The King Street Pilot project is well regarded for the improvement in travel times it brought to transit riders, and for the large jump in ridership on the route. In past articles, I have reviewed the statistics for travel times, but another important aspect is the reliability of headways – the intervals between streetcars. Early results showed a distinct improvement, but this has not been sustained. Moreover, headway reliability outside of the pilot area remains quite erratic, especially near the terminals.
This brings us to the TTC’s assertion that if only the service would depart on time from the ends of its many routes, the problem of irregular service in the middle would look after itself. This is a completely bogus claim on two counts. First off, erratic service at terminals is the norm, and regularly spaced departures usually depend on hands-on service management by supervisors on the street. Second, service has a fair latitude to be considered “on time”, and even with this leeway, gaps and bunches quickly form that exceed TTC targets.
In theory, if travel times are more consistent thanks to the pilot (or any other transit-supportive changes), then it should be easier to keep service properly spaced. Reality is somewhat different from theory.
This article examines headway behaviour at Yonge Street for the 504 King car, as well as the combined service of the King and 514 Cherry cars. Although these are thought of as “blended” services, like all branching TTC routes, there is no co-ordination between the two routes and the Cherry cars fill gaps in the King service by accident, not by design. Beyond the limits of the Cherry cars (Sumach in the east, Dufferin in the west), the King service is as erratic after the pilot’s introduction as it was before.
Also included is a review of 514 Cherry service on the outer ends of the route. Only recently has the service to Distillery and Dufferin Loops become more reliable and the improvement has more to do with revised schedules than with the King Street Pilot.
Format of Charts
In this article, the charts have a similar format to those used in Part I, the review of travel times. Data are presented as percentiles with four separate ranges:
- 100% (red): All headways are at or below this value; in other words, this is the maximum size of a gap between vehicles for the hour and day in question.
- 85% (orange): Most headways lie below this value which can be thought of as the upper end of the wait most riders will see most of the time.
- 50% (green): This is the value at which half the headways are longer and the other half are shorter. It is the median within the data values.
- 25% (purple): One quarter of the headways lie below this value.
If the 25%, 50% and 85% values are all similar, this indicates that the lion’s share of the service is running in a narrow band of headways, and that brings reliability from a rider’s point of view.
If the 25% values hug the axis with values of only a few minutes while the median sits well above this, then the service is running in bunches (short headways followed by longer gaps). These situations will also be reflected in an 85% value that is well above the median (the gaps offsetting the bunches).
Note that in the first section below, only 504 King cars are included. This is intended to show what the service looks like by the time it arrives “downtown” as a comparison to later charts that will show what it looks like beyond the pilot and the overlay of the 514 Cherry service. These areas and services will appear in later sets of charts.
Headways at Yonge Street Westbound
The values during the AM peak bounce around a bit over the past year, and there are a few periods where the maximum headway drops down compared to most months. One of these periods is in late November 2017, just after the pilot began, and another is in March 2018. However, there is no reduction in the 85% values which have stayed fairly constant throughout. This is no particular surprise as the AM peak was not a problem before the pilot and there was little reason to expect much change.
During the midday period, there is a slight change from the onset of the pilot in that the peaks are generally lower, and there is a drop in the median headway starting in mid-December. How much of this is due to the pilot itself is hard to tell pending the addition of warmer-weather data.
The afternoon peak also shows a marked improvement in regularity with lower maxima from the onset of the pilot to mid-December, but then things return to more or less the pre-pilot conditions. It was those early results that raised hopes for the pilot bringing more reliable service, but the effect has not been sustained. Possibly there was a change in mid-December in line management. In other words, the benefit could have less to do with the pilot itself, than with the extra attention the route received during its early weeks.
Early evening service shows a wide gap between median and 85th percentile values, with a 25th percentile hugging the 1-2 minute range. This shows service that is routinely bunched, and the maxima are even worse sitting in the 10-20 minute range than the 85th percentile values.
Late evenings continue this pattern with a mixture of shorter headways (25th percentile in the 3-5 minute range) and long (85th percentile regularly above 10 minutes). Maxima above 20 minutes are not uncommon.
The eastbound charts are similar to the westbound with some improvement after the pilot begins, but the results are very uneven and it is hard to point to a change that is due to the pilot itself.
Late night service shows the effect of a schedule change in mid-February when all streetcar operations shifted from Roncesvalles carhouse to Leslie and Russell in the east end. The more-frequent early evening scheduled service now extends beyond 10 pm eastbound at Yonge although this is commonly a through car immediately followed by one that is headed for a carhouse.
Another way of looking at the data is to track the behaviour of only the 85th percentile, in effect, a value that reflects what most riders see while clipping off the spikes caused by major delays.
For the AM peak, the range of values for each hour from 6-10 am stays generally in the same range.
The PM peak data shows the same drop in mid-November that was associated with the onset of the pilot, and another improvement in March, but there are comparable data in early 2017 and so this cannot be credited to the pilot project. The real trial will come through the warmer months when 2017 data showed an increase in the 85th percentile values.
Eastbound data show similar patterns.
The charts above are in pdf format below.
The Effect of the 514 Cherry Service
The charts above do not include 514 Cherry cars. To get a sense of their contribution to service in the core, I created a consolidation of data for the King car alone and for the combination of King plus Cherry.
The chart below covers only the first three months of 2018 whereas those in the previous section also included all of 2017. This gives the data a bit more “elbow room” so that one can easily see what is happening.
The solid lines are the 504 King data, the same as are charted in the previous section, but over a longer interval. The dotted lines show the effect of the additional 514 Cherry service which pulls down the 85th percentile, and also makes some dents in the 100th percentile peaks.
The result is that riders whose destination lies between Yonge and Dufferin see quite frequent service through the peak 5-6 pm hour.
The pattern is similar through the day, although less so in the late evening when 514 Cherry service is infrequent. The full chart sets are linked below.
The result of the combined services is that even though headways are less than reliable on each route individually, they combine to provide a tolerably frequent service in the central portion of the route. Things may seem all right downtown, but the situation on the outer areas is less attractive, and puts the lie to the TTC’s claims about service spacing from terminals.
Headways Beyond the Pilot Area and the 514 Cherry Overlay
Headways from Dundas West Station inbound are erratic and show little effect from the pilot beyond the already-noted seeming improvement just after implementation. The range of values widens as the day goes on and is particularly bad in the late evening. The 85th percentile routinely lies above the 10 minute line with the 100th percentile regularly well above 20 minutes. This shows not just the erratic service on the central part of the route, but also the effects of short turns at Roncesvalles & Queen.
The midday chart is shown here, and the full set linked below.
A bit further in on the route, at Jameson, the service includes cars that were short turned, and the headway percentiles are lower than at Dundas West Station. Even so, high maximum headways are still common particularly in the evening.
The situation from Broadview Station inbound is similar to that at Dundas West. East end service suffers from short turns at Parliament and, to a lesser extent, at Church.
The 514 Cherry service began in mid-June 2016, and service at the terminals has been erratic since opening day. The King Street pilot produce some improvement in the short term in November-December 2017 (as it also did for 504 King), but the headway reliability declined after the first month. Schedule changes in mid-February 2018 have substantially improved service quality at the terminals during many time periods.
One must ask why it took the TTC so long to address service quality on this route with appallingly wide headways a common situation during almost all operating periods. This shows how service on the tail ends of the route south of King really didn’t matter, but raises the larger question of why a route touted as serving a new neighbourhood and a tourist destination was treated as an afterthought, as extras for the King street line.
Note that the charts for 514 Cherry have a different maximum Y-value of 60 minutes compared to 30 minutes used for 504 King. This was necessary to display the very common high values found in the Cherry route’s headway data.
Westbound from Distillery Loop
Service from Distillery Loop starts out well enough in the AM peak although gaps of twenty or more are distressingly common in a peak period service.
By midday, however, the service on Cherry Street completely falls apart with very wide gaps. A few effects are quite obvious in the chart below:
- For a period at the beginning of 2017, service was much better-behaved than usual.
- A schedule change in May 2017, intended to provide greater running time at the expense of wider headways, did not tame the erratic nature of service which, if anything was worse after than before.
- The benefit of the King Street Pilot is evident from mid-November, but it ends abruptly in mid-December. There is no explanation for this other than possibly the removal of added line supervision that had been laid on for the pilot.
- The schedule change in February 2018 brought both lower scheduled headways and a much more reliable service.
The PM peak shows many of the same patterns as the midday service except that the schedule change in February 2018 was not as substantial during this period of the day.
The early evening shows some improvement similar to the midday service.
Late evening has a few oddities in its chart:
- The embargo on service south from King during 2017 shows up as a wide notch in the chart. For a time, one car might venture down to Distillery Loop just after 10 pm, and this shows up as a 100% value for the headway. Service, such as it was, by the WheelTrans shuttle bus does not show up here because these vehicles were not tracked as part of the 514 service.
- With the schedule change in February 2018, some cars after 10 pm run in to Leslie Barns from Distillery Loop. Typically these leave the loop close behind a car headed downtown, and this shows up as a very short 25th percentile of the infrequent service during this hour.
Weekends have different patterns because there are no “peak” periods but they do share with weekdays a well behaved service early in the day followed by more erratic service in the afternoons and evenings.
Eastbound from Dufferin Loop
The situation at Dufferin Loop eastbound is somewhat different in that very wide headways are more common here because the erratic nature of service is compounded by short turns before cars even reach Liberty Village. There is a noticeable improvement with the new schedules in February 2018.
As at the east end of the route, the AM peak is relatively well-behaved, although with many days showing maximum headways of 20 minutes or more. There is a quite noticeable improvement since the schedule change in February 2018.
Midday shows a pattern we saw in the east end as well with a marked decline in service quality following schedule changes in May 2017, a brief improvement with the King Street Pilot in November 2017, and finally a return to more reasonable headways with the February 2018 schedules. Although the orange 85% line only peeks out from behind the red 100% line, note that it was routinely above the 20 minute mark from May 2017 to February 2018.
PM peak service shows many of the same patterns as earlier periods and again with very common values above the 20 minute line.
Early and late evenings show the improvement due to new schedules in February 2018.
The full charts sets are linked here:
Schedule History for 514 Cherry
- June 19, 2016: Service begins with CLRVs scheduled
- September 4, 2016: A mix of CLRVs and LFLRVs is scheduled, but headways and running times are unchanged.
- May 7, 2017: Service scheduled for LFLRVs. Weekday midday headway widened from 10′ to 15′ and round trip time reduced by the removal of recovery time. Headways and running times unchanged during other periods.
- February 18, 2017: Weekday midday and early evening service improved by restoration of the pre-May 2017 CLRV headway and recovery time. Sunday afternoon service improved.