Achieving Pareto Optimality in a Decentralized Design Environment
Tomonori Honda, Francesco Ciucci, Maria C. Yang · 2009
As engineering systems grow in complexity, the teams that design them require increasingly disparate expertise and must operate in a distributed fashion. At the same time, the teams that design subsystems need to compete and compromise with each other for a limited set of resources. Thus, it becomes crucial to establish a system-level understanding of the trade-offs between subsystems. However, there is little research regarding formal design methods for determining rational designs in a decentralized environment. Lewis and others have developed an effective Game Theoretic approach based on Decision Theory to locate a Nash Equilibrium design with minimum information sharing. This paper presents a design technique to balance tradeoffs between subsystem and system performance by minimizing information sharing between subsystems, thus converging to a set of Pareto Optimal designs. In this research, designers pass a quadratic approximation of each subsystem's objective functions at each iteration. This approach is illustrated by case examples of a pressure vessel and an airplane that show how Pareto Optimal Designs may be obtained in a decentralized design environment.