Establishing the validity of cycle path capacity assumptions in the Highway Capacity Manual

ABSTRACT

Cycle mode share increase is widely desired, but highway design practitioners lack the numerical tools to deliver infrastructure, instead relying on design standards and intuition, with little literature basis. As a case in point, the US Highway Capacity Manual (which is well used internationally) has developed levels of service for cycle infrastructure that are, at their core, based on an assumption of noninteraction between multiple cyclists. This paper uses a modified implementation of the Social Force Model to test the validity of this assumption. Necessary changes such as the consideration of acceleration characteristics and minimum maintainable speed are included. The resulting model produces valid outcomes in keeping with established traffic flow properties, reflecting three-phase traffic flow theory and the ability for the stochastic elements in traffic flow to cause flow breakdown. The developed simulation indicates that there is a fundamental difference in outcome if cyclists are assumed to have a fixed speed versus one they can change given their surroundings. This difference in outcomes is found to exist within the range of literature design flow capacities for bicycle infrastructure and also yields emergent outcomes that align closely with those known behaviors of highway vehicles, which intuitively transfer to cyclists. These findings reinforce the standing need for large-scale empirical studies to determine the basic numerical and behavioral parameters for cyclists, upon which all design ultimately rests.

KEYWORDS:

• Agent-based modelling
• bicycles
• cycle infrastructure
• simulation
• social force model

Acknowledgments

The authors acknowledge the use of the IRIDIS High Performance Computing Facility (and its associated support services at the University of Southampton), upon which the presented simulations were run.

Funding

This work was financially supported by an EPSRC Doctoral Training Centre grant (EP/G03690X/1; http://www.epsrc.ac.uk) through the University of Southampton Institute for Complex Systems Simulation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Notes

¹ Rouphail et al. (1998) note that the numbers considered are not absolute capacity measures, but the interpretation is the equivalent of considering a highway maximum operational flow to not be the measure of moving vehicle capacity, in contrast to its capacity as a parking lot of the same dimension.

² While a theoretical capacity of 3300 bicycles per hour for a cycle path of 1m width (and 4700 bicycles/h for a 1.8 m path) is noted at one point in CROW (2007), the parts of that document focus on the infrastructure revolve around the comfort values.

³ I.e. that which would be calculated at a 0.0 m from the location of the bicycle. The exact number is a function of the parameters chosen.

⁴ E.g., the rule-based lane selection algorithm in PTV Group (2009), detailed in Barcelo (2010).

⁵ Testing of the model indicated that beyond the length used in this paper for this particular scenario (parallel-sided path), output parameters were substantially unchanged; however, the run-time of the simulation would rapidly become intractable.