<p>This investigation is motivated by the problem of optimal design of cooling elements in modern battery systems used in hybrid/electric vehicles. We consider a simple model of two-dimensional steady-state heat conduction generated by a prescribed distribution of heat sources and involving a one-dimensional cooling element represented by a closed contour. The problem consists in finding an optimal shape of the cooling element which will ensure that the temperature in a given region is close (in the least squares sense) to some prescribed distribution. We formulate this problem as PDE-constrained optimization and use methods of the shape-differential calculus to obtain the first-order optimality conditions characterizing the locally optimal shapes of the contour. These optimal shapes are then found numerically using the conjugate gradient method where the shape gradients are conveniently computed based on adjoint equations. A number of computational aspects of the proposed approach is discussed and optimization results obtained in several test problems are presented.</p> / Master of Science (MSc)

Dans cette thèse, nous calculons la forme d’un objet immergé d’aire donnée qui minimise la résistance de vague. Le corps, considéré lisse, avance à vitesse constante sous la surface libre d’un fluide qui est supposé parfait et incompressible. La résistance de vague est la traînée, c’est-à-dire la composante horizontale de la force exercée par le fluide sur l’obstacle. Nous utilisons les équations de Neumann-Kelvin 2D, qui s’obtiennent en linéarisant les équations d’Euler irrotationnelles avec surface libre. Le problème de Neumann-Kelvin est formulé comme une équation intégrale de frontière basée sur une solution fondamentale qui intègre la condition linéarisée à la surface libre. Nous utilisons une méthode de descente de gradient pour trouver un minimiseur local du problème de résistance de vague. Un gradient par rapport à la forme est calculé par la méthode de variation de frontières. Nous utilisons une approche level-set pour calculer la résistance de vague et gérer les déplacements de la frontière de l’obstacle. Nous obtenons une grande variété de formes optimales selon la profondeur de l’objet et sa vitesse. / In this thesis, we compute the shape of a fully immersed object with a given area which minimises the wave resistance. The smooth body moves at a constant speed under the free surface of a fluid which is assumed to be inviscid and incompressible. The wave resistance is the drag, i.e. the horizontal component of the force exerted by the fluid on the obstacle. We work with the 2D Neumann-Kelvin equations, which are obtained by linearising the irrotational Euler equations with a free surface. The Neumann-Kelvin problem is formulated as a boundary integral equation based on a fundamental solution which handles the linearised free surface condition. We use a gradient descent method to find a local minimiser of the wave resistance problem. A gradient with respect to the shape is calculated by a boundary variation method. We use a level-set approach to calculate the wave-making resistance and to deal with the displacements of the boundary of the obstacle. We obtain a great variety of optimal shapes depending on the depth of the object and its velocity.

Le travail présenté dans cette thèse consiste à modéliser les transports de particules dans un écoulement de Stokes en tenant compte des caractéristiques des particules, du régime d’écoulement de fluide qui les transporte et de l’effet de confinement qui pourrait être lié par exemple à la prise en compte d’une structure d’un milieu poreux. Le phénomène de transport que nous étudions est basé sur le mécanisme d’effet cliquet qui apparaît quand on soumet le système à un pompage alternatif dans le temps. L’étude de la dynamique de la particule est effectuée en utilisant la méthode de continuation qui permet de suivre la solution périodique selon le paramètre désiré et d’identifier les types de bifurcation ainsi que les lieux de ces points critiques. Cette étude nécessite la connaissance de la force de traînée exercée sur la particule et que nous calculons en utilisant la méthode des équations intégrales de frontières. / The work presented in this Phd Thesis is the modelling of particle transport in a Stokes flow taking into account the characteristics of the particle, the regime flow, and the effect of the confinement which could be related for example to the consideration of a porous media structure. The phenomena of the transport is based on the mecansim of ratchet effect which appears when the system is undergoes an alternative pumping. The study of the particle dynamics is performed using the continuation method which allows to follow the periodic solutions according to the desired parameter, to identify the types of bifurcation and critical points. This study requires the knowledge of the drag force exerted on the particle and which we compute using the boundary integral equation.

Dans cette thèse, nous menons à l'aide de la méthode des équations intégrales des analyses de sensitivité de solutions ou de spectres de l'équation de conductivité par rapport aux variations géométriques ou de paramètres de l'équation. En particulier, nous considérons le problème de conductivité dans des milieux à forts contrastes, le problème de perturbation du bord d'une inclusion de conductivité, le problème de valeurs propres du Laplacien dans des domaines perturbés et le problème d'ouverture de gap dans le spectre des cristaux photoniques.

Cette thèse a consisté à élaborer une méthode qui permet de résoudre l’équation intégrale comportant comme inconnues les courants et les charges introduite récemment par Taskinen et Ylä-Oijala par une méthode d’éléments frontière sans aucune contrainte de continuité au niveau des interfaces des éléments aussi bien pour les courants que pour les charges. Nous avons d’abord montré comment on pouvait construire cette équation de façon simple et similaire à celle des formulations intégrales usuelles en imposant au problème intérieur relatif au système de Picard, qui est en fait une extension du système de Maxwell, des conditions aux limites adéquates. Pour des géométries régulières de l’objet diffractant, nous avons établi de façon théorique la stabilité et la convergence des schémas numériques ci-dessus en montrant que cette équation peut être décomposée sous la forme d’un système elliptique coercif et d’un opérateur compact dans le cadre des fonctions de carré intégrable.Toute cette étude a été confirmée par des tests numériques tridimensionnels. Comme pour les équations intégrales usuelles de seconde espèce, le cadre théorique valable pour des surfaces régulières ne l’est plus pour des surfaces avec des singularités. L’utilisation formelle de cette équation,pour des surfaces singulières, a donné des résultats entachés d’erreur. Nous avons mis en évidence l’origine des instabilités numériques à l’origine de ces erreurs lorsque les géométries sont singulières en développant une version bidimensionnelle de cette équation. Cette version nous a permis en particulier de montrer que les instabilités étaient dues à des oscillations parasites concentrées autour des singularités de la géométrie. Dans ce cadre nous avons pu mettre en oeuvre plus aisément des approches pour supprimer ou atténuer ces oscillations parasites ou leur effet sur les calculs en champ lointain. Nous avons montré qu’un procédé d’augmentation des degrés de liberté pour la charge par rapport au courant pouvait sensiblement réduire ces instabilités. A la suite de l’amélioration observée sur les résultats dans le cas 2D, nous avons transposé cette procédure au cas tridimensionnel. A travers divers tests, nous avons constaté l’amélioration de la qualité de l’approximation amenée par la procédure de stabilisation / The objective of this thesis was to develop a method that solves the integral equation whose unknowns are the currents and the charges, recently introduced by Taskinen and Ylä-Oijala, by a boundary element method without any continuity constraint at the interfaces of the elements,for both the unknowns. We first show how to construct this equation in a simple way, similar tothe usual integral formulations, through imposing to the internal problem related to the Picard system,which is an extension of the Maxwell system, appropriate boundary conditions. For regular geometries, we have established a theoretical background ensuring the stability and the convergence of numerical scheme, by proving that this equation can be decomposed in a coercive elliptic and a compact parts in the context of square integrable functions. Our study was validated by three-dimensional numerical tests. In the case of usual integral equations of the second kind, the theoretical background for smooth surfaces is no longer valid when the surfaces is singular. The formal use of this equation for singular surfaces gave erroneous results. We pointed out the origin of numerical instabilities bydeveloping a two-dimensional version of this equation. This version has allowed us to show that the instabilities were due to parasitic oscillations accumulating on the geometrical singularities. In this context, we have implemented some approaches to reduce this parasitic oscillations on the calculations in the far field.We have shown that the method of increasing the freedom degrees for the charges relatively to the current could significantly reduces these instabilities. As a result, we have implemented this procedure in three-dimensional case. Throughout various tests, we noted the improvement on the approximation brough bay to the stabilization procedure

No description available.

This work addresses the question of the efficient numerical solution of time-domain boundary integral equations with retarded potentials arising in the problems of acoustic and electromagnetic scattering. The convolutional form of the time-domain boundary operators allows to discretize them with the help of Runge-Kutta convolution quadrature. This method combines Laplace-transform and time-stepping approaches and requires the explicit form of the fundamental solution only in the Laplace domain to be known. Recent numerical and analytical studies revealed excellent properties of Runge-Kutta convolution quadrature, e.g. high convergence order, stability, low dissipation and dispersion. As a model problem, we consider the wave scattering in three dimensions. The convolution quadrature discretization of the indirect formulation for the three-dimensional wave equation leads to the lower triangular Toeplitz system of equations. Each entry of this system is a boundary integral operator with a kernel defined by convolution quadrature. In this work we develop an efficient method of almost linear complexity for the solution of this system based on the existing recursive algorithm. The latter requires the construction of many discretizations of the Helmholtz boundary single layer operator for a wide range of complex wavenumbers. This leads to two main problems: the need to construct many dense matrices and to evaluate many singular and near-singular integrals. The first problem is overcome by the use of data-sparse techniques, namely, the high-frequency fast multipole method (HF FMM) and H-matrices. The applicability of both techniques for the discretization of the Helmholtz boundary single-layer operators with complex wavenumbers is analyzed. It is shown that the presence of decay can favorably affect the length of the fast multipole expansions and thus reduce the matrix-vector multiplication times. The performance of H-matrices and the HF FMM is compared for a range of complex wavenumbers, and the strategy to choose between two techniques is suggested. The second problem, namely, the assembly of many singular and nearly-singular integrals, is solved by the use of the Huygens principle. In this work we prove that kernels of the boundary integral operators $w_n^h(d)$ ($h$ is the time step and $t_n=nh$ is the time) exhibit exponential decay outside of the neighborhood of $d=nh$ (this is the consequence of the Huygens principle). The size of the support of these kernels for fixed $h$ increases with $n$ as $n^a,a<1$, where $a$ depends on the order of the Runge-Kutta method and is (typically) smaller for Runge-Kutta methods of higher order. Numerical experiments demonstrate that theoretically predicted values of $a$ are quite close to optimal. In the work it is shown how this property can be used in the recursive algorithm to construct only a few matrices with the near-field, while for the rest of the matrices the far-field only is assembled. The resulting method allows to solve the three-dimensional wave scattering problem with asymptotically almost linear complexity. The efficiency of the approach is confirmed by extensive numerical experiments.

Ο στόχος της παρούσας διδακτορικής διατριβής είναι διττός: η ανάπτυξη και η εφαρμογή αριθμητικών τεχνικών για την επίλυση προβλημάτων που εμπίπτουν στην περιοχή του Μη-Καταστροφικού Ελέγχου. Συγκεκριμένα αναπτύχθηκαν η Μέθοδος των Συνοριακών Στοιχείων (ΜΣΣ) και η Μέθοδος των Τοπικών Ολοκληρωτικών Εξισώσεων χωρίς Διακριτοποίηση για την αριθμητική ανάλυση στατικών και μεταβατικών προβλημάτων στο πεδίο της ελαστικότητας και της αλληλεπίδρασης ελαστικού με ακουστικό μέσο στις δύο διαστάσεις. Σημαντικό μέρος της διδακτορικής διατριβής αποτέλεσε η ανάπτυξη προγράμματος ηλεκτρονικού υπολογιστή, το οποίο επιλύει τα προβλήματα στα οποία πραγματεύεται το παρόν σύγγραμμα. Η διδακτορική διατριβή αποτελείται από τρεις ενότητες. Στην πρώτη ενότητα γίνεται πλήρης περιγραφή της απαραίτητης θεωρίας για την κάλυψη και κατανόηση των αριθμητικών ΜΣΣ αλλά και των Τοπικών Μεθόδων χωρίς Διακριτοποίηση (ΤΜχΔ). Στη δεύτερη ενότητα εφαρμόζονται οι προαναφερθείσες αριθμητικές μέθοδοι για την επίλυση στατικών και δυναμικών (στο πεδίο συχνοτήτων) διδιάστατων προβλημάτων, ώστε να πιστοποιηθεί η ακρίβεια και η αξιοπιστία των εν λόγω μεθοδολογιών. Τέλος, στην τρίτη ενότητα οι αριθμητικές ΜΣΣ και ΤΜχΔ εφαρμόζονται για την επίλυση προβλημάτων κυματικής διάδοσης που εμπίπτουν στο πεδίο του Μη-Καταστροφικού Ελέγχου. Πιο συγκεκριμένα μελετήθηκε η κυματική διάδοση σε ελεύθερες επίπεδες πλάκες και σε κυλινδρικές δεξαμενές αποθήκευσης υγρών καυσίμων. / The aim of this doctoral thesis is twofold: the development and implementation of numerical techniques for solving wave propagation problems in Non-Destructive Testing applications. Particularly, the Boundary Element Method (BEM) and the Local Boyndary Integral Equation Method are developed, so as to numerically solve static and transient problems on the field of elasticity and fluid-structure interaction in two dimensions. A major part of the present research is the construction of a computer program for solving such kind of problems. This textbook consists of three sections. In the first section, a thorough description on the theory of the BEM and the Local Meshless Methods (LMM) is done. The second section is dedicated for the numerical implementation of the BEM and LMM for solving steady state and time-harmonic two dimensional elastic and acoustic problems, in order to verify the accuracy and the ability of the proposed methodologies to solve the above-mentioned problems. Finally in the third section, the wave propagation problems of traction-free plates and cylindrical fuel storage tanks is studied, from the perspective of Non-Destructive Testing. The numerical methods of BEM and LMM are implemented, as well as spectral methods are utilized, for drawing useful conclusions on the wave propagation phenomena.

This work addresses the question of the efficient numerical solution of time-domain boundary integral equations with retarded potentials arising in the problems of acoustic and electromagnetic scattering. The convolutional form of the time-domain boundary operators allows to discretize them with the help of Runge-Kutta convolution quadrature. This method combines Laplace-transform and time-stepping approaches and requires the explicit form of the fundamental solution only in the Laplace domain to be known. Recent numerical and analytical studies revealed excellent properties of Runge-Kutta convolution quadrature, e.g. high convergence order, stability, low dissipation and dispersion. As a model problem, we consider the wave scattering in three dimensions. The convolution quadrature discretization of the indirect formulation for the three-dimensional wave equation leads to the lower triangular Toeplitz system of equations. Each entry of this system is a boundary integral operator with a kernel defined by convolution quadrature. In this work we develop an efficient method of almost linear complexity for the solution of this system based on the existing recursive algorithm. The latter requires the construction of many discretizations of the Helmholtz boundary single layer operator for a wide range of complex wavenumbers. This leads to two main problems: the need to construct many dense matrices and to evaluate many singular and near-singular integrals. The first problem is overcome by the use of data-sparse techniques, namely, the high-frequency fast multipole method (HF FMM) and H-matrices. The applicability of both techniques for the discretization of the Helmholtz boundary single-layer operators with complex wavenumbers is analyzed. It is shown that the presence of decay can favorably affect the length of the fast multipole expansions and thus reduce the matrix-vector multiplication times. The performance of H-matrices and the HF FMM is compared for a range of complex wavenumbers, and the strategy to choose between two techniques is suggested. The second problem, namely, the assembly of many singular and nearly-singular integrals, is solved by the use of the Huygens principle. In this work we prove that kernels of the boundary integral operators $w_n^h(d)$ ($h$ is the time step and $t_n=nh$ is the time) exhibit exponential decay outside of the neighborhood of $d=nh$ (this is the consequence of the Huygens principle). The size of the support of these kernels for fixed $h$ increases with $n$ as $n^a,a<1$, where $a$ depends on the order of the Runge-Kutta method and is (typically) smaller for Runge-Kutta methods of higher order. Numerical experiments demonstrate that theoretically predicted values of $a$ are quite close to optimal. In the work it is shown how this property can be used in the recursive algorithm to construct only a few matrices with the near-field, while for the rest of the matrices the far-field only is assembled. The resulting method allows to solve the three-dimensional wave scattering problem with asymptotically almost linear complexity. The efficiency of the approach is confirmed by extensive numerical experiments.

No description available.