Improving the robustness of the railway system in Brussels

Thijs Dewilde, Peter Sels, D Cattruysse, Pieter VansteenwegenGhent 2012

Abstract

In order to improve the robustness of a railway system in station areas, this paper introduces an iterative approach to successively solve the route choice problem in station areas to optimality and to improve this solution by applying some changes to the timetable in a tabu search environment. Using a discrete event simulation model, the performance of our algorithms is evaluated based on a case study for the Brussels' area. The railway network of the Brussels' area is introduced and its relevance is emphasized. Computational results indicate an improvement in robustness of about 10%, a decrease in knock-on delay of more than 15%, and a 25% reduction in the number of trains that are confronted with conflicts.

Introduction

In complex railway networks, station areas often serve as the primary bottlenecks. Traditional railway planning typically follows a sequential "timetable-first" approach: planners first establish a fixed timetable and subsequently assign specific tracks and routes. While this ensures feasibility under ideal conditions, it often neglects robustness—the system's inherent ability to absorb minor, unavoidable disturbances (such as extended passenger boarding times) without triggering a cascade of "knock-on" delays across the network.

This study focuses on the Brussels railway corridor, one of the most congested junctions in Europe. The North-Central-South axis handles approximately 80-85 trains per hour, all of which must funnel through the six platforms of Brussels Central station. With capacity nearly saturated, even a 30-second delay can cause intersecting paths to conflict, leading to widespread inefficiency. This paper proposes an iterative framework to design schedules that minimize real travel time by maximizing operational buffers.

Methodology

The researchers developed a two-stage iterative algorithm designed to maximize the "time span" or buffer between trains sharing the same infrastructure. The framework alternates between two specialized modules:

  1. Route Choice Module: This module employs an independent-set method on a conflict graph. It identifies the mathematically optimal set of routes through a station for a given timetable, specifically aiming to maximize the minimum time gap between any two trains.
  2. Timetabling Module: Operating within a Tabu Search environment, this module suggests minor adjustments to train arrival and departure times. These small shifts are designed to de-conflict the schedule, providing the routing module with more flexibility to find robust paths.
  3. The Iterative Loop: The system cycles between these modules, using the output of the routing optimization to inform the next set of timetable tweaks. This process continues until the solution converges, which typically occurs within a few iterations.

To validate the effectiveness of these algorithms, the team utilized a discrete event simulation model. This allowed them to test the optimized schedules against the stochastic "chaos" of real-world operations, such as the reality that planned 1-minute dwell times often extend to 1.5 minutes in practice.

Case Study: The Brussels Junction

The study specifically targets the North-Central-South axis in Brussels. This area is characterized by a high density of traffic and a complex "crisscross" of lines. The physical constraints of Brussels Central station, which acts as a narrow funnel for the entire Belgian rail network, make it an ideal testbed for robustness optimization. The model accounts for the physical reality of the infrastructure, ensuring that proposed routes and times are practically executable.

Key Results and Insights

The application of the iterative approach yielded substantial improvements over traditional feasible-only planning methods:

  • 10% Improvement in Robustness: The network demonstrated a significantly higher capacity to absorb small disturbances without affecting the overall schedule.
  • 15%+ Reduction in Knock-on Delays: By increasing the buffers between trains, the secondary delays caused by an initial late arrival were curtailed by more than 15%.
  • 25% Fewer Conflicts: The number of instances where a train was forced to slow down or stop due to another train occupying its path dropped by a quarter.

These results suggest that robustness should be treated as a primary objective during the planning phase rather than an afterthought. Furthermore, the study highlights that massive infrastructure overhauls may not always be necessary; intelligent, localized shifts in a timetable can significantly alleviate bottleneck congestion.

Limitations and Future Directions

While the framework is highly effective for managing "small disturbances," its performance during major "black swan" events—such as total power failures or major infrastructure breakdowns—remains to be explored. Additionally, while the current focus is on the scheduling and planning phase, there is significant potential in integrating this iterative logic into real-time Traffic Management Systems (TMS) to allow for dynamic adaptation during operations. Future research may also address the computational overhead required to scale this approach from a localized bottleneck to a full national network.

Summary updated

Abstract

In order to improve the robustness of a railway system in station areas, this paper introduces an iterative approach to successively solve the route choice problem in station areas to optimality and to improve this solution by applying some changes to the timetable in a tabu search environment. Using a discrete event simulation model, the performance of our algorithms is evaluated based on a case study for the Brussels' area. The railway network of the Brussels' area is introduced and its relevance is emphasized. Computational results indicate an improvement in robustness of about 10%, a decrease in knock-on delay of more than 15%, and a 25% reduction in the number of trains that are confronted with conflicts.
BibTeX
@inproceedings{dewilde2012improving,
  author = {Thijs Dewilde and Peter Sels and D Cattruysse and Pieter Vansteenwegen},
  title = {{Improving the robustness of the railway system in Brussels}},
  booktitle = {Ghent},
  year = {2012},
}
Figure