<div>This dissertation focuses on a consistent and conservative Phase-Field method for multiphase flow problems, and it includes both model and scheme development. The first general question addressed in the present study is the multiphase volume distribution problem. A consistent and conservative volume distribution algorithm is developed to solve the problem, which eliminates the production of local voids, overfilling, or fictitious phases, but follows the mass conservation of each phase. One of its applications is to determine the Lagrange multipliers that enforce the mass conservation in the Phase-Field equation, and a reduction consistent conservative Allen-Cahn Phase-Field equation is developed. Another application is to remedy the mass change due to implementing the contact angle boundary condition in the Phase-Field equations whose highest spatial derivatives are second-order. As a result, using a 2nd-order Phase-Field equation to study moving contact line problems becomes possible.</div><div><br></div><div>The second general question addressed in the present study is the coupling between a given physically admissible Phase-Field equation to the hydrodynamics. To answer this general question, the present study proposes the <i>consistency of mass conservation</i> and the <i>consistency of mass and momentum transport</i>, and they are first implemented to the Phase-Field equation written in a conservative form. The momentum equation resulting from these two consistency conditions is Galilean invariant and compatible with the kinetic energy conservation, regardless of the details of the Phase-Field equation. It is further illustrated that the 2nd law of thermodynamics and <i>consistency of reduction</i> of the entire multiphase system only rely on the properties of the Phase-Field equation. All the consistency conditions are physically supported by the control volume analysis and mixture theory. If the Phase-Field equation has terms that are not in a conservative form, those terms are treated by the proposed consistent formulation. As a result, the proposed consistency conditions can always be implemented. This is critical for large-density-ratio problems.</div><div><br></div><div>The consistent and conservative numerical framework is developed to preserve the physical properties of the multiphase model. Several new techniques are developed, including the gradient-based phase selection procedure, the momentum conservative method for the surface force, the boundedness mapping resulting from the volume distribution algorithm, the "DGT" operator for the viscous force, and the correspondences of numerical operators in the discrete Phase-Field and momentum equations. With these novel techniques, numerical analyses ensure that the mass of each phase and momentum of the multiphase mixture are conserved, the order parameters are bounded in their physical interval, the summation of the volume fractions of the phases is unity, and all the consistency conditions are satisfied, on the fully discrete level and for an arbitrary number of phases. Violation of the consistency conditions results in inconsistent errors proportional to the density contrasts of the phases. All the numerical analyses are carefully validated, and various challenging multiphase flows are simulated. The results are in good agreement with the exact/asymptotic solutions and with the existing numerical/experimental data.</div><div> </div><div><br></div><div>The multiphase flow problems are extended to including mass (or heat) transfer in moving phases and solidification/melting driven by inhomogeneous temperature. These are accomplished by implementing an additional consistency condition, i.e., <i>consistency of volume fraction conservation</i>, and the diffuse domain approach. Various problems are solved robustly and accurately despite the wide range of material properties in those problems.</div>

We present a discretization method of the Navier–Stokes Cahn–Hilliard equations which offers an impressing simplicity, making it easy to implement a scalable parallel code from scratch. The method is based on a special pressure projection scheme with incomplete pressure iterations. The resulting scheme admits solution by an explicit Euler method. Hence, all unknowns decouple, which enables a very simple implementation. This goes along with the opportunity of a straightforward parallelization, for example, by few lines of Open Multi-Processing (OpenMP) or Message Passing Interface (MPI) routines. Using a standard benchmark case of a rising bubble, we show that the method provides accurate results and good parallel scalability. / Wir stellen eine Diskretisierungsmethode der Navier-Stokes-Cahn-Hilliard-Gleichungen vor, welche es erlaubt, mit wenig Aufwand einen einfachen, skalierbar parallelen Code zu implementieren. Die Methode basiert auf einem Druckprojektionsschema mit unvollständigen Druckiterationen was eine Lösung durch eine explizite Euler-Methode erlaubt. Somit sind alle Unbekannten entkoppelt, was eine sehr einfache Implementierung mit einer unkomplizierten Parallelisierung ermöglicht, zum Beispiel durch Open Multi-Processing (OpenMP) oder Message Passing Interface (MPI) Routinen. Anhand eines Standard-Benchmark-Falls einer aufsteigenden Blase zeigen wir, dass die Methode genaue Ergebnisse und eine gute parallele Skalierbarkeit liefert.

The aluminothermic (AT) process utilises a self-propagating high-temperature synthesis (SHS) type reaction for producing primarily thermite steel and alumina slag at high temperatures during the welding of rails. In this work, an investigation on the early stages of the aluminothermic process, the separation of AT reaction products and mould filling has been carried out, using both experimental and computational methods to predict the time duration of a complete separation and to obtain a better understanding of the internal multiphase flow within the crucible and mould. The decomposition of AT reaction products after the combustion and the subsequent mould filling by thermite steel and alumina slag have been simulated numerically, using a diffusive phase field and volume-of-fluid model. However, to minimize numerical errors on the input parameters of the high- temperature multiphase flow, a careful review on transport properties has been made. Missing data, e.g. the contact angle of thermite steel on waterglass-bonded mould and crucible wall material has been investigated experimentally. Being further necessary for the prediction of the separation time of AT reaction products in compacted thermite, results on the propagation front velocity show a decreasing trend with increasing initial compact temperature. Further, the combustion front velocity is used for a subsequent analysis of the separation time, which is obtained from the phase distribution of thermite steel, alumina slag and intermetallic compounds, using a combustion front quenching (CFQ) methodology. Moreover, geometric modifications on the crucible and mould have been developed for a reduction in changeover time, as well as an optimized multiphase flow field. Their performance during crucible discharge and mould filling has been verified numerically. Furthermore, alumina slag inclusions have been tracked within the mould using a volume-of-fluid approach with their final positions being verified through an authentic welding. / Während des aluminothermischen (AT) Prozesses findet eine SHS-Reaktion Anwendung, um primär Thermitstahl und Aluminiumoxidschlacke bei hohen Temperaturen für das Verschweißen von Bahnschienen herzustellen. In dieser Arbeit wurden Anfangsstadien, welche die Separation der AT-Reaktionsprodukte sowie das Füllen der Gießform einbeziehen, unter Anwendung von sowohl experimentellen als auch numerischen Verfahren untersucht. Damit konnte die Zeitdauer einer kompletten Separation ermittelt und ein genaueres Verständnis der Mehrphasenströmung in Tiegel und Gießform erlangt werden. Die Separation der AT-Reaktionsprodukte nach der aluminothermischen Reaktion und die anschließende Formfüllung wurden mit einem diffusen Phasenfeld und einem Volume-of-Fluid-Modell numerisch berechnet. Für die Minimierung numerischer Fehler in den Eingangsgrößen dieser Hochtemperatur-Mehrphasenströmungen wurde eine intensive Literaturrecherche durchgeführt und fehlende Parameter, wie zum Beispiel die Kontaktwinkel von Thermitstahl auf Wasserglas gebundenem Form- und Tiegelmaterial, wurden experimentell ermittelt. Messungen der Reaktionsfrontgeschwindigkeit in gepresstem Thermit sind notwendig für eine Vorhersage der Separationszeit der AT-Reaktionsprodukte, und die Ergebnisse zeigen einen linear abfallenden Trend mit zunehmender Anfangstemperatur des verdichteten Materials. In dieser Arbeit wurde die Geschwindigkeit der Reaktionsfront verwendet, um aus der Phasenverteilung von Thermitstahl, Aluminiumoxidschlacke und intermetallischen Verbindungen als Ergebnis des CFQ-Experimentes die Separationszeit in verdichtetem Thermit zu approximieren. Es wurden Modifikationen an Tiegel und Gießform erprobt, die für eine Verbesserung der internen Strömungsführung sowie für die Reduzierung der Umrüstzeit sorgen sollen. Die Effizienz dieser Veränderungen wurde anschließend mit numerischen Methoden überprüft. Des Weiteren konnten durch eine Realschweißung die numerisch vorhergesagten finalen Positionen von Schlackeeinschlüssen innerhalb der Gießform verifiziert werden.

Interactions between three immiscible phases, including incompressible viscoelastic structures and fluids, form standard constellations for countless scenarios in natural science. The complexity of many such scenarios has motivated various research efforts in scientific computing. This work presents novel numerical approaches for two specific of these ternary fluid-structure interaction constellations. The potential of these approaches is demonstrated by diverse applications. First, a phase field model is developed describing the interaction between a fluid and a viscoelastic solid. For this purpose, a Navier-Stokes-Cahn-Hilliard system is considered together with a hyperelastic neo-Hookean model. Based on this, an arbitrary Lagrangian-Eulerian (ALE) method is implemented to simulate the indentation of the solid material in the context of atomic force microscopy, capable of predicting physical parameters. Next, the second approach is developed to describe the interaction between a two-phase fluid and a viscoelastic solid, where fluid and solid are defined on separate domains but aligned at the interface between them. The previously introduced phase field model is used to represent the fluid and an ALE method is used for the motion of the grid, where the fluid-solid interface moves with flow velocity. A unified system is solved in all subdomains, which includes both the balance of mass and momentum and the balance of forces at the fluid-solid interface. Simulations of static and dynamic soft wetting are subsequently presented, in particular a contact line moving over a substrate with oscillating stick-slip behavior. This work combines the advantages of phase field and ALE methods for meaningful simulations and emphasizes validity and numerical stability in all approaches.

<div> Porous lithium ion battery (LIB) electrodes are comprised of electrochemically active material particles that store lithium and a surrounding conductive binder, liquid electrolyte, carbon black mixture that facilitates ionic and electronic transport. Typically, lithium diffusivity is several orders of magnitude smaller in the active material as compared to the surrounding electrolyte, making the electrode microstructure a governing factor in determining the balance between its lithium storage capacity and transport rate. Here, the effects of microstructure on the performance of LIBs are systematically analyzed at three length scales - the single particle length scale, the spatially resolved multiple particle length scale, and the porous electrode layer (homogenized) length scale. At the single particle length scale, a thermodynamically consistent variational framework is presented to examine the effects of crystallographic anisotropy, crystallographic texture, grain size, and grain morphology on the LiNi_1/3Mn_1/3Co_1/3O₂ (NMC111) chemistry. The theory was extended to the spatially resolved multiple particle length scale and the porous electrode layer length scale to explain the microstructural origin of experimentally observed instances of apparent phase separation in NMC111. At the electrode length scale, a data driven framework is presented to evaluate the electrochemical performance of a wide range of particle morphologies and battery architectures. Specifically, microstructural characteristics of 53 356 microstructures are assessed, and strategies to optimize electrode design parameters such as active particle morphology, spatial orientation, electrode porosity, and cell thickness are presented.</div><p></p>

The phase-field formulation for fracture based on the framework of representative crack elements is extended to transient thermo-mechanics. The finite element formulation is derived starting from the variational principle of total virtual power. The intention of this manuscript is to demonstrate the potential of the framework for multi-physical fracture models and complex processes inside the crack. The present model at hand allows to predict realistic deformation kinematics and heat fluxes at cracks. At the application of fully coupled, transient thermo-elasticity to a pre-cracked plate, the opened crack yields thermal isolation between both parts of the plate. Inhomogeneous thermal strains result in a curved crack surface, inhomogeneous recontact and finally heat flow through the crack regions in contact. The novel phase-field framework further allows to study processes inside the crack, which is demonstrated by heat radiation between opened crack surfaces. Finally, numerically calculated crack paths at a disc subjected to thermal shock load are compared to experimental results from literature and a curved crack in a three-dimensional application are presented.

<p dir="ltr">Scientific discovery of new knowledge from data can enhance our understanding of the physical world and lead to the innovation of new technologies. AI-driven methods can greatly accelerate scientific discovery and are essential for analyzing and identifying patterns in huge volumes of experimental data. However, current AI-driven scientific discovery pipeline suffers from several inefficiencies including but not limited to lack of <b>precise modeling</b>, lack of <b>efficient learning methods</b>, and lack of <b>human-in-the-loop integrated frameworks</b> in the scientific discovery loop. Such inefficiencies increase resource requirements such as expensive computing infrastructures, significant human expert efforts and subsequently slows down scientific discovery.</p><p dir="ltr">In this thesis, I introduce a collection of methods to address the lack of precise modeling, lack of efficient learning methods and lack of human-in-the-loop integrated frameworks in AI-driven scientific discovery workflow. These methods include automatic physics model learning from partially annotated noisy video data, accelerated partial differential equation (PDE) physics model learning, and an integrated AI-driven platform for rapid analysis of experimental video data. <b>My research has led to the discovery of a new size fluctuation property of material defects</b> exposed to high temperature and high irradiation environments such as inside nuclear reactors. Such discovery is essential for designing strong materials that are critical for energy applications.</p><p dir="ltr">To address the lack of precise modeling of physics learning tasks, I developed NeuraDiff, an end-to-end method for learning phase field physics models from noisy video data. In previous learning approaches involving multiple disjoint steps, errors in one step can propagate to another, thus affecting the accuracy of the learned physics models. Trial-and-error simulation methods for learning physics model parameters are inefficient, heavily dependent on expert intuition and may not yield reasonably accurate physics models even after many trial iterations. By encoding the physics model equations directly into learning, end-to-end NeuraDiff framework can provide <b>~100%</b> accurate tracking of material defects and yield correct physics model parameters. </p><p dir="ltr">To address the lack of efficient methods for PDE physics model learning, I developed Rapid-PDE and Reel. The key idea behind these methods is the random projection based compression of system change signals which are sparse in - either value domain (Rapid-PDE) or, both value and frequency domain (Reel). Experiments show that PDE model training times can be reduced significantly using our Rapid-PDE (<b>50-70%)</b> and Reel (<b>70-98%</b>) methods. </p><p dir="ltr">To address the lack of human-in-the-loop integrated frameworks for high volume experimental data analysis, I developed an integrated framework with an easy-to-use annotation tool. Our interactive AI-driven annotation tool can reduce video annotation times by <b>50-75%</b>, and enables material scientists to scale up the analysis of experimental videos.</p><p dir="ltr"><b>Our framework for analyzing experimental data has been deployed in the real world</b> for scaling up in-situ irradiation experiment video analysis and has played a crucial role in the discovery of size fluctuation of material defects under extreme heat and irradiation. </p>

La cristallisation sous tension (SIC) du caoutchouc naturel (NR) a fait l’objet d’un nombre considérable d’études depuis sa découverte il y a près d’un siècle. Cependant, il existe peu d’informations dans la littérature concernant le comportement du caoutchouc à des vitesses de sollicitation proches des temps caractéristiques de cristallisation. L’objectif de cette thèse est alors de contribuer à la compréhension du phénomène de cristallisation sous tension grâce à des essais dynamiques à grandes vitesses. Pour répondre à cet objectif, nous avons développé une machine de traction permettant de déformer des échantillons d’élastomères à des vitesses de sollicitation pouvant aller jusqu’à 290s-1. Les essais ont été réalisés sur quatre NR avec des taux de soufre variables, deux NR chargés comportant des taux de noir de carbone différents. Nous avons également étudié un matériau synthétique à base de polyisoprène (IR) afin de comparer ses performances à celle du NR. Les essais dynamiques étant relativement difficiles à interpréter, un travail conséquent a donc été d’abord réalisé à basse vitesse. En outre, l’approche expérimentale proposée a été couplée à une approche thermodynamique de la SIC. Les mécanismes généraux associés à la cristallisation que nous identifions sont les suivants: lors d’une traction, la cristallisation consiste en l’apparition de populations cristallines conditionnée par l’hétérogénéité de réticulation des échantillons. Cette cristallisation semble nettement accélérée dès lors que ce cycle est réalisé au-dessus de la déformation de fusion. Nous attribuons ce phénomène à un effet mémoire dû à un alignement permanent des chaînes. Enfin, l’effet de la vitesse est décrit théoriquement en intégrant un terme de diffusion des chaînes dans la cinétique de SIC. Cette approche couplée à des essais mécaniques suggère que la SIC est essentiellement gouvernée par la cinétique de nucléation. Lors des tests dynamiques, la combinaison de l’effet mémoire et d’une accélération de la fusion pendant le cycle entraine une nette diminution voire une disparition de l’hystérèse cristalline. En outre, l’auto-échauffement, qui augmente progressivement avec la fréquence du cycle, tend à supprimer l’effet mémoire en provoquant le passage du cycle en dessous de la déformation de fusion. Lors de ces essais dynamiques, la SIC semble favorisée pour le matériau le moins réticulé. Nous attribuons cet effet au blocage d’enchevêtrements jouant le rôle de sites nucléants pour la SIC. Le matériau chargé semble avoir une moins bonne aptitude à cristalliser à hautes vitesses, par rapport à l’élastomère non chargé, en raison d’un auto-échauffement important à l’interface entre charges et matrice. Enfin, nous notons une convergence des cinétiques de cristallisation du caoutchouc naturel et synthétique à grande déformation et grande vitesse de sollicitation, que nous attribuons à la prédominance du terme énergétique d’origine entropique dans la cinétique de nucléation. / Strain induced crystallization (SIC) of Natural Rubber (NR) has been the subject of a large number of studies since its discovery in 1929. However, the literature is very poor concerning the study of SIC when samples are deformed with a stretching time in the range of the SIC characteristic time (around 10msec-100msec). Thus, the aim of this thesis is to contribute to the understanding of the SIC phenomenon thanks to dynamic tensile tests at high strain rates. To meet this goal, we have developed a dynamic tensile test machine allowing stretching samples of elastomers at strain rates up to 290 s-1. The tests are carried out on four NR with different sulphur amount, two NR with different carbon black filler amounts. We also studied a synthetic rubber made of polyisoprene chains (IR) able to crystallize under strain. Dynamic tests are relatively difficult to interpret; a significant work has thus been first performed at slow strain rate. Moreover, the experiments are coupled with a thermodynamic approach. First, the general mechanisms associated to the crystallization are identified as follows: during mechanical loading or during cooling in the deformed state, SIC is the result of successive appearance of crystallite populations whose nucleation and growth depend on the local network density. Crystallization is enhanced when the cycle is performed above the melting stretching ratio. This phenomenon is attributed to a memory effect due to a permanent alignment of the chains. Finally, the effect of the strain rate is theoretically described thanks to a diffusion term. This approach, coupled with experiments suggests that SIC is mainly governed by the nucleation kinetics. For the dynamic test, the combination of the memory effect and the acceleration of the melting during the cycle lead to a reduction or even disappearance of the crystalline hysteresis. In addition, self-heating, which progressively increases with the frequency of the cycle, causes the delay of the melting stretching ratio. This well explains why the crystallinity index decreases at the minimum stretching ratio of the dynamic cycles when the frequency increases. We finally compared the ability of our different rubbers to crystallize at high strain rates. SIC is enhanced for the weakly crosslinked rubber. This might be related to the dynamics of its free entanglements, these ones acting as supplementary crosslinks at high strain rates. Then, a filled rubber is compared to the unfilled one. We found that the filled sample has a lower ability to crystallize at high strain rates as compared to the unfilled one. This is likely due to the strong self-heating at the interface between the fillers and the rubbery matrix. Finally, we observe a convergence of crystallization kinetics in natural and synthetic rubbers at high strains and high strain rates. This is attributed to the predominance of the entropic energy in the nucleation kinetics in these experimental conditions.

La cristallisation induite par déformation des élastomères est un processus cinétique qui conduit à la formation de nano-cristallites thermodynamiquement stables. La présence de ces nano-cristallites au sein de la phase amorphe modifie considérablement les propriétés mécaniques des élastomères cristallisables. Ces élastomères ont en effet la propriété intéressante d'être auto-renforçants. L’objectif de ce travail est de développer un modèle physique capable de décrire localement l’évolution de la microstructure sous l’effet d’un champ de contrainte élastique durant la cristallisation. Dans ce but, un modèle de champ de phase est élaboré et mis en œuvre dans le cadre de la mécanique des milieux continus en couplant thermodynamique et mécanique avec une dynamique de transition de phase d’Allen-Cahn. La description thermodynamique de la cristallisation induite par déformation à petite échelle est basée sur la fonctionnelle d’énergie libre du système amorphe-cristal. Les conséquences du choix de cette formulation sont discutées, on étudie en particulier les effets de contraintes élastiques sur l’équilibre des phases en volumes ainsi que sur la cinétique de croissance des domaines cristallins au sein de l’amorphe. L’introduction de l’élasticité du réseau des contraintes topologiques induite par les enchevêtrements et/ou les nœuds de réticulation dans le modèle de champ de phase a permis de mettre en évidence l’existence d’un état stable de cristallites formées (modèle énergétique) mais aussi des instabilités de croissance (modèle cinétique). Sur la base de ces deux modèles, cinétique et énergétique, nous avons étudié systématiquement l’influence des contraintes topologiques sur la cinétique de croissance et nous montrons que cette cinétique est en effet contrôlée par l’accumulation de contraintes élastiques à l’interface. La prise en compte de l’élasticité du réseau des contraintes topologiques dans l’approche thermodynamique de cristallisation prédit une augmentation de la tension de surface et par conséquent un arrêt du mécanisme de croissance en donnant lieu à la formation de cristallites stables. Enfin, nous avons adopté le modèle énergétique pour modéliser le couplage entre nucléation, croissance et déformation cyclique. Pour valider le modèle local proposé une comparaison entre les résultats des simulations par champ de phase et les données expérimentales issues de la caractérisation d’un caoutchouc naturel réticulé est effectuée et nous montrons qualitativement l’accord entre l’expérience et le modèle. / Natural rubber NR and more generally elastomer presents unique physical properties that are very important for many engineering applications. Strain induced crystallization of elastomer presents a major interest because it improves considerably the mechanical properties. In fact, the presence of crystallites within the amorphous phase in a polymer network induces a strengthening of this material, giving NR a self-reinforcement character. In this thesis, we develop a mesoscopic model to describe the crystallization of elastomers under strain. In this context, we present a kinetic model using a new physical approach: a phase field model. This model combines the crystallization thermodynamics with the local stress field. The thermodynamic description of the phase transition is based on a Gibbs free energy functional F which contains all energy contributions of the system: the bulk contributions (enthalpy and entropy) and surface tension. To understand the experimental observation of nanometer size crystalites, an explicit account of the topological constraints induced by both entanglements and/or crosslinks is necessary. We investigated two limiting mechanisms, a kinetic limitation of the growth, and an energetic limitation. Based on both the kinetic and the energetic approaches, we have systematically studied the influence of topological constraints on the growth process. We have shown that the growth process is affected by the accumulation of elastic stress at the interface. The kinetic model predicts the existence of instabilities during the growth. These instabilities induce a heterogeneous dynamical growth which leads to the formation of dendrite like structures. On the contrary, the energetic approach predicts an exponential increase of the surface tension during the growth that limits the size of the crystallites very efficiently. In the last part we investigated elastomer crystallization under cyclic deformation. To this end, we coupled the previous energetic model with the nucleation process. Finally the simulation data are compared with experimental measurements.

The investigation of failure in brittle materials, subjected to dynamic transient loading conditions, represents one of the ongoing challenges in the mechanics community. Progresses on this front are required to support the design of engineering components which are employed in applications involving extreme operational regimes. To this purpose, this thesis is devoted to the development of a framework which provides the capabilities to model how crack patterns form and evolve in brittle materials and how they affect the quantitative description of failure. The proposed model is developed within the context of diffusive interfaces which are at the basis of a new class of theories named phase field models. In this work, a set of additional features is proposed to expand their domain of applicability to the modelling of (i) rate and (ii) pressure dependent effects. The path towards the achievement of the first goal has been traced on the desire to account for micro-inertia effects associated with high rates of loading. Pressure dependency has been addressed by postulating a mode-of-failure transition law whose scaling depends upon the local material triaxiality. The governing equations have been derived within a thermodynamically-consistent framework supplemented by the employment of a micro-forces balance approach. The numerical implementation has been carried out within an updated lagrangian finite element scheme with explicit time integration. A series of benchmarks will be provided to appraise the model capabilities in predicting rate-pressure-dependent crack initiation and propagation. Results will be compared against experimental evidences which closely resemble the boundary value problems examined in this work. Concurrently, the design and optimization of a complimentary, improved, experimental characterization platform, based on the split Hopkinson pressure bar, will be presented as a mean for further validation and calibration.