Method for estimating corrosion of underground pipes based on inspection data of telecommunication facilities

Abstract

An extended period has passed since many infrastructures have been constructed, and it is crucial issue for infrastructure to be aged. Corrosion of steel materials has a significant impact on infrastructure, so accurate prediction of corrosion is important for maintaining safe infrastructure. It is known that the corrosion rate varies because of the environment around steel materials. Corrosion rates, especially in the subsurface, vary widely, and studies were conducted to determine corrosion trends in different soils and environments. Hence it is difficult to estimate the corrosion in underground structures in a generic manner due to the analysis cost. In this study, we examine constructing the corrosion prediction method, which that can predict using years buried of facilities and environmental data prepared universally in Japan.

We attempted to construct the method using the data from Japanese telecommunication pipelines. This data set contains 19080 data for internal surface inspection of underground pipeline in six regions of the country and 1704 data for corrosion. First, the data set was combined with topographical classification, subsurface geology, soil classification, river basin, and land use classification. We made cross-tabulations on these five categorical variables and selected an effective variable by calculating Cramerﾒs coefficient of association. Next, we created corrosion prediction formulas for each category by making histograms against years buried.

In this study, the categorical variable that was effective for corrosion was only the river basin. It is inferred that underground pipelinesﾒ inner surface is in contact with water for a long term, which contains underground water. In addition to this, underground water quality is considered to be related to a river. As a result, we could make formulas to estimate corrosion at each condition, dividing the data set between river basins that suffered from severe corrosion and slight corrosion. Assuming the Sigmoid function, there were decades of difference on corrosion ratio between severe corrosion area and slight corrosion.