Numerical Analysis of Directly Injected Hydrogen into a High-Pressure Chamber

Due to the historical use of fossil fuels, the transportation sector has been one of the major producers of greenhouse gas (GHG) emissions and harmful pollutants. Because of its high specific energy content and the potential for zero carbon-based emissions, Hydrogen Internal Combustion Engines have emerged as an option for decarbonizing heavy-duty transportation. However, injecting high-pressure gas into pressurized combustion chambers induces complex compressible flow phenomena, including choked flow and under-expanded supersonic jet structures, which challenge conventional modeling approaches. Accurate prediction of the resulting fuel-oxidizer mixing process is critical for optimizing engine performance and emissions. This study conducts a numerical investigation of transient hydrogen injection into a high-pressure argon environment, benchmarking a 2D axisymmetric Computational Fluid Dynamics (CFD) model against high-fidelity experimental optical measurements. Utilizing Ansys Fluent with a density-based solver, coupled with the k-omega SST turbulence model and species transport equations, simulations were performed at injection pressures of 6 MPa and 10 MPa into a 1 MPa ambient chamber. The simulation successfully captured fundamental compressible physics, including Mach disk formation and significant expansion cooling near the nozzle exit. Validation results revealed a strong dependency on the nozzle pressure ratio (nPR). At 6 MPa (nPR=6), the model achieved good agreement with experimental data, predicting tip penetration depth within 10% . However, at 10 MPa (nPR=10), while axial penetration depth predictions remained within the 10% error margin, they were consistently underestimated, and radial dispersion was significantly under-predicted. These findings quantitatively demonstrate the limitations of the 2D axisymmetric assumption at high energy levels, concluding that accurately resolving mixing in high-nPR injection systems may require capturing inherently three-dimensional turbulent entrainment mechanisms.