The prediction of temporal and spatial concentration profiles of gas cloud is of great importance in safety issues. Existing models for predicting dense gases release are used for flat plates, and they cannot usually involve the complex environments, hence in this article, computational fluid dynamics have been used for prediction of dense gases behavior. The selection of turbulence model shows its significance in the results of computational fluid dynamics. In order to select the best turbulence model, the models of k–ɛ and k–ω have been studied. Experimental data of the test no. 26 of Thorney Island Series data have been extracted. The results show that k–ɛ realizable model is closest to the experimental data. This model has the closest and most appropriate prediction of the spatial and temporal profile. The model is also able to predict the phenomenon of gravity slumping associated with dense gas dispersion.