Available recovery time prediction in case of an accident scenario for NPP component

In nuclear power plant operations, there is a possibility that a stimulating event can initiate undesired accident. Therefore, prediction of the available recovery time (the time within which restorative measures can be taken without compromising the threshold safety limits) is a serious challenge to avoid any accident scenario. Recent advances in sensor technology has made it possible to continuously monitor the plant component parameters. Fuzzy logic based artificial intelligence data-driven systems compare real time operational parameters with the pre-stored reference failure database to provide an effective estimate of the recovery time available. This paper demonstrates the prediction of available recovery time in case of an accident scenario in U-tubes of Nuclear Power Plant (NPP) heat exchanger. When a failure scenario evolves, its evolution pattern is compared with the reference failure database using fuzzy similarity analysis. A reference failure database consisting of actual accidental history is not feasible. In this research, an effort has been made to generate reference failure database employing Computational Fluid Dynamics (CFD) tool of commercial code ANSYS 16.2. Reference failure database consists of data collected from multiple U-tube temperature based failure scenarios. The validity of this procedure is checked by estimating recovery times for several test patterns. Moreover, the actual and predicted recovery times have been compared for the test patterns. A framework has been presented, in which temperature threshold is detected and a comparison is made between the evolving patterns and the reference database. This study gives a roadmap for the implementation of fuzzy logic prediction to enhance the safety of Nuclear Power Plant (NPP) components.