A methodology to incorporate uncertainty into R&D cost and performance data

Abstract

This paper describes and demonstrates a methodology developed for the Electric Power Research Institute that combines modeling, sensitivity analysis and subjective probability encoding to analyze the cost differences between projects competing for R&D funds. The methodology involves two distinct stages: a deterministic stage and a probabilistic stage. The deterministic stage sets ground rules for comparison, compiles variables and ranges, specifies a cost model, identifies key variables by sensitivity analysis, and reviews the outcome by participants involved. The probabilistic stage determines probability distributions for the key variables and then calculates probability distributions for project costs based on the cost model and distributions of key variables. The paper then demonstrates this methodology by describing a case study step-by-step. The case study is an EPRI comparison of an integrated gasificationcombined cycle plant with a conentional coal plant with flue gas desulfurization. Finally a critique of the methodology is given.