Pittsburgh Bus Bunching (2016)(bunching.github.io) |
Pittsburgh Bus Bunching (2016)(bunching.github.io) |
Why do buses bunch?: http://setosa.io/bus/
The solution is also very simple. Stop the waiting time at each bus stop depending on the number of passengers that get on or off. Slacken the timetable, so that every few stops the bus waits for a minute to match the schedule.
On bus routes with very frequent service, for instance every 10 minutes or more frequent, there is no longer really any need to have a timetable - it will be a bigger win for the customer to just make sure the buses don't bunch than it is to actually publish a timetable. In this case, you may want to implement a negative feedback loop based on vehicle tracking. Most buses are remotely tracked now anyway, so it shouldn't be too hard to return a signal to each bus saying either "carry on as normal" or "you're too close to the bus in front, wait a bit at the next stop".
This is basically the one problem that needs to be solved for buses to not be rubbish, and it is so simple to do so.
If you check out their site[2] you can click on the Route Map button and see the buses run in real time.
[1] https://urbanmilwaukee.com/2014/12/03/eyes-on-milwaukee-new-...
Except for the little hiccup of organizational politics in government bureaucracies that needs to be solved or side stepped as a prerequisite. If you can do that tons of "hard" problems are easy to solve.
That's one way to infuriate the passengers on the waiting bus.
This particular piece emphasizes what anyone who attended CMU could tell you - if you're going to take the 62C anytime during rush hour, you better learn to hold your breath.
Edit: kudos to doing this in R. I’ve done other city analysis stuff in R and I keep coming back to it for one off things like this.
Just teach the bus drivers to let the empty busses overtake the crowded one.
It won't solve the pauses when there will be no bus coming, but at least it increases the passenger comfort by a good margin and the average speed by which the bunch travels increases too.
On mobile so can't inspect.
[0]: https://d3js.org/
The worst bunching occurs downtown and right in front of the University of Pittsburgh.
And, Pitt is going to cause bunching at discrete times when classes end.
The explanation given to me by a former driver is very simple. When a driver is very hungover or in bad form for one reason or another they poll the other bus so they get light duty.
Experienced bus travelers would often know when there'd be two buses in a row, too. And of course the bus drivers would point it out.
I wonder what it would look like if departure times were enforced in the don't-leave-before direction. Then giving enough time between scheduled stops that e.g. 90% of buses traveling that stretch can make it in time for that day/time combination. I realize that doesn't mean that 90% of trips will be on-time due to how failures will cascade, but it should eventually catch up due to the slack provided.
The big question is whether people will accept the increased latency (and total trip times) for less variance in waiting time and trip time.
The problem is that the route timing is designed for a specific traffic and ridership level. The actual time to run the route can vary by a factor of 3 or more depending on whether it’s before, during, or after rush hour.
The route timing is designed for after rush hour traffic levels, when it takes about 45-50 min to travel from my stop to my office stop. During rush hour, this trip can take 90 minutes. Before rush hour, it can take 20 on a good day.
So now, if I leave before rush hour, a commute that could be 20 minutes becomes 50 minutes. On a bus that is scheduled to arrive every 7 minutes - so bunching means waiting 15-20 minutes for a bus: LESS commuter time wasted by bunching than is spent sticking strictly to the schedule.
On buses that run even more frequently (one every 2-3 or 3-5 min during rush hour), this would be ridiculous. People aren’t catching the bus on a schedule - they just walk out the door whenever they’re ready, and expect a bus to turn up within a few minutes.
On a bus that’s scheduled to run once every 20 min (and you can wait an hour for a bus if they bunch), people are more invested in the schedule. I don’t actually trust the schedule at all unless within a few stops of the start of the route, and usually use the “just walk out the door” method anyway; but would appreciate and use the schedule if it were reliable.
So: as a bus rider I might support this, but they would have to fix the schedules. Create and enforce bus lanes. Run more buses - so that even when you get more riders than usual they can carry all the passengers at rush hour without being so packed that it takes 5 min to cram each new passenger aboard and half the bus has to debark to let someone off the middle. Then a little bit of traffic is less likely to create bunching.
> Pittsburgh, however, as an unusual misfeature in that the single bus lane goes the wrong way against one way traffic (3 lanes of cars going one way--1 lane of bus going the opposite) for a non-trivial amount of length flagged as "hotspots".
But yes, that going against the flow doesn't make overtakings any easier.
Socialising the idea to the passengers is also probably difficult.
But the bunching itself is not a problem. The problems are the effects it causes like:
1. crowded busses
2. arrival delays
Overtanking has a major impact on the first effect as you have stated too, the 'crowds on the buses quickly equalize'. The impact on the second effect depends on some other parameters. If all busses are crowded there will be no effect at all, but if only a few busses are crowded (the front busses) and the busses behind them are empty, there can be a positive effect on the arrival delays too (the worst case drop off time will be lower).
So I would not say 'it does not help' as it certainly decreases the negative effects.
I mean, where I live, busses manage to overtake in a one lane for each direction scenario (which is theoretically complicated, but works out in practice as traffics lights can give you chances to use the opposite lane). Nevertheless, what is important is that the front bus waits until the busses behind him have taken the lead. Otherwise there will be very few opportunities to safely overtake the front bus.
That would mostly work, but the problem in Pittsburgh at least is the bus bunching happens between related but not quite identical routes. For example, the 61A, 61B, 61C, and 61D all run down Forbes Ave, and frequently get bunched there. But then later in their routes, they fork: two go one direction, and two go the other direction. For many passengers, it doesn't matter which one they take. But for other passengers, it has to be a 61A etc.
As is hinted at by the paper, bunching is generally caused by factors such as traffic which makes it hard to actually eliminate. Transit agencies have tried various solutions to the problem and it is very much a work in progress. One thing I have noticed is that transit agencies tend to have sub-optimal data for their service so they tend to be slow at fixing scheduling related issues.
Pittsburgh, however, as an unusual misfeature in that the single bus lane goes the wrong way against one way traffic (3 lanes of cars going one way--1 lane of bus going the opposite) for a non-trivial amount of length flagged as "hotspots".
This means that while automobile drivers using the bus lane is a self-correcting problem, a bus cannot pass another bus in those areas.
(I call the wrong way bus lane a "misfeature" because it regularly results in out-of-towners winding up dead because they didn't look both directions on a "one way" street. This is particularly tragic when the out-of-towner is someone who is visiting Children's Hospital because their child is being treated for some very aggressive disease.)
I suspect a lot of bunching is caused by buses having to perform complicated manoeuvres around car traffic because of poor stop positioning. In Barcelona what many lines do is skip stops right before left-hand turns (the line isn't even listed at that stop). This way the bus can switch lanes well ahead of time and avoid having to wait for traffic to go though all lanes. This works because most lines stop every two blocks, so you never have to walk too far for the next stop.