Numerical Study On The Effects Of Bed Disturbances And Mass Exchanges On Sediment-laden Flows

Numerical simulation has become an indispensable normal tool for studying the process of sediment transport of flow.The numerical simulation of sediment-laden flow includes five elements:governing equations,initial conditions(such as initial bedform),boundary conditions,numerical methods and empirical relationships(such as empirical formulae of material exchange).The five elements contain their own uncertainties.This thesis focuses on the uncertainties of initial bedform disturbance,numerical methods and empirical relation of the sediment-laden flow simulation.From the open-channel flow to the turbidity current,the qualitative and quantitative analysis of the uncertainties existed in the model was carried out.The main research work and conclusions are as follows:(1)Natural topography of water flow is uneven,but most of the sediment-laden flow numerical simulations assume a smooth initial topography.It will be closer to the natural conditions if some initial disturbance was added to the bedform,but its numerical results to the impact are not yet known.This thesis first introduced the numerical simulation of the open channel flow on the irregular washable bed.The evolution of the sediment-laden flow on the bed with initial perturbation and the bed deformation were carried out.The simulation is carried out by coupled non-capacity modelling frameworks and calculated by the finite-volume method based on the structured grid to study the evolution of the sediment-laden flow on the bed with initial perturbation(a single hump or multiple irregular small humps on flat bed)and the deformation of the bed.This coupled model is compared with the decoupled model of Fagherazzi and Sun(2003)in simulating the formation of cyclic steps at the natural state.The results show that the open channel flow model has good feedback for the initial disturbance of the washable bed and can reproduce the cyclic steps well.However,the formation speed and the shape of cyclic steps are different between the coupled model and the decoupled model.Sediment-laden flow includes open channel flow and turbidity current.The similarity is that the governing equations are both hyperbolic partial differential equations.Its solution may have discontinuities or shocks.They need the TVD(Total Variation Diminishing)scheme to automatically capture shocks or discontinuities.The difficulty of this algorithm is to construct a scheme that perfectly reproduces the hydrostatic condition on complex topography(i.e.,well-balanced).The difference is that the effective gravity acceleration of the open channel is constant,but the effective gravity acceleration of the turbidity current varies with the change of the fluid sediment concentration.This paper explores whether the well-balanced algorithm for turbidity current is suitable for open channel flow.Take the reconstruction of cyclic steps simulation in Fagherazzi and Sun(2003)as an example.When the simulation model of the open-channel flow is reconstructed by the turbidity current coupled mathematical model,the simulation results are worse than that of the open-channel flow that is not reconstructed.The deposition of the first step upstream is more serious,and the steps transport to the upstream slower.This shows that it is better to use the open channel flow mathematical model directly.In addition to some of the empirical relation differences in the two sets of models,it is mostly due to the differences in numerical algorithms.The TVD scheme requires the dispersion of the source and pressure terms of the governing equations to achieve well-balanced.The gravity acceleration of the turbidity current is different from that of the open channel flow,so the well-balanced algorithm cannot be used universally.Therefore,it is necessary to study the well-balanced algorithm which is adapted to both the turbidity current and the open channel flow.(2)There are many references to the choice of material exchange empirical relation in the open channel flow simulation,but few references in the turbidity current simulation.The simulations of the typical sink experiment of turbidity current on complex bedform with numerical models were carried out to analyze the applicability and numerical uncertainties of empirical relationships for water and sediment entrainment.Different empirical formulae of material exchange between the turbidity current and its surroundings were organized.Four empirical formulae of water entrainment and five empirical formulae of sediment erosion are taken into account here.Lock-release and constant-flux turbidity currents are numerically reproduced here by a fully coupled layer-averaged model(TEM).The effects of empirical formulae on the sediment erosion and deposition are compared and the advantages and disadvantages of different empirical formulae are analyzed.The results simulated by ew01 are slightly better than those simulated by other empirical water entrainment formulae.Laboratory lock-release turbidity currents are mainly depositional.For constant-flux turbidity currents,the calculated and measured thickness of deposits are in better agreement when using Es 87 and Es93.The reason is probably that turbidity current data are used for the calibration of both Es87 and Es93.These results have some reference for the selection and optimization of the empirical formulae that used in numerical model to study the effects of turbidity current on the bed topography.Water flows on the bottom bed,leading to the bed deformation and playing an important role in deforming the bed.Multiple uncertainties exist in the process of numerical simulation of sediment-laden flow structure and sedimentation.There are many factors that influence these uncertainties.Such as mathematical model,algorithm construction,empirical relationship,bed topography and other boundary conditions.In this thesis,the above problems are explored and studied in order to reduce the uncertainties in the related numerical simulations.