Elevation valleys are suspected to play a role in pipeline corrosion. One proposed mechanism is water accumulation at these valley locations. However, from our literature review, no studies have been published that show the relationship between valley locations and corrosion anomalies using real-world inspection data at a large scale. This topic has been researched rather indirectly, typically by the study of water accumulation in simulated settings, using two immiscible fluids of different densities, frequently using a combination of some sort of oil and water. These studies are not designed to analyze corrosion effects on the pipe, limiting their ability to predict and quantify where corrosion is more likely to occur, which is a critical aspect of pipeline maintenance.

In this paper, we examine corrosion clustering around valley locations using inline inspection (ILI) data originating from operating pipelines, which support the causality path of corrosion assisted by water accumulation at valley locations. We quantify statistically significant excess corrosion count density for each assessment by comparing elevation valleys to non-valley locations and to local elevation maxima (peaks). A permutation method gives the likelihood of randomly observing the valley corrosion density at non-valley locations. We studied 1,083 ILI assessments of 557 pipeline systems, totaling 28,067 assessment miles, and 11,834 miles of independent systems. A strong association between elevation valleys and significant excess corrosion density was demonstrated in several pipeline systems. Peaks did not show a statistically significant correlation. For assessments with over 100 corrosion counts, the corrosion density was found to be up to 18 times higher at valleys than in the rest of the assessed systems, and 44% of the systems had at least one assessment that presented a significant pattern of corrosion clustering around valleys.

The stability of the generated signal was studied over time for systems with multiple assessments, confirming the consistency of the results. The consistency of this metric applied to a large dataset justifies its use as an indicator of anomalous internal corrosion in a system, potentially helping pipeline operators to identify systems that are or may soon be affected by excess internal corrosion.