RELIABILITY ENHANCEMENT OF SEISMIC RISK ASSESSMENT OF NPP AS RISK MANAGEMENT FUNDAMENTALS PART III: SENSITIVITY ANALYSIS FOR THE QUANTIFICATION OF EPISTEMIC UNCERTAINTY ON FRAGILITY ASSESSMENT

摘要

This study focused on uncertainty-assessment frameworks, the utilization of expertise, and the development of relevant software to improve the reliability of seismic probabilistic risk assessment (SPRA) and promote its further use. In addition, a methodology was developed for quantifying the uncertainty associated with the final SPRA results within the risk management framework of nuclear power plant facilities. This research aimed at contributing to (1) the development of probabilistic models for uncertainty quantification and software, (2) the aggregation of expert opinions on structure fragility estimation and development of implementation guidelines on epistemic uncertainty, and (3) the study of the applicability of newly proposed SPRA models to plant models. In particular, we focused on the second goal. Two different groups of experts, those in the field of civil engineering and those in the field of mechanical engineering, were consulted in the study. With these groups, we conducted a pilot study on the use of expert-opinion elicitation for the identification and quantification of fragility assessment parameters. Some sensitivity analyses were performed using a three dimensional reactor-building model and a conventional evaluation model, and the results were provided to the experts for expert-opinion elicitation. The results of the sensitivity analyses were related to the uncertainty evaluation of the buildings and soil to evaluate the fragility of the equipment. In this paper, those results will be shown in comparison with a conventional evaluation model.