Abstract
This paper presents a unified methodology to handle variability and ignorance by using probabilistic and possibilistic techniques respectively. The methodology has been applied to the safety assessment of geological disposal of high level radioactive waste. Ignorances associated with scenarios, models and parameters were defined in terms of fuzzy membership functions derived through a series of interviews to the experts, while variability was formulated by means of probability density functions (pdfs) based on available data sets. The exercise demonstrated the applicability of the new methodology and, in particular, its advantage in quantifying ignorances based on expert opinion and in providing information on the dependence of assessment results on the level of conservatism. In addition, it was shown that sensitivity analysis can identify key parameters contributing to uncertainties associated with results of the overall assessment. The information mentioned above can be utilized to support decision making and to guide the process of disposal system development and optimization of protection against potential exposure.
