Search

Mathematical models of varying complexity for representing turbo- generators and associated excitation and prime mover control systems are developed. Special reference is made to the application of these models for digital computer solution to determine turbogenerator transient performance-under large system disturbances. A parametric study, of the effect of differing generator designs and control systems configurations, on the transient behaviour is performed, using the digital computer model, In particular, the capabilities of the latest high speed control systems, such as, thyristor exciters and electrohydraulic governors, are assessed, Novel methods of stabilisation of these systems and the effects of both single and combined control action are investigated. In choosing the most appropriate computer-model, the aim was to achieve the highest possible complexity, taking into account the limitations on available machine data, A full account of the results of fault tests on a full scale turbogenerating system with thyristor excitation, in which the author participated, is given, These were intented to verify some of the control policies developed, and provide an accuracy check on the computer programme to give confidence in the modelling technique for future simulkations. Fast response instrumentation for obtaining specialised feedback signals for the site tests is described, together with a full set of equipment for laboratory investigation on a model power system. Although a number of difficulties with the instrumentation were encountered in the tests, the results serve to illustrate the capabilities of fast thyristor excitation systems, and the high degree of accuracy obtained by the digital computer simulation techniques.

Engineering design often requires the optimisation of multiple objectives, and becomes significantly more difficult and time consuming when the response surfaces are multimodal, rather than unimodal. A surrogate model, also known as a metamodel, can be used to replace expensive computer simulations, accelerating single and multi-objective optimisation and the exploration of new design concepts. The main research focus of this work is to investigate the use of a neural network surrogate model to improve optimisation of multimodal surfaces. Several significant contributions derive from evaluating the Cascade Correlation neural network as the basis of a surrogate model. The contributions to the neural network community ultimately outnumber those to the optimisation community. The effects of training this surrogate on multimodal test functions are explored. The Cascade Correlation neural network is shown to map poorly such response surfaces. A hypothesis for this weakness is formulated and tested. A new subdivision technique is created that addresses this problem; however, this new technique requires excessively large datasets upon which to train. The primary conclusion of this work is that Cascade Correlation neural networks form an unreliable basis for a surrogate model, despite successes reported in the literature. A further contribution of this work is the enhancement of an open source optimisation toolkit, achieved by the first integration of a truly multi-objective optimisation algorithm.

In this thesis Direct Numerical Simulation (DNS) is used to investigate the development of turbulent premixed flame kernels during the early stages of growth typical of the period following spark ignition. Two distinct aspects of this phase are considered: the interaction of the expanding kernel with a field of decaying turbulence, and the chemical and thermo-diffusive response of the flame for different fresh-gas compositions. In the first part of the study, three-dimensional, repeated simulations with single-step chemistry are used to generate ensemble statistics of global flame growth. The surface-conditioned mean fluid-velocity magnitude is found to vary significantly across different isosurfaces of the reaction progress variable, and this is shown to lead to a bias in the distribution of the Surface Density Function (SDF) around the developing flame. Two-dimensional simulations in an extended domain indicate that this effect translates into a similar directional bias in the Flame Surface Density (FSD) at later stages in the kernel development. Properties of the fresh gas turbulence decay are assessed from an independent, non-reacting simulation database. In the second part of this study, two-dimensional simulations with a detailed 68-step reaction mechanism are used to investigate the thermo-diffusive response of pure methane-air, and hydrogen-enriched methane-air flames. The changes in local and global behaviour due to the different laminar flame characteristics, and the response of the flames to strain and curvature are examined at different equivalence ratios and turbulence intensities. Mechanisms leading to flame quenching are discussed and the effect of mean flame curvature is assessed through comparison with an equivalent planar flame. The effects of hydrogen addition are found to be particularly pronounced in flame kernels due to the higher positive stretch rates and reduced thermo-diffusive stability of hydrogen-enriched flames.