Bike sharing is spreading across the nation. One of the highest profile programs has been in New York City, and while there has been griping about various aspects of the Citi Bike program, it has by some measures been successful. According to the NY Times, the system has attracted over 70,000 annual members and has handled over more than 42,000 trips on a peak day (The Balancing Act That Bike-Share Riders Just Watch, Aug 14).
One of the challenges that has come up with the system is how to balance the supply and demand of bikes between stations. If you want to understand the nature of the problem, check out this (way cool) graphic over at the New Yorker, which plots the inventory of bikes at different stations over a month. There is generally a flow of bikes to lower Manhattan has people head to work. As the accompanying article puts it (A Month of Citi Bike, Jul 26),
It’s possible that the Citi Bike system may be too successful for its own good. As the program becomes a more popular method of commuting, the workday leaves some areas bereft of bikes, making it more difficult for those with reverse or off-hour commutes to participate in the program. Citi Bike crews do redistribute the bikes, but the empty areas on the map show how challenging it is to balance their availability across the stations.
That gets us back to the Times article on just how the rebalancing of bikes takes place.
The city relies primarily on a fleet of box trucks, hauling bikes by the dozens when cyclists cannot find a way to share effectively.
Directions come from Sunset Park, Brooklyn, delivered by dispatchers with an interactive station map and an evolving sense of traffic patterns. Crews travel in pairs. Loading a truck can take 45 minutes, to say nothing of replenishing a station nearby.
Cyclists tend to work more quickly.
“If we bring 37 bikes,” one worker said beside an Eighth Avenue station last week, “by the time we’re gone there’s two left.”
To combat this problem, the program has leased three spaces as staging areas for bikes that can be brought to nearby stations that tend to empty quickly: an office building beside Penn Station; Pier 40, along the Hudson River; and a lot at Delancey Street on the Lower East Side. The plan follows a model tested at other international hubs, like Waterloo Station in London.
As we have written about before, the need to rebalance capacity is a common problem in bike sharing systems. It is not unlike a typical rush-hour commute. As long as there are residential areas and commercial areas, having people with similar work times will lead to traffic jams in one direction in the morning and in the other in the evening. With bike sharing, distinct residential and commercial areas lead to bikes draining from one area to the other. In principle, they should reverse over the whole day but that may not happen. I may be disinclined to leave my car in my employer’s lot over night but am less invested in a shared bike — particularly when the weather turns bad.
But why is balancing bikes so hard? There are several issues here. One is that bikes are not used on a fixed schedule. Citi Bike does not always have the exact same number of people using the bikes at Penn Station every morning at 7:30AM. Now it can build a forecast but that is going to be complicated for several reasons. First, it’s a growing service. As more people sign up, demand patterns can shift. Second, there is a data censoring issue. Citi Bike only sees when someone rents a bike, not when they try to rent a bike. If Penn Station runs out of bikes at 7:45AM, they cannot see how many potential users arrive at 7:50. To make matters worse, demand will be responsive to supply. If bikes are always available at 7:45, signing up will be more attractive to more commuters. However, improved availability at one location may just result in stealing demand form another. That is, some of the demand on, say, 29th St may be from commuters who would really prefer to grab their bike at Penn Station but hoof over to 29th St when they can’t get a bike at the station. Increasing availability at Penn Station may cause demand at 29th St to drop.
A final point. The use of bike staging areas is an interesting twist. If demand is highly focused on a few locations, this make some sense since bringing out more bikes from the storage area in the morning should be more cost-effective than expanding storage capacity on the sidewalk all day. However, it does have a paradoxical side since bikes in the staging area are not directly available to customers. Effectively, they are improving service at key stations by making some bikes temporarily unavailable.