Hydrogen Concentration Prediction and Hydrogen Explosion Threat Determination in Severe Accidents

The hydrogen is very explosive, and when an explosion occurs, it damages multiple defense walls and a large amount of radioactive material is released outside the containment. Therefore, it is important to predict the concentration of hydrogen generated in the melted core in severe accidents. Since it is limited to obtain the accident data of nuclear power plants, Modular Accident Analysis Program4 (MAAP4) code for optimized power reactor 1000 (OPR1000) was used. The initial conditions of MAAP simulations were simulated according to whether or not the Passive Autocatalytic Recombiner (PAR) was operated. In this study, the hydrogen concentration under severe accidents caused by Loss of Coolant Accidents (LOCAs) was predicted by using only limited measurement signals using Deep Neural Network (DNN), one of the artificial intelligence methods. If the reactor operators can use the DNN model to predict quite accurate hydrogen concentration within containment in case of severe accidents, it will help them prevent the hydrogen explosion.