• A Gaussian Random Field was used to model pipeline corrosion and its variability • Spatial and temporal assessment of corrosion was proposed based on ILI data • Reliability was analyzed from a leak to support further pipe intervention decisions • The assessment incorporates uncertainty from ILI consecutive measurements • Critical segments can be spatially identified using the mean time to failure Corrosion in pipelines is inherently uncertain. Given the extensive infrastructures, not only a temporal but also a spatial assessment is required. Estimating and predicting the corrosion process in pipelines is essential for maintaining structural integrity and operational safety. Traditional methods typically focus on temporal assessments to predict corrosion degradation, and few studies examine how corrosion defects may interact spatially. This study presents a probabilistic degradation model using a Gaussian Random Field (GRF) to account for spatial and temporal variability of corrosion, using depth increments obtained from a matching approach. The GRF is simulated using the Cholesky decomposition for the covariance matrix, based on Gaussian and Exponential semivariogram fittings. These simulations were used jointly with a complementary Gamma Process to evaluate a leak limit state based on the defect depth. The proposed approach was applied to a 6 km section of a pipeline, which allows identifying the Mean Time to Failure when an isotropic (37 to 47 years), and an anisotropic (31 to 41 years) behavior is selected. This work seeks to support further decision-making regarding maintenance and monitoring processes to prevent any failure and consequences from a loss of containment.
Meneses-Gelves et al. (Sun,) studied this question.