This article is a follow-up to an early April review of the gains and losses brought by the St. Clair right-of-way and subsequent “transit priority” measures. When it was published, the TTC had just changed schedules on this route to shorten trip times in response to repairs on several traffic signal locations where that “priority” function was not working. Did these repairs actually have an effect? How well did the line operate with less running time?
Updated May 23, 2015 at 12:40 pm:
Data for weekend operations has been added to the end of this article.
- There was a definite increase in travel times in fall 2014 as compared with summer 2010 when the right-of-way operations west of St. Clair West Station began. The location and severity of the problem varied along the route with notable effects between Bathurst and Oakwood where there are many traffic signals.
- The amount of running time added in the October 2014 schedules was slightly more than the actual increase in average running time over the route from 2010 to 2014.
- Travel times were reduced after the repair of transit priority functions at several signals along the route, notably in sections with many traffic signals. This did not completely reverse the longer running times of 2014.
- Short-turns as a proportion of all service declined with the new schedules in place.
- The increased supervision produced more reliable headways.
- The improvements of October 2014 have been slightly reduced with the new April 2015 schedules that clawed back much of the additional running time.
For reference, here is a table from the first article showing changes in schedules for the route since 2007.
Generally speaking, the April 2015 changes reduced one-way running time by 6 minutes during most periods with a few, minor adjustments to terminal “recovery time”. Those “recovery” values are as much a rounding factor to make the headway come out to an even value as they are a calculated amount of time needed to deal with random delays.
A Note About The Charts
The charts tracking the values of link times (times between two points on a route) and headways (the space between cars) have been modified from the ones used in the previous article, and they are in the same format used for a recent article about the Dufferin bus.
- The data are now presented grouped by hours rather than half hours. This cuts the number of pages required to show the full day’s data in half, and smooths out the effect of having a smaller number of data points for each interval.
- The Standard Deviation values for the data are displayed on the same pages as the average values so that the degree to which the underlying values vary from the averages can be directly compared with them. Low SD values for link times indicate that travel between points takes a reliable amount of time, while high values show that some latitude is needed in the schedule to deal with a range of circumstances. For headways, low SD values indicate that most vehicles are fairly evenly spaced while higher SD values show that spacing is erratic and bunched.
The following sets of charts show the link times for trips between Yonge and Keele Streets. These locations are chosen to be clear of the nearby terminals at St. Clair Station and Gunn’s Loop where layover times could confuse the data.
Running times over the route changed from 2007 to 2015:
- April 2007: A right-of-way only on the eastern portion of the route
- July 2010: Full right-of-way
- Fall 2014: Degraded running times thought to be due to loss of working transit priority at some traffic signals
- April 2015: Repaired transit priority at most locations
- Trips on were shorter in 2010, higher in 2014, and back to 2010 values in 2015.
- The difference in 2014 from 2010 or 2015 is less than five minutes.
- There was a marked increase in mid-October 2014 for trips starting before 8:00 am.
- SD values remain in the 2-3 minute range except for two spikes in July 2010 associated with service disruptions.
- The difference between 2014 and 2015 data is lower than in the AM peak, and values have not returned to 2010 levels.
- SD values are less well behaved for April 2015 showing greater variation in trip times.
- As with the midday values, afternoon trip times have improved slightly, but have not returned to 2010 levels.
- The pattern seen earlier in the day continues.
- The rise in times in mid-October 2014 that showed up in early morning data also appears in late evening data. This suggests that something happened at that time to affect the travel of cars when traffic was light and they would normally make good time along the route. I will return to this in the detailed examination of the route below.
- The eastbound data behaves similarly to the westbound numbers with a larger improvement in running times in the AM peak than in the PM.
- The mid-October 2014 change in running time noted in the westbound data shows up in mid-October late evening eastbound data.
The amount of time lost, on average, in 2014 relative to 2010 was less than the amount by which schedules were extended in fall 2014. Similarly, the amount recouped in 2015 was less on average than the amount by which the scheduled times were reduced.
Looking at the route section by section, certain portions showed greater differences in running times than others between 2014 and 2015.
Yonge to Tweedsmuir
This is the eastern section of the route that was first converted to right-of-way operation in 2007. Even with that right-of-way, higher travel times were seen even in 2007. During the early days of this operation, there were many problems with signal priority, and the situation improves by 2010. By 2014, much of this improvement has been lost, and it is not totally regained in 2015.
Bathurst to Oakwood
There are many traffic signals between Bathurst and Oakwood, and regular riders will know this as an area where signal delays are common. This is a section where the change in travel times, even early in the morning and late at night, is evident.
Oakwood to Dufferin
This section shows consistent travel times over the period. These data are typical of segments on the route where changes in travel times over the years were low.
A Side-Effect of Extra Running Time
Trip times early in the morning showed a jump in mid-October as I mentioned above. For westbound trips, this is illustrated in the following charts.
The running times for some segments and time periods increased noticeably at Week 3, October and remained at these levels showing that they were not one-time effects of a major delay or construction. What is going on here that would affect travel times especially from 6:00 to 7:00 am when traffic is relatively quiet?
The change at that point was that the scheduled running time was increased and operators would take longer simply to stay on time even when the extra time was not required. This shows up even in the link from Tweedsmuir to Bathurst where the primary source of “delay” is the layover at St. Clair West Station. It is most pronounced in the hour beginning at 6:00am, and to a lesser extent at 7:00am. By noon, the effect has completely disappeared.
This is an example of the need to look “under the data” to see what is actually happening.
When the TTC changed the schedules to reflect actual running times in fall 2014, they also implemented a new management strategy with route supervisors to dispatch streetcars from the terminals. The combination of these changes was intended to improve service reliability and minimize short-turns.
The effect of the dispatching from St. Clair Station and Gunns Loop is evident in the following charts.
Westbound from Yonge, two changes are obvious:
- The standard deviation in headways drops well below the level of previous months. This shows that even for the same average headway, the spacing between cars has improved.
- The average headway in the hour after both of the peak periods is considerably better. This results from fewer short turns allowing more service to run through to Yonge Street rather than turning back west from St. Clair West Station.
Eastbound from Keele, the effect is similar with lower SD values and a lower average headway in the periods following the peaks when many short turns would typically occur to “sort out” the service.
Looking at the service on the inner part of the route, the improvement in SD values is still evident, but there is no change in the average headways in the post-peak hours. This contrasts with the data at the terminals confirming that short turns were, until mid-October 2014, commonly used as a response to problems that developed during the peak period. With the added running time, not to mention a change in tactics to use short turns only when absolutely necessary, riders have improved service.
The April 2015 data, however, do show that SD values at the terminals are both increasing and becoming more erratic. This suggests either that the reduction in running time has reduced recovery capabilities of the line, or that efforts at dispatching on a regular headway have fallen off during some time periods, or that a combination of these factors is at work.
The degree to which short turns are used on the route can be seen by comparing the number of vehicles leaving the terminals inbound with the number beyond common short turn points. On the St. Clair route, this is westbound from Bathurst (picking up short-turns at St. Clair West Station) and eastbound at Dufferin (picking up short-turns at Earlscourt Loop).
The values in these charts are ratios. A value of “1” indicates that the same number of vehicles left the terminal as left the short turn point further inward on the route. Some specific exceptions should be noted:
- In 2007, there was an AM peak scheduled short turn at Earlscourt Loop and so the ratio of service counted at Keele and at Dufferin are legitimately different.
- In the morning, many cars enter service westbound from St. Clair West Station and they do not appear in the counts westbound from Yonge. The 6:00 am ratios will always be less than 1 as a result. A similar effect occurs at the end of the PM peak and late in the evening when some cars run in without operating east to Yonge.
- Some of the very low values for westbound service such as early July 2010 reflect a major service disruption in a week with only 1 weekday (July 2, 2010 was a Friday). Single events have a larger effect on stats for short weeks than for those with the usual five weekdays.
- A value of greater than 1 can occur because the counts are taken for the hour at each location and it is possible for more cars to pass, say, Yonge westbound than Bathurst during the same hour.
The ratios improve quite noticeably in mid-October when the new schedules and improved supervision came into play. Some decline is evident in April 2015 suggesting that short turning has become somewhat more common, but not as widespread as in September 2014.
Weekends [Added May 23, 2015]
Weekend results show a similar pattern to weekday numbers with an initial drop from 2007 to 2010 running times as the full right-of-way went into operation. Fall 2014 saw times return more or less to 2007 levels. By April 2015, running times had improved again.
Of note in the months from September to November 2014 is the gradual increase in trip times. Schedules were changed in mid-October, and so the three months represent, a “pre”, “mixed” and “post” change configuration. Some of the additional running time on the new schedules was a direct result of the schedules, not of conditions on the route.
The primary effects are during the afternoon, especially by contrast with pre-right-of-way conditions in 2007 when left turning traffic was a major nuisance.
In these charts, data are consolidated by month rather than by week because there are fewer days and hence a lower number of data points for each grouping than on the weekday charts above.
The purple line (April 2007) generally runs highest on each chart, although some of the fall 2014 values vie with it for top spot. The turquoise line (July 2010) shows the condition just after the full right-of-way operation resumed. The red line (April 2015) shows the current status.
Spikes in the Standard Deviations correspond to delays that were big enough to influence the SD values and, in some cases, the averages as well. The spike in “Saturday Eastbound” values at 13:00 is caused by a long delay on July 24, 2010 when three cars were held eastbound east of Old Weston Road. Two of these were the only cars to complete a trip from Keele to Yonge within that hour, and so they pushed up the average and SD values. This is an example of the effect of smaller data samples where a few values well outside of the typical range can skew the consolidated numbers.