Numerical study on spray characteristics of liquid ammonia based on variable model constants

•A variable discharge coefficient formula is designed to consider the effect of flash boiling on nozzle flow rate.•The KH-RT breakup model constant is adjusted to improve the spray characteristics prediction accuracy.•Low prediction errors for spray characteristics under various conditions is achieved.•The applicability of the variable model under wider conditions is verified. Ammonia as a carbon-free fuel for internal combustion engines can reduce carbon emissions in the transportation sector. For ammonia engines, liquid ammonia direct injection allows precise injection control and theoretically enables higher thermal efficiency. However, liquid ammonia is susceptible to flash boiling and is accompanied by significant variations in spray characterization under different conditions due to its low boiling point, which brings the challenge of numerically predicting liquid ammonia sprays. Therefore, the aim of this study is to achieve high-precision numerical predictions of liquid ammonia sprays at various ambient pressures and fuel temperatures. A formula of variable discharge coefficient that varies with ambient pressure and fuel temperature was first designed to account for the impact of spray resistance caused by liquid ammonia vaporization within the nozzle on the injector flow characteristics during various flash boiling stages. The results indicate that the spray penetration prediction error is less than 1.5 % for all ambient pressures at a fuel temperature of 308 K. A sensitivity analysis of the KH-RT breakup model constants was subsequently conducted, and a variable formula, based on the constant significantly affecting spray penetration, was developed to correct the spray characteristics prediction at higher temperatures. The spray penetration prediction error was less than 5 % for all ambient pressures at a fuel temperature of 338 K, with the prediction error approaching 0 % during the flare flash boiling stage. Finally, simulations were conducted under fuel temperatures of 293 K and 350 K to further validate the applicability of the variable constants model.