In emergency events where hazardous heavy gases are injected into supply air, the rapid prediction of heavy gas dispersion is significantly important. The linear superposition relationship based on a fixed flow field offers the advantage of fast predictions; however, the buoyancy due to the density difference destabilizes the flow field. In this work, the applicability of a linear relationship based on transient accessibility index in predicting heavy gas dispersion delivered from supply air was studied. The dimensionless transient concentrations predicted by the linear model were compared with those of CFD (computational fluid dynamics) simulation. The numerical results from two heavy gases, carbon dioxide (CO2) and hydrogen sulfide (H2S); two supply mass fraction concentrations, 4E-4 and 4E-2; and two air distributions, ceiling supply side down return (CSD) and side up supply side down return (SUSD), were analyzed. The results showed that the flow field and heavy gas concentrations at a low supply concentration of 4E-4 were slightly different from those of the passive gas. The air jet exhibited sinking characteristics at a high supply concentration of 4E-2. Significant prediction deviations using the linear model mainly occurred at a few positions surrounding the supply air jet and in the upper space for the CSD and SUSD. An acceptable accuracy was achieved with average deviations ranging from 7.8%–15.5%. High heavy gas density and heavy gas concentration in supply air increased the prediction deviation. This study provides support for the rapid assessments of emergency scenarios in the context of ventilation decisions.