April 11, 2013

Giving timetables a reliability Lift

Written by 
  • Print
  • Email

Lift, Italy, is pioneering a new approach to timetable planning which is reducing the impact of delays on operations, as Marco Chiandoni explains.

TRIESTE has a long railway tradition. The former port of the Austrian Empire is the terminus of the iconic Semmering railway, which connects the city with Vienna and was the first railway to be selected as a Unesco world heritage site.

The city is also home to Trieste University which is continuing this railway legacy as the modern-day home of one of Italian State Railways' three former engineering schools which was established in the region when it became part of Italy at the end of World War I.

Research at the institution in recent years has focused on railway operations, and has resulted in the establishment of a spin-off company, the Railway and Traffic Laboratory (Lift). Founded by Professor Giovanni Longo, the company is applying the results of this research initially in multiple case studies and now increasingly in software tools that are being adopted by various operations planners all over the world.

In 2011 the research group received the Young Researcher Award from the International Association of Railway Operations Research (Iaror) for presenting an innovative method that incorporates variables such as driving styles and stop times into timetable planning. The algorithm, which is simple and applicable even on a large scale, estimates the probability of traffic conflicts, and as a result delays within a few seconds during the timetable planning process, providing planners with an estimation of the robustness of a particular timetable.

Lift has also developed an approach based mainly on real traffic analyses and micro-simulation which considers the trade-off between the number of operated services and corresponding service reliability. This system estimates the impact on system performance and compares the reliability of different timetable alternatives.

Micro-simulation is a widely-used method to support railway planning. At Lift, thanks to the data management tool Treno, which integrates analyses of real operational data and automates several manual steps, the time required to perform a study shrinks, while the level of detail improves.

The approach has already been used to understand the impact of proposed infrastructure improvements or the layout of new lines on timetable planning.

The goal of all infrastructure managers and railways when planning infrastructure projects is to maximise the benefits of investments by selecting a flexible layout. As a result new lines and stations are often very compatible with different operational concepts. However, Swiss Federal Railways (SBB) has often planned new lines specifically for a given timetable, which is a radically different approach. Longo says to account for these varying philosophies, Lift's idea is to integrate the simulation results of different scenarios such as timetable and infrastructure with multi-criteria approaches that identify the optimal solution.

"If the railway requires a high-speed line, we start by asking the desired running time and the number of services and then develop several scenarios," he says.

This method was used to define an incremental improvement plan for Algerian Railways' core network. Starting from the anticipated demand and the expected long-term volume of services, the system developed and tested a network configuration to identify the best fit to meet these requirements at minimal cost and under different timetable options in order to assess its flexibility. The interventions were then divided into a number of steps in order to achieve the greatest benefit while avoiding unnecessary expense.

"For example we found that it would be possible to obtain more than half the expected 400% capacity increase in the first decade, while the running time reductions, which are achieved through new line sections, will be achieved more gradually," Longo says.

It is possible to use the same approach to assess the minimum number of interventions required to operate a given number of services during special events which require more trains to meet demand, such as Expo 2015 in Milan. Here Lift has estimated the individual and combined impact on traffic of over 20 interventions such as signalling improvements, changes to the layout of stations, installation of faster switches, construction of flyovers, as well as new stops.

Each intervention was simulated according to the real behaviour of trains, incorporating delays and stop times with the corresponding effect on traffic reliability under a given timetable. Lift says that the more the simulated delays were reduced by an intervention, the more the intervention was considered effective.

Lift2Infrastructure manager Italian Rail Network (RFI) fully accepted the results of the study, continuing the design of the selected set of interventions and discarding others which were shown to have a detrimental effect, including a stop in a strategic location in the city. Located between a major station and a junction it showed a significant impact on traffic reliability and therefore was dismissed despite its potential benefits.

Lift similarly estimates the reliability that could be obtained by adopting a new timetable. In 2009 the team was asked to estimate the reliability of the Milan - Rome Frecciarossa high-speed services running on the new Milan - Bologna high-speed line and their effect on other trains.

"With over 1000 trains, 13,000 itineraries and 5000 signals on a 500km line the simulation was a real challenge, especially concerning dispatching within Milan Central station, where up to six simultaneous movements are possible although no flyover is provided," says Lift president Mr Giorgio Mederossi.

The model was first carefully calibrated using real operational data to accurately reproduce the dispatching rules used by operators. Secondly, the new timetable was simulated under those conditions in order to estimate the punctuality that could realistically be obtained with the new infrastructure and timetable. A series of delay scenarios was also defined and simulated to estimate the decrease in reliability under certain conditions, and in particular the effect caused by delays on a specific line on other services.

The results of the simulation showed a higher robustness of the 2010 timetable compared with the 2009 version with the lower anticipated impact of traffic conflicts resulting in higher expected punctuality levels.

While these results are proof of the quality of the new timetable, the possibility of variations in the behaviour of the dispatcher and the initial delays were also used to simulate the most frequent delay scenarios.

By estimating their impact in presence of different dispatching criteria it is possible to suggest to the infrastructure manager ways to minimise their impact on operations and as a result reduce delays.

Lift is now working in South America, where it is currently designing the operations on the planned Line 2 of Metro Lima. In Norway its approaches and tools are allowing infrastructure manager Jernbaneverket to analyse the performance of timetables and accurately estimate the impact of infrastructure improvements. And with railways constantly looking to identify ways to get more from what they have, these tools could be invaluable in these operators' efforts to achieve high availability and reliability.

Get the latest rail news

IRJ Rail Brief newsletter covers global railway news