FSN-based fault modeling in CANDU® stations

• Diagnosis of Neutron Overpower Protection (NOP).
• Detector performance analysis.
• Diagnosis of aging and degradation of NOP systems.
• Fault modeling using Fault Semantic Network (FSN).
• Fault forecasting using intelligent computational methods.

Over the past several years a number of domestic CANDU® stations have experienced issues with neutron detection systems that challenged safety and operation. Intelligent troubleshooting methodology is required to aid in making risk-informed decisions related to design and operational activities, which can aid current stations and be used for the future generation of CANDU® designs. This paper presents fault modeling approach using Fault Semantic Network (FSN) with risk estimation. A case study of challenges with liquid zone level control at one of the domestic CANDU® stations is used to demonstrate this approach. This work will further continue into the development of fault forecasting using intelligent computational methods, such as Genetic Algorithms (GA), Fuzzy Logic, and Artificial Neural Network.