A6. Automated Elementary Reaction Mechanism Generation Incorporating Thermochemistry, Fall-off, and Chemical Activation Reactions of OH with Olefins

Jeffrey M. Grenda and Joseph W. Bozzelli

Corporate Strategic Research
ExxonMobil Research & Engineering Company
Annandale, NJ 08801 USA



One of the major challenges of practical industrial modeling problems such as those characteristic of the development of fuel processors for fuel cell vehicles or for understanding flame propagation or autoignition in internal combustion engines is the construction of detailed chemical kinetic models based on principles of thermochemical kinetics, to represent the gas phase oxidation chemistry. Manually constructed mechanisms are extremely time and labor intensive efforts to develop, and are formidable even for simple model fuels. This is not surprising given the large number of species and reactions which can occur, the requirement of accurate estimations of thermodynamic properties and reaction rates, and the need to determine which species and reaction pathways are significant.

One method of considering larger molecular weight fuel chemistry which we have worked on extensively [1-4] is to computationally determine required chemical species and chemical reactions. This methodology automatically constructs detailed elementary step chemical kinetic mechanisms using structural-based reaction algorithms using planar graph methods [5]. We utilize this approach to both construct mechanisms from scratch as well as to complement our manually developed mechanisms.

An automated approach is presented for computational chemical kinetic mechanism generation with accurate thermochemical, and kinetic parameters which also includes analysis of chemical activation and unimolecular reactions processes, plus fall-off. The composite method is used to build kinetic models with a comprehensive set of chemical reactions and species. An example illustrating the approach for reaction of OH radicals with olefins in combustion systems is presented.

In this paper we briefly outline the methodology which has been employed to for this composite mechanism generation process. Example use of the automated generation code is the application of OH radical reactions with C3 to C5 olefins with varied branching. The analysis includes a wide temperature range, for application to several regimes in the combustion process. These OH + unsaturates examples have reasonable complexity to challenge the code including stabilization of chemically activated adducts and subsequent unimolecular reaction to several channels, as well as important reactions of the initially formed - energized adduct to new products.

The incorporation of updated thermochemical properties from recent computational chemistry literature further enhances the application, which shows a need for several important channels that are routinely neglected in combustion mechanisms.

High pressure limit kinetic parameters for input to the calculations are obtained from evaluation of literature and from ab initio or density functional calculations and use of canonical Transition State Theory. Quantum RRK analysis is utilized to obtain k(E) and master equation or modified strong collision analysis is used for evaluation of pressure fall-off in this complex bimolecular chemical activation reaction system.

Computational Mechanism Generation

The algorithm and underlying computational approaches for computationally constructing kinetic mechanisms has been described previously. [1-4] The methodology automatically constructs detailed, elementary step, kinetic mechanisms using structural-based reaction algorithms. We now include estimation of both pressure and temperature effects using a reasonably rigorous treatment of chemically activated reaction pathways. The mechanisms are constructed by using reaction rules to determining permutations of possible products from reactants, identifying species structures and generating required nomenclature, estimating the required reaction rates, and identifying the species and reactions that are kinetically significant [4].

In order to generate chemical kinetic mechanisms computationally, there are several levels of information which must be either automatically estimated in the course of the generation procedure or retrieved from databases. These include kinetic rate parameters, thermo-chemical and some physical properties, nomenclature, and reaction connectivity.

Kinetic rate estimation rules of different levels of complexity have been included for a wide variety of reaction families for both hydrocarbon pyrolysis and oxidation systems. Example reaction families include inter- and intra-molecular abstraction, recombination/dissociation, molecular elimination, and addition/beta-scission. A large number of additional reaction families are included for with subsequent isomerization and elimination pathways. The kinetic rules include structural details such as bond type and location, resonant stabilization, as well as special case reaction rates for radical species such as H, O, OH, CH3, etc. An interface with reaction rate archives has also been developed to utilize well understood or experimentally measured rate values. [1]

These approaches are coupled with newly developed automated group additivity software to estimate thermochemical and physical properties of molecules and radicals [5]. The accurate estimation of themochemical and transport properties for both radical and molecular species is critical for systems which may include wide temperature ranges over which thermodynamic properties must be well characterized. The automated thermo property estimation has been extended such that species are entered using one of several optional methods; these options range in complexity from straightforward character-string based nomenclature to detailed bond connectivity tables. These are used to computationally construct the species atomic structure and determine the appropriate group additivity parameters to estimate the thermodynamics of the species. The complexity of species may vary from small alkanes or oxygen containing hydrocarbons to multi-ring aromatics. Properties for radical and biradical species are calculated by applying bond dissociation increments to a stable parent molecule to reflect the loss of an H atom [8]. Calculation of reduced frequencies sets for estimation of density of states for use with kinetic analyses are also automatically generated. A detailed archival database functionality is built in to maintain a consistent and accurate catalog of species. The automated approach has been used to date to estimate the thermodynamic properties of well over a million different species.

1. Grenda, J.M.; Dean, A.M.; Bozzelli, J.W. "Recent Advances in Automated Kinetic Mechanism Generation" WPP at the 28th International Symposium on Combustion (Edinburgh, Scotland, Aug 2000) (manuscript in preparation)

2. Grenda, J.M.; Dean, A.M.; Bozzelli, J.W. "An Automated Method for Treating Chemically Activated Reactions Using Computational Mechanism Generation " presented at the 22nd SIAM International Conference on Numerical Combustion (Amelia Island, Fl, March 2000).

3. Grenda, J.M.; Dean, A.M.; Green, W.H.; Peczak, P.K. "Recent Advances in Automated Kinetic Mechanism Generation" , 2nd International Conference on Chemical Kinetics, (NIST, July 1997).

4. Susnow, R.G.; Dean, A. M.; Green, W. H., Peczak, P.; Broadbelt, L. J. J. Phys. Chem. A 1997, 101, 3731-40.

5. Broadbelt, L.J.; Stark, S.M.; Klein, M.T. Computers & Chemical Engineering 1996, 20, 113.

6. Grenda, J.M.; Dean, A.M.; Bozzelli, J.W. "Automated Group Additivity Estimation of Thermodynamic Properties for Use in Computational Mechanism Generation" WPP presented at the 27th International Symposium on Combustion (Boulder, CO , Aug 1998). (manuscript in preparation)

8. Lay, T.H.; Bozzelli, J.W.; Dean, A.M.; Ritter, E.R., J. Phys. Chem. 1995, 99, 14514.