A numerical study of ammonia combustion in spark-ignition and reactive-fuel pilot-ignition engines

Decarbonizing internal combustion engines (ICEs) requires the use of fuels produced from renewable energy, with easy storage and characterized by a combustion process with zero carbon dioxide (CO2) emissions. Ammonia (NH3) perfectly fits all these requirements. However, its use as fuel for ICEs calls into question many of the consolidated aspects related to ignition and flame propagation processes studied during the last decades. NH3 differs from conventional hydrocarbon fuels for a higher minimum ignition energy and auto-ignition temperature, as well as for a lower combustion speed and energy density. Experimental investigations carried out in both metal and optical engines proved the feasibility of NH3 operation as pure fuel in spark-ignition (SI) engines or in reactive-fuel pilot-ignition (RFPI) engines with a pilot injection of a high-reactivity fuel. In this work, computational fluid dynamics (CFD) methodologies consolidated with conventional fuels are applied to simulate a selection of operating points on such experiments. The flame area model (FAM) from Weller is employed for the SI operation, while the tabulated well-mixed (TWM) model is used for the RFPI mode. The effects from a NH3-air dilution, a spark-timing advance and an increase in injection duration are studied to identify the main challenges related to the NH3 combustion modelling in ICEs. The results show that numerical models capture the measured trend of spark-timing and injection duration variations at both stoichiometric and lean NH3-air mixtures. However, for the SI mode, aspects such as the laminar-to-turbulent transition stage and the heat release rate dependency on the ignition time require further modeling improvements. Similarly, for the RFPI mode, the auto-ignition delay of the dual-fuel mixture and the turbulent flame speed are numerically underestimated. Therefore, all these aspects represent challenges that need to be addressed in CFD models to improve NH3 ignition and combustion prediction.