Margin allocation and trade-off in complex systems design and optimization

Abstract

Presented is an approach for interactive margin management. Existing methods enable a fixed set of allowable margin combinations to be identified, but these have limitations with regard to supporting interactive exploration of the effects of: 1) margins on other margins, 2) margins on performance and 3) margins on the probabilities of constraint satisfaction. To this purpose, the concept of a margin space is introduced. It is bi-directionally linked to the design space, to enable the designer to understand how assigning margins on certain parameters limits the allowable margins that can be assigned to other parameters. Also, a novel framework has been developed. It incorporates the margin space concept as well as enablers, including interactive visualization techniques, which can aide the designer to explore the margin and design spaces dynamically, as well as the effects of margins on the probability of constraint satisfaction and on performance. The framework was implemented into a prototype software tool, AirCADia, which was used for a qualitative evaluation by practicing designers. The evaluation, conducted as part of the EU TOICA project, demonstrated the usefulness of the approach.