Study On Key Technical Problems In Modeling Of Hyper-Concentrated Flows

By analyzing observed data, some key technical problems in modeling of hyper-concentrated flows are studied. These problems include the sediment- carrying capacity, integrated roughness coefficient, cross section deformation and recovery saturation coefficient, etc. The main contributions of this thesis are abstracted as follows.(1) The existing studies and observed data show that the important factors affecting the sediment-carrying capacity include the sediment concentration, flow velocity, temperature, median diameter and heterogeneous factor of sediment grains. The influencing effect of these factors on the sediment- carrying capacity is probably not the same if the sediment concentration in the flow is higher or lower than a certain value. This certain value is 70kg/m~3 or so, which is called as the critical sediment concentration. Besides, a model based on BP neural networks is developed to calculate the sediment-carrying capacity. The calculated results indicate that the model has a good ability to estimate the sediment-carrying capacity.(2) The integrated roughness coefficient of the sediment-laden flow has a strong relation with the Froude number and can be calculated if the Froude number is given. It also has a good relation with sediment concentration and exhibits increasing first and then decreasing with the increase of concentration. However, it has a bad relation with the flow discharge and water level. The main factors affecting the integrated roughness coefficient of the hyper- concentrated flow include the sediment concentration, bed material gradation and Froude number. By taking these three affecting factors as the input variables and the integrated roughness coefficient as the output variable, a model based on BP neural networks is developed. The applications show that the present model can be used to calculate the integrated roughness with a high accuracy.(3) Though the depth-averaged velocity has a direct proportion to the flow depth, it needs to be modified if it is used in a mathematical model. The lateral distribution of sediment concentration in a cross section can be calculated by Zhang Hongwu's formula. However, the coefficient in the formula is not a fixed value any more, but increases with the increase of the sediment concentration. The median diameter of sediment grains on a sampling point has a direct proportion to the sediment concentration.(4) The variations of the main channel area and width-to-depth ratio of a cross section are related not only with the flow and sediment condition but also with the boundary condition. The main channel area and channel morphology have a good relation with the flow and sediment condition if the boundary condition changes only in a slight extent. Therefore, the existing empirical formulae based on the flow and sediment condition should be cautiously used to simulate the change of river width or morphology. By analyzing field data and observations of physical models of the lower Weihe River and Sanmenxia reservoir, the allocation of the erosion or deposition materials along the wetted boundary of a cross section should adapt to the incoming flow and sediment as much as possible. In addition, the methods to modify the cross section and to simulate the widening of a cross section are put forward.(5) By assuming that the ratio of the bottom sediment concentration to the depth-averaged concentration is a function of the depth-averaged concentration, a new equation is deduced to express the channel bed variation. The new equation clearly shows that the recovery saturation coefficient is not the ratio of the bottom concentration to the depth-averaged concentration, but a comprehensive coefficient, and it could be far less than one.It is expected that the above research results could be used to improve the ability of the numerical models to simulate the hyper-concentrated flow.