Uncertainty in Risk Assessment: The Representation and Treatment of Uncertainties by Probabilistic and Non-Probabilistic Methods

7

Discussion

While the risk assessment framework based on a probabilistic representation of uncertainties (probabilistic risk assessment, PRA) has proved to be a useful tool in a wide range of applications, a growing number of researchers and analysts are pointing at the limitations of the probability-based approaches for assessing risk in some circumstances. The main point advocated is that the knowledge and information (or lack of such) available for the analysis cannot be properly reflected by probabilities in all situations. To address these situations, approaches other than purely probabilistic ones have been suggested, including those described in the preceding chapters. Much of the development has been of a technical nature, with less emphasis on principles and guidelines for use in practical risk assessment. In this chapter, we provide a discussion raising some concerns and give constructive criticisms on the various approaches proposed. The discussion is taken from or based on Flage et al. (submitted), who identify and discuss five directions of development for uncertainty treatment in the context of risk analysis:

1. Probability

2. Non-probabilistic representations with the interpretation of lower and upper probabilities

3. Non-probabilistic representations with interpretations other than lower and upper probabilities

4. Hybrid approaches for combining probabilistic and non-probabilistic representations

5. Semi-quantitative methods.

In the following we use these five ...