Ce travail de thèse porte sur le développement et l’optimisation de capteurs dans le cadre du contrôle non destructif (CND) par courants de Foucault (CF). Le manuscrit présente plusieurs réalisations de capteurs CF souples gravés sur film Kapton. Un premier volet décrit l’évaluation des paramètres électromagnétiques (conductivité électrique σ et perméabilité magnétique µ) des matériaux typiquement rencontrés en CND par CF. Des méthodes conventionnelles pour estimer σ et µ ont été investiguées et mises en œuvre : il s’agit de la méthode à quatre pointes et du perméamètre. Néanmoins, ces méthodes présentent des difficultés pratiques au regard de l’état de surface (peinture, corrosion,…) et de la géométrie de l’échantillon. Deux capteurs ont donc été conçus : le premier composé d’une bobine d’émission et d’une bobine de réception afin d’évaluer la conductivité des matériaux purement conducteurs, puis le second composé d’un bobinage émetteur et d’une GMR pour évaluer la perméabilité magnétique. La conception des motifs et des résultats expérimentaux sont présentés dans le manuscrit. Le second volet décrit le développement d’un imageur CF flexible statique. L’imageur est un capteur multiélément composé de 576 récepteurs disposés en matrice permettant d’inspecter la surface d’une pièce sans déplacement du capteur par rapport à cette dernière. Le contrôle par l’imageur statique permet d’obtenir une image pixélisée de la surface sous le capteur. L’imageur a été optimisé de manière à détecter un défaut surfacique d’au moins 1 mm de long et d’orientation donnée quel que soit son emplacement vis-à-vis des bobines réceptrices. La conception du capteur et son évaluation expérimentale sont donnés dans le manuscrit. / The work of this thesis focuses on the development and the optimization of probes for non-destructive testing (NDT) by Eddy Currents (EC). The manuscript presents several achievements of flexible EC probes engraved on Kapton film. The first part describes the evaluation of the electromagnetic parameters (electrical conductivity σ and magnetic permeability µ) of materials typically encountered in NDT by EC. Conventional methods to estimate σ and μ have been investigated and implemented: it is the four-point probe and the permeameter. However, these methods present practical difficulties relating to the surface condition (paint, corrosion,…) and the sample geometry. Two probes have therefore been designed: the first is composed of a transmitting and a receiving coil in order to evaluate the conductivity of purely conductive materials, and the second is composed of a transmitter coil and a GMR for evaluate the magnetic permeability. Design patterns and experimental results are presented in the manuscript. The second part describes the development of a flexible static EC imager. The imager is a multielement probe composed of 576 receivers arranged in a matrix allowing to inspect the surface of a structure under test without moving the probe relative to the sample surface. The inspection by the static imager provides a pixelated image of the surface under the probe. The imager has been optimized to detect a surface defect of at least 1 mm long of given orientation regardless of its location relative to the receiver coils. The design of the probe and its experimental evaluation are given in the manuscript.

The continuously increasing demand for connecting electric grids with remote renewable energy sources such as wind power and photovoltaic cells has rekindled interest in high voltage direct current (HVDC) multi-terminal networks. Although HVDC networks have numerous benefits, their adoption relies entirely on the availability of HVDC circuit breakers which, compared to traditional alternating current circuit breakers, have to operate in a time frame of milliseconds. This thesis deals with the design of ultra-fast electro-mechanical actuators based on the so-called Thomson coil (TC) actuator. The simulation of a (TC) actuator constitutes a multi-physical problem where electromagnetic, thermal, and mechanical aspects must be considered. Moreover, it is complex since all those variables are co-dependent and have to be solved for simultaneously. As a result, a multi-physics simulation model that can predict the behavior and performance of such actuators with a high degree of accuracy was developed. Furthermore, other actuator concepts were also investigated and modeled in light of searching for a drive with a superior efficiency. The theory behind the force generation principles of two different types of ultra-fast electromechanical actuators, the TC and the double sided coil (DSC), were compared by the use of static, frequency, and comprehensive transient multi-physics finite element simulation models. Although, simulation models serve as a powerful tool for modeling and designing such state of the art actuators, without validation, they are weak and prone to errors since they rely on approximations and simplifications that might not always hold. Therefore, a prototype was built in the laboratory and the model was validated experimentally. Finally, it is important to note that the drives in this thesis are intended to actuate metallic contacts. As such, their behavior and performance upon mechanical loading was studied. Furthermore, some scaling techniques were applied to boost their performance and efficiency. / <p>QC 20130422</p>

A magnetostrictive material elongates when it is subjected to a magnetic field. This effect can then be used to design powerful actuators. The department of electromagnetic engineering has been working with magnetostricitve material and their applications since the 1980s and is presently engaged in a project focusing on magnetostrictive transducer utilisation for the aeronautic field. The focus of the presented work has been to develop and improve methods and tools supporting the development of magnetostrictive actuators. The axial-radial model was previously developed at the department and is well suited for circular cross sections of magnetostrictive rods. It is, however, common to laminate the magnetostrictive rods resulting in rectangular cross sections. The use of Cauer circuits allows modelling of the shielding effect. This shielding effect results in non-homogenous magnetisation and stress in both rectangular and circular cross sections of the rod. A model based on Cauer circuits, including a hysteresis model based on experimental data, was developed during the project. Furthermore, it is demonstrated how figures of merit and the use of finite element methods can be used to find optimised designs in a systematic and computational efficient way. The modified generalised Fabry factor and the magnetisation inhomogeneity coefficient are two proposed new figures of merit. A Magnetostricitve material is characterised through an experimental procedure. Usually, magnetostrictive material exhibit large hysteresis. An important part of the material characterisation is the post-processing of the measurement data, including a de-hysterisation procedure. In the thesis, a de-hysterisation method which ensures energy consistent data is presented. Energy consistent material data is essential to achieve energy consistent simulations of magnetostrictive systems. It is also demonstrated how the knowledge at the department can be utilised in international projects. In an ongoing project, the department is engaged in two sub tasks. In one of these sub tasks a high torque actuator is to be developed for the helicopter industry. The developed magnetostrictive models are used to perform system simulations of such actuator systems. In the other sub task a device for power harvesting from vibrations is analysed. It has now been shown how to adapt the load impedance in order to extract maximal electric power from the device. / QC 20101115

[pt] Esta dissertação apresenta um projeto de sistema portátil de localização de corpos estranhos metálicos não ferromagnéticos por meio de mapeamento magnético. O sistema baseia-se na indução de correntes parasitas no corpo estranho por um estágio de excitação e na medição por um magnetômetro GMI comercial de elevada resolução (25 pT). A topologia do instrumento é baseada no desacoplamento dos estágios de excitação e medição, com uma configuração que produz linhas de campo magnético primário elevadas na região do corpo estranho e tênues na região do sensor. Esta característica supera as limitações de um instrumento previamente desenvolvido, permitindo aumentar o campo magnético primário de excitação sem saturar o magnetômetro. O projeto é baseado em simulações computacionais, considerando dois tipos de excitação e duas orientações para o eixo de sensibilidade do magnetômetro. Diretrizes internacionais para os limites de exposição à radiação não-ionizante, aspectos biometrológicos, construtivos e elétricos também foram levados em consideração no projeto. A análise de desempenho das configurações mais promissoras confirmou a viabilidade do instrumento de medição proposto, otimizando a operação linear do magnetômetro durante o procedimento de medição e contribuindo para a futura construção de um protótipo de sistema de medição completo, com características de desempenho e segurança asseguradas para a aplicação biomédica pretendida. / [en] This dissertation presents a portable system for the localization of nonferromagnetic foreign metal bodies by magnetic mapping. The system is based on the induction of eddy currents in the foreign body by an excitation stage and the measurement by a high resolution commercial GMI magnetometer (25 pT). The instrument topology is based on the decoupling of the stages of excitation and measurement, with a configuration that produces a high primary magnetic field in the foreign body region and a weak one in the sensor region. This feature surpasses the limitations of a previously developed instrument, allowing to increase the primary excitation magnetic field without saturating the magnetometer. The project is based on computational simulations, considering two types of excitation and two orientations for the magnetometer sensitivity axis. International guidelines for exposure limits to non-ionizing radiation, biometrological, constructive and electrical aspects were also taken into consideration in the project. The performance analysis of the most promising settings confirmed the viability of the proposed measurement instrument, optimizing the linear operation of the magnetometer during the measurement procedure and contributing to the future construction of a complete measurement system prototype, with performance and safety characteristics ensured for the intended biomedical application.

Typischerweise wird die Rotorposition aktiver Magnetlager mit einer Lageregelung mit unterlagerter Stromregelung geregelt. Dieser Ansatz erreicht jedoch insbesondere bei axialen Magnetlagern mit massivem Kern seine Grenzen, daWirbelströme und Sättigungserscheinungen im Magnetkreis eine Diskrepanz zwischen den messbaren und kraftbildenden Spulenströmen hervorrufen. Die in der Folge erhebliche Verminderung der Lagersteifigkeit kann durch eine alternative Flussdichteregelung vermieden werden, da so unmittelbar die kraftbildende Komponente gestellt wird. In diesem Artikel stellen wir eine Regelungsvariante auf Basis eines fraktionalen Flussdichteschätzers vor, der ohne zusätzliche Sensorik auskommt und sich somit auch für bestehende Systeme implementieren lässt. Anhand von berechneten Frequenzgängen zeigen wir das große Verbesserungspotenzial dieser neuen Variante in Bezug auf Regelgüte und Stabilität im Vergleich zu einer klassischen Lageregelung mit unterlagerter Stromregelung.

Seit jeher sind Wirbelströme ein fester Bestandteil der Leistungsbilanz und Verlustberechnung in zahlreichen elektromagnetischen Energiewandlern. In aktiven Magnetlagern und Aktoren haben sie jedoch häufig einen zusätzlichen Einfluss auf die Kraftdynamik, da die einhergehende Feldverdrängung parasitäre Magnetisierungsströme hervorbringt, welche die meist strombasierten Kraftregler beeinträchtigen. Besonders betroffen sind die in dieser Dissertation beispielhaft betrachteten magnetischen Axiallager mit ihrer dreidimensionalen Flussführung, für welche die sonst übliche und effektive Blechung des Kerns unwirksam wird. Aus diesen Gründen werden regelungsbasierte Lösungen angestrebt. Bekannte fortschrittliche Topologien nutzen mitunter aufwendige Regler und Beobachter, wobei der direkte physikalische Bezug zu den mechanischen Parametern Steifigkeit und Dämpfung meist verloren geht. Diese Analogie zu mechanischen Lagern ist jedoch essentiell für eine einfache Inbetriebnahme des Magnetlagers, ein Grund, weshalb sich viele alternative Topologien nicht nachhaltig durchsetzen konnten und die dezentrale kaskadierte Lageregelung mit unterlagerter Stromregelung noch immer als weit verbreiteter Industriestandard gilt. Die in Axiallagern eingeschränkte Stabilität, Dynamik und Bandbreite aufgrund der Wirbelstromeffekte wird dabei zu Gunsten der einfacheren Anwendbarkeit toleriert. Diese Arbeit stellt ein fraktionales Kompensationsglied in Gestalt eines Flussschätzers vor, welches im Rückführungszweig der unterlagerten Regelung die Folgen der Wirbelströme dort herausrechnet, wo sie physikalisch wirken. Die resultierende modellbasierte Flussregelung erhält somit sämtliche physikalische Bezüge und die gute Anwendbarkeit, bei gleichzeitig verbesserten Regelungseigenschaften, sodass diesbezüglich keine Kompromisse notwendig sind. Das zugrundeliegende Modell leitet sich aus der Lösung der Diffusionsgleichung für den massiven Kern ab und stellt zunächst ein transzendentes fraktionales System dar, welches nicht direkt in einer Regelung anwendbar ist. Über Kettenbruchentwicklungen und Frequenzganganalysen ist es jedoch möglich, eine rationale Systembeschreibung zu ermitteln, die in Form einer digitalen Biquad-Filter-Kaskade auch in bestehende Mikroprozessor-Regelungen echtzeitfähig implementierbar ist. Die Arbeit dokumentiert das Vorgehen für eine Vielzahl von Randbedingungen und berücksichtigt verschiedene denkbare Einschränkungen, die in praktischen Anwendungen erwartbar sind. Der messtechnische Funktionsnachweis zeigt eine nahezu vollständige Kompensation der Wirbelstromeffekte in der unterlagerten Regelung, während sich die Bandbreite der Lageregelung nachweislich mindestens vervierfacht bei einem um bis zu 90 % Überschwingen gegenüber dem Industriestandard. / Eddy currents have always been part of loss calculations and power balances in numerous electromagnetic energy converters. In active magnetic bearings and actuators they additionally have a great influence on the force dynamic, as the concomitant magnetic skin effect provokes parasitic magnetizing currents that impair the usually current-based force controllers. Thrust bearings with their three-dimensional flux propagation, which serve as example in this thesis, are especially affected, due to the ineffectiveness of the commonly applied lamination of the iron core. For these reasons, control-based solutions are desired. Known advanced control topologies employ possibly intricating controllers and observers, which hardly preserve the direct physical reference to mechanical parameters like stiffness and damping. However, this analogy to mechanical bearings is essential for a simple bearing operation. That is one reason why many alternative topologies could not been established sustainably and the decentralized cascaded position control with subordinated current control is considered as the indisputable industry standard. Its limitation of stability, dynamic and bandwidth, caused by the eddy current effects in thrust bearings, is only tolerated, in favor of a superior applicability. This thesis introduces a fractional-order compensation element in the form of a flux estimator that compensates the eddy currents effects, where they physically occur, to wit, within the feedback path of the subordinated control. Hence, the resulting flux control maintains all physical references and the simple applicability, but does not compromise on the control characteristic in this regard. The underlying model is derived from the solution of the diffusing equation that describes the nonlaminated core. It firstly constitutes a transcendental fractional-order system, which cannot be directly applied to a bearing control. However, by the use of continued fraction expansions and frequency analysis, a rational system description is determinable, which can be implemented as biquad filter cascade for real-time application even in existing microprocessor controls. This work documents the procedure for a variety of boundary conditions while considering various possible restrictions, which are to be expected in practical applications. The experimental proof of concept reveals a nearly complete compensation of the eddy current effects in the subordinated control. The bandwidth of the outer position control is at least quadrupled, while the overshoot can be reduced by up to 90 % compared to the industry standard.

No description available.

No description available.

Die elektromagnetische Dynamik von axialen Magnetlagern ist im besonderen Maße von den eingesetzten Materialien abhängig. Axiale Flussverläufe machen eine Blechung von Stator und Rotor unwirksam und hohe induzierte Spannungen rufen im Magnetkreis wirbelstrombedingte Gegenfelder hervor. Zusätzliche kompensierende Magnetisierungsströme lassen den messbaren Strom der Steuerspule dem kraftbildenden Hauptfluss vorauseilen. Steigende Anforderungen an die Regeldynamik erfordern daher den Einsatz schwach elektrisch leitfähiger Kernmaterialien (Soft Magnetic Composites) oder eine Berücksichtigung der auftretenden Wirbelstromeffekte in der Regelstrecke unter Anwendung von Systemen gebrochenrationaler Ordnung. Beide Optionen werden in diesem Beitrag aus analytischer Sicht gegenübergestellt und ihre Anwendungsfälle diskutiert.

Ce travail de thèse porte sur le développement d’outils de modélisation pour le contrôle non destructif (CND) par courants de Foucault (CF). Il existe actuellement une tendance à la mise en œuvre de capteurs souples qui représentent une solution pertinente pour inspecter des pièces ayant une surface complexe. L’objectif principal de cette thèse est l’élaboration de techniques permettant la modélisation de ce type de capteurs au sein de la méthode des éléments finis (MEF).Lors de la modélisation d’un capteur souple avec la MEF, trois problématiques se manifestent. La première concerne le maillage des milieux fins qui apparaissent dans ce type de configuration (distance capteur-pièce contrôlée, bobine plate, revêtement fin...). Le maillage de ces régions par des éléments simpliciaux peut poser des problèmes numériques (éléments déformés quand un maillage grossier est considéré et grand nombre d’inconnues quand un maillage fin est utilisé). La deuxième problématique concerne le déplacement du capteur. Si les différents sous-domaines géométriques (air, pièce, capteur...) sont convenablement remaillés pour chaque position du capteur, le temps total pourra être pénalisant. La troisième problématique relative à la modélisation d’un capteur souple porte sur l’imposition du courant dans des bobines déformées.Une comparaison de différentes approches nous a conduit à retenir la méthode overlapping, qui permet de considérer simultanément des milieux fins avec maillages non-conformes. Cette méthode permet d’effectuer le recollement de deux surfaces non planes et/ou de géométries différentes. La méthode overlapping a été implantée dans deux formulations duales (magnétique et électrique) en 2D et 3D intégrées dans le code de calcul DOLMEN (C++) du LGEP. La méthode overlapping a été validée pour la prise en compte de plusieurs types de régions minces (air, milieux conducteurs, milieux magnétiques, bobines plates...). La modélisation des capteurs souples nécessite aussi l’implantation d’une technique permettant d’imposer correctement les courants dans un inducteur de forme arbitraire. Une technique a été sélectionnée et programmée, d’abord pour des bobines classiques (volumiques) déformées puis pour des bobines plates flexibles. Afin de valider les développements réalisés, différentes configurations de test ont été traitées et les résultats comparés avec des solutions de référence analytiques ou expérimentales. / The objective of this thesis is the development of modeling tools for eddy current testing (ECT). Currently there is a tendency to use flexible sensors which represent a viable solution for inspecting parts with a complex surface. The main objective of this thesis is the development of techniques for taking into account this kind of sensors within the finite element method (FEM).When modeling a flexible sensor with the FEM, three issues have to be considered. The first one is related to the mesh of thin regions that appear in this kind of configuration (sensor-inspected part distance, thin coating...). The meshing of these regions with simplicial elements can cause numerical problems (distorted elements when a coarse mesh is considered and high number of unknowns when a fine mesh is used). The second issue is related to the displacement of the sensor on the part surface. If the different subdomains (air, part, sensor ...) are properly remeshed for each position of the sensor, the required time can be penalizing. The third issue, related to the modeling of a flexible sensor, is the computation of the current in distorted sensor coils.A comparison of different approaches has led to select the overlapping element method, which allows to simultaneously consider the thin regions with nonconforming meshes. This method allows to perform the connection of two surfaces which can be non-planar and/or have different geometries. The overlapping method has been implemented in two dual formulations (magnetic and electric) in 2D and 3D and integrated in the computation code (C++) DOLMEN of LGEP. The overlapping method has been validated for several kinds of thin regions (air, conductive regions, magnetic regions, flat coils...). The modeling of flexible sensors also requires the establishment of a technique for properly imposing the current in an inductor of arbitrary shape. A technique has been selected and implemented for conventional (volumic) distorted coils but also for flexible flat coils. Different test configurations have been considered in order to validate the developments and the results have been compared with analytical references or experimental solutions.