In terms of cost and maintenance the polyphase induction motor has advantages compared with its direct current counterpart, but the control of the output torque and speed is more involved. The advent of high power thyristors has allowed a new appraisal of the induction motor for variable speed drives (e.g. winches, cranes, and traction), and some of the well established, but inefficient methods (in terms of cost, control, and power consumption) have been given a new look. This thesis compares some of these methods of control, namely variable frequency, stator voltage, and rotor control and suggests a combination of stator voltage and rotor control to increase the versatility of the machine and to extend the torque/speed envelope of operation without the complexity of a fall inverter system. It is shown that the harmonics produced by a square wave inverter output can be reduced by increasing the number of pulses per half cycle output if the individual pulses have a given relationship. The influence of this harmonic reduction in the output of the inverter fed induction motor is investigated. An equivalent circuit is developed to represent the induction motor when fed with voltage waveforms of high harmonic content. This equivalent circuit is solved using a finite difference technique and is shown to give first order results with a variety of voltage waveforms. First order performance predictions are shown to be possible of an induction motor system, stator fed from a three phase alternating current supply via "back-to-back" pairs of thyristors using a simple chart method. The chart in this method gives a "derated" value of current available from the production of output torque, based on the uncontrolled theoretical performance of the motor and the given firing angle of the thyristors and the power factor of the motor at the given speed. The prediction of the performance of an induction motor with a thyristor direct current chopper in the rotor circuits in both the motoring and braking regions is achieved using the concept of an effective rotor resistance. The procedure for the calculation of this resistance from given parameters of known values is given, and the theoretical performance compared with that of an experimental system.

Despite large amounts of numerical and theoretical work on the finite element method one application, that of shells, is still not adequately understood. Shell elements are significantly less efficient than elements for other applications. Not surprisingly, the curvature function is the primary contributor to the inefficiencies and therefore most of the thesis is concerned with it. Numerous curved beam formulations are studied for an initial numerical assessment of the problem. From an examination of the results and element strain expressions it is possible to identify how curvatures are responsible for the poor accuracy. More rigorous mathematical investigations follow and finally methods of improving the performance of elements with curvature are considered.

Since the early days of infrarenal aortic reconstruction, renal failure has remained an important cause of post-operative morbidity and mortality. Biochemical testing of renal function following aortic surgery showed evidence of post-operative renal impairment in 47 to 60 per cent of patients. Biochemical testing of renal function is inaccurate. This study was undertaken to examine the immediate post-operative effects of aortic surgery on renal function and also the long term effects six months after surgery using radionuclide tests; which show minor changes of renal function not detectable biochemically and test the function of individual kidneys. Glomerular filtration rate (GFR) was measured using 51Cr labelled EDTA clearance. Effective renal plasma flow (ERPF) was measured using 125I labelled Hippuran clearance. The individual kidney function was assessed using a 99Tc labelled DMSA renal scanning and 99Tc labelled DTPA renography. Two age matched groups groups of patients were used as controls, 1. Patients with arteriopathy who were treated conservatively. 2. Patients undergoing major colonic resection. The control groups allowed comparison of those patients undergoing aortic surgery with a similar group of patients who suffered from the same disease but were treated without surgery. The second control group of patients undergoing major colonic surgery allowed comparison with those patients having aortic surgery. The radionuclide tests of renal function were carried out pre-operatively, 10 to 14 days post-operatively and six months later on all aortic patients. The control patients had the tests done twice. The results indicate that the changes in renal function are specific to aortic surgery. 40 per cent of patients when examined at six months showed a significant deterioration of their renal function. The effects of the operative parameters on the renal function are also discussed.

Containment buildings are required to protect the public from the effect of radioactivity released by an accident to the nuclear power plant. they must be designed to minimize leakage through cracks in building wall and so methods are needed which can predict the geometry of the cracks including their widths. This study provides a realistic model of bond action between deformed reinforcing bar and concrete which uses interface element based on measured bond characteristics combined with radial constraints to simulate wedging action. It is shown to be capable of accurately predicting stress, strain, and crack geometry in typical reinforced concrete components for a variety of loading conditions. Of particular note is the ability of the model to simulate the deterioration in bond brought about by prior application of cyclic load of high amplitude. Similitude requirements for physical models of reinforced concrete structures are examined in the light of available experimental results and predicted values. The need to correctly scale the fracture energy to ensure proper representation of the bond characteristics is demonstrated. The study ends with an examination of the suitability of higher order elements which adopt a softening characteristic to represent the effect of a primary crack and its use in analysis of a typical nuclear reactor containment building.

The first part of this Ph.D. thesis describes the significance and basis of QT dispersion in the light of recent clinical findings. A system is then described which was developed for the analysis of QT dispersion using standard twelve lead Electrocardiogram (ECG) paper records. In the first part of the system, ECG records are scanned and stored on computer as '.PCX' files. These files are pre-processed to remove random noise using a set of heuristic rules. An image processing technique known as 'thinning' is used to retrieve the lineal structure of the ECG waveforms. Digital data values are then extracted from these 'thinned' waveforms. The data recovery system was validated by a series of tests one of which showed that the cross-correlation coefficient between original and digitised data was greater than 0.99. A graphical interface incorporating a patient data base was designed to allow for the display and user-interactive measurement (using a cursor) of the recovered waveforms. The next part of the thesis describes the design of an 'automatic algorithm' for the detection of characteristic points of the ECG including the QRS onset, R wave peak and T wave end positions. A transformed signal which is the three-point averaged derivative of the ECG is used to detect significant areas of QRS activity. Peaks corresponding to the R waves are detected and the ECG waveforms are then segmented as a function of heart rate. A normalised threshold is found and used to detect the QRS onset position. T wave peak and end detection is carried out by the study of "area maps" in the transformed signal. Once area maps corresponding to the T wave are determined, the T wave end is detected by using a three point moving average window and a threshold which takes into account the T wave slope. Manual, user-interactive and automatic measurements revealed that (i) the user-interactive system measurements were easily performed by cardiologists and this technique resulted in more accurate and reproducible measurements than the standard manual method; and (ii) values of QT and RR intervals obtained by the automatic algorithm agreed well with measurements made by ten physicians. The final part of the study attempted to find a suitable parameter to characterise QT dispersion. The importance of QT dispersion was then investigated by undertaking a clinical study, whose results supported the value of QT dispersion as a risk indicator of patients prone to ventricular arrhythmias. Use of the automatic algorithm in finding QT dispersion gave the best discrimination of patient groups.

The work described in this thesis was undertaken to obtain a unified treatment to the controller size reduction problem in advanced robust control system design. A common feature in state-space solutions to advanced control system design, such as parametrizations of all stabilizing controllers and Hinfinity suboptimal controllers, is that a free parameter matrix is contained in the parametrization to give the designer freedom in designing the required controllers. However, this free parameter can provide unnecessarily high order controllers. This thesis presents a new methodology for controller size reduction. The methodology utilizes the parametrization of all stabilizing controllers and Hinfinity suboptimal controllers, and then generates a set of low-order stabilizing controllers and a set of low-order Hinfinity suboptimal controllers, respectively. The central idea is to achieve a low-order realization of a full-order controller, by deriving and solving two simultaneous matrix equations in order to eliminate unobservable states. Orthogonal canonical forms are employed to solve these simultaneous equations. A consequence of the algorithms employed is that the order of the controller is reduced from n + nq (or n + no) to nq (or no), where n is the order of the weighted plant and nq (or no) is the order of the free parameter. In design applications, a possible solution to the problem of combining the objectives of robust stability and performance requirements is to use a loop shaping design procedure based on normalized coprime factor plant descriptions. The methodology obtained for low-order Hinfinity suboptimal controllers is extended, with slight modifications, to one and two degree-of-freedom loop shaping design procedures. The results are illustrated by numerical examples. Finally, a practical industrial problem of designing a low-order controller for a tetrahedral robot is considered by applying the methodology developed in the thesis.

Over the last ten years a consensus of the proposed mechanisms of water treeing has started to emerge. A brief review of the current state of understanding is presented as a composite theory which involves chemical, physical and electro-physical effects. This theory treats water tree growth as the movement of a hydrophilic-hydrophobic interface into the polymer. Water-filled microvoids are formed with chemically-stable hydrophilic inner surfaces connected by free-volume routes also converted to have a hydrophilic nature and weakened by chain scission. Solvated ions may travel through such routes by diffusion as the pathways open and close due to electro-kinenc effects. It is concluded from this review that there is very little data on the dynamics of water tree inception. Consequently results are presented which concentrate on the inception behaviour rather than the growth. At high-frequencies or for very long trees, field-assisted diffusion may dominate and the growth rate of long trees may accelerate. Such long trees may also shield smaller trees underneath them and slow their growth thereby producing a bimodal distribution of lengths.

Scuffing is a severe form of surface damage which limits the performance of lubricated sliding machine components. It has no generally accepted criterion, although recent advances in theory and experiment suggest failure is history dependent. The theory concerns components that operate in nominal line contact and have surface roughness with a lay parallel to sliding. However, the predictions are independent of sliding speed, the observations consistent with micro-ehl could be attributable to other temperature dependent effects, some variations in contact frequency are difficult to interpret, and running-in remains unpredictable. In the present work, the capabilities of an existing twin disc machine were significantly enhanced. It was found that temperature dependent effects other than micro-ehl were relatively minor. Changes in contact frequency were associated with transitions between ehl, mixed lubrication and micro-ehl, but were dependent on the presence of a distinguishable "mainscale" wavelength in the surface roughness. A prior surface smoothing process of running-in delayed the onset of scuffing, and surface blackening accompanied running-in in mixed lubrication. Surfaces machined to have a dominant mainscale wavelength without secondary features showed little change. To investigate the effects of surface speeds on scuffing, operating conditions were restricted to test histories that induced mixed lubrication. Two regimes of behaviour were identified within a sliding speed/rolling speed domain. At higher sliding speeds, discs scuffed close to transition from ehl to mixed lubrication. At lower speeds, they ran-in and did not scuff until relatively severe, micro-ehl conditions were imposed. A theoretical heat transfer model was matched to the boundary between the two regimes. Refinement of the model suggested the criterion for scuffing in mixed lubrication to be that the load and speeds allow asperities sufficient time within the conjunction to experience a temperature rise large enough to cause local welding.

This thesis considers the problem of developing sliding mode output tracking controllers for uncertain systems when output information alone is available. Two different approaches to controller design are proposed. The first approach is an observer based scheme which utilizes integral action. A new framework is proposed for the design of a class of sliding mode observers. The attainment of a canonical form, which is central to the framework, is a necessary and sufficient condition for the existence of a class of observers insensitive to matched uncertainty. The results supersede previous work in this area which necessitated checking the validity of a structural constraint between the state space matrices of the system. A formal analysis is undertaken of the combined plant/observer dynamics obtained when using a sliding mode control law incorporating integral action. It is demonstrated that the combined system is quadratically stable, despite the presence of a class of bounded matched uncertainty. Furthermore, the control law and the observer system can be designed independently; in other words, the 'separation principle' for linear systems also holds for this class of uncertain system and controller/observer pair. The second approach considers an output feedback stabilization problem where the class of hyperplanes and control laws considered is restricted to those which require only output information. The analysis is performed in essentially the same framework as that developed for the sliding mode observer. It enables the class of systems considered by other workers in this field to be extended and provides a practical realizable controller which requires no additional assumptions. These results are used in a model-reference framework to obtain a tracking controller. Successful attempts to implement these new theoretical ideas on an experimental furnace at the Gas Research Centre at Loughborough are documented. A single-input single-output sliding mode scheme to control temperature is described. Details are given of a more ambitious, multivariable scheme to regulate both temperature and excess oxygen by the manipulation of the fuel and air flows.

A computer simulation of the ion drift problem in air has been developed to predict the potential, electric field and space charge density in various geometries applicable to wire-duct electrostatic precipitators. The development of the design of the model is explained and its implementation in a fourth generation language is described. The generality of the overall design allows new features to be included. The model uses existing analytical expressions derived using the variational functionals for Poisson's equation and the Galerkin residual for current continuity which are solved iteratively. The resulting system of simultaneous equations are represented in a matrix formulation. Routines in the program: construct the matrices; set the boundary conditions; identify 'Known' and 'unknown' node parameters; iteratively solve the system of equations; test for convergence and output the solution. The Galerkin residual using a field dependent mobility has been derived and a one dimensional simulation for cylindrical geometry has been developed. Predictions are compared for linear relationships between field magnitude and ion mobility. The feasibility of extending this to other relationships is considered. The design and implementation of a three dimensional model with constant mobility is described. The grid can be fitted to a variety of geometries, with plane and curved electrodes, using a functional for equipotential surfaces. No further adaptations to the model are required to make predictions for these different geometries. Charge injection is modelled using a fixed field injection law. Predictions for plane-plane and cylindrical geometries are compared to analytical solutions. The need to improve the efficiency of the program is identified. Possible options to achieve this are discussed so that the model may be developed as a design tool for novel discharge electrodes in electrostatic precipitators.