Analysis of railroad tank car releases using a generalized binomial model. (November 2015)

• Record Type: Journal Article
• Authors: Liu, Xiang; Hong, Yili
• Is Part Of: Accident analysis and prevention. Volume 84(2015)
• Page Start: 20
• Page End: 26
• Publication Date: 2015-11
• Subjects: Petroleum crude oil -- Tank car release -- Train accident -- Generalized binomial model -- Probability; Accidents -- Prevention -- Periodicals; Accident Prevention -- Periodicals; Accidents -- Prévention -- Périodiques; 363.106
• DOI: 10.1016/j.aap.2015.08.004
• Languages: English
• ISSNs: 0001-4575

Abstract

Highlights: Develop a generalized binomial model to analyze interdependent tank car releases within the same train accident. Use the methodology to estimate the probability of a large, multiple-car release incident. Provide new insights into railroad hazardous materials safety and risk analysis. Abstract: The United States is experiencing an unprecedented boom in shale oil production, leading to a dramatic growth in petroleum crude oil traffic by rail. In 2014, U.S. railroads carried over 500, 000 tank carloads of petroleum crude oil, up from 9500 in 2008 (a 5300% increase). In light of continual growth in crude oil by rail, there is an urgent national need to manage this emerging risk. This need has been underscored in the wake of several recent crude oil release incidents. In contrast to highway transport, which usually involves a tank trailer, a crude oil train can carry a large number of tank cars, having the potential for a large, multiple-tank-car release incident. Previous studies exclusively assumed that railroad tank car releases in the same train accident are mutually independent, thereby estimating the number of tank cars releasing given the total number of tank cars derailed based on a binomial model. This paper specifically accounts for dependent tank car releases within a train accident. We estimate the number of tank cars releasing given the number of tank cars derailed based on a generalized binomial model. The generalized binomial model provides a significantly better description for the empirical tank car accident data through our numerical case study. This research aims to provide a new methodology and new insights regarding the further development of risk management strategies for improving railroad crude oil transportation safety.