Abstract
Interest in Concurrent Engineering (CE) has increased as industry looks for more efficient means of product design. Design optimization methods that facilitate the CE approach are an important aspect of current research. Among the methods that have been proposed is the Concurrent Subspace Optimization (CSSO) method, which allows the optimization problem to be decomposed into coupled subproblems. These subproblems may correspond to the different disciplines involved in the design process or to participating organizational design or manufacturing groups. The decomposition allows each discipline to apply their own optimization criteria to the problem. While this method may not be as computationally efficient as other methods, it allows the design process to conform to the departmental divisions that already exist in industry.
The method development to date has focused on continuous systems only. However, problems that can not be modeled as continuous systems, such as those involving the placement of active controllers in CSI applications, would benefit from a method that allows the use of discrete parameters. The paper presents a decomposition method (based on CSSO) for the optimal design of mixed discrete/continuous systems. The method is applied to the design of a composite plate for minimum weight, with design variables contributed from sizing variables (continuous) and material combinations (discrete).
Metric and Bregman projections onto affine subspaces and their computation via sequential subspace optimization methods
