Triple Peaks and Clear Commutes: A Look at Dublin's Bike Sharing Scheme July 24, 2017 by Philip Lowney

Launched in 2009, the Dublin bike sharing scheme, latterly known as Coca Cola Zero Dublin Bikes, and soon to be known as Just Eat Dublin Bikes, has been tremendously successful. To get the wheels rolling (get it?) on this new site, I'm exploring the scheme's behaviour during one week in July, from the 17th to the 23rd. To get a better understanding of what happens at the weekend, I've also included some other Saturdays and Sundays from July.

Please note that the images and statistics provided below are a sample of those provided through this site's own dashboards and maps. You can browse real-time and historical statistics for Dublin and other schemes for yourself by selecting it from the top menu. Comments and suggestions are welcome!

The Weekday Triple of Cycling Activity

Below is a snapshot of the estimated number of bikes in transit each weekday (for an explanation of the estimate, see the Notes at the bottom):

What stands out is the reliable triple-peak of activity every day:

  • A morning and evening rush, peaking around 400-500 bikes,
  • A lunchtime spike shortly after 1 PM, with about two-thirds the level of activity seen in the commuting hours,
  • Finally an evening rush of similar magnitude to the morning, tapering off more slowly than the morning began.

The North-West/South-East Axis of Weekday Commuting

In order to examine commuting activity, I've put together the following illustration of change in the allocation of bikes & free slots during the two commuting periods. For each weekday, two net-change heat-maps are provided, one from 7am to 9am in the morning, and another at 4pm and 6pm in the evening. Each station is colour-coded indicating net change: Blue means stations are losing bikes, and red means they're gaining bikes.

  7 AM - 9 AM 4 PM - 6 PM
Monday 17th
Tuesday
Wednesday
Thursday
Friday 21st

A few things stand out from this:

  • Each weekday, bikes flow out of stations in the North and West of the city in the morning towards a core in the South-East quadrant of the city centre, and in the opposite direction in the evening.
  • The city centre sees less net change than the outer areas - this could be a due to a more even balance between arrivals and departures.
  • In the evening, there is less of a rush to get home towards the East of the city than there was to get to work in the morning (this area is less red than it was blue earlier in the day, save for the commuting epicentre of Heuston station). This corresponds with the first graph here too, where we saw a slower falloff in bikes in transit in the evening than we did a gain in the morning.

The overall trend is a minor surprise for me as a person living in Dublin for years: I always think of the city centre as an area centred on perhaps the Ha'penny bridge or College Green. This may be the case, but from the perspective of cycling commutes it is more centred on Merrion Square or Lower Baggot Street. Perhaps it reflects the greater concentration of office space in this part of the city, however that's just a hunch. You can browse these stats for yourself here.

Weekday Wanderings at One

What kind of trends do we see at lunchtime? To help answer, I've created another set of net-change maps over the course of five weekdays. For each weekday, two maps are provided: one in the run-up to a lunchtime peak, from midday to 1:15pm (approximate peak of lunctime activity), and another from 1:15 PM to 2:30 PM.The objective here is to bisect the activity at lunchtime - one where the number of bikes is increasing and another where it's decreasing. This yields the following (again, red is gaining bikes, blue is losing them):

  Noon - 13:15 13:15 - 14:30
Monday 17th
Tuesday
Wednesday
Thursday
Friday 21st

A few things stand out:

  • There seems to still be an element of the morning/evening flow - people leaving the D2 area at the start of lunch and returning at the end. However, it's less pronounced than before.
  • Activity is more dispersed throughout the city,
  • The western section of the scheme has the least net change, implying less activity

Put together, these makes sense: At lunch, few people probably cycle all the way back home. However, many of us might use lunch time to do an errand, requiring us to go into the city centre. We might also be heading to a nearby shop or café, which are widely distributed. So lunchtime is a significant time of activity with its own behaviours, disbursed accross the office-centric and shopping-centric areas of the city.

A Relaxed Weekend

What follows is another illustration of estimated bikes in transit, from the weekends of the 15th, 22nd and 29th of July. I've put together two trend graphs, each comparing Saturdays and Sundays respectively:

Clearly, activity at the weekend is different to weekdays, specifically:

  • There is about half the level of maximum bikes in circulation than the weekends
  • There is really just one gradual rise and fall in activity

Conclusions

  • Dublin's bike scheme is used extensively to get people to and from work
  • The central axis of commuting activity runs from the north and west of the city towards the south-east quadrent
  • Weekday activity is composed of a reliable triad featuring the morning and evening commutes and a significant tertiary peak for lunch
  • For the weekdays studied, estimated peaks of bikes in circulation during the week is 400 to 500.
  • Weekend activity is composed of a single gradual increase in activity from mid-morning through to the evening.
  • During the weekends studied, estimated peaks of bikes in circulation is 150-300, with Saturdays being busier than Sundays.

Notes

3rd Party Data Source

Real-time bike station information is provided by JC Decaux, the company which provides the infrastructure for Dublin Bikes. It is presented through an API (Application Programming Interface) which can be found here. This service provides the location information for each station, along with realtime information on the number of bikes and free slots in a station. It does not provide a 'bikes in transit' figure, which I estimate instead.

Collection Methodology

The data presented above is a result of continually polling the service provided by JC Decaux, and collecting the responses into a database to create a view of activity over time. The service is polled once per minute.

Estimating Bikes in Transit

Each time the scheme is polled for real-time information, SchemeStats generates a new estimate for the number of bikes in circulation. For Dublin, this is:

Estimated Bikes in Circulation = [Maximum Number of Available Bikes in Circulation in Previous 24 Hours] - [Sum of bikes available in all stands now]

In other words, assuming that there was a few minutes during the last 24 hour period whereby nobody was cycling, and that that period will thus have the maximum number of available bikes, we subtract the currently available bikes in the latest poll from that figure to estimate the bikes in circulation.

This is an imperfect figure and does not account for at least the following unknown variables:

  • Bikes removed from or added back into circulation due to maintenance
  • Bikes taken onto the back of the truck for redistribution