ABSTRACT
The original MS is an integrated system giving that many feedback and feed-forward activities among subsystems. In order to lesson the burden of the feedback and feed-forward computation, auxiliary design variables and compatibility constraints are introduced in the distributed system approach. The auxiliary design variables are “copies” of the computed linking variables from upper-level to lower-level or vice versa (Kroo et al., 1994). Inconsistency exists among these sharing variables and responses among each distributed systems in the middle of optimization process. The measure of such inconsistency is called discrepancy functions and the compatibility constraints are used to ensure convergence is achieved. Once MS is decomposed into several subsystems in MDO, the goal of upper-levels is to minimize the system overall objective function, while the lower-levels (i.e., subsystems) attempt to minimize discrepancy functions that are the inconsistency (i.e., deviations between targets and responses) among the original variables (i.e., linking variables) and the auxiliary variables.
