We study positive solutions to nonlinear elliptic systems of the form: \begin{eqnarray*} -\Delta u =\lambda f(v) \mbox{ in }\Omega\\-\Delta v =\lambda g(u) \mbox{ in }\Omega\\\quad~~ u=0=v \mbox{ on }\partial\Omega \end{eqnarray*} where $\Delta u$ is the Laplacian of $u$, $\lambda$ is a positive parameter and $\Omega$ is a bounded domain in $R^n$ with smooth boundary $\partial\Omega$. In particular, we will analyze the combined effects of the nonlinearities on the existence and multiplicity of positive solutions. We also study systems with multiparameters and stronger coupling. We extend our results to $p$-$q$-Laplacian systems and to $n\times n$ systems. We mainly use sub- and super-solutions to prove our results.

We study computational theory and methods for finding multiple unstable solutions (corresponding to saddle points) to three types of nonlinear variational elliptic systems: cooperative, noncooperative, and Hamiltonian. We first propose a new Lorthogonal selection in a product Hilbert space so that a solution manifold can be defined. Then, we establish, respectively, a local characterization for saddle points of finite Morse index and of infinite Morse index. Based on these characterizations, two methods, called the local min-orthogonal method and the local min-max-orthogonal method, are developed and applied to solve those three types of elliptic systems for multiple solutions. Under suitable assumptions, a subsequence convergence result is established for each method. Numerical experiments for different types of model problems are carried out, showing that both methods are very reliable and efficient in computing coexisting saddle points or saddle points of infinite Morse index. We also analyze the instability of saddle points in both single and product Hilbert spaces. In particular, we establish several estimates of the Morse index of both coexisting and non-coexisting saddle points via the local min-orthogonal method developed and propose a local instability index to measure the local instability of both degenerate and nondegenerate saddle points. Finally, we suggest two extensions of an L-orthogonal selection for future research so that multiple solutions to more general elliptic systems such as nonvariational elliptic systems may also be found in a stable way.

An analytic boundary layer model for thermal convection with a finite-strength plate and depth-dependent viscosity is developed. The model includes a specific energy balance for the lithosphere and accounts for coupling between the plate and underlying mantle. Multiple solutions are possible with three solution branches representing three distinct modes of thermal convection. One branch corresponds to the classic boundary layer solution for active lid plate tectonics while two new branches represent solutions for sluggish lid convection. The model is compared to numerical simulations with highly temperature dependent viscosity and is able to predict both the type of convection (active, sluggish, or stagnant lid) as well as the presence of single and multiple solution regimes. The existence of multiple solutions suggests that the mode of planetary convection may be history dependent. The dependence of mantle viscosity on temperature and water concentration is found to introduce a strong dynamic feedback with plate tectonics. A dimensionless parameter is defined to quantitatively evaluate the relative strength of this feedback and demonstrates that water and heat transport may be equally important in controlling present-day platemantle dynamics for the Earth. A simple parameterized evolution model illustrates the feedback and agrees well with our analytic results. This suggests that a simple relationship may exist between the rate of change of water concentration and the rate of change of temperature in the mantle. This study concludes by investigating the possibility of a magnetic field dynamo in early solar system planetesimals. The thermal evolution of planetesimals is modeled by considering melting, core formation, and the onset of mantle convection and then employing thermal boundary layer theory for stagnant lid convection (if possible) to determine the cooling rate of the body. We assess the presence, strength and duration of a dynamo for a range of planetesimal sizes and other parameters. We find that a minimum radius of O(500) km is required for a thermally driven dynamo of duration O(10) My. The dependence of the results on model parameters is made explicit through the derivation of an analytic solution. / Earth and Planetary Sciences

D'un point de vue industriel, la mise en place de nouvelles architectures de systèmes mécaniques nécessite un long processus de conception permettant de définir et d'anticiper le comportement. Dans le cas particulier des systèmes aéronautiques tels que les moteurs d'avions, un certain nombre de pièces sont particulièrement sensibles car elles doivent répondre à des impératifs stricts en termes d'encombrement, de performance et de tenue mécanique. Dans ce contexte, la prévision du comportement vibratoire revêt une importance particulière puisqu'elle permet d'évaluer le niveau des sollicitations cycliques appliquées sur le système et guide ainsi la détection en amont d'éventuels problèmes de fatigue des matériaux. La plupart du temps, des modèles numériques sont utilisés pour représenter les structures, et le comportement est simulé en résolvant un ensemble d'équations. Pour atteindre un niveau de détail répondant au besoin industriel, ces modèles peuvent être particulièrement gros, et la résolution des équations associées demande des ressources et des temps de calcul considérables. De plus, pour rendre compte au mieux des comportements observés expérimentalement, il est souvent nécessaire de prendre en compte des phénomènes non-linéaires, ce qui augmente encore la difficulté. Les travaux présentés dans ce manuscrit concernent cette problématique du comportement vibratoire des structures non-linéaires et s'orientent autour de deux axes : la réduction de modèle et le calcul des solutions multiples. L'objectif du premier axe est de contribuer à la construction de modèles numériques non linéaires réduits utilisables en conception de systèmes industriels et de proposer des outils d'exploitation et d'interprétation de ces modèles. En particulier, on considère le cas des méthodes de projection de Galerkin et on montre qu'elles sont à même de construire des modèles réduits réalistes. Des méthodes complémentaires de réduction de modèles sont également présentées dans le cas particulier de la recherche de solutions par la méthode de la balance harmonique (HBM) : on s'intéressera en particulier à des méthodes de sélection d'harmoniques. Après avoir comparé les différentes méthodes proposées sur un exemple simple de poutre non-linéaire, elles sont appliquées à un modèle de structure industrielle représentant une aube d'hélice d'open rotor. Le second axe de ces travaux concerne le calcul de solutions multiples pour les systèmes dynamiques non-linéaires. Une particularité de ces systèmes est en effet de présenter plusieurs configurations stables pour un état de sollicitation donné. Il s'agira ici de proposer des méthodes de calcul permettant de dresser la liste exhaustive des solutions possibles. Le travail présenté se concentre sur la recherche de solutions périodiques par la méthode de la balance harmonique pour des systèmes possédant des non-linéarités polynomiales. Ces restrictions conduisent à la résolution de systèmes polynomiaux pour lesquels il existe des méthodes permettant de calculer l'ensemble des solutions. En particulier, on propose l'utilisation originale de méthodes basées sur le calcul de bases de Groebner pour la résolution de systèmes polynomiaux issus de la mécanique. Les différentes méthodes présentées sont illustrées et comparées sur des exemples simples. Les résultats montrent que même pour des systèmes simples, le comportement dynamique peut être très complexe. / In an industrial context, the design of new mechanical systems requires long design processes in order to define and to anticipate the behavior of all the constitutive parts. In the particular case of aeronautical structures such as plane engines, design is especially critical since they have to meet various and strict needs (life duration, performances . . .). Then, anticipating vibratory behavior is very important as this provides information about cyclic solicitations and fatigue. Most often, numerical models are used to mimic the structure and mechanical behavior is simulated by solving a set of differential equations. In the case of industrial structures, such models can be quite large and their resolution very time-consuming. Moreover, in order to model experimental behavior realistically, it is often necessary to take nonlinear phenomena into account and thus increase the required computational effort. The work presented in this PhD deals with the study of mechanical nonlinear systems. It focuses on two principal directions : model reduction and multiple solutions computation. The goal of the first direction is to contribute to the building of numerical reduced order models usable in industrial context and to propose tools to exploit an interpret them. Particularly, Galerkin projection methods are investigated in the context of nonlinear systems reduction, showing that those methods are, under certain conditions, able to give a reliable picture of full system behavior. In the case of the harmonic balance method, complementary methods are also proposed to reduce the size of the algebraic equations system by using harmonic selection techniques. The presented methods are firstly illustrated and compared on a simple nonlinear beam example ; they are then applied to an industrial model of open rotor blade. The second direction of this work deals with the computation of multiple solutions arising in nonlinear dynamical systems. Indeed, it has been shown that such systems can present different stable configurations for a given solicitation. The objective here is to provide tools for computing such multiple solutions. We only consider the case of periodic solutions for systems with polynomial nonlinearities, treated with harmonic balance method. These hypotheses enable one to search for multiple states as solutions of polynomial algebraic systems of equations, for which some methods exist to compute the entire set of solutions. In particular, we propose to use methods relying on Groebner basis computation, in order to compute the whole set of solutions. The proposed methods are illustrated and compared on simple examples, showing that even such simple systems can present very complex dynamical behavior.

L'objectif de ce travail est d'étudier les détails des écoulements et des transferts thermiques induits par convection naturelle au sein d'un canal (simulation bidimensionnelle) et au sein d'une cavité cubique (simulation tridimensionnelle) munis de blocs chauffants. La chaleur dégagée par les blocs résulte d'une génération volumique uniforme de la chaleur, d'une température chaude constante ou d'un flux surfacique uniforme. On utilise l'air (Pr=0.72) comme fluide. Les paramètres de l'étude sont le rapport des conductivités thermiques du bloc solide et du fluide (0.1≤k*≤200), le nombre de Rayleigh (〖10〗^4≤Ra≤〖10〗^7) et la hauteur relative des blocs (1/8≤B≤1/2). La détermination des conditions de validité du modèle à blocs isothermes et du modèle à blocs libérant un flux surfacique uniforme, en fonction du rapport des conductivités thermiques et des autres paramètres du problème, compte parmi les principaux objectifs de ce travail. Les résultats obtenus montrent que l'écoulement et le transfert thermique ainsi que les conditions de validité des modèles mentionnés sont fortement influencés par les paramètres de contrôle et par la multiplicité de solutions trouvée en régime convectif. / The object of this work is to study the details of the flow and heat transfer induced by natural convection in a channel (2D simulation) and in a cubic cavity (3D simulation) containing heating blocks. The heat released by the blocks results from a uniform volumetric heat generation, a constant hot temperature or a uniform surface flux. Air (Pr=0.72) is used as working fluid. The parameters of the study are the thermal conductivities ratio of solid blocks and fluid (0.1≤k*≤200), the Rayleigh number (〖10〗^4≤Ra≤〖10〗^7) and the relative height of the blocks (1/8≤B≤1/2). Among the main objects of this work is the determination of the conditions of validity of the model with isothermal blocks and the model with blocks releasing a uniform surface flux, as functions of the thermal conductivities ratio and the other parameters of the problem. The results obtained show that the flow and heat transfer and the conditions of validity of the models mentioned are strongly affected by the control parameters and the multiplicity of solutions found in the convective regime.

Orientador: Dílson Amâncio Alves / Banca: Carlos Roberto Minussi / Banca: Francisco Villarreal Alvarado / Banca: Vivaldo Fernando da Costa / Banca: Walmir de Freitas Filho / Resumo: O fluxo de carga convencional é considerado inadequado para a obtenção do ponto de máximo carregamento devido a singularidade da matriz Jacobiana. Os métodos da continuação são ferramentas eficientes para a solução deste tipo de problema, e diferentes parametrizações são utilizadas para evitar a singularidade da matriz. Neste trabalho apresentase uma técnica de parametrização geométrica que possibilita o traçado completo das curvas PV sem os problemas de mal condicionamento. A técnica proposta associa a robustez com a simplicidade e a facilidade de compreensão. A singularidade da matriz Jacobiana é eliminada pela adição da equação de uma reta que passa por um ponto no plano formado pelas variáveis perdas de potência ativa totais e o fator de carregamento, dois parâmetros físicos de fácil compreensão. A técnica, aplicada aos sistemas do IEEE (14, 30, 57, 118 e 300 barras) e ao sistema brasileiro sul-sudeste (638 e 787 barras), mostra que as características do fluxo de carga não só são preservadas, mas também melhoradas. Diversos testes são realizados para proporcionar a comparação do desempenho do esquema de parametrização proposto para o método do fluxo de carga continuado. / Abstract: The conventional Newton's method has been considered inadequate to obtain the maximum loading point of power systems due to the Jacobian matrix singularity. Continuation methods are efficient tools for solving this kind of problem, and different parameterizations are used to avoid the matrix singularity. This paper presents a new geometric parameterization scheme that allows the complete tracing of the PV curves without ill-conditioning problems. The proposed technique associates the robustness to the simplicity and easy understanding. The Jacobian matrix singularity is overcome by the addition of a line equation, which passes through a point in the plane, determined by the real power losses and loading factor variables, two parameters with clear physical meaning. The application of this new technique to the IEEE systems (14, 30, 57, 118 and 300 buses) and to the Brazilian systems south-southeast (638 and 787 buses) shows that the characteristics of the conventional Newton's method are not only preserved but also improved. Several tests are carried out to compare the performance of the proposed parameterization scheme for the continuation power flow method. / Doutor

Made available in DSpace on 2014-06-11T19:30:50Z (GMT). No. of bitstreams: 0 Previous issue date: 2008-11-06Bitstream added on 2014-06-13T19:00:48Z : No. of bitstreams: 1 garbelini_e_dr_ilha.pdf: 758680 bytes, checksum: 9f08146d542679d2c1da479912ffd576 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O fluxo de carga convencional é considerado inadequado para a obtenção do ponto de máximo carregamento devido a singularidade da matriz Jacobiana. Os métodos da continuação são ferramentas eficientes para a solução deste tipo de problema, e diferentes parametrizações são utilizadas para evitar a singularidade da matriz. Neste trabalho apresentase uma técnica de parametrização geométrica que possibilita o traçado completo das curvas PV sem os problemas de mal condicionamento. A técnica proposta associa a robustez com a simplicidade e a facilidade de compreensão. A singularidade da matriz Jacobiana é eliminada pela adição da equação de uma reta que passa por um ponto no plano formado pelas variáveis perdas de potência ativa totais e o fator de carregamento, dois parâmetros físicos de fácil compreensão. A técnica, aplicada aos sistemas do IEEE (14, 30, 57, 118 e 300 barras) e ao sistema brasileiro sul-sudeste (638 e 787 barras), mostra que as características do fluxo de carga não só são preservadas, mas também melhoradas. Diversos testes são realizados para proporcionar a comparação do desempenho do esquema de parametrização proposto para o método do fluxo de carga continuado. / The conventional Newton’s method has been considered inadequate to obtain the maximum loading point of power systems due to the Jacobian matrix singularity. Continuation methods are efficient tools for solving this kind of problem, and different parameterizations are used to avoid the matrix singularity. This paper presents a new geometric parameterization scheme that allows the complete tracing of the PV curves without ill-conditioning problems. The proposed technique associates the robustness to the simplicity and easy understanding. The Jacobian matrix singularity is overcome by the addition of a line equation, which passes through a point in the plane, determined by the real power losses and loading factor variables, two parameters with clear physical meaning. The application of this new technique to the IEEE systems (14, 30, 57, 118 and 300 buses) and to the Brazilian systems south-southeast (638 and 787 buses) shows that the characteristics of the conventional Newton’s method are not only preserved but also improved. Several tests are carried out to compare the performance of the proposed parameterization scheme for the continuation power flow method.