Ce travail présente une méthode mixte expérimentale-numérique traitant des mesures mécaniques réalisées sur structures pour l’identification de propriétés matériau viscoélastiques linéaires. Les caractéristiques de ces matériaux sont complexes et dépendent de la fréquence et nous cherchons à les identifier par une approche directe et non paramétrique. Pour cela, deux méthodes d’identification adaptées au traitement des mesures de champs de réponses en fréquences sont présentées. Cellesci utilisent les données fréquentielles, sans analyse modale expérimentale. Les champs expérimentaux sont obtenus par vibromètrie laser et les champs numérique par la méthode des éléments finis. Nous utilisons un modèle volumique adapté à l’étude des structures minces et épaisses pour l’analyse numérique. Les applications se basent sur des plaques simples ou multicouches. Les résultats sont corrélés avec des mesures DMA et des identifications à partir de données modales identifiées. / This work deals with a mixed experimental numerical method for linear viscoelastic material parameter identification. These parameters are complex and frequency dependent. Identication is performed using a direct and non parametric approach. In this goal, two identification methods adapted to frequency response fields measurements treatment are presented. These datas are durectly exploited, without any experimental modal analysis. Experimental fields are obtained using a vibrometer and numerical ones with finite element method. A brick element dedicated to thin/thick structures analysis is employed. Appications are performed using simple plates or sandwich plates. Results are also compared to DMA measurements and identified modal parameters.

Le soudage multipasse est mis en œuvre pour recharger des tuyauteries présentant localement des sous-épaisseurs. La simulation numérique facilite le choix des nombreux paramètres de soudage. La réduction des modèles permet d'accélérer ces choix. Dans ce travail, nous nous sommes intéressés aux cas pour lesquelles il est difficile de réaliser intégralement la simulation du soudage, faute de temps ou par manque de moyens de calcul. Ce sont des simulations hors d'atteintes. Or, les prévisions manquantes ne permettent pas la mise en œuvre d'une méthode de décomposition orthogonale aux valeurs propres pour extraire une base réduite de modes empiriques à partir des données produites par simulation numérique. Nous proposons donc soit un modèle directionnel bien adapté au soudage, soit une étape d'extrapolation des données de simulations par décalage spatial des prévisions calculées. Ces deux approches sont complémentaires de la méthode d'hyper-réduction, dans laquelle les équations de bilan sont restreintes à un maillage réduit. Ces méthodes permettent de démarrer une simulation numérique du soudage avec un modèle éléments finis, puis de poursuivre cette simulation par un modèle hyper-réduit. Cela évite d'avoir à réaliser de nombreuses études paramétriques et permet de traiter des simulations qui sont hors d'atteintes. Ce mémoire se termine par un chapitre traitant du cas de rechargement d'un tube, pour lequel EDF a mis en œuvre un essai instrumenté. / Multi-pass welding is used to recharge pipes with local sub-thickness. Numerical simulation facilitates the selection of many welding parameters. Reducing the order of models speeds up these choices. In this work, we were interested in cases where it is difficult to carry out the entire welding simulation due to time constraints or lack of calculation means. These computations are called out of reach simulations. However, the missing forecasts do not allow the implementation of a orthogonal decomposition method to extract a reduced basis of empirical modes from the data produced by numerical simulations. To overcome this difficulty, we propose either a directional model well adapted to welding, or a step of extrapolation of the simulation data by spatial shift of the already calculated forecasts. These two approaches are complementary to the hyper-reduction method, in which the balance equations are restricted to a reduced mesh size. These methods allow to start a numerical simulation of welding with a finite element model, then to continue this simulation with a hyper-reduced model. This avoids the need for numerous preliminary parametric studies and allows simulations that are out of reach. This manuscript ends with a chapter dealing with the case of reloading a tube, for which EDF has carried out an instrumented test.

O principal objetivo deste trabalho é estudar as singularidades e hipersingularidades existentes nas formulações: singular - clássica - e hipersingular no Método dos Elementos de Contorno (MEC). Também é proposto um esquema residual h-adaptativo para a solução numérica do problema físico governado pela equação de Laplace. Usa-se malha poligonal, juntamente, com funções de interpolação - distribuição - de forma, dos tipos: constantes e lineares. Para controlar o erro a posteriori, é considerado o valor do resíduo, fora dos pontos de colocação. Também é testada uma técnica de quadratura numérica chamada adaptativa, específica para subelementos, no sentido de verificar se a precisão no cálculo das integrais com singularidades é melhorada. O uso de funções hierárquicas é discutido na forma de um algoritmo para atualização da matriz principal do sistema linear. / The main purpose of this work is to study the existing singularities and hypersingularities in the Boundary Element Method (BEM) with singular - classical - and hypersingular formulations. Also, an h-adaptive residual scheme for the numerical solution of the physical problem, driven by Laplace equation, is proposed. Boundary polygonal mesh, with constant and linear interpolation - distribution - shape functions together are used. To control the a posteriori error, is considered the residue value outside the collocation points. Also, a sub-element specific adaptive numerical quadrature technique, in an effort to verify if the precision when dealing with integrals possessing singularities is increased, is tested. The use of hierarchical functions is discussed, as an algorithm to update the linear system main matrix.

Do ponto de vista da hidrodinâmica clássica, o problema de impacto hidrodinâmico configura-se como um problema de contorno com fronteiras móveis cuja posição deve ser determinada simultaneamente à solução da equação de campo. Essa característica traz dificuldades para obtenção de soluções analíticas e numéricas. Nesse sentido, o presente trabalho propõe o desenvolvimento de um método numérico específico para analisar o problema de impacto hidrodinâmico de corpos sólidos rígidos contra a superfície livre da água. A solução da equação dinâmica não linear do problema de impacto depende da determinação do tensor de massa adicional a cada instante de tempo, o qual depende da posição e atitude do corpo no instante considerado. Um método variacional específico é empregado, através do qual os coeficientes de massa adicional são determinados com erro de segunda ordem, na posição considerada. Tal método é exemplo de técnicas numéricas dessingularizadas, através das quais o potencial de velocidade é aproximado em um espaço finito-dimensional formado por funções-teste derivadas de soluções potenciais elementares, tais como pólos, dipolos, anéis de dipolos, de vórtices, etc. O problema potencial de impacto hidrodinâmico, que se caracteriza pela dominância das forças de inércia, é formulado admitindo-se a superfície líquida como equipotencial, o que permite a analogia com o limite assintótico de frequência infinita do problema de radiação de ondas causada pelo movimento de corpos flutuantes. O método desenvolvido é então aplicado ao caso de impacto vertical de corpos axissimétricos, formulando o problema sob o chamado modelo de von Kármán generalizado (GvKM). Nesse modelo as condições de contorno na geometria exata do corpo são satisfeitas, porém os efeitos do empilhamento de água junto às raízes do jato, que se forma ao longo da intersecção com a superfície livre, não são considerados no caso geral. Resultados numéricos do coeficiente de massa adicional para uma família de esferoides são apresentados e tabulados para o pronto uso em análise e projeto. Além disso, considerações acerca da inclusão do efeito de empilhamento de água junto às raízes do jato, ou seja, da elevação da superfície livre são também feitas para o caso de esferas, fazendo uso de abordagens analíticas encontradas na literatura especializada. / In terms of classical hydrodynamics, the hydrodynamic impact problem is characterized as a boundary problem with moving boundary which position must be determined simultaneously with the solution of the field equation. This feature brings difficulties to get analytical and numerical solutions. In this sense, the purpose of this work is to present a variational method technique specifically designed for the hydrodynamic impact problem of axisymmetric rigid bodies on the free surface. The solution of the nonlinear dynamic equation of the impacting motion depends on the determination of the added mass tensor and its derivative with respect to time at each integration time step. This is done through a variational method technique that leads to a second-order error approximation for the added mass if a first-order error approximation is sought for the velocity potential. This method is an example of desingularized numerical techniques, through which the velocity potential is approximated in a sub-space of finite dimension, formed by trial functions derived from elementary potential solutions, such as poles, dipoles, and vortex rings, which are placed inside the body. The potential problem of hydrodynamic impact, characterized by the dominance of inertial forces, is here formulated by assuming the liquid surface as equipotential, what allows the analogy with the infinity frequency limit in the usual free surface oscillating floating body problem. The method is applied to the vertical hydrodynamic impact of axisymmetric bodies within the so-called Generalized von Kármán Model (GvKM). In such approach, the exact body boundary condition is full-filled and the wet correction is not taken into account. Numerical results for the added mass coefficient for a family of spheroids are presented. Moreover, considerations are made on the effects of the free surface elevation for the specific case of an impacting sphere, through analytical approaches.

La maîtrise du bruit et des vibrations est un objectif fréquemment rencontré dans le domaine industriel. Qu’il s’agisse de questions de confort ou de sécurité, les domaines d’applications sont nombreux et variés : transport, BTP, ingénierie civile et militaire… Dans cette thèse, un problème de vibroacoustique interne avec couplage fluide-structure est étudié. Il s’agit d’une cavité remplie de fluide dont les parois sont constituées d’une structure sandwich viscoélastique. Les difficultés numériques associées à ce modèle portent sur la non linéarité du matériau et sur les propriétés des opérateurs matriciels manipulés (conditionnement, non symétrie). Le calcul des vibrations du système dissipatif couplé nécessite une valeur initiale, choisie comme la solution du problème conservatif. Cette solution n’étant pas aisée à déterminer, deux solveurs aux valeurs propres basés sur la Méthode Asymptotique Numérique (MAN) sont proposés pour résoudre le problème des vibrations libres du système conservatif. Associant des techniques de perturbation d'ordre élevé et de continuation, la MAN permet de transformer le problème non linéaire de départ en une suite de problèmes linéaires, plus simples à résoudre. Les solutions obtenues sont ensuite utilisées comme point initial pour déterminer la réponse libre du système dissipatif. Un solveur de Newton d’ordre élevé, basé sur les techniques d’homotopie et de perturbation est développé pour résoudre ce problème. Enfin, le régime forcé est étudié. Pour toutes les configurations envisagées, les résultats obtenus mettent en évidence des performances numériques améliorées par rapport aux méthodes classiquement utilisées (Arpack, Newton…). / Noises and vibrations control is a common objective in the industrial field. Whether it is a question of comfort or safety, the fields of application are numerous and varied: transport, building, civil and military engineering… In this thesis, a vibroacoustics interior problem with fluid-structure coupling is studied. A cavity filled of fluid whose walls are made of a sandwich viscoelastic structure is considered. The numerical difficulties associated with this model relate to the non-linearity of the viscoelastic material and the properties of the matrix operators used (conditioning, non-symmetry). The calculation of the vibrations of the coupled dissipative system requires an initial value, chosen as the solution to the conservative problem. Since this solution is difficult to determine, two eigenvalue algorithms based on the Asymptotic Numerical Method (ANM) are proposed to solve the problem of free vibrations of the conservative system. Combining high order perturbation and continuation techniques, ANM transforms the initial non-linear problem into a set of linear problems that are easier to solve. The solutions obtained are then used as the initial point to determine the free vibrations of the dissipative problem. A high order Newton solver, based on homotopy and perturbation techniques, is developed to solve this problem. Finally, the forced harmonic response of the damped system is computed. For all the configurations tested, the results obtained show improved numerical performance compared to the methods conventionally used (Arpack solver, Newton algorithm…).

L'objectif ultime de ce travail est de fournir des méthodologies robustes et efficaces sur le plan des calculs pour la modélisation et la résolution des problèmes de contact adhésifs basés sur le potentiel de Lennard-Jones (LJ). Pour pallier les pièges théoriques et numériques du modèle LJ liés à ses caractéristiques nondéfinies et non-bornées, une méthode d'adaptativité en modèle est proposée pour résoudre le problème purement-LJ comme limite d'une séquence de problèmes multiniveaux construits de manière adaptative. Chaque membre de la séquence consiste en une partition modèle entre le modèle microscopique LJ et le modèle macroscopique de Signorini. La convergence de la méthode d'adaptativité est prouvée mathématiquement sous certaines hypothèses physiques et réalistes. D'un autre côté, la méthode asymptotique numérique (MAN) est adaptée et utilisée pour suivre avec précision les instabilités des problèmes de contact à grande échelle et souples. Les deux méthodes sont incorporées dans le cadre multiéchelle Arlequin pour obtenir une résolution précise, tout en réduisant les coûts de calcul. Dans la méthode d'adaptativité en modèle, pour capturer avec précision la localisation des zones d'intérêt (ZDI), une stratégie en deux résolutions est suggérée : une résolution macroscopique est utilisée comme une première estimation de la localisation de la ZDI. La méthode Arlequin est alors utilisée pour obtenir une résolution microscopique en superposant des modèles locaux aux modèles globaux. En outre, dans la stratégie MAN, la méthode Arlequin est utilisée pour supprimer les oscillations numériques, améliorer la précision et réduire le coût de calcul. / The ultimate goal of this work is to provide computationally efficient and robust methodologies for the modelling and solution of a class of Lennard-Jones (LJ) potential-based adhesive contact problems. To alleviate theoretical and numerical pitfalls of the LJ model related to its non-defined and nonbounded characteristics, a model-adaptivity method is proposed to solve the pure-LJ problem as the limit of a sequence of adaptively constructed multilevel problems. Each member of the sequence consists of a model partition between the microscopic LJ model and the macroscopic Signorini model. The convergence of the model-adaptivity method is proved mathematically under some physical and realistic assumptions. On the other hand, the asymptotic numerical method (ANM) is adapted to track accurately instabilities for soft contact problems. Both methods are incorporated in the Arlequin multiscale framework to achieve an accurate resolution at a reasonable computational cost. In the model-adaptivity method, to capture accurately the localization of the zones of interest (ZOI), a two-step strategy is suggested: a macroscopic resolution is used as the first guess of the ZOI localization, then the Arlequin method is used there to achieve a fine scale resolution. In the ANM strategy, the Arlequin method is also used to suppress numerical oscillations and improve accuracy.

L'étude par continuation des solutions périodiques et quasi-périodiques est appliquée à plusieurs modèles issus du violon. La continuation pour un modèle à un degré de liberté avec friction régularisée permet de montrer la préservation, par rapport à la friction de Coulomb, des bifurcations de cycle limite (une vitesse maximale et une force minimale permettant le mouvement de Helmholtz) et de propriétés globales de la branche de solution (croissance de l'amplitude avec la vitesse, décroissance de la fréquence avec la force normale). L'équilibrage harmonique est évalué sur la friction régularisée et a des propriétés de convergence intéressantes (erreur faible, monotone, à décroissance rapide). La continuation sur un modèle à deux modes donne accès aux solutions de registres supérieurs, dont la stabilité coïncide avec l'expérience. La valeur retenue pour l'inharmonicité peut modifier fortement le diagramme de bifurcation. Une nouvelle méthode de continuation des solutions quasi-périodiques est proposée. Elle associe l'EH étendu à deux pulsations avec la Méthode Asymptotique Numérique. Une attention particulière est portée à la rapidité des calculs, face à la croissance rapide de la taille des systèmes à inverser. Un modèle de friction prenant en compte la température au point de contact est reformulé à l'aide d'une dérivée fractionnaire. Nous proposons une méthode de continuation de solutions périodiques de systèmes contenant des dérivées ou intégrales fractionnaires. Nous établissons une condition suffisante pour que les cycles asymptotiques du cadre causal (Caputo) soient solutions du cadre que nous avons choisi. / The continuation of periodic and quasi-periodic solutions is performed on several models derived from the violin. The continuation for a one degree-of-freedom model with a regularized friction shows, compared with Coulomb friction, the persistence of limit cycle bifurcations (a maximum bow speed and a minimum normal force allowing Helmholtz motion) and of global properties of the solution branch (increase of amplitude with respect to the bow speed, decrease of frequency with respect to the normal force). The Harmonic Balance Method is assessed on this regularized friction system and shows interesting convergence properties (the error is low, monotone and rapidly decreasing). For two modes the continuation shows higher register solutions with a plausible stability. A stronger inharmonicity can greatly modify the bifurcation diagram. A new method is proposed for the continuation of quasi-periodic solutions. It couples a two-pulsations HBM with the Asymptotic Numerical Method. We have taken great care to deal efficiently with large systems of unknowns. A model of friction that takes into account temperature of the contact zone is reformulated with a fractional derivative. We then propose a method of continuation of periodic solutions for differential systems that contain fractional operators. Their definition is usually restricted to causal solutions, which prevents the existence of periodic solutions. Having chosen a specific definition of fractional operators to avoid this issue we establish a sufficient condition on asymptotically attractive cycles in the causal framework to be solutions of our framework.

<p>Quantum dynamics is the study of time-dependent phenomena in fundamental processes of atomic and molecular systems. This thesis focuses on systems where nature reveals its quantum aspect; e.g. in vibrational resonance structures, in wave packet revivals and in matter wave interferometry. Grid based numerical methods for solving the time-dependent Schrödinger equation are implemented for simulating time resolved molecular vibrations and to compute photo-electron spectra, without the necessity of diagonalizing a large matrix to find eigenvalues and eigenvectors.</p><p>Pump-probe femtosecond laser spectroscopy on the sodium potassium molecule, showing a vibrational period of 450 fs, is theoretically simulated. We find agreement with experiment by inclusion of the finite length laser pulse and finite temperature effects.</p><p>Complicated resonance structures observed experimentally in photo-electron spectra of hydrogen- and deuterium chloride is analyzed by a numerical computation of the spectra. The dramatic difference in the two spectra arises from non-adiabatic interactions, i.e. the interplay between nuclear and electron dynamics. We suggest new potential curves for the 3²Σ⁺ and 4²Σ⁺ states in HCI⁺.</p><p>It is possible to guide slow atoms along magnetic potentials like light is guided in optical fibers. Quantum mechanics dictates that matter can show wave properties. A proposal for a multi mode matter wave interferometer on an atom chip is studied by solving the time-dependent Schrödinger equation in two dimensions. The results verifies a possible route for an experimental realization.</p><p>An improved representation for wave functions using a discrete set of coherent states is presented. We develop a practical method for computing the expansion coefficients in this non-orthogonal set. It is built on the concept of frames, and introduces an iterative method for computing a representation of the identity operator. The phase-space localization property of the coherent states gives adaptability and better sampling efficiency.</p>

The continuous trend in gas turbine design towards lighter, more powerful and more reliable engines on one side and use of alternative fuels on the other side renders flutter problems as one of the paramount challenges in engine design. Flutter denotes a self-excited and self-sustained aeroelastic instability phenomenon that can lead to material fatigue and eventually damage of structure in a short period of time unless properly damped. The design for flutter safety involves the prediction of unsteady aerodynamics as well as structural dynamics that is mostly based on in-house developed numerical tools. While high confidence has been gained on the structural side unanticipated flutter occurrences during engine design, testing and operation evidence a need for enhanced validation of aerodynamic models despite the degree of sophistication attained. The continuous development of these models can only be based on the deepened understanding of underlying physical mechanisms from test data. As a matter of fact most flutter test cases treat the turbomachine flow in two-dimensional manner indicating that the problem is solved as plane representation at a certain radius rather than representing the complex annular geometry of a real engine. Such considerations do consequently not capture effects that are due to variations in the third dimension, i.e. in radial direction. In this light the present thesis has been formulated to study three-dimensional effects during flutter in the annular environment of a low-pressure turbine blade row and to describe the importance on prediction of flutter stability. The work has been conceived as compound experimental and computational work employing a new annular sector cascade test facility. The aeroelastic response phenomenon is studied in the influence coefficient domain having one blade oscillating in various three-dimensional rigid-body modes and measuring the unsteady response on several blades and at various radial positions. On the computational side a state-of-the-art industrial numerical prediction tool has been used that allowed for two-dimensional and three-dimensional linearized unsteady Euler analyses. The results suggest that considerable three-dimensional effects are present, which are harming prediction accuracy for flutter stability when employing a two-dimensional plane model. These effects are mainly apparent as radial gradient in unsteady response magnitude from tip to hub indicating that the sections closer to the hub experience higher aeroelastic response than their equivalent plane representatives. Other effects are due to turbomachinery-typical three-dimensional flow features such as hub endwall and tip leakage vortices, which considerably affect aeroelastic prediction accuracy. Both effects are of the same order of magnitude as effects of design parameters such as reduced frequency, flow velocity level and incidence. Although the overall behavior is captured fairly well when using two-dimensional simulations notable improvement has been demonstrated when modeling fully three-dimensional and including tip clearance.

Quantum dynamics is the study of time-dependent phenomena in fundamental processes of atomic and molecular systems. This thesis focuses on systems where nature reveals its quantum aspect; e.g. in vibrational resonance structures, in wave packet revivals and in matter wave interferometry. Grid based numerical methods for solving the time-dependent Schrödinger equation are implemented for simulating time resolved molecular vibrations and to compute photo-electron spectra, without the necessity of diagonalizing a large matrix to find eigenvalues and eigenvectors. Pump-probe femtosecond laser spectroscopy on the sodium potassium molecule, showing a vibrational period of 450 fs, is theoretically simulated. We find agreement with experiment by inclusion of the finite length laser pulse and finite temperature effects. Complicated resonance structures observed experimentally in photo-electron spectra of hydrogen- and deuterium chloride is analyzed by a numerical computation of the spectra. The dramatic difference in the two spectra arises from non-adiabatic interactions, i.e. the interplay between nuclear and electron dynamics. We suggest new potential curves for the 32Σ+ and 42Σ+ states in HCI+. It is possible to guide slow atoms along magnetic potentials like light is guided in optical fibers. Quantum mechanics dictates that matter can show wave properties. A proposal for a multi mode matter wave interferometer on an atom chip is studied by solving the time-dependent Schrödinger equation in two dimensions. The results verifies a possible route for an experimental realization. An improved representation for wave functions using a discrete set of coherent states is presented. We develop a practical method for computing the expansion coefficients in this non-orthogonal set. It is built on the concept of frames, and introduces an iterative method for computing a representation of the identity operator. The phase-space localization property of the coherent states gives adaptability and better sampling efficiency.