This thesis introduces various control system designs for wave energy converters. It describes optimal conditions for maximizing the energy absorbed from wave energy converters. Feedback realisation of complex-conjugate control and velocity-tracking control are used to create these conditions. Also, passive loading control and latching control, which correspond to non-optimal conditions, are both discussed here. Several different ways of overcoming the non-causal problem of optimal conditions are also discussed. Firstly, prediction of wave elevation or excitation force can be used to solve the non-causal problem. Several predictive approaches’ performances are compared. A new approach, called multi-steps predictive identification, has been shown that have better performance than other approaches. Secondly, the prediction error method, which is used to find a constant approximation of a model’s performance can be also used to overcome the non-causal problem. The most important aim of this project is to maximize absorbed energy. The design of a Power Takeoff (PTO) device of a wave energy conversion system here involved direct optimization of the parameters of a mechanical network using the Nelder–Mead method can be linked to power absorption performance. Approximate transfer functions are realized with different admittances are here compared. Real ocean data sets were tested in terms of admittances. Their advantages and disadvantages of each method are presented in this thesis.

Wireless sensor networks (WSNs) have received much attention in recent years. Such networks comprise spatially distributed sensors to monitor various parameters. Space-based wireless sensor networks (SB-WSNs) consisting of tiny, low power, inexpensive satellites flying in a fleet with a close formation can offer a wide range of applications. Since communication is typically the major factor in power consumption, the activity of the transceiver should be reduced to increase the nodes’ lifetime. To understand the network power behaviour, a space-based wireless sensor network consisting of 40 nodes was designed as an experimental testbed. Several tests were undertaken to investigate the nodes’ lifetime and the packet loss with various sleep/wake up methods. The study found that the nodes with shorter paths to the sink benefit from improvement in their lifetime. In contrast, the other nodes with routes including many hops obtain less enhancement in their lifetime and high packet loss. To further reduce the power consumption, a novel sleep/wake up technique where the nodes have different sleep periods based on their locations has been proposed and tested. This modification enhanced the network operation time by 24% and increased the total delivered packets by 51% compared to when the nodes stay active all their duty cycle. Another concern was uneven power consumption due to the extra packet-relaying duties imposed on central nodes. This was addressed first by altering the connectivity of the network, and then by adding extra nodes dedicated to this task. The proposed sleep/wake up scheme was extended further through the adoption of transmission power control (TPC) and the introduction of multiple sinks. Both mechanisms were used to decrease the power budget required to deliver a packet from source to destination by reducing the number of hops in the paths. This improves the nodes’ lifetime and the total amount of collected data. Findings in this research have direct relevance to the use of commercial off the shelf (COTS) nodes in a SB-WSN and will provide an impetus for accurate estimation of the performance and design of such a network.

This thesis presents recovery control schemes that enable a quadcopter unmanned aerial vehicle (UAV) model to cope with a faulty actuation system. First, the computational aspects of the design of fixed-order H1 controllers are investigated along with the performance they provide for the quadcopter UAV. Double-loop control structures are developed to control the translational velocities of the UAV subject to two different intermittent actuation problems. Fixed-order H1 controllers are designed for the nominal and the faulty modes of operation. These closed-loop modes are modelled as a switched system for which stability is analysed using minimum dwell time theory. Average dwell times are also computed by exploiting multiple Lyapunov-like functions that account for the delays in the detection of a fault. The other key contribution of this thesis is the design of a switched recovery control scheme that does not require the explicit detection of the faults. Sufficient conditions are given in terms of linear matrix inequalities (LMIs) coupled with a scalar, and depend on modified Lyapunov-Metzler inequalities. The switched recovery scheme developed consists of jointly designed state feedback gains switched according to a min-switching strategy that preserves closed-loop stability and satisfy a prescribed H1 or H2 performance. Finally, the inherent fast switching issue of the min-switching strategy is treated at the expense of conservative reformulated LMIs conditions. Furthermore, the state-dependent switched control scheme is extended to output feedback case. Simulation results demonstrate the potential of the developed switching recovery control schemes to overcome various actuation faults.

A study was made of a series of steels containing varying amounts of carbon and vanadium (Table l). The yield and impact characteristics were compared using the relationships between grain size and yield strength and transition temperature as established by PETCH. The precipitation characteristics were studied by micro-hardness testing the ferrite formed under both isothermal and continuous cooling, and the effect of controlling the finishing temperature on rolling on the mechanical properties of one of the steels was also investigated. The results showed an increase in ?i, which increases the yield strength by 2 tons in-2 for the addition of 0.2% V when sufficient carbon is present (0.25%). This also gave a rise in transition temperature of 9°C, the majority of the increase being caused by the initial additions (0.02%). The relationships obtained were all non-linear, with the raising of the austenitising temperature giving a continuous increase in ?i. Grain refinement occurred in the 0.25% C steels, giving a reduction in transition temperature sufficient to counteract the rise due to the vanadium additions together with an increase in yield of 4 tons in-2. When insufficient carbon was present, grain refinement did not occur and the yield strength was lowered by 0.2% V showing that lowering of ?i due to the loss of carbon from solid solution is greater than the increase due to VC. The impact transition temperature showed a similar rise to the 0.25%C steeds. Nitrogen has a greater effect on ?i giving a larger increase in yield strength and transition temperature. A very fine grain size was obtained, but the reduction in transition temperature did not completely offset the rise due to ?i. Vanadium carbide precipitation followed the mode of known theory, secondary hardening being present with 0.25%C and 0.2% V; overageing, however was rapid. Nitrogen gave secondary hardening at lower vanadium and carbon levels, overageing being much slower. All high carbon steels gave hardening peaks on continuous cooling; rapid overageing again occurred. The final constant hardness increased with increasing cooling rate and it is probable that the highest strength occurs between 2000 and 7000°C/hr. Controlled rolling is beneficial if rolling is continued after precipitation occurs resulting in high yield strength but low transition temperature.

This thesis is concerned with a project on mixer-settler extraction systems, as a particular example of many liquid- liquid mass transfer processes. The project was aimed at a better practical understanding of the working of such systems and to develop theoretical models, which could then be used for both design and control. The construction of a system of mixer-settler tanks is described, together with the appropriate facilities for pumping and analysis of the process chemicals. The working pilot plant has been extensively tested to discover its characteristics, when subjected to changes in both flow and boundary concentration. Mass balance principles are applied to each contactor stage to develop equations, which include in their description, the flows in each stage. These flow dynamic relationships are extended to include the settler. Laplace transform analysis is used to produce a simplified model of the flow dynamics. An alternative approach, using sensitivity functions, yields a simplified transfer function model that includes a time delay term. This model can adjust to the process dynamics, in real time. A digital process model, based on the mass balance equations has been developed to work on-line using a standard process control computer, under the constraints normal to such equipment. Both modelling approaches are compared against experimental results. The importance of the time varying nature of the process parameters is recognised, and possible adaptive modelling schemes examined. A new parameter adjustment method is developed using correlation techniques. Finally, modelling is considered in the context of automatic control.

A theoretical and experimental investigation of the dynamic behaviour of preloaded cylindrical shells including the effects of meridional cracking has been carried out. A Donnell-type equation is derived to study preloaded cylindrical shell vibrations. The solution is obtained using the Galerkin Method for the following initial external loads: (i) Axial compression combined with torsion (ii) Bending moment (iii) Axial compression combined with bending moment (iv) Periodic axial compression (v) Periodic bending moment. A simply supported cylindrical shell was tested under axial compression, bending moment and axial compression combined with bending moment. The results are in fair agreement with the present theoretical solution. The analytical study of the vibrations of cracked shells is carried out by introducing the modifications to the strain and kinetic energy functions expressed in terms of normal co-ordinates. It is shown that cracks reduce the natural frequency and change the nodal configuration associated with the lowest natural frequency. The results of the experiments are in excellent agreement with the theoretical predictions.

In modern engineering there is an urgent need for a deeper understanding of the nature of surface texture and its influence upon the functioning of the element of which it forms a part. Of particular importance, in this connection, is the behaviour of surfaces in stationary and sliding contact. Investigations of the contact of surfaces, on the one hand, and the evolution of methods of surface specification and characterisation, on the other, have developed more or less independently. This thesis attempts to bridge the gap between these two areas of study. The main emphasis of the work has been upon random surfaces which are produced by a significant proportion of modern manufacturing methods. The theories used have been drawn from those employed in the study of other types of random processes. Both these theories, and the experimental evidence used to support them have been usually presented in digital form; therefore some emphasis has been placed upon the problems involved in the analysis of data presented in this form. The theoretical analysis is concerned with the representation of a surface profile as a random signal and the significance of this for the properties of surfaces of significance in their contact. This then allows the development of a theory of the movement of a second body over such a random profile. The friction and wear of random surfaces is tackled through the analysis of results obtained from well instrumented experiments; this suggests that a stochastic approach to the tribology of random surfaces is well justified. Finally an attempt has been made to provide a broad fundamental analysis of the generation of such surfaces. In this way it is hoped that the work provides a basis for the classification or the typology of surfaces in terms both of their functional behaviour and of the relationship of this to the details of their generation.

The author has investigated the properties of Hamiltonian circuits in a class of trivalent planar graphs and he has attempted, with partial success, to establish conditions for the existence of Hamiltonian circuits in such graphs. Because the Hamiltonian circuits of a trivalent planar graph are related to the four-colourings of the graph some aspects of the four- colour problem are discussed. The author describes a colouring algorithm which extends the early work of Kempe, together with an algorithm based on the Heawood congruences which enables the parity of the number of four-colourings to be determined without necessarily generating all of the four-colourings. It is shown that the number of Hamiltonian circuits has the same parity as the number of four-colourings and that the number of Hamiltonian circuits which pass through any edge of a trivalent planar graph is either even or zero. A proof is given that the latter number is non-zero, for every edge of the graph, whenever the family of four-colourings has either of two stated properties. The author describes two original algorithms, independent of four-colourings, which generate a family of Hamiltonian circuits in a trivalent planar graph. One algorithm embodies a transformation procedure which enables a family of Hamiltonian circuits to be generated from a given Hamiltonian circuit, while the other generates directly all Hamiltonian circuits which include a chosen edge of the graph. In a new theorem the author proves the existence of Hamiltonian circuits in any trivalent planar graph whose property is that one or more members of a family of related graphs has odd parity.

The modes of behaviour of a representative two-bar assembly with unequal areas and lengths under the simultaneous action of sustained mechani- al load and cyclic thermal gradient are investigated analytically. Three types of material behaviour are used: perfect plasticity, linear kinematic hardening and linear isotropic hardening. These simple models exhibit much of the behaviour of interest in design of structural components subjected to repeated thermal loads: elastic shakedown, reversed plasticity and ratcheting. The analyses provide closed form expressions for the mechanical-thermal load bounds of the various regimes of deformation. The cyclic plastic behaviour of the structure is developed and analytical results are derived for the transient and steady state values of plastic strain. The results are applicable for a wide range of geometrical, material and loading parameters. Comparisons between perfect plasticity, kinematic and isotropic hardening models provide qualitative estimates of the cyclic inelastic behaviour of actual structural components which can be simulated by means of a two-bar assembly. The results also point out those load combinations at which thermal ratcheting experiments are more likely to yield the most useful informations. In the field of new constitutive relations, a single state variable theory of inelastic deformation is developed on the basis of the Bailey-Orowan concept of creep as the outcome of two competing mechanisms: strain hardening and thermal softening. The resulting theory is capable of representing primary creep, creep recovery, there reemergence of primary creep following a sudden increase in stress, effects of rest periods and past deformation history and strain rate sensitivity. The theory is not capable, however, of reproducing the features of material behaviour under reversed loading conditions. An attempt to describe the cyclic phenomena of metals on the basis of the two-state variable concept is presented. The material behaviour is characterized by means of the current size of the yield surface and a dimensionless parameter which represents the shape of the plastic hardening curve. The transient growth laws for these two parameters are developed from phenomenological data on annealed OFHC copper. The predictions of the model are in close agreement with the experimental cyclic hardening behaviour of copper. The model is used to obtain the inelastic response of a two-bar assembly subjected to cyclic thermal load and the results are compared to the closed form solutions of linear hardening models. Finally, a modification is suggested to the structure of the proposed model in the light of recent work on the application of the state variable concept to the theory of plasticity. It is argued that the second parameter in the model should be taken as the current coordinates of the yield surface of the material.

This thesis presents a simple and conservative method for the analysis of inelastic cyclic thermal stress problems. It is developed for the analysis of severe cyclic thermal shocks combined with mechanical loads in thin walled tubes. The methods are intended to predict both the onset of ratchetting and the strains accumulated by small excursions into the ratchetting regime. The text begins by summarising the development of the upper bound shakedown theory-the main tool in this approach. Previous methods for simplified analysis using the upper bound method are reviewed and some ideas adopted. Building on this previous work, a new method of predicting the shakedown bound is developed. This is based on the upper bound analysis of particular ratchet mechanisms, resulting in a limiting equation for each mechanism. These equations are used to produce ratchet bounds for several thermal stress examples. These results are compared with the bounds produced by design code rules currently used for cyclic thermal stress. One observation reached in the prediction of these bounds was the lack of supporting evidence available. Therefore, a set of experiments were performed to show the bounds corresponding to two of these mechanisms. These used a new resistance heating method for applying the thermal cycles. The results are conservatively predicted by upper bound analysis. However, in some regions the effects of cyclic hardening reduces the strain accumulated, making the predictions over conservative. To investigate this effect some simple hardening rules are applied to the shakedown bounds. These are developed to give a prediction of accumulated strains within the ratchet region. Comparison is made both with the experiments and a finite element computer model. Finally, these predictions of shakedown and cyclic hardening controlled strain are combined to give a complete picture of cyclically heated tubes below the creep range.