Numerical simulation and PSO-based analysis of NOₓ, CO, and CO₂ emissions in MILD combustion of methane/ammonia/hydrogen blends with steam addition

• CFD study of methane MILD combustion with H₂, NH₃, and steam additives • Novel chemical mechanism combines GRI-Mech 2.11 with Stagni et al. • H₂ raises furnace temperature; NH₃ and steam reduce peak flame temperature • H₂ and NH₃ lower CO₂; steam increases CO; additives impact NO differently • PSO finds optimal blend (20% H₂, 19% steam) cutting CO 88%, CO₂ 29%, NO 73% This study presents a computational investigation of MILD (Moderate or Intense Low-oxygen Dilution) combustion of methane blended with hydrogen (H₂) and ammonia (NH₃), with steam (H₂O) introduced into the oxidizer. A hybrid chemical mechanism was developed by combining GRI-Mech 2.11 with the kinetics of Stagni et al. for NH₃/H₂ to capture the radical pathways governing pollutant formation. Turbulence–chemistry interaction was modeled using the Eddy Dissipation Concept (EDC) within a RANS framework, while radiative heat transfer was treated with the Discrete Ordinates (DO) method using WSGGM properties. All simulations were run to steady convergence to ensure accuracy and consistency. Analysis of Variance (ANOVA) showed that all three additives significantly influenced temperature and emissions. Hydrogen increased flame temperature and reactivity, while ammonia and steam lowered peak temperatures, with steam further affecting flame stability. Hydrogen and ammonia both reduced CO₂ emissions, but tended to increase NO. Steam, in contrast, effectively suppressed NO. CO decreased with hydrogen addition but increased with ammonia and steam. To achieve simultaneous reduction of pollutants, a Particle Swarm Optimization (PSO) framework was applied. The optimal mixture, containing 20% hydrogen and 19% steam, reduced CO by 88%, CO₂ by 29%, and NO by 73% compared with baseline methane combustion. The novelty of this work lies in integrating an extended NH₃/H₂ mechanism with detailed CFD of MILD combustion and multi-objective optimization using PSO and ANOVA.