Combustion characteristics of the hydrogen and ammonia co-combustion under engine-relevant conditions
Abstract
• Hydrogen and ammonia combustion mechanism is simplified with using machine leaning. • Flame structure of hydrogen and ammonia is studied under engine-relevant conditions. • Hydrogen could accelerate the ammonia combustion process due to its chemical effect. • Main way to generate NO is dehydrogenation reaction involving HNO. Ammonia-hydrogen blends are promising zero-carbon fuels for internal combustion engines; however, simulating their combustion using detailed chemical kinetic mechanisms imposes a massive computational burden. In this study, a Principal Component Analysis (PCA)-based machine learning approach is employed to develop a highly efficient reduced kinetic mechanism (50 species, 315 reactions) for ammonia-hydrogen co-combustion. The reduction strategy strictly preserves chemical fidelity by protecting critical NO X precursors. Quantitative validation against experimental data demonstrates that the reduced mechanism accurately predicts laminar flame speeds across diverse engine-relevant conditions, with mean relative errors (MRE) generally below 8%. Furthermore, comprehensive sensitivity and Rate of Production (ROP) analyses reveal a critical kinetic crossover region at approximately 40% hydrogen blending, where the radical pool fundamentally shifts from an NH X -dominated regime to an H/O/OH-dominated regime. Quantitatively, the dehydrogenation reaction is identified as the dominant route for NO formation, contributing between 25% and 40% depending on the operating conditions. Ultimately, this study provides both a robust reduced kinetic tool and deep quantitative insights into the competing reaction pathways, offering crucial guidance for optimizing zero-carbon engine combustion and mitigating pollutant emissions.
Fields of Study & Subjects
Advanced Combustion Engine Technologies
Combustion and flame dynamics
Biodiesel Production and Applications
Publication details
- Journal / source
- Results in Engineering
- Publisher
- Elsevier BV
- Volume
- 30
- Pages
- 110724–110724
- Publication date
- 26 April 2026
- Language
- EN