Predicting and managing system interactions at early phase of the product development process

Abstract

The activity of designing and developing large, complex, discrete, physical, and engineered products faces the challenges in the physical product system, the organization of people, and the larger systems in which the product resides-the natural and societal systems. This thesis defines system interactions as the interactions amongst design variables within the physical product. Knowing system interactions early in the product development process is critical for project management, design concept selection, and system architecture decisions. However, existing methods that address the system interactions issues, such as the Design Structure Matrix (DSM), are good analysis tools, but cannot be used during conceptual synthesis when the most important decisions about the system designs are made. System level knowledge is defined as the knowledge concerning system interactions. System level knowledge is organizational knowledge that resides in the collective minds of members in the organization. System level knowledge is critical to the success of the design of large systems, yet is often missing due to its empirical nature. A knowledge management framework was proposed in this thesis and tested in industry cases from Ford and CVC. This thesis developed a method to predict and analyze system interactions at early phase of the design process. The method transforms an Axiomatic Design's Design Matrix (DM) into a DSM based on solving systems of linear equations using substitution. Since a DM is more easily constructed during early design phases, we can use this method to obtain a DSM during concept design.

Consequently, the advantages of the DSM system analysis tools and methods can be applied to make better decisions on system design, system architecture, and project management. The method was tested using two industry cases at CVC and Johnson and Johnson Ortho-clinical Diagnostics. Both case studies showed that the method was effective in real engineering projects. Further observations in the case studies also revealed that a DSM could also be easily converted back into a DM. The interchangeability between DSM and DM allows engineering organizations to predict system interactions early on in a project, while capturing and managing system level knowledge throughout the product lifecycle.