Development of a decision making support system for the technical support center based on analysis of characteristics of severe accident situations in NPPs

Abstract

After the Fukushima accident, several unexpected issues including human factor issues emerged, which had not been seriously considered before the accident

some of these issues regard the simultaneous damage to multiple units by a tsunami, unavailability of precise procedures for controlling accidents, destruction of communication lines during and after the accident, and extremely stressful conditions borne by the operators. For example, the challenges of group decision-making (DM) in unexpected severe accident situations are one of the human factor issues that emerged after the Fukushima accident. Before the core damage, the operators follow the abnormal operating procedure (AOP) or emergency operating procedure (EOP) depending on the conditions. If the operators cannot prevent an accident after the AOP or EOP, the reactor core may be damaged. Consequently, the responsibility for plant control is transferred from the operators of the main control room (MCR) to those of the technical support center (TSC)

the latter must follow severe accident management guidelines (SAMGs). DM for choosing a suitable strategy is crucial at this point because the SAMGs are broad guidelines

unlike the EOP, they do not provide a precise correct answer on how to prevent an accident sequence. Therefore, several approaches have been studied to support human performance under stressful conditions

these methods (e.g., fuzzy-based DM support systems, expert systems for diagnosing the plant state, and neural network-based systems for predicting accident sequences) were based on the accident mitigation domain. Although these systems work sufficiently well, both rule-based expert systems and neural network-based systems have critical limitations. Accident scenario designers must input accident sequences and each symptom directly into the database of rule-based expert systems. However, it is impossible to cover all possible scenarios because all severe accidents have been beyond design-based accidents (BDBAs). In addition, the neural network-based system cannot explain why such a result was produced. In this paper, a group DM support system that can support the DM process of TSC members in unexpected severe accident situations such as the Fukushima accident is proposed. The system was designed by considering new human factor issues revealed by the Fukushima accident. The proposed system has five major characteristics for supporting the DM process of TSC members: support for group DM, multiple-criteria DM, human-oriented DM, symptom-based DM, and real-time DM. With these five design requirements in mind, a group DM support system that can be used in severe accidents was developed. In addition, an experiment with real operators was conducted by simulating a total loss of component cooling water (TLOCCW) accident with four operators of the KHNP and a regulator of the KINS. As a result of the case study, the developed group DM support system is expected to perform reliably in the accident mitigation domain. The accident mitigation strategy was selected by three other KHNP operators who did not participate in this experiment, and the group DM support system recommended the most proper strategy through an algorithm as well. It is significant in that there is no big difference. The developed DM support system uses the natural language of the decision makers as the input and can shorten the group DM duration with a semi-automatic algorithm. In addition, the opinions of all members of the TSC can be collected by excluding unnecessary obstructive human factors. The proposed system can be used in uncertain situations that lack precise procedures and information. This DM support system will promote the group DM process of TSC members in the case of a severe accident and play an important role in reducing the DM duration and preventing other accidents. Finally, the proposed DM support system will improve operator support systems (OSSs) in the accident mitigation domain of nuclear power plants (NPPs).