Graph-Based Generation and Reduction of Complex Chemical Reaction Networks.

The steep growth of chemical reaction networks, quickly reaching sizes that hinder kinetic analysis, calls for effective reduction methods. Existing methods rely heavily on detailed rate data and are routinely validated only on selected published mechanisms, thus, limiting their scope. We now introduce first a bipartite network generation model developed to reproduce structural and kinetic features of both published and enumerated combustion mechanisms, enabling the large-scale simulation and benchmarking of reduction strategies. Building on this foundation, a topology-based algorithm (MolRank) was developed. It ranks the redundancy of reactions and species, allowing network pruning using only thermodynamic data. Validation across thousands of simulated networks and its application to combustion reaction networks show that MolRank can prune redundant species and reactions on a very large scale. We suggest this framework as a scalable route for testing and applying reduction methods, opening new opportunities for the analysis of combustion, atmospheric, and biochemical reaction systems.