Until recently, the only realistic form of treatment available to patients in end stage heart failure was transplantation. In the last few years, the possibility of diverting skeletal muscle from its normal function to perform a cardiac assist role has emerged as a potential alternative to transplant surgery. The introduction of an Skeletal Muscle Ventricle (SMV) to the circulation is a potential long-term hazard, as the patient's blood comes into contact with the non-endothelialised surfaces of the wall of the new ventricle and the connecting conduits. This may trigger a cascade of events leading to deposition of thrombus, whose formation is dependent on the nature of the blood flow. The potential problem of haemostasis may arise in the apex of the artificial ventricle, where little mixing and large residence times may occur. There is therefore a strong need for carrying out flow analysis studies to address in detail the questions of haemostasis and thrombogenesis and in this context to evaluate possible candidate SMV configurations. Research on the dynamics of the flow inside model SMVs has been carried out on physical and numerical models with the objective of aialysing the effect of the size and shape of the ventricle and inlet/outlet orientation of the duct. Due to the physiological limit on the power available to pump the blood out of the ventricle, the efficiency of these potential assistance devices has to be maximized. It is also necessary to minimize the risks of haemolysis and thrombogenesis, which are both related, in different ways, to the level of shear stress on the wall and within the flow. A common feature of these flows is the formation of vortex rings. Vortices enhance mixing, and this is a useful process to encourage in an SMV, as it could assist in the mixing of the blood components and in the reduction of apical residence time. Being able to predict accurately the dynamics of the vortices is therefore important, as this will affect the prediction of residence times and shear stresses at the wall and within the flow. It is also very important to know whether numerical codes can predict vortex ring dynamics from both qualitative and quantitative points of view. In order to study the dynamics of the formation of these vortices, first, mathematical models were studied. The general purpose CFDS-FLOW3D code was used in all numerical simulations. Initial investigations of this research project concerned a progressive validation of the numerical solution predicted by the code when the domain where the flow is calculated had moving boundaries. Firstly, comparisons were made with the analytical solution for expanding/contracting pipes. An adapted compliant SMV model was then generated with a truncated apex using sinusoidally prescribed motion of the wall. With this model, two vortex rings could be predicted as in the experiments. The spherical-end model also gave good agreement with experimental flow patterns (ludicello et al., 1994). Frequency-dependent studies were carried out over the range of cardiac values using single- and multi-block versions of the code. A further validation exercise involved the use of sigmoidal filling curves in the in vitro models (Shortland et a!., 1994). Experimental data provided by such studies were used to drive the wall motion in the numerical simulations, and parametric studies of several simulation parameters were carried out. Flow field features and trajectories of the vortex paths were compared with the experiments for different filling curves, with reasonable agreement. However, because shear stress discontinuities occurred in the predictions a strict volume-defined analytical model was constructed for wall movement with smooth spatial and temporal behaviour reproducing experimental filling curves. Numerical predictions showed not only an improvement in the qualitative features of the flow compared with the experiments, but also a quantitative improvement in the prediction of the vortex core paths. Also the shear stress discontinuities were no longer evident. In order to be able to estimate residence times, instantaneous streamlines and particle tracks were produced. Analysis of shear stresses in the fluid and generation of particle pathlines for residence calculation in 3-D geometries will be carried out in the next feature for model candidates for the final SMV design. Some of the material published during the course of the project is included in APPENDIX 1. In this thesis, attention is paid to the SMV fluid dynamics. However, SMV behaviour is a coupled fluid-solid problem. Future work will be carried out in the muscle modelling. To this end, a careful review has been carried out, and is included in the thesis. Implications for future work are also discussed.

Many modem combat aircraft have S-bend intake ducts supplying air to the engine compressor. At high Mach number and/or extreme manoeuvre conditions such ducts can produce excessive flow distortions at the engine face due to flow separating at the first bend of the duct. It has been proposed that vortex generators may be usefully employed in such intake ducts to enhance the homogeneity of the pressure distribution at the engine face, and hence, reduce the unsteady loading on the engine. Both vane and air jets have been tested experimentally as vortex generators and each has been found to reduce the flow distortion that would otherwise reach the engine face. The objective of this thesis was to construct a local numerical model reflecting the physical geometry and conditions of the fully turbulent flow field in the proximity of a vortex generator array. The location of the array is approximately at the first bend downstream in an S-bend intake duct. In this project, five different model geometries were tested. Two were used for model verification and the remaining three for investigation of the local flow structure in the vicinity of the vortex generators within the duct. Two of the local duct models neglected any curvature effects (referred to as flat plate models). The third duct model, referred to as a sector model took into account the circular nature of the duct's cross section. The flow is assumed to be incompressible and fully turbulent and was solved using the Finite Volume, Navier-Stokes Code CFX 4 (CFDS, AEA Technology, Harwell) on a non-orthogonal, body-fitted, grid using the k-c turbulence model and standard wall functions. The behaviour of the longitudinal vortices produced by the vanes and airjets is presented in terms of circulation and peak vorticity decay, peak vorticity paths in cross-stream and streamwise direction, cross-stream vorticity profiles, cross-stream shear stress distribution and streamwise and cross-stream velocity profiles. Negligible difference in results was observed for the flat plate models with and without the jet inlet tube; neither did we see significant differences between the flat plate model and the sector model, since the airjet momentum was not drastically altered. Comparing the predicted results provided by vanes and air jets reflected major differences in vortex circulation between the two but the enhancement in transverse skin friction was of similar magnitude. Experiments also showed that both types of vortex generator provided like enhancement of the flow field. The optimum pitch and skew angle configuration for the air jets in terms of maximum enhancement of the flow field was predicted with 30° pitch and 75° skew angle.

The main concern of this thesis is to develop and advance the knowledge of new hierarchical algorithms for optimal control of interconnected nonlinear systems. To achieve this, four basic hierarchical structures are developed by taking into account the manner in which real process measurements taken from interaction inputs are incorporated and utilized in the model-based optimal control problem. The structures are iterative in nature, and are derived using the dynamic integrated system optimization and parameter estimation (DISOPE) technique to take into account model-reality differences that may have been deliberately introduced to facilitate the solution of the complex nonlinear problem or due to uncertainty in the model used for computation. Three of the four basic hierarchical structures are used as a basis for developing hierarchical optimal control algorithms using a linear quadratic model formulation. Two approaches are used in the coordination problem of the algorithms, price coordination approach and the direct coordination approach. The algorithms are then implemented using two techniques, the single loop and the double loop techniques. All the algorithms are implemented in software and a simulation study is carried out using two examples to investigate their effectiveness and convergence properties. The optimality of the solution provided by the structures and the algorithms described in this research work are established. In addition, convergence analysis is carried out to provide sufficient convergence conditions of the double loop algorithms. Suggestions for future research as a continuation of the work presented in this thesis are also made.

This thesis covers the work and considerations necessary for the design of VHF mobile radio systems, and is directed particularly at the requirements of the emergency services of the United Kingdom. It was occasioned by an international regulatory decision to re-allocate part of their existing operating frequencies. This work provides both a structure for such a design, and a record of events. It considers the various elements of design, initiallv from a theoretical base which is developed to the extent where practical trials and external constraints overtake the theoretical analysis. The constraints can take the form of compatibility with the eXisting system, cost. timescale. and political considerations. The concept of a design flow chart is shown to lead to many circular arguments of the chicken-and-eg~ variety to the extent that the resultant diagram is termed in interaction diagram since its main function is to show how the design elements interface with each other. It consists of many feed-back loops; and design completion is achieved when a traverse of all paths of the interaction diaram can be made without meeting discontinuities at the interfaces. The topic of area coverage is examined in depth and the multiple transmitter common channel (Quasi synchronous) mode is shown to be efficient from both the time and the spectrum economy viewpoints. There are benefits also from operational considerations. The section shows that a reuse factor of 3 should be practical, which is of interest to mobile radio users in general. and cellular radio telephony systems in particular. The choice of type of modulation is quickly shown to lie between FM, AM and SSB. All three are analysed from single station and quasi synchronous aspects: the theoretical predictions of the superiority of AM, for the system under study, is confirmed by subjective trials from which design parameter limits are obtained. Examination of equipment hardware. mobile and base station, shows that system design aspects are severely constrained by intermodulation products due to passive site non linearities. Measurements indicate that these will be troublesome up to at least the 11th order. A frequency assignment plan covering mobile and fixed services for the country is derived and the dominating influence of the intermods is shown to result in a workable, but unconventional version.

This thesis concerns an experimental investigation of hydraulic fracturing when the water pressure is increased rapidly in a borehole and development of a possible simple method for identifying erodible clayey soils. Case histories of hydraulic fracturing in embankment dams and boreholes are reviewed. It is found that hydraulic fracturing in dams is often associated with rapid reservoir filling and zones of low stresses. Previously proposed criteria for hydraulic fracturing are outlined. It is found that no existing theory adequately explains fracturing observed in laboratory tests. Characteristics of dispersive clays and the methods of identifying them are reviewed. It is found that no single test provides results that have a satisfactory level of reliability. Hydraulic fracture was examined in the laboratory on samples of puddle clay from the core of Cwmernderi Dam in modified Bishop and Wesley triaxial cells. Various sets of tests examined the effects of sample geometry, loading rate and overconsolidation on the initial fracturing pressure and examined the effects of subsequent reconsolidation time on refracturing pressures. For the analysis of these tests results, simple criteria for hydraulic fracturing are proposed. The test results suggest that hydraulic fracturing for undrained conditions occurs when the drained strength of the soil element at the edge of a borehole is reached. For fully drained conditions, fracturing occurs when the soil element at the edge of the borehole reaches the no-tension cut-off. A new test, the cylinder dispersion test, was developed and a series of tests were carried on samples of puddle clay from several British dams and on various other soils. The results of these tests demonstrate the major influence of pore water chemistry on the true cohesion and dispersion properties of the soils.

The behaviour of natural soils is highly influenced by structural features arising from their geological history and was recognised to lie outside current frameworks which account only for the stress-volume state of the soil. The objective of the research was to compare the shear behaviour and stiffness of two naturally cemented sands: a calcarenite with relatively low densities, weak particles and strong bonding and a silica sandstone with high densities, strong particles and weak bonding. Comparative studies were also undertaken on their corresponding reconstituted soils and on an artificially cemented carbonate sand. Testing was performed in stress path controlled triaxial systems over a wide range of confining pressures. Identification of the yield surface was found to be an essential feature to describe the shear behaviour of the soils examined. The determination of the yield points of such stiff soils required internal measurements of stresses and strains, with an accuracy higher than currently achieved in soil testing. For reliable determinations of stiffness and yielding the uniformity of strains in the samples had to be guaranteed. Therefore several parts of the equipment were redesigned or modified and new sample preparation and setting up procedures were developed. Comparison of results from natural and reconstituted soils showed that bonding increased the stiffness and the stress-strain linearity. For the naturally cemented sands the maximum shear modulus was found to vary with state only when the soils were sheared after isotropic yielding. Undrained loading-unloading probes showed that the linear behaviour was reversible and that plastic strains were accompanied by a progressive deterioration of bonding resulting initially in a reduction of the yield stresses and finally, after sufficient cycling in a decrease of the maximum shear modulus. The combined influence of bond strength and specific volume on yielding could be accounted for when normalizing the yield stresses by an equivalent pressure taken on the state boundary in isotropic compression. This type of normalization also allowed the full state boundary to be identified. The yield surface was found to be the boundary limiting the domain governed by bonding and for the calcarenite occupied a larger portion of the permissible states than for the silica sandstone. States between the yield surface and the state boundary surface were controlled by bond degradation and either particle crushing for the calcarenite or dilation for the silica sandstone. After accounting for differences in states, both naturally cemented sands showed peak strengths which were higher than those for the reconstituted soils. For the calcarenite the peak strengths simply resulted from cohesion, as the peak stress ratios were reached on the yield surface. Only when stress ratios at yielding were lower than the critical state stress ratio the strength was truly frictional and coincident with the critical state. For the silica sandstone, in contrast, the peak strengths were found to be frictional except at the lowest confining pressures. The peak stress ratios were reached at states above the yield surface and were associated with a maximum rate of dilation. The dilatancy and strength of the intact soil were higher than those of the reconstituted soil at comparable states and were interpreted as resulting from differences in fabric and from the delayed volumetric response induced by the presence of bonding at the early stages of shearing. The case of the silica sandstone showed that only when density is a predominant factor in comparison with bonding then different modes of shear behaviour follow the theory of Critical State Soil Mechanics.

In this research the application of artificial intelligence techniques for on-line process control and fault detection and diagnosis are investigated. The majority of the research is on using artificial intelligence techniques in on-line fault detection and diagnosis of industrial processes. Several on-line approaches, including a rule based controller and several fault detection and diagnosis systems, have been developed and implemented and are described throughout this thesis. The research results obtained demonstrate that rule based controllers can be an alternative in situations where conventional mathematical modelling fails to give a high level of automation. The research on on-line fault detection and diagnosis emphasises the use of deep knowledge based approaches. Therefore, two on-line fault detection and diagnosis systems based on qualitative modelling have been implemented. For the first one only single abrupt faults have been considered while the second one can cope with single and multiple simultaneous abrupt faults. In order to overcome the problems associated with the inherent ambiguity of qualitative reasoning, a fuzzy qualitative simulation algorithm, which allows a semiquantitative extension to qualitative simulation, has been investigated. The adoption of fuzzy sets allows a more detailed description of physical variables, through an arbitrary, but finite, discretisation of the quantity space, and also allows common-sense knowledge to be represented rough the use of graded membership. Further research concerning self-reasoning has been one for qualitative model based diagnosis approaches. A self-learning system which can find any inappropriate settings of fault detection and diagnosis parameters and also learn fault symptoms from on-line sampled data, has been developed. Through machine learning techniques, the system can adjust fuzzy membership functions of the process variables automatically, as well as build the knowledge base on-line very efficiently. In order to cope with incipient faults and transient behaviour of the process under concern, a distributed online fault detection and diagnosis system, consisting of a knowledge based approach coupled with a fuzzy neural network, has been developed. The fault detection task is performed through the knowledge based approach. A systematic methodology for formulating fault detection heuristic rules from knowledge of system structure and component -functions has been investigated. Since structural decomposition corresponds to plant topology, such a method could be easier to implement. A fuzzy neural network approach has been used for fault diagnosis. This system combines the advantages of both fuzzy reasoning and neural networks. In order to speed up the fuzzy neural network training task, an extension of the classical backpropagation learning algorithm is also investigated. The research results achieved with this fault detection and diagnosis system reveal a very good performance and reliability provided that the training data is available.

The research described in this thesis investigated some applications of Cam-clay in numerical analyses. Single element analysis using the program CASIS was conducted to model undrained and drained triaxial compression tests to investigate the relative importance of the critical state soil parameters and streis history on the calculated soil behaviour. The values of these parameters covered the range of most commonly occurring natural soils. The investigation was extended to finite element modelling of plate loading tests using the computer program CRISP to examine the relative importance of the critical state soil parameters and stress history on the calculated undrained plate-soil behaviour. The capability of finite element coupled consolidation analysis of boundary value problems was demonstrated by the simulation of laboratory hydraulic fracturing experiments. The effects of rate of cavity water pressure increase, bore size, confining pressure and overconsolidation ratio on fracturing pressures were examined. The results of the parametric study of triaxial compression tests and plate loading tests were quantified in terms of stiffness (or tangent modulus), strength (or bearing pressure) and strain (or settlement) at the initial, yield, peak and ultimate states in the stress-strain (or pressure-settlement) response and allowed the comparison of these quantities for changing values of the soil model parameters. The studies revealed that the parameter x/A was the most influential because the initial states of soil were dependent on the values of ic and A and this dependence affected the subsequent stress-strain behaviour. In the prediction of movements of soil structures, the parameters x, A and a' are the most influential and their values should be critically evaluated for input in numerical analyses. In prediction of stability, the values of x, A and M are the most important ones. The work showed that the hydraulic fracturing phenomenon can be modelled by finite element coupled consolidation analysis. The computer program CRISP was validated against closed-form solutions and was found to predict the hydraulic fracturing phenomenon with a reasonable. degree of accuracy when the results were compared with the experimental data obtained by Mhach (1991). It was found that the rate of cavity water pressure increase had a significant influence in the prediction of fracturing pressure. The confining pressure was an important factor whereas the effects of bore size and overconsolidation ratio had some effects but were not as significant as the first two.

Although there are many computer codes available for analysis of fluid transients, only. a few are known to be applicable to line break situations and their scope is limited. There is therefore,s till a big potential for development work in the subject. Discrepancies between different models which have been developed have mainly centred on the assumptions used in developing the basic partial differential equations of flow, and subsequent simplifications; the thermophysical model used; representation of various terms in the equations such as the friction term; and the numerical method of solution of the basic partial differential equations. A previous model developed by Tiley (1989), overestimated the actual wave speeds and had problems of instability of the solution. A new approach, in which the three basic partial differential equation of flow are derived, based on the assumption of an unsteady quasi-one-dimensional flow of a real gas through a rigid constant cross-section area pipe, and using the Gamma Delta method is used. No further simplification is made on the basic equations. Significant improvements have been made on the type of equation of state, thermodynamic model, heat transfer approximation and friction factor representation. The QUANT software for thermodynamic and transport properties of real gases is used. A flow dependent explicit equation of Chen (1979) is used to calculate the frictional force and heat transfer is calculated using the concept of recovery factor and adiabatic wall temperature. Numerical solution of the basic equations is performed using the third-order Warming- Kutler-Lomax method, the second-order MacCormack method and the method of characteristics. A pc based computer coding with the C language is used. The QUANT software has successfully been incorporated with the programme. The full benefits of the software could not be realised with linebreak problems due to limitation of the range within which it gives output at present, but satisfactory results have nevertheless been attained. An improved and more accurate way of calculating the break boundary condition has been used. A non-uniform grid spacing has been used, which allow fine grid spacing in the vicinity of the break in order to enable accurate modelling of the rapid transients occurring in that part. Two different models for calculating the heat transfer i. e. one for the case of pipes exposed to the atmosphere and buried pipes have been incorporated with the model. Experimental data from full-scale pipeline tests is used to validate the computer models. Results from the computer model simulations show good agreement with the experimental data. The MacCormack method has been found to be unsuitable for modelling transient flow following linebreak in high-pressure gas pipelines. The method of characteristics has proved to be the method of solution for such applications. A better understanding of the flow following a break in high-pressure gas pipes is achieved, especially the decompression behaviour at the break boundary. Data gathered from feasibility studies conducted in the late 1980's for a pipeline in Tanzania is used to validate the steady state analysis model and to simulate a linebreak in the pipeline. Results of the computer simulation are discussed and recommendations made on the suitability the pipeline design. Additional work is recommended on refining and further testing of the computer programmes and using the Gamdeleps method which covers all the three phases region i.e. gas, liquid and gas/liquid.

The requirements for power generation systems have been reviewed together with the various energy sources available for them. Geothermal energy has been examined in more detail and the principal methods of recovering power from it which are currently employed are discussed. A novel method for improved power recovery from geothermal sources called the Trilateral Flash Cycle (TFC) system is described which has the special requirement of an efficient two-phase expander. Optimum results are obtained from this cycle if a working fluid is used which leaves the expander as dry saturated vapour. A binary mixture of hydrocarbons was therefore sought which by variation of the constituent proportions, would satisfy this requirement for a range of inlet temperatures when the condensing temperature is constant. Methods of estimating mixture properties are reviewed and the chosen thermodynamic model, as well as a computational procedure for evaluation of vapour-liquid equilibria of organic binary mixtures at high pressures, are described. This is based on the Redlich-Kwong- Soave cubic equation of state. By this means a mixture of n-pentane and 2,2-dimethylpropane (neopentane) was found to be the most suitable for the TFC system for expander inlet temperatures between 150-180'C. Temperature-entropy (T-S) diagrams of this organic binary mixture were obtained for several compositions. Bubble and dew pressures of (n-pentane + 2,2-dimethylpropane) have been determined experimentally for five different compositions at six different temperatures, (333.15 K, 353.15 K, 373.15 K, 393.15 K, 413.15 K, and 433.15 K). Vapour pressures of pure n-pentane and pure neo-pentane were also determined at these temperatures. The critical point of neo-pentane was measured to assess the accuracy of the isothermal compression apparatus used. Theoretical predictions were found to be in good agreement with experimental measurements.