Algoritmo genético como ferramenta auxiliar no pré-dimensionamento de estruturas protendidas

• Ano de defesa: 2019
• Autor(a) principal: Narques, Tarniê Vilela Nunes
• Orientador(a): Carvalho, Roberto Chust
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia Civil - PPGECiv
• Palavras-chave em Português: Algoritmo genético; Pré-dimensionamento; Otimização estrutural; Concreto protendido
• Palavras-chave em Inglês: Genetic Algorithm; Pre-sizing; Structural optimization; Prestressed concrete
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA CIVIL::ESTRUTURAS
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/11908

Considering the need of searching for more suitable solutions on the Civil and Structural Engineering ambit, this current research aims to assess the efficiency of the Genetic Algorithm when applied as an auxiliary tool for preliminary dimensioning, given that, in this context, the codes are more simplified once they work with a smaller number of variables, facilitating the method insertion in the project routines of the technical field. Two Genetic Algorithms were developed: one to optimise a post-tractioned prestressed concrete walkway with a “T” cross section, and another to optimise roof beams with an “I” cross section from a shed made of a precast concrete system. In order to evaluate the efficiency of the optimizing method, the solutions found by the algorithms were compared to already executed real solutions of both elements. The Genetic Algorithm, even with simplified coding, has found more economical solutions than the real solutions already implemented, in a relatively short time when compared to results from other research, validating the applicability of the method as an auxiliary tool for pre-dimensioning. Both algorithms of optimisation were structured on real coding, being implemented on them a different Crossover Operator from those traditionally adopted in structural optimization research through Algorithms.