L'énergie photovoltaïque est à nos jours l'une des sources intermittentes les plus développée. Plusieurs années de recherche confèrent une importante maturité à la fois aux modules et aux systèmes de extraction et traitement de son électricité. Cependant, il lui reste encore un important obstacle à franchir avant son utilisation à large échelle : la présence des ombres. Alors que plusieurs solutions ont été déjà proposées pour ce problème, la recherche sur l'ombre en tel que phénomène complexe reste embryonnaire. Cette thèse a pour but de combler ce besoin à la fois en étudiant la présence d l'ombre et en y proposant une nouvelle réponse. L'étude de l'ombre comprend la proposition d'une théorie sur l'intermittence qui prend en compte des aspect à la fois électriques et optiques. A travers de cette théorie, une relecture de la littérature est aussi proposée et donne lieu à une classification des solutions existantes en séries ou parallèles. Les solutions séries utilisent plusieurs structures d'électronique de puissance pour extraire l'énergie d'un nombre plus restreint des cellules photovoltaïques et par conséquent y confinent l'impact de l'ombre. En contre partie, les solutions parallèles utilisent des structures spéciales pour redistribuer le courant parmi les cellules ombrées et illuminées, ce qui amène à l'effacement de l'ombre. La nouvelle réponse à l'ombre proposée dans ce travail s'agit d'une structure parallèle à forte potentiel d'intégration monolithique. Inspirée de son équivalent pour les batteries, cette nouvelle topologie est applicable à plusieurs échelles mais sa commande reste un aspect à maîtriser. Son concept de base est validé au niveau d'un seul module photovoltaïque par la création d'un prototype et une validation expérimentale. Sa commande y est ensuite développée, testée et validée. Le système est capable de détecter la présence de l'ombre, choisir la meilleure stratégie pour la mitiger et l'implémenter en toute autonomie. Le résultat final est une augmentation de la puissance de sortie d'environ 40% dans certains cas. / Photovoltaic energy rates among the most mature renewable sources currently available in the market. However, its growing use in urban environment has met with an important obstacle: shadows. Their study present a two-fold challenge: understanding what they are and how they can be mitigated. While many authors have proposed different solutions for this problem, very few have tried to understand the shadow in its complexity. This thesis seeks, at the same time, a comprehensive view on the shadow itself while proposing a new solution to mitigate it. The comprehensive view of the shadow is proposed through an intermittency theory, where its optical properties and electrical consequences are taken into account. This theory provides the elements to review the current literature into a new perspective. The available solutions are, then, divided into two families: series and parallel. Series solutions employ several structures, each extracting the power of a reduced number of photovoltaic cells. As a consequence the impact of the shadow is restricted. Parallel solutions use few structures to redistribute the current between shaded and unshaded photovoltaic cells, thus sharing the impact of the shadow. The new solution proposed to mitigate the shadow is a parallel system called PV Equalizer. Inspired from its battery equivalent, it has a different topology with a high integration potential, easily scalable but seemingly difficult to control. To prove its concept, a study is conducted to determine its functions. It is found to be capable of not only mitigating but also detecting the shadow. These functions are characterized and their results used to conceive a control algorithm. Finally, this algorithm is tested and validated in a prototype under real operating conditions. The system detected the presence of the shadow, chose the best way to mitigate it and raised the power output by roughly 40 %.

Dissertation (PhD) -- Stellenbosch University, 2001 / ENGLISH ABSTRACT: Millisecond pulse forming networks (PFNs) as applicable to electro-thermal chemical (ETC) loads fall into the <1 MJ energy bracket. The energy requirements of these loads require specialised power sources involving staged energy conversion. For the <1MJ energy bracket, capacitive storage systems are usually employed. However, these systems exhibit low volume energy density and for volume sensitive applications; alternatives need to be considered. Inductive storage supplies form a sub-group of the static supplies that have theoretically superior volume energy density characteristics. This thesis documents the execution of a project concerned with the volumeoptimisation of an inductive storage supply. The system is composed of a three stage energy conversion chain. A prime power source (low power) charges an intermediate storage (IS) which is characterised by its medium power delivery capabilities. Energy is then transferred from the IS to the storage inductor which is characterised by its high power delivery capabilities. When sufficiently charged, the energy is then transferred to the load. Where pulse forming is required, the inductor storage must necessarily be modular. Switching elements to control the energy flow are also required. Work performed at Soreq, Israel, is used as the starting point. A topology variation of the XRAM topology presented by Soreq, original to this thesis, is presented and all its functioning modes are analysed. An existing volume model is analysed and expanded to incorporate heretofore unmodelled yet non-negligible considerations. The volume model generalises the effect of system modularity, subsystem technologies and allows for the incorporation of practical construction issues into the design process. The aim is to develop a 500 kJ, 80 kA, 20 kV system with a volume not exceeding lm3. This volume must include the IS, storage inductor and full switch volume. The optimisation algorithm and system topology developed in this thesis is validated through the construction and testing of a 1.2 kA, 2.5 kV 4 module prototype. A potential full ratings system, composed of contemporary device technologies and exhibiting a volume of just over 0.8m3, is proposed. / AFRIKAANSE OPSOMMING: Millisekonde pulsvormingsnetwerke, soos toegepas op elektrotermies-chemiese laste, val in die <1 MJ energievlak. Die energievereistes van hierdie tipe las vereis gespesialiseerde kragbronne wat die gestoorde energie in verskillende stadiums aan die las beskikbaar stel. Tans word kapasitiewe stelsels gewoonlik vir toepassings wat minder as 1 MJ energie benodig gebruik. ‘n Nadeel van hierdie stelsels is egter hulle relatiewe lae energiedigtheid. Vir toepassings waar lae volume van belang is, moet altematiewe metodes ondersoek word. Pulskragbronne wat van ‘n induktiewe energiestoor gebruik maak vorm ‘n deel van die klas van statiese kragbronne met hoe energiedigtheid. Hierdie tesis handel oor die optimering, in terme van volume, van ‘n induktiewe pulskragbron. Die stelsel bestaan uit drie stadiums, wat die energie van een vorm na ‘n ander omskakel en sodoende die vorm van die puls wat aan die las gelewer word, beheer. A lae-drywing primere kragbron laai ‘n medium-dry wing intermediere energiestoor. Energie word dan van die intermediere energiestoor na ‘n hoe-drywing stoorinduktor oorgedra. Nadat die induktor volgelaai is, word die energie aan die las oorgedra. Indien pulsvorming benodig word, kan van ‘n modulere induktor gebruik gemaak word. Vaste-toestand skakelelemente word gebruik om die energievloei te beheer. Navorsing wat by Soreq, in Israel, uitgevoer is, word as die vertrekpunt vir die studie gebruik. ‘n Verandering aan die XRAM topologie word voorgestel en die werking daarvan word in detail geanaliseer. ‘n Bestaande volume model word ondersoek en uitgebrei om ‘n aantal nie-weglaatbare verskynsels in aanmerking te neem. Die nuwe volume model maak voorsiening vir modulariteit, die effek van substelseltegnologie en ‘n aantal praktiese oorwegings wat in die ontwerp van die stelsel ‘n rol speel. Die finale doel is om ‘n 500 kJ, 80 kA, 20 kV stelsel in ‘n volume van 1 m3 in te pas. Hierdie volume van 1 m3 moet die intermediere energiestoor, die stoorinduktor, asook die skakelaars, bevat. Die optimeringsalgoritme en stelseltopologie wat ontwikkel is, word eksperimenteel deur middel van ‘n 1.2 kA, 2.5 kV, 4 module prototipe geverifieer. Laastens word aangetoon hoe ‘n finale stelsel, gebaseer op huidige skakelaartegnologie, met ‘n totale volume van 0.8 m3 moontlik in die toekoms ontwikkel kan word.

This document explores the isolated multiple-input single ended primary inductor converter (IMISEPIC) and discusses its application. This thesis proposes the following control methods such as current feed-forward control, voltage feedback control and maximum power point control to analyze the IMISEPIC. Zero-ripple technique is also applied to IMISEPIC in order to increase the converter’s life-time. Design strategy and concerns about the IMISEPIC are also presented, and simulations and circuit experiments are conducted to verify the analysis. Finally, the discussion about control limitation is used for future design consideration. / text

The availability of high speed and high power switching devices, such as the IGBT, has opened the opportunity for an increasing number of grid-connected inverter applications that have historically been unachievable. Recently, the number of inverter applications has surged, with now the focus being on increasing the relative performance and power capability. Such applications include UPSs, dynamic voltage restorers, STATCOMs, frequency converters and distributed grid sources such as solar panels. The inverter switching frequency limits its associated bandwidth and hence performance. Every application can benefit by reduction of the extent of this limitation. While state of the art devices like IGBTs enable such applications, the onus is now on developing high bandwidth digital controllers; the ability to connect multiple devices together to achieve power scaling; and having the confidence that the applications will work with other systems on a grid.% Solutions for for improving the inverter performance, ability to scale the power and operation compliance with other grid-connected devices are sought. Constraints and limitations imposed by the hardware and traditional continuous-time derived controllers are identified. A discrete-time direct design controller is then developed specifically for digital controllers, that for the same inverter configuration, achieves twice the bandwidth of a well-tuned traditional controller. An important feature of a controller is having the configurability of being able to choose inverter bandwidth over stability margin. To provide power scaling above that of a single switching module, investigations are performed on the suitability of actively paralleling inverter modules. Both the use of the developed discrete direct design controller and the identification of potential inter-module instabilities for a particular configuration enables the application of paralleled inverters. The operation is confirmed through the application of a sixteen paralleled module inverter system. Finally, a graphical analysis technique is introduced for analysing complex grids that may include inverter systems. The graphical technique demonstrates stability constraints with a range of sources and loads, including both inverters and rotating machines, which historical analysis techniques have been unable to do.

A novel, variable-gain switched-capacitor DC-DC converter is designed, constructed and tested. The proposed converter minimizes many of the problems which have traditionally hindered switched-capacitor DC-DC converters. The converter has high efficiency, strong regulation and low output voltage ripple across a wide variation in the line and load. The converter utilizes an optimized switching configuration that contains the maximum number of ideal conversion ratios for the given number of capacitors driven by a two-phase clock. The switched-capacitor converter is controlled by a gain-hopping feedforward control scheme in conjunction with duty-cycle, pulse-width modulation feedback control. The proposed control technique enhances the efficiency and regulation capability of switched-capacitor DC-DC converters, which are typically limited when there is a large variation in the line. Because the converter is optimized, programmable and capable of providing buck and/or boost operation (stepping-up and/or stepping-down the input voltage), the new switched-capacitor DC-DC converter is well-suited for a variety of applications and operating conditions. In addition, a novel algorithm based on graph theory and network analysis is developed which enumerates all possible ideal conversion ratios for a given switched-capacitor DC-DC converter structure. In particular, this algorithm can be used as a design tool to greatly improve the operation of multi-gain switched-capacitor converters, where the aim is to maximize the number of ideal conversion ratios while minimizing the number of switches and capacitors. Furthermore, the structure of all attainable positive, ideal conversion ratios of a two-phase switched-capacitor DC-DC converter, utilizing up to five capacitors, is enumerated. As a result, the design process for switched-capacitor converters is greatly simplified and a suitable converter structure can be more easily selected for a given application. / Thesis (Master, Electrical & Computer Engineering) -- Queen's University, 2013-04-03 23:27:24.183

With superior longevity, approximately 5 times that of compact fluorescents (CFLs), and high efficacy, around 1.5 times that of CFLs, LEDs are now attracting vast attention from both academic and industrial sectors. Unfortunately, current power supply drivers for LEDs have the following drawbacks: (1) for a two-stage configuration, the power factor correction (PFC) circuit can help LEDs achieve good operating performance but contain too many components and are large in size, have low efficiency and relatively high cost; (2) a single-stage configuration can perform well in PFC and efficiency, however reliability issues occur due to the use of the electrolytic capacitor. In this thesis, the theoretical analysis and implementation of two high power factor, soft-switched, electrolytic-capacitor-less LED drivers are presented. The two drivers solve the aforementioned issues while minimizing its size and cost. The detailed theoretical analysis illustrates the advantages of the presented circuits and provides insight into their design and operation. The simulated and experimental implementations verified the performance of both circuits, which achieve a high power factor, indicating that the drivers have good operating performance. Elimination of the electrolytic capacitors improves the LED drivers’ reliability. In addition, with the help of soft-switching capability, high efficiency is achieved. Simulation and experimental results are presented to support all merits of the two circuits. / Thesis (Master, Electrical & Computer Engineering) -- Queen's University, 2013-04-30 13:22:28.471

Switched reluctance motor (SRM) has become a competitive selection for many applications of electric machine drive systems recently due to its relative simple construction and its robustness. The advantages of those motors are high reliability, easy maintenance and good performance. The absence of permanent magnets and windings in rotor gives possibility to achieve very high speeds (over 10000 rpm) and turned SRM into perfect solution for operation in hard conditions like presence of vibrations or impacts. Such simple mechanical structure greatly reduces its price. Due to these features, SRM drives are used more and more into aerospace, automotive and home applications. The major drawbacks of the SRM are the complicated algorithm to control it due to the high degree of nonlinearity, also the SRM has always to be electronically commutated and the need of a shaft position sensor to detect the shaft position, the other limitations are strong torque ripple and acoustic noise effects.

a control, power quality and stability perspective. This is chiefly due to the future influence on power system behaviour resulting from the rapid cumulative growth of grid integrated wind power, and the improved control characteristics afforded by modern variable speed wind power generators. There is consequently strong motivation to enhance the inherent control robustness, power quality and fault-ride-through characteristics of modern wind power generators. By so doing, the attributes essential for power system operation regarding security of supply, reliability, and power quality can be assured. The work presented in this thesis employs a two-fold methodology in order to improve the inherent control, power quality and fault tolerance characteristics of a modern wind power generator based on a 2MW medium-voltage directly-driven permanently-excited architecture employing full-scale power conversion. Firstly, due to the complexity of modern wind power generators, accurate and complementary detailed non-linear (circuit orientated) and linear analytical (state-space based) wind generator models are developed. Collectively, these facilitate a wide range of detailed transient and smallsignal, control, stability and fault analysis studies. Ultimately, this facilitates the means by which advanced AC voltage controls are developed that significantly extend the wind power generator stable operating range for grid strength variations (grid impedance variation). Secondly, a supercapacitor based electrical energy storage system is designed and integrated within the developed wind generator models in order to facilitate the means by which fault-ridethrough characteristics and power quality can be improved. Fault-ride-through characteristics are ultimately improved by absorbing a proportion of generated power in the integrated storage system during grid-side faults. Power quality is ultimately improved by effectively buffering wind speed fluctuations in the integrated storage system so that a ‘smoothed’ version of the generated power results at the wind generator terminals.

M.Ing. (Electrical and Electronic) / Please refer to full text to view abstract

Le développement de la recharge sans contact de batteries comporte divers avantages pour les véhicules électriques. Cette solution est facile à utiliser, robuste et résistante aux intempéries par rapport aux câbles généralement utilisés. Le principe est basé sur le couplage magnétique entre un émetteur et un récepteur. L'objectif de cette thèse est de contribuer à proposer une norme pour permettre l’interopérabilité, c’est-à-dire, permettre à plusieurs émetteurs de fonctionner avec des récepteurs de différents fournisseurs. Comme le système doit aussi être tolérant au positionnement et doit respecter les recommandations concernant l’exposition humaine, de nombreuses configurations doivent être envisagées. Dans cette thèse, une modélisation avancée et fiable du système complet est proposée. La méthode des éléments finis est exploitée pour déterminer les caractéristiques électriques du coupleur inductif (inductances propres et mutuelles, facteur de couplage) dans différentes configurations de positionnement et d’interopérabilité. Ces valeurs permettent le dimensionnement du convertisseur à résonance. A ce stade différentes topologies de compensation sont considérées. Un modèle analytique au premier harmonique est mis en œuvre pour comparer les topologies et déterminer la fréquence de résonance globale du système. Un modèle circuit du système complet est ensuite développé pour évaluer précisément les courants et tensions. Enfin, un algorithme de régulation basé sur une méthode MPPT (Maximum Power Point Tracking) est évalué pour le réglage automatique de fréquence. A partir des courants calculés à la fréquence de résonance pour un point de fonctionnement nominal et grâce au modèle éléments finis incluant le châssis du véhicule le champ magnétique rayonné est calculé et comparé aux valeurs limites recommandées. A chaque étape de la modélisation, la sensibilité du système aux paramètres de configuration (positionnement, interopérabilité) est analysée. Des mesures effectuées au niveau du coupleur inductif et sur le système complet sont aussi utilisées dans l’analyse et permettent de valider le modèle / Development of contactless battery charging is an opportunity for electric vehicles. Compared to regular plugin cables, this solution is easy to use, robust and weather resistant. The power is transferred thanks to the magnetic coupling of inductive coils and a reduced magnetic circuit. The aim of this thesis is to contribute to propose a standard that would make possible to couple emitters with receivers from different suppliers, that is, to insure interoperability. As the system should also be tolerant to positioning and should respect human exposure recommendations, many configurations must be tested. In this thesis, an advanced and reliable modeling of the whole system is proposed. Using the finite element methods, the electrical characteristics (self, mutual inductances and coupling factor) of the inductive coupler are computed for different geometric and interoperability configurations. These values allow the dimensioning of the resonant converter. At this stage, different compensation topologies are considered. It is shown that the global resonant frequency can be derived and the topologies compared from a classical first harmonic approximation and analytical model. Then, a circuit model of the full system is developed in order to evaluate precisely the currents and voltages. Finally, the performance of a Maximum Power Point Tracking as frequency regulation algorithm is evaluated. From the currents computed at resonant frequency for the nominal operating point and the finite element model of the coupler, including the chassis of the vehicle, the radiated magnetic field is evaluated in order to check safety compliance. At each step of the modeling, the sensitivity of the system to the configuration parameters (positioning, interoperability) is analyzed. Measurements at the coupler level and for the full system are also used in this analysis and allow validating the model