The injection of nanoparticles is a promising and novel approach to enhancing oil recovery in depleted fields. Nanoparticles have one dimension that is smaller than 100 nm and have many unique properties that are useful when it comes to oil recovery. Their small size and the ability to manipulate particle properties are a couple of the advantageous properties. The small size of nanoparticle allows them to easily pass through porous media. Manipulating nanoparticle properties allows for wettability modifications or controlled release of chemicals at a precise location in the formation. Injection of nanoparticle dispersions for secondary or tertiary recovery in corefloods has yielded positive results. Field tests using nanoparticles have also yielded positive results with increased oil recovery. While there has been a sizable amount of work related to corefloods, limited investigation has been reported using micromodels. Micromodels are valuable because they allow for pore scale viewing of the oil recovery, which is not possible with corefloods. In this research both polydimethylsiloxane (PDMS) and glass microfluidic devices were fabricated to test the EOR potential of different types of nanoparticles. Much of the work described in this thesis involved the use of a dead-end pore geometry to trap oil. First the pore space was filled with oil and then waterflooded. This left some oil trapped in the dead-end pores. PDMS micromodels proved difficult to trap oil in the dead-end pores; because of this glass micromodels were tested. After trapping oil, a nanoparticle dispersion was injected into the pore space to test the potential of the dispersion to reduce the residual oil saturation in the dead-end pores. The nanoparticle dispersion was injected at different flow rates (1 [mu]l/hr to 50 [mu]l/hr) to test the effect of flow rate on residual oil recovery. / text

Pour ce travail, nous utilisons un micromodèle à haute complexité et à structure fixe pour faire une série d’expériences en variant la vitesse d’injection, la qualité de la mousse, les distributions de taille de bulles d’injection, et la méthode d’injection. Nous mettons en œuvre un suivi individuel de bulles pour associer les propriétés d’écoulement aux propriétés de taille de bulles ainsi que les caractéristiques structurelles du milieu poreux. Nous proposons de nouveaux outils pour décrire l’écoulement d’un point de vue global et local de différentes manières. Nous établissons des comportements spécifiques à chaque taille de bulle, en montrant que les bulles des mousses piégées sont plus probables d’être de taille inférieure aux tailles de bulles moyennes, alors que les mousses en mouvement accèdent elles-mêmes à différents chemins d’écoulement selon les tailles de bulles. Les bulles plus volumineuses s’écoulent en majorité dans des chemins préférentiels à haute vitesse, généralement parallèles au gradient de pression, mais les petites bulles sont transportées en supplément à l’intérieur de chemins transversaux liant les chemins préférentiels. Ailleurs, pour nos données nous démontrons l’importance supérieure de la fraction de mousse piégée vis-à-vis de la densité de bulles quant à l’explication microscopique de la viscosité apparente, malgré une contribution des deux. Nous caractérisons structurellement les zones piégées à répétition, comme étant soit des zones à faible coordination de pore, de faible taille de seuil d’entrée, d’orientation de seuil désavantageuse, ou une combinaison de ceux-ci. Les zones à fort écoulement échappent à une caractérisation en termes de paramètres de structure locale et nécessitent une considération de l’information des différents chemins traversant la totalité du modèle. À ce but, afin de décrire les zones à fort écoulement, nous développons un modèle générant des chemins, utilisant une représentation en graphe du milieux poreux, basé sur une décomposition initiale en pores et seuils, qui intègre seulement les notions de taille de seuil et d’orientation de seuil relatif au gradient de pression pour caractériser les chemins. / In this work, we use of a high-complexity micromodel of fixed structure on which we perform a series of experiments with varying injection rates, foam qualities, inlet bubble size distributions and injection methods. We perform individual bubble tracking and associate flow properties with bubble size properties and structural characteristics of the medium. We propose new tools describing the local and global flow in different ways. We establish specific behaviors for different bubble sizes, demonstrating that trapped foams are more likely to have smaller than average bubble sizes, while flowing bubbles also tend to segregate in different flow paths according to bubble size. Larger bubbles tend to flow in high-velocity preferential paths that are generally more aligned with pressure gradient, but smaller bubbles tend to access in supplement transversal paths linking the different preferential paths. Furthermore, for our data we establish the pre-eminence of the trapped foam fraction over bubble density within the microscopic explanation of apparent viscosity, although both contribute to some degree. We structurally characterize consistently trapped zones as areas with either low pore coordination, low entrance throat size, unfavorable throat orientation or a combination thereof. High-flow zones however cannot be characterized in terms of local structural parameters and necessitate integration of complete path information from the entire model. In this regard, in order to capture the high-flow zones, we develop a path-proposing model that makes use of a graph representation of the model, from an initial decomposition into pores and throats, that uses only local throat size and throat orientation relative to pressure gradient to characterize paths.

Le stockage géologique du CO2 est une stratégie prometteuse pour limiter la concentration de CO2anthropique dans l’atmosphère. Les aquifères salins (AS) ont été identifiés comme des optionsviables car ils possèdent de grandes capacités potentielles de stockage. Toutefois, les processusrelatifs au piégeage du CO2 souffrent d’un manque de connaissances fondamentales car il existe peude méthodes d’expérimentation rapides et reproductibles, travaillant dans les conditions du stockagegéologique. Ainsi, nous avons développé des microréacteurs haute pression, véritables laboratoiresgéologiques sur puce (GLoCs), recréant les conditions de porosité et de perméabilité des AS pour :(i) Mesurer la solubilité du CO2 dans l’eau et les saumures via un couplage microsystèmes /spectroscopie Raman ;(ii) Etudier les mécanismes d’invasion du CO2 dans les formations géologiques, incluantnotamment les écoulements diphasiques en milieux poreux, les séparations de phases etla précipitation des carbonates. / CO2 geological storage is a promising strategy to control the anthropogenic CO2 concentration in theatmosphere. Deep saline aquifers (DSA) were identified as viable options since they exhibit largestorage capacity. However, processes inherent to CO2 trapping suffer from a lack of fundamentalknowledge, since there are too few fast and reproducible experimental approaches able to work atgeological storage conditions. Therefore, to address these limitations, we have developed highpressure microreactors, so-called “geological labs on a chip” - GloCs – allowing mimicking porosityand permeability conditions of DSA for:(i) Measuring solubility of CO2 in water and brine through the combination of microsystemsand confocal Raman spectroscopy,(ii) Studying invasion mechanisms of CO2 in geological formations, including in particularbiphasic flows in porous media, phase separation and carbonates precipitation.

En vue de contrôler les émissions de gaz à effet de serre, il est envisagé d’injecter du CO2 dans des réservoirs géologiques. Or le CO2 n'est pas un gaz inerte. En modifiant la composition chimique de l'eau in situ, il est à l'origine d'interactions roche/fluide. Ces réactions géochimiques impactent les propriétés d'écoulement. Aussi, pour s'assurer de la viabilité et de la pérennité du stockage, les opérateurs ont besoin de simulations tenant compte de ces écoulements réactifs. Cependant les paramètres de l'équation macroscopique de transport utilisée sont affectés par les réactions surfaciques. Or, ces spécificités dues au transfert de masse ne sont pas prises en compte actuellement. De même, la loi perméabilité-porosité (K-F) n’est estimée que semi-empiriquement. Le but de cette thèse a été de développer une méthode pour obtenir les coefficients macroscopiques précédents et les relations K-F, en résolvant les équations gouvernant les phénomènes à l'échelle du pore. Pour ce faire, nous avons utilisé l'approche réseau de pores. L'avantage du modèle réseau est qu'il prend en compte explicitement la structure tout en conceptualisant cette dernière à un ensemble de pores et de canaux à la morphologie simplifiée (sphères, cylindres). L'étude est basée sur deux changements d'échelles successifs : du local au pore, puis du pore à la carotte. Le problème de transport réactif est résolu pour des éléments basiques, analytiquement ou numériquement. Puis, en faisant appel aux solutions précédemment trouvées, le transport réactif est traité sur l'ensemble du réseau. Notre model fut validé par des observations sur micromodèles, puis à l'aide d'une expérience d'altération acide / The geological storage of CO2 is considered as an attractive option to reduce the greenhouse gas emissions in the atmosphere. CO2 is not an inert gas, however. Its dissolution in brine forms a weak acid that has the potential to react with the host rock formation. The induced pores structure modification impacts the flow properties. Thus, to ensure the viability and sustainability of CO2 storage, operators need simulations that take into account the specificities of reactive transport. However, the macroscopic coefficients of the reactive transport equation are modified from the values of an inert tracer by surface reactions. These specificities due to mass transfer are currently not considered. Similarly, the permeability-porosity (K-F) relationship is only estimated semi-empirically. The aim of this thesis was to develop a method to obtain the macroscopic coefficients and the K-F laws, by solving the equations governing the pore-scale phenomena. To do this, we used the Pore Network Modelling approach (PNM). The advantage of the PNM is that it explicitly takes into account the pore structure, while conceptualizing the latter to a set of pores and throats whose morphology is simplified into spheres or cylinders for instance. The study is based into two successive upscalings: from local-scale to pore-scale, then from pore-scale to core-scale. The reactive transport problem is solved for basic elements, analytically or numerically. Then, using the solutions previously found at the pore scale, the reactive transport phenomena are treated throughout the network. Our model was validated by observations on micromodels and by a comparison with an acid-induced alteration experiment

[pt] Estima-se que 50 por cento das reservas mundiais de petróleo e gás sejam mantidas em reservatórios carbonáticos naturalmente fraturados. Um dos maiores desafios neste tipo de formações é a sua heterogeneidade. Além da presença de fraturas que conectam longitudinalmente o meio poroso, vugs em diferentes escalas e distribuições estão espalhados por toda a matriz porosa. Essas cavidades tornam as características do escoamento de fluidos significativamente diferentes daquelas dos reservatórios convencionais de estrutura porosa homogênea, e trazem a necessidade de avaliar propriedades petrofísicas equivalentes para o meio heterogêneo. Neste estudo, uma abordagem microfluídica é usada para determinar as curvas de permeabilidade relativa de água e óleo e os perfis de distribuição das fases em micromodelos 2D de meios porosos vugulares. Experimentos de injeção simultânea de água-óleo em estado estacionário foram realizados a diferentes fluxos fracionários de água, monitorando a dinâmica da queda de pressão e visualizando o deslocamento de fluidos na escala de poros. A aquisição de imagens em tempo real por microscopia de fluorescência permitiu examinar a evolução da saturação das fases. A comparação direta entre as curvas de permeabilidade relativa dos meios porosos vugulares com aquela da matriz porosa mostrou que a incorporação de vugs leva a (i) maior permeabilidade absoluta equivalente, especialmente com cavidades mais longas e em maior número, (ii) aumento da ocupação de óleo na matriz porosa, devido à invasão de água menos eficiente, e (iii) maior permeabilidade relativa à água, que flui preferencialmente pelo espaço vugular. Esses resultados são consistentes com a natureza molhável ao óleo dos micromodelos, uma vez que os vugs oferecem menor resistência capilar ao fluxo da fase não molhante. Nossa abordagem microfluídica de baixo custo provavelmente nos permitirá estudar sistematicamente configurações mais complexas de meios porosos heterogêneos / [en] It is estimated that 50 percent of world s oil and gas reserves are held in naturally fractured carbonate reservoirs. One of the biggest challenges in this type of formation is its heterogeneous nature. Besides the presence of fractures that longitudinally connect the porous medium, vugs at different scales and distributions are scattered throughout the porous matrix. These cavities cause fluid flow characteristics to significantly differ from those of conventional homogeneous pore structure reservoirs and bring the need to evaluate equivalent petrophysical properties of the heterogeneous medium. In this study, a microfluidic approach is used to determine the water and oil relative permeability curves and phase distribution profiles in 2D micromodels of vugular porous media. Steady-state water-oil injection experiments were performed in these devices at different fractional flows, while monitoring the dynamics of the pressure drop and visualizing the fluid displacement at the pore scale. Live-image acquisition through fluorescence microscopy made it possible to examine the evolution of the saturation of water and oil phases. The direct comparison between the relative permeability curves of well-characterized vugular porous media and their porous matrix showed that the incorporation of vugs leads to (i) higher equivalent absolute permeability, especially with longer cavities and higher vug density, (ii) increased oil occupancy in the porous matrix, due to less efficient water invasion into the porous matrix, and (iii) higher relative permeability to water, which flows preferentially through the vugular space. These results are consistent with the oil-wet nature of micromodels, since the vugs are offering less capillary resistance to the flow of the non-wetting phase. Our low-cost microfluidic approach will likely allow us to systematically study more complex vugular-fractured systems.