This research project aims at proposing a topology optimization framework to solve high resolution 3D topology optimization problems considering local stress constraints and manufacturing uncertainties. In order to achieve such a goal, the stress-constrained formulation presented in Da Silva, Beck and Sigmund (2019) is employed, which follows the three-field density approach developed in Sigmund (2009) and Wang et al. (2011) to handle uniform manufacturing uncertainties during the optimization process. The main difficulties associated with such a formulation comes from the local nature of the local failure criterion and the high computational cost related to the large 3D discretized design spaces. These difficulties are handled as follows: 1) in order to deal with the large number of stress constraints, the augmented Lagrangian method is employed; and 2) in order to deal with the high computational cost, the powerful topology optimization framework recently developed by Aage, Andreassen and Lazarov (2015) based on the freely available library for high performance and scientific computing PETSc, will be exploited.
News published in Agência FAPESP Newsletter about the scholarship: