Abstract
Since the number of aged water pipes in our country has been gradually increasing, we have to develope evaluation methods for the degree of decrepitude in the pipes. Cast iron and ductile iron pipes comprise 58 percent of the water supply pipes in Japan. Corrosion pit depth was known as a symptom of decrepitude in the pipes. However, the previously proposed method for corrosion pit depth prediction was not able to show probability distribution of corrosion pit depth, and it does not utilize zero corrosion data (ca. 16 % in the whole data). Therefore, we aimed to show statistical distribution of corrosion pit depth by applying a bootstrap method to the 5589 pipe corrosion data that have been collected for ca. 50 years.
Firstly, we classified the soil environments based on the difference in pipe corrosion rates at a significance level of 5 %. As a result, the soil envirouments were grouped into four by the soil types and the soil resistivity. Secondly, the statistical distribution of corrosion pit depth as a function of buried years for each of the four classifications was constructed using a bootstrap method, which is a non-parametric statistical method that can make use of zero-corrosion data. By this way, the probability distributions of corrosion pit depths and the years for the pits to reach the pipe wall thickness were estimated for sand over 45 years, silt 40 years, and clay 30 years. Finally, even in the same soil environments, the corrosibity was found to be influenced by the distance from marine clay by statistical analysis. As it was found that the soil environments in the vicinity of marine clay was more corrosive than those away from marine clay, it is recommended to investigate the presence marine clay within a ca. 500 m periphery.
