The goal of this study was to determine the economic feasibility of battery energy storage system (BESS). Three major economic benefits derived from BESS using were studied: 1. Energy Purchase Shifting, 2. Distribution Feeder Deferral, 3. Outage Avoidance. The economic analysis was based on theoretical modeling of the BESS and distribution system. Three simulation models were developed to quantify the effects of different parameters, such as: BESS round-trip efficiency, life span, rated power, rated discharge time, marginal cost of electric energy, 24 h feeder load profile, annual load variation, feeder load growth rate and feeder length. An optimal battery charging/discharging method was presented to determine the differential cost of energy (DCE). The annual maximum DCE was calculated using stochastic probability analysis on seasonal load variation. The net present value was evaluated as the present value difference between two investments: first, the distribution feeder upgrade without BESS deferral, and second, with BESS deferral. Furthermore, the BESS’s contributions under different outage strategies were compared. It was determined that feeder length is the most significant parameter. The economics of the studied system becomes favorable when the feeder length exceeds a critical value.

The purpose of this thesis is to examine the technical losses in Electric Distribution Systems, the sources of the losses, minimum levels of the losses, ways to decrease the losses and current applications in Turkey. The wrong and weak parts of the current applications are determined and emphasized. Ways to decrease losses in Distribution Systems are advised. The energy resources in the world are decreasing rapidly. There is a rapid growth in consumption. It is a must to use existing resources in most efficient way because there is no unlimited energy source. Losses in the electric distribution systems are one of the most important subjects because the most of the technical losses in electric systems occur in the distribution systems.

Driven by technical, economic and environmental benefits for different stakeholders in the power industry, the electric distribution system is currently undergoing a major paradigm shift towards having an increasing portion of its growing demand supplied via distributed generation (DG) units. As the number of DG units increase; microgrids can be defined within the electric distribution system as electric regions with enough generation to meet all or most of its local demand. A microgrid should be able to operate in two modes, grid-connected or islanded. The IEEE standard 1547.4 enumerates a list of potential benefits for the islanded microgrid operation. Such benefits include: 1) improving customers’ reliability, 2) relieving electric power system overload problems, 3) resolving power quality issues, and 4) allowing for maintenance of the different power system components without interrupting customers. These benefits motivate the operation of microgrid systems in the islanded mode. However the microgrid isolation from the main grid creates special technical challenges that have to be comprehensively investigated in order to facilitate a successful implementation of the islanded microgrid concept. Motivated by these facts, the target of this thesis is to introduce new analysis and operational control algorithms to tackle some of the challenges associated with the practical implementation of the islanded microgrid concept. In order to accomplish this target, this study is divided into four perspectives: 1) developing an accurate steady-state analysis algorithm for islanded microgrid systems, 2) maximizing the possible utilization of islanded microgrid limited generation resources, 3) allowing for the decentralized operation of islanded microgrid systems and 4) enabling the islanded microgrid operation in distribution systems with high penetration of plug-in electric vehicles (PEVs). First for the steady-state analysis of islanded microgrid systems, a novel and generalized algorithm is proposed to provide accurate power flow analysis of islanded microgrid systems. Conventional power flow tools found in the literature are generally not suitable for the islanded microgrid operating mode. The reason is that none of these tools reflect the islanded microgrid special philosophy of operation in the absence of the utility bus. The proposed algorithm adopts the real characteristics of the islanded microgrid operation; i.e., 1) Some of the DG units are controlled using droop control methods and their generated active and reactive power are dependent on the power flow variables and cannot be pre-specified; 2) The steady-state system frequency is not constant and is considered as one of the power flow variables. The proposed algorithm is generic, where the features of distribution systems i.e. three-phase feeder models, unbalanced loads and load models have been taken in consideration. The effectiveness of the proposed algorithm, in providing accurate steady-state analysis of islanded microgrid systems, is demonstrated through several case studies. Secondly, this thesis proposes the consideration of a system maximum loadability criterion in the optimal power flow (OPF) problem of islanded microgrid systems. Such consideration allows for an increased utilization of the islanded microgrid limited generation resources when in isolation from the utility grid. Three OPF problem formulations for islanded microgrids are proposed; 1) The OPF problem for maximum loadability assessment, 2) The OPF for maximizing the system loadability, and 3) The bi-objective OPF problem for loadability maximization and generation cost minimization. An algorithm to achieve a best compromise solution between system maximum loadability and minimum generation costs is also proposed. A detailed islanded microgrid model is adopted to reflect the islanded microgrid special features and real operational characteristics in the proposed OPF problem formulations. The importance and consequences of considering the system maximum loadability in the operational planning of islanded microgrid systems are demonstrated through comparative numerical studies. Next, a new probabilistic algorithm for enabling the decentralized operation of islanded microgrids, including renewable resources, in the absence of a microgrid central controller (MGCC) is proposed. The proposed algorithm adopts a constraint hierarchy approach to enhance the operation of islanded microgrids by satisfying the system’s operational constraints and expanding its loading margin. The new algorithm takes into consideration the variety of possible islanded microgrid configurations that can be initiated in a distribution network (multi-microgrids), the uncertainty and variability associated with the output power of renewable DG units as well as the variability of the load, and the special operational philosophy associated with islanded microgrid systems. Simulation studies show that the proposed algorithm can facilitate the successful implementation of the islanded microgrid concept by reducing customer interruptions and enhancing the islanded microgrid loadability margins. Finally, this research proposes a new multi-stage control scheme to enable the islanded microgrid operation in the presence of high PEVs penetration. The proposed control scheme optimally coordinates the DG units operation, the shedding of islanded microgrid power demand (during inadequate generation periods) and the PEVs charging/discharging decisions. To this end, a three-stage control scheme is formulated in order to: 1) minimize the load shedding, 2) satisfy the PEVs customers’ requirements and 3) minimize the microgrid cost of operation. The proposed control scheme takes into consideration; the variability associated with the output power of renewable DG units, the random behaviour of PEV charging and the special features of islanded microgrid systems. The simulation studies show that the proposed control scheme can enhance the operation of islanded microgrid systems in the presence of high PEVs penetration and facilitate a successful implementation of the islanded microgrid concept, under the smart grid paradigm.

Este trabalho de pesquisa apresenta o desenvolvimento de uma metodologia de diagnóstico automático de falhas do tipo curto-circuito de baixa impedância em circuitos alimentadores de média tensão de sistemas reticulados de distribuição de energia elétrica tipo spot network. O diagnóstico compreende a identificação do tipo de curto-circuito ocorrido e a sua localização. A metodologia está baseada no treinamento e uso de Redes Neurais Artificiais (RNAs). Os parâmetros para treinamento das RNAs são obtidos a partir de padrões de comportamento elétrico de curtos circuitos monofásicos, bifásicos e trifásicos à terra, obtidos por simulação de um circuito de distribuição real localizado na cidade de Brasília. Para obtenção do comportamento elétrico do circuito de distribuição foi utilizado o aplicativo de simulação ATP (Alternative Transient Program); para estruturação, treinamento e testes das RNAs da metodologia de diagnóstico foi utilizado o software Matlab (aplicativo de RNA). Os principais resultados dos testes das RNAs da metodologia são apresentados. Tais resultados demonstram a viabilidade científica e tecnológica de se aplicar a metodologia desenvolvida como ferramenta de diagnóstico em tempo real de auxílio dos departamentos de engenharia de operação e manutenção de concessionárias elétricas. Este trabalho também apresenta as principais possibilidades de continuidade desta pesquisa científica e tecnológica baseada em redes neurais artificiais na área de diagnóstico automático de sistemas reticulados. / This work presents the development of an automatic failure diagnostic methodology for low impedance short circuit in mid voltage feeders of distribution spot networks systems. The developed methodology has the feature to identify the type of short circuit and its location. An Artificial Neural Network technique was employed. The parameters used to train the Artificial Neural Networks are obtained based upon patterns in simulations of real cases for short circuit behavior in mono-phase, bi-phase and tri-phase to ground configuration. The input data for the simulation was based on a real distribution circuit belonging to the Power Utility CEB located in Brasília-Brazil. The simulation program ATP (Alternative Transient Program) was used to obtain the electric behavior of the circuit in the distribution network. As for the Artificial Neural Network simulation, trainings and tests Matlab was employed. As a main contribution the results of this work shows the technical feasibility to apply such methodology as a important real time diagnostic tool to support the system operation and maintenance departments of power utilities that uses spot network topologies. Furthermore, it is presented the possibilities to continue this research related to automatic diagnostics for network distribution systems based on Artificial Neural Networks technique.

Este trabalho de pesquisa apresenta o desenvolvimento de uma metodologia de diagnóstico automático de falhas do tipo curto-circuito de baixa impedância em circuitos alimentadores de média tensão de sistemas reticulados de distribuição de energia elétrica tipo spot network. O diagnóstico compreende a identificação do tipo de curto-circuito ocorrido e a sua localização. A metodologia está baseada no treinamento e uso de Redes Neurais Artificiais (RNAs). Os parâmetros para treinamento das RNAs são obtidos a partir de padrões de comportamento elétrico de curtos circuitos monofásicos, bifásicos e trifásicos à terra, obtidos por simulação de um circuito de distribuição real localizado na cidade de Brasília. Para obtenção do comportamento elétrico do circuito de distribuição foi utilizado o aplicativo de simulação ATP (Alternative Transient Program); para estruturação, treinamento e testes das RNAs da metodologia de diagnóstico foi utilizado o software Matlab (aplicativo de RNA). Os principais resultados dos testes das RNAs da metodologia são apresentados. Tais resultados demonstram a viabilidade científica e tecnológica de se aplicar a metodologia desenvolvida como ferramenta de diagnóstico em tempo real de auxílio dos departamentos de engenharia de operação e manutenção de concessionárias elétricas. Este trabalho também apresenta as principais possibilidades de continuidade desta pesquisa científica e tecnológica baseada em redes neurais artificiais na área de diagnóstico automático de sistemas reticulados. / This work presents the development of an automatic failure diagnostic methodology for low impedance short circuit in mid voltage feeders of distribution spot networks systems. The developed methodology has the feature to identify the type of short circuit and its location. An Artificial Neural Network technique was employed. The parameters used to train the Artificial Neural Networks are obtained based upon patterns in simulations of real cases for short circuit behavior in mono-phase, bi-phase and tri-phase to ground configuration. The input data for the simulation was based on a real distribution circuit belonging to the Power Utility CEB located in Brasília-Brazil. The simulation program ATP (Alternative Transient Program) was used to obtain the electric behavior of the circuit in the distribution network. As for the Artificial Neural Network simulation, trainings and tests Matlab was employed. As a main contribution the results of this work shows the technical feasibility to apply such methodology as a important real time diagnostic tool to support the system operation and maintenance departments of power utilities that uses spot network topologies. Furthermore, it is presented the possibilities to continue this research related to automatic diagnostics for network distribution systems based on Artificial Neural Networks technique.

Submitted by Wagner Junior (wagner.junior@unioeste.br) on 2018-09-05T19:37:48Z No. of bitstreams: 8 Nicholas_Eduardo_Lopes_dos_Santos_2018.pdf: 1134120 bytes, checksum: de6da7ce734348755e0ca7b2b4212a63 (MD5) cluster_nodes_16_size_48.csv: 59426 bytes, checksum: f775d4775a1523f204043c946c2d57ac (MD5) cluster_nodes_16_size_200.csv: 236341 bytes, checksum: 40ab39add5f0d778c6539d28c894de41 (MD5) cluster_unit_1_size_48.csv: 5973 bytes, checksum: 6eac99d981bebc23d28c41ac1d577e2c (MD5) cluster_unit_1_size_200.csv: 16244 bytes, checksum: ef59fdb0e49418e1fe6aed35061d3b43 (MD5) cluster_unit_2_size_48.csv: 9118 bytes, checksum: 320e7f72fe0f6e9ffc39e6497b369be1 (MD5) cluster_unit_2_size_200.csv: 37612 bytes, checksum: 189ba1ad3d47ba41af4d26373be61f36 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2018-09-05T19:37:48Z (GMT). No. of bitstreams: 8 Nicholas_Eduardo_Lopes_dos_Santos_2018.pdf: 1134120 bytes, checksum: de6da7ce734348755e0ca7b2b4212a63 (MD5) cluster_nodes_16_size_48.csv: 59426 bytes, checksum: f775d4775a1523f204043c946c2d57ac (MD5) cluster_nodes_16_size_200.csv: 236341 bytes, checksum: 40ab39add5f0d778c6539d28c894de41 (MD5) cluster_unit_1_size_48.csv: 5973 bytes, checksum: 6eac99d981bebc23d28c41ac1d577e2c (MD5) cluster_unit_1_size_200.csv: 16244 bytes, checksum: ef59fdb0e49418e1fe6aed35061d3b43 (MD5) cluster_unit_2_size_48.csv: 9118 bytes, checksum: 320e7f72fe0f6e9ffc39e6497b369be1 (MD5) cluster_unit_2_size_200.csv: 37612 bytes, checksum: 189ba1ad3d47ba41af4d26373be61f36 (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2018-05-04 / Fundação Araucária de Apoio ao Desenvolvimento Científico e Tecnológico do Estado do Paraná (FA) / Traditionally, the maximum demand forecast in the planning horizon is used to determine the additions and reinforcements required for the expansion of a distribution system. The alternatives obtained with this model are considered overestimated because it considers a constant demand behavior along the planning horizon. Therefore, the search for a better representation of the demand in the planning models has been carried out, among them, the scenarios reduction based on clustering. This paper presents an analysis about the possibility of obtaining alternatives of expansion of lower cost in the planning of a distribution system representing the demand to be met and the distributed generation that can not be dispatched by more than one level. It is based on a planning model that considers a better representation of the load and generation that is solved through classical optimization techniques. The scenario reduction, used in the model, is based on levels groups that are defined and obtained whith k-means algorithm application. In addition to presenting expansion alternatives for different levels groups, it also shows that the processing times resulting from the combinatorial problem are acceptable. / Tradicionalmente, utiliza-se a demanda máxima prevista no horizonte de planejamento para determinar as adições e reforços necessários para a expansão de um sistema de distribuição. As alternativas obtidas com esse modelo são consideradas superestimadas, pois considera um comportamento de demanda constante ao longo do horizonte de planejamento. Diante disso, a busca por uma melhor representação da demanda nos modelos de planejamento tem sido realizada, dentre elas, a redução de cenários baseada em clusterização. Este trabalho apresenta uma análise sobre a possibilidade de obtenção de alternativas de expansão de menor custo no planejamento de um sistema de distribuição representando a demanda a ser atendida e a geração distribuída não despachável por mais de um patamar. Ela baseia-se em um modelo de planejamento que considera uma melhor representação da carga e da geração que seja resolvido através de técnicas de otimização clássica. A redução de cenários, utilizada no modelo, é baseada em grupos de patamares aos quais são definidos e obtidos com a aplicação do algoritmo k-médias. Além de apresentar alternativas de expansão para diferentes grupos de patamares, também mostra que os tempos de processamento, resultantes do problema combinatorial, são aceitáveis.