Avaliação da técnica Anderson Acceleration sobre o problema de carbonatação em torres de concreto de aerogeradores

• Ano de defesa: 2021
• Autor(a) principal: Rocha, Larissa Mendes Hermógenes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Engenharia de Energias Renováveis; Programa de Pós-Graduação em Energias Renováveis; UFPB
• Palavras-chave em Português: Anderson acceleration; Método de Picard; Carbonatação do concreto; Método dos volumes finitos; Torres de concreto; Picard’s Method; Concrete carbonation; Finite volume method; Concrete towers; CNPQ::ENGENHARIAS
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/22079

With the growing energy demand, the installed capacity of wind energy has grown every year. The need to install the turbines in higher towers to make better usage of the speed of the wind has made concrete a good option as a building material. One of the problems that most affect the durability of concrete is carbonation, which destroys the protective layer in the steel reinforcement bars, in the reinforced concrete structures. Due to the complex mathematical model of carbonation, numerical methods can help to better understand the phenomenon. This work aims to solve numerically, via Finite Volume Method, the mathematical model of carbonation, and to use the Anderson Acceleration technique to accelerate the convergence of the iterative method for solving the system, the Picard’s method. To accomplish that the two-dimensional mathematical model and its numerical formulation were developed, which was implemented in Matlab ®. The implementation performance was evaluated using a benchmark case, and further, three numerical experiments were used to show the gain with the acceleration technique. In this work, it was shown that the developed computational implementation is capable of producing solutions in agreement with those in literature and Anderson Acceleration was effective in decreasing the number of iterations of the 2D examples, but not for the 1D benchmark.