Topology optimization of a benchmark artifact with target stress states using evolutionary algorithms

Abstract

Additive manufacturing enables extended freedom in designing structural components. In order to reduce manufacturing costs, the product quality has to be assessed early in the process. This can be done by benchmark artifacts which represent critical quality measures of the part in production. As yet there is no integral approach to design a benchmark artifact that characterizes the quality of additively manufactured components based on structural properties. As a first investigation, this study introduces a method to optimize the topology of a benchmark artifact that represents pre-defined critical stresses. In this way, structural properties of an additively manufactured part can be efficiently characterized. The approach includes a basic example with trivial target stresses for which a reference solution is a priori known. Non-trivial target stresses were investigated to present structural solutions close to application. Evolutionary optimization algorithms were used for solving the multi-objective formulation of the problem. An appropriate formulation of the optimization problem was identified to generate plausible solutions robustly. It included additional constraints to the variation of stresses in the neighborhood of the pre-defined stresses as well as a scaling factor of all element densities. A comparative optimization with gradient methods exhibited solutions inferior to the proposed approach.