Supersonic fuel injection is a critical physical process in high-Mach-number propulsion systems, and its efficient mixing and combustion characteristics directly impact engine performance. This paper addresses the engineering bottleneck of high grid complexity in traditional numerical simulations caused by direct modeling of injector orifices. We innovatively introduce the source term method from film cooling applications in aeronautical engines into the field of supersonic injection research, proposing two novel source term loading approaches (surface source and point source). Through benchmark cases of co-flow direct injection and transverse jet, the effectiveness of these methods is systematically validated. The co-flow direct injection test demonstrates that the surface source loading method can more accurately reproduce the near-field fuel jet distribution characteristics at the nozzle exit. The predicted temperature field and OH radical concentration show high consistency with results from direct numerical simulations (DNS) of physical orifices. While the point source loading exhibits robustness in predicting macroscopic wave patterns, it shows discrepancies in capturing near-field physical features, leading to deviations in localized combustion behavior from real-world conditions. The transverse jet validation reveals that the surface source loading successfully replicates the momentum injection and turbulent generation mechanisms of physical orifices through continuous source term application. Its performance metrics—including jet penetration depth, mixing front propagation trajectory, and consistency in combustion modes—outperform the point source approach. The point source loading, however, causes spatial heterogeneity in combustion modes during multi-orifice injection due to imbalanced initial momentum distribution, deviating significantly from experimental observations. Engineering applicability analysis indicates that the surface source method aligns highly with experimental data in predicting near-field flow structures, combustion reaction zone localization, and wall pressure distributions. It effectively balances simulation accuracy and computational efficiency, offering a new pathway for high-fidelity engineering simulations.