| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 12.11 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
O aumento da dependência global de fontes de energia renováveis tem reforçado a ne-
cessidade de uma monitorização e manutenção eficazes de turbinas eólicas. Um dos
componentes mais críticos destes sistemas são as pás das turbinas, sujeitas a várias so-
licitações e potenciais danos estruturais ao longo do tempo. Esta dissertação aborda o
desafio de desenvolver um modelo de teste para a simulação e posterior identificação
de dano em pás de turbinas eólicas, com foco no projeto do protótipo físico e no apri-
moramento dos métodos de deteção de dano. O problema é particularmente relevante
devido às exigências operacionais sobre as pás, que estão expostas a condições ambientais
severas. Os danos, se não forem detetados atempadamente, podem resultar em perdas
de eficiência ou até em falhas catastróficas. Embora os métodos de deteção tradicionais
sejam úteis, podem faltar-lhes precisão na identificação de anomalias em fases iniciais,
tornando esta questão desafiante e importante no âmbito da manutenção da energia
eólica. Para resolver este problema, a solução proposta envolve o desenvolvimento de um
modelo computacional da pá, baseado no perfil aerodinâmico NREL S809, otimizado tanto
para a eficiência aerodinâmica quanto para a integridade estrutural. O trabalho, além
de fornecer um projeto detalhado, introduz uma nova abordagem para a identificação
de dano, baseada na técnica "Cloud of Spheres", como alternativa ao método baseado na
Distância de Mahalanobis. Esta técnica oferece a capacidade para detetar anomalias em
espaços de dados multidimensionais, permitindo uma solução mais intuitiva e robusta
para a monitorização da integridade estrutural. Os resultados, demonstram o potencial
da metodologia implementada. A técnica "Cloud of Spheres" mostrou-se eficaz em dis-
tinguir entre estados danificados e não danificados das estruturas testadas, validando a
sua aplicabilidade em futuros protótipos físicos e cenários de teste. As implicações desta
investigação são substanciais, pois estabelecem uma base sólida para avanços futuros nas
tecnologias de deteção de dano, contribuindo para sistemas de energia eólica mais fiáveis
e eficientes.
The increasing global reliance on renewable energy sources has heightened the need for effective monitoring and maintenance of wind turbines. One of the most critical components of these systems is the turbine blades, which are subject to various structural loads and potential damage over time. This dissertation addresses the challenge of developing a test model for simulating and subsequently identifying damage in wind turbine blades, focusing on the design of the physical prototype and the improvement of damage detection methods. The problem is particularly relevant due to the operational demands on the blades, which are exposed to harsh environmental conditions. If damage is not detected in a timely manner, it can lead to efficiency losses or even catastrophic failures. While traditional detection methods are useful, they may lack precision in identifying early-stage anomalies, making this issue challenging and significant in the context of wind energy maintenance. To address this problem, the proposed solution involves developing a computational model of the blade, based on the NREL S809 aerodynamic profile, optimized for both aerodynamic efficiency and structural integrity. In addition to providing a detailed design, this work introduces a new approach to damage identification, based on the "Cloud of Spheres"technique, as an alternative to the Mahalanobis Distance- based method. This technique offers the ability to detect anomalies in multidimensional data spaces, providing a more intuitive and robust solution for structural health monitoring. The results demonstrate the potential of the implemented methodology. The "Cloud of Spheres"technique proved effective in distinguishing between damaged and undamaged states of the tested structures, validating its applicability in future physical prototypes and testing scenarios. The implications of this research are substantial, as they lay a solid foundation for future advancements in damage detection technologies, contributing to more reliable and efficient wind energy systems.
The increasing global reliance on renewable energy sources has heightened the need for effective monitoring and maintenance of wind turbines. One of the most critical components of these systems is the turbine blades, which are subject to various structural loads and potential damage over time. This dissertation addresses the challenge of developing a test model for simulating and subsequently identifying damage in wind turbine blades, focusing on the design of the physical prototype and the improvement of damage detection methods. The problem is particularly relevant due to the operational demands on the blades, which are exposed to harsh environmental conditions. If damage is not detected in a timely manner, it can lead to efficiency losses or even catastrophic failures. While traditional detection methods are useful, they may lack precision in identifying early-stage anomalies, making this issue challenging and significant in the context of wind energy maintenance. To address this problem, the proposed solution involves developing a computational model of the blade, based on the NREL S809 aerodynamic profile, optimized for both aerodynamic efficiency and structural integrity. In addition to providing a detailed design, this work introduces a new approach to damage identification, based on the "Cloud of Spheres"technique, as an alternative to the Mahalanobis Distance- based method. This technique offers the ability to detect anomalies in multidimensional data spaces, providing a more intuitive and robust solution for structural health monitoring. The results demonstrate the potential of the implemented methodology. The "Cloud of Spheres"technique proved effective in distinguishing between damaged and undamaged states of the tested structures, validating its applicability in future physical prototypes and testing scenarios. The implications of this research are substantial, as they lay a solid foundation for future advancements in damage detection technologies, contributing to more reliable and efficient wind energy systems.
Descrição
Palavras-chave
Pás de turbinas eólicas Identificação de dano Modelação computacional Monitorização estrutural baseada em Cloud of Spheres
