Surface integrity (SI) is the integrated surface behavior and condition of a material after being modified by a manufacturing process; it describes the influence of surface properties and characteristics upon material functional performance. As the leading-edge field of manufacturing research, SI finishing/machining and the consequent machining-induced complex combination of surface roughness, residual stress, work-hardening, macro and microstructure transformation, strongly affect the fatigue and stress behavior of machined parts. This kind of influence is particularly sensitive and pronounced in the difficult-to-machine materials, which are typically chosen for the most critical applications in the automobile, aerospace and nuclear industry. Thus, well-designed SI processing requirement and accurate SI evaluation model are essential to control and ensure the surface quality and functional performance for these key parts. In this thesis, an SI descriptive model for quantitative characterization and evaluation of surface integrity is proposed based on five principal SI characteristics. Considering the nature of surface integrity, a conceptual framework of an SI model for machined parts is established, in which the SI model is constructed based on the correlations between SI manufacturing processes, SI characteristics and final functionality. This model offers a theoretical basis and guideline for controlling SI characteristics and improving fatigue properties for machined parts. An empirical model for estimating the SI-characteristics-caused effective stress concentration factor (SCF) is established with fatigue life as the evaluating indicator. For a typical difficult-to-machine material, GH4169 superalloy, usually used in internal combustion engines, its grindability and the influence of processing parameters on the five principal SI characteristics are investigated in detail. The correlations between the processing parameters and the SI characteristics, between the processing parameters and the fatigue properties, and between the SI characteristics and the fatigue properties, are analyzed based on an orthogonally-designed grinding experiment and corresponding rotary bending fatigue testing for GH4169 samples within the selective range of grinding processing parameters. The feasibility and effectiveness of the proposed model for estimating the SI effective SCF are also validated by the experimental results, and this has actually offered an equivalent and convenient means for evaluation of SI and fatigue properties. Finally, the conclusions and contribution of the research are discussed, and potential future work to build on this research is identified.

Carbon nanomaterials are an active frontier of research in current nanotechnology. Single wall Carbon Nanotube (SWNT) is a unique material which has already found several applications in photonics, electronics, sensors and drug delivery. This thesis presents a summary of the author’s research on functionalisation of SWNTs, a study of their optical properties, and potential for an application in laser physics. The first significant result is a breakthrough in controlling the size of SWNT bundles by varying the salt concentrations in N-methyl 2-pyrrolidone (NMP) through a salting out effect. The addition of Sodium iodide leads to self-assembly of CNTs into recognizable bundles. Furthermore, a stable dispersion can be made via addition polyvinylpyrrolidone (PVP) polymer to SWNTs-NMP dispersion, which indicates a promising direction for SWNT bundle engineering in organic solvents. The second set of experiments are concerned with enhancement of photoluminescence (PL), through the formation of novel macromolecular complexes of SWNTs with polymethine dyes with emission from enhanced nanotubes in the range of dye excitation. The effect appears to originate from exciton energy transfer within the solution. Thirdly, SWNT base-saturable absorbers (SA) were developed and applied to mode locking of fibre lasers. SWNT-based SAs were applied in both composite and liquid dispersion forms and achieved stable ultrashort generation at 1000nm, 1550nm, and 1800 nm for Ytterbium, Erbium and Thulium-doped fibre laser respectively. The work presented here demonstrates several innovative approaches for development of rapid functionalised SWNT-based dispersions and composites with potential for application in various photonic devices at low cost.

No description available.

No description available.

Depicting the transitions where differing material textures meet on a terrain surface presents a particularly unique set of challenges in the field of real-time rendering. Natural landscapes are inherently irregular and composed of complex interactions between many different material types of effectively endless detail and variation. Although consumer grade graphics hardware is becoming ever increasingly powerful with each successive generation, terrain texturing remains a trade-off between realism and the computational resources available. Technological constraints aside, there is still the challenge of generating the texture resources to represent terrain surfaces which can often span many hundreds or even thousands of square kilometres. To produce such textures by hand is often impractical when operating on a restricted budget of time and funding. This thesis presents two novel algorithms for generating texture transitions in realtime using automated processes. The first algorithm, Feature-Based Probability Blending (FBPB), automates the task of generating transitions between material textures containing salient features. As such features protrude through the terrain surface FBPB ensures that the topography of these features is maintained at transitions in a realistic manner. The transitions themselves are generated using a probabilistic process that also dynamically adds wear and tear to introduce high frequency detail and irregularity at the transition contour. The second algorithm, Dynamic Patch Transitions (DPT), extends FBPB by applying the probabilistic transition approach to material textures that contain no salient features. By breaking up texture space into a series of layered patches that are either rendered or discarded on a probabilistic basis, the contour of the transition is greatly increased in resolution and irregularity. When used in conjunction with high frequency detail techniques, such as alpha masking, DPT is capable of producing endless, detailed, irregular transitions without the need for artistic input.

A large and complex structure such as a car body-in-white typically consists of several major components that are produced from thin metal sheets. The components are joined together by different types of mechanical joints such as rivets, spot welds and bolted joints. These mechanical joints are highly influenced the overall dynamic behaviour of structures. The finite element method has been widely used for predicting the dynamic behaviour of structures. However, to model the local effects (such as slip, loosing, or clearance effect) arising from the joints is cumbersome and time consuming. Moreover, the predicted model is often found to be inconsistent with the measured data. The discrepancies are believed to be arisen from the invalid assumptions about the model and properties data of the initial finite element model. This thesis puts forward the idea of using a simple and practical bolted joint modelling considering local effects of the area of the bolted joints of thin metal sheet structures. CFAST element and initial stress ratio are used in modelling the bolted joints and the local effects of the mating area between bolt, washer and surface of the structure. The properties of the parameters of the CFAST element and the initial stress ratio are used in the model updating procedure. The advantage of this technique allows the local effects of the bolted joints to be modelled in a simple way and it proved to be successful in modelling bolted joints. The influence of the stiffness of suspension springs which is used in simulating free-free boundary conditions in the experimental work especially for the structure that is made from thin metal sheet is investigated as well in this thesis. In the investigation, CBUSH element is used to model the suspension springs and the stiffness of the spring is taken into account as the updating parameter for the finite element model of the full welded structure with free-free boundary conditions. In this thesis, the use of the simple and practical modelling is adopted in the development of the finite element model of the full welded structure that consists of ten components made from thin metal sheets joined by spot welds and bolted joints. The model is updated using the results obtained from the experiments via the application of two model updating methods. They are iterative method and response surface method (RSM). In the iterative method, the NASTRAN SOL200 is used to improve the finite element models of the components and of the welded structures. The work in the iterative method is divided into two parts. The first one, the finite element models of components are updated in order to reduce the discrepancies of the natural frequencies before they are assembled together. Meanwhile, the second part is the updating process of the finite element models of the welded structures by concentrating on joint modelling and updating. Finally, the response surface method (RSM) is used in updating the model of the full welded structure with fixed boundary conditions due to bolted joints. The Latin hypercube sampling is used to generate numerical samples. The accuracy and efficiency of both methods (iterative method and RSM) are presented and discussed.

Flight dynamics analysis using computational models is a key stage in the design of aircraft. The models used in industry consist of two main parts. The first is a tabular aerodynamic model which is essentially a large database of aerodynamic data. The tabular aerodynamic model is a highly dimensional database containing aerodynamic loads and moments for different parameter combinations. In order to reduce the size of the tables, a number of assumptions are made. These include having sufficient resolution of the parameter space to capture the variation in the flow dynamics; decoupling certain parameters to reduce the dimensionality; using a single dynamic derivative, assuming independence from the flow conditions; and finally neglecting flow history effects which are dominant during manoeuvres with highly unsteady flow phenomena. Secondly is the use of dynamic derivatives to simulate unsteady motion effects. These are calculated using small--amplitude forced oscillatory motions. In order to accelerate their computation, frequency domain methods are used. The Linear Frequency Domain and Harmonic Balance are two such methods used in this work. As part of the frequency domain calculations, linear solvers are used to provide solution to the frequency domain problem. These solvers use preconditioners to accelerate the time to solution. An alternative method of preconditioning is proposed in this work based on the first and second order spatial discretisation Jacobian matrices. It is shown that there is significant speed up achieved by varying the proportions of the first and second order terms in the preconditioner matrix. In order to assess the performance of the tabular models, an initial assessment is carried out using a hierarchy of manoeuvres of increasing complexity. For each test case, the replay from the tabular model is compared with the fully unsteady time--accurate CFD solution. This is in line with a framework proposed in the literature. It is shown that the tabular model performs well through the linear aerodynamic regime, although breaks down where history effects become significant. The assessment continues with a study of each of the assumptions used to formulate the tables. Again a hierarchy of test cases of increasing complexity is used. Also used are both forced and free--response manoeuvres. It is shown that the resolution and coupling assumptions have little impact on the performance of the tabular model. The use of a single dynamic derivative is not shown to have an impact either, although it is suggested that for more complicated manoeuvres, this could be important. Finally, the most significant error is introduced through neglecting history effects. It is shown that for manoeuvres where history effects dominate, such as those at the extremes of the flight envelope, the tabular model is not sufficient to effectively model the aerodynamics during these manoeuvres.

Automatic segmentation of anterior segment optical coherence tomography (AS OCT) images provides an important tool to aid management of ocular diseases. Having precise details about the topography and thickness of an individual eye enables treatments to be tailored to a specific problem. OCT is an imaging technique that can be used to acquire volumetric data of the anterior segment of the human eye. Fast automatic segmentation of this data, which is not available, means clinically useful information can be obtained without the need for time consuming error-prone manual analysis of the images. This thesis presents newly developed automatic segmentation techniques of OCT images. Segmentation of 2D OCT images is first performed. One of the main challenges segmenting 2D OCT images is the presence of regions of the image that generally have a low signal to noise ratio. This is overcome by the use of shape based terms. A number of different methods, such as level set, graph cut, and graph theory, are developed to do this. The segmentation techniques are validated by comparison to expert manual segmentation and previously published segmentation techniques. The best method, graph theory with shape, was able to achieve segmentation comparable to manual segmentation. Good agreement is found with manual segmentation for the best 2D segmentation method, graph theory with shape, achieving a Dice similarity coefficient of 0.96, which is comparable to inter-observer agreement. It performed significantly better than previously published techniques. The 2D segmentation techniques are then extended to 3D segmentation of OCT images. The challenge here is motion artefact or poor alignment between each 2D images comprising the 3D images. Different segmentation strategies are investigated including direct segmentation by level set or graph cut approaches, and segmentation with registration. In particular the latter requires the introduction of a registration step to align multiple 2D images to produce a 3D representation to overcome the presence of involuntary motion artefacts. This method produces the best performance. In particular, it uses graph theory and dynamic programming to segment the anterior and posterior surfaces in individual 2D images with shape constraint. Genetic algorithms are then used to align 2D images to produce a full 3D representation of the anterior segment based on landmarks or geometric constraints. For the 3D segmentation, a data set of 17 eyes is used for validation. These have each been imaged twice so a repeatability measurement can be made. Good repeatability of results is demonstrated with the 3D alignment method. A mean difference of 1.77 pixels is found between the same surfaces of the repeated scans of the same eye. Overall, a new automation method is developed that can produce maps of the anterior and posterior surfaces of the cornea from a 3D images of the anterior segment of a human eye. This will be a valuable tool that can be used for patient specific biomechanical modelling of the human eye.

In the last decades, we have assisted to an important expansion of the number of indicators for measuring the development of a country− from the GDP per capita, households’ consumption indicators, demographic and medical indicators, schooling rates to governance indexes. This has produced in a first time the development of composite indicators to explain and synthesise the spatial and temporal changes of these different indicators− the Human Development Index (HDI) and its adjusted versions, Multidimensional Poverty Index (MPI), or the Water Poverty Index (WPI), to provide policy makers simple figures to help them in their decisions. The main difficulty faced by the researchers was to explain complex behaviours through single indicators. This research develops a framework to explain and contribute to the better understanding of the relationships between the existing single and complex indicators in the domain of Water Supply and Sanitation (WSS) in Developing Countries. This framework is based on the Bayesian Networks modelling method (Castelletti & Soncini-Sessa, 2007a), (Giné Garriga et al., 2009), (Dondeynaz et al., 2013). In addition to building this analytical framework, this research also aims at measuring and analysing the distribution and the influence of Official Development Assistance (ODA) in recipient countries. The approach chosen is global, targeting cross-countries analysis and comparison to capture the principal key variables of water supply and sanitation coverage expansion and its benefits for the country development. Therefore, this research proposes a methodological framework using Bayesian models for analysing water supply and sanitation access levels together with governance, human development (education, health, and income), water resources, the uses of these resources and the ODA. The research outputs could support national decision making and/or donors’ strategies, in particular the European Union. Variables and data are collected at national country scale for 101 developing countries observations in a new database (WatSan4dev) for year 2004. Five country profiles are identified and ranged around five main thematic axes using multivariate and clustering analyses. The countries from profiles 4 and 5 were the least favoured in terms of development and access to WSS, therefore should benefit from ODA support. However, countries from profile 5 received rather low ODA inputs in 2004, possibly as shown from the models because of their relative instability and poor governance. The modelling approach is led by the principles of robustness and replicability and took into account data availability and nature using Bayesian Networks. It is found that WSS access is strongly associated to country development (+35 % probability change) that is first sensitive, as expected, to the income level. The urbanisation level is the second strong factor associated to development with the limit of slums development. Health care and advanced governance complete these key factors. Lastly, WSS is sensitive to ODA CI where high-level ODA is estimate to benefit first to poor (45%) and middle (34%) development countries at 79% probability. This modelling allowed, in addition, running probabilistic scenarios to test hypotheses and measure the probable changes on WSS and the development. The methodological process, the outputs of multivariate analysis, the five countries profiles, the Bayesian modelling as well as examples of scenarios are described and analysed. The reference date is first 2004. The analytical and modelling process is then applied to the 2000-2008 period.

The work carried out during this project used a computational Fluid Dynamics code to generate aerodynamic tabular models and aircraft manoeuvre simulations. As an outcome of this work, a validation of the aerodynamic prediction tools and an assessment of tabular models for aircraft flight dynamics applications was made. The Stability and Control Unmanned Combat Air Vehicle has been used as a demonstration case. Validation of computational fluid dynamics methods was carried out for highly nonlinear flow topologies using wind tunnel measurements. Integral data, pressure tap measurements and particle image velocimetry information was compared against the predictions over two configurations. Each one had a different leading edge shape distributed along the span of the model. One was sharp throughout with varying leading edge thickness and the other one was mainly rounded. Results showed a good agreement in longitudinal force and moment predictions for low angles of attack. High angles were dominated by a double vortex structure which was very sensitive to incidence angle and leading edge shape. Some wind tunnel effects were noticed in the measurements when predictions were made with and without sting. Overall the numerical predictive capabilities for low and high angles of attack were deemed good for the purpose of flight dynamics model generation. Two methods for predicting manoeuvering flight aircraft loads are presented in this thesis. A tabular aerodynamic model based on numerical predictions was generated for the sharp configuration. Kriging interpolation was used to populate a model consisting of tables of lateral and longitudinal aerodynamic characteristics. Further to this, longitudinal dynamic derivatives were predicted for the test case in hand using forced oscillation numerical predictions. Aircraft geometric characteristics were approximated based on real aircraft data. A set of controls were designed and implemented for the purpose of manoeuvering flight predictions. A code was implemented to predict realistic aircraft manoeuvres based on an existing program. At the core of this method was a commercial optimisation Matlab code called DIDO. Using this and the nonlinear, six degree of freedom equations of motion, purposedly designed aircraft manoeuvres were predicted. The motions were then replayed using time-accurate simulations and the predicted loads were compared against the tabular predictions. In this manner, the validity of the tables of aerodynamic data were benchmarked against a more reliable and expensive numerical method. The static based predictions showed good agreement with the replays for slow manoeuvres at low angles of attack. As manoeuvres became more aggressive, noticeable disagreement was present in the aircraft loads, particularly in the lateral characteristics during periods of large rates of change in attitudes. Hysteresis effects during manoeuvering flight were seen to produce large spreads in data in the angle of attack domain which the predicted dynamic derivatives were unable to capture.