Evaluating Fragility of Interdependent Design Spaces to Quantify the Risk of Space Reduction Decisions in Set-Based Design

Abstract

The litany of decisions made over the course of any marine vessel design have significant impacts on the design’s outcome. Iterative approaches, such as point-based design (PBD), make precise decisions on vessel characteristic values as necessitated by the sequence in which interdependent design activities are performed, but these decisions make PBD susceptible to inefficient rework cycles. To reduce the frequency and magnitude of rework, vessel char- acteristic assignments often coincide with previous designs of similar vessels, even if those designs are suboptimal. Contrarily, convergent approaches, such as set-based design (SBD), allow for more ambiguous definitions of vessel characteristics by making decisions that re- move unfavorable values while keeping ranges of values open. With this approach, SBD depends less on experience to reduce variable sets because that experience is supplemented by various sampling and modeling tools unique to each discipline. However, each discipline cannot thoroughly sample their design space, and it is common for modeling and requirement changes to transpire over the course of a design. Design managers need to reduce design spaces to meet project timelines and budgets, but in doing so, they risk having unexplored areas or design changes reveal important information that would have impacted those reduc- tions. Design managers currently lack any method that quantifies and regulates this risk. A novel way to do so may be through the creation of a framework and metrics rooted in Information Theory. Before developing a strategy to assess the risk of space reductions, this work explores the sequential decisions that make up a conventional PBD approach through a simple polynomial model. The formalization of designer influence and consequential re- work path properties relating to decisions made earlier in a sequence prompts a shift in approaching design through an iterative method to a convergent SBD method. Background research is conducted on entropy metrics from Information Theory that can evaluate design space fragility, and the creation of a framework with these metrics is proposed to quantify the risk of space reductions. Plans are set to test the framework and metrics with a new SBD problem and to develop a multidisciplinary adaptive sampling approach that alleviates high-risk reductions.