Doctor of Philosophy
School of Civil Mining and Environmental Engineering
Dissanayake, Kalyani, Experimental and numerical modelling of flow and sediment characteristics in open channel junctions, Doctor of Philosophy thesis, School of Civil Mining and Environmental Engineering, University of Wollongong, 2009. http://ro.uow.edu.au/theses/3555
Open channel confluences are present in many natural and man-made waterways. The dynamics of the flow in and around the junction are complex; in particular, immediately downstream of the junction, the flow develops a zone of separation on the inner wall, with accompanying secondary re-circulation patterns. The structure of this complex flow is a function of several parameters such as flow rates in both channels, angle of confluence, channel geometry including longitudinal slope and bed discordance, boundary roughness and intensity of turbulence and has a major influence on bed erosion, bank scouring, etc. If in addition, one or both streams are sediment-laden, the structure of the downstream flow becomes even more complex due to additional variables such as variation in sediment particle size and sediment concentration. This makes detailed experimental investigation of such flows very challenging.
In order to investigate the junction flow behavior, laboratory experiments and numerical simulations were performed in an equal-width, equal-depth and 90° flat bed open channel junction. Two separate computational codes PHOENICS and CFX were used for numerical simulations. Water heights, water velocities and sediment particle tracks were computed for different flow ratios and feed concentrations.
For investigating the junction flow behavior experimentally, a laboratory scale open channel junction was designed and constructed at the hydraulics laboratory of the University of Wollongong. Experiments were conducted for clean water and sediment laden flows with different flow ratios and feed concentrations. The downstream Froude number was kept constant (0.37) for all experiments. In sediment laden flow experiments, Corvic vinyl was introduced uniformly to the branch channel as sediment and then captured at the downstream end of the main channel to facilitate clean water flow through the main channel and sediment laden flow through branch channel. Water heights and turbidity were measured at different locations of the main channel utilizing point gauges and a custom made optical turbidity probe respectively.
Numerical predictions showed higher water levels upstream of the junction followed by a sudden drop of water levels immediately downstream of the junction. This phenomenon is accompanied by flow separation at the inner bank. Higher velocities were generated adjacent to the outer bank and velocities were diminished towards the inner bank. The separation zone length and width were diminished with increasing flow ratio q* (q*=main channel flow / total flow). Using a ‘body-fitted’ computational mesh, conforming to the shape of the free surface, and carrying out a ‘water-only’ simulation imposing free slip boundary condition for the free surface produced accurate velocity patterns near the bed and the free surface, showing a good agreement with experimental results.
In laboratory experiments higher sediment concentrations were observed adjacent to the inner wall immediately downstream of the junction, indicating particle deposition in the low-velocity separation region. It was observed that with increasing source sediment concentration from the branch channel, the turbidity downstream of the confluence increased while covering a larger area across the width of the main channel. Low sediment concentrations were observed upstream of the junction in all experiments. Higher turbidity gradients exist close to the junction whereas the turbidity gradients gradually diminish along the downstream of the main channel.
The sediment concentrations across the main channel were controlled by the location of the shear layer. This layer was moved towards inner wall with increasing discharge ratios showing higher sediment concentration adjacent to inner wall of the main channel. For lower discharge ratios q*=0.25 and q*=0.417, sediment particles were dispersed across the entire channel width of the main channel while in higher discharge ratios q*=0.583 and q*=0.75, flow from the main channel occupied most of the cross section and therefore branch channel sediment was confined to a small area adjacent to the inner wall. Similar scenario was observed in simulated particle paths as more particle tracks were shifted towards the outer wall direction for lower discharge ratios q*=0.25 and q*=0.417 than for higher discharge ratios q*=0.583 and q*=0.75. The shape factor of the separation zone (defined as the ratio of maximum separation zone width S w to separation zone length S L ) was found to vary between 0.12 to 0.15 for all experimental conditions tested. However the shape factor for clean water is found to be lower compared to sediment laden flow at all q* values.
The current study provides new data contributing to a better understanding of flow and sediment dynamics at channel junctions. The application of this new knowledge will lead to improved design of river bank protection works and urban flood and erosion control structures adjacent to the junction of branching channels.