The 11th International Conference on Modeling and Diagnostics for Advanced Engine Systems, Chiba, Japan

Dr. Yizhuo Feng represented our team at the The 11th International Conference on Modeling and Diagnostics for Advanced Engine Systems, Chiba, Japan, From:15/12/2025 to: 18/12/2025 .(Conference/Paper Link).

Hydrogen internal combustion engines (H2ICE) have emerged as a potential solution for carbon-free transportation. Due to its low volumetric energy density, the hydrogen injection system is a critical component for H2ICE. Solenoid injectors with an outward-opening pintle, which generate a hollow-cone spray, are an attractive option because of the high flow rate and flexible injection control. However, the computational fluid dynamics (CFD) simulation for the hydrogen injection and mixing process presents challenges, particularly regarding the fuel properties and the hollow-cone jet characteristics, as well as the supersonic flow. This study provides a detailed assessment of numerical modelling for a high-pressure hydrogen jet with an outward-opening injector within the Reynolds-averaged Navier-Stokes (RANS) framework. At first, a simplified 2D geometry domain was applied to carry out the simulations, considering the high computational cost for high-speed gaseous jet simulations. The numerical framework was evaluated based on a quasi-steady 2D simulation. The results indicate that the spatial discretisation scheme significantly influences the development of the hydrogen jet. The second-order discretisation should be employed for the momentum transport equation, as the first-order scheme leads to an unexpected Coandă effect on the injector pintle bottom. However, the other scalar transport equations with second-order discretisation may introduce local numerical dispersion errors. A mesh dependency study indicates that the development of the hydrogen jet core is highly sensitive to the grid scale, resulting in a significant effect on the hydrogen mass flow rate. It was confirmed that the 0.125 mm base grid with an adaptive mesh refinement (AMR) embedding achieves a good balance between mesh convergence and computational efficiency. Moreover, a transient injection simulation using a real injector geometry was conducted and validated against experimental results with various ambient pressure conditions in a constant volume chamber. The predicted jet penetration shows a good agreement with the experimental measurement at 3 and 5 bar ambient pressure conditions. Nevertheless, an AMR strategy based on mass fraction may be ineffective during injection under higher ambient pressure conditions, resulting in a local coarse mesh and a Coandă effect along the chamber wall, which can lead to a significant underestimate of the penetration. Instead, a velocity-based AMR strategy can embed the mesh refinement more appropriately during the hydrogen injection, yielding better predictions.