Conceptual Design of an Accident Prevention System for Light Water Reactors Using Artificial Neural Network and High-Speed Simulator

Abstract

Eight years after the Fukushima accident, the last missing pieces of the jigsaw puzzle for light water reactor (LWR) safety have been put together. In the United States, the nuclear power industry has implemented diverse and flexible strategies to prevent and mitigate severe accidents. In this technical note, the author presents a conceptual design of an online accident prevention system (APS). The proposed concept takes advantage of the fact that the progression of a severe accident caused by an unplanned evolution of the fission dynamics in LWRs, which may be due to mechanical failures, human errors, or external events, progresses significantly slower than events in many other industries, such as chemical explosions or transportation accidents. The APS will make rapid diagnostics of any ongoing event by artificial intelligence and subsequently make immediate predictions using a high-speed simulation code. Should the severe accident lead to core degradation or off-site release, the operators will use all available means including diverse and flexible coping strategies (known as FLEX) to prevent it from happening. Full development and implementation of this APS will greatly enhance nuclear safety in the fight against global warming.

Keywords:

• Diverse and flexible coping strategies
• high-speed simulation code PCTRAN
• sensor fault detection
• artificial neural network
• online accident prevention system