Weak Nodes Detection in Urban Transport Systems: Planning for Resilience in Singapore

Michele Ferretti, Gianni Barlacchi, Luca Pappalardo, Lorenzo Lucchini, Bruno Lepri

Abstract

Urban transport planning has traditionally prioritized efficiency, often at the expense of resilience. This study introduces ACHILLES, a web-based simulation tool that identifies "Urban Achilles Heels"—specific weak points in a city's transport network that, if disrupted, cause disproportionate damage to overall connectivity. By modeling Singapore’s bus network as a multiplex network, the researchers demonstrate that while the city is remarkably resilient, it remains highly sensitive to the loss of "weak ties" that serve as critical bridges for specific urban zones.

Introduction: From Efficiency to Resilience

Traditional urban transport planning focuses on moving the maximum number of people as quickly and cheaply as possible under normal conditions. However, as cities face increasing threats from climate change, infrastructure failures, and other disasters, efficiency alone is insufficient. Resilience—the ability of a network to absorb stress and recover—is becoming a critical metric for urban survival.

Urban mobility is a "wicked problem" characterized by social complexity. A key challenge lies in identifying which specific routes or zones are most fragile. For instance, if a flood impacts a specific intersection in Singapore, it is vital to know whether the entire city's bus network will be compromised or if the impact will remain localized. Identifying these "weak nodes" is essential for effective humanitarian aid and disaster preparedness.

Methodology: The Urban Multiplex Network

The core of this research is the Urban Multiplex Network (UMN). Unlike a simple graph where all connections are treated equally, a multiplex network treats every bus line as a separate "layer."

The Urban Multiplex Network (UMN) model treats different transport layers as interconnected graphs.

Key technical components include:

  • Connection Intensity: A metric combining the actual flow of people (mobility flux) with the "selectiveness" of a zone (the number of neighbors it shares with others).
  • Layer Relevance: This calculates the contribution of a specific transport layer (e.g., a single bus line) to the reachability of a zone.
  • Heel-ness: This namesake metric measures a node's importance by analyzing its weakest connection and determining how many other nodes become unreachable if that connection is severed.
  • ACHILLES Tool: A web-based application built with Flask and WebGL that allows planners to execute "what-if" scenarios, such as adding or removing stops and lines, to observe real-time impacts on city-wide resilience.

Data Sources

The study focuses on Singapore, utilizing four primary data streams:

  1. Transport Data: Static and real-time data on bus lines, stops, and GPS traces from the LTA Data Mall.
  2. Administrative Boundaries: Shape-files defining 323 urban zones.
  3. Mobility Flows: Origin-Destination (O-D) matrices estimated from mobile network data (provided by DataSparkAnalytics).
  4. Flooding Data: Forecasts of flood-prone areas from Singapore’s Public Utilities Board (PUB) to ground simulations in real-world disaster scenarios.

Key Insights: The Strength of Weak Ties

A primary finding of the study is the "strength of weak ties." Borrowing from sociological theory, the authors discovered that routes with low connectivity scores—"weak ties"—are often the most vital for network stability. Removing these "bridges" causes the network to degrade much faster than removing high-traffic "strong" routes.

The ACHILLES interface visualizing connectivity through 3D extrusions during a simulation.

Furthermore, the study emphasizes that resilience is not merely redundancy. Simply having numerous bus lines is ineffective if they all follow the same path. True resilience comes from "connection redundancy"—a diversity of layers that allow a zone to remain connected even if one mode or route fails. The ACHILLES tool visualizes these vulnerabilities using 3D extrusions on a map, where the height of a zone represents its connectivity value.

Results and Resilience Thresholds

The simulations revealed that Singapore’s bus network is exceptionally robust, with all nodes remaining reachable even after a random removal of up to 30% of links. However, targeted vulnerability is a significant concern. When "weak ties" (nodes with high Heel-ness) were specifically targeted, global connectivity dropped significantly faster.

Specifically, removing the 90% "strongest" links left 97% of the city reachable. Conversely, removing the 90% "weakest" links resulted in nearly 20% of urban zones becoming completely isolated. The tool successfully identified four specific vulnerable areas in Northern Singapore that act as critical "Achilles Heels" in the event of localized flooding.

Conclusion and Future Directions

While the current model is limited to the bus network, the authors acknowledge that a multimodal model including MRT and taxis would provide a more comprehensive picture. Future work will also aim to account for dynamic commuter behavior during disasters. The framework is designed to be site-agnostic, making it potentially valuable for cities with more fragile infrastructure where identifying "Achilles Heels" is even more critical.

Summary updated

BibTeX
@inproceedings{ferretti2018weak,
  author = {Michele Ferretti and Gianni Barlacchi and Luca Pappalardo and Lorenzo Lucchini and Bruno Lepri},
  title = {{Weak Nodes Detection in Urban Transport Systems: Planning for Resilience in Singapore}},
  booktitle = {DSAA},
  year = {2018},
  url = {https://openreview.net/forum?id=HIWYlU9Au2U},
}
Figure