Estimating Landslide Induced Probability of Failure to Pipelines Using a Structured Reductionist Approach

Abstract

Much of North America, and indeed much of the global landscape, is comprised of either locally or regionally steep slopes, river valleys, and weak or unstable geology. Landslides and ground movements continue to impact pipelines that traverse these regions. Pipeline integrity management programs (IMP’s) are increasingly expecting quantitative estimates of ground movement or pipe failure as part of pipeline risk management systems. Quantitative analysis usually relies on one or more of statistics, physical models, and expert judgment. Statistics incorporate ground and pipe behavior (for hazard and vulnerability respectively) over a broad area to infer local probabilities. They carry the weight of big data, but the local application is almost certainly incorrect (variability even for regions exceeds 2 orders of magnitude). Detailed geotechnical (hazard) and soil-pipe interaction and stress (vulnerability) models provide rigorous results, but require substantial effort and/or expert judgment to parameterize the inputs and boundary conditions. We present herein a structured tool to calculate probability of failure (PoF) using expert judgment supported by known, instrumented or observable conditions and statistics (where available). We provide a series of tables used as a basis for nodal calculations along a branch path of a decision tree, and discuss the challenges and results from actual application to over 100 sites in the Interior Plains. The method is intended to be a practical informative approach based on, and limited by, data inputs. It is a flexible fit for purpose assessment that takes advantage of the best available data, however, the method relies on the user to articulate a level of confidence in, or the basis of the results.