Open Collections

Dey, Sandip

University of British Columbia

This thesis entails developing analysis methods for evaluating seismic risk to oil and gas pipelines due to permanent ground deformation resulting from fault rupture. To address this, three modules are identified as core part of this research. Firstly, a finite element (FE) model for oil and gas pipeline crossing a seismic induced fault line is developed. Subsequently, uncertainty in input parameters is introduced in the developed FE model and analyzed in a multi-fidelity approach. Finally, probabilistic regional ground deformation due to fault rupture is assessed and integrated with the pipe-soil structural model. A detailed FE analysis technique to study the behavior of buried continuous pipelines crossing fault movements is developed and established with appropriate evaluations. A non-linear sand constitutive model is adopted and implemented in commercial FE package ABAQUS. The adopted material model is first evaluated using available tri-axial test results. This material model is thereafter suitably calibrated for a large-scale test based on direct shear soil test data for that experiment. The pipe-soil FE model is then evaluated against full-scale experimental results. Subsequently, structural response of a buried continuous steel pipeline undergoing fault rupture deformation is studied in a systematic manner. A detailed and efficient analysis framework for design of buried continuous steel pipelines crossing faults is proposed and explained with a case study using Taguchi method for design of experiments. Further to this, uncertainty in input parameters, such as pipe diameter, pipe wall thickness, young’s modulus, yield strength, ultimate strength, sand confining pressure, sand relative density, operating temperature and operating pressure is considered using a multi-fidelity approach which relies on analyses results from a small number of detailed structural analyses and large numbers of simplified structural analyses. The approach efficiently relies on both types of results to predict structural performance accurately considering uncertainty and variability in geometrical and material properties. Significant computational cost savings is achieved where by 500 analyses are completed using 15 minutes computational time low-fidelity (LF) models instead of 48 hours computational time high-fidelity (HF) models with a 3.20GHz eight-core processor with 32 GB RAM for 500 analyses. Finally, extension of the structural analysis framework to a regional scale is completed. This is done by calculating fault rupture induced regional ground displacements from available analytical techniques using probabilistic approaches and thereafter integrating it with the developed structural analysis uncertainty quantification framework.

Text

2021-11-26

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/80273

Civil Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/