Why do buses bunch?
Unfortunately, there is a natural tendency for buses to bunch like this -- almost a "gravitation" effect which is caused by a combination of factors. The first is the reinforcement of delays due to boarding and alighting. When a bus begins to fall behind, then at each bus stop it starts to face larger crowds than otherwise. These larger crowds require higher "dwell times", causing the bus to fall further behind. What's worse is that the more passengers there are riding the bus, the longer it takes for additional passengers to board. There are some scientific studies of this effect, but it should be fairly intuitive to the observant: just watch next time as riders attempt to board an already-crowded bus and you'll see that it takes much longer to get moving again. Also, the more riders on board the bus, the more likely the Stop Request will be pressed and the less likely the bus will be able to skip any stop (this also makes dynamic "non-stopping" or "expressing" less effective).
|How two buses end up leap-frogging each other under heavy demand.|
Ultimately the delays may add up so much that the next bus arrives and passes the crowded one. Unless passengers have opted to wait for the next bus (for example, by consulting their smartphones), it is likely that the new bus will be relatively uncrowded and unencumbered, making it much easier for it to catch up. But once the new bus passes the delayed bus, then it starts to encounter bus stops with higher than expected passenger loads, and begins to slow down in the same way. So it begins to "gravitate" back towards the first bus -- which may be moving more smoothly by now. Some of this could be avoided by sending the bus "express" but then that must be communicated to passengers on-board, giving them plenty of time to alight if necessary. On a bus it seems that delay cancels out any gain from going "express."
|Clearance time adds random delays|
Other sources of delay include increased "clearance time" caused by failure of passing motorists to yield to the bus that is merging back into the travel lane. Although the number of passengers riding isn't directly connected with increased clearance times, there is an indirect relationship: the more often the bus has to pull out of traffic, the more likely it is to spend an inordinate amount of time merging back into the lane. Also, times of heavy automotive traffic -- preventing the bus from merging back -- often correspond to times of heavy bus ridership (e.g. rush hour).
|Optimizing traffic signals for cars can hurt bus performance|
Finally, I would like to mention traffic signal timing. Traffic signals act as a capacity regulator insofar as they only permit bus movements a limited number of times per hour. Generally, that limited number is high enough to allow the needed capacity. But the timing of lights along a corridor can wreck bus schedules when poorly coordinated. For example, a set of signals that is optimized for automotive flow will probably not serve buses very well, because buses travel in a different pattern from private cars -- instead the bus will find itself stuck at every red light along the way, leading to delays that will probably tip over into the dwell-time "gravitation" trap described above.
What can be done? It seems that the root cause of the problem is the variability of dwell and clearance times. These also happen to be areas where train service is typically superior to bus service; although there is nothing inherent about buses which forces this distinction. The difference is simply that most train designs allow fast boarding and alighting (not the Green Line sadly), and that they do not have to deal with the clearance time issue at all. So, the first place to start, I believe, is to find ways of bringing these advantages to bus service as well.
Continued in part 2.