Long-Term Trends on Dublinbikes April 11, 2022 by Philip Lowney


  • Overall usage is down since the beginning of the pandemic and has not recovered.
  • Weekend usage has largely held up, maintaining its pre-covid profile.
  • Peak weekday usage in 2020/2021 was down up to 75% on 2018/2019.
  • Average behaviour in 2021 was largely the same as 2020.
  • It is a better time to use the scheme than ever, with far less empty/full stations.
  • Resources available in the scheme (bikes/stations) have changed little over the last four years.


SchemeStats.bike has been collecting real-time data from the Dublinbikes for almost five years, with previous analyses here describing a lively scheme very much in demand. By contrast, shortly after the pandemic began, we observed a predictable slump in weekday usage. Today I will examine further how the scheme changed over the two years of the pandemic, and to what extent there has been a reversion to pre-covid behaviour.

As a quick recap, that pre-covid norm was defined by:

  • Very high utilisation during the morning and evening commuting periods.
  • Stress on the quantity of empty and full stations towards the end of commuting windows.
  • A moderate rise in activity during weekday lunchtimes.
  • A focus of commuting destinations in the south-east quadrant of the scheme, encompassing the office-dense part of the scheme in Dublin 2.
  • Moderate usage at the weekends, with a single gradual increase in activity peaking mid-afternoon.
  • Weekend usage was far more disparate than weekday usage.

The data presented here has been gathered from June 2017 to February 2022 inclusive. While SchemeStats aims for 100% coverage, there were some occasions where we missed this goal. A coverage report is included at the appendix to this post.


Trend 1: Resources have Remained Stable

The resources available to the scheme consist of stations and bikes. Looking back, we can see that the size of the fleet has remained roughly the same at around 1400 bikes in use. The number of open stations increased by 10% or so in 2018 but otherwise have been stable too, currently at 110.

Scheme Resources in Bikes and Stations

Figure 1 Scheme Resources Since 2017

There was certainly a slight withdrawal of bikes from the scheme in April 2020, but this was likely an understandable response to the stress the pandemic put on maintenance etc.

Trend 2: Aggregate Usage is Down

SchemeStats does not have access to the quantity of rentals per day, however it is possible to convert the estimated figures we produce for real-time estimated bikes in transit into a usage metric. We do this by integrating the figures for bikes in transit over time to generate a ‘minutes cycled’ figure. This is not the same as the number of journeys travelled, but it is a valid way of viewing the usage of the scheme – that is, how many millions of minutes did people spend cycling on it per month.

This exercise, when performed both for total usage and weekend usage, yields the following chart. While the month-by-month data is variable, we can clearly see that from March 2020 onwards there was a very significant downturn in usage which has not recovered. Interestingly, when weekends are looked at on their own, there isn’t the same clear decline in usage, meaning most of the drop-off has been accounted for by an absence of commuting activity.

Aggregate Scheme Usage

Figure 2 Aggregate Scheme Usage over Time

Readers will note of course the discontinuity in the chart above in late 2018 and early 2019. There were some issues with our polling machinery over this time, and our real-time coverage of the scheme’s activity was below an 90%. As aggregate usage figures depend on high-resolution real time data, we simply omitted these months from the chart.

Trend 3: Weekdays are Less Intense

SchemeStats has a ‘day profiles’ tool which allows us to view average values for a given metric, for any day/days of the week, over arbitrary periods. Below, we’ve used that tool to look at the average quantity of estimated bikes in transit on weekdays for each of the four complete years data has been collected.

Average Weekday Profiles for Bikes in Transit

Figure 3 Average Weekday Bikes in Transit 2017-2021

There are some immediate observations one can make from this:

  • The average data for 2018 and 2019 are remarkably similar, both exhibiting the same triple-peak of activity during the morning, lunchtime and evening commuting periods. In both years too, we see that the evening rush is less intense but a little further spread out.
  • The profiles for 2020 and 2021 are remarkably like one another, but dramatically different to the pre-pandemic years. In both years we see a severe drop-off on commuting period usage, explaining the aggregate reduction in traffic shown above.
  • 2021, being almost identical to 2020, illustrates that despite being well into the pandemic, a recovery in commuting habits had not been manifested.

The chart above is based on year-long average data for each trendline, so it could be the case that the average is hiding a recovery. To better explore this, we can obtain a different perspective by simply getting the peak weekday usage every day and averaging that value per month. This will help show how variable or otherwise the data is, and we get the following:

Average Weekday Peak Usage

Figure 4 Peak Weekday Usage, Averaged Per Month

We can conclude from Figure 4 that the profiles in Figure 3 were informed by a stable decline in usage during the week, with peak totals of bikes in transit at once declining by up to 75%.

Trend 4: Weekends Have Held, with Some Extraordinary Days

By contrast with the trend for weekday usage, weekend usage over the entire four year period has remained remarkably stable, even including extremely busy weekend usage in the last two years. Here are the same two types of charts applied to weekend data:

Average Weekend Profiles (facebook and Twitter Blog Post Image)

Figure 5 Average Weekend Bikes in Transit 2017-2021

Average Weekend Peak Usage (facebook and Twitter Blog Post Image)

Figure 6 Peak Weekend Usage, Averaged Per Month

Trend 5: It is Now Easier to Find and Park a Bike Than Ever

The preceding four trends have described a scheme for which the resources have not changed but the demand has declined significantly. On the other hand, the fact that the scheme is less taxed than before means that it is easier to get both bikes and parking spots. To illustrate this, we graphed the peak weekday quantity of empty and full stations for the morning and evening commuting periods, averaged per month, over the whole period the scheme has been in use.

Average Weekday Peak Empty and Full Stations (facebook and Twitter Blog Post Image)

Figure 7 Peak Weekday Full & Empty Stations, Averaged Per Month

As one can see, we had a situation before the pandemic where the combined quantity of empty or full stations in the morning commute reached about 55 by the end of the commute – half the total stations. The evening commute too saw roughly 40 stations empty or full at peak times. This was explained before by the fact that usage in the evenings was somewhat less intense than the mornings.

From March 2020 onward however, the numbers of empty and full stations declined significantly, the morning figure improving by around 75%. Keeping in mind these are peak daily figures, it means that since Covid began one is very unlikely to encounter either empty or full stations most of the day.

Animating Change

To better understand how the fall off in activity looks day to day, it might be useful to plot a time-lapse of the scheme in use over a representative period in each of the years under study. What follows is an animation of the Live Map screen from SchemeStats, rendered in 3D mode, illustrating a week’s worth of activity on the second week in June for 2018-2022. This week was chosen because it is mid-year and there was no public holiday. Cylinders in the video represent stations, with their heights proportional to the number of bikes docked there. Stations turn red when they become empty or full.

The time-lapse illustrates in term short-term what the analyses above indicated were long-term trends – that is:

  • 2018 and 2019 are clearly busier than 2020 and 2021.
  • The axis of commuting from the northwest to southeast remains in the pandemic years, but the quantity of trips taken at peak hours has clearly reduced.
  • Far fewer stations are becoming empty or full during the pandemic.

Final Thoughts

The Dublinbikes scheme pre-pandemic was an integral part of the commuting infrastructure in Dublin. Since the pandemic began, it has seen greatly reduced demand for use during the week and has seen sustained demand for weekend trips. Despite the pressures the pandemic brought to bear on all infrastructure, stations remained open and online throughout the pandemic and the bicycle fleet remained constant. The scheme continues to exhibit a weekday ‘triple peak’ of usage at morning, lunch and evening periods, albeit at significantly reduced levels compared to pre-pandemic norms. At time of writing, reduced demand means that there has never been a better time to be an active user of the scheme, as the likelihood of finding a full or empty station has declined significantly.

As we emerge into a different phase of the pandemic, it will be interesting to see to what extent the scheme recovers. A couple of factors may be working against it. First, given the importance of commutes in the usage share before Covid, partial working-from-home arrangements could result in a permanent reduction in demand. Second, recent years have seen the arrival of dock-less alternatives which might be eating into the market.

However, it should also be said that the scheme has made itself part of the fabric of daily life for many Dubliners, and prior to the pandemic showed a popularity and resilience that many wouldn’t have expected. As sustainable, low-carbon options become ever more part of the transport discussion, 2022 will hopefully see a significant return to form for a scheme that’s served the city well.

A Note from the Author

I really hope you find the tools and blog posts provided here useful. This website is a hobby I undertake in my spare time. If you have enjoyed using this site, please consider sharing it on a platform of your choice using the social buttons provided. Knowing this project is useful to people spurs me on to provide further functionality and content. If you have any recommendations, please don’t hesitate to get in touch via the feedback link above.


Appendix - Coverage

The following table describes the percentage of time covered to a resolution of at least one poll every 10 minutes over the period examined by this post.

Month Coverage
2017-06 99.0625
2017-07 99.8297
2017-08 88.4162
2017-09 99.6782
2017-10 99.4758
2017-11 99.9699
2017-12 100
2018-01 98.1519
2018-02 99.7892
2018-03 98.7119
2018-04 97.2315
2018-05 95.2643
2018-06 97.0463
2018-07 96.9982
2018-08 99.6841
2018-09 97.8519
2018-10 99.9238
2018-11 87.669
2018-12 0
2019-01 43.6873
2019-02 99.9678
2019-03 99.9216
2019-04 14.3472
2019-05 71.8705
2019-06 99.9097
2019-07 97.3947
2019-08 100
2019-09 97.3773
2019-10 93.1295
2019-11 99.8171
2019-12 99.9194
2020-01 100
2020-02 99.2816
2020-03 98.3221
2020-04 100
2020-05 100
2020-06 99.5787
2020-07 100
2020-08 100
2020-09 99.7222
2020-10 96.8952
2020-11 99.9745
2020-12 99.1823
2021-01 99.5453
2021-02 99.8884
2021-03 97.9682
2021-04 99.8958
2021-05 94.4489
2021-06 99.8472
2021-07 100
2021-08 99.888
2021-09 100
2021-10 99.9597
2021-11 99.7801
2021-12 100
2022-01 99.6416
2022-02 99.4345
2022-03 99.8723