Temperature–based quantitative detection of micro–internal leakage in heat–cycle isolation valves of nuclear power plants

The internal leakage of heat-cycle isolation valves in nuclear power plants impairs plant economics and may trigger environmental contamination or safety issues due to medium release. Non-destructive, on-line and quantitative detection methods are therefore urgently required. Among existing techniques, the temperature method is the most promising for quantitative on-line leakage detection of high-temperature valves; however, the coupled multi-physics nature of micro-leakage and their invisibility make both identification and quantification challenging. Taking actual plant pressure and temperature as boundary conditions, artificial valve defects were designed and a fluid-thermal-solid coupled CFD model was built. Temperature-characteristic parameters extracted from the simulated outer-wall temperature field were correlated with leakage rate by combining the least-squares method with multiple linear regression. Experimental validation showed predicted leakage rates agree with measured values within a mean relative error of 6% and the minimum detectable leakage is 2 L/h, outperforming traditional models. The temperature-based predictor therefore offers practical value for enhanced online monitoring and predictive maintenance in nuclear power plants.