A hazard-based predictive approach for onshore gas transmission pipelines using historical failures

Gas transmission pipelines are typically large-scale cross-country engineering structures that can be divided in similar segments based on their attributes and working conditions. The implementation of probabilistic reliability models and subsequent mitigation strategies can be considerably assisted by historical failure data available in different databases from around the world. Statistical analyses and derivations of failure rates in literature are usually performed in terms of the rate of occurrence of failures (ROCOF). This study proposes a robust combination of a known parametric hybrid empirical hazard model and a data processing technique, namely, the nonlinear quantile regression, for reliability analysis and prediction. As for most typical large-scale infrastructures, the hazard function of gas transmission pipelines is expected to have the form of a ‘bath-tub’ curve, as illustrated in Figure 1. The presented method is capable of efficiently addressing such mixed failure distributions, typically a combination of exponential and Weibull distributions. It can also deal with scarce and inconsistent failure datasets, either left-censored or right censored as they are usually met in practice. Implicit in the proposed method is the selection of an adequate segmentation length so that the pipelines can be divided into multiple segments that operate under similar conditions. The method then provides inferences on the complete lifecycle reliability of a reference pipe segment of the region under study. The non-linear quantile regression methodology ensures that parameters of the probabilistic model are estimated in a statistically valid, repeatable and non-subjective way for a given quantile or equivalently, a given probability of exceedance. For the illustration of the usefulness of these models in an industrial context, onshore gas transmission pipelines data for the period from 2002–2014 were obtained from the United States Department of Transportation Pipeline and Hazardous (Image Presented) Materials Safety Administration (PHMSA) database and analyzed. Results obtained are to verify the applicability of the proposed methodology when analyzing rupture incidents of gas transmission pipelines and also to improve the knowledge regarding the state of onshore gas transmission pipelines reported in PHMSA from 2002 to 2014. The practical implications to pipeline operators are mainly the validation of models that predict the structural reliability of pipes based on their failure mechanisms and the implementation of informed optimal maintenance strategies based on relative risk prioritization.