Au cours de la dernière décennie, les nanoparticules métalliques ont trouvé de nombreuses applications dans divers domaines tels que l'optique, la photonique, la catalyse, la fabrication de matériaux, les énergies renouvelables, l'électronique, la médecine et même les cosmétiques. Les nouveaux développements de ces applications nécessitent des méthodes de synthèse de nanoparticules fiables donnant une grande quantité de nanoparticules aux propriétés spécifiques. Les méthodes à base de plasma, tels que des décharges d'étincelles et d’arcs sont parmi les plus prometteuses car elles permettent une augmentation considérable de la vitesse de production et une diminution des coûts. Le contrôle de ces processus est cependant toujours difficile et nécessite de nombreuses études détaillées, à la fois expérimentales et théoriques. Dans cette thèse, les décharges d'étincelles sont étudiées numériquement. L'objectif principal est de mieux comprendre les principaux mécanismes impliqués dans la décharge d'étincelle avec un faible écartement d’électrodes et sous pression atmosphérique. Ensuite, sur la base de la modélisation détaillée proposée, la quantité de nanoparticules produites ainsi que leur distribution en taille est prédite et est comparée avec les résultats expérimentaux correspondants. Dans le modèle proposé, seules les conditions initiales, la géométrie du système et les propriétés du matériau sont utilisés comme paramètres d'entrée. Une décharge d’étincelle unique est divisée en plusieurs unités selon les échelles spatiales et temporelles des processus physiques comme suit: modèles de (i) flux plasma, (ii) décharge, (iii) hydrodynamique, (iv) couche cathodique, (v) érosion d’électrode et (vi) formation de nanoparticules. Les résultats du modèle combiné sont ensuite comparés à la fois avec d'autres résultats théoriques et à des résultats expérimentaux. Enfin, les possibilités d'optimisation de la production de nanoparticules par décharge d'étincelles sont proposées sur la base de la variation des paramètres expérimentaux et sur l'analyse de la quantité de particules produites et de leur taille moyenne / During last decade, metal nanoparticles have found many applications in various areas, such as optics, photonics, catalysis, material manufacturing, renewable energy, electronics, medicine and even cosmetics. Further development of these applications requires reliable nanoparticle synthesis methods providing a large amount of nanoparticle with required properties. Plasma-based methods, such as spark and arc discharges are among the most promising since they allow a considerable increase in the production rate and a decrease in costs. The control over these processes is, however, still challenging and requires many detailed studies, both experimental and theoretical. In this thesis, spark discharge is investigated numerically. The main objective is to better understand main mechanisms involved in spark discharge with a short gap under atmospheric pressure. Then, based on the proposed detailed modeling, the amount of the produced nanoparticles, their size distribution should be predicted and compared with the corresponding experimental results. In the proposed model, only initial conditions, geometry of the system and material properties are used as input parameters. A single spark event is divided into several units according to localization and time scales of physical processes as follows: (i) streamer model, (ii) discharging model, (iii) hydrodynamic model, (iv) cathode layer model, (v) electrode erosion model and (vi) nanoparticle formation model. The results of the combined model are then compared both with other theoretical and experimental results. Finally, possibilities of optimization the nanoparticle production by spark discharge are proposed based on the variation of the experimental parameters and on the analysis of the resulted particle yield and mean size

This project was made by student from UmeåUniversity and a request from the universityETH in Zürich, Switzerland. In this research project the electrical strengthof different natural gases and mixtures was investigated, and the aim was to finda gas or gas mixture with a natural origin or strongly attaching gases that couldreplace SF6 (Sulfur Hexafluoride). The gases were tested with breakdown experiments,one of those test was called lightning impulse breakdown test. The mainpart of this project was to investigate triggered spark gaps, which could be used inlightning impulse breakdown test. These spark gaps were made in a previous thesis,but have proved to not be reliable, therefore an investigation was needed. In thelab, a breakdown test setup, made up of a rectifying circuit and a transformer, wasused. In this project voltages up to 140kV were used. The two main parts of theproject were the spark gap unit and circuit analyzing and the spark gap characterization.These two parts contained test to see if the spark gap worked as it shouldor if there were any problems with it. The results from the tests showed that therewere problems with the spark gap, but these problems could be corrected or avoidedthrough controls of the spark gap before use.

Tantalum carbide (TaC) and hafnium carbide (HfC) possess extremely high melting points, around 3900 oC, which are the highest among the known materials. TaC and HfC exhibit superior oxidation resistance under oxygen deficient and rich environments, respectively. A versatile material can be expected by forming solid solutions of TaC and HfC. However, the synthesis of fully dense solid solution carbide is a challenge due to their intrinsic covalent bonding which makes sintering challenging. The aim of the present work is to synthesize full dense TaC-HfC solid solutions by spark plasma sintering with five compositions: pure HfC, HfC-20 vol.% TaC (T20H80), HfC- 50 vol.% TaC (T50H50), HfC- 80 vol.% TaC (T80H20), and pure TaC. To evaluate the oxidation behavior of the solid solutions carbides in an environment that simulates the various applications, an oxygen rich, plasma assisted flow experiment was developed. While exposed to the plasma flow, samples were exposed to a temperature of approximately 2800 oC with a gas flow speed greater than 300 m/s. Density measurements confirm near full density was achieved for all compositions, with the highest density measured in the HfC-contained samples, all consolidated without sintering aids. Confirmation of solid solution was completed using x-ray diffraction, which had an excellent match with the theoretical values computed using Vegard’s Law, which confirmed the formation of the solid solutions. The solid solution samples showed much improved oxidation resistance compared to the pure carbide samples, and the T50H50 samples exhibited the best oxidation resistance of all samples. The thickness of the oxide scales in T50H50 was reduced more than 90% compared to the pure TaC samples, and more than 85% compared to the pure HfC samples after 5 min oxidation tests. A new Ta2Hf6O17 phase was found to be responsible for the improved oxidation performance. Additionally, the structure of HfO2 scaffold filled with molten Ta2O5 was also beneficial to the oxidation resistance by limiting the availability of oxygen.

This thesis describes experiments to investigate high-altitude relight of a lean direct injection (LDI) combustor. The features that make LDI technology less polluting in terms of NOx compared to conventional combustors are expected to impede relight performance. Therefore an improved understanding of ignition behaviour is required to ensure that stringent relight requirements can be satisfied. Realistic operating conditions are simulated in a ground-based test facility. The application of laser diagnostics presents particular difficulties due to the large quantities ofliquid fuel that impinge on the combustor walls during relight. Advances are made in the application of planar laser-induced fluorescence (PLIF) to monitor fuel placement in a combustor under these conditions. A novel apparatus is developed to deliver a laser sheet to the combustion chamber while protecting all optical surfaces from contamination. The PLIF images are compared with the cold flow field obtained from CFD modelling. These results indicate that fuel becomes trapped inside the central recirculation zone in highconcentrations. High-speed flame imaging performed simultaneously with the PLIF measurements provides important insights into the motion and breakup of flame during relight. An algorithm developed to track the flame activity reveals that the initial spark kernel is convected downstream, before breaking apart and moving upstream towards a recovery origin close to the fuel injector. Analysis of many ignition events has revealed several distinct modes of ignition failure.

In this thesis work, we propose a Map-Reduce variant of the Growing Hierarchical Self Organizing Map (GHSOM) called MR-GHSOM, which is capable of handling mixed attribute datasets of massive size. The Self Organizing Map (SOM) has proved to be a useful unsupervised data analysis algorithm. It projects a high dimensional data onto a lower dimensional grid of neurons. However, the SOM has some limitations owing to its static structure and the incapability to mirror the hierarchical relations in the data. The GHSOM overcomes these shortcomings of the SOM by providing a dynamic structure that adapts its shape according to the input data. It is capable of growing dynamically in terms of the size of the individual neuron layers to represent data at the desired granularity as well as in depth to model the hierarchical relations in the data. However, the training of the GHSOM requires multiple passes over an input dataset. This makes it difficult to use the GHSOM for massive datasets. In this thesis work, we propose a Map-Reduce variant of the GHSOM called MR-GHSOM, which is capable of processing massive datasets. The MR-GHSOM is implemented using the Apache Spark cluster computing engine and leverages the popular Map-Reduce programming model. This enables us to exploit the usefulness and dynamic capabilities of the GHSOM even for a large dataset. Moreover, the conventional GHSOM algorithm can handle datasets with numeric attributes only. This is owing to the fact that it relies heavily on the Euclidean space dissimilarity measures of the attribute vectors. The MR-GHSOM further extends the GHSOM to handle mixed attribute - numeric and categorical - datasets. It accomplishes this by adopting the distance hierarchy approach of managing mixed attribute datasets. The proposed MR-GHSOM is thus capable of handling massive datasets containing mixed attributes. To demonstrate the effectiveness of the MR-GHSOM in terms of clustering of mixed attribute datasets, we present the results produced by the MR-GHSOM on some popular datasets. We further train our MR-GHSOM on a Census dataset containing mixed attributes and provide an analysis of the results.

As GPGPU computing becomes more popular, it will be used to tackle a wider range of problems. However, due to the current state of GPGPU programming, programmers are typically required to be familiar with the architecture of the GPU in order to effectively program it. Fortunately, there are software packages that attempt to simplify GPGPU programming in higher-level languages such as Java and Python. However, these software packages do not attempt to abstract the GPU-acceleration process completely. Instead, they require programmers to be somewhat familiar with the traditional GPGPU programming model which involves some understanding of GPU threads and kernels. In addition, prior to using these software packages, programmers are required to transform the data they would like to operate on into arrays of primitive data. Typically, such software packages restrict the use of object-oriented programming when implementing the code to operate on this data. This thesis presents GPUMap, which is a proof-of-concept GPU-accelerated map function for Python. GPUMap aims to hide all the details of the GPU from the programmer, and allows the programmer to accelerate programs written in normal Python code that operate on arbitrarily nested objects using a majority of Python syntax. Using GPUMap, certain types of Python programs are able to be accelerated up to 100 times over normal Python code. There are also software packages that provide simplified GPU acceleration to distributed computing frameworks such as MapReduce and Spark. Unfortunately, these packages do not provide a completely abstracted GPU programming experience, which conflicts with the purpose of the distributed computing frameworks: to abstract the underlying distributed system. This thesis also presents GPU-accelerated RDD (GPURDD), which is a type of Spark Resilient Distributed Dataset (RDD) which incorporates GPUMap into its map, filter, and foreach methods in order to allow Spark applicatons to make use of the abstracted GPU acceleration provided by GPUMap.

The quasicrystals are characterized by unusual rotational symmetries that are not observed in the crystalline materials, which is the cause of their interesting material properties. Because of that a particular attention was paid to quasicrystalline structures in the literature research. The research also contains a description of electron beam technology, spark plasma sintering method and introduction to the problematics of wear. As the default materials for the experimental part were selected Titanium Grade 2 powder and Cristome A5 powder which was partially composed of quasicrystalline phase. The first series of samples was sintered only from powder Cristome A5. The second series was sintered from the mixture of 80 % Titanium Grade 2 powder and 20 % Cristome A5 powder. For the compaction of samples spark plasma sintering technology was selected. Samples were then systematically modified by electron beam and subjected to pin on disc tests. Samples modified at 750 °C had the best wear resistance. Samples modified at 1150 °C contained increased amount of quasicrystalline phase.

The aim of this thesis is to show how to design and implement an application for network traces analysis using Apache Spark distributed system. Implementation can be divided into three parts - loading data from a distributed HDFS storage, supported network protocols analysis and distributed data processing. As a data visualization tool is used web-based notebook Apache Zeppelin. The resulting application is able to analyze individual packets as well as the entire flows. It supports JSON and pcap as input data formats. The goal of the application is to allow Big Data processing. The greatest impact on its performance has the input data format and allocation of the available cores.

The goal of this diploma thesis was to create four laboratory exercises for the subject "Parallel Data Processing", where students will try on the options and capabilities of Apache Spark as a parallel computing platform. The work also includes basic setup and use of Apache Kafka technology and NoSQL Apache Cassandra database. The other two lab assignments focus on working with a Travelling Salesman Problem. The first lab was designed to demonstrate the difficulty of a task where the student will face an exponential increase in complexity. The second task consists of an optimization algorithm to solve the problem in cluster. This algorithm is subjected to performance measurements in clusters. The conclusion of the thesis contains recommendations for optimization as well as comparison of running with different number of computing devices.

The diploma thesis deals with the preparation of the composite material based on the NiSi system using powder metallurgy supplemented by the sintering with the usage of SPS method (spark plasma sintering). Theoretical part contains general information about the mechanical-chemical process and sintering, while materials and methods used for experimental observation are explained in a separate chapter. Experimental part explains the procedure of the experiment and selected parameters of individual processes. In the experiment, two powder mixtures were created in order to form the NiSi phase in the maximum possible amount of powder material. After successfully reaching presence of the NiSi phase in the range of 87 to 89 wt. %, both mixtures were used to produce sintered samples at temperatures from 700 to 900 °C. Experiments showed the best results for sample number 2, which was sintered at 900 °C for 4 minutes. Resulting porosity was 0.9 % and hardness reached a maximum value of 718 HV 1. However, all sintered samples show cracks at room temperature associated with increased brittleness of the material.