Development of a Quantitative Framework for Regulatory Risk Assessments: Probabilistic Approaches

The Swedish regulators have been active in the field of performance assessment for many years and have developed sophisticated approaches to the development of scenarios and other aspects of assessments. These assessments have generally used dose as the assessment end-point and have been based on deterministic calculations. Recently introduced Swedish regulations have introduced a risk criterion for radioactive waste disposal: the annual risk of harmful effects after closure of a disposal facility should not exceed 10-6 for a representative individual in the group exposed to the greatest risk. A recent review of the overall structure of risk assessments in safety cases concluded that there are a number of decisions and assumptions in the development of a risk assessment methodology that could potentially affect the calculated results. Regulatory understanding of these issues, potentially supported by independent calculations, is important in preparing for review of a proponent's risk assessment. One approach to evaluating risk in performance assessments is to use the concept of probability to express uncertainties, and to propagate these probabilities through the analysis. This report describes the various approaches available for undertaking such probabilistic analyses, both as a means of accounting for uncertainty in the determination of risk and more generally as a means of sensitivity and uncertainty analysis. The report discusses the overall nature of probabilistic analyses and how they are applied to both the calculation of risk and sensitivity analyses. Several approaches are available, including differential analysis, response surface methods and simulation. Simulation is the approach most commonly used, both in assessments for radioactive waste disposal and in other subject areas, and the report describes the key stages of this approach in detail. Decisions relating to the development of input PDFs, sampling methods (including approaches to the treatment of correlation), and determining convergence may all affect calculated results. The report discusses the key issues for each stage and how these issues can be addressed in implementing probabilistic calculations and considered in reviews of such calculations. An important element of an assessment, whichever approach is adopted for undertaking calculations, is the communication of results. The report describes the use of outputs specific to probabilistic calculations, such as probability distribution and cumulative distribution functions, and also the application of general types of output for presenting probabilistic results. Illustrating the way in which a disposal system may evolve is an important part of assessments, and the report describes the issues that must be considered in using and interpreting risk and dose versus time plots