Automatic mechanism generation tools like the Reaction Mechanism Generator (RMG) leverage computational power and quantum chemistry to explore reaction pathways. By extrapolating known thermochemical and kinetic data, RMG creates large-scale predictive models without relying on parameter fitting to limited experimental data. This detailed approach allows the model to better extrapolate to untested conditions, offering a less biased and more comprehensive representation of chemical systems. In our study, we used RMG to generate an ammonia combustion model and validated it against experimental data. One key advantage of RMG is its ability to identify previously unexplored reactions, enhancing our understanding of combustion. However, the process also revealed unphysical species and reactions not extensively studied, prompting further refinement to ensure thermodynamic consistency. We also contribute high-level theoretical calculations for the rate coefficients of the NH + HO₂ ⇌ HNOOH ⇌ HNO + OH reaction pathway, which plays a key role in radical propagation and NOx formation.
The improved model incorporates high-level theoretical calculations to update rate coefficients, reducing uncertainty and enhancing accuracy for predicting radical concentrations and pollutant formation. The resulting mechanism shows better agreement for simulated laminar burning velocities at fuel-rich conditions for NH3-H2 blends compared to existing models. This work demonstrates how combining RMG, quantum chemistry, and master equation methods can generate robust, predictive combustion models. By expanding reaction networks and incorporating new pathways, we provide a more reliable framework for ammonia combustion, contributing to more efficient and environmentally friendly systems.