Skip to main content
Article
A Clustering Based Method to Evaluate Soil Corrosivity for Pipeline External Integrity Management
International Journal of Pressure Vessels and Piping
  • Ayako Yajima, University of Akron Main Campus
  • Hui Wang, University of Akron Main Campus
  • Robert Y. Liang, University of Akron Main Campus
  • Homero Castaneda, University of Akron Main Campus
Document Type
Article
Publication Date
2-1-2015
Disciplines
Abstract

One important category of transportation infrastructure is underground pipelines. Corrosion of these buried pipeline systems may cause pipeline failures with the attendant hazards of property loss and fatalities. Therefore, developing the capability to estimate the soil corrosivity is important for designing and preserving materials and for risk assessment. The deterioration rate of metal is highly influenced by the physicochemical characteristics of a material and the environment of its surroundings. In this study, the field data obtained from the southeast region of Mexico was examined using various data mining techniques to determine the usefulness of these techniques for clustering soil corrosivity level. Specifically, the soil was classified into different corrosivity level clusters by k-means and Gaussian mixture model (GMM). In terms of physical space, GMM shows better separability; therefore, the distributions of the material loss of the buried petroleum pipeline walls were estimated via the empirical density within GMM clusters. The soil corrosivity levels of the clusters were determined based on the medians of metal loss. The proposed clustering method was demonstrated to be capable of classifying the soil into different levels of corrosivity severity.

Citation Information
Ayako Yajima, Hui Wang, Robert Y. Liang and Homero Castaneda. "A Clustering Based Method to Evaluate Soil Corrosivity for Pipeline External Integrity Management" International Journal of Pressure Vessels and Piping Vol. 126-127 (2015) p. 37 - 47
Available at: http://works.bepress.com/robert_liang/6/