Railroad Track Geometric Degradation Analysis: A BNSF Case Study

Abstract

Planning and optimizing maintenance actions for rail failures are possible through understanding rail behaviors, which can be captured by statistical models. In this study, a degradation analysis of railroad tracks is presented for three types of geometry failure modes. Using Burlington Northern Santa Fe (BNSF) data set, effective degradation stress factors are identified and inspection intervals are studied to attenuate the effect of hidden maintenance actions. Parameters of a general log linear Weibull model have been determined to estimate the failure probability at different stresses. To check the prediction performance, the models were applied on an independent test set, and the corresponding area under curve (AUC) were reported. Finally, a comprehensive sensitivity analysis was implemented to check the effect of different stress factors on residual life of a defected rail. This type of analysis is crucial in alerting railroad companies to fix a defect and to schedule preventive maintenance actions to reduce the probability of a track failure.