(273b) Fuzzy Decision-Making for Sustainability Performance Improvement of Complex Systems | AIChE

(273b) Fuzzy Decision-Making for Sustainability Performance Improvement of Complex Systems


Siddiqui, A. - Presenter, Wayne State University
Huang, Y., Wayne State University
Industrial sustainability problems are always complex because of the existence of complicated interrelated factors that affect system’s economic, environmental, and social performance and, more critically, the pervasive presence of various types of uncertainty that is associated with accessible data and information, and is reflected in problem understanding and possessed knowledge. These have made decision making for sustainability performance assessment and improvement very challenging, especially for predictive solution derivation. In theory, uncertainty can be classified into two types: aleatory and epistemic. While the former refers to the inherent variation associated with physical systems and the environment and is objective and irreducible, the latter is caused by the lack of knowledge and information and is subjective and reducible. Thus, sustainability under uncertainty has been a very active research area in sustainability science and engineering. Methodologically, decision making by resorting to fuzzy systems theory may offer a unique perspective in systematically handling epistemic uncertainty.

In this paper, we introduce a predictive fuzzy decision-making methodology for deriving optimal solutions under uncertainty for multistage sustainability performance improvement. By this methodology, a large-scale industrial system is decomposed into a number of subsystems. For each subsystem, fuzzy decision making is made to satisfy its set sustainability goal for a number of development stages. The decisions derived in this lower layer of the decision hierarchy will then be analyzed and coordinated at the higher layer for the entire system’s goal achievement in multiple stages. The coordinated decisions will be transmitted back to the lower layer for each subsystem to adjust its previous decision. This recursive process will be terminated when the sustainability goal for the entire system is achievable. The methodology is systematic, computationally efficient, and suitable for analyzing sustainability status of any system, and predicting its short-to-long-term behavior. The efficacy of the methodology will be demonstrated through a case study on the short-to-long-term sustainability of an industrial system involving chemical manufacturing, metal finishing, and automotive manufacturing.