Formoptimierung bezeichnet die Bestimmung der Geometrischen Gestalt eines Gebietes auf dem eine partielle Differentialgleichung (PDE) wirkt, sodass bestimmte gegebene Zielgrößen, welche von der Lösung der PDE abhängen, Extrema annehmen. Bei der Diskret Adjungierten Methode wird der Gradient einer Zielgröße bezüglich einer beliebigen Anzahl von Formparametern mit Hilfe der Lösung einer adjungierten Gleichung der diskretisierten PDE effizient ermittelt. Dieser Gradient wird dann in Verfahren der numerischen Optimierung verwendet um eine optimale Lösung zu suchen. Da sowohl die Zielgröße als auch der Gradient für die diskretisierte PDE ermittelt werden, sind beide zunächst vom verwendeten Netz abhängig. Bei groben Netzen sind sogar Unstetigkeiten der diskreten Zielfunktion zu erwarten, wenn bei Änderungen der Formparameter sich das Netz unstetig ändert (z.B. Änderung Anzahl Knoten, Umschalten der Konnektivität). Mit zunehmender Feinheit der Netze verschwinden jedoch diese Unstetigkeiten aufgrund der Konvergenz der Diskretisierung. Da im Zuge der Formoptimierung Zielgröße und Gradient für eine Vielzahl von Iterierten der Lösung bestimmt werden müssen, ist man bestrebt die Kosten einer einzelnen Auswertung möglichst gering zu halten, z.B. indem man mit nur moderat feinen oder adaptiv verfeinerten Netzen arbeitet. Aufgabe dieser Diplomarbeit ist es zu untersuchen, ob mit gängigen Methoden adaptiv verfeinerte Netze hinreichende Genauigkeit der Auswertung von Zielgröße und Gradient erlauben und ob eventuell Anpassungen der Optimierungsstrategie an die adaptive Vernetzung notwendig sind. Für die Untersuchungen sind geeignete Modellprobleme aus der Festigkeitslehre zu wählen und zu untersuchen. / Shape optimization describes the determination of the geometric shape of a domain with a partial differential equation (PDE) with the purpose that a specific given performance function is minimized, its values depending on the solution of the PDE. The Discrete Adjoint Method can be used to evaluate the gradient of a performance function with respect to an arbitrary number of shape parameters by solving an adjoint equation of the discretized PDE. This gradient is used in the numerical optimization algorithm to search for the optimal solution. As both function value and gradient are computed for the discretized PDE, they both fundamentally depend on the discretization. In using the coarse meshes, discontinuities in the discretized objective function can be expected if the changes in the shape parameters cause discontinuous changes in the mesh (e.g. change in the number of nodes, switching of connectivity). Due to the convergence of the discretization these discontinuities vanish with increasing fineness of the mesh. In the course of shape optimization, function value and gradient require evaluation for a large number of iterations of the solution, therefore minimizing the costs of a single computation is desirable (e.g. using moderately or adaptively refined meshes). Overall, the task of the diploma thesis is to investigate if adaptively refined meshes with established methods offer sufficient accuracy of the objective value and gradient, and if the optimization strategy requires readjustment to the adaptive mesh design. For the investigation, applicable model problems from the science of the strength of materials will be chosen and studied.

There are several algorithms for solving the linear system of equations that arise from the finite element method with linear or near-linear computational complexity. One way is to find an approximation of the stiffness matrix that is such that it can be used in a preconditioned conjugate residual method, that is, a preconditioner to the stiffness matrix. We have studied two preconditioners for the conjugate residual method, both based on writing the stiffness matrix in block form, factorising it and then approximating the Schur complement block to get a preconditioner. We have studied the stationary reaction-diffusion-advection equation in two dimensions. The mesh is refined adaptively, giving a hierarchy of meshes. In the first method the Schur complement is approximated by the stiffness matrix at one coarser level of the mesh, in the second method it is approximated as the assembly of local Schur complements corresponding to macro triangles. For two levels the theoretical bound of the condition number is 1/(1-C²) for either method, where C is the Cauchy-Bunyakovsky-Schwarz constant. For multiple levels there is less theory. For the first method it is known that the condition number of the preconditioned stiffness matrix is O(l²), where l is the number of levels of the preconditioner, or, equivalently, the number mesh refinements. For the second method the asymptotic behaviour is not known theoretically. In neither case is the dependency of the condition number of C known. We have tested both methods on several problems and found the first method to always give a better condition number, except for very few levels. For all tested problems, using the first method it seems that the condition number is O(l), in fact it is typically not larger than Cl. For the second method the growth seems to be superlinear.

Scientific computing is used frequently in an increasing number of disciplines to accelerate scientific discovery. Many such computing problems involve the numerical solution of partial differential equations (PDE). In this thesis we explore and develop methodology for high-performance implementations of PDE solvers for shared-memory multiprocessor architectures. We consider three realistic PDE settings: solution of the Maxwell equations in 3D using an unstructured grid and the method of conjugate gradients, solution of the Poisson equation in 3D using a geometric multigrid method, and solution of an advection equation in 2D using structured adaptive mesh refinement. We apply software optimization techniques to increase both parallel efficiency and the degree of data locality. In our evaluation we use several different shared-memory architectures ranging from symmetric multiprocessors and distributed shared-memory architectures to chip-multiprocessors. For distributed shared-memory systems we explore methods of data distribution to increase the amount of geographical locality. We evaluate automatic and transparent page migration based on runtime sampling, user-initiated page migration using a directive with an affinity-on-next-touch semantic, and algorithmic optimizations for page-placement policies. Our results show that page migration increases the amount of geographical locality and that the parallel overhead related to page migration can be amortized over the iterations needed to reach convergence. This is especially true for the affinity-on-next-touch methodology whereby page migration can be initiated at an early stage in the algorithms. We also develop and explore methodology for other forms of data locality and conclude that the effect on performance is significant and that this effect will increase for future shared-memory architectures. Our overall conclusion is that, if the involved locality issues are addressed, the shared-memory programming model provides an efficient and productive environment for solving many important PDE problems.

Capacity analysis is a common preliminary step in the design of offshore foundations. Inaccuracies in traditional capacity analysis methods, and the advancement of numerical modelling capabilities, have increasingly led designers to optimise foundations using more complex methods. In this thesis, the ultimate limit state capacity of a range of foundation types is investigated using finite element limit analysis. Novel three-dimensional finite element limit analysis software is benchmarked against analytical solutions and conventional displacement finite element analysis. It is then used to find lower and upper bounds of foundation capacity, with adaptive mesh refinement used to reduce the bound gap over successive iterations of the solution. Rigid foundations subjected to short term loading on clay soil are analysed. The undrained soil is modelled as a rigid--plastic von Mises material, and attention is given to modelling any normal and/or shear stress limits at the foundation/soil interface. Shallow foundations, suction anchor foundations, and hybrid mudmat/pile foundations are considered. Realistic six degree-of-freedom load combinations are applied and results are reported in the form of normalised design charts, and tables, that are suitable for use in preliminary design. Relationships between loading combinations and failure mechanisms are also explored. A number of case studies based on authentic foundation designs are analysed. The results suggest that finite element limit analysis could provide an attractive alternative to displacement finite element analysis for preliminary foundation design calculations.

Les travaux de thèse présentés dans ce mémoire s’inscrivent dans le cadre du projet ANR BARPPRO. Ce programme de recherche vise à étudier l’influence d’une barrière de protection face à une explosion en régime de détonation. L’objectif est d’établir des méthodes de calcul rapides de classement des zones d’effets pour aider les industriels au dimensionnement des barrières de protection. L’une à partir d’abaques, valable pour des configurations en géométrie 2D, sur des plages spécifiées de paramètres importants retenus, avec une précision de +/- 5%. L’autre à partir d’une méthode d’estimation rapide basée notamment sur les chemins déployés, valable en géométrie 2D et en géométrie 3D, mais dont la précision estimée est de +/- 30%. Afin d’y parvenir, l’étude s’appuie sur trois volets : expérimental, simulation numérique et analytique. La partie expérimentale étudie plusieurs géométries de barrière de protection à petites échelles pour la détonation d’une charge gazeuse (propane-oxygène à la stoechiométrie). Les configurations expérimentées servent à la validation de l’outil de simulation numérique constitué du solveur HERA et de la plateforme de calcul TERA 100. Des abaques d’aide au dimensionnement ont pu être réalisés à partir de résultats fournis par l’outil de simulation (3125 configurations de barrière de protection, TNT). L’étude des différents phénomènes physiques présents a également permis de mettre en place une méthode d’estimation rapide basée sur des relations géométriques, analytiques et empiriques. L’analyse de ces résultats a permis d’établir quelques recommandations dans le dimensionnement d’une barrière de protection. Les abaques et le programme d’estimation rapide permettent à un ingénieur de dimensionner rapidement une barrière de protection en fonction de la configuration du terrain et de la position de la zone à protéger en aval du merlon. / This thesis is a part of the ANR BARPPRO project. This research program studies this influence of the protection barrier during an explosion detonation. The goal of this project is to establish fast-computation methods of area classification effects to help the industrial to design the protection barrier on the SEVESO sites. One from abacus, for configurations in 2D geometry on specified parameters used, with an accuracy of +/- 5%. The other from a fast-running method based on broken lines for configurations in 2D and 3D geometries, but the accuracy is +/- 30%. This study includes three approaches: experimental, numerical simulation and analytical approaches. The experimental part studies several geometries of the protection barrier for a gaseous explosion (stoichiometric propane-oxygen mixture) at small scales. The experimental configurations used to validate the numerical simulation tool constituted of the HERA software and the TERA 100 supercomputer. The overpressure charts were able to generate from the numerical results (3125 configurations of the barrier for a TNT charge). The analysis of these results allows to establish different recommendations in the design of the protection barrier. The study of the different physical phenomena present has also helped to set up a fast-running method based on the geometrical, empirical and analytical relations. All these tools will enable an engineer to analyze and estimate the evolution of overpressure around the barrier as a function of the site’s dimensions.

OLIVEIRA, Suzana Matos França de. Simulação de malha triangular: um estudo sobre a adaptatividade da malha. 2013. 73 f. Dissertação (Mestrado em ciência da computação)- Universidade Federal do Ceará, Fortaleza-CE, 2013. / Submitted by Elineudson Ribeiro (elineudsonr@gmail.com) on 2016-07-12T17:36:10Z No. of bitstreams: 1 2013_dis_smfoliveira.pdf: 12369786 bytes, checksum: f3aac3380bd894f258de70c6024bf91f (MD5) / Approved for entry into archive by Rocilda Sales (rocilda@ufc.br) on 2016-07-22T12:31:55Z (GMT) No. of bitstreams: 1 2013_dis_smfoliveira.pdf: 12369786 bytes, checksum: f3aac3380bd894f258de70c6024bf91f (MD5) / Made available in DSpace on 2016-07-22T12:31:55Z (GMT). No. of bitstreams: 1 2013_dis_smfoliveira.pdf: 12369786 bytes, checksum: f3aac3380bd894f258de70c6024bf91f (MD5) Previous issue date: 2013 / In the last decades, cloth animation has been the focus of much research, due to demands from the entertainment industry and from e-commerce. That type of animation is most often the result of a physics-based simulation and has a great computational cost. This work investigates how to reduce the computational cost of the simulation, by refining the mesh only in regions that need a fine level of detail. The fabric model consists of a triangular mesh and uses a spring-mass-damper system to compute the forces among the particles, which are located at the mesh’s vertices. The collision detection depends on the arrangement of the cloth model and the objects in the scene. The mesh is refined or simplified, taking into account the spring compression, collision and curvature, so the simulation uses a better mesh every time step. Therefore, this work’s main objective is to study the dynamic behavior of cloth, using a remeshing procedure in order to adapt the mesh. / A animação de tecido vem sendo estudada abundantemente nas últimas décadas por haver demanda na indústria do entretenimento bem como no comércio eletrônico de roupas. Esse tipo de animação, na maioria das vezes, é feita com base em simulação física, havendo muito gasto computacional. Esse trabalho tenta usufruir de vários modelos para diminuir esse gasto. É feito uma modelagem do tecido com uma malha triangular e usa-se um modelo massa-mola-amortecedor para simular as forças entre as partículas, que são os vértices dessa malha. Dependendo da disposição do modelo do tecido e dos objetos da cena, são detectadas colisões entre eles. A malha é discretizada ou simplificada, levando em consideração a compressão, a colisão e a curvatura das molas, para que seja usada uma malha boa em cada passo ao longo da animação. Portanto, o objetivo principal desse trabalho é estudar o comportamento do tecido utilizando o modelo de remalhamento para adaptar essa malha.

The numerical simulation of fluid flow involving complex geometries is greatly limited by the required spatial grid resolution. These flows often contain small regions with complex motions, while the remaining flow is relatively smooth. Adaptive mesh refinement (AMR) enables the spatial grid to be refined in local regions that require finer grids to resolve the flow. This work describes an approach to parallelization of a structured adaptive mesh refinement (SAMR) algorithm. This type of methodology is based on locally refined grids superimposed on coarser grids to achieve the desired resolution in numerical simulations. Parallel implementations of SAMR methods offer the potential for accurate simulations of high complexity fluid flows. However, they present interesting challenges in dynamic resource allocation, data-distribution and load-balancing. The overall efficiency of parallel SAMR applications is limited by the ability to partition the underlying grid hierarchies at run-time to expose all inherent parallelism, minimize communication and synchronization overheads, and balance load. The methodology is based on a message passing interface model (MPI) using the recursive coordinate bisection (RCB) for domain partition. For this work, a semi-implicit projection method has been implemented to solve the incompressible Navier Stokes equations. All numerical implementations are an extension of a sequential Fortran 90 code, called "AMR3D", developed in the work of Nós (2007) .The efficiency and robustness of the applied methodology are verified via convergence analysis using the method of manufactured solutions. Validations were performed by simulating an incompressible jet flow and a lid driven cavity flow. / A simulação numérica de escoamentos envolvendo geometrias complexas é fortemente limitada pela resolução da malha espacial. Na grande maioria dos escoamentos, há pequenas regiões do domínio onde o fluido se movimenta de forma complexa gerando gradientes elevados, enquanto que no restante do domínio o escoamento é relativamente calmo". O Refinamento Adaptativo de Malhas (Adaptive Mesh Refinement - AMR), possibilita que o refinamento da malha espacial seja mais apurado em regiões especificas, enquanto que nas demais regiões o refinamento pode ser mais grosseiro. O presente trabalho consiste no desenvolvimento de uma metodologia de paralelização para a solução das equações de Navier-Stokes em malhas adaptativas bloco-estruturadas (Structured Adaptive Mesh Refinement - SAMR) utilizando a interface MPI (Message Passing Interface) e o método de bisseção por coordenadas RCB (Recursive Coordinate Bisection) para o balanço de carga. Implementações de métodos SAMR em processamento paralelo oferecem a possibilidade de simulações precisas de escoamentos de elevada complexidade. No entanto, apresentam desafios interessantes quanto à dinamicidade na alocação e distribuição dos dados e no balanceamento de carga. Cabe ressaltar que a é ciência total das aplicações envolvendo métodos SAMR em processamento paralelo é fortemente dependente da qualidade do particionamento dinâmico de domínio, efetuado em tempo de execução, para que se garanta os menores custos de comunicação e sincronização possíveis, além de uma boa distribuição da carga computacional. Neste trabalho, utilizou-se o esquema semi-implícito proposto por Ceniceros et al. (2010) para avanço temporal. Todas as implementações foram efetuadas como uma extensão do código AMR3D", proposto por Nós (2007). A é ciência e a robustez do método proposto são verificadas por meio do método das soluções manufaturadas. As validações foram feitas por meio da simulação do escoamento em uma cavidade com tampa deslizante e de um jato incompressível. / Doutor em Engenharia Mecânica

Este é o primeiro trabalho que apresenta simulações tridimensionais completamente adaptativas de um modelo de campo de fase para um fluido incompressível com densidade de massa constante e viscosidade variável, conhecido como Modelo H. Solucionando numericamente as equações desse modelo em malhas refinadas localmente com a técnica AMR, simulamos computacionalmente escoamentos bifásicos tridimensionais. Os modelos de campo de fase oferecem uma aproximação física sistemática para investigar fenômenos que envolvem sistemas multifásicos complexos, tais como fluidos com camadas de mistura, a separação de fases sob forças de cisalhamento e a evolução de micro-estruturas durante processos de solidificação. Como as interfaces são substituídas por delgadas regiões de transição (interfaces difusivas), as simulações de campo de fase requerem muita resolução nessas regiões para capturar corretamente a física do problema em estudo. Porém essa não é uma tarefa fácil de ser executada numericamente. As equações que caracterizam o modelo de campo de fase contêm derivadas de ordem elevada e intrincados termos não lineares, o que exige uma estratégia numérica eficiente capaz de fornecer precisão tanto no tempo quanto no espaço, especialmente em três dimensões. Para obter a resolução exigida no tempo, usamos uma discretização semi-implícita de segunda ordem para solucionar as equações acopladas de Cahn-Hilliard e Navier-Stokes (Modelo H). Para resolver adequadamente as escalas físicas relevantes no espaço, utilizamos malhas refinadas localmente que se adaptam dinamicamente para recobrir as regiões de interesse do escoamento, como por exemplo, as vizinhanças das interfaces do fluido. Demonstramos a eficiência e a robustez de nossa metodologia com simulações que incluem a separação dos componentes de uma mistura bifásica, a deformação de gotas sob cisalhamento e as instabilidades de Kelvin-Helmholtz. / This is the first work that introduces 3D fully adaptive simulations for a phase field model of an incompressible fluid with matched densities and variable viscosity, known as Model H. Solving numerically the equations of this model in meshes locally refined with AMR technique, we simulate computationally tridimensional two-phase flows. Phase field models offer a systematic physical approach to investigate complex multiphase systems phenomena such as fluid mixing layers, phase separation under shear and microstructure evolution during solidification processes. As interfaces are replaced by thin transition regions (diffuse interfaces), phase field simulations need great resolution in these regions to capture correctly the physics of the studied problem. However, this is not an easy task to do numerically. Phase field model equations have high order derivatives and intricate nonlinear terms, which require an efficient numerical strategy that can achieve accuracy both in time and in space, especially in three dimensions. To obtain the required resolution in time, we employ a semi-implicit second order discretization scheme to solve the coupled Cahn-Hilliard/Navier-Stokes equations (Model H). To resolve adequatly the relevant physical scales in space, we use locally refined meshes which adapt dynamically to cover special flow regions, e.g., the vicinity of the fluid interfaces. We demonstrate the efficiency and robustness of our methodology with simulations that include spinodal decomposition, the deformation of drops under shear and Kelvin-Helmholtz instabilities.

Le calcul d'écoulements dans les générateurs de vapeur des réacteurs à eau pressurisée est un problème complexe, faisant intervenir différents régimes d'écoulement et plusieurs échelles de temps et d'espace. Un scénario accidentel peut être caractérisé par des variations très rapides pour un nombre de Mach de l'ordre de l'unité. A l'inverse en régime nominal l'écoulement peut être stationnaire, à bas nombre de Mach. De plus quelque soit le régime considéré, la complexité de la géométrie d'un générateur de vapeur conduit à modéliser le faisceau de tubes par un milieu poreux, d'où le problème de couplage à l'interface avec le milieu fluide.Un schéma de correction de pression tout-Mach en volumes finis colocalisés a été introduit pour les équations d'Euler et de Navier-Stokes. L'existence d'une solution discrète, la consistance du schéma au sens de Lax et la positivité de l'énergie interne ont été démontrées. Le schéma a été ensuite étendu aux modèles diphasiques homogènes du code GENEPI développé au CEA. Enfin un algorithme Multigrille-AMR a été adaptée pour permettre de mettre en oeuvre notre schéma sur des maillages adaptatifs.Concernant la seconde problématique, une extension de la loi de Beavers-Joseph a été proposée pour le régime convectif. En introduisant un saut d'énergie cinétique à l'interface, on retrouve une loi de type Beavers-Joseph mais avec un coefficient de glissement non-linéaire, qui dépend de la vitesse fluide à l'interface et de la vitesse Darcy. La validité de cette nouvelle condition d'interface a été évaluée en réalisant des calculs de simulation numérique directe à différents nombres de Reynolds. / The numerical simulation of steam generators of pressurized water reactors is a complex problem, involving different flow regimes and a wide range of length and time scales. An accidental scenario may be associated with very fast variations of the flow with an important Mach number. In contrast in the nominal regime the flow may be stationary, at low Mach number. Moreover whatever the regime under consideration, the array of U-tubes is modelled by a porous medium in order to avoid taking into account the complex geometry of the steam generator, which entails the issue of the coupling conditions at the interface with the free-fluid.We propose a new pressure-correction scheme for cell-centered finite volumes for solving the compressible Navier-Stokes and Euler equations at all Mach number. The existence of a discrete solution, the consistency of the scheme in the Lax sense and the positivity of the internal energy were proved. Then the scheme was extended to the homogeneous two-phase flow models of the GENEPI code developed at CEA. Lastly a multigrid-AMR algorithm was adapted for using our pressure-correction scheme on adaptive grids.Regarding the second issue addressed in this work, an extension to the Beavers-Joseph law was proposed for the convective regime. By introducing a jump in the kinetic energy at the interface, we recover an interface condition close to the Beavers-Joseph law but with a non-linear slip coefficient, which depends on the free-fluid velocity at the interface and on the Darcy velocity. The validity of this new transmission condition was assessed with direct numerical simulations at different Reynolds numbers.

With the rapid increase in available compute power and memory, and bolstered by the advent of efficient formulations and algorithms, the role of 3D full-wave computational methods for accurate modelling of complex electromagnetic (EM) structures has gained in significance. The range of problems includes Radar Cross Section (RCS) computation, analysis and design of antennas and passive microwave circuits, bio-medical non-invasive detection and therapeutics, energy harvesting etc. Further, with the rapid advances in technology trends like System-in-Package (SiP) and System-on-Chip (SoC), the fidelity of chip-to-chip communication and package-board electrical performance parameters like signal integrity (SI), power integrity (PI), electromagnetic interference (EMI) are becoming increasingly critical. Rising pin-counts to satisfy functionality requirements and decreasing layer-counts to maintain cost-effectiveness necessitates 3D full wave electromagnetic solution for accurate system modelling. Method of Moments (MoM) is one such widely used computational technique to solve a 3D electromagnetic problem with full-wave accuracy. Due to lesser number of mesh elements or discretization on the geometry, MoM has an advantage of a smaller matrix size. However, due to Green's Function interactions, the MoM matrix is dense and its solution presents a time and memory challenge. The thesis focuses on formulation and development of novel techniques that aid in fast MoM based electromagnetic solutions. With the recent paradigm shift in computer hardware architectures transitioning from single-core microprocessors to multi-core systems, it is of prime importance to parallelize the serial electromagnetic formulations in order to leverage maximum computational benefits. Therefore, the thesis explores the possibilities to expedite an electromagnetic simulation by scalable parallelization of near-linear complexity algorithms like Fast Multipole Method (FMM) on a multi-core platform. Secondly, with the best of parallelization strategies in place and near-linear complexity algorithms in use, the solution time of a complex EM problem can still be exceedingly large due to over-meshing of the geometry to achieve a desired level of accuracy. Hence, the thesis focuses on judicious placement of mesh elements on the geometry to capture the physics of the problem without compromising on accuracy- a technique called Adaptive Mesh Refinement. This facilitates a reduction in the number of solution variables or degrees of freedom in the system and hence the solution time. For multi-scale structures as encountered in chip-package-board systems, the MoM formulation breaks down for parts of the geometry having dimensions much smaller as compared to the operating wavelength. This phenomenon is popularly known as low-frequency breakdown or low-frequency instability. It results in an ill-conditioned MoM system matrix, and hence higher iteration count to converge when solved using an iterative solver framework. This consequently increases the solution time of simulation. The thesis thus proposes novel formulations to improve the spectral properties of the system matrix for real-world complex conductor and dielectric structures and hence form well-conditioned systems. This reduces the iteration count considerably for convergence and thus results in faster solution. Finally, minor changes in the geometrical design layouts can adversely affect the time-to-market of a commodity or a product. This is because the intermediate design variants, in spite of having similarities between them are treated as separate entities and therefore have to follow the conventional model-mesh-solve workflow for their analysis. This is a missed opportunity especially for design variant problems involving near-identical characteristics when the information from the previous design variant could have been used to expedite the simulation of the present design iteration. A similar problem occurs in the broadband simulation of an electromagnetic structure. The solution at a particular frequency can be expedited manifold if the matrix information from a frequency in its neighbourhood is used, provided the electrical characteristics remain nearly similar. The thesis introduces methods to re-use the subspace or Eigen-space information of a matrix from a previous design or frequency to solve the next incremental problem faster.