A dissertation submitted to the Faculty of Engineering and the Built Environment, University of the Witwatersrand, Johannesburg, in fulfilment of requirements for the degree of Master of Science in Engineering Johannesburg, 2017 / South Africa is a water scarce country and dams play a large role in the infrastructure of our country by providing water for many purposes. With the growth of the country, new dams are needed and existing dams require rehabilitation. Often, increasing a spillways capacity forms part of the rehabilitation required at dams. Therefore, one of the main aims of this study is to look at an option for increasing a spillways discharge capacity. The labyrinth and Piano key weirs (PKW) were investigated together with a combination of the two spillway types. Different geometric attributes were combined and varied to develop a new design that would assist in improving discharge capacity while reducing the cost, producing an economically viable option. Thirty five physical models were built and tested in this regard. Comparison was made with the standard PKW design in terms of discharge capacity, hydraulic efficiency, length ratio (L/W) and cost related to concrete volume required. It was concluded that efficiency, discharge capacity and cost can be improved with this new combined design however more geometric variations need to be studied to ascertain these results. / MT 2018

The use of surface piercing propellers (SPPs) shows promise for high speed operation by virtually eliminating appendage drag, which can be as much as 30 percent of the total drag on a vehicle at high speeds. The scarcity of available systematic test data has made reliable performance prediction difficult. The primary objective of this research is to obtain experimental performance prediction data that can be used in SPP design. In a series of open water tests in a non-pressurized towing tank facility, force transducer measurements were taken at tip immersion ratios from 0.5 to .33, yaw angles from 0° to 30° and inclination angles from 0° to 15° over a range of advance ratios from 0.8 to 1.8. Force transducer measurements were taken for thrust, torque, side forces and moments. These results will help develop a baseline for the verification of SPP performance prediction. / by Justin M. Lorio. / Thesis (M.S.C.S.)--Florida Atlantic University, 2010. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2010. Mode of access: World Wide Web.

Bottom of the mountainous rivers is generally composed of natural gravels. Flow depth in such rivers is generally shallow, with the ratio of water depth to size of bed materials (known as relative submergence) rarely higher than 20. In this type of flow, gravels intrusion induces significant spatial variation of the flow characteristics near bed region, which is known as roughness layer. The simultaneous effects of natural gravels and water surface cause formation of complicated flow structure which is to some extent different from the flow with high relative submergence (flow with relative submergence higher than 40). Despite abundance of studies in shallow flows, there are only a limited number of studies concerning spatial organization of near bed flow field for such type of flow, with also contradictory results. The spatial organization of near bed turbulent flow characteristics is also important for transport of fine sediment. Transport of fine sediments is generally correlated to the asymmetry of vertical velocity. Asymmetry of vertical velocity also arises from a quasi-cyclic process of upward motion of low-velocity fluid parcels (ejection) and downward motion of high-velocity parcels (sweep), together known as bursting process. Spatial organization of bursting process and asymmetry of vertical velocity in near bed and respect to bed topography has not been inscribed properly. In heterogeneous flows, the use of spatially averaged turbulent transport equations, known as Double Averaged Navier-Stikes equations (DANS), is common. In DANS equations viscous drag, form drag and correlation of spatial fluctuation of time averaged velocities (known as form induced stresses) are explicitly expressed. Despite prevailing usage of DANS equations in study of gravel bed flow, examination of vertical velocity has not been performed appropriately by applying double averaging method. Also, the role of form induced stresses in vertical momentum flux has not been highlighted. In present thesis, Stereoscopic Particle Image Velocimetry at near bed horizontal layer and Digital Particle Image Velocimetry in vertical planes are employed together with laser scanning of bed elevations to study flow field and turbulence structure over a coarse immobile gravel bed in submergence conditions ranges from 5 to 10. Spatial organization of flow characteristics at the near bed region is analyzed respect to bed topography. This analysis is also composed of spatial distribution of bursting process and vertical momentum flux. Moreover, vertical profiles of double averaged turbulent flow characteristics and form induced stresses with different relative submergences are compared. Results show that near bed flow field is characterized by a strip structure induced by secondary currents. Such structure tends to be disrupted by the effect of gravel protrusions. To better analyze the interaction between the flow field and gravel bed protrusions, cross-correlations of different velocity components and bed elevations in a horizontal layer just above gravel crests are computed. These results show that upward and downward flows occur not randomly on the bed, but in correspondence to upstream and downstream side of gravels. Also, turbulent momentum flux is directed downward in the downstream side of gravel crests and it is directed upward in upstream side of gravel crests. This is due to prevalence of ejection and sweep events respectively in upstream and downstream sides of gravel crests. These results are in agreement with formation of separation and reattachment zones around gravel crests. Moreover, spatial distribution of sweep and ejection events are organized in streamwise elongated strips with high and low values which are consistent with presence of secondary currents cells. Results obtained by double averaging method show that relative submergence affects the normalwise double averaged turbulence intensity profiles all along the flow depth, while only a weak effect, limited to the near bed region, is noticed on streamwise double averaged turbulence intensity profiles. Logarithmic law parameterization of double averaged velocity profiles shows that parameters change considerably with relative submergence and, in some cases, no clear log-law region was found. These results challenge application of log-law in such type of flow. Analysis of the vertical velocity shows that far from the bed, vertical turbulence momentum flux is upward, while below gravel crests it is downward. This behavior is resulted by prevalence of ejection events far from the bed and sweep events below gravel crests. Results show that vertical momentum flux resulted by form induced component is not significant, except below gravel crests which are upward in to the water column. A limited number of qualitative observations in the real case of fine sediments presence in the matrix of rough bed is in agreement with the results of turbulent flow characteristics. Sand ribbons are clearly formed due to secondary currents. Also, fine materials are mostly deposited and eroded respectively in downstream and upstream sides of gravel crests. The results of present study show that in general some regions actively participate in transport, while the other regions do not participate in the transport. From this basis, Rouse criterion has been developed by considering spatial variation of vertical momentum flux.

Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Ocean Engineering, 1977. / Microfiche copy available in Archives and Engineering. / Includes bibliographical references. / M.S.

No description available.

The present research investigates nonlinear barge motions through analyses of coupled multi-degree-of-freedom (MDOF) deterministic and stochastic models. Roll-Heave-Sway and other lower-ordered models are developed to predict the nonlinear motions and analyze the stability of a class of ship-to-shore cargo barges. The governing equations of motion contain coupled rigid body Roll-Heave-Sway relations, hydrostatic and hydrodynamic terms. The rigid body relationships are a part of the general six-degree-of-freedom model. Hydrostatic terms include effects of the barge's sharp edge and of relative Roll-Heave states. Hydrodynamic terms are in a "Morison" form. The characteristics of the excitation wave field are based on linear wave theory. Predictive capabilities of the Roll-Heave-Sway and the Roll-Heave models are investigated. System parameters are calibrated to match experimental test results using several regular wave test cases. Potential theory predictions provide initial estimates of several key system parameters. With the identified system parameters, numerical predictions obtained from time domain simulations of both models are compared with experimental test results for a random wave case, and compared to each other to investigate the coupling effects of sway on roll and heave motions. Reliability against capsizing of a barge in random seas is investigated using stochastic analysis techniques. With the Markov process assumption, the barge response density to random waves is derived as a solution to the corresponding Fokker-Planck equation. The path integral solution technique is employed to obtain numerical solutions for the Roll-Heave and the Roll models. A quasi-2DOF model is introduced to improve the accuracy of the 1DOF Roll model. The reliability of a barge in a variety of sea conditions is analyzed as a first passage problem using the quasi-2DOF model. Mean times to reach specified capsizing probabilities for a barge operating in sea states 1 through 9 are obtained. / Graduation date: 2003

The main idea behind the work presented in this thesis is to investigate if it is possible to find a mechanism that leads to surface magnetic field concentrations and could operate under solar conditions without postulating the presence of magnetic flux tubes rising from the bottom of the convection zone, a commonly used yet physically problematic approach. In this context we study the ‘negative effective magnetic pressure effect’: it was pointed out in earlier work (Kleeorin et al., 1989) that the presence of a weak magnetic field can lead to a reduction of the mean turbulent pressure on large length scales. This reduction is now indeed clearly observed in simulations. As magnetic fluctuations experience an unstable feedback through this effect, it leads, in a stratified medium, to the formation of magnetic structures, first observed numerically in the fifth paper of this thesis. While our setup is relatively simple, one wonders if this instability, as a mechanism able to concentrate magnetic fields in the near surface layers, may play a role in the formation of sunspots, starting from a weak dynamo-generated field throughout the convection zone rather than from strong flux tubes stored at the bottom. A generalization of the studied case is ongoing. / <p>At the time of the the doctoral defence the following paper was unpublished and had a status as follows: Paper nr 7: Submitted</p>

Droplet splatter dynamics is an important study in the field of forensics since a crime event can produce many blood stains. Understanding the origins of the blood stains from pure observations is very difficult because much of the information about the impact is lost. A theoretical model is therefore needed to better understand the dynamics of droplet impact and splatter. We chose to explore a fluid modelling method known as Smoothed Particle Hydrodynamics (SPH) to determine whether it is capable of modelling droplet splatter accurately. Specifically, we chose to investigate an SPH version of a non-Newtonian pressure correction method with surface tension. Three experiments were performed to analyze the different aspects of SPH. From the results of the experiments, we concluded that this method can produce stable simulations if an artificial viscosity model is included, a third-order polynomial kernel is used and the pressure boundary condition on surface particles are non-zero. / UOIT

Magnetohydrodynamics (MHD) studies the dynamics of an electrically conducting fluid under the influence of a magnetic field. Many astrophysical phenomena are related to MHD, and computer simulations are used to model these dynamics. In this thesis, we conduct MHD simulations of non-radiative black hole accretion as well as fast magnetic reconnection. By performing large scale three dimensional parallel MHD simulations on supercomputers and using a deformed-mesh algorithm, we were able to conduct very high dynamical range simulations of black hole accretion of Sgr A* at the Galactic Center. We find a generic set of solutions, and make specific predictions for currently feasible observations of rotation measure (RM). The magnetized accretion flow is subsonic and lacks outward convection flux, making the accretion rate very small and having a density slope of around $-1$. There is no tendency for the flows to become rotationally supported, and the slow time variability of the RM is a key quantitative signature of this accretion flow. We also provide a constructive numerical example of fast magnetic reconnection in a three-dimensional periodic box. Reconnection is initiated by a strong, localized perturbation to the field lines and the solution is intrinsically three-dimensional. Approximately $30\%$ of the magnetic energy is released in an event which lasts about one Alfv\'en time, but only after a delay during which the field lines evolve into a critical configuration. In the co-moving frame of the reconnection regions, reconnection occurs through an X-like point, analogous to the Petschek reconnection. The dynamics appear to be driven by global flows rather than local processes. In addition to issues pertaining to physics, we present results on the acceleration of MHD simulations using heterogeneous computing systems \cite. We have implemented the MHD code on a variety of heterogeneous and multi-core architectures (multi-core x86, Cell, Nvidia and ATI GPU) using different languages (FORTRAN, C, Cell, CUDA and OpenCL). Initial performance results for these systems are presented, and we conclude that substantial gains in performance over traditional systems are possible. In particular, it is possible to extract a greater percentage of peak theoretical performance from some heterogeneous systems when compared to x86 architectures.

Magnetohydrodynamics (MHD) studies the dynamics of an electrically conducting fluid under the influence of a magnetic field. Many astrophysical phenomena are related to MHD, and computer simulations are used to model these dynamics. In this thesis, we conduct MHD simulations of non-radiative black hole accretion as well as fast magnetic reconnection. By performing large scale three dimensional parallel MHD simulations on supercomputers and using a deformed-mesh algorithm, we were able to conduct very high dynamical range simulations of black hole accretion of Sgr A* at the Galactic Center. We find a generic set of solutions, and make specific predictions for currently feasible observations of rotation measure (RM). The magnetized accretion flow is subsonic and lacks outward convection flux, making the accretion rate very small and having a density slope of around $-1$. There is no tendency for the flows to become rotationally supported, and the slow time variability of the RM is a key quantitative signature of this accretion flow. We also provide a constructive numerical example of fast magnetic reconnection in a three-dimensional periodic box. Reconnection is initiated by a strong, localized perturbation to the field lines and the solution is intrinsically three-dimensional. Approximately $30\%$ of the magnetic energy is released in an event which lasts about one Alfv\'en time, but only after a delay during which the field lines evolve into a critical configuration. In the co-moving frame of the reconnection regions, reconnection occurs through an X-like point, analogous to the Petschek reconnection. The dynamics appear to be driven by global flows rather than local processes. In addition to issues pertaining to physics, we present results on the acceleration of MHD simulations using heterogeneous computing systems \cite. We have implemented the MHD code on a variety of heterogeneous and multi-core architectures (multi-core x86, Cell, Nvidia and ATI GPU) using different languages (FORTRAN, C, Cell, CUDA and OpenCL). Initial performance results for these systems are presented, and we conclude that substantial gains in performance over traditional systems are possible. In particular, it is possible to extract a greater percentage of peak theoretical performance from some heterogeneous systems when compared to x86 architectures.