Statistical Modeling of Pitting Corrosion in Buried Pipelines Taking Into Account Soil Properties

Small leaks caused by external pitting corrosion are the leading cause of failure in oil and gas pipelines in many regions of Mexico. Because of this, the need for realistic and reliable pitting corrosion growth models that are capable of accounting for the chemical and physical properties of soils and pipeline coatings is especially great. In this work, maximum pit depths and soil and coating data that were gathered at excavation sites across southern Mexico are used to investigate the impact of soil and pipe characteristics on pitting corrosion in buried pipelines. Soil field-measurements included resistivity, pH, pipe-to-soil potential, humidity, chloride, bicarbonate and sulphate levels, redox potential, soil texture and coating type. Together with the local physical chemistry of the soil and the coating characteristics, the maximum pit depth and pipeline’s age were recorded at more than 250 dig sites. The time dependence of the maximum pit depth was modeled as y_max = β(t−t₀)^α, with β and α being positive constants, t being the pipe’s age and t₀ the pit initiation time. A multivariate regression analysis was conducted with y_max as the dependent variable, while the pipeline’ age and the soil and pipe properties were used as the independent variables. The optimal dependence of β and α on these variables was found and predictive models were proposed to describe the time dependence of the average maximum pit depth and growth rate on soil and pipe properties. Besides the creation of a generic model fitted to all the gathered data, a model was proposed for each one of the three soil types identified in this study: clay, clay-loam and sandy-clay-loam. It is shown that the application of the proposed model allows for prediction of corrosion pit growth more accurately than previous models and that this improvement positively impacts on integrity management plans that address the threat posed by external pitting corrosion.