The mechanics of shear-driven flushing of a dense fluid from a canyon is investigated through a series of laboratory experiments. Such a flow has many environmental applications including the removal of dense pollutants trapped in urban canyons and the purging of salt water from bed depressions at river mouths. While there is an extensive literature on flow in canyons and cavities for the neutrally buoyant case, the problem of flushing a dense fluid from canyon has received considerably less attention. To understand the dynamics of the mixing and to quantify the buoyant contaminant flushing rate, a broad range of experimental results using multiple techniques to quantify the flushing rate are presented.;First the effect of the fluid density in the canyon, which is parameterized in terms of the flow Richardson number, is examined. For a square canyon a total of 26 experiments were conducted for Richardson numbers ranging from 0.08 to 4.5. The effect of the buoyancy on the in-canyon flow structure is examined and the regime diagram showing the flow observed in different ranges of Richardson number is presented. Also the flushing time and decay rate of the dense fluid is quantified. Three different mean stratifications are observed, namely well-mixed, continuously stratified, and two-layer. Increasing the Richardson number decreases the rate of flushing from the cavity. Curve fits through the measured buoyancy over time were used to calculate the flushing rate.;The effect of canyon aspect ratio on the flushing of a dense fluid is considered through four series of finite release experiments for different aspect ratios (η=0.45, 0.75, 1, 2). A total of 64 experiments are conducted over a range of Richardson numbers. The effect of the canyon width on the observed flow regime, in-canyon vortex structure and buoyancy stratification is demonstrated. Empirical equations for the initial decay rate are also determined for all aspect ratios and their results are compared with the existing data in the literature. For high Richardson numbers which is a two-layer stratification, the lower dense layer in the stratification acts as a soft bottom to the canyon, alters the effective aspect ratio of the canyon and changes the flow dynamics. Narrower canyons were found to flush more slowly as the large scale vortex that drives most of the mixing is restricted laterally and, therefore, does not penetrate deep into the canyon.;An experimental investigation of the effect of the upstream surface roughness is also provided in order to highlight the impact of the upstream boundary properties on the flushing. This is the first such systematic investigation of the role of turbulence intensity on flushing of fluid form a canyon. Three series of tests for different upstream surface roughness (d84 = 1.14, 0.83 and 0 cm) have been conducted. It is shown that the initial purging of dense fluid is driven by the mean flow. However, at later stages, the turbulence within the flow is the dominant cause of mixing.;Finally an analytical model is developed to describe both continuous and finite release flows. A series of continuous release experiments were run in order to measure the exact rate of mixing for a broad range of source buoyancies and layer depths. The results of this two-layer model coupled with a revised equation for non-two layer flow regimes shows a good agreement with the all the experimental data presented in this study. It is shown that all the mixing rate data from the finite release and continuous release experiments collapse onto a single line when parameterized in terms of a layer Richardson number.
展开▼
