Energy resource management in smart grids with intensive use of electric vehicles: heuristic and deterministic approaches
Ficheiros
Data
2017
Autores
Título da revista
ISSN da revista
Título do Volume
Editora
Resumo
A presente tese contribui para o aumento de conhecimento na área dos sistemas elétricos de energia e da inteligência computacional ao propor soluções inovadoras para o problema da gestão de recursos energéticos em smart grids. Estas soluções ajudam a viabilizar o crescimento e a proliferação sustentável de novas entidades de negócio associadas ao setor de energia, nomeadamente os agregadores de energia.
A escassez de recursos fósseis e a crescente preocupação climática devido ao uso generalizado desses recursos para a produção de eletricidade, motivou a implementação de novas medidas políticas e incentivos para contrariar o problema. Estas medidas têm motivado inúmeras iniciativas relacionadas com smart grids e contribuído para elevados investimentos em recursos de energia renovável. Embora o futuro dos sistemas de energia pareça bastante promissor, a elevada penetração de recursos energéticos de origem renovável tem colocado novos desafios na indústria e uma crescente complexidade na operação. Este problema é agravado pela provável penetração em massa de veículos elétricos num futuro próximo.
Na gíria anglo-saxónica, o termo virtual power player refere-se a entidades de agregação de energia. Estas entidades terão um papel importante para atenuar o problema da imprevisibilidade, já que permitem uma elevada flexibilidade de operação. Além disso, tornam possível um maior uso de recursos de fonte de energia renovável integrando pequenos produtores e consumidores que não têm acesso ao mercado grossista. O modelo de negócio dos virtual power players deverá ultrapassar um valor de mercado de 5 mil milhões de dólares nos próximos anos, tornando-se assim uma atividade altamente atrativa. O apoio à decisão no que diz respeita à gestão de recursos energéticos é crucial para a sobrevivência destas entidades de forma a reduzirem o custo de operação, aumentaram o lucro, reduzirem o impacto climático e reduzirem os riscos. As soluções existentes não estão preparadas para lidar com uma gestão integrada de recursos energéticos, especialmente em elevado número. Além disso, não existem propostas para avaliar as perdas financeiras com o aumento da disponibilidade em smart grids, importante sobretudo com o aumento da produção intermitente. Outro aspeto importante, é que os modelos existentes tem ignorado importantes fontes de incerteza e utilização da demand response como forma de atenuação da incerteza. A avaliação do desempenho computacional dos modelos propostos é outra vertente que tem sido sistematicamente ignorada. Estas são algumas das limitações das propostas atuais identificadas durante o trabalho efetuado.
A presente tese visa colmatar grande parte destas limitações relacionadas com a gestão de recursos energéticos. As contribuições de uma forma agregada resultam num inovador sistema de apoio à decisão, chamado Advanced Computational Environment for Leveraging Energy Resource mAnagement in smart grids (ACELERA). O ACELERA é composto por um leque diversificado de metodologias que, em conjunto, contribuem para lidar com a complexidade do problema de gestão de recursos energéticos. Estas metodologias incluem meta-heurísticas, métodos híbridos, determinísticos (ou matemáticos) baseados em decomposição, que contribuem para melhorar a escalabilidade e o desempenho do problema visado. Além da diversidade de técnicas, a tese propõe modelos de escalonamento de recursos de energia integrados, isto é, consideram veículos elétricos, sistema de armazenamento de energia e demand response. É de salientar que, programas de demand response adaptados a veículos elétricos foram propostos e implementados de forma a explorar o potencial da smart grid. Da mesma forma, foi desenvolvido um modelo multiobjetivo para avaliar as perdas financeiras com o aumento da disponibilidade da reserva que foi implementado na tese, assim como, um modelo de otimização estocástico que considera diversas fontes de incerteza. Importa frisar que um método denominado multi-dimensional signaling foi proposto e implementado nesta tese para melhorar o desempenho das meta-heurísticas nos modelos de gestão de recursos energéticos propostos. Por fim, é de salientar que o ACELERA foi desenvolvido a pensar em desempenho e de forma a usufruir da tecnologia disponível atualmente, por isso, é compatível com sistemas computacionais distribuídos e que funcionam com processamento paralelo.
As metodologias de apoio à decisão desenvolvidas na tese foram testadas e analisadas em cenários realísticos. Os resultados conseguidos nestas condições são promissores e validam a hipótese de que a gestão de recursos energéticos (diversificados e em grande escala) é possível e pode ser conseguida no contexto de smart grids.
This thesis provides contributions to the complementary fields of power systems and computational intelligence by proposing innovative solutions for the energy resource management problem in the context of smart grids, therefore, supporting innovative business models in the field, such as energy aggregators. The uncertainty concerning availability of fossil fuels and the rising climate effects caused by its widespread use in electricity generation has led to several policies and incentives to attenuate these problems. These measures contributed to huge investments in renewable energy sources and motivated many smart grid initiatives. Although the future panorama of modern power systems looks very promising, the large-scale integration of renewable sources of intermittent nature poses new challenges, strains and limitations in the current power industry. These problems are aggravated by the likely mass penetration of electric vehicles, which is expected to contribute significantly to the growing complexity already in place. Virtual power players are defined as energy aggregator entities that can provide a highly flexible generation and demand as required by the smart grid paradigm. In addition, virtual power players can allow to achieve high integration of renewable energy supply and raise value for small producers and consumers that cannot negotiate directly in the wholesale market. Indeed, it is expected that virtual power players market will steadily grow to more than 5 billion dollars in the next few years, making it a highly interesting sector. However, these entities require adequate decision support tools to overcome the complex challenges and deal with the large number of energy resources. Indeed, energy resource management is crucial for virtual power players to reduce operation costs, increase profits, reduce carbon emissions and improve the system stability. The existing tools are not prepared to deal with integrated energy resource management and large-scale use of distributed energy resources. Moreover, models proposed to evaluate the financial losses from increasing minimum reserve in smart grids have been missing, which is important with the increasing penetration of intermittent generation. In addition, models have been ignoring important sources of uncertainty and demand response as a way to mitigate its risks. The evaluation of the models computational performance is another critical issue that has been largely ignored. These are just a few of the main limitations identified in current proposals. This thesis proposes solutions to address a significant part of the identified issues. The thesis aggregated contributions, ultimately result in an innovative decision support system, named Advanced Computational Environment for Leveraging Energy Resource mAnagement in smart grids (ACELERA). The ACELERA is composed of a reasonable set of diversified methodologies that together contribute to handle the complexity of energy resource management, including metaheuristics, hybrid and deterministic decomposition methods, which mitigate the issues of performance and scalability. Moreover, it incorporates models that deal with integrated management of distributed energy resources, including electric vehicles, energy storage systems and demand response. Besides, demand response programs shaped for electric vehicles have been developed and implemented to fully explore the capabilities of the future smart grid. Likewise, a multi-objective model to evaluate the financial losses from increasing minimum reserve in smart grids has been developed and implemented as well as a stochastic model to address the uncertainty. It is also important to remark that a multi-dimensional signaling method to improve the performance of metaheuristics in solving the optimization problems is proposed in this thesis. Lastly, the ACELERA is compatible and capable of parallel and distributed computing, making the best use of available technology and able to deliver ultimate performance. The developed decision support methodologies have been tested and validated in realistic scenarios. The promising results achieved under realistic conditions support the hypothesis that the large-scale energy resource management with the wide use of diversified energy resources can be achieved in the context of smart grids.
This thesis provides contributions to the complementary fields of power systems and computational intelligence by proposing innovative solutions for the energy resource management problem in the context of smart grids, therefore, supporting innovative business models in the field, such as energy aggregators. The uncertainty concerning availability of fossil fuels and the rising climate effects caused by its widespread use in electricity generation has led to several policies and incentives to attenuate these problems. These measures contributed to huge investments in renewable energy sources and motivated many smart grid initiatives. Although the future panorama of modern power systems looks very promising, the large-scale integration of renewable sources of intermittent nature poses new challenges, strains and limitations in the current power industry. These problems are aggravated by the likely mass penetration of electric vehicles, which is expected to contribute significantly to the growing complexity already in place. Virtual power players are defined as energy aggregator entities that can provide a highly flexible generation and demand as required by the smart grid paradigm. In addition, virtual power players can allow to achieve high integration of renewable energy supply and raise value for small producers and consumers that cannot negotiate directly in the wholesale market. Indeed, it is expected that virtual power players market will steadily grow to more than 5 billion dollars in the next few years, making it a highly interesting sector. However, these entities require adequate decision support tools to overcome the complex challenges and deal with the large number of energy resources. Indeed, energy resource management is crucial for virtual power players to reduce operation costs, increase profits, reduce carbon emissions and improve the system stability. The existing tools are not prepared to deal with integrated energy resource management and large-scale use of distributed energy resources. Moreover, models proposed to evaluate the financial losses from increasing minimum reserve in smart grids have been missing, which is important with the increasing penetration of intermittent generation. In addition, models have been ignoring important sources of uncertainty and demand response as a way to mitigate its risks. The evaluation of the models computational performance is another critical issue that has been largely ignored. These are just a few of the main limitations identified in current proposals. This thesis proposes solutions to address a significant part of the identified issues. The thesis aggregated contributions, ultimately result in an innovative decision support system, named Advanced Computational Environment for Leveraging Energy Resource mAnagement in smart grids (ACELERA). The ACELERA is composed of a reasonable set of diversified methodologies that together contribute to handle the complexity of energy resource management, including metaheuristics, hybrid and deterministic decomposition methods, which mitigate the issues of performance and scalability. Moreover, it incorporates models that deal with integrated management of distributed energy resources, including electric vehicles, energy storage systems and demand response. Besides, demand response programs shaped for electric vehicles have been developed and implemented to fully explore the capabilities of the future smart grid. Likewise, a multi-objective model to evaluate the financial losses from increasing minimum reserve in smart grids has been developed and implemented as well as a stochastic model to address the uncertainty. It is also important to remark that a multi-dimensional signaling method to improve the performance of metaheuristics in solving the optimization problems is proposed in this thesis. Lastly, the ACELERA is compatible and capable of parallel and distributed computing, making the best use of available technology and able to deliver ultimate performance. The developed decision support methodologies have been tested and validated in realistic scenarios. The promising results achieved under realistic conditions support the hypothesis that the large-scale energy resource management with the wide use of diversified energy resources can be achieved in the context of smart grids.
Descrição
Tese de Doutoramento em Engenharia Eletrotécnica e de Computadores
Palavras-chave
Veículos elétricos , Otimização estocástica , Smart grid