A framework for knowledge capture and a study of development metrics in collaborative engineering design
This dissertation presents a knowledge management framework, collaborative design methods, and software technologies that improve multidisciplinary, distributed engineering design. In this setting, team member coordination is non-trivial and a potential source of costly breakdowns. The difficulties arise during the capture, sharing and reuse of design knowledge, intents, decisions, and rationale. Four research phases are described. In the first two phases, we define a framework to support distributed teamwork. Collaborative methods for consensus building via private and public product models are formalized. Individual designers use generalized objects to capture knowledge and create domain-specific product interpretations while a history of their work is stored in a centralized database. We describe implementation of the framework as software prototype called VisionManager, which is evaluated on projects in mechatronic, aerospace, and facilities engineering. VisionManager's in-situ reuse mechanisms and context-dependent manipulation of the design database improve the design process by eliminating time-consuming activities related to communicating and understanding design issues. In the third research phase, we instrument VisionManager's design database to support studies of the product model evolution. Reuse, thrashing, backtracking, and redesign are quantified in experimental data collected from a facilities design course. This inspires the final phase of research---an investigation into metrics for software projects in the computer games industry. Using data from three 1-year projects, correlations are established between measurements of work and quality on a subsystem-basis. Time series analysis of daily progress reveals periodic patterns in the team's development efforts. Daily interactions among engineers, measured via concurrent work on game subsystems, correlate with each engineer's perceived value to the team. Overall, this research demonstrates that external observers can use design history to form an understanding of how a team performs. This dissertation makes three contributions to design knowledge management and its use in team-based engineering. (1) It proposes, prototypes and evaluates a modeling framework to support collaboration and distributed knowledge management for design teams. (2) It demonstrates the use of design history as a source of insight into team design processes. (3) It identifies multiple correlations between historic design data and team performance in the domain of software engineering.