Development and validation of a three-dimensional numerical model for application to river flow.

摘要

This thesis describes a computational fluid dynamics (CFD) model for general hydraulic flows in ducts, open-channels, and rivers. The work is motivated by the need for an accurate, efficient, and robust solver for typical river applications. The model employs the three-dimensional (3D), Reynolds-average Navier-Stokes (RANS) equations, in conjunction with a turbulence model, formulated in generalized curvilinear coordinates. An implicit time-marching scheme is used to solve the governing equations with the PISO algorithm for the velocity-pressure coupling. The convective terms are discretized using a second-order deferred correction scheme. The shear terms and pressure gradients are discretized using a second-order central difference scheme. A patched multiblock method is developed to handle complex geometry. Typical boundary conditions encountered in river flows are implemented.; A rigorous numerical verification procedure is followed to assess the accuracy of the numerical model. The model is used to calculate laminar flow through a 90° curved square duct and turbulent flow through an S-shaped open-channel. The computational efficiency and accuracy of the model are assessed. Good agreement between the model prediction and laboratory measurement is obtained.; This model is then used to investigate the physics of 90° open-channel junction flows. Experimental data of Shumate (1998) are used to validate the model performance. The comparisons indicate that the model captures the flow features observed in the experiments, including water surface profiles, separation zones, and secondary currents.; Application is further made to calculate the flow through a reach of Chattahoochee River near Holcomb Bridge. The region contains three bridge piers, an old water intake, a submerged rock weir, and six water intakes. The complex geometric features of this reach make it a challenging test case for a numerical model. The results are compared with velocity distributions at 0.6 water depth and water surface velocity vector field from a 1:16 laboratory model. Good agreements were obtained.; The present work advances the status of CFD methods for complex flows in hydraulic engineering. The success of two applications demonstrates the model's potentiality in simulating complex hydraulics problems, including river reaches.