A Problem Decomposition Method for Conceptual Design

In this paper, based on findings from situated cognition, we argue that human reasoning is characterized by an ability to dynamically re-organize knowledge available in an experience, and re-construct it for use within the same experience. This paper describes a computational method that models such a knowledge re-organization process. We show its application on the Design Structure Matrix (DSM) formulation of an automotive design problem to demonstrate a design decomposition strategy. The method infers decomposition decisions by uncovering information from syntactic representations of design problems, i.e. it reformulates the problem by a purely syntactical analysis of a formulation. In general, it infers global, implicit design knowledge from local, explicit information in the design representation, and shows that a single problem representation can encode multiple layers of ‘design meaning’. Designers can use the method to explore problem decomposition decisions for large/complex systems in an interactive manner.