An Algorithm for Automated Design Variable Selection in Structural Shape Optimization With Intrinsically Defined Geometry
Abstract A major difficulty encountered in the shape optimization of structural components is the selection of an adequate set of shape design variables. The quality of the solution and the value of the optimal objective function depend on the chosen set of design variables. This paper presents an algorithm for the automated selection of intrinsically defined design variables to solve two-dimensional structural shape optimization problems. The algorithm arrives at a sufficient set of design variables by solving a series of optimization problems. Using the results of intermediate solutions, the algorithm adaptively refines the set of design variables until the solution converges. The algorithm specifies the addition and deletion of design variables and makes use of a model compatibility constraint to determine whether the analysis model must be updated. Two examples are presented which illustrate the effectiveness of the algorithm.