Robust topology optimization of compliant mechanisms with uncertainties in output stiffness

Autor(es)

Cardoso, E. L. ^[1] ; da Silva, G. A. ^[2] ; Beck, A. T. ^[2]

Afiliação do(s) autor(es)

^[1] Santa Catarina State Univ, Dept Mech Engn, Joinville, SC - Brazil

^[2] Univ Sao Paulo, Sao Carlos Sch Engn, Dept Struct Engn, Sao Carlos, SP - Brazil

• Tipo de documento: Artigo Científico
• Fonte: INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING; v. 119, n. 6, p. 532-547, AUG 10 2019.
• Citações Web of Science: 2

Resumo

This work addresses the topology optimization approach to design robust compliant mechanisms with respect to uncertainties in the output stiffness, when compared to the traditional deterministic approach. To this end, two formulations are proposed: probabilistic and nonprobabilistic. The probabilistic formulation minimizes a joint objective function of expected output displacement plus a measure of its standard deviations, for given statistical distribution of the output stiffness. The nonprobabilistic formulation is written as minimization of a joint function of the median of output displacements, plus the width of the intervals that contains the extreme values of the output displacements, for a given interval of output stiffness. The Monte Carlo simulation method is used to evaluate expected values and standard deviations of output displacements in the probabilistic formulation and to assess results obtained with the deterministic approach. It is shown that both formulations lead to designs where output displacements are less sensitive to variations of output stiffness when compared to the traditional deterministic approach. Furthermore, as an additional benefit, it is observed that large variations of output stiffness can hinder the appearance of one-node connected hinges, usually found in the deterministic design of compliant mechanisms. (AU)

• Processo FAPESP: 17/01243-5 - Otimização estrutural baseada em confiabilidade
• Beneficiário: André Teófilo Beck
• Modalidade de apoio: Auxílio à Pesquisa - Regular