L’objectif de cette thèse est de contribuer à la détection de défauts dans des pièces réalisées par le procédé de fabrication additive de déposition métallique par laser (DML), en vue d’un contrôle in situ. Le contrôle in situ envisagé porte sur les derniers cordons déposés par le procédé (acier 316L et Inconel 718), pour lesquels des porosités et fissures peuvent apparaître (~ 100 µm). L’inspection est effectuée par méthode ultrasons-laser (UL), c’est-à-dire tout optique et non intrusive (régime thermoélastique). Le procédé DML engendre une microstructure dite à gros grains et des rugosités de surface qui rendent le contrôle ultrasonore plus délicat. En effet, ces caractéristiques microstructurales provoquent la diffusion des ondes élastiques dans le milieu. Ce travail de thèse participe donc à la compréhension de ces phénomènes et de leurs influences pour mieux détecter des défauts subsurfaciques dont les dimensions sont proches des longueurs d’ondes acoustiques. Ainsi, l’inspection en surface a été optimisée par la conception d’un montage optique, permettant de favoriser la génération de l’onde de Rayleigh, par une ligne source laser fine (~200 µm) et présentant des fronts raides. En conséquence, le contenu spectral de l’onde a été augmenté vers les hautes fréquences, c’est-à-dire jusqu’à 10 MHz. La longueur d’onde de Rayleigh (λR) a pu être ainsi diminuée à une valeur proche de celle des défauts recherchés, de l’ordre de 700 µm. Ensuite, les limites de ce dispositif ont été mises en évidence sur des pièces fabriquées par le procédé DML. Ces pièces comportent des défauts usinés, des entailles et des trous génératrice dont les dimensions sont inférieures à λR, leur surfaces est rugueuse (diffraction multiple de l’onde de Rayleigh). Puis, l’inspection de défauts réels subsurfaciques est étudiée. Ces défauts sont créés soit par une variation d’un des paramètres du procédé, le hatch (distance inter-cordon), ou encore par l’utilisation d’une poudre métallique de mauvaise qualité. En dégradant volontairement ces paramètres, les échantillons présentent alors des taux de porosités (Φ) pouvant s’étendre de 0,5% à 10%. Enfin, deux méthodes de corrélations ont été exploitées pour discriminer les taux de porosités : à travers l’étude du degré de ressemblance et l’Analyse en Composantes Principales (ACP). La première méthode, simple et rapide à mettre en œuvre, ne permet cependant pas de distinguer les taux de porosités. En revanche, l’ACP indique qu’il est possible de discriminer tous les défauts. Pour finir, des tests préliminaires ont été effectués afin de montrer qu’il est possible d’effectuer des mesures par méthode UL sur surface rugueuse, tout en conservant un bon rapport signal sur bruit, et ce sans moyenner les signaux, au cours d’un déplacement robotisé. / This thesis contributes to the detection of flaws in laser metal deposition (LMD) additive manufacturing process samples towards an in situ control of the process. The in situ control foreseen concerns the last layers deposited by the process (316L and Inconel 718 steel) for which porosities and cracks may appear (~ 100 µm). The inspection is performed by laser-ultrasonics (LU), an all-optical and non-invasive technic. Experiments are conducted in thermoelastic mode. The LMD process gives rise to coarse grain microstructure and surface roughness, that make the control tricky. Indeed, those characteristics generate scattering elastic waves. Thus, this work contributes to the better comprehension of those phenomena and their effects of the detection on subsurface flaws which dimensions are close to the acoustic wavelength. Surface inspection has been optimized by designing an optical setup, improving the Rayleigh wave generation by using a thin laser line source (~200 µm). Hence, the spectral content has been increased until it reaches 10 MHz, that is to say a wavelength equals to 700 µm (λR). Then, the limits have been brought out through the inspection of LMD samples. Those samples include sided drilled holes and notches which dimensions are less than λR, on rough surface (multiple scattering of Rayleigh wave). Moreover, real subsurface flaws inspections have been studied. The flaws have been created by the instability of one of the process parameter: the hatch (distance between each layer), or by a bad powder quality. By intentionally deteriorating those parameters, the samples, then, have porosity rate (Φ) with ranging from 0.5% to 10%. Finally, two correlation methods have been studied to discriminate porosity rate: through the study of the normalized cross correlation function and by Principal Component Analysis (PCA). The first method, simple and fast to implement, does not allow distinguishing the porosity rates. Nevertheless, the PCA indicates that it is possible to discriminate all the flaws. Finally, preliminary tests have been carried out to show that LU measurements is feasible on rough surfaces, while keeping a good signal-to-noise ratio, without averaging, during an automated displacement.

Depuis plus de 60 ans, les résines composites à usage dentaire n’ont cessé d’évoluer vers des matériaux toujours plus fonctionnels et esthétiques. Toutefois, l’obtention du matériau de restauration idéal présentant des propriétés mécaniques et physico-chimiques optimales reste encore aujourd’hui un défi de taille. La nature et la proportion des monomères, le type, la taille et le taux de charges, la qualité des liens entre charges et matrice et la présence de porosités au sein des matériaux apparaissent comme des éléments pouvant influencer ces propriétés. Afin d’avancer dans le compréhension de l’impact de ces différents facteurs, plusieurs études ont été entreprises à l’aide de trois résines composites commerciales de viscosités différentes : une résine de viscosité élevée (Filtek P60 – 3M ESPE), une résine de viscosité moyenne (Grandio – Voco) et une résine de viscosité faible (Filtek Supreme XTE – 3M ESPE). L’influence de la manipulation clinique des matériaux sur le taux et le volume des porosités a été évaluée à l’aide de la tomographie 3D à rayons X. La manipulation augmente le pourcentage et diminue le volume moyen des porosités. En outre, plus le matériau est fluide et plus il présente de porosités en son sein. Deux groupes d’échantillons ont été réalisés afin d’évaluer l’influence de la porosité sur le comportement mécanique et physico-chimique des matériaux : l’un à partir des matériaux simplement extrudés des seringues et l’autre au sein duquel nous avons réalisé une adjonction artificielle de porosités. La contrainte de polymérisation à été déterminée à l’aide d’une machine de traction, l’absorption/solubilité a été mesurée en respectant le cadre ISO 4049, le comportement mécanique vrai a été évalué en traction à l’aide du système VidéoTractionTM, les comportements apparents en compression et flexion ont été étudiés respectivement à l’aide d’une machine de compression et d’une machine de flexion 3 points. Les différents constituants organiques et minéraux apparaissent comme des éléments déterminants dans le comportement physico-chimique et mécanique des matériaux. La porosité initiale est également un facteur prépondérant dans l’explication dudit comportement. Son influence n’est cependant pas proportionnelle à son taux puisque l’ajout de porosités artificielles n’influence pas significativement les résultats / For the past 60 years, dental resin composites have been constantly evolving, becoming increasingly functional and aesthetic materials. However, obtaining the ideal dental restorative material, with optimum mechanical and physicochemical properties, is still a significant challenge today. The nature and proportion of the monomers, the type, size and content of fillers, the quality of the bonds between the fillers and the matrix and the presence of porosities inside materials all emerge as aspects that can influence these properties. To improve our understanding of the impact of these various factors, a number of studies have been undertaken using three commercially-available resin composites with different viscosities: one high-viscosity resin (Filtek P60 – 3M ESPE), one moderate-viscosity resin (Grandio – Voco) and one low-viscosity resin (Filtek Supreme XTE – 3M ESPE). The influence of clinical handling of materials on the rate and volume of porosities has been evaluated by 3D X-ray computed tomography. Handling increases the porosity percentage and reduces the average porosity volume. Furthermore, the more flowable the material is, the more porosities it has within it. Two groups of samples were prepared in order to evaluate the influence of porosity on the mechanical and physicochemical behavior of materials: one group of samples consisted of materials simply extruded from syringes, while additional porosities were added artificially in the other samples. The polymerization stress was determined using a tensile testing machine, the absorption/solubility was measured in accordance with ISO 4049, the true tensile mechanical behavior was assessed using the VidéoTractionTM system, and the apparent compressive and flexural behaviors were studied using a compression test machine and a 3-point flexural test machine, respectively. The various organic and mineral components appear to be key elements in the physicochemical and mechanical behavior of materials. The initial porosity is also a predominant factor in terms of explaining this behavior. However, its influence is not proportional to its rate, since the addition of artificial porosities does not significantly affect the results

AA6061 Al alloys modified with addition of Mn, Cr and Cu were homogenized at temperatures between 350 ºC and 550 ºC after casting. STEM experiments revealed that the formation of α-Al(MnFeCr)Si dispersoids during homogenization were strongly affected by various factors such as heating rate, concentration of Mn, low temperature pre-nucleation treatment and homogenization temperature. Through analysis of the STEM results using an image software Image-Pro, the size distributions and number densities of the dispersoids formed during different annealing treatments were quantitatively measured. It was revealed that increasing the heating rate or homogenization temperature led to a reduction of the number density and an increase in size of the dispersoids. The number density of dispersoids could be markedly increased through a low temperature pre-nucleation treatment. A higher Mn level resulted in the larger number density, equivalent size and length/width ratio of the dispersoids in the alloy. Upsetting tests on two of these Mn and Cr-containing AA6061 (Al-Mg-Si-Cu) Al alloys with distinctive Mn contents were carried out at a speed of 15 mm s-1 under upsetting temperature of 450 ºC after casting and subsequent homogenization heat treatment using a 300-Tone hydraulic press. STEM experiments revealed that the finely distributed α-Al(MnFeCr)Si dispersoids formed during homogenization showed a strong pinning effect on dislocations and grain boundaries, which could effectively inhibit recovery and recrystallization during hot deformation in the two alloys. The fractions of recrystallization after hot deformation and following solution heat treatment were measured in the two alloys with EBSD. It was found that the recrystallization fractions of the two alloys were less than 30%. This implied that the finely distributed α-dispersoids were rather stable against coarsening and they stabilized the microstructure by inhibiting recovery and recrystallization by pinning dislocations during deformation and annealing at elevated temperatures. By increasing the content of Mn, the effect of retardation on recrystallization were further enhanced due to the formation of higher number density of the dispersoids. STEM and 3-D atom probe tomography experiments revealed that α-Al(MnFeCr)Si dispersoids were formed upon dissolution of lathe-shaped Q-AlMgSiCu phase during homogenization of the modified AA6061 Al alloy. It was, for the first time, observed that Mn segregated at the Q-phase/matrix interfaces in Mn-rich regions in the early stage of homogenization, triggering the transformation of Q-phase into strings of Mn-rich dispersoids afterwards. Meanwhile, in Mn-depleted regions the Q-phase remained unchanged without segregation of Mn at the Q-phase/matrix interfaces. Upon completion of α-phase transformation, the atomic ratio of Mn and Si was found to be 1:1 in the α-phase. The strengthening mechanisms in the alloy were also quantitatively interpreted, based on the measurements of chemical compositions, dispersoids density and size, alloy hardness and resistivity as a function of the annealing temperature. This study clarified the previous confusion about the formation mechanism of α-dispersoids in 6xxx series Al alloys. Four-point bend fatigue tests on two modified AA6061 Al alloys with different Si contents (0.80 and 1.24 wt%, respectively) were carried out at room temperature, f = 20 Hz, R = 0.1, and in ambient air. The stress-number of cycles to failure (S-N) curves of the two alloys were characterized. The alloys were solution heat treated, quenched in water, and peak aged. Optical microscopy and scanning electron microscopy were employed to capture a detailed view of the fatigue crack initiation behaviors of the alloys. Fatigue limits of the two alloys with the Si contents of 0.80 and 1.24 wt% were measured to be approximately 224 and 283.5 MPa, respectively. The number of cracks found on surface was very small (1~3) and barely increased with the applied stress, when the applied stress was below the yield strength. However, it was increased sharply with increase of the applied stress to approximately the ultimate tensile strength. Fatigue crack initiation was predominantly associated with the micro-pores in the alloys. SEM examination of the fracture surfaces of the fatigued samples showed that the crack initiation pores were always aspheric in shape with the larger dimension in depth from the sample surface. These tunnel-shaped pores might be formed along grain boundaries during solidification or due to overheating of the Si-containing particles during homogenization. A quantitative model, which took into account the 3-D effects of pores on the local stress/strain fields in surface, was applied to quantification of the fatigue crack population in a modified AA6061 Al alloy under cyclic loading. The pores used in the model were spherical in shape, for simplicity, with the same size of 7 μm in diameter. The total volume fraction of the pores in the model were same as the area fraction of the pores measured experimentally in the alloy. The stress and strain fields around each pore near the randomly selected surface in a reconstructed digital pore structure of the alloy were quantified as a function of pore position in depth from the surface using a 3-D finite element model under different stress levels. A micro-scale Manson-Coffin equation was used to estimate the fatigue crack incubation life at each of the pores in the surface and subsurface. The population of fatigue cracks initiated at an applied cyclic loading could be subsequently quantified. The simulated results were consistent with those experimentally measured, when the applied maximum cyclic stress was below the yield strength, but the model could not capture the sudden increase in crack population at UTS, as observed in the alloy. This discrepancy in crack population was likely to be due to the use of the spherical pores in the model, as these simplified pores could not show the effects of pore shape and their orientations on crack initiation at the pores near surface. Although it is presently very time-consuming to calculate the crack population as a function of pore size and shape in the alloy with the current model, it would still be desirable to incorporate the effects of shape and orientation of the tunnel-shaped pores into the model, in the future, in order to simulate the fatigue crack initiation more accurately in the alloy.

Ce travail concerne l'utilisation de la technique de Fracturation Hydraulique (FH) pour exploiter l'énergie géothermique des réservoirs profonds de roches sèches chaudes (HDR). La fracturation hydraulique est réalisée par injection de fluides géothermiques dans des réservoirs partiellement fracturés de faible perméabilité. Les fluides à haute pression sont destinés à faire évoluer les fissures et leur connectivité. Les valeurs de débit/pression auxquelles les fluides géothermiques doivent être pompés, ainsi que le calendrier de pompage pour initier la fracturation hydraulique, dépendent principalement des conditions géostatiques existantes (contraintes géostatiques, pression fluide et température initiales de l'HDR) ainsi que des propriétés des fissures de l'HDR (longueur, épaisseur, densité et distribution directionnelle initiales moyennes de fissures). Tous ces éléments, en sus de leurs effets sur la stabilité des forages, sont analysés dans cette recherche. Des modèles de fracturation, qui sont capables de suivre l'évolution des fissures dans toutes les orientations spatiales possibles, sont utilisés pour obtenir le tenseur anisotrope de perméabilité. Ces modèles sont intégrés dans un code domestique d'éléments finis qui est développé pour résoudre des problèmes aux limites thermo-poroélastiques. Pour supprimer/diminuer les oscillations qui accompagnent les solutions paraboliques et/ou hyperboliques lors de la convection forcée, plusieurs techniques de stabilisation ont dû être implémentées. / The application of the Hydraulic Fracturing (HF) technology to exploit geothermal energy from Hot Dry Rocks (HDR) reservoirs is addressed. HF is achieved by extensively pumping geothermal fluids to already existing fractured HDR reservoirs of low permeability. High fluid pressures are expected to drive cracks to evolve and connect. The newly created burgeoning hydraulic conduits should supposedly enhance the permeability of the existing HDR reservoirs. The flow rate/pressure values at which geothermal fluids should be pumped, as well as the pumping schedule to initiate HF, depend primarily on the existing geostatic conditions (geostatic stresses, initial HDR pressure and temperature) as well as on HDR fracture properties (initial mean fracture length, mean fracture aperture, density and orientational distribution of fractures). While these components, in addition to their effects on borehole stability, are scrutinized in this research, focus is on the evolution during circulation processes of the fracture properties. A fracturing model that is capable of tracking fracture evolution in all possible spatial orientations is used to obtain the time course of the anisotropic permeability tensor. This evolving property is integrated into a domestic finite element code which is developed to solve thermo–poroelastic BVPs: emphasis is laid on the efficiency of the doublet flow technique where a fluid gains thermal energy by circulating through the HDR reservoir from the injection well to the production well. The spurious oscillations in the hyperbolic solutions of the approximated finite element approach that are commensal with the phenomenon of forced heat convection are healed/mitigated through several stabilization approaches.