QUANTITATIVE EVALUATION OF THE IMPACT OF HUMAN RELIABILITY IN RISK ASSESSMENT FOR NUCLEAR POWER PLANTS.

摘要

The role of human beings in the safe operation of a nuclear power plant has been a matter of concern. This study describes methods for the quantitative description of that role and its impact on the risk from nuclear power plants.; The impact of human errors was calculated by observing the changes in risk parameters, such as core melt probability, release category probabilities, accident sequence probabilities and system unavailabilities due to changes in the contribution to unavailability of human errors, within the framework of risk assessment methodology. It was found that for operational pressurized water reactors the opportunity for reduction in core melt probability by reducing the human error rates without simultaneous reduction of hardware failures is limited, but that core melt probability would significantly increase as human error rates increased. More importantly, most of the dominant accident sequences showed a significant increase in their probabilities with an increase in human error rates. Release categories resulting in high consequences showed a much larger sensitivity to human errors than categories resulting in low consequences. A combination of structural importance and reliability importance measure was used to describe the importance of individual errors.; The multiple sequential failures (msfs) during testing, maintenance and calibration were analyzed and distinguished from other types of multiple failures by considering the relationship among the failures. A model was developed for describing msfs by taking into account the process involved in such failures. The model increments the conditional failure probabilities by certain amounts from their lower bounds (independent failure probabilities). This approach provides important insights into the influence of dependence between failures on system reliability. This model can be used effectively to choose an optimum system considering the individual failure probability, dependence factor and the amount of redundancy in a system. It was observed that in many cases it may be better to reduce the individual failure probability and to use a different type of system, rather than trying to decrease the dependence between the failures.