Fault tolerant… public transit

I was going to write a proper blog post today, but I got stuck in a good old fashioned TTC delay for about forty minutes this morning (this also means I haven’t proofread this post nor thought too much about the topic). Some poor folks that left later than I may have been held up by multiple delays. Indeed, a single delay on the TTC seems to invite further delays. I.e., though long delays aren’t (that) frequent, once there is one delay, multiple long delays seem to crop up. If the TTC has statistics on the number of chains of delays that appear within half an hour of each other, I’d love to see if I’m just experiencing a memory bias (also likely). However, I’m not here to rant today, but to put some thoughts down I had during my bonus (cramped) quiet time today.

The current state of affairs

The government thinks that we Torontonians and neighbouring regions are willing to foot what looks like, from various sources, an approximately 1% increase in the TTC’s total operating budget per year by designating it an essential service.  What does this buy us?  Well, suppose it can completely eliminate both legal and illegal strikes; that means that the time spent by union workers striking, based on historical data since 1952, will drop from 0.3% of the time to 0% ((Source: http://en.wikipedia.org/wiki/Toronto_Transit_Commission , accessed March 10, 2010.  This includes Pope John Paul II’s visit to Toronto.)).  This is equivalent to an average of a 5 minute delay per day — which I’d estimate to be below the average due to unexpected bottlenecks ((Since the beginning of March, I’ve experienced two major subway delays lasting fifteen minutes or longer: a forty minute delay this morning and an hour-long delay last Tuesday for an average of 10 minutes of delay per day, including the days I didn’t commute, let alone smaller random delays.  This is not atypical.)).  Aside from the one-time wildcat strike of 2006, relative to random delays, one can easily plan for alternate transportation during a strike and employers should be more understanding during these times (actually, some people plan around the TTC’s nonereliability random delays by doing exactly the same thing).  Thus, if we can apply that same 1% increase in budget towards reducing random delays, we could come out ahead.  Yes, I’m ignoring people who don’t know anyone with cars that can’t afford to get to work if the TTC isn’t running.

Positive feedback loop

One of the simplest ways to deal with random failures is to build in redundancy.  We obviously can’t afford to build a fully redundant, parallel transit system to the existing TTC infrastructure — that would approximately double the TTC’s costs!  But if we look more closely at why a single delay causes a cascade effect, maybe we can deal more effectively with them.

The TTC already has a clue:  as trains reach capacity, it takes longer to board them because people start trying to squeeze in, blocking the doors.  It also takes people longer to enter and exit each train, meaning trains end up standing still in a station for longer, still unable to let everyone board.  Without any fancy queuing theory, we can already see that if we decrease capacity (by decreasing the number of trains moving through a station in a given period of time) while being unable to deal with the previous traffic, one has a negative feedback cycle halted only by the finite number of commuters and an eventual decrease in arrival time.  One way the TTC has attempted to deal with the problem is by shepherding passengers at the busy Bloor-Yonge station so that they don’t all try to enter via the same doors and by closing the doors while people are trying to get on.  I think this is a good change, but it only gets us so far — and does very little to deal with delays that shutter tracks or cause greater queues for trains.

However, I think this isn’t the only reason for additional delays once the prime (un-)mover happens.  Commuters may begin to feel sick and definitely more agitated when delays start, increasing the likelihood of delays at different points along the subway line.  This, too, combines with the previously mentioned positive feedback loop, causing increasing delays.  On Remembrance Day 1998, I recall at least three passenger assistance alarms activated during my commute to school, including one on my particular car because someone started to feel sick from being confined in the subway.  On another occasion, someone threw up next to me on her first ever journey on the TTC during a longer-than-usual delay because of the build-up of body heat and crowded conditions ((Ainsley, close your eyes!!!)) (yes, she threw up on someone because we were packed like sardines, but that someone wasn’t me).  Talk about a ride to remember.


If we can reduce the demand on the subway during times where trains are nearly full, we can reduce boarding delays and, if this system runs even when the subway fails, we’ve achieved some level of success.  One way to do this is by implementing something on a full-time basis that we already do in exceptional situations: shuttle buses.  Unfortunately, these emergency shuttle buses take time to mobilize and probably require paying drivers for overtime.  If we employ them on a regular basis whenever subway capacity is expected to approach maximum capacity, we can improve the reliability of the subway.  I estimate that the platform build-up at York Mills this morning was about one bus-load every five minutes (every two trains).

Currently, the TTC manages over 100 bus routes (though some aren’t particularly frequent).  Assuming we have the buses for a regular shuttle (which we may, considering the emergency shuttle buses) for both subway lines and running them regularly during the five hours or so per weekday that the subways are highly congested, that’s an approximate increase in bus capacity of about 1% and a rise in total costs of less than 1% per year, the estimated cost of making the TTC an essential service.  Because this service would run regularly rather than deployed as needed, shorter delays won’t compound into longer ones as quickly, the aftereffects of long delays will dissipate more quickly, and there is no latency in deployment.

Yes, what I’m suggesting will increase the TTC’s operating costs, but it would provide greater benefits to most people than deeming the TTC an essential service and at lesser cost.  In fact, we already have something similar to what I’m proposing: the 97 Yonge route.  If we promote an express version of this bus (such that it stops at the same points as the subway) that arrives frequently, I think we could enhance the daily commute of many southern Ontarians.

Leave a Reply

Your email address will not be published. Required fields are marked *