Rail accident analysis using large-scale investigations of train derailments on switches and crossings: Comparing the performances of a novel stochastic mathematical prediction and various assumptions. (September 2019)

• Record Type: Journal Article
• Authors: Dindar, Serdar; Kaewunruen, Sakdirat; An, Min
• Is Part Of: Engineering failure analysis. Volume 103(2019)
• Page Start: 203
• Page End: 216
• Publication Date: 2019-09
• Subjects: Derailment -- Turnout component failures -- Hierarchical Bayesian analysis -- Freight transportation -- Spatial analysis; System failures (Engineering) -- Periodicals; Fracture mechanics -- Periodicals; Reliability (Engineering) -- Periodicals; Pannes -- Périodiques; Rupture, Mécanique de la -- Périodiques; Fiabilité -- Périodiques; Fracture mechanics; Reliability (Engineering); System failures (Engineering); Periodicals; Electronic journals; 620.112
• DOI: 10.1016/j.engfailanal.2019.04.010
• Languages: English
• ISSNs: 1350-6307

Abstract

Abstract: Each day tens of turnout-related derailment occur across the world. Not only is the prediction of them quite complex and difficult, but this also requires a comprehensive range of applications, and managing a well-designed geographic information system. With the advent of Geographic Information Systems (GIS), and computers-aided solutions, the last two decades have witnessed considerable advances in the field of derailment prediction. Mathematical models with many assumptions and simulations based on fixed algorithms were also introduced to estimate derailment rates. While the former requires a costly investment of time and energy to try and find the most fitting mathematical solution, the latter is sometimes a high hurdle for analysists since the availability and accessibility of geospatial data are limited, in general. As train safety and risk analysis rely on accurate assessment of derailment likelihood, a guide for transportation research is needed to show how each technique can approximate the number of observed derailments. In this study, a new stochastic mathematical prediction model has been established on the basis of a hierarchical Bayesian model (HBM), which can better address unique exposure indicators in segmented large-scale regions. Integration of multiple specialized packages, namely, MATLAB for image processing, R for statistical analysis, and ArcGIS for displaying and manipulating geospatial data, are adopted to unleash complex solutions that will practically benefit the rail industry and transportation researchers. Highlights: All estimates seem to be incapable of calculating an estimate for a low number of derailments. the assumptions seldomly yield a precise estimate of the derailment rates under any uncertainty Some assumptions which relied on turnout counts, are observed to deviate from the observations the assumptions regarding turnout counts are a weak spot even when being generated mathematically on the basis of a concrete belief.