Session: 16-01-01: Government Agency Student Poster Competition

Paper Number: 150896

150896 - Co-Design Exploration Framework for Multilevel Decision Support Under Uncertainty 

In conventional manufacturing, such as steel production, the fabrication of the end product entails a sequence of manufacturing processes. The properties and performance of the final product depends on the processing history of the materials and the microstructure formed at each process. To achieve the desired product performance, it is essential to concurrently design and explore the material microstructures and processing techniques. This requires considering the multilevel interactions among the material, the product, and the manufacturing processes while also managing the uncertainties involved. A multilevel design requires coordinating the interactions between the levels and identifying solutions that meet the requirements set by the designers at each level of the multiple design levels. This necessitates the capability to co-design, which involves simultaneously exploring a wide range of "satisficing" solutions across multilevel by managing uncertainties supporting robust design decision-making. The co-design approach allows decision-makers or designers at multiple design levels to exchange their knowledge, information, and resources collaboratively with each other in order to achieve a simultaneous exploration of the design space.

In this poster, we present a co-design exploration framework for multilevel decision support and management of uncertainty. Utilizing the framework, we model the interactions and couplings between the levels, which enables the formulation and simultaneous decision-based design exploration of multilevel decision support problems. The framework integrates the coupled-compromise Decision Support Problem (c-cDSP) with the Error Margin Index (EMI) and Design Capability Index (DCI) metrics for robust design along with the interpretable Self-Organizing Maps (iSOM) to support simultaneous multilevel design space exploration. The c-cDSP supports the designer to formulate multiple conflicting goals across multiple levels. The EMI and DCI metrics support the formulation of robust design goals for different sources of uncertainty. The iSOM is a machine learning based visualization tool for visualizing and exploring high dimensional data using 2 dimensional plots. Designers are able to use the framework to (i) formulate multilevel decision support problems under uncertainty considering the interactions between the levels, (ii) simultaneously visualize and explore the multilevel design spaces, and (iii) identify "satisficing" robust solutions in the early stages of design. A hot rod rolling problem is used to test the efficacy of the framework. We showcase the use of the framework for co-design exploration during the thermo-mechanical processing of hot rolled steels. The interactions between the material microstructure during the dynamic and metadynamic phases of recrystallization, with the steel properties such as flow stress, are considered in the problem. The framework is generic and supports the co-design exploration of engineered systems characterized by multilevel interactions, couplings and multidisciplinary decision makers for decision support under uncertainty.

Presenting Author: H M Dilshad Alam Digonta Florida Institute of Technology

Presenting Author Biography: A doctoral student at Florida Institute of Technology. Working in the Systems Realization Lab at Florida Tech under the supervision of Dr. Anand Balu Nellippallil. His doctoral work is supported by the National Science Foundation under Grant No. 2301808.

Authors:

Anand Balu Nellippallil Florida Institute of Technology
H M Dilshad Alam Digonta Florida Institute of Technology