Otimização ao impacto de estruturas do tipo honeycomb via funções de base radial

• Ano de defesa: 2019
• Autor(a) principal: Rodrigues, Matheus Toneli
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Mecânica e de Materiais; UTFPR
• Palavras-chave em Português: Otimização estrutural; Deformações (Mecânica); Veículos - Medidas de segurança; Placas metálicas - Testes de impacto; Métodos de simulação; Engenharia mecânica; Structural optimization; Deformations (Mechanics); Vehicles - Safety measures; Plates, Iron and steel - Impact testing; Simulation methods; Mechanical engineering; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA; Engenharia Mecânica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/4157

Crashworthiness design has become an important topic in automotive industry due to growing concerns with vehicle safety. However, the use of impact energy absorbers should not increase significantly the vehicle weight, because of the high standards of fuel consumption that must be satisfied. Within this context, this thesis proposes a method to optimize the out-of-plane crush behavior of honeycombs, well-known structures for their high-energy absorption capacity and low weight. Due to high computational costs involved in crush simulations, the method proposed employs metamodeling techniques to approximate the response of a finite element model built using the commercial software ABAQUS. In order to find optimal honeycomb configurations, the surrogate model is sequentially improved from the outcome of new impact simulations (infill points), using two approaches: the minimization of a surrogate model predictor and the expected improvement method (EI). The expected improvement method was initially developed to be used in combination with Kriging, but it is applied with radial basis functions (RBF) in the present work. Honeycomb’s cell size, cell shape and thickness are the design variables. The optimization results show a significant improvement compared to the initial design in terms of specific energy absorbed, while peak force values are maintained at low levels. Moreover, the hexagon cell shape seems to have a higher out-of-plane performance compared to rectangle and reentrant auxetic cells. Concerning the proposed surrogate-based optimization method, the algorithm shows a satisfactory performance, solving different singleobjective optimization problems from a reduced number of finite element simulations. This demonstrates the benefit of applying RBF with infill strategies when dealing with large computational time problems, such as crush simulations. Lastly, a multiobjective optimization using only the RBF predictor (without sequential sampling) is carried out to seek simultaneously for optimal solutions with maximum specific energy absorption and minimum peak force.