Error propagation of optimal system design in a hierarchical enterprise

Abstract

Increased computing power has helped virtual engineering become common practice amongst product development firms. However, while capabilities increase, the desire to simulate even larger systems has increased as well. To deal with the complexity and size of these systems, several techniques have been developed to decompose the system into smaller, more tractable subsystems. The drawback of this approach is a substantial decrease in computational efficiency. Therefore the use of simplified models is encouraged and often required to reach convergence.In this thesis, a test model is introduced where different forms of error can be introduced at each level. Error derived from both measurement inaccuracy and modeling inaccuracy is examined coupled with the effect of system constraints as well. A hierarchical decomposition method is selected for its similarity to a typical enterprise organizational structure. In this manner, the results of the examination should be applicable to both system engineering methods and enterprise level problems. The direction of error propagation within the hierarchical decomposition is determined and the effects of robust design considerations and simple system constraints are revealed.