Contents

1. Introduction
2. Animating Cork
3. Day Profiles
4. Daily Flows
5. Station Performance
6. Summary
7. Final Thoughts

1 Introduction

Cork’s Coca Cola Zero Bikes scheme was launched in December 2014 by the An Rothar Nua consortium, which also manages schemes in Galway and Limerick. Lapp’s Island, the commercial core of the city, is also the centre-point of the scheme with nearly all of the stations either on the island itself or on the quays on the opposite side of the river Lee. Running across an east-west axis from Kent Station to University College Cork, its thirty-three stations form a dock-based scheme similar in functionality to that previously discussed in Dublin.

With approximately a third the quantity of stations as Dublin despite less than a fifth of the population, the residents of Ireland’s second largest city are proportionally better served with stations than its capital, at least notionally. I’m eager to see how the Cork scheme compares to Dublin, which as we’ve seen is well-utilised for commuting.

SchemeStats.bike has been collecting data on the Cork scheme since 2017, with data kindly made available by An Rothar Nua via a real-time web service. Data is polled once per minute and collected for exploration by the tools on this site. The data in this post pertaining to day profiles, journey estimates, and station performance are based on three months of data ranging from July 1st to September 30th, 2018.

2 Animating Cork

What follows is a time-lapse animation of the population of each station over a one-week period from Monday September 17th to Sunday September 23rd 2018. There are five minutes of elapsed time between frames. Note that during the early morning period the service queried by SchemeStats.bike reports zero bikes in each station, hence the apparent absence of bikes during that period of the day:

Some observations from the video are as follows:

• The scheme clearly is not as active as Dublin – we do not see the same broad swathes of empty stations emptying and filling up rapidly each day.
• There is not a very clear pattern of commuting from one area to another taking place.
• Stations seem relatively well balanced, with few being either empty or full most of the time.
• Kent Station and Fitzgerald’s park, each on opposite ends of the scheme geographically, appear to be among the most active, with large rapid swings in bikes each weekday.

3 Day Profiles

A ‘Day Profile’ as provided by this site’s analysis tools is defined as the average trend for a given metric (bikes in circulation, available bikes, etc) for a given day of the week. We can plot the mean values for a metric along with its ‘spread’ (standard deviation) and the maximum and minimum values over the course of the day.

The scheme’s stations close at night. For this reason, day profile information below only pertains to the period from 5 AM to midnight each day.

3.1 Bikes in Circulation

Bikes in Circulation is an estimated figure which is calculated on the fly by this site as data comes in. More information about how this is calculated in the About section, but it must be said that this unavoidably includes those bikes being transported on the back of lorries for redistribution. That being said, it’s still a useful gauge of roughly how many bikes are in use at a given time.

Below is a plot of the mean estimated bikes in circulation over the three-month period studied for each day of the week:

Although the data is fuzzy, no doubt offset somewhat in a couple of places by a large pickup of bikes, it’s still possible to make some reasonable observations:

• The quantities of bikes in circulation are comparatively small compared to Dublin, even when accounting for the relative size of the scheme. Concretely, a maximum of 25 bikes in circulation, as viewed in the morning, represents a little under 10% of the total, whereas in Dublin the comparative ‘utilization’ is 30% during the morning peak.
• There is a difference between the weekday and weekend trends, suggesting a significant proportion of usage is for commuting purposes.
• Weekdays have a milder version of the triple peak we observed in Dublin, albeit on a smaller scale:
  • Mornings see a steep rise in bikes in transit, with numbers starting to rise from 7:30 AM and returning to a winding down by around 9:30 AM.
  • There is a very marginal rise in the numbers of bikes in circulation around 1 PM, suggesting some small usage for running lunchtime errands etc.
  • We see an evening peak in circulation, starting at around 4.30PM and winding down gradually over the rest of the evening. As in Dublin, the evening roll-off is more gradual than the morning commute’s increase.
• Something is different about Wednesday and Friday afternoons between about 3.45 PM and 5 PM with an apparent increase in circulation around this time to about 25 bikes in circulation. More on this below.
• Saturday and Sundays by contrast exhibit a single gradual increase in circulating bikes peaking at around 3pm before slowly tapering off.

3.2 Accounting for Wednesday and Friday Afternoons

As mentioned above, there is something different about Wednesday and Friday afternoons. To account for this, I want to determine what the spread of the data is for the data in question, and to do this I’m going to plot the two days that look unusual alongside Monday for a ‘control’. In addition to the mean values as above, I’m going to also graph the following:

• The maximum and minimum values: These are represented by the vertical bars with the horizontal caps on either end showing the max and min numbers of bikes in circulation at that time of day over the 3 months.
• The standard deviation: This is a measure of the degree to which the sample data is reliably close to the mean – wide standard deviations mean that the trend is not reliable regardless of the mean. This is represented by the coloured bands around the line depicting the mean.

The graph is as follows:

Okay so it looks a little weird, but what it’s telling us is that for most of the day, Wednesdays and Fridays are typical, but something happens on at least one occasion both days to throw the figures out during the mid-afternoon, as reflected in the wider dispersion of maxima and mimina and the standard deviation. Either Wednesday and Friday just aren’t reliable days, which I can’t see a reason for, or something happened over the period studied to throw off the mean figures. I’m guessing that it was just a re-distribution took a good few bikes off the road at one time, artificially inflating the figure for bikes in circulation.

Greater credence is lent to this by looking at a slightly different period. Here is the graph for Wednesdays and Fridays for August only:

This late-afternoon peak disappears for August. The morning aberration aside (to which I assign the same cause), I believe this data to be indicative of a normal trend for these days, that is, they fall in to the standard weekday pattern.

3.3 Empty & Full Stations

Let’s turn our analysis over to the day profiles for the number of empty and full stations. Firstly, the numbers of empty stations looks as follows:

And finally switching our attention to the number of full stations we observe the following:

Some observations about these figures:

• These metrics compare favourably with Dublin, which sees significant rises in the numbers of empty and full stations during the commuting period.
• The average for the number of full stations is less than one: This apparent blip in the statistics simply implies that there is usually a maximum of one full station even at the busiest time of day.
• Empty stations are very steady at about three. This represents one tenth of all stations.
• Commuting hours do not make any significant impact on the data – except that there appears to be one full station during the mornings and evenings, about half of the time.
• Different week days show no notable difference in the quantity of empty stations.

4 Daily Flows

As previously noted, the scheme exhibits a moderate peak of bikes in circulation in the morning and evening. Let’s attempt to determine where bikes are going during these two commuting periods.

4.1 The Morning Commute

What follows is the average net change in the number of bikes in each station over the three month period under study during the morning commuting period of 07:30 to 09:30 AM, Monday to Friday. Red stations are gaining bikes and blue stations are losing them:

The top gainers of bikes are as follows:

And the top stations losing bikes are below:

4.2 The Evening Commute

For the sake of this calculation I’m calling the evening commuting period 16:30 to 19:30, as the evening wind-down takes longer than the morning rise. This graph yields the following – again, red stations gaining bikes, blue losing them:

The top gainers of bikes are as follows:

And the top stations losing bikes are below:

5 Station Performance

Finally, let’s try and get a view of overall station performance over the three-month period. I’m going to look at the percentage of time that each station spends empty or full in aggregate, from 06:00 each morning to midnight, for the full week.

Note that in the statistics below two stations are discounted. These are Brian Boru Bridge, which was closed for the period, and College of Commerce, which was closed for most of the period.

5.1 Percentage of Time Empty

Let’s plot this data on the map – those stations in darkest red are to be discounted:

And now for the bottom 5 performing stations:

5.2 Percentage of Time Full

Finally re-drawing the map, this time for percentage of time full:

And once again, the bottom 5 performers:

5.3 Station Performance Observations

In line with the earlier figures for day profiles of the number of empty and full stations, we see that the Cork scheme exhibits comparatively good statistics for time full and empty. One stands a very good chance of obtaining a bike most of the day in most of the stations.

6 Summary

• Typical daily numbers of peak estimated bikes in circulation is around 25-30. This accounts for less than 10% of the available bikes at peak times, compared to approximately 30% for the capital.
• Weekdays follow a trend composed of a peak in usage in the morning and a similar peak in the evening, commensurate with commuting times.
• Friday evenings are quieter than other weekday evenings. This matches the trend seen in Dublin, indicate of people perhaps staying in town to socialize or what have you.
• The number of empty and full stations remains relatively static throughout the day, with commuting hours not impacting highly on this metric. This suggest the scheme is operating within its capacity.
• Typically, there are approximately 2-3 stations empty at any point in time, and either zero or one stations full. Taking the combined figure of 4 stations empty or full typically, representing 12% of stations, this compares favourably with Dublin for which the comparative value is around 35-45% at the time of day when the scheme is most stretched.

7 Final Thoughts

Mostly in the blog I intend focussing more on the data than trying to explain why a scheme behaves the way it does. This lets me focus on my strengths while hopefully facilitating others in having an informed conversation about the future evolution of our bike schemes. However, I feel compelled to offer a couple of opinions on what I’ve seen in the analysis above:

• Although Cork’s scheme is less utilized than Dublin, this is not a solid basis on which to conclude expansion is not merited. With Dublin as an example of a more highly utilised scheme, I would advocate focusing on the differences between the two cities’ schemes and trying to address them in order to obtain a greater uptake.
• Obvious differences come to mind: Cork’s smaller and it’s somewhat hillier than Dublin on the northern side of the city. However, the most important difference I can see is in terms of station distribution. Cork has almost all its stations in the commercial core, whereas in Dublin there are significant quantities of bikes in both the core and less commercial areas. This facilitates the morning commute, with bikes available both near people’s houses and their places of work. I would recommend serious consideration be given to putting stations in future in nearby residential locations and seeing if that results in an improved uptake.

If you’ve read this far and want to see further posts on other schemes both in Ireland and further afield, please consider sharing this using the social media links provided. This website and blog are something I do in my spare time, and what spurs me forward is knowing that my work is being read. Thanks for your support!