Uncertainty in risk analysis: an alternative approach through pessimisation

Abstract

Probabilistic risk analysis (PRA) was originally devised for situations where the spread of consequences is dominated by random variables, such as those describing the weather conditions when a release takes place, which belong to a well understood and measurable frequency distribution. Careful interpretation is required, however, in other applications, where the spread of the variables represents uncertainty due to lack of data and the probability distributions are to some extent arbitrarily assigned. Practical computational difficulties can also arise in situations where the risk appears to be concentrated into a narrowly defined region around a particular set of parameter values. An iterative approach to PRA is suggested, in which a readily available computer subroutine is first used to locate any such peaks and determine the sensitivity of the risk to each of the variables near to the peak. The results are then used as a guide for reappraisal of the input data and, in particular, to ensure that the final analysis correctly represents expert opinion in the most important areas that have been identified. The method is demonstrated with a simple mathematical model of dispersal from a nuclear waste repository.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 5 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.