عنوان مقاله [English]
Most physical systems in nature have nonlinear dynamics and the system linearization of these physical systems is only a simplified assumption. The atmospheric motions, with the same philosophy, have a turbulent structure and ommiting the turbulent motions in the atmosphere is only a problem simplification assumption. When a buoyant jet of a chimney enters the atmosphere, it behaves like a turbulent flow, in which the atmospheric turbulence and self-generated turbulence of the plume play major parts. The present survey aimed at demonstrating the effects of turbulence on plume dynamics through computer simulation. The well-known turbulent flow simulation methods commonly used to simulate plume dynamics and atmospheric processes are: Direct numerical simulation (DNS), Reynolds averaged Navier-Stokes (RANS) and large eddy simulation (LES), which most distinguishing feature is their way of parameterizing the turbulence. As far as computational requirements, accuracy and turbulence simulation, the former two models are the two extremes while LES occupies an intermediate position between them, directly simulating the large-scale eddies and parameterizing the less important sub-grid scale (SGS) dissipative processes using sub-grid models (SGM). Most often, LES can predict the unsteadiness and intermittency of the turbulence structure, which is the most important feature of a buoyancy-driven jet. It should be noted that it is not efficient to employ full LES method when tackling an issue with certain unimportant zones. Furthermore, in the case of strong turbulent motions, the scale of flow structures near the rigid bodies are small and LES method method requires very fine grids that can increase its computational cost as large as DNS. To surmount this drawback, a hybrid RANS-LES method with a new mixed scale sub-grid parameterization model was applied to simulate the turbulent plume dynamics in ANSYS Fluent 14.5 software. The effectiveness and the accuracy of the mentioned turbulence simulation method was demonstrated through simulation study and experimental data in the neutral atmospheric conditions. Comparing the simulation results of the RANS method, the default hybrid RANS-LES method with static sub-grid scale parameterization and the new RANS-LES method with dynamic mixed scale parameterization indicated that the mean temperature profile at stack downstream was more accurately predicted by the new hybrid method. The root mean square error of plume rise estimation of the RANS method, the default RANS-LES method and the new hybrid RANS-LES method were 0.0437, 0.054 and 0.0323, respectively. It was further demonstrated that the Briggs integral plume rise model could not properly predict the plume rise in the presence of turbulence, because it does not consider the updraft and downdraft turbulence-induced motions in the atmosphere. Ultimately, we checked the capability of the new hybrid method to resolve the substantial parts of the turbulent motions. The turbulent energy transfer from the energy containing scales to inertial sub-range followed the well-known law. The capability of the new hybrid method in predicting the mean profile and the turbulent structure can be employed in the study of the effects of turbulent parameters on plume rise in different atmospheric stability classes.