This study focuses on the evaluation of seismic safety of masonry buildings in Turkey by using fragility curves. Fragility curves for masonry buildings are generated by two behavior modes for load bearing walls: in-plane and out-of-plane. By considering the previous research and site investigations, four major parameters have been used in order to classify masonry buildings with in-plane behavior mode. These are number of stories, strength of load-bearing wall material, regularity in plan and the arrangement of walls (required length, openings in walls, etc.). In addition to these four parameters, floor type is also taken into account for the generation of fragility curves by considering out-of-plane behavior mode. During generation of fragility curves, a force-based approach has been used. In this study there exist two limit states, or in other words three damage states, in terms of base shear strength for in-plane behavior mode and flexural strength for out-of-plane behavior mode. To assess the seismic vulnerability of unreinforced masonry buildings in Turkey, generated fragility curves in terms of in-plane behavior, which is verified by damage statistics obtained during the 1995 Dinar earthquake, and out-of-plane behavior, which is verified by damage statistics obtained during the 2010 Elazig earthquake, is combined. Throughout the analysis, ground motion uncertainty, material variability and modeling uncertainty have also been considered. In the final part of the study, a single-valued parameter, called as &lsquo / vulnerability score&rdquo / , has been proposed in order to compare the seismic safety of unreinforced masonry buildings in Fatih sub province of Istanbul and to assess the influence of out-of-plane behavior together with the in-plane behavior of these existing masonry buildings.

There are many advantages of masonry construction like widespread geographic availability in many forms, colors and textures, comparative cheapness, fire resistance, thermal and sound insulation, durability, etc. For such reasons, it is still a commonly used type of residential construction in rural and even in urban regions. Unfortunately, its behavior especially under the effect of earthquake ground motions has not been identified clearly because of its complex material nature. Hence, the masonry buildings with structural deficiencies belong to the most vulnerable class of structures which have experienced heavy damage or even total collapse in previous earthquakes, especially in developing countries like Turkey. This necessitates new contemporary methods for designing safer masonry structures or assessing their performance. Considering all these facts, this study aims at the generation of a new performance-based technique for unreinforced brick masonry structures. First, simplified formulations are recommended to estimate idealized capacity curve parameters of masonry components (piers) by using the finite element analysis results of ANSYS and regression analysis through SPSS software. Local limit states for individual masonry piers are also obtained. Then, by combining the component behavior, lateral capacity curve of the masonry building is constructed together with the global limit states. The final step is to define seismic demand of the design earthquake from the building through TEC2007 method. By using this simple technique, a large population of masonry buildings can be examined in a relatively short period of time noting that the performance estimations are quite reliable since they are based on sophisticated finite element analysis results.

Les présents travaux, à caractère numérico-expérimental, visent à approfondir la connaissance relative au comportement de murs maçonnés, principalement ceux renforcés par matériaux composites vis-à-vis de sollicitations dans leur plan (flexion composée). Ils s'inscrivent dans le cadre de la réhabilitation du patrimoine bâti vis-à-vis du risque sismique notamment du fait de la reconsidération du zonage en France rendu depuis peu plus exigeant. Aussi, d'un point de vue technologique, cette thèse vise à apprécier, évaluer et hiérarchiser l'intérêt et les potentialités de solutions de renforcement mobilisant des matériaux composites, à base polymère ainsi que des composites textile-mortier de nouvelle génération, couplés à des ancrages mécaniques ayant vocation à mieux valoriser ce type d'options. Deux échelles d'analyse ont été retenues dans le cadre de la partie expérimentale. A l'échelle du matériau, dans le but de caractériser finement les matériaux constitutifs de la maçonnerie de briques de béton creux et de générer des jeux de données aussi pertinents que fiables, notamment en prévision de la modélisation numérique, des essais de compression uni-axiale et de push-out à l'échelle « réduite » ont été conduits et ont permis de souligner, en accord avec la littérature, la nécessité de tenir compte de l'interaction brique-mortier, de consolider la compréhension des mécanismes d'endommagement et de rupture des éléments de maçonnerie tout en mettant en lumière l'importance relative des dispersions obtenues. A l'échelle du composant de structure, une campagne expérimentale de flexion composée portant sur six murs, dont un mur témoin, a été conduite sous sollicitations de flexion composée dans le plan des murs avec pour impératif la nécessité de restituer les conditions limites et de sollicitations sous séisme, tout en limitant le champ de l'étude au volet statique monotone en vue d'éprouver les solutions valorisées (matériaux composites et ancrage).Cette partie a permis, audelà de la mise en avant des bonnes dispositions en termes de capacité portante, d'une part, de caractériser comparativement le comportement de ces éléments tant à l'échelle globale (déplacement, capacité de déformation et de dissipation d'énergie, etc.) qu'à l'échelle locale (endommagement, déformations localisées, etc.) via une instrumentation judicieuse, et d'autre part de cerner l'importance des ancrages mécaniques vis-à-vis des sollicitations étudiées. L'approche numérique, de type éléments finis, a été mise à profit dans un deuxième temps pour tenter de restituer le comportement des murs (à l'échelle locale et globale). Sur la base d'une lecture bibliographique critique c'est l'approche micromécanique qui a eu nos faveurs. La modélisation a été conduite en trois dimensions (3D) à l'aide du logiciel ANSYS. Ainsi, les briques et le mortier sont modélisés indépendamment mais liés parfaitement. Deux variantes ont été proposées pour la modélisation de la maçonnerie saine et leur adéquation a été évaluée. Le premier modèle s'appuie sur un couplage entre le modèle de béton d'Ansys en traction et un comportement multilinéaire en compression pour modéliser le mortier, les briques sont supposées pourvues d'un comportement élasto-plastique bilinéaire pour lequel la résistance en compression de la brique est le seuil de la phase élastique. Dans le deuxième modèle, plus en phase avec les constats expérimentaux, seul le comportement des briques est modifié en introduisant un comportement post-pic adoucissant. En ce qui concerne la modélisation des murs renforcés par matériaux composites, ces derniers (FRP et TRC) ont été considérés comme parfaitement liés au substrat de maçonnerie. Toutefois, si le renfort de type FRP est modélisé par un comportement homogène orthotrope, le TRC, rarement modélisé jusqu'à lors, est simulé via deux approches (homogène et hétérogène) dans le but d'apprécier leur pertinence... / This study, using both experimental and numerical approaches, will help to better understand the behaviour of masonry walls. It especially focuses on walls reinforced with composite materials under in-plane loading conditions. In France, more stringent seismic design requirements for building structures have taken effect. So, this research has been initiated in an effort to define reliable strengthening techniques. The selected reinforcement materials are (1) – fiber reinforced polymer (FRP) strips using E-glass and carbon fabrics and (2) – an emerging cementbased matrix grid (CMG) system. The composite strips are mechanically anchored into the foundations of the walls to improve their efficiency. The experimental program involves different levels of analysis. Small-scale models of block masonry structures, carried out with less than ten bricks, are tested. The objective is to obtain a coherent set of data, characterizing the elementary components (hollow bricks and mortar) and their interface, in order to provide realistic values of the parameters required in the foreseen modelling. Shear bond strength has been obtained from triplets and 7-uplets and compressive masonry strength from running bond prisms. These experimental results improve the knowledge of the main damage mechanisms and failure modes of masonry but they suffer from high scattering. At laboratory (large) scale, six walls have been submitted to shear-compression tests - five of them are reinforced and the last one acts as a reference. All the walls share the same boundary and compressive loading conditions, which are chosen to ensure a representative simulation of a seismic solicitation. Nevertheless, masonry walls performances and anchor efficiency are only evaluated under monotonic lateral loadings. A comparative study on global behavior (displacements, deformation capacity, energy dissipation,…) as well as on local mechanisms (local strains, damage,…) is carried out and highlights in particular that strengthened walls exhibit a high increase in shear resistance. Moreover, a 3D finite-element analysis using ANSYS has been performed to help understand the behaviour of unreinforced and strengthened walls. A micro-mechanical approach is adopted: bricks and mortar are modelled separately and linked together by a perfect bond. The Ansys concrete model, capable of cracking, is coupled with a multi-linear plasticity model in compression to describe mortar joints. In a first attempt, bricks exhibit a bilinear behavior law where the brick compressive strength is the elastic threshold; but this model fails to reproduce the resistances of the strengthened walls. To compensate for these overestimations and capture the experimental resistances, a post-pic softening behaviour is preferred for the bricks. To model strengthened walls, all composite strips are supposed to be perfectly linked with the masonry and a linear elastic law is used for the FRP reinforcements. TRC strips are either described by means of a linear law or represented using a heterogeneous approach where matrix and textile grids are modelled separately. In this case, grids are represented using a smeared approach and are embedded within the matrix mesh. So, displacement compatibility is totally satisfied between the textile and the cementitious matrix. The proposed numerical model tends to underestimate walls capacity deformation but ultimate loads and failure modes are in coherence with experimental results. Finally, existing analytical methods have been applied to assess unreinforced and strengthened walls performances. The results are then compared with the experimental data and a critical review is proposed. Existing models could be refined by taking into account more realistic behaviour laws and by relying on energy approaches to better reproduce dissipative mechanisms of TRC materials