A series of full-scale tests were undertaken to examine the effectiveness of the use of geosynthetic materials to reduce lateral soil loads on buried pipelines subjected to transverse ground movements. The testing program consisted of measuring lateral soil loads on steel pipes buried in trenches simulating different native soil and backfill material configurations. The effectiveness of lining the inclined surface of the trench (i.e. “trench slope”) with two layers of geotextile as a method of soil load reduction depends on the formation of good slippage at the geotextile interface. Pipes buried in relatively soft native soil can penetrate into the native soil during lateral displacement, thus causing the geotextile-lining to be ineffective as a reducer of lateral soil loads. Although there is more opportunity for slippage at the geotextile interface when the trench is in relatively stiff soil, the soil loads on the pipe seem to still increase when the pipe moves in close proximity to the trench slope; this effect is likely due to the increased normal pressures on the pipe arising as a result of the presence of the stiff trench in the vicinity of the pipe.

A research program has been undertaken to study the behaviour of buried steel pipelines subject to lateral horizontal ground movements, and to provide appropriate data to calibrate and validate numerical model(s). A large sand chamber (2.5 m W × 3.8 m L × 2.5 m H) available at the University of British Columbia was employed to conduct full-scale lateral pullout tests on steel pipelines, with different diameters and buried in sand simulating different overburden ratios. Numerical analyses were performed using finite-difference-method-based software with the soil response simulated using Mohr-Coulomb and hyperbolic elastic constitutive models. The input parameters for the initial computer modeling were based only on element testing results. The numerical predictions, using the two soil constitutive models, are compared with the results of lateral pullout tests. The numerical model, after validation with full-scale test results can be used to predict soil loads on pipe for different overburden ratios, pipe sizes and soil properties.

Coseismic fault displacement has been recognized as a critical hazard to natural gas transmission pipelines crossing earthquake faults. Prior studies on pipeline response to fault displacement were limited to specific types of pipes and faults, which were indeed constrained by the computational resources. As part of an effort to develop a Bayesian model to relate ground displacement in pipeline strain, we analyze more than 217,000 finite element models of gas pipeline fault crossings, which consider the pipe–soil interactions with both pipe and soil material nonlinearities. Such an enormous number of simulations are selected based on comprehensive sensitivity analyses and cover the most important parameters of gas pipelines in terms of combinations of the following: (1) pipe dimensions and materials, (2) soil properties, and (3) style of faulting and characteristics of fault movements. We devise an automated workflow for input generation–simulation submission–output extraction, by utilizing more than 10,000 cores high-performance super-computing facilities. Finally, we examine the pipeline response for every combination, by investigating the evolution of maximum compressive and tensile strains in the axial direction along the pipe over the fault displacement. These numerical analyses resulted in a comprehensive database of pipeline fragilities crossing earthquake fault for seismic risk analysis of natural gas infrastructure in an earthquake region.

Herein, we utilized machine‐learning (ML) and data‐driven (regression) techniques to tackle a critical infrastructure engineering problem—namely, predicting the seismic response of natural gas pipelines crossing earthquake faults. Such a 3D nonlinear problem can take up to 10 h to solve by performing finite element analysis (FEA), considering the length of the pipeline and a large number of pipe and soil elements. However, the ML and data‐driven techniques can learn the projection rule of input‐output and predict the pipeline response instantaneously given a set of input features. In addition, the well‐trained ML model can be implemented for regional‐scale risk and rapid post‐event damage assessments. In this study, the input for ML comprised approximately 217K nonlinear FEAs, which covered a wide range of combinations of soil, structural and fault properties and yielded critical pipe strain responses under fault‐rupture displacements. We adopted various regression models and physics‐constrained neural networks, which can accurately and rapidly predict the tensile and compressive strains for a broad range of probable fault‐rupture displacements. Performances of various ML and conventional statistical models were systematically examined. Not surprisingly, neural networks exhibited the best performance for this multi‐output regression problem, in which R2 > 0.95 was achieved for a wide range of fault displacement (FD) levels. Further, we used the trained neural network with 14.5 million Monte‐Carlo‐generated input samples to predict the maximum tensile and compressive strain curves of pipelines. This new dataset aimed at filling the missing input‐output points from the 217K FEAs, and improved the accuracy of the prediction of probability of failure for natural gas pipelines under FD hazards.

No abstract