Designing Better Design Teams: Studying Relative Contribution in Engineering Design with Proficient Heterogeneous Computational Agents

Teams are the fundamental unit of performance for modern organizations, especially in engineering design. To increase organizational performance, teams must be constructed to maximize their effectiveness. However, many questions exist about how to compile a team or how individual team members affect team performance. A laboratory experiment is conducted to determine the relative contribution of different proficiency team members in configuration design. The team’s highest proficiency designer is found to have the largest effect on team performance, and enhancing the most proficient member of a team is more likely to contribute to increased team performance than enhancing the least proficient member by the same amount. This study also examined the behavioral differences between different proficiency designers. It determined that relative to lower-proficiency designers, high-proficiency designers spent less time exploring the design space, and each proficiency level had distinct action selection heuristics.

These two behavioral differences are computationally modeled, and the Proficient Simulated Annealing Design Agents (PSADA) model is introduced. The PSADA model is shown to mimic different proficiency human designer behavior and produces agents that perform at different proficiency levels. When agents are put into teams to design collectively, the results from the human subject study are reproduced to show that configuration design teams are most dependent on the proficiency of their best member. Because proficiency is not a static characteristic, learning algorithms are introduced to create the Learning PSADA (LPSADA) model. Lower proficiency LPSADA agents are shown to learn to behave like higher proficiency designers and improve their performance. In short, they learn to become higher proficiency designers.

Finally, the PSADA model is integrated with a computational model for organizational personnel selection to explore how organizations can optimize performance by utilizing this deeper understanding of proficiency’s role in engineering configuration design. The PSADA model is used to create team performance models based on the proficiency of the team members. These models are used as objective functions for an optimization algorithm for personnel selection to optimize organizational performance metrics like product quality and personnel hour usage. This optimized personnel selection method outperforms other personnel selection methodologies.

This dissertation deeply examines designer proficiency in engineering configuration design, assesses how the configuration design proficiency of each team member affects team performance, and explores how organizations can utilize a greater understanding of the relative contribution of team members to maximize team and organizational performance through multiple team personnel selection.