An improved SST k−ω model for pollutant dispersion simulations within an isothermal boundary layer

• Published in: Journal of wind engineering and industrial aerodynamics 2018-08, Vol.179, p.369-384
• Main Authors: An, K., Fung, J.C.H.
• Format: Article
• Language: English
• Subjects: CFD simulations; Pollutant dispersion; Sensitivity study; Turbulence model; Wind tunnel measurements
• ISSN: 0167-6105; 1872-8197
• DOI: 10.1016/j.jweia.2018.06.010

The selection of turbulence models has always been one of the important aspects of computational fluid dynamics model simulations of wind flow and pollutant dispersion problems. The commonly industrially adopted steady Reynolds-averaged Navier–Stokes turbulence models are the realizable k−ε model and Menter's shear stress transport k−ω model. In this study, we demonstrated that the realizable k−ε model and Menter's SST k−ω model produced non-satisfactory correlation results with the experimental data, in terms of the normalized pollutant concentration for a cuboid obstacle. Here, an improved formulation of Menter's SST k−ω turbulence model, particularly for the wind flow and pollutant dispersion simulations, is applied. The modifications have been added in a systematic manner which include the addition of damping functions in the closure equations, the inclusion of the turbulent cross-diffusion term in the omega equation, and the re-establishment of model constants and parameter profiles. Considerable improvement in the correlation between the prediction and the experimental data up to a value of 0.78 with matching wind and turbulent kinetic energy profiles was obtained by the proposed model. Significant reduction of more than 80% in NSME and a satisfactory FAC2 value of 0.54 are also achieved with the improved model. •Less accurate wind & pollutant predictions are observed with RKE and SST k−ω models.•Improved formulation of SST k−ω model for wind & pollutant simulations is applied.•Promising wind and TKE profiles are obtained with the Improved SST k−ω model.•Improved model is capable to achieve correlation values up to 0.78 for concentration.•More than 80% reduction in NSME is obtained with the adoption of the improved model.