Cette thèse analyse les influences des interactions sociales sur les comportements économiques des ménages. Les estimations sur données ivoiriennes permettent de montrer deux types de résultats. Premièrement, la probabilité d'exercer sur un segment du marché du travail croît avec la proportion des individus du groupe qui exercent sur ce segment. Deuxièmement, la réversibilité temporelle de la consommation est rejetée pour certains types de dépenses tels que l'Alimentation et le Transport : les habitudes de consommation jouent un rôle important. De plus, les dépenses de Logement et de Transfert sont les plus sensibles à l'environnement social et donc les plus sujettes à des usages ostentatoires. Par ailleurs, un modèle théorique permet d'expliquer que la consommation ostentatoire se justifie par le besoin de reconnaissance sociale de l'individu dans la société.

Le procédé d'élaboration par chauffage des composites à matrice thermodurcissable est limité par les temps caractéristiques correspondant à la diffusion de la chaleur ; l'utilisation du chauffage par hystérésis diélectrique a montré la solution qu'il était possible d'apporter à cette limitation. Toutefois, la réduction de la durée de l'opération thermique est fonction de la mise forme du matériau résultant du fluage de la matière.<br />La thèse comprend :<br />- premièrement les analyses électromagnétiques, chimique et thermique des conditions de mise en uvre du chauffage par hystérésis diélectrique pour des objets en matériau composite à fibres de verre unidirectionnelles,<br />- deuxièmement les caractérisations physiques et chimiques de la matrice (résine époxydique associée au durcisseur) en fonction de la température et en fonction des paramètres de son écoulement. Ces caractérisations comprennent la mesure de viscosité, la mesure de l'avancement de la réaction par DSC, les caractérisations diélectriques, la caractérisation des mélanges réactifs par GPC,<br />- troisièmement, l'application de ces analyses à la maîtrise du procédé.<br />La thèse conduit à une synthèse de l'ensemble des paramètres physiques et à leurs interdépendances pendant la mise en forme des matériaux.

Cette thèse porte sur le frottement et le blocage des écoulements de grains secs. Notre étude rhéologique porte sur des matériaux modèles dans des géométries de cisaillement simple, annulaire et de plan incliné, et combine expériences et des simulations numériques discrètes. Nous montrons que l'état de cisaillement d'un matériau granulaire est entièrement déterminé par la connaissance du nombre inertiel sans dimension I, traduisant le rapport des forces d'inertie sur les forces de confinement. Les faibles I correspondent au régime quasi statique et les grands I au régime dynamique. La loi de comportement se déduit alors de la variation de la compacité et du frottement effectif avec ce nombre. En cisaillement à vitesse imposée, nous observons une transition progressive entre les régimes quasi statique et dynamique. Dans ce régime intermédiaire, la loi de comportement est indépendante des caractéristiques mécaniques du matériau (faiblement ou fortement dissipatif) et reste qualitativement identique dans les diverses géométries. On observe une localisation du cisaillement dans le cas où la distribution de contrainte est hétérogène, et des intermittences de l'écoulement dans la limite quasi statique. En cisaillement à force imposée, nous montrons que la transition fluide solide (blocage) est brutale, hystérétique et présente de fortes analogies avec celle observée pour des fluides thixotropes (pâtes, mousses, etc.). Les expériences, tant en cisaillement annulaire que sur plan incliné, montrent qu'il existe une plage de taux de cisaillement qui ne peut être atteinte de façon stable (bifurcation de viscosité). Ces différentes observations nous ont conduit à mesurer par simulation l'évolution de la structure du réseau de contacts lors de l'arrêt, et à modéliser l'hysteresis de seuil et la bifurcation de viscosité en tenant compte de la force de piégeage induite par la rugosité. Nos observations montrent aussi l'intérêt de modèle de rhéologie non locale dans la descritpion du phénomène de blocage.

Surface characteristics such as topography and critical dimensions serve as important indicators of product quality and manufacturing process performance especially at the micrometer and the nanometer scales. This paper first reviews different technologies used for obtaining high precision 3-D images of surfaces, along with some selected applications. Atomic force microscopy (AFM) is one of such methods. These images are commonly distorted by convolution effects, which become more prominent when the sample surface contains high aspect ratio features. In addition, data artifacts can result from poor dynamic response of the instrument used. In order to achieve reliable data at high throughput, dynamic interactions between the instrument's components need to be well understood and controlled, and novel image deconvolution schemes need to be developed. Our work aims at mitigating these distortions and achieving reliable data to recover metrology soundness. A summary of our findings will be presented. / Singapore-MIT Alliance (SMA)

A new dynamic model is presented for piezoelectric tube actuators commonly used in high-precision instruments. The model captures coupling between motions in all three axes such as bending motion due to a supposedly pure extension of the actuator. Both hysteresis and creep phenomena are included in the overall actuator model permitting modeling nonlinear sensitivity in the voltage to displacement response. Experimental data on hysteresis and creep are presented to support the modeling. Experiments and model predictions show that due to coupling a voltage Vz corresponding to vertical displacement will produce lateral displacement that acts as a disturbance to the main lateral response. The resonance frequency for the lateral dynamics is inherently lower than that of the longitudinal dynamics. Therefore, Vz is expected to contain frequencies that may excite the lateral resonance. Accordingly, this out of bandwidth disturbance will not be well compensated for either in open or closed loop control of the actuator. In order to preserve performance in open loop actuator control and stability and performance in closed loop control, a large reduction in the bandwidth of vertical motion would be required to avoid exciting the first bending mode. / Singapore-MIT Alliance (SMA)

Thesis (Ph. D.)--University of Washington, 2006. / Vita. Includes bibliographical references (p. 145-151).

Confining walls induce qualitative changes in adsorbed fluids. Among the most intriguing phenomena is hysteresis, where a pore fills with fluid at a greater pressure than it empties. The causes and mechanisms by which this occurs are intensely investigated yet still poorly understood. Ordered mesoporous silicas, recently discovered materials with well-defined pore size distributions, provide an opportunity to deepen our understanding of the fundamental physics of the interaction of fluids with complex solids.In part of this computational investigation we examine idealized pores. In agreement with other recent studies, we find that in 'inkbottle'-shaped pores, where a large cavity is accessible to the bulk fluid only by constrictions, there is no evidence of the long-hypothesized phenomenon of `pore blocking', where the constrictions inhibit fluid desorption from the cavity. We find that even in these simple systems the mechanism of hysteresis depends on pore characteristics, fluid properties and external conditions.For silicas containing cylindrical holes of nearly uniform diameter, such as MCM-41, the state-of-the-art is to consider only a single pore, but the poor qualitative agreement of theoretical with experimental results has improved little as wall representations of increasing sophistication have been developed. Using only a one-dimensional potential, we reproduce features of isotherms, including in the hysteresis region, by averaging over a narrow distribution of pore sizes. The qualitative behavior is shown to be a collective phenomenon not representative of any individual pore. Adding surface roughness and a constriction to the pores yields results quantitatively nearly indistinguishable from experiments.For materials larger than MCM-41, a continuum simulation proves too computationally taxing. Thus, a lattice model with adjustable fineness of site spacing is developed. It is found that a surprisingly low level of fineness is needed for confined systems to closely approximate continuum results. This model is applied to cubically symmetric materials, such as MCM-48 and SBA-16, finding that simulations are able to reproduce much of the qualitative behavior seen experimentally, but the lack of existing knowledge of the nature of silica walls proves to be a limiting factor.

The Greenland Ice Sheet (GIS) contains enough water volume to raise global sea level by over 7 meters. It is a relic of past glacial climates that could be strongly affected by a warming world. Several studies have been performed to investigate the sensitivity of the ice sheet to changes in climate, but large uncertainties in its long-term response still exist. In this thesis, a new approach has been developed and applied to modeling the GIS response to climate change. The advantages compared to previous approaches are (i) that it can be applied over a wide range of climatic scenarios (both in the deep past and the future), (ii) that it includes the relevant feedback processes between the climate and the ice sheet and (iii) that it is highly computationally efficient, allowing simulations over very long timescales. The new regional energy-moisture balance model (REMBO) has been developed to model the climate and surface mass balance over Greenland and it represents an improvement compared to conventional approaches in modeling present-day conditions. Furthermore, the evolution of the GIS has been simulated over the last glacial cycle using an ensemble of model versions. The model performance has been validated against field observations of the present-day climate and surface mass balance, as well as paleo information from ice cores. The GIS contribution to sea level rise during the last interglacial is estimated to be between 0.5-4.1 m, consistent with previous estimates. The ensemble of model versions has been constrained to those that are consistent with the data, and a range of valid parameter values has been defined, allowing quantification of the uncertainty and sensitivity of the modeling approach. Using the constrained model ensemble, the sensitivity of the GIS to long-term climate change was investigated. It was found that the GIS exhibits hysteresis behavior (i.e., it is multi-stable under certain conditions), and that a temperature threshold exists above which the ice sheet transitions to an essentially ice-free state. The threshold in the global temperature is estimated to be in the range of 1.3-2.3°C above preindustrial conditions, significantly lower than previously believed. The timescale of total melt scales non-linearly with the overshoot above the temperature threshold, such that a 2°C anomaly causes the ice sheet to melt in ca. 50,000 years, but an anomaly of 6°C will melt the ice sheet in less than 4,000 years. The meltback of the ice sheet was found to become irreversible after a fraction of the ice sheet is already lost – but this level of irreversibility also depends on the temperature anomaly. / Das grönländische Inlandeis (GIS) besteht aus einem Wasservolumen das ausreicht, um den globalen Meeresspiegel um 7 Meter ansteigen zu lassen. Es ist ein Relikt der vergangenen Eiszeit, das in einer zunehmend wärmer werdenden Welt stark in Mitleidenschaft gezogen werden könnte. In der vorliegenden Dissertation ist ein neues Verfahren zur Modellierung des Antwortverhaltens des Inlandeises auf Klimaänderungen entwickelt und angewendet worden. Die Vorteile des neuen Verfahrens im Vergleich zu den bisherigen Verfahren sind, (i) dass es über einen groen Bereich von Klimaszenarien (sowohl für die ferne Vergangenheit als auch für die Zukunft) anwendbar ist, (ii) dass es die wesentlichen Rückkopplungsprozesse zwischen Klima und Inlandeis enthält und (iii) dass es wegen seiner guten Rechenzeiteffizienz Simulationen über sehr lange Zeitskalen erlaubt. Das neue Modell (REMBO) ist für die Modellierung des Klimas und der Massenbilanz an der grönländischen Oberfläche entwickelt worden und stellt ein verbessertes Verfahren im Vergleich zu den bisherigen dar. Die Entwicklung von GIS über den letzten glazialen Zyklus ist mittels eines Ensembles von verschiedenen Modellversionen simuliert worden. Anschließend ist die Tauglichkeit der Modellversionen durch Vergleich mit Beobachtungsdaten des gegenwärtigen Klimas und der Oberflächenmassenbilanz, sowie mit paleoklimatischen Rekonstruktionen von Eisbohrkernen verifiziert worden. Der Anteil von GIS am Meeresspiegelanstieg während des letzten Interglazials ist im Bereich von 0.5 bis 4.1 m berechnet worden, was konsistent mit bisherigen Schätzungen ist. Von den Ensemblesimulationen sind diejenigen ausgewählt worden, deren Ergebnisse gut mit den Daten übereinstimmen. Durch die Auswahl von geeigneten Modellversionen sind gleichzeitig die Unsicherheiten der Parameterwerte begrenzt worden, so dass sich nun mit dem neuen Verfahren die Sensitivität von GIS auf Klimaänderungen bestimmen lässt. Mit den ausgewählten Modellversionen ist die Sensitivität von GIS auf langfristige Klimaänderungen untersucht worden. Es zeigt sich, dass das GIS ein Hystereseverhalten besitzt (d.h., eine Multistabilität für gewisse Klimazustände) und dass ein Temperaturschwellwert existiert. Bei Überschreiten des Schwellwertes bleibt das GIS nicht erhalten und wird langsam eisfrei werden. Der Temperaturschwellwert der globalen Mitteltemperatur relativ zur vorindustriellen Mitteltemperatur ist im Bereich 1.3-2.3°C ermittelt worden und liegt damit deutlich niedriger als bisher angenommen. Die Zeitdauer bis zum völligen Abschmelzen zeigt ein nichtlineares Verhalten hinsichtlich einer Erwärmung über den ermittelten Schwellwert. Eine Erwärmung von 2°C relativ zur vorindustriellen Zeit führt zu einem Abschmelzen nach 50.000 Jahren, aber eine Erwärmung um 6°C lässt das Inlandeis bereits nach 4.000 Jahren abschmelzen. Ein weiteres Ergebnis ist, dass der Abschmelzvorgang irreversibel werden kann, nachdem ein gewisser Anteil des Inlandeises abgeschmolzen ist – jedoch ist die Irreversibilität eines Abschmelzvorganges auch von der Temperaturanomalie abhängig.

In the recent years significant progresses in thedevelopment of high-temperature superconductors have been made.It is realistic to believe that power applications, based onthese conductors, in a few years will become available. To beable to utilise the conductors in an optimum way, theunderstanding of their behaviour under application-likecondition is essential. One important parameter that has to beoptimised is the power loss, which means that mathematicalmodels of these losses have to be developed. In a typicalapplication the superconductor is utilised in a coilconfiguration where the actual magnetic field is considerablehigher than for a straight structure. For power frequencies thelosses are dominated by hysteresis losses and flux flowlosses. In this thesis, mathematical models of the hysteresis andthe flux flow losses as a function of a transport current, anexternal magnetic field, the temperature and the frequency havebeen developed. The transport current and the magnetic field,which are assumed to be proportional to each other, includeboth an ac and a dc component. The models of the hysteresislosses are based on the critical state theory, and for twoidealised geometries, an infinite slab and a thin strip, newexact closed form equations have been derived. The equationsfor the two idealised geometries are then superimposed tofacilitate the description of a more realistic geometry, i.e. asuperconducting tape with a finite width and thickness. Themodel of the flux flow losses is valid for a tape shapedconductor and is based on both measurements and reasonablephysical assumptions. For the development and the validation ofthe models, a calorimetric measurement set-up has been used.From a limited number of relatively simple measurements, thedeveloped models can be adjusted to a certain superconductor,and the power losses for the actual superconductor can bepredicted in considerable more complicated cases. <b>Keywords:</b>high-temperature superconductor, hysteresislosses, flux flow losses, critical state model, calorimetricmeasurements

The kinetic mechanism of an α/β hydrolase fold epoxide hydrolase from potato, StEH1, has been studied with the aims of explaining the underlying causes for enantio- and regioselectivity, both being important for product purity. Further effort has been laid upon understanding the causes of a hysteretic behavior discovered in the measurements leading to Paper I. The enantioselectivity was investigated with substrates differing only in substituent size at one carbon of the oxirane ring structure. In catalysis with trans-stilbene oxide and styrene oxide, enantioselectivity is the result of differences in alkylation rates. In pre-steady state measurement with trans-2-methylstyrene oxide (2-MeSO), a rate-limiting step involving slow transitions, referred to as hysteresis, was discovered. With this substrate enantioselectivity is proposed to be a consequence of the catalytic rate of (1R,2R)-enantiomer being more influenced by the hysteretic behavior than was the rate of the other enantiomer. In steady-state measurements with (1R,2R)-2-MeSO, at different temperatures and pH, hysteretic cooperativity was displayed. It can be concluded that this behavior is dependent on the relationship between kcat and the rate of transition between two Michaelis complexes. From the differences in pH dependence of kcat/KM in formation of the two diols resulting from low regioselectivity in catalysis of (1R,2R)-2-MeSO, it is suggested that hysteresis is a result of the substrates placed in different conformational modes within the active site cavity. Regioselectivity is proposed to be the result of specific interactions between the catalytically important Tyr and the substrate, with a link between KM-values and degree of regioselectivity. Furthermore, the hysteretic kinetic model proposed can explain hysteresis, cooperativity and regioselectivity resulting from StEH1 catalyzed hydrolysis of (1R,2R)-2-MeSO.