[Paper Review] Demand-driven design of bicycle infrastructure networks for improved urban bikeability

Christoph Steinacker, David-Maximilian Storch+2 more|arXiv (Cornell University)|Jun 14, 2022

Transportation Planning and Optimization34 references29 citations

TL;DR

This paper proposes a demand-driven framework for designing efficient bicycle infrastructure networks by iteratively removing bike paths from a fully equipped network based on cyclists' route choices and safety preferences. The method dynamically adapts to demand distribution, yielding highly bikeable networks with minimal path removals—demonstrating that route choice modeling significantly improves network efficiency compared to static percolation approaches.

ABSTRACT

Cycling is a crucial part of sustainable urban transportation. Promoting cycling critically relies on a sufficiently developed bicycle infrastructure. However, designing efficient bike path networks constitutes a complex problem that requires balancing multiple constraints while still supporting all cycling demand. Here, we propose a framework to create families of efficient bike path networks by explicitly taking into account the demand distribution and cyclists' route choices based on safety preferences. By reversing the network formation process and iteratively removing bike paths from an initially complete bike path network and continually updating cyclists' route choices, we create a sequence of networks that is always adapted to the current cycling demand. We illustrate the applicability of this demand-driven planning scheme for two cities. A comparison of the resulting bike path networks with those created for homogenized demand enables us to quantify the importance of the demand distribution for network planning. The proposed framework may thus enable quantitative evaluation of the structure of current and planned bike path networks and support the demand-driven design of efficient infrastructures.

Motivation & Objective

• To address the challenge of designing cost-effective, safe, and convenient bicycle infrastructure networks that balance budget, connectivity, and user demand.
• To overcome limitations of static network design by incorporating realistic cyclist route choice behavior based on safety and convenience preferences.
• To quantify the impact of demand distribution on optimal bike path network structure and performance.
• To develop a flexible, computationally efficient framework for evaluating and designing bike path networks that adapt to real-world mobility patterns.

Proposed method

• Starts from a complete bike path network where every street segment has a bike path, then iteratively removes paths based on their impact on cyclists' perceived travel distance.
• Uses a cyclist preference graph where edge weights combine physical distance and safety/convenience penalties (e.g., higher for high-traffic streets without bike paths).
• Models route choice as shortest path computation on the preference graph, minimizing perceived distance while favoring safer, separated routes.
• At each step, removes the street segment whose removal causes the smallest increase in average perceived travel distance across all trips.
• Recomputes shortest paths after each removal to reflect updated network conditions and dynamic route adaptation.
• Applies the framework to real-world cities (Hamburg, Dresden) using empirical bike-sharing data and compares results to homogeneous demand and static percolation methods.

Experimental results

Research questions

• RQ1 How does incorporating realistic cyclist route choice behavior—driven by safety and convenience—improve the efficiency of bike path network design?
• RQ2 To what extent does the spatial distribution of cycling demand influence the structure and performance of optimal bike path networks?
• RQ3 Can a dynamic, demand-adaptive network generation framework outperform static percolation methods that rely on fixed importance rankings?
• RQ4 How much improvement in bikeability can be achieved with minimal infrastructure investment when route choice behavior is modeled dynamically?
• RQ5 What is the trade-off between cost (number of bike paths) and performance (bicycle accessibility and safety) in different urban network topologies?

Key findings

• The dynamic framework generates bike path networks with significantly higher bikeability than static percolation methods, especially when demand is heterogeneous.
• Even with only 20–30% of street segments equipped with bike paths, the dynamic approach maintains high bikeability, indicating strong resilience and efficiency.
• The framework reduces the fraction of cycling trips on non-bikeable streets more effectively than static methods, particularly when safety penalties are included in the model.
• For Hamburg, the full simulation took ~14 minutes with empirical demand and ~15 minutes with homogeneous demand, while Dresden required ~29 minutes under homogeneous demand due to increased trip diversity.
• The dynamic approach outperforms penalty-weighted static percolation in both bikeability and perceived travel distance, with relative improvements visible in the ratio of dynamic to static performance.
• The framework is robust across cities with varying network topologies and demand patterns, showing consistent performance gains regardless of city size or street classification consistency.