Rate-based screening of pressure-dependent reaction networks

Computer tools to automatically generate large gas-phase kinetic models find increasing use in industry. Until recently, mechanism generation algorithms have been restricted to generating kinetic models in the high-pressure limit, unless special adjustments are made for particular cases. A new approach, recently presented, allows the automated generation of pressure-dependent reaction networks for chemically and thermally activated reactions (Grenda et al., 2000; Grenda and Dean, in preparation; Grenda et al., 1998; see Refs. [1-3]). These pressure-dependent reaction networks can be quite large and can contain a lar-e number of unimportant pathways. We thus present an algorithm for the automated screening of pressure-dependent reaction networks. It allows a computer to discover and incorporate pressure-dependent reactions in a manner consistent with the existing rate-based model generation method. The new algorithm works by using a partially-explored (or "screened") pressure-dependent reaction network to predict rate constants, and updating predictions as more parts of the network are discovered. It requires only partial knowledge of the network connectivity, and allows the user to explore only the important channels at a given temperature and pressure. Applications to vinyl + O-2, 1-naphthyl + acetylene and phenylvinyl radical dissociation are presented. We show that the error involved in using a truncated pressure-dependent network to predict a rate constant is insignificant, for all channels whose yields are significantly greater than a user-specified tolerance. A bound for the truncation error is given. This work demonstrates the feasibility of using screened networks to predict pressure-dependent rate constants k(T, P). (C) 2001 Elsevier Science B.V. All rights reserved.