Question? Leave a message!




Computer-Aided Construction of Chemical Kinetic Models

Computer-Aided Construction of Chemical Kinetic Models
ComputerAided Construction of Chemical Kinetic Models: Towards Predictive Chemistry William H. Green MIT Dept. of Chemical Engineering DOE Combustion Meeting Potomac, MD, May 2014 1 Why do we want predictive fuel chemistry models • Historically, Fuels have changed pretty slowly… Engines evolved faster, using moreorless static fuels. Fuel was a constant, not a variable… • But now, fuels are changing – To reduce greenhouse gas, laws require renewable fuels – Conventional light sweet petroleum is getting more more expensive – Possible to make many future fuels with synthetic biology chemistry • If fuel is a variable, we need to be able to predict how engine+fuel system depends on fuel composition, not just on engine operating conditions. – Much faster to explore new concepts on computer than by building prototype engines and synthesizing hundreds of barrels of a new fuel for engine tests • Scientifically, we don’t really understand something until we can make quantitative predictions…and discrepancies between those predictions and experiments reveal opportunities to learn something new 2 Challenge: Lots of alternative fuels to evaluate And many types of engines, too Not practical to synthesize test every possible future fuel in every possible future engine… …rapid, easy predictions are valuable and needed 3 Automated Predictive Chemical Kinetics: The Concept Motivation • Rapid, Easy Predictions are valuable needed – Assess alternative fuels feedstocks – Design new processes and engines • Hydrocarbon/Fuel chemistry is complicated – many reactions, species: need to automate 4 Fuel Chemistry models are very big • Many chemical intermediates between fuel and CO . 2 • Multiple competing pathways • Nonlinear, sometimes nonmonotonic kinetics • Real fuels (gasoline, jet, diesel) have many chemical species – So even initial conditions are complicated • Models built by hand: 1,000 species, 5,000 reactions – Human tries to include just the more important reactions. • Models built by computer: 500 species, 10,000 reactions – RMG software typically considers 30,000 species and 100,000 reactions • most of them are found to be numerically negligible 5 We need this black box reactor/engine design emissions in Computer detail chemistry builds and solves the knowledge engine fuel combustion (clearly documented) performance simulation fuel composition error bars operating conditions Similar issues for fuelformation chemistry e.g. in refinery 6 Commercial software can solve detailed kinetic simulations… …if one can supply the fuel’s reaction mechanism. Simulation predictions Diff. Eq. solver Very long Simulation list of Interpreter reactions equations (CHEMKIN, with rate Cantera, KIVA, dY/dt = … parameters GTPower) 7 How we construct fuel chemistry models Chemistry knowledge Simulation predictions Unambiguous documentation of assumptions about how molecules react Diff. Eq. solver Very long list of Simulation Interpreter reactions equations (CHEMKIN, with rate Cantera, KIVA, dY/dt = … parameters GTPower) 8 Chemical Kinetic Modeling Challenges • Identify all important reactions species – But not unimportant species reactions: how to distinguish • Estimate all reaction rate coefficients (and properties, e.g. thermochemistry) to sufficient accuracy. – We use Functional Group extrapolations Quantum Chemistry • Large models pose numerical and computer problems – Very challenging for humans to handle, interpret, debug… …SO WE TRY TO AUTOMATE EVERYTHING We build on prior efforts by combustion research community, e.g. Comprehensive Chemical Kinetics 35 (1997) Advances in Chemical Engineering 32 (2007) 9 RMG algorithm: Faster pathways explored further, growing the model Before: “Current Model” inside. RMG decides whether or not to add species to this model. Final model typically After: 500 species, 8000 rxns OpenSource RMG software. Download from rmg.sourceforge.net 10 Kinetic Model Predictions Rely on Quantum Chemistry for Thermo, Rates • Functional Group approximation – Compute a few examples of each reaction type with quantum, then use same barrier, A factor for analogous reactions. • Most of our calculations at CBSQB3 level – Geometries, Vibrational Frequencies from DFT – Single point energies at stationary points at higher level – Extrapolation to Basis Set Limit • Recent calculations use F12 methods – Explicit dependence on distance between every pair of electrons – Much faster basis set convergence • Most calculations rely on common approximations – RigidRotor HarmonicOscillator approximation – Conventional Transition State Theory (dividing surface at saddle point) – Simple corrections for internal rotors and tunneling – Modified Strong Collision approx. for k(T,P) Are computed thermo, rates accurate enough 11 flow reactors Testing Accuracy of Model Predictions vs. Experiment: CEFRC data on Butanols Pyrolysis (shock tube) Flame RCM Speeds MBMS Shock With collaborations from tube other institutes like Flame Univ. Ghent, NIST. Speed s Rare Situation where detailed data available at many different conditions 12 We used RMG to build a mechanism for butanol pyrolysis and combustion. Four isomers, very different octane numbers. nbutanol secbutanol isobutanol tertbutanol Octane number = 86 Octane number = 98 Octane number = 100 More reactive Less reactive RMG considered about 30,000 possible species, selected as important: • 372 chemical species • 8,723 reactions Important k’s with quantum chemistry Shamel S. Merchant, E.F. Zanoelo, R.L. Speth, M.R. Harper, K.M. Van Geem and William H. Green, Combustion Flame (2013) 13 Model predicted butenes yield from butanols pyrolysis accurately 15 1Butanol (1Butene) isoButanol (isoButene) 2Butanol (1Butene) 2Butanol (2Butene) Pyrolysis, tertButanol (isoButene) 10 T 1000 K P 2 bar t seconds 5 Experiments: Van Geem et al. Univ. Ghent 0 0 5 10 15 Experimental butene yield = wt 14 Predicted butene yield = wtThe kinetic model also quantitatively predicts formation of aromatics from butanols (via rather complicated reaction sequences) Data from K. Van Geem, Ghent pyrolysis of isobutanol 1000 K, 2 atm, 2 seconds 15 When you change reaction conditions, need to add reactions, compute rate coefficients Predictions of 1000 K pyrolysis model for 1467 K OH concentration H O concentration 2 16 Microsecond H O formation at 1400 K 2 sensitive to different reactions than longtime product formation at 1000 K. Bond scissions dehydration Stanford pyrolysis of nbutanol 9 Early Times 10 sec t=3 ms So… computed improved estimates of bond scissions 17 dehydrations based on quantum chemistry…. Quantum calcs for k(T,P) sensitive at 1467 K significantly changes predictions OH concentration H O concentration 2 18 Using quantum calcs (rather than rough estimates by analogy) for most sensitive high T reaction k(T,P) improves predictions of shock tube pyrolysis: It is always a good idea to do the quantum calcs for the Sensitive numbers in the model OH concentration H O concentration 2 Experimental data: Stranic et al., Combust. Flame (2012). See also RosadoReyes et al., J. Phys. Chem. A (2013). 19 Advanced Light Source allows quantification of dozens of species including key radicals in flames • Flames are analyzed with molecular beam timeofflight mass spectrometry • Photoionization with tunable synchrotrongenerated VUV photons allows identification of species • by mass • by ionization energy Data measured by Nils Hansen (Sandia) at Advanced Light Source (LBNL). Hansen et al. PCCP (2012) Hansen et al. Combust. Flame (2013) 20 Often only a few smallmolecule reactions are P dependent. But in MBMS flames, Many Rates Strongly Pdependent 5,398 “Normal” k(T,P)’s Chemistry in our butanols model… MBMS Flame B. M. Wong, D. M. Matheu, and W. H. Green. J. Phys. Chem. A 2003, increasing pressure dependence 107, p. 62066211. 21 increasing pressure dependence Enols sensitive to Hatomcatalyzed chemically activated ketoenol tautomerization + +  “Chemicallyactivated” = “product reacts faster than thermalized”  Chemical activation is a major complication in automated reaction generation: keep track of “wellskipping” reactions, compute k(T,P)  Instead of 2 possible products, 10 are formed.  Instead of 2 Transition States, must compute 16 TS’s. 22 Speciation profiles confirm predictive capabilities: nbutanol flame • Major species are predicted accurately • A more powerful test is provided by comparing modeled and experimental profiles of intermediate species • Originally one significant deviation, due to typo in thermo for one radical (C H ) 4 5 Dozens of additional species traces, variety of flames: all show comparably good agreement. For isobutanol we worked in predictive mode, with similar level of agreement with expt. Significant uncertainties due to calibrations, uncertainties in T(z), boundary conditions at burner, probe 23 Our model predicted Hansen’s MBMS measurements on isobutanol very well (see Hansen et al., Combust. Flame 2013) However, Experimental Data consistent with a model prediction does NOT prove the model is correct 24 Although they all match MBMS data, Literature Models for isoButanol Flame Significantly Differ MIT Model (Green group) Mi KA lano UST Model Model (Fr (Sarathy assoldati et et al al.). ) Many parameters in detailed kinetic models: just because it matches experiment does not mean it is the truth 25 Slide from Nils Hansen Flame Speed predictions look good – but are they accurate enough What accuracy needed Experimentallyderived numbers not completely consistent, involve tricky extrapolation to zero strain. Error bars may be underestimated Model Prediction Model very sensitive to HCO + H O = H + CO + H O 2 2 This k alone leads to uncertainties in predicted flame speed comparable to deviations on this plot. Data from Veloo Egolfopoulos (343 K) , and W. Liu … C.K. Law (353 K), both in Proc Combust Inst (2011). 26 Model quantitatively predicts highT ignition delays for all butanol isomers conditions Stranic et al., Combust. Flame, 2012, 159 (2), 516527. 27 Model also predicts lowT ignition delays in air at conventional conditions fairly well, and the dependence of t on fuel In these RCM experiments by Weber CJ Sung, P N :O ratio is 2 2 held fixed (as in air), and so O is also 2 effectively fixed. Most expts are done this way. Then changes in f are mostly changing fuel. 28 We don’t know everything: model completely mispredicts O sensitivity of lowT ignition 2 delay of butanols Const. Fuel Model: No O 2 In Air dependence Model predicts Expts: 1.5 fuel dependence τ O 2 29 Early stages of ignition: positive feedback versus damping (OH loss channels) We don’t understand O2 dependence in butanols Low T ignition. So we have backed up to propane (with Steve Klippenstein) 30 Discrepancy Led To Discovery of New Type of Reaction of Peroxides, Dominant at Low T, longer t (JSR or liquid phase) Jalan et al., J. Am. Chem. Soc. (2013) Green Truhlar Groups 31 It’s not just butanols…Green group recently made predictive models for many other fuels where fewer data are available • Diisopropyl Ketone (large collaboration with Sandia etc.) – See Allen et al., Combust. Flame (2014) • Ethylamine (Buesser et al., about to be submitted) • JP10 (in collaboration with Ghent, Aerodyne) • Decane mixtures w/isooctane toluene (surrogate jet fuel) • Cineole (in collaboration with Sandia) • Pentanol (in collaboration with Xi’an, Ghent) • IsoPentanol (in collaboration with Ghent) • Thioethers Thiols (organosulfur) pyrolysis 32 • Phenyldodecane pyrolysis Summary • Kinetic models based on quantum chemistry + rate estimates are predictive for huge range of combustion/oxidation/pyrolysis experiments. – Big models can be built and refined pretty quickly. – Experimentalists + Theorists team very effective. • Quick to identify and resolve discrepancies, whether due to model or expt. – Method useful for assessing proposed new biofuels • We are working to extend method to more complicated fuels, accelerate process of building models, add more heteroatom chemistry. • This approach is great for quickly seeing what is predicted by current knowledge of combustion chemistry. • Sometimes discover serious Discrepancies. • Discrepancies motivate ongoing work to improve on current knowledge, and to develop better methods. 33 Acknowledgements Most of the theoretical methods and computer programs described here, and this whole approach, were developed with funding over many years from this DOE Basic Energy Sciences Chemical Physics program managed by Mark Pederson Wade Sisk. The modeling work reported here was done primarily by MIT student Shamel Merchant, with help from other members of Green group. This work benefited greatly from collaborations with other groups supported by DOE Basic Energy Sciences: Stephen J. Klippenstein, Ronald K. Hanson, ChihJen Sung, Donald G. Truhlar, Fokion N. Egolfopoulos, Chung K. Law, Nils Hansen We also gratefully acknowledge our collaborators Kevin Van Geem Guy Marin at U. Ghent, Richard West at Northeastern U., and C. Franklin Goldsmith at Brown U. 34