Project proposal regarding quantification quality of Nordic PSAs.
Recent publications and workshop initiatives for next generation PSA (“Open PSA”) have shown that current fault tree quantification techniques using a limited (truncated) set of minimal cut sets combined with rare-event approximation or Min Cut Upper Bound (MCUB) easily yield results with unknown error, even unknown sign of the error. Especially importance measures can be underestimated because a large portion of basic events may not even contribute to the truncated model. In risk-informed applications such errors can cause incorrect ranking of systems, components, structures and human actions, and yield incorrect allowed configuration times ACT. Serious unknown risks or unnecessary costs can be involved. Other causes for concern are methods for handling of large probabilities and success branches of event trees (and negations and not gates in general). Originally importance measures have been defined only for coherent models (with no negations). The credibility of PSA and the quality of decisions can be improved significantly by identifying and correcting these weaknesses.
The work consists of the following tasks:
- Project plan and questionnaire for Nordic nuclear utilities (plants)
- Assessment of the errors/accuracy of Nordic PSA quantification methods and importance measures
- Identification of deficiencies and needs for improvement in quantification
- Identification of solutions and experimenting with potential solution techniques
- Software developers to be advisors on code features and to implement software changes and to help in testing suggested improvements
- Recalculation of PSA results with improved techniques
- Summary and conclusions