O objetivo dessa pesquisa foi determinar parâmetros qualitativos da madeira de Pinus caribaea var. bahamensis em 3 classes de diâmetro com uma classe comum, nos plantios 1969 e 1992, da Floresta Estadual de Manduri do Instituto Florestal - SP. Foram coletadas 3 árvores por classe e retiradas toras de 3 m. de altura. De cada tora foi retirada a prancha principal e desta retiradas amostras da região da base e do topo. Foram realizadas medições do lenho inicial e tardio de cada anel de crescimento para construção do gráfico de distribuição dos anéis de crescimento de cada árvore. Também foi realizada a amostragem paralela à medula, a amostragem acompanhando o anel de crescimento e na primeira tora a amostragem acompanhando o anel no topo da tora para a realização dos ensaios físicos e mecânicos. As curvas das isopropriedades puderam então ser determinadas. Nas árvores de mesma idade e diâmetro diferentes, as variações das dimensões das traqueídes foram observados nos 3º, 6º, 9º, 12º e 15º anel de crescimento e entre o lenho inicial e tardio. Entre as diferentes idades foi determinado se as variações no 16º(último anel de crescimento) das árvores jovens foram correspondentes ao 16º(anel de crescimento) nas árvores adultas. Também foi determinado se havia variações das dimensões das traqueídes nos últimos anéis de crescimento nas duas idades. Os resultados demonstraram: a)- as características anatômicas foram compatíveis para a espécie; b)- as dimensões das traqueídes variaram no sentido radial, entre idades diferentes e mesma classe e mostraram nas árvores jovens o menor valor de comprimento e espessura da parede e o maior valor de diâmetro tangencial e lume das traqueídes; o crescimento variou entre o 9º e 12º anel de crescimento,. c) a variação entre o lenho inicial e tardio nos 16º anéis de crescimento das árvores de menor idade comparada com os das árvores adultas, não apresentou um padrão definido; (d)- na comparação entre os últimos anéis de crescimento das árvores de idades diferentes o lenho inicial e tardio apresentou um padrão mais homogêneo. As curvas de isopropriedades mostraram o mesmo padrão de distribuição tanto das isopropriedades físicas como das mecânicas. Os tipos de amostras adotadas não apresentaram variações entre si. Os valores médios de cada isopropriedade no sentido axial mostraram os maiores valores de isodensidade aparente nas árvores adultas; para isoproporção de lenho inicial os maiores valores foram verificados nas árvores mais jovens e foram decrescentes no sentido radial; para isoproporção de lenho tardio e para isomódulo de elasticidade e ruptura, os maiores valores foram observados nas árvores adultas e foram crescentes da medula para a casca. Entre as idades estudadas verificam-se diferenças de distribuição das isopropriedades. O conhecimento da distribuição das curvas de isopropriedades possibilita um melhor aproveitamento de madeira. / The objective of this research has been to determine qualitative parameters of Pinus caribaea var. bahamensis wood in three diameter classes with a common base in 1969 and 1992 plantations from Floresta Estadual from the Manduri Instituto Florestal SP. three trees per class were collected and 3-m high logs were removed. From each log the main plank was removed and from this samples were taken from the base and top regions. Measurements of the early and late wood from each growth ring were made so that the distribution graph of the growth rings from each tree could be made. In order for the physical and mechanical tests to be made samplings were taken parallel to the pith, following the growth ring, and, in the first log, a sampling following the ring at the top of the log. The isoproperty curves could then be established. In the trees of same age and different diameters the variations in tracheid dimensions were examined at the 3rd, 6th, 9th, 12th, and 15th growth ring, and between early and late wood. Among different ages it was determined whether the variations at the 16th (last growth ring) of the young trees were equivalent to the 16th (growth ring) in adult trees. It was also determined whether there was any variation in tracheid dimensions in the last growth ring of the two ages. The results showed that: a) the anatomical characteristics were compatible for the species; b) the tracheid dimensions varied in the radial direction among different ages and same class, and showed in the young trees the lowest value for length and wall width, and the largest value for tangential diameter and tracheid lumen; the growth varied between the 9th and the 12th growth ring; c) the variation between early and late wood in the 16th growth rings of the youngest trees, as compared to adult trees did not show a definite pattern; d) in the comparison among the last growth ring of trees of different ages the early and late wood showed a nore homogeneous pattern. The isoproperty curves showed the same distribution pattern, both of the physical and mechanical isoproperties. The kinds of samples used did not show variations from each other. The average values for each isoproperty in the axial direction showed the greatest values of apparent isodensity in adult trees; concerning the isoproportion of early wood, the greatest values were found in the youngest trees and decreased in the radial direction; concerning the isoproportion of late wood, and elasticity and rupture isomodule the greatest values were found in adult trees and decreased from pith to bark. Among the ages studied differences in the distribution of isoproperties can be found. The knowledge of the distribution of the isoproperty curves permits an improved use of the wood.

L'invalidité de la mesure quantique est une propriété des phénomènes quantiques.Elle est reliée au fait que la mesure de systèmes quantiques peut les affecter d'une façon que ne peut pas être décrite au sein de la physique classique.Cette thèse étudie la question de l'invasivité des mesures quantiques à travers l'inégalité de Leggett-Garg et d'une autre inégalité basée sur la condition de non perturbation (\ non-disturbance condition "). La violation de ces inégalités témoigne de l'invasivité des mesures quantiques. Dans un premier temps, nous _étudierons un modèle pour la violation de l'inégalité de Leggett-Garg, qui permettra une caractérisation opérationnelle de l'invasivité de la mesure quantique à travers un paramètre appelé la mesurabilité du système physique. Ce paramètre contrôle la violation de l'inégalité de Leggett-Garg et peut être reliée à des tests expérimentaux de cette inégalité. De cette façon, ce paramètre permet la compréhension et l'interprétation de ces violations. Nous avons également étudie, via ce modèle, la relation entre l'invasivité et une définition particulière de la « macroscopicité", associée la »taille " de systèmes de spin. Nous avons ensuite étudie une application de l'invasivité de la mesure quantique dans le cadre des protocoles pour l'estimation de paramètres en métrologie quantique. Une relation générale entre l'information de Fisher et les corrélations quantiques temporelles a été établie, et permet la caractérisation de la robustesse au bruit de scenarios de métrologie. Cette relation sert de ligne directrice pour la connexion entre l'invasivité de la mesure quantique et des scenarios (quasi-)optimaux en métrologie. Nous avons également établi une relation entre l'invasivité de la mesure quantique et une définition de la cohérence macroscopique. Pour finir, nous avons proposé un protocole pour tester la non-invasivité de mesures, basé sur la condition de non perturbation, pour des systèmes de spin de taille arbitraire. Cette inégalité permet de s'assurer contre la possibilité que sa violation soit due à des perturbations classiques de la mesure. Nous avons montré que la valeur maximale pour la violation de l'inégalité correspond au nombre de particules qui constitue le système / Quantum measurement invasiveness is a feature of quantum phenomena, i.e. associated with the fact that measurements can affect quantum systems in a manner which cannot be described by classical physics. In this thesis we will investigate measurement invasiveness through the Leggett-Garg inequality and another inequality based on the non-disturbance condition, both of which, when violated, witness measurement invasiveness. First, we will study a model for the violation of the Leggett-Garg inequality, which will allow us to provide an operational characterization of measurement invasiveness through a parameter called the measurability of the physical system. This parameter controls Leggett-Garg inequality violation and can be associated with experimental tests of this inequality, helping one to understand and interpret them. We will also investigate, through this model, the relationship between measurement invasiveness and a specific definition of macroscopicity, related to the \size" of spin systems. We will then seek to study an application of measurement invasiveness in the context of protocols for parameter estimation or quantum metrology. A general relationship between the Fisher information and temporal quantum correlations will be established, allowing one to characterize the robustness of metrological scenarios against the presence of noise. This relationship will also serve as a guideline for a connection between measurement invasiveness and (nearly-) optimal metrological scenarios. We will also establish a relationship between measurement invasiveness and a definition of a measure of macroscopic coherence. Lastly, we will propose a protocol for testing measurement noninvasiness based on the non-disturbance condition for spin systems of arbitrary size. This inequality allows one to argue against the possibility of its violation being due to the classical disturbance of measurements. We will show that the maximum value for the violation of this inequality Corresponds to the number of particles which constitutes the system.

La croissance démographique dans les zones urbaines représente un problème pour la planification des transports. La surcharge des systèmes de transport urbains entraîne des coûts monétaires importants et des problèmes environnementaux. Des mesures politiques sont alors nécessaires pour réduire le niveau de congestion et accroître l'efficacité des systèmes de transport. À court terme, les simulateurs de trafic pourraient constituer un outil puissant pour la conception de solutions innovantes. Mais les simulateurs de trafic classiques sont exigeants sur le plan informatique pour les applications à grande échelle. De plus, la mise en place du scénario de simulation est complexe. Une modélisation de trafic agrégée pourrait être une bonne solution (Daganzo-2007, Geroliminis-2008). Le réseau routier des villes est divisé en régions, où un diagramme fondamental macroscopique bien défini (MFD) régule les conditions de circulation à l'intérieur de chacune. Le MFD concerne le débit et la densité de trafic moyens dans une région. Malgré que l’idée d’agréger le réseau de la ville soit simple, il soulève plusieurs défis qui n’ont pas encore été abordés. Jusqu'à aujourd'hui, seule (Yildirimoglu-2014) propose un cadre d'affectation dynamique du trafic pour les réseaux régionaux et les modèles MFD. Ce cadre est basé sur le modèle Logit multinomial et ne traite pas explicitement des distributions de longueurs de parcours. De plus, leur structure ne considère pas que les utilisateurs sont différents les uns des autres et ont des objectifs et des préférences différents pour leurs voyages. L'objectif de cette thèse est double. Tout d'abord, l'influence du comportement des utilisateurs sur la performance globale du réseau routier d’une ville est étudiée. Cette analyse se concentre sur la vitesse moyenne du réseau et ses capacités internes et de sortie, en comparant différents modèles tenant compte des différents types de comportement des utilisateurs par rapport à l'équilibre utilisateur déterministe et stochastique. En second lieu, un cadre innovant et complet d’affectation dynamique du trafic pour les modèles multirégionaux basés sur le MFD est proposé. Ce cadre est divisé en plusieurs étapes et repose sur les connexions entre la ville et les réseaux régionaux. Dans un premier temps, des méthodes systématiques de mise à l’échelle sont proposées pour rassembler les voies régionales. Dans un deuxième temps, quatre méthodes sont discutées pour calculer les distributions de longueurs de parcours pour caractériser ces chemins régionaux. Dans la troisième étape, un modèle de chargement de réseau qui considère les distributions de longueurs de parcours explicitement calculées et l’évolution des vitesses moyennes régionales est proposé. Enfin, ce cadre d'affectation dynamique du trafic est étendu pour prendre en compte les usager qui ont une aversion au regret ou une rationalité imparfaite. Cette thèse s'inscrit dans le cadre d'un projet européen ERC intitulé MAGnUM: approche de modélisation du trafic multi-échelle et multimodal pour la gestion durable de la mobilité urbaine. / The population growth in urban areas represents an issue for transportation planning. This overload of urban transportation systems, leading to significant monetary costs and environmental issues. Policy measures are then needed to decrease the level of congestion and increase the efficiency of transportation systems. In a short term, traffic simulators might be a powerful tool that helps to design innovative solution. But, the classical traffic simulators are computationally demanding for large scale applications. Moreover, the set up of the simulation scenario is complex. An aggregated traffic modeling might be a good solution (Daganzo, 2007; Geroliminis and Daganzo, 2008). The city network is divided into regions where a well-defined Macroscopic Fundamental Diagram (MFD) regulates the traffic conditions inside each one. The MFD relates the average traffic flow and density inside a region. Despite the idea of aggregating the city network is simple, it brings several challenges that have not yet been addressed. Up to today, only Yildirimoglu and Geroliminis (2014) proposed a dynamic traffic assignment framework for regional networks and MFD models. This framework is based on the simple Multinomial Logit model and does not explicitly deal with trip length distributions. Moreover, their framework does not consider that users are different from each other and have different purposes and preferences for their travels. The goal of this PhD dissertation is to twofold. First, the influence of the users behavior on the global network performance is investigated. This analysis focus on the network mean speed and its internal and outflow capacities, comparing different models that account for different kinds of users behavior against the Deterministic and Stochastic User Equilibrium. Second, an innovative and complete dynamic traffic assignment framework for multi-regional MFD-based models is proposed. This framework is divided into several milestones and is based on the connections between the city and regional networks. In a first step, systematic scaling-up methods are proposed to gather the regional paths. In a second step, four methods are discussed to calculate the distributions of trip lengths that characterize these regional paths. In the third step, a network loading model that considers distributions of trip lengths that are explicitly calculated and the evolution of the regional mean speeds is proposed. Finally, this dynamic traffic assignment framework is extended to account for bounded rational and regret-averse users. This PhD is part of a European ERC project entitled MAGnUM: Multiscale and Multimodal Traffic Modeling Approach for Sustainable Management of Urban Mobility.

Methods for and computed results of including the physics and spatial attributes of microstructures are presented for a number of materials applications in devices. The research in our group includes applications of computation of macroscopic response of material microstructures, the development of methods for calculating microstructural evolution, and the morphological stability of structures. In this review, research highlights are presented for particular methods for computing the response in: 1) ferroelectric materials for actuator devices; 2) coarse-graining of atomistic data for simulations of microstructural evolution during processing; 3) periodic and non-periodic photonic composites; and 4) re-chargeable battery microstructures. / Singapore-MIT Alliance (SMA)

Flow of traffic on freeways and limited access highways can be represented as a series of kinemetic waves. Solutions to these systems of equations become problematic under congested traffic flow conditions, and under complicated (real-world) networks. A simplified theory of kinematics waves was previously proposed. Simplifying elements includes translation of the problem to moving coordinate system, adoption of bi-linear speed-density relationships, and adoption of restrictive constraints at the on- and off-ramps. However, these simplifying assumptions preclude application of this technique to most practical situations. This research explores the limitations of the simplified theory of kinematic waves. First this research documents a relaxation of several key constraints. In the original theory, priority was given to on-ramp merging vehicles so that they can bypass any queue at the merge. This research proposes to relax this constraint using a capacity-based weighted fair queuing (CBWFQ) merge model. In the original theory, downstream queue affects upstream traffic as a whole and exiting traffic can always be able to leave as long as it gets to the diverge. This research proposes that this diverge constraint be replaced with a contribution-based weighted splitting (CBWS) diverge model. This research proposes a revised notation system, permitting the solution techniques to be extended to freeway networks with multiple freeways and their ramps. This research proposes a generalization to permit application of the revised theory to general transportation networks. A generalized CBWFQ merge model and a generalized CBWS diverge model are formulated to deal with merging and diverging traffic. Finally, this research presents computational procedure for solving the new system of equations. Comparisons of model predictions with field observations are conducted on GA 400 in Atlanta. Investigations into the performance of the proposed CBWFQ and CBWS models are conducted. Results are quite encouraging, quantitative measures suggest satisfactory accuracy with narrow confidence interval.

The utility of dental calculus as a proxy for diet and mouth use is explored for the Middle Holocene Cis-Baikal region of Central Siberia based on two methods: a macroscopic analysis of severity and a microscopic analysis of particles within deposits. The study area was inhabited by two culturally and biologically distinct cultures, the Early Neolithic (EN) Kitoi culture (8,000 to 7,000/6,800 cal B.P.) and the Late Neolithic-Early Bronze Age (LN-EBA) Isakovo-Serovo-Glaskovo (ISG) cultural complex (6,000/5,800 to 4,000 cal B.P.), separated by a period of cultural transition marked by a cessation in formal cemetery use. Data were collected from four cemetery sites, two dating to the EN and two dating to the LN-EBA. Nonparametric testing of calculus severity revealed that, for adult males and juveniles, lakeshore populations displayed greater affinity to each other than to their contemporaneous cultural counterpart populations living along riverine systems in the Angara River Valley. Trends within the EN cemetery Shamanka II contrasted to the other cemetery populations, with noticeably larger deposits in anterior quadrants and significant sexual distinctions. The proportion of protein to carbohydrates consumed is known to influence calculus formation, but both cultural groups lived on a diet based predominately on meat sources so dietary ratios alone do not adequately explain the differences distinguished. A complex multifactorial model involving microregional differences in resources/environment, foraging patterns, individual variation, and dental wear patterns provides at least a partial explanation for the results observed. A wide range of particles were recovered during the microscopic analysis of calculus, albeit in low concentrations. The low starch grain counts were consistent with a diet based predominately on meats but still provide some of the first direct evidence for plant consumption in the Cis-Baikal, including possible plant processing by cooking or grinding based on damage evident on the grains. Other particles recovered may provide evidence of mouth use activities or palaeoenvironmental influences. Together, the two components of this analysis offer strong evidence that dental calculus is a useful tool for reconstructing hunter-gatherer lifeways but also highlight the limitations of conducting this type of research on previously excavated and potentially contaminated material.

The ionomer-membrane interface in a membrane electrode assembly connects the catalyst and membrane and allows hydrated protons to move between the catalyst and membrane. The continuous operation of the polymer membrane electrolyte fuel cell at high temperature and/or in frequent freeze/thaw cycles leads to membrane degradation and delamination of the interface, which lower the proton conductivity. In this dissertation, we modeled the chemical degradation and proton conductivity of perfluorosulfonic acid (PFSA) ionomers by ab initio calculations and macroscopic modeling. All ab initio calculations were performed using Gaussian 03 suites of program by employing B3LYP/6-311++G** method/basis set. The macroscopic modeling involves nonequilibrium thermodynamics. The results show that PFSA membranes can degrade both via side-chain and backbone in the presence of hydroxyl radical. The energetics of homolytic bond cleavage show that the C–S bond in the side-chain is the weakest link and breaks exothermally in the presence of hydroxyl radical. The C–S bond in the membrane fragment radical can break at low activation energy. The side-chain degradation also leads to the split of the backbone into two parts. The backbone degradation starts with the reaction of –COOH impurities in the backbone with the hydroxyl radical, which has the lowest activation energy, and follows an “unzipping mechanism”. The reactions in this mechanism are exothermic. The channels in the interface were modeled as cylindrical pores and the anionic charges were fixed on the pore wall. The analytical expression of proton conductivity was derived from the evolution equations for mass and momentum of hydronium ions by using an order of magnitude analysis. The results show that the conductivity increases with increasing water content and pore radius. The conductivity usually increases on decreasing the separation distance between sulfonates on the length and decreases with decreasing sulfonates separation distance on the circumference. The conductivity of the two pores, one of the interface and the other of the membrane, is closer to the conductivity of the pore with the lowest conductivity and its magnitude depends on the relative radius and length of the pores.

Malgré son indéniable succès pour décrire les systèmes physiques à l'équilibre thermodynamique (grâce à la distribution de Boltzmann, reflétant la maximisation de l'entropie, et permettant la construction systématique de potentiels thermodynamiques), la physique statistique n'offre pas de cadre général pour étudier les phénomènes hors d'équilibre, i.e dans lesquels on observe un courant moyen non nul d'une grandeur physique (énergie, charge, particules...).L'objectif de la thèse est de décrire de tels systèmes à l'aide de modèles très simples mais qui retranscrivent néanmoins les principales caractéristiques physiques de ceux-ci. Ces modèles sont constitués de particules se déplacant de manière aléatoire sur un réseau unidimensionnel connecté à des réservoirs et soumises à un principe d'exclusion. L'enjeu est de calculer exactement l'état stationnaire du modèle, notamment le courant de particules, ses fluctuations et plus particulièrement sa fonction de grande déviation (qui pourrait jouer le rôle d'un potentiel thermodynamique hors d'équilibre).Une première partie de la thèse vise à construire des modèles dits intégrables, dans lesquels il est possible de mener à bien des calculs exacts de quantités physiques. De nouveaux modèles hors d'équilibre sont proposés grâce à la résolution dans des cas particuliers de l'équation de Yang-Baxter et de l'équation de réflexion. De nouvelles structures algébriques permettant la construction de ces solutions par une procédure de Baxtérisation sont introduites.Une deuxième partie de la thèse consiste à calculer exactement l'état stationnaire de tels modèles en utilisant l'ansatz matriciel. Les liens entre cette technique et l'intégrabilité du modèle ont été mis en lumière au travers de deux relations clef: la relation de Zamolodchikov-Faddeev et la relation de Ghoshal-Zamolodchikov. L'intégrabilité a aussi été exploitée au travers des equations de Knizhnik-Zamolodchikov quantiques, afin de calculer les fluctuations du courant, mettant en lumière des connexions avec la théorie despolynômes symétriques (polynômes de Koornwinder en particulier).Enfin une dernière partie de la thèse porte sur la limite hydrodynamique des modèles étudiés, i.e lorsque la maille du réseau tend vers zero et que le nombre de constituants du système tend vers l'infini. Les résultats exacts obtenus sur les modèles à taille finie ont permis de vérifier les prédictions de la théorie des fluctuations macroscopiques (concernant les fluctuations du courant et du profil de densité dans l'état stationnaire) et de l'étendre à des modèles comprenant plusieurs espèces de particules. / Although statistical physics has been very successful to describe physical systems at thermal equilibrium (thanks to the Boltzmann distribution, which reflects the maximization of the entropy, and allows one to construct in a systematic way thermodynamic potentials), it remains elusive to provide an efficient framework to study phenomena that are out-of-equilibrium, i.e displaying non vanishing current of physical quantities (energy, charge, particles...).The goal of the thesis is to describe such systems with very simple models which retain nevertheless their main physical features. The models consist in particles evolving randomly on a one dimensional lattice connected to reservoirs and subject to hard-core repulsion. The challenge lies in computing exactly the stationary state of the model, especially the particle current, its fluctuations and more precisely its large deviation function (which is expected to play the role of an out-of-equilibrium thermodynamic potential).In the first part of the thesis we construct models, called integrable, in which we can perform exact computations of physical quantities. We introduce several new out-of-equilibrium models that are obtained by solving, in specific cases, the Yang-Baxter equation and the reflection equation. We provide new algebraic structures which allow us to construct the solutions through a Baxterisation procedure.In the second part of the thesis we compute exactly the stationary state of these models using a matrix ansatz. We shed light on the connection between this technique and the integrability of the model by pointing out two key relations: the Zamolodchikov-Faddeev relation and the Ghoshal-Zamolodchikov relation. The integrability is also exploited, through the quantum Knizhnik-Zamolodchikov equations, to compute the fluctuations of the particles current, unrevealing connections with the theory of symmetric polynomials (the Koornwinder polynomials in particular).Finally the last part of the thesis deals with the hydrodynamic limit of the models, i.e when the lattice spacing tends to $0$ and the number of particles tends to infinity. The exact results obtained for a finite size system allow us to check the validity of the predictions of the macroscopic fluctuations theory (concerning the fluctuations of the current and the density profile in the stationary state) and to extend the theory to systems with several species of particles.

Les travaux présentés dans cette thèse visent à proposer des méthodes de routage d’une population de masse à travers un réseau perturbé dont les données varient dans le temps pour l’aide à la conception de plan d’évacuation. Ce problème s’illustre parfaitement en cas de catastrophe d’origine humaine ou naturelle où les populations (potentiellement) impactées par ces sinistres doivent quitter leur lieux de vie pour une période pouvant aller d’un à plusieurs jours. Dans la littérature, ces routages de masse sont souvent modélisés comme des problèmes de flots dynamiques dont l’objectif est de minimiser la durée globale du transfert des individus depuis un certain nombre de points de départs dangereux vers des points d’arrivé sûrs. Toutefois, peu de travaux prennent en compte la notion de sécurité durant ce routage et encore moins le déploiement d’agents (policiers, sapeur-pompiers, ambulanciers,...) pouvant sécuriser et/ou faciliter le déplacement des personnes. / The work presented in this thesis aims to propose methods for routing a mass population through a disturbed network whose data vary over time. This problem can be illustrated by disasters due to humans or natural events where people (potentially) affected have to leave their living places for a period of one to several days. In the literature, mass routing are often modeled as dynamic flow problems whose objective is to minimize the overall duration of the evacuation process from a set of gathering points towards another set of shelter locations. However few papers take into account the concept of safety during this routing nor deploying task forces that can secure or facilitate this process. In this context, the safety security can be seen as a danger affecting the quality of life of people we organize the trip. In this context, the safety can be seen as a danger that influence the health of the people we are trying to evacuate. Indeed, this hazardous event can be related to a radioactive cloud, a fire, a tsunami, an earthquake or a flooding which make some of paths becoming dangerous or less usable by evacuees.

Les travaux présentés dans ce manuscrit visent à étudier la propagation d'un défaut interfacial de géométrie connue soumis à un cisaillement macroscopique à partir d'une méthodologie expérimentale développée durant la thèse qui consiste à i) élaborer un système revêtu céramique/métal dont la morphologie d'interface est contrôlée, ii) introduire un défaut d'interface par la technique de choc laser, iii) soumettre le système revêtu pré-fissuré à un cisaillement macroscopique grâce à une machine de fatigue biaxiale coplanaire et iv) mesurer in situ l'évolution de ce défaut. Le système revêtu subit donc une sollicitation dynamique par la technique de choc laser et quasi-statique par les essais biaxiaux. La morphologie d'interface, paramètre influant sur la fissuration, est également étudiée. Un dépôt d'alumine pure est directement projeté par plasma sur un substrat métallique, un superalliage base cobalt Haynes 188 et un acier inoxydable 304L, sans sous-couche.Un premier travail d'analyse du défaut introduit par choc laser en fonction des paramètres laser et de la morphologie d'interface a tout d'abord été réalisé. Le défaut d'interface résultant se caractérise par une zone délaminée circulaire de quelques millimètres de diamètre et d'une cloque formée par la couche de céramique de quelques dizaines de micromètres de hauteur. Ces dimensions caractéristiques ont été mesurées à partir de techniques d'observation non destructives : profilométrie 3D, observation optique et thermographie infrarouge. La fissuration par choc laser a été étudiée par éléments finis grâce à un modèle de type contact cohésif pour l'interface.La propagation du défaut soumis à un cisaillement macroscopique a été caractérisée expérimentalement grâce aux observations optiques et à la technique de stéréo-corrélation d'images. L'analyse par élément finis du problème a permis d'accéder aux modes de sollicitation le long du front de fissure et de donner une première explication quant aux formes délaminées obtenues expérimentalement. Cette analyse s'appuie sur un modèle de zone cohésive dont les conditions aux limites imposées sont déterminées à l'aide des mesures de déplacement obtenues par corrélation d'images. En particulier, il a été mis en évidence que l’ouverture du front de fissure (mode I), induit par le flambage de la couche et par le chargement macroscopique, favorise la propagation du délaminage qui reste pilotée essentiellement par le cisaillement local (mode II et III). L'influence du cisaillement macroscopique dans le plan de la couche déposée sur la propagation du délaminage interfacial a ensuite été étudiée à partir de trois cas de chargement. Une analyse par éléments finis basée sur la mécanique linéaire de la rupture dans un matériau homogène a permis de déterminer l'influence du cisaillement macroscopique sur le chargement local le long du front de fissure. / The work presented in this manuscript aims to investigate the growth of an interfacial flaw, whose geometry is known, under macroscopic shear loading. An experimental methodology is thus developed in which i) a ceramic/metal coated system with controlled interface roughness is processed, ii) an interfacial flaw is introduced using the laser shock technique, iii) a macroscopic shear loading is applied on the coated system using a biaxial in-plane testing device and iv) interfacial crack growth and buckling are measured in situ. Hence, both dynamic and quasi-static loadings are applied on the coated system by respectively the laser shock technique and biaxial testing. The interface roughness, which affects the crack growth, is also considered in the study. A pure alumina coating is deposited by air plasma spraying on a metallic substrate, polycrystalline cobalt base superalloy Haynes 188 and stainless steel 304L substrates, with no bond coat.First, the flaw resulting from the propagation of a laser shock wave has been analyzed according to the laser parameters and the interface roughness. An interfacial flaw is characterized by a circular delamination with a diameter of a few millimeters and a circular blister with a height of a few tens of micrometers. These characteristic dimensions have been measured thanks to non destructive techniques: 3D profilometry and image analysis based on optical observations and infrared thermography. A finite element analysis has been carried out to investigate the crack behavior under laser shock wave propagation using a cohesive contact to account for the interface behavior.The interfacial flaw growth under macroscopic shear loading has been characterized with optical observations and the digital image stereo-correlation technique. The related finite element analysis enabled to identify the local loading along the crack front and gave a first explanation about the shapes of the delaminated area observed experimentally. This analysis relies on a cohesive zone model whose applied boundary conditions are established from the displacements measured by digital image correlation technique. By this way, the delamination growth was revealed to be mostly driven by local shear (mode II and III) and the crack opening (mode I), induced by the buckling of the deposited layer and the macroscopic shear, makes the delamination growth easier. Finally, the influence of the macroscopic shear loading on the interfacial delamination has been studied from three different macroscopic shear loadings. The finite element analysis based on linear elastic fracture mechanics in a homogenous material has allowed to study the influence of the macroscopic shear loading on the local loading along the crack front.