Validation of a Model Formula for Predicting the Long-Term Corrosion Rate of Carbon Steel Pipes in Freshwater

In Japan, conduits buried underground protect telecommunication cables, and it is desirable to continue to use these conduits as much as possible. Hence, it is important to grasp the deterioration in considering the use limit. There are several kinds of conduits, and carbon steel conduits are mainly utilized especially. Though the carbon steel conduits outer surface contacting with the soil is coated with asphalt jute or polyethylene, the inner surface is protected only with synthetic resin paint. Since the conduit inner surface is connected to the manhole, it may be filled with stagnant water such as groundwater or rainwater. Therefore, it is necessary to predict the long-term corrosion rate of carbon steel conduits in fresh water. We carried out a test to measure the average corrosion depth converted from the weight loss by immersing test pieces made from a carbon steel conduit in distilled water at 60 °C and in a manhole stagnant water as a sample. The test pieces were obtained by cutting the steel pipe into 50 mm, removing the coating film on the inner and outer surfaces, shaping it to a thickness of 3 mm, and applying a urethane coating film on the outer surface so that bare steel was exposed only on the inner surface. This test continued for about 1500 days, and test pieces were brought out every 45 days for the specimens immersed in distilled water and every 90 days for the specimens immersed in manhole. The corrosion products were removed using 10% ammonium dihydrogen citrate heated to 80 °C, and weight loss of each test piece was measured. Using this result, the applicability of the model equation proposed by Ozawa et al. in 2017 was considered. Ozawa et al. developed a model to predict the corrosion depth of carbon steel pipes in neutral static water based on temperature, salinity, and Larson Skold Index (LSI). The model assumes that the corrosion depth with time is determined by the initial corrosion rate, time and film resistance coefficient. Salinity and temperature determine the initial corrosion rate, and LSI and temperature determine the film resistance coefficient. In order to apply this model, the pH, chloride ion and sulfate ion of manhole stagnant water were measured by ion chromatography. The pH was 7.8, Cl- was 2.9 mg/l, and SO42- was 5.3 mg/l. The value of hydrogen carbonate ion has not been measured yet, and 207.5 mg/l, which is the average value measured in 1960 in Tokyo 45 manholes, is used. At this time, the film resistance coefficient becomes 2909.3. The temperature and salinity of the manhole reservoir water were 20 ° C and 0.05 ‰, respectively. The model was applied to distilled water with salinity of 0.01 ‰ and LSI with 1. The results of observed corrosion depth is shown on the vertical axis, and the results are shown in the log diagram with the predicted values obtained from the model equation of Ozawa et al. on the horizontal axis. The corrosion depth on day 45 after immersion in distilled water and the corrosion amount on day 90 after immersion in manhole reservoir water agree with the measured value predicted from the model equation, but the observed corrosion depth tends to be larger than the value derived from the model equation after that. The original model equation showed good agreement when compared with the test results at approximately 500 hours. It may be considered that this model equation is applicable in 45 days (1080 hours), but it is highly possible that the corrosion depth becomes larger than the model equation in the long term. In addition, the corrosion rate of the specimen immersed in the manhole deviates more from the model equation. Figure shows a photograph in which corrosion products were observed with an electron microscope on the test specimen as of 1321 days, and it can be seen that cracks of several μm in width were generated just above the base metal and at 150 μm to 300 μm positions, respectively. There is a possibility that the corrosion rate becomes faster than expected due to the influence of the cracks. In addition, though the stagnant water is presumed as still water in the usual time, the flow with the inflow of the water may be generated depending on the season. Considering these factors, it is possible to improve the long-term prediction accuracy of carbon steel conduits. Figure 1