Atmospheric dispersion of a heavier-than-air gas near a two-dimensional obstacle

Flow over a two-dimensional obstacle and dispersion of a heavier-than-air gas near the obstacle were studied. Two species, one representing air and the other representing the heavier-than-air gas were treated. Equations for mass and momentum were cast in mass-averaged form, with turbulent Reynolds stresses and mass fluxes modeled using eddy-viscosity and diffusivity hypotheses. A two-equation k- turbulence model was used to determine the effective turbulent viscosity. Streamline curvature and buoyancy corrections were added to the basic turbulence formulation. The model equations were solved using finite difference techniques. An alternating-direction-implicit (ADI) technique was used to solve the parabolic transport equations and a direct matrix solver was used to solve the elliptic pressure equation.Mesh sensitivities were investigated to determine the optimum mesh requirements for the final calculations. It was concluded that at least 10 grid spaces were required across the obstacle width and 15 across the obstacle height to obtain valid solutions. A non-uniform mesh was used to concentrate the grid points at the top of the obstacle.Experimental measurements were made with air flow over a 7.6 by 7.6 cm obstacle in a boundary-layer wind tunnel. Smoke visualization revealed a low-frequency oscillation of the bubble downstream of the obstacle. Hot-wire anemometer data are presented for the mean velocity and turbulent kinetic energy at the mid-plane of the obstacle and the mid-plane of the downstream recirculation bubble. A single hot-wire probe was found to be suitable for determining mean streamwise velocities with an accuracy of 11 %. The downstream recirculation bubble was unsteady and had a length range from 3 to 8 obstacle lengths.The experimental results for flow over the obstacle were compared with numerical calculations to validate the numerical solution procedure. A sensitivity study on the effect of curvature correction and variation of turbulence model constants on the numerical solution was conducted. Calculations that included the curvature correction model gave a downstream recirculation bubble length of 5.9 obstacle lengths while excluding the correction reduced this length to 4.4.In the second part of the study, numerical calculations were performed for the dispersion of a heavier-than-air gas in the vicinity of the two-dimensional obstacle. Characteristics of an adiabatic boundary layer were used in these calculations. The densities of the contaminant gases were 0, 25 and 50% greater than the air density. Calculations were performed with the contaminant injection source upstream and downstream of the obstacle.Use of the pressure gradient model reduced the size of the dense gas cloud by as much as 12%. The curvature correction model also affected the cloud expanse by reducing the effective turbulent viscosity in the downstream recirculation bubble. The location of the injection source had the largest impact on the cloud size. The area of the cloud within the 5 % contour was three times larger for downstream injection than for upstream injection.