No description available.

Underwater explosive loading resulting from rapid phase transitions (RPT's) is one example of a fast transient dynamic event. The leakage of substances such as LPG or LNG when stored underwater can cause an RPT. These substances are often stored in a combination of very low temperatures and high pressures with respect to the surrounding fluid (seawater) and their leakage can cause the equivalent of an underwater explosion. Such containers are usually found to be part of a much larger 'storage field' of containers. An RPT occurring in one container will cause underwater explosive loading on neighbouring containers. By simulating an RPT using explosive charges experiments were initially designed using theoretical and empirical techniques. The fluid and structural response of a prototype container subject to symmetric and axisymmetric underwater explosive (UNDEX) loading was then examined experimentally. Theoretical predictions using the finite element hydrocode LS-DYNA and boundary element code USA-DYNA3D were undertaken and compared with experimental observations. Several non-destructive techniques were employed to estimate dynamic collapse buckling criteria from both experimental and theoretical results. The experimental work concluded that the critical regions of the prototype container were the apex and the base under both forms of loading. The quality of the numerical predictions varied dependent on the form of the loading. In some cases the fluid and structural responses were overpredicted, and in others underpredicted. Within the limitations of these numerical procedures it was possible to predict a conservative estimate of a critical charge size under axisymmetric UNDEX loading using LS-DYNA. The critical stand off distance was also estimated from experimental results under symmetric UNDEX loading. The use of numerical approaches to predict fluid-structure interaction as successful for the shock phase of an underwater loading and both LS-DYNA and USA-DYnA3D have been validated for shock loading. Bubble loading simulations proved unsuccessful. Suggested improvements are proposed to increase the application of, and enhance the reliability of, the techniques used in this work.

This thesis is mainly concerned with the study of flow and pressure in silos during filling and discharging. It describes the numerical and experimental studies undertaken to investigate the effects of filling modes and inserts on the flow and pressure in silos. The numerical investigations were performed using two different finite element codes: SILO from Sweden and the commercial Abaqus Code. Finite Element (FE) models were developed to simulate a progressive filling process, and used to investigate the development of filling pressures along the walls of a steep hopper and a shallow hopper. Two filling modes were simulated: a concentric filling mode and a distributed filling mode. The FE results obtained were compared with the predictions from classic theories, and for the case of the shallow hopper, also with the results measured in large scale experiments. The investigation of the effect of inserts on silo flow using the FE code SILO shows that the insert converts the discharge pattern from funnel flow to mass flow in a silo with a modest inclination hopper. Parametric study further exhibited that a combination of a careful positioning of the insert and a reduction of wall friction between the solid and hopper walls would improve the silo discharge pattern in a silo with a shallower hopper. The residence times of tracers were measured to identify the flow pattern in the experiments with a double cone insert and compared with the numerical predictions.

When solid particles are suspended in the fluid and are not in a jammed state, a fruitful approach to modelling the system can be to describe it as a system of particles interacting both with each other and with an external field. In the specific case when the particles are far enough apart, the dominant interactions between particles are those mediated by the surrounding fluid rather than direct particle-particle interactions, possibly only when the particles are touching. One of the most important phenomena observed in this regime is particle roping – rather than being evenly dispersed throughout the fluid, particles congregate in one or more ‘ropes’ aligned with the flow direction. This can be a serious problem in coal fired power stations, which require coal dust to be evenly distributed to operate at maximum efficiency. This thesis presents a basic numerical study of particle-fluid-particle interactions under conditions characteristic of the roping phenomenon found after bends in the pneumatic transport systems of coal fired power plants. The main objectives of this work are to: 1. Obtain a pair potential hydrodynamic force field from computational fluid dynamics (CFD) simulations of two fixed spherical particles at low particle Reynolds number; 2. Estimate the magnitude of errors introduced by the pair potential approximation by comparing the two fixed spherical particles results with CFD simulations of systems of three fixed spherical particles; and 3. Use many-particle Monte Carlo simulations to investigate the conditions under which clustering or roping occurs.

The work contained within the body of this thesis is concerned with the isothermal aerodynamic study of multiple jet systems with special reference to flame interactions. The type of jets used were jets with or without recirculation (swirl) and simple flame interactions were shown to occur for the configurations studied. The main purpose of the study was to show the nature and degree of the aerodynamic interference and to relate these to the factors which govern flame length, stability and combustion intensity. The thesis also contains a study of the turbulence and mean flow characteristics of a swirling jet of variable swirl. The instrument used to measure the mean velocity and turbulence quantities was the hot wire anemometer, and since for the case of a strongly swirling jet each component of the mean velocity and the normal and shear stresses are significant, a technique was evolved capable of separating out each of the 9 individual terms associated with the above, namely ¯u,¯( v), ¯w, ¯(u'v'), ¯(u'w'), ¯(v'w'), ¯(〖u'〗^2 ), ¯(〖v'〗^2 ), ¯(〖w'〗^2 ). The method of analysis is based upon a new velocity voltage relationship which is accurate for all types of probe and/or flow velocity range. Previous methods of analysis were restricted to low turbulence levels where typically the local turbulence intensity (¯(〖u'〗^2 ) ½ / ¯u, ), is not more than 20%. The new method is shown to be valid until the onset of flow reversal which depends upon the type of waveform that the fluctuations in velocity take. The magnitude of the measured quantities are shown to be independent of the type of waveform and assuming that the fluctuations closely resemble a triangular waveform the maximum turbulence levels measurable are 57%.

No description available.

No description available.

No description available.

No description available.

In general, two phase flow through safety relief valves (SRV) is a complex flow and dominated by heat, momentum and mass transfer between phases. It is even more complex for flows in a safety valve where turbulence, flow separation and shock waves associated with compressible flow occur. Unfortunately, the literature indicates that there is no generic model with sufficient accuracy to predict the two phase flow conditions in an SRV. However, existing CFD techniques do provide the potential for predicting the two phase flow in the valve with the possibility of improving the valve design. In this study, two phase flow through an SRV has been investigated experimentally and computationally using a two dimensional mixture model of the two phase flow. Investigations have been carried out on a conventional spring loaded relief valve using a two phase mixture of air and water. The mixture model has been used with the standard k- e turbulence model to predict the two phase flow through the SRV. Quasi steady flow has been assumed to investigate the valve flow-lift and force-lift characteristics. Tests have been carried out for a range of pressures (7- 12 barg) and water mass fraction (0-0.71) and compared with the CFD predicted results. Comparisons have also been made with the ISO 4126-10 sizing model for safety valves. The CFD predictions compare well with the experimental data with an accuracy better than the ISO 4126-10 sizing model. Scaling parameters for force and flow have been presented to assist SRV design under single and two phase flow conditions.