Using Multi Agent Systems to Computationally Study Set-Based Concurrent Engineering and its Interactions with Team Organization and Problem Structure

Concurrent engineering presents opportunities for engineering design teams to develop the parts of a project in parallel and resolve problems as they appear. This allows teams to complete projects faster and create better solutions. This concurrency is limited by the communication needed to coordinate subproblems, and failure to coordinate them requires costly rework, however. Set-based concurrent engineering (SBCE) proposes a method to reduce the required communication and resulting rework by maintaining many candidate designs for each subproblem throughout a project, rather than only one candidate for each subproblem. Maintaining many candidates reduces required communication and rework by ensuring they have compatible options to combine with rather than needing constant coordination and adaptation as the project progresses. Despite these proposed benefits, SBCE meets resistance to adoption as maintaining many design candidates necessitates dividing development between them, even if only one candidate for each subproblem is used in the final product. 

These competing benefits and drawbacks indicate that a greater understanding of SBCE is desired. While SBCE has been received positively in industry, most businesses learn of it through academic study, indicating that it is still in the research phase. Additionally, thorough investigations of SBCE to develop this understanding have yet to be conducted. Human subjects studies to do this would be prohibitively large, however, as the competing benefits and drawbacks require searching of intermediate values of experimental parameters. Therefore, the computational study of SBCE is desired to investigate how it interacts with various problem and team properties for better understanding of it and its applications.

To do this, this work introduces the Point/Set-Organized Research Teams (PSORT) framework and validates it against statements made in industry studies of SBCE. This work then applies the PSORT framework to study how SBCE interacts with various team and problem structures, using problem coupling and varying timespan to further elucidate the impacts of the process. This work suggests that SBCE lowers the need for communication and rework on coupled problems, allowing projects to achieve greater concurrency, and that SBCE can synergize with communication to break through barriers in the design space to create new designs. These benefits are most strongly expressed in large projects with many subproblems and agents. In contrast, SBCE is suggested to be detrimental in smaller and shorter projects where available development effort is limited, or on uncoupled problems where SBCE’s division of development over many designs results in greater quality losses and its benefits to coordination are not needed. These results pose new hypotheses that may serve as the basis for further industry case studies or human subjects studies on new problems.