On the Use of a Bayesian Updating Technique to Get Probabilistic Seismic Hazard Assessment More Rugged

Abstract eng:
Since the basic work of Cornell, many studies have been conducted in order to evaluate the probabilistic seismic hazard (PSHA) of nuclear power plants. In general, results of such studies are used as inputs for seismic PSA. Such approaches are nowadays considered as well established and come more and more used worldwide, generally in addition to deterministic approaches. Nevertheless, some discrepancies have been observed recently in some PSHA, especially from studies conducted in areas with low to moderate seismicity. The lessons learnt from these results lead to conclude that, due to uncertainties inherent to such a domain, some deterministic choices have to be taken and, depending on expert judgment, may lead to strong differences in terms of seismic motion evaluation. In that context, the objective of this paper is to point out some difficulties that may appear in the development of PSHA studies and to propose an approach that may be used to address epistemic uncertainties. The key point, which corresponds to the innovating point of the process, is the use of instrumental experience feedback to update the results of a PSHA. The method used here is based on a Bayesian updating technique including real observations as conditional events, with their own probabilistic distribution. The results presented here point out that a PSHA must be conducted in a real probabilistic spirit that is totally different from a deterministic approach (the choice of “best-estimate” or “median” input data instead of “conservative“ ones is one of the key points). In addition, logic tree procedure, which seems to be the most appropriate way to account for epistemic uncertainties, does not quantify the variability on the physical parameter itself but quantify variability on expert opinion. This may lead to an important bias in a PSHA. Finally, results from PSHA may be strongly different from real seismicity, as recorded, especially depending on previous considerations. Then, the comparison to the instrumental experience data appears to be necessary to address such difficulties. In that context, the use of the Bayesian updating technique presented in this paper may become a necessary tool to address epistemic uncertainties in PSHA and its performances could allow to get PSHA more rugged and consistent with observations.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket