The Dutch rail network system is one of the busiest networks in the world. Small delays can have large implications elsewhere. ProRail has recently introduced a new way to test new timetables: the stability test. This test is being performed with SIMONE, a simulation model for large-scale rail networks. SIMONE enables the assessment of the design on national and regional timetables for stability and robustness in an objective way.

The Dutch rail network system is used intensely and characterised by a complex traffic model where different types of trains run in relative high frequencies. In order to provide enough possibilities to the travellers, there are correspondances at many stations. Due to the intensive usage and the correspondances, a large number of dependencies exist between trains, which results in delays being transferred from one train to another.

ProRail, the organisation for capacity management and rail network safety, uses the DONS-system to produce basic hourly patterns. DONS is an expedient to create conflict free timetables based on market needs, infrastructure and material input. Calculations are made for the running times and dwell times of trains. Next to this, constraints are taken into account regarding the number of trains on different routes, correspondances and headways. Also, the moments of arrival, departure and passing as well as track usage for train movements are specified and necessary slack is applied to cope with variances in the processes.

Stability test.
The strategic and tactical timetable studies compare different variations of timetables with criteria like transport size, travel time, security, environment, costs and quality. Quality here is the degree of feasibility or the stability of the timetable. Important is the sensitivity to relatively small disturbances. Small in this context means that no large interventions are needed like the cancellation of trains, and therefore the plan stays intact. The question is if the planned timetable design has the ability to recover, or to "regulate itself back". Is the plan robust or does it derail after small disturbances?

The concept stability can be measured in different ways. Important indicators, amongst others, are to what extent delays are reinforced or faded out and at what speed. Another important factor is how much of the available slack in the timetable was needed and what part of the network was hit by the delays.

In order to be able to assess the quality, ProRail Innovation has developed together with the Passenger division of Dutch Railways (NS reizigers) and Incontrol Enterprise Dynamics, the stability test, a methodology where simulation stands central. There are several possibilities to disturb the planned processes. Variations can be applied to the processes, the amount of disturbance, and the timeframe when it is being applied. With the stability test, different timetables can be compared with each other. The results could lead to additional experiments, for example to research more closely the influence of a bottleneck.

Simone.
There are few examples at hand where the calculations and analysis of a complete rail network system based on a (simulation) model has been conducted. In general (simulation) studies are carried out on a limited part of the network, like a junction or certain lines. Based on studies conducted with FASTA ProRail has chosen the rough-to-fine approach. According to this approach, the network is viewed from a high abstraction level. The results on the network level could be the starting point to conduct locally oriented studies. The stability test is performed by SIMONE, SImulation MOdel for Networks. SIMONE is jointly developed by ProRail and Incontrol Enterprise Dynamics. The simulation model is used to judge and test the timetables objectively and to support decision making in a more transparent way. With SIMONE, insight is gained in the sensitivity of the disturbances and knowledge about the complexity of the network is acquired.

For example, SIMONE can support:

• Determination of the robustness of a timetable (in how far can a timetable resist disturbances);
• Improvement of the timetable by relating the design standard and the robustness of a timetable in the (simulation) reality;
• Trace and quantification of bottlenecks in a network;
• Analysis of cause and effect relations when delays emerge.

SIMONE is linked to the DONS system in order to generate simulation models of national networks and timetables in an automated way. After generation, these simulation models are instantly ready to use. Figure 2 shows an example of the animation of a SIMONE simulation model.

To emphasise the possibilities of SIMONE, four example studies are highlighted: simulation of a national model and three local studies in Weesp, Hengelo and around Den Bosch.

National model.
SIMONE has been used in the project 'effect', which is part of a larger project called 'evaluation Prorail'. This project was initiated by the Ministry of Transport to determine the effects on supply and demand of the investments in the train infrastructure between 1990 and 2000. The goal of this project was to find the difference between the expansion of the infrastructure and the (imaginary) situation as if the expansions had not taken place. In addition, the timetable of 2000 has been projected on the infrastructure of 1990. The newly created situation is compared with the 'actual' situation in 2000. Further, statements are made on travel time improvements, the associated number of passengers and the stability of the train service. The latter has been determined with the aid of SIMONE. Figure 3 shows a so called network graph which is the result of a calculated scenario by SIMONE.

Network graphs are used during the Prorail analysis. A Network graph contains an overview of the complete network where different performance indicators can be shown. An example is punctuality; the percentage trains that arrive within a certain delay (here 3 minutes) at a station. The colours run from green (trains with delay of less than 3 minutes) to red (trains with delays of more than 3 minutes). It must be noticed that the results in figure 3 are from a fictitious simulation experiment. The graph has no relation to the reality or the Prorail study.

Weesp.
The Capacity Allocation department of ProRail has used SIMONE to assess the role of correspondances at Weesp station in relation to the rest of the network. Weesp is a station where four pairs of trains correspond on each other. Next to this, Weesp is a station located at on of the busiest parts of the network. Hence, the goal was to assess the influence of Weesp on the network.

To answer this question, a model that contained the entire Dutch network was used. Two variants of this model were made: a basic one according to the timetable of 1999 where trains only waited 3 minutes for another connecting train and a variant where the correspondances were active but where trains did not wait for each other.

A series of simulation experiments were conducted, after which the two variants were compared with each other. In the experiments, disturbances were applied to running times and the dwell times and therefore trains were deviating from their timetables. Next to these experiments with general disturbances, there were also experiments with specific disturbances for each specific train.

The study led to the result that, although the network was only partially sensible to the correspondances in Weesp, there was a distinct difference between the variant with and the variant without waiting for correspondances. As well as the positive effects (increase in the punctuality of trains that do not need to wait for the correspondances) as the negative (decrease of reached correspondances) were made clear and weighted.

Hengelo.
In Hengelo the effects of a change in the timetable were researched. The change can be implemented by adding an extra switch at the railway yard. The extra switch enables the simultaneous departure from Hengelo for the stopping train series from Zutphen, Hengelo, Oldenzaal and Zwolle, Hengelo, enschede. By removing the original conflict, it became possible to let the train serie Zutphen, Hengelo, Oldenzaal depart two minutes earlier from Hengelo. By this adjustment in the timetable, extra time is created for the turning in Oldenzaal. The general conclusion of the study was that at first sight the removal of the conflict has negative local effects in Hengelo; trains make up less delays and the percentage of achieved correspondances was significantly smaller. The benefit was reached in Oldenzaal where the train coming from Hengelo has to make a turn in a short time. SIMONE was used to quantify what the effects were under different circumstances and where these effects occureed.

Den Bosch.
The triangle Den Bosch consists of the rail connections between Den Bosch, eindhoven and Tilburg. The Capacity Allocation department of ProRail wanted to know what the effects would be when 6 extra freight trains are added to the timetable. A simulation model was used that contained the entire Dutch network, but only the trains that pass Den Bosch, eindhoven or Tilburg. This accounts for in total 206 trains. With this model, two variants were made: a basic variant with a timetable without the 6 extra freight trains and a variant with the 6 extra freight trains. In figure 1 the difference in the average delay is shown for ten different experiments with so-called noise disturbances. Noise means that all running and dwell times are effected by disturbances are affected with a certain chance. The SIMONE results show that the average delay of all the trains in the situation with 6 extra freight trains has a 10% higher score than in a situation without the 6 extra freight trains. Network graphs demonstrated where the biggest differences between the two variants would occur. Some differences could be directly accounted for from the general knowledge of the timetable and the network, others were more surprising.

Future.
SIMONE started in 1997 as a research study. The first application of SIMONE was the project evaluation of Prorail. Next to this, for their final thesis and under the guidance of ProRail Innovation and the Passengers division of Dutch Railways (NS reizigers), four students have researched fundamental issues such as the relationship between the amount of slack in the timetable and the length of the stretch.

Currently, ProRail and the Passengers division of Dutch Railways use SIMONE. In the future, these organisations will use SIMONE more and more for supporting their decisions on planning, design and usage of the network. The concepts are suitable for the approach of similar problems. Expansion possibilities of SIMONE are the connection with other timetable systems and the detailed modelling of track layouts at the station yards.

Conclusions.
When SIMONE had just been developed it was unsure if the goals could be achieved. The simulation of the entire network is a complex matter. Now, SIMONE can be called a success and SIMONE is deployed for the goals it was originally designed for: an objective and transparent assessment of the stability of a train network and the associated timetable. The results of simulations with SIMONE increase the knowledge about the performance of rail networks from a large number of different angles. Benefits and disadvantages of a certain configuration of a network can be made clearer. The implications of granting certain market demands in term of network performance can now be better weighted. One remark however, is that SIMONE is not used for a long time yet.

SIMONE is applied to the rail network, but it is expected that the use of simulation and the SIMONE methodology can also be applied to the design of complex transport and distribution networks. especially with the contemplation of a complete network, a top down approach is of essential importance in not getting bogged down in details.

This article is also available in Adobe Acrobat PDF format

For more information or if you want to know how we can help you with your challenge, please contact us by e-mail.