide validation of the setup. All experimental products were analyzed either through online sampling in the gas chromatography-refinery gas analyzer (GC-RGA), or via syringe injection to GC/MS/FID/TCD.
Three sets of data were obtained for the oxidation of n-heptane. For cyclohexene, pyrolysis was studied because of its simple unimolecular decomposition reaction pathway, forming ethylene and 1.3-butadiene. All fuels were tested under two conditions with the sampling probe or the sampling probe with a pumping system. Results indicated that the probe that included a pumping system best fitted previous data and simulation results. The conclusion was that the residence time of the end-products in the sampling probe was the main cause of the data discrepancies.
4.1 N-heptane Oxidation
N-heptane was selected because it is one of the primary reference fuels (PRF) for determining research and motor octane number (RON, MON), and because it is a typical diesel fuel. It has been thoroughly tested under a wide range of conditions in several reactors and devices for proving low-temperature chemistry (in Supplemental Material), which can be detected through temperature profiles by jet-stirred reactors [48, 50, 100,
103-105] or via ignition delay times with shock tubes apparatus [106-110] . The detailed 59 mechanism was developed and compiled by Curran in 1998 [111] and Zhang [112]. In this study, n-heptane was tested using experimental parameters similar to Herbinet in Nancy
[50]. All parameters of the JSR in the literature are summarized in Table 4.1 and compared with the facility at KAUST. Only one equivalence ratio was chosen (φ = 1) and repeated under this condition.
Table 4.1: JSR experimental parameter comparison between Nancy and KAUST
JSR in Nancy JSR in KAUST Pressure (Pa) 1.074 × 105 1.013 × 105 Residence time (s) 2 2 Equivalence ratio φ 0.5, 1.0, 2.0, 4.0 1.0 Temperature range (K) 500-1000 480-1100 Fuel concentration (%) 0.5 0.5 Reactor volume (cm3) 95 76 Number of preheating zones 2 1
Carrier gas He N2 Sampling method online and offline online and syringe injection
Nancy Apparatus
In the setup at Nancy, oxidation was under constant pressure, 1.074 × 105 Pa, residence time was two seconds, temperature ranged from 500 to 1100K and equivalent ratios of φ = 0.5, 1.0, 2.0 and 4.0; fuel was diluted in helium with 0.5% fuel concentration.
In the experimental apparatus at Nancy, two pre-heating systems were installed between a vaporizer and a JSR: one system for preventing fuel condensation and the other for the pre- heating zone, which is same as the set point of the furnace temperature. The injection microprobe, upstream, and sampling probe, downstream, were installed on either side of the JSR. 60
KAUST Apparatus
To repeat the experiment, several configurations were tested in order to acquire the same results as the experiment in Nancy [113].
1. Without the sampling probe: experiments were conducted without a sampling
probe to test the significance of the probe. The injection probe was designed
with I.D. 1.6 mm, O.D. 2 mm, and a very small orifice with I.D. 0.7 mm made
of quartz. A glass finger, inserted with a thermocouple, was placed downstream
of the tube and connected to the JSR to measure temperature (Fig. 4.1). Syringe
injection was implemented to introduce the sample to the GC/MS/FID/TCD.
2. With sampling probe: a quartz sampling tube was installed next to the glass
finger (Fig. 5.1 (b)). The main reason for incorporating a sampling probe was
to reduce the residence time of end-products in the furnace, which has a similar
configuration as the injection microprobe. The volume flow rate at the outlet of
the sampling probe was measured at around 10 sccm (standard cubic
centimeters per minute). The residence time in the sampling probe was reduced
to 0.024 s, 1.2% of the residence time in the JSR.
3. With sampling probe and pumping system: a vacuum suction pump was
additionally attached at the exhaust pipeline of the GC system. The volume flow
rate of the sampling probe was improved to around 60 sccm, which exhibits
0.2% of the residence time in the JSR. 61
Pumping device
Figure 4.1: Fuel injection probe I.D. 1.6 mm with 0.7 mm exit. From top to bottom: without
sampling probe; with sampling probe; with sampling probe and pumping device.
4.1.1 Oxidation Experiment of n-heptane
The results with/without a sampling probe are presented in Fig. 4.2, with the experimental and simulation results from Zhang in 2016 [100]. The simulation was 62 performed using the perfect-stirred reactor module and transient solver program, with an end-time of 20 s, using Chemkin-Pro software [114].
In examining the fuel consumption profile, note that at the beginning good consistency is shown with experimental data from the literature [100], though all results deviated from the modeling prediction. Alternative pathway reactions during low temperature were well observed in both tests. With the installation of the sampling probe, the temperature region (650 to 800 K), fits well with experiments from Zhang [100].
However, the results from the experiment without the sampling-probe dropped to only 0.2 to 0.3% of fuel remaining after heating to 700 K or higher. To explain this phenomenon:
When fuel and O2 exit from the reactor to the downstream port without the sampling probe, the volume of the tube is 323 times larger than the volume of the sampling probe and exhibits 3.37 s residence time after leaving the reaction zone. Compared with the sampling probe, this device can be taken as a plugged flow reactor with a higher residence time.
There is an additional reactive temperature zone in which reactions continue to occur, resulting in 40% increased fuel consumption. At this point, experiments without the probe ceased, owing to poor temperature profiles, which began at 700 K.
Next, under a high-temperature regime (900 to 1000 K), the results from this apparatus (with the sampling probe and without a pumping device) showed higher fuel consumption and more O2 remaining than the results in the literature. It can be assumed that residence time in the sampling probe (1.8 s) was still too long, causing some fuel to undergo pyrolysis. An insufficient premixing zone was also suggested as the cause of excessive fuel consumption. As a result, another set of experimental data was collected, 63
combining fuel with O2 and N2 prior to the premixing zone; the results were identical to the previous apparatus (separating fuel and O2 at the entrance of microprobe).
Finally, a pumping device was introduced to achieve lower residence times in the sampling probe. The volume flow rate was increased to 60 sccm, six times higher than previously. Under this circumstance, residence time in the sampling probe was reduced to
15% of the residence time of the mixture in the JSR.
Updated data on n-heptane, O2, CO, and C2H4 appears in Fig. 4.3 and is compared with the literature [100] and the data on the probe without the pumping device. Even though both species were under-predicted by the mechanism from Zhang [100], results from this study have good trajectory, following the published results. More O2 was consumed than in the previous test at high temperature (> 900 K) and n-heptane concentration reached
0.5% between 700 and 750 K. In conclusion, decreasing residence time in the sampling probe has a significant impact on n-heptane and O2 consumption profiles versus different temperatures. 64
0.060 0.005 0.055
0.004 0.050
0.045 0.003
0.040
0.002 0.035
Mole fraction Mole fraction 0.030 Zhang_2016 0.001 Zhang_2016 without probe without probe with probe 0.025 with probe Zhang_2016 0.000 Zhang_2016 0.020 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K)
0.005 0.020 Zhang_2016 Zhang_2016 0.018 without probe without probe 0.004 with probe 0.016 with probe Zhang_2016 Zhang_2016 0.014 0.003 0.012
0.010
0.002 0.008
Mole fraction Mole fraction 0.006 0.001 0.004
0.002 0.000 0.000 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K)
Figure 4.2: Experimental results for n-heptane oxidation with 0.5% fuel concentration, 1 atm, 2 s residence time and equivalence ratio 1. Figure 4.1-a without sampling probe, 4.1-
b with sampling probe, respectively. Configurations for blue and red dots in portrait
position. From left to right, from top to bottom: n-C7H16, O2, CO2, and CO.
65
0.060
0.005 0.055
0.004 0.050
0.045 0.003 0.040
0.002 0.035 Mole fraction
Mole fraction 0.030 Zhang_2016 0.001 Zhang_2016 with probe with probe with probe & pumping 0.025 with probe & pumping Zhang_2016 Zhang_2016 0.000 0.020 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K)
0.020 0.006 Zhang_2016 Zhang_2016 0.018 with probe with probe 0.005 0.016 with probe & pumping with probe & pumping Zhang_2016 Zhang_2016 0.014 0.004 0.012
0.010 0.003
0.008
Mole fraction Mole fraction 0.002 0.006
0.004 0.001 0.002 0.000 0.000 500 600 700 800 900 1000 500 600 700 800 900 1000 Temperature (K) Temperature (K)
Figure 4.3: Comparison of results of n-heptane, O2, CO, and C2H4 (top to bottom, left to
right) with different experimental configurations.
4.2 Cyclohexene Pyrolysis
Cyclohexene pyrolysis has a very simple unimolecular decomposition pathway, the
Retro-Diel-Alder reaction, producing only ethylene and 1,3-butadiene. [101]:
C6H10 → C2H4 + C4H6
Its rate coefficient, adopted in this study, was obtained from the low-pressure flow experiment: k(T) = 1015.15exp(-33,233/T) [115]. This basic decomposition was observed in the JSR experiment in this study in order to check the influence of the pumping system.
In Fig. 4.4, it is apparent that cyclohexene was consumed faster at high-temperatures (> 66
850 K) and products were formed earlier than predicted. Since no premixing problem existed in the pyrolysis study, the only possibility for improvement was to install the pumping system. As a result, another set of experiments were conducted; the results are described in Fig. 4.5. Eventually, species’ profiles matched the simulation results well. The necessity to avoid further reactions in the sampling tube was apparent.
0.0014 C H with probe 0.0010 2 4 C H with probe 0.0012 4 6
0.0008 0.0010
0.0008 0.0006 0.0006
0.0004 Mole fraction 0.0004
Mole Fraction
0.0002 0.0002 0.0000
0.0000 C6H10 with probe -0.0002 800 900 1000 800 900 1000 Temperautre (K) Temperature (K)
Figure 4.4: Experimental results of cyclohexene pyrolysis in JSR. Only ethylene and 1,3- butadiene quantified under 1 atm, 0.1% fuel and 1 s residence time from 750 to 1025 K.
0.0014 C H with probe & pumping 0.0010 2 4 C H with probe & pumping 0.0012 4 6
0.0008 0.0010
0.0008 0.0006 0.0006
0.0004
Mole fraction 0.0004
Mole Fraction
0.0002 0.0002 0.0000
0.0000 C6H10 with probe & pumping -0.0002 800 900 1000 800 900 1000 Temperautre (K) Temperature (K)
Figure 4.5: Experimental results of cyclohexene pyrolysis with sampling probe and
pumping device configuration with identical conditions and modeling as Fig. 4.4. 67
4.3 Iso-octane Oxidation
The iso-octane oxidation mechanism was studied in the JSR in several works of literature [48, 116, 117]. In Dagaut’s article [48], iso-octane was first full-tested in a JSR under 10 atm, mean residence time 1 s, 0.1% fuel with equivalent ratios φ = 0.3, 0.5, 1.0
1.5. In this experiment, iso-octane was oxidized at temperatures from 600 to 1000 K, 1 atm,
φ = 1 and 0.5% fuel with 2 s mean residence time. The main hydrocarbon species detected in GC-RGA FID were CH4, C2H4, C3H6, i-C8H18 and O2, CO in TCD. The kinetic and thermodynamic files for simulation were provided by Atef et al. [102].
Simulation and experimental results are compared in Fig. 4.6. The fuel consumption profile was acquired from 600 to 1000 K to examine negative temperature coefficient (NTC) behavior (600 to 700 K). The set of experimental results without the pumping system did not match the model prediction, especially under high temperature conditions. O2 was over-consumed while CO was over-produced.
However, the experimental results with the pumping system agreed well with simulation results, improving the species concentration profiles; but the pumping system did not greatly improve the effect on the propylene profile and no difference was detected in the ethylene production profile under the intermediate temperature range (800 to 900 K). 68
with probe 0.0040 with probe 0.020 with probe & pumping with probe & pumping iso-octane 0.0035 CH4 CO C3H6 0.015 0.0030
0.0025
0.010 0.0020
0.0015 Mole Fraction
Mole fraction 0.005 0.0010
0.0005 0.000 0.0000
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
0.070 with probe 0.0015 0.065 with probe & pumping
C2H4 0.060 0.0010
0.055
Mole Fraction 0.050 0.0005 Mole Fraction
0.045 with probe with probe & pumping 0.0000 0.040 O2 600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
Figure 4.6: JSR species concentration profiles (points) and simulations, using updated mechanism from Atef et al. [102] for 0.5% fuel and 2 s residence time under atmospheric
pressure. 69
Chapter 5 ETHYL LEVULINATE MODELING AND EXPERIMENT
The primary goal of this study is to optimize the kinetic model of ethyl levulinate using oxidation studies data obtained from the validated JSR setup. The mechanism was developed by Dr. Stephen Dooley and the thermochemistry calculation was performed by
Dr. Manik Kumer Ghosh. In addition to utilizing GC-RGA for sample detection and quantification, some experiments were conducted via syringe injection to GC, coupled with a mass spectrometry detector (MSD), a flame ionization detector (FID) and a thermal conductivity detector (TCD) under several equivalence ratios from 0.5 to 2, N2 as carrier gas, 0.5% fuel at 1.013 × 105 Pa from 600 to 1000 K.
5.1 Ethyl levulinate Oxidation Mechanism
Based on the fuel structure, EL contains both a ketone [118, 119] and an ester functional group [69, 74] resulting in markedly different results from conventional paraffinic hydrocarbons and other oxygenates. The purpose of developing the EL kinetic model is to aid understanding the use of this potential biofuel. The EL kinetic mechanism is built on the top of a gasoline surrogate mechanism [120], composed of C0-C4; primary reference fuels: n-heptane and iso-octane; 1-hexene as olefin and toluene as an aromatic compound. The chemistry of several species such as HCOOC2H5, C2H3COOH, and
C2H3COCH3 were also included in the model.
5.1.1 Kinetic Model
The elementary reaction pathways, their rate constant expressions and the thermochemical properties of the radicals and molecules pertinent to EL combustion, have been prescribed, considering the best knowledge available for each discrete reaction center, as studied in other simpler systems. In general, for the reaction centers comprising the 70 methyl ketone functionality, reaction parameters have been assigned through analogy to those of methyl ethyl ketone. Reaction centers comprising the ethyl ester functionality, analogies to those of the well-studied ethyl propanoate (EP) have been made, principally considering the modelling works of Serinyel et al. [118] and Metcalfe and co-workers [27], respectively. Bond dissociation energy (BDE) of EP is one of the most critical factors for kinetic and thermodynamic parameters calculation. Depicted in Fig. 5.1 (a), the carbon- hydrogen BDE is generally weaker at the α-carbon position in the (C*O)-alkyl moiety of the molecule. Moreover, as C-H bond breaks, the whole molecule is stabilized by the carbonyl group resonance, resulting in a decreased reaction rate with other radical species, delineated in Fig. 5.1 (b).
97.3 85.8 101.1 100.2 89.4 91.8 102.1 α 86.6 94.2
Figure 5.1: (a) Bond dissociation energy of ethyl propanoate (EP) (kJ/mol). Black numbers are C-C or C-O BDE, red numbers are C-H BDE [29]. α-carbon is designated to
the carbon next to carbonyl group (C(=O)). (b) Radical stabilization by the carbonyl
group.
As a radical site at α-carbon forms, the radical is stabilized due to the carbonyl group resonance, slowing down the relative reaction rates. 71
The appropriateness of the analogies imposed above is supported by a similarity in the thermochemical environment local to each reaction center, as depicted in Fig. 5.2. The contributions of the ketone group are adopted from those reported by Hudzik et al. [121] and the ethyl ester contributions are consistent with the finding of El-Nahas et al. [122] for the case of EP. These calculations result in a set of bond dissociation energies for EL, as shown in Fig. 5.2.
(a) 93.6 102.8 82.2 95.2 88.6 84.0 79.9 99.0 90.8 97.0 90.9 97.5
(b) 97.3 101.1 100.2 89.4 85.8 91.8 86.6 102.1 94.2
(c)
101.5 95.6 90.8
Figure 5.2: Chemical structure and bond dissociation energy (BDE) of (a) ethyl levulinate
calculated by Dooley et al. [123], (b) ethyl propanoate (EP) [122] and (c) methyl ethyl
ketone (MEK) [121]. Red numbers represent C-H BDE, black numbers are C-C or C-O
BDE. (Unit: kcal/mol)
To build a complete kinetic mechanism for EL, Dooley suggested that several sub- mechanisms should be served and assembled. The basic chemistry is given by a C0 to C4 mechanism. Moreover, from C5 to C8, hydrocarbon chemistry is provided from the primary reference fuel (linear and branched saturated hydrocarbons), olefins (1-hexene) 72 and aromatics (toluene) from surrogate gasoline mechanism [120]. Furthermore EL and intermediate reactions are included, adopted from various sub-models. Temperature dependent reaction rates are expressed in the model using the modified Arrhenius equation: k(T) = A.Tn.exp(–Ea/RT), where A is the pre-exponential collision frequency factor, T is the temperature, n is the fitting parameter, Ea (cal mol-1) is the activation energy, and R represents the universal gas constant. In most temperature ranges encountered in combustion, temperature dependence can adequately be integrated into A factor (with n =
0). However, additional variation is considered for some over-range reactions in which an appropriate value of n accurately fits the available experimental data [124]. Reaction pathways have been prescribed, generally according to the protocols of Curran et al. [125].
The high-temperature reaction classes considered in the development of this model are listed below.
1. EL unimolecular decomposition. Several decomposition pathways should be
involved to produce smaller radicals. In the mechanism, radical decomposition rate
constants are estimated in the reverse exothermic direction (P+Q => R), allowing
the forward rate constant (R=> P+Q) to be calculated from the equilibrium constant
in the thermodynamic file, and most reverse rate coefficients are assigned to 3×1013
(cm3 mol-1 s-1).
Ethers, esters, and alcohols preferably undergo molecular elimination reactions
[126-128] during oxidation or pyrolysis. Metcalfe et al. [27] presented the high-
pressure limit rate for the EP molecular elimination, mentioned in Chapter 2.3.2:
EP C2H5COOH + C2H4 (R1) 73
with a rate expressions k(T) = 1.6×1013exp(-50,000/RT) s-1. Westbrook et al. [128]
found similar rate expressions with only A factor differences of 2.0×1013 and
1.0×1013 for ethyl acetate and ethyl formate, respectively. In EL, the ethyl ester is
also concluded to undergo elimination, forming levulinic acid (LA) and ethylene
(R2), as the dominant reaction. In this study, the EP rate coefficient was chosen for
simulation in the kinetic mechanism.
2. Hydrogen abstractions from fuel. This class includes all rate coefficients for
abstraction of H atoms from the fuel by small radicals such as H, OH, CH3, CH3O,
CH3O2, HO2, O and O2. Dooley adopted most of the rate coefficients in this class
from EP and MEK, with minor modification to the A factor or activation energy Ea.
3. Fuel radical decompositions. After H abstraction, five fuel radicals proceed, by
further decomposition, to produce smaller species, usually olefins and other
radicals at high temperatures. Along with the translation of this model, most rate
coefficients were developed using both the EP and MEK models. (Note that C-O
beta-scission reactions are also taken into account.)
4. EL radical isomerizations. All rate coefficients (relative to isomerization of the
five fuel radicals) are again assigned as identical, as 1×1013. Actual rates are
determined by thermochemistry of fuel radicals.
5. Ethyl formate formation and consumption. By breaking the C-C bond between
the ketone and carbonyl group, ethyl formate (C2H4OC(O)H) becomes a potential
intermediate and should be considered; its sub-mechanism was developed from
Westbrook et al.[128]. 74
6. Levulinic acid (LA) relative reactions. As one of the major products from EL
decomposition reactions, some LA relative reactions, such as hydrogen abstraction
reactions, are taken from methyl cyclohexane sub-mechanism [129] with minor
adjustments. Beta-scission reactions are using analogy to either EP or MEK sub-
mechanism.
7. Ethyl propanoate and propenoate submodel. As EL decomposes to smaller
radicals, or stable species, some structures are similar to that of EP, or unsaturated
compounds like ethyl propenoate. The decomposition reactions relating to these
specific compounds are taken from analogous reactions in the EP sub-model [27].
8. C2H3COCH3 consumption. Since methyl vinyl ketone has been predicted and
verified as one of the most abundant products from EL oxidation, its further
decomposition reactions (including hydrogen abstractions by small radicals or
unimolecular decomposition), are also considered. Rates are taken from the three-
pentanone oxidation model [130].
5.1.2 Ethyl Levulinate Thermochemistry
The thermodynamic parameters were taken mainly from the surrogate fuel [120].
Regarding the thermochemistry of ethyl levulinate (EL, CH3C(O)CH2CH2C(O)OCH2CH3), levulinic acid (LA, CH3C(O)CH2CH2C(O)OH) and five primary fuel radicals, Dr. Manik
Kumer Ghosh performed the calculations for EL, LA and other relative radicals in the
GAUSSIAN 09 package [131]. The geometries were optimized with B3LYP/6-31G(d,p) theory. The thermochemistry of all species produced by the homolytic cleavage of EL and their consumption products, not described in the EL sub-model, were estimated by group additivity, employing THERM software [132]. This method has been used for over 20 75 years; it is much faster than the higher-level ab initio method and it provides a good estimation.
Overall simulating files were accomplished, including the kinetic mechanism and thermochemistry; modeling computations in this study were carried out using the perfect- stirred reactor code in CHEMKIN PRO [114].
5.2 Ethyl Levulinate Oxidation Experiment
Before systematically quantifying the EL oxidation experiment at different equivalence ratios, several intermediates were detected using GC/MS/FID/TCD; the spectrum appears in Fig 3.18. Some unknown species were identified by MS detector, showing over 85% mass spectra similarity with the NIST library: ethanol C2H5OH, methyl vinyl ketone C4H6O, ethyl acrylate C5H8O2, angelica lactone C5H6O2, ethyl levulinate
C7H12O3 and levulinic acid C5H8O3. Thion [44] performed methyl levulinate reaction pathway theoretical calculations and predicted the production of methyl vinyl ketone and methyl acrylate. Interestingly, methyl vinyl ketone and ethyl acrylate were also formed from the decomposition of EL. These two compounds, and levulinic acid, were the highest peaks recorded by GC/MS analysis after EL oxidation (Fig. 5.3), excepting ethyl levulinate
(peak F) and nitrogen (peak A). 76
GC-MS @ 850 K 4 F
3
)
7
2 G
Int. (x10 C
1 A B D E 0
6 8 10 12 14 16 18 20 22 24 Time (min)
Figure 5.3: GC-MS spectrum of EL oxidation under 850K in JSR unknown species
identified with MS detector. From A to G: A: nitrogen; B: ethanol; C: methyl vinyl
ketone; D: ethyl acrylate, E: angelica lactone, F: ethyl levulinate and G: levulinic acid.
The JSR was next used to validate the mechanism under 1 bar, 600 to 1000 K, 0.5% fuel, 2 s residence time and equivalence ratios from 0.5 to 2. The boiling point of EL is around 206 oC and the vaporizer was heated to 240 oC. Both upstream and downstream transport lines were heated to 230 oC. The experimental results of the EL profile were measured with GC/MS/FID/TCD with a sampling probe and syringe injection method (Fig.
5.4); it has not as yet been detected by GC-RGA, so online analysis is not available. The remainder of the species: ethylene C2H4, methane CH4, hydrogen H2, oxygen O2, carbon monoxide CO and carbon dioxide CO2 were all analyzed and quantified by online analysis.
The pumping system has not yet been implemented with this fuel. 77
EL exhibits a similar consumption profile to ethyl propanoate as the concentration drops significantly at 775 K and is fully consumed after 900 K; but the prediction of ethylene is underestimated and more ethylene was detected from 700 to 850 K. Hydrogen, methane and carbon monoxide consumption profiles are acceptable for φ = 0.5 and 1.
However, oxygen is always under-predicted and the estimation is lower with increasing equivalence ratios. For CO and CO2, experimental signals are nearly twice as great after
900 K, when φ is equal to 2. Similar tendencies was also observed in iso-octane experiments, as underestimation of CO, CO2 occurred when the experiment was conducted without a pumping system. In summary, some rate coefficients in the original kinetic mechanism should be altered, especially those relating to ethylene production.
78
0.006 EL C H 0.08 0.005 2 4 CH4 H 0.004 2 0.06
O 0.003 2
0.04 CO2
0.002 CO
Mole fraction Mole fraction
0.02 0.001
0.000 0.00
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
0.006 EL C H 0.04 0.005 2 4 CH4 H 0.004 2 0.03
O2 0.003 CO 0.02 2 CO
0.002
Mole fraction Mole fraction 0.01 0.001
0.000 0.00
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
0.005 0.020
0.004 0.015 EL 0.003 C2H4 O2 0.010 CH4 CO2 0.002 H2 CO
Mole fraction Mole fraction 0.005 0.001
0.000 0.000
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
Figure 5.4: Temperature profiles of ethyl levulinate (EL) oxidation, with 0.5% fuel, 1 bar,
2 s residence time and equivalence ratio from top to bottom: φ= 0.5, 1, 2.
79
5.3 Kinetic Model Adjustment
Comparing the model validation against experimental results showed some disagreements, therefore further adjustments and improvements to the chemical kinetic model must be made, especially in fuel unimolecular decomposition, H-abstraction and fuel radical derived carbon-carbon (C-C) and carbon-oxygen (C-O) scissions reactions.
The kinetic mechanism was updated by applying new rate coefficients from another member of the levulinic family, methyl levulinate (ML, CH3C(*O)CH2CH2C(O*)OCH3).
Thion et al. [44] calculated the theoretical enthalpy of formation of different species and radicals derived from ML. Thereafter, rate coefficient computation was performed for two classes applied in this model: H atom-abstraction from the ML by OH and CH3 radicals and fuel radical derived C-C and C-O scissions. In addition, rate coefficients for fuel unimolecular decomposition were replaced with analogous rates from methyl butanoate
(MB) [28], ethyl propanoate (EP) [27] and methyl ethyl ketone (MEK) [133] mechanisms.
5.3.1 H-abstraction from The Fuel
In the fuel radicals shown in Fig. 5.5 (EL1J, EL3J, EL4J, EL6J, EL7J) only three rate coefficients were adopted from ML calculation [44]. The rate coefficients are described in Table 5.1, with their source. It was originally assumed that EL → EL1J has the same rate coefficient as EL → EL6J and EL → EL3J has the same rate coefficient as
EL → EL4J. Both sites (C1 and C6) have similar C-H bond dissociation energy (BDE), leading to the identical rate coefficient of EL1J and EL6J. However, C3 and C4 sites were assigned the same rate coefficients, even though C4 has 2.7 kcal/mol higher BDE than C3.
It can also be observed in the EP oxidation model [27] that C2-H is 3.1 kcal/mol BDE lower than CE-H, but is assigned the same rate coefficient. The BDE order (smallest to 80 largest) of EL is C3-H < C4-H < C1-H < C6-H < C7-H. To refine the model, EL + R ↔
EL1J* + RH rate coefficient was replaced with that of ML + R ↔ MLR 6 + RH, EL + R
↔ EL3J* + RH with ML + R ↔ MLR4 + RH and EL + R ↔ EL4J* + RH with ML + R
↔ MLR3 +RH. By plotting the rate coefficients against temperature (Fig. 5.7) the updated rate coefficient order from highest to lowest is: EL3J* > EL6J > EL4J* > EL1J* > EL7J, compared with the previous order: EL3J = EL4J > EL6J = EL1J > EL7J. By following the updates it can be seen that EL6J forms more easily than EL4J and EL1J. However this finding is not consistent with the BDE order, as noted above; therefore, the only change made was to increase the rate coefficient of EL3J, to make the order fit with C-H BDE, which gives: EL3J* > EL4J > EL1J = EL6J > EL7J. In conclusion, only EL + R ↔ EL3J*
+ RH, using ML calculations, was altered in the fuel H-abstraction reaction class in the model. 81
EL EP 1
3 4 7 1 E 5 1 2 3 3 6 2 M
2
EL1J EL3J EL4J
EL6J EL7J
ML6R ML4R ML3R
Figure 5.5: Radicals or fuels nomenclature of ethyl levulinate, ethyl propanoate, and
methyl levulinate.
Table 5.1: Ethyl levulinate and ethyl propanoate mechanism rate coefficients for H
abstraction by OH radical; cm3 mol-1 s-1 cal-1 units.
Reaction A (cm3 mol-1 s-1) n Ea (cal mol-1) Comments
EL + OH → EL1J + H2O 2.34E+07 1.61 -35 same as EL6J 82
EL + OH → EL1J* + H2O 4.06E+03 2.84 65 analogy to ML6R [44]
EL + OH → EL3J + H2O 5.73E+10 0.51 63 same as EL4J
EL + OH → EL3J* + H2O 3.25E+03 2.89 -1,500 analogy to ML4R [44]
EL + OH → EL4J + H2O 5.73E+10 0.51 63 EP model [27]
EL + OH → EL4J* + H2O 2.87E+02 3.06 -2,320 analogy to ML3R [44]
EL + OH → EL6J + H2O 2.34E+07 1.61 -35 EP model [27]
EL + OH → EL7J + H2O 1.76E+09 0.97 1,586 EP model [27]
EP + OH → EP2J + H2O 1.15E+11 0.51 63 BDEC2-H = 94.2 [122]
EP + OH → EPEJ + H2O 1.15E+11 0.51 63 BDECE-H = 97.3 [122]
* parameters from methyl levulinate theoretical kinetic study.
1013 EL1J, EL6J EL3J, EL4J EL7J EL1J* EL3J*
-1 EL4J*
s
-1
12
mol 10
3
k/cm
EL+OH• = ELRJ+H2O 1011 0.8 1.0 1.2 1.4 1.6 1000 K/T
Figure 5.6: Rate coefficient comparison of H-abstraction from fuel. Solid line: original
rates adopted from other molecules; dashed line: rates from the methyl levulinate
theoretical kinetic study [44]. 83
5.3.2 Fuel Radical Derived C-C and C-O Scissions
EL1J, EL3J, EL4J, EL6J and EL7J, are the fuel radicals which are the main
precursors of several intermediates, such as methyl vinyl ketone and ethyl acrylate, via C-
C or C-O beta-scission. The original rate coefficients are analogous to EP [27] and MEK
[133] models. To update the mechanism--particularly for EL1J, EL3J, EL4J radicals--their
C-C beta scission reactions were replaced with those calculated for ML, giving the detailed
rate coefficients delineated in Table 5.2. Comparing the rates of all decomposition
pathways shown in Fig. 5.7, coefficients were reduced in all cases. These rate coefficients
from ML calculations are all included in the new mechanism.
Table 5.2: Rate coefficients for fuel radical derived C-C and C-O scissions, only EL1J,
EL3J and EL4J were altered.
Reaction A (cm3 mol-1 s-1) n Ea (cal mol-1) Sources
EL1J → CH2CO + EP3J 1.00E+14 0.0 3.10E+04 Acetone
EL1J* → CH2CO +EP3J 3.50E+11 0.58 3.52E+04 Analogy to MLR6
EL3J → CH3 + COCHCH2CO2CH2CH3 1.00E+14 0.0 3.10E+04 Acetone
EL3J → C2H3COCH3 + EFF 8.83E+15 -0.67 3.56E+04 EP [27]
EL3J* → CH3 + COCHCH2CO2CH2CH3 8.15E+12 0.3 4.15E+04 Analogy to MLR4
EL3J* → C2H3COCH3 + EFF 4.29E+12 0.3 3.54E+04 Analogy to MLR4
EL4J → CH3CO + EP1D 1.00E+11 0.0 8.42E+03 3-pentanone
EL4J → CH3COCH2CHCO +C2H5O 6.82E+16 -1.34 2.69E+04 3-pentanone
EL4J* → CH3CO + EP1D 6.78E+12 0.22 2.38E+04 Analogy to MLR3
EL4J* → CH3COCH2CHCO +C2H5O 8.45E+11 0.51 4.89E+04 Analogy to MLR3
84
108
106
104
-1
k/ s 102
100
10-2 EL1J C2-C3 scission EL1J* C2-C3 scission
0.8 1.0 1.2 1.4 1.6 1000 K/T
1010
108
106
104
-1
2
k/ s 10
100
10-2 EL3J C1-C2 scission EL3J* C1-C2 scission 10-4 EL3J C4-C5 scission EL3J* C4-C5 scission 10-6 0.8 1.0 1.2 1.4 1.6 1000 K/T
1010
108
106
104
-1
2
k/ s 10
100
10-2 EL4J C2-O3 scission EL4J* C2-C3 scission 10-4 EL4J C5-O3 scission EL4J* C5-O3 scission 0.8 1.0 1.2 1.4 1.6 1000 K/T 85
Figure 5.7: Rate coefficient comparison of EL3J decomposing to smaller species via
beta-scission at different C-C bond positions.
5.3.3 Fuel Unimolecular Decomposition
Concerning fuel unimolecular decomposition reaction class (via C-C and C-O), original rate coefficients were estimated for radical re-combination; subsequently decomposition reactions can be determined from the equilibrium constant from the thermochemistry. These rates can be improved further by comparison to three other oxidation kinetic models: MEK [133] EP [27] and MB [28]. First, the C-C BDE order, from lowest to highest, is C3-C4 < C2-C3 < C1-C2 < O3-C6 < C6-C7 < C4-C5 < C5-O3; the updated model is principally constructed according to this order. For C1-C2 and C2-
C3 bond breaking, rate coefficients were adopted from the MEK model [133]. The C3-C4 bond breaking rate was taken from the methyl butanoate model, and O3-C6, C6-C7, C4-
C5, C5-O3 unimolecular decomposition rate coefficients were from the EP model [27].
These rate coefficients are compared and shown in portrait mode in Fig. 5.8. It is interesting that the rate coefficient for C2-C3 is much higher, giving the order: C2-C3 >> C3-C4 >
C1-C2 > O3-C6 > C4-C5 > C6-C7 > C5-O3. Another unexpected finding is that the order of C4-C5 and C6-C7 is contradictory to the C-C BDE order. This unusual occurrence can also be observed in the EP oxidation model [27]; therefore these rate coefficients were not adjusted. To solve the abnormal C2-C3 rate coefficient, they were given the same rate coefficient as C1-C2 since their C-C BDE are similar (82.2 and 84.0 kcal/mol, respectively), with the same treatment as the rate coefficients of EL1J and EL6J formation.
Table 5.3: Rate coefficients for EL unimolecular decomposition reactions. 86
Reaction A (cm3 mol-1 s-1) n Ea (cal mol-1) Bond position
EL → ELCOJ + CH3 4.04E+90 -21.9 1.22E+05 C1-C2, MEK [133]
EL → CH3CO + EP3J 1.78E+19 -0.73 8.19E+04 C2-C3, MEK [133]
EL → CH3COCH2 + CH2CO2CH2CH3 3.80E+16 0 8.17E+04 C3-C4, MB [28]
EL → CH2CH2COCH3 + C2H5OCO 2.63E+27 -3.23 9.47E+04 C4-C5, EP [27]
EL → CH3COCH2CH2COJ + C2H5O 1.65E+24 -2.04 1.00E+05 C5-O2, EP [27]
EL → CH3COCH2CH2CO2J + C2H5 5.73E+25 -2.76 9.21E+04 O2-C6, EP [27]
EL → CH3 + ELME2 3.39E+21 -1.58 9.21E+04 C6-C7, EP [27]
108 105 102 10-1
-4
-1 10
-7
k/ s 10 C1-C2 C2-C3 10-10 C3-C4 10-13 C4-C5 10-16 C5-O3 O3-C6 10-19 C6-C7 10-22 0.8 1.0 1.2 1.4 1.6 1000 K/T
Figure 5.8: Rate coefficient comparison of EL unimolecular decomposition, including C-
C and C-O bond scission. 87
5.3.4 EL Concerted Elimination Reaction
Note that the rate coefficient of EL = levulinic acid + C2H4 was originally adopted from ethyl propanoate [27]: k(T) = 1.6×1013exp(-50,000/RT). Another rate coefficient was measured by AlAbbad [45] and calculated from a series of shock tube experiments.
Ethylene concentration was monitored using CO2 gas laser absorption technology. Under
1 bar condition, the rate coefficient is expressed as: k(T) = 3×1014exp(-54,574/RT). The fitting was adjusted and used in the model. The rate coefficients from [27] and [45] are compared in Fig. 5.9 in the range of 600 K to 1000 K. Two rate-coefficient expressions are similar, with a cross point around 750 K. Even under lowest and highest temperatures, the rate difference is no greater than an order of magnitude; yet this minor discrepancy may have a significant effect on the prediction of ethylene production. 88
102
101
) -1 100
10-1
10-2
10-3
Rate Coefficient (s Coefficient Rate 10-4
-5 10 13 Metcalfe et al. (1.6x10 exp (50,000/RT)) -41.79 -8.20 10-6 AlAbbad et al. (10 xT exp (33,777/T)) 600 700 800 900 1000 Temperature (K)
Figure 5.9: Rate coefficient comparison in two different studies. Black line is EP oxidation mechanism [27], used in this study. Red line is shock tube experimental results
[45]
5.4 Simulation Results from Updated Kinetic Mechanism
Performance of the original and updated versions are depicted in Fig. 5.10; the disparity is in the ethylene production profile. The updated version shows that the ethylene fits well using the updated rate coefficients, although the EL is consumed earlier. The fuel consumption rate was enhanced and the fuel concentration was over-predicted at stoichiometric and rich conditions. The prediction for other species was similar to the original simulation results.
The reaction pathways of ethyl levulinate were performed and described in Fig.
5.11. The analysis is taken under stoichiometric conditions, atmospheric pressure, with temperature at 800 and 1000 K. At 800 K the fuel is consumed by 20%, the temperature 89 chosen for analysis. EL is mostly consumed by a concerted elimination reaction and over
80% of the fuel decomposed to LA and ethylene. Approximately 10% of fuel is consumed by H abstraction reactions, mostly by OH and H radicals attacking C3 and C4 sites; this is reasonable as C3 and C4 sites have the weakest C-H BDE (Fig. 5.2). The subsequent pathways of LA undergo H abstraction, followed by the production of CH3CO and propenoic acid. However, EL radicals proceed via beta-scissions to smaller, stable and radical species, such as methyl vinyl ketone and ethyl acrylate, observed in GC-MS in Fig.
5.3. At the higher temperature of 1000 K, fuel is almost completely consumed and undergoes further H abstraction. With EL3J there is enough energy to overcome the activation energy barrier of C-C bond breakage, reducing the importance of isomerization from EL3J to EL4J. From the analysis it is evident that EL unimolecular decomposition dominates the consumption of EL at all temperature ranges.
The sensitivity analysis coefficient of 0.5 % fuel under two temperature conditions
(800 and 1000 K, φ = 1) is shown in Fig. 5.12. The sensitivity coefficients for different reactions represent the percentage change of the species concentrations, relative to the baseline concentration, in terms of the dependence of the reaction parameters. The positive coefficient denotes a rate coefficient that promotes fuel formation while the negative coefficient indicates fuel consumption. As expected, fuel is very sensitive to the concerted elimination reaction (EL = C5H8O3 + C2H4) under both temperatures, and causes 80% fuel consumption at all temperature, as seen in the reaction pathway analysis. Also, chain termination reactions (2CH3 = C2H6, CH3 + HO2 = CH4 + O2 and hydroperoxyl radical formation) usually inhibit the consumption of EL, decreasing the overall reactivity; however, chain branching reactions to produce a hydroxyl radical (H2O2 = 2OH and H + 90
O2 = O + OH) exhibit a positive effect on fuel consumption, promoting overall reactivity under both 800 and 1000 K, respectively. At 800 K the occurrence of H-abstraction reactions of EL, such as EL + HO2 and EL + CH3O2, forming fuel radicals EL3J, EL4J, predominate fuel consumption as well. When the temperature rises to 1000 K, the rate coefficients of the basic chemistry of small radicals become more sensitive to fuel consumption. (Note that the importance of ethylene chemistry: C2H4 + OH = C2H3 + H2O and C2H4 + O = CH2CHO + H is also sensitive when the temperature becomes higher.)
Both the sensitivity analysis and reaction pathways analysis reinforce and prove that the accurate rate coefficient of a concerted elimination reaction is proven to be much more important than other reactions. The basic chemistry of ethylene may influence overall reactivity under high-temperatures, so relative rate coefficients should be treated carefully; but levulinic acid chemistry should be re-examined since it is not sensitive to fuel consumption in the current mechanism. 91
0.006
EL 0.08 0.005 C2H4
CH4
0.004 H2 0.06
O 0.003 2
0.04 CO2
0.002 CO
Mole fraction Mole fraction 0.02 0.001
0.000 0.00
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
0.006 EL C H 0.04 0.005 2 4 CH4 H 0.004 2 0.03
O2 0.003 CO 0.02 2 CO
0.002 Mole fraction Mole fraction 0.01 0.001
0.000 0.00
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
0.005 0.020
0.004 0.015 EL 0.003 C H 2 4 CO2 0.010 CH4 O2
0.002 H2 CO Mole fraction Mole fraction 0.005 0.001
0.000 0.000
600 700 800 900 1000 600 700 800 900 1000 Temperature (K) Temperature (K)
Figure 5.10: Updated mechanism of EL oxidation with experimental data identical to Fig
5.4. Solid lines represent updated model. Dashed lines are from previous simulation.
92
+ +
45% 100% 13% 25% 100% 53% . . 6% 6% 5.1% 5.1% + R• + R•
80% 84%
+ 10% 8.9% 50% 28% 52% + R• 22.4%
.
100% 85% . 100% 81% 100% 100%
+
Figure 5.11: Reaction path analysis for EL/O2 mixture at 1atm, 800 K (red) and 1000 K
(black), φ = 1. Percentages represent decomposition percent from parent to daughter
species. 93
Ethyl levulinate, φ = 1, P = 1 bar, T = 800 K
2CH3(+M)<=>C2H6(+M) CH3+HO2<=>CH4+O2 H2O2+O2<=>2HO2 CH2O+HO2<=>HCO+H2O2 CH3O2+CH3<=>2CH3O CH3O2+CH2O<=>CH3O2H+HCO EL+HO2<=>EL4J+H2O2 EL+HO2<=>EL3J+H2O2 EL+CH3O2<=>EL4J+CH3O2H EL+CH3O2<=>EL3J+CH3O2H CH3+HO2<=>CH3O+OH H2O2(+M)<=>2OH(+M) EL<=>C5H8O3+C2H4 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 Sensitivity Coefficient
Ethyl levulinate, φ = 1, P = 1 bar, T 1000 K H+O2(+M)<=>HO2(+M) CH3+HO2<=>CH4+O2 HO2+OH<=>H2O+O2 CH3+OH<=>CH2(S)+H2O C2H3+O2<=>CH2CHO+O H2O2(+M)<=>2OH(+M) C2H4+O<=>CH2CHO+H C2H4+OH<=>C2H3+H2O HCO+M<=>H+CO+M CO+OH<=>CO2+H CH3+HO2<=>CH3O+OH H+O2<=>O+OH EL<=>C5H8O3+C2H4 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 Sensitivity Coefficient
Figure 5.12: Sensitivity coefficients for 0.5 % ethyl levulinate in N2 under (a) 800 and (b)
1000 K at 1 bar, φ = 1.
94
Chapter 6 CONCLUSION AND FUTURE WORK
The goal of this work was to achieve the following objectives: To build and validate a new jet-stirred reactor apparatus to study ethyl levulinate (EL) oxidation; iso-octane, cyclohexene, and n-heptane were used as reference fuels for testing and validation.
Experimental results were also used to validate a chemical kinetic model for a better understanding of fuel chemistry. Some small intermediates were observed and quantified by gas chromatography, combined with a thermal conductivity detector, a mass spectrometer detector and a flame ionization detector. Several species were detected and quantified: ethylene (C2H4), carbon monoxide (CO), carbon dioxide (CO2), propylene
(C3H6), methane (CH4), hydrogen (H2), oxygen (O2) and 1,3-butadiene (C4H6). Some oxygenated species were detected, but are not yet quantified: methyl vinyl ketone (C4H6O), levulinic acid (C5H8O3), ethyl acrylate (C5H8O2), angelica lactone (C5H6O2) and ethanol
(C2H5OH).
The JSR setup was adjusted several times, based on experimental data and compared to results from Zhang et al. [100]. Different sampling methods were incorporated: first without a sampling probe, with the addition of a sampling tube, and finally, adding a sample probe with an installed pumping system. It is indicative of every testing result that the sampling probe with the pumping system were effectively used to avoid further reactions occurring in the sampling probe. Cyclohexene pyrolysis validated the necessity of a pumping system at downstream.
Preliminary simulation of EL oxidation was carried out using the original kinetic model provided by Dr. Dooley, with the concerted decomposition reaction rate coefficient of ethyl propanoate (1.6×1013exp(-50,000/RT)) [27]. An EL oxidation experiment was 95 conducted next; with the validated apparatus the fuel consumption profile was a good match with the simulation results. Primary quantified intermediates were ethylene, CO, and CO2. Other possible oxygenated products include levulinic acid, ethyl acrylate and methyl vinyl ketone, which were observed via GC/MS/TCD/FID. However, ethylene did not fit well with the simulation results under the intermediate temperature regime. To correct the discrepancies between model prediction and experimental results, various rate coefficients were replaced and updated by applying analogies to different compounds, including another concerted reaction rate coefficient from the ethylene time history measurement (3×1014 exp(54,574/RT)) [45]. The experimental data achieved good agreement with the adjusted model, as ethylene production is twice as fast as the original model from
750 to 850 K. The concerted elimination reaction was proven to be the key reaction, over all others, by analyzing reaction pathways and sensitivity.
Future work,
1. The primary objective is quantifying oxygenated species, which can be
accomplished by introducing other GC columns, such as DB-1 or DB-5, and by
changing temperature profiles and oven temperatures. A methanizer for
measuring formaldehyde CH2O is desirable.
2. It is recommended that some rate coefficients in the kinetic model be replaced,
since a few are still assumed to be 1.0×1013. Also, some reactions relative to
primary species, such as ethyl acrylate, should be considered, using an analogy
to ethyl propanoate.
3. EL oxidation under atmospheric pressure was conducted in this study; it is
recommended that EL be tested in other devices at different pressure ranges, 96
such as a shock tube or a rapid compression machine (RCM). Design and
construction of a high-pressure JSR device is required, by doing so, the
mechanism can be testified again and the pressure dependence for some critical
rate coefficients understood.
4. Low-temperature chemistry of EL is currently excluded in the model; a group
from Trinity College Dublin will work on the calculation of rate coefficients for
O2 addition.
97
REFERENCES
1. P. Friedlingstein, R.M. Andrew, J. Rogelj, G. P. Peters, J.G. Canadell, R. Knutti, G. Luderer, M.R. Raupach, M. Schaeffer, D.P. van Vuuren, C. Le Quéré, Persistent growth of CO2 emissions and implications for reaching climate targets. Nat. Geosci., 2014. 7: p. 709-715.
2. A. Demirbas, Importance of biodiesel as transportation fuel. Energy Policy, 2007. 35(9): p. 4661-4670.
3. J.J. Bozell, Chemistry connecting biomass and petroleum processing with a chemical bridge. Science, 2010. 329(5991): p. 522-523.
4. E. Christensen, A.Williams, S. Paul, S. Burton, R. L. McCormick, Properties and Performance of Levulinate Esters as Diesel Blend Components. Energy Fuels, 2011. 25(11): p. 5422-5428.
5. D. Matras, J.Villermaux, Un réacteur continu parfaitement agité par jets gazeux pour l'étude cinétique de réactions chimiques rapides. Chem. Eng. Sci., 1973. 28(1): p. 129-137.
6. D. Gielen, F. Boshell, D. Saygin, Climate and energy challenges for materials science. Nat. Mater., 2016. 15(2): p. 117-120.
7. David A. Guerrieri, P.J.Caffrey, V. Rao, Investigation into the Vehicle Exhaust Emissions of High Percentage Ethanol Blends. SAE Technical Paper, 1995 (950777).
8. S. Kim, B.E. Dale, Environmental aspects of ethanol derived from no-tilled corn grain: nonrenewable energy consumption and greenhouse gas emissions. Biomass Bioenergy, 2005. 28(5): p. 475-489.
9. A.K. Hossain, P.A. Davies, Plant oils as fuels for compression ignition engines: A technical review and life-cycle analysis. Renew. Energy, 2010. 35: p. 1-10.
10. A.P. Chalkley, Diesel engines for land and marine work. 1912.
11. A.S. Ramadhas, S. Jayaraj, C. Muraleedharan, Use of vegetable oils as I.C. engine fuels—A review. Renew. Energy, 2004. 29: p. 727-742.
12. M. Naik, L.C. Meher, S.N. Naik, L.M. Das, Production of biodiesel from high free fatty acid Karanja (Pongamia pinnata) oil. Biomass Bioenergy, 2008. 32: p. 354-357.
98
13. H. Raheman, A.G.Phadatare, Diesel engine emissions and performance from blends of karanja methyl ester and diesel. Biomass Bioenergy, 2004. 27: p. 393- 397.
14. J.B. Heywood, Internal combustion engine fundamentals. McGraw-Hill, 1988.
15. I. Hunwartzen, Modification of CFR test engine unit to determine octane numbers of pure alcohols and gasoline-alcohol blends. SAE technical paper, 1982: p. 1-6.
16. W.-D. Hsieh, R.-H.Chen, T.-L. Wu, T.-H. Lin, Engine performance and pollutant emission of an SI engine using ethanol–gasoline blended fuels. Atmos. Environ., 2002. 36: p. 403-410.
17. A.K. Agarwal, Biofuels (alcohols and biodiesel) applications as fuels for internal combustion engines. Prog. Energy Combust. Sci., 2007. 33: p. 233-271.
18. World fuel ethanol production. 2015. Retrived from http://ethanolrfa.org/resources/industry/statistics/#1454098996479-8715d404-e546
19. T. Gomiero, M.G. Paoletti, Biofuels: Efficiency, Ethics, and Limits to Human Appropriation of Ecosystem Services. J. Agric. Environ. Ethics, 2010. 23: p. 403- 434.
20. F. Ma, M.A. Hanna, Biodiesel production: a review. Bioresour. Technol., 1999. 70(1): p. 1-15.
21. H.J. Harwood, Oleochemicals as a fuel: Mechanical and economic feasibility. J. Am. Oil Chem. Soc., 1984. 61(2): p. 315-324.
22. D.Y.Z. Chang, J.H. Van Gerpen, I. Lee, L.A. Johnson, et al., Fuel Properties and Emissions of Soybean Oil Esters as Diesel Fuel. J. Am. Oil Chem. Soc., 1996. 73(11): p. 1549-1555.
23. Gas Chromatography, Sheffield Hallam University. Retrived from http://teaching.shu.ac.uk/hwb/chemistry/tutorials/chrom/gaschrm.htm.
24. J.V. Gerpen, Biodiesel processing and production. Fuel Process. Technol., 2005. 86: p. 1097-1107.
25. H.A. Dabbagh, F.Ghobadi, M.R. Ehsani, M. Moradmand, The influence of ester additives on the properties of gasoline. Fuel, 2013. 104: p. 216-223.
26. H. O'Neal, S. Benson, A Method for Estimating the Arrhenius A Factors for Four- and Six-Center Unimolecular Reactions. J. Phys. Chem., 1967. 71(9): p. 2903- 2921.
99
27. W.K. Metcalfe, C.T., P. Dagaut, H.J. Curran, J.M. Simmie, A jet-stirred reactor and kinetic modeling study of ethyl propanoate oxidation. Combust. Flame, 2009. 156(1): p. 250-260.
28. S. Dooley, H.J. Curran, J.M. Simmie, Autoignition measurements and a validated kinetic model for the biodiesel surrogate, methyl butanoate. Combust. Flame 2008. 153(1-2): p. 2-32.
29. A. M. El-Nahas, M.V. Navarro, J.M. Simmie, J. W. Bozzelli, H.J. Curran, S. Dooley, W. Metcalfe, Enthalpies of Formation, Bond Dissociation Energies and Reaction Paths for the Decomposition of Model Biofuels: Ethyl Propanoate and Methyl Butanoate. J. Phys. Chem. A, 2007. 111(19): p. 3727-3739.
30. Renewable Fuel Standard program. 2007. U.S. Environmental Protection Agency
31. M. Lewin, E.M. Pearce, Handbook of Fiber Chemistry. 1998.
32. B. Kamm, P.R. Gruber, M. Kamm, D. J. Hayes, S. Fitzpatrick, M.H.B. Hayes, J.R.H. Ross, Chapter 7, Biorefineries-Industrial Processes and Products: Status Quo and Future Directions. 2008.
33. B. Girisuta, L.P.B.M. Janssen, H.J. Heeres, Green Chemicals: A Kinetic Study on the Conversion of Glucose to Levulinic Acid. Chem. Eng. Res. Des., 2006. 84(5): p. 339-349.
34. J. J. Bozell, L. Moens, D.C. Elliott, Y. Wang, G.G. Neuenscwander, S.W. Fitzpatrick, R.J. Bilski, J.L. Jarnefeld, Production of levulinic acid and use as a platform chemical for derived products. Resour. Conserv. Recy., 2000. 28(3-4): p. 227-239.
35. Renewable fuel standard. 2015. Retrived from http://ethanolrfa.org/resources/industry/statistics/#1454098996479-8715d404-e546
36. F.M.A. Geilen, B. Engendahl, A. Harwardt, W. Marquardt, J. Klankermayer, W. Leitner, Selective and Flexible Transformation of Biomass-Derived Platform Chemicals by a Multifunctional Catalytic System. angewandte Chemie, 2010. 122(32): p. 5642-5646.
37. K. Moshammera, S. Vranckx, H.K. Chakravarty, P. Parab, R. X. Fernandes, K. Kohse-Höinghaus, An experimental and kinetic modeling study of 2- methyltetrahydrofuran flames. Combust. Flame, 2013. 160(12): p. 2729-2743.
38. H. Joshi, B.R. Moser, J. Toler, W. F. Smith, T. Walker, Ethyl levulinate: A potential bio-based diluent for biodiesel which improves cold flow properties. Biomass Bioenergy, 2011. 35(7): p. 3262-3266.
100
39. D. Klemm, B. Heublein, H.-P. Fink, A. Bohn, Cellulose: Fascinating Biopolymer and sustainable raw material. Angewandte Chemie, 2005. 44(22): p. 3358-3393.
40. C. Sheng, J.L.T. Azevedo, Estimating the higher heating value of biomass fuels from basic analysis data. Biomass Bioenergy, 2005. 28(5): p. 499-507.
41. D.L. Kaplan, Biopolymers from renewable resources. 1998.
42. T. Lei, Z. Wang, X. Chang, L. Lin, X. Yan, Y. Sun, X. Shi, X. He, J. Zhu, Performance and emission characteristics of a diesel engine running on optimized ethyl levulinateebiodieselediesel blends. Energy, 2016. 95: p. 29-40.
43. Center, H.A.R., Lower and Higher Heating Values of Hydrogen and Other Fuels. U.S. Department of Energy, Energy Efficiency & Renewable Energy, 2015.
44. S. Thion, A.M. Zaras, M. Szori, P. Dagaut, Theoretical kinetic study for methyl levulinate: oxidation by OH and CH3 radicals and further unimolecular decomposition pathways. Phys. Chem. Chem. Phys., 2015. 17: p. 23384-23391.
45. M. AlAbbad, B.R. Giri, M. Szőri, A. Farooq, On the high-temperature unimolecular decomposition of ethyl levulinate. Proc. Combust. Inst., 2017. 36(1): p. 187-193.
46. P. Dagaut, M.Cathonnet, J.P. Rouan, R. Foulatier, A. Quilgars, J.C. Boettner, F. Gaillard, H. James, A jet-stirred reactor for kinetic studies of homogeneous gas- phase reactions at pressures up to ten atmospheres (≈1 MPa). Journal of Physics E: Scientific Instruments, 1986. 19(3).
47. A. Chakir, M. Belumam, J. C. Boettner, M. Cathonnet, Kinetic Study of N- Pentane Oxidation. Combust. Sci. Technol., 1991. 77(4-6): p. 239-260.
48. P. Dagaut, M. Reuillon, M. Cathonnet, High Pressure Oxidation of Liquid Fuels From Low to High Temperature. 1. n-Heptane and iso-Octane. Combust. Sci. Technol., 1993. 95(1-6): p. 233-260.
49. Cleaner Combustion Developing Detailed Chemical Kinetic Models. 2013: Springer London.
50. O. Herbinet, B. Husson, Z. Serinyel, M. Cord, V. Warth, R. Fournet, P.-A. Glaude, B. Sirjean, F. Battin-Leclerc, Z. Wang, M. Xie, Z. Cheng, F. Qi, Experimental and modeling investigation of the low-temperature oxidation of n- heptane. Combust. Flame, 2012. 159(12): p. 3455–3471.
51. S. Singh, W.Grosshandler, P.C. Malte, R.W. Crain Jr., Oxides of nitrogen formed in high-intensity methanol combustion. Symposium (International) on Combustion, 1979. 17(1): p. 689-699. 101
52. U. Burke, W.K. Metcalfe, S.M. Burke, K.A. Heufer, P. Dagaut, A detailed chemical kinetic modeling, ignition delay time and jet-stirred reactor study of methanol oxidation. Combust. Flame, 2016. 165: p. 125-136.
53. C. Togbé, A. Mzé-Ahmed, P. Dagaut, Experimental and Modeling Study of the Kinetics of Oxidation of Methanol−Gasoline Surrogate Mixtures (M85 Surrogate) in a Jet-Stirred Reactor. Energy Fuels, 2009. 23(4): p. 1936-1941.
54. P. Dagaut, J.C. Boettner, M. Cathonnet, Kinetic modeling of ethanol pyrolysis and combustion. J. Chim. Phys., 1992. 89(4): p. 867-884.
55. P. Dagaut, C. Togbé, Oxidation kinetics of butanol–gasoline surrogate mixtures in a jet-stirred reactor: Experimental and modeling study. Fuel, 2008. 87(15-16): p. 3313-3321.
56. B. Galmiche, C. Togbé, P. Dagaut, F. Halter, and F. Foucher, Experimental and Detailed Kinetic Modeling Study of the Oxidation of 1-Propanol in a Pressurized Jet-Stirred Reactor (JSR) and a Combustion Bomb. Energy Fuels, 2011. 25(5): p. 2013-2021.
57. P. Dagaut, S.M. Sarathy, M.J. Thomson, A chemical kinetic study of n-butanol oxidation at elevated pressure in a jet stirred reactor. Proc. Combust. Inst., 2009. 32(1): p. 229-237.
58. S.M. Sarathy, M.J. Thomson, C. Togbé, P. Dagaut, F. Halter, C. Mounaim- Rousselle, An experimental and kinetic modeling study of n-butanol combustion. Combust. Flame, 2009. 156(4): p. 852-864.
59. P. Dagaut, C. Togbé, Experimental and Modeling Study of the Kinetics of Oxidation of Butanol−n-Heptane Mixtures in a Jet-stirred Reactor. Energy Fuels, 2009. 23(7): p. 3527-3535.
60. C. Togbé, A. Mzé-Ahmed, P. Dagaut, Kinetics of Oxidation of 2-Butanol and Isobutanol in a Jet-Stirred Reactor: Experimental Study and Modeling Investigation. Energy Fuels, 2010. 24(9): p. 5244-5256.
61. Z. Serinyel, C. Togbé, G. Dayma, P. Dagaut, An experimental and modeling study of 2-methyl-1-butanol oxidation in a jet-stirred reactor. Combust. Flame, 2014. 161(12): p. 3003-3013.
62. C. Togbé, P. Dagaut, A. Mzé-Ahmed, P. Diévart, F. Halter, F. Foucher, Experimental and Detailed Kinetic Modeling Study of 1-Hexanol Oxidation in a Pressurized Jet-Stirred Reactor and a Combustion Bomb. Energy Fuels, 2010. 24(11): p. 5859-5875.
102
63. G. Dayma, C. Togbé, P. Dagaut, Experimental and Detailed Kinetic Modeling Study of Isoamyl Alcohol (Isopentanol) Oxidation in a Jet-Stirred Reactor at Elevated Pressure. Energy Fuels, 2011. 25(11): p. 4986-4998.
64. C. Togbé, F. Halter, F. Foucher, C. Mounaim-Rousselle, P. Dagaut, Experimental and detailed kinetic modeling study of 1-pentanol oxidation in a JSR and combustion in a bomb. Proc. Combust. Inst., 2011. 33(1): p. 367-374.
65. A. Mzé-Ahmed, K. Hadj-A., P. Diévart, P. Dagaut, Kinetics of oxidation of a reformulated jet fuel (1-hexanol/jet A-1) in a jet-stirred reactor: experimental and modeling study. Combust. Sci. Technol, 2012. 184(7-8): p. 1039-1050.
66. L. Cai, Y. Uygun, C. Togbé, H. Pitsch, H. Olivier, P. Dagaut, S.M. Sarathy, An experimental and modeling study of n-octanol combustion. Proc. Combust. Inst., 2015. 35(1): p. 419-427.
67. S. Gaïl, M.J. Thomson, S.M. Sarathy, S.A. Syed, P. Dagaut, P. Diévart, A.J. Marchese, F.L. Dryer, A wide-ranging kinetic modeling study of methyl butanoate combustion. Proc. Combust. Inst., 2007. 31(1): p. 305-311.
68. S.M. Sarathy, S. Gaïl, S.A. Syed, M.J. Thomson, P. Dagaut, A comparison of saturated and unsaturated C4 fatty acid methyl esters in an opposed flow diffusion flame and a jet stirred reactor. Proc. Combust. Inst., 2007. 31(1): p. 1015-1022.
69. S. Gaïl, S.M. Sarathy, M.J. Thomson, P. Diévart, P. Dagaut, Experimental and chemical kinetic modeling study of small methyl esters oxidation: Methyl (E)-2- butenoate and methyl butanoate. Combust. Flame, 2008. 155(4): p. 635-650.
70. M. H. Hakka, H. Bennadji, J. Biet, M. Yahyaoui, B. Sirjean, V. Warth, L. Coniglio, O. Herbinet, P.A. Glaude, F. Billaud, F. Battin-Leclerc, Oxidation of methyl and ethyl butanoates. Int. J. Chem. Kinet., 2010. 42(4): p. 226-252.
71. G. Dayma, S. Gaïl, P. Dagaut, Experimental and kinetic modeling study of the oxidation of methyl hexanoate. Energy Fuels, 2008. 22(3): p. 1469-1479.
72. G. Dayma, C. Togbé, P. Dagaut, Detailed Kinetic Mechanism for the Oxidation of Vegetable Oil Methyl Esters: New Evidence from Methyl Heptanoate. Energy Fuels, 2009. 23(9): p. 4254-4268.
73. G. Dayma, S.M. Sarathy, C. Togbé, C. Yeung, M.J. Thomson, P. Dagaut, Experimental and kinetic modeling of methyl octanoate oxidation in an opposed- flow diffusion flame and a jet-stirred reactor. Proc. Combust. Inst., 2011. 33(1): p. 1037-1043.
103
74. P.-A. Glaude, O. Herbinet, S. Bax, J. Biet, V. Warth, F. Battin-Leclerc, Modeling of the oxidation of methyl esters—Validation for methyl hexanoate, methyl heptanoate, and methyl decanoate in a jet-stirred reactor. Combust. Flame, 2010. 157(11): p. 2035-2050.
75. M.H. Hakka, P.-A.Glaude, O. Herbinet, F. Battin-Leclerc, Experimental study of the oxidation of large surrogates for diesel and biodiesel fuels. Combust. Flame, 2009. 156(11): p. 2129-2144.
76. S. Bax, M.H. Hakka, P.-A. Glaude, O. Herbinet, F. Battin-Leclerc, Experimental study of the oxidation of methyl oleate in a jet-stirred reactor. Combust. Flame, 2010. 157(6): p. 1220-1229.
77. P. Dagaut, S. Gaïl, M. Sahasrabudhe, Rapeseed oil methyl ester oxidation over extended ranges of pressure, temperature, and equivalence ratio: experimental and modeling kinetic study. Proc. Combust. Inst., 2007. 31(2): p. 2955-2961.
78. P. Dagaut, S. Gaïl, Chemical Kinetic Study of the Effect of a Biofuel Additive on Jet-A1 Combustion. J. Phys. Chem. A, 2007. 111(19): p. 3992-4000.
79. G. Dayma, F. Halter., F. Foucher, C. Togbé, C. Mounaim-Rousselle, P. Dagaut, Experimental and Detailed Kinetic Modeling Study of Ethyl Pentanoate (Ethyl Valerate) Oxidation in a Jet Stirred Reactor and Laminar Burning Velocities in a Spherical Combustion Chamber. Energy Fuels, 2012. 26(8): p. 4735-4748.
80. H.P. Ramirez, K. Hadj-Ali, P. Diévart, G. Dayma, C. Togbé, G. Moréac, P. Dagaut, Oxidation of commercial and surrogate bio-Diesel fuels (B30) in a jet- stirred reactor at elevated pressure: Experimental and modeling kinetic study. Proc. Combust. Inst., 2011. 33(1): p. 375-382.
81. Z. Serinyel, C. Togbé, G. Dayma, P. Dagaut, Experimental and Modeling Study of the Oxidation of Two Branched Aldehydes in a Jet-Stirred Reactor: 2- Methylbutanal and 3-Methylbutanal. Energy Fuels, 2017. 31(3): p. 3206-3218.
82. A. Rodriguez, O. Herbinet, F. Battin-Leclerc, A study of the low-temperature oxidation of a long chain aldehyde: n-hexanal. Proc. Combust. Inst., 2017. 36(1): p. 365-372.
83. S. Thion, C. Togbé, G. Dayma, Z. Serinyel, P. Dagaut, Experimental and Detailed Kinetic Modeling Study of Cyclopentanone Oxidation in a Jet-Stirred Reactor at 1 and 10 atm. Energy Fuels, 2017. 31(3): p. 2144-2155.
84. S. Thion, P. Diévart, P.V. Cauwenberghe, G. Dayma, Z. Serinyel, P. Dagaut, An experimental study in a jet-stirred reactor and a comprehensive kinetic mechanism for the oxidation of methyl ethyl ketone. Proc. Combust. Inst., 2017. 36(1): p. 459-467. 104
85. R. Shahla, C. Togbé, S. Thion, R. Timothée, M. Lailliau, F. Halter, C. Chauveau, G. Dayma, P. Dagaut, Burning velocities and jet-stirred reactor oxidation of diethyl carbonate. Proc. Combust. Inst., 2017. 36(1): p. 553-560.
86. H.J. Curran, W.J. Pitz, C.K. Westbrook, P. Dagaut, J.-C. Boettner, M. Cathonnet, A Wide Range ModelingStudy of Dimethyl Ether Oxidation. Int. J. Chem. Kinet., 1998. 30(3): p. 229-241.
87. C.A. Daly, J.M. Simmie, P. Dagaut, M. Cathonnet, Oxidation of dimethoxymethane in a jet-stirred reactor. Combust. Flame, 2001. 125(3): p. 1106-1117.
88. P.A. Glaude, F. Battin-Leclerc, B. Judenherc, V. Warth, R. Fournet, G.M. Côme, G. Scacchi, P. Dagaut, M. Cathonnet, Experimental and modeling study of the gas-phase oxidation of methyl and ethyl tertiary butyl ethers. Combust. Flame, 2000. 121(1-2): p. 345-355.
89. W.W. Ayass, E.F. Nasir, A. Farooq, S.M. Sarathy, Mixing-structure relationship in jet-stirred reactors. Chem. Eng. Res. Des., 2016. 111: p. 461-464.
90. P. G. Lignola, E. Reverchon, A Jet Stirred Reactor for Combustion Studies: Design and Characterization. Combust. Sci. Technol., 1988. 60(4-6): p. 319-333.
91. D.R. Matras, Rules for the construction and extrapolation of reactors self-stirred by gas jets. Can. J. Chem. Eng., 1975. 53: p. 297-300.
92. O. Herbinet, P.-M. Marquaire, F. Battin-Leclerc, R. Fournet, Thermal decomposition of n-dodecane: Experiments and kinetic modeling. J. Anal. Appl. Pyrolysis, 2007. 78(2): p. 419-429.
93. F. Battin-Leclerc, O. Herbinet, P.-A. Glaude, R. Fournet, Z. Zhou, L. Deng, H. Guo, M. Xie, F. Qi, New experimental evidences about the formation and consumption of ketohydroperoxides. Proc. Combust. Inst., 2011. 33(1): p. 325- 331.
94. O. Herbinet, F. Battin-Leclerc, S. Bax, H.L. Gall, P.-A. Glaude, R. Fournet, Z. Zhou, L. Deng, H. Guo, M. Xie, F. Qi, Detailed product analysis during the low temperature oxidation of n-butane. Phys. Chem. Chem. Phys., 2011. 13: p. 296- 308.
95. Z. Wang, L. Zhang, K. Moshammer, D.M. Popolan-Vaida, V.S.B. Shankar, A. Lucassen, C. Hemken, C.A. Taatjes, S.R. Leone, K. Kohse-Höinghaus, N. Hansen, P. Dagaut, S.M. Sarathy, Additional chain-branching pathways in the low-temperature oxidation of branched alkanes. Combust. Flame, 2016. 164: p. 386-396. 105
96. R.L. Grob, E.F. Barry, Modern Practice of Gas Chromatography. 2004.
97. C. Wang, Parallel GC for Complete Refinery Gas Analysis. 2007.
98. W.K. Cheng, T. Summers, N. Collings, The fast-response flame ionization detector. Prog. Energy Combust. Sci., 1998. 24(2): p. 89-124.
99. J.T. Scanlon, D.E. Willis, Calculation of Flame Ionization Detector Relative Response Factors Using the Effective Carbon Number Concept. J. Chromatogr. Sci, 1985. 23(8): p. 333-340.
100. K. Zhang, C. Banyon, J. Bugler, H.J. Curran, A. Rodriguez, O. Herbinet, F. Battin-Leclerc, C. B'Chir, K.A. Heufer, An updated experimental and kinetic modeling study of n-heptane oxidation. Combust. Flame, 2016. 172: p. 116-135.
101. W. Tsang, C.M. Rosado-Reyes, Unimolecular Rate Expression for Cyclohexene Decomposition and Its Use in Chemical Thermometry under Shock Tube Conditions. J. Phys. Chem. A, 2015. 119(28): p. 7155-7162.
102. N. Atef, G. Kukkadapu, S.Y. Mohamed, M.A. Rashidi, C. Banyon, M. Mehl, K.A. Heufer, E.F. Nasir, A. Alfazazi, A.K. Das, C.K. Westbrook, W.J. Pitz, T. Lu, A. Farooq, C.-J. Sung, H.J. Curran, S.M. Sarathy, A comprehensive iso- octane combustion model with improved thermochemistry and chemical kinetics. Combust. Flame, 2017. 178: p. 111-134.
103. A. Chakir, M. Bellimam, J.C. Boettner, M. Cathonnet, Kinetic study of n-heptane oxidation. Int. J. Chem. Kinet., 1992. 24(4): p. 385-410.
104. H.M. Hakka, R.F.C., A. Pekalski, P.-A. Glaude, F. Battin-Leclerc, Experimental and modeling study of ultra-rich oxidation of n-heptane. Fuel, 2015. 144: p. 358- 368.
105. Philippe Dagaut, M.R., Michel Cathonnet, 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, 1995. 101(1-2): p. 132-140.
106. Jiaxiang Zhang, S.N., Yingjia Zhang, Chenglong Tang, Xue Jiang, Erjiang Hu, Zuohua Huang, Experimental and modeling study of the auto-ignition of n- heptane/n-butanol mixtures. Combust. Flame, 2013. 160(1): p. 31-39.
107. D.F. Davidson, Z.H., G.L. Pilla, A. Farooq, R.D. Cook, R.K. Hanson, Multi- species time-history measurements during n-heptane oxidation behind reflected shock waves. Combust. Flame, 2010. 157(10): p. 1899-1905.
106
108. J. Herzler, L.J., P. Roth, Shock tube study of the ignition of lean n-heptane/air mixtures at intermediate temperatures and high pressures. Proc. Combust. Inst., 2005. 30(1): p. 1147-1153.
109. H.K. Ciezki, G. Adomeit, Shock-tube investigation of self-ignition of n-heptane- air mixtures under engine relevant conditions. Combust. Flame, 1993. 93(4): p. 421-433.
110. C.M. Coats, A.Williams, Investigation of the ignition and combustion of n- heptane-oxygen mixtures. Symposium (International) on Combustion, 1979. 17(1): p. 611-621.
111. H.J. Curran, P. Gaffuri, W.J. Pitz, C.K. Westbrook, A Comprehensive Modeling Study of n-Heptane Oxidation. Combust. Flame, 1998. 114(1-2): p. 149-177.
112. K. Zhang, C. Banyon, J. Bugler, H.J. Curran, A. Rodriguez, O. Herbinet, F. Battin-Leclerc, C. B'Chir, K.A. Heufer, An updated experimental and kinetic modeling study of n-heptane oxidation. Combust. Flame, 2016. 172: p. 116-135.
113. O. Herbinet, B. Husson, Z. Serinyel, M. Cord, V. Warth, R. Fournet, P.-A. Glaude, B. Sirjean, F. Battin-Leclerc, Z. Wang, M. Xie, Z. Cheng, F. Qi, Experimental and modeling investigation of the low-temperature oxidation of n- heptane. Combust. Flame, 2012. 159(12): p. 3455-3471.
114. CHEMKIN-PRO. Reaction Design, 2013. 15131.
115. M. Uchiyama, A. Amano, T. Tomioka, Thermal Decomposition of Cyclohexene. J. Phys. Chem., 1964. 68: p. 1878-1881.
116. P.G. Lignola, F.P.D. Maio, A. Marzocchella, R. Mercogliano, E. Reverchon, JSFR combustion processes of n-heptane and isooctane. Symposium (International) on Combustion, 1989. 22(1): p. 1625-1633.
117. A. D'Anna, R. Mercoguano, R. Barrbella, A. Ciajolo, Low Temperature Oxidation Chemistry of iso-Octane under High Pressure Conditions. Combust. Sci. Technol., 1992. 83: p. 217-232.
118. Z. Serinyel, G. Black, H.J. Curran, J.M. Simmie, A Shock Tube and Chemical Kinetic Modeling Study of Methy Ethyl Ketone Oxidation. Combust. Sci. Technol., 2010. 182(4-6): p. 574-587.
119. S. Thion, P. Diévart, P.V. Cauwenberghe, G. Dayma, Z. Serinyel, An experimental study in a jet-stirred reactor and a comprehensive kinetic mechanism for the oxidation of methyl ethyl ketone. Proc. Combust. Inst., 2017. 36(1): p. 459-467.
107
120. M. Mehl, W.J. Pitz, C.K. Westbrook, H.J. Curran, Kinetic modeling of gasoline surrogate components and mixtures under engine conditions. Proc. Combust. Inst., 2011. 33(1): p. 193-200.
121. J.M. Hudzik, J.W. Bozzelli, Thermochemistry and Bond Dissociation Energies of Ketones. J. Phys. Chem. A, 2012. 116(23): p. 5707–5722.
122. A.M. El-Nahas, M.V. Navarro, J.M. Simmie, J.W. Bozzelli, H.J. Curran, S. Dooley, W. Metcalfe, Enthalpies of Formation, Bond Dissociation Energies and Reaction Paths for the Decomposition of Model Biofuels: Ethyl Propanoate and Methyl Butanoate. J. Phys. Chem. A, 2007. 111(19): p. 3727–3739.
123. Y. Zhang, H.J. Curran, G. Capriolo, S. Dooley, K. Djebbi, S.M. Sarathy, A. Farooq, Ignition delays and a combustion kinetic model of the lignocellulosic biofuel, ethyl levulinate. Personal Communication.
124. F. Dryer, D. Naegeli, I. Glassman, Temperature dependence of the reaction CO + OH = CO2 + H. Combust. Flame, 1971. 17(2): p. 270-272.
125. H.J. Curran, P.Gaffuri, W.J. Pitz, C.K. Westbrook, A comprehensive modeling study of iso-octane oxidation. Combust. Flame, 2002. 129(3): p. 253-280.
126. A.T. Blades, P.W. Gilderson, Kinetics of the Thermal Decomposition of Ethyl Propionate. Can. J. Chem., 1960. 38(9): p. 1412-1415.
127. W.K. Metcalfe, S. Dooley, H.J. Curran, J.M. Simmie , A.M. El-Nahas, M.V. Navarro, Experimental and Modeling Study of C5H10O2 Ethyl and Methyl Esters. J. Phys. Chem. A, 2007. 111(19): p. 4001-4014.
128. C.K. Westbrook, W.J. Pitz, P.R. Westmoreland, F.L. Dryer, M. Chaos, P. Osswald, K. Kohse-Höinghaus, T.A. Cool, J. Wang, B. Yang, N. Hansen, T. Kasper, A detailed chemical kinetic reaction mechanism for oxidation of four small alkyl esters in laminar premixed flames. Proce. Combust. Inst., 2009. 32(1): p. 221-228.
129. J.P. Orme , H.J. Curran, J.M. Simmie, Experimental and Modeling Study of Methyl Cyclohexane Pyrolysis and Oxidation. J. Phys. Chem. A., 2006. 110(1): p. 114-131.
130. Z. Serinyel, N. Chaumeix, G. Black, J.M. Simmie, H.J. Curran, Experimental and Chemical Kinetic Modeling Study of 3-Pentanone Oxidation. J. Phys. Chem. A., 2010. 114(46): p. 12176-12186.
131. M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. 108
Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman, J.V. Ortiz, J. Cioslowski, D.J. Fox, 2009.
132. E.R. Ritter, J.W.Bozzelli, THERM: Thermodynamic property estimation for gas phase radicals and molecules. Int. J. Chem. Kinet., 1991. 23: p. 767-778.
133. Z. Serinyel, G.B., H.J. Curran, J.M. Simmie, A shock tube and chemical kinetic modeling study of methyl ethyl ketone oxidation. Combust. Sci. Technol., 2010. 182: p. 574-587.
134. Olivier Herbinet, W.J.P., Charles K. Westbrook, Detailed chemical kinetic oxidation mechanism for a biodiesel surrogate. Combust. Flame, 2008. 154: p. 507-528.
135. C. Franklin Goldsmith, W.H.G., Stephen J. Klippenstein, Role of O2 + QOOH in Low-Temperature Ignition of Propane. 1. Temperature and Pressure Dependent Rate Coefficients. J. Phys. Chem. A, 2012. 116(13): p. 3325-3346.
136. J. Zádor, C.A. Taatjes, R.X. Fernandes, Kinetics of elementary reactions in low- temperautre autoignition chemistry. Proc. Combust. Inst., 2011. 37(4): p. 371- 421.
109
SUPPLEMENTAL MATERIAL
Low-temperature Oxidation Chemistry
In all engines, including spark-ignition, compression-ignition and homogeneous charge compression ignition engines, fuel reacts with oxygen at low temperatures and gradually approaches to the high temperature region because of piston compression. Note that some high CN species, such as long-chain biodiesels, start to be consumed at much lower temperatures (500 to 600 K), producing small aldehydes and alkenes [134]. Afterwards, the negative temperature coefficient zone emerges and total reactivity decreases as the temperature increases. Eventually, fuel is again consumed and ignites rapidly at high temperature. These pre-ignition phenomena are known as low temperature chemistry.
For some bio-derived fuels with large aliphatic chains attached to a methyl or ethyl ester, low temperature ignition chemistry has been found to show a similar pattern to large n-alkanes [134]. This unusual phenomenon is interpreted as negative temperature coefficient (NTC) [109], where fuel concentration increases with temperature enhancement in a certain temperature range. At low temperatures (from 500 to 700 K), auto-ignition occurs after a series of O2 additions, isomerization and chain branching, as described in
Fig. 8.1. Reactions are initiated from fuel abstraction by O2 to form the fuel radicals R•.
These fuel radicals react with O2 again and are converted, primarily to alkylperoxyl radicals
ROO• , which undergo hydrogen atom internal isomerization to hydroperoxyalkyl radicals •QOOH. The isomerization proceeds via five, six, seven and eight-membered transition state rings, in which the six-membered ring is the most favorable transition state
[111, 135]. Or, ROO• can decompose to cyclic ethers + hydroxyl radicals •OH by breaking the O-O single bond or by forming olefins + hydroperoxyl radicals HO2•. Since the free 110 radical site •QOOH is located at the carbon atom, the main pathway for •QOOH is a second
O2 addition, which forms peroxy hydroperoxyl alkyl radicals •OOQOOH. Eventually, chain branching occurs, mainly by •OOQOOH isomerization, to keto-hydroperoxide and •OH radicals, the major path for low temperature chemistry (Fig. 8.2 blue region). Next, exothermic chain branching forms keto-hydroperoxide, which promotes reactivity at low temperatures and releases more heat that favors the rate of ROO• and •QOOH producing olefins + hydroperoxyl radicals HO2• and cyclic ethers + hydroxyl radicals •OH.
Shown in the orange region in Fig. 8.2, lower reactivity results in propagations of •QOOH and ROO• to more HO2• radicals and olefins (relatively more stable species and radicals than •OH radicals). This competition is responsible for the decreased reactivity and for the NTC phenomena. The NTC region ends when HO2 builds up and self-reacts to form H2O2 + O2, where H2O2 forms 2 •OH radicals (chain branching) and increases reactivity again [135].
At high temperatures (green region in Fig. 8.2), H atom abstraction and unimolecular fuel decomposition are the main pathways where alkyl radicals R• undergo beta-scission to smaller olefins and other species. Furthermore, the •H radical, reacting with O2 to produce •O + •OH, is the dominant chain branching reaction. Two radicals keep the combustion environment reactive, consuming most fuel molecules and generating smaller and more stable species like H2O and CO2. Simulation and experimental results of biodiesels were also compared with long-chain hydrocarbons so that both species can be observed with this series of reaction classes [134]. 111
RH Initiation R• β-scission products
O2 addition
RO2•
Isomerization HO2 • + Olefin
• QOOH OH • + cyclic ether
nd R’ • + β-scission products 2 O2 addition
• OOQOOH
Keto-hydroperoxide +OH
Figure 8.1: General reaction scheme of low temperature oxidation of alkanes (RH)
with long connected secondary carbons [136].
112
Figure 8.2: Experimental (plots) [109] and modeling (lines) results for n-heptane/air mixture ignition delay times under multiple conditions [100]. Graph is divided into green, orange and blue, governed by chain branching to •O + •OH, chain propagation to HO2• +
olefins and chain branching to ketohydroperoxide + •OH, respectively.
Introduction to Kinetics and Thermodynamics
Thermodynamics
Using the thermodynamic properties: heat capacity 퐶, entropy S, and enthalpy H, one can predict and understand the combustion phenomena. As heat transferring to/from the system, heat capacity 퐶푣 is defined as the internal energy change 휕푈 by changing per unit temperature at constant volume:
휕푈 퐶 = ( ) 푣 휕푇 푣
For an isobaric system, the heat capacity 퐶푝 is defined as the enthalpy difference by changing per unit temperature at constant pressure, in the following expression:
휕퐻 퐶 = ( ) 푝 휕푇 푝
Furthermore, at a defined temperature, the enthalpy of the system can be calculated by integration:
푇2 퐻(푇2) = 퐻(푇1) + ∫ 퐶푝푑푇 푇1
Where 퐻(푇1) is a standard enthalpy usually at 273 or 298 K for difference substance.
Moreover, when it comes to a closed system, the second law of thermodynamics define the information about the direction of heat change and determine if the heat transfer 113 is spontaneous or not, as well as reversibility. In a reversible thermodynamic process, when heat transfer to a system happens, a variable entropy ∆푆 based on the expression:
푑푞 푑푆 = 푟푒푣 푇
A reversible pathway between two states 푖 and 푓 of a system can be integrated:
푓 푑푞 ∆푆 = ∫ 푟푒푣 푖 푇
To meet the criterion of second law of thermodynamics, entropy should always be positive from initial to final state when a spontaneous change occurs, giving an expression for an irreversible process:
∆푆푖푟푟푒푣 > 0
Just like enthalpy, this physical quantity can also be integrated for a reversible process:
푇푓 푑푞 푆(푇 ) = 푆(푇 ) + ∫ 푟푒푣 푓 푖 푇 푇푖
Since total entropy can be interpreted as ∆푆푖푛푡 + ∆푆푒푥푡 where ∆푆푖푛푡 is denoted as the entropy change in the system and ∆푆푒푥푡 represents the entropy change from the
푄 surrounding. ∆푆 can be replaced by − (Q is the heat transfer to the system). The 푒푥푡 푇 irreversible process of entropy change can be arranged as below, also named Clausius inequality:
푄 ∆푆 ≥ 푖푛푡 푇
As it is rearranged:
0 ≥ 푄 − 푇∆푆
In an isobaric system, the heat transfer 푄 is equal to enthalpy change ∆퐻 that Gibbs energy is introduced for a convenient expression, 114
∆퐺 = ∆퐻 − 푇∆푆
For better understanding of the spontaneity of a system or a reaction, the conclusion can be made that a reaction is spontaneous if ∆퐺 is negative and ∆퐺 = 0 indicates the criteria for equilibrium.
Reaction Kinetics
Experimentally, many rate coefficients exhibit a straight line while plotting rate coefficient in natural log form against 1/T with a negative slope. The corresponding equation can be written as below, also known as Arrhenius equation, where 퐴 is pre- exponential factor, 퐸푎 is activation energy, 푅 is the universal gas constant and 푇 is temperature.
퐸 ln 푘(푇) = ln 퐴 − 푎 푅푇
The pre-exponential factor, also called frequency factor, is to determine the number of times two molecules colliding with each other. According to collision theory, the activation energy can be defined as the minimum energy required to overcome the potential energy barrier to proceed a reaction. As far as the energy is enough to surpass the barrier, two molecules form a transition-state complex which is relatively unstable with short lifetime.
Eventually, the transition state molecule breaks into products by releasing energy in terms of heat and the equation can be organized as an alternative form with separate temperature coefficient:
−퐸 푘(푇) = 퐴푇푛exp ( 푎) 푅푇
To solve the reverse rate coefficients for thousands of reactions, the thermodynamic aspects of chemical kinetics play an important role. Based on the fundamental concept of 115 thermodynamics, Gibbs energy of activation for a reaction can be determined, relative to an equilibrium constant:
∆퐺 = −푅푇 푙푛퐾
Since an equilibrium constant is the ratio of forward and backward rate coefficients for a reaction, the above equation becomes:
∆퐺 푘 = 푘 exp (− ) 푓 푏 푅푇
푘푓 is the rate coefficient of a forward reaction and 푘푏 for backward reaction. By replacing
Gibbs energy with entropy and enthalpy of activation, the equation can be rearranged as:
∆푆 ∆퐻 푘 = 푘 exp ( )exp (− ) 푓 푏 푅 푅푇
Therefore, combining the chemical kinetics, rate coefficients of backward reactions can be calculated from two thermodynamic quantities: entropy and enthalpy.