Otimização multinivel de vigas de concreto armado via algoritmos genéticos

• Ano de defesa: 2014
• Autor(a) principal: Oliveira, Lucas Façanha de
• Orientador(a): Melo, Antônio Macário Cartaxo de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Engenharia de estruturas; Concreto armado; Algoritmos genéticos; Otimização
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/25938

Reinforced concrete is a composite material that has extensive application in the construction market. It is estimated that 11 billion tons of concrete per year are used in the world. The reinforced concrete beams are structural elements that have applications in buildings, homes, bridges, etc. and are designed via traditional method of trial and error, following standard recommendations concerning to the strength, stability, rigidity and durability. After a predimensioning, run up iterative cycles of analysis and design. In designing, a requirement is used to define the project and the rest are checked. Modifications and stopping are controlled by the designer. In the optimum design method, on the other hand, a criterion of performance is set up (cost, weight, strength, etc.) and the best combination of variables is searched that satisfies all regulatory criteria and results in design with the best performance. In general, it uses cost as a performance measure and the works are distinguished by the nature of design variables and the inclusion of details of armor. The objective of this work is to formulate a model of optimization on two levels to minimize the cost of reinforced concrete beams. In the first level (global) the cost is minimized with the dimensions of the sections (discrete) and the areas of longitudinal reinforcement (continuous) as variables and constraints like strength (ULS), deformation (SLS) and ductility are considered. The detailed transverse reinforcement is already designed at this level. Strategies to include the cost of the armor cuts and anchoring without defined bar sizes are suggested. At the second level (local) the steel lowest volume of arrangement of longitudinal reinforcement is searched, considering the cuts and anchoring. The variables describe the topology of the bars in the sections and restrictions of good design practice, and geometric compatibility are checked. The linear analyzes are performed with classical beam finite element using the software FAST program and the optimization models are solved using genetic algorithm from the software BIOS. The GA parameters are calibrated with test examples and literature’s application examples are made. The sensitivity of the solution is studied for varying cost parameters of the objective function.