Complex Delay Dynamics on Railway Networks: From Universal Laws to Realistic Modeling

Railways are a key infrastructure for any modern country—their state of development has even been used as a significant indicator of a country's economic advancement. Their importance has grown dramatically in recent decades due to increasing railway traffic and government investments aimed at exploiting railways to reduce CO₂ emissions and mitigate global warming.

Despite this critical role, extreme events such as major disruptions and large delays that compromise the correct functioning of the system occur daily. These phenomena have been approached primarily from a transportation engineering perspective, while a general theoretical understanding remains lacking. A better comprehension of these critical situations could undoubtedly improve traffic handling policies.

In this work, we move toward this theoretical understanding by proposing a model of train dynamics on railway networks that aims to unveil how delays spawn and spread throughout the network.

The video below shows a side-by-side comparison of delay dynamics: the left side shows our computational model's predictions, while the right side displays real-world data from the railway network. The close correspondence demonstrates our model's ability to capture the underlying mechanisms of delay propagation.

Inspired by models for epidemic spreading, we model the diffusion of delays among trains as the diffusion of contagion among a population of moving individuals. We built and tested our model using two large datasets of Italian and German railway traffic, collected through APIs designed to provide passengers with real-time information about trains, service status, and delays.

The model adequately reproduces delay dynamics in both systems, demonstrating that it captures the underlying key factors. In particular, our model predicts that the emergence of clusters of stations with large delays is not primarily due to external factors, but mainly to interactions between different trains.

Furthermore, our model provides a quantitative account of the differences between the two railway systems in terms of the probability of contagion and delay dynamics, offering insights that could inform more effective traffic management strategies.