Numerical Simulation of Non-Equilibrium Graded Sediment Transport

Li, Qiang (1995) Numerical Simulation of Non-Equilibrium Graded Sediment Transport. PhD thesis, University of Glasgow.

Abstract

A comprehensive one-dimensional mathematical model for simulating unsteady non-equilibrium graded sediment transport has been developed and verified with experimental and field data. The model framework is based on non-equilibrium sediment transport, involving the interaction between size fractions, separation simulation of suspended-load and bed load movement, and the exchange of particles between four different model layers. The implicit finite difference Preissmann scheme is used in the numerical model. This is known to be stable, flexible and robust. The two step operator splitting method, called the two point scheme, is employed to solve the advection-dispersion equation. A Newton-Raphson iteration method is used to linearise the highly nonlinear equation system. A fully coupled solution technique, called the double block sweep method, is adopted to reflect the strong physical interrelationship between flow and sediment transport components and to suppress computer errors and divergence of the numerical solution, a problem found in uncoupled or partly coupled methods. The numerical dissipation in the Preissmann scheme can be minimised by selecting the proper space and time weighting factors. In general the space weighting factors for all governing equations are centred. For short term simulation, such as flood events, the time weighting factor is taken as 0.55 to reduce numerical dissipation, for long term simulations a value of 1.0 is used. The model has been tested against standard benchmarks to check the stability, numerical dissipation and performance of the code. To solve the governing equations empirical sediment relationships must be employed. The main relation is the evaluation of the fractional sediment transport capacity. In this model van Rijn's sediment transport formulae developed for single-sized sediments have been used and modified for graded sediment using the concept of a hiding function. Two hiding functions for use with van Rijn's formulae were developed based on experimental data from HR Wallingford, United States Waterways Experimental Station and Gibbs & Neill. The first one was developed using Einstein's hiding function defmition which adjusts the Shields threshold condition for each size fraction. The second hiding function was developed using Parker's defmition of a reduced hiding function which adjusts the threshold condition for each size fraction based on the Shields value for the geometric mean size. In the formulations of the two hiding functions the significance of the Froude number has been assessed and accounted for. Two parameters for grain size distribution, mean size and standard deviation, were used to represent the effect of the bed material composition. The two hiding functions have been verified and compared by simulating the experiments of armour development and formation conducted in Aberdeen University. The results indicated that the reduced hiding function gives a satisfactory agreement between observed and calculated values and the hiding function overestimates the threshold conditions for the finer particles in the mixture. As the base data and optimisation technique are identical for each hiding function it is believed that the reason for the difference is related to the physical nature of graded sediment transport. The model has been used to simulate field investigations in Goodwin Creek, USA. Four transport events under flood condition were selected based on the availability of information. The model along with the empirical sediment relationships was therefore tested in a very active mobile bed river with graded sediment transport and a bimodal bed material. In order to compare the performance of van Rijn's formulae with a reduced hiding function, Parker's formula with his reduced hiding function was also used to simulate the same events. The effect of sediment inflows and initial bed material composition on the numerical results were also examined during the numerical simulations. The overall results indicated that real life simulation requires extensive data particularly for initial and boundary conditions. Parker's formula is sensitive to the resistance factor therefore the correct estimation of the resistance factor is a necessary condition for using Parker's formula The model has been applied in a real medium river system, the River Clyde in Scotland, for long term simulations of the river returning to regime following the cessation of dredging. The numerical results from this model have been compared with ones from a previous study in which the regime method was used. A good comparison of the results between these two studies was obtained. The effect of a new tidal weir, which is planned to be built, on the final regime condition and the river environment was also investigated.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Additional Information: Adviser: G Pender
• Keywords: Civil engineering, Geological engineering
• Date of Award: 1995
• Depositing User: Enlighten Team
• Unique ID: glathesis:1995-75734
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 19 Nov 2019 18:28
• Last Modified: 19 Nov 2019 18:28
• URI: https://theses.gla.ac.uk/id/eprint/75734