Low-pressure direct injection of hydrogen through outward-opening hollow-cone injectors is attractive for maximizing usable tank content, yet even at nominal injection pressures near 20 bars, the annular flow chokes and discharges as an under-expanded jet whose shock-induced expansion, vortex dynamics, and mixing govern in-cylinder performance. Here, a computed-tomography-informed, experimentally constrained compressible computational fluid dynamics framework is developed to resolve the transient dynamics, thermodynamics, and mixing of such jets from a production-representative Bosch hollow-cone injector. A time-resolved inlet mass-flow rate boundary condition derived from measured needle-lift and injected-mass data drives the simulations, which are validated against high-speed schlieren over ambient pressures of 1–5 bar (penetration, projected area, morphology) and against local one-dimensional Raman measurements of hydrogen concentration and temperature. Validation against Raman measurements showed that the predicted hydrogen concentration field was largely unaffected by the selected energy formulation, while noticeable differences were observed in the predicted temperature distribution. The jets are governed by upstream and downstream recirculation and coherent vortex rings whose scale and intensity depend on ambient pressure. Crucially, lower ambient pressure increases penetration and projected area but does not improve mixture homogenization, as large coherent vortical structures promote axial convection, allowing locally rich pockets to persist over longer downstream distances. In contrast, higher ambient pressure produces a more confined jet with a shorter characteristic mixing length, promoting more rapid dilution of the rich hydrogen core and faster homogenization toward lean and near-stoichiometric conditions despite the reduced jet penetration.
Hydrogen engines are designed to run at ultra-lean conditions to suppress knocking and nitrogen oxides (NOx). This study aims to systematically assess various types of combustion strategies in achieving ultra-lean combustion. To ensure high-fidelity modeling of ultra-lean hydrogen-air combustion, several important model developments were conducted. In particular, ultra-lean hydrogen flames exhibit increased propensities in wall quenching and pronounced diffusive-thermal (DT) instability. To accurately describe these characteristics in turbulent and high-pressure conditions, a comprehensive combustion model was developed to represent DT-turbulence interaction and flame quenching effects, along with an updated chemical kinetic mechanism with detailed NOx formation. The integrated models were validated against experiments from port-fuel injection (PFI), direct-injection spark-ignition (DISI), and pre-chamber combustion (PCC) configurations. Detailed analysis of simulation results revealed that, relative to the PFI mode, the DT effects are mitigated in DISI conditions due to mixture stratification, whereas stronger effects are observed in PCC operation. Neglecting DT effects in the PCC configuration leads to near-misfire behavior, primarily due to underpredicted combustion rates in the pre-chamber region. Moreover, neglecting flame quenching is found to yield an overprediction of wall heat transfer loss and underprediction of hydrogen slip, especially for the PCC mode. A comparative assessment of the three combustion modes under identical operating conditions indicates that PFI produces the lowest NOx emissions due to the more homogenous lean mixture burning, while PCC results in slightly higher NOx emissions due to elevated combustion temperatures. Nevertheless, comparable peak thermal efficiencies of approximately 47% are achieved across all three combustion modes.