On the capacity of bikeways

May 20, 2016

it has been lot of research carried out on the capacity of roads, transit or pedestrians infrastructures, with results proven empirically. Such don’t really exist for the cycle tracks, but as the success of the London’s “Super cycling highways” shown, it will become a significant matter

NACTO, in a recent publication [8], estimates the capacity of a 2 way bike lane (3 to 3.50m) at 7,500 bike/hour. This number seems to be derived of the Highway Capcity Manual citing very dusty publications [1]. this post argues that the capacity of a bikeway is more in the 1,500 cyclist/hr/lane (where a lane is 1.20m to 1.5m wide)

A short Litterature review

Most of the papers trying to estimate a bike lane’s throughput tend to rely either on mathematical models, experiments or a blend of both:

An example of experiment

An example of experiment leaded by [2] to determinate a bike lane throughput. The experiment above has allowed [2] to conclude that the capacity of a bikelane should be in the 2,500 bike/hr range

The problem of such approaches is they are not (yet) validated by practice (…and in some case, the experiments seem to be more representative of a velodrome typology than an urban bike lane). They also tend to provide a great range of result: One literature study [3] found a capacity of 1,500-5,000 cyclists per hour and traveling speed around 12-20 km/h. Another literature study [4] found a capacity of 2,000-10,000 cyclists per hour for a 2.5 m wide cycle track. It is also important to notice that all these numbers concern an uninterrupted bike lane (e.g bike lane with no intersection).

However, [5] ( as cited by [6] ) reported that the theoretical and practical capacities of a Chinese bicycle lane are about 2000 bicycles/h/ln and 1280 bicycles/h/ln, respectively… That is also in line with an empirical result presented by [7] concerning the Denmark:

Bike lane capacity as a function of its width, as according to [7]

Bike lane capacity as a function of its width, as according to [7]

How much lanes of cyclists fit in 3.5m width bikeway
[7] tend to answer to it:

If a 1.8 to 2 m wide bikeway fit 2 lanes of cyclists, any additional lane could require a 1.20m additional width (notice that the cyclists could have a tendency to ride in quincunx to increase their available lateral room). That is the reason for the suggested significant increase in capacity as soon as an unidrectional bike lane reach 3m in the graph above. (There is also some reasons to believe that a 4 meters bidirectioanl bike lane is not width enough to enable 2 lanes of cyclists in each direction in a sustainable way: see video below)

Validation of the numbers in practice

Up to recently, it was basically no opportunity to validate a bikelane model capacity in real life. China of course has wide and busy bike lanes, but they has never presented a typology directly applicable to Europe or America, be by their different geometry or by the type of vehicle: many trikes (up to recently), and nowadays those bikelane tend to be overwhelmed by sccoter (electric or not) (- 70% in Hangzhou as measured by [6]. Bikes also tend to move much more slowly). However, with the recent opening of the London cycle highway this things could change:

This video represent the cycling traffic on the London’s Blackfriars bridge: The incoming lanes presents the symptom of a bikeway reaching capacity (bike moving slowly, at speed apparently just enough to maintain balance, and the rare occasional take over use the opposite lane)

What is the effective throughput of Black friars Bridge bike lane?

Sure enough, some cycling supporters quickly raised the question while some other provided some numbers. Here are ours

  • there are 37 incoming cyclists crossing a a line (represented by the bottom of the video)
  • there are 24 outgoing cyclists crossing the same line

that represents an “instant” throughput of 11,000 cyclists/hour on a period of 20s, or ~3,300 cyclists/hour/ln in the busiest direction (or the equivalent of 10,000 cyclists/hour per car width lane).

First issue, Instant throughput ≠ Throughput

Traffic tends to not flow homegeneously (move in wave, aka “stop and go” traffic), and a measure on 20s can’t be directly scaled into a more generalized throughput. [7] faced with similar issues applied a correction factor based on freeway traffic observation (by comparing maximum observed traffic volumes in 15 minute intervals and maximum volumes in 10 and 20 sec intervals on freeways): this correction factor is estimated to be 0.63

The estimated maximum throughput per lane observed on the Blackfriar Bridge video above become closer to 2000bile/hr/lane. Still a sustantial number (but already significantly less than the number touted by NACTO), a number also in agreement with theorical number exposed by


Second issue, Uninterrupted bike lane throughput ≠ Interrupted bike lane Throughput

The video below better illustrates the later issue:

There are right and left turning cyclists, which fatally indhers the capacity of a bike lane…and there are signal controlled intersections (a necessity as illustrated by the difficulty of the cyclist, waiting at an intersection, to integrate itself in the main bike lane). As for motor traffic, all these tend to halve the real capacity of a typical interrupted urban bike lane vs an hypothetcal uninterrupted one. So that the real capacity we could measure here tend to be more in the ~1,000 to ~1,700 bike/hr/ln (according to if we apply a correcting factor or not). Those numbers are also in line with results from the field carried by [5] and [7]

Why all that matters?
Notice the 2 double deckers in one lane, seen in less than 30s: Could we conclude that the transit capacity of this lane is 24,000people/hour?

It matters since wrong numbers could lead to wrong decisions on the allocation of the street space, but also on the “right sizing” of a bike infrastructure.
The matter is of importance, essentially in regard of transit:

There is no question that a bike lane can achieve a very high throughput but does a 3.5 meter bike lane can carry as much as people than a tram or a BRT?

According to NACTO, a bus lane (BRT) can carry 4,000 people (8,000 if train): Those number thought slightly optimistic [9] are fairly realistic, and can be verified empirically…Here we infer that a bike lane of similar width (3 to 3.5m) can be competitive with a buslane but not a trams transit system, in term of throughput.

Furthermore we have not touched the whole notion of Level of Service: the mentioned capacity for Transit are design capacity, that is capacity allowing the transit system at its design (optimal) speed. Reaching the capacity of the bike lane as measured on the BlackFriars Bridge tend to infers a degradation of the speed for cyclists (at least by queuing at traffic light). We also didn’t touch the bike parking issue

[1] the third edition of the Highway capacity Manual suggests a capacity range of 1700-2500 bicycle/hr/ln where each lane is 3 to 4 feet. Those numbers are inferred of previous publication (“bikeway planning design and guideline” institute of traffic and transportation engineering University of California at Los Angeles 1972, “Geometric Design” by W. King, C., and Harkens, in Transportation and Traffic Engineering Handbook, Institute of Transportation Engineers, 1972). the 7,500 number touted by the NACTO correspond to the highest range of it considering that a 3.5m wide can feet 3 lanes of bike

[2] “Experimental feature of bicycle flow and its modeling” Jiang R. Hu M. , Wu Q. and Song W.

[3] “Operational Analysis of Uninterrupted Bicycle Facilities”, Allen, P., Rouphail, N., Hummer, J. E., and Milazzo, J. S. Transportation Research Record, 1636 29-36, 1998

[4] “Bicycle Traffic Flow Characteristics: Experimental Re-sults and Comparisons”, Navin, F. P. D. ITE Journal, 64 31-36, 1994

[5] “Capacity and level of service for urban bicycle path in China”, Feng Li, China MUnicpal Engineering magazine 1995, 71: 11-14

[6] “Estimating Capacity of Bicycle Path on Urban Roads in Hangzhou, China”, Zhou D., Xu, C. and Wang D. , Transportation Research Board 94th Annual Meeting, 1995.

[7] Analysis of Bicycle Traffic on One-way Bicycle Tracks of Different Width, Thomas Skallebæk Buch and Poul Greibe, Trafitec,

[8] Transit Street Design Guide, NACTO 2016

[9] The French CERTU suggest a design capacity of 3,500 for aBRT, 6,000 fro a tram. Translink considered a design capacity of 3,000 for a BRT, and 7,000 for a LRT on Broadway.


2 Responses to “On the capacity of bikeways”

  1. Jeff Leigh Says:

    The US DOT Federal Highways Administration developed a tool for calculating the Level of Service of a multi-use path, and this has been applied to segregated paths as well. It was used in this mode to calculate required lane widths for the new Active Bridge relating to the viaducts removal. The following may provide some more detail on this question of capacity.


    • Voony Says:

      Thanks for the link,

      I am not convinced the formula for mulit-use path apply per sei for segregated paths as well:

      in the spreadsheet, it looks a relatively low number of bike (~200 p hr on 8ft) is sufficient to get a level F of service (disregarding how much pedestrian use it): that is probably true from a pedestrian viewpoint, but it is probable that a segregated path of the same width can accomodate much more cyclists with a much more acceptable level of service.

      Similarly, Point grey, road width can comfortably accomodate lot of bike, but the level of service could be quickly compromised by much more modest increase in motor traffic

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: