Reliability assessment of freight wagon passing through railway turnouts using adaptive Kriging surrogate model

authored by: Jun Lai, Kai Wang, Yan Shi, Jingmang Xu, Jiayin Chen, Ping Wang, Michael Beer
Abstract: Railway turnout (RT) is a crucial component of railway infrastructure that consists of several components. Assessing the derailment probability of freight wagons passing through the turnout is crucial for quantifying failure risks and optimizing the performance of the freight wagon-turnout system (FWTS). However, existing assessment methods often require extensive model evaluations and impose substantial computational costs. To address this issue, an efficient reliability analysis method is established for assessing the derailment risk at RTs. Firstly, a dynamic model is developed to capture the wheel-rail dynamic interaction and the numerical model is validated by field tests. Secondly, to reduce the computational cost in the reliability analysis, an efficient adaptive Kriging method based on an error stopping criteria and a learning function is adopted to estimate the failure probabilities under multiple failure modes of wheel derailments. Based on the efficient learning function and convergence criterion, accurate failure probability results can be obtained with a small number of multibody and finite element coupled dynamic simulations. Furthermore, the prediction accuracy of the proposed method in capturing random characteristics for FWTS is evaluated. Finally, the influence of the evolution of rail wear on the failure probability is further discussed.
Organisation(s): Institute for Risk and Reliability
External Organisation(s): Southwest Jiaotong University
University of Liverpool
Tongji University
Type: Article
Journal: International Journal of Rail Transportation
No. of pages: 20
ISSN: 2324-8378
Publication date: 20.01.2024
Publication status: E-pub ahead of print
Peer reviewed: Yes
ASJC Scopus subject areas: Automotive Engineering, Transportation, Mechanics of Materials
Electronic version(s): https://doi.org/10.1080/23248378.2024.2304000 (Access: Closed)