Analysis of Dynamic System Risks Where Pipelines Cross Slow-Moving Landslides

Abstract

Pipelines are often constructed across dormant or normally slow-moving landslides. The potential for pipeline failure in response to landslide movement depends on several factors. These include the likelihoods of different landslide velocities being realized, the probabilities that movements physically impact the pipeline, the capacity of the pipeline to accommodate landslide displacements, and the ability of operators to detect and respond to deteriorating conditions. Each of these factors is difficult to predict but estimating the probabilities of landslide velocity transitions is particularly challenging. In this paper we review a conceptual approach to predict landslide velocity and displacement using Markov chains that combine geomorphic evidence of long-term landslide behaviour with current estimates or measurements of landslide velocity. A framework is proposed to combine time-dependent estimates of landslide displacement and the deterioration of pipeline strain capacity to estimate the probability of pipeline failure over time and in response to potential changes in landslide velocity. The expected efficacy of monitoring programs and trigger action response plans is accounted for in the vulnerability model.