2021

Bhagya Shrithi Grandhi, Technische Universitat Braunschweig
Emmanouil Chaniotakis, University College London
Constantinos Antoniou, Technische Universitat Munchen
Stephan Thomann, Mount Sinai Beth Israel Hospital Sol and Margaret Berger Department of Urology
Felix Laube, Emch + Berger Holding AG

The transportation sector is one of the significant contributors to global warming; immense efforts are being taken to make transport sustainable with railways to be considered one of the solutions. However, their ridership is decreasing over the years, something attributed to its unreliability and lack of punctuality. Railways being complex systems, any disruptions to the operations adversely impact the whole network, causing delays. Though many studies  discuss the effect of an incident to a network, the majority of them mostly focus on a specific parameter of the system with little attention to holistic quantification. This research aims at developing an operational framework which predicts the impact that an incident might have on the network. It allows railway dispatchers to better prepare operations in response to an incident. A sequence of models is created, which predicts two response variables, firstly the incident duration, which in turn is used to predict the second variable, total delay caused by the incident. Various influencing attributes are examined such as weather, network-- and railway--related attributes. The relationship between the attributes and the response variables is studied, to understand the possible incident impact. The base data used for the study is real incident data from Denmark railways. The results show that neural networks provide the best models for both response variables. Also, the operational framework improves the prediction with highest accuracy compared to the standalone models in the study. {Keywords}: Railway Incidents, Incident duration, Total delay,  Operational Framework, Predictive modelling

Mehmet Şirin Artan, Yildiz Technical University
Ismail Sahin (sahin@yildiz.edu.tr), Yildiz Technical University

Despite technological improvements in railway transportation to support reliable train services, uncertainty still plays a vital role affecting the overall performance of the railway system. Timetable is a final product of the planning process resulting in determining the types and number of train services in the next season in the railway network. The rolling stock and train crew requirements are also evaluated and satisfied for appropriate service provision. The timetabling process is under the influence of uncertainties associated with all these components of train services. Particularly, time allowances (i.e., running time supplements and buffer times) are embedded in the timetable to mitigate undesirable effects of time wise fluctuations of train services. In this study we use the Markov chain models to measure the performance of trains and the timetable. The time dependent performance analysis allows us to predict the running time, arrival time and delays at the subsequent activities of trains. The time independent performance analysis, on the other hand, evaluate the long-term performance of timetable corresponding to the time allowances. We have developed the models and test their performances using real-world data collected in the Netherlands railway network. The analysis and evaluation show that the part of the railway considered in this study performs well and the Markov models developed can be used effectively to make various analyses and obtain performance measures.

Yuexin Wang (1290879030@qq.com), Southwest Jiaotong University
Chao Wen, Southwest Jiaotong University
Rui Hu, Southwest Jiaotong University
Ping Huang, Southwest Jiaotong University
Chuanling Xu, Southwest Jiaotong University

The delay of high-speed trains has always affected the quality and efficiency of railway passenger transportation. Investigating the rules of delay recovery is a critical step for dispatchers to make reasonable dispatching decisions in time to reduce train delays. Based on the train operation data on the Wuhan to Guangzhou high-speed railway in China, this study applied a support vector regression (SVR) algorithm into train delay recovery prediction. First, the independent factors that have an impact on delay recovery were determined through Spearman correlation analysis, which are the delay recovery type, train class, delay time, supplement time, and train operation interval. Second, a grid search method was applied to determine the hyper-parameters of the SVR model. Then, the predictive model was trained and tested with the delay recovery cases of high-speed trains. The results show that the predicted delay recovery cases of the model highly coincided with the actual cases. Additionally, the predictive performance of the model under allowable errors was examined, which shows that the accuracy of the model can reach 95.96% with 1 minute allowable error. Finally, comparison analyses show that the proposed model outperforms other widely used train delay recovery models in terms of both regression and classification metrics.

Bowen Gao, Tongji University
Decun Dong, Tongji University
Yusen Wu, Tongji University
Dongxiu Ou, Tongji University

The rescheduling of train timetables under a complete blockage is a challenging process, which is more difficult when timetables contain lots of trains. In this paper, a Mixed Integer Linear Programming (MILP) model is formulated to solve the problem, following the rescheduling strategy that blocked trains　wait　inside stations during a disruption. In addition, while the exact end time of the disruption is known, trains at stations downstream of the blocked station are allowed to depart early. The model aims at minimizing the total delay time and the total number of delayed trains under the constraints of station capacities, activity time, overtaking rules and rescheduling strategies. Because there are too many variables and constraints in the MILP model, a three-stage algorithm is designed to speed up the solution. Several experiments are carried on the Beijing-Guangzhou high-speed railway line from Chibibei to Guangzhounan, with 162 trains (29 cross-line trains and 133 local trains) in the original timetable. Simulation results show that our model can handle the optimization task of the timetable rescheduling problem very well and the three-stage algorithm greatly improves the solving speed of the model. Since all instances can get a better optimized disposition timetable within 450-600 seconds, our approach is acceptable for practical use.

Guei-Hao Chen, Sinotech Engineering Consultants, Inc.
Jyh-Cherng Jong, Sinotech Engineering Consultants, Inc.
Anthony Han, National Chiao Tung University

Crew scheduling is one of the crucial processes in railroad operation planning. Based on current regulations, working and break time requirements, as well as the operational rules, this process aims to find a duty arrangement with minimal cost that covers all trips. Most past studies considered this subject for railroad systems as an optimization problem and solve it with mathematical programming based methods or heuristic algorithms, despite numerous logical constraints embedded in this problem. Few studies have applied constraint programming approaches to tackle the railroad crew scheduling problem. In this paper, we propose a hybrid approach to solve the problem with a constraint programming model for duty generation, and an integer programming model for duty optimization. We have applied these models to the Kaohsiung depot of Taiwan Railways Administration, the largest and the only railroad operator in Taiwan. The encouraging results show that the proposed approach is more efficient than the manual process and can achieve 30% savings of driver cost. Moreover, the approach is robust and provides flexibility to easily accommodate related operational concerns such as minimizing the number of overnight duties. Thus, this hybrid two-phase approach seems to have the potential for applications to the railroad crew scheduling problems outside Taiwan.

Rafael Mendes Borges (r.mendesborges@tudelft.nl), Technische Universiteit Delft
Egidio Quaglietta, Technische Universiteit Delft

The Hyperloop is a concept of a ground transportation system consisting of capsules (called pods) traveling at very high-speeds in near-vacuum tubes. The hyperloop aims to be a fast, cheap, and sustainable alternative to short-haul flights and high-speed rail. The small pod size requires very high frequencies to respond to future high levels of passenger and cargo demands. Media and representatives of the emerging Hyperloop industry acclaim the Hyperloop as a very capacity‑effective transport system, however there is no clear scientific evidence proving that. A theoretical investigation is therefore necessary to understand which capacity the Hyperloop could safely provide and whether that could satisfy the future transport demand. This paper provides a comparative analysis of the capacity that the Hyperloop can offer for several operational scenarios and different signalling systems, including Moving-Block and the advanced concept of Virtual Coupling. Results show that Moving-Block could achieve required transport capacity levels only if pods could use high deceleration rates likely to be unsafe and uncomfortable to passengers. Virtual Coupling is instead observed to be a more satisfactory operational concept that could address transport demand while respecting safety and comfort standards if reliable pod platooning technologies are available.

Pengling Wang (pengling.wang@gmail.com), Eidgenossische Technische Hochschule Zurich
Nikola Besinovic, Technische Universiteit Delft
Rob M.P. Goverde, Technische Universiteit Delft
Francesco Corman, Eidgenossische Technische Hochschule Zurich

The efficient use of regenerative energy directly contributes to reducing the amount of energy to be purchased. This paper presents a periodic timetable fine-tuning method for the utilization of regenerative energy and shaving power peaks with maintaining the structure and robustness of the original timetable. A mixed integer linear programming model based on the periodic event scheduling problem framework is proposed to construct a set of feasible timetables that best on increasing synchronized acceleration and braking events with maintaining the timetable robustness at a certain level. Then a linear programming model is developed to find the optimal timetable that leads to a minimum power peak value. The proposed models were adopted to fine-tune the timetable of 2019 for a Dutch railway network. The results show that the optimized timetable outperforms the original timetable in the utilization of regenerative energy and shaving power peaks. For example, the total energy consumption (traction energy-regenerative energy) decreases by 5.98\%. The 15-minutes power peak value decreases by 8.51\%. As the energy bill is determined by the total energy consumption and the maximum power peak value in some countries, a large amount of energy cost reduction can be expected by using the optimized timetable.

Steven Kirkpatrick, Applied Research Associates Inc
Tyler Dick AR030 AR0405A, University of Illinois, Urbana-Champaign
Xiang Liu, Rutgers University

The analyses of the historical data on derailments indicated that unit trains have, on average, increased derailment severities compared to manifest trains.  In addition, there has been a significant increase in hazmat unit train service over the last decade.  These factors could potentially have an impact on hazmat rail transportation safety.  To address this issue, it is important to understand the factors of unit train operations that result in an increase in the derailment severity.  A model was developed and applied to investigate the factors that influence the severity of a given derailment.  The model developed is a simplified analytical model of the longitudinal kinematic of the residual train behind the point of derailment.  The model includes the important forces acting on the train in a derailment including the braking behavior, blockage forces produced by the derailed cars, rolling resistance, and grade.  The resulting model was validated against data from various derailments.  The model was then applied to evaluate various derailment conditions and provides several insights into the importance of various factors that effect derailment severity.

Matthew Parkes, University of Illinois, Urbana-Champaign
Tyler Dick (ctdick@illinois.edu), University of Illinois, Urbana-Champaign
Adrian Diaz de Rivera, University of Illinois, Urbana-Champaign

In response to multiple derailments involving hazmat trains, in early February 2020 Transport Canada released ministerial order (MO) 20-02, imposing speed restrictions of 20 mph to 25 mph on trains transporting a sufficient quantity of hazardous material. Since much of the North American freight network is used by multiple train types, the extreme speed heterogeneity created by this mandate substantially reduced train performance. Although this order was replaced within two weeks by new speed restrictions, that were in turn replaced in May, MO 20-02 introduced the most extreme levels of train speed heterogeneity. The research team investigated the corresponding capacity effects to better understand the effects of train speed heterogeneity at low speed and inform agencies on future speed restrictions in this range. Using Rail Traffic Controller and General Train Movement Simulator, we quantitatively investigated the capacity loss due to these speed restrictions and found that MO 20-02 can lead to mainline capacity loss in excess of 60%.