Decentralized determination of design variables among cooperative designers for product platform design in a product family

Abstract

Establishing a product family enables companies leverage mass customization to meet demand from different market segments. Product family design environment can be inherently decentralized and requires a number of entities to achieve a common goal. Issues related to product platform design can be modeled as multi-objective optimization with respect to design variables, production cost and other factors. A common drawback in decentralized environments is information sharing which can reduce the quality of the decision especially under partial information or information privacy restrictions. In this paper, we present an agent-based collaborative design model and propose a decentralized coordination mechanism to facilitate decision-making. This is achieved by modeling a collaborative design environment and applying a two-phase augmented Lagrangian-based algorithm such that the decision makers achieve a common goal without having a global view of the entire design space. We demonstrate the efficiency of the proposed mechanism using a General Aviation Aircraft and a Universal Electric Motor product family design problem. The product designers and the platform designer act independently and each has no knowledge of the decision functions of other designers. The computational results showed that quality solutions can be obtained in decentralized environments.