Performative Structural Design Optimization: Generative Algorithm for a Preliminary Study of a Voided Beam

Abstract

In the world of structural design, in most cases, there is a need to control the shape of structural elements and—at the same time—the performance that each one can achieve. With the evolution of structural analysis tools, nowadays it is possible not only to have an immediate investigation of the structure’s performance, but also to search for the best shape by imposing geometric constraints. The aim of this paper is to present an innovative methodology called the performative structural design optimization (PSDO) method, based on the use of algorithm-aided design (AAD). The proposed approach deals with an emptied voided beam; starting from the parameterization of a large-span beam, the search method for the most performing shape is accomplished by multi-objective evolutionary algorithms (MOEAs). The obtained results are characterized by a double optimization: the structure achieved by the hypervolume estimation algorithm for multi-objective optimization (HypE Reduction) (OCTOPUS) represents the starting shape for the application of form-finding, giving so the possibility to obtain different feasible solutions from a single study and to choose the best one in terms of structural behavior.