인공지능 학습을 통한 원자력발전소 정·주기 시험 자동화에 대한 연구

Abstract

Nuclear Power Plants (NPPs) are facilities that require the highest reliability. NPPs are performing a lot of tests during normal operation and Overhaul (O/H). Operators, and maintenance workers of NPPs perform these tests to prove the safety of the NPPs and operate the plant safely. These tests are mainly Surveillance Test (ST), which is based on the Final Safety Analysis Report (FSAR), and the Periodical Test (PT), which is based on the Technical Specification (Tech-Spec). Also, since the Fukushima Dai-ichi accident, there has been an emphasis on the safety of nuclear power plants globally. In addition, small modular reactors (SMR), which have strength in efficiency compared to the normal NPPs, real-time monitoring, and safe operation through Artificial Intelligence (A.I.) learning, have recently been in the spotlight. Furthermore, the International Atomic Energy Agency (IAEA) reported that human error was the main factor of accidents at nuclear power plants, and the Korea Atomic Energy Research Institute (KAERI) also reported that about 24% of accidents at nuclear power plants were caused by human error. To increase safety by reducing human resources and human errors, it is necessary to study the potential of automating the tests carried out in NPPs. To safely operate NPPs, the operator checks the requirements of the FSAR and Tech-Spec. Tech-Spec of NPPs defines Limiting Conditions for Operation (LCO) and Surveillance Requirements (SR) to secure the safety of NPPs during Design Basis Accident (DBA) and satisfy the regulatory requirements of the FSAR. In NPPs, there is an important safety concept is called the Engineered Safety Features (ESFs), which is to limit plant/equipment damage and to mitigate the consequences of the accident. Out of these safety systems, I will focus on the operability test of one of the ESFs called Auxiliary Feedwater System (AFWS). AFWS is a safety system that supplies water to the Steam Generator (SG) until the Shutdown Cooling System (SCS) is connected when the Station Black-Out (SBO) occurs or Main FeedWater System (MFWS) cannot supply feedwater due to any event or accident. In this paper, I classify and analyze the task of the Auxiliary Feedwater Pump test among the numerous tests performed in NPPs, as the initial process to automate the test. In addition, using MARS code and CNS simulator, I will make simple automatic procedure process and verify it with several scenarios.