Ce travail concerne l'analyse de la vulnérabilité des structures de protection et des ouvrages sensibles en béton soumis à des actions dynamiques sévères (impacts, explosions) dues à des risques anthropiques d'origine accidentelle ou non. L'objet est la mise au point d'outils prévisionnels de simulation capables de décrire de manière objective le comportement dynamique du béton. Pour cela, une approche numérique novatrice reposant sur la méthode des Eléments Discrets (MED) est développée. Une première partie de cette thèse concerne la simulation des essais quasi-statiques de compression et traction uniaxiales. Une loi de transfert de moment (LTM) a été introduite pour pallier au problème de fragilité en compression simple. Ensuite, la procédure d'identification des paramètres du modèle modifié a été optimisée pour bien reproduire le comportement macroscopique du béton. Enfin, le modèle a été validé en représentant correctement le comportement quasi-statique de plusieurs types de béton. La deuxième partie du travail traite la simulation des essais de traction dynamique du béton aux barres de Hopkinson. Les résultats ont montré la nécessité de prendre l'effet de vitesse de déformation dû au matériau pour bien reproduire le comportement expérimental. Ensuite, Les paramètres du modèle permettant de reproduire cet effet de vitesse ont été identifiés. Enfin, des essais avec des taux de déformation très élevés ont été simulés et les résultats numériques ont été en accord avec le comportement observé expérimentalement. / This work concerns the analysis of the vulnerability of sensitive concrete structures subjected to severe dynamic actions such as impacts due to natural hazards or human factors. The object is to develop a numerical tool that can describe objectively the dynamic behaviour of concrete. Then, a 3D discrete element method (DEM) was developed and used to perform the analysis. The first part of this thesis focuses on the simulation of quasi-static uniaxial compression and traction tests. A moment transfer law (MTL) was introduced to overcome the problem of brittle compressive behavior. Then, the identification procedure of the modified DEM model has been optimized in order to reproduce very well the macroscopic behaviour of concrete. Finally, the model has been validated by representing properly the real quasi-static behavior of different types of concrete. The second part of the study deals with the simulation of the dynamic Hopkinson traction bar tests of concrete. The results showed that a local rate effect has to be introduced to reproduce the strain rate dependency, which would then be a material-intrinsic effect. Then, the parameters of the model have been identified. Finally, simulations were run at high strain rates and showed consistent results with respect to experimental behaviour.

Mitigation of the blast risk associated with terrorist attacks and accidental explosions threatening critical infrastructure has become a topic of great interest in the civil engineering community, both in Canada and abroad. One method of mitigating blast risk is to retrofit vulnerable structures to resist the impulsive effects of blast loading. A comprehensive re-search program has been undertaken to develop fibre reinforced polymer (FRP) retrofit methodologies for structural and non-structural elements, specifically reinforced concrete slabs and walls, subjected to blast loading. The results of this investigation are equally valid for flexure dominant reinforced concrete beams subject to blast effects. The objective of the research program was to generate a large volume of research data for the development of blast-resistant design guidelines for externally bonded FRP retrofit systems. A combined experimental and analytical investigation was performed to achieve the objectives of the program. The experimental program involved the construction and simulated blast testing of a total of thirteen reinforced concrete wall and slab specimens divided into five companion sets. These specimens were subjected to a total of sixty simulated explosions generated at the University of Ottawa Shock Tube Testing Facility. Companion sets were designed to study one- and two-way bending, as well as the performance of specimens with simply-supported and fully-fixed boundary conditions. The majority of the specimens were retrofitted with externally bonded carbon fibre reinforced polymer (CFRP) sheets to improve overall load-deformation characteristics. Specimens within each companion set were subjected to progressively increasing pressure-impulse combinations to study component behaviour from elastic response up to inelastic component failure. The blast performance of companion as-built and retrofitted specimens was quantified in terms of measured load-deformation characteristics, and observed member behaviour throughout all stages of response. The results show that externally bonded FRP retrofits are an effective retrofit technique to improve the blast resistance of reinforced concrete structures, provided that debonding of the composite from the concrete substrate is prevented. The test results also indicate that FRP retrofitted reinforced concrete structures may survive initial inbound displacements, only to failure by moment reversals during the negative displacement phase. The experimental test data was used to verify analytical techniques to model the behaviour of reinforced concrete walls and slabs subjected to blast loading. The force-deformation characteristics of one-way wall strips were established using inelastic sectional and member analyses. The force-deformation characteristics of two-way slab plates were established using commonly accepted design approximations. The response of all specimens was computed by explicit solution of the single degree of freedom dynamic equation of motion. An equivalent static force procedure was used to analyze the response of CFRP retrofitted specimens which remained elastic after testing. The predicted maximum displacements and time-to-maximum displacements were compared against experimental results. The analysis indicates that the modelling procedures accurately describe the response characteristics of both retrofitted and unretrofitted specimens observed during the experiment.

Background：Aortic stenosis (AS) is the commonest valve disease in the West. Aortic valve replacement (AVR) remains the only available management for AS and results in improved symptoms and recovery of ventricular functions. In addition, it is well known that AVR results in disruption of LV function mainly in the form of reversal of septal motion as well as depression of right ventricular (RV) systolic function. The aim of this thesis was to study, in detail, the early and mid-term response of ventricular function to AVR procedures (surgical and TAVI) as well as post operative patients’ exercise capacity. Methods：We studied LV and RV function by Doppler echocardiography and speckle tracking echocardiography (STE) in the following 4 groups; (1) 30 severe AS patients (age 62±11 years, 19 male) with normal LV ejection fraction (EF) who underwent AVR, (2) 20 severe AS patients (age 79±6 years, 14 male) who underwent TAVI, (3) 30 healthy controls (age 63±11 years, 16 male), (4) 21 healthy controls (age 57±9 years, 14 male) who underwent exercise echocardiography. Results: After one week of TAVI, the septal radial motion and RV tricuspid annulus peak systolic excursion (TAPSE) were not different from before, while surgical AVR had significantly reversed septal radial motion and TAPSE dropped by 70% compared to before. The extent of the reversed septal motion correlated with that of TAPSE (r=0.78, p<0.001) in the patients as a whole after AVR and TAVI (Study I). Compared with controls, the LV twist function was increased in AS patients before and normalized after 6 months of surgical AVR. In controls, the LV twist correlated with LV fractional shortening (r=0.81, p<0.001), a relationship which became weak in patients before (r=0.52, p<0.01) and after AVR (r=0.34, p=ns) (Study II). After 6 months of surgical AVR, the reversed septal radial motion was still significantly lower than before. The septal peak displacement also decreased and its time became prolonged. In contrast, the LV lateral wall peak displacement increased and the time to peak displacement was early. The accentuated lateral wall peak displacement correlated with the septal peak displacement time delay (r=0.60, p<0.001) and septal-lateral time delay (r=0.64, p<0.001) (Study III). In 21 surgical AVR patients who performed exercise echocardiography, the LV function was normal at rest but different from controls with exercise. At peak exercise, oxygen consumption (pVO2) was lower in patients than controls. Although patients could achieve cardiac output (CO) and heart rate (HR) similar to controls at peak exercise, the LV systolic and early diastolic myocardial velocities and strain rate as well as their delta changes were significantly lower than controls. pVO2 correlated with peak exercise LV myocardial function in the patients group only, and the systolic global longitudinal strain rate (GLSRs) at peak exercise was the only independent predictor of pVO2 in multivariate regression analysis (p=0.03) (Study IV). Conclusion: Surgical AVR is an effective treatment for AS patients, but results in reversed septal radial motion and reduced TAPSE. The newly developed TAVI procedure maintains RV function which results in preservation of septal radial motion. In AS, the LV twist function is exaggerated, normalizes after AVR but loses its relationship with basal LV function. While the reversed septal motion results in decreased and delayed septal longitudinal displacement which is compensated for by the accentuated lateral wall displacement and the time early. These patients remain suffering from limited exercise capacity years after AVR.

Concrete structures may be subjected to dynamic loading during their service life. Understanding the dynamic properties of concrete structures is becoming critical because of the increased concern about the dynamic loading of both civilian and military structures, and especially, the recent increase in terrorist attacks on structures. Fiber reinforced concrete (FRC) is known to exhibit superior performance in its post-peak energy absorption capacity, (i.e., toughness) under flexural and tensile loading. However, the behavior of fiber reinforced concrete under compressive impact has not previously been investigated. In the present research, the response of fiber reinforced concrete was investigated over the full strain rate regime, from static loading to high strain rate loading, and finally to impact loading. The compressive toughness of FRC under static loading was studied using an existing Japanese standard (JSCE SF-5). Then, a test method for FRC under compressive impact loading was developed, involving the use of a high speed video camera system to measure the deformation of FRC cylinders under compressive impact. The strain rate sensitivity of FRC in both flexure and compression was also fully investigated. FRC was found to have higher strengths under impact loading (both flexural and compressive) than under static loading. The compressive toughness under impact loading increased due to the high peak load and the high strain capacity. FRC under flexural impact loading showed a greater strength improvement than under static flexure. FRC displays a much higher Dynamic Improvement Factor (DIF) under flexural impact than under compressive impact. It gave an overall higher performance under impact than under static loading. It also exhibited a higher strain rate sensitivity than plain concrete in both compression and flexure. Damage analysis, in terms of loss of strain energy, was carried out based on damage mechanics principles. Damage was found to increase with increasing strain rate. A new constitutive model was proposed to account for the relationship between DIF (Comp) and strain rate and the data derived from the model were found to be consistent with the experimental results.

Previous experimental observations have shown that the pseudoelastic response of NiTi shape memory alloys (SMA) is localized in nature and proceeds through nucleation and propagation of localized deformation bands. It has also been observed that the mechanical response of SMAs is strongly affected by loading rate and cyclic degradation. These behaviors significantly limit the accurate modeling of SMA elements used in various devices and applications. The aim of this work is to provide engineers with a constitutive model that can accurately describe the dynamic, unstable pseudoelastic response of SMAs, including their cyclic response, and facilitate the reliable design of SMA elements. A 1-D phenomenological model is developed to simulate the localized phase transformations in NiTi wires during both loading and unloading. In this model, it is assumed that the untransformed particles located close to the transformed regions are less stable than those further away from the transformed regions. By consideration of the thermomechanical coupling among the stress, temperature, and latent heat of transformation, the analysis can account for strain-rate effects. Inspired by the deformation theory of plasticity, the 1-D model is extended to a 3-D macromechanical model of localized unstable pseudoelasticity. An important feature of this model is the reorientation of the transformation strain tensor with changes in stress tensor. Unlike previous modeling efforts, the present model can also capture the propagation of localized deformation during unloading. The constitutive model is implemented within a 2-D finite element framework to allow numerical investigation of the effect of strain rate and boundary conditions on the overall mechanical response and evolution of localized transformation bands in NiTi strips. The model successfully captures the features of the transformation front morphology, and pseudoelastic response of NiTi strip samples observed in previous experiments. The 1-D and 3-D constitutive models are further extended to include the plastic deformation and degradation of material properties as a result of cyclic loading.

The aim of this thesis is to provide microstructural and mechanical characterisation of α-β titanium alloys exposed to a range of thermo-mechanical conditions, in particular under-going high rate deformation at elevated temperatures, representative of the Linear Friction Welding (LFW) manufacturing process. Three α-β titanium alloys provided by Rolls-Royce are studied: Ti-64 blade, disc and Ti-6246 disc. Ti-64 and Ti-6246 show complex deformation behaviour with strain, strain rate and temperature, especially near the transus temperature, where the low temperature α phase is transformed into the high temperature β phase. The microstructure and mechanical properties evolve in an interconnected fashion, and understanding this mutual influence is necessary to better predict the behaviour of these alloys. Characterisation of the mechanical properties was performed through uniaxial compression tests at strain rates from 0.001 to 3000 s^-1, using an Instron screw-driven machine at quasi-static rates, a servo-hydraulic machine at medium rates and a Split-Hopkinson Pressure Bar and a drop-weight tower at high strain rates. The tests were performed over a range of temperatures from room temperature to 1300 °C. The main focus was on high strain rate and high temperature tests, with the development of a gravity driven direct impact Hopkinson bar, referred as a drop-weight system, which is intended to evaluate the mechanical response of metals to high strain rate loading at temperatures up to c. 1300 °C. The design and principles of operation of the system are presented, along with calibration and validation data. Preliminary tests were performed on stock Ti-64, heated at two rates: 1 and 20 °C s^-1. The evolution of the mechanical properties was analysed, focussing on the strain rate, temperature and phases dependencies. Characterisation of the microstructure was realised by performing interrupted compression tests, first at room temperature, three plastic strains, 4%, 10% and 20%, and two different strain rates, 0.001 and 2000 s^-1; then at 4% plastic strain, a strain rate of 2000 s^-1 and three elevated temperatures, 700, 900 and 1100 °C. A better understanding of the microstructure evolution with strain, strain rates and temperature, including the macrotexture and microtexture of the specimens, was obtained using Electron Backscatter Diffraction (EBSD) to characterise the texture of the undeformed and deformed materials. The better understanding of the flow stress and microstructural evolution of both Ti-64 and its individual α and β phases with various strain rates and temperatures is intended to be used in the development of more accurate models representing the behaviour of these alloys. Predicting the microstructure evolution and then the mechanical properties of a material is essential to optimise the final mechanical properties of the alloys when welded by manufacturing processes such as the LFW process.

Ce travail concerne l'analyse de la vulnérabilité des structures de protection et des ouvrages sensibles en béton soumis à des actions dynamiques sévères (impacts, explosions) dues à des risques anthropiques d'origine accidentelle ou non. L'objet est la mise au point d'outils prévisionnels de simulation capables de décrire de manière objective le comportement dynamique du béton. Pour cela, une approche numérique novatrice reposant sur la méthode des Eléments Discrets (MED) est développée. Une première partie de cette thèse concerne la simulation des essais quasi-statiques de compression et traction uniaxiales. Une loi de transfert de moment (LTM) a été introduite pour pallier au problème de fragilité en compression simple. Ensuite, la procédure d'identification des paramètres du modèle modifié a été optimisée pour bien reproduire le comportement macroscopique du béton. Enfin, le modèle a été validé en représentant correctement le comportement quasi-statique de plusieurs types de béton. La deuxième partie du travail traite la simulation des essais de traction dynamique du béton aux barres de Hopkinson. Les résultats ont montré la nécessité de prendre l'effet de vitesse de déformation dû au matériau pour bien reproduire le comportement expérimental. Ensuite, Les paramètres du modèle permettant de reproduire cet effet de vitesse ont été identifiés. Enfin, des essais avec des taux de déformation très élevés ont été simulés et les résultats numériques ont été en accord avec le comportement observé expérimentalement. / This work concerns the analysis of the vulnerability of sensitive concrete structures subjected to severe dynamic actions such as impacts due to natural hazards or human factors. The object is to develop a numerical tool that can describe objectively the dynamic behaviour of concrete. Then, a 3D discrete element method (DEM) was developed and used to perform the analysis. The first part of this thesis focuses on the simulation of quasi-static uniaxial compression and traction tests. A moment transfer law (MTL) was introduced to overcome the problem of brittle compressive behavior. Then, the identification procedure of the modified DEM model has been optimized in order to reproduce very well the macroscopic behaviour of concrete. Finally, the model has been validated by representing properly the real quasi-static behavior of different types of concrete. The second part of the study deals with the simulation of the dynamic Hopkinson traction bar tests of concrete. The results showed that a local rate effect has to be introduced to reproduce the strain rate dependency, which would then be a material-intrinsic effect. Then, the parameters of the model have been identified. Finally, simulations were run at high strain rates and showed consistent results with respect to experimental behaviour.

Les argiles compactées sont utilisées dans de nombreuses applications, notamment en géotechnique et en géotechnique de l’environnement, en raison de leur faible perméabilité, et de leurs propriétés de rétention notamment. Cependant, une fois en place, ces matériaux pourraient être exposés à des sollicitations thermiques et/ou hydriques, à long et très long terme. L’objectif principal de ce travail est de quantifier expérimentalement l’impact de ces sollicitations sur la compressibilité d’une argile compactée, et plus particulièrement son fluage. Avec cet objectif, des cellules œdométriques à température contrôlée entre 5 et 70°C ont été développées. Deux types d’œdomètre à succion contrôlée par les méthodes osmotique et solutions salines ont été employés dans une gamme de succion comprise entre 0 et 20,8 MPa, et à une température constante de 20°C. Ces dispositifs ont permis d’étudier le fluage jusqu’à une contrainte verticale de 3600 kPa. L’étude s’est concentrée sur le comportement d’une argile moyennement gonflante compactée. Les résultats obtenus ont tout d’abord montré que la contrainte de préconsolidation apparente σ’p diminue à mesure que la température augmente. Le coefficient de fluage Cαe augmente avec la température, cet effet étant plus particulièrement marqué à des contraintes plus élevées. Une relation linéaire entre le coefficient de fluage Cαe et l’indice de compression incrémental C*c a été observée dans la plage de contraintes considérée et le rapport (Cαe /C*c) dépend de la température. Ensuite, deux approches expérimentales complémentaires (essais de fluage par paliers ou à vitesse de déformation contrôlée) ont mis en évidence la dépendance des caractéristiques de fluage vis-à-vis de la succion du sol. Par ailleurs, la contrainte de préconsolidation apparente σ’p augmente avec l’augmentation de la vitesse de déformation έv et de la succion. En revanche, l’indice de compression Cc et le coefficient de fluage Cαe varient d’une manière non monotone avec une valeur maximale sous une succion de 3,5 et de 2 MPa, respectivement. L’évolution de ces paramètres apparaît fortement liée à la structure interne du sol. L’analyse de la variation de σ’p avec έv et de Cαe avec Cc a montré que la relation Δlog σ’p /Δlog έv = Cαe/Cc est également valable pour le sol argileux compacté étudié dans les cas saturés et non saturés / Compacted clays are used in many applications, including geotechnical and environmental geotechnical applications, due to their low permeability and retention properties. However, once in place, these materials could be exposed to thermal and/or water variations in the long and very long term. The main objective of this work is to experimentally quantify the impact of these variations on the compressibility of a compacted clay, and in particular its creep. With this objective, oedometric cells with controlled temperatures between 5 and 70°C were developed. Two types of oedometers with suction controlled by osmotic and saline methods were used in a suction range of 0 to 20.8 MPa, and at a constant temperature of 20°C. These devices were used to study creep up to a vertical stress of 3600 kPa. The study focused on the behavior of a moderately swelling compacted clay. The obtained results first showed that the yield stress σ’p decreases as the temperature increases. The creep coefficient Cαe increases with temperature, this effect being particularly marked at higher stresses. A linear relationship between the creep coefficient Cαe and the incremental compression index C*c was observed within the stress range considered and the ratio (Cαe /C*c) is temperature dependent. Then, two complementary experimental approaches (creep tests by steps or at controlled strain rate) highlighted the dependence of creep characteristics on soil suction. In addition, the yield stress σ’p increases with increasing strain rate έv and suction. In contrast, the compression index Cc and the creep coefficient Cαe vary in a non-monotonic manner with a maximum value under suction of 3.5 and 2 MPa, respectively. The evolution of these parameters appears to be strongly related to the internal structure of the soil. Analysis of the variation of σ’p with έv and Cαe with Cc showed that the relationship Δlog σ’p /Δlog έv =Cαe/Cc is also valid for the studied compacted clayey soil in saturated and unsaturated states. In conclusion, the results of this work allowed information to be gathered for better understanding the compressibility and creep behavior of compacted clayey soils as a function of temperature and suction

This study investigates the effect of strain rate on fracture properties of Ferritic-Pearlitic Ductile Iron. A series of dynamic three point bending tests, with various load application rates, are conducted on Charpy V-notch specimens, in room temperature and approximately -18 °C. The tests are performed in a custom-made fixture and during the tests, force and displacement data are recorded. A XFEM (Extended Finite Element Method) model of the test setup has been established and material data from the tests are used as input to the model. The test results show a strong dependency of the strain rate regarding the force needed for crack initiation. Moreover, it can be concluded that low temperature makes the material very brittle, even at low load application rates.

Above true strain rates of $10^4$ s$^{-1}$ FCC metals exhibit a rapid increase in strength. Understanding of the physical mechanisms behind this strength transition is hindered by the number and interdependence of candidate mechanisms. Broadly, contributions to strength can be split into `instantaneous' effects and the more permanent `structural' ones. In this thesis a series of experiments are presented which are designed to separate the two types of contribution. Chapter 2 outlines the basics of dislocation plasticity, based on the seminal works of Taylor and Orowan. It then progresses on to discuss recent experimental and theoretical work on the understanding of slip as avalanche behaviour. Chapter 3 summarises traditional modelling approaches for instantaneous strength contributions which are routinely applied below $10^4$ s$^{-1}$. It then continues on to outline a number of different approaches which have been adopted to attempt to explain and model the strength transition. Chapter 4 outlines the methods used in the earliest stages of the study: Instron and split Hopkinson pressure bar methods. Both methods are well established, and cover the majority of the range of rates under study. Emphasis is made on minimising experimental sources of error, and subsequently accounting for those which are unavoidable. Finally, the specimen material is introduced and is shown to be fit for purpose. Chapter 5 presents a set of mechanical tests of specimens at strain rates between $10^4-10^5$~s$^{-1}$. The softening of the specimens with increased temperature is observed to increase with strain rate, both in absolute terms and when normalised to the 300 K measurement for each strain rate. The observations are most easily explained if the strength transition is due to an increase in early stage work hardening, however, some anomalous behaviours remain. Chapter 6 introduces a new experimental technique; direct impact Hopkinson pressure bars, required to perform experiments shown to be necessary by the results of Chapter 5. Photon Doppler velocimetry is applied to the projectiles used in experiments, removing one of the most significant flaws of the technique, and creating a more confident basis with which to perform further experimental work. Chapter 7 presents a series of `jump tests' at ambient temperatures. Specimens are deformed at strain rates ranging from $10^{-2}$ to $10^5$~s$^{-1}$ to a fixed strain of 0.1, then reloaded to yield at a strain rate of $10^{-1}$. The yield point at reload is shown to have the same rapid upturn as seen when the specimens were deforming at high rates, providing strong evidence that the increase in strength is due to changes in the underlying dislocation structure, rather than a dynamic effect, as it remains even when the high strain rate is removed. Chapter 8 continues on from the conclusions of Chapter 7. Jump tests are expanded to a variety of temperatures and strains, to provide a more complete characterisation of metal behaviour. No dramatic change in the saturation of work hardening is observed to coincide with the increase in early stage work hardening. Chapter 9 discusses discrepancies between contemporary high rate models and recent developments in the understanding of plasticity being an avalanche process. Potential consequences of incorporating avalanche plasticity into high rate models are explored. Particular attention is paid to Brown's observation that based on quasi static observations of avalanche behaviour, the formation of dislocation avalanches will begin to fail at strain rates of approximately $10^4$ s$^{-1}$. Consequences of the progressive breakdown of avalanche behaviour are discussed with respect to the experimental observations presented in earlier chapters. In Chapter 10, we will discuss the key conclusions of the work. Finally, a number of avenues are proposed for building upon the current work both theoretically and experimentally.