No description available.

Les standards consacrés à la conception des oléoducs se concentrent principalement sur les chargements opérationnels, tels que les pressions internes et externes, et les procédures d'analyse de défauts actuelles n'exploitent pas les capacités d'écrouissage du matériau. Pourtant, dans des conditions extrêmes, les oléoducs peuvent être soumis à des contraintes au-delà de la limite d'élasticité jusqu'à atteindre 2.5% de déformations plastiques. Ici, les procédures proposées par ExxonMobil et PRCI basées sur des critères en déformation sont présentées, et l'utilisation de l'éprouvette SENT (Single Edge Notched Tension) pour caractériser la ténacité est étudiée, en comparant les différentes procédures d'essais recommandées. Puis, une importante campagne expérimentale a été réalisée pour caractériser deux aciers pour oléoducs à température ambiante et à basses températures. Les comportements mécaniques des matériaux de base et d'apport ont été identifiés grâce à l'utilisation de l'analyse inverse, et il est montré que le modèle d'endommagement GTN permet de modéliser finement les essais sur éprouvettes de laboratoire. Enfin, deux essais sur structures (pression et flexion, puis pression et traction) ont été réalisés de manière à comparer les approches globales et le modèle d'endommagement GTN. Ce dernier démontre une bonne transférabilité de l'éprouvette vers la structure. / Pipeline design codes and standards traditionally focus on the operational loadings such as internal and external pressures that are likely to exist over the entire lifetime of the pipeline. Existing Engineering Critical Assessments are mostly based on stress considerations, where the design margin is given as a percentage of the yield strength. In extrem environments, pipelines may experience stresses beyond the yield and plastic deformations up to 2.5 %. In such conditions, strain-based design procedures apply. In this work, a literature review of the existing strain based methods is proposed, including ExxonMobil and PRCI multi-tier approaches. The use of the Single Edge Notched Tension (SENT) specimen to measure the material toughness is then studied, benchmarking the recommended testing procedures from literature. A comprehensive experimental campaign was carried out to fully characterize two actual line pipes at room and low temperatures. The mechanical behavior of parent and weld materials are identified using an inverse analysis, and GTN damage model is shown to allow accurate modeling of the laboratory testings. Finally, two full scale tests (pressure + bending or pressure + tension) were carried out to benchmark the global approaches and GTN damage model. The latter showed a very good transferability from specimens to the structure.

Dans cette thèse, plusieurs approches de modélisation de composites renforcés par des fibres sont proposées. Le matériau étudié est le béton fibré, et dans ce modèle, on tient compte de l’influence de trois constituants : le béton, les fibres, et la liaison entre eux. Le comportement du béton est analysé avec un modèle d’endommagement, les fibres d'acier sont considérées comme élastiques linéaires, et le comportement sur l'interface est décrit avec une loi de glissement avec l’extraction complète de la fibre. Une approche multi-échelle pour coupler tous les constituants est proposée, dans laquelle le calcul à l'échelle macro est effectué en utilisant la procédure de solution operator-split. Cette approche partitionnée divise le calcul en deux phases, globale et locale, dans lesquelles différents mécanismes de rupture sont traités séparément, ce qui est conforme au comportement du composite observé expérimentalement. L'identification des paramètres est effectuée en minimisant l'erreur entre les valeurs calculées et mesurées. Les modèles proposés sont validés par des exemples numériques. / In this thesis, several approaches for modeling fiber-reinforced composites are proposed. The material under consideration is fiber-reinforced concrete, which is composed of a few constituents: concrete, short steel fibers, and the interface between them. The behavior of concrete is described by a damage model with localized failure, fibers are taken to be linear elastic, and the behavior of the interface is modeled with a bond-slip pull-out law. A multi-scale approach for coupling all the constituents is proposed, where the macro-scale computation is carried out using the operator-split solution procedure. This partitioned approach divides the computation in two phases, global and local, where different failure mechanisms are treated separately, which is in accordance with the experimentally observed composite behavior. An inverse model for fiber-reinforced concrete is presented, where the stochastic caracterization of the fibers is known from their distribution inside the domain. Parameter identification is performed by minimizing the error between the computed and measured values. The proposed models are validated through numerical examples.

[pt] A falha mecânica conhecida como fadiga é caracterizada pela iniciação e/ou propagação de trincas, causada por forças variáveis. Suas metodologias tradicionais calculam uma tensão elástica uniaxial equivalente que atua no componente, a fim de compará-la com os dados experimentais de comportamento mecânico do material do componente sob cargas uniaxiais. Esta hipótese pode levar a resultados não conservativos, por considerar que o material é igualmente sensível a tensões normais e cisalhantes, o que é falso em várias aplicações práticas. Portanto, dados torcionais e multiaxiais são necessários para melhor prever a vida em fadiga dos componentes. Para executar estes experimentos, o presente trabalho propõe uma variedade de projetos de componentes e metodologias de montagem para que se possa usar em uma máquina de tração-torção Instron 8874 uma garra hidráulica originalmente projetada para uma máquina tração pura Instron 8501. É proposto um método simplificado para estimar, por controle de deslocamento, as propriedades de fadiga de baixo ciclo em cisalhamento (gama)N, evitando assim a necessidade de usar equipamentos caros e diferentes tipos de corpos de prova. Este método é usado para caracterização das ligas Aço SAE 1020 e Alumínio 6351-T6 e os dados levantados são comparados com as propriedades medidas de fadiga de baixo ciclo em tração (epsilon)N, identificando assim se o material é mais sensível a tensões normais ou cisalhantes. Um programa numérico é usado para ajustar as curvas (epsilon)N e (gama)N nos dados experimentais, e seus procedimentos de implementação são discutidos. Por fim, são propostos e calibrados modelos de fadiga multiaxial de plano crítico mais adequados para cada material testado, com base nos dados medidos. / [en] The mechanical failure known as fatigue is characterized by the formation and/or propagation of cracks caused by variable forces. Its traditional methodologies normally calculate an equivalent uniaxial tensile stress acting on the component, in order to compare it with the known experimental mechanical behavior data of the component s material measured under uniaxial loads. This assumption can lead to non-conservative results because it considers the material to be equally sensitive to shear and tensile stresses, which is not true in a wide range of practical applications. Therefore, torsional and multiaxial experimental data is necessary to better predict the fatigue life of components. To execute those experiments, the present work proposes a variety of component designs and assembly methodologies to use on an Instron 8874 axial-torsional testing machine a hydraulic grip originally designed for an Instron 8501 uniaxial testing machine. Furthermore, a simplified method to estimate shear (gamma)N low-cycle fatigue properties via displacement-controlled experiments is proposed to avoid the need of using expensive equipment and different specimen designs, and used for characterization of SAE 1020 Steel and 6351-T6 Aluminum alloys. This data is compared with the measured tensile (epsilon)N low-cycle fatigue properties to identify if these materials are tensile or shear sensitive under multiaxial loading conditions. A numerical computing code is used to fit (epsilon)N and (gamma)N curves to the experimental data, and its implementation procedures are discussed. Finally, the most suitable critical-plane multiaxial fatigue models are proposed and calibrated for each material tested, based on the measured data.

Indonesia is a developing country that suffers from earthquakes and windstorms and where at least 60% of houses are non-engineered structures, built by unskilled workers using masonry and timber. The non-engineered housing units developed in urban region are also vulnerable to seismic hazard due to the use of low quality of material and constructions method. Those structures are not resistant to extreme lateral loads or ground movement and their failure during an earthquake or storm can lead to significant loss of life. This thesis is concerned with the structural performance of Indonesian low-rise buildings made of masonry and timber under lateral seismic load. The research presented includes a survey of forms of building structure and experimental, analytical and numerical work to predict the behaviour of masonry wall and traditional timber frame buildings. Experimental testing of both masonry and timber have been carried out in Indonesia to establish the quality of materials and to provide material properties for numerical simulations. The experimental study found that the strength of Indonesia-Bali clay brick masonry are below the minimum standard required for masonry structures built in seismic regions, being at least 50% lower than the requirement specified in British Standard and Eurocode-6 (BS EN 1996-1-1:2005). In contrast, Indonesian timber materials meet the strength classes specified in British Standard/Eurocode- 5 (BS EN 338:2009) in the range of strength grade D35-40 and C35).Structural tests under monotonic and cyclic loading have been conducted on building components in Indonesia, to determine the load-displacement capacity of local hand-made masonry wall panels and timber frames in order to: (1) evaluate the performance of masonry and timber frame structure, (2) investigate the dynamic behaviour of both structures, (3) observe the effect of in-plane stiffness and ductility level, and (4) examine the anchoring joint at the base of timber frame that resists the overturning moment. From these tests, the structural ductility was found to be less than two which is below the requirement of the relevant guidelines from the Federal Emergency Management Agency, USA (FEMA-306). It was also observed that the lateral stiffness of masonry wall is much higher than the equivalent timber frame of the same height and length. The experimental value of stiffness of the masonry wall panel was found to be one-twelfth of the recommended values given in FEMA-356 and the Canadian Building code. The masonry wall provides relatively low displacement compared to the large displacement of the timber frame at the full capacity level of lateral load, with structural framing members of the latter remaining intact. The weak point of the timber frame is the mechanical joint and the capacity of slip joint governs the lateral load capacity of the whole frame. Detailed numerical models of the experimental specimens were setup in Abaqus using three-dimensional solid elements. Cohesive elements were used to simulate the mortar behaviour, exhibiting cracking and the associated physical separation of the elements. Appropriate contact definitions were used where relevant, especially for the timber frame joints. A range of available material plasticity models were reviewed: Drucker-Prager, Crystalline Plasticity, and Cohesive Damage model. It was found that the combination of Crystalline Plasticity model for the brick unit and timber, and the Cohesive Damage model for the mortar is capable of simulating the experimental load-displacement behaviour fairly accurately. The validated numerical models have been used to (1) predict the lateral load capacity, (2) determine the cracking load and patterns, (3) carry out a detailed parametric study by changing the geometric and material properties different to the experimental specimens. The numerical models were used to assess different strengthening measures such as using bamboo as reinforcement in the masonry walls for a complete single storey, and a two-storey houses including openings for doors and windows. The traditional footing of the timber structures was analysed using Abaqus and was found to be an excellent base isolation system which partly explains the survival of those structures in the past earthquakes. The experimental and numerical results have finally been used to develop a design guideline for new construction as well as recommendations for retrofitting of existing structures for improved performance under seismic lateral load.