报告详情
EVOLUTION AND STATUS OF SIMULATION CODES USED IN CRUD DEPOSITION MODELING IN PRESSURIZED WATER REACTORS
编号：102
访问权限：公开
更新：2024-09-23 23:13:10
浏览：127次
口头报告
摘要
Nuclear Power Plants contribute about a fourth of global low-carbon electricity production, mainly through Pressurized Water Reactors (PWRs), which comprise about 70% of all active reactors worldwide. Chalk River Unidentified Deposit (CRUD) poses significant challenges for PWRs. CRUD, which is formed from the buildup of fouling deposits in the Reactor Coolant System (RCS), might lead to increased hydraulic resistance, increased occupational radiation exposure, CRUD Induced Power Shift (CIPS), and CRUD Induced Localized Corrosion (CILC). These CRUD deposits on the reactor core ultimately impact the operation and cause significant economic losses. Given the challenges associated with observing CRUD deposition in real-time, the modeling of CRUD growth and effects has gained significant interest over the past decades, as various simulation tools have been developed to predict CRUD formation, transport, and deposition in PWRs. This study examines some important simulation codes used in CRUD deposition modeling in PWRs, categorizing them into three generations: legacy, transitional, and state-of-the-art codes. Legacy Codes, such as VIPRE-1, laid the foundation with basic thermal-hydraulic models. Transitional Codes, such as BOA, introduced more sophisticated models and improved computational efficiency; meanwhile, State-of-the-Art codes, such as the VERA suite of codes, provide high-fidelity multi-physics simulations. However, the high-fidelity multi-physics simulations offered by these State-of-the-Art codes often require extensive computational resources and run time.
Additionally, this study discusses the role of machine learning in CRUD modeling, especially the use of convolutional neural networks to build surrogate models. These machine learning models emulate these high-fidelity simulations with acceptable accuracy while operating much faster and requiring fewer computational resources. However, limited research has applied machine learning techniques for CRUD modeling. This leaves a considerable void in the existing literature.
Additionally, this study discusses the role of machine learning in CRUD modeling, especially the use of convolutional neural networks to build surrogate models. These machine learning models emulate these high-fidelity simulations with acceptable accuracy while operating much faster and requiring fewer computational resources. However, limited research has applied machine learning techniques for CRUD modeling. This leaves a considerable void in the existing literature.
关键词
CRUD, Deposition, Simulation Codes, Machine Learning, Modeling
报告人
CYRIL NDIP NDE FRU
Harbin Engineering University Lei Jin
Student Harbin Engineering University全部评论
重要日期
-
会议日期
09月23日
2024
至09月25日
2024
-
09月24日 2024
报告提交截止日期
-
09月25日 2024
注册截止日期
主办单位
Harbin Engineering University (HEU)
联系方式
- Prof. WANG Xiang
- xi******@hrbeu.edu.cn
历届会议
-
2016年09月26日 中国 Chengdu,China
8th International Symposium on Symbiotic Nuclear Power Systems for 21st Century -
2016年09月26日 中国 Chengdu,China
第八届21世纪和谐核电系统国际会议 -
2010年08月23日 中国 哈尔滨市
第三届21世纪和谐核电系统国际会议（ISSNP2010）
发表评论