Under-expanded hydrogen jet fires pose a critical safety challenge, yet remain difficult to model with low-Mach CFD solvers. This study compares two strategies for representing the high-pressure jet with Fire Dynamics Simulator (FDS): (i) an Eulerian, notional nozzle inlet based on fully expanded conditions, and (ii) a Lagrangian, momentum-carrying particle source that transfers mass and momentum to the resolved field. Validation against Sandia vertical jet experiments examines flame length, centerline temperature, and radiative heat flux. A systematic parametric study for particle diameter, size distribution, injection offset, and the basis of inlet conditions for the Lagrangian approach quantifies the sensitivities and identifies practical settings that balance accuracy and cost. The Lagrangian approach reproduces the measured flame length evolution and centerline temperatures and matches the reported radiant fraction; the Eulerian case underpredicts core temperature and radiation. Feature-importance analysis isolates injected particle diameter as the dominant control parameter. Grid/cost comparisons indicate a practical compromise at grid size of 0 . 10 m . • Eulerian and Lagrangian CFD of hydrogen jet flames. • Lagrangian method compares favorably against experimental data. • Sensitivity analysis identifies particle diameter as key parameter.
Alibakhshian et al. (Sun,) studied this question.