Towards automated development of real fuel combustion models: from quantum chemistry to chemical kinetics.

Abstract

The development of predictive chemical kinetic models for real fuel combustion requires the integration of accurate quantum chemistry methods, statistical dynamical rate theories, thermochemical property estimation, and systematic mechanism construction. In this perspective, we discuss the current state of this pipeline as applied to combustion chemistry, from wave function and density functional theory methods for the characterization of potential energy surfaces, through transition state theory, its variational and multi-structural extensions, and master equation approaches for pressure-dependent kinetics, to the estimation of thermochemical properties via quantum chemistry, group additivity, and, increasingly, machine learning. We highlight representative case studies, including from our group, involving large and flexible species, multi-well and multi-channel reactions, and species with pronounced multi-reference character, which illustrate both the maturity and the remaining limitations of these methodologies. We further discuss automatic mechanism generation and reduction strategies that translate elementary-reaction-level accuracy into tractable kinetic models for large fuels. Throughout, we identify open challenges, including the treatment of large chemical systems with numerous conformers, the reliable description of multi-reference electronic structure, the quantification and propagation of uncertainties, and the assumptions inherited from small-molecule studies that may not hold for large hydrocarbons. We conclude by discussing how artificial intelligence and machine learning methodologies are starting to accelerate and, in some respects, reshape each stage of this pipeline, and what this implies for the future development of predictive and transferable combustion kinetic models.

Abstract as supplied by the indexing source. © the authors / publisher.