Les milieux poreux sont omniprésents dans le quotidien des hommes du 21ème siècle que ce soit par hasard ou par nécessité. Certains permettent le transport d’une ou plusieurs phases fluides immiscibles, une situation à la fois courante et critique dans de nombreuses applications industrielles. Le transport d’une phase mouillante, l’imbibition, a été largement étudiée mais un certain nombre de problèmes restent ouverts. Dans cette thèse, nous nous sommes concentrés sur trois d’entre eux. La fragmentation de phase induite par la gravité dans une jonction asymétrique a été étudiée. La dynamique et un critère d’apparition ont été prédits analytiquement puis confirmés expérimentalement. Ensuite, l’évolution de l’épaisseur d’un front d’imbibition dans un réseau de capillaires a été simulée numériquement. Les résultats ont permis de lever une controverse de la littérature. Enfin, l’imbibition dans un milieu constitué de lames flexibles déformables a été explorée numériquement et expérimentalement. Il a été montré que la déformabilité des lames induit leur coalescence et agit sur la dynamique de l'imbibition. / Porous media are ubiquitous in today’s society due to chance or necessity. Some of them allow the transportation of one or several immiscible fluid phases simultaneously. This property is both usual and critical in many industrial processes. The transport of a wetting phase, imbibition, has been widely studied but lots of issues remain opened. In this manuscript, three of them are addressed. Gravity induced phase fragmentation has been studied in an asymmetric pore junction. Fragmentation criterion and dynamic are predicted analytically and confirmed experimentally. Then, the broadening of an imbibition front in a capillary network has been investigated numerically. Results resolved a controversy in the literature. Finally, imbibition in a medium made of flexible sheets has been studied both numerically and experimentally. It showed that sheets flexibility induces their clustering and influences the imbibition dynamic.

A thesis submitted to the Faculty of Engineering and the Built Environment, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Doctor of Philosophy, 2016 / Solar powered gas micro-turbines present opportunities for off-grid power generation. Due to the intermittent nature of the solar energy supply, existing Solar Gas Turbine (SGT) plants employ hybridisation with fossil fuels to generate dispatchable power. In this work sensible heat and latent heat storage solutions are investigated as a means of increasing the solar share of a SGT cycle, thus reducing the consumption of diesel fuel. The sensible heat storage concept was based on a pressurised packed bed of spherical ceramic particles, using air as the heat transfer fluid. An axisymmetric, two-phase heat transfer model of the system was developed, based on the continuous solid phase approach. The model was successfully validated against experimental data from a packed bed of alumino-silicate particles over the temperature ranges of gas turbine cycles (350-900 °C and 600-900 °C). The validated numerical model was utilised to conduct a parametric design study of a six hour (1.55 MWhth) storage system for a gas micro-turbine. The results show that a high storage efficiency and high utilisation factor can be achieved when combining sensible heat storage in alumina with fossil fuel hybridisation, with somewhat lower values without hybridisation. An analysis of different inventory geometries showed that a packed bed of spherical particles is best suited to pressurised sensible heat storage. The latent heat storage concept was based on a pressurised packed bed of Encapsulated Phase Change Material (EPCM) particles. Sodium sulphate was identified as a suitable phase change material for the gas turbine cycle. The sensible heat storage model was extended to account for intra-particle temperature gradients and phase change within the particles. The intra-particle phase change model was validated against published experimental data for a single EPCM sphere heated and cooled by convection. The full EPCM storage model was further successfully validated against experimental data from a packed bed of macro- encapsulated sodium sulphate particles with alumina shells, up to a temperature of 950 °C. A comparison of the two storage concepts for a 7 m3 bed shows that a packed bed of en- capsulated sodium sulphate particles would have a 36% lower energy storage capacity than a bed of solid alumina particles. This is due to the limited melt fraction in the EPCM bed when a temperature limit is placed on the base. Increasing the packed bed volume to 10.5 m3 would provide a comparable thermal performance to the 7 m3 solid alumina bed, at a 12% lower storage mass. A hybrid three-layer packed bed is proposed to increase the volumetric energy storage density. Modelling shows that this concept could provide a small increase of 5.3% in the amount of energy discharged above 850 °C, compared to the solid alumina particles only.

O escoamento bifásico de gás-líquido é encontrado em muitos circuitos fechados que utilizam circulação natural para fins de resfriamento. O fenômeno da circulação natural é importante nos recentes projetos de centrais nucleares para a remoção de calor. O circuito de circulação natural (Circuito de Circulação Natural - CCN), instalado no Instituto de Pesquisas Energéticas e Nucleares, IPEN / CNEN, é um circuito experimento concebido para fornecer dados termo-hidráulicos relacionados com escoamento monofásico ou bifásico em condições de circulação natural. A estimativa de transferência de calor tem sido melhorada com base em modelos que requerem uma previsão precisa de transições de padrão de escoamento. Este trabalho apresenta testes experimentais desenvolvidos no CCN para a visualização dos fenômenos de instabilidade em ciclos de circulação natural básica e classificar os padrões de escoamento bifásico associados aos transientes e instabilidades estáticas de escoamento. As imagens são comparadas e agrupadas utilizando mapas auto-organizáveis de Kohonen (SOM), aplicados em diferentes características da imagem digital. Coeficientes da Transformada Discreta de Cossenos de Quadro Completo (FFDCT) foram utilizados como entrada para a tarefa de classificação, levando a bons resultados. Os protótipos de FFDCT obtidos podem ser associados a cada padrão de escoamento possibilitando uma melhor compreensão da instabilidade observada. Uma metodologia sistemática foi utilizada para verificar a robustez do método. / Two-phase flow of liquid and gas is found in many closed circuits using natural circulation for cooling purposes. Natural circulation phenomenon is important on recent nuclear power plant projects for decay heat removal. The Natural Circulation Facility (Circuito de Circulação Natural CCN ) installed at Instituto de Pesquisas Energéticas e Nucleares, IPEN/CNEN, is an experimental circuit designed to provide thermal hydraulic data related to single and two-phase flow under natural circulation conditions. This periodic flow oscillation behavior can be observed thoroughly in this facility due its glass-made tubes transparency. The heat transfer estimation has been improved based on models that require precise prediction of pattern transitions of flow. This work presents experiments realized at CCN to visualize natural circulation cycles in order to classify two-phase flow patterns associated with phase transients and static instabilities of flow. Images are compared and clustered using Kohonen Self-organizing Maps (SOMs) applied on different digital image features. The Full Frame Discret Cosine Transform (FFDCT) coefficients were used as input for the classification task, enabling good results. FFDCT prototypes obtained can be associated to each flow pattern, enabling a better comprehension of each observed instability. A systematic test methodology was used to verify classifier robustness.

O presente trabalho é um estudo comparativo entre a modelagem de motores de indução bifásicos simétricos com o uso da transformação dq0, que é uma modelagem matemática clássica na análise dos motores de indução bifásicos e a modelagem através da teoria do vetor espiral, proposta na década de 80 para os motores de indução trifásicos. A modelagem clássica é apresentada detalhadamente, bem como a modelagem por vetores espirais. Foram realizadas simulações para comparar os resultados entre as diferentes abordagens. / This work is a comparative study between the classic modeling using dq0 transformation and the spiral vector theory applied to study symmetrical two-phase induction machine. The spiral vector theory was proposed by Dr. Sakae Yamamura to study electrical machines in the middle of 1980s. To compare both types of modeling, simulations were performed to solve differential equations.

O escoamento bifásico de gás-líquido é encontrado em muitos circuitos fechados que utilizam circulação natural para fins de resfriamento. O fenômeno da circulação natural é importante nos recentes projetos de centrais nucleares para a remoção de calor. O circuito de circulação natural (Circuito de Circulação Natural - CCN), instalado no Instituto de Pesquisas Energéticas e Nucleares, IPEN / CNEN, é um circuito experimento concebido para fornecer dados termo-hidráulicos relacionados com escoamento monofásico ou bifásico em condições de circulação natural. A estimativa de transferência de calor tem sido melhorada com base em modelos que requerem uma previsão precisa de transições de padrão de escoamento. Este trabalho apresenta testes experimentais desenvolvidos no CCN para a visualização dos fenômenos de instabilidade em ciclos de circulação natural básica e classificar os padrões de escoamento bifásico associados aos transientes e instabilidades estáticas de escoamento. As imagens são comparadas e agrupadas utilizando mapas auto-organizáveis de Kohonen (SOM), aplicados em diferentes características da imagem digital. Coeficientes da Transformada Discreta de Cossenos de Quadro Completo (FFDCT) foram utilizados como entrada para a tarefa de classificação, levando a bons resultados. Os protótipos de FFDCT obtidos podem ser associados a cada padrão de escoamento possibilitando uma melhor compreensão da instabilidade observada. Uma metodologia sistemática foi utilizada para verificar a robustez do método. / Two-phase flow of liquid and gas is found in many closed circuits using natural circulation for cooling purposes. Natural circulation phenomenon is important on recent nuclear power plant projects for decay heat removal. The Natural Circulation Facility (Circuito de Circulação Natural CCN ) installed at Instituto de Pesquisas Energéticas e Nucleares, IPEN/CNEN, is an experimental circuit designed to provide thermal hydraulic data related to single and two-phase flow under natural circulation conditions. This periodic flow oscillation behavior can be observed thoroughly in this facility due its glass-made tubes transparency. The heat transfer estimation has been improved based on models that require precise prediction of pattern transitions of flow. This work presents experiments realized at CCN to visualize natural circulation cycles in order to classify two-phase flow patterns associated with phase transients and static instabilities of flow. Images are compared and clustered using Kohonen Self-organizing Maps (SOMs) applied on different digital image features. The Full Frame Discret Cosine Transform (FFDCT) coefficients were used as input for the classification task, enabling good results. FFDCT prototypes obtained can be associated to each flow pattern, enabling a better comprehension of each observed instability. A systematic test methodology was used to verify classifier robustness.

Bromelina é um nome coletivo para enzimas proteolíticas encontradas no talo, fruto e folhas do abacaxi (Ananas comosus Merr). A bromelina possui propriedades anti-inflamatórias, de debridamento, entre outras. Para a produção da bromelina deve-se, preferencialmente, usar resíduos do abacaxi, visto que os produtos do fruto têm aplicação comercial. Este trabalho teve como objetivo a extração de bromelina a partir de cascas de abacaxi através de sistema de duas fases aquosas (SDFA), e sua aplicação em hidrogel polimérico. Foram realizados estudos de estabilidade da bromelina comercial, em que se observou maior estabilidade em pH 5,0 com menor perda da atividade relativa em todas as temperaturas estudadas (20, 30, 40 e 50°C). O estudo da extração da bromelina em SDFA formado por polietileno glicol (PEG) e ácido poliacrílico (PAA) (com auxílio da análise de variância de parâmetros como rendimento, fator de purificação e coeficiente de partição) proporcionou rendimento de 335% e fator de purificação de 25,8. Os hidrogéis poliméricos à base de PEG estudados apresentaram-se flexíveis, com pouca elasticidade e taxa de absorção superior a 1000%. Hidrogel carreado de bromelina pelo método de turgescência proporcionou a maior liberação da enzima, assim como a maior atividade (80% da bromelina liberada em 24 h e 278 ± 89 U/mL). / Bromelain is a collective name for the proteolytic enzymes found in the stem, fruit and leaves of pineapple (Ananas comosus Merr.). Bromelain possesses anti-inflammatory properties, debridement, among others. For bromelain production one should preferably use the waste materials, whereas pineapple fruit products have commercial application. This study aimed to extract bromelain from pineapple peels using aqueous two-phase system (ATPS), and its application in polymeric hydrogels. Stability studies of commercial bromelain were performed, which found greater stability at pH 5.0 with minor loss of relative activity at all temperatures studied. The study of bromelain extraction in ATPS composed by polyethylene glycol (PEG) and poly acrylic acid (PAA) (with assistance of variance analysis of parameters such as yield, purification factor and partition coefficient) showed yield 335% and purification factor of 25.8. The PEG-based hydrogels studied presented flexibility, low elasticity and swelling ratio higher than 1000%. Hydrogel containing bromelain, loading by embedding (solvent sorption) method, yielded the highest enzyme release, as well as the highest activity (80% bromelain released over 24 h and 278 ± 89 U / mL).

A partição das enzimas glicose-6-fosfato desidrogenase (G6PD) e uroquinase (UK) em sistemas micelares de duas fases aquosas foi investigada, teórica e experimentalmente. Inicialmente, a partição de G6PD em sistema micelar de duas fases aquosas composto pelo tensoativo não-iônico C10E4 (óxido de n-deciltetraetileno) foi estudada. Observou-se uma partição governada primariamente por interações repulsivas estéricas, do tipo \"volume de exclusão\", sendo a G6PD recuperada preferencialmente na fase inferior do sistema, pobre em micelas, resultando em valores de KG6PD inferiores a 1,0. Os tensoativos catiônicos CnTAB (brometos de alquiltrimetilamônio, n = 8, 10 e 12) foram adicionados ao tensoativo C10E4, de modo a formar sistemas micelares mistos (não-iônico/catiônico) de duas fases aquosas e atrair a G6PD de carga efetiva negativa para a fase superior, rica em micelas positivamente carregadas. Os coeficientes de partição obtidos nos sistemas C10E4/CnTAB/tampão foram no mínimo 2,5 vezes maiores que os correspondentes no sistema C10E4/tampão mantendo-se o volume de exclusão constante. De uma forma geral, o sistema micelar misto C10E4/C10TAB/tampão forneceu o melhor KG6PD = 7,7, com um rendimento de G6PD na fase superior de 71%. A adição de ligantes de afinidade (inibidores enzimáticos) ao sistema de duas fases aquosas C10E4/tampão para purificação de UK e G6PD também foi estudada. Não foram observadas diferenças significativas no perfil de partição da UK na presença ou não do ligante p-aminobenzamidina, o qual apresentou partição uniforme entre as duas fases do sistema, sendo obtidos valores de KUK ~ 0,70 em ambos os casos. Para a G6PD, um pequeno incremento no KG6PD foi observado na presença dos ligantes de afinidade cibacron blue e procion red, porém resultando ainda em valoresfnferiores a 1. Portanto, a utilização de inibidores enzimáticos livres em solução como ligantes de afinidade para purificação de UK e G6PD em sistema micelar C10E4/tampão de duas fases aquosas não resultou em diferenças significativas no perfil de partição das enzimas. Por fim, um sistema de purificação em dois estágios, sendo o primeiro constituído de sistema micelar não-iônico (C10E4) de duas fases aquosas e o segundo de sistema micelar misto (C10E4/ C10TAB) de duas fases aquosas foi aplicado para a purificação de G6PD presente em homogeneizado celular de Saccharomyces cerevisiae, sendo obtido um rendimento final de G6PD de 59%, com um fator de purificação de 1,3. / The partitioning of the hydrophilic enzymes glucose-6-phosphate dehydrogenase (G6PD) and urokinase (UK) in two-phase aqueous micellar systems was investigated, both experimentally and theoretically. Initially, the partitioning of G6PD in two-phase aqueous micellar system composed of the nonionic surfactant C10E4 (n-decyl tetra(ethylene oxide) was studied. It was observed a partitioning behavior governed primarily by repulsive, steric, excluded-volume interactions, which drove the enzyme preferentially to the bottom, micelle-poor phase, resulting in KG6PD values smaller than one. The cationic surfactants CnTAB ((alkyltrimethylammonium bromide, n = 8, 10, and 12) were mixed wíth the nonionic surfactant C10E4 to form two-phase aqueous mixed (nonionic/cationic) micellar systems. The measured G6PD partition coefficients in the C10E4/CnTAB/buffer systems were at least 2.5 times larger than those in the corresponding C10E4/buffer system, clearly demonstrating that the net negatively charged G6PD was attracted electrostatically to the top, mixed micelle-rich phase, containing a greater number of positively-charged C10E4/CnTAB mixed micelles. Overall, the two-phase aqueous mixed (C10E4/C10TAB) micellar system yielded the highest KG6PD = 7.7, with a G6PD yield in the top phase of 71%. The addition of affinity ligands (enzyme inhibitors) to the two-phase C10E4 /buffer systems was studied for UK and G6PD purification. No significant differences in the UK partitioning behavior was observed in the presence or not of the ligand p-aminobenzamidine, since the ligand was found to partition evenly between the two phases of the system. Partition coefficient values of about 0.70 were obtained in both cases. For the G6PD, the partition coefficient was found to be slightly higher in the presence of either cibacron blue or procion red as affinity ligand, however still smaller than one. Therefore, the use of competitive inhibitors free in solution as affinity ligands for UK and G6PD purification in two-phase C10E4 /buffer systems did not provide significant differences in the enzymes partitioning behaviors. Finally, a two-step purification process, combining a two-phase aqueous nonionic (C10E4) micelar system, followed by a two-phase aqueous mixed (C10E4/C10TAB) micellar system, was employed for the purification of G6PD present in a Saccharomyces cerevisiae cell homogenate. It was obtained a G6PD final yield of 59%, and a purification factor of 1,3.

A lithium vaporizer for a high-power magnetoplasmadynamic (MPD) thruster is modeled using a parallel approach. A one-dimensional, thermal-resistive network is developed and used to calculate the required vaporizer length and power as a function of various parameters. After comparing results predicted by this network model with preheat power data for a 200 kW MPD thruster, we investigate performance over a parameter space of interest for the Advanced Lithium-Fed, Applied-field, Lorentz Force Accelerator (ALFA2) thruster. Heater power sensitivity to cathode tube emissivity, mass flow rate, and vapor superheat are presented. The cold-start heater power for 80 mg/s is found to range from 3.38 to 3.60 kW, corresponding to a vaporizer (axial) length of 18 to 26 cm. The strongest drivers of vaporizer performance are cathode tube emissivity and a conduction heat sink through the mounting flange. Also, for the baseline ALFA2 case, it is shown that increasing the vapor superheat from 100 K to 300 K has the effect of lowering the vaporizer thermal efficiency from 57% to 49%. Also, a finite-volume computational fluid dynamic (CFD) is implemented in FLUENT 6.2 which includes conjugated heat transfer to the solid components of the cathode. This model uses a single-fluid mixture model to simulate the effects of the two-phase vaporizer flow with source terms that model the vaporization. This model provides a solution of higher fidelity by including three-dimensional fluid dynamics such as thermal and momentum boundary layers, as well as calculating a higher resolution temperature distribution throughout the cathode assembly. Results from this model are presented for three mass flow rates of interest (40 mg/s, 80 mg/s, and 120 mg/s). Using a fixed power and length taken from the conceptual ALFA2 design, the dryout point ranges from 12.3-17.6 cm from the base of the cathode assembly for 40 mg/s and 80 mg/s, respectively. For the 120 mg/s case, the two-phase flow never reaches dryout. Finally, results two modeling approaches are compare favorably, with a maximum disagreement of 13.0 percent in prediction of the dryout point and 4.2 percent in predictions of the exit temperature.

La phase d’allumage est une composante critique à prendre en compte lors de la conception et du dimensionnement d’une chambre de combustion aéronautique, en particulier lorsque de nouvelles technologies ou architectures sont envisagées dans l’objectif de réduire les émissions de polluants causées par la combustion de carburants d’origine fossile. Il est donc primordial d’atteindre une compréhension détaillée du processus complexe qu’est l’allumage dans des conditions réalistes afin d’être en mesure de choisir les meilleures géométries qui assurent un fonctionnement fiable, stable et sûr des moteurs tout au long de leur cycle de vie. Des simulations aux grandes échelles de l’allumage circulaire d’une chambre de combustion annulaire avec injection de carburant liquide sont réalisés pour trois points de fonctionnement et comparées avec les données expérimentales en termes de structure de flamme et de délai d’allumage. Une analyse détaillée des trois séquences d’allumage numériques permet d’identifier certains aspects clés de la propagation de la flamme dans le mélange froid diphasique. Enfin, les pertes thermiques aux parois sont prises en compte, dans l’objectif d’évaluer la capacité de la simulation à retrouver la forte chute de la vitesse de propagation observée expérimentalement lorsque les parois sont à température ambiante. / Ignition is one of the critical issues that arise in the design and dimensioning of aeronautic combustors, in particular when new technologies are envisioned to reduce the amount of pollutants generated by the combustion of fossil fuels. It is therefore important to achieve a detailed understanding of this complex process in realistic conditions in order to enable informed design choices leading to reliable, stable and safe operation of the engines.Large eddy simulations of the light-round with two phase injection are carried out for three operating conditions and compared to experimental data in terms of flame structure and global duration. The liquid phase is described with a mono-disperse Eulerian approach.A detailed analysis of the three numerical light-round sequences allows to identify some key aspects of the flame propagation in the two-phase mixture. Interactions between the flame, the flow field and the liquid sprays create heterogeneities in the liquid repartition and wakes on the downstream side of the swirling jets formed by the injectors, with notable effects on the motion of the leading point and on the absolute flame velocity.Finally, heat losses at the walls are accounted for during the light-round in order to assess the simulation's ability to retrieve the marked slowdown of the flame propagation observed experimentally when the quartz walls are at ambient temperature.

Cette thèse de doctorat s’intéresse aux écoulements diphasiques métastables typiques de certains transitoires accidentels qui pourraient intervenir dans les centrales nucléaires. Ces phénomènes sont difficiles à traiter en raison de la complexité topologique de l’écoulement, des transferts entre phases et du couplage fort entre les caractéristiques thermodynamiques et les aspects mathématiques.Les méthodes aujourd’hui en usage dans l’industrie ne décrivent pas complétement la complexité de ces écoulements car elles s’appuient sur des modèles trop simples. En fait ces méthodes ne prennent pas en compte le déséquilibre thermo-chimique entre l’eau liquide et sa vapeur. Par ailleurs, les méthodes hyperboliques proposées récemment dans la littérature pour la simulation des écoulements métastables ne peuvent pas être appliquées dans l’industrie car elles utilisent des lois d’état simples qui ne sont pas adaptées pour les calculs industriels.Le but de cette thèse est de développer une nouvelle approche qui couple les méthodes hyperboliques modernes à des équations d’état précises. Le produit final de ce travail est un nouveau modèle pour l’analyse industrielle des écoulements diphasiques métastables qui associe de nouvelles techniques pour le calcul des transferts interfaciaux et des propriétés de l’eau et de sa vapeur. De plus, cette approche est d’un coût abordable pour les configurations industrielles.Les méthodes développées dans cette thèse ont été systématiquement vérifiées avec des solutions exactes et validées en utilisant des données expérimentales de la littérature. / This Ph.D. thesis deals with the metastable two-phase flows typical of accidental transients that could occur in nuclear power plants. Those phenomena are of difficult treatment due to the topological difficulty of the flow, the interphase transfers and the strong coupling between thermodynamic features and mathematical aspects.The methods today in use in industry do not fully describe the complexity of these flows because based on too simple models. In fact, they do not take into account the thermo-chemical disequilibrium between liquid and vapor water. On the other hand, the hyperbolic methods recently proposed in the literature for the simulation of metastable flows can not be used in the industry because based on simple equations of state that are not adequate for industrial calculations.The purpose of this Ph.D. thesis is to develop a new approach that couples the modern hyperbolic methods to accurate equations of state. The final product of this work is a new model for the industrial analysis of metastable two-phase flows that incorporates novel techniques for the calculation of interfacial transfers and of steam-water properties. Moreover, it is computationally affordable for its use in industrial configurations.The methods developed in this thesis have been sistematically verified against exact solutions and validated using experimental data of the literature.