When the St. Clair right-of-way went into operation after an extended construction period and a lot of political upheaval, streetcar operation was scheduled to be faster than the old mix-traffic model. The TTC even produced a before & after comparison that is still posted on their Planning page (scroll all the way down to “Miscellaneous Documents”).
Alas, the 512 St. Clair is now scheduled to operate more slowly than in pre-right-of-way times. This article reviews the evolution of the route since July 2010 when it fully opened from Keele to Yonge to early 2021.
This is a long article, and I will not be offended if some readers choose not to delve into the whole thing. My intent in part was to show the level of analysis that is possible with a large amount of data stretching over a decade, and also to examine the issue in some detail.
As a quick summary:
- Scheduled travel speeds for the 512 St. Clair car have slowed since the right-of-way opened in July 2010, and they are now below the pre-right-of-way level in 2006.
- There was an improvement in 2010, but this has been whittled away over the decade with progressively slower schedules.
- Separately from travel times, scheduled terminal recovery times have increased from 2010 to 2020 especially during off peak periods. This does not affect speed as seen by riders, but it does show up in longer terminal layovers. This recovery time now accounts for a non-trivial portion of total time on the route.
- Driving speeds are slower in 2020 (pre-pandemic) than in 2010. This is a characteristic across the route, not at a few problem locations, and is probably due to differences in how the new Flexity cars are operated compared to the predecessor CLRVs. A few location, notably the constricted underpass between Old Weston Road and Keele Street, have seen a marked decline in travel speeds over the decade.
- Many locations have “double stop” effects where streetcars stop nearside for a traffic signal, and again farside to serve passengers. Transit signal “priority” clearly needs some work on this route.
It is important to stress that this gradual decline in speed does not invalidate the right-of-way itself. Routes without reserved lanes have fared worse over the past decade, and St. Clair would certainly be slower today without them. The big challenge, especially with pandemic-era ridership declines, is to maintain good service so that wait times do not undo the benefit of faster travel once a car shows up.
The charts below show the scheduled speed over the line from 2010 to 2021 with 2005 (pre-construction) shown at the left side as a reference point. The information is broken into two charts to clarify situations where there are overlaps.
In 2005, the AM and PM peak values were the same, but from 2010 onward the PM peak had a slower scheduled speed. In the off-peak, the midday and early evening speeds are the same from 2010 until 2018 after which midday speeds drop considerably.
The big dips in the charts correspond to periods of construction when travel times were extended to compensate.
The transition from CLRV to Flexity service began in 2018, and by September it was officially recognized in the schedule.
Schedules are one thing, but what is the actual “on the ground” behaviour of the route. Here are two charts showing the evolution of travel times between the two terminals westbound in the 8-9 am and the 5-6 pm peak hours. Regular readers will recognize the style of the charts, but there are several points worth mentioning.
- The data run from July 2010 when the right-of-way was completely open to February 2021, although there are gaps. I did not collect data in every month over the period. However, the overall pattern is fairly clear. Unfortunately, I did not collect any data between July 2010 and September 2014 and yet there is a clear jump between the two.
- Travel times build up to late 2019 and remain high to January 2020. Then comes the pandemic and the times fall, but not by much (the change is much more noticeable on other routes that operate in mixed traffic).
- There are upward spikes in values. A few of these are caused by delays that affect several cars so that even the median value (green) rises. However, if only one car pulls onto the spare track at St. Clair West and lays over, this pushes the maximum (red) way up while leaving the other values lower. (Layovers can also occur at Oakwood Loop, and at Earlscourt Loop eastbound.)
- Occasional downward spikes of the minimum values (blue) do not represent supercharged streetcars, but rather bus extras that ran express for at least part of their trip.
- When comparing these value to the scheduled speeds above, there are subtle differences:
- The scheduled speed is based on end-to-end travel including arrival and a short layover, notably for passenger service at St. Clair Station. “Recovery time” (about which more later) is not included in the scheduled speed calculation.
- The travel time is measured between two screenlines: one is in the middle of Yonge Street, and the other is just east of Gunn’s Road so that the entire loop is west of the line. This does not include any terminal time at either end, but does include layovers, if any, at St. Clair West Station Loop.
Here are the corresponding charts for eastbound travel.
Full chart sets including midday and evening travel times are in the pdfs linked below for those who are interested.
These charts show changes have occurred, but where and why?
How Much Time Is Spent at St. Clair West Station?
St. Clair West Station presents an opportunity for a layover, although this is limited to the degree that eastbound and westbound cars might push each of the “out of the way”. More recently, with the introduction of Flexity streetcars, have they moved more slowly through the station?
Here are charts for peak hour travel times between Tweedsmuir (the street just east of the east portal) and Bathurst. In both cases the screenlines are in the middle of the intersection and so the times shown here include dwells at stops and traffic signals.
The median values rose from about 4 to 5 minutes from 2010 to 2021 although the effect was slightly greater for eastbound than westbound trips.
The spikes in values correspond to cars laying over within the station for an extended period. These are less common in data for later years in the series.
The full chart sets are here:
How Much Time Is Spent At Gunn’s Loop?
A common problem on many routes is that vehicles accumulate at terminals because they arrive early. Trips across the route might have a median time of 30 minutes, but with a span from 25 to 35 minutes. Schedules are based on the high end of values (the 95th percentile), and this leaves most cars with more time than they actually need. The situation is compounded by reduced travel times in the pandemic era although on St. Clair this effect is small compared to routes running in mixed traffic.
The time spent at Gunn’s Loop varies over roughly the same range of times as the travel times, a band of values roughly 10 minutes wide. Spikes correspond to vehicles laying over (usually buses). Periods where the times rise above typical levels correspond to schedules with added running time for construction delays.
Looking only at the peak periods, the values stayed fairly consistent, but in the mid-evening, something odd happens at Thanksgiving weekend, 2020. There is a jump in median values that persists to the end of the chart.
(The large spike in 2018 was caused by a late-evening transition from bus to streetcar operation for overhead upgrades along the route. Buses tended to arrive at Gunn’s Loop and layover before entering service.)
The travel time chart across the route shows a similar change. Note that I have used the previous hourly interval because the chart captures vehicles leaving Yonge Street between 8 and 9 pm. They will reach Gunn’s Loop later, and so the effect of shorter travel times and longer terminal layovers will show up there later.
Note also that for the period of the September schedules, travel times are higher and more tightly clustered. There is a corresponding dip in terminal times. This pattern only exists for the mid-evening period and only for one schedule period (Labour Day to Thanksgiving).
Although there was a new schedule implemented on Thanksgiving weekend, there was no change in the headways, travel time or terminal recovery time. Below are the scheduled service summaries showing that the only change was that temporary operation of the route from Hillcrest ended. This should have no effect on travel times.
The full chart set is here:
How Much Time Is Spent At St. Clair Station Loop?
The pattern of changes in terminal time at the east end of the route is similar to that at the west end shown above. However, the total time in these charts is higher probably for two reasons:
- At Gunn’s Loop, the terminal area for the time calculations is bounded by a screenline just east of the loop. At St. Clair Station, the screenline is at Yonge Street and the loop is much larger.
- St. Clair Station has better “creature comforts” than Gunn’s Loop, and if an operator has time to spare, it is a more likely location for a layover.
Here are the peak hour charts. There is a marked rise in terminal times at the beginning of the pandemic corresponding to reduced passenger demand and faster travel times across the route.
The mid-evening pattern at Gunn’s Loop also shows up at St. Clair Station.
A few notes:
- The high values in April 2018 were caused by the streetcar-to-bus transition and the early arrival of a bus at the station. Only the maximum value spikes, not the 85th percentile indicating that this was only one vehicle considering the low number of observations within one hour.
- The dip in terminal times for the September 2020 schedule period corresponds to the jump in eastbound travel times similar to the westbound data shown above.
The full chart set is here:
Separately from travel time on a route, TTC schedules provide for recovery time at terminals. This concept has evolved over the years, and is partly bound up with the labour contract. The TTC did not want to guarantee a break at the end of each trip as a scheduled event, especially if drivers would be entitled to take their break regardless of being on time or late. Moreover, a guaranteed recovery time would add to vehicle requirements assuming headways remained the same. For example, on a route with a five minute service, a guaranteed break of 5 minutes at each end of the line would cost two extra drivers and vehicles.
Instead, “recovery time” was usually a result of making the headways come out evenly. For example, if a trip took 57 minutes, but an even headway was desired, this would be achieved by adding 3 minutes of recovery time. This is particularly important for branching services where each branch’s round trip time must be a multiple of the headway (or of half the headway where vehicles alternate between branches). The amount of “recovery” time had nothing to do with actual driving conditions and everything to do with making the schedule work out properly. Zero recovery times were not uncommon.
In recent years, the TTC has changed to scheduling service so that short turns are almost impossible in response to political pressure. While this is a noble goal, simply making running and recovery times longer is not necessarily a productive use of vehicles, especially when little attention is given to managing service and providing evenly spaced vehicles. Indeed, having too much time can encourage operators to pay less attention to the schedule because they know there is padding enough to make up for any delays or extra layover time they might take.
The TTC has no published standard for the ratio of recovery to travel time, nor of how this might be adjusted to account for varying conditions on routes.
As we saw earlier in the article, the scheduled speed has declined on the St. Clair route, but what has also happened is that recovery times are now considerably larger, especially in the off-peak. Considering that the route is roughly half an hour each way, the recovery time has grown quite large, an ironic situation for a route with a private right-of-way over almost all of its length. There are much longer streetcar and bus routes that do not have such generous schedules. As of February 2021, there are periods when over 20 per cent of the total time for a round trip is dedicated to “recovery”. This is a waste of resources.
Driving Speed Comparisons
Driving speed is not the same as scheduled speed over a route. The scheduled value is an average over the distance, and is selected to allow for almost the worst case of slow trips caused either by weather, minor delays or inexperienced operators. If the scheduled times are enforced (which for the most part they are not), then vehicles travel at a lower speed than they would otherwise, riders get slow trips, and operators are forced to waste time all along a route. This is not exactly a good advertisement for transit as an alternative to driving.
Buses tend to drive faster than streetcars for various reasons including the conditions bus drivers face on most major bus routes – fast suburban arterials – and an absence of restrictive operating practices that have accumulated on the streetcar system over many years under the rubric of “safety”. (That issue is a discussion for another time.) St. Clair is less subject to these practices because it is straight and has few junctions where speed restrictions through special trackwork apply.
In any event, the speed at which vehicles travel at any location is based on a combination of traffic conditions and operator experience, plus any added constraints.
The charts below are a bit “messy” and I will tease them apart to reveal more detail. They compare average driving speeds along the route in each direction by hour of the day and looking at six separate two-week periods from 2010 to 2021. The dates have been chosen to avoid the effect of construction and other major disruptions.
In the westbound charts below, travel is left-to-right from St. Clair Station to Gunn’s Loop. The solid lines show the average speed point by point along the route (each point is 10m wide), whereas the dotted lines are interpolated across the route to show the overall pattern. The pink line (2010) is the highest while the purple line (2020) is generally the lowest.
Each dip in the chart corresponds to a location where vehicles stop. The vertical lines represent streets, and they lie in the middle of the crossing. Notable at many locations is the “double stop” effect of a nearside stop for a traffic signal followed by a farside stop for passenger service. We talk a lot about “transit priority” in Toronto, but are not always reliable in the delivery.
Here is the chart for the pm peak hour:
Finally, here is mid-evening.
Eastbound data show similar patterns. Note that these charts should be read from right to left, the direction of travel.
The full chart sets with data by hour from 6 am to midnight are linked below.
You can watch the evolution of driving speeds over the day by stepping from page to page, effectively a flip-chart animation.
2010 vs 2020 Speeds
To show the change over the past decade, the charts below contain the same data, but with only the 2010 and 2020 data shown. A common observation across these charts is that the peaks are higher in 2010 than in 2020. This probably corresponds to the different characteristics and driving styes for the Flexity streetcars compared to the CLRVs which drivers tended to push up to higher speeds between stops.
Small changes accumulate across the route, but there is a quite noticeable difference between Old Weston Road and Keele. This is consistent across the day.
Eastbound data show the same relative slowdown from Keele to Old Weston, as well as a comparable dip between Laughton and Caledonia that does not appear in westbound data.