This paper presents a robust and artist driven method for fracturing a surface polyhedral mesh via fracture maps. A fracture map is an undirected simple graph with nodes representing positions in UV-space and fracture lines along the surface of a mesh. Fracture maps allow artists to concisely and rapidly define, edit, and apply fracture patterns onto the surface of their mesh. The method projects a fracture map onto a polyhedral surface and splits its triangles accordingly. The polyhedral mesh is then segmented based on fracture lines to produce a set of independent surfaces called fracture components, containing the visible surface of each fractured mesh fragment. Subsequently, we utilize a Voronoi-based approximation of the input polyhedral mesh’s medial axis to derive a hidden surface for each fragment. The result is a new watertight polyhedral mesh representing the full fracture component. Results are aquired after a delay sufficiently brief for interactive design. As the size of the input mesh increases, the computation time has shown to grow linearly. A large mesh of 41,000 triangles requires approximately 3.4 seconds to perform a complete fracture of a complex pattern. For a wide variety of practices, the resulting fractures allows users to provide realistic feedback upon the application of extraneous forces.

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

Nous proposons dans ce mémoire de thèse une méthodologie permettant la résolution du problème de la décomposition de Hodge-Helmholtz discrète sur maillages polyédriques. Le défi de ce travail consiste à respecter les propriétés de la décomposition au niveau discret. Pour répondre à cet objectif, nous menons une étude bibliographique nous permettant d'identifier la nécessité de la mise en oeuvre de schémas numériques mimétiques. La description ainsi que la validation de la mise en oeuvre de ces schémas sont présentées dans ce mémoire. Nous revisitons et améliorons les méthodes de décomposition que nous étudions ensuite au travers d'expériences numériques. En particulier, nous détaillons le choix d'un solveur linéaire ainsi que la convergence des quantités extraites sur un ensemble varié de maillages polyédriques et de conditions aux limites. Nous appliquons finalement la décomposition de Hodge-Helmholtz à l'étude de deux écoulements turbulents : un écoulement en canal plan et un écoulement turbulent homogène isotrope. / We propose in this thesis a methodology to compute the Helmholtz-Hodge decomposition on discrete polyhedral meshes. The challenge of this work isto preserve the properties of the decomposition at the discrete level. In our literature survey, we have identified the need of mimetic schemes to achieve our goal. The description and validation of our implementation of these schemes are presented inthis document. We revisit and improve the methods of decomposition we then study through numerical experiments. In particular, we detail our choice of linear solvers and the convergence of extracted quantities on various series of polyhedral meshes and boundary conditions. Finally, we apply the Helmholtz-Hodge decomposition to the study of two turbulent flows: a turbulent channel flow and a homogeneous isotropic turbulent flow.

[pt] A otimização topológica tem se desenvolvido bastante e possui potencial para revolucionar diversas áreas da engenharia. Este método pode ser implementado a partir de diferentes abordagens, tendo como base o Método dos Elementos Finitos. Ao se utilizar uma abordagem baseada no elemento, potencialmente, cada elemento finito pode se tornar um vazio ou um sólido, e a cada elemento do domínio é atribuído uma variável de projeto, constante, denominada densidade. Do ponto de vista Euleriano, a topologia obtida é um subconjunto dos elementos iniciais. No entanto, tal abordagem está sujeita a instabilidades numéricas, tais como conexões de um nó e rápidas oscilações de materiais do tipo sólido-vazio (conhecidas como instabilidade de tabuleiro). Projetos indesejáveis podem ser obtidos quando elementos de baixa ordem são utilizados e métodos de regularização e/ou restrição não são aplicados. Malhas poliédricas não estruturadas naturalmente resolvem esses problemas e oferecem maior flexibilidade na discretização de domínios não Cartesianos. Neste trabalho investigamos a otimização topológica em malhas poliédricas por meio de um acoplamento entre malhas. Primeiramente, as malhas poliédricas são geradas com base no conceito de diagramas centroidais de Voronoi e posteriormente otimizadas para uso em análises de elementos finitos. Demonstramos que o número de condicionamento do sistema de equações associado pode ser melhorado ao se minimizar uma função de energia relacionada com a geometria dos elementos. Dada a qualidade da malha e o tamanho do problema, diferentes tipos de resolvedores de sistemas de equações lineares apresentam diferentes desempenhos e, portanto, ambos os resolvedores diretos e iterativos são abordados. Em seguida, os poliedros são decompostos em tetraedros por um algoritmo específico de acoplamento entre as malhas. A discretização em poliedros é responsável pelas variáveis de projeto enquanto a malha tetraédrica, obtida pela subdiscretização da poliédrica, é utilizada nas análises via método dos elementos finitos. A estrutura modular, que separa as rotinas e as variáveis usadas nas análises de deslocamentos das usadas no processo de otimização, tem se mostrado promissora tanto na melhoria da eficiência computacional como na qualidade das soluções que foram obtidas neste trabalho. Os campos de deslocamentos e as variáveis de projeto são relacionados por meio de um mapeamento. A arquitetura computacional proposta oferece uma abordagem genérica para a solução de problemas tridimensionais de otimização topológica usando poliedros, com potencial para ser explorada em outras aplicações que vão além do escopo deste trabalho. Finalmente, são apresentados diversos exemplos que demonstram os recursos e o potencial da abordagem proposta. / [en] Topology optimization has had an impact in various fields and has the potential to revolutionize several areas of engineering. This method can be implemented based on the finite element method, and there are several approaches of choice. When using an element-based approach, every finite element is a potential void or actual material, whereas every element in the domain is assigned to a constant design variable, namely, density. In an Eulerian setting, the obtained topology consists of a subset of initial elements. This approach, however, is subject to numerical instabilities such as one-node connections and rapid oscillations of solid and void material (the so-called checkerboard pattern). Undesirable designs might be obtained when standard low-order elements are used and no further regularization and/or restrictions methods are employed. Unstructured polyhedral meshes naturally address these issues and offer fl exibility in discretizing non-Cartesians domains. In this work we investigate topology optimization on polyhedra meshes through a mesh staggering approach. First, polyhedra meshes are generated based on the concept of centroidal Voronoi diagrams and further optimized for finite element computations. We show that the condition number of the associated system of equations can be improved by minimizing an energy function related to the element s geometry. Given the mesh quality and problem size, different types of solvers provide different performances and thus both direct and iterative solvers are addressed. Second, polyhedrons are decomposed into tetrahedrons by a tailored embedding algorithm. The polyhedra discretization carries the design variable and a tetrahedra subdiscretization is nested within the polyhedra for finite element analysis. The modular framework decouples analysis and optimization routines and variables, which is promising for software enhancement and for achieving high fidelity solutions. Fields such as displacement and design variables are linked through a mapping. The proposed mapping-based framework provides a general approach to solve three-dimensional topology optimization problems using polyhedrons, which has the potential to be explored in applications beyond the scope of the present work. Finally, the capabilities of the framework are evaluated through several examples, which demonstrate the features and potential of the proposed approach.

Cette thèse porte sur la modélisation et la discrétisation d’écoulements Darcy dans les milieux poreux fracturés. Nous suivons l’approche des modèles, dits à dimensions hybrides, qui représentent les réseaux de fractures comme des surfaces de codimension 1 immergées dans la matrice. Les modèles considérés prennent en compte les interactions entre matrice et fractures et permettent de traiter des fractures agissant comme conduites ou comme barrières, ce que nécessite de prendre en compte les sauts de pression aux interfaces matrice-fracture. Dans le cas des écoulements diphasiques, nous proposons des modèles, qui prennent en compte les sauts de saturations aux interfaces matrice-fracture, dû à la capillarité. L’analyse numérique est menée dans le cadre général de la méthode de discrétisations gradients, qui est étendue aux modèles considérés. Deux familles de schémas numériques, le schéma Vertex Approximate Gradient et le schéma Volumes Finis Hybrides sont adaptées aux modèles à dimensions hybrides. On prouve via des résultats de densité que ce sont des schémas gradients, pour lesquels la convergence est établie. En diphasique, l’existence d’une solution est obtenue en passant. Plusieurs cas tests sont présentés. En monophasique, on observe la convergence sur des différents types de mailles pour une famille de solutions dans un milieux fracturé hétérogène et anisotrope. En diphasique, nous présentons une série de cas tests afin de comparer les modèles à dimensions hybrides au modèle de référence, dans lequel les fractures ont la même dimension que la matrice. A part quantifier le gain en performance de calcul, ces tests montrent la qualité des différents modèles réduits. / This thesis investigates the modelling of Darcy flow through fractured porous media and its discretization on general polyhedral meshes. We follow the approach of hybrid dimensional models, invoking a complex network of planar fractures. The models account for matrix-fracture interactions and fractures acting either as drains or as barriers, i.e. we have to deal with pressure discontinuities at matrix-fracture interfaces. In the case of two phase flow, we present two models, which permit to treat gravity dominated flow as well as discontinuous capillary pressure at the material interfaces. The numerical analysis is performed in the general framework of the Gradient Discretisation Method, which is extended to the models under consideration. Two families of schemes namely the Vertex Approximate Gradient scheme (VAG) and the Hybrid Finite Volume scheme (HFV) are detailed and shown to fit in the gradient scheme framework, which yields, in particular, convergence. For single phase flow, we obtain convergence of order 1 via density results. For two phase flow, the existence of a solution is obtained as a byproduct of the convergence analysis. Several test cases are presented. For single phase flow, we study the convergence on different types of meshes for a family of solutions. For two phase flow, we compare the hybrid-dimensional models to the reference equidimensional model, in which fractures have the same dimension as the matrix. This does not only provide quantitative evidence about computational gain, but also leads to deep insight about the quality of the proposed reduced models.