Thesis deals about solution of non-steady flow in hydraulic systems, which have one dominant component of velocity. Such systems can be arbitrarily structured and they are not limited by number of elements. Computation is based on Lax-Wendroff method and enables considering of variable sound peed as function of static pressure and properties of pipe material. It means, that hydraulic system can be very various. Numerical method is described in detail and description is also focused on sensitivity of method for time step and length step. It can be very imported for evaluation of numerical viscosity, which is compared with second viscosity of fluid. Hammer is working title of software, which was developed on the basis of written numerical procedures. This software enables fast computation of flow in pipe-line systems.

Les maladies cardiovasculaires représentent la première cause de mortalité dans le monde. L’hypertension artérielle et l’hypertrophie du ventricule gauche en constituent deux facteurs de risque souvent asymptomatiques. Dans ce contexte, la pression aortique s’impose comme un indicateur de la santé cardiovasculaire reflétant non seulement les propriétés biomécaniques des artères centrales mais aussi la post-charge ventriculaire. La mesure de ce paramètre hémodynamique est donc cruciale pour le diagnostic et la prise en charge thérapeutique. L’objectif de ce travail est d’estimer non-invasivement la pression aortique en couplant l’Imagerie par Résonance Magnétique Cardiovasculaire (IRM-CV) à la modélisation de l’écoulement sanguin. Nous avons ainsi opté pour un modèle biomécanique unidimensionnel (1D) réduit, prenant en compte l’interaction du sang avec la paroi élastique de l’aorte. Ce segment 1D est couplé en sortie à un modèle Windkessel, dit 0D, modélisant l’ensemble du réseau vasculaire en aval de l’aorte. Les paramètres locaux du modèle 1D ainsi que ses conditions aux limites (conditions d'entrée et paramètres du modèle 0D) sont déterminés à partir des données acquises non-invasivement par IRM-CV. Cela permettrait des prédictions 1D spécifiques à chaque sujet/patient. Nous avons évalué la validité du modèle sur des fantômes d’aorte saine et pathologique. Un banc expérimental compatible avec l’IRM a été mis en œuvre; il reproduit au mieux l’écoulement aortique in vivo. Des mesures de pression invasives ont été confrontées à celles prédites par le modèle 1D. Nous avons également évalué la sensibilité du modèle 1D aux paramètres d’entrée. / Cardiovascular (CV) diseases remain the most common cause of death worldwide. Hypertension and left ventricle hypertrophy are two major risk factors associated to such diseases. In this context, aortic blood pressure is considered as a biomarker of increased CV risk and, more generally, a CV health indicator. In fact, it encodes information about biomechanical properties of central elastic arteries and represents left ventricle afterload. Assessment of this hemodynamic parameter is thus crucial for CV disease diagnosis and for evaluating therapeutic benefits. The aim of this work is to non-invasively assess the aortic blood pressure by coupling CV Magnetic Resonance Imaging (CV-MRI) to blood flow modeling. We thus developed a reduced one-dimensional (1D) flow model taking into account the fluid-structure interaction. A Windkessel (or zero-dimensional (0D)) model describing the arterial tree downstream of the aorta was coupled to the 1D segment. Both 1D model parameters and boundary conditions (inlet condition and 0D model parameters) were non-invasively determined using CV-MRI data. This gives the opportunity of deriving subject- or patient-specific blood flow models. To validate our approach, we applied our model to both healthy and pathologic aorta phantoms. Each phantom was mounted in an experimental setup reproducing as well as possible in vivo aortic flow and compatible with the MRI environment. Invasive pressure recording was compared to predicted pressure waves. We also performed a sensitivity analysis of our reduced 1D model.

Materiais enrijecidos são encontrados cotidianamente na engenharia, como na construção de aeronaves, veículos, navios e manufaturas diversas. Na Engenharia Civil, são exemplos de materiais enrijecidos o concreto armado e o solo reforçado. A solução aproximada para esse tipo de problema pode ser encontrada pelo uso de métodos numéricos como o Método dos Elementos Finitos (MEF) e o Método dos Elementos de Contorno (MEC). Particularmente, o MEC é muito vantajoso em casos onde se tem elevados gradientes de tensão, como em problemas da mecânica da fratura, além de modelar de forma natural meios infinitos, sendo muito utilizado na análise de interação solo-estrutura ou de túneis. No entanto, como no MEC somente o contorno é discretizado, torna-se necessária a utilização de um acoplamento numérico para que o efeito mecânico dos enrijecedores seja considerado. Para esse acoplamento, podem ser utilizados diversos métodos numéricos, sendo neste trabalho realizado com o MEF e o MEC em sua forma unidimensional (MEC 1D). Dentro desse contexto, o acoplamento com o MEC 1D destaca-se por vantagens, como a compatibilidade dos métodos e a redução de aproximações. Com isso, este trabalho tem como principal objetivo o desenvolvimento e comparação de soluções numéricas para o problema de domínios enrijecidos, no âmbito do MEC aplicado a problemas bidimensionais. Os enrijecedores foram considerados como elementos de treliça, sendo realizado inicialmente na sua forma mais usual, modelando-os por meio do MEF. A seguir foi implementada nova formulação para o acoplamento, na qual os enrijecedores são modelados por meio do MEC 1D. A implementação do MEC 1D foi validada pela comparação de resultados com soluções analítica e do programa computacional FTOOL. No caso do acoplamento, os resultados de ambas formulações foram comparados com resposta do programa computacional ANSYS. Foram avaliados quatro exemplos, sendo dois isotrópicos e dois anisotrópicos. Foram aplicadas diferentes condições de carregamentos, apoios e materiais, sendo utilizadas diversas discretizações e graus de aproximação nos enrijecedores. Verificou-se os efeitos da mudança do grau de aproximação, mantendo-se constante o número de graus de liberdade. Os resultados obtidos foram similares às respostas do ANSYS, foram mecanicamente equivalentes, não havendo diferença relevante em custo computacional. O acoplamento com o MEC 1D levou a resultados estáveis e, em geral, melhores do que com o MEF. No caso de aproximação quadrática, o acoplamento com o MEF frequentemente levou a resultados instáveis. Com o aumento do grau de aproximação, os resultados de ambos métodos se aproximam, se tornando muito próximos com aproximação do quarto grau. Verificou-se que próximo às pontas dos enrijecedores ocorrem concentrações de tensão, havendo maior influência da discretização e grau de aproximação adotado nestas áreas. / The reinforced materials are commonly encountered in engineering applications, as in the construction of airplanes, vehicles, ships and general manufactures. In Civil Engineering, the reinforced concrete and the reinforced soil are examples of reinforced materials. The solution of this kind of problem can be found by the use of numerical methods as the Finite Element Method (FEM) and the Boundary Element Method (BEM). Particularly, the BEM use in cases with high stress gradients, such as in fracture mechanics problems, is very advantageous. The BEM models infinite domains in a natural way, being largely used in analysis as soil-structure interaction and tunnel modelling. However, as in the BEM only the boundary is discretized, became necessary the use of a numerical coupling to consider the mechanical effect of the stiffeners. For the coupling many numerical methods can be used, in this study it was realized with the FEM and with the BEM in its unidimensional way (BEM 1D). In this context, the coupling BEM/BEM 1D stands out for its advantages, like the compatibility between the methods and the reduction of approximations. Therefore, the main objective of this study is the development and the comparison of numerical solutions for reinforced media problems, considering the framework of the bidimensional BEM problems. The coupling was implemented considering the stiffeners as truss elements, initially it was implemented in the most usual way, by modelling the stiffeners through the FEM. In the following, a new coupling formulation was proposed, in which the stiffeners are modelled through the BEM 1D method. The BEM 1D implementation was validated by comparing the results with analytical and numerical solutions, the last one obtained with the software FTOOL. Both coupling strategies were compared with the solution obtained by the software ANSYS. Four examples were evaluated, with two isotropic and two anisotropic domains. Different boundary conditions, supports and materials were applied, as many discretization and the approximation degree of the stiffeners were tested. The effects of changing the approximation degree in both coupling strategies keeping the degrees of freedom constant were analysed. The results obtained with the implemented algorithms were mechanically coherent, and they were similar to ANSYS results. For all examples, there is no significant computational costs differences between the two coupling strategies. However, the coupling with the BEM 1D leaded to more stable results and better approximations. It was observed that the MEF results were instable for many results, mainly in the quadratic approximations. When the approximation degree rises, the methods tend to converge to equivalent results, becoming very close in fourth degree approximation. Lastly, it was observed stress concentration in the stiffeners ends. In these regions, the discretization and the approximation degree have large influence to the numerical response.

Materiais enrijecidos são encontrados cotidianamente na engenharia, como na construção de aeronaves, veículos, navios e manufaturas diversas. Na Engenharia Civil, são exemplos de materiais enrijecidos o concreto armado e o solo reforçado. A solução aproximada para esse tipo de problema pode ser encontrada pelo uso de métodos numéricos como o Método dos Elementos Finitos (MEF) e o Método dos Elementos de Contorno (MEC). Particularmente, o MEC é muito vantajoso em casos onde se tem elevados gradientes de tensão, como em problemas da mecânica da fratura, além de modelar de forma natural meios infinitos, sendo muito utilizado na análise de interação solo-estrutura ou de túneis. No entanto, como no MEC somente o contorno é discretizado, torna-se necessária a utilização de um acoplamento numérico para que o efeito mecânico dos enrijecedores seja considerado. Para esse acoplamento, podem ser utilizados diversos métodos numéricos, sendo neste trabalho realizado com o MEF e o MEC em sua forma unidimensional (MEC 1D). Dentro desse contexto, o acoplamento com o MEC 1D destaca-se por vantagens, como a compatibilidade dos métodos e a redução de aproximações. Com isso, este trabalho tem como principal objetivo o desenvolvimento e comparação de soluções numéricas para o problema de domínios enrijecidos, no âmbito do MEC aplicado a problemas bidimensionais. Os enrijecedores foram considerados como elementos de treliça, sendo realizado inicialmente na sua forma mais usual, modelando-os por meio do MEF. A seguir foi implementada nova formulação para o acoplamento, na qual os enrijecedores são modelados por meio do MEC 1D. A implementação do MEC 1D foi validada pela comparação de resultados com soluções analítica e do programa computacional FTOOL. No caso do acoplamento, os resultados de ambas formulações foram comparados com resposta do programa computacional ANSYS. Foram avaliados quatro exemplos, sendo dois isotrópicos e dois anisotrópicos. Foram aplicadas diferentes condições de carregamentos, apoios e materiais, sendo utilizadas diversas discretizações e graus de aproximação nos enrijecedores. Verificou-se os efeitos da mudança do grau de aproximação, mantendo-se constante o número de graus de liberdade. Os resultados obtidos foram similares às respostas do ANSYS, foram mecanicamente equivalentes, não havendo diferença relevante em custo computacional. O acoplamento com o MEC 1D levou a resultados estáveis e, em geral, melhores do que com o MEF. No caso de aproximação quadrática, o acoplamento com o MEF frequentemente levou a resultados instáveis. Com o aumento do grau de aproximação, os resultados de ambos métodos se aproximam, se tornando muito próximos com aproximação do quarto grau. Verificou-se que próximo às pontas dos enrijecedores ocorrem concentrações de tensão, havendo maior influência da discretização e grau de aproximação adotado nestas áreas. / The reinforced materials are commonly encountered in engineering applications, as in the construction of airplanes, vehicles, ships and general manufactures. In Civil Engineering, the reinforced concrete and the reinforced soil are examples of reinforced materials. The solution of this kind of problem can be found by the use of numerical methods as the Finite Element Method (FEM) and the Boundary Element Method (BEM). Particularly, the BEM use in cases with high stress gradients, such as in fracture mechanics problems, is very advantageous. The BEM models infinite domains in a natural way, being largely used in analysis as soil-structure interaction and tunnel modelling. However, as in the BEM only the boundary is discretized, became necessary the use of a numerical coupling to consider the mechanical effect of the stiffeners. For the coupling many numerical methods can be used, in this study it was realized with the FEM and with the BEM in its unidimensional way (BEM 1D). In this context, the coupling BEM/BEM 1D stands out for its advantages, like the compatibility between the methods and the reduction of approximations. Therefore, the main objective of this study is the development and the comparison of numerical solutions for reinforced media problems, considering the framework of the bidimensional BEM problems. The coupling was implemented considering the stiffeners as truss elements, initially it was implemented in the most usual way, by modelling the stiffeners through the FEM. In the following, a new coupling formulation was proposed, in which the stiffeners are modelled through the BEM 1D method. The BEM 1D implementation was validated by comparing the results with analytical and numerical solutions, the last one obtained with the software FTOOL. Both coupling strategies were compared with the solution obtained by the software ANSYS. Four examples were evaluated, with two isotropic and two anisotropic domains. Different boundary conditions, supports and materials were applied, as many discretization and the approximation degree of the stiffeners were tested. The effects of changing the approximation degree in both coupling strategies keeping the degrees of freedom constant were analysed. The results obtained with the implemented algorithms were mechanically coherent, and they were similar to ANSYS results. For all examples, there is no significant computational costs differences between the two coupling strategies. However, the coupling with the BEM 1D leaded to more stable results and better approximations. It was observed that the MEF results were instable for many results, mainly in the quadratic approximations. When the approximation degree rises, the methods tend to converge to equivalent results, becoming very close in fourth degree approximation. Lastly, it was observed stress concentration in the stiffeners ends. In these regions, the discretization and the approximation degree have large influence to the numerical response.

The spintronic materials graphene and FeRh are of great scientific and technological interest due to their unique properties. Graphene's remarkable electronic transport and low spin interaction suggest it could be a near-perfect spin-transport material, while the equiatomic alloy FeRh undergoes a first-order antiferromagnetic (AF) to ferromagnetic (FM) phase transition when heated through a critical temperature ~370 K. Combining these materials could lead to a single multifunctional spin injection, transport and detection device in which a range of stimuli - heat, magnetic field, strain etc. - could be used to manipulate the device state. However, realisation of such a multifunctional device is extremely challenging. This thesis describes the progress made in developing a novel method of spin injection into graphene, and details a study of the metamagnetic phase transition in FeRh nanowires suitable for use as spin injection and detection electrodes. The measured values of spin lifetime and spin diffusion length in graphene are an order of magnitude lower than those predicted theoretically. In this project, a novel 1D contact geometry was investigated to determine whether the dwelling of spins underneath tunnel barrier contacts was the cause of the discrepancy. Although these devices exhibited very high charge carrier mobility - indicating successful device fabrication, defect-free graphene flakes and low levels of contamination - no spin signals were observed. Through a thorough investigation of this unexpected result it was determined that the quality of the graphene/- ferromagnetic interface was limiting the polarisation of injected spin current. The use of FeRh as a novel spin injection and detection material was investigated through magnetic force microscopy imaging of the AF and FM phases during heating and cooling sweeps. The results from FeRh full-films showed a strong dependence on surface morphology, as certain surface types were observed to favour the FM phase. These behaviours were confirmed in patterned nanowire devices, which indicated that the dependence on surface topology dominated over spatial confinement effects. In order to perform these studies a magneto-transport measurement system capable of performing measurements over a wide temperature range 2 K - 500 K in a rotatable magnetic field of up to 750 mT was developed. The noise base of the completed system was measured at just 10% above the theoretical minimum level.

The development of a high-resolution coupled one-dimensional/two-dimensional hydrodynamic model of Charles City, Iowa is presented in this study as part of a larger Iowa Flood Center initiative to create a library of steady inundation maps for communities in Iowa which have a high risk of flooding. Channel geometry from bathymetric surveys and surface topography from LiDAR were combined to create the digital elevation model (DEM) used in numerical simulations. Coupled one- and two-dimensional models were used to simulate flood events; the river channel and structures were modeled one-dimensionally, and the floodplain was modeled two-dimensionally. Spatially distributed roughness parameters were estimated using the 2001 National Land Cover Dataset. Simulations were performed at a number of mesh resolutions, and the results were used to investigate the effectiveness of re-sampling simulation results using higher- resolution DEMs. The effect of removing buildings from the computational mesh was also investigated. During 2011, the stream channel geometry is being changed as part of a recreational park in downtown Charles City. After incorporating the planned changes to the stream channel, the model was used to create a library of steady inundation maps which are available on the Iowa Flood Center website.

The study of benthic community structure and protein expression pattern of four benthic species in effluent areas of sewage treatment plants in Kaohsiung, Taiwan Dwi Candra Pratiwi Advisor: Dr. Li-Lian Liu Institute of Marine Biology, National Sun Yat-sen University, Kaohsiung, Taiwan, R.O.C. Abstract In order to gain better understanding on the impacts of sewage treatment plant (STP) effluent on marine environment, the present study was undertaken to investigate the structure of benthic community and protein expression patterns of 4 benthic species in STP effluent areas in Kaohsiung, Taiwan. Six sites were selected, 3 near Jhong Jhou STP, 2 near NSYSU STP and one reference site. Samples were collected in December 2009 (winter) and July 2010 (summer) by bottom trawling. In winter, temperature was 26.8oC; salinity and pH were 32. 2‰ and 8.1. In summer, environmental data were 33.7oC, 31.5‰ and 8.0, respectively. In all, 5 phyla, 31 families and 48 different species were captured in two seasons. Proteomic approach by one-dimensional sodium dodecyl sulfate polyacrylamide gel electrophoresis (1D SDS-PAGE) was conducted on 3 fishes, Leiognathus splendens, Apogon fasciatus, Engyprosopon multisquama, and one crab Portunus hastatoides. Environment characteristics and structure of benthic community had no significant difference among the 6 sampling sites. Protein expression patterns based on the four examined species also indicated no significant difference among sites. The complexity of pollutants in Kaohsiung Harbor is probably the main reason affecting all sampling sites. In other words, contaminated area in Kaohsiung Harbor may cover a wide range, at least III over the present sampling distance, i.e. 8km which causes no difference in community structure and protein expression pattern among control, Jhong Jhou STP and NSYSU STP sites. Key words: 1D SDS-PAGE, Leiognathus splendens, Apogon fasciatus, Engyprosopon multisquama, Portunus hastatoides

abstract: There has been a surge in two-dimensional (2D) materials field since the discovery of graphene in 2004. Recently, a new class of layered atomically thin materials that exhibit in-plane structural anisotropy, such as black phosphorous, transition metal trichalcogenides and rhenium dichalcogenides (ReS2), have attracted great attention. The reduced symmetry in these novel 2D materials gives rise to highly anisotropic physical properties that enable unique applications in next-gen electronics and optoelectronics. For example, higher carrier mobility along one preferential crystal direction for anisotropic field effect transistors and anisotropic photon absorption for polarization-sensitive photodetectors. This dissertation endeavors to address two key challenges towards practical application of anisotropic materials. One is the scalable production of high quality 2D anisotropic thin films, and the other is the controllability over anisotropy present in synthesized crystals. The investigation is focused primarily on rhenium disulfide because of its chemical similarity to conventional 2D transition metal dichalcogenides and yet anisotropic nature. Carefully designed vapor phase deposition has been demonstrated effective for batch synthesis of high quality ReS2 monolayer. Heteroepitaxial growth proves to be a feasible route for controlling anisotropic directions. Scanning/transmission electron microscopy and angle-resolved Raman spectroscopy have been extensively applied to reveal the structure-property relationship in synthesized 2D anisotropic layers and their heterostructures. / Dissertation/Thesis / Doctoral Dissertation Materials Science and Engineering 2018

Submitted by Fernanda Rodrigues de Lima (fernanda.rlima@ufpe.br) on 2018-07-27T22:51:25Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Pedro Paulo Dantas de Souza Paiva.pdf: 5292653 bytes, checksum: 177d836a13787596c993b64d01ede96a (MD5) / Approved for entry into archive by Alice Araujo (alice.caraujo@ufpe.br) on 2018-08-08T17:09:39Z (GMT) No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Pedro Paulo Dantas de Souza Paiva.pdf: 5292653 bytes, checksum: 177d836a13787596c993b64d01ede96a (MD5) / Made available in DSpace on 2018-08-08T17:09:39Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Pedro Paulo Dantas de Souza Paiva.pdf: 5292653 bytes, checksum: 177d836a13787596c993b64d01ede96a (MD5) Previous issue date: 2017-05-26 / O VHTR é um reator nuclear térmico, moderado a grafite e refrigerado por hélio. Para seu desenvolvimento, há a necessidade de utilização de ferramentas computacionais eficientes para a análise de aspectos de modelagem, operação e segurança. A proposta deste trabalho é estudar o comportamento do VHTR por meio de análise paramétrica, alterando-se modelo de turbulência, perfil de geração de energia nos blocos combustíveis e a influência de modificações na própria geometria. Busca-se também avaliar a implementação de uma metodologia simplificada que reduza o esforço computacional e a duração de uma simulação. Procedeu-se à análise do escoamento do fluido refrigerante através dos canais refrigerantes e canais by-pass em uma seção de 1/12 de uma coluna de blocos combustíveis, utilizando-se diferentes modelos de turbulência. Os resultados obtidos com essas simulações foram comparados àqueles obtidos por meio de correlações do número de Nusselt descritos na literatura. Observou-se que a simulação na qual se utiliza o modelo -ε possibilita a obtenção de resultados que convergem bem com aqueles fornecidos pelas correlações, para ambos os tipos de canais. O modelo -ω proporciona bons resultados para os canais refrigerantes e, o SSG, para o canal by-pass. Utilizou-se geometria contendo canais by-pass de diferentes dimensões, além de uma que possuía apenas os canais refrigerantes, sem canal by-pass. Verificou-se que a existência de um escoamento by-pass induz a um aumento no gradiente de temperatura no bloco combustível. Realizaram-se estudos comparativos entre os resultados obtidos em simulações realizadas com diferentes perfis de geração de energia térmica (uniforme e senoidal) nos canais combustíveis. Verificou-se que, quando há a mesma geração de energia térmica total no bloco combustível, a máxima temperatura constatada em cada um dos materiais é menor para o caso da geração de energia com perfil senoidal. Quando utilizado, no perfil senoidal, um fator radial de pico (1,25), há um aumento considerável na temperatura de todos os materiais, possibilitando a ocorrência de regiões em que a temperatura pode ultrapassar o limite usualmente aceito para o combustível do reator (1250°C) em operação normal. O canal refrigerante localizado no centro do bloco combustível tem diâmetro inferior aos demais canais existentes nesse bloco. Para verificar a hipótese de que a existência de um gradiente de temperatura no bloco combustível, com a temperatura mais elevada ao centro e a temperatura mais baixa estando na periferia desse bloco, deve-se fortemente à menor dimensão desse canal central, realizaram-se simulações computacionais utilizando-se uma geometria com canal central de diâmetro igual ao dos demais. A condição de entrada escolhida para essa nova estrutura foi, primeiramente, o mesmo fluxo mássico total e, depois, a mesma diferença de pressão entre entrada e saída verificados na simulação da geometria padrão. Os resultados obtidos confirmam a hipótese aventada. Realizou-se simulação utilizando uma metodologia simplificada, que consiste em uma análise unidimensional do fluido em um canal refrigerante acoplada à análise tridimensional da condução do calor no grafite e nos canais combustíveis. Os resultados obtidos com a metodologia simplificada apresentaram excelente convergência com aqueles obtidos com a simulação completamente tridimensional, e em um tempo de simulação cerca de 45 vezes menor. / The VHTR is a thermal, graphite moderated and helium cooled reactor. For its development, there is a need to use efficient computational tools to analyze aspects of modeling, operation and safety. The purpose of this work is to study the behavior of the VHTR by means of parametric analysis, altering the turbulence model, the energy generation profile in the fuel blocks and the influence of modifications in the geometry itself. It also seeks to evaluate the implementation of a simplified methodology that reduces the computational effort and duration of a simulation. The coolant flow through the coolant channels and by-pass channels were analyzed in a 1/12 section of a fuel block column using different turbulence models. The results obtained with these simulations were compared to those obtained by means of Nusselt number correlations described in the literature. It was observed that the simulation using the -ε model makes it possible to obtain results that converge well with those provided by the correlations, for both types of channels. The -ω model provides good results for the coolant channels and the SSG for the by-pass channel. Geometry was used with by-pass channels of different dimensions, besides one that had only the cooling channels, without by-pass channel. It has been found that the existence of a by-pass flow induces an increase in the temperature gradient in the fuel block. Comparative studies were performed between the results obtained in simulations carried out with different profiles of thermal energy generation (uniform and sinusoidal) in the fuel channels. It was verified that when there is the same total thermal energy generation in the fuel block, the maximum temperature observed in each of the materials is smaller for the generation of energy with sinusoidal profile. When a peak radial factor (1,25) is used in the sinusoidal profile, there is a considerable increase in the temperature of all materials, allowing the occurrence of regions where the temperature may exceed the limit usually accepted for reactor fuel (1250° C) in normal operation. The coolant channel located in the center of the fuel block has a smaller diameter than the other channels in this block. To verify the hypothesis that the existence of a temperature gradient in the fuel block, with the highest temperature at the center and the lowest temperature being at the periphery of this block, is due to the smaller dimension of this central channel, computer simulations were performed using a geometry with a central channel with the same diameter as the others. The input condition chosen for this new structure was, firstly, the same total mass flow and then the same pressure difference between input and output found in the simulation of the standard geometry. The results confirm the hypothesis. Simulation was performed using a simplified methodology, which consists of a one-dimensional analysis of the coolant flow in a coolant channel coupled to the three-dimensional analysis of heat conduction within the graphite and fuel channels. The results obtained with the simplified methodology were in excellent agreement with those obtained with the fully three-dimensional simulation and the time required to complete the simulation with the simplified methodology was about 45 times less than the time that the fully three-dimensional simulation lasted.

Ce travail a été réalisé dans le cadre d’une convention Cifre entre Snecma et le LMFA. L’objectif de ce travail est de développer une méthodologie permettant un calcul Navier Stokes 3D instationnaire du champ aérodynamique dans le compresseur au cours du pompage profond, tout en ayant un temps raisonnable de restitution des résultats. A l’aide de ce champ 3D, une réponse vibratoire des aubes pendant le cycle de pompage pourra être calculée. Le pompage est une instabilité aérodynamique qui s’initie dans le compresseur mais qui est influencé, à travers les pertes induites, par tout le système dans lequel le compresseur est intégré. La méthodologie développée pendant cette thèse permet de modéliser l’intégralité du système (banc d’essai ou moteur). Pour cela, l’écoulement dans le compresseur est calculé en 3D alors que le reste du système est calculé en 1D. Un couplage 1D/3D a donc été mis en place, puis testé sur un cas test académique. La méthode de couplage est ensuite appliquée au calcul du pompage d’un compresseur axial. Le cas d’application turbomachine choisi se base sur le banc d’essai CREATE installé au LMFA et contenant un compresseur haute vitesse de 3.5 étages. Dans un premier temps, la méthode est appliquée sur le dernier étage de CREATE seul. Le pompage profond obtenu sur ce compresseur mono-étage fait l’objet d’une analyse globale puis locale. Il ressort de cette analyse que le cycle de pompage est guidé par la dynamique du volume situé à l’aval du compresseur. Le volume de tranquilisation à l’amont du compresseur induit sur le compresseur une oscillation de pression et de débit à une fréquence environ dix fois plus élevée que la fréquence propre du pompage. L’analyse locale montre que le pompage se déclenche par une progression de l’écoulement de jeu vers les bords d’attaque du rotor. A l’aide du champ de pression statique sur les aubages pendant le pompage, un calcul de structure est mené sur le rotor et le stator. Une réponse vibratoire visible est obtenue après la chute de débit seulement si la fréquence propre du premier mode de vibration est suffisamment basse. Dans une seconde étape, la méthode de couplage est appliquée au pompage des 3.5 étages du compresseur CREATE et son banc d’essai. Le cycle obtenu est comparé aux résultats expérimentaux et à un autre calcul de pompage qui modélise l’intégralité du banc d’essai par les équations de Navier-Stokes 3D. La physique du cycle obtenu est cohérente avec les résultats expérimentaux et numériques pour les phases d’entrée en pompage, d’inversion et de rétablissement du débit. La dernière phase de repressurisation du système est beaucoup plus courte, ce qui en tenant compte des conclusions tirées de la première étape, indique que les cycles comparés correspondent probablement à des positions (angle de fermeture) de vanne aval différentes. / This work is a result of collaboration between Snecma and the LMFA. The purpose of this work is to develop a model able to simulate surge occurrence and development in compressors. This model must provide an unsteady 3D aerodynamic field inside the compressor and must have reasonable computation costs. With this 3D field, it will be possible to compute a vibratory response of blades during surge cycle. Surge is influenced by all the system around the compressor due to pressure losses in the system. With the methodology developed in this study, the whole system may be computed (test bench or engine). For that, flow in the compressor is computed in 3D, while the rest of system is computed with 1D model. A 1D/3D coupling strategy has been set up and validated on an academic test case. Coupling method is then applied for surge modeling of an axial compressor. The turbomachiny case is based on CREATE test bench, installed at LMFA. This bench contains a high speed, compressor of 3.5 stages. In a first step, method is applied only to the last stage of CREATE. A global and local analysis focuses on the deep surge obtained on this one stage compressor. Surge cycle is driven by downstream plenum aerodynamic. Pressure and mass flow oscillation with a frequency ten times higher than surge frequency is due to upstream plenum. Local analysis shows that surge occurence is due to progression of tip clearance flow toward rotor leading edges. With the help of static pressure field on blades during surge, structural computations are done on rotor and stator. A significant vibratory response is obtained after mass flow drop only if the first vibratory mode has a frequency low enough. In a second step, 1D/3D coupling is applied to surge modeling of 3.5 stages of compressor CREATE. Surge cycle obtained is compared to experimental data and to another computation which model the whole test bench with 3D Navier-Stokes equations. Physics of our surge cycle is coherent with these results but our cycle is faster and deeper.