Thesis (MScEng)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: The air-to-air refuelling of large aircraft presents challenges such as a long fuel transfer time, slow aircraft responses and a large distance between the aircraft CG and the receptacle position. This project addresses some of these issues by adding a control system to keep the receiver aircraft in the correct position relative to the tanker to enable fuel transfer. This project investigates different control strategies which are designed to control the A330-300 during refuelling at one trim condition. The controllers are based on a mathematical aircraft model which was derived from a simulation model received from Airbus. The first set of controllers uses the aircraft actuators directly. Controllers that are based on the CG dynamics and the receptacle dynamics are compared. Due to the large distance between the CG and the receptacle it was found to be essential to control the receptacle position, and not only the CG position. Also, a controller that is based on a model of the receptacle dynamics performs better. The second set of controllers uses the aircraft manual control laws as an inner loop controller. This set of controllers and the last direct actuator controller use the same axial controller that uses the engine thrust to control axial position. It was found that both the direct actuator controller and the manual control laws controller are able to keep the receptacle within the disconnect envelope in moderate turbulence. In both sets of controllers the axial controller fails to keep the receptacle reliably within the disconnect envelope in light turbulence. From the results it is concluded that both the direct actuator control and manual control laws can be used to successfully control the receptacle position in the normal and lateral positions as long as the receptacle kinematics are included in the control design. Using only the engine thrust for axial control is insufficient. Several recommendations are made to improve the axial control and also how these results can be used in future work. / AFRIKAANSE OPSOMMING: Die lug-tot-lug brandstof hervulling van groot vliegtuie het uitdagings soos ’n lang hervullingstyd, stadige vliegtuig dinamika en ’n groot afstand tussen die hervullingspoort en die vliegtuig massamiddelpunt. Hierdie projek spreek sommige van hierdie uitdagings aan deur ’n beheerstelsel by te voeg wat die vliegtuig in die korrekte posisie relatief tot die tenker hou vir brandstofoordrag om plaas te vind. Hierdie projek ondersoek verskillende beheerstrategieë wat ontwerp is om die A330- 300 te beheer by ’n enkele gestadigde toestand. Die beheerders is gebaseer op ’n wiskundige vliegtuigmodel wat vanaf ’n simulasiemodel afgelei is. Die simulasiemodel is vanaf Airbus verkry. Die eerste stel beheerders beheer direk die vliegtuig se beheeroppervlakke. Beheerders wat onderskeidelik die massamiddelpunt en die hervullingspoort beheer word vergelyk. Daar is gevind dat dit essensieel is om die hervullingspoort te beheer en nie slegs die massamiddelpunt nie, as gevolg van die groot afstand tussen hierdie twee punte. Die tweede stel beheerders gebruik die vliegtuig se eie beheerwette as ’n binnelusbeheerder en vorm self die buitelus. Albei stelle beheerders gebruik dieselfde aksiale beheerder wat enjin stukrag gebruik om die aksiale posisie te beheer. Daar is gevind dat beide stelle beheerders die hervullingspoort binne die ontkoppelingsbestek kan hou in die normale en laterale rigtings tydens matige turbulensie. In beide stelle beheerders is dit die aksiale beheerder wat faal om die hervullingspoort betroubaar in posisie te hou, selfs in ligte turbulensie. Vanaf die resultate word afgelei dat beide die direkte beheerder en die buitelusbeheerder gepas is om die laterale en normale posisiebeheer toe te pas mits die dinamika van die hervullingspoort in ag geneem word. Om slegs stukrag te gebruik vir aksiale beheer is nie voldoende nie, en verskeie voorstelle word gemaak om die aksiale beheer te verbeter in toekomstige navorsing.

Locally generated content comprises nearly 80% of front page content for most American newspapers, with Associated Press and other wire service content comprising between 15% and 20% across various circulation categories. However, newspapers with daily circulation between 30,000 and 50,000 provide a sharp distinction – a “curve” or “groove” – publishing an average of barely 60% locally generated content on their front page, pulling nearly 40% from wire services. The anomaly declines somewhat for publications with daily circulation between 50,000 and 100,000. This statistical content analysis of 42 American newspapers of diverse circulation sizes also reveals a universal focus on domestic politics, which comprises nearly a third of all locally generated and wire content, and reveals key commonalities across circulation categories and regional boundaries. / text

<p>The compressible turbulent boundary layer developing over atwo-dimensional bump which leads to a supersonic pocket with aterminating shock wave has been studied. The measurements havebeen made with hot-wire anemometry and Particle ImageVelocimetry (PIV).</p><p>A method to calibrate hot-wire probes in compressible ow hasbeen developed which take into account not only the ow velocitybut also the inuence of the Mach number, stagnation temperatureand uid density. The calibration unit consists of a small jetow facility, where the temperature can be varied. The hot wiresare calibrated in the potential core of the free jet. The jetemanates in a container where the static pressure can becontrolled, and thereby the gas density. The calibration methodwas verfied in the at plate zero pressure gradient turbulentboundary layer in front of the bump at three different Machnumbers, namely 0.3, 0.5 and 0.7. The profiles were alsomeasured at different static pressures in order to see theinuence of varying density. Good agreement between the profilesmeasured at different pressures, as well as with the standardlogarithmic profile was obtained.</p><p>The PIV measurements of the boundary layer ow in front ofthe 2D bump showed good agreement with the velocity profilesmeasured with hotwire anemometry. The shock wave boundary layerinteraction was investigated for an inlet Mach number of 0.69.A lambda shock wave was seen on the downstream side of thebump. The velocity on both sides of the shock wave as measuredwith the PIV was in good agreement with theory. The shock wavewas found to cause boundary layer separation, which was seen asa rapid growth of the boundary layer thickness downstream theshock. However, no back ow was seen in the PIV-data, probablybecause the seeding did not give enough particles in theseparated region. The PIV data also showed that the shock wavewas oscillating, i.e. it was moving approximately 5 mm back andforth. This distance corresponds to about five boundary layerthicknesses in terms of the boundary layer upstream theshock.</p><p><b>Descriptors:</b>Fluid mechanics, compressible ow,turbulence, boundary layer, hot-wire anemometry, PIV, shockwave boundary layer interaction, shape factor.</p>

In der vorliegenden Arbeit wird, ausgehend vom Stand der Wissenschaft und Technik des Lichtbogendrahtspritzens, das Verwenden von chromreichen Fe-Basislegierungen zur Herstellung von korrosionsbeständigen und gleichzeitig tribologisch geeigneten Zylinderlaufbahnbeschichtungen für Verbrennungsmotoren untersucht und qualifiziert. Der Einfluss der gewählten Werkstoffe und die durch die Beschichtung resultierenden Schichteigenschaften werden anhand von Ergebnissen der Tribologie- und Korrosionstests verglichen. Die Bewertung liefert einen Beitrag zum Verständnis der Mechanismen bei der Herstellung solcher Funktionsschichten. Es werden sowohl materialografische als auch diverse andere Charakterisierungsmethoden herangezogen. Ergänzt werden die Ergebnisse durch die Voraussage des motorischen Verhaltens und der anschließenden motorischen Qualifizierung des neuen Beschichtungssystems sowie die darauf folgenden Befundungen, welche der Bestätigung der theoretischen und im Laborumfeld ermittelten Erkenntnisse zum tribologischen und korrosiven Verhalten solcher Systeme dienen. Die Arbeit schließt mit einer Diskussion und einem Ausblick und gibt Empfehlungen für weiterführende Vertiefungen auf diesem Gebiet. / In the presented work it has been researched and qualified based on the knowledge about twin wire arc spray deposition states, the utilization of iron base chromium materials as a corrosion resistance and tribological useful cylinder liner coating for engine applications. The influence of the material selection and the produced twin wire arc coating properties has been compared with the results of the tribology and corrosion test. The comparison allows the understanding of the circumstances and relation by producing such functional coatings. Several materialographic-, tribological and corrosion tests have been done. In addition, the testing results were confirmed by a real engine application and a final analyses of the coating after engine run which delivers more information to understand the theory and laboratory testing results. The work discusses the achieved results and gives a forecast for further investigations in this field of science.

La tendance actuelle dans le domaine du transport ferroviaire est d'intégrer des modules de puissance de plus en plus puissants dans des volumes de plus en plus réduits. Cela pose des problèmes, notamment en termes de fiabilité, car lors de leurs cycles de fonctionnement, les interrupteurs à semi-conducteurs et leur environnement immédiat sont soumis à des contraintes électro-thermo-mécaniques plus sévères. Cela peut entraine leur destruction et donc la défaillance de la fonction de conversion d'énergie. L'objectif principal de cette thèse est de décrire des modèles et des outils de simulation multi-physiques afin de caractériser ces contraintes. Nous avons choisi comme cas d'étude les fils de connexion dits «wire bonding». Ces fils sont, en effet, considérés comme l'un des points faibles en ce qui concerne la durée de vie des modules de puissance, utilisés dans les systèmes embarqués notamment dans le ferroviaire. Dans ce contexte multi-physique, nous avons développé des modèles, numériques, éléments finis, analytiques, 3D ou 1D, afin de déterminer les contraintes thermomécaniques lors du passage du courant dans ces fils. A travers les modèles décrits et les résultats de simulation présentés, nous caractérisons le comportement des fils d'un point de vue électrique, thermique, magnétique ou mécanique. Plus précisément les interactions électromagnétiques, électrothermiques, électromécaniques ou thermomécaniques entre modèles et entre outils de simulation sont discutées. Les résultats sont confrontés aux mesures thermiques et de déplacement. Ces dernières sont réalisées par le biais de prototypes expérimentaux. Le mode de sollicitation utilisé est dit actif. Un régime de courant, continu ou alternatif, est appliqué au système. La réponse thermique et mécanique du système est alors obtenue. Les conclusions de cette étude permettent d'une part de mieux caractériser le comportement électro thermomécanique des fils de bonding et de mieux comprendre l'origine des modes de défaillance de cette technologie d'interconnexion. D'autre part, une démarche d'utilisation des modèles et outils logiciels multi physiques pour une simulation électro thermomécanique est présentée / The trend in the field of railway transport is to integrate increasingly powerful power modules in smaller volumes. This is problematic, especially in terms of reliability: during their cycles of operation, the semiconductor switches and their immediate environment are subject to tougher electro-thermo-mechanical stresses. This can lead to their destruction and then, to the failure of the energy conversion function. The main goal of this work is to describe the models and multi-physics simulation tools to characterize these stresses. We chose as a case study the connection wire called “wire bonding”. These wires are, indeed, considered one of the weaknesses of the life time of the power module used in embedded systems, particularly in railway applications. In this multi-physics context, we have developed numerical, finite elements, analytical, 3D or 1D models to determine the thermo-mechanical stresses during the current flow through the wires. Thanks to the models described and the simulation results presented, we characterize the behavior of the wire for an electrical, thermal, magnetic or mechanical point of view. More precisely, the electro-magnetic, electro-thermal, electro-mechanical or thermo-mechanical interactions between models and between simulation tools are discussed. The results are compared to thermal and displacement measurements. They are realized thanks to experimental prototypes. The type of solicitation is called active. A system of direct or alternating current is applied to the system. The thermal and mechanical response of the system is obtained. The conclusions of this study allow, on the one hand, characterizing the electro thermo-mechanical behavior of wires bonding and understanding the origin of the failure modes of this technology. On the other hand, a way of using models and multi-physics software tools for an electro thermo-mechanical simulation is presented

Na indústria de petróleo e gás as tubulações são formas eficientes de transportar esses produtos ou seus derivados das regiões de exploração para as refinarias e em alguns casos até o consumidor. O principal tipo de aço utilizado na fabricação dessas tubulações são os aços ARBL que devido à boa relação entre resistência à tração e tenacidade a baixas temperaturas geram uma boa relação custo/benefício na utilização e na montagem de tubulações. Neste trabalho é apresentado um estudo sobre a soldabilidade do tubo API 5L X80 construído com aço ARBL. Os corpos-de-prova utilizados foram anéis de 508 mm de diâmetro externo com espessura de parede de 19,0 mm. A soldagem foi executada com o processo eletrodo revestida no experimento 1, em um par de anéis, e o outro par de anéis foi soldado, na raiz, com o processo de soldagem MAG-TC e nos passes de enchimento e acabamento foi utilizado o processo arame tubular auto-protegido. Os resultados dos ensaios de resistência à tração, dobramento lateral e de Nick-Break demonstram que ambos os procedimentos de soldagem são qualificados pela norma API 1104. Associando o resultado do ensaio de Charpy V a 0°C e associando-o a quantificação das frações volumétricas de microconstituintes MA e a caracterização microestrutural é possível observar que os valores de energia absorvida na ZAC tem maior relação com o mecanismo de refino de grão, que melhora os valores obtidos, do que com a quantidade de fração volumétrica do microconstituinte MA. / In the oil and gas industries, pipelines are efficient ways of transporting these products, or its derivatives, from exploration regions to refineries and, in some cases, up to the final consumer. The main type of steel used in manufacturing these pipelines are HSLA steels that, due to the good relationship between tensile strength and toughness at low temperatures, create a cost/benefit ratio in the use and installation of these pipelines. This paper presents a study on the weldability of API 5L X80 HSLA pipe steel. The samples used were pipe rings with 508 mm outside diameter and a wall thickness of 19.0 mm. The root welding process was performed using SMAW in Experiment 1 with filing and finishing passes with SMAW. In the second experiment, the root pass were welded with GMAW-CT, and filling and finishing passes with FCAW-SS. Tensile test results, lateral bending tests and \'Nick-Break\' tests showed that both welding procedures were qualified according to API 1104 Standard. Associating these results with Charpy V test at 0°C and with quantification of volume fractions of MA microconstituent is possible to observe that the values of absorbed energy in the HAZ are more closely related to the mechanism of grain refinement, which improves the values obtained with the volume fraction of the amount of microconstituent MA.

Un nouveau procédé de fabrication additive de pièces métalliques, basé sur le procédé de soudage à l’arc appelé CMT (Cold Metal Transfert), est étudié dans l’objectif de réaliser des pièces en alliage d’aluminium Al-5Si. Un banc de fabrication additive basé sur le principe des imprimantes 3D open source, sur lequel a été intégré le procédé CMT, a été spécialement développé. Le procédé CMT permet de contrôler la fusion d’un fil d’aluminium et son dépôt sous la forme de gouttelettes sur la surface de construction, formant après solidification des « cordons » qui peuvent être superposés pour fabriquer des pièces. L’influence des paramètres du procédé sur les phénomènes de transfert de matière et de chaleur lors de la fusion du métal et de son dépôt sur la surface de construction, ainsi que sur les caractéristiques géométriques des cordons déposés, dans le cas de dépôts mono-cordon, puis dans le cas de murs formés par la superposition d’un grand nombre de cordons, est étudiée. Plusieurs défauts géométriques ont été observés, et les conditions de leur apparition analysées, grâce notamment à l’utilisation d’une caméra rapide. La compréhension des relations entre paramètres procédé, mécanismes de transfert de chaleur et de matière, et géométrie des cordons, a permis de corriger ces défauts en identifiant puis modifiant les paramètres procédé responsables de leur apparition. Enfin, une méthode de contrôle en ligne du procédé, basée sur l’analyse des signaux de tension et d’intensité produits par le générateur de soudage au cours du phénomène de dépôt, qui permet de détecter précocement l’apparition de défauts, et ainsi de modifier les paramètres procédé avant qu’ils ne s’amplifient, a été proposée. / A new additive manufacturing process for metallic parts, based on the arc welding process known as CMT (Cold Metal Transfer), is studied with the objective of building parts with the aluminium alloy Al5Si. A workbench for additive manufacturing based on the 3D printers open-source principle, on which the CMT generator was integrated, was specially developed. The CMT process allows to control the aluminium wire melting and its deposition under the form of droplets on the building surface, forming, after solidification, beads that can be superposed for the parts construction. The process parameters influence on the material transfer and heat transfer during the metal melting and deposition on the build surface, as well as on the geometric characteristics of the deposed beads, in the case of mono-layer deposits, and in the case of multi-layer walls, is studied. Many geometric defects were observed, and their apparition conditions analysed, thanks in particular to the use of a high-speed camera. The understanding of the relations between the process parameters, the melting and heat transfer mechanisms, and the beads geometry, allowed the defects correction by identifying and modifying the process parameters responsible of their apparition. Finally, an on-line control method for the process, based on the analysis of the voltage and current signals produced by the welding generator during the deposition phenomena, making possible the early detection of defects, and then the modification of the process parameters before they are amplified, has been proposed.

WAAM is a relatively unexplored additive manufacturing method. Although research in this area has been performed for some years and the hardware is relatively cheap, the method is not widely used. As the name suggest, it uses wire and an arc welding equipment to deposit beads on top of each other to create a geometry. As WAAM is a near net-shape method, the parts must be machined to its net-shape after the beads has been deposited. BAE Systems Hägglunds AB are investigating the use of WAAM in an industrial robot cell and this Master’s thesis has been written with the purpose of enabling the use of WAAM for manufacturing parts at the company. This report investigates how a part is prepared for WAAM and near net-shape machining. A formula for approximating the cost of manufacturing a part is investigated. A software for slicing a .STL file for generating a toolpath is developed in Matlab. The software then exports the toolpath to a code that the robot can read. It can also generate a digital model of the work piece for net-shape machining through CATIA macro. A model for calculating the cost of using the WAAM-cell once the toolpath for a part is known is presented. The investigated areas and the developed software are then applied to a part, and the results of the report is discussed.

In der vorgelegten Arbeit wurden zwei Arten von Quantendrahtstrukturen untersucht, die mittels Molekularstrahlepitaxie (MBE) hergestellt wurden. Erstens ist dies eine laterale Quantendrahtstruktur, die sich entlang einer Mesakante durch selektives Wachstum auf strukturierten GaAs (311)A-Substraten ausbildet. Zunächst wurden vertikal gestapelte Quantendrähte mit starker elektronischer Kopplung realisiert. Weiterhin wurden, unter Nutzung des amphoteren Einbaus von Si, p-i-n-Leuchtdioden mit einem einzelnen Quantendraht in der aktiven Zone hergestellt, die sich durch selektive Ladungsträgerinjektion in die Quantendrähte auszeichnen. Die Leuchtdioden wurden weitergehend mittels Mikrophotolumineszenz(µ-PL), Kathodolumineszenz (CL) und Elektronenstrahl-induziertem Strom (EBIC) charakterisiert. Zur Erklärung der selektiven Elektrolumineszenz (EL) wurde ein Modell, basierend auf der lateralen Diffusion von Elektronen und Löchern, vorgeschlagen. Für verspannte Systeme wurde der Einfluss von atomarem Wasserstoff auf das Wachstum von (In,Ga)As auf GaAs (311)A und die Bildung von lateralen Quantendrähten untersucht. Atomarer Wasserstoff spielt dabei die Rolle eines Surfaktanden und unterdrückt deutlich die Bildung von dreidimensionalen Inseln. Zweitens wurde das Wachstum von verspannten (In,Ga)As-Schichten auf GaAs (100) untersucht. Es zeigte sich, dass die dreidimensionale Inselbildung durch die Wachstumskinetik bestimmt ist, und ein Übergang von symmetrischen zu asymmetrisch verlängerten Inseln bei Erhöhung der Wachstumstemperatur auftritt. Dieser Prozess wird durch das Zusammenspiel von Oberflächen- und Verspannungsenergie bestimmt, wobei die experimentellen Befunde in guter Übereinstimmung mit den theoretischen Arbeiten von Tersoff und Tromp sind. Ausgehend von asymmetrischen (In,Ga)As-Inseln wurden selbstorganisierte Quantendrähte hergestellt, deren Homogenität und Länge sich durch Wachstum einer Vielschichtstruktur deutlich erhöhen. Strukturell wurden die (In,Ga)As-Quantendrähte mittels Rasterkraftmikroskopie (AFM), Röntgendiffraktometrie (XRD) und Transmissionselektronenmikroskopie (TEM) untersucht. Der laterale Ladungsträgereinschluss in den Quantendrähten zeigte sich deutlich in polarisationsabhängigen Photolumineszenz- und Magnetophotolumineszenzmessungen. / The present work focuses on two types of quantum wire structures which were grown by molecular beam epitaxy (MBE). First, the sidewall quantum wires based on the selective growth on mesa stripe patterned GaAs(311)A are studied. Single stacked sidewall quantum wires with strong electronic coupling have been fabricated. p-i-n type LEDs of the quantum wires employing the amphoteric Si incorporation for p- and n-type doping on GaAs(311)A have been fabricated. Strong selective carrier injection into the quantum wires is observed in electroluminescence (EL) measurements. The samples are characterized by micro-photoluminescence (µ-PL), cathodoluminescence (CL), as well as electron beam induced current (EBIC) measurements. To account for the highly selective EL, a model is proposed, which is based on the lateral diffusion of electrons and holes resulting in self-enhanced carrier injection into the quantum wires. Atomic hydrogen effects in the growth of (In,Ga)As on GaAs(311)A and its application to the sidewall quantum wire are investigated. It is found that atomic hydrogen suppresses island formation. Atomic hydrogen delays the relaxation by islanding thus playing the role of a surfactant. Second, the growth of (In,Ga)As layers on GaAs(100) is investigated showing that the formation of coherent 3D islands is a kinetically limited process. The transition from square-shaped islands to elongated islands is observed by changing the growth temperature for the growth of (In,Ga)As single layers. The elongation of the islands is a tradeoff between the surface free energy and the strain energy. A quantitative comparison between the experimental results and the theoretical work done by Tersoff and Tromp shows a good agreement. Self-organized quantum wires based on elongated discolation-free islands have been fabricated. The uniformity of the quantum wires is greatly improved by a superlattice growth scheme which also makes the wires much longer. The structural characterization of the quantum wires is performed by atomic force microscopy (AFM), x-ray diffractometry (XRD), and transmission electron microscopy (TEM). The lateral carrier confinement in the quantum wires is confirmed by polarization dependent PL and magneto-PL measurements.

Thomas Kin Fong Lei. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2000. / Includes bibliographical references (leaves 64-66). / Abstracts in English and Chinese. / LIST OF FIGURES --- p.IV / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Model-based Control of SMA Wires --- p.3 / Chapter 2.1 --- Model Identification of SMA Wires --- p.3 / Chapter 2.1.1 --- Temperature-Current Relationship --- p.3 / Chapter 2.1.2 --- Stress-Strain Relationship --- p.5 / Chapter 2.1.3 --- Martensite Fraction-Temperature Relationship --- p.8 / Chapter 2.2 --- Model-based Position Control of Two Linking SMA Wires --- p.9 / Chapter 2.3 --- Summary --- p.12 / Chapter 3 --- Neural-fuzzy-based Control of SMA Wires --- p.13 / Chapter 3.1 --- Adaptive Neuro-fuzzy Inference System (ANFIS) --- p.13 / Chapter 3.1.1 --- ANFIS Architecture --- p.13 / Chapter 3.1.2 --- Hybrid Learning Algorithm --- p.16 / Chapter 3.2 --- Generalized Neural Network (GNN) --- p.20 / Chapter 3.2.1 --- GNN Architecture --- p.20 / Chapter 3.2.2 --- Approximation of the GNN --- p.22 / Chapter 3.2.3 --- Backpropagation Training Algorithm --- p.24 / Chapter 3.2.4 --- Complexity Reduction of the GNN --- p.25 / Chapter 3.2.5 --- Error Bound of In-exact Reduction of the GNN --- p.29 / Chapter 3.3 --- Neural-fuzzy-based Position Control of Four Linking SMA Wires --- p.32 / Chapter 3.3.1 --- ANFIS-based Position Control of Four Linking SMA Wires --- p.32 / Chapter 3.3.2 --- GNN-based Position Control of Four Linking SMA Wires --- p.35 / Chapter 3.3.3 --- Performance Comparison of ANFIS and GNN Algorithms --- p.37 / Chapter 3.4 --- Summary --- p.39 / Chapter 4 --- SMA Actuated Artificial Face --- p.40 / Chapter 4.1 --- Muscles of the Human Face --- p.40 / Chapter 4.2 --- The Software Part: facial model --- p.41 / Chapter 4.3 --- The Hardware Part: artificial face and peripheral interface --- p.43 / Chapter 4.3.1 --- SMA Actuated Artificial Face --- p.43 / Chapter 4.3.2 --- Peripheral Interface --- p.45 / Chapter 4.4 --- Position Control on the Artificial Face --- p.47 / Chapter 4.4.1 --- Model-based Position Control on Artificial Face --- p.48 / Chapter 4.4.2 --- Neural-fuzzy-based Position Control on Artificial Face --- p.49 / Chapter 4.4.3 --- Comparison of the Model-based and Reduced GNN Control of Artificial Face --- p.49 / Chapter 4.5 --- Experimental Result --- p.50 / Chapter 5 --- Conclusion --- p.52 / Appendix1 --- p.53 / Appendix2 --- p.55 / Appendix3 --- p.56 / Appendix4 --- p.58 / Bibliography --- p.64