Bayesian Benefit-Risk Analysis for Sustainable Process Design

Abstract

A Bayesian model for assessing uncertain and variable economics of proposed environmentally conscious and other manufacturing processes from available information is described in this paper. Economic risks may stem from environmental, health, and safety liabilities; potential product losses; and other intangibles. Benefits and costs may be uncertain due to lack of information and experience within the available time and budget, or may be inherently variable in nature. Bayesian inference allows the integrated use of available numeric data, related data, summary statistics, and professional judgment. The Principle of Maximum Entropy is used as a basis for incorporation of subjective information. A Bayesian Pareto incident-size distribution is used to model size distributions for individual costs. Sensitivity analysis is used to help identify primary elements of project economic risk. Probability distributions for planning period costs, exceedance probabilities, and other risk indicators for alternative paint-stripping projects were computed and compared for an example industrial facility. The relative importance of variability and uncertainty as a function of planning-period length is probed.