P91 and P92 steels are two ASTM grades of steel which have been used in high temperature applications, such as fossil-fuelled power stations, nuclear power stations and chemical plants. Operating under creep conditions, i.e. high temperature and/or high stress, the welds made from these steels are potential failure locations and, therefore, life limiting for the entire plant. In this thesis, the results of creep and creep crack growth (CCG) tests, which were carried out on P91 and P92 welds, are reported. These tests were carried out on welds constituents, i.e. parent materials (PM) and weld metal (WM), and across-welds, for the P91 material at 650°C and for the P92 material at 675°C. For the cross-weld tests, interest was focused on the Type IV region, a narrow zone at one end of the heat affected zone (HAZ) at the side of the PM. Also reported, in this thesis, are the results of the Finite Element analyses for predicting the creep and creep crack growth in the P91 and P92 materials. For the P91 material, CCG tests were carried out on PM and cross-weld CT specimens, at 650°C. The results of the CCG tests for P91 show that the CCG rates for the cross-weld CT specimens are about ten times higher than those for the PM specimens. For P92, uniaxial and notched bar creep tests were carried out on the PM and WM, at 675°C. Cross-weld uniaxial, notched bar and waisted specimens were also tested in order to characterise the creep behaviour of the P92 HAZ material. Impression creep tests were also carried out on the P92 HAZ material. The results of the CCG tests for P92 show that the CCG rates for the cross-weld CT specimens are about two times higher than these for the PM CT specimens. The fracture mechanics parameter, C*, was used to correlate the CCG rates in the P91 and P92 CT specimens. The load line displacement rates and the CCG data, for the CT specimens, were used to calculate C* values according to ASTM E 1457-00 (2001). The Reference Stress was also found to correlate the CCG rates in the P91 CT specimens. Further, FE analyses were carried out to obtain the values of C*, based on the steady state value of the contour integral C(t). Stationary crack and growing crack CT models were used to obtain the C* values. Damage mechanics theory and equations were used to predict the creep and CCG for the P91 and P92 materials using the commercial FE package, ABAQUS. Both the Kachanov and the Liu and Murakami damage models were used. In order to use these models, material properties have to be determined. The results of creep and creep rupture tests were used to determine those properties. It was found that both damage models could be used to predict the creep behaviour of the tested materials. However, the Liu and Murakami model was favoured over the Kachanov model in predicting the CCG in the CT specimens. The effect of the material multiaxiality is also highlighted.

Wind borne debris is one of the principal causes of building envelope failure during severe storms. It is often of interest in windstorm risk modelling to estimate the potential flight trajectories and impact energy of a piece of debris. This thesis presents research work aimed at the development and validation of a numerical model for the simulation of plate-type windborne debris. While a number of quasi-steady analytical models are available at present, these models are unable to account for the fluid-plate interaction in highly unstable flows. The analytical models are also limited to simple launch flow conditions and require extensive a-priori knowledge of the debris aerodynamic characteristics. In addition, the use of Euler angle parametrisations of orientation in the analytical models results in mathematical singularities when considering 3D six degree-of-freedom motion. To address these limitations, a 3D Computational Fluid Dynamics (CFD) model is sequentially coupled with a quaternion based singularity-free six degree of freedom Rigid Body Dynamics (RBD) model in order to successfully simulate the flight of plate-type windborne debris. The CFD-RBD model is applied to the numerical investigation of the flow around static, forced rotating, autorotating and free-flying plates as well as the treatment of complex launch conditions. Key insights into the phenomena of plate autorotation are highlighted including the genesis of the aerodynamic damping and acceleration torques that make autorotation possible. The CFD-RBD model is then validated against measurements of rotational speed and surface pressure obtained from recent autorotation experiments. Subsequently a general 3D spinning mode of autorotation is demonstrated and the CFD-RBD model is extended to include plate translation in order to simulate windborne debris flight. Using the CFD-RBD flight model, a parametric study of windborne debris flight is carried out and four distinct flight modes have been identified and are discussed. The flight results are contrasted against available free-flight experiments as well as predictions from existing quasi-steady analytical models and an improved quasi-steady force model based on forced rotation results is proposed. The resulting CFD-RBD model presents the most complete numerical approach to the simulation of plate-type windborne debris, directly simulating debris aerodynamics, and incorporates complex launch flow fields in the initial conditions.

Understanding the correlation between the mix proportions, micro structural characteristics, and macro-scale properties of concrete (i.e. the process-structure-properties relationship) is fundamental to achieving a more advanced understanding of how to apply and optimise this abundant engineering material. Although, concrete has been traditionally evaluated by its physico-mechanical and functional properties; development of advanced and effective inspection techniques during the last decade has demonstrated that fundamental macro-level properties of concrete depend, to a great extent, on its properties at the micro- and nano levels. This research was intended to make a quantitative assessment of impact load-induced micro-structural damage in concrete and, more particularly, to investigate the influence of ITZ micro-and nano local properties (as influenced by aggregate characteristics) on the impact load-induced cracking behaviour of concrete. Five different types of concrete mixtures were designed with the same total water cement ratio either by using natural aggregates as reference or by totally replacing the natural coarse aggregate with unconventional aggregates (such as copper slag, blue brick, sintered fly ash and tyre rubber) having significant differences in strength, shape and surface texture, porosity and roughness, and elasticity. A range of advanced techniques including X-ray diffraction, mercury porosimetry, 3D X-ray computed tomography coupled with digital image analysis, laser surface profilometry, 3D nanotech vertical scanning interferometry and scanning electron microscopy fitted with energy-dispersive X-ray spectrometer were used to characterize the aggregates and the concrete micro-structures. Based on the results obtained a possible mechanism for the micro-structural damage in concrete was proposed. Poorer aggregate characteristics alone could be responsible for a greater ITZ deterioration after loading but the results demonstrated that in fact, the aggregate causes a change in the ITZ conditions and it is these altered ITZ conditions that have a major effect on overall mix behaviour and govern the damaging process of concrete under impact loading. It was also concluded that the presence of a weak and porous ITZ has two opposite effects on the failure process. First, the chemical and porosity heterogeneities within the ITZ can cause fluctuations/disordering in the cracking (fracture) path, resulting in an increase in the tortuosity and corresponding fracture energy dissipation. Second, a weak and porous ITZ transfers less stress from the matrix to the aggregate particles. This leads to a lower compressive strength but increased toughness due to micro crack path lengthening and energy dissipation. Finally, the effect of the aggregate on the surface area roughness of the ITZ was established for the first time in the concrete literature. The roughness number of the area near the ITZ was found to positively correlate with dissipated surface fracture energy. An increase in the roughness number is associated with an increase in the dissipated fracture energy. The significance of this correlation however, lies in the fact that the rougher near – ITZ fraction of the bulk paste is more resistant to cracking at the macro level. Findings from this study will lead to a better understanding of the impact load-induced micro-structural damage phenomena. In addition, the micro-structural data from SEM and X-ray CT obtained during impact and mechanical testing of the concrete mixtures could be used to develop a multi-scale finite element model to simulate and predict the behaviour and fracture damage of concrete subjected to dynamic loading.

The thermo-mechanical fatigue (TMF) of power plant components is caused by the cyclic operation of power plant due to startup and shutdown processes and due to the fluctuation of demand in daily operation. Thus, a time-dependent plasticity model is required in order to simulate the component response under cyclic thermo-mechanical loading. The overall aim behind this study is to develop a material constitutive model, which can predict the creep and cyclic loading behaviour at high temperature environment, based on the cyclic loading test data of the P91 and the P92 steels. The tests on all specimens in the study were performed using the Instron 8862 TMF machine system with a temperature uniformity of less than ±10°C within the gauge section of the specimen. For the isothermal tests on the P91 steel, fully-reversed, strain-controlled tests were conducted on a parent material of the steel at 400, 500 and 600˚C. For the P92 steel, the same loading parameters in the isothermal tests were performed on a parent material and a weld metal of the steels at 500, 600 and 675°C. Strain-controlled thermo-mechanical fatigue tests were conducted on the parent materials of the P91 and the P92 steels under temperature ranges of 400-600°C and 500-675°C, respectively, with in-phase (IP) and out-of-phase (OP) loading. In general, the steels exhibit cyclic softening behaviour throughout the cyclic test duration under both isothermal and anisothermal conditions. The cyclic softening behaviour of the P91 steel was further studied by analyzing stress-strain data at 600°C and by performing microstructural investigations. Scanning electron microscope (SEM) and transmission electron microscope (TEM) images were used to investigate microstructural evolution and the crack initiation of the steel at different life fractions of the tests. The TEM images of the interrupted test specimens revealed subgrain coarsening during the cyclic tests. On the other hand, the SEM images showed the initiation of microcracks at the end of the stabilisation period and the cracks were propagated in the third stage of cyclic softening. A unified, Chaboche, viscoplasticity model, which includes combined isotropic softening and kinematic hardening with a viscoplastic flow rule for time-dependent effects, was used to model the TMF behaviour of the steels The constants in the viscoplasticity model were initially determined from the first cycle stress-strain data, the maximum stress evolution during tests and the stress relaxation data. Then, the initial constants were optimized using a least-squares optimization algorithm in order to improve the general fit of the model to experimental data. The prediction of the model was further improved by including the linear nonlinear isotropic hardening in order to obtain better stress-strain behaviour in the stabilisation period. The developed viscoplasticity model was subsequently used in the finite element simulations using the ABAQUS software. The focus of the simulation is to validate the performance of the model under various types of loading. Simulation results have been compared with the isothermal test data with different strain ranges and also the anisothermal cyclic testing data, for both in-phase and out-of-phase loadings. The model’s performance under 3-dimensional stress conditions was investigated by testing and simulating the P91 steel using a notched specimen under stress-controlled conditions. The simulation results show a good comparison to the experimental data.

Vehicle bodies are generally constructed from galvanized steel, which, together with phosphate and e-coat paint treatments, ensures corrosion resistance. The use of these materials alone cannot provide adequate corrosion protection to certain features that are inherent to vehicle body construction but are also vulnerable to corrosion, such as cut edges of panels and creviced joints. The use of further corrosion protection measures, (e.g. sealers, lacquers and waxes), is undesirable because they require additional manufacturing processes, increase weight and reduce recyclability of the vehicle. The potential benefits of using zinc-magnesium alloy coated steel (ZMG) as a substitute for conventional galvanized steel were investigated in this work. Cyclic corrosion testing in sodium chloride and acid rain-based environments was conducted on panels of ZMG and conventional galvanized steel and the resistance of each material to red rust initiation and propagation was assessed. ZMG offered approximately a 3-fold improvement in red rust resistance compared to galvanized steel in the sodium chloride test but ZMG's corrosion benefit was attenuated in the acid rain environment. Cyclic corrosion testing was also conducted on painted test panels incorporating geometric features; enhanced edge and crevice corrosion resistance was also observed for panels constructed from ZMG. Corrosion products formed in each environment were characterized using a suite of analysis techniques and mechanisms to explain the enhanced corrosion resistance of ZMG were proposed based on these products and on the literature. An inhibiting corrosion protection mechanism was suggested for ZMG whereby cathodic activity was retarded via the precipitation of insulating, sparingly soluble magnesium hydroxide. Further inhibition of cathodic activity has been attributed to the specific oxide layer (possibly magnesium oxyhydroxide doped with zinc) present at the ZMG surface. The observed efficacy of the corrosion protection mechanisms suggests that ZMG may allow improvement of the vehicle body corrosion protection system for vehicle weight and recyclability targets.

Two novel approaches to particle size measurement are investigated; these are designated as Particle Movement Displacement Distribution (PMDD) method and Separated Multiple Image Technique (SMIT). An advantage of these methods compared with the established particle sizing methods of Static Light Scattering (SLS) and Dynamic Light Scattering (DLS) is that PMDD and SMIT do not suffer from the intensity weighting problem that affects SLS and DLS. The performance of the PMDD method is examined through computer simulations and through analysis of pre-existing experimental data. The SMIT method is investigated through computer simulations and through the construction and use of an optical system. The ability of both methods was measured through the assessment of an ‘area error’ measure which gives an estimate of the accuracy of a recovered particle size distribution. This area error measure varies between 0 and 2; with 0 corresponding to a perfectly recovered distribution. Typically a good inversion of DLS data can achieve an area-error value of 0.32 to 0.34 and this figure (along with the recovered mean particle size and standard deviation of the distribution) was used to judge quantitatively the success of the methods. The PMDD method measures the centre of individual particles in each image. A vector histogram is formed based on the connection between the centres in the first image and the centres in the next image. This vector histogram contains information about the particle size distribution. A maximum likelihood data inversion procedure is used to yield a particle size distribution from this data. The SMIT method is similar to the Static Light Scattering (SLS) method, but it combines angular dependent intensity method and individual visualisation method together to recover individual particle sizes without an intensity weighting. A multi-aperture mask and wedge prisms are utilised in this method to capture particle images formed from light scattered into a number of selected directions. A look-up table is then used to recover the individual particle sizes, which are then formed into a histogram. For the PMDD method, computer simulation results established the optimum values for parameters such as the time interval between frames, the exposure time and the particle concentration and also investigated the effects of different noise sources. For mono-modal size distributions, the PMDD method was shown through computer simulation to be capable of recovering a particle size distribution with an area error of around 0.27 which compares well with the typical DLS result. PMDD results were also recovered from mono-modal experimental data with mean particle sizes close to the manufacturers quoted particle mean size. However, recovery of bi-modal distributions was found to be not so successful; for bi-modal distributions, the recovered distributions generally had only a single peak, which, of course gives a very poor area-error figure. This result compares poorly with the particle tracking method ‘Nano Particle Tracking Analysis’ which is able to recover bi-modal distributions. For this reason further research was concentrated on an image intensity method (SMIT). For the SMIT method, computer simulation results established the optimum values for parameters such as the particle concentration and also investigated the effects of different noise sources and of aberrations in the optical system. The SMIT method was shown through computer simulation to be capable of recovering particle size distributions for both mono-modal and bi-modal distributions. The area error values obtained were in the range 0.24 to 0.45, and most of the results are good compared to the DLS value. The major problem with the SMIT method was found to be the presence of a small number of recovered particle radii much larger (or smaller) than the true sizes. These errors were attributed to ambiguities in the look-up table system used to convert the relative intensity data values into particle sizes. Two potential methods to reduce the influence of these ambiguities were investigated. These were, firstly by combining Brownian motion movement data from tracking individual particles over a few frames of data, and secondly by combining an estimate of the total scattered intensity from a particle with the normal SMIT data to constrain the look-up procedure. In computer simulations both approaches gave some improvement but the use of the total scattered intensity method gave the better results. In a computer simulation this method managed to improve the area-error from 0.37 for SMIT alone to 0.25 for SMIT combined with this extra information. Based on the success of these computer simulation results, an experimental SMIT system was constructed and tested. It was found necessary to first calibrate the optical system, to account for the different optical transmission coefficients of the different prisms/optical paths. But using a particle sample with particles of known size to calibrate; other particle sizes were successfully recovered from the experimental data using the original SMIT data processing. A majority of the recovered particle radius were close to the manufacturers quoted particle mean radius. Attempts to implement the total intensity approach to enhance the SMIT were found not be successful due to the difficulty in measuring the small displacements in particle positions required with sufficient accuracy. A possible alternative design to overcome this problem is suggested in the future work section 7.2.

This thesis describes the investigation and development of damage modelling for composites materials at their micro-scale (e.g. fibre, matrix). A damage model for elastic materials, based on a "local" damage approach, has been introduced to predict failure onset and simulate the post-failure behaviour of unidirectional threedimensional representative volume elements (RVE) or unit cells with hexagonal distribution of the fibres over the cross section. The damage model consists of three parts: an elastic model, a failure criterion and the post-failure behaviour. Modifications of von Mises criteria and Maximum Principal Stress criterion have been considered to evaluate failure in the matrix whilst for the fibre in general the Maximum Principal Stress criterion has been used. The damage model has been implemented into the commercial code ABAQUS with subroutines in FORTRAN (UMAT and USDFLD). The material properties in the residual stress analyses are considered temperature dependant to simulate the volumetric contraction during the manufacturing process. Hence, the overall residual stress introduced from curing was determined by considering two ontributions: volume shrinkage of matrix resin from the crosslink polymerization during isothermal curing and thermal contraction of both resin and fibre as a result of cooling from the curing temperature to room temperature. Finally, three different typologies of 3D unit cells have been investigated. The first kind of micro-model is based on a symmetric distribution of the fibres and the unit cells have two phases, i.e.: matrix and fibre. The second typology of unit cells is still based on a uniform architecture but include a three-dimensional interphase between fibre and matrix. As in real composites at their constituent level fibres are randomly distributed. The mutual distance between fibres represents a critical factor for the ultimate mechanical properties of the micro-composites. Hence the last kind of micro-models account for this non-uniform position of fibres within the RVE although they consists of only two phases. FEM analyses have indicated that predicted damage initiation and evolution are clearly influenced by the presence of residual stresses in all the three different typologies of unit cells analysed. The numerical analyses on the numerical models have proved that, in general, the overall mechanical properties are strongly influenced by the presence of residual stress, fibre volume fraction, fibre distribution and interphasial properties. In particular on transverse tensile loading, residual stresses produces beneficial results in terms of ultimate strength while in the case of longitudinal loading (parallel to the fibres) the matrix, due to the high compressive stress, undergoes a premature failure although.

Thermal barrier coatings (TBCs) are usually applied on high temperature gas turbine components. They reduce the need for additional cooling of the exposed surfaces and improve the durability of the underlying materials. However, the lack of reliable lifting methods limits their applicability in the design of turbine components and so they are usually employed as additional protection for components that already meet the design requirements. In order to develop failure models and equations of practical interest, the mechanical behaviour and degradation of properties of coatings at elevated temperature needs to be understood. Several phenomena such as the growth of an oxide layer, degradation of bond coats, creep and thermal expansion mismatch between the different layers that compose the TBC contribute in the development of stresses at high temperature. The effect of thermal cycling has been covered in previous research, giving rise to models that explained how accumulated cyclic inelastic strains occurred in the bond coat and oxide layer due to the thermal expansion mismatch. This favoured the wrinkling of the oxide layer and the concentration of stresses, which could eventually cause crack nucleation, growth and failure of the coating. The research contained in this thesis focuses mainly on the development of stress concentrations during high temperature exposure. A coupled micro-structural-mechanical constitutive model was implemented in order to take into account the processes the coatings undergo at high temperature. High tensile stresses, perpendicular to the oxide-top coat interface, which may induce crack nucleation within the oxide layer at high temperature, were obtained.

Chalcogenide glasses offer transmission from the far visible to the far - infrared (IR) wavelength range. They exhibit photosensitivity and have high linear and nonlinear refractive indices. There are many potential applications involving near- and mid- infrared light such as laser delivery, optical data storage and all-optical switching. Two types of optical waveguides based on chalcogenide glasses were developed in this project: (1) The fabrication of planar optical waveguides in thin As40Se60 glass films was carried out via a hot embossing pressing technique. Previous work had shown it possible to fabricate optical waveguides in polymers using hot embossing techniques. Nevertheless using hot embossing to pattern waveguides in a thin chalcogenide glass film did not receive much success. In the present work, single-mode optical rib waveguides operating at telecommunication wavelengths were successfully patterned in a thermally evaporated As40Se60 glass thin film on a Ge17As18Se65 chalcogenide glass substrate. This experimental line demonstrated a fast and economic way of producing planar waveguides in thin chalcogenide glass films. (2) For the first time, a one-layer, solid micro-structured optical fibre (MOF) was successfully drawn from As40Se60 and Ge10As23.4Se66.6 (atomic %) chalcogenide glasses for operation in the near- to mid-infrared. This experimental line showed a new and flexible route to micro-structuring of mid-infrared fibre for operation in the near- to mid-infrared, presenting an all-solid (i.e. glass-glass) alternative to air-glass micro-structuring. The principal advantage of the new approach is mechanical rigidity. A sufficiently large refractive index step between the component glasses exists to enable structures that rely on photonic bandgap effects for their operation to be realised in future work. Underpinning the development of these optical waveguides, the refractive index dispersion of bulk chalcogenide glasses As40Se60, Ge10As23.4Se66.6 and Ge17As18Se65 was measured using ellipsometry from 0.3 μm to 2.3 μm wavelength in this project. Also, the refractive index of thin As40Se60 films was measured and compared with that of bulk As40Se60 samples. Finally a Se precursor purification process was developed to enhance the purity of the end-glasses.

Coal combustion by-products generated from coal-fired power plant and cause enormous problems for disposal unless a way can be found to utilize these by-products through resource recovery programs. The implementation of air act regulations to reduce NOx emission have resulted millions of tonnes of pulverised fuel ash (PFA) accumulated with high percentage of unburned carbon made it un-saleable for the cement industry. Moreover, alternative fuels such as biomass and import coals were suggested to reduce gas emissions but on the other hand PFA marketability was reduced. The main objective of this study was thus to utilise high carbon PFA into value added products. Through this work, the relationships beside the factors that could influence the carbon content in the PFA and reduce it in terms of producing raw material useful for different applications were explored. These factors were extensively investigated through thermogravimetric analyses, surface area measurements, microscopy and optical studies, and particle size distribution analyses. Five high unburned carbon PFAs were selected as feedstocks for PFA beneficiation, cement tests, and carbon activation. In order to beneficiate a high carbon PFA, incipient fluidisation was selected as the preferred route being a dry separation method which does not expose the carbon to potential contaminants that may alter its reactivity or physical properties. Enriched PFAs (i.e. depleted carbon) were separated and then cement tests were conducted in different mixture ratios (PFA/cement) throughout different time scales. These tests were demonstrated by using samples derived from biomass co-firing and import coals. The PFA/cement mixtures achieved good strength and workability via standard values. Unburned carbon (i.e. enriched carbon) streams were activated using steam at temperature 850 C and time from 60-300 minutes. For all unburned carbons investigated in this project, the surface areas of their activated counterparts increased to reach maximum level after three hours and four hours compared with other works. But this increase dropped back according to the reduction of the pore widening. Consequently, the surface area exhibited a high level of low carbon burn-out for the carbon sourced from biomass co-firing (1435 m2/g and 38 wt.%, respectively). This was revealed due to the carbon gasification and pore widening level. In addition, optical studies showed that the carbon types changed in a different manner during the activation.