Enhanced fragility analysis of buried pipelines through Lasso regression

Abstract

Buried pipelines are one of the critical lifeline structures, and recently, efforts have been directed toward their probabilistic risk assessment. This paper explores the fragility analysis of buried pipelines due to permanent fault displacement. Although several studies have been carried out for the fragility analysis of buried pipelines, they are conditioned only on one significant input parameter. Unlike previous studies, the fragility curves presented in this paper are multi-dimensional, i.e., conditioned on all the significant input parameters. The fragility curves are generated using a machine learning technique called Lasso regression. This paper also explores the relative importance of various uncertain parameters on the fragility estimates. The fragility analysis results suggest that the fault displacement and fault–pipe crossing angle are the most important parameters.