This thesis describes the development of optimised fluorescent dye-doped tracer droplets for gas phase particle image Velocimetry (PIV), to study mixing in multi-phase flows. The use of these tracer droplets in applications where flare can be an obstacle to obtaining velocity flow data is also demonstrated. In PIV, micron-sized tracer particles are normally required to accurately follow the flow while in the same time providing optimum fluorescence signal for proper image capture. Thus, there is a requirement to identify dyes with high quantum yield that can be dissolved in suitable nebulisable solvents at high concentrations and to investigate the effect of high concentration on fluorescence properties, such as fluorescence concentration quenching effects that could lead to a decrease of the fluorescence signal from tracer droplets. The selection criteria of candidate dyes and the study of their fluorescence properties by steady-state spectrofluorometry are presented. Bis-MSB and DCM were identified to be the optimum blue and red emitting dyes and to offer high solubility in <i>o</i>-xylene and DMSO respectively. A novel experimental approach employing stabilised emulsions to emulate the fluorescence properties of micron-sized tracer droplets has been developed. The development of a single-colour-camera PIV system that can image micron-sized and spectrally distinct fluorescent tracers in a two-phase flow is reported. The use of dye-doped microemulsions in a novel micro-PIV seeding methodology for full field velocity measurements in microfluidic devices is presented. This approach gives improved particle image contrast and reduced motion parallax uncertainty, when compared to conventional solid seed particles. Results of micro-PIV measurements in T- and Y-junction microfluidic chips are presented. Channel velocity profiles were found to agree with CFD simulations.

A discussion is given of numerical techniques for modelling fluid flows with free surfaces, covering the basic principles behind the schemes and their respective advantages and disadvantages. This leads to the adoption of a two dimensional "Volume of Fluid" (VOF) method for the simulation of wave impacts on vertical walls. The VOF method is described in detail, including enhancements to various sections of the algorithm. The "water piston" theory of air pockets is given, together with a scheme to incorporate the effects into the model. The results of a test problem are given which illustrates the ability of the VOF method to simulate free surface flows, and another test case validates the approach to air pocket modelling. Finally a series of comparisons with experimental results of wave impacts is given; these include time series of surface elevations and wall pressures. The results illustrate the difficulty in making direct comparisons with experiment due to the problem of ensuring identical wave conditions and show that, given this constraint, the VOF model is capable of accurate simulation of impact events.

The aim of this study was to examine a fully turbulent, particle laden flow in a square duct. Velocities were measured by particle image velocimetry, using both direct beam and fibre optic illumination systems. Small bubbles of diameter approximately 2<i>μm </i>(density~1.2<i>K g m</i>^-3) were used as seeding, and three sizes of large glass particles, with diameters of 59±51<i>μm, </i>173±53<i>μm, </i>and<i> </i>471±120<i>μm </i>(density=2500<i>K g m</i>^-3), were introduced into the flow. Measurements were obtained for all four particle sizes within five parallel planes, and from these results mean flow profiles, turbulent stresses and spectra, and concentration profiles were all examined. The results compared well with the simulations and experiments of other authors, and where such comparison was not possible some interesting trends were observed. In fully developed turbulent flow in a square pipe there are eight spanwise circulation cells, which move fluid away from the walls along the midwall bisectors and then back into the corners. It was possible to accurately estimate how much the spanwise velocity of the glass particles would lag that of the fluid, using empirical data and an approximate form of Stokes law. The glass particles were also found to have a relatively slow streamwise velocity in the centre of the circulation cells.

No description available.

In this study, four calibration experiments were conducted for the DEM validation: filling of a model silo, confined compression in a cylinder, rod penetration into a granular bulk and silo discharging through an outlet. DEM simulations were conducted using two commercial software: PFC3D and EDEM, the latter being the DEM software that originated from Edinburgh University. A set of 8 benchmark tests were carried out to validate the codes and to evaluate the fundamental aspects of DEM. Following that, a large number of DEM simulations were conducted and comparisons between DEM simulations and experiments were made. For corn grains: the study shows that 4-sphere representation together with the measured corn properties produced satisfactory match with experiments for silo filling (normal wall pressure distribution), confined compression (normal wall pressure distribution, load transfer to boundary surfaces, and silo design parameter <i>K</i> and <i>m</i>), rod penetration (force-displacement response) and silo discharge (mass flow rate and angle of repose). For glass beads: the DEM simulations also gave good agreement with experiments for silo filling (normal wall pressure distribution), confined compression (normal wall pressure distribution, load transfer to boundary surfaces), rod penetration (force-displacement response) and silo discharge (mass flow rate). These findings provide sound verification that DEM is capable of producing quantitative predictions of the problems studied. They also suggest that very accurate representation of the non-spherical particle shape may not be necessary to produce satisfactory predictions and capturing the linear dimensions of a particle may be adequate. Two DEM results that produced larger discrepancies with experiments are filling density (~17% lower for corn grains and ~8% lower for glass beads) and loading stiffness (stiffer response). Plausible explanations for these are given in this thesis, which should be explored further.

No description available.

The experimental study was conducted to explore the flow behaviour of particulate solids in silos and to provide the experimental data for the theoretical work. Silo discharge tests using polypropylene pellets in a half cylindrical model silo were conducted under different discharge eccentricities. In addition, some experimental data from a research project on a full scale silo at the University of Edinburgh were also analysed and used in this thesis. In these tests, the main observations relating to flow were the residence times of the markers seeded in the solid during filling (residence time is the time a particle takes to move from its initial position to the outlet). The top surface profile was also measured at regular intervals during discharge. The residence time measurements are analysed using several approaches to infer the approximate location of the flow channel boundary in both silos. These techniques demonstrate how residence time measurements can be used to infer the flow patterns in silo discharge. In addition to the above, the transparent front wall in the half cylindrical model silo enabled a record to be made of the development of the flow channel boundary and the trajectories of individual particles. This data provided further insight into the complex flow patterns. The direct measurements of the changing flow channel boundary are also used in the thesis to evaluate the dilation of the solid which occurred in the flowing material. The dilation, or decrease in the bulk density, is believed to play a significant role in determining the solids flow pattern. The theoretical work presented starts with the development of a simple engineering model to predict solids flow in funnel flow silos. This model uses the kinematic theory for steady state flow, subject to simple assumptions concerning the top surface profile, top surface flow geometry and particle trajectory. The flow of iron ore pellets in the full scale silo and polypropylene pellets in the model silo during concentric emptying are predicted using the simple engineering model. The predictions for both silos are plausible and are in good agreement with the experimental observations.

The effects of drag-reducing polymer additives on oscillatory fluid flow, particularly the damping of small-amplitude gas bubble pulsation and liquid column oscillation, were investigated. The damping of bubble pulsation was investigated both analytically and experimentally. The analytical investigation showed that the elasticity of the viscoelastic liquids opposed the viscous effects on the damping of bubble pulsation. On the experimental side, an apparatus was designed to produce gas bubbles of uniform size. The performance of the apparatus was satisfactory. The experimental results of the effects of Polyox additives on the damping of bubble pulsation were in agreement with the theoretical analysis. However, the results of the effects of the Separan additives would suggest the presence of anomalous extensional viscosity effects on the damping of bubble pulsation. The effects of the polymer additives on bubble formation and sound radiation were investigated. The effects of the polymer additives on the damping of liquid column oscillation were investigated experimentally. It was found that the steady-state rheological properties of the visco-elastic liquids can adequately describe their liquid column oscillatory flow. The absence of the effects of the solutions elasticity is attributed to the smallness of their relaxation time in comparison with the time period of the liquid column oscillation.

Experimental investigations have been carried out on hollow glass micro-spheres with average diameter 60μm in two dimensional rectangular air-fluidized beds. The particle bulk and effective densities were 110kg/m³ and 200 kg/m³ respectively. Fluidized bed expansion and pressure drop across the bed were studied. Despite theoretically belonging to Group C (cohesive, difficult to be fluidized) according to Geldart’s powder classification, these microspheres fluidized easily and the beds expansion was extremely high when the gas velocity was increased above the minimum fluidization value. Using diffusing-wave spectroscopy (DWS), granular temperatures were determined throughout the gas fluidised bed for superficial velocities up to and somewhat beyond the minimum bubbling velocity, which was more than 4 times the minimum fluidisation velocity. The granular temperature was found to increase smoothly with superficial velocity between the minimum fluidisation and minimum bubbling velocities, contrary to the results of earlier workers who also used DWS, but consistent with the results of other workers using less direct methods. It was found that the granular temperature at a set superficial velocity increased with height above the distributor whilst the pressure gradient remained constant. This, combined with increasing laser light transmission, suggested particle clustering with the cluster size decreasing with height. Granular temperature was found to vary symmetrically about the bed centreline where it was also maximal, corresponding to velocity profiles observed by others for similar bed geometries. Preliminary studies of particle microdynamics in a laboratory scale circulating fluidized bed were also made.

This thesis presents a renormalization group method to tackle the problem of reducing the number of degrees of freedom necessary to describe fluid turbulence. Attempts to apply renormalization group methods to fluid turbulence have been in general based on an existing formulation of a multivariate normal model of turbulence. This model is an appropriate zero-order model for perturbation theories of turbulence in the context of the moment-closure problem because of the Gaussian distribution of the zero-older field for which moments of all orders can be expressed in terms of pair correlations. However, this is not a relevant attribute for renormalization group methods. A critical review of renormalization group methods based on the multivariate normal model is presented. An alternative approach was developed by McComb and Watt [Phys. Rev. A, 46, 4797 (1992)], who introduced a formal conditional average. The rescaling two-field theory leads to the derivation of a recursion relation, which eliminates finite blocks of turbulent velocity modes while maintaining the form invariance of the dynamical equation. In the present thesis, it is shown that the theory was heuristic in some respects and there were inadequate explanations in the procedure of that theory. A new formulation of renormalization group method, based on an alternative interpretation of the two-field theory, is presented here. A new zero-order model field is proposed for the Fourier modes in which there is no coupling between the high-wavenumber band of modes being eliminated and the remaining low-wavenumber modes. This model has the property that high-wavenumber modes can be eliminated without the need for a conditional average. The model field is then made the basis of a formal perturbation series which recovers the results of the two-field theory in a way which eliminates certain ambiguities, and allows one to see clear relationship between a turbulent velocity field and the zero order model field. The results are the systematic derivations of an equation for the high wavenumber modes that exhibits form invariance under the renormalization group transformation, and an expression for the effective viscosity for use in the computational simulation of homogeneous and isotropic turbulence. A value for Kolmogorov constant of α = 1:6 is obtained, when the fixed point for the effective viscosity is numerically calculated.