A value of information methodology for multiobjective decisions in quantitative set‐based design. Issue 6 (22nd July 2021)

• Record Type: Journal Article
• Authors: Shallcross, Nicholas J.; Parnell, Gregory S.; Pohl, Ed; Goerger, Simon R.
• Is Part Of: Systems engineering. Volume 24:Issue 6(2021)
• Page Start: 409
• Page End: 424
• Publication Date: 2021-07-22
• Subjects: design maturation decisions -- multiobjective decision analysis -- set‐based design -- uncertainty reduction -- value of information; Systems engineering -- Periodicals; 620.0011
• DOI: 10.1002/sys.21593
• Languages: English
• ISSNs: 1098-1241

Abstract

Abstract: Engineering complex systems is an exercise in sequential multiobjective decision making under uncertainty. One method for handling this complexity and uncertainty is set‐based design (SBD). SBD is a concurrent engineering and management methodology that develops, analyzes, and matures numerous design options, reducing risk and delivering higher value to the stakeholders and end users. SBD accomplishes this through controlled design space convergence which reduces uncertainty and prevents premature design decisions. While SBD has been the subject of numerous scholarly articles, there is limited research providing quantitative methodologies that inform decisions enabling design maturation and convergence. We present a value of information (VOI) based methodology for multiobjective decision problems, and demonstrate its applicability for SBD decisions. We apply Bayesian decision models and information value to inform multiobjective modeling and design maturation decisions. Research contributions include: 1) a framework integrating VOI into the SBD process, 2) a multiobjective VOI method assessing a higher‐resolution model's ability to reduce uncertainty, and 3) a means of informing modeling decisions by comparing multiple high resolutions models, given their usage cost and their potential to deliver information value. Finally, we demonstrate the inherent issues associated with premature decisions and traditional point‐based design approaches which run the risk of selecting an alternative that later proves infeasible.