Efficient Representation and Reduction of Extreme Uncertainty in Environmentally Benign Design and Manufacture

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

The primary goal of this project is to determine which uncertainty representations and corresponding decision methods are best suited for decision-making under the extreme uncertainty common in Environmentally Benign Design and Manufacture. Three methods will be developed and compared: utility-based statistical decision theory; an extension of utility theory based on Probability Bounds Analysis; and Information-Gap Decision Theory. In these methods, the uncertainty due to natural random behavior in physical processes (aleatory uncertainty) and the uncertainty due to a lack of information (epistemic uncertainty) are considered explicitly. The challenge is to determine which uncertainty representation is best suited for a particular combination of aleatory and epistemic uncertainties. This is achieved through computational experiments, which will be validated in industrial case studies. In addition, a method is developed to guide the efficient gathering of additional information when currently available information is insufficient to support a particular decision in Environmentally Benign Design and Manufacture. This research will directly impact society by pursuing research that aims to reduce the environmental impact of industry while increasing its efficiency, creating an economic and ecological win-win situation. It directly supports industry by developing new models, methods and tools that can be used to assess and reduce the environmental impacts during design and manufacture. At the fundamental research level, this research will result in theoretical contributions toward the characterization of uncertain information and knowledge, and the ability to make decisions based on this uncertain information. The research results will be integrated into several undergraduate and graduate courses.