Rate Constant Estimation for Large Chemical Kinetic Models and Application to Biofuels

Transcription

1 Rate Constant Estimation for Large Chemical Kinetic Models and Application to Biofuels ICCK 2001, MIT July 28, 2011 William J. Pitz, Henry J. Curran, Charles Westbrook, Marco Mehl, S. M. Sarathy and Taku Tsujimura, P.. Box 808, Livermore, CA This work performed under the auspices of the U.S. Department of Energy by under Contract DE-AC52-07NA27344

2 Development of chemical kinetic models for fuels Fundamental experimental measurements Ab initio calculations oh 7/13/ 0 rucich 1o 1 0 0g e e e e e e e e e e e e e e+00 4 c3h8+oh<=>nc3h7+h2o 1.054e e+03 C1C2 base chemistry Thermodynamic database Reaction rate constants reactions h+o2<=>o+oh 9.65E E+04 o+h2<=>h+oh 5.080e e+03 oh+h2<=>h+h2o 2.160e e+03 o+h2o<=>oh+oh 2.970e e+04 h2+m<=>h+h+m 4.577e e+05 h2/ 2.5/ h2o/ 12/ co/ 1.9/ co2/ 3.8/ o+o+m<=>o2+m 6.165e e+00 h2/ 2.5/ h2o/ 12/ ar/.83/ co/ 1.9/ co2/ 3.8/ ch4/ 2/ c2h6/ 3/ he/.83/ o+h+m<=>oh+m 4.714e e+00 h2/ 2.5/ h2o/ 12/ ar/.75/ co/ 1.5/ co2/ 2/ ch4/ 2/ c2h6/ 3/ he/.75/ Detailed chemical kinetic model for practical fuels Reaction rate rules High temperature mechanism Reaction class 1: Unimolecular fuel decomposition Reaction class 2: H atom abstractions from fuel Reaction class 3: Alkyl radical decomposition Reaction class 4: Alkyl radical + 2 = olefin + H 2 Reaction class 5: Alkyl radical isomerization Reaction class 6: H atom abstraction from olefins Reaction class 7: Addition of radical species to olefins Reaction class 8: Alkenyl radical decomposition Reaction class 9: lefin decomposition 2

3 Need reaction rate rules for many chemical classes of fuels Alkanes Alkenes Cycloalkanes Aromatics Alcohols Methyl esters (biodiesel compounds) Carbenes (aldehydes, ketenes) Special structures in intermediate species: Alkylhydroperoxides Alkylperoxy H 3

4 Need reaction rate rules for many types of reaction steps R -RH H Fast High Temperature Combustion Long Chain Alkanes + 2 H + + H 2 + H Reactivit ty Low T Mechanism NTC Hi T Mechanism + 2 H + H Reactor Temperature - H H Degenerate Branching Path H + + 4

5 Assign reaction rate rules by reaction classes High temperature mechanism Reaction class 1: Unimolecular l fuel decomposition Reaction class 2: H atom abstractions from fuel Reaction class 3: Alkyl radical decomposition Reaction class 4: Alkyl radical + 2 = olefin + H 2 Reaction class 5: Alkyl radical isomerization Reaction class 6: H atom abstraction from olefins Reaction class 7: Addition of radical species to olefins Reaction class 8: Alkenyl radical decomposition Reaction class 9: lefin decomposition 5

6 Reaction classes for low temperature reactions Low temperature mechanism Reaction class 10: Alkyl radical addition to 2 (R + 2 ) Reaction class 11: R + R R 2 = R + R R Reaction class 12: Alkylperoxy radical isomerization Reaction class 13: R 2 + H 2 = RH + 2 Reaction class 14: R 2 + H 2 2 = RH + H 2 Reaction class 15: R 2 +CH 3 2 = R + CH Reaction class 16: R 2 + R 2 = R + R + 2 Reaction class 17: RH = R + H Reaction class 18: R Decomposition Reaction class 19: QH = Cyclic Ether + H Reaction class 20: QH = lefin + H 2 Reaction class 21: QH = lefin + Aldehyde or Carbonyl + H Reaction class 22: Addition of QH to molecular oxygen 2 Reaction class 23: 2 QH isomerization to carbonylhydroperoxide + H Reaction class 24: Carbonylhydroperoxide decomposition Reaction class 25: Reactions of cyclic ethers with H and H 2 6

7 Reaction rate rules make the assignment of reaction rate constants manageable Class 2 Fuel + (H, H, CH 3, H 2 ) => fuel radical + (H 2, H 2, CH 4, H 2 2 ) H- atom abstraction rate rules for alkanes C-H type A (cm 3 mol -1 s -1 ) n E A (cal) E ,756 H e , E , E ,586 H E E E ,154 CH E , E E ,160 H E , E ,090 7

8 Class 1 Reaction rate rule issues: fuel decomposition reactions Alkanes Set by reverse reaction Exothermic direction C-C C bond breaking most important C-C-C-C <=> C. + C-C-C. C-C-C-C <=> C-C. + C-C. Some variations in forward rate constants, even though you think they should be all the same 8

9 Class 2 Reaction rate rules for H-atom abstraction from alkanes Fuel + (H, H, CH 3, H 2 ) => fuel radical + (H 2, H 2, CH 4, H 2 2 ) C-H type A( (cm 3 mol -1 s -1 ) n E A (cal) E ,756 H e , E , E ,586 H E E E ,154 CH E , E E ,160 H E , E ,090 9

10 1.E-10 New Argonne n-butane+oh data for = radical+h2o H + alkanes Class Sirvaramakrishnan et al (Argonne) n-butane + H = butyl + H2 1.E-11 Droege and Tully 1986 (Sandia) LLNL-NUIG Reaction rate rules 1.E-12 10

11 1. E E E New measured and calculated rate constants for H + alkanes are higher at T > 900K Class 2 n propane+oh = radical+h2o Sivaramakrishnan 2009 theory Sivaramakrishnan i 2009 experiments LLNL NUIG reaction rate rule Squares: Droege and Tully 1986 experiments Sirvaramakrishnan 2009 propane + H = propyl + H2 LLNL-NUIG rate rule (based on Tully) 11

12 Ab initio calculations show higher rates due to higher primary rate Class 2 n propane+oh propane+oh= radical+h2o k cm 3 mole 1 s sec 1 1.E+13 Argonne s calculated cross-over LLNL primary LLNL secondary Argonne primary (ab initio) Argonne secondary (ab initio) Tully primary (Experimental) Tully secondary (Experimental) Tully s measured crossover propane + H = propyl + H2 1.E /T[K] 12

13 H-atom abstraction from the fuel: H2 + alkanes Uncertainty in rate of a factor of 3-6 Class 2 C3H8+H2 => ic3h7+h22 CSM Ignition very sensitive to this rate constant under RCM conditions NUIG 13

14 Fuel + H2 shows high sensitivity when the fuel is hydrogen Class 2 Sensitivity results under conditions in rapid compression machine: H22+H H2+H2 [ms] Ignition delay bar end of compression phi=1 H2/2/N2/Ar = 12.5/6.25/18.125/ Sensitivity: Rate Constants x 2 H22+H<=>H2+H2 Baseline H+2<=>+H H+2+M<=>H2+M H22+M<=>H+H+M Baseline H2+H<=>H+H H2+H<=>H2+2 H+H2<=>H+H2 +H2<=>H+H H22+H<=>H2+H H22+H<=>H2+H2 H2+H2<=>H22+2 H22+<=>H+H2 Most sensitive reaction: H22+H<=>H2+H2 1.E+14 1.E+13 New rate constant fit: New fit H22+H <=> H2+H2 Baulch et al Temperature of Sensivity analysis (963 K) Tsang and Hampson, 1986 Ellingson 2007 Baldwin and Walker /Tc [1/K] Important branching sequence at high pressure: H2+H2=>H+H22 H22=>H+H Retarding reaction: H2+H2=>H22+2 Log k 1.E+12 1.E+11 1.E+10 New computed rate, Ellingson 2007 Fit, new H2 mechanism /T[K] 14

15 Class 3: alkyl radical decomposition. Improvements for iso-octane Class mes [ms] 100 Stoichiometric mixtures 10 atm ion Delay Ti atm Ignit Dashed Previous mechanism on website Solid: Updated version K/T 15

16 More accurate estimate for iso-octyl radical decomposition rate constant: Class 3 1.E+13 1.E+12 1.E+11 1.E+10 cc8h17 => tc4h9 + ic4h8 LLNL original Colorado School of Mines ab binitioiti Klippenstein ab initio Log k 1.E+09 1.E+08 1.E+07 1.E+06 1.E+05 1.E+04 Milano generic for alkyl radicals 1.E /T[K] 16

17 Low temperature reactions: Effect of R-2 bond strength varies with bond type and controls amount of low temperature chemistry Class 10 R+2 R2 Values used in LLNL models 0 Bond dissociation energy ( H 298 ):R R- => R del H

18 Low temperature chemistry: R QH isomerizations Class 12 6 Member ring isomerization H H K 6 = 2.5E+10 exp(-20450/rt) 5 Member ring isomerization H H K 5 = 2.0E+11 exp(-26450/rt) 18

19 R 2 isomerizations: Rate constants from computational chemistry (Dean, Carstensen et al. Colorado School of Mines) Class 12 5-member TS 6-member TS H R H R secondary primary tertiary primary tertiary Activation energy depends on ring size and overall thermochemistry Amenable to rule generation 19

20 Significant differences in CSM vs. LLNL rate constants: R2 isomerization Class 12 CBS-QB3 results generally lower than LLNL values for 5-member TS CBS-QB3 results much higher than LLNL values for 6-member TS Mainly due to higher A-factors (much higher than alkyl isomerizations) Differences lead to significantly different reaction pathways 20

21 Mechanisms for fuels are built in a hierarchical manner and increase rapidly in size with fuel size Hydrogen Methane 8 Species - 20 Reactions 30 Species Reactions Propane 100 Species Reactions CH 4 2 H H 2 H 2 C 2 H 6 C 2 H 5 CH 3 2 CH 3 CH 3 H H 2 H 2 H C 2 H 4 CH 3 CH 3 H C 2 H 3 CH 2 CH 2 H 2 C 2 H 2 HC H 2 2 Aromatics C H Soot C 2 21

22 Fuels Size and Mechanism Size n-alkane C 8 H 18 C 10 H 22 C 12 H 26 C 14 H 30 C 16 H 34 Mechanism Size (Detailed Mechanism) 700 Species 3150 Reactions 950 Species 4050 Reactions 1250 Species 5150 Reactions 1650 Species 5150 Reactions 2100 Species 8150 Reactions 22

23 Application of rules to biofuels Biodiesel Large methyl esters Alcohols Iso-pentanol Butanol Aromatics lefins 23

24 Biofuels Biodiesel New types of biofuels Biomass derived from algae and other single cell organisms rapeseed Algal pilot scale bioreactor in Lawrence, Kansas From: Smith, Sturm, denoyelles and Billings, Trends Ecol. Evol. (2010) 24

25 Algal oil-derived fuels contain additional esters From: Marchese and B. Fisher, "Measurement of Gaseous and Particulate Emissions from Algae- Based Fatty Acid Methyl Esters," SAE

26 Soybean and rapeseed derived biodiesels have only 5 principal p components R triglyceride R Fatty acid methyl esters (FAMEs): + 3 CH 3H Methyl Palmitate (C16:0) methanol % R 3 CH 3 + R Soybean Rapeseed methyl ester H H glycerol H C16:0 C18:0 C18:1 C18:2 C18:3 Methyl Stearate (C18:0) Methyl leate (C18:1) Methyl Linoleate (C18:2) Methyl Linolenate (C18:3) 26

27 Assembled chemical kinetic model for all of the five main components in biodiesel derived from soybeans or rapeseed oil methyl palmitate methyl stearate Built with the same reaction rate rules as our successful methyl decanoate and methyl decenoate mechanism methyl oleate methyl linoleate methyl linolenate 5 component mechanism, approximately 5,000 species 20,000 reactions Model with all 5 components now published and available: Westbrook, Naik, Herbinet, Pitz, Mehl, Sarathy and Curran, "Detailed chemical kinetic reaction mechanisms Lawrence Livermore for soy National and rapeseed Laboratory biodiesel fuels," Combustion and Flame,

28 Experimental validation: New biodiesel model reproduces oxidation of n-decane/methyl palmitate mixture in jet stirred reactor 1.0 Methyl palmitate 0.8 Conversio on n-decane Stoichiometric fuel/ 2 /He mixtures 1 atm 1.5 s residence time Temperature - K Jet stirred reactor data: Hakka et al. Comb Flame

29 Many of the predicted species profiles compare well with experiments: e.g. 1-heptene 8.0E E-05 1-heptene Mole fractio on 4.0E-05 Jet stirred reactor data: Hakka et al. Comb Flame E E Temperature - K 29

30 Biodiesel components ignite in order of number of double bonds Ignitio on delay - ms 100 Engine-like stearate conditions: 13.5 bar linoleate Stoichiometric fuel/air mixtures 10 1 palmitate oleate linolenate /T - K 30

31 Increased number of double bonds reduces low T reactivity of individual components in stirred reactor at diesel conditions Diesel engine conditions of high pressure and fuel-rich mixtures: 50 bar, =2 (Fuel: 200 ppm, residence time = 0.05 s) methyl stearate CN101 methyl oleate CN Simulated conversions of biodiesel components CN101 Conver rsion CN 59 CN 23 methyl linolenate CN Stearate t 101 leate 59 Linolenate 23 Jet stirred reactor Derived cetane numbers from Knothe (2010) Temperature - K 31

32 C = C double bonds reduce low T reactivity s s a v v a s s -C C C C = C C C C- s s a a s s Inserting one C=C double bonds changes the reactivity of 4 carbons atoms in the C chain Allylic C H bond sites are weaker than most others Therefore they are preferentially abstracted by radicals 2 is also very weakly bound at allylic sites and falls off rapidly, inhibiting low T reactivity 32

33 We have seen the same effect in hydrocarbon fuels: hexenes C = C - C - C - C - C 1-hexene C - C = C - C - C - C 2-hexene C - C - C = C - C - C 3-hexene R2 isomerization initiates low temperature reactivity Moving the double bond towards the center of the molecule blocks more R2 kinetics [ms] Ignition delay time Ignition delay times in a rapid compression machine of hexene isomers ( MPa, Φ=1): Hexene 2-Hexene 3-Hexene T [K] Experimental data: Vanhove et al. PCI2005 Simulations: Mehl, Vanhove, Pitz, Ranzi Combustion and Flame

34 Plant and animal fat oils have different fatty acid profiles that affect reactivity in a diesel engine palmitate stearate oleate linoleate linolenate CN With models for all 5 major components, we can now model all these types of biodiesel: Not a surrogate model, but a real biodiesel (B100) model! 34

35 Use Diesel PRF as a scale to compare reactivity of biodiesel compounds Jet stirred reactor Simulated Diesel PRF scale in a PSR Diesel PRF mixtures 1 (n-hexadecane and 2,2,4,4,6,8,8-heptamethylnonane) CN60 CN50 50 bar =2 2 fuel: 200 ppm =0.05s Con nversion CN40 CN20 CN CN50 CN40 CN60 CN Temperature - K As CN increases, reaction in PSR starts at lower temperatures and has a greater extent of low T combustion 35

36 Diesel PRF scale allows assessment of the reactivity of biodiesel from different sources Beef tallow (CN58) Jet stirred reactor Conversio on Simulated reactivity profiles for biodiesel Biodiesel fuels fuels CN60 (PRF) CN20 (PRF) Rapeseed (CN54) Linseed (CN39) Temperature - K linseed beef tallow peanut olive soy rapeseed CN20 CN60 50 bar =2 fuel: 200 ppm =0.05s (PRF) (PRF) 36

37 bservations on reactivity of biodiesel fuels from different oils Methyl ester fuels from different plant and animal fats and oils have different reactivity Detailed composition of these biodiesel fuels determine their reactivity Biggest factor for reactivity variability of biodiesel, large methyl ester fuels is the number of C=C double bonds We can model kinetics of most of these biodiesel fuels using the new biodiesel kinetic mechanism The mechanisms still need refinements and testing, and careful laboratory experiments would be very valuable 37

38 What & Why Isopentanol? A Next Generation BioFuel: Isopentanol (3-Methyl-1-Butanol or 3 Methylbutane-1-ol) is one of biomass derived alcoholic fuel, like Ethanol The challenge of JBEI: To convert all monomer sugars (hexoses and pentoses) released from depolymerization of lignocellulosic biomass into transportation fuels and other chemicals. And the initial targets of JBEI is ethanol, butanol, isopentanol, hexadecane, and geranyl decanoate ester. Higher alcohols such as isopentanol has higher energy density and lower hygroscopicity compared to ethanol. Volatility is moderate like gasoline, Not too high 38

39 Approach Development of Isopentanol reaction mechanism Single-zone Simulations Validation Study of the kinetics involved in the autoignition process Simulate an HCCI Engine Combustion Compare with representative experimental results 39

40 Development of Reaction Mechanism High temperature chemistry: Unimolecular decomposition and H atom abstraction from fuel by activated radicals mainly occur Alcohols have weak C-H bonds at site Low temperature t chemistry: Based on low temp. chemistry of isooctane because isooctane has some similar structures to isopentanol Results showed Too Short Ignition Delay & Too Strong NTC Concerted elimination of H 2 : Concerted elimination forming aldehyde and H 2 from R 2 is so fast that low 2 2 temperature reactions would be slowed down C H 40

41 Schematic Energy Diagram for the Concerted Elimination of H H abstraction by radicals H H + Radicals H H II + H 2 H I C H 41

42 c] Ignition dela ay time [ se Validations of Reaction Mechanism : 0.5 P ini i : 0.7, 2 MPa ST Exp., P: 2.0MPa CV Cal., P: 2.0MPa RCM Exp., E P: 2.0MPa 20MP RCM Cal., P: 2.0MPa ST Exp., P: 0.7MPa CV Cal., P: 0.7MPa RCM Exp., P: 07MP 0.7MPa RCM Cal., P: 2.0MPa c] Ignition dela ay time [ se : 1.0 P ini i : , 0.8, MPa ST Exp., P: 2.3MPa CV Cal., Cal P: 2.0MPa 20MPa RCM Exp., P: 2.0MPa RCM Cal., P: 2.0MPa ST Exp., P: MPa CV Cal., P: 0.8MPa RCM Exp., P: 0.7MPa RCM Cal., P: 0.7MPa ,000/T / [1/K][ / ] 10,000/T 000/T [1/K] Isopentanol model developed in this study can reproduce the experimental data which were acquired under various, T, and P conditions with a shock tube and an RCM Shock tube experiments: Kenji Yasunaga, Fiona Gillespie, and Henry Curran (NUI Galway - Ireland) Lawrence Rapid compression Livermore National machine Laboratory (RCM) experiments: Bryan Weber, Yu Zhang and Chih-Jen Sung (UConn.) 42

43 Developed chemical kinetic model for new biofuel iso-pentanol and compared it to experiments in Sandia HCCI engine Iso-pentanol mechanism HCCI engine experiments: Yang and Dec, Sandia, SAE 2010 New generation biofuel proposed by DE Joint BioEnergy Institute (JBEI) Reaction rate rules on successful isooctane because it has some similar structures Model development and application: LLNL visiting scientist Dr. Taku Tsujimura National Institute of Advanced Industrial Science and Technology, Japan 43

44 Iso-pentanol model predicts correct combustion phasing as load is increased in Sandia HCCI engine Experiments and Calculations: Required T BDC for constant combustion phasing C [deg.c] T BDC Iso-pentanol with EGR Exp. CA10: deg.ca Cal. CA50: deg.ca Exp. CA10: deg.ca m : 0.38 Cal. CA50: deg.ca P in [kpa] 44

45 Iso-pentanol model predicts intermediate heat release that allows high load operation for HCCI Iso-pentanol Experiments Calculations l TDC TDC CA10: deg.ca m : 0.38 no EGR CA50: deg.ca m : 0.38 no EGR HCCI engine experiments: Yang and Dec, Sandia, SAE

46 Developed model for 4 isomers of butanol and compared model predictions to flame experiments at USC butanol mechanism: 4 isomers Flame speed measurements: Egolfopoulos et al. USC H Twin premixed counterflow flames Iso-butanol is a new type of biofuel that can be made directly from cellulose using bacteria 46

47 Lamin nar Flame Velo ocity (cm/s) Butanol mechanism accurately simulates flame speeds important for predicting spark ignition engine combustion 1 Butanol Equivalence Ratio Lamin ar Flame Velo city (cm/s) iso Butanol 10 0 Experimental data: Veloo, Egolfopoulos et al. 2010, Equivalence Ratio fuel/air mixtures 1 atm Lam minar Flame Ve elocity (cm/s) Butanol Lawrence Livermore Equivalence National Ratio Laboratory locity (cm/s) inar Flame Ve Lam tert Butanol Equivalence Ratio 47

48 Butanol model well predicts ignition delay times at pressures and temperatures found in IC engines 1 Rapid compression machine iso-butanol tert-butanol Symbols: experimental data Sung et al., AIAA paper, 2011 Ign nition Dela ay time (s s) butanol n-butanol butanol isomers 15 atm, phi=1, in air /T (1/K) Rapid compression machine University it of Connecticut t 48

49 Chemical kinetic mechanism for larger aromatics C C 2 H 2 The kinetic mechanism of the aromatics has an intrinsic hierarchical structure A new module specific to C8 alkyl aromatics is now under development 49

50 p-xylene mechanism well reproduces species profiles in jet stirred reactor 1.E 01 1.E 03 1.E 02 2 Toluene Mole Fraction 1.E 03 1.E 04 1E 05 1.E 05 p-xylene C C 2 CH 2 Mole Fraction 1.E 04 1.E 05 Benzene Cyclopentadiene Benzaldehyde raction Mole Fr 1.E 06 1.E 02 1.E T [K] CH 4 H 2 1E 04 1.E C 2H 4 1.E 05 C 2 H 4 1.E T [K] Jet stirred reactor P = 1 atm, Ф = 1, τ = 0.1s Experiments: Gail and Dagaut Combustion and Flame 2005 C 2 H 6 1.E T [K] 50

51 rtho-, para- and ethyl-benzene models compare well to ignition delay times measured at pressure and temperatures relevant to engines Ignition delay times in a shock tube for aromatics ion Delay Times [µs] Ignit atm, fuel/air mixtures, =1 rtho-xylene Para-xylene Ethylbenzene Shock tube experimental data: Shen and ehlschlaeger, Combustion and Flame K/T 51

52 Mechanisms are available on LLNL website and by LLNL-PRES

53 Summary Reaction classes and reaction rate rules greatly simplify the task of developing chemical kinetic models and assigning rate constants Continually updating reaction rate rules and adding new rules for new moieties such as those from new biofuels Made a lot of progress in chemical kinetic modeling new classes of compounds like esters and alcohols and difficult compounds to model like aromatics 53

54 Acknowledgements Fokion Egolfopoulos, butanol Jackie Sung, iso-pentanol John Dec and Yi Yang, iso-pentanol 54

55 Acknowledge support from: DE ffice of Vehicle Technologies Gurpreet Singh Kevin Stork DE ffice of Basic Energy Sciences Wade Sisk DD ffice of Naval Research 55