Optimal Pipeline Maintenance Strategies in the United States: Stochastic Reliability Analysis of Gas Pipeline Network Failures

Abstract

The increasing number of corrosion-based incidents in pipeline systems within recent decades has encouraged decision makers to focus on predicting when pipelines will fail, and to improve maintenance strategies. Policymakers and pipeline operators need to frequently evaluate and renew existing failure prediction models and maintenance strategies. Accordingly, this research accomplished the following two objectives: (1) it developed a predictive failure model for internal and external pipeline corrosions, and (2) it developed a model to determine the optimal timeline for maintenance policy, when the corresponding costs can be minimized. For this analysis, data were collected from 11 natural gas transmission pipeline failures that occurred between 2001 and 2011 in the United States, and were categorized into two groups based on their internal and external corrosion levels. Then the homogeneous Poisson process (HPP) and the nonhomogeneous Poisson process (NHPP) were implemented to evaluate the reliability trend of pipelines. The results revealed that the trend of pipeline reliability was deteriorating for both external and internal corrosions, except for a few cases that showed improvement and stationary trends. Additionally, the decade of installation and types of corrosion were recognized as significant causes of pipeline failure. To find the optimal replacement time, a cost-effective model was developed to determine the latest time that maintenance can be scheduled before the cost of the maintenance increases. The results of this study will help decision makers in the pipeline industry accurately predict the number of failures in pipeline networks, and wisely select the most appropriate maintenance policies.