A heuristic sequencing procedure for sequential solution of decomposed optimal design problems

The efficiency of designing a large-scale structure with many variables can often be increased by decomposing the design problem into less complex sub-problems. This article presents heuristic procedures of determining an appropriate design sequence to solve decomposed structural optimization problems using a sequential design (singlepass) strategy. One heuristic procedure exploits the monotonicity of a global constraint, which is included in all the sub-problems. Another procedure is based on the coupling strength of interactions between sub-problems, obtaining the appropriate design sequence and then verifying it with monotonicity analysis. The procedures are applied to two types of structural design problems; the metric for sequencing decomposed sub-problems can be obtained analytically for one design problem and numerically for another one. As a result, the sequencing procedures lead to satisfactory solutions at low computational cost, indicating their value for industrial product development.