TNAG-1506-FCO40-2064-Guangdong-nuclear-power-station-project-at-Daya-Bay-safety-c-1986 — Page 154

APPENDIX 2

THE EXPECTED OUTPUT OF A PILOT RISK STUDY

•

The Pilot Risk Study method which is intended for application to the Daya Bay nuclear plant is designed to provide rough estimates of the following

quantities relevant to an appreciation of the risk profile and its derivation from the plant design.

1) Identification of dominant accident ́sequences leading to core melt

and estimation of their probabilities of occurrence.

2) Identification of dominant accident sequences which lead to core

melt and failure of the containment owing to system failures. Est- imation of the probabilities of occurrence of same.

3) Identification of the physical processes which may lead to cont- ainment failure following a core melt sequence which does not prejudice containment systems. Estimation of probabilities of failure based on past studies and a review of the phenomenology.

4) Estimation of radioactive releases following core melt accidents,

based on past studies.

The information can be used in conjunction with consequence calculations to estimate the probabilities of casualties in specified populations near the plant.

Qualifications

1

Because of its pilot ' nature, the PRS output suffers from some uncertainties additional to those which occur in full Probabilistic Risk Assessments. The most important of these additional uncertainties are

:

i) Because detailed systems analysis is not performed, system behaviour

is estimated on the basis of generic information from similar systems (eg the historical records of similar system failures as given in the Oak Ridge Accident Sequence Precursor Study). If the system under study does not conform to the generic data, an error may arise. This can only be removed by detailed system information and analysis.

ii) Where information is lacking for example where designs are not

finalised - assumptions have to be made, and these may not reflect the eventual state of the plant. However, it is not expected that a future plant would be constructed and operated to significantly lower standards than have been achieved by the generic population of current plants, so the assumption of generic performance should be pessimistic if anything.

-

The uncertainties which affect PRA, regarding human factors, containment phenomenology, source terms etc, will of course affect the pilot study results as the pilot study is based on past PRA for those aspects which cannot be derived directly from experience.