Rotors supported by active magnetic bearings under contact-free levitation have many advantages, such as allowing near frictionless rotation and high rotational speeds. They also provide the designer the capability to achieve increased machine power density. However, magnetic bearings possess limited load capacity and operate under active control. Under certain operational conditions, the load capacity may be exceeded or a transient fault may occur. Touchdown bearings or bushes are required in such systems, from the operational and design points of view, to prevent contact between the rotor and stator laminations causing damage to the system. Rotor/touchdown bearing contact can occur in fault conditions and under external disturbances, even if the magnetic bearings are fully functional. If the rotor makes contact with the touchdown bearings, the ensuing rotor dynamics may result in transient or sustained contact dynamics. These dynamics involve transmitting a range of stresses under distortional strains. Such stresses occur under different contact modes, can be of very short or long duration, and can affect the life span of touchdown bearings. Touchdown bearings thus characterise significant safety and reliability aspects of active magnetic bearing systems. Maximisation of the performance and operational life of magnetic bearing systems dictates the need for minimisation of rotor/touchdown bearing contact. Magnetic bearing forces may have the capability to restore contact-free rotor levitation, though this will require appropriate control strategies to be devised. An understanding of the contact dynamics is required, together with the relationship between these and magnetic bearing control forces. In this thesis, rotor/touchdown bearing contact conditions are identified and investigated dynamically. An active magnetic bearing system with a long flexible rotor is considered. A nonlinear system model is employed, and a speed range covering three of the rotor’s critical frequencies is taken into account. Different types of transient and steady-state contact modes are identified, including non-persistent and persistent trapped contact modes of varying contact force levels and time durations. Design methodology is presented for a force measurement system capable of providing rotor/touchdown bearing contact force related data, based on experimental strain measurement. The system is implemented and a calibration method of assessing magnetic bearing forces based on rotor/touchdown bearing contact is demonstrated. The frequency dependent behaviour of the active magnetic bearing system is considered using evaluated force and phase measurements. Force measurements covering one of the rotor’s critical speeds are experimentally validated using an open-loop control strategy, aimed at attenuating rotor vibration in contact cases. Rotor recovery from a persistent rub contact mode through employing synchronous active magnetic bearing forces is demonstrated and discussed. The range of magnetic bearing control forces capable of contact elimination is also explored for different running speeds. The new control method presented provides insight into potential control methods capable of achieving rotor/touchdown bearing contact recovery utilizing experimental force data in contact conditions. This will contribute towards improving safety and reliability aspects of active magnetic bearing systems, which undergo operational conditions leading to rotor/touchdown bearing contact.

The prevention of hot gas ingress between rotating and stationary discs in gas turbines is big business, with experimental and computational research being common in the sector. Experimental rigs, operating at a fraction of the engine size and in simulated fluid dynamic conditions, model the engine environment. Computational Fluid Dynamics (CFD) is used in both academia and industry to model the flow and heat transfer in a turbine. CFD is expensive, time consuming and requires detailed experimental validation. The engine designer has a need for simpler, faster mathematical modelling methods, ultimately to be used in 1D design codes. The research in this thesis stems from this need for the industrial engine designers to be able to predict the flow, pressure and temperatures in the secondary-air-system. Momentum-integral equations are known to model flow over rotating and stationary discs in isolation. This thesis shows that the momentum-integral equations can be solved together, to successfully model the flow inside a rotor and stator cavity. New momentum-integral equations are derived, free of the incorrect assumption that swirl ratio inside a rotor-stator cavity does not vary with radius. Two cavity models are described based upon the momentum-integral equations: one for a closed cavity and one for a cavity with sealing flow and no ingress. Both are computationally fast and are shown to give good agreement with experimental measurements and CFD results. Detailed flow structures are given for a range of rotor-stator cavity cases and the results of the models allow conclusions about the flow structure to be drawn. It is found that the outer region, where flow leaves the rotor and is entrained by the stator, is not affected by sealing flow. As well as complete cavity models, two other models for specific rotor-stator phenomenon have been derived. The effect of ingress on the swirl ratio in the cavity has been modelled, using a momentum balance approach. The buffer ratio and buffering effect, which quantify how the rotor is protected from ingress, have been defined, modelled and validated against measurements of adiabatic effectiveness for four different seal geometries. The model has allowed the calculation of Φ′min,r, the sealing flow rate where the effectiveness on the rotor reaches 95%.

Modern gas turbine design continues to drive towards improved performance, reduced weight and reduced cost. This trend of aero-engine design results in thinned blade aerofoils which are more prone to aeroelastic problems such as flutter. Whilst extensive work has been conducted to study the flutter of isolated turbomachinery blades, the number of research concerning the unsteady interactions between the blade vibration, the resulting acoustic reflections and flutter is very limited. In this thesis, the flutter of such embedded blade rows is studied to gain understanding as for why and how such interactions can result in flutter. It is shown that this type of flutter instability can occur for single stage fan blades and multi-stage core compressors. Unsteady CFD computations are carried out to study the influence of acoustic reflections from the intake on flutter of a fan blade. It is shown that the accurate prediction of flutter boundary for a fan blade requires modelling of the intake. Different intakes can produce different flutter boundaries for the same fan blade and the resulting flutter boundary is a function of the intake geometry in front of it. The above finding, which has also been demonstrated experimentally, is a result of acoustic reflections from the intake. Through in-depth post-processing of the results obtained from wave-splitting of the unsteady CFD solutions, the relationship between the phase and amplitude of the reflected acoustic waves and flutter stability of the blade is established. By using an analytical approach to calculate the propagation and reflection of acoustic waves in the intake, a novel low- fidelity model capable of evaluating the susceptibility of a fan blade to flutter is proposed. The proposed model works in a similar fashion to the Campbell diagram, which allows one to identify the region (in compressor map) where flutter is likely to occur at early design stages of an engine. In the second part of this thesis, the influence of acoustic reflections from adjacent blade rows on flutter stability of an embedded rotor in a multi-stage compressor is studied using unsteady CFD computations. It is shown that reflections of acoustic waves, generated by the rotor blade vibration, from the adjacent blade rows have a significant impact on the flutter stability of the embedded rotor, and the computations using the isolated rotor can lead to significant over-optimistic predictions of the flutter boundary. Based on the understanding gained, an alternative strategy, aiming to reduce the computational cost, for the flutter analysis of such embedded blades is proposed. The method works by modelling the propagation and reflection of acoustic waves at the adjacent blade rows using an analytical method, whereby flutter computations of the embedded rotor can be performed in an isolated fashion by imposing the calculated reflected waves as unsteady plane sources. Computations using the proposed model can lead to two orders of magnitude reduction in computational cost compared with time domain full annulus multi-row computations. The computed results using the developed low-fidelity model show good correlation with the results obtained using full annulus multi-row models.

This thesis deals with topics concerning both passive and active control of stirred tanks. Regarding passive flow control, the effect of certain turbine blade modifications is investigated, most notably that of the blade perimeter increase in a fractal manner, applied on a conventional radial turbine stirring an unbaffled tank. It is found that the tested modifications show potential for applications, as by applying them, a drop in power consumption, an increase of the bulk turbulence intensity and the mass flow rate, and a suppression of the shed blade vortices' intensity and coherence is achieved. The latter, in particular, is argued to be a potential cause of the above-mentioned drop in torque/power consumption. Additional material from this section are the detailed comparison of fractal and perforated bluff bodies and a characterisation of the form drag distribution of radial turbines stirring unbaffled tanks. The latter was achieved by employing a novel pressure measuring technique. Regarding the active flow control, this thesis focuses on the prediction of stirred tank power consumption in situations where the shaft speed is not constant, but rather time dependent. The motivation for this is that such speed control has been shown to promote mixing in the tank. Employing first principles, qualitative scaling laws and empirical correlations, analytical models for the prediction of the torque response, when the shaft speed undergoes smooth, or step changes are developed. The predictions are then experimentally validated using torque measurements. The above models could find application in the design process of variable speed systems.

Autoignition is a multi-physics phenomenon that occurs in wide-ranging engineering applications, such as diesel engines, gas turbines and jet-engine afterburners. In these applications, autoignition occurs in the presence of flow inhomogeneities and turbulence. Further, liquid fuel autoignition presents a case of chemically reacting flow where processes such as jet break-up, atomisation, droplet evaporation (interfacial heat and mass transfer), turbulent mixing and chemical reaction occur simultaneously in the presence of important flow, mixture and phase inhomogeneities. The multi-scale nature and direct coupling between these various contributing processes, present a challenging fundamental problem which cannot be understood by extrapolation from homogeneous, inhomogeneous (gaseous) and even single droplet combustion studies. This thesis presents an experimental study on autoignition of polydispersed droplets generated by single liquid jets of pure liquid n-heptane and n-pentane injected axisymmetrically from a circular nozzle into a confined turbulent hot coflow of air at atmospheric pressure. Distinct phenomena are identified concerning the emergence of various autoignition regimes --- no autoignition, random spots and continuous flame; the occurrence of these regimes depends on the reaction conditions of air temperature, air velocity, and liquid fuel injection velocity. In the random spots regime, autoignition appears in the form of well-defined localised spots occurring randomly along the length of the reactor. Optical measurements of these random spots are made from which the autoignition lengths/locations are measured and are used to infer average delay (or residence) times from injection. At higher air temperatures and lower liquid fuel injection velocities, autoignition is observed to move closer to the injector; the corresponding delay times also decrease. With increasing air velocity and hence turbulent fluctuations, autoignition moves downstream but the delay times decrease. The shorter delay times are also associated with faster evaporation of the liquid fuel. The results from this work suggest that turbulent inhomogeneous autoignition of liquid fuels cannot be directly predicted from chemical delay times from homogeneous studies and that the effects of evaporation, and turbulent mixing, as quantified by the mixture fraction and conditional scalar dissipation rate must be accounted for. The results also provide insights into the flow conditions leading up to the various observed autoignition phenomena for liquid fuels, and constitute a valuable data-set that can contribute towards developing and validating models of advanced multiphase turbulent combustion and chemically reacting flows.

One of the operational problems encountered with vertical axis wind turbines is their low starting torque. A number of analytical methods were investigated to see whether they could predict the starting performance of vertical axis turbines. The chosen methods used " actuator disc theory" for both single and multiple streamtubes. Two different forms of the multiple streamtube model are applied, one using a single actuator disc and the other using two discs in tandem. The computational analysis of all models simulates the blade aerodynamics throughout the full range of incidence from -180° to 180°. The effects of varying various geometric parameters of the windmill upon the performance of the rotor are investigated to find a design with improved self starting characteristics. The best agreement between theory and experiment was obtained using the multiple streamtube (double disc) method. Savonius rotors have been commonly employed as " starters "for Darrieus turbines. A new analytical method has been developed to model the performance characteristics of the Savonius rotor. In this method the blade is divided up into small elements, and each element is treated as a thin airfoil. The rotor torque and power are computed taking into account the blades' motion, the blade shape and momentum consideration. This method shows good agreement with experimental results for a variety of Savonius rotors.A new experimental technique has been developed to provide information about the variation of torque within a cycle. These results have been used as a check on all the theoretical methods. The agreement between these experimental results and the theoretical methods show that they predict both the time averaged and the instantaneous performance.

Electricity generation is highly carbon intensive, associated with complex externalities. The preceding thirteen decades have witnessed improvements in efficiencies of rotary machines employed' for electricity generation through enhanced metallurgy, understanding of fluid mechanics and thermodynamics; consequently improving their conventional designs without much heed to environmental degradation, understanding of energy resources, and sustainability, until emergence and realisation of disastrous climate changes. This forced the environmentalists and scientists to explore new frontiers, and to harness renewable energies in any possible way. The situation required an innovative approach and unconventional design to satisfy the need of time. The aim of this research is to recommend a novel design for an expander-turbine that is suitable for generating electricity using compressed air as working fluid; investigating material suitability for cryogenic temperatures. The underlying endeavour is to promote reliable quality renewable energy generation that is sustainable, economical, and suggesting zero-carbon solution in commercial and domestic environment. The results of this venture are very promising and are expected to address UK's commitments to Kyoto protocol for reducing carbon footprint in energy generation. The methodology adopted was to investigate efficiency of individual mechanisms in relevant rotary machines that helped in identifying individual portfolio components that may improve efficacy of turbines if put together.

Wind turbines become more cost effective as they grow larger; however the blade mass increases at a greater rate than the power. For a continued size increase, reducing the mass of the blades is necessary. Additionally, lighter blades lower overall turbine costs because the loads on the rest of the structure are decreased. Therefore, the use of lightweight blades can have a significant impact on the cost of wind energy. To achieve blade mass reductions, an alternative structural layout is generated using topology optimisation. The result is a topology which varies along the blade length, transitioning from a structure with trailing edge reinforcement to one with offset spar caps. An alternative beam topology optimisation method is developed that enabled a buckling constraint to be applied. The structural efficiency of the topologically optimised blade is then assessed using shape factors and performance indices, measures which have been expanded to account for asymmetric bending of beams with multiple materials. The utility of shape factors is first demonstrated on six example beam sections before being applied to the blade. To demonstrate application to a more refined design, the performance of a 100m wind turbine blade is assessed , using maps to visualise the structural efficiency. The effect of using carbon fibre and offsetting the spar caps is evaluated, providing a greater understanding of the improved designs. Overall , the results show that wind turbine blades can be improved with structural layouts that take advantage of favourable bend-bend coupling between the out-of-plane and in-plane directions. Because traditional design concepts do not account for bending coupling, a missed opportunity for further mass reduction exists. To this day, the structural topology of the blades has remained fixed despite increasing length and changing loads. Topology optimisation and structural efficiency analysis are shown as methods used to challenge this design convention and reduce blade mass, thereby lowering the cost of wind energy.

Radiation tube burner systems are widely used as in-direct heating systems in heat treatment furnaces. Saving energy through improving the combustion and heat transfer process in the radiation tubes has become a pressing issue in the heat treatment industry over the last few years. The material structure during the process of heat treatment is predominantly determined by the temperature. The heat treatment processes usually require an even temperature distribution inside the furnace over a long duration to achieve the desired material properties. Due to the rising energy costs and the environmental concerns regarding the combustion emissions, reducing energy consumption has become a significant area of concern for the industry. This research is aimed at improving the combustion process in the radiation tubes to enhance the heat transfer to the heat treatment chamber, therefore achieving the objective of making energy savings. As a measure of improving the energy efficiency in the heat treatment furnace, Single End Radiant Tube burners (SET) were used to replace Vtube combustion systems. The energy efficiencies of the SET and V-tube combustion systems were experimentally verified in two full-scale working furnaces. This project carried out a quantitative analysis of the combustion and heat transfer processes in the radiation tubes and the heating chamber using Computational Fluid Dynamics (CFD) simulations. One of the main objectives of the project was to optimize the design of the furnace heating system with the aid of validated CFD simulations. Experimental data obtained in the full-scale working furnaces were used for the validation of the CFD simulations and to provide the boundary conditions for the CFD cases. Experimental instruments were installed in two heat treatment furnaces to measure the fuel flow rates in the radiation tubes, their surface temperatures and the temperature distribution inside the heat treatment furnaces. The fuel flow rates were used to determine the energy efficiencies of the heat systems and the fuel inlet conditions for the CFD cases. The measurements of the outer surface temperatures of the radiation tubes were used to determine an average temperature as the boundary conditions for CFD simulations of the combustion process inside the SET burner. The temperature distributions inside the furnace heating chamber were examined by measuring the temperature at specific points using a thermocouple matrix. This temperature measurement provided data to validate the CFD simulations of heat transfer inside the furnace heating chamber. Three series of CFD simulations were carried out in this project. The cases in the first series of CFD simulations were based on the SET burner. The flow mixing, combustion and heat transfer process in the SET burner were analyzed in the baseline CFD case, and the influence of radiation models on the CFD simulations were investigated. The design parameters, such as the effect of the burner diameter, were also verified in the baseline CFD studies. Using the baseline case as a reference, a numerical study was carried out to explore the applications of advanced combustion technologies, such as high temperature air combustion (HiT AC) and two-stage combustion with preheated air in the SET. The results of the CFD simulations were used to determine the heat flux rates through the SET burner wall into the heating chamber, which were used as the boundary conditions for the CFD simulations of the heating chamber in the second series of CFD cases. A good agreement was found between the numerically predicted temperatures at specified points inside the furnace heating chamber and the experimentally measured temperatures at the same points. This demonstrated that the heat flux rates from the SET burners can be applied in CFD-aided design to optimize the operational conditions of similar or super-size furnaces with confidence. Finally, a case study of CFD-aided design was carried out in the third series of CFD simulations of the heat transfer process inside the super-size furnace chamber. CFD simulations were used to verify if six SET burners are sufficient to provide the required temperature distribution inside the chamber and provide optimum locations for the SET burners.

Wind turbines are widely used for electricity generation. Typically turbines are deployed in farms located either on-shore or off-shore. In these arrangements the flow onto a turbine may be turbulent due to the disruption caused by turbines located further upwind. At onshore locations, turbines are typically smaller but will often be located downwind of structures or terrain which will cause the incident flow to be turbulent. Although wind turbines have been employed commercially for several decades, design tools are based on assumptions of quasi-steady flow and the effect of turbulence on turbine performance is not fully understood. In this study the effects of turbulent flow on wind turbine loading and performance were investigated by means of some sophisticated experimental methods in conjunction with numerical predictions. With this intention, the atmospheric boundary layer was simulated using conventional methods within the wind tunnel in the University of Manchester. The characteristics of the flow were established using cross hot-wire anemometry. The maximum thickness for the simulated atmospheric boundary layer that was produced by an arrangement of a combination of vortex generators, a barrier wall and a group of cubes was found to be over 0.7m. This combination sustained the turbulence intensity to between 3% and 23% and the turbulence length scale between 150mm and 210mm for the downstream flow. Meanwhile, the grid turbulence generator produced a turbulent flow at a cross section a distance of five mesh sizes downstream, with 16% turbulent intensity and with 35mm turbulent length scale across the entire cross section. These flow fields were experienced by a designed 2-Dfoil (chord = 60mm, span = 400mm, 40000 < Re < 75000) with the profile NACA4705 alongside a reference flow with no upstream element. These flow conditions were employed to quantify the effect of turbulence characteristics on lift and drag coefficients of the aerofoil prior to implementing the test case in a rotating frame. In addition, numerical simulations were conducted in order to corroborate the results obtained in the 2-D experiment. Further to this, the experiments were carried out on a rotating frame to observe how the turbulent characteristics of the flow might alter the performance of the miniature wind turbine. The blade Reynolds number in the rotor experiments is less than 105 and so considerably reduced from the Reynolds of a full-scale wind turbine. However, since the boundary layer is turbulent the effect of onset turbulence is expected to be representative. The turbine performance was then supported by implementing the Blade Element Momentum theory in the MATLAB environment. In conclusion, the results confirmed that the unsteadiness in the upstream flow associated with the high level of turbulent intensity can enhance the power coefficient of the turbine as a result of increasing the ratio of lift over drag coefficients. However, the large turbulent length scale can substantially diminish the power coefficient.