A parallel decomposition algorithm in application to structural design

Abstract

During previous years, structural optimization has been recognized as a useful tool for the systematic design of components. On the other hand, the optimization of large-scale systems or structures is impeded by an immense solution effort. In this paper, decomposition and parallelization strategies are presented for that type of problem and applied to the shape optimization of complex shells of revolution. Here, the point of emphasis is laid on the development of a PARallel DEComposition strategy, called PARDEC, for the solution of very complex optimization problems which cannot be solved efficiently by sequential integrated optimization. For that purpose, the global optimization problem will be partitioned into smaller, parallel solvable subproblems. By means of a coordination strategy, these subproblems will be forced to adopt those subsolutions which are optimal and compatible for the global optimization problem. The subtasks ‘structural analysis’ and ‘sensitivity analysis’ within the optimization process are parallelized for rendering sequential integrated optimizations more efficiently. The coordination strategy facilitates the solution of arbitrary problem formulations, including multidisciplinary problems. All realized parallel strategies are implemented on parallel computing systems and are verified by shape optimization of an automotive wheel.