Analysis of Route 512 St. Clair — Part 2: Headways and Link Times

In the previous article, I reviewed the operation of the St. Clair route on Easter Sunday, 2007, as a starting point for a review of the route’s overall behaviour.  In this post, I will turn to data for the entire month that shows overall patterns and the amount of variation we might expect to find.

If the headways (the time between successive cars) range over a wide band, then service is perceived as irregular by riders regardless of what the printed timetable may say, and regardless of the “average” loads riding counts might report over an hourly period.  When headways are a mix of long and short values, the cars on long headways will carry heavier loads and the “average” experience for a rider is that they wait a long time for an overcrowded car.  The half-empty one a few minutes behind is little benefit to anyone, but it brings down the “average” load in the statistics.

Link times (the length of a journey from one point to another) reveal how predictable (or not) the time needed for a car to travel along a route will be.  If link times are consistent, this indicates that external effects (including unusual loads that stretch stop service times) are rare.  Even if the times vary over the course of a day, but do so within a predictable, narrow band, a route should be fairly easy to manage.  If the times vary a lot with no obvious pattern or are scattered within a wide band, then running times and service are hard to manage.

In reviewing St. Clair, I found that the operating environment, as it existed in April 2007, was quite benign compared to routes like King and Queen, the subject of previous analyses here.  This has important implications for the right-of-way project now underway on this route.  Congestion and random delays do play some role, but not an overwhelming one, in service quality.  Reducing the impact of congestion when and where it occurs will be beneficial, but more is needed than just getting autos out of the streetcars’ way to ensure reliable service.


As I mentioned in Part 1, the TTC considers variations from a scheduled headway of 3 minutes to be “on time” performance.  Alas, on St. Clair, the AM peak frequency in April 2007 was 3 minutes and the PM peak was 4.  A variation of 3 minutes on such small base values is meaningless, and service could be “on time” by TTC standards even if it were a complete mess.

Even during off-peak periods, a variation of 3 minutes was a substantial portion of one headway and an “on time” service would appear quite irregular to customers.

The following charts have seven pages each.  These show:

  • Headways and trends for weekdays broken into four pages (weeks 1 to 4)
  • Headways for all weekdays on one page (to show the overall spread in values)
  • Headways and trends for Saturdays
  • Headways and trends for Sundays (includes Good Friday, April 6)

The scheduled headways for April 2007 were:

  • Weekdays:  3’00” AM Peak (6’00” west of Lansdowne), 6’20” Midday, 4’00” PM Peak, 7’30” Early Evening, 8’40” Late Evening
  • Saturdays:  9’30” Early Morning, 8’00” Late Morning, 6’30” Afternoon, 7’45” Evenings
  • Sundays:  11’00” Early Morning, 9’00” Late Morning, 8’00” Afternoon, 8’40” Evenings

Westbound at Yonge
Westbound at Bathurst
Westbound at Dufferin
Westbound at Caledonia

As we saw in the Easter Sunday chart, headways westbound from Yonge tend to lie in a band about 6 minutes wide, although the amount of scatter appears slightly greater late in the week, and even moreso on weekends.  What this shows is that the service leaving St. Clair Station westbound is already running on uneven headways before it makes its way along the line to Keele.

The band smears out as we move from east to west showing the effect of cars bunching into pairs as those carrying short headways catch up with those carrying long headways.  Some unevenness is inevitable, but what we see here is that service starts off uneven and gets worse.

Eastbound at Keele
Eastbound at Dufferin
Eastbound at Bathurst
Eastbound at Yonge

Headways eastbound from Keele behave similarly, and there is a noticeably wider scatter on weekends.  By the time the service reaches Bathurst, the headways are spread out over a 10 minute band, considerably worse than the TTC’s own performance target.  By Yonge the band is even wider.

If there are issues with congestion and random delays, these should show up in the link times.

Link Times

Yonge to Tweedsmuir Westbound
Tweedsmuir to Yonge Eastbound
Bathurst to Keele Westbound
Keele to Bathurst Eastbound

The link times from Yonge to Tweedsmuir (the eastern portal of St. Clair West Station) are quite uniform except for a few days (April 4 and 12) with delays late in the afternoon, and on the weekend of Friday April 20 to Sunday April 22 with extended running times.   Data for that weekend should be ignored as the line was operating partly with buses and many runs did not make full trips.  This sort of condition confuses CIS which attempts to predict where a vehicle is based on the schedule.

Eastbound, the times are more spread out on certain days.  This appears to be due to faulty CIS data for selected runs.  Looking at the raw data, I find that certain runs misreport their location possibly because they are not picking up signpost information.  When this is periodically corrected, their link times go out of whack. 

From Friday noon April 20 through Sunday April 22, the line was split with the east half running streetcars and the west half running buses.  CIS gets very, very confused when the routes don’t operate as scheduled, and the data for that period should be ignored.

From Bathurst to Keele westbound (the part of the route in mixed traffic), there are some service blockages, a slight rise in running time for the afternoon peak, and a noticeable rise in times during weekend midday and afternoon periods. The eastbound pattern is similar.

There are two points worth noting. Overall, running times stay in a fairly narrow band except when they are disrupted by delays. This indicates that congestion is fairly predictable and headways should be manageable by a combination of service spacing and adequate running times.

What is quite noticeable is the amount of time cars spend at terminals and in St. Clair West Station.  The following charts give the round trip times from the intersection close to each terminus and back again.  Although data within the loops can be unreliable, data at the intersections appears to be valid.  This means that the round trip time gives a good picture of the degree to which cars are laying over at the terminals.

Yonge to St. Clair Station
Keele to Gunns Loop

At Yonge, the trend lines for weekdays show that cars have time for a leisurely trip through St. Clair Station.  The “cloud” of values for all weekdays spans a range of over 10 minutes and is considerably longer than the headway at most times of the day.  Pairs of cars will regularly be in the station.  Under these conditions, it is hard to understand how an irregular headway westbound of Yonge can be justified.  The situation is worse on most Saturdays and Sundays.

At Gunns Loop, the layovers don’t seem to be as long on weekdays, although there appears to be plenty of time on weekends for many cars.

The link from Tweedsmuir to Bathurst consists mainly of St. Clair West Station, and the time needed to traverse this link in each direction is an indication of the length of time spent in that loop.

Tweedsmuir to Bathurst Westbound
Bathurst to Tweedsmuir Eastbound

Westbound, this trip takes about 5 minutes quite consistently with a spread of plus or minus at most 2 minutes.  Eastbound the trip is slightly faster, but this may simply be an artifact of the CIS data.  As an interchange, this loop contributes substantially to many through trip times compared to what we would see with a straight-through arrangement.  That’s a point worth noting for future designs and their impact on travel times.

In any event, cars appear to take their major layovers at the terminals, not at St. Clair West.  As mentioned above, data for the weekend of April 20-22 should be ignored as the line was operating with streetcars on the east half and buses on the west likely due to a construction project.

The scheduled round trip running times in April 2007 for St. Clair for the trip from St. Clair Station to Gunns Loop (including scheduled recovery time) were:

  • Weekdays:  AM Peak: 72,  Midday: 70,  PM Peak: 76,  Early Evening: 67,  Late Evening: 61
  • Saturdays:  Early Morning: 57,  Late Morning: 80,  Afternoon: 84,  Early Evening: 62,  Late Evening: 60
  • Sundays:    Early Morning: 55,  Late Morning & Afternoon: 72,  Evening: 60  

Yonge to Keele Westbound
Keele to Yonge Eastbound

As I mentioned above, the exact departure times from the terminals is not always reliable in CIS data, but times at nearby points are.  These two charts show the time for the full trip from Yonge to Keele each way.  Allowing for a reasonable turnaround time at each end, the running times are quite adequate for the route.

Introduction of a full right-of-way will certainly reduce traffic interference where it now occurs, but it is not a panacea for all the ills of the route.  Proper headway management will go a long way to resolving the irregular service now enjoyed by St. Clair’s customers.

In the last part of this series, I will turn to specific days where service is disrupted to review how the line operated under these conditions.

7 thoughts on “Analysis of Route 512 St. Clair — Part 2: Headways and Link Times

  1. I myself noted the variance on this route when I lived near it. It was impossible to rely on at night as I never knew how long it would be for the next car. At Yonge Street in particular I noted odd behaviour that really seemed to depend on the driver.

    Some would sit at the McDonalds side of the loop forever (even while in the streetcar) then pull up to the outbound portion, load quicky, and take off without a second thought. Others would leave the car at the outbound loop for long periods. Others would move the car as soon as thier ‘break’ was over and sit at both platforms for a while.

    It became impossible to predict when a car would go or if there would be one waiting with doors open. The only thing that I noticed being a constant was that if you looked out the window of that McDonalds (which I went to often) at any point of the day, you’d almost certainly see at least one streetcar sitting in the inbound platform.

    Why they can’t have regular departure times from the station is beyond me, and IMHO comes down to either one of two things. A – incorrect timekeeping devices on the cars. or B – lazyness on the part of operators who want an easy ride for the same ammount of cash as the guy behind them who they’ve just screwed royally.

    Steve: To this I would add C: A management attitude that blames irregular service on factors beyond their control while ignoring basic operational issues like regularly spaced service. The +/- 3 minute rule absolves them of actually trying to provide better service by producing a (meaningless) statistic that claims most service is “on time”.


  2. Steve: “The +/- 3 minute rule absolves them of actually trying to provide better service by producing a (meaningless) statistic that claims most service is “on time”.”

    I’m curious, is the 3 minute rule in the labour agreement with the union? I expect it probably is and getting it changed might not be too easy even if everybody wanted it.

    Steve: I believe that it is a management-created number. On suburban bus routes with wider headways, +/-3 can be annoying (especially if you just miss a bus that is running early), but on a frequent city route where the headway is below 10 minutes, it’s a meaningless number.

    Relative to a 6 minute headway, this means that cars could be as close as 3 or as far apart as 9 minutes. Proportionately, that would translate to an acceptable 1-hour gap on a 20-minute service which is obviously intolerable.

    As for labour relations, the issue is to manage the headways. If everyone is decently spaced, even if they’re a bit wider than the schedule, but even, then service would appear better than usual and every car would carry as close to an even load as is possible allowing for occasional surges and genuine delays.


  3. Sorry if this comes across as rude and it likely will but I dont see why an operator would require a supervisor to tell him when it’s time to leave the terminal. I also don’t see how this is good labour relations. It’s fine and dandy when you are early, but when you have to follow that guy you get screwed. Both drivers are in the same union, why would the union only stand up for the guy who decides to leave early?


  4. What is the mechanism used to know when cars or buses leave the terminals? I assume it is either the CIS data or operators report to Transit Control saying “I’m leaving now? In any case why can’t Management (if that’s not a mythical concept at the TTC) not simply check the actual departure times against the schedule and then discuss variences with operators who are constantly off schedule. As Nick points out, it is also to the benefit of drivers if the cars and buses are properly spaced.

    Steve: The process by which operators know their schedule is antique, to put it mildly. Every day, an operator taking a car out of the carhouse writes out their schedule onto a waybill. Perish the thought that the TTC would print these in a form that could be reused from day to day. What are the little notes the operators write on their waybills? How many transfers they issued. Sigh.

    The CIS console shows the current time and the schedule deviation. However, this depends on two important things. First, CIS has to know accurately where the car is, and terminals are particularly problematic. Second, if there is no “schedule” in the sense of set times, but instead the supervisors are actually managing to a headway, then ths CIS information is more or less useless. Once they move to GPS vehicle location, it would be possible to display the headway and operators could use this to maintain spacing, but I’m not holding my breath for that change.


  5. The 3-minute variance explains the frequent bunching of cars before the reserve was built. Two cars arriving together every 6 minutes constitutes on-time service. Reinforces your point about where management’s head is.


  6. Steve said ” The process by which operators know their schedule is antique, to put it mildly. Every day, an operator taking a car out of the carhouse writes out their schedule onto a waybill. Perish the thought that the TTC would print these in a form that could be reused from day to day. What are the little notes the operators write on their waybills? How many transfers they issued. Sigh”

    I’m flabergasted! No doubt this is how the Toronto Street Railway did things in 1861 – the TTC does love its history :->. I suppose the TSR also had separate printed transfers for each route for each day – something else which seems terribly archaic and wasteful – how many unused transfers are recycled every day?


  7. Daily waybills are actually a requirement from the PROVINCIAL Highway Traffic Act. They apply to all highway carriers as well as all urban transit sytems. This ARCHAIC system is not the fault of the TTC but rather of the province. The TTC is actually working towards a system whereby they will provide the daily waybill (this was actually part of the last contract proposal) which would eliminate the need for the initial operator to write up the daily waybill. The other important part of the waybill (which you are obviously ignorant about) is the daily safety/circle check of the vehicle (another HTA requirement) which is recorded on the waybill. Before making some of the comments which have appeared so far, you should check into the background about why certain paperwork is LEGALLY required by the province.

    Steve: Sorry, but NOTHING prevents the TTC from preprinting the schedules and having the operators fill out the variable information, and as for transfer numbers, there is no way anyone can tell me that is an HTA requirement. In Vancouver, they have had preprinted schedules for ages and these include information about time points along a route, not just departure times at terminals. This is essential for the timed transfer system on some routes during the off peak.


Comments are closed.