Orientador: Maria Cristina Dias Tavares / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-16T22:19:27Z (GMT). No. of bitstreams: 1 Vidigal_RodrigoFernandes_M.pdf: 3000237 bytes, checksum: 232f1475b91d13386621a654c4a17690 (MD5) Previous issue date: 2010 / Resumo: Este trabalho apresenta uma análise do comportamento de uma linha de transmissão isolada de um pouco mais de meio comprimento de onda (2600 km) na freqüência fundamental, 60 Hz, alimentada a partir de uma barra infinita para diferentes condições de operação. Os perfis de tensão e de corrente, para diferentes condições de carregamento, são monitorados ao longo da linha. Uma análise do fator de potência da potência injetada no terminal emissor também é realizada e sua influência nos perfis de tensão e de corrente é verificada. As perdas ao longo da linha para diferentes níveis de potência transmitida são medidas e uma alternativa para minimização dessas perdas é apresentada. Uma metodologia de análise sistemática na freqüência fundamental de um sistema elétrico de potência identificando as condições críticas é também apresentada. A presente metodologia é aplicada em um sistema formado por linhas de transmissão de 500 kV existentes no sistema elétrico brasileiro, as quais são interligadas formando um tronco de 2600 km com a finalidade de representar uma linha de transmissão de um pouco mais de meio comprimento de onda. As análises feitas nesse trabalho consistem na verificação do comportamento do sistema na presença de diferentes tipos de curtos-circuitos em diferentes pontos da linha. Após a análise sistemática em regime permanente, as condições extremas podem ser simuladas em um programa de transitórios eletromagnéticos. Deste modo o comportamento da linha isolada alimentada a partir de uma barra infinita na presença de diferentes tipos de curtos-circuitos é discutido e os principais resultados são apresentados nesse trabalho. O tronco de 2600 km formado por linhas de 500 kV existentes no Brasil foi utilizado para simular a manobra de energização. As simulações incluíram diferentes técnicas de energização como o fechamento direto das fases, a utilização de um resistor de pré-inserção e o chaveamento controlado. Para cada alternativa de chaveamento foram analisadas as sobretensões transitórias e o perfil de tensão ao longo da linha. As simulações foram feitas utilizando-se os softwares ATP e PSCAD/EMTDC / Abstract: This work presents an analysis of the behavior of an isolated transmission line with a little more than half-wave length (2600 km) on fundamental frequency, 60 Hz, connected to an infinite busbar for different operating conditions. The voltage and current profiles for different loading conditions are monitored along the line. An analysis of the influence of the power factor of the transmitted energy on the voltage and current profiles along the line is performed. The losses along the line for different levels of transmitted power are measured and a method to minimize these losses is presented. A systematic review methodology in the fundamental frequency of a power system identifying the critical conditions is also presented. This methodology is applied to a system formed by 500 kV existing transmission lines in the Brazilian electrical system, which are interconnected forming an AC-link of 2600 km in order to represent a transmission line of a little more than half-wave length. The analysis made in this paper means to verify the system behavior in the presence of different types of short circuits at different points along the line. After systematic analysis in steady state, the extreme conditions can be simulated in an electromagnetic transient program. Thus the behavior of the isolated line connected to an infinite busbar in the presence of different types of short circuits is discussed and the main results are presented in this paper. The 2600 km AC-link formed by actual Brazilian 500 kV lines was used to simulate the trunk energization. The simulations included different techniques such as direct closure energization, the use of a pre-insertion resistor and the controlled switching. For each alternative it was analyzed the transient overvoltage profile along the line. The simulations were done with the software ATP and PSCAD / EMTDC / Mestrado / Energia Eletrica / Mestre em Engenharia Elétrica

No description available.

Orientador: José Roberto Sanches Mantovani / Banca: Rubén Augusto Romero Lázaro / Banca: Antonio Padilha Feltrin / Banca: Neida Maria Patias Volpi / Banca: Marcos Julio Rider Flores / Resumo: Neste trabalho, propõe-se um algoritmo Branch and Bound não linear para resolver o problema de planejamento e despacho ótimo de fontes de potência reativa em sistemas de energia elétrica. O modelo de planejamento é formulado como um problema de programação não linear inteiro misto, não convexo e de grande porte. Este modelo consiste na minimização dos custos das fontes reativas contínuas e ou discretas que devem ser alocadas no sistema. As restrições consideradas no modelo devem assegurar a qualidade e a confiabilidade do suprimento de energia para os consumidores, mantendo as magnitudes das tensões dentre seus limites pré-estabelecidos, atendendo as demandas de potência ativa e reativa e um conjunto de restrições físicas e operacionais dos equipamentos instalados no sistema tais como: limite nas capacidades de geradores, compensadores síncronos e estáticos e os limites na variação dos "taps" dos transformadores. O aspecto relevante deste trabalho é que o algoritmo proposto resolve diretamente problemas de programação não lineares inteiros misto, resolvendo em cada nó da árvore de Branch and Bound um problema de programação não linear que é o despacho ótimo de fontes reativas, em que as restrições de discretização das variáveis de alocação ou investimento dos bancos de capacitores e indutores são relaxadas. Estes problemas são resolvidos utilizando o método de pontos interiores (MPI) primal-dual preditor-corretor, que geralmente melhora o desempenho do MPI. O algoritmo Branch and Bound não linear proposto possui técnicas eficientes para a escolha do próximo subproblema que deve ser resolvido assim como a variável de separação dos subproblemas e os testes de sondagem. Para contornar os problemas de mínimos locais que são encontrados na resolução dos problemas de programação não linear, os testes de sondagem foram redefinidos... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: A nonlinear Branch and Bound algorithm to solve the problem of reactive optimal dispatch and planning of electrical power systems is proposed in this work. The planning model is formulated as a nonlinear non convex and large scale programming problem. This model consists of minimizing the costs of the reactive continuous and/or discrete sources that must be allocated on the system. The constraints considered on the model must assure the quality and reliability of supplying energy to the users maintaining the voltage magnitudes within the predefined limits, attending the active and reactive power demand and a set of physical and operational constraints of the installed equipment such as: generator capacity limits, synchronous and static compensators and the tap of the transformers. The emphasis of this work is that the proposed algorithm solves the integer nonlinear problems directly. A nonlinear programming problem is solved in each node of the Branch and Bound tree which is the optimal reactive power dispatch where the constraints that adjust the capacitors and inductors banks are relaxed. These problems are solved using the interior point method (IPM) primaldual predictor corrector, which in general improves the performance of the IPM. The nonlinear Branch and Bound algorithm proposed has efficient techniques to choose the next sub problem that must be solved as well as the separation variable of the sub problems and the sounding tests. To minimize the local minima problem, that are found solving the nonlinear programming problem, the sounding test were redefined adding a predefined percent to the current incumbent, obtaining an additional security margin, where in nonlinear problems not always the objective function value to a posterior problem is greater than the predecessor. Results are presented for test systems as IEEE30, IEEE118 and IEEE300. By simulations... (Complete abstract click electronic access below) / Doutor

Economia e Gestão do Ambiente / Master in Environmental Economics and Management / First and foremost, this research project develops a framework for analyzing electric system operators ( ESOs ). A deep literature survey is invoked to inventory the responsibilities of these institutions; then, a graphical model is developed so as any type of operator can be represented. This framework is useful to create comparisons between operators and understand the different nature of the responsibilities they are charged with, in connection with the particularities of their electric sectors. This framework was developed with the precise aim to benchmark the Brazilian operator against similar entities, yet it can be used for any country with an at least functionally unbundled ESO. Canada, Norway and Colombia were chosen due to the high percentage of hyroelectricity in their electricity generation capacity and production matrixes, a characteristic that is similar to the Brazilian case. Nevertheless, the study finds that the Brazilian operator s particularities owe mainly to the uniqueness of the electricity context, despite a few transaction costs between institutions than can possibly be reduced.

This thesis reports findings from a number of modern machine learning techniques applied to electricity market price forecasting. The techniques evaluated were Support Vector Machines, Boosting, Bayesian networks, neural networks and a weekly average method. All techniques were evaluated on seven day into the future forecasting of the Regional Reference (pool) Prices (RRP) for the New South Wales (NSW) region of the Australian National Electricity Market (NEM). Due to highly volatile and non-repetitive nature of the NSW RRP, all complex machine learning methods provided inferior accuracy forecasts compared to a weekly average method. The weekly average method was computationally less expensive and more transparent to the user than any of the machine learning techniques. The Support Vector Machine (SVM) was chosen for its novel application to electricity price forecasting because it is considered to be the next generation to neural networks. The structured SVM training algorithm proved more consistent and reliable than the neural network algorithm. Bayesian networks offer the adaptability of a neural network with the advantage of providing a price forecast with confidence intervals for each half-hour determined from the actual data. The SVM and Bayesian techniques were found to provide acceptable forecasts for NSW demand. An investigation of international electricity markets found that each market was unique with different market structures, regulations, network topologies and ownership regimes. Price forecasting techniques and results cannot be universally applied without careful consideration of local conditions. For instance, price data for the Spanish and Californian electricity markets were investigated and found to have significantly lower price volatility than the NSW region of the NEM. An extensive examination of the NSW RRP showed that the price exhibited no consistent long-term trend. A stationary data set could not be extracted from the price data. Thus, making forecasting unsuited to techniques using large historical data sets. The strongest pattern found for NSW prices was the weekly cycle, so a weekly average method was developed to utilise this weekly cycle. Over 25 weeks of NSW RRP from February to July 2002, the seven day into the future price forecast mean absolute error (MAE) for the SVM technique was 27.8%. The weekly average method was more accurate with an MAE of 20.6% and with a simple linear price adjustment for demand, the error was reduced to 18.1%. The price spikes and uneven distribution of prices were unsuitable for the Boosting or Bayesian network techniques.

No description available.

No description available.

No description available.

No description available.

This thesis reports findings from a number of modern machine learning techniques applied to electricity market price forecasting. The techniques evaluated were Support Vector Machines, Boosting, Bayesian networks, neural networks and a weekly average method. All techniques were evaluated on seven day into the future forecasting of the Regional Reference (pool) Prices (RRP) for the New South Wales (NSW) region of the Australian National Electricity Market (NEM). Due to highly volatile and non-repetitive nature of the NSW RRP, all complex machine learning methods provided inferior accuracy forecasts compared to a weekly average method. The weekly average method was computationally less expensive and more transparent to the user than any of the machine learning techniques. The Support Vector Machine (SVM) was chosen for its novel application to electricity price forecasting because it is considered to be the next generation to neural networks. The structured SVM training algorithm proved more consistent and reliable than the neural network algorithm. Bayesian networks offer the adaptability of a neural network with the advantage of providing a price forecast with confidence intervals for each half-hour determined from the actual data. The SVM and Bayesian techniques were found to provide acceptable forecasts for NSW demand. An investigation of international electricity markets found that each market was unique with different market structures, regulations, network topologies and ownership regimes. Price forecasting techniques and results cannot be universally applied without careful consideration of local conditions. For instance, price data for the Spanish and Californian electricity markets were investigated and found to have significantly lower price volatility than the NSW region of the NEM. An extensive examination of the NSW RRP showed that the price exhibited no consistent long-term trend. A stationary data set could not be extracted from the price data. Thus, making forecasting unsuited to techniques using large historical data sets. The strongest pattern found for NSW prices was the weekly cycle, so a weekly average method was developed to utilise this weekly cycle. Over 25 weeks of NSW RRP from February to July 2002, the seven day into the future price forecast mean absolute error (MAE) for the SVM technique was 27.8%. The weekly average method was more accurate with an MAE of 20.6% and with a simple linear price adjustment for demand, the error was reduced to 18.1%. The price spikes and uneven distribution of prices were unsuitable for the Boosting or Bayesian network techniques.