(731a) Generating Detailed Pyrolysis Models for Large Systems

One technology that stands to fundamentally change the way we model large kinetic systems is automatic mechanism generation software. Automatic generation software, including the Reaction Mechanism Generator (RMG), work by having a computer enumerate possible reactions between species in the model, determine which of these reactions are relevant, and even determine the model parameters by various estimation and ab initio schemes. This software has already successfully modeled various small molecule combustion and pyrolysis systems, as shown by their ability to predict how the system will behave experimentally before ever performing an experiment on the system. With increasing available computational power and better methodologies, the next obvious use case for automatic mechanism generation software is to model systems containing larger molecules that would be difficult to study otherwise. Besides the obvious challenge of computational time needed to enumerate all relevant elementary reactions for systems containing large molecules, there are further challenges that need to be overcome, including obtaining high accuracy kinetic data for large molecules, and validating the whole methodology with experimental data for the hundreds if not thousands of species formed. In this work, we present a detailed kinetic model for pyrolysis of a liquid hydrocarbon mixture, generated using RMG, that predicts the concentrations of major and minor products and their time profiles reasonably well. We discuss how to refine large models like this in a timely manner using efficient sensitivity analysis, how to obtain high accuracy ab initio data for large molecules using modern quantum chemical methods, and how to validate these models against GCxGC-FID/MS data. Some of the remaining challenges associated with this approach are also discussed.