Topology optimization for fail-safe design
In structural mechanics (especially aerospace) applications certain components are required to be fail-safe. This means, that a structure is capable to sustain a certain load even if a primary load path fails. Furthermore, optimization is used increasingly for designing structures, for instance in order to save weight. In conventional design optimization, it is possible to consider fail-safe behavior as a constraint. For that, structural elements are removed during the optimization to check fail-safe behavior. In the framework of topology optimization however, the structure emerges from the optimization. It can only be determined as a post-processing step what elements the structure consists of.First approaches for topology optimization under fail-safe constraints obviate this problem by removing larger parts of the design space instead of structural elements. This results in high numerical effort. The applicability of this approach was only demonstrated for stiffness based optimization problems, but not for considering stress as objective or constraint function.The first objective of the proposed project is the development of a method for automated identification of structural elements within a topology optimization. This method will be embedded in a topology optimization with fail-safe constraints. In each optimization step, the identified structural elements are removed from the design space and the fail-safe conditions are evaluated with respect to maximum stress. Beside the approach to consider fail-safe constraints in an optimization, more efficient, simplifying approaches exist, which allow enforcing a redundant structure in topology optimization. Whether a design obtained from such approach is fail-safe can easily be checked afterwards. However, it is unknown if the obtained design is optimal when considering fail-safe constraints. This can only be determined by comparison with approaches that explicitly take into account fail-safe constraints.Therefore, the second objective of the proposed project is the development and analysis of simplifying, efficient approaches for topology optimization of fail-safe structures.It is investigated under which conditions these approaches provide fail-safe design by applying them to significant examples and comparing these results to the ones obtained with approaches with explicit fail-safe constraints. Furthermore, it will be investigated in which conditions the obtained design are robust with respect to stochastically scattering parameters. Though robust design optimization approaches differ fundamentally from fail-safe approaches, publications show that both approaches tend to provide similar redundant designs.