This paper reports the validation of a three-dimensional numerical simulation of the in-cylinder processes during gas-exchange, injection, and compression in a direct-injection, hydrogen-fueled internal combustion engine. Computational results from the commercial code Fluent are compared to experimental data acquired by laser-based measurements in a corresponding optically accessible engine. The simulation includes the intake-port geometry as well as the injection event with its supersonic hydrogen jet. The cylinder geometry is typical of passenger-car sized spark-ignited engines. Gaseous hydrogen is injected from a high-pressure injector with a single-hole nozzle. Numerically and experimentally determined flow fields in the vertical, central symmetry plane are compared for a series of crank angles during the compression stroke, with and without fuel injection. With hydrogen injection, the fuel mole-fraction in the same data plane is included in the comparison as well. The results show that the simulation predicts the flow field without injection reasonably well, with increasing numerical-experimental disagreement towards the end of the compression stroke. The injection event completely disrupts the intake-induced flow, and the simulation predicts the post-injection velocity fields much better than the flow without injection at the same crank-angles. The two-dimensional tumble ratio is evaluated to quantify the coherent barrel motion of the charge. Without fuel injection, the simulation significantly over-predicts tumble during most of the compression stroke, but with injection, the numerical and experimental tumble ratio track each other closely. The evolution of hydrogen mole-fraction during the compression stroke shows conflicting trends. Jet penetration and jet-wall interaction are well captured, while fuel dispersion appears under-predicted. Possible causes of this latter discrepancy are discussed.

This content is only available via PDF.
You do not currently have access to this content.