The present work focuses on the formulation of a Continuum Damage Mechanics model for nonlinear analysis of masonry structural elements. The material is studied at the macro-level, i.e. it is modelled as a homogeneous orthotropic continuum. The orthotropic behaviour is simulated by means of an original methodology, which is based on nonlinear damage constitutive laws and on the concept of mapped tensors from the anisotropic real space to the isotropic fictitious one. It is based on establishing a one-to-one mapping relationship between the behaviour of an anisotropic real material and that of an isotropic fictitious one. Therefore, the problem is solved in the isotropic fictitious space and the results are transported to the real field. The application of this idea to strain-based Continuum Damage Models is rather innovative. The proposed theory is a generalization of classical theories and allows us to use the models and algorithms developed for isotropic materials. A first version of the model makes use of an isotropic scalar damage model. The adoption of such a simple constitutive model in the fictitious space, together with an appropriate definition of the mathematical transformation between the two spaces, provides a damage model for orthotropic materials able to reproduce the overall nonlinear behaviour, including stiffness degradation and strain-hardening/softening response. The relationship between the two spaces is expressed in terms of a transformation tensor which contains all the information concerning the real orthotropy of the material. A major advantage of this working strategy lies in the possibility of adjusting an arbitrary isotropic criterion to the particular behaviour of the orthotropic material. Moreover, orthotropic elastic and inelastic behaviours can be modelled in such a way that totally different mechanical responses can be predicted along the material axes. The aforementioned approach is then refined in order to account for different behaviours of masonry in tension and compression. The aim of studying a real material via an equivalent fictitious solid is achieved by means of the appropriate definitions of two transformation tensors related to tensile or compressive states, respectively. These important assumptions permit to consider two individual damage criteria, according to different failure mechanisms, i.e. cracking and crushing. The constitutive model adopted in the fictitious space makes use of two scalar variables, which monitor the local damage under tension and compression, respectively. Such a model, which is based on a stress tensor split into tensile and compressive contributions that allows the model to contemplate orthotropic induced damage, permits also to account for masonry unilateral effects. The orthotropic nature of the Tension-Compression Damage Model adopted in the fictitious space is demonstrated. This feature, both with the assumption of two distinct damage criteria for tension and compression, does not permit to term the fictitious space as “isotropic”. Therefore, the proposed formulation turns the original concept of “mapping the real space into an isotropic fictitious one” into the innovative and more general one of “mapping the real space into a favourable (or convenient) fictitious one”. Validation of the model is carried out by means of comparisons with experimental results on different types of orthotropic masonry. The model is fully formulated for the 2-dimensional case. However, it can be easily extended to the 3-dimensional case. It provides high algorithmic efficiency, a feature of primary importance when analyses of even large scale masonry structures are carried out. To account for this requisite it adopts a strain-driven formalism consistent with standard displacement-based finite element codes. The implementation in finite element programs is straightforward. Finally, a localized damage model for orthotropic materials is formulated. This is achieved by means of the implementation of a crack tracking algorithm, which forces the crack to develop along a single row of finite elements. Compared with the smeared cracking approach, such an approach shows a better capacity to predict realistic collapsing mechanisms. The resulting damage in the ultimate condition appears localized in individual cracks. Moreover, the results do not suffer from spurious mesh-size or mesh-bias dependence. The numerical tool is finally validated via a finite element analysis of an in-plane loaded masonry shear wall.

La mayoría de las estructuras de hormigón armado se someten a solicitaciones combinadas de esfuerzos axiles, flexión, cortante y torsión. La fisuración del hormigón, plastificación de las armaduras y otros efectos no-lineales hacen que las secciones transversales de estos elementos presenten un comportamiento anisótropo que deriva en el acoplamiento de los esfuerzos normales y tangenciales. Es decir, esfuerzos normales o momentos flectores pueden producir deformaciones de corte y vice versa. Aunque en algunas ocaciones, esta interacción es considerada de forma simplificada en el dimensionamiento de estructuras, hasta el momento no se ha realizado un análisis profundo de los efectos acoplados en secciones de forma arbitraria bajo cargas 3D generales utilizando modelos de fibras.El objetivo principal de esta tesis es generalizar el análisis de secciones de hormigón armado mediante fibras, de forma que se pueda reproducir la res-puesta no-lineal acoplada frente a esfuerzos normales y tangenciales bajo solicitaciones tridimensionales generales. De igual forma, se pretende obtener, para los esfuerzos cortantes y torsión, la misma capacidad de representación de geometrías y combinación de materiales que ofrecen los modelos de fibras para esfuerzos de flexo-compresión.La primera problemática estriba en representar adecuadamente la cinemática de la sección transversal. Con la excepción de las deformaciones normales contenidas en el plano de la sección, no existe una teoría cinemática que a priori pueda dar la distribución del resto de deformaciones o tensiones en la sección, sin dejar de satisfacer las condiciones de equilibrio interno o continuidad entre las fibras que componen la misma.Por otra parte, para materiales anisótropos, como el hormigón fisurado, en general todos los esfuerzos internos pueden estar acoplados. Además, es preciso considerar la distorsión de la sección transversal para satisfacer el equilibrio entre fibras.El problema se aborda de forma general, considerando una sección de forma y materiales cualesquiera. Se parte del problema diferencial de equilibrio de un sólido con el que se ha podido deducir un sistema de equilibrio entre fibras (equilibrio a nivel sección). Se puede demostrar que éste es complementario al problema estándar de vigas. El sistema complementario permite recuperar información tridimensional que normalmente se pierde al resolver un problema de vigas.Posteriormente, se propone una solución interna del problema complementario, en la que el alabeo y la distorsión de la sección quedan expresados como una función de las deformaciones generalizadas de una viga: deformaciones axil y cortantes, curvaturas de flexión y torsión. No son necesarios grados de libertad adicionales a nivel estructura ni hipótesis a-priori sobre la forma de los campos de deformación o tensión interna.A partir de la formulación teórica, se desarrolla un modelo de elementos finitos plano de la sección transversal. El modelo está preparado para servir como respuesta constitutiva de cualquier tipo de elemento viga en sus puntos de integración. %Se evita así la necesidad de realizar un modelo de elementos sólidos de toda la barra para estudiar la respuesta frente a una combinación general de esfuerzos normales y tangenciales.Se implementan una serie de modelos constitutivos para distintos materiales. En particular, se implementa un modelo constitutivo triaxial para hormigón fisurado, considerando la anisotropía inducida por la fisuración e incluyendo la superficie de rotura según un criterio multiaxial.La formulación seccional es validada mediante varios casos de estudio teóricos y experimentales. La respuesta no-lineal acoplada bajo diversas combinaciones de esfuerzos normales y tangenciales es reproducida con precisión, lo cual queda patente tanto en las curvas esfuerzo-deformación obtenidas como en las matrices de rigidez seccionales.Finalmente, se recopilan las conclusiones derivadas de la presente investigación y seofren recomendaciones para futuros trabajos. / Most RC structures are subjected to combined normal and tangential forces, such as bending, axial load, shear and torsion. Concrete cracking, steel yielding and other material nonlinearities produce an anisotropic sectional response that results in a coupling between the effects of normal and shear forces, i.e. normal force or bending moments may produce shear strains and vice versa. Although this interaction is sometimes taken into account, in a simplified manner, in the design of RC structures, a deep analysis of the coupling effects of RC sections using fiber models has not yet been made for arbitrary shape sections under general 3D loading.The main objective of this thesis is to generalize the fiber-like sectional analysis of reinforced concrete elements, to make it capable of considering the coupled non-linear response under tangential and normal internal forces, from a general 3D loading.Similarly, it is desired to obtain, for torque and shear forces, the same capacity and versatility in reproducing complex geometry and materials combination that fiber-like sectional representations offers for bending and stretching.The first problematic lies in finding a proper representation of the section's kinematics under such general loading. Except for in-plane normal strains, there is no single kinematical theory capable of a-priori representing the correct distribution of the others strains or stresses satisfying, at the same time, inter-fiber equilibrium and continuity. On the other hand, for rather anisotropic materials, such as cracked concrete, all internal forces are, in general, coupled. It is also required that distortion is allowed for the section's kinematics in order to guarantee satisfaction of internal equilibrium.The problem is dealt in a general form considering arbitrary shaped sections and any material behaviour. Starting from the differential equilibrium of a solid, an inter-fiber equilibrium system (equilibrium at the sectional level) was deduced. This system, which is complementary to the standard equilibrium problem of a beam-column, allows to recuperate information of the three-dimensional problem that is generally lost when solving a beam problem.Further, a solution of the equilibrium at the sectional level is proposed in which the section's warping and distortion are posed as a function of the generalized beam-column strains (axial and shear strains, bending and torsion curvatures). No additional degrees of freedom are required at the structural level nor a-priori hypotheses on the distribution of the internal strains or stresses.After the theoretical formulation, a planar finite element model for cross-sectional analysis is developed. The model can be used as a constitutive law for general beam column elements at their integration points.A series of constitutive models have been implemented for several materials. In particular, a triaxial constitutive model for cracked concrete is implemented considering crackinduced anisotropy and a multiaxial failure criterion.The sectional formulation is validated by means of various theoretical and experimental case studies. Non-linear coupled response under normal and tangential internal forces is reproduced with accuracy, as can be seen both in the predicted internal force-strain curves and in the sectional stiffness matrixes.Finally, the conclusions drawn from the current research are summarized andrecomendations for future works are given.

Presented is an overview of a semi-analytical solution which was developed to study the response of internally pressurized elliptical composite cylinders with clamped boundaries. Using a geometrically linear analysis and the solution scheme, the response of a quasi-isotropic elliptical cylinder is compared with the response of a quasi-isotropic circular cylinder in order to study the effects of elliptical geometry. The distinguishing features of the response of an elliptical cylinder are the inward normal displacement of the cross section at the ends of the major diameter that occur despite the outward force of the internal pressure, the presence of circumferential displacements, and the presence of inplane shear strains. These effects lead to spatial variations, including sign reversals, of a number of displacement, strain, and curvature responses. The responses of a quasi-isotropic elliptical cylinder evaluated using a geometrically linear analysis are then compared to the responses evaluated using a geometrically nonlinear analysis. It is shown that geometric nonlinearities tend to flatten certain responses at the ends of the minor diameter, and reduce the magnitude of certain responses in the boundary region. To study the influence of material orthotropy, the responses of axially-stiff and circumferentially-stiff elliptical cylinders evaluated using geometrically nonlinear analyses are examined. It is shown that in some instances material orthotropy can be used to mitigate the influence of the elliptical geometry and make particular responses look like those of a circular cylinder. An evaluation of failure using the maximum stress and Hashin failure criteria and geometrically linear and nonlinear analyses is presented for elliptical cylinders. These failure criteria involve interlaminar shear stresses which are computed by integrating the equilibrium equations of elasticity through the thickness of the cylinder wall. The failure criteria are used to assess the mode of failure (e.g., tensile or compressive fiber or matrix modes), the location of failure, and the pressure at failure. Both criteria predict first failure to occur at the clamped boundaries because of matrix cracking. The predicted failure pressures and circumferential locations are very similar for the two criteria, and the nonlinear analyses predict slightly higher pressures at somewhat different circumferential locations. First fiber failure is also considered. For this failure the two criteria predict similar failure scenarios for the linear analyses, but they differ in their predictions for the nonlinear analyses. Specifically, using the maximum stress criterion, the circumferentially-stiff elliptical cylinder is predicted to fail due to fiber compression, but the Hashin criterion predicts failure to be due to fiber tension, and at a different circumferential location. Also, first fiber failure pressures are at least a factor of two greater than the first matrix failure pressure. / Master of Science

L’étude des structures raidies est un sujet de recherche récurrent. En effet, celles-ci sont présentes dans de nombreuses applications industrielles. Leur utilisation offre de nombreux avantages notamment du point de vue de l’allègement, critère qui est particulièrement important dans l’industrie automobile par exemple. Dans un tout autre domaine, la table d’harmonie de piano constitue un exemple typique de ces structures raidies et les problèmes relevés par les professionnels du domaine sont nombreux. Aujourd’hui, les ressources numériques permettent de prendre en compte de nombreux phénomènes dans les modèles développés avec pour conséquence, une difficulté d’interprétation entre les données d’entrée et de sortie. En considérant moins de paramètres à la fois, les modèles simplifiés présentent alors l’avantage de pouvoir en séparer la participation dans le rendu global. En prenant comme point de départ une plaque simplement supportée rectangulaire orthotrope dite « spéciale » dans laquelle la table est inscrite, la forme de la table d’harmonie est recréée par ajout d’une densité de ressorts ponctuels dans le domaine complémentaire. Par couplage avec des superstructures collées sur chaque face, il est possible de déterminer le comportement vibratoire de l’instrument ainsi que le rayonnement acoustique à partir des impédances de rayonnement d’une plaque simplement supportée bafflée. Ce modèle analytique est par la suite couplé à une corde et résolu dans le domaine temporel et présente alors l’avantage de pouvoir évaluer perceptivement l’impact de modifications structurelles. L’originalité d’un tel calcul tient dans le fait que des forces d’interactions assurent la continuité entre les soussystèmes et deviennent des inconnues du problème, comme dans les problèmes de contacts ou de frottements. Enfin, la prise en compte des petites non-linéarités géométriques de la corde sera faite en les considérant comme des seconds membres des équations du mouvement, ce qui permet de conserver la notion de modeslinéaires / Ribbed structures are a current subject of study and they are often used in many industrial domain. Indeed, there use offers some advantages as for example in automotive domain to reduce the weight of vehicule that is an important actual constrain. In the musical domain ribbed structures are also often used and musical instruments mixed structural, perceptive and subjective considerations. In that domain, one of the main difficulties is to translate the language between the science and the music and to make equivalence between phenomena describe by the musicians and scientific indicators. The piano soundboard is a typical example of these ribbed structures and difficulties reported by piano makers are many. Nowadays, numerical computing resources allow to take into account some phenomena in modeling but as a consequence, it appears a “black box effect” between input and output data. Taking into account less parameters, reduced models offer the advantage to separate the influence of each parameter in the final result. Taking into account as starting point a simply supported rectangular plate with special orthotropy, the edges of the soundboard are described by an addition of punctual springs in the fictive domain. Coupling to it some stiffeners on the both faces, it is possible to calculate the vibratory behavior of the instrument such as the acoustic radiation through the acoustical radiation impedances of the baffled simply supported plate. This simplified analytical modeling is representative of phenomena found in the literature and is also well adapted to parametrical studies. So, it highlights the space localizations of vibration, due to the conception of the instrument, and allows studying the influence of structural parameters on the mobility along the bridge, the acoustical radiated power for example. In order to qualify, with a reduced computing cost, the influence of structural parameters on the perceived sound, this analytical model for ribbed structure is coupled to a string and solved in time domain.

A general procedure for the evaluation of the mechanical properties of structural joints, named component method, is now available from intensive research works at the European level. This procedure allows the analytical prediction of the resistance, but also of the stiffness and the deformation capacity, of structural joints under external forces (axial or shear forces, bending moments ). The component method is nowadays integrated as a reference procedure in two European design codes, respectively for steel structures (EC3 [EN1993]) and steel-concrete composite structures (EC4 [EN1994]). However, its potential scope is much larger and present studies are aimed to apply to situations as joints in fire, joints under seismic loading, joints under exceptional loads (Robustness Project) . More recently, a research project [CTI-2004] has succeeded in applying the component method to the investigation of the elastic behaviour of mechanical joints in timber construction. That is the result of the collaboration of CTIB-TCHN (Belgian Institute for Wood Technology) and University of Liège. The main principle of the component method is the following: identification of constitutive components subjected to tension, compression or shear in the joint; determination of the mechanical behaviour of these individual components; "assembling" components so as to derive the mechanical properties of the whole joint. In the present paper, timber joints with dowel fasteners are considered. Two components may be identified: "dowel" component (dowel fastener in bending and shear); embedding component (timber member in embedding). The "dowel" component is known from past researches, whereas little information is available for the embedding component. EC5 [EN1995] proposes formulation to predict the behaviour for joints composed of these two components; but it only depends on two factors: the dowel diameter and the timber density. The influence probably significant of the grain direction (material strongly anisotropy) and the thickness of the connected members are for instance neglected. Experimental, numerical and analytical investigations have recently been performed by University of Liège in collaboration with CTIB-TCHN so as to propose another formulation more precise for joints. Experimental results, performed by CTIB-TCHN, have been used as reference for the development of numerical model and, then, analytical model. The application of the component method to the prediction of the elastic behaviour of timber joints consist of two steps: "local" investigation on components that is to develop analytical models for the prediction of the elastic behaviour of components; "global" investigation on joints that is to develop analytical models for the prediction of the elastic behaviour of joints. The application of the component method to timber joints with dowel fasteners is a first step towards the use of this concept in future to others mechanical joints (screw, punched metal plate, nail, bolt ). In this case, others components may be derived to cover the field of application expected.

Work occupies computational modelling mechanical behavior elastomers and composits with rubber matrix and their utilization for compute model of tire creation. MATADOR tire 165/65 R13 Axisymetric 2D model was created in two geometric variants. For the computational modeling is applying the Finite element method (FEM). The model was in different variants distinctive grade of modeling material. At first was done inner pressure analyst impact on deformation of each of model. Then was count influence on tire load with angular velocity meanwhile with inner pressure. The impact thickness of tire protector layer on global deformation tyre casing was verified too.

La modélisation du trafic routier dans un réseau dense et de grande étendue nécessite un grand nombre de données, si bien qu'une modélisation par arcs est impossible en pratique. Pour simplifier le problème, une idée est d'agréger les tronçons du réseau en un continuum sur lequel le trafic routier s'écoule comme un fluide surfacique. Cette modélisation est qualifiée de bidimensionnelle. Même si la structure géométrique détaillée du réseau est perdue, une telle modélisation évite la description très fine du trafic sur un réseau dans lequel les points de mesure ne sont pas en nombre suffisant pour permettre une évaluation exhaustive de l'état du trafic. Une série de travaux commencée dans les années 1980 a permis de dégager quelques concepts importants. Cependant, ces travaux n'ont pas résolu les problèmes de modélisation fondamentaux : comment déduire et modéliser des comportements globaux à partir de comportements locaux (flux sur un axe, interactions aux intersections) ? Deux modèles bidimensionnels de trafic sont développés. Le premier modèle est statique. Le trafic s'écoule dans des directions de propagation privilégiées (orthogonales). Le modèle prend en compte l'équilibre entre l'offre de déplacement du réseau et une demande élastique de déplacement des usagers. Les principales sorties sont constituées en chaque point et pour chaque destination par les débits directionnels et les coûts de déplacement. Le deuxième modèle est dynamique. L'écoulement du trafic est décrit au niveau de cellules élémentaires du réseau dans lesquelles on définit les notions d'offre et de demande. À partir d'une loi comportementale obtenue sur un réseau discret, on écrit l'équation dynamique de conservation du trafic routier en tout point d'un réseau anisotrope / Traffic road modelling in the dense network of a wide area needs a large amount of data. It renders a keen modelling unmanageable in practice. To simplify the problem, an idea is to aggregate the network links as continuous medium where traffic road flows as a fluid on a surface. This modelling is called two-dimensional. Even if the detailed geometric structure of the network is lost, such a modelling avoids the traffic keen description on a network where measure points are not numerous enough to allow an exhaustive evaluation of traffic state. A series of articles started in the 80's have highlighted relevant concepts. Nevertheless, these previous works have not solved fundamental modelling issues: how to deduce and model global behaviours basing on local behaviours (flow on an axe, interactions at intersections). Two models are developed. The first model is static. Traffic road flows on privileged directions (orthogonal). The model takes into account the balance between the network trip supply and the users' elastic trip demand. The principal outputs are, for each point of the network and each destination, the directional flows and the trip costs. The second model is dynamical. The description of traffic flows stands at the level of elementary traffic cells, where concepts of supply and demand are defined. With a behaviour law deduced from a discrete network, we establish the conservation dynamic equation of road traffic at each point of an anisotropic network

Treten im Beton mehraxiale Spannungszustände auf, führen diese gegenüber einer einaxialen Beanspruchung zu einer signifikanten Änderung des Materialverhaltens. Neben einer festigkeitssteigernden bzw. -abmindernden Wirkung ergeben sich ebenfalls große Unterschiede im Spannungs-Dehnungs-Verhalten. Zur effizienten Konzipierung von Betonstrukturen unter komplexen Beanspruchungszuständen ist daher die Kenntnis des veränderten Materialverhaltens notwendig. Zur experimentellen Bestimmung des Spannungs-Dehnungs-Verhaltens eines Ultrahochleistungsbetons mit einer einaxialen Druckfestigkeit von über 170 N/mm² wurden mehraxiale Belastungsversuche an würfelförmigen Probekörpern durchgeführt. Die Untersuchung umfasste insgesamt 35 zwei- und dreiaxiale Spannungsverhältnisse unter proportionaler Laststeigerung mit vorrangiger Betrachtung von Zug-Druck-Druck-Beanspruchungen. Für die Einleitung der Zugbeanspruchungen in die Prüfkörper wurde eine neue Methode entwickelt, bei der mittels einbetonierter Schrauben die Belastung auf den Beton übertragen wird. Die Bestimmung des Verformungsverhaltens erfolgte im Inneren der Probekörper mit sechs tetraederförmig angeordneten Faser-Bragg-Gittern. Die somit direkt gemessenen Dehnungen ermöglichen die nachträgliche Berechnung der Komponenten des Dehnungstensors des Bezugskoordinatensystems. Für den untersuchten Ultrahochleitsungsbeton fallen die auf die einaxiale Druckfestigkeit bezogenen mehraxialen Festigkeitswerte mit zunehmendem hydrostatischen Druckspannungsanteil der Beanspruchung geringer aus als bei Normalbetonen. Weiterhin weist das Verformungsverhalten eine größere Sprödigkeit gegenüber Normalbetonen auf, so dass auch unter dreiaxialen Druckspannungszuständen die Probekörper schlagartig versagen. Aus den gemessenen Spannungs-Dehnungs-Linien werden neben den maximalen Festigkeiten die Festigkeitswerte an der Elastizitätsgrenze, der Affinitätsgrenze sowie beim Volumenminimum der Probekörper bestimmt. Zur Approximation dieser charakteristischen Werte wurde eine Grenzflächenbeschreibung entwickelt und an den Versuchsergebnissen kalibriert. Des Weiteren erfolgte die Zusammenstellung einer Datenbank mit in der Literatur verfügbaren mehraxialen maximalen Festigkeitswerten von Betonen mit einaxialen Druckfestigkeiten von 10 N/mm² bis 180 N/mm² und die Kalibrierung des entwickelten Modells zur Grenzflächenbeschreibung in Abhängigkeit der einaxialen Druckfestigkeit. Die bei der Kalibrierung der Grenzfläche für einzelne Betonfestigkeitsklassen bestimmten Freiwerte hängen dabei stark von den vorliegenden Versuchsdaten und speziell vom Wertebereich der hydrostatischen Spannungsanteile der maximalen Beanspruchungen ab. Die Approximation des Spannungs-Dehnungs-Verhaltens der mehraxial beanspruchten Probekörper erfolgt mittels eines schädigungsbasierten Materialgesetzes. Hierbei wird für den anfänglich isotropen Beton zum einen eine lastinduzierte isotrope Schädigung und zum anderen eine lastinduzierte orthotrope Schädigung angenommen, die von den auftretenden Hauptdehnungen abhängig ist. Mit dem entwickelten Materialgesetz werden sehr gute Übereinstimmungen mit den gemessenen Spannungs-Dehnungs-Linien erreicht, so dass sich ebenfalls eine gute Vorhersage der maximalen Festigkeitswerte ergibt. / Concrete under multiaxial stress states shows significant changes of the material behaviour compared to uniaxial loading. Besides strength increasing and decreasing effects, also great differences in the stress-strain behaviour occur. In order to design concrete structures efficiently concerning complex stress states, the knowledge about the modified material behaviour is necessary. To determine experimentally the stress-strain behaviour of an ultra high performance concrete with a uniaxial compressive strength of about 170 N/mm², multiaxial loading tests on cubic-shaped specimens were carried out. Altogether, the investigation contained 35 biaxial and triaxial stress ratios under proportionally increasing load with primarily tension-compression-compression loadings. Applying the tensile load on the specimen, a new method was developed, which uses screws embedded in the concrete to transfer the loading. The deformations were measured by using six tetrahedron-shaped arranged Fibre Bragg Gratings inside the concrete specimen. Subsequently, with the directly measured strains the components of the strain tensor of the reference coordinate system could be determined. For the investigated ultra high performance concrete the increase of the multiaxial strength, referring to the uniaxial compressive strength, decreases compared to normal strength concrete with the increasing hydrostatic stress component of the load. Moreover, the deformation behaviour shows an increased brittleness compared to normal strength concrete, so that even under triaxial compressive stress states the specimens fail abruptly. Besides the ultimate strength, from the measured stress-strain curves the strength at the proportional limit, at the limit of affinity as well as at the minimum volume of the specimen is determined. To approximate these characteristic values, a description of a hypersurface is developed and calibrated with the test results. Furthermore, a database with multiaxial ultimate strength values of concretes with uniaxial compressive strengths between 10 N/mm² to 180 N/mm² available from literature was compiled and a calibration of the developed hypersurface model depending on the uniaxial compressive strength was carried out. Thereby, the obtained values of arbitrary parameters of individual concrete strength classes depend severely on the available test results, especially on the range of values of the hydrostatic stress component of the ultimate strength. The approximation of the stress-strain behaviour of the multiaxial loaded specimens is carried out by means of a damage-based material law. For this purpose, concerning the initially isotropic concrete, a load-induced isotropic and orthotropic damage depending on the principle strains is assumed. With the developed material law, very good accordance with the measured stress-strain curves could be achieved, so that also results in a good approximation of the ultimate concrete strength.

La technique des goujons collés dans les structures bois allie performances mécaniques et esthétisme. Elle répond au besoin de conservation du bâti pour le domaine de la rénovation et au besoin d’assemblages de plus en plus rigides et résistants en construction neuve avec l’avènement des structures bois de grandes ampleurs et l’arrivée de nouveaux produits tels que le bois lamellé croisé (CLT). De nombreuses études de caractérisation mécanique ont été menées au cours de ces 35 dernières années. L’enquête menée auprès des professionnels et experts en vue de l’introduction de cette technique dans l’Eurocode 5 a néanmoins révélé des attentes et lacunes particulières vis-à-vis des connaissances sur le comportement des assemblages multi-tiges, le choix des adhésifs, le comportement à long terme du collage et les techniques de contrôle qualité. Les travaux de cette thèse, menés en collaboration entre l’université de Bordeaux et l’Université Laval (Québec, Canada), ont pour objectif d’apporter des réponses sur le comportement mécanique des adhésifs, et assemblages sous sollicitation thermique et d’initier les développements sur le comportement à long terme et les assemblages multi-tiges. Les principaux résultats sont issus de travaux expérimentaux et numériques valorisés sous la forme d’articles scientifiques. Les essais mécaniques dynamiques menés sur une colle polyuréthane (PUR) et une résine époxy (EPX) entre 30°C et 120°C mettent en avant des différences importantes de raideur et de dégradations entre les deux adhésifs. Exposés à la chaleur, les assemblages sont eux aussi affectés avec des pertes de raideur et de résistance dès 40°C, en amont de la température de transition vitreuse et de la dégradation des adhésifs. Amené à suivre le régime de température extérieur, l’assemblage doit par sa conception être isolé afin de prévenir les risques aux états limites de service (ELS) et aux états limites ultimes (ELU). Très peu de données sont disponibles dans la littérature sur le comportement à long terme des assemblages goujons collés. L’étude du fluage ajoute une dimension temporelle aux problématiques de raideur. La base de données expérimentales collectée dans ces travaux repose sur 12 campagnes d’essais de 1 à 2 mois en conditions climatiques régulées en température et humidité relative, ou variables (service classe 1). Chargement et climat influent sur le glissement, mais aussi sur la durée de vie. Si des disparités sont présentes entre EPX et PUR, les études menées à 20°C et 50°C révèlent à nouveau l’importance de prendre en considération la température, notamment aux ELS. L’étude en régime variable illustre de son côté la sensibilité du système aux variations de température et d’humidité. L’étude de la répartition des efforts en fonction de la raideur des goujons est choisie comme fil conducteur pour investiguer le passage vers des éléments multi-tiges. Un modèle 3D mono-tige aux éléments finis est tout d’abord développé pour prendre en compte l’orthotropie du bois et regarder l’influence du centre de moelle jusqu’à présent négligée. Par la suite ce modèle est étendu aux configurations multi-tiges afin de simuler différentes configurations d’essais. Un prototype d’essai de caractérisation mécanique des assemblages multi-tiges est finalement présenté. / Glued-in rods have successfully been used for connections or reinforcement of timber structures. With the development of tall and large timber buildings and new products such as Cross Laminated Timber (CLT), there is an increasing need for connections that provide high stiffness and strength. Timber connections using glued in rods have a general aesthetic appeal, and take advantage of the structural adhesives that provide a high stiffness and load capacity. During the last 30 years, multiple applications have been developed for renovation and new construction. In parallel, numerous investigations have characterized the mechanical performance of these connections, but harmonized design rules are not available. A lack of knowledge on the multiple-rod connections, on the choice of adhesives, on long-term effects and on the control of quality is identified by a recent study and must be investigated before a new submission to Eurocode 5. This thesis is performed in the framework of collaboration between Université de Bordeaux (France) and Université Laval (Canada). Objectives are focused on the mechanical performance of adhesives and glued-in rod connections under elevated temperature, on the development of creep tests and the investigations of multiple-rod connections. This study combines experimental and finite element modelling results, which are presented in the form of scientific articles. A Dynamic Mechanical Analysis (DMA) on one polyurethane (PUR) and one epoxy (EPX) adhesives followed by static tensile tests on the connections with small-diameter steel glued-in rods have been conducted at different temperatures. High differences in stiffness are observed between the two adhesives. Glued-in rod connections with the EPX and PUR adhesives demonstrate significant losses of stiffness and resistance beyond 40°C, before the glass transition temperature of the adhesive. Following the outside temperature, connections must be insulated to prevent excessive slip and risk of failure at the Serviceability Limits State (SLS) and at the Ultimate Limit States (ULS). Few data on the long-term effects are available in the literature. In this work, 12 campaigns of creep tests, adding time effects to previous work, have been performed. Glued-in rod connections were tested during one to two months in a conditioning room where temperature and relative humidity were controlled or in a room with a variable climate representative of service class 1. Disparities were observed between EXP and PUR. Creep tests conducted at 20°C and 50°C revealed an important role of the temperature, particularly for SLS. Creep tests in variable climate illustrated the sensibility of connections to the temperature and humidity variations. The stiffness and stress distribution in multiple-rod connections were investigated. First, a 3D model was developed to observe the influence of the wood orthotropy on the stiffness and stress distribution in a connection with a single rod. Then, the model was extended to multiple-rod connections to simulate different loads and boundary conditions. Finally, a prototype of a connection with multiple glued-in rods was tested.

Treten im Beton mehraxiale Spannungszustände auf, führen diese gegenüber einer einaxialen Beanspruchung zu einer signifikanten Änderung des Materialverhaltens. Neben einer festigkeitssteigernden bzw. -abmindernden Wirkung ergeben sich ebenfalls große Unterschiede im Spannungs-Dehnungs-Verhalten. Zur effizienten Konzipierung von Betonstrukturen unter komplexen Beanspruchungszuständen ist daher die Kenntnis des veränderten Materialverhaltens notwendig. Zur experimentellen Bestimmung des Spannungs-Dehnungs-Verhaltens eines Ultrahochleistungsbetons mit einer einaxialen Druckfestigkeit von über 170 N/mm² wurden mehraxiale Belastungsversuche an würfelförmigen Probekörpern durchgeführt. Die Untersuchung umfasste insgesamt 35 zwei- und dreiaxiale Spannungsverhältnisse unter proportionaler Laststeigerung mit vorrangiger Betrachtung von Zug-Druck-Druck-Beanspruchungen. Für die Einleitung der Zugbeanspruchungen in die Prüfkörper wurde eine neue Methode entwickelt, bei der mittels einbetonierter Schrauben die Belastung auf den Beton übertragen wird. Die Bestimmung des Verformungsverhaltens erfolgte im Inneren der Probekörper mit sechs tetraederförmig angeordneten Faser-Bragg-Gittern. Die somit direkt gemessenen Dehnungen ermöglichen die nachträgliche Berechnung der Komponenten des Dehnungstensors des Bezugskoordinatensystems. Für den untersuchten Ultrahochleitsungsbeton fallen die auf die einaxiale Druckfestigkeit bezogenen mehraxialen Festigkeitswerte mit zunehmendem hydrostatischen Druckspannungsanteil der Beanspruchung geringer aus als bei Normalbetonen. Weiterhin weist das Verformungsverhalten eine größere Sprödigkeit gegenüber Normalbetonen auf, so dass auch unter dreiaxialen Druckspannungszuständen die Probekörper schlagartig versagen. Aus den gemessenen Spannungs-Dehnungs-Linien werden neben den maximalen Festigkeiten die Festigkeitswerte an der Elastizitätsgrenze, der Affinitätsgrenze sowie beim Volumenminimum der Probekörper bestimmt. Zur Approximation dieser charakteristischen Werte wurde eine Grenzflächenbeschreibung entwickelt und an den Versuchsergebnissen kalibriert. Des Weiteren erfolgte die Zusammenstellung einer Datenbank mit in der Literatur verfügbaren mehraxialen maximalen Festigkeitswerten von Betonen mit einaxialen Druckfestigkeiten von 10 N/mm² bis 180 N/mm² und die Kalibrierung des entwickelten Modells zur Grenzflächenbeschreibung in Abhängigkeit der einaxialen Druckfestigkeit. Die bei der Kalibrierung der Grenzfläche für einzelne Betonfestigkeitsklassen bestimmten Freiwerte hängen dabei stark von den vorliegenden Versuchsdaten und speziell vom Wertebereich der hydrostatischen Spannungsanteile der maximalen Beanspruchungen ab. Die Approximation des Spannungs-Dehnungs-Verhaltens der mehraxial beanspruchten Probekörper erfolgt mittels eines schädigungsbasierten Materialgesetzes. Hierbei wird für den anfänglich isotropen Beton zum einen eine lastinduzierte isotrope Schädigung und zum anderen eine lastinduzierte orthotrope Schädigung angenommen, die von den auftretenden Hauptdehnungen abhängig ist. Mit dem entwickelten Materialgesetz werden sehr gute Übereinstimmungen mit den gemessenen Spannungs-Dehnungs-Linien erreicht, so dass sich ebenfalls eine gute Vorhersage der maximalen Festigkeitswerte ergibt. / Concrete under multiaxial stress states shows significant changes of the material behaviour compared to uniaxial loading. Besides strength increasing and decreasing effects, also great differences in the stress-strain behaviour occur. In order to design concrete structures efficiently concerning complex stress states, the knowledge about the modified material behaviour is necessary. To determine experimentally the stress-strain behaviour of an ultra high performance concrete with a uniaxial compressive strength of about 170 N/mm², multiaxial loading tests on cubic-shaped specimens were carried out. Altogether, the investigation contained 35 biaxial and triaxial stress ratios under proportionally increasing load with primarily tension-compression-compression loadings. Applying the tensile load on the specimen, a new method was developed, which uses screws embedded in the concrete to transfer the loading. The deformations were measured by using six tetrahedron-shaped arranged Fibre Bragg Gratings inside the concrete specimen. Subsequently, with the directly measured strains the components of the strain tensor of the reference coordinate system could be determined. For the investigated ultra high performance concrete the increase of the multiaxial strength, referring to the uniaxial compressive strength, decreases compared to normal strength concrete with the increasing hydrostatic stress component of the load. Moreover, the deformation behaviour shows an increased brittleness compared to normal strength concrete, so that even under triaxial compressive stress states the specimens fail abruptly. Besides the ultimate strength, from the measured stress-strain curves the strength at the proportional limit, at the limit of affinity as well as at the minimum volume of the specimen is determined. To approximate these characteristic values, a description of a hypersurface is developed and calibrated with the test results. Furthermore, a database with multiaxial ultimate strength values of concretes with uniaxial compressive strengths between 10 N/mm² to 180 N/mm² available from literature was compiled and a calibration of the developed hypersurface model depending on the uniaxial compressive strength was carried out. Thereby, the obtained values of arbitrary parameters of individual concrete strength classes depend severely on the available test results, especially on the range of values of the hydrostatic stress component of the ultimate strength. The approximation of the stress-strain behaviour of the multiaxial loaded specimens is carried out by means of a damage-based material law. For this purpose, concerning the initially isotropic concrete, a load-induced isotropic and orthotropic damage depending on the principle strains is assumed. With the developed material law, very good accordance with the measured stress-strain curves could be achieved, so that also results in a good approximation of the ultimate concrete strength.