The Myth About More Service on King

Over the past year, oft we heard the TTC argue that adding service is of no use because vehicles just get stuck in traffic.  More service does not lead to more ridership.  This is hogwash, but they’ve been getting away with it.

Let’s look at just what sort of additional service the King car has enjoyed.

Back in the dark ages of 1990 before a recession and service cuts took their tolll, the King car provided the following service:

  • AM Peak: 2’20” combined service west of Church, 3’30” to Broadview Station
  • Midday: 6’00”
  • PM Peak: 2’43” over the entire route
  • Early Evening: 7’17”
  • Late Evening: 9’20”

The route operated 4,600 vehicle miles or about 7,360 km of service and carried 58,800 passengers per day, and the Lake Shore route did not yet exist.

By 2000, the service looked like this:

  • AM Peak:  A 4’00” headway from Dundas West to Broadview overlaid by a 4’00” headway east from Roncesvalles for 12 trips corresponding to the peak eastbound demand to provide a 2’00” headway for 48 minutes.
  • Midday:  6’20”
  • PM Peak:  4’00”
  • Early Evening:  7’30”
  • Late Evening: 10’00”

The route operated 4,300 vehicle miles or about 6,880 km of service and carried just under 52,000 passengers per day, and the Lake Shore route did not yet exist.

By 2003, the service looked like this:

  • AM Peak:  A 4’00” headway from Dundas West to Broadview overlaid by a 4’00” headway east from Roncesvalles for 16 trips corresponding to the peak eastbound demand to provide a 2’00” headway for 64 minutes.  Three trips on Lake Shore fit into this but they are widely enough spaced that they do not make a big impact on the headway seen by riders waiting in Parkdale or King/Bathurst.
  • Midday:  5’40”
  • PM Peak:  4’12”
  • Early Evening:  7’15”
  • Late Evening: 10’30”

The route operated 4,200 vehicle miles or about 6,720 km of service and carried just under 48,000 passengers per day.

Today, the service is:

  • AM Peak:  Same as 2003
  • Midday 4’45”
  • PM peak 3’45”
  • Early evening 7’00”
  • Late evening 9’00”

The King and Lake Shore routes operate a total of 7,100 km per day and, according to the TTC stats for 2005, carried 47,900 passengers.

[Updated Feb 5] When the TTC says that it added nine cars to the route, this is simply not true in the context of recent service changes.  Relative to 2000, extra eastbound trips have been added to the AM peak, and midday service has been increased.  However, there has not been the magnitude of change one would imply from “nine more cars” because overall the amount of extra service is small.

[Updated Feb 5]  The number of cars in service in the AM peak on King and Lake Shore was 44 in 1990, 38 in 2000, 50 in 2003, and 48 today.  Note that in 2003, the route was under major reconstruction.

The combined 2’00” AM peak service has been in place since 2000, with a few trips added by 2003.  Meanwhile, the wider headway of the afternoon peak remains.  One might reasonably ask whether the prospect of a tedious wait for a trip home is a disincentive to people who might ride the marvellously improved AM peak service.

The improved midday service, implemented in September 2005 as a first step in the Ridership Growth Strategy was worthwhile, but this probably happened after the 2005 riding count was taken, or was in place for only a short time before.

The TTC must stop misrepresenting the issues of service, demand and traffic congestion.  We are deadlocked with a bogus claim that more service doesn’t generate more riding and a demand that exclusive lanes are the only solution to our problems.  The logical conclusion is that we shouldn’t spend more on TTC service because it won’t make any difference.  What, then, is the purpose of the Ridership Growth Strategy?

5 thoughts on “The Myth About More Service on King

  1. These timings, presumably, exclude the fact that MANY 504 cars are short-turned and are based on all cars running the whole route.  I assume that getting accurate figures on the % of cars short-turned is impossible but my rather limited experience seems to indicate there are more and more short-turns.

    You have already written on the currency and accuracy of TTC ridership figures – not very current or accurate being the bottom line!  I suggest the TTC should try to get better figures of ridership so they can properly measure need and see trends.  If one can question the ridership figures there is sure to be someone who will use this to question any remedy based on them!

    Steve:  I am about to start a detailed analysis of operations on many routes including King based on monitoring data from the vehicle tracking system that I received recently from the TTC.  This is going to take a while.  Stay tuned.


  2. Wasn’t 2003 the year King Street trackage was rebuilt?

    Steve:  That is correct.  I will add a note to the main post indicating that this “high water mark” in number of vehicles was due to construction.


  3. Steve, I’m not quite sure how these numbers prove your point. You should take a closer look at them. Convert them into passenger-kilometres travelled – that’s how you measure route performance – then compare the differences in headways which are quite minor for the most part.  You’ll notice a very different trend.  I don’t think a 15 or 30-sec difference in headway – or even a 1-minute difference – can influence ridership in any significant way.

    Steve:  Two points.  There is no way to reliably convert passengers to passenger kilometres.  Yes, on a broad scale, passengers on a route will take average trips of “x” km and therefore the boarding count is a surrogate for km travelled.  However, changes in the km operated are a function of two effects:  implementation of transit signal priority and the temporary increase in trip times to allow for construction delays.

    I concur that small changes in headway will have little effect on riding.  That’s the whole point of this post.  The TTC claims that it “added” service and nothing happened when, in fact, the wave of frequent service eastbound through Parkdale and Bathurst/Niagara have been in place since 2000.  The wave was lengthened slightly, but overall, there has been little change in peak service offered on this route for a long time.

    Claiming that adding service produced no effect is misleading because the change was made years ago.  Ridership had lots of opportunity to settle into that service level, and it’s no surprise that people claim today that they cannot get on because the cars are full.  Some of the development fuelling that crowding wasn’t even built in 2000.


  4. Looking at the headway figures between 1990 and today it appears as though service today is very slightly better than 1990.  The service seen by the customer is worse though, and vehicle bunching appears to have a big part in the problem.

    The TTC may be right that congestion may be the cause, and that adding additional service does not correct the problem because they are addressing symptoms rather than fundamental problems.  The fundamental problem appears to be control, and this is where the problem needs to be addressed.  A brief delay for any reason (Traffic, lights, slow loading, slow operator etc.) will result in in longer headway, heavier loading, longer stops, more stops etc which magnifies the delay.  As the operator falls further and further behind the schedule, the schedule loses all relevence and he just runs the vehicle as best he can. 

    Eventually the service gets so far out of whack that someone steps in and takes drastic action such as short turns and removal from service.  The following vehicle sees the opposite, few customers, infrequent stops etc. and hangs on the back of the first vehicle, effectively taking his vehicle out of service.  New vehicles injected into service will tend to catch up to other vehicles and tend to operate as “following vehicles” unless great care is taken on where they are injected into the fleet.

    The key is precise control of vehicle headway so that problems are not allowed to escalate out of control.  This is not an easy problem to resolve to the required precision, and there are many implications to all parties involved.

    In order to effectively address undesirable situations, it is important that the accent is on underlying problems rather than symptoms.  Just as lack of inadequate funding leads to “prestigious projects”, lack of control leads to poor service.

    Some thoughts: Vehicle location detection (GPS?) tied into transit signals to provide a dynamic headway control to replace fixed scheduling.  This is done quite effectively in the subway.

    Steve:  Actually, the way the subway works is that at various locations, the signals are tied to the schedule.  They allow a train to leave if the scheduled time has been reached.  However, they have no ability to regulate service where everything is running late and so bunches will stay bunches rather than being automatically spaced out.

    The signal at, say, Yonge westbound, does not “know” that the train sitting on the platform is run “43”, only that some train was due to leave at “10:15”.  Whatever train shows up will be dispatched even if its leader is less than two minutes ahead.  Dynamic spacing of service is a tricky business as it requires “look ahead” and “look back” capability, the sort of thing a real, live tower operator would use by looking at the signal board to see where the trains were, and managing them based on skill and experience.

    Also, the surface fleet does not yet have GPS, although this is to be installed as part of the automated stop announcement system.  The much older units now onboard rely on wayside transmitters (“signposts”) that update the monitoring system with a vehicle’s location.  The “in between” locations are calculated from the odometer.  Holding a surface vehicle for time would be meaningless given that they are so often late.  The problem is to ensure proper spacing, not just “on time” performance.  That requires a major rethink of line management.


  5. I can see your point, but the effect of TSP [Transit Signal Priority] on the overall km operated is small since its primary purpose is to maintain or improve “on time” performance, not travel time (e.g. VIVA in York Region, 98 B-Line in Vancouver).  Even if the TTC’s TSP does give unconditional priority (which is probably the case since streetcars aren’t equipped with GPS therefore the phases must be fixed), how many extra “runs” per day can you squeeze out of that?

    Steve:  For those unfamiliar with the system, the TSP in Toronto operates on the basis of extending green time when a transit vehicle is present and/or shortening red time if the vehicle arrives in that cycle of the phase.  The basic cycle is not changed, only the balance between times allocated to each direction.  Since the basic cycle is often 80 seconds, this scheme runs out of steam when the headways are close because there is always (in theory) a transit vehicle present in one direction or another.

    The TTC did reduce running times on several routes after introduction of TSP.  This generally translated to fewer cars running on the same headway rather than the same cars running closer together.  Indeed, TSP was sold on the basis of operating cost savings, not on its ability to improve service.  Since passengers “feel” waiting time at stops far more than riding time in vehicles, this approach is not the most productive improvement one might make.  However, the ethos of the day was to save money and cut total vehicles in service, and the pressure of growing riding was unknown.

    Also, based on the headways, this is what the passenger capacity/hour shows:

    AM Peak: 1270 to 1900
    Midday: 740
    PM Peak: 1630
    Early Evening: 610
    Late Evening: 476

    AM Peak: 2220 (Roncesvalles to Broadview)
    Midday: 935
    PM Peak: 1184
    Early Evening: 634
    Late Evening: 493

    The real big drop is in the PM Peak – that’s probably where the effect of traffic congestion becomes most prominent.

    Steve:  There is no question that there is more congestion in the PM peak than the AM.  This makes the TTC’s claim about adding service that gets lost in congestion even more galling.  The added service was in the AM peak back in 2000, while the PM peak provides far less capacity (and by extension more widely spaced service even without the effect of gaps).


Comments are closed.