Includes bibliography. / Uninterruptible power supplies (UPS) maintain a constant supply of power to a critical load. The distinguishing features are a fixed voltage and frequency, low harmonic content and the ability to supply the load for a period of time after the incoming supply has failed. The document begins by identifying the typical power-line disturbances and their effect on various types of equipment. Together with the power conditioners used to suppress these disturbances, various UPS standby, line-interactive and on-line configurations are introduced. Next, the possibility of modifying a locally manufactured UPS to meet the design specifications is investigated. The performance of the system under load is evaluated, and forms the basis for the following decision. Due to the large number of modifications required and the inflexibility of local UPS topology, an alternative topology is adopted. In the new topology a power factor corrector, constructed around a DC to DC boost converter, interfaces with the incoming AC line. It delivers a half sinusoidal current into the DC bus. An IGBT inverter using sinusoidal unipolar pulse-width modulation regenerates the AC load voltage after filtering through a LC-filter. The DC bus voltage ripple is reduced by synchronising the inverter load and power factor corrector current while the battery pack maintains the DC bus during a power failure. The power factor corrector employs a dedicated analogue controller chip while a MCS-51 microcontroller generates the inverter PWM, provides the remote monitoring facilities, battery charging and performs general support tasks. The total harmonic distortion of the input current is measured at less than 4% while the power factor remained above 0.99 over the entire load range. The ripple regulator reduced the DC bus voltage ripple without any noticeable effect on the load. Under maximum load, the steady state output voltage is maintained during the -20%, +10% variation in the incoming line voltage. However, the transient response fails to meet the 5% design specification. A 0-100% load step results in a 7% drop in the output voltage while the loss of the load results 10% jump in voltage. System efficiency is measured at 85%. It is the lack of processing power, precluding the use of floating point or an optimal control algorithm, which ultimately compromises the performance of the system. It is recommended that the microcontroller be replaced with a 16-bit processor or digital signal processor to provide the extra computational power needed to optimise the UPS response. To improve the voltage regulation, it is recommended that the control include an inner current loop while the switching frequency should be increased to reduce the energy storage in the output filter. Further adjustments and refinements to the topology are suggested in the final chapter.

Bibliography: leaves 105-108. / This thesis investigates the problem of electric long-tenn load forecasting based on weather conditions (specifically temperature) and also investigates load forecasting by segmenting customers according to their pattern of use using clustering techniques in order to produce an effective long-tenn load forecast.

Includes bibliography. / New equipment for producing and following the fatigue process, in metals, has been developed. The specimens, which must be in the shape of tuning forks, are resonated at high amplitudes. This is achieved by attaching small, powerful, samarium-cobalt magnets to the ends of the tynes, enabling them to be driven efficiently by a "U"-core electromagnet. A small, piezoceramic strain gauge provides a method of picking up the vibrations. To maintain resonance, the signal is used in a positive feedback loop, which incorporates an analogue multiplier to provide AGC. This also keeps the amplitude constant at any desired level, throughout the duration of an experiment.

This thesis set out to improve on previous methods of digital control of a three phase fully controlled thyristor bridge, using phase control of the thyristors, and to use this bridge for the purpose of controlling the speed of a separately excited D. C. machine, under the supervision of a small digital computer. Interfacing equipment for this purpose was designed and built. Various methods of speed detection were considered, and a simple digital tachometer designed and built. This tachometer was able to feed a digital number, related to the speed of the machine, to the computer, and was accurate to better than 1%. A direct digital control system was then implemented. The behaviour of the system was first studied under open loop conditions in order to determine the parameters of the system, and to discover any shortcomings in it. The behaviour of the closed loop system was then studied, using z-transform analysis, at sampling rates of from twice to eight times the natural frequency of the machine. Initially, the computer was used as an integral controller. The behaviour of the mathematical model was compared to that of the actual system, and the correlation found to be very good. The control system was then compensated, using compensation routines based on the z-transform analysis, and the behaviour of various compensated systems studied at sampling rates varying from 0,4 to 1 second. Again, the actual system response was found to correspond closely to the designed behaviour. Finally, a dual-machine control system, on a time sharing basis, was implemented, and a degree of interrelation between the speeds of the two machines introduced. The control algorithms developed were able to provide speed control to better than l% accuracy, with a 5% or less overshoot in the step response, a delay of one sampling period but no overshoot in the ramp response, and satisfactory behaviour under load transient conditions. The time, and memory, utilization was such that such algorithms could be extended to the control of several machines, using the single computer, if sufficient input and output facilities were available.

Includes bibliographical references. / This thesis aimed to design a measurement system to obtain B/A readings for different liquids using the isentropic phase method. The resulting system uses a sinusoidal wave source that operates on less than 20kPa to pressurize a sample liquid of less than 0.2 ml. The advantage of the system is that it can obtain measurements using relatively few parameters of the liquid, requiring only density and sound speed, and that it can obtain a result in a relatively short amount of time (less than 15 minutes after system has reached thermal equilibrium).

Bibliography: leaves 109-113. / Various transducers have been developed turbulent shear in the ocean. The airfoil probe with it's piezoceramic beam encapsulated in a rubber airfoil has the simplest electronic processing system. However, the beam is very fragile and displays unpredictable thermal effects. An airfoil probe with semiconductor strain gauges on an aluminium cantilever beam has been developed as an alternative to the piezoceramic beam. The probe was calibrated by exciting it with a known shear generated by the water flow from an oscillating nozzle. During the calibration the thermal sensitivity of the probe was established to be -1.7 % °C-1• The probe, along with it's high gain, low noise processing system, is fitted to a tethered free-fall vehicle. A solid state data logger situated in the vehicle is used to record the data generated by the turbulence probe and a pressure transducer. Field trials at St.Helena Bay and Hout Bay showed that the sensitivity of the system is 20 dB lower than that of similar systems using piezoceramic beams. The system is able to resolve turbulent dissipation levels above 10-8 W kg-1, making it useful in regions characterised by a typical mixed layer (dissipation level of 10-W kg-1); however, it's sensitivity is not adequate for deep sea measurements where dissipation levels may be as low as 10-10 W kg-1.

Includes bibliographical references. / This thesis investigates high-speed digital transmission over a conditioned, voice-grade telephone circuit (M.1020), using a technique known as partial response signaling, or PRS. In particular, the case where 4800 bps, full-duplex transmission is required in a CCI'PT V. 22 type format is investigated. The main v.22 criterion to be adhered to, is that frequency-division multiplexing (FDM) is to be used as the means of separating thetransmit and receive channels. The carrier frequencies should be 1200 Hz and 2400 Hz respectively. The investigation concerns the modulation and demodulation sections only.

Bibliography: leaves 102-106. / A program is presented for the solution of static electromagnetic fields in bounded and unbounded 2- dimensional domains. The program comprises of a mesh generator which descretises the domain into triangular finite elements or if applicable into special elements, called infinite elements which model the bounded and unbounded domains respectively. The potential function is solved for, using a special solution technique which enhances the speed of the program. The program outputs data in the form of potential or flux component distributions along lines of interest.

Bibliography: leaves 156-160. / The extraction of gold from ore that has been mined is the most important part of the process which eventually produces gold bullion. The process most commonly used today is that of carbon-in-pulp gold extraction (CIP). One of the main reasons for this is that it is the most economically efficient method of extracting gold from ore. The process uses activated carbon to absorb gold from a cyanide leach solution. Slurry containing the gold bearing ore and the activated carbon flow in a counter-current fashion. This counter-current flow enables a high percentage of the gold to be recovered. Gold can then be recovered through an elution process. Large amounts of activated carbon are used in the process and a formal multi variable control study of the adsorption section of the CIP process could provide further economic savings by extracting more gold with controlled amounts of carbon. A study was performed to identify the chemical mechanisms involved in the adsorption section of a CIP plant. It was felt that the workings of the process could best be established by designing a simple simulator of the process. The simulator was designed with four reactor tanks, in which the carbon absorbs gold from the leached slurry. The simulator uses a continuous transfer of carbon. In order to fully understand the operation of the HERIG, a simulation study was performed. This simulated model was a simplified version of the actual rig. The level changes of COLD water in the tanks were assumed to be instantaneous and the heat transfer coefficients were assumed to be the same for all four of the tanks. The calculation of the heat transfer coefficients was investigated thoroughly and care was taken to obtain accurate values. The simulator designed was a lumped parameter model. The pipes containing the HOT stream were divided into many small section, in each of which a constant temperature was assumed. A sum of the contributions of all the sections submerged under the COLD water was used to calculate the heat transferred into the COLD water. The COLD water in the tank is stirred continuously and is assumed to be at a constant temperature. The level of COLD water in each tank on the HERIG represents the mass of carbon in each of the tanks on a CIP plant. A change in the HOT water pipe temperature (concentration of Au in slurry) was examined as a function of a change in the level of COLD water in a tank (mass of carbon in a reactor). A steady state and dynamic analysis verified that trends observed from the CIP model were in fact mimicked by the trends observed on the HERIG. It was then decided to perform a formal control study of the HERIG, since the numerous similarities found between the CIP and HERIG enabled relevant conclusions to be drawn about the control of CIP from the control of the HERIG.

Includes bibliography. / The objectives of the project were firstly, to propose and implement the hardware needed for a Machine Vision System that is to operate on the Spiral Ore Concentrator. Secondly, to propose and implement an algorithm that would operate together with the hardware, to find the transition point between the two types of material, also to find the optimum position for the blade separator, while the system operates in a continuous repetitive mode. Finally, to have an on line analysis of the efficiency of the refinement process in the Spiral Ore Concentrator. The hardware of the Machine Vision System was implemented and tested in the laboratory. The hardware included a video recorder, through which a video signal was obtained from the play-back of video material taken of the ore flow in the Spiral Ore Concentrator. The video signal was used as input to an RGB decoder in order to remove the colour modulation. A black and white frame grabber digitized the video signal which was then analyzed by a computer. Based on the fundamental theory of edge detection used in computer vision systems, an algorithm was designed and implemented to detect the transition point between the two types of material. This algorithm was used to find the difference of averages of grey levels between two global neighborhoods within a specified area of interest window. The algorithm gave consistent results and was robust towards surface irregularities. The algorithm, operating in a continuous repetitive mode, gave rapid fluctuations in the determined edge position. Because a motor with a slow response time would be used to control the movement of the blade separator, the determined edge position signal was smoothed by a filter. Based on periodogram analysis of the edge position signal, a smoothing filter was implemented which incorporates a median filter, followed by a fading-memory polynomial filter. These filters gave sufficient smoothing with little lag to step changes in ore concentration. The efficiency of the ore separation was monitored by the determination of losses. These losses consist of, the percentage of black material loss and percentage of white material contamination. Two types of losses could be identified, they were Spraying losses and Filter losses, these were combined to give Total losses. From the true edge position curve, which was obtained by finding the edge position every second pixel point, the Nyquist sampling frequency could be determined. Because of the slow sampling rate, an error in the calculation of the losses was determined from the true edge curve, and the sub-sampled edge curve. Based on this error, it was shown that the machine vision system could multiplex multiple camera inputs.