Case Study in the Application of Probabilistic Framework for Pipeline Failure Estimation due to Landslides

Abstract

Landslides pose a significant threat to the loss of containment and failure of buried oil and gas pipelines. As the transmission pipelines traverse diverse terrain, identification of the susceptible regions for landslides and slope movement requires significant effort and the scale of the system often poses a challenging engineering task for site investigations. The threat posed by the large-scale ground movement to the pipeline integrity is often managed through aerial and satellite-based inspection systems. Due to the granularity of these inspection systems, there are significant uncertainties in the sites inferred as susceptible and unsusceptible to the slope movement activity. Furthermore, there are uncertainties in estimating the probability of pipeline failure due to the slope movement activity, as the pipeline response to the ground movement depends not only the pipeline parameters, but also on the site-specific soil parameters.

This case study presents the application of data-driven approaches for the probability estimation of the landslide susceptibility over a large gas transmission pipeline network in North America. The framework used for validating the estimated probabilities with independent data sets and model verification are discussed. Furthermore, methods of uncertainty propagation from the inspection, modelling, and site investigations to the estimation of the probability of the pipeline failure due to ground movement are also presented.