Research question and state of the art

Planning and evaluation of railway operations are tasks often performed with the help of simulations having the aim of estimating the reliability of the system. Punctuality and delays are important performance indices. Low performance can affect the demand to a high amount, and should be avoided. The simulations can be performed with microscopic simulation models, where everything is modelled in detail, or more general with macroscopic models, depending on the requirements of the application. Simulations are often used to test proposed timetables for evaluating how the timetable would work in real-life operations and to investigate travel time, delays and the resulting capacity utilization. An important aspect for simulations is how to handle incomplete or unknown input data. An example is the departure times of freight trains in Sweden. They do not necessarily depart according to the timetable, but rather when they are ready and there is a free train path in the timetable. Hence, their timetable can be considered unknown. This study presents a method for dealing with incomplete data regarding freight train departure times, such that timetables for simulation purposes can be created. A representative timetable is evaluated by a microscopic and a macroscopic railway simulation model to investigate how they perform when the available input data is incomplete.

When assessing the performance of a railway system, reliability is an important measure. Often, the reliability is estimated by simulation. Microscopic simulation models are commonly used to simulate railway networks. In order to represent reality as exactly as possible, they have a high level of detail for the infrastructure – with exact station layouts, including the placement of switches, signals, etc. – as well as for the trains, which incorporate detailed specifications. For the best representation of a system, Borndörfer et al. (2018) recommend dynamic, synchronous, microscopic, stochastic simulation. Several different microscopic simulation tools are available, e.g. the commercial alternatives RailSys (Bendfeldt, et al., 2000; Radtke & Hauptmann, 2004), and LUKS (Janecek & Weymann, 2010), and the open source simulator OpenTrack (Nash & Hürlimann, 2004). 

However, when larger networks are simulated, the amount of details in microscopic models yields long simulation times and increased complexity for the user. Macroscopic models with a lower level of detail can, therefore, be preferred. A macroscopic model for delay propagation in large networks was presented by Büker and Seybold (2012). Zinser et al. (2018) presented a macroscopic simulation model and performed a case study comparing it to a microscopic simulation approach for infrastructure disruptions, showing promising results for the new model. This model was subsequently named PRISM and further developed (Zinser, et al., 2019), creating what is essentially a macroscopic simulation tool. An adjustable model capable of micro-, meso- and macroscopic simulations was introduced by Cui et al. (2018). Efforts have also been made to integrate microscopic and macroscopic models to simplify data management while improving timetable forecasting (Huber & Wilfinger, 2006).

Method

The specific input needed for the railway simulation models differs between microscopic and macroscopic models. In case of unspecified simulation input data, a full timetable still has to be created to run the simulation. 

The proposed method is to create a timetable which accounts for the variation in freight train departure times compared to their scheduled departure times. Empirical data about the actual departure time for the trains is collected and mapped to a probability distribution. For each train with uncertain departure time in the timetable, a departure time is drawn from the distribution and inserted into the timetable. If the drawn departure time is in conflict with the departure time of another train, the freight train departure time is scheduled to the earliest possible departure after the conflict is resolved. In this way, the freight trains will obtain an actual departure time whose deviation from the scheduled departure time is reflecting the real-life statistics. An advantage of that method is that the results can be used in both simulation tools, and that the new timetable can be considered as the undisturbed one, i.e. can be simulated with additional disturbances. A representative resulting randomized timetable is simulated with the macroscopic simulator PRISM and the microscopic simulator RailSys, and the performance is compared. Focus is on the conflict-solving and on the resulting delays. In addition, the generated timetables are compared to in-tool-generated ones.

Analysis and results

Neither RailSys nor PRISM have well-working tools for generating timetables. However, both offer the opportunity to insert entry delays which with adequate distributions can be used to steer the departure time in the way the new method aims for. That means that simulation can be used in order to create the timetable, which requires a 2-step process. In addition, the created timetable has to be exported into the second tool.  The result from this study is a method to create realistic timetables for simulation purposes when some trains have an uncertain departure time. The uncertain departure times in the timetable are based on statistics of actual departure times for the considered trains. The resulting timetables are suitable for simulation in both microscopic and macroscopic railway simulations and improve the simulation models’ reflection of real operational conditions.

The preliminary results for the comparison of the simulations in both models show that the estimated delays are of similar amount. A run in RailSys is much more time-consuming than in PRISM. This allows PRISM to simulate more timetables with lower effort and in less time than with RailSys. In that way, statistics can be aggregated for the system, for example for comparing it to an adjusted supply of freight trains. With RailSys, however, the simulation results are closer to reality as more details are incorporated in the simulation model.

References

Bendfeldt, J.-P., Mohr, U. & Müller, L., 2000. RailSys, a system to plan future railway needs. WIT Transactions on the Built Environment, Årgang 50, pp. 249-255.

Borndörfer, R. et al. red., 2018. Handbook of Optimization in the Railway Industry. s.l.:Springer.

Büker, T. & Seybold, B., 2012. Stochastic modelling of delay propagation in large networks. Journal of Rail Transport Planning and Management, 2(1), pp. 34-50.

Cui, Y., Martin, U. & Liang, J., 2018. PULSim: user-based adaptable simulation tool for railway planning and operations. Journal of Advanced Transportation, Årgang 2018.

Huber, H.-P. & Wilfinger, G., 2006. Integration-Enhancements for Microscopic and Macroscopic Railway Infrastructure Planning Models. Montreal, Canada, Proceedings of 7th World Congress of Railway Research.

Janecek, D. & Weymann, F., 2010. LUKS-Analysis of lines and junctions. Lisbon, s.n.

Nash, A. & Hürlimann, D., 2004. Railroad Simulation using OpenTrack. Computers in Railways IX, pp. 45-54.

Radtke, A. & Hauptmann, D., 2004. Automated planning of timetables in large railway networks using a microscopic data basis and railway simulation techniques. WIT Transactions on The Built Environment, Årgang 74.

Zinser, M. et al., 2019. PRISM: A Macroscopic Monte Carlo Railway Simulation. Tokyo, Japan, Submitted to the 12th World Congress on Railway Research.

Zinser, M. et al., 2018. Comparison of microscopic and macroscopic approaches to simulating the effects of infrastructure disruptions on railway networks. Vienna, Austria, Proceedings of 7th Transport Research Arena TRA 2018.