Addressing climate change requires substantial reductions in CO2 emissions from the transportation sector, where alternative fuels for internal combustion engines play a crucial role. Hydrogen stands out as a compelling energy carrier capable of enabling low-carbon combustion while leveraging existing engine technologies. Its adoption can support a transition toward fuel-flexible powertrains and deliver rapid decreases in exhaust carbon emissions. This approach is particularly relevant for hard-to-abate segments, where full electrification remains challenging. Building on this perspective, this numerical study investigates the modelling behaviour of a heavy-duty port fuel injection (PFI) internal combustion engine fuelled with hydrogen. Initially, the mixture was assumed to be fully premixed to avoid uncertainties related to injection and mixing processes and to significantly reduce computational cost; this assumption was subsequently validated through selected injection simulations. A methodology was then developed to ensure robust model responses by analysing convergence over three consecutive cycles and by appropriately defining the initial and boundary conditions, as well as mesh resolution. Three representative experimental operating points were investigated: full load, maximum power, and cruise conditions. Two combustion modelling approaches were then compared. ECFM, a flamelet-based model, demonstrated its ability to match experimental data through a calibration process that accounts for turbulence-chemistry interactions via the adjustment of model parameters. In contrast, SAGE is a detailed chemistry solver that employs a kinetic reaction mechanism to directly compute reaction rates, without requiring calibration. The comparison highlighted certain limitations of SAGE arising from its underlying approach, whereas ECFM showed more stable and reliable behaviour, albeit with the need for case-specific calibration.