The study of open channel flow and dam breaking is not a new topic in computational fluid dynamics. However it has only recently started to gain significant attention from researchers using meshless methods, i.e. numerical modelling techniques which do not rely on the use of a mesh to discretise the domain. The research presented here is an attempt to use the meshless method known as smoothed particle hydrodynamics (SPH) to simulate the flow of water down a channel. Hydra, a pre-existing SPH code designed originally for astrophysical simulations, was converted to simulate water flow and then applied to the problem of dam bursting and flow over a weir. The conversion of the code to its new purpose was verified by simple code tests and then extensive validation was performed via the modelling of multiple dambreaks. The validation process can be split into three broad categories: 1) Comparison against the published data gained from other numerical methods both meshless and traditional. 2) Comparison against physical experiments performed by the author. 3) Comparison against independent experimental data found in the literature. Hydra in its newly converted form was satisfactorily applied to the majority of the tests presented to it and the same level of accuracy was achieved as with any of the other codes tested. A limit to the SPH method for performing this type of simulation was proposed based on particle number, smoothing length and initial conditions. A formula for the calculation of the number of ghost particles required to prevent spurious boundary pressures was also proposed. An analysis of various kernels used by different SPH researchers was presented and it was discovered that a relatively simple cubic spline kernel proved sufficient and that increasing complexity did not provide an increase in solution accuracy. The flow of water over a weir was presented next and results compared to published data which utilised a leading mesh based fluid simulation package. Results gained from Hydra simulations showed good downstream water level prediction but overestimation of upstream levels. A steady state solution was achieved within a similar timeframe compared to the grid based method. It was concluded that use of the SPH method and the Hydra code in particular can provide solutions to problems involving water flow down a channel and accuracy on the dambreak tests was equal to any rival codes/methods tested. However when the complexity of the boundaries involved in the model increased there was some evidence that the CFX simulation package could be used to achieve a more accurate solution than Hydra. Suggestions for continuation of research into Hydra as a water flow modelling code are presented in addition to recommendations for improving the experimental methods used.

Beach cusps are swash zone morphological patterns that have been of interest to many scientists and engineers. This study aims to improve understanding of the formation and long-term evolution of beach cusps by numerical simulation using a 2D process-based morphodynamic model, solving the coupled NLSW equations and sediment conservation equation simultaneously. A numerical implementation is applied building on the model of Dodd et al. (2008), which succeeds in simulating the occurrence of beach cusps. The numerical scheme improves the accuracy and stability of the swash zone computation. Results from a comparison between different numerical implementations concludes that the most suitable numerical scheme is the Roe-averaged scheme of Castro Diaz et al. (2008) with Minmod flux-limiter using the Harten and Hyman (1983) entropy fix method and the Hubbard and Dodd (2002) approach for the shoreline boundary condition. Before simulating the 2D beach cusps, the sensitivity of the model parameters and two different types of incoming waves are tested in the ID bed change. The sensitivity test results show that there is a convergence of the results when the minimum computational depth (dtol) ≤ 1 mm. Also the relationship between the bed profile and beach cusp parameters is that a greater maximum tip position (xs,max) is achieved, and more erosion in the tip region occurs when the bed friction coefficient (ƒw), the hydraulic conductivity (K), and dtol are smaller. On the other hand, the effect of scaling the sediment transport coefficient (A) is to scale the rate of change of the bed level, and appears not to lead to qualitative differences. Moreover, the incoming sine wave creates three components of 1D beach profile (long-shore bar, trough, and swash berm) in the computational domain, while the incoming sawtooth wave creates a wider equivalent region, because of wave simply breaking farther offshore. The 2D simulations give approximately the same beach cusp formation as those of Dodd et al. (2008); however, the geometrical parameters (ƒ) from the self-organisation theory are still high when compared with previous field observation and numerical simulations. The evolution of the beach cusps is investigated by Fourier and global analyses (Garnier et al., 2006), and can be divided into three stages: 1D development in the cross-shore profile, 2D small cusp spacing pattern, and 2D final bigger cusp spacing pattern, caused by the coalescing of two small bays and subsequent rearrangement to an equal spacing. However, an unphysical behaviour is found during cusp evolution, which is the reversing behaviour between horn and embayment. It appears that this reversible behaviour is caused by deposition at the embayment head, created from high infiltration and the usage of the velocity-only type of sediment transport equation in the model.

The critical conditions for incipient sediment motion induced by oscillating-grid generated turbulence interacting with a sloping sediment layer were investigated experimentally. Near-spherical monodisperse sediments were used throughout with relative densities of 1.2 and 2.5 and mean diameters(d)ranging between 80 and 1087 μm. Interaction characteristics were analysed in terms of the critical Shields parameter θc, defined using the peak root mean square (r.m.s) horizontal velocity component in the near-bed region. Bed slope effects on θc were investigated by tilting the bed (and the grid) at angles between 0 and the repose limit. In all cases, the grid was aligned to be parallel to the bed surface, so that the oscillation direction is always normal to bed surface. The measured values of θc on a horizontal bed were comparable to the values reported in the traditional Shields diagram with θc seen to increase monotonically for hydraulically smooth bedforms and to be approximately constant for hydraulically rough bedforms. To account for bed slope effects, the measured values of θc were compared with a force-balance model based on the conditions for incipient grain motion on a sloping bed. For hydraulically smooth bedforms, where the bed roughness is small compared to the boundary-layer depth, the model was derived to account for how viscous stresses act to damp the drag and lift forces acting on the near surface sediment. For hydraulically rough bedforms, where this viscous-damping effect is not present, the model assumes the standard approach with the drag and lift forces scaling with the square of the near-bed (inviscid)velocity scale. In both cases the model predicts the bedforms to become more mobile as the bed slope is increased. However, the damping effect of the viscous sublayer acts as a stabilizing influenced for hydraulically smooth bedforms, to reduce the rate at which the bed mobility increases with bed slope. The measured values of θc in the hydraulically rough bedforms were in agreement with the trends predicted by this model. However, measured θc in smooth bed cases were lower than predicted, and fall on the hydraulically rough trend when bed slope is < 20. When the bed slope reaches the repose limit, θc falls between the smooth-bed and rough-bed cases. Measurements of sediment trajectories due to the turbulence interacting with the bed were obtained, for a range of impact conditions. Observations of the sediment trajectories during the interaction show the individual sediment grains to be predominantly displaced in a circular 'splash'. Data showed that the ‘splash’ feature and particle entrainment within the turbulence structure was within one eddy turnover time.

Large scale patterns in the seabed often occur in the nearshore zone of sandy beaches. A widely occurring bed pattern is the crescentic bar. These bed patterns develop under moderate wave conditions, and form a lunate shaped alongshore pattern in front of a coast. Over recent years, knowledge concerning the development, occurrence, and characteristics of these bed patterns has been significantly expanded through field studies and modelling attempts. An example of such a model is the linear stability analysis, which describes the initial development of crescentic bed patterns along an undisturbed beach. To date, comparisons between field measurements and modelling results have been general in nature. The purpose of this research is to investigate whether a linear stability analysis, which is useful for understanding the physics of emerging bed-forms, can be used to make quantitative predictions in the field. To this end a morphodynamical linear stability model (Morfo60, [Calvete et al., 2005]) is used to describe the development of crescentic bed patterns at the coast at the USACE Field Research Facility in Duck, North Carolina, USA. Wave, tide and bathymetry data recorded at Duck over a two month period in 1998 are used to model the development of these morphodynamical patterns. The model predictions are compared with field observations made at Duck, over the same two month period, reported by van Enckevort et al. [2004]. A direct comparison shows that predicted length scales of crescentic bed patterns are similar to those observed. However, the model predictions show more fluctuations than are observed in the field. This is because the model describes the development of crescentic bed patterns starting from an alongshore constant bed, whereas in reality bed patterns already exist in most situations. An algorithm is developed to overcome these fluctuations and identifies the more physically significant model predictions based on large growth rates and consistency in length scales. The moments at which physically significant model predictions occur correspond better with field observations than the original model predictions. The effects of pre-existing bed-forms on the development of crescentic bed patterns are investigated using a non-linear model (Morfo55, [Garnier, 2006]). Results show that pre-existing bed patterns can have significant effects, however, the finally dominant length scale, the linear growth and decay rates, and the migration rate can be accurately described by a linear stability model. Pre-existing length scales that exhibit significant linear growth will remain and undergo further development, whereas length scales that are outside the linear growth rate curve decay and give rise to a bed pattern with a bigger linear growth rate. The conclusions drawn from the research concerning pre-existing bed patterns are applied to improve predictions linear stability model. This results in considerable improvements in the comparison of model predictions with field observations, for certain periods of time.

An integrated horizontally two- and fully three-dimensional numerical model system has been developed based on a combined unstructured and σ-coordinate grid to simulate the flow and water quality process in large water bodies with a focus on the three dimensional behaviours at specific areas. The model is based on the time dependent Reynolds-Averaged Navier-Stokes equations with a non-hydrostatic pressure distribution and a baroclinic force being incorporated in the three dimensional (3D) model. The two sub models interact dynamically during the solution procedure with no time-step restriction due to integration. The main idea is to use a fractional step algorithm for each model and then integrate the two models fraction by fraction. Hybrid 2D-3D finite volume cells have been introduced for the link nodes which are partly in the 2D domain and partly in the 3D domain. Thus an interpolation/averaging procedure at the interface and domain overlapping is no longer needed. The 3D model uses the projection method for pressure calculation. The advection equation is solved by the semi-Lagrangian method. Other components are solved via the finite element - finite volume (FV) method. The water surface is determined implicitly through a global matrix equation created by assembling the domain's matrices. The cell integrals are calculated analytically to eliminate a common source of numerical diffusion due to the use of approximation techniques for the FV integrals. The horizontal gradients of the density and shear stresses are calculated on true horizontal planes, in order to avoid artificial velocity and diffusion in highly stratified flows. Neumann interpolation elements with virtual nodes have been introduced at Neumann type of boundaries for more accuracy. The integrated model has been verified using analytical solutions and benchmark test cases, including the Ekman velocity distribution, wind driven circulation, lock exchange and integrated 2D-3D flows in basin. The results show the model is capable of the model for accurate simulation and implicit 2D-3D integration. Keywords: integrated modelling, hydrodynamic numerical model, non-hydrostatic, unstructured mesh, hybrid finite element finite volume method.

A dam break event is considered, taking place over a uniform sediment bed. Understanding and modelling the erosion that occurs when the fluid behind the dam collapses at release has important applications in coastline morphodynamics / beach erosion modelling. A new coupled two dimensional Navier-Stokes solver and sediment transport model is presented with novel methods for dealing with non-converging solutions to the Navier-Stokes equations and a new adaptation to the Youngs [1982] volume-of-fluid reconstruction scheme. The implementation of a sediment transport model includes a new method for accounting for mass conservation for the transition of sediment between bed and flow as well as a novel method for accounting for the redistribution of material associated with the maintenance of the critical angle of repose or slope limit. The model is validated and then applied to a dam break simulation for various backwater and tailwater conditions. Classical experimental realizations of dam break events have involved the rapid removal of a barrier in a flume [e.g. [Levin, 1952; Dressler, 1954; Bell et al., 1992])]. However, early-time flow analysis encounters two problems with this method. Firstly, the removal of the barrier creates a strong vortex sheet on the face of the static fluid immediately behind the barrier that is not present in either the idealized problem, or the motivating environmental problems. Secondly, the removal of the barrier cannot take place instantaneously and so a brief jet-like flow is initially induced through the opening between the base of the barrier and the sediment layer. We partially circumvent these difficulties with the classical experiments by implementing a novel dam break barrier release, using a barrier similar to the wrapped-fabric design of [Dalziel, 1993], which minimizes the initial vortical disturbance. Three-dimensional stereoscopic Particle Image Velocimetry (PIV) measurements allow us not only to capture the velocity field in the laser-plane, but also perpendicular to it. These planar experimental results are compared to the results of the numerical study and the comparison is shown to be good while the simulation successfully converges.

No description available.

The lattice Boltzmann method (LBM) has become an effective numerical technique for computational fluid dynamics (CFD) in recent years. It has many advantages over the conventional computational methods like finite element and finite difference methods. The method is characterised by simplicity, easy treatment of boundary conditions and parallel feature in programming that makes it ideal for solving large-scale real-life problems. This thesis presents the development and applications of a lattice Boltzmann model for both steady and unsteady two-dimensional axisymmetric flows. The axisymmetric flows are described by three-dimensional (3D) Navier-Stokes equations, which can be solved by three-dimensional (3D) lattice Boltzmann method. If cylindrical coordinates are applied, such 3D equations become 2D axisymmetric flow equations. However, they cannot be solved using the 2D standard LBM. In order to study more complicated axisymmetric flow problems by 2D LBM, in this thesis, firstly, the revised axisymmetric lattice Boltzmann D2Q9 model (AxLAB®) is applied and tested for some benchmark for axisymmetric laminar flows and more complicated flows including 3D Womersley flow and forced axisymmetric cold-flow jets, and flows with swirl such as the cylindrical cavity flows and the swirling flow in a closed cylinder with rotating top and bottom. Secondly, the AxLAB® is extended to simulate turbulent flows and non-Newtonian fluid flows. A well-known power-law scheme is incorporated into the AxLAB® to simulate the non-Newtonian fluid flow: the Taylor Couette flows for Newtonian and non-Newtonian fluids are simulated and compared. The combined effects of the Reynolds number, the radius ratio, and the power-law index on the flow characteristics are analysed and compared with other literatures. All the numerical results are also compared with the existing numerical results or experimental data reported in the literature to demonstrate the accuracy of the model. Thirdly, a further developed AxLAB® is presented to simulate the turbulent flows. The turbulent flow is efficiently and naturally simulated through incorporation of the standard subgrid-scale stress (SGS) model into the axisymmetric lattice Boltzmann equation in a consistent manner with the lattice gas dynamics. The model is verified by applying it to several typical cases in engineering: (i) pipe flow through an abrupt axisymmetric constriction, (ii) axisymmetric separated and reattached flow and (iii) pulsatile flows in a stenotic vessel. All the numerical results obtained using the present methods are compared with experimental data and other available numerical solutions, indicating good agreements. This shows that the improved AxLAB® is simple and is able to predict axisymmetric turbulent, non-Newtonian complicated flows at good accuracy.

Typescript (photocopy). / Includes bibliographical references (leaves 77-78).

[pt] presente dissertação descreve o desenvolvimento de um equipamento destinado a determinar a condutividade hidráulica de meios porosos saturados in situ. O permeâmetro consiste em uma sonda que emprega o método da vazão constante. Após sua inserção pelo modo de cravação. Uma bomba de seringa instalada na superfície do terreno é utilizada para aplicar uma vazão constante enquanto que a carga hidráulica induzida no meio é medida através de um transdutor de pressão piezoresistivo instalado no corpo da sonda. O equipamento permite o escoamento das linhas de drenagem e de medição de carga hidráulica a partir da superfície possibilitando a saturação do meio poroso após a cravação da sonda bem como minimiza as incertezas associadas à medição de pressão. Esta última é alcançada através de um transdutor diferencial de pressão com uma faixa de trabalho de 10kPa que possibilita medições de poropressões bem próximas a zona de injeção. Esta característica permite a realização de ensaios num tempo curto e minimiza o problema de compatibilidade de fluidos decorrente do processo de injeção. / [en] This thesis describes a piece of equipment developed to determine the hydraulic conductivity of saturated porous media. The permeameter consists of a pushed in probe and employs the constant flow rate method. A syringe pump installed on the surface is used to develop a constant rate of flow whereas a piezoelectric transducer installed in the probes body measures the induced change in hydraulic head. The great innovation in this device consists on its ability of saturating all of its drainage lines allowing the media to be saturated and minimizing the errors of pressure measurements. Pressure measurement is carried out using a 10kPa differential pressure transducer that enables porepressure to be measured close to the injection zone. This characteristic enables shorter tests to be carried out and minimize problems associated to fluid compability