| Summary: | Piping in nuclear power plants is subject to corrosion and erosion caused by the interaction with fluids it carries. In order to prevent accidents such as pipe rupture, it is necessary to non-destructively inspect the thickness of the pipe wall and predict the remaining pipe life. In contrast, signals obtained by non-destructive inspection are affected by various uncontrollable and unknown factors, which will result in uncertainty in evaluating pipe wall thickness. Ignoring the uncertainty would lead to a large error in the pipe reliability assessment. Therefore, it is important to develop a reasonable pipe wall thinning management that can takes the uncertainty of non-destructive testing into consideration. Based on this background, this study aimed to develop a piping wall thinning prediction model that accounts for the uncertainty in evaluating pipe wall thickness and to evaluate its applicability to ultrasonic testing. At first, ultrasonic tests were performed to measure the thickness of specimens simulating flow accelerated corrosion. The results of the measurements were statistically analysed to obtain a mathematical model correlating the evaluated and true thickness. A numerical model to predict the reduction of wall thinning with its uncertainty is developed. Compared to the conventional method, the proposed method is able to predict wall thickness with high accuracy.
|
|---|