Pittsburgh's Bus Bunching Bad News Bears

They say bad news comes in threes. Sometimes, too, so do Pittsburgh's buses. In twos and threes, Pittsburgh's bunched buses bear the bad news of crowding (the front bus is often overfull, with subsequent buses nearly empty), excess waiting (bunching creates long waits between bus arrivals), and underutilization (crowding and unpredictable arrivals reduce public transit use). Caused by the cascading effect of variance in passenger arrivals, bunching bedevils public transit planners not just in Pittsburgh but in cities around the globe.

Does bunching follow predictable patterns? This project presents the geotemporal occurrence of bus bunching in Pittsburgh for PAAC routes 61, 71, P1 and G2 during March 2016. The goal is better understanding of where and when bunching occurs, to be used as a foundation for approaching the natural follow-on question: "how do we make this better?"

The chart below shows when and where along a bus's route bunching tends to occur. The data is derived from vehicle location information collected for the month of March 1 - 27, 2016 (data collection details). Observations are defined as "bunched" if observed to be operating within 250m of another bus servicing the same line in the same direction.

The chart shows the frequency of bunching occurrences for buses travelling along their service routes. Compare the bunching frequency of on-street routes such as 61B or 71C to dedicated busway routes P1 or G2.

White vertical lines indicate that no trips departed during the time window, corresponding with published schedules. Red vertical lines indicate the recurrence of bunching for routes at certain departure times, and that bunching tends to persist throughout the duration of the route. Horizontal lines show patterns in where bunching occurs.

One possible hypothesis is that bunching is caused by heavy ridership during rush hour (inbound between 7:00 and 10:00 am, and outbound between 4:00 and 6:00 pm). Contrary to what one might expect, the data suggests that location (horizontal bands) is more deterministic of bunching than time of day (vertical bands).

Routes P1 and G2 operate along dedicated busways. Termed "bus rapid transit," these busways separate buses from on-street traffic. Despite carrying as many passengers as 61x and 71x routes, these buses experience far less bunching.

Bunching Timelapse Heatmap

To see this all in motion, click the "play" button below to see a time lapse of the occurrences of bunching. The timelapse map below shows data for March 1 - 27, 2016 aggregated into a single day.

The heatmap shows spatial frequency of bunching. Since more buses are in operation during rush hour, everything else being equal one would expect to observe more bunching during rush hour (simply by virtue of there being more buses which could bunch).

The timelapse heatmap shows that downtown, Oakland, and the intersection of Murray and Forbes in Squirrel Hill are persistent hot spots for bunching. The movement of bunches through time illustrate that bunches snowball and persist. As above, routes P1 and G2 show comparatively infrequent bunching.

What can be done to reduce the waiting and overcrowding caused by bus bunching in Pittsburgh? The evidence above suggests that one possible solution would be to create new Bus Rapid Transit lines, especially connecting Downtown, Oakland, and Squirrel Hill. A coalition of community partners are advocating for just that. Learn more about Bus Rapid Transit and their efforts by visiting www.GetTherePgh.org.