Measurements of local mass transfer coefficient of Flow Accelerated Corrosion at feeder outlet pipes in CANDU using Plaster of Paris method

Full text of publication follows: Flow Accelerated Corrosion (FAC) is a corrosion mechanism that results in wall thinning in piping systems and components. FAC on carbon steels in pure water has occurred in a number of industry and power plant. A pipe wall thinning due to the FAC at nuclear power plant is just reported in confined to carbon steel pipe on the secondary system which does not equip a radioactive component. Recently, at Canadian Deuterium Uranium type reactor, it is reported that the feeder lines suffer the wall thinning on the primary system to equip 380 feeder pipes. Differently from the fast FAC rate of the secondary system on PWR, FAC rate of feeder pipes on CANDU is not more than the 0.2 mm/year. Although the FAC rate of feeder pipe is relatively slow, the narrow thickness margin of 2.6089 mm is endangered sufficiently by only the slow FAC rate. The FAC is governed by the mass transfer coefficient which is determined by the flow field. To well predict the FAC rate, mass transfer coefficient should be well measured or calculated. New measurement method of mass transfer coefficient is developed to obtain the coefficient for the complex shape like feeder pipe. The method evaluated the naphthalene sublimation method to apply it under the water flow. The naphthalene sublimation method can be used to study mass and heat transfer with confidence for a variety of applications, but with certain restrictions. At present, Plaster of paris method can modified the sublimation method to overcome the restrictions. This method is particularly useful in complex flows on geometries and for flows with large gradients in wall transport rate. The test specimen can be easily prepared by several methods, including dipping, machining, spraying, and casting. The local transfer coefficients can be determined with high accuracy and in detail by automated measurement systems that eliminate most human errors during the measurement process. To evaluate the method, the experiments are performed to compare with other correlations for straight pipe and feeder pipes. The comparisons are good agreement with error of 10 % for the straight pipe and are agreement with the CFX 4 calculation results for the feeder pipe. (authors)