Analysis of 512 St. Clair for Jan-Feb 2010 (Part 1: Headways)

In the previous series of four articles about the St. Clair car, I review operations in January 2010.  Quite noticeable within the data for January was a change in line management style in the last week of the month with a more regular pattern of short turns and generally more reliable service.

On February 14, the TTC implemented new schedules with longer running times in an attempt to overcome problems with the original versions.  Headway reliability on weekdays improved slightly, although the largest change had already come with the new line management scheme.  On weekends, although the worst of the bunching and gapping was reduced, there continued to be wide variation in vehicle spacing suggesting that the level of management necessary to ensure reliable service was not present.

I have now received February 2010 data for the St. Clair route, and this series of articles will review what it shows about the effect of the new schedules on various aspects of the operation.  Here I will review headways at key points, and in future articles I will look at link times, general line management and intersection delays.

To present the headway reliability data for the months of January and February, I have consolidated various charts over that period for easy visual contrast.  Eastbound service is shown at Dufferin and at Tweedsmuir (just east of St. Clair West Station), while westbound service is shown at Yonge and at Bathurst (just west of St. Clair West Station).

2010.JanFeb Yonge WB
2010.JanFeb Bathurst WB
2010.JanFeb Dufferin EB
2010.JanFeb Tweedsmuir EB

Each set of charts contains eight pages, one for each week in the interval.  (Note that the third week of February contains Family Day, a statutory holiday, on the 15th.  This day is included in the Sunday data below.)

For each location, what is quite striking is the extremely broad scatter, especially east of St. Clair West Station, in values during early January.  This settles down and by month-end the service is somewhat more reliable.  I say “somewhat” because there is still a range of about ±5 minutes around the values.  Given that the locations are all near points where service might be being monitored (terminals plus St. Clair West), this is troubling.

Another notable point is that the trend lines through the data are quite consistent to the point that they lie almost on top of each other.  This indicates that large, random delays had little effect on the service.

Consolidating all of the weekdays onto a single chart shows the marked difference between January and February.  In the next two charts, the data for two locations in each direction are presented in one file.

2010.JanFeb WB Weekdays
2010.JanFeb EB Weekdays

At all four locations, the “cloud” of data points for January is much more spread out than for February showing the extremely unreliable service of the route’s early days.  Service east of St. Clair West (Yonge westbound, Tweedsmuir eastbound) is noticeably worse than to the west because of short turning, and this persists, albeit on a reduced scale, in February.

Weekend operations should raise flags at the TTC.  (Note that the Sunday charts below include New Year’s Day, January 1, and Family Day, Monday, February 15.)

2010.JanFeb WB Saturdays
2010.JanFeb EB Saturdays
2010.JanFeb WB Sundays
2010.JanFeb EB Sundays

During January, service west of St. Clair West station was better than east of it, as we have seen on weekdays, but headways much above and below the scheduled value were quite evident.  The situation improves in mid-February, and the effect is more pronounced on Sundays.

Note:  The wider headways shown on Sunday, February 28 in the evening are due to an emergency that blocked service west of St. Clair West Station for roughly an hour starting at 1845.  Some service was provided by shuttle buses extended east from Oakwood, but that is not included in the charts here because most of the buses do not yet have GPS units feeding information to the central system.  Only streetcar runs are shown in the headway charts.

The operations on the January schedule show quite clearly the combined effect of widespread short-turning and wider scheduled weekend headways.  When one might expect a larger proportion of “choice” riders to be trying out the new line, service was appallingly bad.

Equally troubling is the continued scatter of headways on weekdays even with the new schedules.  The TTC considers service to be “on time” when it is ±3 minutes of schedule.  However, on scheduled service of 5 minutes or better, this represents a wide variation in the perceived headway together with bunching and uneven vehicle loads.

4 thoughts on “Analysis of 512 St. Clair for Jan-Feb 2010 (Part 1: Headways)

  1. I got the data up on over the weekend. So if you want to see this data on a map check it out. It also can generate speed/distance and route location charts along with vehicle idleness. I’m in the process of generating service overlays. These overlays figure out how far you can travel in 40 minutes using ttc or walking from 1500 points in toronto and then create a “heat map” to show areas with bad coverage. You can see a demo of this on the site now, in the near future I will have them for every hour. I’m looking forward to getting more routes every month to do this analysis on.

    As always feedback is welcome.


  2. I find this and your previous analyses to be fascinating. But if I understand your methodology correctly I do think you put too much emphasis on average headways and not enough on the variance in headways. No doubt your regular readers are quite aware that average headways reflect nothing other than the number of vehicles on the line and are therefore unrelated to line management issues – but the untutored might miss that. Further, if the line management question is “How can I minimize average passenger wait time, given the number of vehicles I have in service” the answer is “minimize variance in headways”. So a graph that shows a polynomial spline of headways is nearly invariant from day to day does not really “indicate that large, random delays had little effect on the service” so much as that they have the same number of cars out there each day. A graph showing the evolution of headway variance over the typical day would be more informative to the issue at hand I think.

    Steve: On the contrary. I make a point of noting the breadth of the “cloud” of data points around the trend lines. It is quite striking how these vary between the first three months of January before line management was improved, and to a lesser extent with the new schedules in mid-February. When I speak of large random delays, I mean delays of 20 minutes or more which are always sufficient to pull the polynomial line away from the regular pattern. (On some occasions I remove the data points as they cause a totally meaningless attempt at threading an order-6 curve through the data, but in this case I have left things untouched.)

    The small “delays” are due much more to operational style, traffic signals, and a not-inconsiderable amount of time having a chat with route supervisors who seem to induce delays of their own in the service.

    I could produce a graph of variance, but given that there tends to be a fair amount of scatter in all of the data, I think that information is clearly visible to anyone who looks at the graphs. At some point, I have to give my readers credit for some intelligence.


  3. I hope the TTC manages to get this business solved PDQ. Anyone who read yesterday’s Star will know that Royson James has added his two cents worth arguing that the new Transit City lines will just be streetcars on PRW like St. Clair and Spadina and then, wait for it, they are second best and should be scrapped in favour of subway lines even if such lines can only built at the rate of one station per year.

    The man is clearly an idiot but sadly, he does have his pulpit and someone needs to set him straight on the difference between the heritage lines like St Clair and Spadina and the Transit City lines.

    That said, it wouldn’t hurt if the bad publicity from St. Clair was brought to and end.

    On a related note, the other night we crawled up Spadina from Bremner Blvd. to Bloor Station. The streetcar ahead was out of sight and I can only presume that the operator of our car (though not necessarily his passengers) was ahead of schdedule.

    No passenger should have to experience that!


  4. Steve: Move this comment to wherever appropriate if you decide to allow it.

    Today I looked at the Gunn’s Loop area track work. I noticed the eastbound road has only two lanes yet, at a public meeting I attended the traffic planner said there would be an extra long left turn lane from St. Clair onto Gunns Road to handle the traffic going into the new big box development on the old Swift’s site. What happened to that?

    Steve: Not sure, although planned roadwork west of Gunn’s Loop was put on hold pending a final decision on work to extend the streetcar line westward. It’s possible that this extra long lane is part of that future project.


Comments are closed.