The common profiled steel cladding systems used in Australia and its neighboring countries are made of very thin (0.42 mm) high strength steel (G550 with a minimum yield stress of 550 MPa) and are crest-fixed. However, these claddings often suffer from local pull-through failures at their screw connections during high wind events such as storms and cyclones. Past experience and researches have shown that the loss of steel roofs has often occurred due to local pull-through failures of their screw connections under uplift or suction loading. Loss of claddings always led to a progressive collapse of the entire building. This situation is continuing because of the lower priority given to the design of roof and wall cladding systems. At present, steel design codes do not provide guidelines for the design of crest-fixed steel roof or wall claddings. Past research has shown that European and American recommendations for steel claddings cannot be used for Australian crest-fixed cladding systems as they were developed mainly for valley-fixed claddings subjected to gravity loading instead of crest-fixed claddings subjected to wind uplift/suction loading. Therefore at present the design of thin steel cladding systems is based on laboratory tests and is expensive. These situations inhibit the innovative design and advances in the steel cladding industry. Since the local pull-through failures in the less ductile G550 steel claddings are initiated by transverse splitting at the fastener hole, analytical studies have not been able to determine the pull-through failure loads accurately. Therefore in the first stage of this research an appropriate fracture/splitting criterion was developed using a series of large scale and small scale experiments of crest-fixed steel claddings. A shell finite element model of crest-fixed steel cladding was then developed that included the new fracture/splitting criterion and advanced features such as hyperelastic material modelling, contact simulations, residual stresses and geometric imperfections. The improved finite element analyses were able to model the pull-through failures associated with splitting as evident from the comparison of their results with the corresponding full-scale experimental results. An extensive series of parametric studies considering the effects of material properties and geometric parameters of the two commonly used trapezoidal cladding profiles was undertaken using finite element analysis. Appropriate design formulae for the pull-through and dimpling failure load of trapezoidal profiles were then derived for optimization purposes and to simplify the current design method. This will then lead to modification and optimisation of cladding profiles to satisfy the requirements of both strength (safety during cyclones and storms) and economy. This thesis presents the details of large scale experimental studies undertaken and the results including the criterion for the splitting/fracture failure of high strength steel cladding systems. It describes the many advances made in the finite element modelling of crest-fixed steel cladding systems including the effects of localised pull-through and dimpling failures. Finally, it presents a simple design method for trapezoidal steel cladding systems under wind uplift or suction loading.

Operating temperature range is one of the primary design considerations for developing effective disc brake system performance. Very high braking temperatures can introduce effects detrimental to performance such as brake fade, premature wear, brake fluid vaporization, bearing failure, thermal cracks, and thermally-excited vibration [2]. This project is concerned with investigating deficiencies and proposing improvements in brake system Finite Element (FE) models in order to provide high quality descriptions of thermal behaviour during braking events. The work focuses on brake disc/pad models and the degree of rotational freedom allowed for the pad. Conventional models [10] allow no motion/or free motion of the pad. The present work investigates the effect on disc/pad interface temperature and pressure distributions of limited relaxations of this rotational restriction. Models are proposed, developed and validated that facilitate different rotational degrees of freedom (DoF) of the pad. An important influencing factor in friction brake performance is the development of an interface tribo-layer (ITL). It is reasonable to assume that allowing limited rotational motion of the pad will impact the development of the ITL (e.g. due to different friction force distributions) and hence influence temperature. Here the ITL is modelled in the numerical simulations as a function of its thickness distribution and thermal conductivity. Different levels of ITL thermal conductivity are defined in this work and results show that conductivity significantly a1qwffects interface temperature and heat partition ratio. The work is based around a set of test-rig experiments and FE model developments and simulations. For the experimental work, a small-scale test rig is used to investigate the friction induced bending moment effect on the pad/disc temperature. Significant non-uniform wear is observed across the friction surface of the pad, and reasons for the different wear rates are proposed and analyzed together with their effect on surface temperature. Following on from experiment a suite of models is developed in order to evidence the importance of limited pad motion and ITL behaviours. A 2D coupled temperature-displacement FE model is used to quantify the influence of different pad rotational degrees of freedom and so provide evidence for proposing realistic pad boundary settings for 3D models. Normal and high interface thermal conductance is used in 2D models and results show that the ITL thermal conductivity is an important factor influencing the maximum temperature of contact surfaces and therefore brake performance. The interface heat partition ratio is calculated by using the heat flux results and it is confirmed that this value is neither constant nor uniform across the interface surfaces. Key conclusions from the work are (i) that ITL thermal conductivity is an important factor influencing the interface temperature/heat flux distribution and their maximum values, (ii) that allowed motion of the pad significantly affects the interface pressure distribution and subsequently the temperature distribution, (iii) that the transient heat partition in friction braking is clearly quite different to the conventional friction-pair steady heat partition (the heat partition ratio is not uniformly distributed along the interface) and (iv) that the thickness of the ITL increases through braking events, reducing the heat transfer to the disc, and so providing a possible explanation for increasing pad temperature observed over the life time of a brake pad.

Quantitative and qualitative assessment of the seismic risk to highway bridges is crucial in pre-earthquake planning, and post-earthquake response of transportation systems. Such assessments provide valuable knowledge about a number of principal effects of earthquakes such as traffic disruption of the overall highway system, impact on the regions' economy and post-earthquake response and recovery, and more recently serve as measures to quantify resilience. Unlike previous work, this study captures unique bridge design attributes specific to California bridge classes along with their evolution over three significant design eras, separated by the historic 1971 San Fernando and 1989 Loma Prieta earthquakes (these events affected changes in bridge seismic design philosophy). This research developed next-generation fragility curves for four multispan concrete bridge classes by synthesizing new knowledge and emerging modeling capabilities, and by closely coordinating new and ongoing national research initiatives with expertise from bridge designers. A multi-phase framework was developed for generating fragility curves, which provides decision makers with essential tools for emergency response, design, planning, policy support, and maximizing investments in bridge retrofit. This framework encompasses generational changes in bridge design and construction details. Parameterized high-fidelity three-dimensional nonlinear analytical models are developed for the portfolios of bridge classes within different design eras. These models incorporate a wide range of geometric and material uncertainties, and their responses are characterized under seismic loadings. Fragility curves were then developed considering the vulnerability of multiple components and thereby help to quantify the performance of highway bridge networks and to study the impact of seismic design principles on the performance within a bridge class. This not only leads to the development of fragility relations that are unique and better suited for bridges in California, but also leads to the creation of better bridge classes and sub-bins that have more consistent performance characteristics than those currently provided by the National Bridge Inventory. Another important feature of this research is associated with the development of damage state definitions and grouping of bridge components in a way that they have similar consequences in terms of repair and traffic implications following a seismic event. These definitions are in alignment with the California Department of Transportation's design and operational experience, thereby enabling better performance assessment, emergency response, and management in the aftermath of a seismic event. The fragility curves developed as a part of this research will be employed in ShakeCast, a web-based post-earthquake situational awareness application that automatically retrieves earthquake shaking data and generates potential damage assessment notifications for emergency managers and responders. / Errata added at request of advisor and approved by Graduate Office, March 15 2016.

La machine à réluctance variable peut être utilisée dans les véhicules électriques où pour des considérations d’autonomie, le rendement est crucial. En raison du fort champ de fuite dans la région de l’entrefer de la machine à réluctance variable due à sa géométrie particulière à pôles saillants, les pertes « cuivre » peuvent devenir conséquentes. Il est alors recommandé de ne pas placer les conducteurs au voisinage de l’entrefer. Cependant, des instructions concrètes pour la conception d’un enroulement optimal sont manquantes. Généralement, les pertes « cuivre » dans les machines électriques sont la somme des pertes Ohm DC classiques et des pertes additionnelles dites par courants de Foucault. Les pertes DC étant constantes à un point de fonctionnement donné, l’optimisation est axée alors sur la réduction des pertes par courants de Foucault en jouant sur la configuration géométrique de l’enroulement. Dans le cas de calculs répétitifs fastidieux, rencontrés par exemple lors des processus de conception et d’optimisation du bobinage des machines électriques, il y a un intérêt significatif à réduire le temps de calcul. Dans ce travail, on présente trois techniques de réduction du modèle et leurs applications dans les analyses par la méthode des éléments finis. Outre l’influence de la fréquence d’alimentation et de la section du conducteur, plusieurs facteurs liés à la configuration de l’enroulement influent sur les pertes additionnelles par courants de Foucault :i) la position du conducteur dans l’encoche au voisinage de la dent du stator ou de la zone de l’entrefer .ii) la disposition des conducteurs envers les lignes du champ magnétique bidimensionnelles de l’encoche .iii) l’utilisation d’un conducteur massif ou multi filamentaire; les filaments sont connectés en parallèle et peuvent permuter leurs positions périodiquement au sein du conducteur tout au long du bobinage. Dans cette thèse, on étudie principalement l’influence de la disposition géométrique des spires dans l’encoche et du type du conducteur utilisé s’il s’agit d’un conducteur massif, en fils de Litz ou en fils torsadés. Les pertes par courants de Foucault sont la conséquence d’un couplage fort électrique-magnétique entre la densité du courant et la variation en fonction du temps du champ magnétique. En utilisant le modèle de Maxwell, ce couplage est décrit par une équation différentielle à dérivée partielle qui ne peut être résolue simplement. La résolution de cette équation utilisant l’approche analytique n’est possible que sous certaines hypothèses simplificatrices qui peuvent dégrader la fiabilité de la solution. La modélisation par la méthode des éléments finis permet quant à elle de prendre en compte le mouvement du rotor et la non-linéarité du circuit magnétique garantissant ainsi une meilleure précision. Néanmoins, cela conduit à une large capacité de stockage et à un temps de calcul substantiel qui peut entraver tout processus de conception ou d’optimisation. Pour surmonter ce problème, on propose dans ce manuscrit trois techniques de réduction du modèle. Ces techniques assurent une réduction efficace de la taille du système matriciel associé à la modélisation par la méthode des éléments finis et diminuent par conséquent le temps de calcul : i) une réduction spatiale qui évite une modélisation en 3D des conducteurs complexes en fils torsadés et en fils de Litz et propose une modélisation 2D satisfaisante .ii) la technique de la perturbation. iii) la réduction de l’ordre du modèle utilisant la méthode de la décomposition orthogonale aux valeurs propres combinée à la méthode d’interpolation empirique discrète. La comparaison du modèle réduit à un modèle complet de référence montre l’efficacité de la réduction du modèle à réduire le temps de calcul tout en restant en deçà d’une erreur de précision acceptable. / The switched reluctance machine can be used in hybrid or electric vehicle where, for autonomy considerations, energy efficiency is crucial. Because of the strong stray field in the air-gap region of the switched reluctance machine due to its salient pole geometry, the copper losses can become substantial. It is firmly recommended therefore not to place the coil conductors near the air-gap region. Nevertheless, concrete instructions for optimal winding design are missing. The copper losses in electrical machines are subdivided into classical DC ohmic losses and additional eddy current losses occurring due to the time varying magnetic fields penetrating the copper conductors. Based on the fact that the DC losses are constant at a given operating point, the optimization is focused on reducing the eddy current losses by modifying the winding geometry configuration. In the case of tedious repetitive calculations, met for example during design and optimization processes of electrical machine windings, there is a significant interest in reducing the computation time. This work suggests three model reduction techniques and their applications in the finite element analyses.Besides the frequency of the excitation current and the cross section of the coil conductors, several factors related to the winding configuration can affect the addition al eddy current losses:i) the coil conductor position in the winding slot especially near the stator pole or close to the air gapii) the disposition of the coil conductor against the two-dimensional flux lines in the slot windingiii) the subdivision of the solid conductor into multiple parallel strands swapping their positions periodically in the conductor cross section throughout the length of the machine winding.This thesis mainly studies the influence of the geometric coils disposition in the slot windings and the type of the conductor used whether it is solid or stranded, with Litz or twisted wires.The eddy current losses exit through the strong electro-magnetic coupling between the electric current density and the time dependent magnetic flux lines penetrating the conductors; it is described mathematically by a partial differential equation that cannot be solved easily. The analytical approach, which is used practically for a quick resolution of the strong electro-magnetic coupling equation, is only possible under certain simplifying assumptions that deteriorate brutally the reliability of the copper losses calculation. The finite element modeling as for it, allows taking into account the rotor motion and the non-linear behavior of the magnetic circuit, thus ensuring a higher accuracy. However, it leads under these conditions to a substantial calculation time and requires large storage capacity. These constraints are critical and may hinder therefore any process of conception or optimization. In this thesis, we suggest three different model reduction techniques that can be effective in reducing the size of large scale complete finite element models and enable therefore to shorten the computational time:i) the spatial reduction avoiding the 3D modeling which seems required in the case of twisted and Litz wires and suggesting an alternative satisfactory 2D modeling.ii) the perturbation technique.iii) the model order reduction using the proper orthogonal decomposition combined with the discrete empirical interpolation method.The comparison between the reduced model solutions to that of the complete finite element model has proved the effectiveness of the proposed model reduction techniques; they allow shrinking the required computational time while staying below an acceptable error of accuracy.