The Probability of Future Human Actions Affecting a Geological Repository for Nuclear Wastes: Developing a Model that Incorporates Information Science and Historical Perspectives

摘要

A prototype computer model of future human actions affecting a geological repository has been developed based on a Markovian information state model previously outlined by Woo [1,2]. An expert elicitation procedure has been used to develop the model and data for a trial application to the case of a proposed deep underground repository at the Sellafield site. The experts rejected several assumptions that commonly underlie assessment of future human actions, and preliminary analysis indicates that, on this basis, higher risks are forecast than by a Poisson model based on historical evnet frequencies.