Este trabalho propõe um esquema de diferenças finitas compacta de alta ordem para resolver problemas elípticos com coeficientes variáveis em malhas composta. São apresentados a formulação matemática e a dedução do método compacto de quarta ordem aplicado à problemas elípticos bidimensionais, em malha regular e composta. Foi adotado o uso da biblioteca PETSc com os seus pré-condicionadores e métodos numéricos para resolver os sistemas lineares resultantes da discretização do problema. Por fim, testes visando verificar o código foram feitos, utilizando o método de soluções manufaturadas, para mostrar alta eficiência e acurácia do método desenvolvido. / This paper proposes a scheme of compact finite difference higher order for solve elliptic problems with variable coeficients in composite meshes. we present the mathematical formulation and the deduction of the compact method of fourth order applied to two-dimensional elliptic problems in regular and composite mesh . It was adopted using the PETSc library with its pre- conditioners and numerical methods for solving linear systems resulting from discretization of the problem. Finally , tests to verify the code were made using the method of manufactured solutions to show high eficiency and accuracy of the method developed .

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-03-02T17:15:29Z No. of bitstreams: 1 patriciacordeiropereirapampanelli.pdf: 10123132 bytes, checksum: 7b722bf4cc93a2714a196242334a5f40 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-03-06T19:57:33Z (GMT) No. of bitstreams: 1 patriciacordeiropereirapampanelli.pdf: 10123132 bytes, checksum: 7b722bf4cc93a2714a196242334a5f40 (MD5) / Made available in DSpace on 2017-03-06T19:57:33Z (GMT). No. of bitstreams: 1 patriciacordeiropereirapampanelli.pdf: 10123132 bytes, checksum: 7b722bf4cc93a2714a196242334a5f40 (MD5) Previous issue date: 2011-02-25 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Inúmeras áreas utilizam elementos finitos para modelagem e simulação de fenômenos físicos, químicos, biológicos, etc. Neste contexto, a geração de malhas consiste na subdivisão do espaço de acordo com o domínio de interesse. Contudo, a diversidade de geometrias e topologias dificulta esta tarefa. Este trabalho apresenta um método de geração de malhas trivalentes a partir de modelos triangulares. A etapa de pré-processamento consiste em extrair características diferenciais do modelo, como as regiões de curvatura máxima e mínima. O mapeamento no espaço rombóide é obtido através de duas otimizações, ambas de programação inteira mista. Na primeira etapa, um campo de direções é gerado com objetivo de interpolar as direções principais sobre toda a superfície. A segunda otimização consiste em mapear o modelo triangular no espaço bidimensional e homeomorfo ao disco. Este mapeamento permite extrair a malha trivalente no espaço rombóide com distribuição mais uniforme dos vértices. A cobertura trivalente obtida neste trabalho pode ser aplicada na representação de nano-estruturas de carbono dada sua geometria predominantemente hexagonal. Estas estruturas são especialmente interessantes devido á sua extensa aplicabilidade. / Several areas use finite element method for modeling and simulating physical, chemical or biological phenomena. In this context, the mesh generation consists of subdividing the space under the domain of interest. However, the diversity of geometries and topologies difficult this task. This work presents a method for trivalent mesh generation from triangular models. A pre-processing step consists in extracting differential characteristics of the model, as the regions of maximum and minimum curvatures. The rhomboid mapping space is obtained through two optimizations, both of mixed integer programming. In the first stage, a direction field is generated to interpolate the principal directions on the entire surface. The second optimization consists in mapping the triangular model in a two-dimensional homeomorphic space to the disk. This mapping allows to extract the trivalent mesh on the rhomboid space with a more uniform distribution of vertices. The trivalent coverage obtained in this study can be applied on the representation of carbon nano-structures due to its predominantly hexagonal geometry. These structures are specially interesting because of its extensive applicability.

Cette thèse étudie, au niveau continu et au niveau discret sur des maillages polyédriques, les équations de transport tridimensionnelles scalaire et vectorielle. Ces équations sont constituées d'un terme diffusif, d'un terme advectif et d'un terme réactif. Dans le cadre des systèmes de Friedrichs, l'analyse mathématique est effectuée dans les espaces du graphe associés aux espaces de Lebesgue. Les conditions de positivité usuelles sur le tenseur de Friedrichs sont étendues au niveau continu et au niveau discret afin de prendre en compte les cas d'intérêt pratique où ce tenseur prend des valeurs nulles ou raisonnablement négatives. Un nouveau schéma convergeant à l'ordre 3/2 est proposé pour le problème d'advection-réaction scalaire en considérant des degrés de liberté scalaires associés aux sommets du maillage. Deux nouveaux schémas considérant également des degrés de libertés aux sommets sont proposés pour le problème de transport scalaire en traitant de manière robuste les différents régimes dominants. Le premier schéma converge à l'ordre 1/2 si les effets advectifs sont dominants et à l'ordre 1 si les effets diffusifs sont dominants. Le second schéma améliore la précision de ce schéma en convergeant à l'ordre 3/2 lorsque les effets advectifs sont dominants. Enfin, un nouveau schéma convergeant à l'ordre 1/2 est obtenu pour le problème d'advection-réaction vectoriel en considérant un seul et unique degré de liberté scalaire sur chaque arête du maillage. La précision et les performances de tous ces schémas sont examinées sur plusieurs cas tests utilisant des maillages polyédriques tridimensionnels / This thesis analyzes, at the continuous and at the discrete level on polyhedral meshes, the scalar and the vector transport problems in three-dimensional domains. These problems are composed of a diffusive term, an advective term, and a reactive term. In the context of Friedrichs systems, the continuous problems are analyzed in Lebesgue graph spaces. The classical positivity assumption on the Friedrichs tensor is generalized so as to consider the case of practical interest where this tensor takes null or slightly negative values. A new scheme converging at the order 3/2 is devised for the scalar advection-reaction problem using scalar degrees of freedom attached to mesh vertices. Two new schemes considering as well scalar degrees of freedom attached to mesh vertices are devised for the scalar transport problem and are robust with respect to the dominant regime. The first scheme converges at the order 1/2 when advection effects are dominant and at the order 1 when diffusion effects are dominant. The second scheme improves the accuracy by converging at the order 3/2 when advection effects are dominant. Finally, a new scheme converging at the order 1/2 is devised for the vector advection-reaction problem considering only one scalar degree of freedom per mesh edge. The accuracy and the efficiency of all these schemes are assessed on various test cases using three-dimensional polyhedral meshes

[pt] Este trabalho aplica três métodos básicos de malhas auto-adaptativas em problemas de mecânica dos fluidos e transmissão de calor. Um método unidimensional é aplicado a problemas descritos pela equação generalizada de Burgers. Um método bidimensional é aplicado ao problema da. cavidade com tampa móvel para números de Reynolds Iguais a 100, 500 e 1000 e ao problema de convecção natural em uma cavidade para números de Rayleigh iguais a 10 elevado a 3,10 elevado a 4,10 elevado a 5 e 10 elevado a 6. Um terceiro método, fundamentado em um sistema de equações elípticas é proposto e aplicado ao problema de convecção natural na cavidade, para análise de desempenho. Procura-se definir, para cada aplicação, critérios como: a) variável indicada para dirigir a adaptação; b) especiacação da intensidade desejada na adaptação e c) justificação para o uso de malhas auto-adaptativas no problema em estudo. Observa-se, nos problemas unidimensionais, que quando o problema tem uma região única de intensos gradientes, a adaptação da malha funciona convenientemente, especialmente na presença de um termo de fonte. No problema da cavidade com, tampa móvel, verifica-se que a função de corrente deve dirigir a adaptação. Para Re igual a 1000 o método empregado consegue melhorar a descrição geral do problema. O problema de convecção natural apresenta melhores resultados para Ra: 10 elevado a 6 e com a adaptação dirigida pelo gradiente da temperatura que está diretamente relacionado ao termo de fonte do problema (forças de empuxo). São consideradas as características geométricas de ortogonalidade e suavidade das malhas adaptadas. Para isso, compara-se malhas obtidas pelos métodos bidimensionais implementados. Observa-se que o método fundamentado no sistema elíptíco consegue concentrar a malha nas regiões de intensos gradientes, sem danos excessivos naquelas características. / [en] The present work is concerned to the study and implementation of self-adaptive grids to tluid mechanics and heat transfer problems. Initially, a ID method is applied to simple problems described by the generalized Burger-s equation, After that, a 2-D method is applied to the wall driven Cavity with Reynolds numbers equal to 100, 500 and 1000 and to the natural convection problem inside cavltles with Rayleigh numbers equal to 10 to the third power, 10 to the fourth power, 10 to the fifth power and 10 to the sixth power. A third method, based on a system of elliptic equations is proposed and applied to this natural convection problem, for error assessmet. For each application, one seeks to define criteria such as: a) most suitable dependent variable to drive the adaptive technique; b) intensity of the adaptive technique and c) need for the use of the technique at all. From the 1-D problems, it may be concluded that in presence of single regions of strong gradients, the technidne works line, specially shenever a source term is present. From the wall driven cavity problem, it appears that the stream function is most suitable to be the driving force for the adaptation. For instance, for Reynolds number on the order of 1000, the used method works well. The natural convectlon problem indicates best results for high Rayleigh numbers, say 10 to the sixth power, and with the use of the temperature gradient that is straight related to the source term, that is, buoyancy in the case. In the present work, geometric characteristics such as ortogonality and smooth are considered. Among the three methods, studied, it seems that the elliptic one, proposed here, is able to concentrate grid points Whenever needed, without severe penalization to those characteristics.

Cette thèse présente l'étude et le développement de méthodes numériques pour la résolution d'équations de transport, en particulier d'une méthode de remapping bidimensionnel dont un avantage important par rapport aux algorithmes existants est la propriété de conservation de la masse. De nombreux cas-tests permettront de comparer ces approches entre elles ainsi qu'à des méthodes de référence. On s'intéressera en particulier aux équations dites de Vlasov-Poisson et du Centre-Guide, qui apparaissent très classiquement dans le cadre de la physique des plasmas. / This PhD thesis presents the study and development of numerical methods for the resolution of transport equations, in particular a bidimensional remapping method whose main advantage over existing algorithms is the property of mass conservation. Numerous test cases are presented in order to compare these approaches with regard to the others and with reference methods. Focus is made on the so-called Vlasov-Poisson and Center-Guide equations, that appear very classically in the domain of plasma physics.

La plupart des simulations interactives ont besoin d'un modèle de détection de collisions. Cette détection nécessite d'une part d'effectuer des requêtes de proximité entre les entités concernées et d'autre part de calculer un comportement à appliquer. Afin d'effectuer ces requêtes, les entités présentes dans une scène sont soit hiérarchisées dans un arbre ou dans un graphe de proximité, soit plongées dans une grille d'enregistrement. Nous présentons un nouveau modèle de détection de collisions s'appuyant sur deux piliers : une représentation de l'environnement par des cartes combinatoires multirésolutions et un suivi en temps réel de particules plongées dans ces cartes. Ce modèle nous permet de représenter des environnements complexes tout en suivant en temps réel les entités évoluant dans cet environnement. Nous présentons des outils d'enregistrement et de maintien de l'enregistrement de particules, d'arêtes et de surfaces dans des cartes combinatoires volumiques multirésolutions. / Most interactive simulations need a collision detection system. First, this system requires the querying of the proximity between the objects and then the computing of the behaviour to be applied. In order to perform these queries, the objects present in a scene are either classified in a tree, in a proximity graph, or embedded inside a registration grid.Our work present a new collision detection model based on two main concepts: representing the environment with a combinatorial multiresolution map, and tracking in real-time particles embedded inside this map. This model allows us to simulate complex environments while following in real-time the entities that are evolving within it.We present our framework used to register and update the registration of particles, edges and surfaces in volumetric combinatorial multiresolution maps. Results have been validated first in 2D with a crowd simulation application and then in 3D, in the medical field, with a percutaneous surgery simulation.

The paper is concerned with the Nitsche mortaring in the framework of domain decomposition where non-matching meshes and weak continuity of the finite element approximation at the interface are admitted. The approach is applied to singularly perturbed reaction-diffusion problems in 2D. Non-matching meshes of triangles being anisotropic in the boundary layers are applied. Some properties as well as error estimates of the Nitsche mortar finite element schemes are proved. In particular, using a suitable degree of anisotropy of triangles in the boundary layers of a rectangle, we derive convergence rates as known for the conforming finite element method in presence of regular solutions. Numerical examples illustrate the approach and the results.

The paper deals with a combination of the Fourier method with the Nitsche-finite-element method (as a mortar method). The approach is applied to the Dirichlet problem of the Poisson equation in threedimensional axisymmetric domains with reentrant edges generating singularities. The approximating Fourier method yields a splitting of the 3D problem into 2D problems on the meridian plane of the given domain. For solving the 2D problems bearing corner singularities, the Nitsche finite-element method with non-matching meshes and mesh grading near reentrant corners is applied. Using the explicit representation of singular functions, the rate of convergence of the Fourier-Nitsche-mortaring is estimated in some $H^1$-like norm as well as in the $L_2$-norm. Finally, some numerical results are presented.

No description available.

[pt] A simulação numérica é uma ferramenta essencial para a engenharia de reservatórios moderna, em particular no desenvolvimento de campos de óleo marítimos. A maioria das simulações de reservatórios utilizam malhas estruturadas em três dimensões, com tamanho variando de alguns milhares a dezenas de milhões de células. Algumas simulações apresentam um alto custo computacional que pode dificultar os estudos de desenvolvimento de um campo, mesmo com a alta capacidade computacional disponível hoje. Malhas de simulação não-estruturadas são uma alternativa para reduzir o tamanho dos modelos de reservatórios (e, consequentemente, o tempo de execução das simulações), sem sacrificar a qualidade dos resultados. Este trabalho utiliza malhas de Voronoi, também conhecidas como malhas de bissetores perpendiculares, uma vez que suas propriedades permitem simplificar as equações discretizadas do escoamento em comparação com outros tipos de malhas não-estruturadas. Dois passos são críticos para a criação de um modelo não-estruturado de reservatórios a partir de um modelo geológico refinado: geração da malha e transferência de escala das propriedades. A maioria dos métodos propostos para ambas as tarefas utilizam informações de simulações na malha refinada. Embora essa abordagem apresente bons resultados, pode ser muito custosa e precisa ser refeita caso haja alterações significativas nas condições de escoamento. Este trabalho discute técnicas para geração de malha e transferência de escala que não dependam de simulações na escala fina. As técnicas utilizam apenas a distribuição de propriedades de reservatórios e o posicionamento de poços, falhas e outras feições discretas. A abordagem adotada para geração da malha parte de uma disposição regular de pontos que são redistribuídos de acordo com um mapa de espaçamento previamente definido. Dois algoritmos iterativos para redistribuição desses pontos baseados em modelos físicos são propostos. Diversos critérios de espaçamento também são investigados. Dois algoritmos de transferência de escala em malhas não-estruturadas são propostos. Estes métodos se baseiam nas técnicas de Cardwell and Parsons e de renormalização para transferência de escala em malhas estruturadas. Por fim, exemplos representativos são utilizados para demonstrar as potencialidades e eficácia das estratégias propostas. / [en] Numerical simulation represents an essential tool for modern reservoir engineering, especially for the development of offshore oil fields. Most reservoir simulations are performed on three-dimensional structured grids, with a size ranging from a few thousands to tens of millions of cells. Some simulations can have a high computational cost that hinders the field development studies, even using the processing power available nowadays. Unstructured meshes are an effective alternative to reduce the size of reservoir models (and, consequently, the overall simulation time) without sacrificing the quality of the results. In this work, we adopt Voronoi meshes, also known as perpendicular bisector grids, since their properties simplify the discretized flow equations in reservoir simulations when compared to other types of unstructured meshes. Two main steps are critical to creating an unstructured reservoir model from a refined geological model: grid generation and upscaling of the reservoir properties. Most methods employed for both steps rely on information obtained from simulations using fine-scale meshes. Although this approach yields good results, it can be time-consuming and may be optimal only for the specified set of flow conditions. This work discusses the generation of unstructured grids and upscaling techniques that do not require any previous simulations. Instead, they are based only on reservoir property distributions and the location of discrete features such as wells and faults. The proposed grid generation strategy starts from a regular set of points and then redistributes them according to a previously defined spacing map. Two iterative redistribution algorithms based on physical models are presented, and several criteria for spacing maps are also investigated. Two upscaling algorithms for unstructured grids are proposed, based on the Cardwell and Parsons and renormalization techniques for structured meshes. Finally, representative examples are presented to demonstrate the capabilities and effectiveness of the proposed strategies.