Laboratory test of reinforced and prestressed concrete structures have been used widely to explore the behavior of reinforced and prestressed concrete components and structures; Such tests are often time-consuming and costly. However, numerical models have been shown to compare favorably with experiments. Thus, computations are viewed nowadays as efficient alternatives to tests, time-wise and cost-wise. In the research reported in this thesis, finite-element model were used in a study of pretressed structural components in order to correlate levels of pretension with deflection and strain measurements. The two main objectives were to develop a suitable finite element model of prestressed concrete beams and to forecast beam prestension on the basis of deformations resulting from specified simple load, e.g., a uniformly distributed transverse load. A commercial finite-element analysis package (ANSYS 12) was used to set up, use and evaluate the computational model. Furthermore, a finite-difference model was employed in order to ascertain the validity of ANSYS results by comparison with engineering beam theory taking into account the applied pretension. This study demonstrates the potential usefulness of deflection and strain measurements as indicators of the pretension applied or remaining in prestressed concrete beams. / text

La transduction électrostatique est utilisée depuis plusieurs décennies dans les fréquences du domaine audible, principalement sous la forme de microphones membranaires. La transposition du même principe de transduction, mais dans un domaine de fréquence au-dessus du MHz, et par l’utilisation de dispositifs micro-usinés, c'est-à-dire produits à l’aide de technologies de photolithographie, a été proposée à partir de la fin des années 1990. Ces transducteurs, désignés sous l’acronyme cMUT (capacitive micromachine ultrasonic transducers), se composent d’un assemblage de transducteurs élémentaires, chacun possédant une partie mobile conventionnellement appelée diaphragme ou membrane, actionnée par la pression électrostatique. Cette thèse s’inscrit dans le développement de transducteurs de ce type destinés au domaine de l’imagerie médicale ultrasonore. Ce secteur d’application utilise actuellement des transducteurs basés sur des céramiques (ou, dans certains cas précis, des polymères) piézoélectriques. Le cMUT est intéressant dans certains sous-domaines d’application des ultrasons médicaux en raison de sa bonne adaptation à une production en grande série, de son intégration plus facile avec des éléments électroniques, de son faible échauffement et de l’absence de matériaux toxiques dans son processus de fabrication. La partie théorique de cette thèse repose sur une approche de modélisation par différences finies. Un modèle basé sur la théorie des plaques minces est développé pour prendre en compte la mécanique du transducteur élémentaire cMUT (c'est-à-dire d’un seul diaphragme). Ce modèle est ensuite complété par l’intégration de l’effet d’un chargement acoustique par un fluide. De façon à modéliser un transducteur entier, il est nécessaire de prendre en compte le couplage acoustique existant entre les différentes membranes. Pour rendre cela possible, un circuit équivalent, permettant de réduire chaque membrane à un système à un seul degré de liberté, est développé. Il est validé en le comparant au modèle de différences finies dans des cas où celui-ci peut être utilisé. Les travaux expérimentaux ont fait appel principalement aux deux techniques de caractérisation suivantes : les mesures d’impédance électrique, et les mesures de déplacement effectuées par interférométrie laser. Ces mesures ont été utilisées dans une double optique. D’une part, dans un objectif de caractérisation, ils ont permis de vérifier la fonctionnalité des dispositifs fabriqués et d’évaluer leurs performances. D’autre part, en comparant différentes configurations entre elles, ils ont rendu possible une validation expérimentale du modèle qui a été mis au point. / Electro-acoustic transduction based on electrostatic force has a long history in the range of audible frequencies, mainly as membrane-based microphones. Starting in the late 1990’s, it has been proposed to use the same principle in the multi-MHz frequency domain, thanks to micro-machined devices – meaning they’re produced through lithography technology. Such transducers, known as cMUT for capacitive micromachine ultrasonic transducers, are made of an assembly of elementary vertically mobile cells, usually designated as membranes, driven by electrostatic force. This PhD work is part of the development of this kind of transducers designed for medical imaging applications. This area currently uses transducers based on piezoelectric ceramics (or piezoelectric polymers for some peculiar cases). CMUT is an interesting alternative for some subdomains of medical ultrasound applications, due to its volume production ability, its easier integration with electronic elements, its low heat dissipation and the absence of toxic materials.

In this work, we design a linear, two step implicit finite difference method to approximate the solutions of a biological system that describes the interaction between a microbial colony and a surrounding substrate. Three separate models are analyzed, all of which can be described as systems of partial differential equations (PDE)s with nonlinear diffusion and reaction, where the biological colony grows and decays based on the substrate bioavailability. The systems under investigation are all complex models describing the dynamics of biological films. In view of the difficulties to calculate analytical solutions of the models, we design here a numerical technique to consistently approximate the system evolution dynamics, guaranteeing that nonnegative initial conditions will evolve uniquely into new, nonnegative approximations. This property of our technique is established using the theory of M-matrices, which are nonsingular matrices where all the entries of their inverses are positive numbers. We provide numerical simulations to evince the preservation of the nonnegative character of solutions under homogeneous Dirichlet and Neumann boundary conditions. The computational results suggest that the method proposed in this work is stable, and that it also preserves the bounded character of the discrete solutions.

A breed-and-burn molten salt reactor (BBMSR) concept is proposed to address the Generation IV fuel cycle sustainability objective in a once-through cycle with low enrichment and no reprocessing. The BBMSR uses separate fuel and coolant molten salts, with the fuel contained in assemblies of individual tubes that can be shuffled and reclad periodically to enable high burnup. In this dual-salt configuration, the BBMSR may overcome several limitations of previous breed-and-burn (B$\&$B) designs to achieve high uranium utilisation with a simple, passively safe design. A central challenge in design of the BBMSR fuel is balancing the neutronic requirement of large fuel volume fraction for B$\&$B mode with the thermal-hydraulic requirements for safe and economically competitive reactor operation. Natural convection of liquid fuel within the tubes aids heat transfer to the coolant, and a systematic approach is developed to efficiently model this complex effect. Computational fluid dynamics modelling is performed to characterise the unique physics of the system and produce a new heat transfer correlation, which is used alongside established correlations in a numerical model. A design framework is built around this numerical model to iteratively search for the limiting power density of a given fuel and channel geometry, applying several defined temperature and operational constraints. It is found that the trade-offs between power density, core pressure drop, and pumping power are lessened by directing the flow of coolant downwards through the channel. Fuel configurations that satisfy both neutronic and thermal-hydraulic objectives are identified for natural, 5$\%$ enriched, and 20$\%$ enriched uranium feed fuel. B$\&$B operation is achievable in the natural and 5$\%$ enriched versions, with power densities of 73 W/cm$^3$ and 86 W/cm$^3$, and theoretical uranium utilisations of 300 $\mathrm{MWd/kgU_{NAT}}$ and 25.5 $\mathrm{MWd/kgU_{NAT}}$, respectively. Using 20$\%$ enriched feed fuel relaxes neutronic constraints so a wider range of fuel configurations can be considered, but there is a strong inverse correlation between power density and uranium utilisation. The fuel design study demonstrates the flexibility of the BBMSR concept to operate along a spectrum of modes ranging from high fuel utilisation at moderate power density using natural uranium feed fuel, to high power density and moderate utilisation using 20$\%$ uranium enrichment.