Analysis of 512 St. Clair Operations for July 2010 — Part II (Link Times) (Updated)

In the first article of this series, I gave an overview of the data for one day’s operation on the 512 St. Clair route in July 2010.  Here, I will review link times (the time taken to get from one location to another) on the route for the entire month, and compare this with data from April 2007 when the St. Clair car last ran over its full length from Yonge to Keele.

Updated December 29, 2010 at 9:50 am:

A set of charts has been added comparing the running times between Keele and Yonge for April 2007 and July 2010.  See the end of the article for links and commentary.

First we look at the entire route, except for the terminal portions east of Yonge and west of Keele.  This eliminates variations caused by layover time from a review of running time across the route.

2010.07 Westbound From Yonge to Keele

2010.07 Eastbound From Keele to Yonge

[A note re the scale:  I have left the maximum value for the Y-axis on these charts fairly high so that the outliers in the data are visible.  Occasional values that lie off of the chart have been dropped so that they do not pull the trend lines away from the main body of data points.]

Each set of charts contains seven pages holding:

  • Weeks 1 through 4 of the month including trendlines.
  • All weekdays to show the overall pattern.
  • All Saturdays, excluding those where service did not operate over the entire route.
  • All Sundays and holidays, excluding those where service did not operate over the entire route.

Several points are evident here:

  • There is very little effect caused by “peak periods”.
  • Weekday times lie in a band roughly 10 minutes wide.  Given a trip time of roughly half an hour, this shows a considerable amount of variation although it is not concentrated in any particular time of day.
  • Saturdays have running times comparable to or slightly greater than weekdays.
  • Sundays and holidays have the shortest running times.

Comparison to April 2007 data shows what has changed and what has not.

2007.04 Westbound from Yonge to Keele

2007.04 Eastbound from Keele to Yonge

There is not much change in the running times from 2007 to 2010 on weekdays, although a peak period effect is more noticeable especially in week 4 of April 2007 when construction had started.  What is quite striking is the improvement on weekends when traffic restrictions on St. Clair tend to favour cars over transit (turns, parking) and the effect of the midday traffic buildup is obvious.

On a related note, the TTC is improving service on the 512 St. Clair route in January 2011 to the point that afternoon headways will be better on weekends than on weekdays (6’00” M-F, 4’00” Sat, 5’15” Sun).

From a traffic point of view, the busiest part of St. Clair lies west of St. Clair West Station.  Here are the comparative charts for 2010 and 2007.

2010.07 Westbound from Bathurst to Keele

2010.07 Eastbound from Keele to Bathurst

2007.04 Westbound from Bathurst to Keele

2007.04 Eastbound from Keele to Bathurst

Again, these charts show some effect from peak period congestion and longer stop service times, but the big differences come on weekends.

(For a further look at April 2007 data, see my article on link times and headways for that period.)

At St. Clair West Station, the time needed to pass through the loop is fairly consistent.  In the graphs linked below, the interval measured is from Bathurst Street to Tweedsmuir (the east portal).  Note that eastbound trips will include stop service time at Bathurst because this is a farside stop.

2010.07 Westbound from Tweedsmuir to Bathurst

2010.07 Eastbound from Bathurst to Tweedsmuir

At the terminals, cars often have time for an extended layover.  In the charts linked here, the time plotted is the round trip from the nearest major intersection through the loop and back.  This eliminates problems with deciding when a car has actually “turned around” because they do not take layovers in consistent positions.

2010.07 Terminal times at Yonge

2010. 07 Terminal times at Keele

At St. Clair Station, the round trip time from Yonge eastbound to Yonge westbound has a lower bound of about 2.5 minutes.  This is most evident on page 5 of the charts where all weekdays are shown together and only one data point (and that one late at night) lies below this level.  What is quite striking is that the upper level at 10 minutes and beyond is well-populated.  The trend lines hover at about 6 minutes, and the larger values will in part come from variation in travel times across the route described above.

On weekends, the situation is roughly the same.  The weekend of July 17-18 shows very long layover times at St. Clair Station because a truncated route was operating on the regular schedule, and cars had much time to kill on every trip.

The situation is similar for the round trip from Keele west to Gunn’s Loop and return.  Notable events include:

  • The delay on July 5 (discussed in the previous article) where three vehicles held eastbound at Keele due to a delay at Old Weston.  This shows up in the rise in the pink trendline for that date caused by three data points that are off the top of the chart.
  • Long layovers during the AM peak on July 23 caused by the route splitting at St. Clair West Station and cars arriving quite early back at Gunn’s Loop.
  • Extraordinarily long layovers on July 31 during the afternoon and early evening.

I will review these events in more detail in another article.

Although one could claim that the right-of-way was of little benefit, especially considering its cost, what we do see is that the street had very little reserve especially during periods when traffic regulations were relaxed or when construction removed some capacity.  As the street redevelops and local demand increases, the road (like many others in the GTA) would become constrained more of the time and transit would suffer.  That is the heart of the pro-right-of-way argument, that the streetcar line will grow in importance as overall population and travel demand require more of the transit system.

In the next article, I will turn to the question of headway reliability.

Update:  Link Time Comparisons

In response to a comment left in this thread, I have created sets of charts to compare data from 2007 to 2010.  Here are the results.

2010 vs 2007 Westbound from Yonge to Keele

2010 vs 2007 Eastbound from Keele to Yonge

The first page of each set shows the average link times for each hour for six sets of data.  For each year, there are weekday, Saturday and Sunday averages.

The dark and light blue lines contrast the weekday data where the greatest difference shows up in the PM peak.

The purple lines with square and star data points contrast the Saturday data where the difference in running times pre- and post-right-of-way is much more noticeable.  The 2007 Saturday data is the worst of the lot, but it is worth noting that even in 2010, Saturdays are slightly slower than weekdays.

The green and brown lines contrast the Sunday data where, no surprise, the running times are shortest of all.

Following pages show the details for individual days grouped by type of day to allow easy comparison of 2010 to 2007 by flipping between adjacent pages.  Note that the number of data points in a few cases is very small because of short turns that prevented many trips from reaching the ends of the line.  Values over one hour have been dropped from the data.

(The data for Friday, April 20, 2007 from noon onward have been dropped.  Operations were somewhat chaotic and the “old” CIS system, not GPS based, introduced a number of rogue movements into the data.)

What is visible here is that overall the right-of-way has lowered the running time over the route and this benefit is most pronounced for periods when the pre-right-of-way St. Clair was congested.

5 thoughts on “Analysis of 512 St. Clair Operations for July 2010 — Part II (Link Times) (Updated)

  1. I think you are reaching a very important conclusion that has been rumoured but not verified until this work, namely that link times with the dedicated ROW have not improved much between April and July, other than on weekends. Again, I would like a simple table that summarizes this and provides just enough data to demonstrate this. For example, calculate average link times (Yonge-Keele, or Keele-Yonge) during July and April for 5 times: Weekday 7-9AM, weekday 12-2 PM, weekday 5-7 PM, Sat 12-2, Sun 12-2.

    I await the comparative headway analysis with great interest.

    Armed with these results in simple graphics and metrics, we could then do two things: (1) discuss with TTC what operational improvements on St Clair would help ( eg more traffic signal priority?, arming inspectors with NextBus? ) (2) try to reach a higher degree of comfort that the planned surface LRT lines like Sheppard (assuming they survive) will be “better”.

    Steve: I am also planning articles on Spadina, Dundas and Carlton so that we can compare the behaviour or routes of different characters. One important point for St. Clair vs TC lines is the question of stop spacing and the degree to which streetcars are routinely held at traffic signals (whether there is a station there or not). Operating speeds are constrained to some degree by the amount of traffic (motor and pedestrian) routinely crossing the line, and it is impractical to run at a higher speed.

    The average speed, however, is affected as much by the time a vehicle spends stopped as on the speed it travels when it is moving given the nature of urban transit routes.


  2. Hi Steve,

    Great article, I’m curious – do you share your results directly with the TTC at any point? It seems several improvements may result in great saving at little to no cost. To be fair I think there are some intrinsic problems that have no ideal solution and you hinted at several of these.

    On the other hand, is there too much politics (in whatever form) which interferes from you sharing results / suggestions any improvements? I’m inclined to believe this is the case unfortunately.

    Steve: I have offered to meet with TTC staff to share my methodology and my observations, but this offer has not been taken up. There are severe organizational barriers within the TTC: programming work such as I had to undertake to analyze and present the data would be done by the IT department, the people who want the info work in Service Planning, and the people who could benefit from a review of their techniques are in Operations.

    I remember visiting Vancouver during the early days of SkyTrain, and they were producing this sort of analysis from their first day of operation to track how the system behaved and how their control strategies worked (or not). This is not a new concept.

    When the TTC’s vehicle monitoring system was implemented decades ago, this sort of analysis was one of the benefits cited for the new system. However, they ran short of money, and the benefits were not understood by the very groups who would make use of them. The result — that part of the system was never implemented.


  3. It sounds like there is a big difference between here and Vancouver. The Vancouver authority care.

    In organizations which care about the results, the internal barriers that you talk about are removed. I’ve been on tiger teams where you get people from different parts of the organization together to work on a single problem. It’s a lot of effort, but it works. However it needs a group of people who care about the problem and support from senior management to allow it to happen.


  4. Thank you for the comparative charts. My observations:

    -too bad this data is not for the same months, eg July 2007 vs July 2010 rather than April 2007, but we will take what we have

    Steve: It’s a function of what happened when. I asked for April 2007 way back when as it was the last month of operation to Keele, and was clear of the winter storm effects. July 2010 is the start of service. Many of the observations (especially headways as you will see in the next article) don’t need exact month-to-month comparisons to be valid.

    -pleased to see that for all days and periods (except Sat around 21:00), the running times are now lower

    Steve: Saturdays in July 2010 seemed to be afflicted by a lot of activity on St. Clair which may or may not last into the fall/winter season. Only when I get some data for other months will we know. There are seasonal fluctuations on many routes. For example, week nights downtown see a lot of traffic (road and transit) on routes through entertainment districts, but this is sensitive to the day of the week and to the weather.

    -the running time reductions range from 0 mins (typically early morning and late nite) to 6 mins. This is on an average trip time of about 20-25 mins (early morning and late night) and 25-33 mins the rest of time. So I think it is fair to say that trip time reductions of 10-20% are occuring during non-quiet times. Not bad, in view of the facts that the stop spacing, speed limits, and passenger demand have not changed.

    Steve: Based on schedule changes going into effect in January, demand is actually up on weekends.


  5. Well, the use of a driverless system more or less guarantees that all of the tools for monitoring and improving service patterns on a line reside with a single team of engineers instead of being scattered across departments. I’d be more interested to see what kind of analysis the Vancouver authorities do for the bundle of (extremely heavily used) bus services along Broadway, for instance.

    Steve: I agree, but the important point was that they wanted to do this from day one. The TTC could have done this with CIS, but chose not to, and they have had decades.


Comments are closed.