Poli(hidroxibutirato) (PHB) é um polímero com um alto potencial de aplicação industrial, pois é biodegradável, tem propriedades físicas semelhantes ao poli(propileno) e é o mais abundante de sua classe, os poli(hidroxialconoatos). Porém, sua rigidez e fragilidade, combinadas com seu elevado custo de produção, têm limitado sua utilização em larga escala. Nesse sentido o PHB pode ser utilizado para o desenvolvimento de novos materiais como blendas, que são mais viáveis em comparação aos métodos de síntese. Estudos de blendas PHB/PVA por solução reportam boa compatibilidade entre os componentes, porém esse processo apresenta baixa produção, já que ampliação de escala é dificultada. Neste trabalho, blendas de PHB/PVA nas composições de 90:10, 75:25 e 50:50 em massa e plastificadas com glicerol foram obtidas por meio de mistura mecânica, utilizando uma extrusora rosca simples. As técnicas de espectroscopia no infravermelho por transformada de Fourier (FTIR), calorimetria exploratória diferencial (DSC), análise térmica dinâmico-mecânica (DMTA) e microscopia eletrônica de varredura (MEV) foram empregadas para estudar a miscibilidade das blendas. Os principais resultados indicam que as blendas são miscíveis: i) deslocamentos de bandas dos espectros de FTIR das blendas nas regiões de 1277 e 1054 cm-1 indicando interações intermoleculares fortes no sistema, como por exemplo polares e ligações de hidrogênio, ii) Tg única e intermediária às Tgs dos polímeros puros observada nas curvas do DMTA para as blendas, e iii) a ausência de duas fases nas imagens do MEV realizadas para as blendas com 90% de PHB após extração em meio aquoso da possível fase rica em PVA. Também foi possível observar que a cristalinidade do PHB não foi afetada significativamente pela adição de PVA como mostrado pelos resultados de DSC e difratometria de raios X (DRX). / Poly(hydroxybutyrate) (PHB) is a polymer with high potential for industrial applications, because it is biodegradable, has similar physical properties to poly (propylene) and is the most abundant in its class, the poly(hydroxyalconoates). However, its rigidity and fragility, combined with its high production cost, have limited its use in large-scale. Thus, PHB could be used to develop new materials as blends, which are more feasible compared to synthesis methods. Studies of PHB/PVA blends via solution shows good compatibility between the components, nevertheless, this process presents low production, since scaling up is hard to do. In this work blends of PHB/PVA in the compositions (w/w) 90/10, 75/25 and 50/50, and plasticized with glycerol were prepared by mechanical mixing using a single screw extruder. Fourier transform infrared (FTIR) spectroscopy, differential scanning calorimetry (DSC), dynamic-mechanical thermal analysis (DMTA) and scanning electron microscope (SEM) were used to study the miscibility of the blends. The main results indicate that the blends are miscible: i) shifted bands in FTIR spectrum of the blends in 1277 and 1054 cm-1 indicating strong intermolecular interactions in the system, such as polar and hydrogen bonds; ii) single Tg for the blends and intermediate between those of the pure component polymers observed on the DMTA curves; and iii) absence of two phases in the SEM images of blend with 90% PHB after extraction in aqueous medium of the possible PVA-rich phase. It was also possible to observe that the PHB crystallinity was not significantly affected by the addition of PVA as shown in the results of DSC and x-ray diffraction.

Poli(hidroxibutirato) (PHB) é um polímero com um alto potencial de aplicação industrial, pois é biodegradável, tem propriedades físicas semelhantes ao poli(propileno) e é o mais abundante de sua classe, os poli(hidroxialconoatos). Porém, sua rigidez e fragilidade, combinadas com seu elevado custo de produção, têm limitado sua utilização em larga escala. Nesse sentido o PHB pode ser utilizado para o desenvolvimento de novos materiais como blendas, que são mais viáveis em comparação aos métodos de síntese. Estudos de blendas PHB/PVA por solução reportam boa compatibilidade entre os componentes, porém esse processo apresenta baixa produção, já que ampliação de escala é dificultada. Neste trabalho, blendas de PHB/PVA nas composições de 90:10, 75:25 e 50:50 em massa e plastificadas com glicerol foram obtidas por meio de mistura mecânica, utilizando uma extrusora rosca simples. As técnicas de espectroscopia no infravermelho por transformada de Fourier (FTIR), calorimetria exploratória diferencial (DSC), análise térmica dinâmico-mecânica (DMTA) e microscopia eletrônica de varredura (MEV) foram empregadas para estudar a miscibilidade das blendas. Os principais resultados indicam que as blendas são miscíveis: i) deslocamentos de bandas dos espectros de FTIR das blendas nas regiões de 1277 e 1054 cm-1 indicando interações intermoleculares fortes no sistema, como por exemplo polares e ligações de hidrogênio, ii) Tg única e intermediária às Tgs dos polímeros puros observada nas curvas do DMTA para as blendas, e iii) a ausência de duas fases nas imagens do MEV realizadas para as blendas com 90% de PHB após extração em meio aquoso da possível fase rica em PVA. Também foi possível observar que a cristalinidade do PHB não foi afetada significativamente pela adição de PVA como mostrado pelos resultados de DSC e difratometria de raios X (DRX). / Poly(hydroxybutyrate) (PHB) is a polymer with high potential for industrial applications, because it is biodegradable, has similar physical properties to poly (propylene) and is the most abundant in its class, the poly(hydroxyalconoates). However, its rigidity and fragility, combined with its high production cost, have limited its use in large-scale. Thus, PHB could be used to develop new materials as blends, which are more feasible compared to synthesis methods. Studies of PHB/PVA blends via solution shows good compatibility between the components, nevertheless, this process presents low production, since scaling up is hard to do. In this work blends of PHB/PVA in the compositions (w/w) 90/10, 75/25 and 50/50, and plasticized with glycerol were prepared by mechanical mixing using a single screw extruder. Fourier transform infrared (FTIR) spectroscopy, differential scanning calorimetry (DSC), dynamic-mechanical thermal analysis (DMTA) and scanning electron microscope (SEM) were used to study the miscibility of the blends. The main results indicate that the blends are miscible: i) shifted bands in FTIR spectrum of the blends in 1277 and 1054 cm-1 indicating strong intermolecular interactions in the system, such as polar and hydrogen bonds; ii) single Tg for the blends and intermediate between those of the pure component polymers observed on the DMTA curves; and iii) absence of two phases in the SEM images of blend with 90% PHB after extraction in aqueous medium of the possible PVA-rich phase. It was also possible to observe that the PHB crystallinity was not significantly affected by the addition of PVA as shown in the results of DSC and x-ray diffraction.

Yes / Positron emission particle tracking (PEPT) was used to visualise the flow patterns established by mixing in two laboratory-scale anaerobic digesters fitted with mechanical mixing or gas mixing apparatus. PEPT allows the visualisation of flow patterns within a digester without necessitating the use of a transparent synthetic sludge. In the case of the mechanically-mixed digester, the mixing characteristics of opaque sewage sludge was compared to a transparent synthetic sludge at different mixing speeds. In the gas-mixed apparatus, two synthetic sludges were compared. In all scenarios, quasi-toroidal flow paths were established. However, mixing was less successful in more viscous liquids unless mixing power was increased to compensate for the increase in viscosity. The robustness of the PEPT derived velocities was found to be significantly affected by the frequency with which the particle enters a given volume of the vessel, with the accuracy of the calculated velocity decreasing in regions with low data capture. Nevertheless, PEPT was found to offer a means of accurate validation of computational fluid dynamics models which in turn can help to optimise flow patterns for biogas production. / The first author was funded via an EPSRC CASE award in conjunction with Severn Trent Water. The second author was funded via a University of Birmingham Postgraduate Teaching Assistantship award.

京都大学 / 新制・課程博士 / 博士(理学) / 甲第23713号 / 理博第4803号 / 新制||理||1687(附属図書館) / 京都大学大学院理学研究科地球惑星科学専攻 / (主査)准教授 河上 哲生, 教授 田上 高広, 教授 下林 典正 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DFAM

The UASB (Upflow Anaerobic Sludge Blanket) reactor, developed by Prof. Gatze Lettinga and his colleagues in the Netherlands more than thirty years ago, is successfully employed in many sewage treatment plants in countries with tropical and subtropical climate. However, are still reported serious operational issues related to the three-phase separators and problematic aspects involving the complexity of its influent distribution system. The focus of this research was to simplify the introduction and dispersion of the sewage in the reactor as well as enhancing the sludge retention, by replacing the characteristic piping system for a mechanical mixing apparatus and substituting the (conventional) three-phase separators by a high-rate settler, respectively. In essence, it was proposed a system partitioned into two distinct regions: one for reaction (i.e. anaerobic digestion of organic matter), in which the sludge blanket was mechanically mixed through the action of a rotating impeller, and another for the removal of suspended solids by the use of a lamella plate clarifier; a sludge recirculation system was also implanted, to pump the settled biomass back to the digestion compartment. The research was subdivided into three main phases: in the first two phases, bench-scale experiments were performed in order to obtain basic data for the design of the mechanical mixing and sedimentation devices, while the third phase consisted of the application of these technologies in a pilot-scale system with total volume of 20.8 m3 (comprised of a 17.0 m3 digestion chamber and a 3.8 m3 compartment for solids removal), as well as its performance assessment. During the development of the research, domestic sewage was applied to feed the experimental system, which was located at the School of Engineering of São Carlos (Campus I) in São Carlos (São Paulo State, Brazil). Throughout the period of effective operation, which lasted 130 days, four distinct operating modes were tested (characterized by intermittent mixing and varying hydraulic loading rates, ranging between 27.6 and 66.5 m3.d-1) and the system reached maximum removal efficiencies of BOD, COD and TSS at rates of 82%, 72% and 83%, respectively. The high-rate settler presented an impressive efficiency removing suspended solids (rates up to 98%). The research showed that the new \"Anaerobic Activated Sludge Blanket\" system (A2SB) can provide satisfactory performance even operating without the three-phase separator and influent distribution system. / O reator UASB, desenvolvido pelo Prof. Gatze Lettinga e colaboradores na Holanda há mais de trinta anos, é utilizado com sucesso em muitas estações de tratamento de esgoto sanitário em países de clima tropical e subtropical. Entretanto, ainda são relatadas sérias questões operacionais relacionadas aos separadores trifásicos e aspectos problemáticos envolvendo a complexidade de seu sistema de distribuição do esgoto afluente. O foco da presente pesquisa foi o de simplificar a distribuição e dispersão do esgoto no reator, assim como aprimorar a retenção de lodo, mediante substituição do sistema convencional de tubos por misturador mecânico e do (tradicional) separador trifásico por decantador de alta taxa, respectivamente. Em essência, propôs-se reator dividido em duas regiões distintas: uma para reação (i.e. digestão anaeróbia da matéria orgânica), na qual a manta de lodo foi agitada mecanicamente através do movimento rotacional de um impelidor, e outra, para remoção de sólidos suspensos mediante uso de decantador com placas paralelas; um sistema de recirculação do lodo também foi implantado, para bombear a biomassa sedimentada até o compartimento de digestão. A pesquisa foi subdividida em três fases: nas duas primeiras, experimentos em escala de bancada foram realizados visando obtenção de dados básicos para dimensionamento dos sistemas de agitação e de decantação, enquanto que a terceira fase consistiu na aplicação dessas tecnologias em sistema de tratamento, em escala piloto, com volume útil total de 20,8 m3 (composto por uma câmara de digestão de 17,0 m3 e um compartimento de 3,8 m3 para remoção de sólidos), assim como na análise de seu desempenho. Durante a pesquisa, esgoto sanitário foi utilizado para alimentar o sistema experimental, instalado na Escola de Engenharia de São Carlos. Ao longo do período de operação, que durou 130 dias, quatro condições operacionais foram testadas (caracterizadas pela agitação intermitente e diferentes taxas de carregamento hidráulico, que variaram entre 27,6 e 66,5 m3.d-1) e o sistema alcançou eficiências de remoção máxima para DBO, DQO e SST de 82%, 72% e 83%, respectivamente. O decantador de alta taxa apresentou surpreendente eficiência de remoção de sólidos suspensos (taxas de até 98%). A pesquisa demonstrou que o novo sistema A2SB (Anaerobic Activated Sludge Blanket) pode apresentar desempenhos satisfatórios, mesmo operando sem separador trifásico e sistema de distribuição do esgoto afluente.

Chemical modification is an important way to improve the properties of existing polymers, and one of the important examples is the hydrogenation of nitrile butadiene rubber (NBR) in organic solvent by homogeneous catalysis in order to extend its application. This process has been industrialized for many years to provide high performance elastomers (HNBR) for the automotive industry, especially those used to produce components in engine compartments. In the current commercial process, a batch reactor is employed for the hydrogenation step, which is labor intensive and not suitable for large volume of production. Thus, novel hydrogenation devices such as a continuous process are being developed in our research group to overcome these drawbacks. In order to make the process more practical for industrial application, high concentration polymer solutions should be targeted for the continuous hydrogenation. However, many problems are encountered due to the viscosity of the high concentration polymer solution, which increases tremendously as the reaction goes on, resulting in severe mass transfer and heat transfer problems. So, hydrogenation kinetics in high concentration NBR solution, as well as the rheological properties of this viscous solution are very essential and fundamental for the design of novel hydrogenation processes and reactor scale up. In the present work, hydrogenation of NBR in high concentration solution was carried out in a batch reactor. A commercial rhodium catalyst, Wilkinson’s catalyst, was used with triphenylphosphine as the co-catalyst and chlorobenzene as the solvent. The reactor was modified and a PID controller was tuned to fit this strong exothermic reaction. It was observed that when NBR solution is in a high concentration the kinetic behavior was greatly affected by mass transfer processes, especially the gas-liquid mass transfer. Reactor internals were designed and various agitators were investigated to improve the mechanical mixing. Experimental results show that the turbine-anchor combined agitator could provide superior mixing for this viscous reaction system. The kinetic behavior of NBR hydrogenation under low catalyst concentration was also studied. It was observed that the hydrogenation degree of the polymer could not reach 95% if less than 0.1%wt catalyst (based on polymer mass) was used, deviating from the behavior under a normal catalyst concentration. The viscosity of the NBR-MCB solutions was measured in a rotational rheometer that has a cylinder sensor under both room conditions and reaction conditions. Parameters that might affect the viscosity of the solutions were studied, especially the hydrogenation degree of polymer. Rheological properties of NBR-MEK solutions, as well as NBR melts were also studied for relevant information. It is concluded that the hydrogenation kinetics deviates from that reported by Parent et al. [6] when polymer is in high concentration and/or catalyst is in low concentration; and that the reaction solution (HNBR/NBR-MCB solution) deviates from Newtonian behavior when polymer concentration and hydrogenation degree are high.

Chemical modification is an important way to improve the properties of existing polymers, and one of the important examples is the hydrogenation of nitrile butadiene rubber (NBR) in organic solvent by homogeneous catalysis in order to extend its application. This process has been industrialized for many years to provide high performance elastomers (HNBR) for the automotive industry, especially those used to produce components in engine compartments. In the current commercial process, a batch reactor is employed for the hydrogenation step, which is labor intensive and not suitable for large volume of production. Thus, novel hydrogenation devices such as a continuous process are being developed in our research group to overcome these drawbacks. In order to make the process more practical for industrial application, high concentration polymer solutions should be targeted for the continuous hydrogenation. However, many problems are encountered due to the viscosity of the high concentration polymer solution, which increases tremendously as the reaction goes on, resulting in severe mass transfer and heat transfer problems. So, hydrogenation kinetics in high concentration NBR solution, as well as the rheological properties of this viscous solution are very essential and fundamental for the design of novel hydrogenation processes and reactor scale up. In the present work, hydrogenation of NBR in high concentration solution was carried out in a batch reactor. A commercial rhodium catalyst, Wilkinson’s catalyst, was used with triphenylphosphine as the co-catalyst and chlorobenzene as the solvent. The reactor was modified and a PID controller was tuned to fit this strong exothermic reaction. It was observed that when NBR solution is in a high concentration the kinetic behavior was greatly affected by mass transfer processes, especially the gas-liquid mass transfer. Reactor internals were designed and various agitators were investigated to improve the mechanical mixing. Experimental results show that the turbine-anchor combined agitator could provide superior mixing for this viscous reaction system. The kinetic behavior of NBR hydrogenation under low catalyst concentration was also studied. It was observed that the hydrogenation degree of the polymer could not reach 95% if less than 0.1%wt catalyst (based on polymer mass) was used, deviating from the behavior under a normal catalyst concentration. The viscosity of the NBR-MCB solutions was measured in a rotational rheometer that has a cylinder sensor under both room conditions and reaction conditions. Parameters that might affect the viscosity of the solutions were studied, especially the hydrogenation degree of polymer. Rheological properties of NBR-MEK solutions, as well as NBR melts were also studied for relevant information. It is concluded that the hydrogenation kinetics deviates from that reported by Parent et al. [6] when polymer is in high concentration and/or catalyst is in low concentration; and that the reaction solution (HNBR/NBR-MCB solution) deviates from Newtonian behavior when polymer concentration and hydrogenation degree are high.

The UASB (Upflow Anaerobic Sludge Blanket) reactor, developed by Prof. Gatze Lettinga and his colleagues in the Netherlands more than thirty years ago, is successfully employed in many sewage treatment plants in countries with tropical and subtropical climate. However, are still reported serious operational issues related to the three-phase separators and problematic aspects involving the complexity of its influent distribution system. The focus of this research was to simplify the introduction and dispersion of the sewage in the reactor as well as enhancing the sludge retention, by replacing the characteristic piping system for a mechanical mixing apparatus and substituting the (conventional) three-phase separators by a high-rate settler, respectively. In essence, it was proposed a system partitioned into two distinct regions: one for reaction (i.e. anaerobic digestion of organic matter), in which the sludge blanket was mechanically mixed through the action of a rotating impeller, and another for the removal of suspended solids by the use of a lamella plate clarifier; a sludge recirculation system was also implanted, to pump the settled biomass back to the digestion compartment. The research was subdivided into three main phases: in the first two phases, bench-scale experiments were performed in order to obtain basic data for the design of the mechanical mixing and sedimentation devices, while the third phase consisted of the application of these technologies in a pilot-scale system with total volume of 20.8 m3 (comprised of a 17.0 m3 digestion chamber and a 3.8 m3 compartment for solids removal), as well as its performance assessment. During the development of the research, domestic sewage was applied to feed the experimental system, which was located at the School of Engineering of São Carlos (Campus I) in São Carlos (São Paulo State, Brazil). Throughout the period of effective operation, which lasted 130 days, four distinct operating modes were tested (characterized by intermittent mixing and varying hydraulic loading rates, ranging between 27.6 and 66.5 m3.d-1) and the system reached maximum removal efficiencies of BOD, COD and TSS at rates of 82%, 72% and 83%, respectively. The high-rate settler presented an impressive efficiency removing suspended solids (rates up to 98%). The research showed that the new \"Anaerobic Activated Sludge Blanket\" system (A2SB) can provide satisfactory performance even operating without the three-phase separator and influent distribution system. / O reator UASB, desenvolvido pelo Prof. Gatze Lettinga e colaboradores na Holanda há mais de trinta anos, é utilizado com sucesso em muitas estações de tratamento de esgoto sanitário em países de clima tropical e subtropical. Entretanto, ainda são relatadas sérias questões operacionais relacionadas aos separadores trifásicos e aspectos problemáticos envolvendo a complexidade de seu sistema de distribuição do esgoto afluente. O foco da presente pesquisa foi o de simplificar a distribuição e dispersão do esgoto no reator, assim como aprimorar a retenção de lodo, mediante substituição do sistema convencional de tubos por misturador mecânico e do (tradicional) separador trifásico por decantador de alta taxa, respectivamente. Em essência, propôs-se reator dividido em duas regiões distintas: uma para reação (i.e. digestão anaeróbia da matéria orgânica), na qual a manta de lodo foi agitada mecanicamente através do movimento rotacional de um impelidor, e outra, para remoção de sólidos suspensos mediante uso de decantador com placas paralelas; um sistema de recirculação do lodo também foi implantado, para bombear a biomassa sedimentada até o compartimento de digestão. A pesquisa foi subdividida em três fases: nas duas primeiras, experimentos em escala de bancada foram realizados visando obtenção de dados básicos para dimensionamento dos sistemas de agitação e de decantação, enquanto que a terceira fase consistiu na aplicação dessas tecnologias em sistema de tratamento, em escala piloto, com volume útil total de 20,8 m3 (composto por uma câmara de digestão de 17,0 m3 e um compartimento de 3,8 m3 para remoção de sólidos), assim como na análise de seu desempenho. Durante a pesquisa, esgoto sanitário foi utilizado para alimentar o sistema experimental, instalado na Escola de Engenharia de São Carlos. Ao longo do período de operação, que durou 130 dias, quatro condições operacionais foram testadas (caracterizadas pela agitação intermitente e diferentes taxas de carregamento hidráulico, que variaram entre 27,6 e 66,5 m3.d-1) e o sistema alcançou eficiências de remoção máxima para DBO, DQO e SST de 82%, 72% e 83%, respectivamente. O decantador de alta taxa apresentou surpreendente eficiência de remoção de sólidos suspensos (taxas de até 98%). A pesquisa demonstrou que o novo sistema A2SB (Anaerobic Activated Sludge Blanket) pode apresentar desempenhos satisfatórios, mesmo operando sem separador trifásico e sistema de distribuição do esgoto afluente.

With the rise in global energy prices, as well as energy consumption being the largest source of greenhouse gas emissions, biofilm-based systems utilized for wastewater treatment, such as moving bed biofilm reactors (MBBRs), have grown in popularity for their lower energy consumption compared to conventional activated sludge processes. However, this technology requires large amounts of energy to constantly distribute and suspend its biofilm carrier within the reactor by either aerators or mechanical mixers. Many studies have been done on optimizing the aeration systems, but limited research has been focused on the mechanical mixing systems. This master’s thesis project aims to narrow the research and data gaps in MBBR mechanical mixing by conducting scaled-down experiments to study the influence of different mixer configurations on carrier’s flow fields in a reactor tank. The main objective is to determine the conditions for good carrier mixing and their energy use efficiency. Other objectives include determining the effects of scaling on carrier flow fields and whether the experimental results can be used to help develop and validate MBBR computational fluid dynamics (CFD) models. The results showed that good carrier mixing occurred in conditions where (1) mixer height was 3 cm from the bottom of the tank, had no inclinations and positioned along a long wall; (2) mixer flows could maintain their momentum; (3) sufficient length was given for mixer jet streams to develop and widen; (4) there was a dual presence of vertical flow loops and horizontal bulk flow loops; (5) 2 mixers did not result in counter- rotating flows; and (6) 2 mixers were not positioned in one corner of the tank. The effects of scaling did not impact the carrier flow fields and was determine by comparing the experimental results from this study with the results from a previous experiment that utilized a smaller tank. The results from this study could qualitatively match with the results of the CFD model. Limitations that occurred during the study when trying to define “good mixing” were also discussed. Lastly, the thesis ends with stating future work and recommendations. / Med stigande globala energipriser och en energiförbrukning som är den största källan till utsläpp av växthusgaser, har biofilmbaserade system för avloppsvattenrening, t.ex. biofilmsreaktorer med rörlig bädd (MBBR), ökat i popularitet tack vare sin lägre energiförbrukning jämfört med konventionella processer med aktivt slam. Denna teknik kräver dock stora mängder energi för att ständigt distribuera och suspendera biofilmbäraren i reaktorn med hjälp av antingen luftare eller mekaniska blandare. Många studier har gjorts för att optimera luftningssystemen, men begränsad forskning har fokuserats på de mekaniska blandningssystemen. Detta examensarbete syftar till att minska forsknings- och dataluckorna inom mekanisk blandning i MBBR genom att genomföra nedskalade experiment för att studera hur olika blandarkonfigurationer påverkar bärarens flödesfält i en reaktortank.Huvudsyftet är att fastställa villkoren för god blandning av bärare och deras energianvändningseffektivitet. Andra mål är att fastställa effekterna av skalning på bärarnas flödesfält och om de experimentella resultaten kan användas för att utveckla och validera CFD-modeller (Computational Fluid Dynamics) för MBBR. Resultaten visade att god bärarblandning inträffade under förhållanden där (1) blandarhöjden var 3 cm från tankens botten, utan lutningar och placerad längs en lång vägg; (2) blandarflöden kunde behålla sitt momentum; (3) tillräcklig längd gavs för blandarjetströmmar att utvecklas och breddas; (4) det fanns en dubbel närvaro av vertikala flödesslingor och horisontella bulkflödesslingor; (5) 2 blandare inte resulterade i motroterande flöden; och (6) 2 blandare inte placerades i ett hörn av tanken. Effekterna av skalning påverkade inte bärarens flödesfält och fastställdes genom att jämföra de experimentella resultaten från denna studie med resultaten från ett tidigare experiment som använde en mindre tank. Resultaten från denna studie kunde kvalitativt matchas med resultaten från CFD-modellen. Begränsningar som uppstod under studien när man försökte definiera "bra blandning" diskuterades också. Slutligen avslutas avhandlingen med att ange framtida arbete och rekommendationer.

Cette étude s’intéresse à la modélisation de la ségrégation des impuretés lors des procédés de solidification dirigée du silicium pour l’industrie photovoltaïque. Il s’agit d’un problème multi-physiques et multi-échelles qui nécessite des modèles efficaces pour pouvoir traiter des configurations industrielles en 3D. La première partie de l’étude porte sur le développement de fonctions de paroi solutales dans le but d’estimer la ségrégation sans résoudre numériquement la couche limite solutale. Un modèle analytique est alors utilisé pour estimer le paramètre convecto-diffusif à partir de la contrainte de frottement à l’interface solide/liquide. L’approche proposée consiste à coupler une simulation hydrodynamique de la convection dans la phase liquide et un calcul analytique de la ségrégation. Cette démarche est validée sur un cas de référence en 2D. Le modèle développé fournit une estimation pertinente de la concentration dans un lingot solidifié avec une vitesse imposée, y compris sur un maillage ne résolvant pas la couche limite solutale. La deuxième partie de l’étude concerne l’utilisation d’un système de brassage mécanique dans le but de favoriser la ségrégation des impuretés. Des simulations transitoires de brassage sont réalisées avec le logiciel FLUENT. Les résultats sont comparés à des mesures de champ de vitesse par PIV effectuées sur une maquette en eau, afin de valider le modèle hydrodynamique. La simulation de brassage est ensuite couplée à un calcul de ségrégation en régime quasi-permanent qui permet d’analyser l’influence de l’écoulement sur la couche limite solutale. Dans la dernière partie de l’étude, une simulation de solidification dirigée 3D instationnaire en régime de convection forcée, intégrant la thermique du four, est réalisée. Un modèle empirique de forces volumiques est alors proposé pour décrire l’écoulement lié au brassage dans le silicium liquide. Une première tentative d’estimation des ségrégations par l’approche analytique est ensuite mise en œuvre. / The present study focuses on solute segregation during photovoltaic silicon directional solidification. This multi-physics problem involves various spatial and temporal scales. The numerical simulation of this process requires efficient models, especially for 3D industrial configurations. In the first part of the study, solute wall functions are derived from a scaling analysis in order to estimate the segregations without numerical resolution of the solute boundary layer. The method is based on the coupling of an hydrodynamic simulation of convection in the liquid phase and an analytical segregation computation. The developed analytical model provides an estimation of the convecto-diffusive parameter from the wall shear-stress at the solid/liquid interface. A reference case in 2D with imposed solidification rate is used for validation purposes. The developed model provides a meaningful estimation of concentration fields in the ingots. In a second part, we focus on segregation optimization by a mechanical stirrer. Transient stirring simulations, using a sliding mesh technique, are achieved with FLUENT commercial software. Results are compared to PIV velocity field measurements performed on an experimental setup using water. A segregation computation in a quasi-steady regime is then implemented in the stirring simulation. The effect of the stirring parameters are directly observed on the solute boundary layer at the solid/liquid interface. In a third part, a transient solidification simulation, including furnace thermal conditions, is performed in a 3D configuration with forced convection. The flow generated by the impeller is described thanks to an empirical model based on body forces. A first attempt is finally made to retrieve segregations in the ingot with the developed analytical method.