Integrated Approach for Simulation and Prediction of Railway Track Dynamic Responses
Advisor: Professor Erol Tutumluer
Abstract
This PhD study focused on the development of a Train-Track (TT) model and a Trian-Track-
Bridge (TTB) model to address the challenging hanging tie problem of railway track transition
zones with track irregularities and accurately simulate the dynamic responses of bridge approaches
under moving train loads. The study included a comprehensive analysis of field data collected by
Tutumluer et al. (2024), development and optimization of analytical models and algorithms,
comparison of the developed model with existing models and commercial software, and the
development of a track property prediction model based on the fully developed TTB model and
machine learning methods. The primary objective was to develop an accurate simulation model
and gain a deeper understanding of the mechanisms causing hanging tie problems near bridge
abutments. Following the validation of the TTB model with real-world field data, the study
explored various conditions, including different shapes of tie gaps, varying lengths of tie gap zones,
and different locations of bridge approaches, such as bridge entrances and exits. The effects of
train operating speed, tie spacing, and tie types were also investigated. Additionally, a secondary
objective was to develop a predictive model for determining track properties, such as ballast
stiffness and subgrade stiffness. This model aims to predict and monitor track substructure
properties, thereby enabling more efficient scheduling of maintenance activities.
An initial research task of this PhD study involved interpreting detailed field data previously
collected along the US Amtrak Northeast Corridor near Chester, Pennsylvania (Tutumluer et al.,
2024). The data included individual layer deformations of track substructure layers and track
geometry data. Statistical analyses were performed on the field data to quantify transient responses
and performance trends at the bridge approaches. The field data revealed hanging tie issues at the
instrumented bridge approaches, caused by several sequential ties near the bridge abutments
experiencing lack-of-support conditions. These ties, with gaps underneath, exhibited oscillatory
motion due to the dynamic loading from moving train wheels. This nonuniform support worsened
the substructure conditions, leading to significant deformations, including heave due to train
passages.
The TTB model developed for ballasted track systems in this study is applicable to both regular
embankments, bridge entrance and exit locations. Simulation results from the model closely
matched the transient displacements collected in various substructure layers from the field
instrumentation at different periods within a two-year period. The model parameters were carefully
selected based on a comprehensive literature review and property information gathered from
manufacturers. The model also accounted for vertical track irregularities and examined the
dynamic responses at bridge approach locations. Comparisons were made among the prediction
obtained from the validated model and other existing analytical models and available commercial
software.
Finally, the study also developed a track property prediction model based on machine learning
methods applied on the analysis results of the validated train-track-bridge model. Collecting track
property information in the field is often difficult, time-consuming, and expensive, and it disrupts
train services. The predictive model developed in this study, therefore aimed to alleviate these
issues, providing a more efficient approach to maintenance scheduling.
This study identified that track transition zones subjected to differential settlements, impact loads,
and hanging tie issues exhibited an increasing trend in transient vertical deformations over time,
in contrast to open track locations. These transition zones also presented larger tie gaps. The
research further demonstrated that bridge entrance locations experienced more substantial dynamic
responses than bridge exit locations. Additionally, consistent tie gaps were found to result in higher
transient displacements on both ties and ballast at track transitions, while in open track locations,
only the transient displacement on top of ties increased. The length of the gap zone with a
consistent tie gap did not significantly influence transient displacements on ties and ballast, unlike
tie spacing has a significant impact on the track dynamic responses. Moreover, high train operating
speeds had a pronounced effect on transient displacements on ties and ballast, whereas lower
speeds did not exhibit a significant impact.
