Related resources
Search for item elsewhere
University researcher(s)
Modelling of Flood Waves Based on Wave Propagation Algorithms with Bed Efflux and Influx Including a Coupled Pipe Network Solver
[Thesis]. Manchester, UK: The University of Manchester; 2011.
Access to files
-
FULL-TEXT.PDF (pdf)
Layman's Abstract
Abstract of thesis submitted by: Hossein MahdizadehFor the degree of: Doctor of Philosophyand entitled: Modelling of Flood Waves Based on Wave Propagation Algorithms with Bed Efflux and Influx Including a Coupled Pipe Network Solver.Flood propagation over urban areas can cause an interaction between the free-surface flow and large underground pipe networks used for storm drainage and sewage, causing outflows and inflows at the bed. The associated waves may collide with each other and the surface waves. In this thesis the shallow water equations are used to model this type of wave interaction over dry or wet beds with bathymetry gradients and friction terms. The proposed shallow water scheme is solved based on finite volume high-resolution Godunov-type methods. The solver is well-balanced and can accurately balance the source terms and flux-gradients for the steady-state solutions. The solver also utilises a new type of Riemann wave speed to provide depth-positive results over nearly dry beds and dry states. Additionally a new type of source term is introduced in the continuity equation to model pipe inflow and outflow conditions at bed connections. For the standard one-dimensional shallow water equations the numerical results are validated with analytical solutions or other reference solutions provided in the literature. This includes the incipient Riemann problems for nearly dry and dry-states, steady flow over a hump in a rectangular channel and the wave propagation problem. Eventually, the generation of dry bed in the middle, over discontinuous topography is considered. Close agreement is achieved between the shallow water scheme and analytical or reference solutions for the above test cases. For the shallow water problems with influx/efflux source terms comparisons are made with STAR-CD, a commercial Navier-Stokes solver for general fluid flow prediction. The shallow water model is first used to simulate vertical flows through finite gaps in the bed. Next, the interaction of the vertical flows with a dam-break flow is considered for both dry and wet beds. An efflux number, En, is defined based on the vertical efflux velocity and the gap length. A parameter study is undertaken to investigate the effect of the one-dimensional approximation of the present model, for a range of non-dimensional efflux numbers. It is found that the shallow flow model gives sensible predictions at all times provided En<0.5, and for long durations for En>0.5. Dam break flow over an underground connecting pipe is also considered for the one-dimensional efflux problems.To solve two-dimensional problems the shallow water scheme uses the dimensional-splitting method which solves each one-dimensional Riemann problem in the x- and y-directions separately. The cross-derivative terms for second-order accuracy are incorporated by solving another Riemann problem in the orthogonal direction. For two-dimensional problems first the dam-break problems are considered over wet and dry beds. Then, flood propagation over complex terrain is demonstrated. Next, efflux discharge is modelled in isolation over a dry bed and then with dam-break interaction, comparing with STAR-CD results. Again very good agreement is shown between the two-dimensional shallow water model and STAR-CD for the efflux numbers of En<0.5.For modelling the inundation problem over an underground pipe network the solver is coupled with the general underground pipe network solver to calculate the efflux discharge as the flood waves pass through the pipe network. For analysing the pipe network with unknown effluxes an additional set of equations is incorporated into the solution of a general pipe network solver. The shallow water solver coupled to an underground pipe network is then used to simulate dam-break interaction with pipe networks with 9 and 25 nodes to demonstrate the versatility of the method.