The use of probability elicitation in the high-level nuclear waste regulation program

Abstract

Expert judgement elicitation is expected to be used in the performance assessments (PA) of the long-term behavior of high-level waste (HLW) geologic repositories. As a preparation for an effective review of the U.S. Department of Energy (DOE) PA, the Nuclear Regulatory Commission (NRC) is evaluating the mechanics of eliciting expert judgements. One of the objectives of this evaluation is to explore techniques for generating and aggregating probabilistic judgements of future conditions at the proposed HLW repository at Yucca Mountain, Nevada. An actual elicitation was conducted as an aid to these evaluations. This paper documents this probabilistically centered elicitation and subsequent activities to explore aggregation of opinion techniques. Future climate in the Yucca Mountain, Nevada vicinity was selected as the topic for elicitation. Personnel from the NRC and Center for Nuclear Waste Regulatory Analyses (CNWRA) defined the climatic parameters of interest in conjunction with a panel of five expert climatologists. Individual elicitations were performed with each climatologist to produce probabilistic estimates of each parameter at seven points of time in the future. The elicitations employed the fractile technique to generate cumulative probability distributions representing the uncertainty in the predictions. After the individual elicitations, a group session was conducted to explore aggregation and consensus methods.