Oil pipeline infrastructure is crucial to energy security and economic development; however, many systems now operate beyond their intended design lifespan. This study addresses three key limitations in current pipeline incident analysis: the isolated treatment of recurring incidents, inadequate consideration of regional variation, and challenges associated with left-truncated failure data. This study proposes an integrated framework that combines spatial analysis to identify incident hotspots with recurrent event survival analysis to model time-dependent failure risks across regions, using incident data from the Pipeline and Hazardous Materials Safety Administration and network attributes from the US Energy Information Administration (1900–2022). Results show that while 45% of initial failures occur within the first decade of operation, the intervals between subsequent failures progressively shorten, with an average reduction of 14% between successive incidents. A Cox proportional hazards model adapted for left-truncated data quantifies the influence of key risk factors, while a geographically weighted Cox model reveals significant regional variation in the effects of variables such as coating type. This segment-level, spatially explicit framework supports more accurate, region-specific risk assessments and informs targeted maintenance strategies to improve pipeline safety and resilience.
Asaye et al. (Sat,) studied this question.