LOCA 사고시 비정상 안전주입계통 회복을 위한 AI에 기초한 Golden Time 예측

Abstract

After the Fukushima nuclear accident the importance of accident management for nuclear power plants (NPPs) has being increased. For that reason, many countries have focused on the way of improving the NPP safety. If loss-of-coolant accident (LOCA) that is a typical case of design basis accidents (DBAs) happens, emergency core cooling system (ECCS) and depressurization system operating in emergency must work normally. Failure of the safety related systems may leads directly to loss of the reactor core coolability. In this case, DBA may be converted into serious accidents, such as the core uncovery and the reactor vessel (RV) failure. Therefore, predicting the recovery time of the safety injection system (SIS) for recovering reactor core coolability is very important to take initial actions promptly [2]. In this study, we have analyzed the golden time for recovering safety injection system (SIS) when the nuclear reactor lost the coolability by LOCA. Prediction of the golden time for recovering the reactor core coolability was performed by simulating many cases of LOCA accident condition using modular accident analysis program (MAAP) code [3]. Optimized power reactor (OPR1000) was used for target model. Also, the group method of data handling (GMDH) model and support vector regression (SVR) model were applied to predict the golden time for recovering the reactor core coolability [4]. As a result of this study, the proposed GMDH model and SVR model were able to accurately predict the golden time. Therefore, it can be used effectively in the accident situation of the NPP, and it is considered that it is possible to take accident actions quickly and accurately.