A Decision Support Framework for Robust Multilevel Co-Design Exploration of Manufacturing Supply Networks

Decision-making in the design of Manufacturing Supply Networks (MSNs) is complex owing to the need to account for decisions made by groups at multiple levels and the interactions that include potential conflicts and uncertainties. Decisions are made based on information available from simulations that employ computational models. These models are abstractions of reality and are therefore sources of uncertainty. This necessitates that the focus be placed on design space exploration to identify a robust satisficing solution set. Hence, the need to support the efficient exploration of multilevel design/solution spaces, simultaneously. Many of the frameworks that facilitate multilevel design employ optimization-based iterative approaches that are computationally expensive and ill-suited for design exploration. Frameworks that support robust design exploration are limited by their capability to support the efficient exploration of multilevel design/solution spaces, simultaneously. Hence, the requirement to facilitate the 'multilevel robust co-design and solution space exploration' of MSNs. In this paper, we present a framework that allows designers to: i) model robust decision-making and their interactions across multiple levels, and ii) visualize and systematically carry out co-design exploration of multilevel design/solution spaces. In the framework, we combine a Preemptive formulation of coupled Decision Support Problem with robust design constructs and interpretable-Self-Organizing Maps (iSOM) based visualization to facilitate multilevel robust co-design. We use a steel MSN problem with decisions being made at two levels to test the framework. Using the problem, we demonstrate the utility of the framework in supporting designers in modeling multilevel decision-making under uncertainty and their interactions, and efficient co-design exploration of the robust design/solution spaces in MSNs. The proposed framework is based on information and decision flows, making it generic and capable of supporting designers in the robust co-design of multilevel systems.