La formation d'agrégats d'anticorps monoclonaux en solution est difficile à empêcher. Même si la présence de gros agrégats est assez rare, leur existence peut avoir des effets dramatiques dans les systèmes d'injection, en menant à des situations de colmatage partiel ou total de la restriction dans ce dernier. Cela entraîne une injection mal contrôlée ou même une obstruction totale du système d'injection. Très peu est connu sur le rôle de la taille des agrégats et de la pression appliquée sur de tels évènements de colmatage. Dans cette thèse, nous présentons un système microfluidique modèle, imitant les systèmes médicaux d'injection afin de comprendre fondamentalement le colmatage de restrictions d'une taille donnée. Des solutions très concentrées en anticorps monoclonaux nous permettent de créer des agrégats de protéines (plus grands que 50 micromètres) en utilisant un stress mécanique ou thermique. Nous montrerons que le colmatage a lieu quand les agrégats atteignent la taille de la restriction et peut dans certains cas être défait en augmentant la pression appliquée. La possibilité observée d'éjecter des agrégats de la restriction via une augmentation en pression indique le rôle important de la déformabilité des agrégats de protéines, à ce jour complètement inexplorée. Nous réalisons des expériences systématiques pour différentes tailles relatives d'agrégats et de pressions appliquées, et nous mesurons le débit en sortie. Malgré leurs formes et densités différentes, nous pouvons prédire le nombre d'évènements de colmatage pour une taille donnée de restriction par des mesures utilisant le Flow Imaging Microscopy (MFI). De plus, notre système peut détecter l'occurrence de très gros agrégats (très rares) souvent non détectés par d'autres techniques. Avec un modèle mécanique simple, nous pouvons estimer pour la première fois un ordre de grandeur du module d'Young et un diamètre effectif de pores pour des agrégats d'anticorps monoclonaux. Nous avons également développé une autre expérience modèle dans un canal hyperbolique couplé avec un flow focusing afin d'observer la déformation d'agrégats sous écoulement élongationnel. Nous décrirons leur comportement en analysant leurs trajectoires qui sont pour la plupart d'entre eux du tumble et de l'alignement avec l'écoulement. De plus, nous développerons un modèle mécanique qui tient compte de la force de friction dans une expérience modèle contrôlée avec une solution polymérique de PEGDA. Nous étudierons ainsi le rôle d'une différence de pression minimale à appliquer pour remettre la particule en mouvement dans la restriction, et ainsi relier cela aux agrégats de protéines. / The formation of aggregates in solutions of monoclonal antibodies is difficult to prevent. Even if the occurrence of large aggregates is rather rare, their existence can have dramatic effects in injection devices, as they can lead to partial or total clogging of constrictions in the latter. This leads to badly controlled injection or even total obstruction of the device. Little is know on the role of aggregate size and applied pressure on such clogging events. In this thesis, we present a microfluidic model system, mimicking medical injection devices to gain fundamental understanding of the clogging of constrictions of given size. Highly concentrated solutions of monoclonal antibodies allow us to create protein aggregates (bigger than 50 micrometers) using mechanical or heat stress. We show that clogging occurs when aggregates reach the size of the constriction and can in some cases be undone by increasing the applied pressure. The observed possibility to eject aggregates from constrictions via an increase in pressure indicates the important role of protein aggregate deformability, so far completely unexplored. We perform systematic experiments for different relative aggregate size and the applied pressure, and measure the flow-rate. Despite their different shapes and density, we can predict the number of clogging events for a given constriction size by Flow Imaging Microscopy (MFI) measurements. In addition our device can detect the occurrence of very rare big aggregates often overlooked by other detection techniques. With a simple mechanical model where we neglected the friction, we could estimate for the first time an order of magnitude for the Young modulus and a porous diameter for monoclonal antibodies aggregates. We also develop another model experiment with an hyperbolic channel coupled with a flow focusing to observe deformation of the aggregates under extensional flow. We describe their behavior by analyzing their trajectories which are for most of them tumbling and alignment with the flow. Moreover, we develop a mechanical model which took into account the friction force in a controlled model experiment with polymeric solution. We thus investigate the role of a minimal applied pressure to generate the particle movement into the constriction, and then link it with protein aggregates.

Bentonite is regarded as an ideal geomaterial for the engineering barrier system of a deep geological repository (DGR) where nuclear wastes are disposed, as it has several desirable properties for sealing the nuclear wastes, including low permeability, low diffusion coefficient, high adsorption capacity and proper swelling ability. Nevertheless, gas migration in saturated bentonite may undermine the sealing ability of the geomaterial. Previous experimental studies showed that the gas migration process is accompanied by complex hydromechanical (HM) behaviors, such as gas breakthrough phenomenon, development of preferential pathways, build-up of water pressure and total stress, nearly saturated state after gas injection test, localized consolidation, water exchange between clay matrix and developed fractures and self-sealing process. These experimentally observed behaviors should be properly modelled for conducting a reliable performance assessment for the geomaterial over the lifespan of DGR. In this thesis, two different coupled HM frameworks, i.e., one based on double porosity (DP) concept, referred to as coupled HM-DP framework, and the other on phase field (PF) method, referred to as coupled HM-PF framework, are proposed to simulate the gas migration process in saturated bentonite. For the coupled HM-DP framework, the saturated bentonite is assumed as a superposition of a MAcro-Continuum (MAC) and a MIcro-Continuum (MIC). Two-phase flow is only allowed in the MAC, whereas the MIC is impermeable to both water and gas. Nevertheless, the water can transfer between the MIC and the MAC under the water pressure gap. The first coupled HM model in this framework is based on a double effective stress concept. Mechanical behaviors of the MAC and the MIC are respectively governed by Bishop-type effective stress and Terzaghi’s effective stress. The model can well simulate the evolutions of both gas pressure and gas outflow rate, the water exchange between clay matrix and developed pathways, the high degree of saturation and the consolidation of clay matrix. To account for the development of preferential pathways, the damaging effect has been introduced in the framework. In this improved model, Bishop-type effective stress for the MAC is replaced by the independent stress state variables, i.e., net normal stress and suction, since using the net normal stress is beneficial to simulating tensile failure under high gas pressure. Numerical results showed that the damage-enhanced model can well describe the effect of the development of preferential pathways on the build-up of water pressure and total stress. In addition, the proposed hysteretic models for intrinsic and relative permeabilities make the coupled HM framework more flexible to reproduce the experimental results. To explicitly simulate the development of preferential pathways, a coupled HM-PF framework is developed by using Coussy’s thermodynamic theory and the microforce balance law. The coupled HM-PF framework is implemented in the standard Finite Element Method (FEM). To avoid the pore pressure oscillation and enhance the computational efficiency, a stabilized mixed finite element, in which linear shape functions are selected for interpolating all primary variables, is adopted to discretize the whole domain. In the developed framework, swelling pressure (initial stress) is accounted for by introducing a modified strain tensor that is the sum of the strain tensor due to deformation and the strain tensor calculated from the initial stress. The numerical results showed that the developed coupled HM-PF framework can capture some important behaviors, such as the discrete pathways, localized gas flow, built-up of water pressure and total stress under constant volume condition and nearly saturated state in clay matrix. A spatially autocorrelated random field is introduced into the framework to describe the heterogeneous distribution of HM properties in bentonite. The heterogeneity is beneficial to simulating the fracture branching and the complex fracture trajectory. Numerical results showed that some factors, such as Gaussian random field, coefficient of variation, boundary condition and injection rate, have significant influences on the fracture trajectory. At the end of the thesis, the obtained numerical results are synthesized and analyzed. Based on the analysis, the pros and cons of the developed numerical models are discussed. Corresponding to the limitations, some recommendations are proposed for future studies.