Greater Greater Washington has an outstanding piece up about how we measure bus reliability:

Every month, Metro’s customer service committee looks at a presentation on operating statistics, including a chart showing the latest bus “on-time performance” percentage. That percentage reflects the number of buses that arrive within a certain time before or after the published schedule — usually around 73-75%. However, on its own, this number doesn’t give management enough information to effectively maintain bus quality. In addition, it doesn’t capture important elements of bus performance, like bus bunching…

London looks at bus on-time measuring differently. Because the bus lines in London are operated by private contractors, it’s very important for the local transit authorities to accurately measure on-time performance because there are real financial incentives or penalties involved. They measure how often buses pass by certain points on the network and track the “excess waiting time.” All that time you have to wait for a bus that’s running late or is bunched with others is added up and averaged over the route, and the excess waiting is compared to how much you’d normally have to wait assuming you come to the bus stop randomly. The reports for bus performance are published on the web.

Do click on that last link to London’s bus reports (excerpt pictured above).  I think Sound Transit’s ridership reports are pretty transparent, although Metro’s are somewhat less so.  But this blows both of them out of the water.

Metro’s definition of “on-time” is between 1 minute early and 5 minutes late, tighter than WMATA’s standard of 2 and 7 minutes, respectively.  Still, they were able to achieve a 75% on-time performance (pdf) in 2007.  Sound Transit’s definition is less than 10 minutes late, which they’ve achieved about 94% of the time (pdf).

The point that GGW makes, though, that on frequent routes the actual service interval is more important than adherence to any actual schedule.  I know that neither agency’s staff is really looking for more things to do right now, but it’d be nice to start adding this framework for routes like the 3 and the 550.

On the other hand, adherence to this metric and this metric only would incentivize some odd behavior — like having some bunched buses simply idle for a while to fix the interval.  What’s better: making sure the wait at the stop is uniformly short, or preventing people from sitting on idle buses?

10 Replies to “Bus Reliability Metrics”

  1. it’s not entirely clear to me why average on-time performance over service changes is a transparent measure of performance (or what is even going into that aggregate measure).

  2. One thing to consider with the “odd behavior” of holding bunching buses on routes with high frequency is that the action results in at least two significant benefits.

    The first is that it distributes seating capacity better. These types of routes tend to attract passengers more or less as a rate of passengers than a group when compared to low frequency routes. This is likely part that the passengers on a high frequency route don’t have to adhere to a schedule and partly because even the passengers that do try to catch a scheduled bus have allow different amounts of time to catch that bus.

    In practice, the longer the time is between buses the more passengers that accumulate at the bus stops. A bus that is less than the headway behind the previous bus will pick up fewer passengers. A bus that is more than the headway behind will pick up more passengers. Until seats can be physically moved to another bus in route, any bus picking up a low number of passengers has underutilized capacity.

    The second benefit is that holding a bus helps to prevent more bunching downstream. Holding a bus allows that bus to pick up passengers at make stops that otherwise would have been down by the following bus. Reducing the time spent by the following bus used to load passengers and make stops helps keep the following bus on schedule.

    Ideally, this type of control would be done subtlely and early through a reduced driving speed or short waits at major stops and not by having a bus that has already caught up to another bus wait 5-10 minutes.

    1. All this would be much easier if all the buses had GPS. A bus driver that knew her bus was closing in on a late bus up ahead could slow down, pause for a moment at a stop, etc. A bus driver that knows he’s late and that there is another bus coming close behind (less than 2 minutes) could skip a stop where no one wants to get off and make up a minute or two of time, knowing that the passengers at the stop would not be further delayed to any significant extent.

      But without GPS, the bus drivers can’t know their relative positions, so any attempt to avoid bunching isn’t going to work until the buses are within sight of each other–which is almost too late.

      1. I’ll definately second the need for GPS in every coach. I think the decisions should be made centrally though, and any actions taken to aleviate bunching coordinated by a dispatcher who directs drivers.

        I also wonder if rather than idling a second bus when it catches up to an earlier one, if it would be better to make them both switch to a staggered skip-stop system like on 3rd Ave where each one only makes stops at every second bus zone. Might not do as much to fix the bunching right away, but I think it would get more people where they are going (or at least within a couple blocks of it) on time while there is the heavy demand causing the bunching in the first place.

      2. This skip-stop situation naturally occurs a lot of times when a 48-Rainier Beach and a 48-Columbia City bunch near 23rd & Jackson. Bus A will make a stop when Bus B has no one to get off. Bus B passes, and while it waits for people to get on at the next stop, Bus A passes by. This leapfrogging can happen 5 or 6 times before the routes branch.

      3. Ah, but the late bus is probably picking up extra passengers that would have missed it if it had been on time. The better solution is for the bus behind to leapfrog the late bus and pick up some of the passengers.

    2. Perhaps a lesson could be learned from what commonly happens in NYC with the trains, and I’m sure elsewhere.

      If one train is delayed, causing bunching, the first train will switch to the express tracks, skipping a segment of local stops. The conductor will first make an announcement, telling all passengers intending to get off at the local stops to get off and wait for the next train, which is directly behind. The second train picks them up next, and they make their stop with minimal delay, and no one had to wait on an idled train.

      There’s no reason buses couldn’t also do this, with a little communication between drivers.

  3. I know when I’m waiting at a non-timepoint I’ll often determine how long I should be waiting by the interval between buses. If it’s been that long since I arrived at the stop, I’ve waited long enough even if buses are horrendously late.

  4. The problem with holding buses is that it’s *incredibly* frustrating to the people on the bus that’s being held. Yes, there presumably aren’t that many of them because the bus in front presumably picked up more people. But for anyone on a schedule, it means that you can’t even relax when you get on a bus and there doesn’t seem to be traffic, because your bus just might choose to sit in place to let the other bus get ahead.

    1. I just went and read the GGW thread and one of the comments made me realize: I like buses to be on schedule and all, but the real key is getting screens with predicted wait times at stops. If I know the bus is still 10 minutes away, I can get a cup of coffee, find a paper to read, and generally relax in my waiting time. If I have no idea when the bus is coming, the whole wait is much more stressful — 5 minutes with no knowledge of when the bus is coming is about as bad as 10 or 15 predictable minutes.

Comments are closed.