Evaluating Approaches to Simplify Automatically Generated Oxidation Reaction Mechanisms
NAZIFA SAYEED, Jia Jiang, Kelley Barsanti, Gabriel Isaacman-VanWertz, Virginia Tech
Abstract Number: 365
Working Group: Aerosol Chemistry
Abstract
Oxidation in the atmosphere of a single reactive organic compound can generate thousands or tens of thousands of products. Chemical mechanisms used in atmospheric modeling need to be simplified to avoid slowing down simulations and/or making it difficult to interpret or manage output. This project aims to develop and examine a generalizable method for generating oxidation mechanisms of reactive organic compounds that can be efficiently and reasonably implemented into detailed chemical models without significantly biasing the output products. We developed a Python-based automated pipeline to interface with the SAPRC MechGen mechanism generation tool to simulate the oxidation of volatile organic compounds (VOCs) under various environmental conditions. We use this tool to evaluate different methods for simplifying the reaction chemistry, primarily focusing on approaches internal to MechGen that winnow reactions by yields and an external “step-forward” approach that limits the number of generations and allowable number of reactions of each product. To enable the external approach, the system performs automated generation of reaction pathways, cleans and restructures the data, and simplifies the reaction network by replacing transient radicals that are not typically treated in simple chemical models (i.e., alkyl, alkoxy) with their stable and/or peroxy radical products. Output chemistry is provided in a format designed for integration into the F0AM (Framework for 0-dimensional Atmospheric Modeling) chemical box model. We evaluate differences in the product mixture using several different simplification approaches: full multi-generational oxidation using MechGen, directly applied with different thresholds for yields under which reactions are removed, and “step-forward” chemistry using MechGen for each step with different numbers of generations and products considered. The chemistry of the product mixture of a testbed simulation of alpha-pinene oxidation is examined under each scenario, including distributions of elemental composition, aerosol formation, and gas-phase reactivity, and compared to explicit chemical mechanisms previously validated against experimental data.