PhD candidate Rolf van Lieshout of Erasmus School of Economics conducted research into a 'self-organized railway system' that can be relied upon when mass train cancellations are imminent. In this way, an out-of-control situation, such as that on Sunday 3 April when no trains ran, could possibly be prevented. Models show that trains can still run on a large scale if the timetable and centralised control are released at such times.
‘A perfect storm' is how econometrician and researcher Rolf van Lieshout sees the train failure of Sunday 3 April, when the NS was forced to bring train traffic to a complete standstill due to a malfunction in the planning systems. The train failure is a textbook example of exactly the type of situation he has been researching for the past four years: an out-of-control situation, where there are no disruptions to the infrastructure or trains and enough staff are present, but trains are still not running due to a lack of information and overview in the central control centre.
"This national disruption suddenly makes it very easy to explain the relevance of my research. Unfortunately, it is too late to include it in my dissertation itself, but I can certainly use it for my layman's talk," says the researcher, who also experienced disruption himself on 3 April but ultimately made it home by metro from The Hague to Rotterdam.
Winter weather around 2010
Normally, the central NS control room has a complete overview of all planned journeys and keeps track of the current situation of trains and drivers. Van Lieshout's PhD research is aimed at exploring an alternative form of control where a switch can be made during an out-of-control situation when the central control system can no longer keep up and the overview has been lost. The reason was the series of severe winters from 2009 to 2012, which regularly caused total chaos on the railways. "There were a number of days when it froze and snowed heavily, which meant that there were a few hitches every winter. The NS and ProRail therefore sought a solution to limit the disruption."
One of the new protocols that is now being switched to when weather conditions are poor is to scale back frequencies, so that on some routes, for example, only two trains an hour run instead of four. This is a prevenive measure that sometimes turns out to be unnecessary and does cause inconvenience to passengers. As an alternative, the PhD candidate therefore investigated a 'self-organized rail system' in which trains and personnel can be deployed independently of the central control system.
In this system, the timetable is completely abandoned and trains run based only on frequency. "Equipment and staff are allocated to journeys at the end stations on the basis of locally available information. There is no communication between different stations or with the central control centre. One condition, of course, is that the signals can still be operated, so that safety is guaranteed."
Real timetable-free driving
"At the final stations, you then need to have a kind of station manager. This person knows which trains and which drivers are available and then determines where a train must go, what time it leaves and which driver will drive it," explains the researcher. They make the decision based on simple protocols and completely independent of modern technology. For example, if a train has just left The Hague Central Station for Rotterdam, it is better to give the next train arriving at The Hague Central Station another destination. This method of control does result in a lower frequency of trains and delays, but a total out-of-control situation does not occur.
In his research, Van Lieshout started with a simple model of the train network, which he gradually expanded if it turned out that no congestion occurred. The researcher eventually ran a microscopic model on part of the railway network. Such a model is very difficult to make, as it contains all the control elements such as signals and switches, as well as the train's characteristics such as its braking distance. At each subsequent step, the strategies developed turned out to work quite well. "In our experiments, we observed that the train traffic always remained reasonably steady and no major congestion occurred, for example."
According to the researcher, this would be an enormous advantage, because the alternative would be to shut down all rail traffic. However, the model is of limited use in a situation with extreme weather because there will almost certainly be many new disruptions that need to be addressed. For a situation such as that on 3 April, this protocol could serve as a good back-up.
Van Lieshout: "When it's snowing, passengers still have an understanding for it, but this was a beautiful sunny day. Not running any trains at all would not only inconvenience passengers directly, it would also damage the reputation of the NS. I don't know whether it will ultimately be implemented, because it really is a radically different way of running trains. Follow-up research is also needed to translate theory into practice. And do you really want to develop an entire protocol for those rare occasions when this happens? But the chance that this will be continued has increased after 3 April.”