Spatial Statistical Analysis: A “Blind-Approach”

Chapter 5 discussed sources of uncertainty identified during metal loss detection and degradation. These uncertainties affect the extent and location of the corrosion defects reported by the ILI measurements, posing additional challenges for pipeline monitoring. In this chapter, some relevant questions are raised in terms of corrosion spatial variability, such as the following: How the soil influenced the corrosion spatial distribution? Is it possible to describe the spatial variability of defects’ depth? How can an onshore pipeline be divided depending on its actual condition? Is there a spatial correlation among corrosion defects, or are they randomly located? For this purpose, a statistical analysis of the corrosion spatial variability is proposed considering three different scales. The first scale implements the records from the entire pipeline, with a basic soil classification to study the effect of the surrounding soil and identify the main features of deep defects. The objective is to characterize the soil in the surroundings and to provide tools for predicting pipe corrosion based on the reported measurements. The second scale divides the pipeline to identify segments with critical conditions in which further analysis should be implemented. Finally, the third scale evaluates information about the corrosion spatial autocorrelation. The spatial dependencies of the corrosion measurements are characterized based on statistical indicators such as Moran’s I. The objective is to give some insights into how defects are affected by their neighbors.