Safety Performance Improvement for Nuclear Power Plants Using THOMAS and Accident Prediction Function

The environments of nuclear industry are changed by incoming of digital technology. Until now, the nuclear power plant was adhering to analogue system in the large part of system. However the reliability of digital technology is increased, the adopting of digital technology is accelerated in the nuclear industry. It is not exception of the part of monitoring system. Digital based thermal hydraulics online monitoring advisory system of nuclear power plant, as called THOMAS, which is can be developed economically compared with existing monitoring system is used for the decision making tool in the accident condition. We selected the Ulchin 3 and 4 units which is the type of Korea Standard Nuclear Power Plant(KSNP) as reference plant. For nuclear power plants, EOPs (Emergency Operating Procedures) help operators to diagnose and analyze accidents. But it is very difficult that operators diagnose and analyze similar accidents with EOPs in a given short time. There are also possibilities to follow wrong procedures due to complex and extensive procedures. Therefore, it is important to develop a methodology for diagnosing accidents in a short time and reduction of human errors that made by complex signals and indicators. THOMAS has a function of decision making using influence diagram logic. The influence diagram logic is based on total probability and Bayesian theory. And also the accident modeling is based on emergency operating procedure(EOP). The final goal of this system is, in the accident situation, to present the success path to the operator for the recovery of system. In this paper, at first, we will deal briefly with total system of THOMAS. And then 3D visualized window and accident prediction function will be introduced in detail