Abstract
The range of application of structural optimization methods can be considerably enlarged by using decomposition techniques. In this paper a novel procedure is introduced to deal with such problems more efficiently. The mechanical structure resp. system is divided into several subsystems splitting up the design variables, objective functions, and constraints accordingly. The boundary state quantities of the subsystems and the global (i.e. subsystem overlapping) functions are approximated by a sensitivity analysis of the entire system using suitable approximation concepts. It is thus possible to optimize the subsystems independently. Variables, objective functions and constraints can be chosen arbitrarily; all coupling information is obtained from the sensitivity analysis by means of global information. The application of this technique is demonstrated by a two-dimensional shape optimization problem.