Ces travaux portent sur la déformation des aciers au cours de leur solidification, au moyen d'une expérience instrumentée et de sa simulation numérique. L'étude est focalisée sur deux phénomènes induits par la déformation : la fissuration à chaud et la macroségrégation. L'expérience consiste à poinçonner latéralement un lingot de 450 kg, alors que son cœur est encore partiellement liquide.L'expérience est instrumentée thermiquement et mécaniquement. Les lingots sont analysés, visuellement en termes de lieu et de fréquence d'apparition de fissures, et par microsonde pour les ségrégations chimiques.Pour la fissuration à chaud, une simulation numérique 3D par éléments finis est mise en œuvre avec le logiciel Thercast®, dans lequel a été implanté un critère d'amorçage de fissure basé sur la déformation plastique cumulée en fin de solidification, entre deux valeurs critiques de fraction de liquide. La comparaison entre simulations et observations montre le caractère prédictif du critère.La simulation numérique de la macroségré-gation est réalisée avec le logiciel R2sol qui résout simultanément la déformation du solide et l'écoulement du liquide. La simulation montre la redistribution des solutés dans le cœur du lingot sous l'effet de la compression du squelette solide et de l'écoulement du liquide, induits par le poinçonnement. Elle reproduit qualitativement les mesures expéri-mentales mais sous-estime l'amplitude des hétérogénéités de composition chimique. Une discussion des résultats permet de dégager des pistes permettant d'espérer une prédiction quantitative dans le futur.Les deux thématiques étudiées ont mis en relief la nécessité d'une bonne modélisation des phénomènes de microségrégation des alliages multiconstitués. Un modèle a été spécifiquement développé à cet effet. / Experimental and numerical studies of hot tearing and macrosegregation formation during steel solidification are reported. On one hand, an ingot punching test is considered. It consists of the application of a deformation at the surface of a solidifying 450 kg steel ingot. On the other hand, finite element thermo-mechanical modelling of the test is used.For hot tearing analysis, 3D finite element modeling is applied by use of Thercast® software. The time evolution of the strain tensor serves to evaluate the possibility for hot tear formation with a Hot Tearing Criterion (HTC). The HTC compares the local accumulation of strain over a certain solidification interval with the expression of a critical value proposed in the literature. Detailed comparisons reveal an excellent capability of the HTC to predict the formation of hot tears.For macrosegregation analysis, a two-phase formulation has been implemented (R2sol software), in which the velocities of the liquid and solid phases are concurrently solved for. The simulation shows how solutes are redistributed through the central mushy zone of the ingot under the effect of the com-pression of the solid phase resulting from the punching of the solid shell. The simulation proves its capability to reproduce the main experimental trends. However the predicted intensity of macrosegregation is lower than measured. Through discussion and analysis of different numerical sensitivity tests, critical material parameters and model improvements are identified in view of achieving better quantitative predictions in the future.The two topics studied have clearly shown the need for a good modelling of microsegregation phenomena in multicomponent alloys. A numerical model has especially been developed and implemented in the two software packages.

Ce travail de thèse concerne l'étude du filage et des propriétés de fibres composites à base denanotubes de carbone. Les propriétés mécaniques des fibres en cours de solidification et enmouvement dans un bain de coagulation sont évaluées afin d'étudier l'influence de différentsparamètres physico-chimiques impliqués dans leur fabrication. Ces fibres, combinant despropriétés mécaniques et électriques prometteuses, peuvent être obtenues de façon continuegrâce au développement d'un nouveau procédé de filage. Elles présentent de plus un effetoriginal de mémoire de température dont l'origine est étudiée dans cette thèse. Ces fibres sontpotentiellement utiles pour diverses applications: des matériaux à haute absorption d'énergiemécanique à des textiles conducteurs fonctionnels. / This thesis deals with the study the wet-spinning process for the production of carbonnanotube composite fibers and their properties. We have characterized the mechanicalproperties of the fibers during their solidification as they circulate along the pipe of thespinning line. The study of the influence of various chemical parameters allowed us todevelop a new process for the continuous and scalable production of these fibers, whichexhibit unique mechanical and electrical properties. Moreover, they show an original effect oftemperature memory. The origin of this phenomenon is investigated in this work. These fiberscould be used for various applications such as high energy absorption materials or functionalconductive textile.

Les coulées industrielles permettent de distinguer deux types de structures : colonnaires et équiaxes. La mise en place de ces structures a des conséquences importantes sur les autres hétérogénéités, particulièrement les macroségrégations chimiques. Le code SOLID, développé à l’Institut Jean Lamour, permet de décrire de manière couplée la convection naturelle du liquide ainsi que la germination, la croissance et le transport des grains équiaxes. Le travail présenté a pour but de proposer une modélisation de l’apparition et de la croissance des structures colonnaires couplées à celles des grains équiaxes, permettant ainsi de prédire la Transition Colonnaire-Equiaxe (TCE) et Equiaxe-Colonnaire (ECT). La particularité du modèle est de considérer la croissance couplée des structures uniquement au niveau des pointes primaires colonnaires car c’est à cet endroit que les gradients de soluté sont les plus importants. Après validation, le modèle est appliqué à des cas de coulées industrielles de lingots d’acier et comparé à des mesures expérimentales. Il en ressort en premier lieu que sans la modélisation du mouvement des grains équiaxes, les morphologies et les ségrégations de carbone prédites ne correspondent pas à l’expérience. Par la suite, on montre que les résultats obtenus dépendent fortement du scénario d’apparition des grains équiaxes. Une germination hétérogène volumique des grains équiaxes ne permet pas de prédire la TCE expérimentale. En revanche, la fragmentation des grains, associée à un critère pour le début de la fragmentation, prédit une TCE et des ségrégations en carbone en accord avec l’expérience. On montre alors que la masselotte des lingots peut ainsi être une source importante de grains / It is possible to distinguish two main types of structures in castings: columnar and equiaxed. The dynamic set up of these structures has a strong impact on other heterogeneities, especially the chemical macrosegregations. Developed at the Institut Jean Lamour, SOLID is a numerical code that accounts for natural convection as well as the germination, growth and transport of equiaxed grains. The purpose of this work is to model the appearance and the growth of the columnar structures coupled with the description of the equiaxed grains. The model can therefore predicts the Columnar-to-Equiaxed (CET) and Equiaxed-to Columnar (ECT) Transitions. The main characteristic of the model is to consider the coupled growth of both structures only in the zone near the tips of the primary columnar dendrites. It is indeed there that the strongest solute gradients are located. The model is verified by comparing it to experiments and other models of the literature. The model is then applied to the case of industrial steel ingots and compared to experimental measurements. The first result is that without taking into account the movement of the equiaxed grain the results for equiaxed grain morphology and for macrosegregation do not agree with the measurements. Next, we find that the phenomenon considered for equiaxed grain formation is decisive for the CET prediction. When heterogeneous volumic nucleation is considered, we were not able to predict the CET correctly. However, when fragmentation at the columnar front is considered – along with a criterion for the onset of fragmentation – the results agree quite well with the experiments. It is also shown that the hot-top of ingots is consequently an important source of equiaxed grains

Manufacturing of light-weight materials is associated with several types of casting defects during solidification. Porosity defects are common, especially in aluminum and its alloys, which initiate crack propagation and thereby cause drastic deterioration in the mechanical properties. These defects, classified as micro and macro defects (based on their sizes), are mainly governed by release of hydrogen into the liquid at the solid-liquid interface, which triggers the nucleation and growth of hydrogen bubbles in the melt. Subsequently, these bubbles interact with solidifying interfaces such as dendritic arms and eutectic fronts, leading to the formation of pores. Macroscopic defects in the form of voids are created due to solidification shrinkage. The primary focus of the present work is to develop phenomenological models for the evolution of microporosity and microstructures during solidification. The issues outlined above typically occur in multi-phase environments comprising of solid, liquid and gaseous phases, and over a range of length and time scales. Any phenomenological prediction would, therefore, require a multi-phase-scale approach. Principles of volume averaging are applied to equations of conservation to obtain single-field formulations. These are then solved with appropriate interface tracking techniques such as Enthalpy, Level-set, Volume-of-fluid and Immersed-boundary methods. The framework is built up on a standard pressure based incompressible fluid flow solver (SIMPLER algorithm) and coupled modeling strategies are proposed to address the interfacial dynamics. A two-dimensional framework is considered with a fixed-grid Cartesian co-ordinate system. Scaling analyses are performed to bring out the relative effects of various competing parameters in order to obtain further insights into this complex phenomenon. The numerical results and scaling predictions are validated against experimental observations published in literature. In literature, numerical predictions of microporosity mainly include criteria based models based on empirical relations and deterministic/stochastic models based on diffusion driven growth assuming spherical bubbles. The dynamic evolution of non-spherical bubble-metal interface in a three-phase system is yet to be captured. Moreover, several in-situ experiments have shown elongated bubble shapes during the engulfment phase, therefore a criterion to define the dependence on cooling rates and the resulting bubble morphology can possibly deliver further practical insights. We propose a numerical model for hydrogen bubble growth, its movement and subsequent engulfment by a solidifying front, combining the features of level-set and enthalpy methods for tracking bubble-metal and solid-liquid interfaces, respectively. The influx of hydrogen into heterogeneously nucleated bubbles results in growth of bubbles to sizes up to a few hundreds of microns. In the first part of this numerical study, a methodology based on the level-set approach is developed to simultaneously capture hydrogen bubble growth and movement in liquid aluminum. The solidification is first assumed to occur outside the micro-domain providing a specified hydrogen influx to the bubble-in-liquid system. The level-set equation is formulated in such a way as to account for simultaneous growth and movement of the bubble. The growth of a bubble with continuous and fixed hydrogen levels in the melt is studied. The rates of growth of bubble-liquid and solidifying interfaces are compared using an order of magnitude analysis. This scaling analysis explains the thought experiment proposed in the literature, where difference in bubble shapes was attributed to the cooling rate. Moreover, it shows explicit dependence on bubble radius and cooling rate leading to a new criterion for bubble elongation proposed in this thesis. This also highlights the comparison between solidification and hydrogen diffusion time-scales which primarily govern the competitive growth behavior. The bubble-in-liquid model is coupled with microscopic enthalpy method to incorporate effects of solidification and study the interaction of solid-liquid and bubble-liquid interfaces. The phenomena of bubble engulfment and elongation are successfully captured by the proposed model. A parametric study is carried out to estimate the bubble elongation based on different initial bubble sizes and varying cooling rates encountered in typical sand, permanent mold and die casting processes. Although simulation of microstructures has been extensively studied in the literature, very few models address the phenomena of simultaneous growth and movement of equiaxed dendrites. The presence of different flow environments and multiple dendrites are known to alter the position and shape of the dendrites. The proposed model combines the features of the following methods, namely, the Enthalpy method for modeling growth; the Immersed Boundary Method (IBM) for handling the rigid solid-liquid interfaces; and the Volume of Fluid (VOF) method for tracking the advection of the dendrite. The algorithm also performs explicit-implicit coupling between the techniques used. Validation with available literature is performed and dendrite growth in presence of rotational and buoyancy driven flow fields is studied. The expected transformation into globular microstructure in presence of stirring induced flows is successfully simulated. A simple order estimate for time required for stirring is performed which agrees with numerical predictions. In buoyancy driven environment of a settling dendrite, the arm tip speeds show expected higher velocity of the upstream tip compared to its counterpart. The model is extended to study thermal and hydrodynamic interactions between multiple dendrites with appropriate considerations for different orientations and velocities of the dendritic solid entities. The present model can be used for the prediction of grain sizes and shapes and to simulate morphological transformations due to different melt flow scenarios. In the final part, the methodology presented for growth and engulfment of hydrogen bubbles is extended to study the phenomenon of diffusion driven bubble growth occurring in direct foaming of metals. The source of hydrogen is determined by the rate of decomposition of the blowing agent. This is accounted for by a source term in the hydrogen species conservation equation, and growth rate of hydrogen bubbles is calculated on the basis of diffusive flux at the interface. The level-set method is used for tracking the bubble-liquid interface growth, and the macroscopic enthalpy model is used for obtaining heat transfer and solid front position. The model is validated with analytical solution by comparing the front position and the solidification time. The variation of foam density with a transient hydrogen generation source is studied and qualitatively compared with results reported in literature. The modeling strategies proposed in this work are generic and therefore have potential in simulating a variety of complex multi-phase problems.

In this thesis, we present an experimental and numerical study of globularization during reheating of thixocast billet having non-dendritic microstructure. The process of reheating is an important step in the semisolid processing and is essential to control its microstructure and hence its mechanical properties. Material chosen for this study is Aluminum alloy, A356. The primary focus of this study is the heat treatment below eutectic temperature i.e. transformation in solid phase. It is found that during short duration heat treatment, globularization of primary α grains and spheroidization of eutectic Si flakes take place which improves the mechanical properties of semisolid cast products significantly. A prolonged heat treatment is found to degrade the properties of castings since it enhances the porosity and coarsening of Si. The study suggests that a precise heat treatment practice can be designed to improve the semisolid microstructure. A computational model based on Phase field approach has been proposed to study this phenomena. Predictions based on this model are qualitatively compared with corresponding experimental observations. Since eutectics form an important step in multiphase solidification, an attempt has been made to develop an enthalpy based explicit micro-scale model for eutectic solidification. In this preliminary study, growth of adjacent α and β phases in a two dimensional Eulerian framework has been simulated. The model is qualitatively validated with Jackson Hunt theory. Results show expected eutectic growth. This methodology promises significant saving in computational time compared to existing numerical models.

La première partie de ce manuscrit est consacrée à l’étude des microstructures et des microségrégations, héritées de la solidification des alliages ternaires Fe-Al-Ti, Fe-Al-Zr et quaternaires Fe-Al-Ti-Zr. Pour améliorer la compréhension des phénomènes ayant lieu au cours de la solidification et disposer d’un outil permettant de prédire les phases formées au cours de la solidification, une description thermodynamique du coin riche en fer du système quaternaire Fe-Al-Ti-Zr est proposée dans une seconde partie. A partir de l’ensemble des données expérimentales et bibliographiques disponibles concernant les équilibres de phases dans les différents systèmes ternaires, une description thermodynamique de chacun des systèmes ternaires est effectuée. Les systèmes Fe-Al-Zr et Fe-Ti-Zr ont été complètement décrits à l’aide du logiciel ThermoCalc. Le système ternaire Fe-Al-Ti a fait l’objet d’une description partielle. Les résultats obtenus permettent de proposer une première description thermodynamique du coin riche en fer du système quaternaire Fe-Al-Ti-Zr. Des séquences de solidification ont également été calculées à partir de cette description pour les alliages ternaires Fe-Al-Zr et quaternaires Fe-Al-Ti-Zr et comparées aux résultats expérimentaux / The first part of this work deals with the study of the microstructures and micro-segregations phenomena, inherited from the solidification on Fe-Al-Ti, Fe-Al-Zr and Fe-Al-Ti-Zr alloys. To improve our understanding of the phenomena occurring during the solidification process and to dispose of a predictive tool of the phases formed during the solidification process, a thermodynamic modeling of the iron rich corner of the Fe-Al-Ti-Zr quaternary system is proposed on a second part. From the data available in this work and in the literature, a thermodynamic modeling of each of the constituting ternaries systems is performed. Fe-Al-Zr and Fe-Ti-Zr ternaries systems were fully modeled using the ThermoCalc software. The Fe-Al-Ti ternary system is only partially modeled. These results allowed us to propose a first description of the iron rich corner of the Fe-Al-Ti-Zr quaternary system. Solidification sequences were calculated from this thermodynamic model for Fe-Al-Zr ternaries and Fe-Al-Ti-Zr quaternaries alloys and compared to our experimental results

Les ciments phosphomagnésiens sont des ciments inorganiques de la famille des ciments activés par des acides. Malgré le fait qu'ils soient connus depuis le début du XXème siècle, leur utilisation reste assez limitée dans le génie civil. Ces matériaux sont utilisés dans le cadre de réparation notamment pour des réparations routières qui nécessitent un développement rapide des résistances en compression. Plus récemment, ces ciments ont été utilisés pour des applications de stabilisation/solidification (S/S) des déchets et notamment de certains types de déchets chimiquement incompatibles avec le ciment Portland. Toutefois, les défis de mise en œuvre de ce type de ciment sont liés à la nature même de la réaction, très exothermique et très rapide. La phase liante de ces ciments, la k-struvite (MgKPO4.6H2O), est obtenue par un mélange de magnésie (MgO) et de dihydrophosphate de potassium (KH2PO4) en présence d'eau. MgO + KH2PO4 + 5H2O  MgKPO4.6H2OLes mécanismes qui régissent la prise de ce ciment sont encore mal connus et plusieurs théories, mettant en jeu ou non la formation de produits secondaires, ont été proposées. Nos travaux de recherche ont porté sur la compréhension des mécanismes de prise des ciments phosphomagnésiens ainsi que sur l'influence des paramètres de formulation sur les propriétés de ces ciments. L'étude a montré que la formation de la k-struvite passe par une réaction de précipitation-dissolution d'un produit intermédiaire, la newberyite (MgHPO4.3H2O). Les réactions de formation de ces deux produits sont contrôlées par le taux de sursaturation des espèces en solution et le pH du milieu réactionnel. L'étude met également en lumière la forte sensibilité de ces ciments à la quantité d'eau introduite. Avec un rapport E/C très faible comparativement à un ciment Portland (rapport E/C compris entre 0,1 et 0,25), une faible variation (0,02) entraine la ségrégation de la pâte de ciment et une hétérogénéité à la surface du matériau. Parmi les paramètres de formulation, le rapport molaire MgO/KH2PO4 (Mg/P) est le plus important. En effet, il influe sur la quasi-totalité des propriétés du ciment : résistance en compression, temps de prise, chaleur de réaction, fluidité de la pâte, …. L'utilisation de faibles rapports Mg/P entraine une mauvaise tenue à l'eau du ciment, la formation de cristaux à l'intérieur de la microstructure et visibles à la surface de l'échantillon (apparition d'une efflorescence) ainsi qu'un gonflement important de la pâte de ciment (pouvant aller jusqu'à la fissuration des échantillons). A l'issue de l'étude paramétrique une formulation d'une pâte a pu être définie. Des mesures de variations dimensionnelles, ainsi que des mesures de retrait chimique ont été effectuées, afin de mieux comprendre les mécanismes mis en jeu pour induire les phénomènes de gonflement. En support, des analyses de la microstructure (MEB, DRX, ATG) ainsi que des essais de lixiviation viennent compléter la campagne expérimentale. Ces résultats viennent confirmer l'influence d'un rapport Mg/P faible sur le gonflement et la tenue à l'eau du ciment phosphomagnésien. Pour finir, une étude sur l'influence des divers ajouts, dans le but d'améliorer ses performances, a montré que les fillers inertes (sable siliceux ou cendres volantes) retardent le temps de prise et réduisent la chaleur d'hydratation / Magnesium phosphate cements are the most representative cements of the activated-by-acid cements family. Despite the fact that they are known since the early 20th century, their use in civil engineering is fairly limited. These materials are used for road repairs where the fast compressive strength development is an advantage. Recently they have also been used in wastes stabilization/solidification (S/S), especially with wastes incompatible with Portland cement. The challenges of the use of these cements are related to the nature of their formation reaction: fast, very exothermic, with a very short setting time (only a few minutes).The bonding phase, k-struvite (MgKPO4.6H2O), is obtained from magnesium oxide mixed in water with monopotassium phosphate (KH2PO4).MgO + KH2PO4 + 5H2O  MgKPO4.6H2OThe setting mechanisms are still poorly known and various theories, involving or not secondary product formation, have been suggested. Our researches have aimed to understand the setting mechanisms, as well as the influence of the formulation parameters on the properties of the magnesium phosphate cement. Results show that the formation of k-struvite involved a precipitation-dissolution reaction of an intermediate product, the newberyite (MgHPO4.3H2O). Formation reactions of these two products are controlled by the supersaturation rate and the pH of the solution. The study highlights the strong effect of water on the properties of fresh cement paste. With a low mass ratio e/c in comparison of Portland cement (ratio e/c between 0,1 and 0,25), a slight modification of the ratio (0,02) leads to a segregation and a surface heterogeneity of the cement paste. Among the formulation parameters, the molar ratio MgO/KH2PO4 (Mg/P) seems the most important parameter. Indeed, it impacts most of the properties of the magnesium phosphate cement (compressive strength, setting time, reaction heat, paste fluidity …). Low Mg/P ratios lead to poor water resistance, to crystals formation inside the microstructure that can be seen on the surface of the sample (an efflorescence appearance), and to important swelling of the paste, leading to the cracking of the samples. After the parametric study, a magnesium phosphate cement paste has been defined. Dimensional changes and chemical shrinkage measurement were conducted to understand the mechanisms involved in this swelling phenomenon. In support, microstructural analyses (SEM, XRD, TGA) and leaching tests complete the experimental campaign. The results confirm the influence of a low Mg/P ratio on cement swelling and water resistance. Finally, a study on various additions to the paste has been conducted, with the purpose of improving the cement paste performances. It shows that the addition of an inert filler (siliceous sand or fly ashes) has a retarding effect and reduced the reaction heat

Nous avons étudié les phénomènes convectifs et leur interaction dynamique avec la formation des microstructures pendant la solidification dirigée d’alliages étalliquesbinaires.La méthode post-mortem a été utilisée d’abord pour étudier la Transition olonnaire-Equiaxe pendant la solidification dirigée d’échantillons cylindriques d’Al-3,5wt%Ni non affiné sous la Technique de Rotation Accélérée de Creuset. La simulation numérique a été éffectuée et acquérie les résultats en concordance avec les manipulations.La technique in-situ a été appliquée pour comprendre l’évolution en fonction de temps des grains pendant solidification d’Al-4wt%Cu. La caractéstiques tatistiques des grains ont été discutées.La convection d’instabilité déclenchée par la poussée ou la tension superfaciale sous les gradients thermiques verticale et horizontale dans un système de double couches liquide-zone poreuse ont réspectivement étudié par analysis d’instabilité linéaire.L’inhomogénéité de la perméabilité de zone pateuse dendritique a été tenue en compte afin de comprendre son influence sur le début de convection pendant la solidification dirigée d’Al-3,5wt%Li. / We studied the convective phenomena and their dynamical interaction with the formation of the microstructurs during directional solidification of binary metallic alloys.The post-mortem method was used first to study the Columnar-Equiaxed-Transition during the directional solidification of unrefined Al-3.5wt%Ni in cylindric samples under the Accelerated Crucible Rotation Technique. The numerical imulation was carried out and achieved the results in agreement with experiments.The in-situ technique was applied to understand the evolution of equiaxed grains during solidification of Al-4wt%Cu in function of time. The statistical characteristics of equiaxed grains were discussed.The buoyancy-driven and surface-tension-driven instability convection under vertical and horizontal thermal gradients in a liquid-porous double-layered system were respectively investigated through linear instability analysis.The inhomogeneity of the dendritic mush permeability was taken into account in order to understand its influence on the triggering of convection during the directional solidification of Al-3.5wt%Li.

La modélisation multi-échelle des procédés de solidification présente un grand intérêt pour les industries. Toutefois, il est difficile de coupler les phénomènes prenant place à de multiples échelles pour obtenir des simulations quantitatives à grande échelle. Ceci est réalisé en combinant trois méthodes : les éléments finis (FE), un automate cellulaire (CA) et la méthode Parabolic Thick Needle(PTN). La méthode FE permet une résolution des équations de conservation écrites pour des quantités moyennées, ce qui est adapté aux calculs de grands domaines. Elle permet la description macroscopique des transferts de chaleur et de masse. De plus, la méthode CA permet de suivre le développement de l’enveloppe de chaque grain dendritique à une échelle mésoscopique. Le couplage de ces deux méthodes est le modèle CAFE et il a démontré son efficacité pour simuler quantitativement la solidification et notamment la transition colonnaire - équiaxe. Le Dendritic Needle Network (DNN) est une méthode mésoscopique introduite récemment. Celle-ci s’appuie sur la conservation de la masse de soluté à proximité des pointes dendritiques pour calculer avec précision leur cinétique de croissance. Comme cette méthode repose sur l’estimation directe du gradient de composition à l’interface solide/liquide, le régime de croissance n’est plus supposé stationnaire. Nous introduisons la méthode Parabolic Thick Needle PTN reprenant la méthode de croissance du DNN pour une pointe. Elle est implémentée avec une méthode des éléments finis pour résoudre le flux de soluté est largement validé par rapport aux résultats analytiques provenant de la solution d’Ivantsov. Le couplage du CAFE avec la cinétique de croissance provenant du PTN permet d’obtenir un modèle unique de solidification s’appuyant sur 3 échelles. La grille CA gère à la fois la forme des enveloppes des grains et les mécanismes de ramification. Le maillage FE est utilisé pour résoudre les problèmes de flux et de conservation de masse et d’énergie à la fois à l’échelle de la couche de soluté de la pointe et à l’échelle du domaine simulé. Ceci est rendu possible grâce à une stratégie de remaillage anisotrope multi-critères. Diverses simulations démontrent les capacités du modèle. Les pistes d’amélioration sont développées pour espérer, à terme, une simulation 3D d’expériences de laboratoire. / Multiscale modelling of solidification processes is of great interest for industries. However coupling the multiple scale phenomena to reach quantitative large simulations is challenging. This is achieved using a combination of three methods : the Finite Element (FE), the Cellular Automaton (CA) and the Parabolic Thick Needle (PTN). The FE method provides a solution of the conservation equations, written for volume average quantities, that is suitable for large domain size computations. It serves for macroscopic description of heat and mass transfers. Additionally, the CA method tracks the development of the envelope of each individual dendritic grain at a mesoscopic scale. The coupling of these two methods is the CAFE model and was demonstrated to provide efficient and quantitative simulations of the columnar-to-equiaxed transition for instance. The Dendritic Needle Network (DNN) is another mesoscopic method recently introduced. It uses solute mass balance considerations in the vicinity of the tip of the dendrites to compute accurately the growth kinetics. Because it relies on adirect estimation of the composition gradient at the solid-liquid interface, steady state growth regime is no longer assumed. We introduce the Parabolic Thick Needle (PTN) method inspired from the DNN’s computed growth idea for one dendritetip. Its implementation with a FE method to solve the solute flow is extensively validated against analytical results given by the Ivantsov solution. Coupling CAFE with PTN computed growth kinetics provides a unique solidification model. The CA grid handles both the shape of the grain envelopes and branching mechanisms. The FE mesh is used to solve flux and conservation of mass and energy at both the scale of the dendrite tip solute layer and the domain dimensions. It is possible thanks to adaptive remeshing strategies. Various simulations demonstrate the capabilities of the model. The improvement areas are being developed in order to hope, in the long term, for 3D simulation laboratory experiments.

This study aims to improve the Additive Manufacturing (AM) design space for the popular multi-component Ni alloy Inconel 625 (IN625) thorough investigating the microstructural evolution, namely the solidification microstructure and the solid-state phase transformations during the Laser Powder Bed Fusion (LPBF) process. Highly non-equilibrium solidification and the complex reheating conditions during the LPBF process result in the formation of various types of solidification microstructures and grain morphologies which consequently lead to a wide range of mechanical properties. Understanding the melt’s thermal conditions, alloy chemistry, and thermodynamics during the rapid solidification and solid-state phase transformation in AM process will help to control material properties and even produce a material with specific microstructural features suited to a given application. This research helps to better understand the process-microstructure-property relationships of LPBF IN625. First, a set of simple but effective analytical solidification models were employed to evaluate their ability to predict the solidification microstructure in AM applications. As a case study, Solidification Microstructure Selection (SMS) maps were created to predict the solidification growth mode and grain morphology of a ternary Al-10Si-0.5Mg alloy manufactured by the LPBF process. The resulting SMS maps were validated against the experimentally obtained LPBF microstructure available in the literature for this alloy. The challenges, limitations, and potential of the SMS map method to predict the microstructural features in AM were comprehensively discussed. Second, The SMS map method was implemented to predict the solidification microstructure and grain morphology in an LPBF-built multi-component IN625 alloy. A single-track LPBF experiment was performed utilizing the EOSINT M280 machine to evaluate the SMS map predictions. The resulting microstructure was characterized both qualitatively and quantitatively in terms of the solidification microstructure, grain morphology, and Primary Dendrite Arm Spacing (PDAS). Comparing the experimentally obtained solidification microstructure to the SMS map prediction, it was found that the solidification mode and grain morphology were correctly predicted by the SMS maps. Although the formation of precipitates was predicted using the CALculation of PHAse Diagrams (CALPHAD) approach, it was not anticipated from the analytical solution results. Third, to further investigate the microsegregation and precipitation in IN625, Scanning Transmission Electron Microscopy (STEM) using Energy-Dispersive X-ray Spectroscopy (EDS), High-Angle Annular Dark-Field Scanning Transmission Electron Microscopy (HAADF-STEM), Scheil-Gulliver (with solute trapping) model, and DIffusion-Controlled TRAnsformations (DICTRA) method were employed. It was found that the microstructural morphology mainly consists of the Nickel-Chromium (gamma-FCC) dendrites and a small volume fraction of precipitates embedded into the interdendritic regions. The precipitates predicted with the computational method were compared with the precipitates identified via HAADF-STEM analysis inside the interdendritic region. The level of elemental microsegregation was overestimated in DICTRA simulations compared to the STEM-EDS results; however, a good agreement was observed between the Scheil and STEM-EDS microsegregation estimations. Finally, the spatial variations in mechanical properties and the underlying microstructural heterogeneity of a multi-layer as-built LPBF part were investigated to complete the process-structure-properties relationships loop of LPBF IN625. Towards this end, numerical thermal simulation, electron microscopy, nano hardness test, and a CALPHAD approach were utilized to investigate the mechanical and microstructural heterogeneity in terms of grain size and morphology, PDAS, microsegregation pattern, precipitation, and hardness along the build direction. It was found that the as-built microstructure contained mostly columnar (Nickel–Chromium) dendrites were growing epitaxially from the substrate along the build direction. The hardness was found to be minimum in the middle and maximum in the bottom layers of the build’s height. Smaller melt pools, grains, and PDAS and higher thermal gradients and cooling rates were observed in the bottom layers compared to the top layers. Microsegregation patterns in multiple layers were also simulated using DICTRA, and the results were compared with the STEM-EDS results. The mechanism of the formation of precipitates in different regions along the build direction and the precipitates’ corresponding effects on the mechanical properties were also discussed. / Thesis / Doctor of Philosophy (PhD)