This post is a summary of the major issues I have seen so far in the CIS data for the King route.  The supporting detailed analyses will follow in separate posts, but I wanted to get the main issues out early so that readers would see where this is going. 

Acknowledgements and Disclaimers

I wish to thank Bob Boutilier and Steve Perron at the TTC for making available the data that allowed this and many other analyses to come.

The opinions expressed here are my own and do not represent the position of the TTC.  They provided the data.  I did the analysis, and I am sure that there are changes and improvements that will come to light with feedback, official or otherwise.

For those readers who are ATU 113 members, I want to clearly state that my intent is not to point fingers at anyone, but to provide some of the raw material needed to address how service can be improved.  Although CIS records them, I specifically asked that operator badge numbers not be included in the data I received from the TTC. 

Inevitably, some dubious operating practices, most commonly “soaking” (running early so that your vehicle is near-empty and the operator behind is overworked), are clearly visible in some of the charts, but this is fairly rare.  Indeed, I must ask how two vehicles can be left running nose to tail for hours with no intervention.  The responsibility falls at least as much on line management as on the operators involved.  Other problems are evident and far more common.

Headways

The TTC’s goal is that service operate within three minutes of the scheduled headway.  On routes like King with a 2-minute AM peak wave, and headways of 4-to-6 minutes at other times, this is really a meaningless goal.  All the same, headways well beyond the 3-minute target do occur especially beyond common short turn points such as Parliament or Roncesvalles, and especially at off-peak times when a short turn puts a bigger gap into the service.

Unfortunately, a lot of effort seems to go into keeping cars on time at the expense of regular, frequent service over the entire route.   Short turns are the end result of many factors, and I will talk about them a bit at the end of this piece.

One chronic problem, even during off peak periods and days when there is little congestion, is irregular departure times at terminals.  Even on Christmas Day under ideal conditions, service tended to run as pairs of cars.  Although the scheduled service was every 9 minutes, in practice this would swing between 6 and 12 minutes westbound from Broadview Station.  By the time the service reached Parkdale and Roncesvalles Village, the second car might only be 4 minute behind its leader.  The service leaving Broadview is within the TTC’s target, but the result is irregular service even on a day when this should not occur.

The effect on passengers is that most see service running every 12 minutes or worse despite what is advertised.  Assuming that passengers arrive at stops uniformly, most of them will be served by the car with the wider headway.  They will wait longer and be more crowded than any statistics or schedule averaged over an hour will reveal.

If and when the TTC embraces stop information signs showing the anticipated arrival of the next car, it will be intriguing to see the information these display as compared with the schedule.

Congestion

The TTC often blames its problems on congestion saying that it cannot improve service without exclusive transit lanes.  In today’s Toronto Star, we have TTC Chair Adam Giambrone:

“‘Do something about King!’ – I hear that all the time,” he says. “Well, what do you want me to do about it? If I add another streetcar, it just gets eaten up in traffic.” 

Alas, that is the canned TTC staff response we have been hearing for many years, years during which service was mercilessly cut to adapt to reduced subsidies from both the City and Queen’s Park, and when the fleet was allowed to shrink through retirements.  The length of time most streetcar routes get on their schedules is at a level comparable to the early 1990s.  What has happened?  Improved travel times through signal priority have shaved minutes here and there from trips.  The big problem remains that there are not enough cars on the street.

Update:  After I had posted this item, I received the following email from Adam Giambrone:

Steve,
 
For the record, I agree that there is not enough service on King (and some other streetcar routes) and I have said this in public before.  We need at least 10 streetcars maybe a few more to carry everyone.  Having said that, traffic remains a huge problem and as you know we are seriously looking at your idea for a change in how service is provided on King.
 
You can quote me on this!!
 
Adam
Adam Giambrone
Chair – Toronto Transit Commission (TTC)

Congestion is not uniform in time or space.  It happens at different locations at different times of the day and days of the week.  It is not only (or in some cases even) a rush hour problem.  For example:

• Congestion in the AM peak is a minor factor on the King route.  Although there is some increase in trip times through certain neighbourhoods, this is partly due to loading delays with overloaded cars.
• In the afternoon peak, the running times through the core, especially westbound, don’t start to build up until after 5:00 pm and this effect lasts into the early evening.  The problem appears to be the onset of street parking before the level of road traffic has dropped to a level where half of the road space can be given up.
• Congestion in Parkdale is variable and is closely related to operations on the Gardiner Expressway.  When that expressway is blocked, travel times through Parkdale, westbound especially, rise dramatically.
• Congestion in the theatre and club district (west of University Avenue) appears only on certain nights partly depending on weather conditions, the day of the week, and activity at the theatres on King.  This is primarily a westbound effect given the circulation patterns in the area.  There is no provision for this effect in the schedule, and riders on the Broadview and Roncesvalles ends of the line suffer from many short turns any night the club district is busy up to and beyond midnight.
• Early closing of offices before Christmas triggered major delays on the downtown routes because rush-hour level traffic attempted to drive on streets clogged with parked cars. 

All route segments do not suffer from congestion, and some parts have a minimal change in running times during peak periods.  Proposals for reserved lanes need to address both the location and time of day when congestion problems delay streetcars.  Off-peak, when scheduled service is less frequent, requires different aproaches both to parking management and to provision for area and time-specific delays.

Service Reliability

The King route is scheduled with a basic AM peak service every 4 minutes.  To this, the TTC adds a “wave” of cars also on a 4 minute headway scheduled to operate eastbound through Parkdale and King-Niagara corresponding with the inbound peak demand to downtown.

However, for reasons best known to the TTC, these extra cars originate at Russell Carhouse in the east end, make a trip to Dundas West and then come back through downtown.  That’s what it says on the schedule. 

In fact, many of these cars do not enter service on time and some do not enter service at all.  When they come out late, they are short-turned and there is a good chance they will not be on time for their eastbound trip either.  Even if they were, they miss serving the very part of the peak demand for which they are intended.  I believe that these runs are probably operated by volunteers (overtime work) or by drivers on the “spare board” who will in for missing regular drivers. 

When the weather is bad, both are in short supply and there is a lower chance that this type of run will make it onto the street.  Bad weather, of course, drives up TTC riding, and this happens just at the point where more service is likely to be missing.

A similar problem occurs inbound from the Beach in the AM peak, and I will discuss that in more detail when I turn to the Queen car.

Vehicle Assignments

Many of the vehicles providing the 2-minute wave discussed above are supposed to be ALRVs with 50% more capacity.  On paper, the effect is to have 2.5 cars’ worth of capacity every 4 minutes, the equivalent of a 1’36” headway.

In practice, many of the extras operate with CLRVs cutting the capacity of the service from the planned level, even if it did operate on time and at full strength.  This problem also exists on Queen which is supposed to be an all-ALRV service, but which may have CLRVs attempting to handle the headway and demand of an ALRV.

Another fairly common phenomenon is change-offs where a vehicle in service is exchanged for another.  This is most commonly due to equipment problems.  The number of change-offs seen in the CIS data suggests a low level of reliability by the fleet.  However, some runs quite regularly operate, day in, day out, without change-offs while others almost always involve one if not two replacement cars. 

Whether this reflects the use of the “good cars” on base, all-day runs and “bad cars” on peak trippers, or if something else is at work, remains to be seen.  I have not yet been able to look at vehicle assignments across all of the routes to determine if there are certain cars that are always showing up as change-off candidates regardless of where or when they operate.

Either way, change-offs represent vehicles that go out of service when they should be carrying passengers, and high failure rates limit the number of vehicles actually available for service.

Short Turns

I will deal with short turns in detail in a later post but will give an overview here.

Short turns are a symptom of other problems:

• Cars are late because they did not enter service on time, or because they took a longer-than-reasonable terminal layover.
• Cars are late due to congestion either of traffic or passengers.  Fighting congestion is a problem that must be attacked on many fronts, but passengers congestion is totally within the TTC’s control.  It is affected by headway regularity, adequacy of service, and fare collection strategies (all door versus pay-as-you-enter loading).
• Cars are late because a junior operator may not have the experience and confidence needed to get through peak traffic as quickly as other cars on the line.
• An accident or other blockage may hold a group of cars which must be sorted out into something resembling regular service.

There are two major problems with short turns that have been around for decades:

• Service to the termini is considered as expendible with the goal of restoring service downtown.  However, a route like King is functionally at least three separate, overlapping routes from a demand point of view.  It is a Broadview car, it is a King car, and it is a Roncesvalles car.  Very large gaps to both termini result from the common short turns at Parliament eastbound and Roncesvalles westbound.  Service on Broadview is supplemented by the Dundas car (although the current diversion for track construction lessens that benefit), but on Roncesvalles, it’s the King car or nothing at all.
• Cars re-entering service from short turns may do so in the wrong sequence (triggering the need for yet another short-turn) or may do so on a very short spacing before or (more likely) after a through car.  This pair of cars may operate across the city without any apparent attempt to space the service.

The underlying problem is that having cars on time is more important than providing good service.  Some short turns do little more than reverse the order of two runs.

There is no question that some short-turning is inevitable.  However, the TTC needs to find a way that service can be operated without a slavish need to keep operators on their own schedules that have no meaning or importance to the travelling public.  This may require a change in work hours and pay arrangements (paying an operator to be on duty, not to be at location x on run y at time z), and this will obviously need input from the ATU.

Managing a service in real time requires a good information system that tells Transit Control where all vehicles are accurately.  CIS has many failings in that regard, and its replacement must include good tracking and line management capabilities.  This will also create a need for better line management strategies and schedules that reflect actual operating conditions.

I will be posting a number of items dealing with various parts of my analysis over the next few weeks.