عنوان مقاله [English]
Background and Objectives: Predicting the spread of pollutants is essential for rivers protection and management as well as human health and public safety perspectives. Longitudinal dispersion coefficient is conventionally predicted using costly field tracer tests, empirical equation or analytical formulation. However, the results obtained by these traditional methods are valid for only the reach examined or the flow and geometry condition under which the formula presented. In this study, an innovative method is introduced to predict the longitudinal dispersion coefficient in waterways using the digital image processing technique.
Materials and Methods: The experiments were carried out in a recirculating glass-walled laboratory flume of 18m length, 0.9m width and0.6m height with an asymmetric compound channel section. The ratio of flood plain width to main channel width was 1 and the overflow depth was 0.14 cm. Flow velocity measurements were taken using three-dimensional Acoustic Doppler Velcimeter (ADV). Three different tracers including color powder, food color and potassium permanganate solution were tested to find the most suitable tracer and. The tracer was injected using a half-tube filled with the dye solution and released uniformly and instantaneously across the flume width. The injection section was taken sufficiently far downstream of the start of the flume such that the flow was fully developed determined by measured velocity profiles. The spreading of the tracer cloud was recorded at three locations 4.00, 6.44 and 8.88 m downstream of the injection point using three digital video capturing Fujifilm JX420 cameras. Then, the captured videos were used to extract image sequences.
Results: Using the Beer-Lambert law of absorption, which correlates the absorbance to both, the concentrations of the attenuating dye as well as the thickness of the material sample, the depth-averaged concentration of the tracer across the flume width was determined. The longitudinal dispersion coefficient was calculated by the standard method of change of moments. The results showed that the image processing technique could be used as a reliable, accurate and economic method in studying the longitudinal dispersion coefficient. Three different tests were conducted with different relative depths including 0.15, 0.25 and 0.35 in the compound channel.
Conclusion: As the magnitude of the relative depth increases from 0.15 to 0.35, the non-dimensionalized longitudinal dispersion coefficient increases 65 and 56% in the main channel and floodplain, respectively. Finally, an equation is proposed to calculate the longitudinal dispersion coefficient in compound channels based on the experimental data. More researches are needed to extend this method to the field condition.