<p>The Discontinuous Galerkin time domain (DGTD) method is promising in modeling of realistic multiscale electromagnetic systems. This method defines the basic concept for implementing the communication between multiple domains with different scales.</p><p>Constructing a DGTD system consists of several careful choices: (a) governing equations; (b) element shape and corresponding basis functions for the spatial discretization of each subdomain; (c) numerical fluxes onto interfaces to bond all subdomains together; and (d) time stepping scheme based on properties of a discretized</p><p>system. This work present the advances in each one of these steps.</p><p> </p><p>First, a unified framework based on the theory of differential forms and the finite element method is used to analyze the discretization of the Maxwell's equations. Based on this study, field intensities (<bold>E</bold> and <bold>H</bold>) are associated to 1-forms and curl-conforming basis functions; flux densities (<bold>D</bold> and <bold>B</bold>) are associated to 2-forms and divergence-conforming basis functions; and the constitutive relations are defined by Hodge operators.</p><p>A different approach is the study of numerical dispersion. Semidiscrete analysis is the traditional method, but for high order elements modal analysis is prefered. From these analyses, we conclude that a correct discretization of fields belonging to different p-form (e.g., <bold>E</bold> and <bold>B</bold>) uses basis functions with same order of interpolation; however, different order of interpolation must be used if two fields belong to the same p-form (e.g., <bold>E</bold> and <bold>H</bold>). An alternative method to evaluate numerical dispersion based on evaluation of dispersive Hodge operators is also presented. Both dispersion analyses are equivalent and reveal same fundamental results. Eigenvalues, eigenvector and transient results are studied to verify accuracy and computational costs of different schemes. </p><p>Two different approaches are used for implementing the DG Method. The first is based on <bold>E</bold> and <bold>H</bold> fields, which use curl-conforming basis functions with different order of interpolation. In this case, the Riemman solver shows the best performance to treat interfaces between subdomains. A new spectral prismatic element, useful for modeling of layer structures, is also implemented for this approach. Furthermore, a new efficient and very accurate time integration method for sequential subdomains is implemented.</p><p>The second approach for solving multidomain cases is based on <bold>E</bold> and <bold>B</bold> fields, which use curl- and divergence-conforming basis functions, respectively, with same order of interpolation. In this way, higher accuracy and lower memory consumption are obtained with respect to the first approach based on <bold>E</bold> and <bold>H</bold> fields. The centered flux is used to treat interfaces with non-conforming meshes, and both explicit Runge-Kutta method and implicit Crank-Nicholson method are implemented for time integration. </p><p>Numerical examples and realistic cases are presented to verify that the proposed methods are non-spurious and efficient DGTD schemes.</p> / Dissertation

Neste trabalho desenvolveu-se e implementou-se um modelo computacional paralelo multifísica para a simulação do transporte de substâncias e do escoamento hidrodinâmico, bidimensional (2D) e tridimensional (3D), em corpos de água. Sua motivação está centrada no fato de que as margens e zonas costeiras de rios, lagos, estuários, mares e oceanos são locais de aglomerações de seres humanos, dada a sua importância para as atividades econômica, de transporte e de lazer, causando desequilíbrios a esses ecossistemas. Esse fato impulsiona o desenvolvimento de pesquisas relativas a esta temática. Portanto, o objetivo deste trabalho é o de construir um modelo computacional com alta qualidade numérica, que possibilite simular os comportamentos da hidrodinâmica e do transporte escalar de substâncias em corpos de água com complexa configuração geométrica, visando a contribuir para seu manejo racional. Visto que a ênfase nessa tese são os aspectos numéricos e computacionais dos algoritmos, analisaram-se as características e propriedades numérico-computacionais que as soluções devem contemplar, tais como a estabilidade, a monotonicidade, a positividade e a conservação da massa. As estratégias de soluções enfocam os termos advectivos e difusivos, horizontais e verticais, da equação do transporte. Desse modo, a advecção horizontal é resolvida empregando o método da limitação dos fluxos de Sweby, e o transporte vertical (advecção e difusão) é resolvido com os métodos beta de Gross e de Crank-Nicolson. São empregadas malhas com distintas resoluções para a solução do problema multifísica. O esquema numérico resultante é semi-implícito, computacionalmente eficiente, estável e fornece acurácia espacial e temporal de segunda ordem. Os sistemas de equações resultantes da discretização, em diferenças finitas, das equações do escoamento e do transporte 3D, são de grande porte, lineares, esparsos e simétricos definidos-positivos (SDP). No caso 2D os sistemas são lineares, mas os sistemas de equações para a equação do transporte não são simétricos. Assim, para a solução de sistemas de equações SDP e dos sistemas não simétricos empregam-se, respectivamente, os métodos do subespaço de Krylov do gradiente conjugado e do resíduo mínimo generalizado. No caso da solução dos sistemas 3-diagonal, utiliza-se o algoritmo de Thomas e o algoritmo de Cholesky. A solução paralela foi obtida sob duas abordagens. A decomposição ou particionamento de dados, onde as operações e os dados são distribuídos entre os processos disponíveis e são resolvidos em paralelo. E, a decomposição de domínio, onde obtém-se a solução do problema global combinando as soluções de subproblemas locais. Em particular, emprega-se neste trabalho, o método de decomposição de domínio aditivo de Schwarz, como método de solução, e como pré-condicionador. Para maximizar a relação computação/comunicação, visto que a eficiência computacional da solução paralela depende diretamente do balanceamento de carga e da minimização da comunicação entre os processos, empregou-se algoritmos de particionamento de grafos para obter localmente os subproblemas, ou as partes dos dados. O modelo computacional paralelo resultante mostrou-se computacionalmente eficiente e com alta qualidade numérica. / A multi-physics parallel computational model was developed and implemented for the simulation of substance transport and for the two-dimensional (2D) and threedimensional (3D) hydrodynamic flow in water bodies. The motivation for this work is focused in the fact that the margins and coastal zones of rivers, lakes, estuaries, seas and oceans are places of human agglomeration, because of their importance for economic, transport, and leisure activities causing ecosystem disequilibrium. This fact stimulates the researches related to this topic. Therefore, the goal of this work is to build a computational model of high numerical quality, that allows the simulation of hydrodynamics and of scalar transport of substances behavior in water bodies of complex configuration, aiming at their rational management. Since the focuses of this thesis are the numerical and computational aspects of the algorithms, the main numerical-computational characteristics and properties that the solutions need to fulfill were analyzed. That is: stability, monotonicity, positivity and mass conservation. Solution strategies focus on advective and diffusive terms, horizontal and vertical terms of the transport equation. In this way, horizontal advection is solved using Sweby’s flow limiting method; and the vertical transport (advection and diffusion) is solved with Gross and Crank-Nicolson’s beta methods. Meshes of different resolutions are employed in the solution of the multi-physics problem. The resulting numerical scheme is semi-implicit, computationally efficient, stable and provides second order accuracy in space and in time. The equation systems resulting of the discretization, in finite differences, of the flow and 3D transport are of large scale, linear, sparse and symmetric positive definite (SPD). In the 2D case, the systems are linear, but the equation systems for the transport equation are not symmetric. Therefore, for the solution of SPD equation systems and of the non-symmetric systems we employ, respectively, the methods of Krylov’s sub-space of the conjugate gradient and of the generalized minimum residue. In the case of the solution of 3-diagonal systems, Thomas algorithm and Cholesky algorithm are used. The parallel solution was obtained through two approaches. In data decomposition or partitioning, operation and data are distributed among the processes available and are solved in parallel. In domain decomposition the solution of the global problem is obtained combining the solutions of the local sub-problems. In particular, in this work, Schwarz additive domain decomposition method is used as solution method and as preconditioner. In order to maximize the computation/communication relation, since the computational efficiency of the parallel solution depends directly of the load balancing and of the minimization of the communication between processes, graph-partitioning algorithms were used to obtain the sub-problems or part of the data locally. The resulting parallel computational model is computationally efficient and of high numerical quality.

Les méthodes en simulation numérique dans le domaine de l’ingénierie pétrolière nécessitent la résolution de systèmes linéaires creux de grande taille et non structurés. La performance des méthodes itératives utilisées pour résoudre ces systèmes représente un enjeu majeur afin de permettre de tester de nombreux scénario.Dans ces travaux, nous présentons une manière d'implémenter des méthodes itératives parallèles au dessus d’un support exécutif à base de tâches. Afin de simplifier le développement des méthodes tout en gardant un contrôle fin sur la gestion du parallélisme, nous avons proposé une API permettant d’exprimer implicitement les dépendances entre tâches : la sémantique de l'API reste séquentielle et le parallélisme est implicite.Nous avons étendu le support exécutif HARTS pour enregistrer une trace d'exécution afin de mieux exploiter les architectures NUMA, tout comme de prendre en compte un placement des tâches et des données calculé au niveau de l’API. Nous avons porté et évalué l'API sur les processeurs many-coeurs KNL en considérant les différents types de mémoires de l’architecture. Cela nous a amené à optimiser le calcul du SpMV qui limite la performance de nos applications.L'ensemble de ce travail a été évalué sur des méthodes itératives et en particulier l’une de type décomposition de domaine. Nous montrons alors la pertinence de notre API, qui nous permet d’atteindre de très bon niveaux de performances aussi bien sur architecture multi-coeurs que many-coeurs. / Numerical methods in reservoir engineering simulations lead to the resolution of unstructured, large and sparse linear systems. The performances of iterative methods employed in simulator to solve these systems are crucial in order to consider many more scenarios.In this work, we present a way to implement efficient parallel iterative methods on top of a task-based runtime system. It enables to simplify the development of methods while keeping control on parallelism management. We propose a linear algebra API which aims to implicitly express task dependencies: the semantic is sequential while the parallelism is implicit.We have extended the HARTS runtime system to monitor executions to better exploit NUMA architectures. Moreover, we implement a scheduling policy which exploits data locality for task placement. We have extended the API for KNL many-core systems while considering the various memory banks available. This work has led to the optimization of the SpMV kernel, one of the most time consuming operation in iterative methods.This work has been evaluated on iterative methods, and particularly on one method coming from domain decomposition. Hence, we demonstrate that the API enables to reach good performances on both multi-core and many-core architectures.

TEIXEIRA, Daniel Nascimento. Uma técnica de decomposição a priori para geração paralela de malhas bidimensionais. 2014. 94 f. : Dissertação (mestrado) - Universidade Federal do Ceará, Centro de Ciências, Departamento de Computação, Fortaleza-CE, 2014. / Submitted by guaracy araujo (guaraa3355@gmail.com) on 2016-06-15T19:57:36Z No. of bitstreams: 1 2014_dis_dnteixeira.pdf: 17919971 bytes, checksum: 092ad12b33cf64a31552e6a839a5a5bc (MD5) / Approved for entry into archive by guaracy araujo (guaraa3355@gmail.com) on 2016-06-15T19:58:41Z (GMT) No. of bitstreams: 1 2014_dis_dnteixeira.pdf: 17919971 bytes, checksum: 092ad12b33cf64a31552e6a839a5a5bc (MD5) / Made available in DSpace on 2016-06-15T19:58:41Z (GMT). No. of bitstreams: 1 2014_dis_dnteixeira.pdf: 17919971 bytes, checksum: 092ad12b33cf64a31552e6a839a5a5bc (MD5) Previous issue date: 2014 / This work describes a technique of two-dimensional domain decomposition for parallel mesh generation. This technique works for both distributed and shared memory and has the freedom to use any data structure that manages rectangular regions parallel to the axes to decompose the domain given as input, such as a quaternary tree (quadtree) or a binary space decomposition (bsp), for example. Any process of mesh generation that respects the prerequisites established can be used in the subdomains created, for instance, Delaunay or Advancing Front, among others. This technique is called a priori because the mesh on the interface of the subdomains is generated prior to the their internal meshes. The load estimation for each sub-domain in this work is performed with the aid of a refined quadtree, whose level of refinement guides the creation of edges that are defined from the bounderies of only inner cells. This way of estimate load produces results that accurately represent the number of elements to be generated in each subdomain. That contributes to a good partitioning of the domain, making the mesh generation in parallel be significantly faster than the serial generation. Furthermore, the quality of the generated mesh in parallel is qualitatively equivalent to that generated serially within acceptable limits. / Este trabalho descreve uma técnica de decomposição de domínios bidimensionais para geração em paralelo de malhas. Esta técnica funciona tanto para memória distribuída quanto compartilhada, além de permitir que se utilize qualquer estrutura de dados que gere regiões quadrangulares paralelas aos eixos para decompor o domínio dado como entrada. Pode se utilizar por exemplo, uma árvore quaternária (quadtree) ou uma partição binária do espaço (bsp). Além disso, qualquer processo de geração de malha que respeite os pré-requisitos estabelecidos pode ser empregado nos subdomínios criados, como as técnicas de Delaunay ou Avanço de Fronteira, dentre outras. A técnica proposta é dita a priori porque a malha de interface entre os subdomínios é gerada antes das suas malhas internas. A estimativa de carga de processamento associada a cada subdomínio é feita nesse trabalho com a ajuda de uma quadtree refinada, cujo nível de refinamento orienta a criação das arestas que são definidas a partir da discretização das fronteiras das células internas. Essa maneira de estimar carga produz resultados que representam, com boa precisão, o número de elementos a serem gerados em cada subdomínio. Isso contribui para um bom particionamento do domínio, fazendo com que a geração de malha em paralelo seja significativamente mais rápida do que a geração serial. Além disso, a qualidade da malha gerada em paralelo é qualitativamente equivalente àquela gerada serialmente, dentro de limites aceitáveis.

Le projet ANR ICARE, dans lequel s'inscrit cette thèse, vise au développement de méthodes pour l'analyse de structures complexes et de grande taille. Le défi scientifique consiste à investiguer des zones très localisées, mais potentiellement critiques vis-à-vis de la tenue mécanique d'ensemble. Classiquement, sont mis en œuvre aux échelles globale et locale des représentations, discrétisations, modèles de comportement et outils numériques adaptés à des besoins de simulation gradués en complexité. Le problème global est traité avec un code généraliste dans le cadre d'idéalisations topologiques (formulation plaque, simplification géométrique) et comportementale (homogénéisation) ; l'analyse locale quant à elle demande la mise en œuvre d'outils spécialisés (routines, codes dédiés) pour une représentation fidèle de la géométrie et du comportement.L'objectif de cette thèse consiste à développer un outil efficace de couplage non-intrusif pour la simulation multi-échelles / multi-modèles en calcul de structures. Les contraintes de non-intrusivité se traduisent par la non modification de l'opérateur de rigidité, de la connectivité et du solveur du modèle global, ce qui permet de travailler dans un environnement logiciel fermé. Dans un premier temps, on propose une étude détaillée de l'algorithme de couplage global/local non-intrusif. Sur la base d'exemples et de cas-test représentatifs en calcul de structures (fissuration, plasticité, contact...), on démontre l'efficacité et la flexibilité d'un tel couplage. Aussi, une analyse comparative de plusieurs outils d'optimisation de l'algorithme est menée, et le cas de patchs multiples en interaction est traité. Ensuite le concept de couplage non-intrusif est étendu au cas de non-linéarités globales, et une méthode de calcul parallèle par décomposition de domaine avec relocalisation non-linéaire est développée. Cette méthode nous a permis de paralléliser un code industriel séquentiel sur un mésocentre de calcul intensif. Enfin, on applique la méthode de couplage au raffinement de maillage par patchs d'éléments finis. On propose un estimateur d'erreur en résidu explicite adapté au calcul de solutions multi-échelles via l'algorithme de couplage. Puis, sur la base de cet estimateur, on met en œuvre une procédure non-intrusive de raffinement local de maillage. Au travers de ces travaux, un outil logiciel de couplage non-intrusif a été mis au point, basé sur l'échange de données entre différents codes de calcul (protocole Message Passing Interface). Les développements effectués sont intégrés dans une surcouche Python, dont le rôle est de coupler plusieurs instances de Code_Aster, le code d'analyse de structures développé par EDF R&D, lequel sera utilisé dans l'ensemble des travaux présentés. / This PhD thesis, part of the ANR ICARE project, aims at developing methods for complex analysis of large scale structures. The scientific challenge is to investigate very localised areas, but potentially critical as of mechanical systems resilience. Classically, representation models, discretizations, mechanical behaviour models and numerical tools are used at both global and local scales for simulation needs of graduated complexity. Global problem is handled by a generic code with topology (plate formulation, geometric approximation...) and behaviour (homogenization) simplifications while local analysis needs implementation of specialized tools (routines, dedicated codes) for an accurate representation of the geometry and behaviour. The main goal of this thesis is to develop an efficient non-intrusive coupling tool for multi-scale and multi-model structural analysis. Constraints of non-intrusiveness result in the non-modification of the stiffness operator, connectivity and the global model solver, allowing to work in a closed source software environment. First, we provide a detailed study of global/local non-intrusive coupling algorithm. Making use of several relevant examples (cracking, elastic-plastic behaviour, contact...), we show the efficiency and the flexibility of such coupling method. A comparative analysis of several optimisation tools is also carried on, and the interacting multiple patchs situation is handled. Then, non-intrusive coupling is extended to globally non-linear cases, and a domain decomposition method with non-linear relocalization is proposed. Such methods allowed us to run a parallel computation using only sequential software, on a high performance computing cluster. Finally, we apply the coupling algorithm to mesh refinement with patches of finite elements. We develop an explicit residual based error estimator suitable for multi-scale solutions arising from the non-intrusive coupling, and apply it inside an error driven local mesh refinement procedure. Through this work, a software tool for non-intrusive coupling was developed, based on data exchange between codes (Message Passing Interface protocol). Developments are integrated into a Python wrapper, whose role is to connect several instances of Code_Aster, the structural analysis code developed by EDF R&D, which will be used in the following work.

Made available in DSpace on 2014-06-11T19:27:14Z (GMT). No. of bitstreams: 0 Previous issue date: 2008-07-30Bitstream added on 2014-06-13T19:55:34Z : No. of bitstreams: 1 freitas_tc_me_ilha.pdf: 1484818 bytes, checksum: 9f0ca1d5825d93918e44fc9b31aae513 (MD5) / Agência Nacional de Energia Elétrica (ANEEL) / Este trabalho apresenta o estudo, a implementação e a aplicação de duas técnicas de identificação de parâmetros modais utilizando apenas as respostas da estrutura, denominadas: Identificação Estocástica de Subespaço (IES) e Decomposição no Domínio da Freqüência (DDF). A IES é baseada na Decomposição em Valores Singulares (DVS) da projeção ortogonal do espaço das linhas das saídas futuras no espaço das linhas das saídas passadas. Uma vez realizada a DVS da projeção ortogonal é possível obter o modelo de espaço de estado da estrutura e os parâmetros modais são estimados diretamente através da decomposição em autovalores e autovetores da matriz dinâmica. A DDF é baseada na DVS da matriz de densidade espectral de potência de saída nas linhas de freqüências correspondentes a região em torno de um modo. O primeiro vetor singular obtido para cada linha de freqüência contém as respectivas informações daquele modo e os correspondentes valores singulares levam a função densidade espectral de um sistema equivalente de um grau de liberdade (1GL), permitindo a obtenção dos parâmetros do respectivo modo. Os métodos são avaliados utilizando dados simulados e experimentais. Os resultados mostram que as técnicas implementadas são capazes de estimar os parâmetros modais de estruturas utilizando apenas as respostas. / This work presents the study, implementation and application of the two techniques for the modal parameters identification using only response data: Stochastic Subspace Identification (SSI) and Frequency Domain Decomposition (FDD). The SSI is based on Singular Value Decomposition (SVD) of the orthogonal projection of the future output row space in the past output row space. After the completion of the SVD of the orthogonal projection, is possible to get the state space model of the structure and the modal parameters are estimated directly through the eigenvalues and eigenvectors decomposition of the dynamic matrix. The FDD is based on the SVD of the output power spectral density matrix in the frequencies lines around a mode. The first singular vector obtained for each frequency line contains the respective information about this mode and the corresponding spectral density function leads to an equivalent system of one degree of freedom (1 DOF), allowing the calculation of the parameters of the mode. The methods are evaluated using simulated and experimental data. The results show that the techniques implemented are capable to estimate the modal parameters of structures using only response data.

Neste trabalho desenvolveu-se e implementou-se um modelo computacional paralelo multifísica para a simulação do transporte de substâncias e do escoamento hidrodinâmico, bidimensional (2D) e tridimensional (3D), em corpos de água. Sua motivação está centrada no fato de que as margens e zonas costeiras de rios, lagos, estuários, mares e oceanos são locais de aglomerações de seres humanos, dada a sua importância para as atividades econômica, de transporte e de lazer, causando desequilíbrios a esses ecossistemas. Esse fato impulsiona o desenvolvimento de pesquisas relativas a esta temática. Portanto, o objetivo deste trabalho é o de construir um modelo computacional com alta qualidade numérica, que possibilite simular os comportamentos da hidrodinâmica e do transporte escalar de substâncias em corpos de água com complexa configuração geométrica, visando a contribuir para seu manejo racional. Visto que a ênfase nessa tese são os aspectos numéricos e computacionais dos algoritmos, analisaram-se as características e propriedades numérico-computacionais que as soluções devem contemplar, tais como a estabilidade, a monotonicidade, a positividade e a conservação da massa. As estratégias de soluções enfocam os termos advectivos e difusivos, horizontais e verticais, da equação do transporte. Desse modo, a advecção horizontal é resolvida empregando o método da limitação dos fluxos de Sweby, e o transporte vertical (advecção e difusão) é resolvido com os métodos beta de Gross e de Crank-Nicolson. São empregadas malhas com distintas resoluções para a solução do problema multifísica. O esquema numérico resultante é semi-implícito, computacionalmente eficiente, estável e fornece acurácia espacial e temporal de segunda ordem. Os sistemas de equações resultantes da discretização, em diferenças finitas, das equações do escoamento e do transporte 3D, são de grande porte, lineares, esparsos e simétricos definidos-positivos (SDP). No caso 2D os sistemas são lineares, mas os sistemas de equações para a equação do transporte não são simétricos. Assim, para a solução de sistemas de equações SDP e dos sistemas não simétricos empregam-se, respectivamente, os métodos do subespaço de Krylov do gradiente conjugado e do resíduo mínimo generalizado. No caso da solução dos sistemas 3-diagonal, utiliza-se o algoritmo de Thomas e o algoritmo de Cholesky. A solução paralela foi obtida sob duas abordagens. A decomposição ou particionamento de dados, onde as operações e os dados são distribuídos entre os processos disponíveis e são resolvidos em paralelo. E, a decomposição de domínio, onde obtém-se a solução do problema global combinando as soluções de subproblemas locais. Em particular, emprega-se neste trabalho, o método de decomposição de domínio aditivo de Schwarz, como método de solução, e como pré-condicionador. Para maximizar a relação computação/comunicação, visto que a eficiência computacional da solução paralela depende diretamente do balanceamento de carga e da minimização da comunicação entre os processos, empregou-se algoritmos de particionamento de grafos para obter localmente os subproblemas, ou as partes dos dados. O modelo computacional paralelo resultante mostrou-se computacionalmente eficiente e com alta qualidade numérica. / A multi-physics parallel computational model was developed and implemented for the simulation of substance transport and for the two-dimensional (2D) and threedimensional (3D) hydrodynamic flow in water bodies. The motivation for this work is focused in the fact that the margins and coastal zones of rivers, lakes, estuaries, seas and oceans are places of human agglomeration, because of their importance for economic, transport, and leisure activities causing ecosystem disequilibrium. This fact stimulates the researches related to this topic. Therefore, the goal of this work is to build a computational model of high numerical quality, that allows the simulation of hydrodynamics and of scalar transport of substances behavior in water bodies of complex configuration, aiming at their rational management. Since the focuses of this thesis are the numerical and computational aspects of the algorithms, the main numerical-computational characteristics and properties that the solutions need to fulfill were analyzed. That is: stability, monotonicity, positivity and mass conservation. Solution strategies focus on advective and diffusive terms, horizontal and vertical terms of the transport equation. In this way, horizontal advection is solved using Sweby’s flow limiting method; and the vertical transport (advection and diffusion) is solved with Gross and Crank-Nicolson’s beta methods. Meshes of different resolutions are employed in the solution of the multi-physics problem. The resulting numerical scheme is semi-implicit, computationally efficient, stable and provides second order accuracy in space and in time. The equation systems resulting of the discretization, in finite differences, of the flow and 3D transport are of large scale, linear, sparse and symmetric positive definite (SPD). In the 2D case, the systems are linear, but the equation systems for the transport equation are not symmetric. Therefore, for the solution of SPD equation systems and of the non-symmetric systems we employ, respectively, the methods of Krylov’s sub-space of the conjugate gradient and of the generalized minimum residue. In the case of the solution of 3-diagonal systems, Thomas algorithm and Cholesky algorithm are used. The parallel solution was obtained through two approaches. In data decomposition or partitioning, operation and data are distributed among the processes available and are solved in parallel. In domain decomposition the solution of the global problem is obtained combining the solutions of the local sub-problems. In particular, in this work, Schwarz additive domain decomposition method is used as solution method and as preconditioner. In order to maximize the computation/communication relation, since the computational efficiency of the parallel solution depends directly of the load balancing and of the minimization of the communication between processes, graph-partitioning algorithms were used to obtain the sub-problems or part of the data locally. The resulting parallel computational model is computationally efficient and of high numerical quality.

CoordenaÃÃo de AperfeÃoamento de Pessoal de NÃvel Superior / This work describes a technique for generating parametric surfaces meshes using parallel computing, with distributed memory processors. The input for the algorithm is a set of parametric patches that model the surface of a given object. A structure for spatial partitioning is proposed to decompose the domain in as many subdomains as processes in the parallel system. Each subdomain consists of a set of patches and the division of its load is guided following an estimate. This decomposition attempts to balance the amount of work in all the subdomains. The amount of work, known as load, of any mesh generator is usually given as a function of its output size, i.e., the size of the generated mesh. Therefore, a technique to estimate the size of this mesh, the total load of the domain, is needed beforehand. This work makes use of an analytical average curvature calculated for each patch, which in turn is input data to estimate this load and the decomposition is made from this analytical mean curvature. Once the domain is decomposed, each process generates the mesh on that subdomain or set of patches by a quad tree technique for inner regions, advancing front technique for border regions and is finally applied an improvement to mesh generated. This technique presented good speed-up results, keeping the quality of the mesh comparable to the quality of the serially generated mesh. / Este trabalho descreve uma tÃcnica para gerar malhas de superfÃcies paramÃtricas utilizando computaÃÃo paralela, com processadores de memÃria compartilhada. A entrada para o algoritmo à um conjunto de patches paramÃtricos que modela a superfÃcie de um determinado objeto. Uma estrutura de partiÃÃo espacial à proposta para decompor o domÃnio em tantos subdomÃnios quantos forem os processos no sistema paralelo. Cada subdomÃnio à formado por um conjunto de patches e a divisÃo de sua carga à guiada seguindo uma estimativa de carga. Esta decomposiÃÃo tenta equilibrar a quantidade de trabalho em todos os subdomÃnios. A quantidade de trabalho, conhecida como carga, de qualquer gerador de malha à geralmente dada em funÃÃo do tamanho da saÃda do algoritmo, ou seja, do tamanho da malha gerada. Assim, faz-se necessÃria uma tÃcnica para estimar previamente o tamanho dessa malha, que à a carga total do domÃnio. Este trabalho utiliza-se de um cÃlculo de curvatura analÃtica mÃdia para cada patch, que por sua vez, à dado de entrada para estimar esta carga e a decomposiÃÃo à feita a partir dessa curvatura analÃtica mÃdia. Uma vez decomposto o domÃnio, cada processo gera a malha em seu subdomÃnio ou conjunto de patches pela tÃcnica de quadtree para regiÃes internas, avanÃo de fronteira para regiÃes de fronteira e por fim à aplicado um melhoramento na malha gerada. Esta tÃcnica apresentou bons resultados de speed-up, mantendo a qualidade da malha comparÃvel à qualidade da malha gerada de forma sequencial.

Neste trabalho desenvolveu-se e implementou-se um modelo computacional paralelo multifísica para a simulação do transporte de substâncias e do escoamento hidrodinâmico, bidimensional (2D) e tridimensional (3D), em corpos de água. Sua motivação está centrada no fato de que as margens e zonas costeiras de rios, lagos, estuários, mares e oceanos são locais de aglomerações de seres humanos, dada a sua importância para as atividades econômica, de transporte e de lazer, causando desequilíbrios a esses ecossistemas. Esse fato impulsiona o desenvolvimento de pesquisas relativas a esta temática. Portanto, o objetivo deste trabalho é o de construir um modelo computacional com alta qualidade numérica, que possibilite simular os comportamentos da hidrodinâmica e do transporte escalar de substâncias em corpos de água com complexa configuração geométrica, visando a contribuir para seu manejo racional. Visto que a ênfase nessa tese são os aspectos numéricos e computacionais dos algoritmos, analisaram-se as características e propriedades numérico-computacionais que as soluções devem contemplar, tais como a estabilidade, a monotonicidade, a positividade e a conservação da massa. As estratégias de soluções enfocam os termos advectivos e difusivos, horizontais e verticais, da equação do transporte. Desse modo, a advecção horizontal é resolvida empregando o método da limitação dos fluxos de Sweby, e o transporte vertical (advecção e difusão) é resolvido com os métodos beta de Gross e de Crank-Nicolson. São empregadas malhas com distintas resoluções para a solução do problema multifísica. O esquema numérico resultante é semi-implícito, computacionalmente eficiente, estável e fornece acurácia espacial e temporal de segunda ordem. Os sistemas de equações resultantes da discretização, em diferenças finitas, das equações do escoamento e do transporte 3D, são de grande porte, lineares, esparsos e simétricos definidos-positivos (SDP). No caso 2D os sistemas são lineares, mas os sistemas de equações para a equação do transporte não são simétricos. Assim, para a solução de sistemas de equações SDP e dos sistemas não simétricos empregam-se, respectivamente, os métodos do subespaço de Krylov do gradiente conjugado e do resíduo mínimo generalizado. No caso da solução dos sistemas 3-diagonal, utiliza-se o algoritmo de Thomas e o algoritmo de Cholesky. A solução paralela foi obtida sob duas abordagens. A decomposição ou particionamento de dados, onde as operações e os dados são distribuídos entre os processos disponíveis e são resolvidos em paralelo. E, a decomposição de domínio, onde obtém-se a solução do problema global combinando as soluções de subproblemas locais. Em particular, emprega-se neste trabalho, o método de decomposição de domínio aditivo de Schwarz, como método de solução, e como pré-condicionador. Para maximizar a relação computação/comunicação, visto que a eficiência computacional da solução paralela depende diretamente do balanceamento de carga e da minimização da comunicação entre os processos, empregou-se algoritmos de particionamento de grafos para obter localmente os subproblemas, ou as partes dos dados. O modelo computacional paralelo resultante mostrou-se computacionalmente eficiente e com alta qualidade numérica. / A multi-physics parallel computational model was developed and implemented for the simulation of substance transport and for the two-dimensional (2D) and threedimensional (3D) hydrodynamic flow in water bodies. The motivation for this work is focused in the fact that the margins and coastal zones of rivers, lakes, estuaries, seas and oceans are places of human agglomeration, because of their importance for economic, transport, and leisure activities causing ecosystem disequilibrium. This fact stimulates the researches related to this topic. Therefore, the goal of this work is to build a computational model of high numerical quality, that allows the simulation of hydrodynamics and of scalar transport of substances behavior in water bodies of complex configuration, aiming at their rational management. Since the focuses of this thesis are the numerical and computational aspects of the algorithms, the main numerical-computational characteristics and properties that the solutions need to fulfill were analyzed. That is: stability, monotonicity, positivity and mass conservation. Solution strategies focus on advective and diffusive terms, horizontal and vertical terms of the transport equation. In this way, horizontal advection is solved using Sweby’s flow limiting method; and the vertical transport (advection and diffusion) is solved with Gross and Crank-Nicolson’s beta methods. Meshes of different resolutions are employed in the solution of the multi-physics problem. The resulting numerical scheme is semi-implicit, computationally efficient, stable and provides second order accuracy in space and in time. The equation systems resulting of the discretization, in finite differences, of the flow and 3D transport are of large scale, linear, sparse and symmetric positive definite (SPD). In the 2D case, the systems are linear, but the equation systems for the transport equation are not symmetric. Therefore, for the solution of SPD equation systems and of the non-symmetric systems we employ, respectively, the methods of Krylov’s sub-space of the conjugate gradient and of the generalized minimum residue. In the case of the solution of 3-diagonal systems, Thomas algorithm and Cholesky algorithm are used. The parallel solution was obtained through two approaches. In data decomposition or partitioning, operation and data are distributed among the processes available and are solved in parallel. In domain decomposition the solution of the global problem is obtained combining the solutions of the local sub-problems. In particular, in this work, Schwarz additive domain decomposition method is used as solution method and as preconditioner. In order to maximize the computation/communication relation, since the computational efficiency of the parallel solution depends directly of the load balancing and of the minimization of the communication between processes, graph-partitioning algorithms were used to obtain the sub-problems or part of the data locally. The resulting parallel computational model is computationally efficient and of high numerical quality.

La turbine Pelton est une turbine hydraulique dont le fonctionnement se caractérise par l’interaction d’un jet d’eau avec les augets d’une roue. Cette étude a pour but de comprendre les phénomènes influençant le jet et son interaction avec les augets. Pour cela deux actions différentes ont été menées. Une première a visé à caractériser expérimentalement la fragmentation d’un jet de turbine Pelton. La seconde s’est attachée à développer une méthode numérique pouvant mener`à la simulation précise de jets réels de turbines Pelton. La partie expérimentale a permis de déterminer le mode de fragmentation de ces jets (atomisation turbulente), mais aussi l’influence de la rugosité des parois de l’injecteur sur les performances de la turbine. La participation de ce travail à un projet expérimental a permis de montrer l’influence de l’écoulement en sortie d’injecteur sur la fragmentation du jet. Les phénomènes physiques influençant principalement l’évolution du jet ont ainsi été déterminés. La partie numérique a eu pour but de mettre en place une méthode permettant de simuler l’évolution d’un jet de turbine Pelton (fragmentation) et son interaction avec un auget. Etant donnés les progrès de la méthode SPH-ALE pour la simulation d’impact de jets pour les turbines Pelton, il a été décidé d’adapter cette méthode pour les simulations visées. Ainsi une étude du choix de la vitesse des interfaces de problème de Riemann a permis de réaliser un modèle multiphase stable pour les forts rapports de densité (eau-air). Cette méthode s’est avérée garantir les propriétés de continuité de vitesse normale et de pression à l’interface entre les fluides. L’ajout des phénomènes de tension de surface s’est fait par l’adaptation du modèle CSF (Continuum Surface Force) et le développement d’un second modèle nommé Laplace Law Pressure Correction (LLPC).L’intégration du saut de pression dans le solveur de Riemann a nécessité une étude précise du calcul de la courbure et a permis d’améliorer la simulation de loi de Laplace. La méthode numérique a été ensuite validée sur les cas académiques d’onde gravitaire, de rupture de barrage et d’oscillation de goutte. Les ressources en mémoire et le temps de calcul associé à cette méthode ont nécessité la parallélisation du code de calcul. Le caractère lagrangien de la méthode a très largement influencé la méthode de découpe de domaine pour permettre une bonne répartition de la charge de calcul entre les différents processeurs. En conclusion les phénomènes physiques influençant la fragmentation de jets issus d’injecteurs de turbine Pelton sont désormais mieux connus et ils ont pu être introduits dans la méthode numérique. Les prochains développements porteront sur la simulation de jets dont la condition d’entrée s’attachera à être représentative des caractéristiques d’un écoulement en sortie d’un injecteur de turbine Pelton. / A Pelton turbine is characterized by a water jet which is impacting rotating buckets. The main goal of this study is to understand the phenomena which are impacting the jet and its interaction with the bucket. This study was considering two main works. One is considering experiments which allow determining the jet fragmentation. The second part considers development of a numerical code able to reproduce phenomena linked to Pelton jet fragmentation. The experimental part succeeds to associate Pelton jet behavior with mode of jet fragmentation (turbulent dispersion) and shows the impact of hydraulic roughness on Pelton turbine performances. The access to experimental results from a project involving this PhD work, demonstrates the role of the inlet velocity/turbulence profile on the jet fragmentation. The numerical part used the SPH-ALE (Smoothed Particle Hydrodynamics - Arbitrary Lagrange Euler) method to implement physical models linked with jet fragmentation. This choice was done because of its ability to predict pressure fields resulting of the interaction of a water jet and a rotating bucket. A multiphase model was developed based on a modification of the velocity of the interface of Riemann problem. This model does not diffuse the interface and recovers continuity of normal velocity and pressure at the interface between both fluids. Surface tension effect was implemented through an adaptation of the CSF (Continuum Surface Force) model and through amodel called LLPC for Laplace Law Pressure Correction. A study of the computational methods to determine the interface curvature was performed for the integration of the pressure jump in the Riemann solver. Validation was done on academicals test cases as gravity waves, dam breakor droplet oscillations. The numerical code was parallelized to perform large numerical simulations.To conclude, the numerical code integrates physical phenomena which were shown as important in the experiments. The developments will try to perform jet simulation with inlet condition which will be representative of flow conditions at the outlet of a Pelon turbine injector.