O ilhamento intencional quando desenvolvido em modo seguro pode ser considerado uma solução viável à promoção da confiabilidade das redes elétricas na presença de geradores distribuídos. Diante disso, esta tese de doutorado propõe metodologias que definem os ajustes das proteções do ponto de acoplamento comum e do gerador para viabilizar tecnicamente o ilhamento intencional, permitindo também avaliar a probabilidade de sucesso deste quando não planejado por um período de operação da rede. As metodologias propostas consideram os relés de deslocamento de fase ou \"salto de vetor\", relés de taxa de variação de frequência e os relés de sub/sobrefrequência convencionais. Estas metodologias baseiam-se na associação direta do tempo de detecção do ilhamento e do tempo de restrição. Estes tempos são oriundos da capacidade de detecção de ilhamento do relé baseado em medidas de frequência do ponto de acoplamento comum e da proteção do gerador. Os resultados foram consistentes e além de determinar a viabilidade técnica do ilhamento intencional de geradores distribuídos, permitiram avaliar a probabilidade de sucesso do ilhamento intencional. Adicionalmente, as metodologias permitiram definir quando possível os ajustes das proteções mais adequados à operação segura da ilha. / The intentional islanding when perfomed in a safe way can be considered a viable solution to the promotion of the reliability of the electrical networks in the presence of distributed generators. In this context, this doctoral thesis proposes methodologies that define the adjustments of the protections of the point of common coupling and the generator to technically enable the intentional islanding and allow to evaluate the probability of successful islanded operation. The proposed methodologies consider the vector surge or \"vector jump relays\", the rate of change of frequency and conventional under/overfrequency relays for islanding detection purposes, and the under/overfrequency relays to protect the generator. These methodologies are based on the direct association of the islanding detection time and the restriction time, and on their relation with the active power imbalance of the island. The detection and restriction times are derived from the islanding detection capability of the previous-mentioned relays, and from the generator protection, respectivelly. The results showed that the methodologies were able to identify favourable conditions for a successful islanded operation, and allowed assessing its probability of occurrence given different conditions of load and generation power imbalance. Additionally, the methodologies allowed to define, when possible, the adjustments of the most adequate protections for the isolated operation of the microgrid.

O ilhamento intencional quando desenvolvido em modo seguro pode ser considerado uma solução viável à promoção da confiabilidade das redes elétricas na presença de geradores distribuídos. Diante disso, esta tese de doutorado propõe metodologias que definem os ajustes das proteções do ponto de acoplamento comum e do gerador para viabilizar tecnicamente o ilhamento intencional, permitindo também avaliar a probabilidade de sucesso deste quando não planejado por um período de operação da rede. As metodologias propostas consideram os relés de deslocamento de fase ou \"salto de vetor\", relés de taxa de variação de frequência e os relés de sub/sobrefrequência convencionais. Estas metodologias baseiam-se na associação direta do tempo de detecção do ilhamento e do tempo de restrição. Estes tempos são oriundos da capacidade de detecção de ilhamento do relé baseado em medidas de frequência do ponto de acoplamento comum e da proteção do gerador. Os resultados foram consistentes e além de determinar a viabilidade técnica do ilhamento intencional de geradores distribuídos, permitiram avaliar a probabilidade de sucesso do ilhamento intencional. Adicionalmente, as metodologias permitiram definir quando possível os ajustes das proteções mais adequados à operação segura da ilha. / The intentional islanding when perfomed in a safe way can be considered a viable solution to the promotion of the reliability of the electrical networks in the presence of distributed generators. In this context, this doctoral thesis proposes methodologies that define the adjustments of the protections of the point of common coupling and the generator to technically enable the intentional islanding and allow to evaluate the probability of successful islanded operation. The proposed methodologies consider the vector surge or \"vector jump relays\", the rate of change of frequency and conventional under/overfrequency relays for islanding detection purposes, and the under/overfrequency relays to protect the generator. These methodologies are based on the direct association of the islanding detection time and the restriction time, and on their relation with the active power imbalance of the island. The detection and restriction times are derived from the islanding detection capability of the previous-mentioned relays, and from the generator protection, respectivelly. The results showed that the methodologies were able to identify favourable conditions for a successful islanded operation, and allowed assessing its probability of occurrence given different conditions of load and generation power imbalance. Additionally, the methodologies allowed to define, when possible, the adjustments of the most adequate protections for the isolated operation of the microgrid.

Un ilot non-intentionnel peut se former lorsque d’une petite partie du réseau est déconnectée du reste du système électrique, et continue pourtant de fonctionner à l’insu du gestionnaire de réseau en raison de la présence de générateurs locaux. Cet événement n’est pas voulu : le comportement des ilots n’est pas maitrisé, ce qui entraîne des risques potentiels pour les personnes et équipements électriques. L’ilotage non-intentionnel est un phénomène complexe et mal connu, aujourd’hui considéré comme rare, mais dont la probabilité d’apparition tend actuellement, a priori, à s’accroître progressivement. Cet accroissement est dû à l’effet conjugué de plusieurs phénomènes : premièrement, le développement des installations de production de petite et moyenne puissance, raccordées directement au réseau de distribution ; deuxièmement, l’élargissement des seuils de détection qui sont utilisés dans leurs protections de découplage ; et troisièmement, l’implémentation dans ces installations d’automatismes de régulation de puissance active et réactive. Dans ce contexte, la chaire industrielle Smartgrid d’Enedis a financé ces travaux de thèse afin de mieux caractériser ces phénomènes d’ilotage et d’améliorer leur détection.Afin d’atteindre ces objectifs, une première priorité de la thèse a été d’améliorer la compréhension des ilots non-intentionnels au travers d’une approche analytique. Celle-ci a notamment permis la résolution formelle d’un modèle simple capable d’expliquer le comportement en fréquence d’un îlot. Ces résolutions ont été enrichies par de nombreuses expérimentations en laboratoire et par des simulations numériques. Les modèles ont fait l’objet d’une étude bibliographique approfondie, notamment pour le choix des modèles de charge.Une seconde priorité de la thèse a été d’estimer l’efficacité des méthodes passives de détection d’ilotage au travers du calcul de la zone de non-détection (ZND). L'influence de plusieurs charges sur la ZND a été analysé, ce qui a permis d’améliorer la compréhension du phénomène complexe que sont les ilots non-intentionnels. L’efficacité des nouvelles protections anti-ilotage utilisant des seuils sur la dérivée en fréquence (ROCOF) a aussi fait l’objet d’évaluations approfondies. Enfin, les méthodes de calcul de ZND ont été améliorés par rapport aux méthodes préexistantes, ce qui a permis de radicalement améliorer l’estimation faite de l'impact des nouvelles régulations de puissance P(f) et Q(U). / Unintentional islands might appear when a subpart of the grid is disconnected from the power system and keeps working because of local generation. These events are unwanted: their behavior is not controlled and leads to potential risks for people and for electric equipment. Unintentional islanding is a complex phenomenon and badly understood. Nowadays it is considered to be rare, but a priori its occurrence probability is currently steadily increasing. This rising is the consequence of several phenomena: firstly, the development of distributed generation of small and average size directly connected to the distribution grid; secondly, the enlargement of frequency thresholds used by isolating protections; and thirdly, the implementation of power regulations of active and reactive power in the generation unit. Within this context, the Smart Grid chair of Enedis financed this PhD thesis in order to better characterize these islanding phenomena and to improve their detection.To achieve these targets, a first priority of the PhD was to improve the comprehension of unintentional islanding through symbolic calculation. This approach makes it possible to solve a simple model and to explain the evolution of the frequency during an island. These results are completed with several laboratory experiments and by numeric simulations. Models are designed based on a thorough bibliography, in particular for load modeling.A second priority of the PhD was to estimate the efficiency of anti-islanding protections through the computation of non-detection zone (NDZ). The impact of several loads on the NDZ is analyzed to improve the understanding of unintentional islanding, which is a complex phenomenon. The efficiency of new anti-islanding protections using thresholds on the rate of change of frequency (ROCOF) is also thoroughly evaluated. It is shown that the computation method used by the protection plays a key role. At last, the calculation methods of the NDZ is improved regarding previous methods. Among other things, it allows to radically change the estimations of the impact of new power regulations P(f) and Q(U).

Within the past decade, talk has arisen of shifting the utility grid from centralized, radial sources to a distributed network of sources, also known as distributed generation (DG); in the wake of deregulation, the California energy crisis, and northeastern blackouts. Existing control techniques for DG systems are designed to operate a system either in the connected or disconnected (islanding) mode to the utility; thus not allowing for both modes to be implemented and transitioned between. Existing detection and re-closure algorithms can also be improved upon. Dependent upon the method implemented, detection algorithms can either cause distortions in the output or completely miss a disturbance. The present re-closure process to reconnect to the utility is to completely shutdown and wait five minutes. The proposed methods of this study improve upon existing methods, via simulation and hardware experimentation, for DG systems that can intentionally islanding themselves. The proposed, "switched-mode", control allows for continuous operation of the system during disturbances by transitioning the mode of control to reflect the change in the system mode (grid-connected or islanding). This allows for zero downtimes without detrimental transients. The proposed detection method can sense disturbances that other methods cannot; and within 25 ms (approximately 1.5 line-cycles at 60 Hz). This method is an improvement over other methods because it eliminates the need to purposely distort the outputs to sense a disturbance. The proposed re-closure method is an improvement over the existing method due to the fact that it does not require the system to de-energize before re-synchronizing and reconnecting to the utility. This allows for DGs to continuously supply power to the system without having to shut down. Results show that the system is generally ready to reconnect after 2 to 5 line cycles. / Master of Science

Nowadays due to the increase of renewable energy sources, Distributed Generations (DGs) assumed an important role in distribution and sub transmission systems. However, abnormal operating conditions of the distribution system may cause that the DG remains isolated from the main system and connected to part of distribution system feeding its local loads. This operating condition is known as DG islanding. There are many island detection techniques however they can fail in some situation, such as in cases of low power unbalance. This work aims to present three news passive island detection technique for synchronous Distributed Generation that has better performance than the traditional one. The first technique aims to attenuates the oscillations caused in the frequency by short circuits and load switching using Mathematical Morphology. The MM reduces frequency oscillations, and in this way it allows to reduce the thresholds. With smaller thresholds is possible the detection of small power mismatch and in a much smaller time. The second technique is based on Frequency Oscillation Estimation in order to distinguish the islanding from other events that may occur in distribution systems. This island detection algorithm uses a small time window to estimate the oscillation frequency, obtaining faster responses than the existing methods which use larger windows to estimate the damping and frequency of oscillation. The third method aims to recognize de frequency shape through a pattern recognition algorithm. The algorithm uses singular value decomposition and linear discrimination analysis to classify the events. The algorithms performances have been tested considering several generation and load scenarios including short circuits, load and capacitor switching, DG outage and islanding. The proposed methods were compared with the island detection method commonly used, i.e. Rate of Change of Frequency and Under / Overfrequency. The proposed methods are reliable since they do not trip for the most of non-islanding event. Among the tested methodologies, the method of the Frequency Oscillation Estimation has the best performance. Its detection time is less than 40ms and it Non Detection Zone is smallest than the range between + 3.2% and -1.6% of the GD rated power. The mathematical simplicity of EFO is adequate for practical relay implementation and the results show that EFO performs better than the other techniques. / Atualmente, devido ao aumento das fontes de energia renováveis, a Geração Distribuída (GD) assumiu um papel importante nos sistemas de distribuição e subtransmissão. No entanto, condições anormais de operação podem fazer com que a GD seja desconectada do sistema principal e permaneça conectada a uma parte menor do sistema, alimentando apenas suas cargas locais. Esta condição de operação é conhecida como ilhamento não intencional da GD. Existem diversas técnicas de detecção de ilhamento, porém, estas podem falhar em determinadas situações como em casos de baixo desbalanço de potência. Este trabalho apresenta três novas técnicas para detecção de ilhamento de gerações distribuídas com geradores síncronos, as quais apresentam melhores desempenhos que as técnicas tradicionais. A primeira metodologia proposta utiliza a Morfologia Matemática (MM) e visa atenuar as oscilações de frequência causadas por curtos-circuitos e comutação de carga. A MM reduz as oscilações e desta forma permite a redução dos limiares de detecção. Com limiares menores é possível detectar desbalanços de potência menores e em um tempo muito menor. A segunda técnica utiliza a Estimação da Frequência de Oscilação (EFO) e busca distinguir ilhamentos de outros eventos que podem ocorrer no sistema de distribuição. Nesse caso, o algoritmo de detecção de ilhamento utiliza uma pequena janela de tempo para estimar a frequência de oscilação. Assim, obtém uma resposta mais rápida do que os métodos até então existentes, que estimam o amortecimento e a frequência de oscilação. O terceiro visa reconhecer a forma da frequência através de um algoritmo de reconhecimento de padrão. O algoritmo utiliza decomposição em valores singulares e análise linear discriminante para promover a classificação dos eventos. Os desempenhos dos algoritmos foram avaliados com base em diversos cenários de geração e de carga, incluindo curtos-circuitos, chaveamentos de carga e de banco de capacitores, saída de GD e ilhamento. Os métodos propostos foram comparados com os algoritmos de detecção de ilhamentos comumente utilizados, ou seja, a Taxa de Variação da Frequência e Sub/Sobrefrequência. Os métodos propostos se mostraram confiáveis uma vez que não operam para a maioria dos eventos que não se caracterizam como ilhamentos. Dentre as metodologias testadas, o método da estimação da frequência de oscilação é o que apresentou melhor desempenho. Seu tempo de detecção é menor que 40ms e sua Zona de Não Detecção (ZND) é inferior a faixa situada entre +3,2% e -1,6% da potência nominal da GD. A simplicidade matemática da EFO favorece sua implementação prática, e os resultados mostram um desempenho bem superior quando comparado a outras técnicas já implementadas.

This report addresses the parameter settings for a protective relay designed to safeguard a 15 kW wind turbine located in Rimbo, Norrtälje municipality, against unintentional islanding. This type of situation arises when distributed energy resources continue to supply local loads, even if the main grid has been unintentionally disconnected. The previous protection system experienced malfunctions and has been replaced with a new protective relay to ensure reliable protection. The new protective relay introduces two parameters (relay settings) associated with unintentional islanding protection function, namely RoCoF (rate of change of frequency) hysteresis, and delay, which were not defined in the previous protective relay. This report aims to determine suitable relay settings and verify that the relay operates as intended, taking into account both local and national requirements for connected generators. When these requirements did not specify or recommend a specific value, reference was made to practices in neighbouring countries, and consideration was given to the generator owner's benefit. Testing the relay with the chosen settings was crucial to ensure a correct understanding and accurate input of the settings. The testing was conducted in a laboratory environment with carefully defined signals of controlled frequency and voltage, while the relay's trip output was logged. The report includes information about the current settings along with a theoretical background to explain their functionality. It also presents test results and the provided information on parameter configurations. Additionally, recommended settings for RoCoF hysteresis and implementation delay are suggested, aiming to enhance protection against inadvertent operation. The results confirmed that the relay can be used in Rimbo, Norrtälje municipality, according to the parameters described in the report. / Denna rapport behandlar parameterinställningar för ett skyddsrelä som skyddar en 15 kW vindkraftsturbin i Rimbo, Norrtälje kommun, mot oavsiktlig ödrift. Denna typ av situation uppstår när distribuerade energiresurser försätter att mata lokala laster, även om stamnätet har kopplats bort oavsiktlig. Det tidigare skyddssystemet har blivit skadat och ersatt med ett nytt skyddsrelä för att säkerställa ett tillförlitligt skydd. Det nya skyddsreläet introducerar två parametrar (reläinställningar) relaterade till skyddsfunktionen mot oavsiktlig ödrift, nämligen RoCoF (frekvensderivata intervall) hysteres och fördröjning, som inte var definierade i det tidigare skyddsrelät. Denna rapport syftar till att fastställa lämpliga reläinställningar och verifiera att reläet fungerar enligt avsikt, med hänsyn till både lokala och nationella krav för anslutna generatorer. När dessa krav inte specificerade eller rekommenderade ett specifikt värde, refererades det till praxis i närliggande länder och hänsyn togs till generatorägarens fördel. Testning av reläet med de valda inställningarna var av central betydelse för att säkerställa en korrekt förståelse och korrekt inmatning av inställningarna. Testningen utfördes i en laborationsmiljö med noggrant definierade signaler av kontrollerad frekvens och spänning, samtidigt som reläets utlösningsutgång loggades.  Rapporten inkluderar information om de aktuella inställningarna tillsammans med en teoretisk bakgrund för att förklara deras funktion. Den presenterar också resultat från tester och den tillhandahållna informationen om parameterinställningar. Dessutom rekommenderas inställningar för RoCoF-hysteres och fördröjning för implementering, med målet att förbättra skyddet mot oavsiktlig ödrift. Resultaten konstaterade att reläet kan användas i Rimbo, Norrtälje kommun enligt de parametrar som beskrivs i rapporten.

Nowadays islanding has become a big issue with the increasing use of distributed generators in power system. In order to effectively detect islanding after DG disconnects from main source, author first studied two passive islanding methods in this thesis: THD&VU method and wavelet-transform method. Compared with other passive methods, each of them has small non-detection zone, but both of them are based on the threshold limit, which is very hard to set. What’s more, when these two methods were applied to practical signals distorted with noise, they performed worse than anticipated. Thus, a new composite intelligent based method is presented in this thesis to solve the drawbacks above. The proposed method first uses wavelet-transform to detect the occurrence of events (including islanding and non-islanding) due to its sensitivity of sudden change. Then this approach utilizes artificial neural network (ANN) to classify islanding and non-islanding events. In this process, three features based on THD&VU are extracted as the input of ANN classifier. The performance of proposed method was tested on two typical distribution networks. The obtained results of two cases indicated the developed method can effectively detect islanding with low misclassification.

Rapid advances in the smart grid technology are making it possible to tackle a lot of problems in the aged power systems. High-speed data acquisition system, high-voltage power electronic equipment, advanced utility and customer interaction technologies, as well as distributed renewable generation are enabling the revolution in the electric power generation, delivery and distribution. Through the implementation of ubiquitous metering and communication networks, the customers would no longer be a passive receiver of the electrical energy, but instead, an active participant in the power system and electricity market. They can not only sell their own energy to the utility, but also take part in the emergency restoration in the power grid. Nonetheless, some technical barriers are encountered during this revolution, such as difficulties in integrating home automation, smart metering, customer interaction and power system operation into the whole system. This thesis proposes a customer involved load shedding algorithm for both the power system frequency control and the micro-grid islanding. This new algorithm possesses the features of centralized load control and distributed load control, which fully utilizes the advantages of hierarchical communication networks along with the home automation. The proposed algorithm considers the reliability of the power grid as well as the comfort of the electricity users. In the power distribution system, the high-level control centre is responsible for coordinating the local load controllers, whilst the local controller takes charge of frequency monitoring and decision making. In the micro-grid, a centralized control strategy is adopted to better serve the system with the wide set of information available at the micro-grid control centre. The simulation results have demonstrated the correctness and feasibility of the proposed algorithm. Finally, the hardware implementation further tests the validity of the wireless sensor networks serving as the system’s monitoring and communication technology. / Thesis (Master, Electrical & Computer Engineering) -- Queen's University, 2013-09-24 20:01:37.098

Um dos principais problemas que podem ocorrer em sistemas de distribuição ou de subtransmissão de energia elétrica com geração distribuída (GD) é o ilhamento. O ilhamento ocorre quando um ou mais geradores distribuídos alimentam uma porção do sistema que se encontra eletricamente isolada do restante do sistema elétrico. Logo, geradores distribuídos têm de ser equipados com esquema de proteção anti-ilhamento, pois a operação ilhada pode colocar em risco a segurança das pessoas e equipamentos e pode deteriorar a qualidade da energia elétrica suprida aos consumidores locais. Este trabalho está inserido no contexto de proteção de sistemas elétricos de distribuição com GD e propõe uma nova técnica passiva-inteligente de proteção anti-ilhamento para geradores síncronos distribuídos. O método proposto utiliza a mineração de dados do Data Mining of Code Repositories (DAMICORE), o qual possui grandes potencialidades para descobrir correlações em dados complexos. Para realizar essa tarefa, três algoritmos são executados em sequência. São eles: o Normalized Compression Distance (NCD) para calcular a matriz de distância do conjunto de dados; o Neighbor-Joining (NJ) para construir as árvores filogenéticas; e por fim o Fast Newman (FN) para buscar grupos na árvore filogenética nos quais os dados apresentam alto grau de independência. O método de detecção de ilhamento nesta dissertação utiliza 10 características importantes dos sinais de tensão e corrente trifásicas no ponto de acoplamento comum (PAC). Essas 10 características são utilizadas como entrada ao DAMICORE para formular um algoritmo baseado em mineração de dados capaz de detectar ilhamento e de diferenciá-los de outros tipos de distúrbios. Os resultados mostram que o método identifica o ilhamento corretamente, detectando-o rapidamente, especialmente nos casos em que a detecção do ilhamento por relés baseados em medidas de frequência falha ou é lenta. / One of the main issues that can occur into distribution power systems with distributed generation is islanding. Islanding occurs when one or more distributed generators feed a portion of the distribution system that becomes electrically isolated from the rest of the power system. Thus, distributed generators have to be equipped with anti-islanding protection schemes, because the islanded operation may put at risk the safety of people and the equipment. It can also deteriorate the power quality supplied to local consumers. This work is in the context of the protection schemes for distribution power systems with distributed generation and it proposes a new intelligent passive anti-islanding protection scheme for distributed synchronous generators. The scheme is based on the Data Mining of Code Repositories (DAMICORE), which finds the similarity within complex data sets. This task is performed by computing the following algorithms, in sequence: the Normalized Compression Distance (NCD) to calculate a distance matrix among the data sets; the Neighbor-Joining (NJ) to build the phylogenetic trees; and the Fast Newman (FN) to find clusters with high degree of independence by using the phylogenetic trees. The proposed method comprises the extraction of 10 important features of the three-phase voltages and currents measured at the point of common coupling of the distributed generator. These 10 features are then used as input to the DAMICORE in order to formulate a data mining-based algorithm capable of detecting islanding and of distinguishing it from other disturbances. The results show that the method identifies the islanding correctly, detecting it quickly, especially where the detection of islanding by frequency-based relays are slow.

A expansão de geração distribuída nos sistemas de distribuição e de subtransmissão de energia elétrica em todo o mundo tem sua origem na reestruturação do setor elétrico, na necessidade de aproveitamento de diferentes fontes primárias de energia, nos avanços tecnológicos, nos incentivos governamentais e na maior conscientização sobre conservação ambiental. Com tal expansão, surge a necessidade de avaliar os impactos dessa forma de geração na operação das redes elétricas, principalmente das redes de distribuição, as quais foram inicialmente projetadas para operar considerando apenas a subestação de distribuição como fonte de energia. Entre esses impactos, um importante item a co0nsiderar é a detecção de ilhamento. Tal ocorrência é altamente indesejada pelas concessionárias distribuidoras de energia elétrica as quais estabelecem que os proprietários de geradores distribuídos devem instalar um sistema de proteção capaz de detectar ilhamentos. Dentre os métodos empregados para detecção de ilhamento os mais comuns para este propósito são os relés baseados em medidas de freqüência (relé de sub/sobrefreqüência, relé de taxa de variação de freqüência). Tais dispositivos são fortemente dependentes do desbalanço de potência ativa na rede ilhada e podem falhar caso esses desbalanços sejam pequenos. Ressalta-se que o comportamento variável da carga em sistemas de distribuição e a possibilidade de formação de diversas ilhas podem originar diferentes níveis de desbalanço de potência ativa e reativa. Conseqüentemente, pode haver vários cenários de operação em que os relés previamente citados falharão em detectar ilhamento. Neste contexto, esta dissertação de Mestrado propõe metodologias de proteção adaptativa anti-ilhamento aplicadas a relés baseados em medidas de freqüência, que visem a proteção do sistema elétrico e do gerador distribuído na maioria das condições de desbalanço de potência ativa e reativa que possam ocorrer na rede elétrica. Os geradores distribuídos a serem analisados são do tipo síncrono devido a sua grande utilização em pequenas centrais hidroelétricas e térmicas. São propostas duas metodologias: uma baseada em simulações dinâmicas e outra baseada em fórmulas analíticas. Os resultados obtidos foram satisfatórios para vários cenários de operação de um gerador distribuído conectado em uma rede de distribuição. Além disso, as vantagens e limitações de cada método são apresentadas e discutidas. / Distributed generation expansion worldwide has been caused by the de-regulamentation of the electricity sector, by the necessity to explore different energy resources, by technological advances, by government incentives and by environmental concerns associated with energy production. Such expansion brings the necessity to assess he impacts caused by distributed generators on the distribution systems, which were designed to operate considering the utility substation as the only power source. Among these impacts, on that deserves major attention is islanding detection. Utilities recommend that distributed generators\' owners provide efficient anti-islanding protection installed at the point of common coupling between the generator and the grid. Common anti-islanding protection schemes are composed by frequency-based relays (standard frequency relay and/or rate of change of frequency relay). These devices are strongly dependent on the active power imbalance in the islanded system and they can fail to detect an islanding condition if such power imbalance is small. Therefore, it is worth pointing that the variable power consumption of the distribution system\'s loads and the possibility of creating several energized islands can originate different power imbalance levels. Consequently, there may be various operating scenarios in which the frequency-based relays will fail to detect islanding. In this context, this work proposes adaptive methods to adjust frequency-based relays in order to provide effective anti-islanding protection considering different power imbalance levels. Synchronous generators are studied, since their wide application in distributed generation plants, such as small hydro and thermal power plants. Two methods are proposed: one is a simulation-based method and the other is based on analytical formulas. Both methods presented good results considering all the operating scenarios tested with a distributed generator connected to a distribution system. Furthermore, their advantages and limitations are presented and discussed.