Advanced Reliability Analysis at Slope Crossings

Pipelines cross diverse terrain and as a result are subjected to a variety of geotechnical hazards. Depending on the location of the pipeline relative to a geotechnical threat, it may be subjected to external forces which could lead to pipeline deformation or failure. Generally, geotechnical threats manifest as slope movement, subsidence/settlement, seismic waves, or frost heave/thaw settlement. While similar analysis techniques may have tangential applicability to all these threats, this paper focuses on the landslide/slope movement scenario. Here, the authors present an approach for evaluating pipelines in areas where slope movement is known or has the potential to occur. The methodology uses advanced finite element analysis (FEA) and statistical reliability techniques to estimate the probability of failure (PoF) of the pipeline at a given site. A case study where the method was employed is also presented. The presented process serves as an advanced analysis tool within a geohazard reliability program. This in-depth PoF analysis can be conducted after a screening level assessment has highlighted a given site. The data required for the analysis includes, at minimum: basic pipe properties, operational information, inertial measurement unit (IMU) in line inspection (ILI) pipeline centerline data, depth of cover survey data, and some estimation of relevant soil to pipe interaction parameters. Other information that can be incorporated to enhance accuracy and reduce conservatism include geotechnical reports and instrumentation measurements (e.g. slope inclinometers or strain gauges). The uncertainties associated with the inputs are estimated based on standards or subject matter expert (SME) input. Incorporating the defined uncertainties, numerical models are created using the commercially available finite element (FE) analysis software ABAQUS, where the pipe is modeled using pipe beam elements and the soil to pipe interactions is modeled using pipe-soil interaction elements. The FE models are processed using a design of experiments (DoE) approach to define response surfaces for both compressive and tensile strain demands. Strain capacities are estimated using the Dorey (U of A) and CRES (PRCI) models for compressive and tensile strains, respectively. Using the resulting relationships for strain demands and capacities, Monte Carlo simulations are completed using the previously defined uncertainties. The simulated cases where strain demand exceeds capacity produce an estimation of probability of exceedance (PoE). Finally, the PoF is obtained by multiplying the PoE by an estimated likelihood of slope movement occurring and impacting the pipe.