Numerical Modelling Of Turbidity Current And Retrogressive Erosion Processes In Heavily Sediment-laden Reservoirs

Turbidity currents and retrogressive erosion are typical processes of flow and sediment transport in reservoirs in heavily sediment-laden rivers, which are also important physical processes that influence the development of reservoir depositon and the morphological evolution of deposits. In regard to sediment management of reservoirs in heavily sediment-laden rivers, these two phenomena are often utilized as measures to maintain reservoir capacity. Some aspects on the physics of turbidity currents and retrogressive erosion are still not well understood, and the related numerical models are still unable to meet the requirements in practical applications. Turbidity current venting and drawdown flushing (with the creation of retrogressive erosion) are usually combined in reservoir operation. Flow exchanges between the main river and tributaries may also involve in the flushing and routing process, which makes the flow and sediment transport in the reservoir very complex and hard to simulate. Additionally, numerical modelling studies are rarely reported on the process of retrogressive erosion at the confluence of the main river and tributaries with strong two-dimensional characteristics. Therefore, it is important to develop mathematical models on turbidity currents and retrogressive erosion and carry out related studies based on numerical simulation, for a better planning of sediment management and optimization of reservoir operation in heavily sediment-laden rivers.This dissertation presents an integrated model coupling open-channel flow, turbidity current and flow exchanges between main river and tributaries. The model consists of two sets of governing equations for the open-channel flow and turbidity current, which are derived based on the generalized St. Venant equations by taking into account the effect of lateral flow exchanges. These two sets of equations are solved in the finite volume method framework with a Godnunov type scheme, and the solutions are executed in an alternating calculation mode. A plunge criterion is used to determine the location of the upstream boundary of the turbidity current, and to specify the corresponding boundary conditions. For the surface-gradient driven flow exchange, a storage cell method is used to update the water level at confluence and calculate the discharge of flow exchange. For the turbidity current intrusion, a formula is proposed for estimating the discharge of turbidity current intruding into a tributary with the effect of bed slope being considered. As indicated by the formula, the discharge of turbidity current intrusion increases as the depth and sediment concentration of the turbidity current at the confluence increase, and it decreases as the tributary bed steepen. The formula was verified with field measurements of turbidity current intrusion in the Xiaolangdi Reservoir.Turbidity current events in two laboratory experiments with different set-ups were used to test the capabilities of the proposed model. A field-scale application of the coupled model was conducted to simulate two turbidity current events occurring in the Sanmenxia Reservoir, and the method for calculating the limiting height of aspiration was adopted to estimate the outflow discharge after the turbidity currents arrived in front of the dam. The predicted plunge locations and propagating speed of the turbidity currents were in agreement with the measurements. Moreover, the calculated interface evolution processes and the sediment delivery ratios also agreed generally with the observed results. Therefore, the proposed model can help to select the design capacity of the outlets, and optimize the procedure for sediment release in reservoirs.The integrated model was applied to simulate two events of water-sediment regulation conducted in the Xiaolangdi Reservoir in 2004 and 2006. The reservoir drawdown process, turbidity current evolution and sediment venting efficiency predicted by the model were in close agreement with the measurements. The necessity to couple the flow exchanges was demonstrated by comparing the performance of the proposed model with different intrusion discharge formulas and the common one-dimensional model neglecting the main river and tributary interaction. The simulation results also imply that a lower reservoir pool level is prefered to enhance the sediment venting efficiency as it can shorten the trvalling distance of the turbidity current before it reaches the dam. And the flood from the upstream reservoir should be released in time to provide sufficient energy for the turbidity current. The analysis of the simulation results in the reach with delta deposits shows that the characteristics of the erosion are in accordance with the typical diagram of retrogressive erosion, and the simulated evolution process of the bed profile agrees with the theoretical solutions.A two-dimensional morphodynamic model is improved in this study. The TVD-MUSCL reconstruction is improved to eliminate the spurious flow between the wet and dry cells. The reconstructed bed elevation at cell interface is used to calculate the numerical depth at each side of the cell while implementing the DFB treatment of the bed slope source term. This improvement is proved to be able to guarantee the well-balance property of the model. The 2D model was verified against results from a flume experiment on retrogressive erosion.The improved 2D model was then applied to investigate the regtrogressive erosion at the confluence of the main river and the tributary Zhenshui in the Xiaolangdi Reservoir. The simulations were carried out with the reservoir topography at the current state and the state after continuouse trapping of sediment respectively. It is revealed that the difference of bed elevation of the main river and the tributary at the confluence determines the maximum scour depth that can be achieved by retrogressive erosion at the tributary mouth. Using the flood from the tributary is the most effective way to scour the deposits at the tributary mouth, while the method of reservoir drawdown is inferior.