La défaillance des matériaux et structures d'ingénierie peut être considéré comme le résultat d'une interaction complexe entre différents phénomènes physiques tels que la nucléation des cavités, les microfissures, les microvides et d'autres processus irréversibles. Ces micro-défauts se fondent éventuellement en une ou plusieurs macro-fissures conduisant à une diminution de la capacité portante et finalement à une défaillance de la structure considérée. La prévention des défaillances des structures et des composants structurels a toujours été un sujet important et une préoccupation majeure en ingénierie. Cette thèse vise à représenter une défaillance localisée dans des matériaux non linéaires sans dépendance de maillage. Un intérêt particulier sera le cas de l’adoucissement dynamique des déformations. Les phénomènes localisés sont pris en compte en utilisant l'approche des discontinuités embarquées fortes dans laquelle le champ de déplacement est amélioré pour capturer la discontinuité. Sur la base de cette approche, on a d'abord développé un modèle unidimensionnel de barres élasto-plastique capable de représenter une défaillance pour des matériaux ductiles avec un durcissement combiné dans une zone de processus de fracture FPZ et un adoucissement avec des discontinuités fortes encastrées. Les résultats comparant le modèle unidimensionnel proposé aux travaux (semi-) analytiques sont présentés. Il a été démontré que la stratégie proposée offre des solutions indépendantes de maillage. La déformation augmente dans le domaine de l’adoucissement avec une diminution simultanée de la contrainte. Le problème se décharge élastiquement à l'extérieur de la zone d’adoucissement de déformation. L'énergie dissipée se trouve à disparaître. Le modèle a également été comparé à un modèle de dommage unidimensionnel capable de représenter la fracture dynamique de la barre d'endommagementélasto-endommagée dans la zone de traitement de fracture - FPZ et de adoucissement avec de discontinuités fortes encastrées pour trouver un bon accord entre deux modèles. Un modèle d'éléments finis bidimensionnel a été développé, capable de décrire à la fois le mécanisme de dommage diffus accompagné d'un durcissement initial et d'une réponse d’adoucissement ultérieure de la structure. On a analysé les résultats de plusieurs simulations numériques effectuées sur des essais mécaniques classiques sous des charges progressivement croissantes telles que le test Brésilien ou le test de flexion en trois points. Le cadre de dynamique proposé est montré pour augmenter la robustesse de calcul. On a constaté que la direction finale des macro-fissures est assez bien prédite et que l'influence des effets d'inertie sur les solutions obtenues est assez modeste notamment en comparaison entre différentes mailles. Ce modèle bidimensionnel a été étendu plus loin dans le modèle bidimensionnel de discontinuité intégrée en viscodamage pour aider à explorer brièvement la mise en œuvre du schéma de point intermédiaire de second ordre qui peut fournir des résultats améliorés sous limitation de la régularisation visqueuse du modèle de dégâts localisés. / Failure of engineering materials and structures can be considered as a result of a complex interplay between different physical phenomena such as nucleation of cavities, microcracks,microvoids and other irreversible processes. These micro-defects eventually coalesce into one or more macro-cracks leading to a decrease in the load-bearing capability and finally, to failure of the structure under consideration. Prevention of failure of structures and structural parts has always been a critical subject and a major concern in engineering. This thesis aims to represent localized failure in non linear materials without mesh dependency. Of special interest will be the case of dynamic strain-softening. Localized phenomena are taken into account by using the embedded strong discontinuities approach in which the displacement field is enhanced to capture the discontinuity. Based upon this approach, a one-dimensional model for elasto-plastic bar capable of representing failure for ductile materials with combined hardening in FPZ-fracture process zone and softening with embedded strong discontinuities was first developed. Results comparing the proposed one-dimensional model to (semi-) analytical works are presented. It was shown that the proposed strategy provides mesh independent solutions. Strain increases in the softening domain with a simultaneous decrease of stress. The problem unloads elastically outside the strain softening region. The strain energy is found to vanish. The model was also compared with a one dimensional damage model capable of representing the dynamic fracture for elasto-damage bar with combined hardening in fracture process zone - FPZ and softening with strong embedded discontinuities to find a good agreement between two models. A two-dimensional finite element model was developed, capable of describing both the diffuse damage mechanism accompanied by initial strain hardening and subsequent softening response of the structure. The results of several numerical simulations, performed on classical mechanical tests under slowly increasing loads such as Brazilian test or three-point bending test were analyzed. The proposed dynamics framework is shown to increase computational robustness. It was found that the final direction of macro-cracks is predicted quite well and that influence of inertia effects on the obtained solutions is fairly modest especially in comparison among different meshes. This two-dimensional model was expanded further into the two dimensional continuum viscodamage-embedded discontinuity model to help briefly explore the implementation of the second order mid-point scheme that can provide improved results under limitation of viscous regularization of localized failure damage model.

Concrete, which was hitherto considered as a brittle material, has shown much better softening behavior after the post peak load than anticipated. This behavior of concrete did put the researchers in a quandary, whether to categorize concrete under brittle materials or not. Consequently concrete has been called a quasi-brittle material. Fracture mechanics concepts like Linear elastic fracture mechanics (LEFM) and Plastic limit analysis applicable to both brittle and ductile materials have been applied to concrete to characterize the fracture behavior. Because of quasi-brittle nature of concrete, which lies between ductile and brittle response and due to the presence of process zone ahead of crack/notch tip instead of a plastic zone, it is found that non-linear fracture mechanics (NLFM) principles are more suitable than linear elastic fracture mechanics (LEFM) principles to characterize fracture behavior. Fracture energy, fracture process zone (FPZ) size and the behavior of concrete during fracture process are the fracture characteristics, which are at the forefront of research on concrete fracture. Another important output from the research on concrete fracture has been the size effect. Numerous investigations, through mathematical modeling and experiments, have been carried out and reported in literature on the effect of size on the strength of concrete and fracture energy. Identification of the sources of size effect is of prime importance to arrive at a clear analytical model, which gives a comprehensive insight into the size effect. With the support of an unambiguous theory, it is possible to incorporate the size effects into codes of practices of concrete design. However, the theories put forth to describe the size effect do not seem to follow acceptable regression. After introduction in Chapter-1 and literature survey in Chapter-2, Chapter-3 details the study on size effect through three point bend (TPB) tests on 3D geometrically similar specimens. Fracture behavior of beams with smaller process zone size in relation to ligament dimension approaches LEFM. The fracture energy obtained from such beams is said to be size independent. In the current work Size effect law (Bazant et al. 1987) is used on beams geometrically similar in three dimensions with the depth of the largest beam being equal to 750mm, and size independent fracture energy G Bf is obtained. In literature very few results are available on the results obtained from testing geometrically similar beams in three dimensions and with such large depth. In the current thesis the results from size effect tests yielded average fracture energy of 232 N/m. Generally the fracture energies obtained from 2D-geometrically similar specimens are in the range of 60-70 N/m as could be seen in literature. From 3D-geometrically similar specimens, the fracture energies are higher. The reason is increased peak load, could be due to increased width. The RILEM fracture energy Gf , determined from TPB tests, is said to be size dependent. The assumption made in the work of fracture is that the total strain energy is utilized for the fracture of the specimen. The fracture energy is proportional to the size of the FPZ, it also implies that FPZ size increases with increase in (W−a) of beam. This also means that FPZ is proportional to the depth W for a given notch to depth ratio, because for a given notch/depth, (W−a) which is also W(1 − a ) is proportional to W`because (1 − a ) is a constant. WWThis corroborates the fact that fracture energy increases with size. Interestingly, the same conclusion has been drawn by Abdalla & Karihaloo (2006). They have plotted a curve relating fracture process zone length and overall depth the beam. In the present study a new method namely Fracture energy release rate method is suggested. In the new method the plot of Gf / (W−a) versus (W−a) is obtained from a set of experimental results. The plot is found to follow power law and showed almost constant value of Gf / (W−a) at larger ligament lengths. This means that fracture energy reaches a constant value at large ligament lengths reaffirming that the fracture energy from very large specimen is size independent. The new method is verified for the data from literature and is found to give consistent results. In a quasi-brittle material such as concrete, a fracture process zone forms ahead of a pre-existing crack (notch) tip before the crack propagates from the tip. The process zone contains a scatter of micro-cracks, which coalesce into one or more macro-cracks, which eventually lead to fracture. These micro-cracks and macro-cracks release stresses in the form of acoustic waves having different amplitudes. Each micro or macro crack formation is called an acoustic emission (AE) event. Through AE technique it is possible to locate the positions of AE events. The zone containing these AE events is termed the fracture process zone (FPZ). In Chapter-4, a study on the evolution of fracture process zone is made using AE technique. In the AE study, the fracture process zone is seen as a region with a lot of acoustic emission event locations. Instead of the amplitudes of the events, the absolute AE energy is used to quantify the size of the process zone at various loading stages. It has been shown that the continuous activities during the evolution of fracture process zone correspond to the formation of FPZ, the size of which is quantified based on the density of AE events and AE energy. The total AE energy released in the zone is found to be about 78% of the total AE energy released and this is viewed as possible FPZ. The result reasonably supports the conclusion, from Otsuka and Date (2000) who tested compact tension specimens, that zone over which AE energy is released is about 95% can be regarded as the fracture process zone. As pointed out earlier, among the fracture characteristics, the determination of fracture energy, which is size independent, is the main concern of research fraternity. Kai Duan et al. (2003) have assumed a bi-linear variation of local fracture energy in the boundary effect model (BEM) to showcase the size effect due to proximity of FPZ to the specimen back boundary. In fact the local fracture energy is shown to be constant away from boundary and reducing while approaching the specimen back boundary. The constant local fracture energy is quantified as size independent fracture energy. A relationship between Gf , size independent fracture energy GF , un-cracked ligament length and transition ligament length was developed in the form of equations. In the proposed method the transition ligament length al is taken from the plot of histograms of energy of AE events plotted over the un-cracked ligament. The value of GF is calculated by solving these over-determined equations using the RILEM fracture energies obtained from TPB tests. In chapter-5 a new method involving BEM and AE techniques is presented. The histogram of energy of AE events along the un-cracked ligament, which incidentally matches in pattern with the local fracture energy distribution, assumed by Kai Duan et al. (2003), along the un-cracked ligament, is used to obtain the value of GF , of course using the same equations from BEM developed by Kai Duan et al. (2003). A critical observation of the histogram of energy of AE events, described in the previous chapter, showed a declining trend of AE event pattern towards the notch tip also in addition to the one towards the specimen back boundary. The pattern of AE energy distribution suggests a tri-linear rather than bi-linear local fracture energy distribution over un-cracked ligament as given in BEM. Accordingly in Chapter-6, GF is obtained from a tri-linear model, which is an improved bi-linear hybrid model, after developing expressions relating Gf , GF , (W−a) with two transition ligament lengths al and blon both sides. The values of Gf , and GF from both bi-linear hybrid method and tri-linear method are tabulated and compared. In addition to GF , the length of FPZ is estimated from the tri-linear model and compared with the values obtained from softening beam model (SBM) by Ananthan et al. (1990). There seems to be a good agreement between the results. A comparative study of size independent fracture energies obtained from the methods described in the previous chapters is made. The fracture process in concrete is another interesting topic for research. Due to heterogeneity, the fracture process is a blend of complex activities. AE technique serves as an effective tool to qualitatively describe the fracture process through a damage parameter called b-value. In the Gutenberg-Richter empirical relationship log 10N=a−bM, the constant ‘b’ is called the b-value and is the log linear slope of frequency-magnitude distribution. Fault rupture inside earth’s crust and failure process in concrete are analogous. The b-value, is calculated conventionally till now, based on amplitude of AE data from concrete specimens, and is used to describe the damage process. Further, sampling size of event group is found to influence the calculated b-value from the conventional method, as pointed out by Colombo et al. (2003). Hence standardization of event group size, used in the statistical analysis while calculating b-value, should be based on some logical assumption, to bring consistency into analytical study on b-value. In Chapter-7, a methodology has been suggested to determine the b-value from AE energy and its utilization to quantify fracture process zone length. The event group is chosen based on clusters of energy or quanta as named in the thesis. Quanta conform to the damage stages and justify well their use in the determination of the b-value, apparently a damage parameter and also FPZ length. The results obtained on the basis of quanta agree well with the earlier results.