Numerical study and comparison with experiment of dispersion of a heavier-than-air gas in a simulated neutral atmospheric boundary layer

This paper presents a Reynolds-averaged Navier–Stokes simulation of the dispersion of a heavier-than-air gas from a ground level line source in a simulated atmospheric boundary layer. A previously published experimental study has been used to define the computational domain and boundary conditions, as well as to compare with the predicted results. The dispersed material is a mixture of 97% carbon-dioxide and 3% propane by concentration, where the latter gas was used as a tracer in the experiments. The floor of the computational domain was populated with vertical fences in order to simulate a rough surface for boundary layer development, as in the experiments. This also helped in obtaining streamwise homogeneity for mean velocity and turbulence kinetic energy. The results and comparisons with the experimental data are presented for concentration profiles as well as a number of derived parameters, such as entrainment velocity. The cases presented are for three Richardson numbers of 0.1, 7 and 16. Sensitivity tests are carried out to show the effects of boundary conditions at the inlet to the flow domain, turbulence model, namely, the standard k–ε model and RNG k–ε model, and the turbulent Schmidt number. The results showed significant sensitivity to the value of turbulent Schmidt number. By optimizing the value of this parameter, it was possible to obtain close comparisons between the predicted and measured parameters.

► The study presents a well-defined test case, with supporting experimental data, for CFD modeling. ► Including roughness elements in the simulation helped in obtaining longitudinal homogeneity. ► The results for species concentration are sensitive to the value of turbulent Schmidt number. ► The value of turbulent Schmidt number is a function of Richardson number.