BaNTERA: A Bayesian Network for Third-Party Excavation Risk Assessment

Abstract

Third-party damage constitutes a major threat to underground natural gas pipeline safety; in the U.S., between 2016 and 2020, it caused eleven fatalities, twenty-nine injuries, and $124M USD in property damage losses. Several research studies have been carried out to identify the causes and contextual factors leading to third-party damage. However, there is a lack of models that are not only causally-based, but also comprehensive and suitable for modeling the probabilities of a pipe hit and subsequent damage. This paper presents the development process and results of building BaNTERA, a probabilistic Bayesian network model for third-party excavation risk assessment in the U.S. BaNTERA’s capabilities for risk-informed decision support are presented in three ways: verification of the model’s performance, validation of its damage rate predictions with historical industry data, and application in multiple case study scenarios. Preliminary results indicate that BaNTERA offers valuable insight including and beyond a probability estimation of third-party damage. Using the best available industry data and previous models derived from multiple sources, different inference methods can assist in pipeline damage prevention and risk mitigation. As such, BaNTERA represents a promising holistic and rigorous tool for addressing third-party excavation damage in natural gas pipelines.