Prediction of flow-accelerated corrosion in single/dual elbow in pipelines considering geometric parameters and layout effect

Flow-accelerated corrosion (FAC) is the major pipe wall thinning phenomenon to be managed for its potential of catastrophic pipe rupture. Present management rule for pipe wall thinning in Japanese nuclear power plants is based on pipe wall thickness measurements resulting in large number of measurements in each plant. For future improvement of the plant management in Japan, introduction of prediction method or prediction code for thinning rate and residual lifetime evaluation is expected. To predict the maximum wall thinning rate or minimum wall thickness efficiently for each pipe component, one-dimensionally along the pipelines, relative wall thinning trend of each component type, "geometry factor", combined with the effect of upstream component, "declining effect", are essential, which are both derived from hydraulic features. In this study, the geometry factor and the declining effect for pipeline elbows are evaluated considering specific geometric parameters through series of computational fluid dynamics calculations. For the geometry factor, correlation equation is arranged as a function of curvature radius and bending angle of elbows, showing evident effect of each geometric parameter. And for the declining effect, firstly, the effect of plane angle formed by the upstream and downstream elbows is evaluated that the in-plane configuration with the flow streaming forward would be the most conservative condition. Then, based on the in-plane configuration, correlation equation for the declining effect is arranged as a function of the distance between the two elbows. It is found that even when the distance is very short, the remaining portion of the upstream effect on the downstream elbow would be much less than previous knowledge which may improve the overprediction in proximate condition. By implementing these correlations into author's FAC prediction model, residual wall thickness of plant pipe elbows is shown to be predicted mostly within 20 % from the measured data.