Detailed numerical simulation of ammonia/diesel combustion under CI engine conditions

Ammonia is considered a promising carbon-free fuel for marine engines, yet challenges persist in utilizing ammonia as a carbon-free marine fuel due to its inherently low reactivity and significant emissions of NOx and N2O. To address these issues, advanced combustion strategies, such as ammonia/diesel dual-fuel Reactivity-Controlled Compression Ignition (RCCI), have been proposed. This thesis presents both detailed numerical simulations and advanced modeling approaches to thoroughly investigate ammonia/diesel RCCI combustion under compression-ignition engine conditions. Employing Direct Numerical Simulation (DNS), fundamental insights are obtained into ignition mechanisms, flame structures, and emission formation pathways. Complementing the DNS studies, computationally efficient combustion modeling strategies, specifically optimized Flamelet Generated Manifold (FGM) and Chemistry Coordinate Mapping (CCM) models, are developed and validated, enabling practical predictions of complex dual-fuel combustion phenomena.

The DNS analysis reveals a complex interplay among multiple reaction zones, including lean premixed flames (LPF), rich premixed flames (RPF), diffusion flames (DF), and rich ammonia oxidation layers (RAOL). The combustion and emission characteristics are largely governed by the individual properties of these reaction layers and their complex interactions. The LPF significantly influences combustion efficiency and NO formation, whereas the DF primarily serves as the major NO-consuming zone. Moreover, the formation and consumption of N$_2$O are heavily influenced by interactions between LPF, RPF, and RAOL, highlighting the importance of detailed flame-structure interactions in RCCI combustion. Parametric studies identify the effects of ambient pressure, temperature, scalar dissipation rate, and ammonia equivalence ratio on ignition, flame propagation, and pollutant formation.

For practical computational applications, advanced combustion modeling strategies based on DNS results are proposed and evaluated. The enhanced FGM model utilizes optimized tabulation strategies derived directly from one-dimensional RCCI configurations, effectively capturing multiple reaction zone combustion phenomena. Additionally, the CCM model, augmented by direct error control and an optimized parallel load-balancing strategy, significantly improves computational efficiency without compromising accuracy. Both models accurately reproduce complex RCCI combustion dynamics with substantial computational savings, achieving approximately 160 times speedup for the FGM model and several times speedup for the CCM model compared to full DNS calculations.